Eurotyp: 6 Tense and Aspect in the Languages of Europe [Reprint 2011 ed.] 9783110197099, 9783110157529

Citation preview

Tense and Aspect in the Languages of Europe

≥

Empirical Approaches to Language Typology Eurotyp

Editors Georg Bossong Bernard Comrie

Mouton de Gruyter Berlin · New York

20-6

Tense and Aspect in the Languages of Europe

edited by Östen Dahl

Mouton de Gruyter Berlin · New York

2000

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.

앝 Printed on acid-free paper which falls within the guidelines of the 앪 ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication-Data Tense and aspect in the languages of Europe / edited by Östen Dahl. p. cm. ⫺ (Empirical approaches to language typology ; 20-6) One of nine vols. published as part of the Typology of Languages in Europe (Project). Includes bibliographical references and index. ISBN 3 11 015752 7 (alk. paper) 1. Europe ⫺ Languages ⫺ Tense. 2. Europe ⫺ Languages ⫺ Aspect. I. Dahl, Östen. II. Typology of Languages in Europe (Project) III. Series. P380.T46 2000 415⫺dc21 00-029180

Die Deutsche Bibliothek ⫺ Cataloging-in-Publication-Data Tense and aspect in the languages of Europe / ed. by Östen Dahl. ⫺ Berlin ; New York : Mouton de Gruyter, 2000 (Empirical approaches to language typology ; 20 : EUROTYP ; 6) ISBN 3-11-015752-7

© Copyright 2000 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Typesetting: Christoph Eyrich, Berlin. Printing: WB-Druck, Rieden/Allgäu. Binding: Lüderitz & Bauer, Berlin. Printed in Germany.

General preface

The present volume is one of a series of nine volumes in which the results of the European research project “Typology of Languages in Europe” (EUROTYP) are published. The initiative for a European project on language typology came from a proposal jointly submitted to the European Science Foundation (ESF) by Johannes Bechert (University of Bremen), Claude Buridant (University of Strasbourg), Martin Harris (University of Salford, now University of Manchester) and Paolo Ramat (University of Pavia). On the basis of this proposal and following consultations with six experts the Standing Committee for the Humanities of the ESF decided to organize a workshop (Rome, January 1988), in which this idea was further explored and developed. The results of this workshop (published by Mouton, 1990) were sufficiently encouraging for the Standing Committee to appoint a preparatory committee and entrust it with the tasks of drawing up a preliminary proposal, of securing interest and participation from a sufficiently large number of scholars and of finding a suitable programme director. The project proposal formulated and sent out by Simon Dik (University of Amsterdam) as chair of this committee met with very supportive and enthusiastic reactions, so that the Standing Committee for the Humanities recommended the funding of a planning stage and the General Assembly of the ESF approved a year zero (1989) for an ESF Programme in Language Typology. During this planning phase all major decisions concerning the management structure and the organisation of the work were taken, i. e., the selection of a programme director, the selection of nine focal areas around which the research was to be organized, the selection of a theme coordinator for each theme and the selection of the advisory committee. The first task of the programme director was to draw up a definitive project proposal, which was supplemented with individual proposals for each theme formulated by the theme coordinators, and this new proposal became the basis of a decision by the ESF to fund the Programme for a period of five years (1990⫺1994). Language typology is the study of regularities, patterns and limits in crosslinguistic variation. The major goal of EUROTYP was to study the patterns and limits of variation in nine focal areas: pragmatic organization of discourse, constituent order, subordination and complementation, adverbial constructions, tense and aspect, noun phrase structure, clitics and word prosodic systems in the languages of Europe. The decision to restrict the investigation to

vi

General preface

the languages of Europe was imposed for purely practical and pragmatic reasons. In the course of the project an attempt was made, however, to make as much sense of this restriction as possible, by characterizing the specific features of European languages against the background of non-European languages and by identifying areal phenomena (Sprachbünde) within Europe. More specifically, the goals of the EUROTYP project included the following: ⫺ to contribute to the analysis of the nine domains singled out as focal areas, to assess patterns and limits of cross-linguistic variation and to offer explanations of the patterns observed. ⫺ to bring linguists from various European countries and from different schools or traditions of linguistics together within a major international project on language typology and in doing so create a new basis for future cooperative ventures within the field of linguistics. More than 100 linguists from more than 20 European countries and the United States participated in the project. ⫺ to promote the field of language typology inside and outside of Europe. More specifically, an attempt was made to subject to typological analysis a large number of new aspects and domains of language which were uncharted territory before. ⫺ to provide new insights into the specific properties of European languages and thus contribute to the characterization of Europe as a linguistic area (Sprachbund). ⫺ to make a contribution to the methodology and the theoretical foundations of typology by developing new forms of cooperation and by assessing the role of inductive generalization and the role of theory construction in language typology. We had a further, more ambitious goal, namely to make a contribution to lingustic theory by uncovering major patterns of variation across an important subset of languages, by providing a large testing ground for theoretical controversies and by further developing certain theories in connection with a variety of languages. The results of our work are documented in the nine final volumes: Pragmatic Organization of Discourse in the Languages of Europe (edited by G. Bernini) Constituent Order in the Languages of Europe (edited by A. Siewierska) Subordination and Complementation in the Languages of Europe (edited by N. Vincent) Actance et Valence dans les langues d l’Europe (edited by J. Feuillet) Adverbial Constructions in the Languages of Europe (edited by J. van der Auwera) Tense and Aspect in the Languages of Europe (edited by Ö. Dahl)

General preface

vii

Noun Phrase Structure in the Languages of Europe (edited by F. Plank) Clitics in the Languages of Europe (edited by H. van Riemsdijk) Word Prosodic Systems in the Languages of Europe (edited by H. van der Hulst) In addition, the EUROTYP Project led to a large number of related activities and publications, too numerous to be listed here. At the end of this preface, I would like to express my profound appreciation to all organizations and individuals who made this project possible. First and foremost, I must mention the European Science Foundation, who funded and supported the Programme. More specifically, I would like to express my appreciation to Christoph Mühlberg, Max Sparreboom and Genevie`ve Schauinger for their constant and efficient support, without which we would not have been able to concentrate on our work. I would, furthermore, like to thank my colleague and assistant, Martin Haspelmath, and indeed all the participants in the Programme for their dedication and hard work. I finally acknowledge with gratitude the crucial role played by Johannes Bechert and Simon Dik in getting this project off the ground. Their illness and untimely deaths deprived us all of two of the project’s major instigators. Berlin, September 1995

Ekkehard König, Programme Director

ix

Preface This volume contains about twenty papers which represent the work of the EUROTYP Theme Group on Tense and Aspect. (The final versions were submitted in 1997, and no substantial updates have been undertaken since then.) I want to thank here first and foremost the authors of the papers, not only for their work but also for the patience they have shown during the long and complex editing process. In addition to the authors, several other people participated in our group meetings and contributed greatly to the discussions: Joan Bybee, Bernard Comrie, Martin Haspelmath, Ekkehard König, Barbara Moltzer, Vladimir Nedjalkov, Nina Niissalo, Svenka Savi´c, Suzanne Schlyter. Obviously, the number of people who have helped us with information on various languages, in particular by filling out our rather extensive questionnaires, is much larger, making it impossible to enumerate them here. A general thanks is extended to them on behalf of all the members of our group. Also, we thank all people who were helpful in the organization of the meetings, in particular the ESF staff in Strassbourg, who were always been willing to assist us when needed. For contributing to the major task of converting our manuscripts into a neat printed volume, we thank Susan Long, who corrected our English, Bernard Comrie and Georg Bossong, the series editors, and the staff of Mouton de Gruyter. Finally, two persons deserve special mention here: the late Simon Dik, without whom there surely would have been no EUROTYP program, and Larissa Bister, our goddaughter, whose birth coincided with the final group meeting in Helsinki. To them we dedicate this volume. Stockholm, March 2000

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Östen Dahl

x

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

xi

Contents

General preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

General Papers Östen Dahl The tense-aspect systems of European languages in a typological perspective

3

Lars Johanson Viewpoint operators in European languages . . . . . . . . . . . . . . . . . .

27

Pier Marco Bertinetto and Denis Delfitto Aspect vs. Actionality: Why they should be kept apart . . . . . . . . . . . . . 189 Eva Hedin The type-referring function of the Imperfective . . . . . . . . . . . . . . . . 227 Rolf Thieroff On the areal distribution of tense-aspect categories in Europe . . . . . . . . . 265 Future Time Reference Östen Dahl The grammar of future time reference in European languages . . . . . . . . . 309 Eva Hedin Future marking in conditional and temporal clauses in Greek . . . . . . . . . 329 Östen Dahl Verbs of becoming as future copulas . . . . . . . . . . . . . . . . . . . . . . 351 The Perfect Jouko Lindstedt The perfect – aspectual, temporal and evidential . . . . . . . . . . . . . . . . 365

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

xii

xii

Contents

Östen Dahl and Eva Hedin Current relevance and event reference . . . . . . . . . . . . . . . . . . . . . 385 Mario Squartini and Pier Marco Bertinetto The Simple and Compound Past in Romance languages . . . . . . . . . . . . 403 Hannu Tommola On the perfect in North Slavic . . . . . . . . . . . . . . . . . . . . . . . . . 441 Nina Graves Macedonian – a language with three perfects? . . . . . . . . . . . . . . . . . 479 Marja Leinonen and Maria Vilkuna Past tenses in Permic languages . . . . . . . . . . . . . . . . . . . . . . . . . 495 The Progressive Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot The progressive in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . 517 Pier Marco Bertinetto The progressive in Romance, as compared with English . . . . . . . . . . . . 559 Karen H. Ebert Progressive markers in Germanic languages . . . . . . . . . . . . . . . . . . 605 Hannu Tommola Progressive aspect in Baltic Finnic . . . . . . . . . . . . . . . . . . . . . . . 655 Casper de Groot The absentive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693 Case Studies Éva Ágnes Csató Some typological features of the viewpoint and tense system in spoken NorthWestern Karaim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723 Karen H. Ebert Aspect in Maltese . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

xiii

Contents

xiii

Appendices 1

The Future Time Reference Questionnaire . . . . . . . . . . . . . . . . . 789

2

The Perfect Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . . . 800

3

Questionnaire on the Progressive Aspect . . . . . . . . . . . . . . . . . . 810

4

List of abbreviations used in interlinear glosses . . . . . . . . . . . . . . 819

5

List of working papers . . . . . . . . . . . . . . . . . . . . . . . . . . . 822

Indices Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 827 Language index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 834 Author index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 841

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

xiv

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

1

General Papers

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

2

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

3

Östen Dahl

The tense-aspect systems of European languages in a typological perspective

1. General According to the original E UROTYP program proposal, Theme Group 6, with the grammatical categories of tense and aspect as its domain, would focus “on the study of grammaticalization processes as manifested in European languages, and on the identification, description and explanation of tendencies peculiar to the tense-aspect systems of European languages”. The following issues were singled out for special attention in the work of the Theme Group: (i) diachronic paths of development; (ii) identification and explanation of areal phenomena; (iii) in-depth studies of individual languages; (iv) language acquisition. Except for the last item, which was planned more as a possible point of contact with other research projects, these issues have all in fact been in focus in the work of our group, something that is hopefully reflected in this volume. The work of the group relied both on the general tradition of tense-aspect research and on earlier contributions of the group members themselves. I shall comment on these two in turn. It is natural that the linguistic phenomena traditionally subsumed under the labels “tense” and “aspect” should have attracted the attention of scholars early on, given their salience in the grammars of most natural languages and their intimate relationship with central cognitive categories. Anyone who undertakes a study in this area has to cope with the burden of an enormous tradition. Paradoxically, however, our EUROTYP Theme Group had relatively little previous work to build on, compared to some other groups in the program. The reason is that, whereas there is an abundant literature on the tense-aspect systems of individual languages as well as works of a general theoretical character, more directly typologically oriented research on tense and aspect is relatively scarce, in spite of notable exceptions such as Friedrich (1974), Comrie (1976, 1985), Chung & Timberlake (1985), and Ultan (1978). There are a number of obvious factors behind this scarcity: the predominantly semantic nature of the problems and the difficulty in finding a suitable framework in which different systems can be compared. Given that several members of the group had considerable research experience within the field, it was hardly to be expected (nor desired) that their theoretical ori-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

4

4

Östen Dahl

entation would be totally homogeneous. Seniority being the most objective criterion of order, precedence goes no doubt to Lars Johanson’s approach, presented already in his thesis on Turkish aspect (Johanson 1971) and further developed in his contribution to this volume. Since I was given the responsibility to organize the work within the group, it will not be too difficult to discern a bias in the original proposal and the ensuing publications towards the claims made in my earlier work as presented in Dahl (1985), my joint paper with Joan Bybee (Bybee & Dahl 1989), and the methodology used in the investigations behind those publications. Less visible in this volume, although also influential, has also been the Functional Grammar of Simon Dik, the late EUROTYP Chairman. Naturally, the work within the group also builds on the earlier investigations of tense-aspect systems of individual languages or language families by group members such as Pier Marco Bertinetto (Romance, particularly Italian), Karen Ebert (Germanic, particularly Frisian), Casper de Groot (Hungarian), Eva Hedin (Greek), Jouko Lindstedt (Slavic, particularly Bulgarian), Rolf Thieroff (German), and Hannu Tommola (contrastive studies Finno-Ugrian:Slavic). We defined three “focal areas” for the work within the group: (I) Future Time Reference; (II) The Perfect; (III) The Progressive. The last two focal areas thus had a major “gram type” as defined below as their object of study. The first focal area, on the other hand, looked at grammatical marking in a semantically defined domain. Still, of course, there was a salient “gram type” also in Focal Area I, namely the future. Methodologically, the three areas were organized in similar ways, the central empirical tool being a questionnaire. Within Focal Area I, a relatively large number of descriptive sketches of individual languages were produced.1 The rest of this introductory chapter will present, as a general background, an outline of the theoretical assumptions behind my own approach to the typology of tense-aspect systems and some of the typological and areal generalizations that can be made about those systems. In addition, the contents of the volume will be summarized.

2. Notes on the methodology of typological investigations In language typology, methodological issues have been a somewhat neglected area, although lately, questions about language sampling and the use of different kinds of data have become more topical. In large-scale typological research, the following main types of data are available: – primary data elicited by questionnaires and similar methods – primary data from corpora of different kinds – secondary data in the form of previously existing descriptions of the languages in question

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

5

The tense-aspect systems of European languages in a typological perspective

5

All three methods have their advantages and drawbacks. One extremely important consideration in typological research is cost in the broad sense of the word – cost in terms of money, other material and immaterial resources and, above all, time. Crudely expressed, in order to get anywhere at all, you have to have “quick and dirty” methods of data collection. Both the questionnaire method and the method of collecting data via grammars should be evaluated in this light. There is really no conflict between the large-scale approach and the in-depth study of individual languages. A large-scale typological investigation necessarily has to be superficial but gives a perspective on the phenomena found in individual languages that you cannot get by looking at them just one by one. Even if most linguists probably agree about this, large-scale typological investigations are still sometimes met with a certain skepticism, especially with regard to the possibility of bias in the heuristics. It is of course true that in some sense you always have to have some idea what you are looking for already in the beginning of a search, and that your expectations will necessarily bias how you interpret data. (Cf. the famous example of the ozone hole at the South Pole, which was initially neglected because the data were filtered away as being too extreme.) A translation questionnaire, that is, a questionnaire in which native speakers of a language L are asked to translate expressions from some other language into L, samples a grammatical domain in a way that has to be guided by the investigator’s initial knowledge of the domain. However, the validity of this sampling is not untestable: if the questionnaire, when applied to a language, fails to elicit examples of forms listed in descriptions of that language, it is clear that the questionnaire has to be somehow modified. Conversely, the adequacy of a grammatical description is tested when a questionnaire is applied to the language: if forms turn up that are not listed in the grammar, we know that the grammar is not adequate. The translation questionnaire method for investigating the use of grammatical and lexical items in languages has one great advantage, which is perhaps not always appreciated, and that, in my opinion, makes up for a number of its drawbacks. It relies on the notion of translational equivalence, which has the nice property that it is operationally definable and thus independent of any linguistic theory, preconceived or otherwise. An utterance in a language can be said to be translationally equivalent to an utterance in another languages if the two utterances are both given as responses to the task of translating one and the same utterance in a third language. The assumption is that translational equivalence in a large set of contexts will be a strong indicator of synonymy, but synonymy is a notion which can only be dealt with within a given semantic theory, and there is also no guarantee that there are no disturbing factors in the translation process. Translational equivalence thus means that two expressions are actually translated the same way by informants, not that there is necessarily any deeper relationship between them. The relationship between the notions of translational equivalence and synonymy would be rather similar to that be-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

6

6

Östen Dahl

tween acceptability and grammaticality, as these notions are commonly understood in linguistic theory. Let us look at a concrete (albeit constructed) example. Suppose that we compare a Swedish and a German translation of Shakespeare’s works and that we find that the English word mean corresponds to either mena or betyda in Swedish and to either meinen or bedeuten in German. Suppose further that we find that whenever the Swedish translator uses mena, the German uses meinen, and whenever the Swedish translator uses betyda, the German uses bedeuten. We are then entitled to say that with respect to these two corpora, the words mena and meinen (or betyda and bedeuten) are translational equivalents. The point here is that we can make this statement without making any claims about the meanings of the words involved. This does not mean, of course, that the fact that the words are translationally equivalent is irrelevant to a description of their meaning; on the contrary, it forms a good point of departure for a further study of them. In a similar way, finding that two grammatical forms are translational equivalents with respect to certain questionnaire material may be highly relevant to the understanding of these forms, but it does not presuppose that we have characterized the meaning of the forms in question in advance. It is thus possible to speak for example of perfects or progressives in different languages on the basis of translational equivalence data without having a theory of the semantics of the perfect or the progressive. The typologist’s dream is to have large tagged multilingual corpora of translated texts in which the distribution of various items could be compared systematically. A corpus-based investigation would have the advantage of eliminating the risk of bias in the material due to theoretical preconceptions. On the other hand, it is of course much more costly. For most languages of the world, the question is not so much how we could create such corpora but rather whether they will be there for us to investigate in a generation or two. Realistically, then, the translation questionnaire method will be with us for some time.

3. “The Bybee & Dahl approach” In the end of the 70’s, I initiated a data-oriented investigation of tense-aspect systems in a large number of languages. Using a translation questionnaire of about 160 sentences, we gathered data about 64 languages in what could probably most aptly be called a “convenience sample”. At the same time, Joan Bybee (together with Revere Perkins and William Pagliuca) conducted an investigation of verbal morphological categories in a controlled sample of 50 languages, using extant grammatical descriptions as the main source of information. The results from these projects were published simultaneously, in Dahl (1985) and Bybee (1985). In spite of the differences in methodology, the results obtained were strikingly similar. In 1989, Joan Bybee

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

7

The tense-aspect systems of European languages in a typological perspective

7

and I published a joint paper in which we tried to integrate the approaches. Recently, Bybee, Perkins, and Pagliuca have presented their joint work in a monograph (Bybee et al. 1994), which is at present the most complete and up-to-date treatment of tense, aspect and modality in a grammaticalization perspective. Although there are differences in details, and sometimes in emphasis, between the individual works listed here, and between the views of the individual researchers, there is a sufficiently well-developed common core for it to make sense to talk about a “Bybee & Dahl approach”. The interpretation I give in this chapter is my own, however. (Cf. also Chapter 1, “Theoretical Background” in Bybee et al. 1994.) The B&D approach differs from most other treatments of tense and aspect in that the basic units of description are not “the category of tense” and “the category of aspect” but rather what we call grams2, i.e., things like Progressive in English, the Passésimple in French etc. Notions like tense, aspect, and mood are seen as ways of characterizing the semantic content of grams, or domains from which their meanings are chosen, but do not, in the typical case, represent structurally significant entities in grammatical systems. Many, if not most, grams combine elements from several domains in their semantics, and it is the rule rather than the exception that grams that would traditionally be treated as belonging to the same category behave very differently with respect to how they are expressed in a language. The term “gram” is intended to be used on a language-specific level, that is, a gram belongs to the grammar of an individual language rather than to the general theory of human languages. In this volume, we follow the practice introduced in Comrie (1976) and write names of language-specific grammatical items with initial capitals, and this practice applies also to grams. We thus speak, for example, of the English gram Progressive. An important tenet of the B&D approach, however, is that tense-aspect grams can crosslinguistically be classified into a relatively small set of types. In a universal theory of grammar, then, the relevant unit is the crosslinguistic gram type, the manifestations of which at the language-specific level is the individual gram. Such gram types should not be thought of as absolute entities – characters chosen from a universal “gram alphabet” – but rather as the statistically most probable clusterings of properties in “grammatical space”, or alternatively, as relatively stable points along the paths of development that grams take in the course of grammaticalization processes, to be further discussed below. Nor should gram types be identified with “notional” or “semantic categories”. It is true that what keeps the grams of a certain gram type together is primarily their semantics, but it is essential that the gram type is not equal to a notion or concept but is a type of grammatical element, which can also be characterized as to its expressional properties: each gram type has a typical mode of expression, directly related to its position in grammaticalization processes. Later on in this paper, I will discuss a further notion, that of a gram family, which finds its application primarily in areal linguistics. A gram family is basically a set of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

8

8

Östen Dahl

language-specific grams that can be hypothesized to have arisen through one and the same historical process – either by being inherited from a common parent language or as a result of language contact. Gram families, then, differ from gram types in having a location in time and space rather than being universally available, as the latter are.

4. Grammaticalization In the B&D approach, we see the study of the grammaticalization (or grammaticization) processes which give rise to tense-aspect grams as an integral part of the general study of those systems, closely intertwined with and often inseparable from the synchronic description. This is not the place to review the rapidly growing literature on grammaticalization: some recent general works that should be mentioned are Lehmann (1982, 1985), Hopper & Traugott (1993), and Heine, Claudi & Hünnemeyer (1991). I shall instead briefly summarize some of the important properties of grammaticalization processes, as they have been identified in the literature, but from a perspective coloured by my own research experience. The “classical” definition of grammaticalization is the one given already by Antoine Meillet in 1912, viz. that “grammaticalization” denotes those diachronic processes by which lexical items develop into grammatical items. Such an understanding of the notion may seem too narrow, however. The emergence of fixed word order, for instance, would only be subsumable under grammaticalization when the position of morphemes which are on their way to becoming grammaticalized is concerned. Yet, we would want to see such processes as a unitary phenomenon. A more generous definition of grammaticalization would generalize it to all processes by which grammatical phenomena develop.3 With respect to the processes that interest us here, namely those that feed tense-aspect systems, the classical understanding of grammaticalization is adequate for the majority of all cases. I shall therefore concentrate my discussion on those. When a lexical item grammaticalizes, changes affect both its content and its form. There is no unanimity in the literature concerning the nature of the semantic changes that are involved in grammaticalization. According to one popular view, grammaticalization essentially means semantic bleaching, that is, the semantic content of the item is partly or wholly lost. Another view emphasizes the role of semantic processes such as metaphor in grammaticalization. A possible synthesis of these might differentiate between the early stages of a grammaticalization process (e.g., the development of full verbs into auxiliaries), which are in many respects rather like lexical semantic change in general, and where metaphor, metonymy and similar processes may play essential roles, and the later stages (e.g., the development of past tenses from perfects), for which terms like semantic bleaching may be more appropriate (Hopper & Traugott 1993, Chapter 4).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

9

The tense-aspect systems of European languages in a typological perspective

9

Semantic bleaching in general increases the domain of applicability of an item, and thus may lead to an increase in frequency. The same effect may also be the result of another process, namely that of obligatorization. The property of being obligatory in certain semantically or syntactically defined contexts is often mentioned as characteristic of grammatical elements. In the area of tense and aspect, we may for instance contrast two kinds of progressives: those with obligatory use, as the English Progressive construction, and those which are normally only optional, as the progressive constructions found in most other Germanic languages (see Ebert’s paper “Progressive markers in Germanic languages” in this volume). In many cases, the crucial property is not so much absolute obligatoriness as something that could perhaps be labeled independence of relevance considerations, that is, that the use of a certain item is governed by factors other than carrying new and relevant information in the utterance context or not. For instance, past tense morphemes in most European languages may not always be obligatory – there are various uses of the present for past time reference – but they are by and large used irrespective of whether the temporal information they contain is necessary for the intended message or not. In particular, grammaticalized morphemes tend to be used even if they duplicate the information carried by some other element in the sentence (in the case of a tense morpheme, temporal adverbials and the like). In fact, it may sometimes be more difficult to omit a past tense marking when it is in principle redundant than when it is not: the combination of a present tense and a deictic adverbial like yesterday is felt as a deviation from the norm. Both semantic bleaching and the diminished reliance on relevance considerations lead to a general decrease in communicative motivation of an item. Thus, when an item is grammaticalized, its content becomes less significant to the communication. As a concrete example of this we may take the development of perfects into pasts. As noted in Dahl (1985: 138), one may postulate a hierarchy of definiteness in temporal reference, such that, ascending it, the probability of using a perfect diminishes. We may distinguish three or perhaps four steps in this hierarchy, with respect to the point in time at which a situation is located: (i) an indefinite time-point in the past, (ii) a time-point located by an overt time adverbial, (iii) a definite time-point presupposed in the context, (iv) a definite time point defined by a narrative context. The communicative motivation for a tense morpheme marking past time reference is arguably less the more easily the time reference is derivable from the context. The hierarchy thus represents a scale of diminishing communicative motivation. The development of a perfect into a past, as it can be witnessed for instance in present-day German, proceeds along this hierarchy, allowing perfect marking for more and more definite time reference. The connection between grammaticalization and communicative motivation is something that has not always been appreciated fully. One reason may be that there is at least seemingly a conflict between the decrease in communicative motivation

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

10

10

Östen Dahl

and the functionalist idea that much of language change is motivated by factors having to do with the communicative function of the items involved. In my view, it is obvious that if we are adequately to describe the “functions” of grammatical items, we need a much wider interpretation of “function”. I shall return to this question shortly. The diachronic perspective makes possible a re-evaluation of the role of prototypes in the semantics of grammatical items. In Dahl (1985), gram types were said to be definable in terms of their semantics, and the different manifestations of a gram type were supposed to share the same prototype. Diachronically, the prototypical uses ought to be the oldest ones, from which the others have developed. To the extent that grammaticalization involves shifts in meaning rather than just ‘semantic bleaching’, that does not exclude the prototype of a gram changing, however. It is reasonable to assume that English will no longer shares its prototype with its origin, the Germanic verb willan ‘to want’. It is also possible that the processes referred to as ‘semantic bleaching’ also tend to make the prototype less salient. One interesting phenomenon that is best understood in the diachronic perspective is that of what can be called “doughnut grams”4, namely grams whose domain has no focus, that is, no prototypical uses. Doughnut grams are in fact quite frequent, and arise naturally in grammaticalization processes whenever two or more grams are generated out of the same source, at differing times. The older gram then has its centre invaded by the younger one, but keeps the periphery for the time being. Typical cases are progressives that develop into imperfectives and then have to yield their old territory to a new progressive formation, resulting in the seemingly paradoxical situation of a progressive having no progressive uses. Doughnut grams are in fact a special case of the more general phenomenon of residual grams, that is, grams whose domain has been reduced by the invasion of another gram. Such developments, in which grams lose rather than gain territory, may look like counterexamples to the generalizations of grammaticalization theory, but are only apparently so, in that such losses are the secondary result of some other well-behaved grammaticalization process. However, it is not excluded that such secondary processes may result in shifts of meaning, that is, that what was in the earlier situation a secondary use becomes a primary one. This would appear to be the case for example with subjunctives, which are often residual categories with original indicative meaning. Probably as a consequence of the semantic or functional changes, grammaticalized items undergo reduction processes of various sorts. Equally important, however, is the loss of autonomy, which, with a maximally brief characterization, means that what is originally an independent word turns into a modification of another word. ‘Modification’ should be understood here in a very general sense: it may be both linear, i.e., expressed by affixation, or nonlinear, i.e., by various other processes, such as stem alternations (e.g., ablaut and umlaut) or change in prosodic patterns. In fact, decrease in linearity of expression can be seen as one important factor in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

11

The tense-aspect systems of European languages in a typological perspective

11

Figure 1. A donut gram

grammaticalization – nonlinear modifications are characteristic of advanced stages of grammaticalization. An idea that has played an important role in the discussion of grammaticalization is that of its unidirectionality (Hopper & Traugott 1993, Chapter 5). It is not exactly clear, however, what the claim that grammaticalization is unidirectional implies, and what would constitute counterevidence to it. It is often taken to mean that there are no processes by which grammatical morphemes develop into lexical ones. This version of the claim is probably untrue but also fairly uninteresting. A more relevant question is whether the individual processes that instantiate grammaticalization are reversible or not. For instance, consider the following statement: “a category can shift from PROG to IPFV or vice versa”.5 What the unidirectionality hypothesis tells us is that there should be no “vice versa”: imperfectives should not turn into progressives. But this statement again needs qualification. It should be noted that on the whole, it is very hard to exclude in a principled way a certain historical development, that is, to establish conclusively that a synchronic state A can never be followed by a synchronic state B. In particular, if we are talking about the possibility of reversing a certain process, it is hard to exclude that some combination of processes might lead to what looks like such a reversal. We might imagine, for instance, that an imperfective might end up as a progressive as a result of there being a number of new grams which happen to take over exactly the nonprogressive uses of the imperfective. But this still does not mean that we have to imperfective grammaticalization process postulate an inverse of the progressive in the theory. A somewhat more troublesome situation is perhaps the following. Suppose there is a language community in which some kind of grammaticalization takes place, e.g., a progressive develops into an imperfective, but in a geographically restricted fashion, resulting in a dialect split, where the gram in question becomes an imperfective in dialect A but remains a progressive in dialect B. Suppose further that due to factors having to do with prestige and other extralinguistic factors, the speakers of dialect A give up most of their dialectal features and adopt what is essentially dialect B. It

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

12

12

Östen Dahl

seems that in such a situation, the grammaticalization process may be reversed in the sense that the speakers of dialect B stop using the gram in question as an imperfective and revert to the less grammaticalized stage where it is only a progressive. What this constructed example shows is the borderline between language change and language shift is extremely problematic. On one hand, we could argue that it is not dialect A that is changing, rather, its speakers are shifting to dialect B. On the other hand, it is clear that changes that we would like our theory to account for – clear cases of grammaticalization – also often spread in a rather similar fashion, by speakers adopting forms from neighbouring dialects with high prestige. There are in fact attested examples of historical developments in which there seems to be a reversal of a grammaticalization process, and which might be accounted for by an explanation of the “sociolinguistic” kind just sketched. In older stages of High German, the perfect auxiliary could be omitted, especially in subordinate clauses, as in the following example: (1)

German (W. Goethe, Faust I, Vorspiel auf dem Theater) Ihr beiden, die ihr mir so oft, In Not und Trübsal, beigestanden, . . . ‘Ye two that have so often stood by me In time of need and tribulation . . . ’ (G. M. Priest’s translation)

In other languages, e.g., Slavic, auxiliary drop (or perhaps rather: copula drop) shows up as one part of the process by which perfects develop into pasts. In Modern German, however, the process has been reversed in the sense that it is in general no longer possible to omit the perfect auxiliary. (Ironically, the principle that the perfect auxiliary may be omitted in subordinate clauses was borrowed in written Swedish, where it has survived and is still operative.) What we have to conclude, I think, in order to maintain the unidirectionality thesis, is that it has to be seen as operating on a fairly high level of abstraction. We cannot exclude that courses of events that look exactly like the reversal of some grammaticalization process sometimes take place. However, we should still be able to do without such reversed processes as independent constructs in our theory. Given the prominent place of various kinds of reduction – semantic and phonological – in grammaticalization processes, it is somewhat tempting to view grammaticalization in general as “linguistic attrition”. I think it is important also to emphasize the positive aspects of grammaticalization: that the object of study is the build-up of grammar, with the focus on systems of inflection. Such systems are a widespread – although not universal – feature of human languages, and it is reasonable to assume that we cannot explain their existence only in terms of the wearing-down of lexical material. Rather, we have to assume that inflection serves a function of its own in language. Exactly what that function is remains to be elucidated, like the question of whether we are somehow genetically predisposed to learning inflections.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

13

The tense-aspect systems of European languages in a typological perspective

13

One issue that has not yet been taken seriously in the study of grammaticalization is whether the items that are subject to grammaticalization processes are really animals of the same kind, in particular, whether a concept like “morpheme” is suitable to cover them all. The concept of a morpheme, a child of structuralism, more or less tacitly presupposes an “item and arrangement” (IA) model of language. Applying the IA model to inflectional morphology is in a way forcing it to look the way it would look if it were just like syntax. In many ways, the two other models discussed in Hockett’s classic paper (1958), “item and process” (IP) and “word and paradigm” (WP) seem more suited for more complex inflectional systems. If this is the case, however, it is a serious challenge for grammaticalization theory to describe how morphemes are integrated into systems which do not really consist of morphemes. I shall return to this somewhat abstract statement in a while and give a more concrete illustration of what I mean. Bybee et al. make the following statement in the introductory chapter to their book: “We do not take the structuralist position that each language represents a tidy system in which units are defined by the oppositions they enter into and the object of study is the internal system the units are supposed to create. Rather, we consider it more profitable to view languages as composed of substance – both semantic substance and phonetic substance.” (1994: 1). It is of course a little risky to associate very specific positions with a large and heterogeneous movement like that of linguistic structuralism, and it may be debated whether you have to believe in “tidy systems” to be a structuralist. The following points may therefore be seen either as marking categorical differences or just shifts of emphasis between grammaticalization-based theories and traditional structuralist approaches: – A grammaticalization-based approach will tend to emphasize precisely the substantive similarities – both in expression and in content – between grammatical items in different languages. The system-internal relations that characterize an item will be seen as resulting from its substantive properties rather than the other way round. When subscribing to the Saussurean slogan that there is nothing in language but differences, linguists have had such phenomena in mind as the fact that the meaning of an unmarked member of an opposition depends on the meaning of the marked member. More precisely, the nonuse of a certain obligatory marking signals that the conditions for that marking are not fulfilled. However, it is important to see that in order to determine this effect, we have to formulate those conditions in the first place. In other words, the paradigmatic relations do in fact presuppose the substantive properties of the items that enter into them. – Grammaticalization processes tend to give rise to situations that do not easily lend themselves to a description in terms of binary oppositions. Thus, grams tend to expand from a point of origin in a wave-like fashion, (metaphorically speaking) chasing each other along a path of development. If one can talk of an “opposition” between an older and a younger gram on the same path, it is rather a secondary

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

14

14

Östen Dahl

effect of the relative positions of the grams. Indeed, due to the multidimensionality of the grammaticalization process, it may not be possible to establish a systematic semantic difference between two such grams. – Like other work inspired by prototype theory, this approach rejects the idea of “invariant meanings” and does not postulate a sharp borderline between conventional and contextual interpretations.

5. Grammaticalization clines Grams show gradualness both synchronically and diachronically. Synchronically, the use of a gram tends to be obligatory in the central (prototypical, focal) uses and optional in the peripheral ones, with sinking propensity of use as we go outwards. We can then talk about grammaticalization clines, that is, ordered sets of contexts along which the frequency of grams decreases monotonically. Good examples of such clines are the Romance de-andative future constructions, discussed in the introduction to Part II of this volume. Such clines may of course involve several dimensions, and most probably do in the majority of cases. Diachronically, the propensity to use a gram in a given context also rises gradually. But we also have to include the geographical point of view here. Since linguistic changes of the kind exemplified by grammaticalization tend to spread outwards from a centre of innovation, the propensity to use a gram in a certain context will decrease as we move away from that point. Reducing grammatical space and real space to one dimension each, we may display a theoretical model of a grammaticalization cline as in Figure 2. In real life, the slopes will probably be less smooth. Still, we could take the graph to be an idealized model of, for instance, the use of the Passato Prossimo in Italian, as described by Squartini & Bertinetto in their paper in Part III of this volume.

6. Gram types in tense-aspect systems Figure 3 shows the major gram types that tend to show up in tense-aspect systems and the most common grammaticalization paths that connect them. I have divided the gram types into core gram types and peripheral gram types, depending on their typical degree of grammaticalization. The core gram types are those that as a rule have morphological (mainly inflectional) modes of expression, and which are also in general characterized by being more or less obligatory in their central uses. The peripheral ones are predominantly expressed periphrastically. The most common inflectional tense-aspect gram types in the world’s languages are imperfective, perfective, past and future. Indeed, it is rather hard to find an inflectional tense-aspect system that lacks all four of them. Of these, the first three

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

15

The tense-aspect systems of European languages in a typological perspective

15

Figure 2. A grammaticalization cline

Figure 3. Major tense-aspect gram types

definitely have predominant inflectional marking; the future is a bit more questionable in this regard. The imperfective and perfective are problematic in other respects, which we now turn to. In most tense-aspect gram types, the marking relations are fairly clear. We have auxiliaries, particles, and affixes marking for example progressive constructions or forms, but there are no morphemes marking nonprogressivity. When it comes to

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

16

16

Östen Dahl

perfectivity and imperfectivity, on the other hand, we find both perfective and imperfective markers. In structuralist terms, we cannot identify one of the members of the opposition as the unmarked one. Moreover, perfective and imperfective verb forms tend to be distinguished from each other by rather more complex devices than many other items in morphology. In languages from all over the world, we find that morphological processes such as ablaut, consonant gradation, reduplication, infixes etc. are used to create stem alternations between perfective and imperfective forms to an extent not found anywhere else in tense-aspect systems. Also, there is often a considerable amount of lexical idiosyncrasy: you cannot predict from one verb to another how the opposition is going to be realized. In view of all this, I shall introduce a new term for the grammatical entity represented by the distinction between perfective and imperfective: I shall label it a hypergram type, more specifically the perfectivity hypergram type, since it appears to be one level higher than the gram types we are talking of in other places in this book, and may in specific languages be realized as grams of different types. We might of course keep the structuralist term “opposition”, but this might give the wrong associations. The interaction between aspectual and temporal elements in the semantics of the core gram types has far-reaching consequences for tense-aspect systems in general. More specifically, there is a coupling between notional perfectivity and past time reference, and notional imperfectivity and present time reference, in the following sense. States and on-going processes are most naturally thought of as holding at or going on at a specific point in time, at which they can be observed. This point in time will, in the default case, be the time of speaking. Completed events, on the other hand, are typically referred to after being completed. States and on-going processes, then, are connected with present time reference, while completed events are connected with past time reference. This connection shows up in tense-aspect systems in several ways: – Many systems (most of them outside Europe) treat different types of verbs in opposite ways: a zero-marked verb form is interpreted as having present time reference if it is stative and as having past time reference if it is dynamic. – In languages with a distinction between perfective and imperfective verb forms – regardless of the marking relations between them – the perfective forms are in the majority of all cases restricted to past time reference, at least when appearing in asserted main clauses. – In those languages that in addition to the perfectivity hypergram also have a past, it is often (probably in the majority of all such languages) restricted to the imperfective, that is, it is what was called a PASTi in Dahl (1985) and a Past Imperfective in Bybee et al. (1994). In such a case, we get what is called a tripartite system in Bybee & Dahl (1989), which is found in a relatively large number of European languages.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

17

The tense-aspect systems of European languages in a typological perspective

Type 0 (common)

No core categories

Type 1 (common)

IMPERFECTIVE

Type 2 (common)

NON-PAST

Type 3 (common)

Type 4 (less common)

PERFECTIVE

PAST

IMPERFECTIVE

NON-PAST

PERFECTIVE

PAST

IMPERFECTIVE

NON-PAST

17

PAST

PERFECTIVE

NON-PAST

PAST

Figure 4. Combinations of core gram types

Figure 4 shows what combinations of the core gram types, perfective, imperfective, and past, are generally found. A number of languages spoken in the eastern part of Europe have tense-aspect systems that seem to fit Type IV systems, said in Figure 4 to be less common. The nature of “Slavic-style aspect”6 has been discussed in Dahl (1985) and Bybee & Dahl (1989). To sum up the essential points, the system found, e.g., in Russian differs from typologically more common manifestations of perfectivity (i) by being less narrowly tied up with time reference, (ii) by displaying a somewhat different semantics, which seems more closely related to Aktionsart or actionality distinctions, traditionally connected with the inherent semantics of the verb as a lexical item. These properties seem to be connected with the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

18

18

Östen Dahl

historical origin of the perfectivity distinctions in question, in that derivational processes and in particular the morphemes called “bounders” in Bybee & Dahl (1989), that is, telicity-inducing verb particles like up and out in English or prefixes like voz- and vy- in Russian. It seems that there is reason to argue that the “Slavicstyle” systems undergo a grammaticalization process the result of which is that they come closer to the more common perfectivity systems: there is thus a clear difference between Russian and some of the West and South Slavic languages in this regard and an even clearer difference relative to the non-Slavic European languages in which bounders are used as perfectivity markers (Latvian, Lithuanian, Hungarian). The last gram type treated as belonging to the core, the future, is discussed in detail elsewhere in this volume (Part 2), since it was in the centre of interest of one of the Focal Areas of the group. Actually, it is only the more advanced futures that deserve being lumped together with the core gram types; “younger” futures, that is, less grammaticalized grams that mark future time reference, for which we might coin the label futuroids, are better seen as belonging to the periphery of tense-aspect systems. On the periphery, we also find important gram types such as the progressive, the perfect and the habitual. The first two of these made up Focal Areas II and III of our group and accordingly, Parts 3 and 4 of this volume treat them from different perspectives. Here, it should be noted that the progressive and the perfect feed the core gram types, each from its direction: the progressive is a main source for marked imperfectives, while the perfect gives rise to perfectives and pasts, and in addition, to various other gram types, such as indirectives and hodiernal pasts.

7. The areal study of tense-aspect systems We shall here discuss areal phenomena on two levels, which we shall call the microlevel and the macrolevel respectively. The microlevel is the one that has been paid most attention in traditional areal linguistics, which centered on the notion of Sprachbund – a set of languages, geographically close but not necessarily genetically related, in which similar grammatical developments can be found. As will be argued in the introduction to Part II of this volume, Sprachbund phenomena are the rule rather than the exception in grammaticalization processes, in that most of these processes tend to spread over several geographically contiguous languages, giving rise to gram families, as defined above. The genetic distance between the members of a pair of languages involved in such a process may vary from one extreme to another – from closely related dialects to totally unrelated languages. It is plausible that the ease with which a process spreads is inversely correlated to this distance, but there is no reason to see influences that

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

19

The tense-aspect systems of European languages in a typological perspective

19

jump the borders between language families as being different in kind rather than just in degree from intrafamilial influences. Europe is an ideal domain for studying micro-areal phenomena, being of the right size to contain a number of “micro-areas” and having a well-documented linguistic history, making it possible to trace synchronic phenomena backwards. Such studies are right now becoming much easier with the advent of computerized corpora both for older and modern texts, although we have not been able to take advantage of these developments to any greater extent within the EUROTYP program. While the microlevel of areal typology can be defined as a level where the relations and mutual influences between individual languages are still discernible, the macrolevel concerns tendencies within larger groups of languages which may be up to the size of a continent. Areal phenomena of this kind have become the object of systematic study only recently, one reason for the newly awakened interest being the insight that many of the phenomena studied in typology have an uneven geographical distribution, which may distort the typologist’s results if neglected. Tense-aspect turns out not to be an exception in this regard. Using my own database and that created by Joan Bybee and her collaborators in the GRAMCATS project, I looked at the global distribution of the major gram types. (For a fuller account of the investigation, see Dahl 1995). In Figure 5, the distribution of pasts and perfectives/imperfectives in the GRAMCATS sample is plotted. If we lump together pasts, past imperfectives and remoteness markings, we can see a clear concentration of those gram types in a few fairly well-delimited areas, the largest one covering the bulk of the western part of the Old World, excluding in particular West Africa. There are also clear concentrations in Australia, New Guinea and some other parts of Oceania, and more scattered occurrences in the Americas. Perfectivity distinctions are more evenly distributed. A few remarks on the relation between typological samples and areal phenomena are in order here. It can be said that, due to the way it is constructed, a sample like the one presented in Bybee et al. (1994) or the similar but larger sample of Nichols (1992) (which comprises 175 languages) is in fact rather unlikely to do justice to areal phenomena. The choice method makes it improbable that two geographically contiguous languages make it into the same sample. Also, the percentage of languages chosen is very low – the GRAMCATS sample comprises roughly one per cent of the world’s languages, which means that each language in the sample is proxy for about one hundred languages. Any grouping that you can discern and that is large enough not to be due to chance will thus comprise several hundred languages. The conclusion is that no areal phenomenon that covers a smaller number of languages can be detected in this way, which, among other things, means that Europe (with its 175 languages) is not really a possible candidate for an area here. On the other hand, it is under these circumstances all the more remarkable that the areal patterns seen in Figure 5 are so clear. A further conclusion to be drawn, then, is that areal influence

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

20

20

Östen Dahl

Symbol

Past (incl. past imperfectives and remoteness markings)

Perfective/imperfective



 







x

Figure 5. Distribution of pasts and perfectives/imperfectives in the GRAMCATS sample

with regard to tense and aspect is strong on both the micro and the macro levels. Future research will hopefully make it possible to integrate the study of the two levels, that is, see how individual grammaticalization processes are related to larger-scale tendencies.

8. Europe vs. the rest of the world As we noted already, Europe is really too small to come out as an area of its own in a sample like the one used in the GRAMCATS project. It is here that the European bias of Dahl’s 1985 sample turns into something of an advantage, in that Europe is covered well enough in that sample for it to be possible to contrast it with the rest of the world. We also refer to Thieroff’s contribution to this volume. For areal generalizations, it appears most suitable to delimit Europe in the more traditional way where it does not include the Caucasus, since the languages spoken

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

21

The tense-aspect systems of European languages in a typological perspective

21

in that area are quite different in a number of respects. With regard to what we called major TMA gram types above, there are some quite clear tendencies within the European area: – All of Europe, including the Caucasus, belongs to the area in Northern Eurasia where pasts are found overwhelmingly according to the data available. – We can discern an area in Northern Europe where there are no grammaticalized perfectivity distinctions. This area includes the Germanic and the northern FinnoUgrian languages. If we look for perfectives and imperfectives of the typologically most common type (as described in Section 7 above), we notice that most of the Slavic and Baltic languages lack this gram type. Instead we find various varieties of what I labelled “Slavic-style aspect” above in this area. – Furthermore, there is another negatively defined area in Northern Europe, partly overlapping with the previous one, viz. the one where there is no grammaticalized future (see Dahl’s introduction to Part II of this volume), comprising the Germanic languages (except English), the Finno-Ugrian languages, and at least the older stages of the Slavic languages. – With respect to the peripheral gram types, we note the high frequency of perfects, especially of the habeo type, that is, perfects derived from a construction involving a transitive verb for possession. This type is hardly documented at all outside Europe. (For further discussion, see the papers in Part 3.) – A particular area which partly falls within Europe is the one where indirectives develop out of perfects/resultatives (see Johanson’s and Lindstedt’s contributions to this volume). – Fully grammaticalized progressives are not particularly frequent in Europe, with the exception of an “Atlantic” area comprising the Iberian peninsula, the British Isles and Iceland (see the papers in Part 4).

9. The structure of this volume The papers in the volume are organized in five Parts, three of which are devoted to the focal areas of the Theme Group. In addition, there is an introductory section containing papers with a more general orientation and a final section containing two case studies on individual languages. In addition to the present introductory paper, Part 1, “General papers”, contains four papers. In “Viewpoint operators in European languages”, Lars Johanson applies the theoretical model he originally presented in Johanson (1971) to the description of what he calls viewpoint operators in European languages, assumed to constitute the cores of the European tense-aspect (in Johanson’s terms, aspectotemporal) systems. Pier Marco Bertinetto and Denis Delfitto, in “Aspect vs. Actionality: Why they should be kept apart”, discuss the traditional “aspect:Aktionsart”

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

22

22

Östen Dahl

distinction, using the label “actionality” for the latter. Rolf Thieroff, in “On the areal distribution of tense-aspect categories in Europe”, looks at the areal distribution of tense-aspect categories within Europe, based on a sample of 40 languages. In “The type-referring function of the Imperfective”, Eva Hedin presents an account of the perfective:imperfective distinction based on concepts type reference and token reference, arguing that some uses of the imperfective that are usually seen as exceptional are instead basic. In Part 2, which treats the first focal area, Future Time Reference, the introductory paper by Östen Dahl, “The Grammar of Future Time Reference in European languages” presents the main grammatical means for future time reference marking in the European languages. Eva Hedin discusses the factors that govern the choice between different verb forms in conditional and temporal clauses with future time reference in Modern Greek. A particular phenomenon pertaining to Northern Europe is treated in Östen Dahl’s “Verbs of becoming as future copulas”. Part 3, “The Perfect”, contains an introduction by Jouko Lindstedt, “The Perfect – Temporal and Aspectual”, which gives a survey of the general characteristics of this crosslinguistic gram type. Current relevance, a key concept in many treatments of the perfect, is discussed in a paper by Östen Dahl and Eva Hedin. Several papers are devoted to the manifestations of the crosslinguistic gram type perfect in individual languages or language groups. The perfect of Old Slavonic has undergone very different developments in the East and West Slavic languages – discussed in Hannu Tommola’s “On the Perfect in North Slavic” – where it has in general developed into a general past tense, and South Slavic, where it is preserved to a larger extent but has acquired an evidential character. A particularly complex picture is offered by Macedonian, treated in Nina Heikkinen’s “Macedonian – a language with three perfects”. Developments similar to those of South Slavic are found in the FinnoUgrian languages Udmurt and Komi, as described by Marja Leinonen and Maria Vilkuna in “Past tenses in Permian languages”. The situation in Romance, surveyed by Pier Marco Bertinetto and Mario Squartini in “Romance Perfects”, is different but no less varied. Part 4, “The Progressive”, is structured in a similar way. A general introduction is given by Pier Marco Bertinetto, Karen Ebert, and Casper de Groot. Surveys of progressives in different areas are given by Hannu Tommola (“Progressive aspect in Baltic Finnic”), Karen Ebert ("Progressive markers in Germanic languages”), and Pier Marco Bertinetto (“The progressive in Romance, as compared with English”). In addition, Casper de Groot surveys the manifestations in a number of European languages of a hitherto neglected phenomenon – “The Absentive”, a construction which is used to indicate that someone is involved in an activity at a place different from the deictic centre. The two languages whose tense-aspect systems are treated in Part 5, “Case studies”, have in common that they show up in a setting which is rather atypical for their

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

23

The tense-aspect systems of European languages in a typological perspective

23

respective language families. Karaim, as described by Éva Csató in “Some typological features of the viewpoint aspect and tense system in spoken NW Karaim” is a Turkic language spoken in Lithuania, i.e., at a considerable distance from other languages of the same group. Maltese (Karen Ebert: “Aspect in Maltese”) is the only indigenous Semitic language in Europe, and the only descendant of Classical Arabic which uses the Latin alphabet.

10. What we have accomplished In spite of the fact that literally thousands of books and articles have been written about tense and aspect in European languages, our undertaking is unique in that we have tried to put the European tense-aspect systems in a consistent typological and diachronic perspective. In this way, we think we have been able to advance the understanding in particular of the dynamics of those systems, how they develop over time and how this is reflected in the rich patterns of synchronic variation. In addition, we have been able to fill in some blanks on the map, both with regard to some neglected phenomena, such as the absentive, and to some less well-studied languages in different corners of Europe.

Notes 1. See Appendix 4 for a list of the working papers of the Theme Group. 2. In Dahl (1985), the term “category” was used in the sense “gram” is used here. The term “gram” was originally coined by William Pagliuca. Whether “gram” should be seen as an abbreviation of “grammatical morpheme” or not, is perhaps a matter of taste; personally I find that one has to stretch the meaning of ‘morpheme’ a bit too much to do that. 3. One problem here is that “grammaticalization” may not be the best term for this more general notion. I have myself considered trying to propagate “grammatogeny”, but as long as this has not gained general acceptance it may be wiser to keep to the old “grammaticalization” even if we use it in a wider sense. 4. The doughnut metaphor was first used in print by Kemmer (1993). 5. The source of this quotation is Dahl (1985: 93). 6. The label “Slavic-style” is unfortunate in that it implies that all Slavic systems look the same. “North Slavic” would be a more adequate label.

References Bybee, Joan L. 1985

Morphology: A Study of the Relation between Meaning and Form. Amsterdam: John Benjamins.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

24

24

Östen Dahl

Bybee, Joan & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Bybee, Joan L., Revere Perkins & William Pagliuca 1994 The Evolution of Grammar. Tense, Aspect, and Modality in the Languages of the World. Chicago / London: University of Chicago Press. Chung, Sandra & Alan Timberlake 1985 “Tense, aspect, and mood”, in: Timothy Shopen (ed.), Language Typology and Syntactic Description, Volume 3: Grammatical Categories and the Lexicon. Cambridge: Cambridge University, 202–258. Comrie, Bernard 1976 Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. 1985 Tense. Cambridge University Press. Csató, Éva 1992 “On some theoretical and methodological problems of the typological study of tenseaspect categories”, in: E UROTYP Working Papers VI: 1: 29–36. Stockholm: University of Stockholm, Department of Linguistics. Dahl, Östen 1985 Tense and Aspect Systems. Oxford: Blackwell. 1995 Areal tendencies in tense-aspect systems. In Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl, Mario Squartini (eds.), Temporal Reference, Aspect and Actionality. Vol. 2: Typological perspectives. Torino: Rosenberg & Sellier, 11–28. Friedrich, Paul 1974 “On aspect theory and Homeric aspect”, International Journal of American Linguistics, vol. 40, No. 4., Part 2, Memoir 28. Gvozdanovic´c, Jadranka & Theo A. J. M. Janssen (eds.) 1991 The Function of Tense in Texts. Amsterdam: North-Holland. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer 1991 Grammaticalization: A Conceptual Framework. Chicago: The University of Chicago Press. Hockett, Charles F. 1958 “Two models of grammatical description”, Word 10: 210–231. Hopper, Paul J. & Elizabeth Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Jakobson, Roman 1966 “Implications of language universals for linguistics”, in: J. H. Greenberg (ed.) Universals of Language. Cambridge, Mass.: The MIT Press, 263–278. Johanson, Lars 1971 Aspekt im Türkischen. Vorstudien zu einer Beschreibung des türkeitürkischen Aspektsystems. Acta Universitatis Upsaliensis, Studia Turcica Upsaliensia 1. Stockholm: Almqvist & Wiksell. Kemmer, Suzanne 1993 The Middle Voice. (Typological studies in language 23). Amsterdam and Philadelphia: Benjamins. Lehmann, Christian 1982 Thoughts on Grammaticalization: A Programmatic Sketch. Vol. I. (Arbeiten des Kölner Universalien-Projekts 48). Köln: Universität zu Köln, Institut für Sprachwissenschaft. [Revised version published 1995 by LINCOM Europa, München.] 1985 “Grammaticalization: synchronic variation and diachronic change”, Lingua e Stile 20: 203–218.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

25

The tense-aspect systems of European languages in a typological perspective

25

Meillet, Antoine 1912 “L’évolution des formes grammaticales”, Scientia (Rivista di Scienza) 12, No. 26, 6. Nedjalkov, Vladimir Petroviˇc (ed.) 1988 Typology of Resultative Constructions. Amsterdam: Benjamins. Nichols, Johanna 1992 Linguistic Diversity in Space and Time. Chicago: The University of Chicago Press. Ultan, Russell 1978 “The nature of future tenses”, in: J. Greenberg (ed.), Universals of Human Language. Vol. 3. Word structure. Stanford: Stanford University Press.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

26

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

27

Lars Johanson

Viewpoint operators in European languages

1. Introduction 1.1. Purposes The present contribution is a survey of viewpoint operators in European languages. These operators are conceived of as representing different concepts of terminality and operating on different actional contents in order to produce meanings within the semantic space of aspectotemporality. It is assumed that they constitute the cores of the European aspectotemporal systems and that it is possible to establish, by generalization from semantic values signalled by the language-specific categories concerned, a restricted set of basic crosslinguistic distinctions sufficient to determine these cores. It is the intent of this study to present a model for differentiated interlingual comparison in viewpoint dimensions, the main questions being how certain basic categories within the aspectual-actional-temporal field are interrelated and by virtue of what values they contribute to aspectotemporal meanings in different languages. Looking at these questions in a way rather different from traditional approaches, I hope to be able to detect essential regularities in the interaction of linguistic values, to understand major similarities and differences between aspectotemporal systems, and to discriminate between certain properties covered by general typological terms such as “perfective” (PFV), “imperfective” (IPFV), “progressive” (PROG) and “perfect” (PF). The approach is in many respects compatible with the semantic tradition represented by Comrie (1976), though it tries to put more accent on properties of formal coding and on system comparison. While differing from genuinely substantialist proposals (Section 3.2), the study also aims at relating the findings to certain results of more ontologically oriented research in aspectology. The foundations of the model were laid in Johanson (1971), a study of Turkish “verbal aspect” in comparison with similar categories of other languages with special attention to the problem of finding a suitable framework in which aspectotemporal systems can be studied from a typological point of view. The present survey is limited to viewpoint characterization of events as expressed by predications based on finite verb forms, disregarding similar notions represented in non-finite items such as converbs and participles. Rather than dealing with a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

28

28

Lars Johanson

handful of well-known representatives of “Standard Average European”, it aims at covering Europe in a broad geographical sense. On the other hand, it does not go into greater detail, does not take dialectal and social variation into account and largely disregards categories outside the core systems. As European aspectotemporal systems are the best studied in the world, there is abundant material available for comparison. The survey is based on extensive materials, data from grammatical studies of different kinds, my own text analyses, and questionnaires testing the use of categories in certain types of situations. However, it cites a limited number of examples – as far as possible condensed to subjectless forms of the third person singular – and gives relatively few explicit references to individual contributions. Existent descriptions are not always quoted in the traditional grammatical or typological terminology they are couched in. If my interpretations deviate from those offered in some of the sources, the reason is that my questions require partly different answers, since the goal lies beyond the scope of language-specific grammar, in an integrated theory of aspectotemporality. The sole purpose of the discussion is to sketch a general framework in which detailed typological descriptive work can be carried out. As this framework is also meant to elucidate how aspectotemporal structures come to function the way they do, the present paper should also be seen as a contribution to a typology of grammatical change. A well-known problem resides in traditional terminological practices, according to which both viewpoint and actional categories represent “verbal aspect” in some sense. It is the contention of the present study that the two types do not represent semantic distinctions of the same kind. Without engaging in nonsensical discussions of “what aspect is” – as if a term should be protected from heretic definitions – it might be claimed that more precise and less iridescent terms are needed to distinguish viewpoint categories from the actional contents they operate on. In the following discussion, however, viewpoint categories will frequently be referred to by the simple term “aspect”, rather than by the tautological expression “viewpoint aspect” introduced in Smith (1991). While not incompatible with pre-Slavistic aspectual theory, this use of “aspect” is partly at variance with the Slavistic tradition. Given the long-standing privilege of Slavistics to define “verbal aspect”, it may, for example, seem provocative to deal, as I shall venture to do here, with the Bulgarian perfective vs. imperfective duality as an actional rather than an aspectual opposition.

1.2. Three dimensions of aspectual terminality To begin with, it might be useful to comment briefly on certain key notions, which will then be discussed at length in the relevant sections. The basic assumption is that a limited set of distinctions is needed to describe the aspectotemporal cores of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

29

Viewpoint operators in European languages

29

European languages. The material available contains a great number of verbal categories representing different ways of conceptualizing and envisaging events in three dimensions of aspectual terminality. The distinctions are based on the following three notions: Intraterminality, envisaging the event within its limits, intra terminos. Postterminality, envisaging the event after the transgression of its relevant limit, post terminum. Adterminality, envisaging the event in the attainment of its relevant limit, ad terminum. Preliminary examples of these notions are Irish bhí ag scríobh ‘was writing’ (intraterminal), Albanian ka shkruar ‘has written’ (postterminal), and Czech napsal ‘wrote, has written, had written’ (adterminal). There is affinity, though by no means identity, between the three terminality notions and the categories “imperfective” (IPFV), “perfect” (PF), “perfective” (PFV) in current typological work on verbal aspect. The notion of relevant limit will be discussed in Section 5. Aspects pertaining to the three dimensions impose different perspectives on events as described by predications. They do not describe an actional content as such, but express how it is conceptualized as occurrence (or negated occurrence). An event, abbreviated E, is an action conceived of as something being or becoming the case in some world. The term is thus used in a broad sense for a realized portion of an action. It includes both transitional events, which change situations, and non-transitional events, which do not. In many modern studies, the term “event” is restricted to the former category. A non-transitional event characterized by internal dynamics will be referred to as a processual one. The terms “state of affairs” and "situation” will be avoided, since they are too easily understood as a general situation described in a text. An event has three main internal phases: a beginning (first limit, terminus initialis, initium), a course (cursus) and an end (second limit, terminus finalis, finis). A global event consists of one or more single basic events. A basic event may thus be a subevent of a global one. A global event may, even if it consists of a set of subevents, be uni-occasional, taking place on one single occasion, at one undivided time interval. A pluri-occasional global event is a set of identical subevents, distributed over several occasions, at clearly separated intervals along the time axis. Thus, in French écrivait chaque jour ‘wrote every day’, the aspect, an intraterminal viewpoint operator, characterizes a pluri-occasional global event consisting of a set of basic events. Aspects are relational in the sense that they present events by relating their limits to some point of view, an orientation point, abbreviated O. Expressed in localistic terms, such points of view can be situated inside or outside the global event. The viewpoint notions are, however, not defined in dependence of time reference or of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

30

30

Lars Johanson

an identifiable O. Aspects invite the decoder to try to identify a contextually relevant point of time as O, but they do not involve any built-in orientation point.

1.3. General framework Before proceeding to the realizations of viewpoint notions, let me briefly indicate their position in a more comprehensive framework. Aspectotemporality is conceived of as a pluri-dimensional space of linguistic concepts comprising aspect, actionality, and temporality. It is a complex phenomenon, realized by interaction of morphosyntax, lexical semantics and pragmatics. Its parameters are subject to considerable variation, the specific contributions varying across individual systems and thus not being universally predictable. However, it is my claim that an adequate description of the interaction of a rather limited number of aspectual, actional and temporal categories will capture the essence of aspectotemporality in European languages. Viewpoint categories represent terminality notions that mirror basic human ways of perceiving and processing events. They interact semantically with those elements of the predication that express the kind of action concerned. I claim that viewpoint operators operate on actional contents and determine them aspectually. This means that viewpoint and actionality parameters interact to the effect that the meanings of the resulting items are interpreted in terms of the scope of the former parameters over the latter. Actionality represents distinctions concerning the inherent phase structure of predications. Items specifying the actional content have no aspectually determining force by themselves. On the other hand, they may develop diachronically into viewpoint operators. Thus, though aspect and actionality tend to be intertwined and closely allied, they are taken here to represent separate, logically independent features, rather than two cognitively identical parameters, two ways of expressing the same semantic distinctions. I do not assume a “semantic domain of aspectuality” that receives either lexical or grammatical expression. The main function of viewpoint operators is not to select phases present in the meanings of lexemes. The interpretation of an aspectually characterized predication involves different elements of taxis and is also dependent on the contribution of contextual elements such as different time adverbials. A further characterization common to European languages is temporal determination by means of grammatical tenses. The ultimate interpretation of aspectotemporality is heavily dependent on pragmatic needs, notably on requirements of the thematic context. In the present survey, however, the main interest will be directed towards determining the semantic contribution of viewpoint operators to the global interpretation of a predication. Their eminent semantic and syntactic functions at the clause-combining level cannot be dealt with here, since this task also requires that predications based on non-finite items be taken into consideration.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

31

Viewpoint operators in European languages

31

2. Viewpoint operators 2.1. Characteristics of viewpoint categories As has been stressed above, the difference between the operators and their operanda, the objects of aspectual perspectivization, is basic to my approach. The distinction between viewpoint values and internal phase structure meanings inherent in actional content enables us to observe and understand their systematic interactions, affinities, and roles in forming central aspectotemporal systems, and the diachronic shifts among them. It was argued in Johanson (1971) that their interactional meanings are unintelligible unless they are taken to belong to different semantic levels. Seiler accepts these considerations as “arguments décisifs” (1993: 24). Similarly, Bertinetto & Delfitto (this volume) stress the theoretical need to separate aspectual and actional content in order to grasp the intricate interplay of notions belonging to different conceptual domains. The task of aspect is thus not to transfer phase structures to actional contents that lack them as inherent properties. The fact that some phase structures prefer or avoid certain viewpoint operators is a matter of semantic fertility and infertility of the combinations in question and does not prove any equality of actional and aspectual meanings. Viewpoint operators offer different choices for envisaging and presenting events as such, for opening perspectives on them and their internal phase structure, for viewing them in relation to their limits. They cannot specify the kind of event described, contribute to its definition, express ontological characteristics, or change the actional content they are chosen to operate on. What is conceived of as one and the same event is presented in different dimensions of terminality. The actional content is left intact and remains identical under different aspects, different ways of viewing the internal constituency of an event (cf. Comrie 1976: 3). Phases not highlighted by ad-, intra- or postterminality are only latent, “concealed”, but not necessarily left unrealized in an objective sense. The traditional description of actionality and aspect as “objective” vs. “subjective” is misleading. Even the choice of the actional content relies upon the encoder’s conceptualization of the event and does not reflect the real world objectively. Events can be presented in subjective ways by actional categories as well. On the other hand, the choice of viewpoint operators is not subjective in terms of expressing the encoder’s attitude or being primarily subject to stylistic aims. The optionality of aspectual choice is often misunderstood. If the viewpoint is determined by context and situation, the view cannot be totally free: there may then be one natural choice only. The operators are used in various discourse types to present events as related to each other and to certain occasions, as successive or overlapping, as cursus- or limitoriented, etc. The restrictions found in coherent narratives are rather systematic and rather similar in different European languages. As is well known, there is usually

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

32

32

Lars Johanson

more freedom of choice if the event is presented as being isolated from a setting, without direct connection to other events. However, claiming that the aspect choice is not independent of what kind of situation is referred to is not equal to maintaining that aspectual meanings are directly related to extralinguistic reality. Aspects do not signal that events themselves have properties that can be described as “imperfective”, “intraterminal”, etc. Objectivism in this sense is as erroneous as the subjectivism it reacts against. It is necessary to find a viable path between the two extreme simplistic positions, the Scylla of "subjectivism” and the Charybdis of “realism”.

2.2. Viewpoint markers and values Viewpoint operators are expressed by viewpoint markers. Most markers are combined aspectotemporal markers that determine events with respect to both aspectual and temporal coordinates (Section 2.3). The following types of expression may be discerned with respect to interaction with actionality categories: (i) A given viewpoint operator may have a special marker. This type, represented by the French Imparfait écrivait ‘wrote, was writing’, produces clear-cut form-meaning correspondences with respect to interaction with actionality categories. (ii) A combined (portmanteau) marker may combine a viewpoint operator with a meaning component of actionality, thus expressing a certain interactional meaning. Two features that behave independently in one language may be fused in another language. Seen from the viewpoint of the latter, the fusion may look like a single feature. This morphological fusion of two categories, represented by the Russian perfective Past napisal ‘wrote’, produces indistinct form-meaning correspondences. Where no morphosyntactic viewpoint marker is available, there is of course no form- meaning correspondence and no systematic way of conveying a viewpoint content. Adterminality (AD), intraterminality (INTRA) and postterminality (POST) will be dealt with as three different semantic values that aspect items may be marked for, if they enter into corresponding language-specific oppositions based on the features AD, INTRA and POST. The following conceptual and terminological distinctions are needed for the marked and unmarked categories: intraterminality vs. nonintraterminality postterminality vs. nonpostterminality adterminality vs. nonadterminality Compared with the traditional system going back to the Stoa, INTRA and POST items might be said to correspond to the so-called “definite tenses” (xrónoi horisménoi). INTRA items are basically “paratatic” (paratatikoí) and POST items

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

33

Viewpoint operators in European languages

33

“syntelic” (syntelikoí), whereas INTRA and POST items broadly correspond to the so-called “indefinite tenses” (xrónoi aóristoi); see Pohlenz (1959: 45–46). Cooccurrence and competition of items is the basis for the formal coding of values within the oppositions and thus for the comparison of systems. Marked members of the oppositions are items with the values intraterminality (INTRA), postterminality (POST), and adterminality (AD), implying the positive notion. Unmarked members are items with the values nonintraterminality ( INTRA), nonpostterminality ( POST), and nonadterminality ( AD), implying negation of, or neutrality towards, the positive notion on a common basis of comparison (Johanson 1971: 32– 34). Items that are not members of such oppositions are indifferent to the values in question. However, an item may also be naturally characterized by a certain value though lacking a competing item in the same temporal stratum (Sections 2.3, 2.4). This kind of neutralization will be marked with the sign Æ , e.g., INTRAÆ , POSTÆ. Viewpoint values serve as ingredients in processes of semantic composition, representing notions on the basis of which complex values are created. Note, however, that they are not conceived of as freely combinable minimal semantic building blocks. A notional system consisting of three oppositions involving three possible values each as basic parameters would indeed be overgenerating. The definitions of the values are generalizations based on different distinctions that may be empirically recognized in existing European languages. The interrelations of the values and the constraints on their combinability will be empirically determined and discussed below. Particular attention will be paid to the pitfalls of naive compositional morpheme-by-morpheme analysis (cf. Johanson 1974a). Numerous European languages are equipped with both INTRA and POST categories, e.g., Romance, Turkic, Iranian, Modern Greek, Bulgarian, Albanian, Caucasian, eastern Finno-Ugrian. POST categories are also present in several Germanic and western Finno-Ugrian languages, etc. On the other hand, AD categories are only found in certain Slavic languages such as Russian, Polish, and Czech. The following examples demonstrate some possibilities of characterizing a past event expressed by a verb with the lexical meaning ‘write’. (Note that all items have complex values that are not fully specified here.)

INTRA INTRA

POST POST

AD AD

items marked for intraterminality, e.g., Armenian grum e¯ r, English was writing items opposed to intraterminality, e.g., Latin scripsit items marked for postterminality, e.g., Norwegian har skrevet, Finnish on kirjoittanut, Komi gižöma items opposed to postterminality, e.g., Norwegian skrev, Finnish kirjoitti, Lithuanian raše, Latvian rakst¯ıja, Komi gizis items marked for adterminality, e.g., Russian napisal items opposed to adterminality, e.g., Russian pisal

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

34

34

Lars Johanson

2.3. Viewpoint and tense Aspectual and temporal meanings may coexist in one and the same item. The analysis of aspectotemporal relations that will be proposed here differs considerably from certain temporalist approaches which first establish “tenses” in Reichenbach’s sense and then try to explain the unexplicable rest of the system in terms of some “aspectual” concept. The results of such analyses often seem to allege unmotivated differences in parts of the core systems where European languages are indeed astonishingly similar. The analysis suggested in the present paper takes the viewpoint perspectives to be primary forms of perceiving and envisaging events, and their temporalization to be a secondary step. Tense has no natural priority over viewpoint, temporalization being just one possible way of determining events. Viewpoint notions generally seem to be older than temporal ones, and some systems, e.g., those of Romance creoles, function entirely on non-temporalized viewpoint distinctions. My analysis is based on the contention that contemporary European languages give precedence to aspectual relations over temporal ones, the so-called tenses being the result of determination of aspectual values in different temporal strata. Thus, what a tense situates on the time-axis is the perspective on the event rather than the event itself (Johanson 1994). The temporal order relation of anteriority holds between two points of time. One of them is the orientation point O, a primary or secondary deictic centre, typically referred to by adverbials such as English just. The second point is the localization point L. The anteriority relation implies that L – but not necessarily the whole event as such – is prior to O. Temporal and aspectual values combine in hierarchic structures, in which one value has scope over the other. Such structures will be represented by means of bracketing, e.g., PAST ( INTRA) Italian scrisse ‘wrote’, PAST (POST) Lithuanian yra paraše¬ s ‘has written’, PAST ( AD) Belarusan pisa˘u ‘wrote, was writing’. 2.3.1. Orientation The primary O, abbreviated Os (where S = ‘speech act’) is the primary deictic centre, the “present world” or “nunc”, the core of which is the moment of encoding. In general, O will by default be interpreted as Os , if there are no contextual indications to the contrary. Relatedness to Os will be referred to as primary orientation. Primarily oriented anteriority implies that L is temporally situated before Os (“absolute past reference”) and thus indicates the situation of the encoder in relation to L. Primarily oriented anteriority and nonanteriority are typically, though not exclusively, expressed by explicit PAST distinctions. The temporal notion involved has scope over the whole predication and situates aspectual and other perspectives on the event expressed.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

35

Viewpoint operators in European languages

35

The value PAST signals remoteness in the sense of dissociation of L from the deictic centre Os , e.g., wrote, was writing, had written. Many PAST items may even express non-temporal remoteness in a counterfactual sense (Johanson 1971: 51– 52). PAST items that are not intraterminal and operate on transformative actional contents may also be used to express a fictive accomplishment of an event, e.g., Modern Greek éfigha, Albanian ika, Turkish gittim, Hungarian már el is mentem (‘I have already gone’ =) ‘I am going now’. PAST implies that no temporal order between Os and an L is signalled, which does not, however, mean temporal indifference. Present tenses are typically unmarked nonanteriors. In many languages, absence of the normal PAST marker, e.g., Turkish idi, Kalmyk bilä, Maltese kien ‘was’, is understood as PAST by default. Primary orientation produces present time readings. PAST can also have future time reference, “futurate meaning”, often with contextual support of a future time expression, e.g., German ruft morgen an ‘will call tomorrow’. Such uses convey relatively strong certainty regarding the actual occurrence, for example, according to some schedule, without the modal shades of meaning often present in prospective items (see below). If the time established by the PAST tense is Os , the given aspectual perspective is presented as valid at this point. The event referred to may be envisaged intraterminally, postterminally, or adterminally: PAST (INTRAÆ ) PAST (POST) PAST (AD)

“intraterminal-in-present” (with present time reference), e.g., French écrit ‘writes, is writing’ “postterminal-in-present” (with past time reference), e.g., English has written “adterminal-in-present” (with future time reference), e.g., Russian napiset ‘will write’

Temporal relatedness to a secondary deictic centre, O2 , will be referred to as secondary orientation. O2 is a contextually relevant temporal point, independent of Os, subjectively chosen or conditioned by the topic (“topic time”, deictic centre of the “text world”), usually localizable by means of time expressions or dependent taxis, i.e. relatedness to other events denoted (Jakobson 1957: 8). It is established by any given item subject to linear successivity. If it is conceived of as a time axis situated in the past (“tunc”), it coincides with the localization point: L = O2 . The anteriority relation establishes a secondary point of view, at which viewpoint perspectives may be presented as valid. The three main aspects may thus be projected into the past as follows:

PAST (INTRA) PAST (POSTÆ) PAST (AD)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

“intraterminal-in-past”, e.g., French écrivait ‘was writing’ “postterminal-in-past”, e.g., Modern Greek exi ghrápsi ‘had written’ “adterminal-in-past”, e.g., Russian napisal ‘wrote’

36

36

Lars Johanson

European languages differ a great deal from each other with respect to anteriority marking. Many mark PAST rather strictly, which means a high degree of grammaticalization in the sense of generalization. (Events located prior to Os can, however, be referred to by PAST items as “historical presents”.) Less temporalized systems, particularly at historically earlier stages of development, dispense with marking intra- and postterminals for PAST. Numerous Indo-European, Finno-Ugrian, Caucasian, Turkic, Mongolian and other languages possessing PAST items may facultatively employ corresponding PAST items in past narratives, i.e., use simple intra- and postterminals instead of intra- and postterminals-in-past, e.g., Turkish ölüyor instead of ölüyordu ‘was dying’ or ölmü¸s instead of ölmü¸stü ‘had died’. This is, for example, the case in those Slavic languages which have preserved a pluperfect (Maslov 1980: 54, 58–59). There are also reduced systems in which one single item represents the past and pre-past strata. No special PAST (POSTÆ) item is available, but only one general POSTÆ item that is used to cover the pre-past stratum as well. Thus, the Hungarian item írt or the Polish item napisał covers the meanings ‘has written / wrote / had written’ etc. by means of one single anteriority marker; cf. Czech vidˇel ‘saw / has seen / had seen’, etc. The Irish Past tense is normally used in a similar way as a general anterior item, e.g., léigh sé an leabhar ‘has read / read / had read the book’. Some languages possess special devices to mark a remote past, e.g., the Kabardian suffix - a-, added to the stem of the Simple Past: txa- a-šˇc ‘wrote long ago / once’. Another possible perspective to be mentioned in connection with temporal interpretations is that of prospectivity, PRO. It presents a non-realized event as foreseen (expected, intended, predicted, etc.) at some O. This projection into the future can be interpreted as relative posteriority (“conceived time”), though many PRO items have modal (epistemic) shades of meaning, presenting the occurrence as less certain. PAST (PRO) items, “prospectives-in-present”, with readings such as ‘shall, will, is supposed to, is expected to write’, are, e.g., Turkish yazacak, German wird schreiben, Bulgarian šte cˇ ete. With primary orientation, such items denote events foreseen to take place after Os . PAST (PRO) items, “prospectives-in-past”, imply that the event is foreseen at an O2 prior to Os (past future, futurum praeteriti), e.g., Turkish yazacaktı, Modern Greek tha éghrafe, Bulgarian šteše da cˇ ete, Swedish skulle skriva ‘would, should, was to, was going to write’. The event may just have been likely to happen at O2 , or it may be known at Os that it has actually taken place. 2.3.2. Temporal interpretations of viewpoint categories Temporal interpretations may also be suggested by aspectual categories. INTRA and AD have a natural affinity with present time reference, POST, AD, INTRA to past time reference. Intraterminality may be interpreted as simultaneity, postterminality as anteriority, etc. Thus, in Maltese, the postterminality of the Perfect and

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

37

Viewpoint operators in European languages

37

the intraterminality of the Imperfect are interpreted as relative past and non-past reference, respectively (cf. Comrie’s discussion of literary Arabic 1976: 78–81). PAST (POST) items, “postterminals-in-past”, relate a postterminal perspective to a past L = O2 and are thus temporally interpretable in terms of two anteriority relations (“past-in-past”). PAST (PRO (POST)) items, “postterminals-in-future”, imply that an O2 is foreseen, at which the relevant limit of the event is transgressed, and may thus be interpretable as posteriority  anteriority (perfectum futuri), e.g., will have written. PAST (PRO (POST)) items, prospective “postterminals-inpast”, imply that, at a past O2 , a later O3 is foreseen, at which the relevant limit of the event is transgressed. It may be interpreted to the effect that the L of an anteriority relation also serves as the L of a posteriority relation (perfectum futuri praeteriti), e.g., Modern Greek tha íxe ghrápsi ‘would have written’. Temporal interpretations of aspects are partly misleading, since the semantic functions are more complex and never strictly temporal (for Turkish, see Johanson 1994). Thus, a PAST (POST) item known as PF is not a simple past tense. On the other hand, it may develop diachronically into a general past. Many European languages possess generalized pasts going back to PAST (POST) items and used for both primarily and secondarily oriented anteriority, e.g., Hungarian írt, Maltese kiteb ‘has written, wrote’. A common meaning of such anteriors is that at least the relevant limit of the event is anterior to O. The following discussions of viewpoint categories will focus on PAST and PAST items and largely disregard prospective items. With respect to the localization of events, three temporal reference strata will be assumed: a present (non-past) stratum, a simple past stratum, and a pre-past stratum. Among competing items in the simple past stratum are PAST items such as wrote and was writing, but also primarily oriented PAST (POST) items such as has written. Items covering the present and pre-past strata are, due to lack of competition, mostly INTRAo and POSTo items.

2.4. Combination of values Combinations of values will be given in hierarchical notations which indicate the scope that the values have over each other. The formulae do not indicate indifference to values of other items. Examples:

PAST (INTRA) PAST (INTRAo )

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

items marked for anteriority and intraterminality, e.g., Turkish yazıyordu ‘wrote, was writing’. nonanterior, naturally intraterminal items lacking competing items in the same temporal stratum, e.g., French écrit ‘writes, is writing’.

38

38

Lars Johanson

PAST (POSTo)

PAST (

naturally postterminal items marked for anteriority and lacking competing items in the same temporal stratum, e.g., German hatte geschrieben ‘had written’. nonpostterminal, nonintraterminal items marked for anteriority, e.g., English wrote. nonanterior items marked for postterminality and competing with an intraterminal in the same temporal stratum, e.g., Armenian grel e¯ ‘has written’.

POST ( INTRA))

PAST (POST ( INTRA))

2.5. Degrees of focality Another concept needed in order to capture the main distinctions observed within the aspectotemporal field is focality, of which intra- and postterminals may display higher or lower degrees. Both categories originate in the narrow “nunc” perspective of the primary deictic Os , the common starting point of aspectotemporal developments (Johanson 1971: chapter 4, 8; 1993). Events that are current or (at least partly) past at Os are naturally represented by intra- and postterminals respectively. Analogous perspectives applied at secondary orientation points ontogenetically derive from this primary “nunc” situation. Now, psychological interest may focus more or less on the situation prevailing at O. Focality is a scalar notion. Intra- and postterminals show higher or lower focality degrees depending on the relative narrowness of the range of vision determined by “nunc”. A rather rough division of the focality scales will suffice for the purposes of the present overview. I shall first distinguish focality (F) from nonfocality (NF), and then assume two degrees, relatively high (HF) vs. relatively low focality (LF). This yields the following subtypes with respect to intra- and postterminal focality:

INTRAF POSTF

INTRAHF INTRALF INTRANF POSTHF POSTLF POSTNF

(INTRAHF (POSTHF

: :

INTRALF) POSTLF)

: :

INTRANF POSTNF

(high-focal intraterminality), e.g., English was writing. (low-focal intraterminality), e.g., French écrivais ‘was writing, wrote’. (nonfocal intraterminality), e.g., Turkish yazardı ‘wrote, would write, used to write’. (high-focal postterminality), e.g., East Armenian grac e¯ ‘has written’. (low-focal postterminality), e.g., Norwegian har skrevet ‘has written’. (nonfocal postterminality), e.g., South German hat geschrieben ‘has written, wrote’.

It should be stressed that the definitions given for INTRA and POST apply, in their full sense, to focal items only. Nonfocals are atrophic items that do not do full justice

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

39

Viewpoint operators in European languages

39

to the original terminal notions but represent them in a weak or diluted way. In a strict sense, we might rather consider them ex-postterminals and ex-intraterminals. Since they are not oppositive items competing with corresponding minus items, INTRANF and POSTNF items might also be written INTRAo and POSTo. Though the degrees indicated above should not be conceived of as absolute positions, INTRAHF roughly corresponds to “progressives”, INTRALF to “continuous” and “habitual” items, and INTRANF to more general items. Similarly, POSTHF may be said to correspond to “statives” and “resultatives”, POSTLF to “perfects” and “constatives”, and POSTNF to more general items. Focal INTRA and POST items imply “dwelling” in a given intra- or postterminal state. Intra- and postterminal items constantly move on the gliding focality scale: their history is characterized by a successive decrease in focality, defocalization processes. INTRAoLF and POSToLF items are frequently replaced by former high-focals. There are also language-specific oppositions with respect to the degree of focality, FOC. Note that if a INTRA or POST item is high-focal, the corresponding INTRA or POST item has a broader use than one opposed to a low-focal. A INTRAHF and a INTRALF item may fuse into an undifferentiated INTRAF item; a POSToHF and a POSToLF item may fuse into a POSTF item. On the other hand, undifferentiated INTRAF or POSTF items may split into high- and low-focals. The high-focals that tend to replace the low-focals are derived items, frequently of an analytical nature. Focality degrees do not determine whether a given item is more or less “aspectual” or “temporal”. All intra- and postterminals are temporally interpretable viewpoint operators. The scalar nature of focality means that if an intra- or postterminal in language A has a more restricted range of uses than a corresponding item in language B, it may have a higher degree of focality. The functional difference does not necessarily imply that A has a specific feature absent in B, but only that the given common feature is represented to a higher degree in A than in B. Note that covering – being used for – situations that allow characterization by higher focality is not equal to expressing a higher degree. Lower items on the scales often cover ontological situation types represented by higher ones. A lower item in language A may be used for situation types that require a higher item in B. Thus, general presents and pasts such as German schreibt and hat geschrieben may cover, without being focal themselves, situations requiring an English is writing and has written respectively. Even within one and the same language, a lower item X may occasionally be used instead of a higher item Y for one and the same situation, without signalling the higher degree.

2.6. Actional content As we have noted, one of the parameters of aspectotemporality is the actional content. Interaction with actional categories is crucial for the semantic and pragmatic

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

40

40

Lars Johanson

realization of viewpoint operators. The event referred to is expressed by a predication, the nominals of which refer to entities and whose predicate core denotes properties of, and relations between, these entities. The predication is assigned an actional content, which concerns the lexical semantics of the predication and includes various qualitative-quantitative properties as objective content restrictions. The actional content will also be referred to as actionality and actional properties. Actional notions are dealt with in the literature under various headings such as “aspect”, “aspectuality”, “aspectual character”, or “Aktionsart” in the classical sense of “die Art und Weise, wie die Handlung des Verbums vor sich geht” (Brugmann 1904: 493). Note again that, in the present framework, the actional content is not taken to include viewpoint distinctions or to represent “aspectual character” of the kind claimed to “rest ultimately upon the same ontological distinctions” as aspect (Lyons 1977: 706). The actional parameters most relevant for viewpoint distinctions are those of the internal phase structure. Whether explicit (overtly signalled) or implicit (covert), they serve as criteria for dividing actional phrases into aspect-sensitive semantic classes (Section 5) and are thus fundamental to a precise understanding of aspectual realizations. Phase structure properties are not, as viewpoint operators, relational in the sense of presenting events in relation to orientation points. They display a good deal of language-specific variation, but also strong crosslinguistic correspondences. As already mentioned, an aspect may be more or less fertile in combination with a given actional content. Aspectoactional combinations will be written with the sign . Thus, the formula PAST (INTRA)  [t] denotes ‘intraterminality-in-past operating on an actional phrase of a transformative phase structure’. Several European languages make systematic use of special modes of action expressing phase structure properties. The explicit marking of transformativity and nontransformativity will be referred to as T-marking and T-marking (Sections 6.3, 6.5). Russian-type perfectives will be dealt with as combined AD- and Tmarkers.

2.7. Core systems, peripheries, preaspectuals The systems of viewpoint operators hold central positions in the grammars of European languages. However, formal elements found in the cores frequently also take a productive part in more or less comprehensive peripheral systems of periphrastic constructions. The high expandability of some systems (Turkish, Estonian, etc.) may even cause difficulties in delimiting the basic inventories of regular grammatical items. In spite of such difficulties, it seems necessary to distinguish, in principle, between more central items and more peripheral ones. One important criterion of aspect grammaticalization is the degree of generalization of the use, often inadequately referred to as the degree of “obligatoriness”.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

41

Viewpoint operators in European languages

41

Relevant questions are whether a given item only applies to certain parts of the lexicon, to what extent it rules out alternative items or can be replaced by other items, etc. Peripheral items have a less generalized use than central ones. For example, several French aspect-like periphrases are still peripheral, preaspectual items outside “la conjugaison française orthodoxe” (Kuryłowicz 1956: 27). Thus, est en train d’écrire ‘is writing’ is not yet a full-fledged PAST (INTRAoHF ) item of the core system. The same is true of vient d’écrire, venait d’écrire, a été en train d’écrire, and avait été en train d’écrire, corresponding to the English core items PAST (POST) ‘has written’, PAST (POST) ‘had written’, PAST (POST (INTRA)) ‘has been writing’ and PAST (POST (INTRA)) ‘had been writing’. Peripheral items often prefer to operate on actional contents of certain types. On the other hand, they may combine with different tense and mood categories. They may be semantically more specific, but even if they express similar notions of terminality as the grammaticalized aspects, they are usually non-oppositional, not taking part in firmly established aspectual oppositions of the kind mentioned. Central, genuine viewpoint items have a highly generalized use, a wider applicability to actional contents, and possess an aspectotemporally determining force, referring exclusively to events, notably global events. Aspectual characterization typically concerns the global event. For example, adterminality does not operate at the subevent level, whereas actional transformativity does. Since there are diachronic developments leading from peripheral constructions without aspectotemporally determining force to highly grammaticalized viewpoint operators, we may in many cases speak of preaspectual items. They do not reach the degree of generalization expected from aspectotenses. When actional concepts are further grammaticalized to express viewpoint, the development typically includes generalization, compatibility with more actional contents. The more limited the applicability of a preaspectual marker is, the longer its way is to the status of a viewpoint operator. For example, certain items originating in completive modes of action pass through limit-specifying stages before becoming AD operators. Other preaspectual items, cursus-specifying statals and items specifying a posttransformational phase, may develop into high-focal INTRA and POST operators. Thus, Turkic postverb constructions converb  auxiliary tur- ‘stand’ are confined to certain lexeme types as markers of actionality, but are freely combinable as intraterminals (-pˇ tur-a) and postterminals (-p tur-ur), e.g., Nogai yazïp tura ‘writes’, Karachai jazïbdï ‘has written’ (Johanson 1995). It may be difficult to distinguish a preaspectual stage from an early aspectual stage, for example, when a T-marker has just become a AD marker, or while an intraterminal or postterminal is still high-focal (“progressive”, “resultative”). Whereas aspectual characterization typically concerns the global event, there are high-focal intraterminals that also operate on subevents. There are also preaspectual constructions expressing high-focal terminality notions without taking part in aspectual op-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

42

42

Lars Johanson

positions, and lexicalizations based on the notions of intra- and postterminality, e.g., English interesting, interested. In some cases it may even be difficult to distinguish preaspectual stages from stages of subsequent defocalization processes.

2.8. Set and non-set events, linear successivity A few further notions relevant to the use of aspects should be added at this point. First, past items can be interpreted more or less diagnostically or historically, as focusing the attention more on the O (orientation point) or on the L (localization point) of the temporal relation. A central discourse function of historical items is to carry the main row of events in narratives subject to linear successivity. Historicaldiagnostic items represent both the event as related to L and its validity at O. In the historical dimension, an event can be presented as more or less set, occurring in a particular setting, or non-set. A temporally set event is conceived of as taking place on a specific occasion, a sequentially set event as part of a chain of events. In a particular setting defined by a narrative, an event may be presented as linked to a preceding and / or a following event. The principle of linear successivity (Johanson 1971: 246–247) implies that several events presented one after another are most naturally interpreted as proceeding in linear time as a temporal sequence in the sense of ‘did x and [then] did y’. This linkage does not, however, imply that the end of one event necessarily coincides with the beginning of the next one. Though not signalled explicitly, the order of events is suggested iconically by the order of the items. Aspectotemporal items are called propulsive if they allow this interpretation, i.e., produce progression in a narrative, and non-propulsive or ruptive if they are unable to advance the plot and thus disrupt the successivity (‘in sequence’ vs. ‘out of sequence’). However, it is often difficult to decide whether linearity suggests a strict temporal sequence or another kind of ordering of events. An event can also be thought of as isolated from a sequential setting, without direct connection to other events. Events outside strict temporal and sequential settings are often relatively open to aspectotemporal conceptualization and display more variation in this respect than set events. Unlike many aspectologists, I do not consider the aspectual oppositions neutralized when their members are applied to such situations.

2.9. Textual behaviour My analysis differs from attempts at equating viewpoint values with discourse functions, reducing them to certain features of the textual behaviour of the items concerned. Oppositions in the viewpoint dimensions serve to express dynamics in discourse structure, for example, to relate events to each other. None seems to be restricted to

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

43

Viewpoint operators in European languages

43

presenting individual events in particular contexts, to expressing “aspect in the narrow sense” as distinguished from “syntactic aspect” (Galton 1962: 18–21). Viewpoint distinctions are basic to the organization of narrative discourse and contribute to assigning aspectotemporal orientations to series of events, to presenting them as non-transitional or transitional, as temporally successive or overlapping, etc. AD, INTRA and POST past items are typically propulsive, used for narration of sequences of events. INTRA, AD and POST items are typically ruptive. The former often fulfill “foregrounding” functions, forming sequences that carry the main narrative line. The latter are often used as “backgrounding” descriptive or commenting devices (Johanson 1971: 234–254; cf. Weinrich 1964; Hopper 1979). High-focal INTRA and POST items are particularly incompatible with narrative sequence. Aspectual-actional-temporal items of different kinds meet similar discourse exigencies in different languages. Though actionality, aspect and tense are categorically distinct, their interactions contribute to textual functions of essentially the same nature. Descriptions of discourse structure thus give indispensable insights into the functioning of viewpoint categories. Aspectoactional combinations produce more or less cursus-oriented or limit- oriented readings and may suggest, without being modes of action, a dwelling in an event, an entry into it, an exit from it, or its mere occurrence. Members of INTRA and AD oppositions are known to fulfill similar textual functions. For analogies between the Turkish and Russian past tense oppositions yazıyordu ‘wrote, was writing’ vs. yazdı ‘wrote’ and pisal ‘wrote, was writing’ vs. napisal ‘wrote’, see Johanson (1971: 93). These oppositions constitute what was referred to as the “main aspect opposition” (“Hauptaspektgegensatz”), used for situation description and situation change, for example, within the so-called “incidential schema” (“Inzidenzschema”; Pollak 1960: 132–133; cf. Bertinetto, Ebert & De Groot, this volume, fn. 8). It is thus clear that essential functions of viewpoint categories are related to the discourse and cannot be described without discourse analysis. It is of utmost importance to describe the connections of viewpoint distinctions with lexical and propositional meanings, notably with the temporal sequence of phases of the actional content. On the other hand, the uses just mentioned do not cover all cases of aspectual relevance and are not sufficient to determine the values of the items in the sense of the pertinent semantic features involved. Viewpoint operators do not just serve the organization of narrative discourse and cannot, as linguistic categories, be assigned meaning in terms of context-dependent functions only. Though they do contribute to the textual functions mentioned, their semantics cannot be equated with their functions within the textual loom of situations, i.e., they cannot be defined in exclusively discourse-pragmatic terms as a means of expressing phases of actions, situation and situation change, succession and parallelity of events. In spite of all affinities between aspects and textual function types, no precise correspondences have been demonstrated. For example, it has not been possible to set up well-defined

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

44

44

Lars Johanson

aspects on the basis of “foregrounding” and “backgrounding” in narrative discourse. Such distinctions seem too vague to define aspectual values. It is thus necessary to detect and define the semantic values that make the textual behaviour possible (Johanson 1971: 246). The typologist should not only register “broadly similar” functions at the textual level, but also try to determine, interpret and explain the similarities and differences observed. Textual functions are not identical to, but fulfilled by virtue of, specific perspectival values, which should therefore be pinpointed as precisely as possible.

2.10. Relations to traditional categories and terms The specific terminality notions suggested within the present framework are intended to cover the aspectotemporal field in a more precise way than typological categories such as IPFV, PFV, PF, PROG seem to do. This, naturally, does not exclude similarities between the two kinds of notions. For example, it is possible to assume basic IPFV vs. PFV correlations offering the option of presenting or viewing an event as ‘a single whole’ or not, i.e. enabling the encoder to describe it, according to what he or she is concerned with, as a totality or as something unfolding, with specific attention to its internal structure (Comrie 1976: 3, 16). It may then be said that PFV and IPFV items are typically used to characterize events textually in an integral (“bounded”) and non-integral (“non-bounded”) way, respectively. However, the somewhat different idea that aspect is characterizable in terms of completion and non- completion often leads to misinterpretations of the aspectual content. Many so-called PFV items capable of conveying complete single events are aspectually unqualified or less qualified (nonpostterminal and / or nonintraterminal) items. With transformative actional contents, even items void of viewpoint content may imply completion. The present framework differs from traditional approaches by distinguishing types of IPFV and PFV items. It also connects certain isolated traditional categories with each other on the focality scales. Thus, statives and resultatives, which are usually not classified as PFV or IPFV (cf. the “ ” items in Kuryłowicz 1956: 27), appear as high-focal postterminals. PROG items are similarly connected with other intraterminals as high-focal variants. Note that the allegedly typical aspect duality ‘progressivity’ vs. ‘totality’ is not realized explicitly in any language-specific opposition, since the value INTRA, which is needed for progressivity, and the value AD, needed for explicit totality, do not occur in the same aspect systems. In traditional aspectological literature, the terms “perfective” and “imperfective” are used in various meanings, corresponding to different notions as distinguished in the present framework: (i) Morphological categories of the Slavic type (Russian, Bulgarian, etc.), in the following referred to as perfective (pf.) and imperfective (ipf.).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

45

Viewpoint operators in European languages

45

(ii) Crosslinguistic aspect types, in the following referred to as PFV and IPFV. (iii) Viewpoint operators of the Russian type, in the following referred to as adterminal (AD) and nonadterminal ( AD). (iv) Viewpoint operators of the Romance type, in the following referred to as intraterminal (INTRA) and nonintraterminal ( INTRA). (v) Actional markers, in the following referred to as T-markers and T-markers. The terms “bounded” and “non-bounded” will not be employed here, since they are also used in various meanings, sometimes indiscriminately for integral vs. nonintegral textual representation, transformative vs. nontransformative actional content, and telic vs. atelic properties of the events referred to. More terminological precision is needed if we are to avoid talking of “bounded” and “non-bounded” situations expressed by “bounded” and “non- bounded” sentences containing “bounded” and “non-bounded” grammatical categories, etc.

3. Crosslinguistic types of categories 3.1. An external tertium comparationis A few comments are necessary on the problem of setting up crosslinguistic types of categories in the conceptual space of aspectotemporality. Languages obviously delimit and divide this space differently, and there is no universally valid viewpoint system. As no two categories occupy exactly the same position, all analogies will prove approximative. How can one determine the types of distinctions suited for interlingual functional comparison? On the one hand, since grammatical meaning is language-specific, the point of departure must be empirical observations on concrete languages, system-based analyses of meaning and use (semantics, pragmatics), relying on linguistic reality. On the other hand, since linguistic values determined within differently structured systems cannot be compared with each other in a direct way, an external tertium comparationis is required for crosslinguistic research on assignment of aspectotemporal functions.

3.2. Notional and situational classifications To arrive at the necessary generalizations, empirical methods are obviously required. One danger potentially threatening all approaches is that preconceived semantic notions are imputed to a given material, so that linguistic facts are violated and adjusted to a foreign scheme. This may be done by aprioristic application of distinctions peculiar to the linguist’s mother-tongue or some well-known language such as Latin, English, or, as frequently in aspectology, a Slavic language. There have been strong

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

46

46

Lars Johanson

tendencies towards hypostatizing members of specific Slavic oppositions as if they represented fixed values and transferring them to other systems (Rundgren 1963: 55–56). Needless to say, no individual language can be taken as the standard of comparison or as the point of departure for generalization. No aspectotemporal category of Irish, Bulgarian, Albanian, or Tatar is identical to any category of English or Russian. But inductive attempts at establishing, by abstraction, general concepts based on language-specific ones may also be dangerous. Language-specific concepts known under labels such as “imperfective”, “perfect”, etc. are not generalizable beyond particular stages of development of individual languages. Empirical bases of comparison that are too narrow may yield types such as the “perfect” (PF), which has few clear representatives in the whole set of languages compared and often a peculiarly vague status even in languages where it does occur. Typically enough, it has sometimes been characterized as a “free-floating gram”. Numerous attempts at defining crosslinguistic types proceed from the question how certain general types of referential meaning are encoded language-specifically, trying to arrive at generalizations anchored outside language, in some ontological, psychological or logical “reality”. One kind of solution is offered by purely notional systems with intensionally defined distinctive features. Even if there may be reasons to assume, behind the diversity of languages, cognitive categories common to all humans, such assumptions do not legitimate aprioristic approaches. The claim that basically the same linguistic features are common to all languages still lacks substantiation. In default of such evidence, interlingual comparison cannot be used to prove, for example, that an autonomous linguistic feature present in languages A and B must also be present in language C, though not formally signalled there. Preconceived metalinguistic schemes based on the application of notions from logical semantics, on one hand, often predict meanings that are not systematically reflected in natural languages, and, on the other hand, fail to predict meanings that are actually expressed. Many typologists operate with ontological classifications, grouping together language-specific grammatical devices in types and subtypes according to the referential range of their applications, i.e., their use to represent certain types of extralinguistic situations. Such approaches are onomasiological, starting from situation types and trying to systematize the possibilities of expressing them. The problem addressed in our case is: “What devices do speakers of European languages use when verbalizing certain types of “aspectual” situations?”. This may even result in certain situations being defined as “PF situations”, "IPFV situations”, etc., something which consequently motivates questions such as “How does language X behave when it needs to express PF, IPFV, etc.?”. A general problem – seldom dealt with explicitly, and mostly solved intuitively – is how to establish the types in question. A wellargued taxonomy of situational contexts is needed if such substantialist approaches are to yield more than trivial results. Statements to the effect that IPFV items denote

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

47

Viewpoint operators in European languages

47

“IPFV situations”, etc., are of course useless unless clear-cut criteria are given for associating the given grammatical devices with extralinguistic types. Purely ontological classifications may lead far from what is actually coded by the devices in question and fail to capture essential differences between the linguistic categories lumped together. It may even turn out that aspectually relevant languagespecific categories do not fit into any of the types unless some of their essential semantic properties are disregarded. Some are only marginally correlated to the type they are claimed to represent, their main language- specific functions being irrelevant to the definition. Some typologists working in the field of aspect and tense operate with “prototypical uses” that only determine semantic foci and leave the peripheries unspecified. Similarly used language-specific items are subsumed under crosslinguistic types (e.g., the “gram-types” in Bybee & Dahl 1989), the semantic structure of which is conceived of as a “radial structure” with a prototype focus and extensions. Needless to say, such approaches require clear criteria both for “similar uses” and for identifying certain uses as central as compared to the rest. If such clear criteria are really available, we may observe cases in which a given item A displays a broad similarity with an item B in the sense that both are used in very much the same set of situational contexts. On the other hand, less adequate criteria may lead us to ignore semantically important uses outside the foci and thus also essential differences between the broadly similar items. The observation that an item A is similar to an item B with respect to certain uses may prove relatively unessential for the semantics of A. The latter may play a clear-cut language-specific role, but still prove to be just marginally correlated to the type it is supposed to instantiate. If A does not exhibit the similarity required, it may be classified as a “default” category, not correlated to any type at all, without prototypical characterization. This may even befall items that have clear semantic profiles in their specific systems, e.g., the Turkic so-called Aorists (e.g., Turkish yazar ‘writes, will write’). Such cases may indicate that the focal uses postulated are insufficient as a basis of classification and that the crosslinguistic types proposed should be reconsidered and defined in a more differentiated way. The use of items in certain types of situations is no doubt an important part of the study of viewpoint operators. However, the expression-function correspondences remain unclear unless the results are put into a semantic framework where they can be compared to the linguistic values of the items involved. The analysis must be supplemented by a further analysis that makes the semantic connections between interrelated categories visible and intelligible. Bybee & Dahl talk of a “gram” as having inherent semantic substance reflecting the history of its development as much as the place it occupies in a synchronic system (1989: 97). A clear consequence of this is that the “substance” must be defined properly. In order to determine the place of a given item in a synchronic system, it is not sufficient to observe its use in certain types of extralinguistic situations and broadly similar uses of items in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

48

48

Lars Johanson

other languages. However, arguing that contextual uses should be distinguished from grammatical meaning is not tantamount to objecting to situational classifications as such. Two points should be stressed to make this standpoint quite clear: (i) the place an item occupies in its system is certainly not a sufficient basis for typological research, and (ii) there is certainly no contradiction between grammatical meaning and use, if the latter is captured adequately. I assume that “gram-type” approaches and system-based ones yield complementary results and consider both necessary for the typological description of European languages (cf. Csató 1992).

3.3. A flexible framework connecting conceptual content with language-specific structures In much contemporary work on aspect, category types are defined in absolute terms and established as fixed functional stations – PFV, IPFV, PROG, PF, etc. – to which aspectual realizations in different languages are linked. The positions are usually not clearly ordered in relation to one another and, if interconnected at all, are connected at best by paths of diachronic development. Such classifications in terms of fixed points may hide important differences between related categories and also fail to account for important common features. In particular, they may create the impression that languages outside the Standard Average European type exhibit less clear-cut categories. Thus, according to current definitions, Turkish lacks both a PF and a PROG, though it obviously possesses closely related categories. The range of variation within the space of aspectotemporality rather calls for approaches that account more properly for differences and similarities and make the category types comparable to each other, intra- and interlingually, rather than representing them as isolated, unconnected points in the space. The kind of framework argued for here should be a more flexible one: a pluri-dimensional space of viewpoint values with definitions formulated in relational and partly scalar terms. In such a framework, Turkish might, for example, possess postterminals with a clear affinity with the PF type and intraterminals that differ from the PROG type by a lower focality degree. For a typology within the space of aspectotemporality, general cognitive-conceptual reference is the necessary tertium comparationis on the basis of which languagespecific categories can be evaluated and compared. A typologist dealing with European viewpoint categories should compare their reference fields, determine which language-specific categories are “broadly similar” with respect to these fields, and set up possible reference types. A linguistically based conceptual network is needed that not only covers certain fixed points in the space of aspectotemporality but can also capture intermediate positions whose relevance is obvious, for example, from diachronic developments of viewpoint operators.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

49

Viewpoint operators in European languages

49

Deductive typological analyses should make use of results gained in inductive research, in detailed system-based descriptive work. Attempts at setting up basic types will yield better results if more attention is paid to immanent structures, concepts systematically expressed in languages by overtly marked or formally detectable covert categories. The basis should be linguistic rather than extralinguistic reality in the sense that the elements constituting the types are the ones typically found in natural languages. Statements on functional similarities and differences between categories should rest on what they actually signal in their systems. It should be asked by virtue of what structural qualities they are similar and / or different. The claim that each language, as a first step, be described in its own terms does not represent a relativist view incompatible with crosslinguistic comparison. Only this procedure will enable us to show that both the conceptual space of aspectotemporality and its structuring are largely common to languages of different types, that the connections of cognitive-conceptual content with language-specific structures are far from random, that the number of connection types is not unlimited, that the relevant language-specific content categories form – on the basis of considerable similarities with respect to their reference fields – a restricted set of grammeme types, and that it might thus be possible to find prototypical connections of cognitive categories with language-specific structures. The obviously strong constraints upon the structures of central aspectotemporal systems of European languages seem rather promising in this respect. An approach of the kind suggested here should avoid common typological fallacies such as equating crosslinguistic and language-specific categories and reducing the levels of description to the effect that important typological features of the languages compared cannot be captured (Csató 1992: 31–32). It should use variation in order to recognize invariants, pertinent common functional denominators. It should establish linguistic values without a conceptual realism that hypostatizes them. The values themselves, however, should be capable of being projected on extralinguistic reality. The goal would be a system of relevant conceptual coordinates determined by different configurations of values. Classifications of categories according to this coordinate system would yield different crosslinguistic types. No such type would be semantically identical to any individual category belonging to it. The values established would by definition be too general to predict the exact uses of the items.

3.4. Viewpoint values The functions of the items studied are products of interacting aspectual, actional and temporal values but also of other factors involved in the communication. The values are conceived of as unique combinatory potentials, relatively context-independent meanings, unifying different uses at a higher level of abstraction. In order to spec-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

50

50

Lars Johanson

ify values, the linguist must try to subtract determining factors, identify meanings common to the majority of uses and formulate them as adequately as possible. The combinatory variants produced by determining factors are instantiations of the relatively invariant meaning. Thus, context-dependent readings of intraterminals such as ‘simultaneity’, ‘inherence’, ‘manner’, and ‘instrumentality’ are all submeanings derivable from the value INTRA. Values in this sense have little in common with caricature pictures of “neat structuralist meanings” of a simplistic kind. They are determined by observing and describing systematic interactional processes. When establishing them, the linguist must neglect certain differences in favour of overall resemblances, but this does not imply that the differences might be overlooked in the total description. The precise language-specific functions and their distribution depend on the number of items and their oppositional configurations. The differences show up in the clearest way in discourse types that allow for maximal competition of the items involved. In contexts where one particular viewpoint operator is the only natural choice, its central meaning is most readily discerned. There may also be marginal uses in which the value seems weak or absent. If a common feature posited fails to unify the uses, prototype semantics, with as adequately defined foci as possible, may prove useful. The values of the items arise within the oppositions they enter into. Since the interaction of values is crucial, no item can be treated “in its own right”. The individual category as such, without connections to others, is not a relevant entity for the study of grammatical meaning. The distinction between semantically marked and unmarked categories is also necessary; see Johanson (1971: 28–36) and Comrie’s remarks on the effects of the functioning of a category as the marked or the unmarked member of a binary aspect opposition (1976: 21, 112). The asymmetry arising from this distinction will have important consequences for the analysis. The marked member, e.g., a INTRA item, represents a differentia specifica on the basis of a common genus proximum, while the unmarked member, e.g., a INTRA item, takes up the space left over by the marked one, representing the absence of the positive concept and thus both a negative and a neutral value (Johanson 1971: 32–35). The marked item signals the plus value, whereas the opposing item lacks this value and gets its weaker values by default. All values derive their precise meanings from the context, but the values of unmarked categories are more dependent on the context than those of the marked ones. Unmarkedness of this kind is a well-known linguistic phenomenon. If, to cite a simple example, the values [young] and [male] are assigned to English duckling and German Enterich ‘drake’ respectively, duck and Ente may be assigned both negative values – [ young] ‘old’ and [ male] ‘female’, respectively – and the corresponding neutral values, i.e., ‘duck’ regardless of age or sex. There is nothing circular about an analysis reckoning with a neutral value in this systematic and predictable sense. The neutral value is expected to realize itself

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

51

Viewpoint operators in European languages

51

in contexts where the feature in question is less relevant or irrelevant. It does not artificially eliminate contradictions to any rule, since it is itself part of the rule. Functions and their distribution thus depend on the number of items interacting. A semantic feature of an item A may depend on the existence of a contrasting item B, so that A cannot be described adequately without regard to B. It is, as we have seen, important whether a PAST item contrasts with other PAST items, e.g., with an adterminal (Russian pisal), with an intraterminal (Turkish yazdı), with a postterminal (Norwegian skrev), with both an intra- and a postterminal (English wrote), or with none at all (South German hat geschrieben). But we have also noted that it is sometimes difficult to decide on questions of competition and contrast with other items, since many categories are less grammaticalized in the sense of having a less generalized use. This is often the case with high-focal intra- and postterminals on the threshold of aspectual function, e.g., German war am Schreiben ‘was writing’. Another case already mentioned is the neutralization arising when an item is naturally characterized by a certain value but lacks competition in the same temporal stratum (ADo , INTRAo , POSTo). Thus, the natural viewpoint of Os -oriented non-pasts is intraterminality, a perspective derived from the primary deictic “nunc” situation, but a contrastive value INTRA can only arise with a competing INTRA item in the present stratum. Similarly, if an item covering the simple past or pre-past stratum has a natural affinity with postterminality but lacks a competitor there, it does not signal POST as a contrastive value. The relationship between aspectotemporal items and the situation types they may cover, i.e., be used for, presents interesting problems. The following sections will include some discussion on possible generalizations concerning the ways of expressing objective situations by means of items signalling certain features. Most of the many unsolved problems concerning “broadly similar” categories cannot, however, be discussed at length here, e.g., questions such as “How does language A, void of category X, express what is typically denoted by X in language B?”. Remember that covering situations that allow characterization by a certain feature is not equal to expressing that very feature. Language A may use Y for situation types that require X in language B. Y may well cover situational areas represented by X without possessing the same value. Thus, in languages lacking INTRA aspect markers, items indifferent to this value may refer to ongoing situations. Y may be semantically more general than the “broadly similar” X, i.e. also be used in cases where B requires Z. This is possible even if Y does not signal any of the values connected with X and Z and is indifferent to both of them. Similarly, within one single language, several items may be used for one and the same situation without possessing the same values. My position thus differs essentially from approaches in which items are, regardless of questions of distinctiveness, assigned PFV, IPFV or PF values if only they occur in certain “perfective”, “imperfective” or “perfect” contexts.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

52

52

Lars Johanson

3.5. Diachronic developments Within various current hypotheses of grammaticalization, functions of grammatical categories are defined in terms of the dynamics of their development. This procedure does not, of course, contradict an analysis in terms of synchronic functional oppositions: the two approaches are indeed complementary to each other. The synchronic part of the task cannot be left out, since pointing to certain positions along diachronic paths naturally requires clear criteria for determining the respective functions. In Bybee & Dahl (1989: 97), the “inherent semantic substance” of the individual “gram” is thought to reflect the history of its development. This may be a correct observation, but it should also be emphasized that the examination of an item in terms of the dynamics of its development presupposes proper synchronic analyses at the relevant stages. Needless to say, in order to decide whether or not it has left the function ‘x’ and is on the path of becoming a ‘y’ item, the linguist must first have defined both ‘x’ and ‘y’. As the functions along a path may be subject to essential changes, it is not uninteresting to determine if a given item is used as ‘tense’, ‘aspect’, ‘mood’, etc. Bybee & Dahl do not consider it necessary to define what “overarching categories” of this kind a given “gram” belongs to. This statement should rightly be interpreted to the effect that it is not always possible to classify a given item unequivocally as belonging to one single category. It does not, however, exclude the necessity of analyzing the meaning of the item in terms of temporal, aspectual and modal elements. The history of European viewpoint operators involves different and often complicated grammaticalization processes. Most of them can be shown to have undergone substantial diachronic changes of different kinds during their careers. The functional developments in the aspectual-actional-temporal field tend to proceed along rather similar lines. Observations of the development of various aspectotemporal systems have led linguists to assume panchronic chains of functional shifts, which will be commented on below. All items tend to extend their uses, losing specific meaning features and assuming more general functions to cover more situational contexts. New items are often introduced to take over the more specific former functions of the old ones. New items entering the dimensions of intra- and postterminality are mostly observed to start their careers as high-focals. The expressions of these functions are more often renewed than others, and, though all European languages possess the necessary material resources, some of them carry out such renewals more often than others. Precise semantic criteria are needed to judge the degree of grammaticalization of individual items. One task is to distinguish between the operators and their actional sources. As is well known, concepts from the wide field of actionality (descriptive, phasal moods of action) are frequently abstracted und grammaticalized to express viewpoint notions. These gradual processes involve intermediary stages that are sometimes difficult to determine adequately. The boundary between actional and

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

53

Viewpoint operators in European languages

53

aspectual function may even be blurred by formal fusion (Section 2.2). In general, however, the stages exhibit unmistakable characteristics. The different values within the aspectual-actional-temporal field may be encoded by many various morphosyntactic means. The modes of expression include complex predicates, auxiliaries, adverbs, case marking, flexion, derivation, and merger with the verb stem (Seiler 1993: 21). Morphology often plays an undeservedly central part in the discussions on functions. Since formal items change their functions, no values are tied to specific modes of expression, and statements on allegedly typical expressions of aspect and actionality do not always correspond to the facts. However, the formal development of viewpoint markers typically goes from lexical constructions via periphrastic constructions to inflectional ones. New items emerge from the lexical potential – transformative, frequentative, iterative, completive, stative, and other items – and existent verb forms. If the points of departure are known, the itineraries leading to viewpoint operators are also largely predictable. Such processes will also be briefly commented on in the following sections.

4. Actional content 4.1. Ontological classification of events Viewpoint operators apply to actional contents of different types and do not themselves signal any ontological properties of events. Aspectual values should not be hypostatized and interpreted in terms of actionality. For example, though IPFV – in the sense of INTRA or AD – is often taken to express durativity, iterativity, habituality, continuativity, etc., such readings depend on the actional content itself and not on the view applied to the event. Events of different ontological types may, with certain restrictions that will be discussed below, be envisaged intraterminally, postterminally or adterminally. Iterative and pluri-occasional (“habitual”) readings of aspectual items must thus be distinguished from explicit means of signalling such meanings. Repetition is neither a viewpoint value itself, nor systematically linked to any such value. A set of repeated events can be envisaged as INTRA, INTRA, AD, AD, POST, POST, or represented without any aspectual characterization. Serial readings, henceforth indicated by [ser], are suggested implicitly, or signalled explicitly by modes of action or by contextual elements such as adverbial modifiers denoting cyclic time (daily, every year), frequency (often, seldom) and habituality (usually, always). Whether the number of occurrences is undetermined or overtly quantified (‘X times’) may have consequences for the choice of aspect; for example, Russian AD may be used in the latter case. A pluri-occasional global event may be conceived of as a state in the sense of a habit with an undetermined number of occurrences. This habituality

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

54

54

Lars Johanson

is compatible with different aspects, and the fact that INTRA and AD are often preferred to express it does not mean that it is part of the IPFV semantics. In French lisait chaque jour, Turkish her gün okuyordu, Bulgarian vseki den cˇ eteše ‘read every day’, a global event, consisting of repeated portions of ‘reading’ distributed over separated intervals, is envisaged intraterminally at an O2 . However, habits can also be envisaged in other ways. Thus, in the sentence just cited, English prefers the Simple Past read, because its intraterminal item was reading signals high focality. (For focality degrees and pluri-occasionality, see 7.3.) Type-referring, potential, or dispositive readings of aspectual items must also be distinguished from explicit signals of such meanings. Viewpoint operators do not themselves signal such distinctions, e.g., differences between generic and nongeneric reference, events conceived of as types and as tokens. It is difficult to follow Hedin’s proposal (this volume) that IPFV – in the sense of our INTRA and AD categories – is type-referring and thus not used to envisage particular instantiations of events in time. However, strong defocalization of INTRA items may produce dispositive and other similar modal readings (7.8). It is often claimed that PFV – in the sense of AD and INTRA – expresses transitional (situation-changing) events (“achievements”, “accomplishments”), whereas IPFV – in the sense of INTRA and AD – expresses non-transitional events (“processes”, “states”). However, it is not a pertinent function of viewpoint operators to signal such ontological categories. Something that might be conceived of as a “state” or a “change” can be viewed in various aspectual perspectives. Definitions of the kind mentioned follow from equating aspect values with discourse functions. Narrative settings suggest sequences of transitional and non-transitional events. A transitional event leads to a change in the state of affairs, a leap into a new situation in the relevant text world. A non-transitional event occurs without producing such a change. Though it is a typical discourse function of IPFV items such as the Russian imperfective Past or the Romance Imperfect to stand for non-transitional events, they may also be used for transitional ones. If PFV is taken to signal transition, many linguistic facts become difficult to account for. Though AD is dynamic and tends to express changes on the basis of a given state, this is not always the case with INTRA items. The latter not only indicate that something ‘becomes the case’, but can also refer to non-transitional events, to something that ‘remains the case’ or simply ‘is the case’. They may well express ‘states’ or ‘processes’ prevailing for a certain time, e.g., French a dormi deux heures, Modern Greek kimíthike dhío óres, Turkish iki saat uyudu, Bulgarian pospa dva cˇ asa ‘slept for two hours’. The conflation of aspectual and ontological meaning may lead to confusing classifications. A consequence of Lyons’ analysis (1977: 709–710) is that the French passé simple régna ‘reigned’ in régna pendant trente ans ‘reigned for thirty years’ could be characterized as a “process verb” with respect to “aspectual character”, since it is “durative”, but as an “event verb” with respect to “aspect proper”, since it is “punctual” (cf. Bache’s

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

55

Viewpoint operators in European languages

55

justified critique 1982: 63). With INTRA items, the main factor is not ‘change’ or ‘transition’, but absence of the intraterminal perspective.

4.2. Modes of action Modes of action, expressed by periphrastic or derivational markers, have functions similar to adverbial elements. They modify the meaning of the basic actional phrase, deriving new actional contents from more basic ones. The markers may be preverbs, as in Indo-European languages, or, as in Turkic and Mongolian, postverbs, consisting of a converb suffix and a desemanticized auxiliary verb, e.g., Kalmyk biˇcˇj av‘copy’ (‘take writing’). They prefer actional contents of certain semantic types, not displaying the degree of generalization typical of aspectotenses. The actional properties signalled are of a qualitative or quantitative nature. Descriptive and procedural markers specify the kind or manner of development, e.g., a certain kind of ‘writing’ such as ‘rewrite, copy’: Norwegian skrive om, English re-write, Russian pere-pisat’, Lithuanian per-rašyti. Quantificational markers signal properties of frequency, duration and degree of accomplishment such as iterative, frequentative, semelfactive, durative, delimitative, perdurative, attenuative, completive. Some are of particular relevance for the realizations of aspect. Iteratives signal that the action consists of repeated acts and are often used to exper-rašyti ‘rewrite’, Chepress pluri-occasionality, e.g., Lithuanian per-rašin˙eti mala ‘drink’. Special devices for signalling pluri-occasionality (‘hachen miyla bituality’, ‘nonactuality’, etc.) should be distinguished from pluri-occasional readings of low- or nonfocal intraterminals, which may also cover the referential areas of higher items (7.3.2). Slavic languages use secondary imperfective formations such ipf. cˇ itat’ ‘reads’, e.g., cˇ ityval ètu knigu as Russian cˇ ityvat’ ‘read repeatedly’ ‘has (on several occasions) read in this book’. Such explicit [ser] markers may also combine with other modes of action, e.g., poˇcityvaet ‘repeatedly reads a little’ poˇcitaet ‘reads a little’. Certain Slavic languages make systematic use of iteraipf. psát ‘write’. Bulgarian tives derived from imperfectives, e.g., Czech psávat possesses one such verb, which may function as a specialized pluri-occasionality marker, biva ‘(usually) is’. Some devices, e.g., the Lithuanian -dav- frequentatives of the type rašydavo ‘used to write’, are clearly pluri-occasional and not ‘habitual’ in a sense that would include events without separated localization intervals, e.g., used to live there. Some other devices, e.g., the English used to periphrasis, may also cover events which are not pluri-occasional, do not qualify as habits in any normal sense of the word, but rather represent permanent properties of the subject referent, e.g., The Temple of Diana used to stand at Ephesus (Comrie 1976: 28, cf. Macaulay 1978). Among similar actional devices are the Irish constructions with bíonn ‘is usually’ and bíodh

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

56

56

Lars Johanson

‘used to be’. Many other devices signal both pluri-occasional and permanent actions, e.g., Armenian periphrases with the auxiliary linel ‘be, be repeatedly, usually be’, Karachai alïwˇcandï ‘usually takes’, alïwˇcan edi ‘used to take’, Kalmyk irdg ‘usually comes’, irdg bilä ‘used to come’. Several pluri-occasional devices are restricted to the past stratum, e.g., English used to and Yiddish fleg periphrases, Lithuanian -davfrequentatives, Turkic finite items in -a tur an. Some languages possess special markers of dispositive meaning, interpretable as pluri-occasionality, habituality, potentiality, or future time reference. Maltese ikun is formally a nonfocal Imperfect of ‘be’, and its semantic properties derive from this source. Compare Turkic items of the type bolur ‘may be, is possible’, developed from “Aorists” of bol- ‘become, be’. Combinations of PLUR (pluri-occasionality) and DISP (disposition) markers with T-marking and with INTRA and POST operators will be commented on below. Delimitative and perdurative modes of action include in their actional content a crucial limit as a measure of minimal-maximal extension. Delimitatives pose a temporal limit to the action: ‘for [not longer than] a certain period of time’, often with the meaning ‘spend [a certain period of time] V-ing’, e.g., Russian poˇcitat’ ‘read for a while’. Perduratives express an action carried out ‘a whole entity / period through’, e.g., Russian proˇcitat’ ‘read through’, progovorit’ ‘talk for an entire period of time’. Both Russian types can be imperfectivized to express iterativity, e.g., poˇcityvat’ ‘read repeatedly for a while’, proˇcityvat’ ‘read through repeatedly’. Continuative modes of action signal the continuation of a given action, ‘keep (on) / continue V-ing’, e.g., Kalmyk umš-jˇa ‘go on reading’, Yiddish haltn in eyn shraybn ‘keep on writing’ (Ebert, this volume). Certain modes of action are preaspectual items, developing diachronically into viewpoint operators. Completives may play important roles in T-marking, explicit marking of transformativity (Section 6.3), and develop into AD items. They do not specify a final phase, but signal ‘V thoroughly, to completion’, e.g., Gothic gafulljan ‘fill to completion’, Hungarian meg-ír- ‘write (and finish writing), write to completion’, German auf-essen ‘eat up’. Some European languages such as Slavic, Baltic, Hungarian, Kartvelian, and Ossetic make systematic use of completives for T-marking. Phasal modes of action are not perspectival and relational in the sense of presenting the limits of an event in relation to orientation points. Many of them specify one inherent phase of the undifferentiated actional content denoted by the corresponding unmarked actional phrase, i.e., select the beginning, the course or the end. The selection is often done by means of phasal verbs such as begin, proceed, finish, or special lexicalizations. Many languages possess phasal pre- and postverbs which disambiguate ambiguous actional contents by excluding certain readings, e.g., English sit down, Kalmyk unt-jˇ od- ‘fall asleep’, unt-jˇ kevt- ‘sleep’, Russian u-znat’ ‘get to know’. Phases of particular cognitive saliency or social relevance are more likely to

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

57

Viewpoint operators in European languages

57

be specified than others. Note that a phasal mode of action picks out a part of the potential content of a given actional phrase and that this part can be conceived of as denoting an action of its own, to which, for example, new phasal modes of action can also apply. The dynamic initial phase of an actional content may be distinguished from the subsequent statal phase by means of ingressive, initium-specifying markers meaning ‘enter, begin, come to perform the action’ (‘start V-ing’, ‘begin to V’), e.g., Lithuanian imti rašyti ‘begin writing’. Pre- and postverb constructions are often found with actional contents of a cognitively salient initium. Thus, Serbian do-znati specifies the entrance into the state of ‘knowing’: ‘get to know, come to know, learn, acquire knowledge’. Other examples: Russian u-videt’ ‘catch sight of’, po-ljubit’ ‘take a liking to’, za-plakat’ ‘start to cry’, Hungarian le-ül ‘sit down’, Tatar tot-ïp al- ‘seize’. Egressive, finis-specifying meanings may be expressed by phasal verbs meaning ‘finish’, etc. There are, however, few if any egressive pre- or postverb constructions specifying the dynamic end phase of an actional content in the sense of ‘conclude, leave the action’. Egressives differ from completives, which do not just specify a final phase: ‘finish writing’ does not denote the same action as ‘write to completion’. Statal or progressive, cursus-specifying modes of action operate on actional contents conceived of as having a salient cursus and exclude limit-oriented readings. They often go back to iteratives or duratives and may combine such functions with statal functions, ‘be busy V-ing’, e.g., Swedish hålla på och skriva ‘keep writing, be writing’. Some also allow perdurative, continuative, or habitual interpretations. Some are based on locative metaphors, using elements meaning ‘at’ or ‘in’, e.g., German am Schreiben sein, Danish være ved at skrive. Others are locomotive constructions based on metaphors of movement (‘move, go, run, come V-ing’), e.g., Italian periphrases with andare ‘go’, venire ‘come’ or Tatar complexes with yörˇe- ‘move, run, go’. Some are postural verb constructions based on body position metaphors (‘stand’, ‘sit’, ‘lie’, etc.), e.g., Italian periphrases with stare ‘be (situated)’, Swedish sitta (och) ‘sit (and)’, Tatar tˇor- ‘stand’, utïr- ‘sit’, yat- ‘lie’, Kalmyk kevt- ‘lie’. The auxiliaries either preserve some of their lexical meanings, delimiting the action to certain body positions, or they are desemanticized and thus interchangeable. Statals play important parts as T-markers (Section 6.5). Poststatal markers express an evolutional stage following upon the basic action, ‘just have V-ed’, e.g., French venir de, Catalan acabar de, Icelandic vera [ny] búinn að, signalling that the event is immediately prior to an O. Such actional items are often based on locative or movement metaphors (‘be after doing’, ‘come from doing’) and may be observed as preaspectuals developing diachronically into POST operators, e.g., Welsh mae wedi yn darllen ‘is after being in reading’ ‘has read’, Irish tá tar eis a scríobh ‘is after writing’ ‘has written’. Prestatal markers express a stage prior to the basic action, ‘be about to V’, ‘tend to V’, etc.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

58

58

Lars Johanson

5. Internal phase structure 5.1. Aspect-sensitive actional categories Actional content parameters of particular relevance for viewpoint realizations are subsumed under the internal phase structure (IPS). They do not concern the perspective applied to an event, but constitute aspect-sensitive actional categories basic to the use of viewpoint operators as terminality categories. In their interaction with aspect grammar, they clearly show their categorial independence within the field of aspectuality. Phase structure properties such as [t], [tf] and [mom] are implicit or explicit features of the internal constituency of the actional content. The following is an attempt to determine basic phase structure values in European languages and to distinguish overt as well as covert actional categories on the basis of their way of reacting to aspects. The resulting categories show strong similarities across the languages under study. Though distributed in different ways, the distinctions mirror important differences with respect to the cognitive relevance of the phases of actions.

5.2. Categorization The following categorization covers relevant phase distinctions in a variety of European languages. IPS category Transformative [t] Finitransformative [mom] [ mom] Initiotransformative Non-transformative [ t] [dyn] [ dyn]

The actional content is conceptualized as implying transformation as implying final transformation without a salient cursus with a salient cursus as implying initial transformation without transformation as dynamic as static

This scheme allows five basic categories to be distinguished: (i) momentaneous finitransformatives, (ii) non-momentaneous finitransformatives, (iii) initiotransformatives, (iv) dynamic nontransformatives, and (v) non-dynamic nontransformatives. The five classes may be ordered according to their degree of limit-orientation: [tf, mom], [tf, mom], [ti], [ t, dyn], [ t, dyn]. The classification goes back to a categorization of Turkish actional phrases based on formal tests (Johanson 1971: 194–233). It differs considerably from the classifications proposed by Vendler (1967) and Dowty (1972), which concern situation types conveyed by the whole sentential context and do not distinguish between viewpoint and actionality. Breu’s and Sasse’s division of aspectually relevant actions into

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

59

Viewpoint operators in European languages

59

five classes (Breu 1984, Sasse 1991a, 1991b) starts from processual, stative and terminative actions and divides the last two classes further into two subclasses each. My basic distinction is the one between transformativity [t], divided into [tf] and [ti], and nontransformativity. The main difference from most other approaches is that I classify linguistic units expressing actions rather than actions as such. The units classified are not verbs, but actional phrases, consisting minimally of a verbal lexeme, which may change their phase structure by way of recategorization (Section 6). The three possible phases – the two limits (initium, finis) and the intermediate cursus – show different degrees of saliency in the types mentioned. For example, each actional content has a relevant limit (¢), which varies according to the phase structure type. With nontransformatives, it is identical to the initial limit of the action. With transformatives, it is the crucial limit (ª), at which the transformation takes place. Graphically: [tf] [ti] [ t]

ª ¢

ª

5.2.1. [¦t] The features will now be discussed in some detail. The basic classificatory criterion in natural languages of different types is transformativity. An actional phrase is transformative [t] if the action designated by it has a natural evolutional turning point, a crucial initial or final limit ª. Depending on the actional phrase, this limit may be the end or beginning of the action or even constitute the whole action. A nontransformative [ t] actional phrase does not imply any such limit. Transformativity is not a vague notion of “some change in the world” and does not simply mean ‘containing an endpoint’, which might apply to any event. The actional content of transformatives comprises a culmination point at which a transformation takes place. They typically refer to telic (“desinent”, “bounded”, “cyclic”, “terminative”) actions, which by nature contain an inherent final limit indicating an evolutional minimum-maximum, and, if fully achieved, reach this built-in endpoint. Non- momentaneous transformatives thus have a heterogeneous and dynamic actional content. Note, however, that telic actions may be referred to by both initioand finitransformatives. The crucial limit may be the “left” or “right” boundary of the actional content expressed by the actional phrase. On the other hand, transformativity is not tantamount to telicity. The terms transformative and nontransformative refer to properties of the actional phrases, whereas the terms telic and atelic will be reserved for properties of the actions themselves. For example, an initiotransformative actional phrase denotes both an initial telic and a following atelic action.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

60

60

Lars Johanson

Since viewpoint operators present limits of the events as attained (AD), transgressed (POST), or concealed (INTRA), information concerning the presence of a crucial limit in the actional content is important for the interpretation. However, this presence does not necessarily mean that the crucial limit is focused upon and that the cursus is less important. Whether the crucial limit is highlighted or not, is a strictly aspectual matter. The complete representation of a telic event may produce a change leading to new states of affairs, but the use of a transformative actional phrase to describe an event does not necessarily imply that the transformation takes place. Even if the action expressed is not considered to be fully carried out unless the crucial limit is reached, this does not mean that it is conceived of as non-occurring. Even combined with an element meaning ‘almost’, transformatives do not necessarily imply – as nontransformatives do – that the action does not take place at all. The ‘almost’ element refers to the attainment of the crucial limit and does not exclude the occurrence of possible portions of action preceding that limit. Reference to telic events does not necessarily include the endpoint. Any event can be presented from within, so that the finis is not envisaged. Transformativity is not identical to PFV in the sense of AD or INTRA. It should be stressed that even items completely void of aspectual meaning may suggest completion if the actional content is transformative. An implicit [t] distinction underlies the old Indo-European actional classification of Aorist and Present stems, originally without any special markers. Both stem types were indifferent towards intraterminality and constituted the nonpostterminal member of an opposition with the postterminal Perfect. Languages may possess items that are indifferent towards intraterminality but typically interpreted as ‘ongoing’ with [ t] actional contents and as ‘accomplished’ with [t] actional contents. For example, Nenets exhibits a neutral item with this natural differentiation. (For pidgin and creole items of this kind, see Bickerton 1975.) 5.2.1.1. [tf] Transformatives may differ from each other with respect to which phase constitutes the crucial limit. With finitransformatives [tf], the end of the actional content is conceptualized as the inherent evolutional turning point, with whose attainment a leap into a new state occurs. A [tf] content as expressed by actional phrases such as reach or die is conceived of as moving towards a natural conclusion. Though it is heading for completion, it is not necessarily envisaged as completed. The inner goal of the action must be distinguished from the endpoint of the event. The use of a [tf] actional phrase that signals full achievement does not necessarily imply that the corresponding event is fully achieved. The action is not fully carried out unless the final limit is reached, but it can be conceived of as going on before this point, e.g., viewed during the cursus leading up to it. Note that ingressive verbs signalling an entering phase, e.g., Russian za-igrat’ ‘start playing’, are also [tf] actional phrases.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

61

Viewpoint operators in European languages

61

If a past tense is used for an event described with a [tf] item, it means that its finis does not occur later than at O. If the action has already been carried out, it cannot go on any more. Thus, [tf] items do not occur in constructions such as ‘has V-ed, and is still V-ing’ (Johanson 1971: 198) or with continuative expressions such as ‘go on V-ing’. Due to the inherent culmination point, the feature [tf] may be less fertile with INTRAHF (Section 10.2.1.5). Finitransformatives normally react negatively in tests concerning gradual realization. The indivisible, ‘all-or-nothing’ content is mostly incompatible with adverbials implying occurrence in portions. Even if the action may take up a certain amount of time, they also reject temporally delimiting duration adverbials of the type ‘for X time’, e.g., *reach the house for two hours. This criterion distinguishes finitransformatives from nontransformatives and initiotransformatives. However, they readily combine with mensural expressions of the type ‘in X time’, expressing the total indivisible action including its crucial limit, e.g., reach the house in two hours. This criterion distinguishes them from nontransformatives. The feature [tf] is relevant in all European languages, e.g., Classical Greek árnymai ‘acquire, gain’, Lithuanian ¬ieiti ‘enter’, German gewinnen ‘win’, Tatar ül- ‘die’. In earlier literature, Indo-European [tf] distinctions were mostly discussed as “perfective” vs. “imperfective” distinctions. Their presence in the German verb system was first discussed by Jacob Grimm and Hermann Paul. 5.2.1.1.1. [TF, MOM] The cursus of finitransformatives may be more or less relevant. The telic events they refer to may be conceived of as momentaneous [mom] or temporally extended [ mom]. In the first case, only the transforming final limit is salient. In the second case, the cursus is thought of as a process leading up to that limit. Terms such as “punctual” and “durative” will be avoided, since they are easily misleading; cf. Comrie’s clarifying discussion of ‘punctuality’ (1976: 41–44). “Durativity” is often used for a considerable temporal extension, and sometimes even for INTRA. Momentaneous finitransformatives [tf, mom] imply abrupt transformation without preliminaries, without any salience of the cursus leading to it. The action is conceived of as absolutely indivisible. Though even events of very short duration have an extension in time, initium and cursus seem irrelevant and appear to merge with the transforming finis, e.g., drop, explode, sneeze, Icelandic byrja ‘begin’, Bulgarian skokna ‘jump’, Modern Greek vrísko ‘find’ (in the concrete sense), anakalípto ‘discover’, East Armenian patahel ‘occur’. Such actions typically correspond to Vendler’s “achievements” or to actions expressed by Breu’s and Sasse’s “totally terminative” verb class. It is highly dubious whether the actions denoted by these actional phrases might be regarded as telic, since their three phases practically coincide. As has been empha-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

62

62

Lars Johanson

sized above, however, transformativity is not equal to telicity. The actional phrases in question signal a transformation and are thus transformative. They do not behave like initiotransformatives, but are clearly finitransformative according to the criterion that they cannot occur in ‘has V-ed and is still V-ing’. All [tf, mom] actional phrases are naturally compatible with momentaneous time adverbials, e.g., reached the house two hours ago. In default of a salient course, they do not combine with ingressives or egressives (‘begin / stop V-ing’). Due to the same fact, the INTRA perspective is of limited use with them. The interaction usually results in imminential and propinquitive meanings (10.2.1.3). In a AD language, the only natural operator is AD. Formally corresponding AD partners lack or imply repetition, [ser]. For quantitative reinterpretation, see 6.4. 5.2.1.1.2. [TF, - MOM]

The feature combination [tf, mom] refers to actions of some duration, the cursus of which is cognitively relevant and may be conceptualized as preliminaries leading up to the transforming finis, e.g., English die, Modern Greek paghóno ‘freeze’, Russian razbudit’ ‘waken’, East Armenian ka˙rucel ‘build’. The actional content may be more or less processual, implying successive transformations, e.g., grow, improve. The actions expressed typically correspond to Vendler’s “accomplishments” or to those denoted by Breu’s and Sasse’s “gradually terminative” verb class. It may certainly be discussed whether actions such as ‘arrive’ and ‘die’ are momentaneous or not in extralinguistic reality. Thus, such actions are often used as prototypical examples of Vendler’s “achievements”. The basis of the present classification is, however, the empirical observation that actional phrases expressing such actions allow pretransformational phases of some duration in their actual linguistic behaviour. Many European languages have few [tf, mom] verbs, but readily create corresponding expanded actional phrases (6.1). Since the actions are both goal-oriented and conceived of as having a certain duration as a totality, the actional phrases may occur in the question ‘How long does it take to V?’ and thus combine with adverbials expressing in what time a given event is carried out (‘in X time’). Non-momentaneous finitransformatives are fertile with various aspects. INTRA envisages the preliminaries without the transformation, e.g., Turkish geliyordu ‘was coming’. AD envisages the attainment of the crucial limit, i.e., the very transformation, e.g., Russian vstretil ‘met’. INTRA and AD disregard limits, e.g., Turkish geldi ‘came’, Russian vstreˇcal ‘met, was meeting’. 5.2.1.2. [ti]

The crucial limit may also be the beginning of the action. Initiotransformatives [ti] are, like finitransformatives, actionally heterogeneous, but conceptualize an initial evolutional turning point as an inherent part of the actional content. They combine

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

63

Viewpoint operators in European languages

63

the concept of entering a state with that of the state itself, “marquent un point de départ avec une ouverture possible sur un développement ultérieur” (Seiler 1993: 28). Such actions, which involve a transformative beginning of the cursus, correspond to the ones expressed by Breu’s and Sasse’s “inchoative-stative” verb class. An initiotransformative such as Turkish otur- denotes two evolutionally coupled phases: a transformational ‘sit down’ and a resulting posttransformational ‘sit’. It can thus occur in constructions such as ‘(has) V-ed and is still V-ing’. The first phase stands for a telic and dynamic action, the second one for an atelic and static action. The verb may thus correspond to both a finitransformative and a nontransformative of another language, e.g., Russian sest’ ‘sit down’, sidet’ ‘sit’. It is clear that initiotransformatives constitute a class of their own and should not be mistaken for a subclass of [tf] or [ t]. They are neither ingressives (inchoatives, inceptives, etc.) nor statives. They cover both a telic action and an atelic action, i.e., what may, in some other language, be expressed by two actional phrases, one [tf]  one [ t]. In this sense, initiotransformatives are certainly ambiguous, but their ambiguity is systematic, distinguishing them from all other classes. They involve a cognitively significant initium just in the same way as finitransformatives involve a cognitively significant finis, and thus do not deserve the designation “two-phase verbs” more than non-momentaneous finitransformatives do. The former imply a transformation leading to a state, the latter a state leading to a transformation. Just as non-momentaneous [tf] actional phrases have a preliminary (pretransformational) and a transformational phase, [ti] actional phrases possess a transformational and a posttransformational phase. In their initial readings, [ti] actional phrases are, like [tf] items, compatible with momentaneous time adverbials. In their statal readings they are, like [ t] items, compatible with temporally delimiting duration adverbials. As for phasal verbs, continuatives and egressives may combine with their statal reading (‘go on V-ing’, ‘stop V-ing’), whereas there are heavy constraints on the use of ingressives to specify their initium (*‘begin V-ing’), at least when the initial transformative phase is momentaneous. [ti] verbs are not equally well represented in all European languages. Examples of [ti] are English hide, Czech opˇret se / opírat se ‘lean  be leaning’, Classical Greek órnymai ‘get in motion  move’, Modern Greek stékome ‘stop  stand still’, katalaváino ‘understand (= become aware of  be aware of)’, krívo ‘hide (= put out of sight  keep out of sight)’, Romanian cunoa¸ste ‘come to know  know’, Maltese libes ‘put on  wear’, Turkish tut- ‘grasp  hold’, Tatar awïr- ‘fall ill  be ill’, Hungarian fekszik ‘lie down  lie’, Persian nešastan ‘sit down  sit’, Talysh nïšte ‘sit down  sit’, hïte ‘fall asleep  sleep’, Nogai oltïr- ‘sit down  sit’, Kalmyk su‘sit down  sit’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

64

64

Lars Johanson

5.2.2. [ t] Nontransformative [ t] contents are actionally homogeneous, without a salient initial or final phase, e.g., English cry, dance, know, run, want, work, write, Classical Greek phér¯o ‘carry’, Russian dut’ ‘blow’. The atelic actions they describe have, when represented as events, their natural limits, but none is conceptualized as crucial. All three phases are equally relevant, and a possible limitation is external (6.2). Needless to say, [ t] is never identical to IPFV in the sense of INTRA or AD. To specify initium and finis of [ t] contents, ingressives and egressives (‘begin / stop V-ing’) are required. As a [ t] actional content lacks a culminating point, it is, when represented as events, actually taking place from the moment it begins. Thus, to use a past tense for an event described with a [ t] item, it is enough that its initium is prior to O. Predications such as English X has written, X wrote, Russian X pisal, Turkish X yazdı mean that X has already carried out a portion of the action at O, e.g., written something. The event must at least have begun, but may be still going on or already finished. This means that [ t] actional phrases may occur in constructions such as ‘has V-ed and is still V-ing’. The action can be interrupted at any point of its course and still be said to have already taken place. On the other hand, there is no natural point beyond which it would not be prolongable. Nontransformatives are naturally compatible with durative expressions indicating that the action is carried out for a certain time (‘for X time’). They are also compatible with points of time and may combine with momentaneous adverbials. This is due to the presence of a relevant limit in the actional content. Though [ t] items lack a crucial limit, they do possess a limit of relevance for aspectual realizations: the initium. For ‘initial attraction’, see 7.2.2 and 10.2.2.1.1. 5.2.2.1. [ t, dyn]

The feature ‘dynamicity’ [dyn] is inherent to transformatives, but it is also a subclassifying criterion for nontransformatives with respect to aspect reagence. Most languages account for the distinction [ t, dyn]. Dynamic nontransformatives stand for less time-stable actional contents than non-dynamic ones, and have relatively well discernible cursus with clear beginnings and ends, e.g., burn, eat, grow, look, play, sew, sing, speak, walk, wash, Russian pisat’ ‘write’, myt’ ‘wash’, pit’ ‘drink’, pomogat’ ‘help’, stradat’ ‘suffer’, Modern Greek dhiavázo ‘read’, dhulévo ‘work’, East Armenian šnˇcel ‘breathe’, zbosnel ‘walk’. The actions expressed more or less correspond to Vendler’s “activities” or Breu’s and Sasse’s “(processual) actions” (ACTI). ‘Concreteness’ and ‘agentivity’ are frequent though not necessary features. While [ t] actional contents are homogeneous in the sense of lacking initial or final transformations, [ t, dyn] contents are dynamic in the sense of internal processual evolution. They often involve some progress observable in gradually produced effects

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

65

Viewpoint operators in European languages

65

and may then combine with expressions of speed, e.g., write very fast. Many actions imply little if any internal progress. Dynamic nontransformatives easily combine with INTRA and INTRA values, and the dynamicity is particularly well suited for INTRAHF items (“progressives”). In AD languages, AD is the natural choice, whereas the use of AD implies transfer to [t], e.g., Russian napisat’ ‘write’, vymyt’ ‘wash’, vypit’ ‘drink’, pomoc’ ‘help’, postradat’ ‘suffer’. Thus, the AD distinction clearly manifests itself with verbs of this kind, without the tendencies towards lexical differentiation observed with [tf, mom] verbs (Forsyth 1970: 53). The difference between AD and AD only resides in the view of the event: its presentation in the attainment of the crucial limit (AD), or as mere occupation with the action, without reference to any limit ( AD). Russian [ t, dyn] verbs expressing actions that imply little internal progress often lack perfective partners, e.g., iskat’ ‘search for’, mesti ‘sweep’, tancevat’ ‘dance’, upravljat’ ‘govern’, šumét’ ‘make noise’ or indeterminate motion verbs such as guljat’ ‘stroll’ and teˇc’ ‘flow’. Even in languages lacking AD distinctions, many verbs of the types cited above, e.g., eat, write, may vacillate with respect to their [t] conceptualization, the finis being potentially conceivable as a crucial limit (without explicit external limitation by an object). 5.2.2.2. [ t,

dyn]

With non-dynamic nontransformatives [ t, dyn], the actional content is conceptualized as static, homogeneous, lacking internal processual evolution. It covers relatively unchanging, time-stable physical, psychical and social states – properties, relations, knowledge, possession, etc. – with less clearly discernible cursus and limits, and is often less concrete and less agentive than [ t, dyn] actional contents, e.g., be blind, contain, remain, Icelandic eiga ‘possess’, þekkja ‘know’, Portuguese viver ‘live’, Romanian costa ‘cost’, Modern Greek lípo ‘be lacking’, aksízo ‘be worth’, Russian znaˇcit’ ‘mean’, prinadležat’ ‘belong’, sostojat’ ‘consist’, naxodit’sja ‘be located’, uvažat’ ‘respect’, East Armenian karo anal ‘be able’, nsanakel ‘denote’. The actions expressed approximately correspond to Vendler’s “states” or Breu’s and Sasse’s “totally stative” class. In default of internal evolution, [ t, dyn] actional contents are incompatible with expressions of speed. As they have a low preference for countability, they are often incompatible with expressions of repetition. They may also avoid combining with ingressives and egressives. Actional contents implying little internal progress exclude gradual expressions, e.g., *sit little by little. [ t, dyn] actional contents readily combine with INTRA, which presents them from a viewpoint located within their course, e.g., Modern Greek íksere, Turkish biliyordu ‘he knew’. Due to their lack of dynamicity they are infertile with INTRAHF (10.2.2.5). Combinations with INTRA are often rather limited. Thus, Romanian verbs such as cântari ‘weigh’ are normally not used in the INTRA past,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

66

66

Lars Johanson

and Modern Greek verbs such as aníko ‘belong to’, periéxo ‘contain’, periméno ‘wait’ lack Aorist forms. Not unexpectedly, AD is excluded. Thus, Russian [ t, dyn] verbs such as stojat’ ‘stand’ and ždat’ ‘wait’ lack perfective partners. Perfectives such as prostojat’ ‘stand through a certain period of time’, and podozdat’ ‘await’ represent special modes of action (Section 4.2) and are not normal AD aspectual partners of stojat’ and ždat’.

6. Actional recategorization 6.1. Recategorization processes An actional phrase is minimally a verb lexeme taken in its most concrete and quantitatively simple sense, referring to a single basic event. It may be assigned one or more of the features discussed above and classified accordingly. Attention must be paid to different semantic readings and valency differences that affect the internal phase structure. The problem is extremely complex, so much the more as actional values can also change pragmatically. Restriction to the lexeme level is impracticable, since verbs are seldom context-free. Though the central syntactic role of the verb and its morphology has often led to the assumption that aspect and actionality relate to the simple verb, strict lexeme classifications are impossible, even languagespecifically. One and the same lexical item may prove ambivalent in tests, showing both [ t] and [t] properties, e.g., dine for two hours vs. dine in two hours. In particular, many [tf, mom] verbs do not represent their class in a clear-cut way without disambiguating complements. In the following, it will be assumed that minimal actional phrases may change their basic phase structure by way of recategorization. This is thought to take place according to certain principles that were discussed in Johanson (1971: 198–220) and supposed to be valid beyond the particular purpose of classifying Turkish actional phrases. The compositional process of recategorization starts from the syntacticsemantic minimum of a verb in its most concrete and quantitatively simple sense and proceeds to account for the actional effects of more abstract and quantitatively complex readings in successively expanding syntagms containing various obligatory and facultative complements. Note that this conception is not equal to the traditional view of the actional content of the very verb as “vacillating according to the context”. The basic phase structure may be transformativized or nontransformativized. The former change implies actional heterogenization, the latter homogenization. The specification of the actional content may be overtly signalled by T-marking and T-marking. T- and T-markers may be derivational elements closely tied to the verb, auxiliaries, parts of complex predicates, case-marking devices, adverbials, etc. The actional values expressed by T- and T-marking are often referred to as “im-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

67

Viewpoint operators in European languages

67

perfectivity” and “perfectivity” respectively, though they are not aspectual in the sense of viewpoint categories. Exceptions are portmanteau markers (2.2), which combine a viewpoint and a phase structure value, thus expressing an interactional meaning in themselves.

6.2. Transformativization Nontransformatives may be limited to [t] by mensural units expressing a specific quantity and thus defining the minimal-maximal extension of the action. A crucial limit, external to the verb meaning itself, is set with regard to the subject referent, an object referent or the goal of a motion. The decisive factor is the undivided reference. The limitation lies in the verb- external entity, which is totally involved in the action – totally covered, affected, created, consumed, destroyed, etc. – and thus specifies its crucial limit. The action leads to a transformation because there is an end to the entity. Limiting elements will not be dealt with as “context”, but as part of the actional phrase. The limitation may also be spatial. With [ t, dyn] verbs such as go, run, walk, a crucial limit may be set by a mensural expression, e.g., a kilometre, or by the goal of the motion, a materially limiting entity expressed by a direction adverbial, e.g., to the beach. Undivided reference (‘all the way to’) is decisive for the limitation to [tf]. Similarly, [ti] items can be recategorized to [tf] by adverbials that restrict their content to one of the two possible phases. In Turkish sandalyeye otur- ‘sit down on the chair’ and Hungarian az ágyra fekszik ‘lie down on the bed’, the items otur- ‘sit down  sit’ and fekszik ‘lie down  lie’ are recategorized as [tf]. In sandalyede otur- ‘sit on the chair’ and az ágyon fekszik ‘lie on the bed’, the actional contents are homogenized to [ t] (cf. Csató 2000). As far as verb complements are concerned, the decisive point is, again, whether their referents are quantified as undivided entities or not. The crucial limit may be set in relation to a subject or object referent with certain properties. A [ t, dyn] verb such as write primarily refers to an atelic event. It may be limited to [tf] by a totally affected object referent, expressed by a nominal such as a letter, the letter, letters, the letters, two letters, a set of letters. With undivided reference, the resulting action is telic, i.e., it cannot be considered fully achieved unless the relevant amount of letters is produced. The present article is not the adequate framework for dealing in detail with how complements and their case-marking relate to limitation, and how limitation relates to referentiality, specificity and definiteness as part of general problems of “transitivity” and information structure. There are certain – though often rather unsystematic – affinities between undivided reference and object definiteness, between [ t] and indefinite objects, between [t] and definite objects. Even the indefinite objects in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

68

68

Lars Johanson

build a house and bake a cake offer crucial limits, whereas the definite article in play the piano does not. Many European languages lack a definite article, the presence vs. absence of which may signal specificity, non-specificity, definiteness and indefiniteness. Russian pisat’ ‘write’ has the basic features [ t, dyn]. Writing is an atelic action unless it has an object as a goal. More important than goal-directedness, however, is the undivided reference to the object. Reference to an entity that is conceived of as totally produced (pis’mo ‘letter’), referring, sets a crucial limit to it. With AD napisal pis’mo, the event is viewed in its attainment of this limit, which can be translated as ‘wrote a / the letter’. For the question of multiple entities, see 6.4. Differences between the definite and the indefinite conjugation in some Finno-Ugrian languages may also contribute to differentiating [ t].

6.3.

T-marking

A nontransformative actional content may be transformativized by means of special phase structure markers, T-markers, which, focusing on the finis or the initium, explicitly signal the notion of a crucial limit which the basic actional phrase does not contain. T-markers may turn nontransformatives into finitransformatives, e.g., English sit [t]  sit down [tf], or initiotransformatives into finitransformatives, specifying the initial phase of the content of the unprefixed verb, e.g., Hungarian fekszik ‘lie down  lie’ [ti]  le-fekszik ‘lie down’ [tf]. T-marking categories also include the above-mentioned delimitative and perdurative modes of action, which imply a crucial limit, e.g., Russian poˇcitat’ ‘read for a while’, proˇcitat’ ‘read through’. Though [ t] features are very often implicit, most European languages also use explicit T-marking devices. The use is more or less generalized. Some languages, e.g., Baltic (Lithuanian, Latvian), Hungarian, Kartvelian, Ossetic, most Turkic languages, and some Slavic languages such as Bulgarian, employ T-marking rather systematically. T-marking mostly starts with phase specification and limitation of transitive actional phrases containing objects (e.g., write a letter). Preverbs are most commonly used as T-markers, e.g., Classical Greek pheúg¯o  dia-pheúg¯o ‘flee’, Latin facio  ef-ficio ‘yield’, Gothic fulljan  ga-fulljan ‘fill up’, Lithuanian rašyti  pa-rašyti ‘write down’, Latvian rakst¯ıt  uz-rakst¯ıt ‘write down’, las¯ıt  iz- las¯ıt ‘read (and finish reading)’, Hungarian ír  meg-ír ‘write (and finish writing), write down’. Bulgarian mostly uses preverbs or the suffix -nto turn imperfectives into perfectives, e.g., piša ‘I write’, ‘I am writing’  napiša ‘I write up’. Kartvelian uses T-marking preverbs that do not change the lexical meaning of the verb, e.g., Georgian da- for cers ‘writes’ and mo- for kvdeba ‘dies’. Modern Georgian offers a choice between anª unmarked Present, e.g., cªer ‘tu l’écris’, ª au terme”, and a T-marked prefixed Present, expressing the action “vue par rapport

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

69

Viewpoint operators in European languages

69

e.g., da-cer ‘tu l’écris’ = ‘tu mènes à bout l’action d’écrire’ ‘tu l’écriras’ (Vogt ª 1971: 175). The preverbs used mostly go back to expressions of movement, often directional adverbs meaning ‘away’, ‘down’, ‘into’, ‘through’, ‘up’, etc., e.g., Classical Greek apo- ‘[away] from’, dia- ‘through’, kata- ‘down’, syn- ‘[together] with’. In Svan, the T-marker is a directional preverb. Ossetic, which is rather similar to Georgian in its T-marking system, has genuinely Iranian preverbs with primary spatial-directional functions, e.g., fe (‘away’; pati-). Postverbs are used as T-markers in Turkic and Mongolian: Kalmyk -jˇ ork- (conu¯ cˇ k- ‘drink up’. They are common in all verb  ‘throw’)  -ˇck-, e.g., u¯ - ‘drink’ Turkic languages of Europe except Standard Turkish. The postverbs mostly go back to dynamic verbs such as ‘give’, ‘put’, ‘reach’, ‘send’, ‘take’, ‘throw’, e.g., Chuvash bˇet- ‘end’, ïl- ‘take’, par- ‘give’, xur- ‘put’, s´it- ‘reach’. T-markers thus go back to lexemes with dynamic meaning components. They may more or less preserve the original lexical meaning or add an additional actional meaning, from which the [t] notion derives. The [t] meaning is often combined with some additional specification of the content with respect to direction or manner of realization, e.g., Hungarian ki-jön ‘come out’, German er-jagen ‘hunt down’, Tatar eˇ šläp bˇetˇer- ‘work (and finish working)’. T-marking is often performed by completives, consumatives, and exhaustives, signalling that the object referent is effected or affected thoroughly, to completion, totally consumed (e.g., eat up). The additional meanings tend to fade away in favour of pure T- marking. In archaic Classical Greek, the lexical meanings of T-marking preverbs such as apo- ‘[away] from’ are often rather well preserved. The possibility of substituting apo-thane˜ın for thane˜ın ‘die’ in Attic Greek indicates that the lexical meaning has been lost. This development may even lead to the loss of the simplicia. T-marking is often referred to as “perfectivization” and confused with marking of AD. Thus, Lithuanian verbs provided with T-markers such as pa- are usually called “perfectives”, though they only signal transformativity (cf. Maslov 1985: 15). T-marked items are functionally similar to AD items by signalling a crucial limit, but they do not, as the latter, imply the actual attainment of this limit. T-marking only specifies the actional phrase. In Functional Grammar of the Simon Dik tradition, it should, as argued in Johanson (1996), be taken to belong to the innermost layer of /-operators, operating immediately on the predicate. T-marking may be said to represent a preaspectual stage, since it may develop diachronically into viewpoint marking. This shift has taken place in Slavic languages such as Russian, where T-marking not only implies ‘a crucial limit to attain’, but also views this limit as attained in the sense of AD. Since T-marking does not signal PFV in the sense of AD or INTRA, its absence should not be confused with IPFV in the sense of INTRA or AD. Thus, Hungarian T-unmarked past items as in írta a levelet ‘wrote the letter’, írt egy levelet ‘wrote a letter’, levelet írt ‘wrote letters’ do not display INTRA or AD

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

70

70

Lars Johanson

meanings and may, for example, readily combine with ‘for X time’ adverbials. The combination PAST times T is not a past intraterminal, but just suggests that the crucial limit was not attained at the relevant time interval, e.g., Finnish kirjoitti kirjetta ‘was [occupied with] writing a letter’ (partitive); cf. kirjoitti kirjeen ‘wrote a letter’ (T-unmarked). In systems with T-marking, [ t] items may get conative interpretations without a INTRA opposition, e.g., sterben ‘die’ as against er-sterben in 19th century German, e.g., Ich sterbe, sterbe und kann nicht ersterben ‘I am dying, dying, and cannot pass away’ (Goethe). Similarly, a Classical Greek [t] item such as épeithe ‘persuaded’ may suggest conation (‘tried to persuade’). The distinction between T-marked and T-unmarked items yields the same effect in Georgian, e.g., in the Aorist items of the sentence A o, a o da ver gaa o ‘Il essaya de l’ouvrir, sans résultat’ (Vogt 1971: 187). As we noted, T-markers are used with various degrees of generality. Some languages have developed consistent transformativizing systems, where T-markers form highly grammaticalized modes of action used more or less obligatorily with [t] actional phrases. For example, Hungarian verbs that are clearly transformative in their quantitatively basic meaning are T-marked, e.g., meg-hal ‘die’ (6.8). Among the Kartvelian languages, Svan applies obligatory T-marking to transformatives. Still, T-marking generally tends to be rather irregular in that not all [t] actional phrases take part in it. T-marking may also be more or less fertile with different aspectotemporal categories (10.2.1, 10.3.2). Note that T-marking may combine with iterativity or pluri-occasionality markers. Bulgarian secondary imperfectives derived from perfectives, e.g., napisva ‘usually writes up’  pf. napiše ‘writes up’, form a special actional type denoting plurioccasional global events with telic subevents. Compare Russian iterative perduratives such as proˇcityvaet ‘repeatedly reads through’  proˇcitaet ‘reads through’. Lithuanian exhibits T-marked frequentatives such as parašydavo ‘used to write (to completion)’.

6.4. Nontransformativization Transformatives may be actionally homogenized by suppression of the effect of the crucial limit and thus turned into nontransformatives (Johanson 1971: 194– 201). Nontransformativization may come about in different ways, most frequently by quantitative reinterpretation. In these cases, the actional content does not suggest a single-action reading [ser], but is quantitatively interpreted as repeated – as an action composed of a series of identical actions – and thus gets a serial reading [ser]. Such actional phrases are used to represent global events containing subevents. [ser] readings may be suggested by overt markers indicating the in-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

71

Viewpoint operators in European languages

71

volvement of multiple entities (number of subject and object referents), by quantifying adverbials, etc., but they are also possible without such markers. Serialization may turn transformatives into nontransformatives: [t]  [ser] = [ t]. The actional content then no longer involves one single transformation, but recurrent transformations all through its duration, which has a homogenizing effect. Iteratives are normally nontransformative. An actional phrase with actants involving multiple entities such as French tous mouraient ‘all were dying’ (Johanson 1971: 206) may thus be nontransformative: a collective ‘dying’ may be conceived of as atelic. When [tf, mom] actional contents such as ‘explode’ are realized as [ser], they normally lose their transformativity. The change of the phase structure to [ t] naturally changes the applicability of viewpoint operators. For example, high-focal intraterminals prefer to operate on [ ser] actional phrases (10.2.1.5). The feature [ser] is always present in pluri-occasional (“inactual”, habitual, etc.) meanings. Note that, since [ser] is not part of the context, but of the actional phrase itself, we do not say that the applicability of aspects to certain verbs changes under pluri-occasional readings. The limiting effect of “accusative” objects in Finnish and Estonian is cancelled by [ser] interpretation. The same is true of Lithuanian T-marked actional phrases: in kasdien paraš˙e po viena¬ laiška¬ ‘writes (and finishes writing) a letter every day’, the global event, envisaged in its course, is a sequence of events, each characterized as transformative. PAST (INTRAo) operates on the global event, while the T-marking refers to the actionality of each subevent. Slavic secondary imperfectives, which have a nontransformativizing effect, frequently imply [ser]. Though the notion of a crucial limit is preserved in the verb meaning, it is only valid for each sub-action and suspended as a feature of the whole action expressed. The formation of Bulgarian imperfectives from perfectives (with -a-, -va-, -ava-, -uva-) is a productive nontransformativizing device, the products of which often occur with pluri-occasional (habitual) meanings, e.g., napis-va-m ‘I (usually) write up’. Though [ t] might be said to be typical of actional phrases expressing repeated events, plurality is not homogenizing as such. Even a [ser] actional phrase may be limited. The plurality may be exhaustive, involving a whole set of entities, e.g., boil eggs [ t] boil all the eggs [t], and definite articles as markers of identifying reference may support such readings. Thus, write letters tends to be interpreted as [ t] because of its indefinite plural object. The corresponding object nominal in Russian pisat’ pis’ma may easily be interpreted as referring, in a limiting way, to a complete particular set (‘write and finish writing a / the [whole set of] letter[s]’). Still, the decisive factor is not the reference to a whole set of entities. Actional phrases with actants involving multiple entities may also be transformative: a collective ‘dying’ can, for example, also be conceived of as telic. The decisive question is about undivided vs. divided reference to the set of entities: does the global event as a whole have a “desinence” – a built-in endpoint, an inherent final limit indicating an evolu-



$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

72

72

Lars Johanson

tional minimum-maximum – or not? A good deal of indeterminacy may be expected with respect to the conceptualization of such cases. Finally, negation of [t] actional phrases often has a nontransformativizing effect. An actional content consisting of the absence of a certain telic action is usually conceived of as lacking a shape which might culminate in a crucial limit.

6.5.

T-marking

Nontransformativization may be signalled by T-marking modes of action, which overrule the idea of a crucial limit in the meaning of the basic actional phrase and thus exclude limit-oriented interpretations. T-marking can be performed by Slavic secondary imperfectives derived from prefixed perfective stems by means of suffixes that go back to iterative markers. In the meaning of these verbs, the actional notion of a crucial limit is suspended as a relevant feature. They still preserve their iterative function in some Slavic and Baltic languages, e.g., Lithuanian per-raš-in˙eti ‘rewrite, copy (repeatedly)’. But Slavic secondary imperfectives may also function as T-markers without implying [ser], e.g., Czech vy-hazuji ‘I throw out, I am throwing out’. Ossetic transformatives may be nontransformativized by means of the element -cæi. Turkic and Mongolian languages use postverb constructions to specify the statal phase of ambiguous actional awïr-ïp tˇor- ‘be ill’, Kalmyk unt- ‘fall phrases, e.g., Tatar awïr- ‘fall ill  be ill’ unt-jˇ kevt- ‘sleep’. The [ t] meaning may also be combined asleep  sleep’ with a specification of the manner of realization, e.g., Turkish yaz-ıp dur- ‘keep on writing’ (durativity). T-marking can also be carried out with case-marking on object nominals. Transitive [t] actional phrases may have intransitive [t] counterparts, e.g., German (etwas) durchbóhren [t, mom] / (etwas) dúrchbohren [t, mom] ‘bore through, pierce (something)’ vs. durch (etwas) bohren ‘bore through (something)’, Swedish skriva (något) ‘write (something)’ vs. skriva på [ skriva p] (något) ‘be engaged in writing (something)’. Some Finno-Ugrian languages employ systematic Tmarking by means of the partitive as opposed to the total (“accusative”, formally genitive or nominative) object case. Such oppositions are usually said to distinguish “limited” from “non-limited” (“total”, “resultative”) actions. Since “limitation” here means divided reference to the object, the definition does not contradict our analysis. Limitation in our sense means that the actional content has an inherent limit defined by the extension of the object. The partitive serves as a T-marker with homogenizing effect, e.g., Finnish lukea kirjaa ‘read (parts of) the book’. By contrast, the total object case implies an action that includes a crucial limit, lukea kirjan ‘read (and finish reading) the book’. Similarly, the total object case in Estonian küpsetas koogi ‘baked a cake’ (cf. Metslang & Tommola 1995: 305) might be analysed as the











$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

73

Viewpoint operators in European languages

73

unmarked case representing the natural transformativity of ‘bake a cake’, i.e., with a built-in crucial limit, whereas the partitive suppresses this limit in küpsetas kooki ‘engaged in baking a cake’. T- and T-markers may interact to produce differentiated actionality systems. Transformatives need nontransformativizing devices and vice versa. It is thus not surprising if T-markers occur with items carrying T-markers. Several T-markers go back to iteratives, which often start from [tf, mom] verbs and then diffuse to other types. Interestingly enough, the originally iterative Slavic -aj- derivates, which developed into T-markers, almost always occur with T-marked verbs.

6.6. Recategorization options The recategorization options mentioned are roughly summarized in the following graphic, which shows the paths of SER(ialization) to [ t] by means of [ser], HOM(ogenization) to [ t] by other means, and LIM(itation) to [tf]:

ti

HOM



t

tf

LIM



LIM



SER



ti t

SER



tf

6.7. Interaction with time adverbials A few words should be said here about the interaction of time adverbials with actional values, i.e. as modifiers of the actional phrase. The compatibility of time adverbials with actionality and aspect is a complex matter, to which I shall return in several sections. One basic question is whether certain adverbials refer to the global event, its subevents, or some other interval, e.g., an aspectual orientation point. The stereotype ‘X Time’ will be used for any quantified unit of time, and ‘tx’, ‘ty’ for different instants of time (Bertinetto & Delfitto, this volume). Compatibilities with certain time adverbials can be used as criteria for distinguishing [ t]. The adverbials themselves do not indicate such actional values, but their meanings interact with the phase structure in various ways. The temporal delimitation performed by certain adverbials differs from the material limitation discussed above. Time adverbials have no limiting effect [t]  [t]. Not all temporally delimited events are telic and expressed by [t] actional phrases.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

74

74

Lars Johanson

‘In X time’ expressions do not measure the action, but the time needed for its accomplishment. Since this implies totality, they only combine with transformatives and may thus serve as a diagnostic tool for identifying the actional feature [t]. They do not delimit the action, but indicate how long it takes to reach its crucial limit, e.g., mow the lawn in (= within) ten minutes. ‘X times’ expressions quantify the number of occurrences overtly and may thus have a delimiting effect, e.g., ‘go out four times’. ‘For X time’ expressions are mensural, indicating temporal extension. Despite this temporally delimiting function, they do not limit [ t] actional phrases. They might, on the contrary, even seem to have a nontransformativizing effect in cases such as mow the lawn for ten minutes. However, mow the lawn is an actional phrase with no intrinsic transformativity to lose. The adverbial for ten minutes serves as a criterion for determining it as nontransformative. Basic transformatives must undergo quantitative reinterpretation to become compatible with ‘for X time’, e.g., win for a long time: [t]  [ser] [ t, dyn]. The type ‘from tx to ty’ is delimiting in a similar way. Both types are not only used to measure the event itself, but may also indicate an O, the interval of an aspectual view (2.3.1). ‘Until tx’ expressions are temporally delimiting, either referring to the extension of the action or to the time it takes to accomplish it. With [ t], they measure the action, e.g., drink until dawn, and with [tf, mom], they may measure the accomplishment time, e.g., German das Problem bis zehn Uhr lösen ‘solve the problem by ten o’clock’. They neither transformativize nor nontransformativize, and cannot serve as [t] criteria. Abtemporal expressions of the types ‘since tx’ and ‘since X time’ determine the beginning of an event still going on at an O, and may combine with [ t] or [t]. Speed adverbials combine with [t] and [ t, dyn] actional phrases, e.g., kill slowly, write fast. Basically [t, mom] actional phrases can combine with them after quantitative reinterpretation, e.g., tick quickly: [t, mom]  [ser] [ t, dyn]. Graduality adverbials combine with [t] and [ t, dyn] actional phrases capable of implying internal progress, e.g., burn down gradually (but *dance little by little). Adverbials meaning ‘already’ express a state already obtaining at O (‘already being in the state at O’). Those meaning ‘still’ express a state that has not yet ceased to obtain at O (‘remaining in the state at O’). Both are only compatible with states, which does not, however, mean that they combine with [ t] actional phrases only.

6.8. Generalized T-marking As a stage on the way leading from lexicon to grammaticalized aspect, T-marking, overt marking of the [t] distinction, may be employed in a generalized way, [t] and [ t] verbs being grouped together to form fixed actional pairs. Thus, while Indo-European verbs were largely assigned “tense classes” according to their [t]

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

75

Viewpoint operators in European languages

75

actionality, many Classical Greek verbs acquired counterparts of the opposite actionality. More and more Aoristic [t] counterparts to Presentic [ t] verbs were created. Verbs with similar meanings and different phase structures combined to form suppletive pairs integrated in the [ t] duality. The difference obviously concerned the actional content, not aspect as “la vue d’un procès” (Seiler 1993: 27). The actional function of the Aorist was increasingly taken over by preverbial items, e.g., diéphyge and katéphyge, whereas simple items such as éphyge ‘fled’ were limited to a new “constative” use of the Aorist (cf. 7.2.2, 7.7.2). The T-marking was rather general, but did not comprise all Aorists. T-marked items later expanded to encroach on the functional territory of the simple verbs, causing their partial loss. Some modern European languages, e.g., Italo-Croatian, show similar tendencies to generalize T-marked verbs and to lose their unprefixed counterparts (Breu 1992: 114–116). In Slavic, T-marking is rather dominant. The perfective vs. imperfective duality created actional systems of paired verbs early on. The lexical meanings of the markers largely faded, and the [ t] functions were generalized: dynamic [t] contents were systematically distinguished from [t] contents, e.g., nap˘ısati ‘write down’ from p˘ısati ‘write’. Items without T-marking were increasingly interpreted as [t]. Some Slavic perfective vs. imperfective distinctions, e.g., the Bulgarian one, have retained much of this early stage of semantic delevopment. A system of this kind is also present in modern Hungarian. Basically finitransformative actional phrases require T-marking, e.g., meg-hal ‘die’. The unmarked counterpart hal can only refer to an atelic action of a global event with [ser] reading: [tf]  [ser] [t]. Mensural complements specifying the total extension need T-marking for limitation, e.g. ír egy levelet ‘write a letter’  meg-ír egy levelet ‘write a letter (to completion)’, a levelet írja ‘write the letter’  meg-írja a levelet ‘write the letter (to completion)’. With nonmensural complements, the T-marker marks the crucial limit of each subevent, not of the global event, e.g., leveleket ír ‘write letters’  leveleket ír meg ‘write letters (to completion)’. Even type-referring actional phrases may be T-marked: levelet ír ‘do letter-writing’  levelet ír meg ‘do letter-writing (to completion)’. Even in languages where T-marking has become relatively systematic, it may be optional or lacking in certain cases. Given the asymmetrical nature of the actional distinctions, T-unmarked actional phrases are not always necessarily classified as [t]. Classical Greek explicit T-marking did not comprise all Aorists. The clearly transformative nature of an actional phrase may make T-marking superfluous, e.g., Lithuanian ¬ieiti ‘enter’, mirti ‘die’. Modern Slavic languages possess nontransformativa tantum lacking [t] counterparts, e.g., Bulgarian govorja ‘I speak, am speaking’. Actional phrases with a potentially limiting object may dispense with Tmarking to signal that the object is nonlimiting, e.g., Bulgarian PAST (INTRA)  [t] pišeše nova simfonija ‘was writing a new symphony’ (Lindstedt 1985: 163).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

76

76

Lars Johanson

Hungarian may refrain from marking actional phrases containing non-specific complements, e.g., levelet kapott ‘received a letter / letters’, új munkába kezdett ‘began a new work’. Directional adverbials may limit an actional phrase without obligatory T-marking, e.g., lefekszik az ágyra or az ágyra fekszik ‘lies down on the bed’ (6.2). Some cases of lacking T-marking concern the relation between global events and subevents. It was noted that a basically transformative Hungarian actional phrase may dispense with T-marking if it gets a [ser] reading recategorizing it to [ t]. Thus, hal ‘die’ requires T-marking if the actional content is [ ser], e.g., meg-halt ‘(s)he died’, whereas the T-marker is optional under the reading [t]  [ser] [ t], e.g., százával haltak ‘hundreds died’. The simple item refers to the actionality of the global event, not of the single subevents. The situation is similar in Italo-Croatian, which has in principle preserved its [t] distinctions, e.g., in Present and Imperfect, but shows tendencies towards dismantling the system. The younger generation is giving up T-marking to signal transformativity of the subevents of a [ser] global event, e.g., rivaša ‘arrived (repeatedly)’ (Breu 1992: 110). Generalized and systematic T-marking was the starting point of the evolution of Slavic AD aspect (Section 9.4). The change implied that the [t] and [ t] verbs grouped together developed into pairs signalling viewpoint distinctions. Interestingly enough, a similar path led from the Classical Greek preaspectual actionality duality to the Modern Greek INTRA aspect (Section 7.2).

7. Intraterminality 7.1. Definition The most widespread viewpoint opposition in European languages is that of intraterminality vs. nonintraterminality. Intraterminality, INTRA, envisages the event from an internal point of view, intra terminos: within its limits, after its beginning and before its end. INTRA items view the event in its course and are unable to grasp it in its totality. Since the viewpoint is inside the event, the initium, the finis and the full cursus do not appear in the range of vision (Johanson 1971: 101). In narrative discourse, this internal way of viewing events makes them non-propulsive, incapable of advancing a plot. INTRA items are used to describe processes and states as observed from orientation points temporally included in them. INTRA is thus an instruction to situate the cursus of the event as overlapping an O. The event and O are not coextensive, O being included in the event time. The extension of the event is otherwise left unspecified: it may or may not last beyond O. As with all viewpoint operators, the realization of the phases that are not aspectually highlighted is by no means negated, but only latent, excluded from the view. INTRA items tell us nothing about a pos-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

77

Viewpoint operators in European languages

77

sible attainment of the final limit after the point of aspect and thus do not signal objective non-completion. Nonintraterminality, INTRA, as opposed to INTRA, does not present the event from an inside point of view, but rather envisages it from outside, without special regard to limits. It neither expresses nor excludes completion, but tacitly suggests, unless the contrary is signalled, the occurrence of the event as an unanalyzed totality including the endpoint. This integral meaning makes INTRA items propulsive in narrative discourse. INTRA may also imply the attainment of a crucial limit, if there is one to reach. Completion in this sense is the effect of transformative actional phrases. The role of O should not be misstated to the effect that INTRA only signals an orientation interval for the event, while the INTRA denotes the very event. Both INTRA and INTRA items obviously refer to the event itself. INTRA items apply to events of various ontological types. The reason for using them is not that the event is durative or repeated. The event may be going on in a concrete and uninterrupted sense, have a short duration, be observed as ongoing over a longer stretch of time, be pluri-occasional, etc. Even telic events of a very short duration can be conceived of as viewed ‘from within’. Terms such as “durative” or “nonpunctual” for ‘intraterminal’ should, however, be avoided, since they are too easily interpreted as ‘temporally extended’, ‘being in a prolonged state’, etc. The massive attention given to ‘duration’ and ‘punctuality’ in literature on aspect has often caused confusion with notions concerning longer and shorter duration. INTRA oppositions do not primarily concern the question of duration. The introspective manner of presentation emanates from the primary deictic “nunc” perspective. This natural vantage point for observing current events is always located in the middle of what happens, only allowing one to perceive an event from inside, not in its totality (Johanson 1971: 100–101, 130–131, 1994: 249–251). This view is typical of the discourse type of the synchronic report, which focuses on Os and does not capture any simultaneous event as a whole (Johanson 1971: 77–78). Without temporalization, an intraterminal may refer to ongoing events in any temporal stratum. Inside a narrative with a past-time O2 axis, it is naturally understood as a past item. Otherwise, it is mostly understood as non-past by default. A PAST (INTRA) item may thus be regarded as an instruction to situate the cursus of the event as overlapping an O which is not posterior to Os . Unless there are contextual indications to the contrary, O will be interpreted as coinciding with Os , e.g., is finishing the letter. The narrow “nunc” perspective is naturally intraterminal, inherent to Os -oriented non-past items. This has important consequences for present time reference, as the restrictions concerning AD show (Section 9). AD cannot be applied to events current at “nunc”, as it disregards the cursus and envisages the future attainment of the crucial limit. INTRA items are also incapable of envisaging a present cursus and are thus prima facie interpreted as ‘past’. The claim that

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

78

78

Lars Johanson

“the Present can normally only have IPFV meaning” is thus correct in the sense that Os -oriented PAST items not marked for AD- or INTRA imply a natural intraterminal perspective. However, the lack of a competing INTRA item in the same temporal stratum causes a neutralization, the product of which is denoted as INTRAo . PAST (INTRAo ) values can be assigned to Present tenses such as Modern Greek vréxi ‘it rains, it is raining’, Georgian s inavs ‘sleeps, is sleeping’, Armenian spasum e¯ ‘waits, is waiting’, Bulgarian cˇ ete ‘reads, is reading’, Ossetic dzuri ‘speaks, is speaking’, Basque mintzatzen naiz ‘I speak, I am speaking’. Genuine INTRA oppositions of competing items are restricted to the past strata.

7.2. Intraterminality oppositions Since intraterminality is a primitive category, INTRA oppositions are found in many languages. Even all known creole systems have INTRA and INTRA items, though less temporalized. The Arabic opposition INTRA yaktubu ‘writes, is writing, was writing’ (“Imperfect”, al-muda¯ r¯ı ) vs. INTRA kataba ‘has written, wrote’ (“Per as the typical example of “un système aspectal fect”, al-m¯ad¯ı) is often regarded  pur” (Kuryłowicz 1956: 27). PAST (INTRA) oppositions are found in huge areas throughout Europe, e.g., in English, all Romance languages, South Slavic, Modern Greek, Maltese, Basque, all Turkic languages, most Iranian languages (except Ossetic), Kalmyk, eastern Finno-Ugrian, Kartvelian and several other Caucasian languages. The oppositions are less developed in Germanic and western Finno-Ugrian languages. Most Slavic languages have given up their earlier intraterminality oppositions (7.9). Today, PAST (INTRA) is relevant in Eastern South Slavic and neighbouring Serbian varieties. Where only the so-called Aorist has been preserved formally, e.g., in certain Serbo-Croatian varieties, it does not represent INTRA, since it no longer takes part in a PAST (INTRA) opposition. With intraterminality-in-past, the “nunc” perspective is applied to past events by shifting the point of view from Os to an O2 . PAST (INTRA) is an instruction to situate the cursus as overlapping a past O2 . PAST (INTRA) oppositions are known under different more or less felicitous designations, e.g., ‘cursive’ vs. ‘constative’, ‘imperfect’ vs. ‘aorist’, ‘inaccompli’ vs. ‘accompli’, ‘paratatic’ vs. ‘aoristic’ oppositions. As is well known, PAST (INTRA) oppositions are often subsumed together with PAST (AD) oppositions under a general IPFV vs. PFV dichotomy. It is one of the aims of the present contribution to point out some substantial differences between these two types of opposition. Examples of PAST (INTRA) oppositions are Hittite appiskit ‘was taking’ vs. epta ‘took, has taken’, Italian cantava ‘sang, was singing’ vs. cantò ‘sang’, Spanish hablaba ‘talked, was talking’ vs. habló ‘talked’, Catalan veia ‘was seeing’ vs. va veure ‘saw’, Romanian fugea ‘ran, was running’ vs. fugi ‘ran’, Bulgarian cˇ eteše

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

79

Viewpoint operators in European languages

79

‘read, was reading’ vs. cˇ ete ‘read’, Albanian hapte ‘opened, was opening’ vs. hapi ‘opened’, Romany džahs ‘went, was going’ vs. džajas ‘went’, Modern Greek éghrafe ‘wrote, was writing’ vs. éghrapse ‘wrote’, Armenian grum e¯ r ‘wrote, was writing’ vs. grec ‘wrote’, Gagauz alïrdi ‘took, was taking’ vs. aldï ‘took’, Chuvash vuª laˇccˇ eˇ ‘read, was reading’ vs. vulˇceˇ ‘read’, Karachai ala edi ‘took, was taking’ vs. aldï ‘took’, Kumyk bara edi ‘went, was going’ vs. bardï ‘went’, Kalmyk umsjala ‘read, was reading’ vs. umšv ‘read’, Tati bæbaftæn bu ‘wove, was weaving’ vs. baft ‘wove’, Lezgian fizwaj ‘was going’ vs. fena ‘went’, Sami læi båradæmen ‘was eating’ vs. båradii ‘ate’, Mordvin lovnilˇe ‘read, was reading’ vs. lovnosˇe ‘read’, Low (Meadow) Mari ludeš ïle ‘read, was reading’ vs. ludo ‘read’, Udmurt puko val ‘sat, was sitting’ vs. puksiz ‘sat down, sat’, Komi-Zyryan munö völi ‘went, was going’ vs. munis ‘went’. In a similar way, Basque analytical verbs oppose PAST (INTRA) to PAST ( INTRA) items, e.g., mintzatzen nintzen ‘I spoke, was speaking’ vs. mintzatu nintzen ‘I spoke’. The few synthetical verbs form a similar but less generalized PAST (INTRA) opposition besides an “aspectually neutral” Past tense (Haase 1994: 289). Such language-specific oppositions have much in common. As shown by examples like French régna / régnait pendant trente ans ‘reigned for thirty years’ (Lyons 1977: 709), they obviously signal different aspects in the sense of viewpoint notions. For example, the Turkic oppositions are very similar to the Romance ones. The corresponding distinction present in Basque analytical verbs “is the same as in the Romance languages” (Haase 1994: 282). Kartvelian makes a basic difference between items derived from a INTRA (xazovani) and a INTRA (certilebrivi) stem. Thus, ª ª “à l’opposition, the Georgian PAST (INTRA) opposition also largely corresponds en français littéraire, entre l’imparfait et le passé simple” (Vogt 1971: 182). It is clear that the viewpoint operators of these oppositions basically operate on global events. The aspectual values have scope over the actional ones. This is also the case when, as in systems of the Bulgarian type, both the aspect of the global event and the actionality of its subevents are expressed overtly. There is often formal parallelism between PAST (INTRAo) and PAST (INTRA) items, the latter being formed from Present stems by means of past auxiliaries, e.g., Tatar bara ‘is going, goes’ vs. bara idˇe ‘was going’, Udmurt mïnä ‘is going, goes’ vs. mïnä val ‘was going’, Romany džal ‘is going, goes’ vs. džahs ‘was going’ (cf. his ‘was’), Talysh håndedä ‘is reading, reads’ vs. håndedä be ‘was reading’, Kabardian ejˇ ‘reads, is reading’ vs. ejˇert ‘was reading’, Chechen molu ‘drinks, is drinking’ vs. molu-ra ‘was drinking’. Kartvelian Imperfects are formed from the INTRA stem by means of a d-suffix. In Indo-European, the Imperfect and the Present are also formed from one common INTRA (“Present”) stem. The items signalling intraterminality show varying degrees of grammaticalization and generalized use. In the languages dealt with here, the INTRA distinction is generalized (“obligatory”) in the past tense. There are, however, also systems in which

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

80

80

Lars Johanson

INTRA items are used so restrictively that the opposing items seem to be rather neutral PAST items, e.g., Estonian PAST ( INTRA) tegi ‘did’ as opposed to PAST (INTRA) oli tegemas ‘was doing’. Preaspectual or early aspectual distinctions of this kind will be commented on under 7.7. 7.2.1. Uses of PAST (INTRA)

A INTRA item suggests that the event is being the case at an O. Any instant during the cursus is a possible vantage point O. An O2 may be more or less easily identifiable on the internal time axis of the discourse, but the value INTRA is not dependent on this identifiability. PAST (INTRA) items may also be used without an identifiable O2 . The event may or may not be conceived of as occurring on a specific occasion or as part of a particular setting. Since a PAST (INTRA) item does not envisage the finis, it does not tell us whether or not the event continued beyond O2 . A PAST (INTRA) item may present a uni-occasional event as an ongoing process at a past O2 , e.g., Italian pioveva, Turkish ya˘gmur ya˘gıyordu, Modern Greek évrexe ‘it was raining’. This “continuative” use is said to be one of the main functions of IPFV items. A PAST (INTRA) item may also present a pluri-occasional event – as a global event containing repeated subevents – in the same way, e.g., Italian veniva ogni giovedì, Turkish her per¸sembe günü geliyordu ‘came every Thursday’. This ‘habitual’ use is supposed to be another main function of IPFV items. In certain languages, e.g., Bulgarian, the actionality of the subevents may additionally be signalled by T-marking (Section 6.3). Repetition is, however, not a pertinent feature of INTRA, and INTRA items are not necessary to express repetition. Hedin (this volume) emphasizes that, though the ‘habitual’ use has led to the conclusion that repetition is part of the IPFV semantics, mere repetition does not call for IPFV. O2 may be part of a longer interval expressed by a temporal adverbial such as in Turkish PAST (INTRA) Dün ya˘gmur ya˘gıyordu or Italian Ieri pioveva ‘Yesterday it was raining’. The corresponding PAST ( INTRA) items ya˘gmur ya˘gdı or ha piovuto ‘it rained’ do not suggest any included O2 . Often, [ dyn] actional phrases expressing permanent states combine with PAST (INTRA), e.g., Turkish Ev, Roma’da bulunuyordu, Italian La casa si trovava a Roma ‘The house was situated in Rome’. Here, the state is viewed as being the case (valid, obtaining) at a temporally included O2 . Frequential expressions do not block the use of PAST (INTRA) items, e.g., French J’y allais cinq fois par semaine, Turkish Haftada be¸s defa oraya gidiyordum, Modern Greek Píjena ekí pénde forés tin evdhomádha ‘I went there five times a week’. The decisive factor is that the event was the case during an interval that serves as a point of introspection. In narrative discourse, PAST (INTRA) items play an important part within the so-called incidential schema of overlapping events, denoting an event that has already begun and is taking place when another event begins, e.g., éghrafe ‘was writ-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

81

Viewpoint operators in European languages

81

ing’, pˇrechod´zowach so ‘was walking’ and udïl’ ‘was sleeping’ in Modern Greek Éghrafe éna ghráma ótan bíka ‘He was writing a letter when I entered’, Upper Sorbian Pˇrechod´zowach so po lˇesu. Nadobo stupich na hada ‘I was walking in the forest. Suddenly I stepped on a snake’ and Mordvin Zahar sovas’ kudov. Elena ešˇco udïl’ ‘Zahar entered the house. Elena was still asleep’ (Serebrennikov 1960: 229). Sequences of PAST (INTRA) items may express complex situations in which several events are going on simultaneously. 7.2.2. Uses of PAST ( INTRA)

PAST (

INTRA) items are by no means restricted to single events of limited duration, as often assumed in the traditional literature. ‘Nondurative’ is not a pertinent feature of their use. However, durative events are not viewed in their course, but rather summarized, e.g., Kalmyk PAST( INTRA)) gazet umšv ‘read newspapers’ as against PAST (INTRA)) gazet umšžala ‘was reading newpapers’. Nor is ‘noniterative’ a pertinent property. PAST ( INTRA) items are readily employed in sentences with global events containing repeated subevents, e.g., Italian è venuto ogni giovedì, Turkish her per¸sembe günü geldi ‘came every Thursday’, or with explicit reference to a definite number of times, e.g., French (il) y est allé cinq fois, Turkish be¸s defa oraya gitti, Modern Greek píje ekí pénde forés ‘went there five times’. The exact delimitation of an event naturally combines with a INTRA view, though a INTRA view can also, as we have seen, be applied. In y allait cinq fois par semaine ‘went there five times a week’, the global event contains an indefinite number of subevents, each constituting a weekly procedure consisting of five occurrences of a basic event. It is the global event that is characterized aspectually from the viewpoint of a temporally included O2 . The overt quantification of occurrences is not decisive for the aspect choice. Moreover, ‘situational change’ or ‘the view of the event as a completed whole’ are not pertinent features of the nonintraterminal varieties of PFV. PAST ( INTRA) items of various languages are readily used for events that are neither transitional nor temporally delimited, e.g., Turkish geçen yıl çok çalı¸stı, Modern Greek pérsi dhúlepse polí ‘worked much last year’. INTRA does not signal completion, but typically suggests totality with transformatives. As for entities expressed by objects, INTRA as such does not specify whether or not they are totally affected (finished, consumed, etc.). Such contextual meanings presuppose undivided reference to the entity in question, i.e., require that the referent is totally involved in the actional content as its minimal-maximal mensural unit. As unmarked members of their oppositions, INTRA items possess a negative and a neutral value. The negative value of a PAST ( INTRA) item enables it to present an event as ‘becoming the case’, i.e., situation-changing. With certain actional contents it may suggest ‘initial attraction’, with others ‘final attraction’, realizations of-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

82

82

Lars Johanson

ten interpreted as “ingressivity” and “resultativity”, respectively. In such cases, the Modern Greek Aorist, as Seiler remarks, “provoquera toujours l’impression d’une forte coupe” (1952: 74). The neutral value of INTRA disregards any situational change, presents the event in an integral way and is thus a kind of “natural” aspect for the actional phrase. The neutral value allows complexive, constative, summarizing uses. It was said in traditional literature that the Classical Greek Aorist, in its complexive function, expresses the view of the event in its totality, contracting the cursus, as it were, into one point, surveying it with one glance from the beginning to the end, and concentrating a set of subevents into one point (Brugmann 1900: 476). When required, the negative value is activated, contrasting with the INTRA value. Similar phenomena are observed in many languages, e.g., in the Akkadian ‘constative’ ikšud, a INTRA (‘noncursive’) item with the negative value of a forme historique, ‘he conquered, and then ’ as well as a neutral value, ‘he has conquered, he conquers in general’ (Rundgren 1963: 57). However, the use of a PAST ( INTRA) item ultimately depends on the number and nature of the interacting items present in the system. If it only interacts with a PAST (INTRA) item, it covers a wider functional range than if it also interacts with a PAST (POST) item. PAST ( INTRA) items only interacting with a PAST (INTRA) denote past events both in a historical and a diagnostic way, e.g., Turkish yazdı, Maltese kiteb ‘wrote, has written’. PAST ( INTRA) items are main devices in narrative discourse. Thus, in literary French, the Passé simple (Simple Past) fulfills the function of a propulsive tense, on which the advancement of the plot rests. PAST ( INTRA) items may denote sequences of temporally successive events, conceived of as linked to each other, e.g., Latin veni, vidi, vici ‘I came, I saw, I conquered’, but also temporally overlapping events, e.g., Ali konu¸stu, biz de dinledik ‘Ali spoke and we listened’, Portuguese A Maria contou uma história e o Pedro ouviu-a ‘Maria told a story and Pedro listened to it’ (Oliveira & Lopes 1995: 100). A PAST (INTRA) item is used when one of the events includes the O2 of the other, e.g., Turkish Ali konu¸stu, biz de dinliyorduk ‘Ali spoke and [while he was speaking] we listened’. Within the incidential schema of overlapping events, PAST ( INTRA) items denote an event that begins while another event is already taking place, e.g., stupich ‘stepped’ in the Upper Sorbian example cited above. 7.2.3. Allegedly propulsive uses of PAST (INTRA)

In several European languages, PAST (INTRA) seems to intrude into the domain of items capable of viewing an event as a totality and to have a propulsive effect with regard to the progression of events in the sequential taxis of narratives. Such uses (“narrative imperfect”, “imparfait pittoresque”, “imperfectum rupturae”) are wellknown in stylistically marked contexts, often in specific discourse types. A question

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

83

Viewpoint operators in European languages

83

often posed is whether the INTRA value is neutralized here or the orientation of the opposition is inversed to the effect that PAST (INTRA) is its unmarked member. An Imperfect such as Italian cantava ‘was singing, sang’ may indeed seem to be an unmarked Past tense as opposed to cantò (Vincent 1988: 300). In the present framework, these uses are still considered realizations of the value INTRA and not neutral “simple denotative” uses. The PAST (INTRA) items still refer to a state obtaining at an O2 , though with a technique of presentation that imitates, for the purpose of narrative intimacy and vividness, the one-dimensionality of the synchronic report (Johanson 1971: 262–265). The event is shown as a synchronically observed scene without limits, which creates an impression of direct perception. The INTRA view is not tied to a given O established by the narrative, but the plot is advanced by changing points of orientation, moving along the time axis and allowing for successive insertions of the intraterminal perspective. The point of view is mobile, whereas the view of the event remains introspective. This use of INTRA may suggest a direct leap into the cursus and thus create an ‘in medias res’ effect, an unexpected perspective on the event as already going on, e.g., French A minuit il mourait ‘At midnight, he was (already) dying’ (Pollak 1960: 145– 151) or East Armenian K’iˇc heto na ijˇnum e¯ r depi jor ‘Shortly afterwards he was (already) walking down to the gorge’ (Kozintseva 1995: 282). The dwelling before the finis produces an effect of non-advancement, and the aspectotemporal situation may be interpreted as ‘already obtaining’ or as ‘still obtaining’. The former reading is by no means ingressive. Turkish birdenbire uyuyordu ‘was suddenly asleep’ does not express the initium, but offers, by contrast to uyudu ‘fell asleep, slept’, a sudden introspection into the cursus. For combinations of soudain ’suddenly’ with French Imperfect, see Sten (1952: 261–262). Languages void of PAST (INTRA) items may achieve a similar effect by using a [ t] verb instead of a corresponding ingressive [t] verb (specifying the initial phase of the same action). German Plötzlich schlief er ein ‘Suddenly he fell asleep’ denotes the beginning of the action ‘to sleep’, whereas Plötzlich schlief er ‘Suddenly he was asleep’ does not (Johanson 1975: 147). A certain time interval may constitute the focus of interest and serve as O. Its overt expression may then function as the topic of the sentence, whereas the following comment conveys what is characteristic of O, i.e. what is ‘already’ or ‘still’ the case there, e.g., Italian PAST (INTRA) Due secoli fa nasceva a Bonn L. van Beethoven ‘Two centuries ago, L. van Beethoven was born in Bonn’ (Bertinetto 1987: 75). Hedin (this volume) cites Modern Greek cases in which a time interval is focused on, for example, in a curriculum vitae: PAST (INTRA) To 1945 pandrevótan ‘In 1945, (s)he got married’ as opposed to PAST ( INTRA) To 1945 pandréftike ‘(S)he got married in 1945’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

84

84 7.2.4.

Lars Johanson

INTRA choice with time expressions

The choice of INTRA items with time expressions is a complex matter that cannot be dealt with at length here. The situation is hardly ever so simple that a certain adverbial occurs exclusively with INTRA or INTRA items. Time adverbials as contextual markers offer numerous problems that can only be hinted at here, in particular, whether they refer to O, the global event, or its subevents. Even the role of temporally delimiting adverbials depends on which of these three units they delimit. Both INTRA and INTRA occur with expressions of ‘duration’ and ‘punctuality’ (for Modern Greek, see Seiler 1969: 6). Punctual expressions such as ‘suddenly’ are compatible with INTRA if they refer to the point of introspection, e.g., Turkish Birdenbire gülüyordu ‘Suddenly [it was observed that] (s)he was laughing’. Compare the restricted combinability with AD. Intraterminals naturally cooccur with O-referring adverbials such as ‘just’, ‘right now’, ‘right then’, and with adverbials expressing that the initium is not later than O (‘already’) or that the finis is not earlier than O (‘still’): is already writing, is still writing. Compare the similar use of AD. With ‘X times’ expressions, PAST ( INTRA) is required to represent the whole global event including its final limit, e.g., Turkish üç kere bana yazdı ‘wrote to me three times’. Note that PAST (INTRA) is excluded in such cases, whereas both AD and AD items can be used. With ‘in X time’ expressions, which measure accomplishment time and identify the actional phrase as [t], PAST ( INTRA) items are usually chosen, e.g., repaired the car in three hours. They reject PAST (INTRA) items, since the indication of the outer measures of an event is not reconcilable with its presentation from an internal point of view (Johanson 1971: 229). Compare the similar use of AD. ‘In X time’ can only cooccur with a basically [t] actional phrase, if the latter is reinterpreted  [ser] [ t] and the adverbial refers to each subevent. Then, PAST (INTRA) is also possible, e.g., French toujours réparait la voiture en trois heures ‘always repaired the car in three hours’. ‘For X time’ expressions, which measure the temporal extension and identify the actional phrase as [ t], prefer PAST ( INTRA), e.g., played chess for one hour, repaired the car for three hours. Exact indications of the outer measures of the event are mostly infertile with a INTRA view, e.g., *When I entered, they were playing chess for an hour. Compare the natural use of PAST ( AD) to present an event as extending over a period of time. This choice shows an essential difference between IPFV items of the INTRA and AD types. It is thus wrong to claim that ‘for X time’ necessitates a PFV view. ‘For X time’ may cooccur with INTRA items if the actional content has a [ser] reading and the adverbial can be taken to refer to each subevent, e.g., French Ils jouaient toujours aux échecs pendant une heure ‘They always played chess for an hour’. Low-focals are generally used in such cases. Some high-focals can only be used for uni-occasional events, for example, the Italian Copula  Gerund periphra-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

85

Viewpoint operators in European languages

85

sis, which is thus incompatible with durative adverbials. Languages with a focality opposition may use a higher item to envisage each subevent as ongoing, e.g., They were playing chess every evening for three years. If ‘for X time’ measures the total extension of a pluri-occasional global event, INTRA is again natural, e.g., Turkish Senelerce, her hafta bana mektup yazdı ‘For years and years he wrote to me every week’. ‘Until tx’ expressions are temporally delimiting in a similar way and choose INTRA, e.g., played chess until dinner. INTRA is possible if the actional content gets a [ser] interpretation. The adverbial then determines the duration of each occurrence, e.g., Turkish Her ak¸sam, saat ona kadar s¸ atranç oynuyorlardı ‘Each evening they played / were playing chess until ten o’clock’. Here, INTRA applies to a pluri-occasional global event composed of temporally delimited but aspectually not characterized subevents. Again, languages with a focality opposition may use a higher item to envisage each subevent as ongoing, e.g., Each evening they were playing chess until ten o’clock. The type ‘from tx to ty’ behaves similarly, if it measures the event itself, e.g., played chess from 9 p.m. to midnight. If it indicates O, the intraterminal viewpoint, it is of course compatible with INTRA, e.g., From 9 to 10 p.m. (= as long as observed), they were playing chess. In many languages, ‘since tx’ expressions refer to an event that is still going on and viewed intraterminally at O, ‘tx’ determining its initium, e.g., Italian ballava da mezzogiorno, Spanish bailaba desde el mediodía ‘had been dancing since noon’. The abtemporal expression ‘since X time’, which measures the time between the initium and O, may be used similarly (Bertinetto, this volume). Compare the corresponding uses of AD.

7.3. Degrees of focality Intraterminals display higher and lower focality degrees. Focality concerns the concentration (focus) of the psychological interest on the situation obtaining at O, the core of “nunc”. All intraterminals refer to events that are relevant within an interval including O, but they may differ with respect to the relative narrowness of the range of vision determined by “nunc” (Johanson 1971: 130–134). The ‘presentness’ is more or less focal: from the narrowest idea of an interval confined to the immediate proximity of O to broader ideas of an expanded and even infinitely expandable period of time. Focality is a scalar notion, and focality values are relative. The focality values of individual items are also subject to constant diachronic change. If focality degrees mirror the temporal validity of events, it is only in a relative sense. For example, the temporariness conveyed by high-focals is not restricted to events going on at O in the most concrete and unbroken sense. I am writing a book can be felicitously

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

86

86

Lars Johanson

uttered at a moment when no concrete writing or preparatory work is going on. Are you still playing the guitar? may be asked at a moment of total musical inactivity and only imply a relative temporariness, delimiting an expanded ‘presentness’ from a still broader one. A higher focality degree may also be chosen to express a higher internal dynamicity, intensity or actional density of what is going on at O. Such effects depend on the conceptualization of the actional content. For example, actional density is less likely to be produced with [ dyn] actional phrases. There have been many attempts to classify items of different focality degrees in European languages and at defining them in a general way by means of ontological and situational characteristics. The results are unsatisfactory, partly due to the lack of clear criteria for the contexts in which the types are supposed to occur. It has not been possible to set up situational types in terms of the focality degrees used to describe them, not even to establish absolute functional stations such as “progressive” in a straightforward way. Focality oppositions do not distinguish ‘processes’ (nontransitional events involving internal gradual change) from ‘states’ (non-transitional events not involving any internal gradual change). The ontological and situational characterization basically depends on actionality and is ultimately a matter of interaction of aspect – of higher and lower focality degrees – with different types of actional content. As a rough classification of ‘presentness’ ascribable to events going on at O, it might be assumed that the range of vision can be (i) narrow, (ii) expanded, or (iii) open. It may thus concern (i) uni-occasional events, basically confined to the immediate proximity of O and actually performed there; (ii) uni- or pluri-occasional events, not confined to the immediate proximity of O, but actually performed there; or (iii) uni- or pluri-occasional events, in principle being the case at O, but not actually performed there. The tentative focality scale adopted here has three positions or cardinal degrees: relatively high focality (HF ), relatively low focality (LF ) and nonfocality (NF ). The classification is based on notions developed in Johanson (1971), where different degrees of focality (“Prägnanz”) were discerned, and mirrors empirically well-known variants of intraterminals. It is not incompatible with the one found in Comrie (1976: 25), where an “imperfective” aspect is subdivided into “habitual” and “continuous”, the latter category being further subdivided into "nonprogressive” and “progressive”. High-focal roughly corresponds to “progressive”, low- focal to “nonprogressive continuous”, and nonfocal to “habitual”. However, my classification only applies to IPFV items of the intraterminal kind. European nonadterminals do not distinguish linguistically relevant focality degrees. Terms such as “progressivity” and “habituality” will be avoided here, since they are often used in aspectological literature in a substantial sense (e.g., for short-term and long-term states) that has little in common with focality degrees. The meaning of “habituality” in everyday language might also give rise to confusion, suggesting

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

87

Viewpoint operators in European languages

87

that ‘habit’ and ‘repetition’ are necessary features. Nonfocality does not only cover INTRA operating on [ser] actional phrases. On the other hand, as noted above, habits do not necessarily require INTRA presentation as obtaining at an O. The focality degrees will not be referred to as different "aspects”. They are neither special viewpoint operators nor actional content categories. Since the same is obviously true of Comrie’s subtypes, there is no justification for Bache’s assessment of “progressivity” and “habituality” as inherent meanings in the sense of “Aktionsarten”, or for his remark that Comrie’s subcategories are “all definable in terms of inherent meanings rather than as subdivisions of imperfectivity” (1982: 60). 7.3.1. Focality types Intraterminals of high focality, INTRAHF items, focus on the core of “nunc” and events going on there, suggesting a ‘narrow presentness’. They tend to stand for uni-occasional events, i.e., one single occasion of performance, and are particularly suited to convey the impression of internal dynamics, gradual movement towards the finis. These properties are typical of PROG (“progressive”), often considered to be a subtype of IPFV, e.g., was writing. According to Dik, the English Progressive expresses one facet of what may be covered by the “imperfective” of other languages, though the semantic content gets a more specific interpretation (1989: 188–189). Since clear criteria for “progressive situations” are not available, we shall not try to establish a high-focal category delimited from other intraterminals on the basis of situations. The question ‘What is X doing right now?’ may be interpreted in terms of different degrees of narrowness and answered with more than one item. It is not uncommon that two items of different focality degrees both qualify as PROG according to a given situational criterion or that an item qualifying for PROG in one test fails in the next one. Intraterminals of low focality, INTRALF items, do not focus strongly on the core of “nunc” but denote, in the sense of an ‘expanded presentness’, single or repeated, uni-or pluri-occasional events going on there. What is often referred to as a “general present” is characterized by low-focal intraterminality and does not constitute an independent aspect. As noted above, nonfocal PAST items are non-oppositive INTRAo items representing the intraterminal notion in a rather feeble way. This subtype of intraterminality is also represented by Imperfects such as French écrivais or Turkish yazıyordu ‘was writing, wrote’. It is apt to present an event as a ‘continuous unfolding process’ and thus corresponds to Comrie’s negatively defined “continuousness”, namely an “imperfectivity” that is not “habituality” (1976: 26). However, I do not consider it possible to establish an absolute low-focal category on the basis of unequivocal situational criteria. Nonfocal intraterminals, expressing INTRANF , do not focus on the core of “nunc” but denote, in the sense of an ‘open presentness’, single or repeated, uni-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

88

88

Lars Johanson

or pluri-occasional events obtaining there. The event is conceived of as being valid in a regular or characteristic way, without being actually performed at the very moment of introspection. It may thus also be an extratemporal event, an event-type or a general property. Nonfocals roughly correspond to Comrie’s “habituals” which “describe a situation which is characteristic of an extended period of time” (1976: 27–28). Note, however, that it is not the objective extension in time that decides whether the event is ‘characteristic’ or not, as Bache’s interpretation of Comrie’s classification seems to suggest (1982: 61). Nonfocal non-pasts, PAST (INTRAoNF ) items, thus exhibit uses that clearly burst the narrow introspective view. These so-called “general” or “unmarked” Presents can be used for uni-occasional events in temporally limited progress at Os , pluri-occasional events, temporally unlimited events, events referred to as a type, universally valid facts (gnomic use), past events (historical use), fictitious events, foreseen, scheduled, planned, intended events (futurate use), potential events, ability to perform a given action, etc. In all this, they are rather similar to PAST ( AD) items. 7.3.2. Coverage of situations Focality degrees are distributed across language-specific items in rather different ways. As for non-past items, there are, for example, “concrete” Presents representing high and low focality, “usual” or “general” Presents representing low focality or nonfocality, combined “Present-Futures”, general items covering the whole range of current, habitual and timeless events, and “potential” Presents representing nonfocality with modal values (e.g., disposition). For example, both the Georgian general Present and the Turkish potential Present may express timeless events: Georgian Dedamica mzis garšemo brunavs, Turkish Dünya, güne¸s etrafında döner ‘The earth ª turns round the sun’. A lower item on the focality scale is semantically more general than a higher one, indifferent towards the notion of a higher focality. It may thus cover referential areas of a higher item, i.e. be used in cases that allow characterization in terms of higher focality or require a higher item in some other language. If a language has two items of different focality, the lower one may roughly cover ongoing events that are more precisely expressed by the higher one, e.g. English wrote was writing. A relatively low INTRA item in language A may cover situations that require a higher item in language B, e.g., French écrivait English was writing. This is also true of items English was that are indifferent towards intraterminality, e.g., German schrieb writing. Nonfocals may roughly cover situations more accurately expressed by lowand high-focals, e.g., French fume, Modern Greek kapnízi ‘smokes’. Many nonfocals with partly modal functions may also cover higher areas. All these items must, however, be distinguished from special markers of pluri-occasionality, disposition,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

89

Viewpoint operators in European languages

89

etc. (Section 4.2). The latter cannot cover any higher referential areas, as a praesens / imperfectum usuale, generale or potentiale can. Though intraterminals generally operate on global events, high-focals may operate on the subevent level, i.e. apply to subevents of a pluri-occasional global event, e.g., Each time we flew home, he was reading a book, She always calls me when I am working, Portuguese O João está sempre a fumar ‘John is always smoking’ (Oliveira & Lopes 1995: 109, footnote 16). If two items of different focality degrees are available, the lower one may apply to the global event and the higher one to subevents (see “Prägnanzüberlagerung”, ‘focality superimposition’, Johanson 1971: 267–268). Operating on the subevent level is typical of preaspectuals and early high-focals, e.g., Finnish Hän soittaa aina, kun olen siivoamassa ‘(S)he always calls when I am cleaning’. This behaviour is rather natural given the fact that the items develop from modes of action, which modify single basic actions.

7.4. Focality oppositions Numerous European languages exhibit thoroughly grammaticalized focality oppositions. The area of oppositions in this rigorous sense includes most Romance, Turkic, Iranic, and Caucasian languages, English, Icelandic, Irish, Basque, Maltese, Kalmyk, etc. The absolute degrees vary: some oppositions concern higher, others lower focality. On the other hand, many European languages such as Slavic, Finno-Ugrian and most of the Germanic ones make little use of highly grammaticalized devices for distinguishing focality degrees. However, many of them possess preaspectual devices representing lower degrees of grammaticalization and not integrated into the core of the aspectotemporal systems. It may sometimes seem difficult to estimate the degree of grammaticalization and thus to delimit the areas of focality oppositions. The items signalling higher focality show varying degrees of generalized use (“obligatoriness”) and combinability with tenses. Many are used rather restrictively. If all such devices are considered, the areas of focality oppositions will practically cover the whole of Europe. Some of them will be commented upon under 8.8–9. In the non-past stratum, English progressives express high-focal intraterminality within the opposition PAST (INTRAo (FOC)) is writing vs. PAST (INTRAo ( FOC)) writes. In the simple past stratum, however, the high-focal is the only INTRA item of the opposition: PAST (INTRA) was writing vs. PAST ( INTRA) wrote. The opposing simple Present and Past tenses are not “perfectives”, just items characterized by a lower degree or absence of focality. The highly developed English Progressive system is unique within the Germanic group, but Irish has a strikingly similar basic system. In the non-past stratum, there is an opposition of a high-focal progressive Present, e.g., PAST (INTRAo (FOC)) tá ag scríobh ‘is writing’, and a so-called “usual” or “habitual” Present tense, e.g., PAST

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

90

90

Lars Johanson

(INTRAo ( FOC)) scríobhann ‘writes’. In the simple past stratum, however, the high-focal is, as in English, the only INTRA item: PAST (INTRA) bhí ag ól ‘was drinking’ vs. PAST ( INTRA) d’ól ‘drank’. Similarly, in Basque, analytical verbs form one single INTRA item in the simple past stratum, PAST (INTRA), e.g., joaiten nintzen ‘I went, was going’ vs. PAST ( INTRA) joan nintzen ‘I went’. In the non-past stratum, the few synthetical verbs left from an older system may oppose a PAST (INTRAo (FOC)) item to a PAST (INTRAo ( FOC)) one, e.g., egoiten naiz ‘I am staying’ vs. banago ‘I stay’ (Haase 1994: 285). The situation is similar with some preaspectual or early INTRA distinctions, e.g., Estonian PAST (INTRAo (FOC)) on tegemas ‘is doing’ vs. PAST (INTRAo ( FOC)) teeb ‘does’, but PAST (INTRA) oli tegemas ‘was doing’ vs. PAST ( INTRA) tegi ‘did’. Kuryłowicz assumes a corresponding Hittite opposition appiskizzi ‘is taking’ vs. epzi ‘takes’ (1956: 26). The types just mentioned are, however, rather marginal in the overall European picture. Focality oppositions more frequently involve two INTRA items in both strata, a phenomenon sometimes addressed as grammaticalization of IPFV and PROG in one and the same language. For example, Romance languages except French and Romanian employ, apart from their INTRAo Presents and INTRA Imperfects, thoroughly grammaticalized devices signalling higher focality, e.g., items of the type Auxiliary  Gerund / Infinitive. There are thus PAST (INTRAo (FOC)) vs. PAST (INTRAo ( FOC)) oppositions such as Italian sta cantando ‘is singing’ vs. canta ‘sings, is singing’, Spanish está hablando ‘is speaking’ vs. habla ‘speaks, is speaking’, and Portuguese está a falar or est’a falando ‘is speaking’ vs. fala ‘speaks, is speaking’. There are corresponding PAST (INTRA (FOC)) vs. PAST (INTRA ( FOC)) oppositions such as stava cantando ‘was singing’ vs. cantava ‘sang, used to sing, was singing’, estaba hablando ‘was speaking’ vs. hablaba ‘spoke, used to speak, was speaking’. The PAST high-focals are used rather extensively in the languages where they occur. The situation is somewhat different in Turkic and Iranian languages. For example, Turkish has oppositions between higher and lower focal items, e.g., dü¸smekte[dir] ‘is falling’ vs. dü¸süyor ‘is falling, falls’, dü¸smekteydi ‘was falling’ vs. dü¸süyordu ‘was falling’; compare the corresponding Kirmanji oppositions dikevîye ‘is falling’ vs. dikeve ‘is falling, falls’, diketîye ‘was falling’ vs. diket ‘was falling’. The lower items are very frequently used. All Turkic and most of the Iranian languages also display PAST (INTRAo (FOC)) vs. PAST (INTRAo ( FOC)) oppositions including still lower focals, e.g., Azerbaijani yazïr ‘is writing’ vs. yazar ‘writes, will write’, Tati mibaftæn ‘is weaving’ vs. mibafu ‘weaves, will weave’. Corresponding PAST (INTRA (FOC)) vs. PAST (INTRA ( FOC)) oppositions are Azerbaijani yazïrdï ‘was writing’ vs. yazardï ‘wrote, used to write, would write’, Tati mibaftæn bu ‘was weaving’ vs. mibaft ‘wove, would weave’, Talysh håndedä be ‘was reading’ vs. ahåndi ‘read, used to read, would read’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

91

Viewpoint operators in European languages

91

Focality oppositions are found in many other languages. Examples of PAST (INTRAo (FOC)) oppositions are Sami læ båradæmen ‘is eating’ vs. bårat ‘eats, is eating’, Kalmyk umšjˇana ‘is reading’ vs. umšna ‘is reading, reads’. Northeast Caucasian (Nakh-Dagestanian) examples are Chechen (Nakh) moluš vu ‘is drinking’ vs. molu ‘is drinking, drinks’, Khvarshi kulše goli ‘is throwing’ vs. kulše ‘is throwing, throws’, Akhvakh qwarere godi ‘is writing’ vs. qwariri ‘writes’ and Archi zari šiarši vi q’onq’ ‘I am writing a book (just now)’ vs. zari šiar q’onq’ ‘I write a book (in general)’. Certain Caucasian languages use different copulas – an “esse concretum” and an “esse abstractum” – for different focality degrees, e.g., Tabasaran PAST (INTRA (FOC)) licurayi licuri ayi ‘was going’ vs. PAST (INTRA ( FOC)) licuji licuri vuyi ‘went (generally)’. Note that, in spite of their material similarity to English Progressives, the Lithuanian so-called Continuatives (periphrases ‘be’  Present participle) do not regularly convey high-focal intraterminal meaning, but rather tend to express imminence, ‘be about to’. Thus, buvo beraša¬ s is normally interpreted as ‘was about to write’, and not as ‘was (already) writing’. Renewal of focality is often observed earlier in PAST than in PAST items. Thus, Gagauz has one PAST item of higher and one of lower focality, verer ‘gives, is giving’ vs. verir ‘gives, will give’, whereas there is only one established PAST item, veri(r)di ‘gave, was giving’. In some languages, PLUR (pluri-occasionality) and DISP (disposition) markers may combine with members of focality oppositions. The marked member of the English opposition PAST (PLUR (INTRA (FOC))) used to be writing vs. PAST (PLUR (INTRA ( FOC))) used to write signals that the first actant referent was, on each occasion, found in the middle of the given event. Similarly, Irish pluri-occasionality constructions can distinguish “between a progressive event which is taking place at the moment of speaking and a similar event which is performed regularly over a longer period of time” (Ó Baoill 1994: 202), e.g., PAST (PLUR (INTRAo (FOC))) bíonn ag scríobh ‘is usually writing’ and PAST (PLUR (INTRA (FOC))) bhíodh ag scríobh ‘used to be writing’. In Lithuanian, PLUR may also combine with the so-called Continuative, e.g., b¯udavo beraša¬ s ‘used to be about to write’, b¯udavo beparaša¬ s ‘used to be about to write to completion’. The Maltese DISP (disposition) marker ikun signals, with the Progressive INTRAo (FOC), disposition to high-focal intraterminality, e.g., PAST (DISP (INTRAo (FOC))) ikun qed jikteb ‘will be writing’, and, with the Imperfect INTRAo ( FOC), it signals disposition to low-focal intraterminality, e.g., PAST (DISP (INTRAo ( FOC))) ikun jikteb ‘he will write’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

92

92

Lars Johanson

7.5. Oppositions of higher and lower focality The degree of focality signalled in the oppositions may be higher or lower. While no absolute graduation is possible, a rough distinction between relatively high focality (HF) and relatively low focality (LF) may be useful. In some oppositions, the higher item basically stands for a narrow presentness and is not readily used for a expanded or open presentness, e.g., Icelandic PAST (INTRAo (FOCHF )) er að lesa bókina ‘is reading the book’, er sofandi ‘is asleep’ vs. PAST (INTRAo ( FOC)) leser bókina ‘is reading, reads the book’. Some Romance focals are more restricted than others. For example, Italian copula  Gerund items seem to represent a higher focality than motion verb  Gerund or copula  locative Infinitive items. Turkic languages of Europe display oppositions of a relatively low focality, e.g., Turkish PAST (INTRAo (FOCLF)) yazıyor ‘writes, is writing’ vs. PAST (INTRAo ( FOC)) yazar ‘writes, will write’; PAST (INTRA (FOCLF)) yazıyordu ‘was writing, wrote’ vs. PAST (INTRA ( FOC)) yazardı ‘wrote, would write’ (Johanson 1971: chapter 5, 1994: 261–262). A similar picture is found in some non-Turkic languages standing under strong Turkic influence, e.g., Lezgian: PAST (INTRAo (FOCLF )) fizwa ‘goes, is going’ vs. PAST (INTRAo ( FOC)) fida ‘goes, will go’; PAST (INTRA (FOCLF)) fizwaj ‘was going, went’ vs. PAST (INTRA ( FOC)) fidaj ‘went, would go’. The unmarked FOC items represent a highly general, ‘open presentness’, which includes pluri-occasionality and also type-reference, genericity, extratemporality, potentiality, disposition, inclination, tendency, intention, etc. The nonfocal value is thus partly modal. While being neither Habituals nor Futures, the items in question are, by virtue of their value, easily interpreted as having habitual or future time reference. This is a type found in several European languages. A relatively low-focal Maltese intraterminal, signalled by the Imperfect  qed, is opposed to a nonfocal intraterminal, expressed by the bare Imperfect, e.g., PAST (INTRAo (FOC)) qed jikteb ‘he is writing’ vs. PAST (INTRAo ( FOC)) jikteb ‘he writes, will write’; PAST (INTRA (FOC)) kien qed jikteb ‘he was writing’ vs. PAST (INTRA ( FOC)) kien jikteb ‘he wrote, would write’. The nonfocal is often taken for a habitual, though it also gets modal and future time readings. It is a defocalized item similar to the Turkic type mentioned. In some cases, the qed-marked item is used in a still less focal sense, i.e. for pluri-occasional subevents, albeit with a meaning of temporariness, e.g., qed jorqod ‘is (usually) sleeping (now as opposed to earlier)’. This occurs with initiotransformatives forming high-focal POST items of the type rieqed ‘has fallen asleep’ = ‘is asleep, is sleeping’ (see 10.3.1.2). Northeast Caucasian (Nakh-Dagestanian) languages such as Andi, Avar, Archi, and Lezgian possess similar nonfocals of the type PAST (INTRAo ( FOC)). The Lezgian -da form has general and partly modal functions interpretable as ‘habitual’,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

93

Viewpoint operators in European languages

93

‘future’, etc. It has therefore been characterized as a “Future” with a “future / habitual polysemy” (Haspelmath 1994: 276). The above-mentioned Khvarshi item kulse ‘throws, will throw’ displays analogous uses. In several Iranian languages such as Talysh and Tati, the PAST (INTRA ( FOC)) items get similar habitual and modal (potential, counterfactual) readings, e.g., Talysh ahåndi ‘read, used to read, would read’. The further the focal item develops along the defocalization path (7.8), the stronger the modal meaning of the unmarked member will be. Certain languages distinguish more than two degrees of focality. In particular, high focality can be renewed, when the lowest item has developed so far along the defocalization path that it rather belongs altogether to the modal area. Some Turkic languages of the area have renewed high focality by means of periphrases with auxiliary verbs such as yatïr ‘is lying’ and turur ‘is standing’. Besides the Nogai opposition PAST (INTRAo (FOC)) yazayatïr ‘is writing’ vs. PAST (INTRAo ( FOC)) yazadï ‘is writing, writes’, there is a third lower item that corresponds formally to Turkish yazar but surpasses it with respect to modal content, PAST (MOD) yazar ‘will, may, tends to, is likely to write’. Compare Karachai ala turadï ‘is taking’, aladï ‘is taking, takes’, alïr ‘will [etc.] take’. There are corresponding PAST items, e.g., Karachai ala tura edi ‘was taking’, ala edi ‘was taking, took’, alïr edi ‘would [etc.] take’. Turkish also has, besides yazıyor and yazar, a third item which, at least in certain discourse types, represents a higher degree of focality, yazmakta(dır) ‘is writing’, yazmaktaydı ‘ was writing’. Similar trichotomies may be said to exist in Daghestanian languages, if their so-called continuatives, signalling ‘is, was still V-ing’ are taken to represent a high degree of focality. Albanian may express different degrees of focality by (i) copula  intraterminal gerund, (ii) the combination po  Present, and (iii) the plain Present. The narrowest “nunc” is expressed by the high-focal (i), e.g., është duke punuar ‘is just working’. Degree (ii) is defined as nongeneral (“merkmalhaft nicht-generell”, Buchholz & Fiedler 1987: 168) and is functionally rather close to the Turkish Present tense in -iyor. There are also PAST items corresponding to both (i) and (ii).

7.6. Sources and shapes of intraterminals Intraterminals, like postterminals, usually emerge as high-focals by way of grammaticalizing statal expressions. The creation seems to be based on relatively few models, largely identical to the construction types of statal, progressive, or cursus-specifying modes of action mentioned under 4.2. In general, the verb takes a non-finite – participial, gerundial or infinitival – form, mostly provided with a copula. Typically, the point of departure is a nominal form of the verb, e.g., writing, a preaspectual item outside the verb conjugation and originally usable as an adjective. The aspectualization turns it into a genuinely verbal

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

94

94

Lars Johanson

form, e.g., is writing, directly related to write. Compare the preaspectual development to postterminals (8.9). Some examples are the Late Latin esse periphrases, the English Progressive, e.g., Old English wæs wrítende ‘was writing’, Icelandic er sofandi ‘is asleep’, as well as the types represented by Khvarshi kulše goli ‘is throwing’ and Chechen moluš vu ‘is drinking’ (intraterminal converb  ‘is’). The diachronic development often seems to start with meanings of permanence and to pass through iterative stages. The relation between the viewpoint and the cursus is often expressed by locative metaphors of inessive or adessive nature (‘be in’, ‘be at’), e.g., the older English construction is at V-ing, Finnish on lukemassa ‘is reading’, Danish er ved at arbejde ‘is working’, Dutch is aan het lezen ‘is reading’, Icelandic er að lesa ‘is reading’, Basque egoiten da ‘is staying’, Portuguese está a ir or está indo ‘is going’, Turkish çalı¸smakta[dır] ‘is working’, Armenian gnum e¯ ‘is going’, Talysh kårdä ( kårdedä) ‘is doing’, Lezgian fizwa ‘is going’, Bagvalal ig’iyax ira ‘is doing’. Inessive constructions meaning ‘in[side]’ obviously provide the most natural expression of intraterminality. In Basque, the copula is ‘be’ with intransitives and ‘have’ with transitives (verbs requiring the ergative). Typical statives such as the postural verbs ‘stand’, ‘sit’ and ‘lie’ are often used. Many ‘be’ copulas actually originate in such verbs; see Italian sta cantando ‘is singing’ and the adessive Portuguese type está a escrever ‘is writing’. Some Turkic intraterminals go back to actional periphrases based on body position metaphors and consisting of converb in -a  turur ‘is standing, is situated’, e.g., Chaghatay qïladur ( *qïla turur) ‘is doing’, Chuvash yulat’ ( *qala turur) ‘is staying’. Kalmyk uses bä¯ ‘stand, be’ combined with the converb in -d. As we have noted, some modern Turkic languages of Europe have renewed high focality by means of yatïr ‘is lying’. The Maltese focalizing element qed goes back to qieg ed, a participle of qag ad ‘stay, sit’. Compare similar actional constructions in Germanic, e.g., Danish sidder og skriver ‘sits / is writing’, Swedish står och tittar ‘stands / is looking’. Metaphors of movement are also well suited to express a ‘dwelling’ in the cursus. Romance languages such as Italian, Spanish, Catalan, and Portuguese use desemanticized locomotive items of the type ‘go’, ‘come’  Gerund. Turkish alıyor ‘is taking’ goes back to an actional periphrasis converb  yorïr (‘moves’). Azerbaijani alïr and Gagauz alär are of similar origin. Older intraterminals are usually known to us as synthetical formations and often exhibit more reduced, less transparent shapes with less clearly identifiable sources. Synchronically, some Present items even appear to have zero-markers, e.g., Adyghe sä-txï ‘I write’. The etymologies of all these older items will not be dealt with here. Suffice it to say that some of them began their career as iteratives, duratives, intensives, etc., and then became statals and T-markers. Indo-European seems to have renewed its present formations successively by means of statal items. Classical Greek so-called Present stems, e.g., épheuge ‘escaped, was escaping’, are opposed

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

95

Viewpoint operators in European languages

95

to Aorist stems, e.g., éphyge ‘escaped’. On this basis, Modern Greek has developed a full-fledged intraterminality opposition, enlarging the old stem formation to the effect that most verbs have a INTRA stem opposed to a INTRA stem, e.g., ghrafvs. ghraps- ‘write’. There are also suppletive pairs such as trogh- vs. fagh- ‘eat’. The Latin Imperfect leg¯ebat ‘was reading’ has its regular reflexes in all Romance languages, e.g., Portuguese lia. The Slavic imperfective type in -ajo¬ , which came to oppose the old Present, was of iterative origin. Iranian languages form intraterminality with elements such as mi-, di- prefixed to the stem, e.g., Kirmanji PAST (INTRA) dixebitî ‘was working’ vs. PAST ( INTRA) xebitî ‘worked’. Turkic languages formed older intraterminals in -(V)r, e.g., Turkish alır ‘takes’ (later on generally renewed by means of periphrases).

7.7. The way to intraterminality Many European languages display items that have progressive-like meanings without being full-fledged aspectual INTRAHF items. The development of high-focal intraterminals often starts with preaspectuals that preserve older iterative, durative or continuative meanings. Numerous progressive-like constructions in Germanic, Finno-Ugrian, Turkic, etc. exhibit uses atypical of grammaticalized INTRAHF items. Some periphrases express ‘be in the course of’, e.g., French être en train de, Romanian a fi în curs de, Danish være i gang med. Others use verbs meaning ‘hold’, e.g., Norwegian bokmål jeg holder på med å skrive, Yiddish ikh halt in shraybn ‘I am writing’. Hungarian displays a progressive-like device based on inversion, e.g., ment ki ‘was going out’ vs. kiment ‘went out’. It is created by placing the T-marker of a transformative in a postverbal position and adding a specific accent pattern (‘ment ‘ki), e.g., éppen írta alá a levelet ‘was just signing the letter’ vs. éppen aláírta a levelet ‘just signed the letter’. The Ossetic T-marker -cæi also has similar progressivelike uses, e.g., racæicïdi ‘was going out’ vs. racïdi ‘went out’. When a preaspectual progressive-like item develops into an early aspectual INTRAHF item, there are stages where it exhibits both actional and aspectual features. For cases of apparent ambiguity in Turkic, see Johanson (1995). The process in question must also be distinguished from subsequent defocalization processes (7.8), which, in some cases, might even seem to take the reverse direction. Examples of these bidirectional tendencies will be given below. Some languages possess special high-focal continuative items signalling that an intraterminal state has begun and is continued, ‘is, was still V-ing’, e.g., Catalan segueix treballant ‘is still working’ (seguir ‘follow, continue’  Gerund). Daghestanian items of this kind carry markers such as Andi -guža, Archi -mat, Lezgian -z-ma, e.g., Lezgian ksuzma ‘is still falling asleep’ as against the non-continuative high-focal ksuzwa ‘is falling asleep’ (Haspelmath 1994: 269). Kalmyk uses non-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

96

96

Lars Johanson

past and past continuatives meaning ‘keeps, kept V-ing’, ‘is, was still V-ing’, e.g., ködlä bilä ‘was still working’. The Lithuanian so-called Continuatives have different functions, but high-focal continuativity in the non-past stratum may be expressed by combining be- with the particle te-, e.g., teb˙era (be)raša¬ s ‘is still writing’ (Mathiassen 1996). Preaspectuals are less grammaticalized in the sense of having a less generalized use and being easily replaced by unmarked items to refer to ongoing events, even within the so-called incidential schema. This is the case with most Germanic progressive periphrases. Not even items such as war am Essen ‘was eating’ in spoken German varieties are generalized enough to be included in the core systems. Their optional use for ongoing events indicates that they are less grammaticalized (Johanson 1975: 150). Among similar items are Baltic Finnic inessive constructions, e.g., Estonian oli tegemas ‘was doing’, and the Hungarian progressive device based on inversion, e.g., ment ki ‘was going out’. Several Romance items, e.g., the Italian stava per periphrasis, are often easily replaced by simple Imperfects of lower focality. Some other typical differences may be observed. Preaspectuals often combine with ‘in X time’ adverbials, whereas INTRAHF items do not. Many preaspectuals such as Latin esse periphrases, Germanic postural statals, Swedish hålla på att / och constructions, Turkic postverb constructions with tur- ‘stand’, may express durativity in addition to their statal meaning, whereas INTRAHF items do not. Preaspectuals may also apply to [ dyn] actional contents, whereas INTRAHF items require dynamicity. Combination with [ dyn] may yield derogative connotations of impatience with the intensity or perseverance of the event. Such meanings result from durativity, i.e. an actional notion. Preaspectual postural verb constructions may preserve some of the original meaning of the auxiliary and thus be restricted to actions performed in a certain body position, e.g., Dutch zit te kijken ‘is sitting and watching’, Finnish istuu työskentelemässä ‘is sitting and working’. In INTRAHF items, the auxiliaries are desemanticized and thus may be interchangeable. For Turkic postverbial constructions, see Johanson 1992: 30, 1995. Some preaspectuals tend to be closely connected with agentivity, intention, and purpose. INTRAHF items, as genuine viewpoint operators, do not require the event to be purposeful and thus easily combine with non-intentional actional phrases. There are locative-purposive items implying ‘dwelling somewhere in order to V’, in which an intentional notion (‘with the purpose of V-ing’) is added to the meaning of ongoing event (‘be V-ing’), e.g., the Finnish construction Copula  -ma-Infinitive  Inessive. For so-called absentive constructions (‘be away V-ing’), see Ebert (this volume), De Groot (this volume), and Tommola (this volume a). Locative-purposive constructions are by definition infertile with non-agentive, non-intentional actional phrases. Imminential constructions such as German war im Begriff zu schreiben or Lithuanian buvo beraša¬ s ‘was about to write’ often seem to reside in the immediate neighbourhood of high-focal intraterminals or to move in their direction.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

97

Viewpoint operators in European languages

97

While preaspectual items may occur in habitual contexts, i.e. be compatible with pluri- occasionality, INTRAHF items tend towards uni-occasionality. The copula  gerund constructions of the Ibero-Romance languages are not restricted to unioccasionality and thus preserve the old stage found in Late Latin esse periphrases (Dietrich 1973). The same seems to be true of the Finnish locative-purposive construction. However, a subsequent defocalization will increase the tendency towards pluri-occasionality, which is an example of the bidirectional tendencies mentioned above. 7.7.1. Peripheral periphrases Whereas the distribution of INTRA items is restricted, their components may exhibit a relatively high combinability and occur in various peripheral items of a lower degree of grammaticalization. For example, the elements of Late Latin esse periphrases have higher combinability than the high-focal items of modern Romance languages (Dietrich 1973). As mentioned above, some languages use intraterminal elements referring to events that are foreseen to be the case at a posterior O2 , i.e., PAST (PRO) items may combine with items of intraterminal nature, e.g., English will be writing, Modern Greek tha ghráfi ‘will be writing’. Similar combinations are possible with certain Romance progressives, e.g., Italian stare  Gerund, the Lithuanian Continuative, the Turkish -mekte olacak periphrasis, etc. Peripheral items of this kind cannot be dealt with at length in the present overview. Note, however, that auxiliaries involved in complex constructions may themselves take part in INTRA and POST oppositions. The latter will be dealt with in Section 8.9. As is well known, intraterminal participles also take part in lexicalization processes of adjective formation, e.g., English charming. The whole combinatorics of such elements must be omitted here. One important point should, however, be mentioned. In the INTRAHF items dealt with so far, the copula is a non-dynamic auxiliary (‘is’, ‘was’). They imply non-dynamic, inessive readings in the sense of ‘dwelling in the cursus’. Some languages exhibit apparently similar constructions with dynamic auxiliaries (‘becomes’, ‘became’), in the illative sense of ‘entering into the cursus’, e.g., English get going, Turkish gelir ol- ‘set out’ (literally ‘get coming’). The dynamic statals resulting from these combinations have a rather modest distribution in modern European languages. The Turkish type does not, in the contemporary language, express an entering into a uni-occasional event and is thus not high-focal (Johanson 1971: 190–193). Similar differences are also found in certain Romance constructions. In the Portuguese periphrasis estava a escrever ‘was writing’, which expresses a past dwelling in an intraterminal state, the auxiliary is a non-dynamic PAST (INTRA) item. In esteve a escrever ‘wrote’, however, the auxiliary is a dynamic PAST ( INTRA) item. Constructions of the latter kind are indeed not intraterminals, since the event is

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

98

98

Lars Johanson

not envisaged in progress at an O, but nonintraterminals, rather functioning as delimitatives in the sense of ‘come to spend a certain time V-ing’. They can consequently be used with mensural ‘until tx’ and ‘for X time’ adverbials, e.g., Spanish estuvo escribiendo todo el día ‘wrote the whole day’. Dynamic postterminal constructions will be dealt with in Section 8.9. 7.7.2. The Greek development The Greek development from a preaspectual actionality duality into a INTRA opposition is particularly interesting. The Classical Greek Present vs. Aorist distinction was, as we have noted, originally an actional one, but it developed into an aspectual one, the nontransformatives turning into intraterminals and the transformatives into nonintraterminals. The Classical Greek duality is not yet an IPFV opposition in the sense of a viewpoint distinction. The Aoristic [t] items imply a crucial limit. The [tf] items imply a final one and may thus produce a so-called “effective meaning”, e.g., áreto ‘gained, acquired’, épeise ‘persuaded’. The [ti] items imply an initial limit and may thus yield an “ingressive meaning”, e.g., edákryse ‘burst out crying’, ekhár¯e ‘got joyful’, ebasíleuse ‘became regent’. On the other hand, the [ t] items do not imply any limit and thus have a more general meaning of mere occupation with the action, e.g., edákrye ‘cried, was crying’, ékhaire ‘rejoiced’, ebasíleue ‘reigned’. Classical Greek [ t] items are therefore more easily used for single accomplished events than Modern Greek INTRA items. Even [ t] items such as épeithe ‘persuaded / has persuaded’ only denote engagement in the action and do not signal either conation (‘tried / has tried to persuade’) or full accomplishment (‘convinced / has convinced’, ‘succeeded / has succeeded in persuading’). The interacting Aorists have a marked [t] meaning and consequently a more limited frequency. Thus, [ t] epeskeúaze rather means ‘was / has been busy with the production’, whereas [t] epeskeúase means ‘produced / has produced’. These past items are the only eventoriented ones, since the Perfect was still a high-focal postterminal. In early narratives, we often find sets of [t] items denoting discrete events, e.g., éteke ‘gave birth to’ or élipen ‘handed over’, and a concluding [ t] item, e.g. tíkte ‘gave birth to’ or le˜ıpe ‘handed over’. The [ t] items are not inceptives (Koller 1951), since they do not indicate the beginning of the actions expressed by tíktein ‘give birth to’ and leípein ‘leave, let’. They are more likely to highlight the consequences of the event at Os (Ruipérez 1954: 86) or, more precisely, to suggest the opening of a consequence (“une ouverture possible sur un développement ultérieur”; Seiler 1993: 28). The use is based on the simple fact that the Imperfect has the default value [ t], whereas the Aorist is transformative, i.e. [tf] or [ti]. Since transformatives by definition imply a transition, the Aorist may suggest a situational change in the propulsive sense: ‘did (and then)’, which is not possible with the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

99

Viewpoint operators in European languages

99

unmarked, nontransformative Imperfect. The unmarked [ t] item is also used for constative functions, since Classical Greek lacks a low-focal POST item for such purposes. The development in Homeric, Classical, and Hellenistic Greek cannot be dealt with here, but it should be noted that Modern Greek exhibits a clearly different opposition with a wider applicability and essentially changed values. The actionality of the verb pairs has developed into corresponding viewpoint values, though the orientation of the system is inverted. The Imperfect has developed into a INTRA item with a narrower use, the Aorist into a INTRA item with a broader use. This is a more definite change to a genuine viewpoint opposition limited to “la vue d’un procès” (Seiler 1993: 27).

7.8. Defocalization of intraterminals High-focal intraterminals tend to develop into less focal items, acquiring more general and finally modal (prospective, subjunctive, etc.) functions along the line INTRAHF INTRALF INTRANF MOD. Cases of semantic fusion in the sense of a general INTRAF item and other combined items are frequent. This development is a defocalization drift and not a process of perfectivization. It is often accompanied by morphological reduction. As we have noted, newly created marked items are more likely to be expressed periphrastically than older and unmarked ones, which often have reduced shapes and less clearly identifiable sources. The diachronic development of high-focals expanding their range of uses to assume lower functions is rather well-known. When items of higher degrees are degraded, the higher focality is renewed by new items that leave the old items with lower functions. Innovations of higher focality encroaching on the area of PAST (INTRAoLF) items constantly lead to changes of whole systems. Old “worn out” intraterminals may also finally disappear. Compare the similar encroachment upon PAST (POSTLF) items by high-focal postterminals (8.10). The development from high-focals into more general Presents and Imperfects is known from many languages. Maltese jikteb ‘writes’, together with kiteb ‘has written, wrote’ reflecting the West Semitic INTRA opposition (cf. Standard Arabic yaktubu vs. kataba), goes back to a INTRAHF item, a participial stative of nominal origin, which later on penetrated the low-focal area of a former Imperfect, thus assuming the whole INTRA range. It is now a INTRALF item, the high-focality having been renewed by qed jikteb ‘is writing’. The Slavic prip˘ekajo¬ ‘burn’, originally an iterative and then a renewer of higher focality, drove the old Present pripeko¬ into the modal domain of prospectivity (Kuryłowicz 1956: 28). In Irish, Scots Gaelic and Welsh, periphrases have pushed the old simple Presents into the areas of lowand nonfocality. In Turkic, former high-focals have become low-focals, pushing the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

100

100

Lars Johanson

former low-focals, the so-called Aorists, into the nonfocal and modal domains (Johanson 1971: 131–139). In Armenian, the originally high-focal lsum e¯ ‘is hearing’ has been degraded to a low-focal and relegated the old Present ls¯e ‘hears’ into nonfocal and modal functions. Persian m¯ıkunad ‘is doing, does’ has similarly pushed the former low-focal kunad into the modal area. The Lezgian intraterminal in -z(a)wa, e.g., fizwa ‘is going’, which once renewed high focality, is now a low-focal, while the item in -da, e.g., fida ‘will go, goes’, has assumed nonfocal (‘habitual’) and even modal (‘future’) functions, possibly under Azerbaijani influence (Haspelmath 1994: 276). Old intraterminals have vanished in several languages. In Basque, the older Present item is used with only about a dozen of verbs (Haase 1994: 279, 284.). The old Kartvelian so-called Permansive, belonging to the Aorist system, once had lowand nonfocal functions (habitual, general, timeless events), but is absent in modern literary Georgian, where the lower functions are now fullfilled by the Imperfect. Defocalization often seems to take place earlier with PAST than with PAST items. Turkic PAST items of lower focality have generally preserved more indicativity than the corresponding PAST items. High-focal intraterminals are mostly confined to certain types of internal phase structure. While preaspectual progressives of an actional nature may also apply to [ dyn] actional contents, genuine INTRAHF items start with [dyn] actional phrases and are generally infertile with [ dyn]. However, when an INTRAHF item begins to step down the focality scale, it tends to tolerate weaker dynamicity again. When it has expanded to cover all actionality classes including [tf, mom], it has eo ipso ceased to be a high-focal. Similarly, while preaspectual progressives are often also used with non-agentive actional contents, INTRAHF items seem to start in agentive contexts. The compatibility with non-agentive actional phrases increases further down the focality scale. Finally, while preaspectuals may be compatible with pluri-occasionality, INTRAHF items tend towards uni-occasionality. In the course of a subsequent defocalization, a tendency towards pluri-occasionality is observed. These cases are examples of the bidirectionality mentioned, involving one development from actionality into INTRAHF and a reverse development within the defocalization process. It has already been noted that operating on the subevent level is typical of preaspectuals developed from modes of action, which modify single basic actions, and that this option may also be observed with INTRAHF items. Lower intraterminals operate on global events. When a former low-focal yields its central function to an expanding high-focal, it may be left with a residue that is nonfocal and even more or less modal (Johanson 1971: 138–139). It implies disposition (potentiality, inclination, attitude, etc.) and is thus easily interpretable as having habitual or future time reference, though it is neither a Habitual nor a Future. As has already been noted, Turkic languages of Europe possess dispositive items such as Turkish yazar ‘will write, tends to write’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

101

Viewpoint operators in European languages

101

PAST markers turn them into dispositive pasts and counterfactuals, e.g., yazardı ‘would write’. Items situated at the end of the defocalization path display strongly modal meanings, e.g., Tatar barïr ‘will go’ (vs. bara ‘is going, goes’), Chuvash yulˇe ‘will remain’ (vs. yulat’ ‘is remaining, remains’). As we have also noted, the Lezgian so-called Future in -da (Haspelmath 1994: 271, 276) is claimed to exhibit “future / habitual polysemy”. This is not surprising, since it is a nonfocal item with modal functions interpretable as future time reference. It would be more astonishing if it were a Future that expressed habituality. As Haspelmath remarks, the corresponding PAST item “cannot, of course, express past future situations”, and thus only has “the past habitual reading” (1994: 271). This is, however, a restriction we would not expect from a Future. Genuine PAST (PRO) items usually have PAST (PRO) counterparts, prospectives-in-past such as Turkish yazacaktı ‘would write, was going to write’, Bulgarian šteše da cˇ ete ‘would read, was going to read’. Tendencies towards defocalization are observed in different parts of the linguistic map of Europe. Many of today’s low-focals are results of the development PAST (INTRAHF ) PAST (INTRALF). Turkish, whose -iyor items have undergone this development, is far from unique in this respect, but rather typical of an eastern area in which defocalization and renewal of focal intraterminality have taken place in particularly systematic and observable ways.

7.9. Loss of INTRA categories

INTRA distinctions emerge,

undergo various changes and may finally vanish. When studying such developments, it is important to distinguish changes of formal markers from cases where functional oppositions as such cease to exist. A phenomenon of the former kind, observed in Romance and other languages, is that a former PAST (POST) item, e.g., French passé composé, takes over a nonintraterminal function from an older PAST ( INTRA) item, e.g., French passé simple (Section 8.10.2). A phenomenon of the latter kind is the loss of the PAST (INTRA) distinction, signalled by the Imperfect and the Aorist, in most Slavic languages. This destruction started with the loss of the Imperfect, whereas the Aorist has usually been preserved longer as a formal item, before being replaced by a former PAST (POST) item. The PAST (INTRA) distinction is only maintained in South Slavic languages, Bulgarian, Macedonian and conservative Serbo-Croatian varieties, e.g., Bulgarian gledaše ‘was looking’ vs. gleda ‘looked’, Serbo-Croatian pisaše ‘was writing, wrote’ vs. pisa ‘wrote’. Its use in written Croatian is stylistically marked as archaic. Though some Serbo-Croatian dialects have preserved the Aorist, the Imperfect has usually disappeared and the opposition therefore ceased to function. Thus, the spoken standard Serbian “Aorist” is just a PAST item. There are even signs of a merger between

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

102

102

Lars Johanson

Aorist and Present tense forms. Interestingly enough, Italo-Croatian has preserved the Imperfect but lost the Aorist. Here, the old opposition is carried on by means of an item developed from a former PAST (POST) item and taking over the PAST ( INTRA) part. This situation is obviously due to contact with Italian, which has a similar INTRA distinction. The PAST (INTRA) distinction also seems to be endangered in spoken Romanian, where both the Aorist (perfectul simplu) and the Imperfect (imperfectul) tend to be replaced by the former PAST (POST) item (perfectul compus). The Sorbian PAST (INTRA) distinction remained relatively long, but is now lost. The early reduction of verbal categories in East and West Slavic languages and the loss of the PAST (INTRA) distinction laid the ground for the development of the AD distinctions found there. The intraterminal and nonintraterminal items of the past strata vanished. The inherently intraterminal non-past items left underwent a defocalization process down to the nonfocal stage. The postterminal categories to be dealt with in Section 8.10 underwent a corresponding defocalization process down to the nonfocal stage. On this basis, the new idiosyncratic viewpoint distinction of adterminality was established (9.4).

8. Postterminality 8.1. Definition The next viewpoint opposition to deal with is that of postterminality vs. nonpostterminality. Postterminality, POST, envisages the event after the transgression of its relevant limit, post terminum. Transgressing the limit means going beyond it and not only reaching it in the adterminal sense. Nonpostterminality, POST, disregards this view. Postterminals are used to describe events as observed from different orientation points situated after the initium or the finis, depending on the phase structure type. The initium is the relevant limit of initiotransformatives and nontransformatives, and the finis is the relevant limit of finitransformatives. POST items may thus present an event from an orientation point outside its cursus. Postterminality is, like intraterminality, a basic mode of presentation emerging from the narrow perspective of the primary “nunc”, and as such a natural way of conceptualizing the flow of events. In its focal forms, the postterminal perspective is tied to an O at which no event is grasped in its totality. No past phase of the event appears in the range of vision so that it might be regarded directly. This indirect way of presentation implies that the event is, entirely or partly, already out of sight. As with all aspects, however, the phases not envisaged are only “concealed”, i.e. latent. Postterminals do not negate the realization of what they exclude from their more or less narrow view.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

103

Viewpoint operators in European languages

103

POST focuses the attention on a situation obtaining beyond the relevant limit, where the event, whether totally or partially past, is still relevant in one way or another, i.e. extends right up to O, has effects relevant to O, or allows a conclusive judgment at O. The relevance of the event to O does not mean that it terminates at O. Nor does the general POST definition include notions such as ‘change of situation’, ‘state’, ‘result’, or ‘recentness’. If POST items suggest a change leading to a new situation, this is an effect of transformativity and dynamicity. Whether or not the postterminal situation is conceptualized as a state depends on how strongly it is focused, i.e. on the focality degree. If the action itself is telic and resultative, the postterminal situation can be said to result from it. Even if the event has already disappeared from the range of vision, it may have left traces, vestiges, effects observable there. Thus, the well-known fact that children usually acquire PAST (POST) items (e.g., British English and Swedish Perfects) earlier than other past tenses is explained by the naturalness of the postterminal perspective on past events with perceptible effects. However, this kind of resultativity is not part of the general definition. Nor is it decisive whether such possible effects of the event have already been overruled. Finally, it is also unessential whether the event is temporally located immediately prior to O or in a more remote past. POST items of some focality are compatible with and often accompanied by relational adverbs indicating the diagnostic dimension and meaning ‘already’ = ‘not later than O’, e.g., has already left. In languages lacking focal postterminals, such lexical markers give a diagnostic interpretation to other past items. 8.2. Postterminals as anteriors While POST can of course be interpreted in temporal terms, it cannot simply be defined as anteriority relative to O, in the sense of “the position of E relative to R” (Reichenbach 1947: 297). It does not just form ‘relative tenses’ denoting events that are completed before O. Whether the whole event or just part of it may be interpreted as having occurred prior to O depends on the internal phase structure. The feature POST is an instruction to situate the relevant limit of the event posterior to an O, the point of relevance. While POST has a natural affinity with the representation of past events, it is not equal to the feature PAST. A PAST (POST) item can, by virtue of its viewpoint value, refer to an event partly or wholly prior to Os . It consequently involves a component of anteriority as opposed to PAST (INTRA) items, which allows it to compete with PAST ( POST) items in the simple past stratum. On the other hand, it does not signal remoteness in the sense of PAST items. Focal POST items typically remain in the realm of the given deictic centre and do not establish own deictic centres, as PAST ( INTRA) and PAST ( POST) items do.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

104

104

Lars Johanson

Comrie’s description of the “perfect” (PF) as a relation between a state and a previous situation (1976: 52, 56, 62) may seem to concern an event-external time relation and thus to be a tense definition according to his own criteria. It has been criticized as contradictory that an aspect is taken to describe something as being prior to some point on the time-axis. The contradiction disappears if it is not the event itself that is situated anterior to Os , but if the postterminal perspective on it is situated at Os . The definition of PF as a POST item does not vitiate the distinction between aspect and tense. Anterior items can be interpreted more or less diagnostically or historically, i.e. as focusing more on the O (orientation point) or on the L (localization point) of the temporal relation. Postterminals tend to be diagnostic and are thus mostly used as non-narrative devices. Focal PAST (POST) items are rather typical of the synchronic report, and their occurrence in narrative discourse is relatively low. Focal POST items are, due to their indirect way of viewing events, non-propulsive, illsuited to advance the plot in the way historical items do. When used in narratives, they typically represent events which, at “topic time” (O2 ), are accessible to retrospection only and thus do not belong to the narrative event chain at a given discourse level. This of course does not mean that the events they refer to are unessential for the narrative in question. With decreasing focality, POST items become increasingly historical, without focusing explicitly on the localization point. They often refer to events outside strict temporal and sequential settings and are less inclined to combine with expressions that specify the occasion on which the event has taken place.

8.3. Postterminality oppositions Postterminals are extremely widespread. Even all known creole systems exhibit POST and POST items, though the latter are less temporalized. In the temporalized European systems, items expressing natural viewpoints such as intra- and postterminality get contrastive values only if they have competing categories to interact with in the same temporal stratum. This is, for example, the case when a language has both a primarily oriented postterminal-in-present and a corresponding nonpostterminal past item. Such oppositions are found in many European languages: English, North Germanic, Modern Greek, Albanian, Bulgarian, Macedonian, Iranian, Armenian, Kartvelian, Finno-Ugrian, Turkic, etc. A competing nonpostterminal past item opposed to a PAST (POST) item will be written as PAST ( POST). Examples of the oppositions of the type PAST (POST) vs. PAST ( POST) are Icelandic hefur talað ‘has spoken’ vs. talaði ‘spoke’, Faroese er komin ‘has come’ vs. kom ‘came’, Norwegian har skrevet ‘has written’ vs. skrev ‘wrote’, Swedish har rest bort ‘has gone on a journey’ vs. reste bort ‘went on a journey’, Sami læ båradan

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

105

Viewpoint operators in European languages

105

‘has eaten’ vs. båradii ‘ate, has eaten’, Estonian on teinud ‘has done’ vs. tegi ‘did’, Vepsian tonu ‘has brought’ vs. toi ‘brought’, Low (Meadow) Mari ludïn ‘has read’ vs. ludo ‘read’, vozen ‘has written’ vs. vozïš ‘wrote’, High (Hill) Mari siren ‘has written’ vs. sirïš ‘wrote’, Udmurt mïnäm ‘has gone’ vs. mïnüz ‘went’, Komi-Zyryan munöma ‘has gone’ vs. munis ‘went’, Latvian ir rakst¯ıjis ‘has written’ vs. raksti¯ıja ‘wrote’, Standard Italian ha cantato ‘has sung’ vs. cantò ‘sang’, Spanish ha hablado ‘has spoken’ vs. habló ‘spoke’, Catalan ha vist ‘has seen’ vs. va veure ‘saw’, Basque mintzatu dut ‘I have spoken’ vs. mintzatu nuen ‘I spoke’, Bulgarian xodil e ‘has gone’ vs. xodi ‘went’, Modern Greek éxi pési ‘has fallen’, épese ‘fell, has fallen’, Albanian ka vrarë ‘has killed’ vs. vrau ‘killed’, Armenian grac e¯ ‘has written’ vs. grec ‘wrote’, ª ‘read’, Sorani n¯us¯ıwa ‘has written’ vs. n¯us¯ı ‘wrote’, Kalmyk umšjˇ ‘has read’ vs. umšv Karachai al andï ‘has taken’ vs. aldï ‘took’, Lezgian fenwa ‘has gone’ vs. fena ‘went’, Chechen mella ‘has drunk’ vs. melira ‘drank’, Georgian damiceria ‘I have  written it’ vs. davcere ‘I wrote it’.  PAST (POST) items may differ a great deal from each other with respect to how systematically they are used. Thus, the Lithuanian item yra raše s ‘has written’, is relatively rare in everyday usage, the Simple Past raše ‘wrote’, being the dominant item in the simple past stratum. This is an effect of a relatively high degree of focality of the PAST (POST) item (Section 8.5). Similarly, since the Irish PAST (POST) a scríobh ‘has just written’, the PAST ( POST) item is a high-focal item, tá éis is generally also used in cases where English uses the Perfect or the Pluperfect, e.g., scríobh ‘has written / had written / wrote’. The artificial language Esperanto has corresponding devices, e.g., PAST (POST) estas skribinta ‘has written’ vs. PAST ( POST) skribis ‘wrote’, though it would be difficult to determine the range of their use. Postterminals-in-present, often referred to as Present Perfect, Anterior Present, etc., invite the addressee to situate the relevant limit of the event before a non-past O. Unless there are contextual indications to the contrary, O will by default be identified as Os, i.e. as having primary orientation. Postterminals used with this orientation convey events whose relevant limit has been transgressed at Os (E* before Os ), e.g., Modern Greek éxi fíji ‘has left’. Thus, by virtue of their viewpoint value, they may refer to an event prior or partly prior to Os , and, as such, compete with PAST ( POST) items in the simple past stratum. PAST ( POST) items are characterized by the absence of postterminality, which makes them more event-oriented and suitable as narrative pasts. Since their perspective is not tied to an O that narrows the range of vision, they freely survey the flow of events and may refer directly, in a historical way, to an event at the very interval of its realization, independently of its relevance to any simultaneous or following O. If a PAST ( POST) item is not intraterminal, it may present the historical event including its limits in an integral way, suggesting, unless the contrary is signalled, the occurrence of the event as an unanalyzed totality. Note that this is possible without

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

106

106

Lars Johanson

the marked totality explicitly signalled by AD items, which highlight the crucial limit. A historical PAST ( POST) item emancipates the event from the deictic centre of Os , creating a deictic centre of its own, an O2 somewhere in the text world. PAST ( POST) items are thus typically used for situating events and often refer to events conceived of as occurring on specific past occasions. Consequently, they are less compatible with markers of indefinite, unrestricted or generalized time (‘ever’, ‘always’, ‘never’, etc.) or with O-implying adverbs meaning ‘already’, ‘not yet’, ‘just’, ‘now’, etc., e.g., Norwegian allerede, French déjà, Italian già, Spanish ya, Finnish jo, Lithuanian jau ‘already’. However, they readily combine with temporal expressions specifying the interval of realization, e.g., Norwegian døde i går ‘died yesterday’. PAST ( POST) items do not break the linear successivity, since they are propulsive, capable of narrating chains of events and advancing the plot in narrative discourse. They may, of course, also refer to isolated events outside such chains. The decisive point is that they present an event in a more direct way than focal PAST (POST) items do. PAST ( POST) items are thus not “simple Pasts”, usable for all past events. The tasks just sketched clearly result from the value POST, i.e. independence of an O situated after the relevant limit. They differ considerably from items that may present past events both in a historical and a diagnostic way, e.g., Hungarian lopott, Maltese seraq, Turkish çaldı ‘stole, has stolen’. Two types of PAST ( POST) items will be distinguished below, one plain type and one PAST ( POST ( INTRA)) type (11.6.2–3). Some languages exhibit similar oppositions in the pre-past stratum, distinctions between a non-dynamic-diagnostic item (“plusquamperfectum status”) and a dynamic-historical one (“plusquamperfectum actionis”). One Romance example is the obsolete French opposition Pluperfect vs. Past anterior, e.g., avait fait vs. eut fait ‘had done’. As the loss of the Simple past also led to the loss of the Past anterior, the Pluperfect was the only item to cover the whole pre-past stratum (diagnostic and historical functions). In many spoken varieties, however, the Past anterior has been replaced by a “passé surcomposé” a eu fait (literally ‘has had done’), which, according to Harris, may be used as an optional “perfectivity” marker, e.g., quand il a eu payé ‘as soon as he had finished paying’ (1988: 229). Albanian has an analogous distinction between a Pluperfect and an “Aorist II”, e.g., kishte hapur vs. pat hapur ‘had opened’ (Buchholz & Fiedler 1987: 130–131). Similar oppositions are Udmurt mïnäm val vs. mïnüz val ‘had gone’, veram val vs. veraz val ‘had said’, Chuvash kilnˇecˇ cˇ eˇ vs. kilseˇccˇ eˇ ‘had come’, Turkish gitmi¸sti ‘had gone’ vs. gittiydi ‘went (once, long ago, etc.)’, Lezgian fenwaj ‘had gone’ vs. fenaj ‘went earlier’ (cf. fenwa ‘has gone’ vs. fena ‘went’; Haspelmath 1994: 267).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

107

Viewpoint operators in European languages

8.4.

107

POSTo items

The O of the postterminal perspective may also, like the O of the intraterminal perspective, be separated from the time of encoding and situated at any other point on the time-axis. For example, it may be determined relative to another event (taxis). If the vantage point is a foreseen O2 , the perspective is projected into the future, e.g., French Ce soir, j’ai terminé la lettre ‘This evening, I will have finished the letter’, Dutch Wanneer X komt, is Y al vertrokken ‘When X arrives, Y will have left’. In such cases, the postterminal-in-present is a PAST (POSTo) item, as it lacks a competing historical item with a contrastive value. Postterminals-in-past (referred to as Pluperfects, Past Perfects, Anterior Pasts, Past Anteriors, Preterit Anteriors, etc.), convey events whose relevant limit is transgressed at a past O2 , e.g., Classical Greek egegráphei ‘had written’, Modern Greek íxe ghrápsi ‘had written’, Icelandic hafði talað ‘had spoken’, Sami læi båradan ‘had eaten’, Estonian oli teinud ‘had done’, Vepsian ol’i tonu ‘had brought’, Low (Meadow) Mari vozen ïle ‘had written’, ludïn ïle ‘had read’, Udmurt veram val ‘had said’, Komi-Zyryan munöma völi ‘had gone’, Latvian bija rakst¯ıjis ‘had written’, Latin cant¯averat ‘had sung’, Catalan havia vist ‘had seen’, Italian aveva cantato ‘had sung’, Romanian mersese ‘had gone’, Portuguese partira or tinha partido ‘had left’, Bulgarian beše daval ‘had given’, Tatar bar an idˇe ‘had gone’, Nogai barïp edi ‘had gone’, Karachai al an edi ‘had taken’, Chuvash kalan˘acˇ cˇ eˇ ‘had said’, Turkish yemi¸sti ‘had eaten’, Albanian kishte vdekur ‘had died’, Serbian bio je isao ‘had gone’, Kirmanji hatibû ‘had come’, Sorani n¯us¯ı bu ‘had written’, Chechen mell¯era ‘had drunk’. In general, so-called Pluperfects do not have the same semantic structure as socalled Perfects. They differ from them by signalling PAST, but they are not just “Perfects related to the past”, temporally transposed PAST (POST) items, the only difference residing in the orientation points (O2 vs. Os ). Most European Pluperfects are PAST (POSTo) items since they cover the whole pre-past stratum and lack a competing historical item with a contrastive value. They thus do not share the special properties of the corresponding PAST (POST) items, which have a more restricted use due to the competition with PAST ( POST) items. For example, the Portuguese Pluperfect is not subject to the restrictions typical of the Perfect (8.10.2, 10.3.2). While a PAST (POST) item competes with a PAST ( POST) item, there are seldom oppositions of the type Perfect-in-past vs. Preterite-in past (for exceptions, see 8.3). There may also be more substantial differences between the non-past and past items in question. Thus, Turkish has a Pluperfect in -mi¸sti, e.g., gitmi¸sti ‘had gone’, whereas the corresponding simple finite form in -mi¸s is a predominantly indirective item, e.g., gitmi¸s ‘has [apparently] gone’. A PAST (POSTo) item invites the addressee to situate the relevant limit of the event before a past O2 (E* before O2 ). In X had left when Y came, the transgression

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

108

108

Lars Johanson

has taken place at the O2 marked by the adverbial when he came, whereas in X left when Y came the relevant limit, due to the PAST ( POST) value of left, coincides with the O2 . A time expression combined with a PAST (POSTo) item may refer diagnostically to the O2 , e.g., At two o’clock, X had [already] written the letter, or historically to the event time, e.g., X had written the letter at two o’clock. PAST (POSTo) items are often used in narratives, when the chain of events is broken by an explanation or a regress that conveys a past or partly past event. PAST (POSTo) items may also be employed without a secondary orientation to an O2 . This allegedly “expressive” use “instead of” a simple Past indicates non-relevance or even nonvalidity at Os (8.5.2). Some reduced systems use one single general POSTo item to cover both the simple past and the pre-past stratum, e.g., Hungarian írt ‘has written, wrote, had written’. The lack of a PAST marker may be made up for by lexical disambiguaters meaning ‘earlier’, ‘just (before)’, ‘already’, etc. Pluperfects are lacking in many Slavic languages, in particular in West, East, and Western South Slavic. Some of them may still use PAST markers in certain registers, e.g., Ukrainian buv xodiv ‘had gone’, Belarusan pryjšo˘u by˘u ‘had come’, Polish kupił był ‘had bought’ (almost extinct). Numerous languages that do possess a Pluperfect often substitute simple postterminals for it in past narratives (for Slavic, see Maslov 1980: 54, 58–59). A PAST marker of a Pluperfect may itself be a PAST (INTRA) item, e.g., French avait ‘had’, était ‘was’, a PAST ( POST) item, e.g., Danish havde ‘had’, var ‘was’, or a PAST ( POST ( INTRA)) item, e.g., English had, was. It may also be a PAST (POSTo) item, in which case the main verb is a PAST (POSTo) item as well, e.g., Maltese kien kiel ‘had eaten’, Yiddish iz gehat avekgeforn ‘had gone away’, hot gehat geshribn ‘had written’, Swiss German (Züritüütsch) isch ggange gsii ‘had gone’, Serbian bio je išao ‘had gone’. Languages in which certain varieties have generalized Past tenses of the PAST (POSTo) type, while others have not (8.10.2), may exhibit two corresponding Pluperfects, e.g., German hatte geschrieben and hat geschrieben gehabt ‘had written’.

8.5. Degrees of focality Anterior items can be interpreted more or less diagnostically or historically. A diagnostic reading is O-oriented, focusing the attention on O and stressing the relevance of the event at this point, often supported by adverbials of the type ‘already’. A historical reading is event-oriented, focusing on L and stressing the relevance of the event at the time of its realization, answering questions such as “When did it occur?”. However, the two dimensions are seldom neatly distinguished by morphosyntactic means. Most anterior items represent both sides, the reported event and its validity at O.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

109

Viewpoint operators in European languages

109

Postterminals tend towards diagnostic readings, referring to events that are relevant to an interval including O, e.g., has broken the window vs. the more eventoriented broke the window. However, they can also be more O-oriented themselves, focusing less rigidly on the postterminal situation. They thus display higher and lower degrees of focality, concentration (focus) of the psychological interest on the situation obtaining at O, the core of “nunc”. The range of vision may extend from the narrowest idea of an interval confined to the immediate proximity of O to much broader views. It has already been emphasized that focality is a scalar notion, that focality values are relative, and that the choice also depends on the conceptualization of the actional content. Still, it might be roughly assumed that events whose relevant limits precede O may be presented by POST items in a (i) narrow, (ii) expanded or (iii) open way. The range may be (i) confined to O, the event being conceived of as relevant at O only; (ii) not confined to O, while the event is conceived of as still relevant at O; or (iii) not confined to O, while the event is not necessarily conceived of as relevant at O. Since the scalar focality of POST items depends on the relative narrowness of the range of vision, it is reminiscent of the focality of INTRA items. For example, socalled “resultatives” and “progressives” might be claimed to represent high focality in the postterminal and intraterminal dimensions respectively. In both dimensions, the basic question is to what degree the event is conceived of as part of the “O world”. Since focality values are relative, we do not assume absolute functional stations such as “resultative”, “perfect”, etc. Nevertheless, a tentative focality scale with three cardinal degrees will be adopted here: relatively high focality (HF), relatively low focality (LF), and nonfocality (NF). Note again that the nonfocals are non-oppositive POSTo items that represent the postterminal notion in a weak way. The interpretation of a POST item in terms of diagnostic and historical orientation depends on its focality degree. By and large, high-focals might be said to correspond to “statives” and “resultatives”, low-focals to “perfects” and “constatives”, and nonfocals to more general pasts. However, we assume gliding transitions and areas of overlap between the postulated types. The focality degrees are distributed to items in different ways, and individual items may also represent more than one degree. All language-specific POST items are subject to constant functional shifts along the focality scale. Labels such as “perfect” and “resultativity” will be avoided, since they are often used in a more substantial sense. It seems impossible to establish absolute categories on the basis of unequivocal situational criteria. Our scheme does not reflect any attempt at classifying postterminals by means of fixed ontological and situational characteristics. Such interpretational types are a matter of interaction of aspect (of higher and lower focality) with different types of actional content (Section 10). One word on focality degrees and coverage of situations must be added. Since a POST item of lower focality is more general than a higher one, it may be used in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

110

110

Lars Johanson

cases where some other language permits or requires a higher item. It may cover referential areas of a higher item, i.e. be used instead of it to refer to a given situation, without expressing a higher degree of focality. In languages that do not distinguish focality degrees, items indifferent towards such distinctions may cover situations that would require higher focality in another language, e.g., German hat geschrieben English has written. A nonfocal item in language A may roughly cover situations that would require a low-focal item in language B, e.g., French a écrit English has written. Lower focality does not exclude higher interpretations, e.g., stative readings of sentences such as has arrived = ‘is now here’ or Spanish ha comprado una casa ‘has bought a house’ = ‘owns a house’. To render the static situation described in is already asleep / awake, individual languages may use high-focals, low-focals, nonfocals, or items indifferent to postterminality. A general Past covers the whole range of anteriority. The South German Past tense may cover situations that require focality in other languages. Thus, ist verschwunden ‘has disappeared / disappeared’ may correspond to three Swedish items representing the simple past stratum, PAST (POSTHF ) är försvunnen ‘is gone’, PAST (POSTLF) har försvunnit ‘has disappeared’, and PAST ( POST) försvann ‘disappeared’. Higher items may also cover lower levels in the sense of involving them logically. Thus, Russian vypito ‘is drunk up’ implies that the liquid in question has been drunk or was drunk on some occasion(s). 8.5.1. High-focal postterminality

POSTHF items are strongly O-oriented, putting high focus on the postterminal state obtaining at O after the transgression of the relevant limit of the event. Their range of vision is narrow, restricted to what is still relevant of the event in the “O world”. High-focals are basically restricted to transformatives, i.e. [tf] and [ti] actional phrases. The high focality of the Indo-European Perfect is attested in Classical Greek, e.g., téthn¯eke ‘is dead’, gégraphe ‘has written’. Other examples of PAST (POSTHF ) items are Hittite appan harzi ‘has taken’, Lithuanian yra raše¬ s ‘has writ§ Romany hi mulo ‘is dead’, Turkish ölmü¸s ten’, yra at˙eje¬ s ‘has come [and is here]’, bulunuyor ‘has died, is dead’, Karachai ketibdi ‘is gone’, East Armenian gnacac e¯ ‘is gone’, Kalmyk untsn ‘is asleep’, Akhvakh cˇ ankohe gudi ‘has fallen ill’, Agul aq’unaa ‘is / has done’, Basque joana da ‘is gone’. Corresponding PAST (POSToHF) items are, for example, Hittite appan harta ‘had taken’, Lithuanian buvo atvyke¬ s § ‘had arrived [and was there]’, Romany his mulo ‘had died, was dead’, Turkish ölmü¸s bulunuyordu ‘had died, was [already] dead’, Karachai ketib edi ‘was gone’, Kalmyk suusn bilä ‘was sitting’, Eastern Armenian gnacac e¯ r ‘was gone’. What is often referred to in terms such as “l’état résultant d’une action accomplie” (Kuryłowicz 1956: 26) is thus not taken here to be an independent aspect but just to represent a high degree of focal postterminality. High-focals denote a dwelling in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

111

Viewpoint operators in European languages

111

a postterminal state created by transgressing the relevant limit of the event and still prevailing at O. They often imply that something – the referent of the subject or of an object – has such properties at O that it can be concluded that the relevant limit, leading to this state, has already been transgressed at that point. However, readings with respect to ‘result’ vary according to lexical contents. For example, as Ruipérez shows, the pertinent value of the Classical Greek perfect cannot be “la consideración del estado resultante” (1954: 60). Since high-focals signal property-like states, the number of verbs occurring with them is usually limited to a rather restricted set. Among verbs of a suitable actional content are many denoting spatial and mental transformations, e.g., ‘come’, ‘go out, ‘hang up’, ‘lean’, ‘open’, ‘pass by’, ‘put’, ‘sit down’, ‘call to mind’, ‘convince’, ‘decide’, ‘discover’, ‘fall asleep’, ‘get accustomed’, ‘get tired’, ‘persuade’, ‘understand’, ‘wake up’. The posttransformational state may be reversible or not, e.g., German ist geöffnet ‘is opened’, ist zerrissen ‘is torn’. The postterminal property-like states tend to be perceptible. For example, the event may be manifested by nothing more than traces left at O. This is the source of the well-known indirective meanings connected with many postterminals (8.7). High-focals may also be used for states not preceded by any corresponding real telic event (“quasi-resultative use”), e.g., Russian Dom okružen lesom ‘The house is surrounded by forest’.



PAST ( POST)HF items

8.5.1.1.



PAST ( POST)HF items indicate that the posttransformational state obtains at Os

and may thus get present-like readings, in particular with initiotransformatives, e.g., Classical Greek Perfect dédie ‘[has got frightened and] is frightened’, kékt¯etai ‘[has acquired and] owns’, Archi ustullit q’owdili wi ‘is sitting on the chair’ (Kibrik 1977: 195). As is well known from school grammars, the Classical Greek Perfect corresponds to a Latin “perfectum praesens” and not to a “perfectum historicum”. The Irish high-focal easily gets non-past interpretations, e.g., tá leabhar scríofa aige ‘has written a book’ or ‘has a book written’ (Ó Baoill 1994: 208). The high-focal Old Georgian Perfect is often considered a Present tense. Svan PAST (POST)HF items behave syntactically like Presents (Deeters 1930: 181–182). High-focal postterminals of Baltic languages are usually called “Compound Presents”. Analogously, PAST (POST)HF items indicating that the state obtains at a past O2 may get imperfective-like readings. The syntactic behaviour of high-focals thus tends to be similar to that of intraterminals. However, POSTHF items differ from INTRAHF items by their reluctance to be used for future time reference. The highest focal uses imply a dwelling in the postterminal state without regard to the preceding telic event, however recent. Less high uses allow reference to the event itself. While paying due attention to such differences, we shall not try to set up absolute categories such as “statives” vs. “resultatives” (Nedjalkov 1983) on the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

112

112

Lars Johanson

basis of situational criteria (use in “stative situations” and “resultative situations”). In many languages, however, certain verbs possess special adjective-like forms, totally diagnostic POSTHF items expressing ‘frozen’ states without any dynamic element, e.g., English is gone, French est maigri ‘is emaciated’, Bulgarian e umoren ‘is tired’, Irish tá briste ‘is broken’, Turkish asılıdır ‘is suspended’, Georgian ceria ‘scriptum ª est’. It can thus be concluded that, if the only focal of a given language is a high-focal, the opposing PAST ( POST) has a wide range of use, including functions typical of a “historical perfect”. For example, the Classical Greek Aorist PAST ( POSTHF ) ekt©¡esato, opposed to PAST (POSTHF ) kékt¯etai ‘has acquired, owns’, corresponds to both PAST (POSTLF) has acquired and PAST ( POSTLF) acquired in English. The highest postterminals are naturally compatible with temporal qualifiers of the type ‘still’, ‘as before’, indicating that the state has not ceased to obtain at O (‘remaining in the posttransformational state at O’), e.g., is still gone. They also combine with abtemporal ‘since X time’ expressions, e.g., German ist seit gestern verhaftet ‘is arrested since yesterday’. Like low-focals, they also occur with qualifiers signalling ‘being in the posttransformational state at O already’. Some languages have special high-focal continuative items signalling that the postterminal state has begun and is continued, e.g., Archi abumat bi ‘continues to have snowed’ = ‘has not thawed’ (Kibrik 1983: 114); cf. Lezgian continuative postterminals signalling ‘still’ / ‘any more’ (Haspelmath 1994: 273). High-focal postterminals avoid expressions referring to the historical dimension. They less readily combine with expressions signalling the event time, e.g., German *ist gestern verhaftet ‘was arrested yesterday’. If such expressions do occur, the Oorientedness remains, e.g., Dutch is in 1703 gesticht ‘was founded in 1703’, Swedish är född 1936 ‘was born in 1936’, Modern Greek íne xtisméno to 1900 ‘was built in 1900’. They are mostly incompatible with expressions of iteration (e.g., several times), since the events denoted are, as a rule, uni-occasional. High-focal postterminals may also less readily combine with agentive complements, e.g., the Finnish construction ‘be’  -ttu. Even if an agent is actualized, the O-orientedness is preserved, e.g., Russian napisano inostrancem ‘has been / is written by a foreigner’. The interaction of high-focals with initio- and finitransformatives will be discussed in Section 10.2.1.5. 8.5.1.2. Diathetic conditions The diathetic conditions, which vary considerably, will not be dealt with in detail here. In sentences based upon intransitive POSTHF items, the first actant of the basic valency pattern is realized as subject, e.g., English is gone, Classical Greek béb¯eke ‘has gone’, Icelandic er dottið ‘has fallen [and is lying]’, Lithuanian yra at˙eje¬ s ‘has come [and is here]’, Russian ušedšij ‘is gone’, Macedonian e umren ‘has

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

113

Viewpoint operators in European languages

113

died, is dead’, Romany hi mulo ‘has died, is dead’, Karaim öl’g’˙an’ ‘has died, is dead’, cˇ ïxkan ‘is gone (out)’, Kalmyk üksn ‘has died, is dead’. With transitives, the second actant of the basic valency pattern may become first actant and thus function as subject without any other overt voice shift, e.g., English is closed = ‘has been closed’, Classical Greek érr¯oge ‘is broken’ (rh©¡egnymi ‘break’), hést¯eke ‘stands’ (híst¯emi ‘set’), Lithuanian yra užrašyta ‘is / has been written down’ (užrašyti ‘write down’), Russian postroen ‘is / has been built’ (postroit’ ‘build’), Czech je naps˘an ‘is / has been written’, Macedonian e razbudeno ‘is awake’ (razbud‘wake’), Romany hi xalo ‘is / has been eaten’ (xa- ’eat’), East Armenian p’akac e¯ ‘is / has been closed’ (p’akel ‘close’), Georgian ceria ‘is / has been written’ (cerª ‘write’), Archi ebˇxuli bi ‘is lying’ (eˇxas ‘put, lay’).ª This diathetic versatility is mirrored in the distribution of intransitive and transitive transformatives in Indo-European essive constructions with postterminal participles, e.g., German ist gestorben ‘has died’ (from sterben ‘die’) vs. ist getötet ‘is / has been killed’ (from töten ‘kill’). French verbs such as mourir ‘die’ and tuer ‘kill’ exhibit the same ambiguity. It is also typical of Iranian ergative languages such as Kirmanji and Talysh. The split ergative systems of all Indo-Iranian languages are restricted to postterminals (focal and defocalized). Since the object of transitives gets the same case-marking as the subject of intransitives, postterminal forms of transitives are often claimed to be ‘passives’. Romany postterminal participles show the same diathetic ambiguity, getting an active or passive interpretation according to the actant structure of the verb (intransitive / transitive). They are thus not passive participles, and it is not adequate to claim that the states sit, stand, lie can only be expressed by means of passive constructions (Holzinger 1993: 194). Items such as hi bešdo ‘is sitting’, hi tardo ‘is standing’, hi tšido ‘is lying’ do not require passive interpretations (‘is / has been put, laid’, etc.). A similar diathetic versatility is also found in Mongolian, e.g., Moghol¯ı tebatxsan bi ‘has fallen asleep [= is asleep]’ vs. biˇciksambi ‘is written’ (Johanson 1974b), and in Old Turkic high-focals; cf. reflexes such as Modern Turkish yatık ‘lying’ vs. yırtık ‘torn’. High-focals with inverse orientation do not necessarily go back to passives; nor do they per se indicate ergative structures. Thus, in languages of the type dealt with here, either the first or the second actant of the basic valency pattern may turn up as first actant with transitive POSTHF items. The decisive criterion for the choice is which participant is most clearly affected by the transformation. The change is mostly discernible in the state of second actant referents. The effects of verb contents such as ‘close’, ‘eat up’, ‘fill’, ‘hang up’, ‘wash’, ‘write down’ are usually less characteristic of the agents than of the patients, e.g., ‘window’, ‘cake’, ‘bottle’, ‘picture’, ‘shirt’, ‘word’. The promotion to subjecthood results from the semantic role of a patient affected by the transformation and thus being bearer of a property typical of the postterminal state. There may also, however, be ambivalence with one and the same verb, e.g., Romany hi pilo ‘is drunk,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

114

114

Lars Johanson

has got drunk / is drunk, has been drunk up’, Mongolian idegsen ‘he who has eaten’ / ‘that which has been eaten’ (Grœnbech & Krueger 1955: 33). In most modern Indo-European languages, however, first actants of transitive patterns do not remain first actants with high-focals. Thus, transitive constructions mostly get passive interpretations. The situation is different in languages whose postterminals take part in systematic diathetic oppositions, e.g., Turkish ölmü¸s bulunuyor / öldürmü¸s bulunuyor / öldürülmü¸s bulunuyor (‘is in the state of having died / having killed / having been killed’). In Indo-European languages, originally possessive-passive constructions, implying that the first actant possesses the result of the telic action (‘possesses something being in a postterminal state’), have later been transformed into active high-focals, e.g., Latin habet epistolam scriptam ‘has finished the letter’ = ‘has the letter in a written-down state’), Old Icelandic hefir bókina lesna ‘has finished the book’ (‘has the book in a read-through state’), Swedish har artikeln skriven ‘has finished the article’. This type has undergone later defocalization in most languages, e.g., French Passé composé a écrit la lettre ‘has written, wrote the letter’ (8.10.2). The pattern also seems to have been copied into Basque, where the Perfect shows close structural similarity with the habeo factum item of Late Latin. There is a comparable development in ergative languages lacking a possessive verb meaning ‘have’, where the possessor of the original nominal construction appears as the first actant in an oblique case. 8.5.2. Low-focal postterminality Low-focal postterminals, POSTLF items, imply lower degrees of focality than highfocals. They tie the event less tightly to the “O world”, focusing less strongly on the postterminal phase as a state persisting at the core of “nunc”. They are more eventoriented, expanding the view in the historical direction and directing more attention to the event itself. Whereas high-focals operate on transformatives, low-focals also operate on [ t] actional phrases and may denote single and repeated, uni- and pluri-occasional events. At the same time, they express relevance of the event to the situation obtaining at O. This relevance may be of different kinds: temporal closeness, importance with respect to the effect, some other affinity or psychological link to O. Obviously, such meanings are largely non-objective. Low-focal postterminality does not correspond to a certain situational type (such as some “PF situation”), and, indeed, attempts at finding unequivocal situational criteria to establish an absolute low-focal category seem futile. It is not a pertinent feature of low-focals to express, as is often claimed for PF, a present result of a past event. On the one hand, POSTLF items certainly express the event itself, and, on the other hand, the feature ‘result’ is, again, dependent upon actional factors. Readings of ‘completion’ are restricted to transformatives. The idea that POSTLF items refer to two temporal strata simultaneously (“Doppelzeitigkeit”) seems rather inaccurate. Postterminality

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

115

Viewpoint operators in European languages

115

is the value that determines precisely the connection of a past or partly past event with Os . The PAST (POSTLF) subtype of postterminals is found in many European languages, not only in Germanic ones such as English, Scandinavian, North German and Dutch, or Romance ones such as Italian, Spanish and Portuguese, but also in Modern Greek, Albanian, Bulgarian, Macedonian, Armenian, Georgian, etc., as well as in Iranian, Finno-Ugrian, and Turkic languages. Examples: English has gone, Standard Italian ha cantato ‘has sung’, Catalan ha escrit ‘has written’, Modern Greek éxi érthi ‘has come’, Albanian ka shkruar ‘has written’, Armenian grel e¯ ‘has written’, Basque joan da ‘has gone’, Finnish on lukenut ‘has read’, Estonian on teinud ¯ ‘has spoken’, Votian on j¯onnu ‘has drunk’, High (Hill) ‘has done’, Ingrian ono länd Mari siren ‘has written’, Low (Meadow) Mari lum vozïn ‘snow has fallen’, KomiZyryan kulöma ‘has died’, Talysh håndäše ‘has read’, Kirmanji xwendîye ‘has read’, Georgian dauceria ‘has written it’, Chechen vaxana ‘has gone’, Nogai yaz an ‘has ª written’, Azerbaijani yazmïš ‘has written’, Chuvash kilnˇe ‘has come’, Kalmyk irlä ‘has come’. If the only focal of a given language is a low-focal, the opposing PAST ( POST) has a relatively narrow range of use, excluding functions typical of a “historical perfect”, e.g., English PAST ( POST) acquired as opposed to PAST (POSTLF) has acquired. Compare Classical Greek PAST ( POST) ektesato, which, as was stated above, corresponds to both English items. But low-focals can also have a relatively higher and lower position on the focality scale, implying more or less Os -relevance. Some of them, e.g., the Modern Greek and Armenian ones, are relatively high and used more restrictively than the English one, without being genuine high-focals. The Modern Greek Perfect represents a rather late renewal of focal postterminality (Aerts 1965), has a limited use in several varieties, and does not exhibit any tendencies to be used historically, i.e. in an event-oriented sense. PAST (POST)LF items allow varying contextual interpretations. Alongside their affinity with the present, they function as past tenses, opposed to a PAST ( POST) and sometimes also to a PAST (INTRA) item. The temporal reading is based on the event being past or partly past at Os (E* before Os ). Low-focals are incompatible with adverbials of the type ‘still’, indicating ‘remaining in the posttransformational state at O’, e.g., *has still gone (cf. high-focal is still gone), Armenian *de˙r grel e¯ ‘*has still written’. The types ‘already’ (‘the relevant limit is not later than O’) and ‘not yet’ (‘the relevant limit is later than O’) are allowed. Due to their rather strong diagnostic orientation, PAST (POSTLF) items do not yet present the event in a direct way, but in a rather static manner lacking narrative dynamism. It is typical of their textual behaviour that they are not used for main narrative sequences, but rather for expressing past or partly past events relevant to the discourse “nunc”. On the other hand, they are frequent in conversational discourse. In spite of their relative event-orientedness, PAST (POST)LF items are typically

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

116

116

Lars Johanson

used to refer to isolated events, detached from event chains and other clear settings. They often occur when the order of events is less important, when less attention is paid to the individual occasion, or when no particular occasion is referred to. The latter case includes type-reference to past events and summarizing characterizations in terms of experience concerning several or all possible past occasions. This does not mean that PAST (POSTLF) items refer to an “unspecified past”. They certainly refer to events occurring on specific occasions, but they present them in a less specified way. Non-specificity of this kind may also be suggested by other means, e.g., relatively unqualified devices such as PAST ( INTRA), PAST ( AD), and more general PAST items. The relatively strong Os -dependence thus restricts the use of PAST (POSTLF) items to locate events on specific occasions, to specify their realization intervals. Their combinability with specific past time adverbials varies across languages. It is excluded in English (*has arrived yesterday at six o’clock), and mostly avoided in Scandinavian and Finnish, except in some inferential uses. Though the toleration of more event-oriented expressions in some other languages, e.g., in nearly all Romance languages, may be seen as a sign of lower focality, it does not deplete the meaning of Os -relevance. Many PAST (POSTLF) items are more tolerant of definite localization and thus seem to be less focal than the English Perfect. However, even the relatively high Lithuanian postterminal may be used with locating adverbials such as vakar ‘yesterday’. Estonian on teinud ‘has done’ seems to be more tolerant than its Finnish counterpart, whereas the latter differs from the English Perfect by combining more easily with adverbials indicating a point of time (Tommola 1994: 227). The Bulgarian Perfect allows “perfect of action” uses such as e došel, snošti k˘asno ‘has arrived, late in the night’ (Maslov 1981: 254) and is even considered an “unmarked past” by some linguists (Janakiev 1962: 432, Aronson 1967: 88). In any case, it has been shown that its basic meaning cannot be ‘resultative’ or ‘definite’ (see Friedman 1977: 55–61). One kind of current relevance is recentness, temporal closeness to Os , sometimes explicitly marked by a relational adverbial of the type ‘just’, e.g., has (just) left. But recentness is not part of the definition of low-focal postterminality. Readings such as ‘recent past’, ‘recent change’, ‘new situation’, or ‘hot news’ (McCawley 1971, Comrie 1976: 56–61, Anderson 1982, Dahl & Hedin, this volume), claimed to be typical of PF, are just contextual interpretations of PAST (POSTLF) items. Not all languages possessing a PAST (POST) choose this item to render such meanings. Other items may be preferred, unless it is essential to stress that the effect of the recent event still persists at Os. To convey ‘hot news’ in the sense of has (just, already) left, languages such as Modern Greek, Bulgarian, and Armenian tend to use Aorist Pasts, i.e. PAST ( POST ( INTRA)) items. The sensitivity of PAST (POST) items to temporal distance varies across languages (see, e.g., Squartini & Bertinetto, this volume) and may be a factor in the defocalization drift (8.10.2). A well-known

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

117

Viewpoint operators in European languages

117

example of a specific kind of sensitivity is the hodiernal PAST (POST) type. Thus, in Spanish ha hablado esta mañana ‘spoke this morning’ and Catalan aquesta nit ha dormit bé ‘slept well last night’, PAST (POST ( INTRA)) is used instead of the PAST ( POST ( INTRA)) items habló and dormí, since the period in which the events were performed is conceived of as not yet finished. To avoid the Os -relevance interpretation suggested by a PAST (POSTLF) item, a PAST (POSTo) item without a secondary orientation can be used, e.g., Modern Greek Pluperfect in O papús mu íxe pandreftí téseris forés ‘My grandfather married four times’ (Hedin, this volume; cf. 1987: 22). This use signals non-relevance or even non-validity at Os : the event is not valid any more, is overruled by another past event (‘since the event, other events have occurred’), or its effect is cancelled, possibly by a “twoway action”. In such cases, AD languages naturally tend to use PAST ( AD) items, which do not highlight any limit. 8.5.2.1. O-overlapping uses Contextual readings such as “inclusive” (O-overlapping), “resultative”, and “experiential” depend on the actional content, i.e. result from interaction with different phase structures. O-relevance may exist in the sense that the event has begun in the past and stretches right up to O and overlaps it. Inclusive readings in this sense are possible with nontransformatives and initiotransformatives. The combinations POST  [ t] and POST  [ti] only imply that O, the point where the relevant limit is viewed as transgressed, is situated after the initium. The event has started before O, but it is unclear whether it has taken place in its totality or whether O is included in its duration. For PAST (POST)LF items, this means that the event may still be carried on at Os , the present moment of coding (“perfect of persistent situation”; Comrie 1976: 60, 62). Languages possessing a focal postterminal generally use it in such cases, often together with ‘for X time’ or ‘since X time’ adverbials of duration, e.g., Finnish on asunut täällä jo kolme vuotta ‘has lived here for three years’. The Os -overlapping interpretation may be made unequivocal with contextual devices such as ‘already’. On the use of PAST (POST (INTRA)) items, e.g., English has / had been V-ing paraphrases, to suggest Os-overlapping readings, see 11.4. 8.5.2.2. “Resultative” and constative uses One possible kind of Os -relevance is ‘resultativity’, implying a change that has yielded a result present at Os , e.g., Modern Greek éxi érthi ‘has come (and is here)’. Low-focals are aspectual because they express a postterminal perspective, not because they may have resultative realizations. The difference between resultative and non-resultative readings resides in actional differences. If a resultative reading is suggested, it is due to a [t] phase structure, possibly expressed by means T-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

118

118

Lars Johanson

marking, plus suitable additional lexical properties. Besides being telic, the event must also yield an effect that is valid at Os . For example, has woken up and has broken the window suggest valid effects on the referents of the first and second actants respectively. Such uses are reminiscent of the ‘resultativity’ of high-focals and are restricted to the same types of verbs, e.g., Russian postaret’ ‘grow old’, Armenian hognel ‘get tired’. Combination with [ t] yields non-resultative readings, e.g., Finnish on lukenut sitä kirjaa ‘has read in this book’, with T-marking by means of the partitive object. A further kind of Os -relevance is the constative use, based on a conclusion at Os regarding the event. It includes the “experiential” meaning, which is considered to be one of the main PF functions, implying that “a given situation has held at least once during some time in the past leading up to the present” (Comrie 1976: 58). Constative postterminals present uni- or pluri-occasional events in a complexive, summarizing way: ‘has done (altogether, always, all life long, so far, hitherto, before, ever, often, on some / any occasion, never)’, e.g., has met him (at least once), has married twice (so far), Catalan Vostè ha viscut sempre aquí? ‘Have you always lived here?’, Latvian Vai tu [visp¯ar] esi bijis Pariz¯e? ‘Have you [ever] been to Paris’? (as opposed to Vai tu biji Pariz¯e? ‘Were you in Paris’?). There is no reason to consider this use exclusively “temporal” or less “aspectual” than others. Constative lowfocals are often accompanied by adverbials of the types ‘ever’ or ‘so far’. Thus, Armenian PAST (POSTLF) items usually cooccur with markers of indefinite time such as erbev¯e ‘some time’, orev¯e zamanak ‘ever’ (Kozintseva 1995: 283). Many European languages use low-focal postterminals as constatives. Some of them possess rather specialized constative items. Thus, Turkic postterminals in -ipdir, -mišdir, and -gändir present events in a complexive, summarizing way (often with elements of evidentiality and / or remoteness). The Kalmyk item in -la has similar uses, e.g., Ter neg däkjˇ u¯ lˇencd xarhla ‘I once (some time) have met / met him in the street’ (Benzing 1985: 75). Constative readings are often products of the interaction of low-focal postterminals with [ t] actional phrases, e.g., has seen much. They may, however, also arise with [t] actional phrases, in which case a telic event is similarly summed up at Os . The main element of Os -relevance is then just a conclusion regarding the event rather than the posttransformational state. If constatives with [ t] actional phrases seem to be more event-oriented in the sense of “Perfects of Action”, it is only because they offer no posttransformational state to highlight. They may be used, as is, for example, the Bulgarian Perfect, to suggest that the event is not tied to a specific setting. The same is indeed true of PAST (POST)LF items with [t] actional phrases, but this phenomenon is mostly ignored by grammarians because of the more salient posttransformational (“resultative”) readings possible with such combinations. Certain low-focals are basically restricted to constative uses. The Portuguese PAST (POST)LF item does not exhibit any strongly posttransformational (“re-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

119

Viewpoint operators in European languages

119

sultative”) readings, e.g., *tem escrito a carta ‘*has written the letter’ (Oliveira & Lopes 1995: 108), and is rather limited in its use, basically preferring [ t, dyn] actional phrases, including cases of quantitative reinterpretation [t]  [ser] [ t]. Constative uses are also typical of many varieties of American Spanish. The Turkish finite item in -mi¸s is not a “resultative” Perfect but has, besides its inferential uses, also low-focal constative functions (Johanson 1971: chapter 8). The corresponding Azerbaijani -miš, however, has more “perfect-like” uses than the Turkish item, probably due to influence from the Persian PAST (POST)LF item, e.g., karda ast ‘has done’ = yapmïš (Johanson 1988: 249). The same is true of some Kipchak-Turkic -gän items. Certain languages possess special negated low-focal constatives with the meaning of a perfectum nondum facti ‘has not yet done’. 8.5.3. Nonfocal postterminality Nonfocal postterminals, POSTNF items, do not focus on O, but denote a wideopen anteriority, not tying the event to the “O world” and not implying any particular O-relevance. PAST (POSTNF) items are PAST (POSTo) items, since they lack competing nonpostterminals, general anteriors covering the whole diagnostichistorical range, usable as general retrospective items with primary as well as secondary orientation. Differing from focals by being suited for direct event-orientation, they are used as propulsive items in narratives and combine freely with specific time expressions. They denote single or repeated, uni- or pluri-occasional past events. With respect to actionality, the same tendencies are found as with focals: [t] actional phrases are more suited to highlight the effect of the event than [ t] actional phrases. PAST (POSTNF) items require special lexical markers to distinguish the diagnostic dimension from the historical one. The use of specific time expressions suggests historical readings, e.g., German hat um sechs Uhr gefrühstückt ‘had breakfast at six o’clock’, whereas devices of the ‘already’ type exclude them, e.g., hat schon gefrühstückt ‘has (already) had breakfast’, Czech už napsal ten dopis ‘has (already) written the letter’. Constative readings may be suggested by expressions of the types ‘once’, ‘ever’, ‘already’ (e.g., German schon, schon einmal, jemals, Hungarian már, valaha), e.g., Hungarian Láttad már a kutyámat? ‘Have you ever seen my dog?’. Nonfocals are found in most Slavic languages, in South German, Yiddish, Hungarian, several Romance varieties such as colloquial French, northern Italian dialects, colloquial Romanian, Romansh, etc. Some examples from verbs meaning ‘write’ are Serbo-Croatian je pisao, Italian ha scritto, German hat geschrieben, French a écrit, Maltese kiteb. Some are general POSTo items that also cover the pre-past stratum, e.g., Polish przeczytał ‘has read, read, had read’, Hungarian írt ‘has written, wrote, had written’. Very many Pluperfects are nonfocals, i.e. PAST (POSToNF) items,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

120

120

Lars Johanson

e.g., Dutch had gewerkt ‘had worked’, Estonian oli kirjutanud ‘had written’, Talysh håndäš be ‘had read’, Kirmanji kiribû ‘had done’. Nonfocals emerge from focal postterminals due to loss of nonpostterminals (8.10.2). They function as pasts capable of primary orientation and also maintain their original postterminal capability of secondary orientation. While sometimes called “temporal Perfects”, they are not low-focal postterminals any more. Even Slavic past items taking part in AD oppositions, e.g., Russian (na)pisal ‘has written, wrote, had written’, are former focal postterminals that have lost their focality and become PAST (POSTo) = PAST items. These PAST items might thus also be represented as PAST (POSTo (AD)) and PAST (POSTo ( AD)) items, respectively.

8.6. Oppositions of higher and lower focality Postterminals may form language-specific oppositions with respect to the degree of focality: PAST (POST (FOC)) vs. PAST (POST ( FOC)). The degrees signalled may be higher or lower. Examples of FOC oppositions in which the higher item is a PAST (POSTHF) one (‘the posttransformative state is still prevailing at Os ’) and the lower item a PAST (POSTLF) are English is gone vs. has gone, Irish tá leabhar scríofa aige ‘has a book written’ vs. tá tar eis leabhar a scríobh ‘has [just] written a book’, Icelandic er dottið ‘has fallen (and is now lying)’ vs. hefur dottið ‘has fallen’, French est maigri ‘is emaciated’ vs. a maigri ‘has grown thin’, Finnish ovat väsyneitä ‘are tired’ vs. ovat väsyneet ‘have got tired’, Swedish är bortrest ‘is gone on a journey’ vs. har rest bort ‘has gone on a journey’, North Macedonian e dojden ‘has come (and is here)’ vs. ima dojdeno ‘has come’, Albanian ështe i vdekur ‘is dead’ vs. ka vdekur ‘has died’, Karachai ketibdi ‘has left (and is now gone)’ vs. ketgendi ‘has gone’, Armenian grac e¯ ‘has written’ vs. grel e¯ ‘has written’, Lezgian fenma ‘has, is still gone’ vs. fenwa ‘has gone’, Chechen vaxana bu ‘is gone’ vs. vaxana ‘has gone’, Basque joana da ‘is gone’ vs. joan da ‘is, has gone’. There are also FOC oppositions with passive items such as North German ist gestohlen ‘is stolen (and now missing)’ vs. ist gestohlen worden ‘has been stolen’. Passive constructions of higher focality often contrast with active ones of lower focality, e.g., English is convinced vs. has convinced, Georgian icereba igi ‘this is writª ‘[(s)he] is / has been ten’ vs. dauceria ‘has written it’, Finnish [hänet] on kammattu ª combed’ vs. on kammanut ‘has combed’. Items of different voice do not, however, form proper oppositions on a common basis of comparison. It was stated above that most European Pluperfects are general PAST (POSTo) items, since competing nonpostterminals are lacking. They are thus both diagnostic and historical, exhibiting static, experiential and other uses. Nevertheless, they may be more or less focal. Corresponding focality oppositions of PAST (POSTo) items

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

121

Viewpoint operators in European languages

121

are, for example, English PAST (POSTo (FOC)) was gone vs. PAST (POSTo ( FOC)) had gone, French était maigri ‘was emaciated’ vs. avait maigri ‘had grown thin’, Karachai ketib edi ‘had left (and was still gone)’ vs. ketgen edi ‘had gone’, Armenian gnacac e¯ r ‘was gone’ vs. gnacel e¯ r ‘had gone’, Lezgian fenmaj ‘had, was still gone’ vs. fenwaj ‘had gone’. In Kalmyk, past forms of higher focals -sn bilä contrast with forms of the postterminal in -jˇ, which, however, mostly lack the PAST marker bilä ‘was’. While items such as the Latin Perfect cantavit have lost their postterminality and turned into nonfocals, focality might seem to be more resistent in Pluperfects such as cantaverat ‘had sung’, but the latter is a general PAST (POSTo) item, lacking contrasting items and thus covering all pre-past situations. Though the corresponding former PAST (POST) items have developed into nonfocals, Pluperfects of this kind may be largely restricted to diagnostic uses, highlighting the posttransformational state, rather than being used to refer in a historical way to pre-past events. Thus, Western South Slavic Pluperfects are nowadays restricted to [t] actional phrases and diagnostic uses (cf. Gvozdanovi´c 1995).

8.7. Indirective postterminals The problem of postterminal and ex-postterminal items conveying indirectivity must be mentioned here, though it cannot be dealt with at length. As is well known, certain languages possess evidential categories signalling that a given utterance is based on indirect evidence. Indirectives, often referred to as “reportive”, “narrative”, or “relative” items, are used to express different kinds of dissociation from the event, cognitive or emotional distance to it, non-testimonial reference, assumption, doubt, irony, surprise, etc. While they play central parts in Altaic, Uralic, and several other languages, they are marginal in Indo-European, though represented in certain regions such as the Baltic, Pontic, Caucasian, and Central Asian areas. Certain types of indirectives are closely connected with postterminality. Focal postterminals often get more or less clear readings of indirective evidentiality. PAST (POST) items convey an indirect perspective, expressing the postterminal state emerging from a past or partially past event, and may thus also be used indirectively for past events whose occurrence is only inferred from information available at Os. For example, PAST (POST) items of the eastern Finno-Ugrian languages are characterized as “perfecta praesentis” etc. (Serebrennikov 1960: 59, 165), but have additional indirective meanings, e.g., Komi-Zyryan munöma ‘has [apparently] gone’. The connection between the postterminal state and the event is then established by way of inference: the situation at “nunc” is such that it may be concluded that the relevant limit of the event has already passed. The indirect view creates an element of distance and uncertainty concerning the actual realization of the event.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

122

122

Lars Johanson

The source of information, the basis on which the conclusions regarding the past or partly past event are drawn, may vary. It may be inference from perception or present knowledge, a conclusion based on given facts, effects, vestiges, or hearsay, reports of other persons. This semantic link between indicative and modal uses is a general typological one. Many focal postterminals activate this latent property and exhibit both aspectual and more or less vague inferential uses. Focal postterminals tend towards indirective readings in various Indo-European languages, from Sanskrit down to modern languages such as Armenian and Scandinavian. For inferentiality suggested by PF items in different languages, see Comrie (1976: 108–110). A recent example is the new Macedonian possessive Perfect consisting of ‘have’  participle, which may also get modal readings, e.g., ima napraveno ‘has [presumably] done’ (cf. Graves, this volume). In some Italo-Albanian varieties, the construction corresponding to the literary Albanian Perfect has modal meanings expressing possibility or probability, e.g., ka vdekur ‘may have died’, ‘has presumably died’ (Breu 1991: 55). In Balkan varieties of Romany, the postterminal active participle may be used as an inferential past item. Similar phenomena may be found in the use of the Perfect of certain Kurdish varieties, obviously under Turkish influence. But the indirective shades of meaning occurring with focal postterminals are generally vague and unstable. The indirective capacities are often overstated in the literature, for example, in claims regarding presumptive or inferential uses of Scandinavian Perfects “about a probable event in the past” (Haugen 1976: 80). The modal meaning is often produced by other factors. Tendencies towards indirectivity are often supported by inferential or dubitative expressions, e.g., particles such as Swedish nog ‘probably, presumably’ or Armenian erevi ‘probably’ used with the low-focal postterminals in question. However, certain languages possess comparatively clear-cut and stable indirectives developed from postterminals. Such items are often found in Turkic, Mongolian, Finno-Ugrian, Kartvelian, etc. They have left the focal stage of postterminality and developed into nonfocal indirectives that may also be used historically and serve, in narratives accounting for unwitnessed past events, as propulsive units of the discourse basis. Unlike focals, they thus also combine freely with specific time expressions. Since references to the very circumstances of the event, notably the temporal localization, are not favoured by a more or less strongly O-dependent focality, they tend to block indicative readings and to urge indirective interpretations. Thus, the indirect postterminal perspective is reinterpreted as indirective meaning. This development may already start with inferential uses of low-focals, but in the nonfocal indirective types, the ability to represent the historical dimension is fully developed. This development is typical of Turkic finite items in -gän, -miš, -iptir, etc., which have indirective and certain constative uses. The situation is partly similar in Mongo-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

123

Viewpoint operators in European languages

123

lian. The Kalmyk low-focal in -jˇ may refer to past events in an inferential or reportive way and also express the typical additional meanings of unexpectedness, etc. Former Perfects of the Bulgarian and southwestern Macedonian area show similar functions and are thus obviously not PF items any more. The Albanian so-called admirative goes back to a focal postterminal, differing from it by the reverse order of the ‘have’ auxiliary and the participle. It displays the expected nuances of emotional distance, disbelief, irony, astonishment, and is also used narratively. The presence of an indirective marker of non-attested evidence means that the opposing items get more categorical readings. Moreover, if a postterminal has turned into a nonfocal indirective, it lacks a nonpostterminal item to compete with in the non-modal past stratum. Thus, Turkic PAST ( INTRA) items in -di often lack pure postterminals to compete with and thus have a rather wide range of use. Since Turkish gitmi¸s mainly displays indirective uses, e.g., ‘went / has gone’  ‘apparently’, the item gitti ‘went / has gone’ is employed for almost all past events that do not call for indirective and / or intraterminal marking. The opposition with indirective items does of course not imply that -di only expresses events witnessed by the speaker. In the literary language, the addition of -dir (‘is’ turur ‘stands) to -mi¸s eliminates the indirectivity and creates an indicative item in -mi¸stir. Since -dir corresponds to the Bulgarian copula e, the formal and functional situation seems to be rather similar in the third person: xodil = gitmi¸s, xodil e = gitmi¸stir (Johanson 1971: 305). In spoken varieties, however, -mi¸stir is a presumptive item, which allows for an opposition within the evidential field: gitmi¸stir ‘went, has gone’  ‘presumably’ vs. gitmi¸s ‘went, has gone’  ‘obviously, reportedly’. A similar phenomenon in Latvian is that past indirectives used in dubitative functions carry the auxiliary esot, but may omit it in narrative functions. Many languages, Turkic, Mongolian, Finno-Ugrian, Caucasian, Nuristan languages, Tibetan, etc., possess indirective particles (reportive particles, distance particles) that may produce indirectives corresponding to different aspectotemporal and modal items. Turkic languages possess generalized, temporally indifferent indirectivity markers of the types imiš ärmiš and ekän ärkän, which should not be confused with the postterminal types. Bulgarian bil may function as a generalized marker of indirectivity in a rather similar way. The overall structure of the Bulgarian indirective system indicates considerable Turkic influence, non-Oghuzic patterns being the most probable primary models (Johanson 1998). The Baltic Finnic and Baltic indirective markers are somewhat different, though also of a participial nature. They produce special PAST and PAST items, e.g., of the Estonian indirective mood (“kaudne kõneviis”) sõitvat ‘reportedly goes’, olevat sõitnud ‘has reportedly gone’. The Latvian “relative mood” is based on the finitely used intraterminal participle in -ot ( * -ant), e.g., PAST rakstot ‘reportedly writes’, PAST esot rakst¯ıjis ‘has reportedly written’. Lithuanian has corresponding items, PAST raša¬ s, PAST esa¬ raše¬ s.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

124

124

Lars Johanson

8.8. Sources and shapes of postterminals Postterminals have various though largely similar sources. Many can be traced back to periphrases, which tend to turn into more reduced and synthetic shapes as a concomitant of the semantic development. The old Indo-European Perfect, a high-focal postterminal, was already synthetic in Old Indic and Classical Greek. In some Turkic languages, the analytic item -ib turur (postterminal converb  turur ‘stands’) has boiled down to -b. Analytic and synthetic items may coexist in one system. Thus, Arumanian has a synthetic PAST (POST) item, but an analytic PAST (POSTo) item. Most European postterminals can be traced back to periphrastic essive or possessive constructions, based on the metaphorical models ‘dwelling in a postterminal phase’ (essive) and ‘having something in a postterminal phase’ (possessive). There is no evidence that they go back to completives or to constructions with auxiliaries derived from directional and movement verbs. There are also dynamic versions, ‘entering into a postterminal phase’ (essive) and ‘getting something into a postterminal phase’ (possessive). The postterminal phase is signalled by a non-finite verbal form, e.g. a participle or a converb. Typically, the point of departure of the development is a nominal form of the verb, a preaspectual item usable as an adjective, an adverb or a noun, and not part of the verb conjugation, e.g., Latin scriptum, Hungarian írva ‘written’. The aspectualization turns it into a genuinely verbal form, e.g., scriptum est ‘is written’, directly related to scribere ‘write’. A further example of an item going back to a deverbal nominal form outside the conjugation is the Armenian item in -ac (Kozinceva 1983: 215). Compare the preaspectual development into intraterminals (7.7). Certain languages use locative metaphors such as ‘being (just) after the event’, e.g., Irish tá éis a scríobh ‘has just written’ (= ‘is after writing’). An older item of possessive origin is the Akkadian Stative (Rundgren 1961: 59). The auxiliary of possessive items is often a conjugated transitive verb of the ‘have’ type, whereas essive items are mostly provided with a copula that is a conjugated form of an auxiliary ‘be’, ‘be situated’, ‘stand’, e.g., Hittite eš-, Latin esse, Old English beo-, Old Slavic ¯ Lezgian awa, Sami læt. byti, Armenian em, Turkic tur-, Kalmyk bä-, Essive items with different “nomina perfecti”, postterminal participles, converbs, gerunds, are used in practically all languages, e.g., Indo-European, Finno-Ugrian, Turkic, Mongolian, Caucasian, Basque. A typical model is the Archi combination postterminal suffix -w-  converb  copula, i.e. PAST (POST) -li i, PAST (POSTo) -li edi. Some examples of active readings are Dutch is / was vertrokken ‘has / had left, Finnish on / oli tullut ‘has /had come’, Latvian ir / bija rakst¯ıjis ‘has / had written’, Sorbian je / be pil ‘has / had drunk’, Livonian um / vol’ lu’ggˇen ‘has / had read’, Abaza dcaxjatI ‘has gone’, dcaxjan ‘had gone’, Kirmanji ketîye ‘has fallen’, ketibû ‘had fallen’, Talysh avïm gätä ‘I have taken it’ vs. avïm gätä be ‘I had

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

125

Viewpoint operators in European languages

125

taken it’ (ergative), Lezgian ksanwa ‘has fallen asleep’, ksanwaj ‘had fallen asleep’. Slavic has in general used the old participle in -l, but some modern varieties employ other active postterminal participles, e.g., in the North Russian focal poluˇcivši ‘has received’ (Trubinskij 1984: 216). Some items of the essive type are diathetically ambiguous (8.5.1.2). Some get passive interpretations, e.g., Polish jest napisany ‘is / has been written’, Czech je napsán ‘is / has been written’, German ist gestohlen ‘is / has been stolen’, Faroese er gjørt ‘is done’, Irish tá léite ‘is read’. In Georgian, some postterminals are periphrases with a passive participle  copula ‘be’, whereas others have developed from stative passives, e.g., miceria ‘is written for me’. Many modern languages, Baltic, Finnoª Ugrian, Turkic and others, consistently distinguish active and passive postterminal participles, Latvian ir / bija uzrakst¯ıjis ‘has / had written down’, ir / bija uzrakst¯ıts ‘has been / had been written down’; cf. active vs. passive distinctions in participles such as Estonian -nud vs. -tud, Tatar -gän vs. -ˇelgän, Esperanto -ita vs. -inta. IndoEuropean postterminal participles may not only be diathetically ambiguous, but also have present or past interpretations according to their internal phase structure and thus seem temporally ambiguous (10.3.2). Auxiliaries of essive items often occur in reduced shapes, mirroring their level of grammaticalization. They may also be lacking altogether. The copula of the old Perfect is lost in most Slavic languages, e.g., Polish pisał (Old Polish pisał jest), preserved in Bulgarian, Macedonian, and Sorbian, and can be omitted in certain cases in some languages, e.g., Serbo-Croatian. The copula may often be absent in postterminals-in-present but present, at least optionally, in postterminals-in-past, e.g., Ukrainian xodiv ‘has gone’ vs. buv xodiv ‘had gone’, Vepsian tonu ‘has brought’ vs. ol’i tonu ‘had brought’, Low (Meadow) Mari vozen ‘has written’ vs. vozen ïle ‘had written’, Maltese kiteb ‘has written’ vs. kien kiteb ‘had written’, Kalmyk irsn ‘has come’ vs. irsn bilä ‘had come’. Finally, it may be lacking in the 3rd person of postterminals-in-present but be present elsewhere. This is usually the case with Turkic non-verbal copulas developed from personal pronouns, e.g., in -gän, -miš, -p items. With postterminals-in-present, the absence and presence of the 3rd person copula may also distinguish indirective meanings from non-indirective ones. Compare the similar situation in Bulgarian and the copulaless form of the Latvian “relative mood” as a reportive Past, e.g., rakst¯ıjis ‘is reported to have written’. The possessive type presents the referent of the first actant as the possessor of the posttransformational state. A deverbal nominal denoting a posttransformational state is reinterpreted as a verbal structure and incorporated into an inflectional paradigm. One widespread subtype makes use of a possessive verb meaning ‘have’. The type has broken a window is thus the result of a reanalysis of has a broken window. The subject = possessor remains in the nominative case. Representatives of this ‘have’ auxiliary are, e.g., Latin habere, Old Icelandic hafa, Old English habb-, Italian avere, Portuguese ter, Sardinian áere, Romanian avea, Macedonian ima, Modern

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

126

126

Lars Johanson

Greek éxo, Albanian ka. The auxiliary may be contracted, e.g., Romanian a are, and even omitted. The latter is possible in subordinated clauses in earlier German and still in modern Swedish, e.g., Swedish brevet jag skrivit ‘the letter I have / had written’. The possessive type is common to many European languages, e.g., spoken Latin habet scriptum ‘has written’, French a donné ‘has given’, Italian ha cantato ‘has sung’, Dutch heeft gelezen ‘has read’, Modern Greek éxi ghrápsi ‘has written’, Albanian ka hapur ‘has opened’, Macedonian ima dojdeno ‘has come’. Some Slavic languages, notably colloquial varieties, exhibit more recent items of this type. (See Maslov 1988: 80–85; for Polish, Czech, Slovak, Serbo-Croatian, see Galton 1976; for northern Russian dialects, see Panzer 1984: 115–127). One of the Georgian Perfect types is a combination of a participle with ‘have’, e.g., mokluli m˙qavs ‘I have killed him’. In this paper, we shall not discuss contact-inducedªdevelopments of the possessive type in European languages, but just note that it is even possible that periphrases such as the Old Icelandic and Basque ones are structural copies of the Latin habet factum type. The Basque low-focal analytical construction (present auxiliary  participle) may be an innovation originating from spoken Latin. A second possessive subtype, presenting the possessor as the first actant, is found in ergative languages lacking a possessive verb meaning ‘have’. Like the first one, this subtype also goes back to a deverbal nominal denoting a posttransformational state, reanalyzed as a verbal structure and incorporated into an inflectional paradigm. The possessor of the nominal construction appears as the first actant marked by an oblique case, e.g., Kirmanji min tu dîtî ‘I saw you’. Older Indo-European languages display a rather clear-cut distribution of essive and possessive items. At their high-focal stages, the items in question are restricted to certain actional phrases compatible with the original meanings of the paraphrases. The essive type is primarily used with intransitives, and the possessive type with transitives, e.g., Hittite p¯an ešzi ‘is gone’, p¯an ešta ‘was gone’, hatran harzi ‘has   with the first written’, hatran harta ‘had written’. The auxiliaries tend to agree   actant. Examples of this are the Latin high-focal with hab¯ere, the Old English one with habb- ‘have’, the Old Icelandic ones with hafa and the corresponding esse, beoand vera items. At later stages of development, generalization phenomena can be observed, and the distribution of essive and possessive items varies considerably across languages. Transitives get passive meanings with the essive type, e.g., is written. Transformative intransitives often choose the essive type, e.g., Dutch is gearriveerd ‘has arrived’. Nontransformative intransitives may choose the possessive type, e.g., Dutch heeft geslapen ‘has slept’, or the essive type, e.g., Yiddish iz geslofn ‘has slept’. (Yiddish, however, does not use the essive type for all nontransformative intransitives.) The possessive type has generalized in English and Swedish low-focals, e.g., Swedish har försvunnit ‘has disappeared’, har skrivit artikeln ‘has written the article’. If essive constructions with intransitives are available, they are high-focal, e.g., English is

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

127

Viewpoint operators in European languages

127

gone, Swedish är försvunnen ‘has disappeared, is lost’. Compare high-focal possessive constructions with transitives such as Swedish har artikeln skriven ‘has finished the article, has the article in a written state’.

8.9. Peripheral postterminal items Many components found in postterminals show a high combinability and also occur in constructions that indicate a ‘dwelling in’ or ‘entering into’ a posttransformational phase without functioning as viewpoint operators of the core systems. These less generalized peripheral items exhibit uses atypical of fully grammaticalized postterminal items and must be assigned POSTo values, as they do not compete with nonpostterminals in their respective temporal strata. Some are only actional, specifying postterminal states related to hypothetical actions rather than expressing events. Certain preaspectual items may be difficult to distinguish from viewpoint operators. For some Turkic cases of apparent ambiguity, see Johanson (1995). Baltic Finnic has preaspectual high-focals consisting of a non-finite item (past participle or infinitive  case) and an auxiliary going back verbs meaning ‘become’, ‘remain’, etc. As for prospective postterminals, PAST (PRO (POSTo)) items (“Future perfect”, “Futurum exactum”, “Future anterior”, etc.), instruct the addressee to situate the relevant limit of the event prior to an O2 foreseen at Os , e.g., Latin scripserit ‘will have written’, Icelandic mun hafa talað ‘will have spoken’, French aura vu ‘will have seen’, Modern Greek tha éxi ghrápsi, Bulgarian šte e dal ‘will have given’, Italian avrà cantato ‘will have sung’, Romanian va fi venit ‘will have come’, Portuguese terá ido ‘will have gone’, Maltese ikun kiel ‘will have eaten’, Armenian gnacac e¯ linelu ‘will be gone’, Turkish ölmü¸s olacak ‘will have died’. Correspondingly, PAST (PRO (POSTo)) items instruct the addressee to situate the relevant limit of the event before an O foreseen at a past O2 , e.g., Modern Greek tha íxe ghrápsi ‘would have written’, Bulgarian štese da e cˇ el ‘would have read’, Armenian gnacac e¯ r linelu ‘would be gone’, Turkish gitmi¸s olacaktı ‘would have gone’. Since no competing nonpostterminals are available, these combinations are not simply PAST (POST) items transposed into different time strata. They are absent in many languages, notably in those possessing defocalized PAST (POSTo) items. As was noted above, auxiliaries involved in complex constructions may themselves take part in INTRA and POST oppositions. Examples of the latter are the English distinctions has been vs. was and has become vs. became. The PAST (POST) items may, as expected, be used to avoid the implication of a specific occasion and do not readily combine with specific time expressions, whereas the competing PAST ( POST) items are historically oriented. Combined with items signalling a INTRA or POST perspective, the PAST (POST) auxiliary expresses that the perspective on the event has been valid. The PAST ( POST) auxiliary suggests

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

128

128

Lars Johanson

that the perspective was valid on some past occasion. For example, it is possible to distinguish between ‘has dwelled in an intra- or postterminal state’ and ‘dwelled in an intra- or postterminal state’, e.g., Estonian on olnud tegemas ‘has been doing’ vs. oli tegemas ‘was doing’, Swedish har varit gjord vs. var gjord. Correspondingly, dynamic auxiliaries taking part in POST oppositions allow the speaker to distinguish between ‘has entered’ and ‘entered’ an intra- or postterminal state, i.e. to express that the perspective on the event has become valid or became valid on some past occasion. Certain peripheral constructions involving POSTo elements have developed into more central items. Most of them are inessive and illative constructions based on the non-dynamic and dynamic copula types ‘be’ and ‘become’ and implying ‘dwelling in’ and ‘entering into’ a postterminal state. Dynamic intraterminal constructions were mentioned in Section 7.7.1. Only a few possible postterminal counterparts can be hinted at here. Illative-dynamic poststatals have a modest distribution in European languages, often restricted to lexicalizations, e.g., become interested, get married. There are also genuinely paradigmatic items. Combinations implying ‘entered a postterminal state’ (‘became having V-ed’) are East Armenian gnacac e av ‘went’, Turkish açmı¸s oldu ‘opened’. The dynamic auxiliaries are PAST ( POST) items of verbs meaning ‘be’, ‘become’, or ‘have’. The combinations in questions are not dissimilar to older past anteriors such as French eut fait ‘had done’ or Italian fu uscito ‘had gone out’, whose auxiliaries were PAST ( POST ( INTRA)) items. Passive combinations are, e.g., Estonian sai tehtud, Swedish blev gjord, Turkish yapılmı¸s oldu ‘got done’. Dynamic items of this kind often have the meaning of ‘eo ipso’, e.g., Turkish gitmi¸s oldu ‘thus / therewith went’ (Johanson 1971: 312). Unlike the corresponding non-dynamic items, they are incompatible with adverbials meaning ‘already’, which refer to a state obtaining at O. This is also true of the Romance anterior past types, e.g., French fut parti ‘had left’. Combinations provided with postterminal auxiliaries and implying ‘has entered a postterminal state’ are, for example, East Armenian gnacac e¯ e el ‘has gone’, Estonian on saanud tehtud ‘has become done’. Among combinations implying ‘had entered a postterminal state’ are for example, literary Italian ebbe cantato ‘had sung’, Turkish gitmi¸s olmu¸stu ‘had gone’, East Armenian gnacac e¯ r e el ‘had gone’. As was noted above, PLUR and DISP markers (signalling pluri-occasionality, habituality, disposition) may combine with POST in the sense of ‘has usually done’ and ‘tends to have done’, e.g., Lithuanian b¯udavo raše¬ s ‘used to have written’, Armenian gnacac e¯ linum ‘is usually gone’, Maltese ikun kiteb, Turkish yazmı¸s olur ‘will / tends to have written’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

129

Viewpoint operators in European languages

129

8.10. Defocalization of postterminals Postterminals are subject to successive functional shifts with gradually diminishing degrees of focality, increasing historical orientation, weaker O-relevance. Items characterized by a more diagnostic orientation get more event-oriented functions. This defocalization drift is taken here to include the diachronic steps (1) high fo(2) low focality (3) nonfocality. Semantic fusion in combined items cality is often observed. High-focals tend to develop into less focal items, acquiring increasingly general (historical) and finally modal (irreal, conditional) functions or disappearing. In particular, innovations of higher focality encroaching on the area of PAST (POSTLF) items lead to constant system changes; compare the similar encroachment upon PAST (INTRAoLF) items by high-focal intraterminals (7.8). Well-known examples of the development are found in Classical Greek, where the old high-focal Perfect developed into a diagnostic  historical item, or in Persian, where a high-focal [man¯a] krtam ‘I have done’ developed via a low-focal [man] kart into a nonfocal kardam (Kuryłowicz 1956: 29–30). The evolution of Romance Perfects includes corresponding diachronic stages; see, for example, the scheme laid down in Harris (1982). Turkic defocalization developments have been dealt with in Johanson (1971, 1993, etc.). The origin of Maltese kiteb ‘has written, wrote’, opposed to jikteb ‘writes’ (cf. Standard Arabic – INTRA kataba and INTRA yaktubu) is an old POSTHF item (corresponding to the Akkadian so-called Permansive), which later on defocalized to a PAST (POSTNF ) item and pushed the old Imperfect (yaktub) into modal functions. The high focality has been renewed by participial POSTHF items of the type rieqed ‘is asleep’ (with a limited set of verbs). Standard Arabic creates POST ( INTRA) items by means of the particle qad, e.g., qad kataba ‘has written’. Graves (this volume) presents three Macedonian items of a postterminal nature or origin, occurring side by side and opposed to each other in various ways in the individual dialects: (1) a recent essive item formed with a passive participle, (2) a possessive construction, and (3) an old essive item formed with an active participle. These constructions do not constitute three PF items in a synchronic sense, but hold different positions along the focality scale. In the North, item 1 is high-focal, e.g., umren e ‘is dead’, 2 is limited to possessive cases, and 3 is low-focal. In the South-West, item 1 is losing its focality, 2 is low-focal, and 3 is a nonfocal indirective. Item 2 is generally gaining ground from 3 and has replaced it in some dialects. Defocalization is, as we have noted, due to semantic generalization, leading from narrower to increasingly broader uses. Items developing along such lines become capable of covering more and more situations, taking over functions of older, lower items, gaining ground from them and eventually superseding them. Even if higher focals thus may seem to “take the place” of lower ones, they do not replace them

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

130

130

Lars Johanson

in the sense of assuming the same values. A higher item with relatively specialized functions typically extends its use to subsume both its own old functions and certain functions of a lower item. Narrow uses are typical of young items created to renew the expression of a certain function. High-focals represent earlier stages of semantic development than lower items, which have a wider functional range and the ability to cover functions typical of higher items. It would be misleading to refer to defocalization as perfectivization. The development of PF items into “perfective pasts” in languages with a PFV vs. IPFV opposition (Bybee & Dahl 1989: 74) is not a result of defocalization. First, the development of a PAST (POST) item into a PFV of the nonintraterminal kind depends on the presence of an opposing INTRA item and does not belong to the defocalization process. Secondly, defocalization never leads to the emergence of PFV items of the adterminal kind. This seems natural, since a device denoting the transgression of a limit is not likely to become a device denoting the mere attainment of the same limit. There is no defocalization path leading from a PF item to a perfective of a Slavic type. Certain Slavic T-marked postterminals have developed into AD pasts, but their adterminality derives from their transformativity, not from the former postterminal (“Perfect”) value. A further reason to distinguish PFV items of the adterminal and the nonintraterminal kind is thus that their paths of development are quite different. 8.10.1. From high-focal to low-focal Some of the numerous shifts from high-focal to low-focal postterminality have already been mentioned. For example, the Old Georgian Perfect was high-focal, whereas the modern Georgian one is more event-oriented and vacillates between diagnostic and historical uses. Armenian exhibits two items representing the two stages of development (cf. Maslov 1988). It was also noted that, when higher focals are defocalized, high-focal postterminality is often renewed by new periphrastic items; compare the analogous renewal of high-focal intraterminality. Thus, the former high value of Persian krtam ‘done’ was renewed by kartak am ‘I have done’, which developed into karda am (Kuryłowicz 1956: 29–30). The latter item does not express “l’action perfective pure et simple”, but is simply a low-focal PAST (POST) item. The Turkic items in -gän and -miš seem to have been renewers of focality, possibly filling the place of an older finite item in -ip (Kormušin 1984: 44). As they turned into low-focals, high focality was renewed by periphrases such as -ip turur, -miš turur and -gän turur, which later on developed into constative low-focals themselves (Johanson 1993, 1995). In most Turkic languages, the originally high-focal postterminal in -miš is not used any more. Interestingly enough, however, several modern Iranian languages, when accommodating copies of Turkic verbs to native

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

131

Viewpoint operators in European languages

131

morphosyntactic frames, use do and be constructions plus a -miš form of the copied (typically transformative) verb, e.g., Zaza kandırmiš kärd ‘persuaded’ (cf. Doerfer 1993). The Mongolian items in -jˇugu, -gsan and -luga also seem to have been high-focals. In Literary Mongolian, the item in -luga expressed “a fact of completed action, and the situation thus established” (Grønbech-Krueger 1955: 36). The corresponding Kalmyk item in -la exhibits constative, complexive functions. Whereas Turkic, Mongolian, Iranian, and some Caucasian languages display rich renewing developments of focal postterminals, Germanic, Romance, Slavic and Finno-Ugrian languages are rather unproductive in this respect. For example, since the postterminal periphrasis esm˘ı  participle in -l was defocalized and finally replaced the Aorist, no North Slavic language has developed any reasonably grammaticalized new focal category. It has already been stated that, when high-focals become low-focals, they are not restricted to [t] actional phrases any more, but rather operate freely on all actional types. An interesting fact is the weak defocalization of the Calabrian and Sicilian Perfects, which have retained a higher degree of focality and still only operate on transformatives. The ambivalent diathetic orientation has been given up in some languages. Compare, for example, the Old Turkic type yïrtïq ‘torn’ (preserved in Turkish yırtık) with modern Turkic passive forms such as Turkish yırtılmı¸s ‘torn’. In several Indo-European languages, possessive ‘have’ items have expanded to operate on more actional types than they did as high-focals, gradually encroaching on the territory of essive items, e.g., English is gone has gone. There are several examples of this development in Romance languages, e.g., Spanish somos idos hemos ido ‘we have gone’ (cf. also Catalan, Portuguese and Romanian).The defocalization of the Indo-European possessive type was also accompanied by other syntactic changes. The participle lost its adjectival character and its agreement with the object, became part of the verb phrase and mostly changed its place in the sentence, e.g., Old Icelandic hefir bókina lesna ‘has the book in a read-through state’ hefir lesit bókina ‘has read the book’. This development is also assumed for Late Latin: habet litteras scriptas ‘has the letter in a written-down state’ habet scriptum litteras ‘has written the letter’. Constative readings are already found in the highest parts of the low-focal sector. Graves (this volume) reports that even the new Macedonian high-focal may, though not characteristically, be used with experiential meaning. Thus, if PAST (POST) items are observed to have “non-resultative” uses with [ t] actional phrases, we cannot exclude the possibility that they have retained a higher focality degree; cf. Tommola’s discussion (this volume b) of past tenses in Old Russian.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

132

132

Lars Johanson

8.11. From low-focal to nonfocal Focal postterminals are, as we have noted, found in large parts of Europe. They are by no means limited to some maritime areas in the western parts of the continent, but are also present in Finno-Ugrian, Turkic, Caucasian, and other languages. On the other hand, the drift to nonfocality is also a widespread phenomenon and not an areal feature limited to a coherent inner part of the western half of the continent. This ultimate defocalization coupled with the disappearance of a former PAST ( POST) item (“preterite loss”) has been described in various ways. It is traditionally often said that an original “perfect” comes to be used as a “perfectum historicum” and finally turns into a “historical tense” or a “past tense”. If focal postterminality is thought to be reference to two temporal strata, transition from low-focality to nonfocality is taken to be a loss of this double reference. The development is sometimes also conceived of as a retrospective expansion of the present space of time. The change from focal postterminals to pasts is a universal tendency also reflected in ontogenetic development, namely in child language acquisition (see Antinucci & Miller 1976, cf. Givón 1982: 151). The generalization of a low-focal and the disappearance of the corresponding POST item means the loss of the POST opposition. Postterminals involved in this defocalization process may be more or less progressive. An item X is more progressive than an item Y, if it takes on Os -independent functions in more contexts than Y does. There are generally transitional stages with promiscuous uses of the former POST vs. POST items and with residual higher functions observable in the POST items. The tendency towards total defocalization is observed in many older languages including Old Indic, Latin, etc. As for the Classical Greek Perfect, the tendency led, from the Alexandrian period on, to promiscuous use and transition to a nonfocal item. The Perfect took a step down the focality scale, assumed event-oriented functions, fused with the Aorist and was superseded by it (Browning 1983: 30). The result of similar developments is found in several modern European languages. As for Slavic, it is present in Russian, Belarusan, Ukrainian, Czech, Polish, Rusyn, and also largely characteristic of Slovene, Croatian as well as most Serbian varieties. The development has been delayed in Sorbian, possibly under German influence. However, in Lower Sorbian, the former PAST ( POST) item has practically disappeared. Upper Sorbian and literary Lower Sorbian exhibit promiscuous use, the socalled Perfect being a PAST (POSTNF) item that can always be substituted for the former PAST ( POST) item (Faßke 1981: 262–263). Among Germanic languages, German and Yiddish are most strongly affected. The Dutch Perfect is found in the lowest part of the low-focal sector, being more progressive than the English or Scandinavian counterparts but more conservative than the Southern German one, since it has not yet ousted the PAST ( POST) item. The Afrikaans counterpart is clearly

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

133

Viewpoint operators in European languages

133

more progressive. Romance postterminals exhibit various degrees of defocalization. Their progressivity seems to increase along a line stretching from Galician and Portuguese to Castilian Spanish, Catalan, and Occitan and to Italian, French, Romanian, and Romansh. Thus, the Spanish type ha hablado ‘has spoken’, in the middle of this continuum, is observed to occupy more and more of the functional territory of the type habló ‘spoke’. On Sardinian, see Bossong (1993). Nonfocals are also found in Hungarian, Maltese, Romany and varieties of Albanian. In several other languages, postterminals seem to be encroaching on the territory of nonpostterminals, e.g., the Estonian Perfect, possibly due to Russian influence. The functional distribution often exhibits a good deal of regional variation. In several languages, the loss of POST oppositions predominantly affects certain regional varieties, e.g., South German, North Italian, Daco- and Istro-Romanian, or Gheg dialects of Albanian. In dialects tending towards defocalization, the POST opposition is mostly rare in colloquial language (Italian, Serbo-Croatian, Romanian) but preserved in more formal registers and stylistically marked contexts. The old Catalan PAST ( POST) item (escriví ‘wrote’) is also mostly limited to written and literary registers. However, the latter reduction is not due to defocalization of the PAST (POST) item (ha escrit ‘has written’), but to the introduction of the PAST ( POST) periphrasis va  infinitive (va escriure ‘wrote’). Moreover, if a POST opposition is given up, it is not always the former PAST (POST) item that generalizes. It is also possible that the former PAST ( POST) item develops into a more general item referring to past events both historically and diagnostically (Section 11.6.1). With the change to nonfocality, the meaning of a postterminal generalizes to cover the widest range of diagnostic and historical interpretations. The loss of the specific meaning component of Os relevance makes the items compatible with more contexts. However, the generalization does not lead to the precise expression of more functions. If a relatively specialized item X extends its use to subsume both its old functions and the functions of an item Y, this generalization leaves both the old X and Y functions without adequate expressions. Unless focality is renewed by some other item, the nonfocal may also continue to cover cases of Os relevance. Thus, even if the Russian Past might be claimed to cover meanings typical of a “Perfect” (Maslov 1980: 51, 53), these meanings are not explicitly expressed. It can only be concluded that this generalized Past is also used in cases where some other language would use a focal postterminal. A nonfocal can never replace a PAST ( POST) item in the sense of taking over its value. While it may refer to the event historically, it is no genuine ‘temps historique’. Only if focal postterminality is renewed by other means may it evolve into a nonpostterminal past. The statement that a low-focal postterminal, e.g., in the sense of a PF item, develops into a nonfocal is problematic if the latter is taken to be a PFV item. Thus, in the Functional Grammar as initiated by Dik (e.g., 1989), this would mean that a higher

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

134

134

Lars Johanson

operator (“perfect”) develops into a lower one (“perfective”), which is at variance with other known diachronic tendencies. But this is only a pseudo-problem. First, nonfocal postterminals are, per se, not PFV items in an adterminal or a nonintraterminal sense. Secondly, even if they develop into PFV items in new oppositions, the problem of a higher operator developing into a lower one does not arise. In a Functional Grammar model of this kind, PFV should, as I have argued, be taken to be a higher operator on a par with PF (Johanson 1996). When low-focals lose their focality, the development seems to start with [t] actional phrases, whereas residual focal uses may be observed with [ t] actional phrases. For the development of the Old French Perfect, see Schwenter (1994). The uses with [ t] actional phrases have constative readings, which may mean that constativity is the last domain given up by focal postterminals. Note that the American English Perfect, while obviously losing some of its focal uses, is still stable in constative functions. The relatively weakly defocalized Portuguese Perfect, which was formerly also used with [t] actional phrases, is now essentially restricted to [ t]. The situations are partially similar in Galician and varieties of American Spanish. Constative readings may thus be transitory to the nonfocality stage. Constative interpretations are possible with all low-focal items and with both [t] and [ t] actional phrases. On the other hand, constative interpretations without “resultative” shades of meaning are typical of POSTLF  [ t]. It is obvious that they are only a result of interaction with actional values and do not represent any independent cardinal stage in the defocalization drift. Sensitivity to temporal distance may be decisive for the first step leading from POSTLF to POSTNF. This step is often taken in contexts referring to events temporally remote from Os . Comrie supposes a gradual “relaxation of the degree of recentness required for the use of the Perfect” to have been a key part of the Romance development (1976: 61). Postterminals of the lowest focality degree are often used in ways that seem to motivate definitions as “remote Past”, “Tempus der Ferndistanz”, or “mythic Past”, e.g., Turkic -miš and -gän items, constatives in -ipdir etc., Kalmyk constatives in -la. There have been similar losses of postterminal oppositions of the pre-past stratum. Varieties in which a former PAST (POST) is defocalized are also likely to exhibit corresponding general pluperfects. However, the loss of the opposition may be accompanied by compensatory developments. As we noted, the loss of the French Simple past led to the loss of the Past anterior. In certain spoken varieties, however, the latter has been replaced by the “passé surcomposé” a eu lu. Similar products are the South German Pluperfect hat geschrieben gehabt ‘had written’, the Serbian type bio je išao ‘had gone’, and the Albanian so-called Perfect II, typical of Gheg varieties whose former Perfect has become a nonfocal, e.g., ka pasë qenë ‘had been’ (Buchholz & Fiedler 1987: 133). In varieties of this kind, other types of supercompound items for the pre-past and even prepre-past stratum are also found (literally

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

135

Viewpoint operators in European languages

135

‘had had V-ed’), e.g., French avait eu vu ‘had seen’, German hatte gelesen gehabt ‘had read’, Albanian Gheg dialects kishte pasë ardhë ‘had come’.

9. Adterminality 9.1. Definition The last viewpoint oppositions to be dealt with in more detail are those of adterminality vs. nonadterminality, present in North, East and Western South Slavic languages. They are of an aspectual nature, though intricately interrelated with actionality. AD distinctions represent rather atypical varieties of the general PFV vs. IPFV distinction assumed by many linguists. Adterminality (AD) is signalled by perfectives. However, the presence of a viewpoint value AD in a given verb presupposes a partner verb with the same lexical meaning, e.g., Russian AD postroit’ vs. AD stroit’ ‘build’. Thus, not all morphological perfectives signal AD. Hermann even avoided the term “Aspekt” because of its use for Slavic perfectives and imperfectives, which are not always, as he noted, “subjektiv geschieden” (1927: 228). As already stated, many perfective verbs represent modes of action and are thus not AD partners. On the other hand, many verb pairs are genuine AD partners although they may be translated into other languages by different lexemes, for example, Russian dobivat’sja by English strive after and dobit’sja by attain. The opposition is absent in verbs that do not form aspectual pairs. Such verbs are referred to as imperfectiva tantum and perfectiva tantum (e.g., Belarusan zaminac’ ‘disturb’ and abnarodvac’ ‘publish’, respectively) or are even claimed to represent both aspects (e.g., arandavac’ ‘lease’). Adterminality, AD, envisages the event ad terminum, in the attainment of the relevant limit of its actional content. Since it operates on transformatives, this limit is a crucial one, generally the terminus finalis. By contrast, nonadterminality, AD, disregards the attainment of a relevant limit. AD denotes that the transformation is brought about, whereas AD does not deny or exclude it. Note that, as always with viewpoint notions, the phases that are not highlighted are only latent, not necessarily inexistent. Adterminality is not identical to transformativity, which only implies a limit to attain but not the very attainment of this limit. Whereas [t] implies the presence of the crucial limit in the actional content, AD signals its realization in an event. The difference is thus by no means a gradual one. Much confusion has been caused by the compatibility of [t] and AD as well as their special interconnections in Slavic systems. As noted above, it is necessary to distinguish the viewpoint operator (aspect) from the operandum (actional content) but possible to assume that [t] and AD may merge into portmanteau markers. Thus, Russian-type perfectives

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

136

136

Lars Johanson

are combined AD- and T-markers, expressing AD  [t]. The corresponding imperfectives signal neither AD nor [t], thus standing for a nonadterminal perspective on events. By definition, systems of this kind lack the combination *AD  [ t]. How does our definition relate to other assumptions regarding aspect oppositions of the Russian type, as discussed from Jakobson (1932) on? The [t] actional content of perfectives in the sense of a possible transformation rather seems to correspond to what is sometimes called the ‘event-unit’ (Barentsen 1985: 59–60: “handelingseenheid”). AD has a certain affinity with the feature ‘totality’, which presents the ‘event-unit’ as one total whole. But the AD opposition does not concern ‘completion’ to the effect that AD characterizes an event as unfinished. AD neither affirms nor negates completion. The aspectological literature offers numerous examples of AD verbs in sentences implying that the event has really been brought to an end, e.g., Polish czytał te¬ ksia¬ z˙ke¬ ‘has read this book’. Nor is a well-defined ‘result’ of the event a pertinent feature of AD. Both AD and AD may get a more or less resultative reading according to the context. The idea of asymmetry in the markedness structure – AD being the semantically marked and AD the semantically unmarked member of the opposition – is of basic importance and conforms to a long tradition going back to A. X. Vostokov (Jakobson 1932). For example, it excludes the possibility that morphologically unmarked imperfectives such as Russian pisat’ ‘write’ are taken to possess zero markers signalling a positive aspectual content. AD represents negation of the AD value and irrelevance towards it, often simply implying occupation with the event. The AD past is often said to have a ‘simple denotative’ or ‘general factual’ function, the event being referred to in a generalized manner. Comrie attests that Russian AD items can be used when “the speaker is simply interested in expressing the bare fact that such and such event did take place, without any further implications, and in particular without any implication of progressive or habitual meaning” (1976: 113). This statement does not mean that imperfectives signal ‘totality’ and is thus not incompatible with Comrie’s totality view of perfectivity. According to Dahl (1985: 76), Comrie’s claim that IPFV pays essential attention to the internal structure clashes with the idea that it sometimes expresses “the bare fact that such and such an event did take place”. AD is widely used for events dissociated from a sequential setting, e.g., Czech Tuhle knihu cˇ etla ‘She has read this book’. It is only natural that the direct adterminal view of the limit is more fertile in sequential settings, but there is no reason to claim that the opposition is neutralized outside them. Since the attained transformation signalled by AD suggests a transition to a new situation beyond the transformation, AD is typically used for temporally situated events conceived of as linked to a preceding and / or following event, as leading from one situation to another. On the other hand, AD is rather indifferent to situational change. The relevance of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

137

Viewpoint operators in European languages

137

sequential connection as a feature of discourse organization will be further discussed below. It is, however, important to emphasize that AD distinctions are relevant with non-set events as well, the decisive point being whether the event is envisaged in its very attainment of the crucial limit or not. It is often claimed that AD typically denotes single events and that negation is more characteristic of AD than of AD. Such features concern the actional content, the object of aspectual characterization. As we have seen, a serial reading as well as a negation may lead to recategorization [t] [ t]. Since [ tt] is not typical of actional phrases expressing single actions, AD does not typically refer to global events with subevents. However, perfectives of certain Slavic languages, e.g., Czech, behave differently with respect to the representation of repeated events and may also be used for [ser] readings.

9.2. Temporalization A PAST (AD) item normally situates the AD perspective at an L that is anterior to Os and coincides with the attainment of the crucial limit, e.g., Russian napisal ‘wrote, has written’, Belarusan proˇcyta˘u, Polish przeczytał ‘read, has read, had read’. PAST ( AD) items, which are not concerned with any crucial limit, only signal anteriority, e.g., Russian pisal ‘wrote, has written, was writing, has been writing, had written, had been writing’, Belarusan cˇ yta˘u, Polish czytał ‘read, has read, was reading, has been reading, had read, had been reading’. Most Slavic languages only possess one single past tense, which thus has several English translation equivalents. The non-past stratum is more complicated. One problem concerns presentness related to Os . PAST situates the aspectual perspective at a point that is not anterior to Os . With PAST ( AD) items, this point can coincide with Os , the natural vantage point from which an ongoing event is observed, e.g., Russian pišet pis’mo, Polish pisze list ‘is writing, writes a letter’. The final limit is not included in the view offered here, no matter how extended the event is. PAST ( AD) items such as SerboCroatian piše ‘writes, is writing’ are characterized by the natural intraterminality of Os -presentness and might thus even be represented as a PAST (INTRAo ( AD)) items. Like other nonfocal PAST (INTRAo) items, they display usages that burst the narrow “nunc” perspective, being used for uni-occasional events in progress at Os , pluri-occasional events, temporally unlimited events, events referred to as a type, universally valid facts, past events, ficticious, scheduled, intended, potential events, etc. PAST (AD) cannot be applied to events in progress at Os . The meaning of aspectotemporal items of this structure is incompatible with the description of something going on at encoding time. AD cannot be applied to events current at “nunc”, as it is incapable of envisaging a present cursus. It might be thought that the presen-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

138

138

Lars Johanson

tation of the event as a totality necessarily implies Os -posteriority of the crucial limit to attain. More adequately expressed, however, adterminality envisages the very attainment of a future crucial limit and thus projects that limit directly into the future. It is a well-known fact that perfective presents of Russian, Polish, Czech, etc. do not express current events any more, but are regularly used with future – not necessarily ‘immediate’ future – time reference, which has become their main function. The socalled incapability of presentness of PAST (AD) items (“Gegenwartsunfähigkeit des perfektiven Präsens”) has been taken to be a central criterion of aspect. AD is totally incapable of primary deictic presentness, cannot be anchored in the present time sphere. Some kinds of presentness which will not be discussed here, namely those represented by so-called historical and performative presents, are less dependent of the Os perspective and may allow a PAST (AD) presentation envisaging the crucial limit. Thus, PAST (AD) items actually serve as a special kind of prospectives. If they interact, as Russian napišet ‘will write’, with periphrastic imperfective futures such as budet pisat’ ‘will write, will be writing’, they may be said to function as PAST (PRO (AD)) items, opposed to PAST (PRO ( AD)) ones. Compare the Modern Greek Aoristic Future, e.g., tha ghrápsi ‘will write’, which is a PAST (PRO ( INTRA)) item opposed to a PAST (PRO (INTRA)) item, e.g., tha ghráfi ‘will write, will be writing’. It differs considerably from the Slavic PAST (AD) type of future time reference in signalling prospectivity by means of a special marker tha and not directly envisaging the future attainment of a crucial limit.

9.3. Combinability The application of AD oppositions is impossible in certain contexts. Thus, the use of AD is blocked with phasal verbs such as ‘begin’, e.g., Russian *naˇcinaet napisat’ pis’mo ‘starts writing a letter’. It is possible to begin an action that contains a crucial limit, but not to begin the attainment of this limit. This is clear evidence of the aspectual nature of AD. The combinability of AD with time expressions has already been briefly commented on in Section 7.2.4. AD items exhibit certain restrictions with punctuality but may combine with punctual expressions that do not refer to the crucial limit, e.g., Russian v dva cˇ asa rabotal ‘was working at two o’clock’. However, they do not combine with expressions meaning ‘suddenly’, e.g., *vnezapno stojal tam ‘stood there suddenly’. This is one of several differences between IPFV items of the INTRA and AD types. AD items may, like intraterminals, cooccur with adverbials expressing that the initium is not later than O (‘already’) or that the finis is not earlier than O (‘still’), Russian uže pisal ‘was already writing’, vse ešˇce pisal ‘was still writing’. However,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

139

Viewpoint operators in European languages

139

AD in uže napisal would suggest pre-past: ‘had already written’. ‘In X time’ expressions (Russian za X vremja), which measure accomplishment time and identify the actional phrase as [t], require PAST (AD) items, which subsume the value [t], e.g., Russian poˇcinil mašinu za tri cˇ asa ‘repaired the car in three hours’. Compare the similar use of PAST ( INTRA). Abtemporal ‘since X time’ expressions (Russian uže X vremja), which measure the time between the initium and O, require AD, e.g., Russian uže dva goda tam žil ‘had been living there for two years’. Compare the corresponding uses of INTRA items. ‘X times’ expressions may be limiting, and PAST (AD) can thus be used to represent the whole global event with its final limit, e.g., Russian napisal mne tri raza ‘wrote to me three times’. Compare the similar use of PAST ( INTRA). However, if the final limit is unimportant to highlight, PAST ( AD) may also be used, e.g., pisal mne tri raza, which would of course be impossible with a PAST ( INTRA) item. ‘For X time’ expressions, which measure the temporal extension and identify the actional phrase as [ t], require, in languages such as Russian and Polish, AD items, which subsume the value [ t]. To present a past event as extending over a period of time, Russian uses PAST ( AD), e.g., dolgo pisal ‘wrote for a long time’. As we have seen, however, this is a case where PAST ( INTRA) items are used, since a INTRA view is infertile with exact indications of the outer measures of the event. Bulgarian, as expected, uses the imperfective T-unmarked Aorist, e.g., stoja d˘algo na prozoreca ‘stood for a long time at the window’. It was noted above that these different choices show an essential incongruity between IPFV items of the INTRA and AD types and that ‘for X time’ expressions by no means necessitate a “perfective” view. On the other hand, a durative event may be presented as a totality by means of delimitative or perdurative modes of action: ‘spent [a certain time period] V-ing’. 9.4. The way to AD It has already been noted that the sources of AD systems are transformatives and nontransformatives, represented morphologically by T-marked perfective and Tunmarked imperfective verb stems respectively, and grouped together into pairs. AD items are thus formed by means of prefixes or suffixes from stems that eo ipso become AD items. There are also cases of suppletivism, e.g., Polish AD wzia¬ c´ vs. AD bra´c ‘take’. Secondary imperfectives (see below) are formed with suffixes from perfective stems or with vowel changes in the primary stem, e.g., Polish zarobi´c  zarabia´c ‘gain’. Many aspectologists assume a stable basic distinction of perfectivity vs. imperfectivity, present in all older and modern Slavic languages. With respect to grammatical values, this alleged stability or unity is highly questionable. There is no

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

140

140

Lars Johanson

unitary “Slavic style aspect”, and general definitions of a Slavic perfectivity such as “l’action dans sa totalité, comme un point, en dehors de tout devenir” (Saussure 1916: 161–162) conceal substantial differences. Modern Slavic languages use formal perfectives for both T-marking and adterminality. While the dichotomy serves AD distinctions in languages such as Russian, it is at least known to be “less aspectual” in certain West and South Slavic languages. It will be argued in the present contribution that the type found in Bulgarian belongs to the domain of actionality. The two types represent different semantic values and cannot be said to be essentially equivalent. There is thus also little justification for assuming a “Sprachbund”, an areal group of Slavic, Baltic, German, and Hungarian, which would have grammaticalized “perfectivity” in a uniform way. The Slavic languages do not represent a unified type, and German does not even employ T-marking in the systematic way as Hungarian, Bulgarian, Lithuanian, Georgian or Ossetic does. The systems mentioned certainly display basic morphological analogies. For example, the Slavic formation types are partly paralleled in Baltic languages. Of course, a “Sprachbund” does not need to be homogeneous, and may display different degrees of development in its parts. However, the different interpretations of the morphological data at the semantic level leave us with a poor basis for establishing linguistically interesting common features. T-marking, even if it is systematically applied, remains within the domain of actional content. On the other hand, T-markers may develop diachronically into viewpoint operators. The exclusively T-marking perfectives represent an earlier stage of semantic development than the AD-signalling ones found in certain Slavic languages. The Russian-type development led from (i) lexeme derivation to (ii) Tmarking to (iii) aspect formation. A mode of action without aspectotemporally determining force developed into a viewpoint operator; an actional distinction turned into a AD opposition. The perfectives extended their function from marking actional phrases for [t] to determining them aspectually as AD. The T-markers came to mark AD  [t], a development into what Mourek (1895) called “true perfectivity”. The marked category not only implied a crucial limit to attain but also envisaged the attainment of this limit. The viewpoint operator imposed an additional semantic constraint on the meaning it operated on. A portmanteau marker combined transformativity, which just aims for totality, with adterminality, which actually involves it. The unmarked items became nonadterminals, implying no crucial limit to attain and consequently no attainment. This decisive step is often ignored in the discussion. It has not been taken in Hungarian, Latvian, Lithuanian, Kartvelian, Ossetic, etc. Though the T-marking systems of these languages are often referred to as perfective vs. imperfective distinctions, they have not developed into AD oppositions. The Bulgarian distinction is still an actional one, the perfective functioning as a T-marker (marker of “pre-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

141

Viewpoint operators in European languages

141

del’nost’ ”; Tommola 1984). The presence of an imperfective Aorist and a perfective Imperfect in Bulgarian has sometimes been used as an argument against the aspectual nature of the Imperfect vs. Aorist opposition. On the contrary, it should rightly be considered evidence against treating the perfective vs. imperfective dichotomy as a viewpoint distinction. Similarly, the Hungarian T-marking preverbs are not aspectual markers or “perfectivizers” in an adterminal sense. The distinction between megír ‘writes (to finish writing)’ and ír ‘writes’ is one of actional content. It has been suggested that a Russian-type aspect system is in the process of development in Hungarian. Comrie, who rightly considers Hungarian less developed than Russian with respect to a perfective vs. imperfective opposition system, supposes that the Hungarian preverbs are developing into markers of perfectivity (1976: 93– 94). However, Csató (1994: 232–237) shows that T-marking is decisive in assigning correct interpretations to the Hungarian aspectotemporal forms and “provides a better characterization of the function of Hungarian verbal prefixes (which derive transformatives) than does their characterization as perfectivizing” (Comrie 1994: 299). Though the difference between actional and aspectual categories is by no means gradual, there may of course be gradual diachronic developments from pure Tmarking into AD. The starting point of the Slavic development is a Proto-Slavic system comprising the aspectotemporal categories Present, Perfect, Imperfect, Aorist, and, in addition, a rather unrestricted [t] duality (perfective vs. imperfective). This system was restructured in most Slavic languages with the loss of the Imperfect, the Aorist, and the PAST (INTRA) opposition holding between these two items. Moreover, the Perfect mostly lost its PAST (POST) function and developed into a PAST item. The [t] distinction, semantically very different from the INTRA and POST distinctions, could not compensate for their loss or, as is sometimes claimed, take over their roles. In languages such as Russian, the [t] distinction developed into a AD distinction, in the past strata in the following way: (i) pisal PAST (POST)  [ t] PAST ( AD); (ii) napisal PAST (POST)  [t] PAST (AD). These changes seem rather natural in the light of the affinities between certain actional and aspectotemporal categories. Thus, INTRA items such as Imperfects have affinities with [ t], and INTRA items such as Aorists with [t]. As already stated, [t] was constitutive for the Classical Greek Aorist. T-marking may be more important with some aspectotemporal categories than with others. It is hardly astonishing if the affinities also lead to mergers. A system of free combinability of viewpoint operators and actional contents may change to the effect that [t] is largely restricted to INTRA, and [ t] to INTRA. Thus, the Slavic Aorist is, in languages that have preserved it, predominantly perfective. In Sorbian, the Aorist is used with perfectives and the Imperfect with imperfectives (Šewc 1968: 171–172). These past tenses hold an intermediary position between an older Slavic type, represented by

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

142

142

Lars Johanson

Bulgarian, and a new type, represented by Russian. Their development is a parallel to the merger of transformativity with adterminality in languages of the Russian type.

9.5. Differences between AD- and T-markers The two types of perfective vs. imperfective distinctions, as represented, for example, by Russian and Bulgarian, differ clearly in their functions. AD signals aspect based on actionality, since AD presupposes and subsumes the actional value [t]. This does not imply any dominance of actional values; the aspectual values always have scope over the actional ones. The Bulgarian-type use of perfectives and imperfectives is purely actional and concerns the [t] parameter. Here, too, as we shall see, aspectoactional parameters interact to the effect that the viewpoint values (INTRA, POST) have scope over the actional ones. A few comments should be made here on the use of morphological perfectives and imperfectives as AD and T and the differences between them. First of all, adterminal items display a much more general and systematic use than T-markers do. It has already been noted that one indication of a completed grammaticalization process is a higher degree of generality of use. There are thus never one-to-one correspondences between adterminals and Tmarkers. For example, though Ossetic T-marked items are often claimed to be equivalent to Russian perfectives, fæk-kodta ‘did’ corresponds to sdelal, whereas the simplex kodta corresponds to both delal and sdelal (Miller 1962: 139). Another difference mentioned above is the blocking of AD with phasal verbs such as ‘begin’, as it is not possible to begin the very attainment of a crucial limit. A T-marker easily combines with such verbs, since it is quite natural to start an action directed towards a crucial limit. Thus, Polish verbs expressing the beginning or end of an activity or process combine with imperfectives. The Lithuanian perfective exhibits far fewer such constraints, and Hungarian T-marked actional phrases also rather freely occur with phasal verbs, e.g., Lithuanian e˙ m˙e atprasti ‘began to give up (a habit)’, Hungarian kezdett megbolondulni ‘began to go crazy’. Many speakers also accept Hungarian kezdett megírni ‘started to write (to completion)’. All this shows us, again, that the Lithuanian and Hungarian items in question are not AD items. 9.5.1. Capability of future time reference and incapability of presentness It was also stated above that PAST (AD) cannot apply to an event in progress and gets future time reference if related to Os, e.g. Russian napišet pis’mo ‘will write a / the letter’, perejdet most ‘will cross the bridge’, Czech napíše ‘will write’. It should, however, be noted that all finitransformatives easily suggest future time reference with PAST items. At Os , a telic event can only be viewed in its course or during

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

143

Viewpoint operators in European languages

143

its ongoing preliminaries, and this view implies that a possible accomplishment can only occur in the future. In other words, a goal-oriented event is bound to the Os perspective with the natural implication of posteriority. This kind of future reference is, for example, a natural reading of Classical Greek Present forms of [t] verbs, e.g., néomai ‘go, come’. This is universally valid for the non-past of finitransformatives, in particular for [tf, mom] actional phrases. The possible preliminaries are observed in the deictic present, but the crucial limit must lie, if anywhere, in the future. The question “What is going on right now?” is answered more satisfactorily by nontransformative items. Thus, to express ‘is writing a letter’, Lithuanian uses raso laiska, i.e., with the T-unmarked (“imperfective”) item rašo rather than the T-marked (“perfective”) parašo. Finnish uses the T-marked lukee kirjaa ‘is reading / reads the book’ (with partitive object), whereas the T-unmarked lukee kirjan ‘reads and finishes reading the book’ (with the object in the total case) has future time reference. However, the decisive criterion of AD is not the capability of future time reference but the total incapability of presentness. The notion of adterminality in the primary deictic “nunc” perspective means a more direct future projection. The object of observation is, again, not an event moving towards a crucial limit, but it is the very attainment of that limit. Note that Lithuanian perfectives such as parašyti ‘write (to completion)’ do not share the total “incapability of presentness” typical of Russian or Polish perfectives such as napisat’, napisa´c. Their main non-past function is not to express prospectivity. The Latvian so-called perfective Present behaves similarly. In the same way, the Present of a T-marked Hungarian actional phrase just suggests that the possible completion can only take place in the future, e.g., megírja a levelet ‘writes / is writing (to complete) the letter’. Similarly, the Bulgarian perfective Present does not display the future time reference typical of PAST (AD) items. To express prospectivity, languages such as Lithuanian, Hungarian and Bulgarian use special items. Hungarian uses the fog Future with both transformatives and nontransformatives, e.g., PAST (PRO)  [t] meg fogja írni a levelet ‘will write (to complete) the letter’. Bulgarian possesses a special Future, consisting of the particle šte  Present and predicting events of both transformative and nontransformative actional contents, e.g., PAST (PRO)  [t] šte doide ‘will come’. Georgian PAST items determined by T-markers get readings in the sense of ‘does (to completion)’ and tend, as expected, towards future time reference, e.g., dacer ‘tu l’écriras’, in which case they may be translated by Russian perfectives, e.g., ª napišeš’ ‘you will write’ (Vogt 1971: 183–184). Hungarian non-past items such as átmegy a hídon ‘crosses the bridge’ can, like German überschreitet die Brücke, express an ongoing telic event, projecting its conclusion into the future. If the Russian perfectives were T-markers, we would expect them to behave similarly. However, perejdet most ‘will cross the bridge’ adds the AD meaning, directing the attention directly to the event in its actual conclusion. Note also that the Hungarian and Ger-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

144

144

Lars Johanson

man items, unlike the Russian ones, combine with explicit PRO markers, e.g., át fog menni a hídon and wird die Brücke überschreiten ‘will cross the bridge’. 9.5.2.

AD-marking of global events and T-marking of subevents

A further important criterion is how perfectives relate to global events and possible subevents. Adterminality envisages the whole event as such, whether it consists of one single event or subevents. T-marking, however, may also be used for the actional characterization of the subevents. As we have seen, a transformative actional phrase may be quantitatively reinterpreted as [ser], which removes the crucial limit and makes it more capable of presentness. If it still contains a T-marker, it is in order to assign transformativity to the subevents. Thus, in their primary reading as denoting Os -oriented single events, Finnish and Estonian transformative actional phrases with total objects are interpreted as having future time reference, e.g., Estonian küpsetab koogi ‘will bake a cake’. When a [ser] reading removes the limitation and makes the whole actional phrase more capable of presentness, the T-marking total object specifies the actional properties of the subevents. If Russian perfectives were T-markers, they would be expected to function similarly. However, they do not signal [t] of subevents, but operate, as AD markers, on the actionality of the global event. It is the attainment of the crucial limit of the global event – not of the subevents – that is envisaged by a AD operator. Os -oriented T-marked non-past items typically refer to an ongoing global event that is not completed itself but consists of telic subevents. Thus, the Bulgarian combination PAST (INTRAo )  [t] mainly expresses repeated events. Similarly, Lithuanian, Latvian, and Hungarian perfectives can refer to ongoing global events, the T-marking only signalling the transformativity of the subevents, e.g., Latvian katru dienu uzraksta pa v¯estulei, Hungarian minden nap megír egy levelet ‘writes (and finishes writing) a letter every day’. Here, the function of the “perfectives” is not to denote habituality, but to characterize the subevents actionally. The pluri-occasional reading arises from the [ser] reinterpretation. The pluri-occasional global event, seen in progress at Os , is a set of events, each of which is actionally characterized as [t]. The Bulgarian, Lithuanian, Latvian, and Hungarian perfectives are thus not AD categories. Only T-marking perfectives may express the telic character of subevents. For the sentence just cited, the imperfective Present would be the only choice in Russian: každyj den’ pišet odno pis’mo. Here, the AD aspect operates on a [ser] actional content used to denote a pluri-occasional global event. There is no device to mark the subevents for [t] actionality. Adverbials expressing unlimited repetition such as ‘every day’ (Lithuanian kasdien, Russian každyj den’, etc.) combine with T-marked actional phrases but not with PAST (AD) items. If Russian perfectives were strictly T-marking, they would also be used to assign transformativity to subevents. If the Bulgarian, Lithuanian, Latvian,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

145

Viewpoint operators in European languages

145

and Hungarian “perfectives” were adterminals, they would only be used to envisage the attainment of the endpoint of the global event, which is incompatible with the present perspective.

9.5.3. Imperfectivization Slavic secondary imperfectives derived from prefixed stems may, as we have seen, function as T-markers. Obligatory secondary imperfectivization is sometimes considered the major criterion for Slavic aspect grammaticalization, and it indeed seems to provide arguments for the presence of AD oppositions. In languages such as Russian, T-marked items of this kind have developed into aspectual imperfectives. From a simple imperfective pisat’ ‘write’, a perfective perepisat’ ‘rewrite, copy’ is derived by means of a transformativizing preverb that also modifies the lexical meaning. The non-past item perepisˇet ‘will rewrite’ exhibits the incapability of presentness to be expected from a PAST (AD) item. As should also be expected, a secondary imperfective item pere-pis-yv-at’, derived from pere-pisat’, does not share this incapability, since it is a AD item: perepisyvaet ‘rewrites’. The Lithuanian T-marked per-rašyti ‘rewrite’ corresponds to both AD perepisat’ and AD pere-pisyvat’ and is thus obviously not a AD item. The item per-raš-in˙eti ‘rewrite, copy (repeatedly)’ is still an iterative here. Georgian uses for ‘rewrite, copy’ a T-marked verb, which, like all finitransformatives, naturally suggests future reference in non-past items, e.g., gadacers ‘will rewrite’. The Rusª item with full capability of sian secondary imperfective Present, however, is a AD presentness, e.g., pere-pisyvaet ‘rewrites, is rewriting’. The Georgian lexeme is just a transformative, and it even lacks a lexically adequate nontransformative equivalent, the only [ t] counterpart available being the simple cers ‘writes, is writing’. ª

10. Interaction of viewpoint operators with phase structure types 10.1. Typical realizations The interdependence of viewpoint operators and the actional contents they apply to has been emphasized above. The choice of aspectoactional combinations produces different types of interactional readings. Inherent phase structure properties in terms of [t], [dyn], [mom] and other distinctions determine the realizations. The typical realizations of post-, intra- and adterminality with finitransformative, initiotransformative and nontransformative actional phrases can be roughly illustrated as follows:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

146

146 [tf] [ti]

Lars Johanson

Postterminality

Intraterminality



ˆ ...... 



......ˆ ...... 

ˆ ............



......ˆ ...... 

ˆ ............ [t]

Adterminality



......ˆ ......

ˆ 

ˆ 

ˆ

The relevant limit of nontransformatives is marked with , and the crucial transformational limit of transformatives with . The symbol ˆ stands for the viewpoint, and dots ( ) mark the variable space within which it may be situated. Grammatical viewpoint categories always interfere with actional semantics, and it is important to describe the specific connections. Different aspects operating on one and the same actional content produce different actional readings. As we have noted, some of the interactions in narrative discourse allow textual interpretations of ‘entry’, ‘dwelling’ and ‘exit’. Such readings are, as has been stressed, not identical to the functions of phasal modes of action, e.g., ingressivity. The latter signal properties of a given action as such, whereas aspects signal the conceptualization of an event as an occurrence of the same action, as action taking place. Viewpoint operators do not select certain phases, as phasal devices do. They do not pick out certain portions of the actional content, but may just give prominence to phases typical of the respective actional classes. Thus, actional distinctions cannot compensate for viewpoint distinctions, as their functions are different. For example, a system without a INTRA or AD opposition does not express any PFV vs. IPFV duality in a systematic way. Certain complementary distributions of viewpoint operators over actional classes do not prove any functional identity. Even elaborate systems of actional devices cannot make up for the lack of viewpoint operators. They may have roughly similar effects with respect to the representation of non-transitional and transitional events in discourse, but cannot express viewpoint meanings. Viewpoint semantics does not only serve to describe events in terms of ‘situation’ vs. ‘situation change’.

10.2. Realizations of INTRA The INTRA opposition imposes viewpoint values on actional phrases of different phase structures. It has been stressed above that [t] is not equal to INTRA and that, for example, the Bulgarian imperfective vs. perfective and Imperfect vs. Aorist distinctions are of different kinds. If perfective and Aorist are both defined as expressing “boundedness”, whereas imperfective and Imperfect are both taken to express “non-boundedness”, the combinations "imperfective Aorist” and “perfec-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

147

Viewpoint operators in European languages

147

tive Imperfect” get contradictory meanings that are impossible to account for. The combinations of INTRA and [t], the latter sometimes signalled by presence vs. absence of T-marking, do not represent complex aspectual meanings, but different predictable aspectoactional realizations of viewpoint values operating on different actional contents. The interactional meanings become unintelligible unless the basic categories are defined independent of each other. 10.2.1.

INTRA

INTRA highlights the cursus of the actional phrase meaning in a much more explicit way than the AD kind of IPFV. The view ‘from within’ yields textual readings of ‘dwelling in the event’. PAST (INTRA) items are thus not propulsive, but have a ruptive effect in narratives, denoting events that form a temporal or explicatory background for the main events. Their textual behaviour is often similar to that of PAST ( AD) items. Interaction of this cursus- oriented aspect with inherent phase structures may yield different processual, progressive, stative, preliminary and other readings. 10.2.1.1.

INTRA  [

t]

Intraterminality is most naturally applied to nontransformatives. The values INTRA and [ t] are highly compatible, though by no means identical. INTRA needs a cursus of some saliency to operate on, which is the case with [ t] items. Events expressed with [ t] items may of course be conceived of as having limits, though none is crucial. INTRA  [ t] views the event within these limits, without any emphasized beginning or end. It is not enough to define the result of this interaction as a mere ‘state’, since introspection is an essential element. Still, PAST (INTRA)  [ t] may exhibit a textual behaviour roughly similar to that of PAST ( AD)  [ t]: cf. Modern Greek éghrafe ghrámata ‘wrote / was writing letters’ and Russian pisal pis’ma. INTRA signals that something is going on at an O. The combination PAST (INTRA)  [ t] tells us that a portion of the action is already achieved at Os. Dik’s observation on English progressives holds for INTRA  [ t] in general: if it is truly said that, at some point of time, an atelic state of affairs obtains, for example, he is painting, it may be concluded at some later point that this state of affairs “has obtained”: he has painted (1989: 94). Whether a past event expressed by PAST (INTRA) can afterwards be interpreted as having occurred or not is a matter of actional content, the decisive factor being compatibility with the notion of partial realization. INTRA combines with both [ t, dyn] and [ t, dyn] actional phrases. With the latter, it denotes a static dwelling in the event, without internal progression, e.g., French savait, Turkish biliyordu ‘knew’. With [ t, dyn] actional phrases, it may

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

148

148

Lars Johanson

yield readings of a dynamic dwelling in the event, with possible internal progression, e.g., French travaillait, Modern Greek dhúleve, Turkish çalı¸sıyordu ‘was working’. Note that these readings are not necessarily progressive in the sense of a high-focal intraterminality. INTRA can operate on an actional content that is homogenized in the sense of [tf]  [hom] [ t], e.g., Finnish on korjaamassa kelloa ‘is repairing the clock’, where the actional phrase korjata kelloa is homogenized by the partitive, and the INTRAHF operator marked by -massa. It can also operate on an actional content that is homogenized in the sense of [tf]  [ser] [ t], in which case the intraterminally viewed global event consists of repeated and possibly pluri-occasional subevents. Pluri-occasionality yields processual readings of so-called habitual iteration, to which we shall return. As we have noted, INTRA is not meaningful with specifications of the entire temporal extension, e.g., French *écrivait deux heures. (An adverbial such as pendant deux heures would rather measure the period of intraterminal observation.) The Bulgarian imperfective Imperfect, which signals PAST (INTRA)  [ t], is not used in sentences denoting uni-occasional events with delimitative ‘for X time’ adverbials. This restriction is only intelligible if the Imperfect is defined as INTRA and not simply as IPFV, since AD items freely combine with ‘for X time’ adverbials. With pluri-occasional [ser] readings, imperfective Imperfects are also compatible with ‘for X time’ expressions, if the specification of the temporal extension refers to each subevent, e.g., peeše pesenta tri minuti ‘used to sing the song for three minutes [on each occasion]’ (Lindstedt 1985: 205). INTRA  [ t] does not combine with ‘in X time’ adverbials when the reference is uni-occasional. Under [ser] readings, however, the combination is possible: in this case, the accomplishment time of each telic subevent is measured, e.g., Bulgarian PAST (INTRA)  [t]  [ser] [ t] peeše pesenta za tri minuti ‘used to sing the song in three minutes [on each occasion]’. 10.2.1.2.

INTRA  [ti]

Since [ti] actional phrases denote both a transformation and a following posttransformational phase, the combination INTRA  [ti] signals an introspective view of an event excluding its crucial starting point, the very transformation. PAST (INTRA)  [ti] may thus highlight the cursus phase of the lexical meaning, the posttransformational statal phase of the actional content, e.g., Modern Greek katalávene ‘understood (= was aware of)’, krivótan ‘was hiding, hid (= kept himself / herself out of sight)’, Turkish duruyordu ‘was standing, stood’. These readings contrast with INTRA readings suggesting initial attraction (10.2.2.1.1). Note that the non-dynamic posttransformational phase highlighted by INTRA  [ti] combinations may be objectively identical to the one highlighted by corre-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

149

Viewpoint operators in European languages

149

sponding POST  [ti] combinations, e.g., is hiding = has hidden, understands = has understood, Classical Greek rhigé¯o ‘I shudder [with fear] = érriga ‘I am terrified’, Turkish oturuyor ‘is sitting’ = oturmu¸stur ‘has sat down’ (cf. 10.3.1.2). The two options may differ with respect to focality. As was noted bove, Maltese qedmarked intraterminals are less focal than high-focal POST items of the same initiotransformative verb, e.g., qed jorqod ‘is (usually) sleeping’ vs. rieqed ‘has fallen asleep, is asleep’. Note also that the distinction possible between (i) hid and (ii) was hiding becomes impossible if the INTRA item is high-focal and thus does not apply to all actional phrases. The English past item understood is ambiguous between (i) and (ii), since *was understanding is blocked for (ii). In languages lacking INTRA distinctions, [ti] actional phrases are systematically ambiguous, e.g., German erkannte (i) ‘became aware of’, (ii) ‘was aware of’.

INTRA  [tf] The combination INTRA  [tf] signals an introspective view of an event excluding its crucial endpoint, the very transformation. It is found in all INTRA languages. 10.2.1.3.

Thus, the Georgian “determined Imperfect” denotes an event “qui se déroulait dans le passé vers un but, un terme” (Vogt 1971: 189). Among the Kartvelian languages, Svan readily applies T-marking to intraterminal transformatives. INTRA  [tf] is also expressed by the T-marked Bulgarian perfective Imperfect. However, Tmarking is relatively dispensable in Present tenses. Thus, Lithuanian may use the simple Present, e.g., miršta ‘dies’, ¬ieina ‘enters’. The Kartvelian use of T-marking preverbs in the Present emerged rather late. The combination does not tell us whether the crucial limit is actually reached after the viewpoint or not. This fact has often been misunderstood. It has been called an “imperfective paradox” that an utterance containing an “imperfective telic” verb can be true even if it is known that the endpoint was never reached (Dowty 1977). It would be more correct to state that if a telic event expressed by a [tf] item is said to be in progress at some point, it cannot later be concluded with certainty that it has been achieved (cf. Dik 1989: 94). The combination INTRA  [tf] may thus produce preliminary readings of partial realization, preparation for completion, imminence, inclination, propinquitivity, conativity, etc. (Johanson 1971: 202–206). The readings vary according to the semantics of the actional phrase and the context. It may seem that intraterminality is incompatible with momentaneous finitransformatives and that the latter are thus monoaspectual, combinable with INTRA only. There appears to be little need for INTRA  [tf, mom], an inside view of an actional content consisting of a limit alone, without a salient cursus to look into. The combination INTRA  [tf, mom] is nevertheless used, referring to the time immediately before the crucial limit of which the action consists. It may thus have

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

150

150

Lars Johanson

imminential meaning, implying that the event is just about to occur, or propinquitive meaning, implying non-achievement. Such meanings are often produced by highfocal intraterminals (Section 10.2.1.5). Intraterminality is far more frequently applied to non-momentaneous finitransformatives, e.g. Modern Greek Imperfect PAST (INTRA)  [tf, mom] péthene ‘was dying, was close to death’, Turkish bo˘guluyordu ‘was drowning’, Georgian kvdeboda ‘il mourait’ (as opposed to PAST ( INTRA) mokvda ‘il mourut’; Vogt ª ª the event in a 1971: 182). The combination INTRA  [tf, mom] envisages phase preceding the crucial limit and only shows preparatory preliminaries. Questions concerning completion are aspectually irrelevant. Turkish Ali bo˘guluyordu ‘Ali was drowning’ leaves it open what happened after the introspection, whether the event was thwarted or not. Knowledge to the effect that the participant did not drown suggests a propinquitive reading of ‘was on the verge of’. The combination is often used conatively, denoting the mere attempt to reach the crucial limit. Verbs such as ‘try’ may be used to translate it. Conative readings are favoured by actional phrases with the feature [controlled], implying that the first actant referent has the power to influence the realization e.g., the Georgian imperfectum de conatu in vaˇ erebdi mas ‘j’essayais de le convaincre’ (Vogt 1971: 182). Italo-Croatian uses a construction copied from the Italian stava per periphrasis, e.g., stojaša za umbri, ma nija umbra ‘was dying, but did not die’ (Breu 1992: 117). Compare the above-mentioned Russian imperfectives with conative readings, e.g., ubeždal ‘tried to convince’. Neither the propinquitive-conative readings nor the interpretation of successful achievement is conveyed by INTRA  [tf, mom] as such. If the actional phrase may be interpreted as [tf, mom] or as [tf, mom], it may remain unclear whether the event has already begun or is only intended, e.g., Lithuanian jau buvo beišeina s ‘was about to go out’, Italo-Croatian stojaša za partit ‘was on the point of leaving’ (Breu 1992: 117). Note that the Hungarian T-unmarked Past mostly does not get preliminary readings with [tf] actional phrases, e.g., *a beteg halt ‘the patient was dying’. This would be natural if it were an IPFV item opposed to the T-marked Past, e.g., a beteg meghalt ‘the patient died’. A similar effect is, however, possible with spatially limiting adverbials and intraterminality signalled by a T-marker placed after the verb, e.g., ment ki az udvarra ‘was on his way out to the courtyard’. Compare similar cases of INTRA operating on spatial limitation such as Modern Greek píjene éna xiljómetro ‘was walking a kilometre’ (vs. píje éna xiljómetro ‘walked a kilometre’; Hedin, this volume). 10.2.1.4.

INTRA  [t]  [ser]

[ t]

Intraterminality may, as we have noted, operate on [ser] actional phrases denoting pluri-occasionality. The reading [ser] may be due to a quantitative reinterpretation without any overt signal. The objective property of repetition is not a sufficient

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

151

Viewpoint operators in European languages

151

reason to use a INTRA operator, but the decisive criterion is the inside view, which can be applied to both uni- and pluri-occasional global events. The global event is presented by INTRA regardless of the actional properties of the subevents. If the latter are telic and basically represented by [t] actional phrases, the combination INTRA ? [t]  [ser] [ t] is used. Examples of PAST (INTRA) ? [tf]  [ser]  [ t] are Turkish her gün saat altıda kalkıyordu, Bulgarian stavaše vseki den v šest caša ‘got up every day at six o’clock’. The basic phase structure may be [tf, mom], e.g., Romanian exploda o bomba ‘a bomb used to explode’, Modern Greek to évriske ‘used to find it’. The basic structure may also be [ti], e.g., Turkish her gün orada oturuyordu ‘sat / was sitting there every day’. T-marking languages may use the presence and absence of marking to characterize the subevents. Absence basically stands for global events whose subevents are atelic. Thus, Bulgarian T-unmarked PAST (INTRA) items, imperfective Imperfects, are regularly used for intraterminally viewed global events consisting of atelic subevents. An intraterminal perspective on a global event consisting of telic subevents is usually expressed by T-marked INTRA items. The distribution of functions in this aspectoactional interaction is clear-cut: INTRA marks the global event aspectually, and T marks the actionality of each subevent for [t]. Thus, a Bulgarian or Macedonian T-marked PAST (INTRA) item, a perfective Imperfect, may apply an intraterminal perspective to a whole set of unrestrictedly repeated telic subevents, e.g., in Bulgarian subordinate clauses such as veski p˘at štom ja pogledneše ‘each time he glanced at her’. Bulgarian PAST (INTRAo )  [t], the perfective Present, typically expresses pluri-occasional telic events. T-marked Imperfects are typically used in the same way in Kartvelian, e.g., Georgian and Svan. Italo-Croatian uses T-marked PAST (INTRA) items to express an intraterminal view on a sequence of telic events, e.g., Sa vržaša jašuˇc, rivaša nonda, sa skinaša a njimi ponesaša mblika gor ‘He used to ride away, arrive, get off and bring milk up to them’ (Breu 1992: 110). A similar view on global events with repeated telic subevents is achieved with T-marked Present items, e.g., Bulgarian sutrin izleze, veˇcer se v˘arne ‘leaves in the morning and returns in the evening’ (Stankov 1976: 31); compare the Lithuanian example already quoted: kasdien paraš˙e po viena¬ laiška¬ ‘writes (and finishes writing) a letter every day’. The Lithuanian or Bulgarian kind of “perfectivity” is thus by no means incompatible with present time reference. There is nothing contradictory about a sequence of telic events viewed in the natural intraterminality of a Present tense. On the other hand, it is clear that AD languages cannot use their quite different kind of “perfectivity” in such cases, but only PAST ( AD) items. However, the type of aspectoactional interaction discussed here is relatively limited. Bulgarian T-marked habitual Imperfects are rather rare and can be replaced by unmarked ones (Maslov 1959: 272). Thus, the latter may also be used for intratermi-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

152

152

Lars Johanson

nally viewed global events consisting of telic subevents. Italo-Croatian T-marked Imperfects are replaced by unmarked ones in the speech of the younger generation (Breu 1992: 115–116). Such simplifying steps towards the expression of pure aspect at the expense of actional characterization of subevents bring the systems closer to the Romance and Turkic ones. This simplification is formally opposite to the one that has taken place in AD languages, where the T-marking perfectives are formally maintained but have come to function as AD operators. On the other hand, the developments are similar inasmuch as AD also expresses aspect without regard to actional properties of possible subevents. 10.2.1.5. Combinations with INTRAHF Intraterminals of different focality degrees interact differently with certain actional contents. High-focals are especially sensitive to the phase structure properties [ser], [dyn] and [mom]. INTRA items naturally start as high-focals with [ t, dyn] and [ti] actional phrases. If they spread to combine with the remaining phase structure classes, [tf, mom], [ t, dyn] and [tf, mom], they are eo ipso lowfocals. INTRAHF items are usually more restricted with respect to [ser] than low- and nonfocals. Preaspectuals often reject pluri-occasional readings, e.g., German war am Trinken ‘was drinking (on one occasion)’. The English progressive was drinking tends more to uni-occasionality than Spanish bebía or Turkish içiyordu ‘was drinking, drank’. If used pluri-occasionally, it suggests a narrower presentness in the sense of temporariness, e.g., was drinking coffee every day. INTRAHF prefers to operate on [dyn] actional phrases. High-focal intraterminality is particularly fertile with dynamic contents implying internal evolution. INTRAHF items are often strongly preferred to represent ongoing dynamic – transitional or dynamic non- transitional, i.e. processual – events implying gradual change, particularly progress observable in gradually produced effects, e.g., is improving. Such combinations are often accompanied by expressions of graduality and speed, meaning ‘little by little’, ‘slightly’, ‘rapidly’, etc. To represent ongoing events expressed by [ dyn] actional phrases, items of lower focality are sufficient, e.g., the English Simple Present. A simple present is often sufficient to imply a narrower “nunc” if the event is of a mental nature. Irish so-called stative verbs such as sílim ‘I think, I am thinking’ may have PROG meaning without being used in the special INTRAHF form. Combinations such as English *was knowing, Maltese *qed jaf ‘*is knowing’ are usually blocked. Most preaspectual items show the same constraint, e.g., German *war am Wissen ‘was knowing’. Even in cases of rather narrow temporariness, many Romance and Germanic INTRAHF items refuse to operate on [ dyn] verbs of spatial location such as ‘hang’, ‘sit’, ‘stand’. The same holds for postural verbs in some Turkic languages (Johanson 1995). Preaspectual progressives

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

153

Viewpoint operators in European languages

153

used with [ dyn] actional contents often have derogative connotations. Some highfocals avoid passives, since non-agentive roles are often linked with [ dyn] actional contents. When the expression of high-focal intraterminality is renewed in a language, it does not affect [ dyn] actional phrases. On the other hand, when older intraterminals lose focality, becoming more and more restricted to nonfocal or modal uses, INTRA  [ dyn] combinations tend to retain low-focal functions longer than others. The former East Armenian Present tense is modal today, except with a few verbs, e.g., gitem ‘I know’. The present-day use of Turkish so-called Aorist, formerly a category of higher focality, is often less modal with [ dyn] verbs than with others, e.g., ister ‘wants’, bilir ‘knows’. Similarly, the Lezgian -da form preserves its older meaning with verbs such as k’anda ‘wants’, cˇ ida ‘knows’ (Haspelmath 1994: 276). Such residual meanings obviously result from aspectoactional interaction. It is, however, necessary to stress the relative nature of the ability of high-focals to operate on more or less dynamic actional contents. Some relatively high focals more readily accept less dynamic actional contents than others. The Icelandic vera ‘be’  Present Participle periphrasis is less restrictive than the vera að  Infinitive construction, e.g., er sofandi ‘is asleep’. Portuguese high-focals are relatively open, e.g., está a gostar ‘is liking’ (Oliveira & Lopes 1995: 107). Maltese high-focals combine with most [ dyn] actional phrases to express temporariness. The same holds for certain English high-focals, e.g., is feeling young. There is no absolute ‘unmarked present time context’ in which high-focals are obligatory. If high-focals of two languages differ in their use in certain cases of low actional dynamicity, this means that the defocalization drift has advanced further in one of the items, not necessarily that the events denoted are conceptualized differently in terms of ‘state’, ‘process’, etc. High-focal intraterminality is fertile with non- momentaneous transformatives and less fertile with momentaneous ones, though it may yield imminential or propinquitive readings (see below). The combination INTRAHF  [tf, mom] occurs in many languages with preliminary readings: a preliminary phase preparing for the attainment of the crucial limit is viewed as going on at an O, e.g., Icelandic er (alveg) að sofna ‘is (just) falling asleep’. But INTRAHF  [t] may also refer to events likely to occur in the immediate future without any meaning of ongoing preliminaries, the intraterminal viewpoint being situated immediately prior to the crucial limit (‘is / was close to V-ing’, ‘is / was about to V’). In some languages, INTRAHF items are the only device for expressing imminence with transformatives. Imminential readings often result from INTRAHF operating on [mom] actional contents, e.g., Icelandic er að fara ‘is leaving / about to leave’, Danish er ved at køre ‘is departing / about to depart’, Portuguese está a ganhar ‘is winning / about to win’; cf. Comrie’s ‘prospective’ aspect (1976: 64–65). Transformative phasal verbs such as begin and finish either do not combine with high-focals at all or get imminential read-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

154

154

Lars Johanson

ings, e.g., Icelandic vera að  byrja ‘begin’. Some high-focals are ambiguous with respect to imminence and ongoing process, e.g., is leaving, whereas others are unequivocally imminential, e.g., German ist im Begriff abzufahren ‘is [on the point of] leaving’. Certain preaspectuals prefer the feature [controlled] to express an ongoing process, whereas non-agentive and non-purpose actional phrases tend to express imminence, e.g., Swedish Tåget håller på att gå ‘The train is about to leave’ (not *‘already leaving’). High-focal intraterminals may thus yield prospective meaning and eventually develop into prospectives. Preaspectual high-focals often suggest propinquitive readings (‘be close to V-ing, but not V’), usually with [t] actional phrases, e.g., Danish var ved at dø, men ‘was on the verge of dying, but ’. The Lithuanian construction ‘was’  beprefixed Present Participle typically expresses a thwarted event (Mathiassen 1996). For Yiddish ikh halt baym shraybn ‘I am about to write’, see Ebert, this volume. 10.2.2.

INTRA

Nonintraterminals, which combine with all kinds of actional phrases, are unmarked for intraterminality and thus have negatively defined values. They present an integral view of the event, suggesting, unless the contrary is stated otherwise, that the event is carried out and comes to an end. I have referred to them as “pseudoperfectives”, since they differ essentially from adterminals, “perfectives proper”, but exhibit a partly similar textual behaviour (Johanson 1996). PAST ( INTRA) items are propulsive and thus used in narrative texts to indicate the succession of events. Interacting with INTRA and POST items, they may express situation changes occurring on the basis of ongoing situations. On the other hand, they are not explicitly limit-oriented as AD items are. They do not give any special prominence to limits in the meaning of the actional phrase and may indeed suggest rather cursus-oriented readings. This fact is often ignored in attempts at reducing AD and INTRA to a common PFV denominator under the assumption that IPFV items are cursus-oriented and PFV items limit-oriented. 10.2.2.1.

 [t] INTRA and [t] are, though not identical, highly compatible with each

INTRA

The values other. As INTRA does not deny the inclusion of the crucial limit into the view, the combination PAST ( INTRA)  [t] thus typically refers to accomplished past telic events. Transformative INTRA items are partly similar to adterminals in their textual behaviour. Nonintraterminality is sufficient to produce this similarity. In narrative discourse, INTRA  [t] may suggest readings of ‘entry’ into the event and ‘exit’ from it (“terminal attraction”; Johanson 1971: 211), but there may also be more cursus-oriented readings.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

155

Viewpoint operators in European languages

10.2.2.1.1.

INTRA

155

 [TI]

With initiotransformatives, INTRA items may suggest, among other interpretations, an initial attraction, putting an accent on the initial crucial point and highlighting the transformation. PAST ( INTRA)  [ti] combinations with initium-oriented ‘entry’ readings are, for example, Turkish yattı ‘lay down’, bildi ‘(suddenly) knew’, Modern Greek katálave ‘understood (= became aware of)’, kríftike ‘hid (= put himself / herself out of sight)’, East Armenian atec ‘(suddenly) hated, began to hate’. These readings contrast with the correspondingª INTRA readings dealt with above (10.2.1.2). Some Italo-Croatian verbs, e.g., znat ‘know’ and imat ‘have’, have copied the [ti] phase structure of corresponding Italian verbs, e.g., sapere, avere, and thus get analogous initium-oriented readings with PAST ( INTRA) items (‘learned’, ‘got’), where other Slavic languages such as Russian or Serbian use inceptive modes of action. Verbs such as cˇ ut ‘hear’ und vit ‘see’ are also used as [ti] verbs, e.g., sa cˇ uja ‘heard’, sa vidija ‘caught sight of’ (Breu 1992: 115). As was stressed above, however, initial attraction should not be confused with explicitly inceptive or ingressive modes of action. It is an important fact that PAST ( INTRA)  [ti] items also have more cursus-oriented readings without special attention given to the initium, e.g., Turkish oturdu ‘sat (for a period of time)’. It would thus be wrong to claim that it is the function of INTRA items to select the initium with verbs of this type, as is often assumed for PFV items. The idea that the INTRA opposition produces pairs of stative vs. ingressive meaning appears to be an unjustified objectivization of the aspect distinction in question. 10.2.2.1.2.

INTRA

 [TF]

With finitransformatives, INTRA items may suggest, among other interpretations, a final attraction, putting a certain accent on the final crucial point and thus highlighting the transformation. PAST ( INTRA)  [tf] combinations with finis-oriented ‘exit’ readings are, for example, Modern Greek péthane ‘died’, éxtise éna spíti ‘built a house’, Italo-Croatian ponija ‘brought’, Maltese are˙g ‘went out’, Turkish kitabı okudu ‘read the book’, Portuguese escreveu a carta ‘wrote the letter’. These are all examples of INTRA  [tf, mom]. As for the combination INTRA  [tf, mom], it can, since there is no salient cursus, only refer to the crucial limit of which the event consists, e.g., Portuguese ganhou ‘won’, Modern Greek vríke ‘found’. But it is not monoaspectual, as the combination INTRA  [tf, mom] is also possible. INTRA  [tf] combinations do not produce PFV items of the AD type, explicitly highlighting the finis. They just report that the event has obtained, not excluding the crucial limit from the view. The narrative behaviour of PAST ( INTRA)  [tf] is partly similar to that of PAST (AD) items. Both types are propulsive and

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

156

156

Lars Johanson

may express a transition to a new situation. This is considered a main PFV property. But INTRA is not a PFV of the AD type, highlighting the crucial limit. PAST ( INTRA)  [tf] does not select the final phase, as PFV pasts are supposed to do. The accent on the finis should not be confused with implicitly signalled egressivity. The possible interpretation of the event as an accomplished whole results from the inherent phase structure. The Maltese Simple Past expresses completion with transformatives only. Modern Greek éghrapse éna ghráma ‘wrote a letter’ corresponds to Russian PAST (AD) napisal pis’mo only in the case of undivided object reference. Hungarian Pasts operating on [tf] actional phrases differ clearly from PAST (AD) realizations and are interpreted by De Groot as “complete / perfective” situations (1984). A past tense of a finitransformative requires no qualified terminality to suggest readings of ‘completion’, ‘accomplishment’, ‘achievement’. Readings of ‘result’, ‘consumption’, ’exhaustiveness’, etc. are also effects of contextually determined transformativity (Johanson 1971: 208–213). The question whether [tf] is supported by explicit T-marking or not is relevant for languages such as Lithuanian, Bulgarian, Macedonian, Italo-Croatian, and Georgian. Thus, the Bulgarian perfective Aorist contains two features that can be assigned distinct functions: the viewpoint operator INTRA gives a nonintraterminal view of an event whose actionality is specified by T-marking. The Georgian Aorist, a PAST ( INTRA) item opposed to the PAST (INTRA) Imperfect, exhibits two variants, of which only one is a T-marked transformative, e.g., dacera ‘he wrote ª today). The (and finished writing)’, and an unmarked form cera (of very limited use ª item is thus part of two oppositions, one aspectual and one actional:

INTRA  [ t] cerda ‘was writing’ INTRA  [t] dacera ‘wrote’ vs. INTRA  [ t] cera ‘wrote’ The opposition between T-marked and T-unmarked INTRA items is typical of Georgian and Laz (Zan), but not of Svan, which is consistent in T-marking transformatives. T- marking of PAST ( INTRA)  [tf] combinations is sometimes not needed, since the actional phrase is obviously [tf] anyhow, e.g., Lithuanian mir˙e ‘died’, prad˙ejo ‘began’. However, T-marking is often required with actional ª

ª

ª

phrases denoting actions that may be either telic or atelic, e.g., Lithuanian paraš˙e ‘wrote’ (and finished writing)’ as opposed to a less goaldirected raš˙e ‘wrote’. As was noted above, Italian influence has led to decreasing T-marking of some verb categories in Italo-Croatian. Note again that a T-marker, unlike the viewpoint operator AD, operates on basic events. In cases of INTRA  [t]  [ser] [ t], i.e., when nonintraterminality is applied to the expression of a global event consisting of repeated occurences of telic subevents, T-marking naturally signals the actionality of each subevent, e.g., Bulgarian napravi tova njakolko p˘ati ‘did this several times’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

157

Viewpoint operators in European languages

157

As we have noted, ‘in X time’ adverbials, which measure accomplishment time, naturally combine with transformatives but do not cooccur with [ t] actional phrases under uni-occasional readings. When PAST ( INTRA) is applied to an actional phrase determined by an ‘in X time’ adverbial, the actionality is thus usually [tf], e.g., did it in ten minutes. Delimitative and perdurative modes of action require a specification of the time interval filled with the event and thus produce transformative actional phrases meaning ‘spend [a certain time period] V-ing’, the basic action being atelic. The time-specifying expression is no mensural adverbial of the ‘for X time’ type. For example, the T-marker of the Bulgarian perfective Aorist signals the [tf] actionality of a temporally delimited atelic event, e.g., PAST ( INTRA)  [tf] pospa dva cˇ asa ‘slept for two hours’. Since the actional phrase is [tf], corresponding Russian delimitatives and perduratives only choose the AD aspect, e.g., PAST (AD) tam prožil tri goda ‘spent three years there’. Compare PAST ( AD) items based on nontransformatives and provided with ‘for X time’ adverbials, e.g., žil tri goda tam ‘lived there for three years’. 10.2.2.1.3.

INTRA

[

T]

Nonintraterminals may also operate on nontransformatives, which is not possible with the adterminal type of PFV items. INTRA  [ t] suggests integral readings of realization of a portion of the actional content without special attention to limits. Even an event lacking salient limits has its initium, cursus and finis; none is excluded from the view. As a result of the weak terminality value of INTRA and the vague phase structure of [ t], the combination allows various initium-, cursus- and finisoriented readings, e.g., ‘(and then) read’, ‘read (for some time)’, ‘read (and then)’. It thus has a propulsive function in narrative discourse. In its general textual behaviour, however, it is more similar to the nonadterminal type of IPFV than to the adterminal type of PFV. Even INTRA  [ t, dyn] is indeed a possible combination, e.g., French contint ‘contained’, Turkish bekledi ‘waited’, Modern Greek élipse ‘lacked’. Static verbs are often claimed to be “monoaspectual”, i.e., not to combine with PFV items. The combinability may be reduced, e.g., in Modern Greek, where a few [ t, dyn] verbs such as kséro ‘know’, periéxo ‘contain’, periméno ‘wait’ lack Aorist forms (Sasse 1991b: 15), but, on the whole, the constraints concern the AD type of PFV rather than the INTRA type. Since INTRA items lack the marked holistic nature supposed to be typical of PFV, they cannot stress totality as the adterminal counterparts do. There is thus no basis for a clear-cut ‘progressivity’ vs. ‘totality’ duality with [ t, dyn] actional phrases. INTRA  [ t] expresses a portion of nontransformative actional content without any marked phase meaning. Due to the negative viewpoint value, it may get slightly limit-orientated readings, a weak initial or final attraction. Thus, items such

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

158

158

Lars Johanson

as Turkish a˘gladı and Albanian qau (from a˘gla- and qan ‘weep’) may put a certain accent on the initium, ‘(and then) cried, burst into tears’, and thus be translated by inceptives, e.g., Russian zaplakal ‘began to cry’. The combination often expresses the mere occurrence of a past event of some duration (‘for a certain time’), e.g., Modern Greek dhúlepse ‘worked’, Turkish okudu ‘read’, Italian scrissi ‘wrote’, Maltese kiteb ‘wrote’. Italo-Croatian T-unmarked PAST ( INTRA) items may get ‘for a while’ readings, e.g., dop ja sa rabija (jena mal) ‘then I worked (a little)’ (Breu 1992: 115). This is a kind of cursus-orientation that does not present the cursus from a point of view established in the middle of it. INTRA  [ t] does not specify whether the entity denoted by a possible direct object is totally affected (finished, consumed, etc.) or not. Undivided reference to the entity is neither signalled nor excluded. Without a limiting object, the actional phrase remains nontransformative and cannot be T-marked. In the Bulgarian senˇ tence Cetoxme knigi ot dva do cˇ etiri cˇ asa ‘We read books from two to four o’clock’ (Lindstedt 1985: 149), the [ t] actional content of cˇ eta remains T-unmarked, since it is not limited to [t] either by the object or by the adverbial specifying the temporal extension of the event. The event is characterized as having occurred in an integral and not goaldirected way, which excludes T-marking. There is no reason for “perfective” interpretations of such cases. The INTRA type of PFV, e.g., Modern Greek PAST ( INTRA) éghrapse ghrámata ‘wrote letters’, is thus rather similar to the AD type of IPFV, e.g., Russian PAST ( AD) pisal pis’ma. This is quite natural, since both represent an unmarked way of presenting a [ t] event, which is not the case with the AD type of PFV. PAST ( INTRA) [ t] is naturally compatible with ‘for X time’ expressions of temporal extension, specifying the length of the portion of action, e.g., Maltese raqad il-˙gurnata koll-ha ’slept the whole day’, Kalmyk xoyr cˇ asas avn tavn cˇ as kürtl gazet umšv ‘read newspapers from two to five o’clock’. Note that the actional content is not limited by such expressions but remains nontransformative. A wellknown example is Classical Greek ebasíleuse triákonta ét¯e ‘reigned for thirty years’, French régna trente ans, Bulgarian caruva trijset godini, Turkish otuz yıl hüküm sürdü, etc. With durational ‘for X time’ adverbials, T-marking languages such as Bulgarian and Georgian typically dispense with T-marking and use unmarked, unqualified PAST ( INTRA)  [ t] items. PAST ( INTRA)  [ t] may represent a nonintraterminally viewed atelic global event consisting of an unlimited or limited series of atelic or telic subevents. Readings with telic subevents presuppose recategorization [t]  [ser] [ t], e.g., Bulgarian cˇ esto se napiva ‘often got drunk’ (Ivanˇcev 1971: 133–134), na dva p˘ati minava ‘passed twice’ (Maslov 1981: 250). The reason for choosing PAST ( INTRA) is certainly not that the series was discontinued afterwards, which is equally possible with PAST (INTRA), but rather that the global event is not viewed from inside. As was stressed above, not even habitually repeated events need to be

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

159

Viewpoint operators in European languages

159

presented from an internal point of view but can also be envisaged in an integral way. INTRA  [ t] naturally excludes T-marking. As for actions that may be conceived of in different ways, T-unmarked verbs are used to distinguish atelic conceptualizations from goaldirected ones, e.g., Lithuanian uses raš˙e ‘wrote’, Bulgarian jado ‘ate’. Lack of T-marking only concerns the actional content. It may also mean that transformativity is not an essential or fertile notion with a given actional content and context. Bulgarian plain Aorists such as spa ‘slept’ are simply nonintraterminals suggesting, as opposed to Imperfects, an integral view of the event. Verbs that refuse T-markers have contents that are less fertile with transformativity. An item such as objadva ‘had dinner’ simply lacks a T-marker signalling transformativity, while its aspectual meaning arises from the INTRA value. Telicity of the event thus does not necessarily require T-marking of the actional phrase that denotes it. T-unmarked PAST ( INTRA) items are ideal for expressing general-factual meanings without stressing goal-directedness and accomplishment, for signalling mere occurrence or occupation with the action (‘has been / was engaged in V-ing’). T-marking may be dispensed with in constative, experiential, characterizing, type-referring utterances, in contexts where notions of completion and situation change are not essential (‘has [some time in the past] V-ed’). PAST ( INTRA) without T-marking in a case such as Bulgarian Dnes uˇciteljat me nakazva ‘Today the teacher punished me’ (Stankov 1980: 97) implies that the event is referred to in an integral, generalized way and that no crucial limit needs to be pointed out in the given context. T-unmarked PAST ( INTRA) items may also stand for telic events with subsequently cancelled results, e.g., Koj e otvarjal cˇ antata mi? ‘Who opened my bag [which is no longer open]?’ (Stankov 1976: 14–15). There is no need to stress the attainment of a transformation the result of which has been reversed. Note that PAST ( AD) items, which do not signal any crucial limit either, may be used to refer to telic events in similar ways. The situation may be similar in languages lacking systematic INTRA and AD distinctions. In Hungarian, an accomplished telic event may well be referred to without T-marking, e.g., tegnap írta a levelet ‘wrote the letter yesterday’. The T-marked version megírta a levelet ‘wrote (and finished) the letter’ adds the goal-directedness of the accomplished action. Tunmarked items are allowed, if the notion of goal-directedness is unessential.

10.3. Realizations of POST Next, the interactions between POST items and inherent phase structure types will be commented on. Their main discourse-pragmatic functions in narratives are ruption and retrospection for purposes of temporal or explicatory backgrounding. It

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

160

160

Lars Johanson

should be kept in mind that viewpoint operators are not defined as devices selecting phases of the lexical meaning of actional phrases. With finitransformatives, the phase highlighted by POST is never part of the lexical meaning. Thus, ‘have died’ is not part of die. With initiotransformatives, however, the posttransformative viewpoint may well be situated in the statal phase of the lexical meaning. Thus, ‘having sat down’ is part of the content of Turkish otur- ‘sit down  sit’. Most of the cases discussed below are realizations of high-focal postterminality with certain phase structure types. Some of them will be compared to corresponding combinations with high-focal intraterminality. The different values and their realizations are interrelated and understandable within the respective systems, but they cannot easily be explained by means of fixed functional stations based on situation types. If such analyses lead to the result that closely related realizations turn up at different stations such as ‘resultative’ and ‘progressive’, their interconnections become invisible. POST items will be largely disregarded, as they combine freely with different actional phrases and the resulting combinations do not exhibit any striking peculiarities. Note, however, that PAST ( POST) items (French passé simple, Italian passato remoto, etc.), which are often taken to be PFV items, readily combine with [ t] actional phrases, whereas this is excluded with PFV items of the adterminal kind.

10.3.1.

POST  [t]

Postterminality is particularly suited for transformatives. Thus, Armenian, POSTHF  [t] items typically combine with [t] verbs such as halvel ‘thaw’, barkanal ‘get angry’, as well as with expanded [t] actional phrases containing basic [ t] verbs as anel ‘do’, grel ‘write’. T-marking languages may use T-marked items. However, as we have noted, T-marking may be more or less important with different aspectotemporal items. Marking is often less relevant for highly focal postterminals that pay no or little attention to the action behind the postterminal state. The Classical Greek Perfect could easily dispense with T-markers void of lexical meaning. Whereas the Aorist apéthane ‘died (off)’ from (apo)thn©¡eisk- ‘die (off)’ was T-marked, the Perfect might lack T-marking: téthn¯eke ‘has died, is dead’. Under the influence of other forms, however, T-markers were often attached to the Perfect forms as well. Similarly, the Georgian passive high-focal postterminal ceria ‘it is written’ ª always carries a remains unmarked, whereas the more event-oriented Perfect almost T-marking preverb. As we have noted above, when low-focals lose their focality, the development typically begins with [t] actional phrases.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

161

Viewpoint operators in European languages

161

POST  [tf] POST  [tf] produces a perspective post terminum finalem. Since POST items

10.3.1.1.

can denote an event whose relevant limit is situated earlier than that of the preceding item, the combination with [tf] may reverse the linear order of events. If the action is conceptualized as having an inner goal, completive readings are produced. The inner goal may be signalled by T-marking, e.g., Bulgarian perfective Perfect PAST (POST)  [tf] napisal e ‘has written’. PAST (POST)  [t] suggests that the event has effects relevant to a past O2 . High-focals refer to the posttransformative state, e.g., Classical Greek kékl¯etai ‘[has been named and] is called’, gégraptai ‘is [already] written’, egégrapto ‘was [already] written’, kékt¯etai ‘has acquired’ ‘possesses’, Maltese jismu ‘is named’, fieq ‘is healed’, East Armenian lavacac e¯ ‘is healed’, Karachai ketibdi ‘has left (and is still gone)’, Romany his anphandlo ‘was tied’. Low-focals may, even when they denote telic events, lack T-marking in contexts where the constative, summarizing, type-referring statement of the occurrence of the event itself is more essential than the notion of completion, e.g., Bulgarian viždal e ‘has (at least once) seen it’. This is typical of postterminally characterized events not conceived of as occurring in particular settings. Compare the analogous use of PAST ( AD) items instead of PAST (AD) ones and of T-unmarked PAST ( INTRA) items instead of T-marked ones.

10.3.1.2.

POST  [ti]

Combinations of postterminality with initiotransformatives produce a perspective post terminum initialem. A [ti] actional phrase denotes both a transformation and a following posttransformational phase. The POST  [ti] combination is often misunderstood as an interaction of PFV with “stative” verbs. However, [ti] verbs are neither stative nor inchoative. The combination is only intelligible as the effect of a postterminal view of an actional content that combines a dynamic transformative phase with a non-dynamic posttransformational one. The latter may thus be objectively identical to the phase highlighted by a corresponding intraterminal. The relation is particularly clear with high-focals. POSTHF  [ti] expresses a state that has come about through an initial transformation. This state may correspond to the cursus of the same actional phrase, the state highlighted by a INTRA  [ti] item. A statement meaning ‘is in the state of having sat down, is seated’ may refer to the same objective situation as a statement meaning ‘is in the state of sitting’. Similar equations are ‘has learnt’ = ‘knows’, ‘has got fond of’ = ‘likes’, ‘has stood up’ = ‘is standing’, ‘had lain down’ = ‘was lying’. Some Maltese initiotransformatives have special high-focal participial items that focus on the posttransformational state obtaining at O, thus covering situations that might alternatively be expressed by high-focal INTRA items, e.g., rieqed ‘has fallen asleep’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

162

162

Lars Johanson

= ‘is asleep’. English examples are has hidden = is hiding, has leaned = is leaning, has understood = understands. Verbs occuring in POST  [ti] combinations often express both the acquisition and the possession of positions and properties, e.g., ‘get, have’, ‘become similar, resemble’, ‘call to mind, remember’, ‘catch sight, look’, ‘come to believe, believe’, ‘get accustomed, be accustomed’, ‘fall asleep, be asleep’, ‘get filled, be full’, ‘grasp, hold’, ‘learn, know’, ‘get hope, hope’. As we have noted, [ti] verbs are frequent in some European languages and rare in others. Classical Greek Perfects representing a posttransformational state are, e.g., dédorke ‘looks, is looking’, gég¯ethe ‘is glad’, éoike ‘resembles’. They often denote the same phase as the corresponding Present items, e.g., dérketai ‘looks’, g¯ethe˜ı ‘is glad’, or lack a corresponding Present (éoike). Similar examples are Armenian nstac e¯ ‘has sat down’ = ‘is sitting’ from nstel ‘sit down, sit’, barkacac e¯ ‘has got angry’ = ‘is angry’ from barkanal ‘get, be angry’, kangnac e¯ ‘has stood up’ = ‘is standing’ from kangnel ‘stand up, stand’, Archi osdili i ‘is standing’ from ocis ‘stand still’ and axuli i ‘is lying’ from axas ‘lie down’ (Kibrik 1983: 113), Albanian ka qëndruar ‘has stopped’ = ‘is standing’ (= qëndron), Talysh hïtä be ‘had fallen asleep = was sleeping’ (Miller 1953: 173) from hïte ‘fall asleep, sleep’, nïšta be ‘has sat down = was sitting’ from nïšte ‘sit down, sit’. A Turkish example is oturmu¸stur ‘has sat down’ = oturuyor ‘is sitting’ from otur‘sit down, sit’. Among other Turkish verbs are dur- ‘stop, stand’, inan- ‘come to believe, believe’, yat- ‘lie down, lie’, uyu- ‘fall asleep, sleep’. Corresponding Kalmyk verbs are kevt- ‘lie down, lie’, med- ‘get to know, know’, s¯u- ‘sit down, sit’, unt- ‘fall asleep, sleep’, zogs- ‘stop, stand’, e.g., unt-sn ‘is asleep’, untsn bilä ‘was asleep’, suusn ‘is sitting’, suusn bilä ‘was sitting’. Compare Maltese so-called Imperfects with certain [ti] verbs, e.g., jaf ‘knows’, jixbah ‘resembles’ (Borg 1981: 157) and Romany items such as hi bešdo ‘is sitting’, hi tardo ‘is standing’, hi tšido ‘is lying’. High-focal “present state” uses of POST  [ti] may, as is well known, develop into Present items. Germanic PreteritePresents, formed from originally strong verbs, have Present forms going back to POSTHF  [ti], e.g., Gothic wait ‘knows’, Old Icelandic veit ‘knows’, á ‘possesses’, mun ‘remembers’, skal ‘shall’, English can, Norwegian vet (cf. Classical Greek o˜ıde ‘knows’). The effect of high-focal postterminality on [ti] verbs as well as the close connection with corresponding intraterminals remains unintelligible as long as the analysis is based on translations into languages lacking [ti] verbs. PAST (POST) and PAST (POST) items combined with [ti] are often taken to be non-pasts and simple pasts respectively, whereas the corresponding combinations with [tf] are regarded as past and past-before-past items respectively. The Kalmyk postterminal item in -jˇ is sometimes claimed to express both past and present events (“Geschehenes”, “Geschehendes”) without clear rules. This variation is, however,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

163

Viewpoint operators in European languages

163

regular and due to the inherent phase structure. Finitransformatives such as ömsjˇ or irjˇ are interpreted as ‘has / had dressed’ = ‘is / was dressed’ (compare Turkish giyinmistir / giyinmisti) or ‘has / had come’ = ‘is / was (t)here’ (compare Turkish gelmi¸stir / gelmi¸sti). On the other hand, initiotransformatives such as s¯uˇj are interpreted as ‘has / had sat down’ = ‘is / was sitting’ (compare Turkish oturmu¸stur / oturmu¸stu). ¯ is sometimes thought to have both In the same way, the Kalmyk item in -ad bänä PF and progressive meaning (‘has done’, ‘is doing’), depending on the verb type, but it always expresses a dwelling in the postterminal state, the only variable being the phase structure. Similarly, the Maltese postterminal active participle expresses a posttransformational state with certain verbs. With [ti] verbs such as raqad ‘fall asleep, sleep’ and rikeb ‘mount, ride’, we get PAST (POSTHF)  [ti] realizations, e.g. rieqed ‘is sleeping’ ‘has fallen asleep’, riekeb ‘is on horseback’ ‘has mounted’. The same situation applies to motion verbs meaning ‘get moving, move on’ such as mexa ‘set out, walk’. Thus, miexi ‘is walking’ represents a POSTHF aspect of a [ti] actional content and refers to a posttransformational phase of this content, obtaining as a state at Os . It may be tempting to claim that the participle encodes ‘progressivity’ in such cases. However, although progressive interpretations are suggested by the English translations (‘dressed’ ‘wearing’, etc.), this is not a progressive item in the sense of a high-focal intraterminal. POSTHF items differ from INTRAHF items by lacking internal dynamicity and not being capable of future time reference. POST  [ti] combinations competing with INTRA  [ti] ones lack the graduality necessary for implying internal dynamicity. For example, in Armenian, only the latter type combines with expressions such as aveli u aveli ‘more and more’, kamac-kamac ‘little by little’. ª

ª

POST  [ t] When a POST operator applies to a [

10.3.2.

t] actional content, the initial limit is the relevant one. With the combination PAST (POST)  [ t], the event has at least begun prior to Os and may also overlap it. PAST (POST)  [ t] suggests a past O2 . If an actional phrase such as read a book is interpreted as [ t], lacking an inner goal of the action, its combination with PAST (POST) implies has read (some pages of) a book. T-marking languages dispense with T-marking in such cases. Bulgarian thus uses its imperfective Perfect for PAST (POST)  [ t]. But there are also certain constraints on the combination POST  [ t]. The ban on high- focal postterminals with nontransformatives has been mentioned already. There is little need for representing a state post terminum if a crucial limit is lacking. In a few languages, however, items known as high-focal postterminals can also be used with [ t] verbs, e.g., Kalmyk kelsn bilä ‘was speaking’. The relevant limit is, as expected, the initium. The nontransformatives are thus treated very much like initio-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

164

164

Lars Johanson

transformatives. Similar situations are found in several Eurasian languages outside the boundaries of Europe. Certain traces in Indo-European languages are also reminiscent of this phenomenon. Here, some postterminal participles may get different temporal readings according to their internal phase structure. Passive postterminal participles may get nonanterior readings with [ t] verbs, e.g., German ist geliebt ‘is loved’, but anterior readings with [t] verbs, e.g., ist gefunden ‘has been found’. If we use Esperanto as a metalanguage, this gives the following German–Esperanto equivalences: ist gefunden = estas trovita ‘has been found’, war geschrieben = estis trovita ‘had been found’, ist geliebt = estas amata ‘is loved’, war geliebt = estis amata ‘was loved’. The participles of [ t] verbs are thus clearly postterminal in the sense of the initium being the relevant limit. (Note, however, that languages such as English and French have developed passives with which [t] verbs also get nonanterior readings, e.g., is found, est trouvé.) The applicability of POST to [ t] increases with lower focality degrees, but there are often restrictions on combinations with [ t, dyn] actional phrases. This is, for example, true of the Modern Greek possessive Perfect. POST items are also excluded with Archi verbs such as hubus ‘blow’, qebus ‘dance’, and arhas ‘think’ (Kibrik 1983: 113). As was noted above, when low-focals lose their focality, [ t] actional phrases are usually the last to be affected. Thus, the Portuguese and Galician Perfects are largely restricted to [ t], with constative uses summarizing the event at Os . These are clear examples of an interaction with an actional value. As we have seen, [ t] combinations often produce constative (“experiential”) readings of lowfocals. Finally, defocalized PAST (POSTo ( INTRA)) items such as the Maltese Perfect are, of course, fully combinable with [ t] actional phrases. They are often classified as PFV items, though their properties are very different from those of PAST (AD) items.

10.4. Realizations of AD There is a good deal of interdependence between AD and [t] actionality. As we will see, there is no freedom of choice if the global event is referred to with a [tf, mom] or a [ t] actional phrase. The former case requires AD, the latter case AD. Adverbials expressing temporal punctuality such as Russian vdrug ‘suddenly’ naturally require AD. Incompatibility with such adverbials is, however, only typical of the AD variety of IPFV, not of the INTRA variety (Section 7.2.4). 10.4.1.

AD  [t]

Adterminality suggests strong limit-orientation, envisaging the event in its attainment of the crucial limit. It thus presupposes [t].

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

165

Viewpoint operators in European languages

165

AD is particularly suited to operate on finitransformatives: AD  [tf, mom] gives strong prominence to the finis and may also yield readings that imply an exit from the event. The actional phrase denotes an action heading for completion, and the viewpoint operator signals the attainment of it. The combination does not favour [ser] readings. An actional content referring to the lowest quantity of a telic event requires AD, e.g., Polish umarł ‘died’. Russian perfectives such as poslat’ ‘send’, razbudit’ ‘waken’ or vstat’ ‘get up’ express a total event and its conclusion. The AD meaning requires that the actional content of napisal pis’ma be interpreted as [tf] including a limiting object, a set of letters referred to as a whole: ‘wrote (and finished) (the) letters’ (undivided reference to the set). As we have noted, [tf] does not combine with ‘for X time’ adverbials. Thus, AD  [tf] napisal pis’mo ‘wrote the (whole) letter’ cannot be modified by adverbials such as dolgo ‘for a long time’ or dva cˇ asa ‘for two hours’. Since [tf, mom] actional contents lack a salient cursus, AD  [t, mom] envisages nothing more than the crucial limit itself in the moment of transformation. There is no AD option. AD  [ti] gives strong prominence to the initium, signalling the attainment of the initial crucial limit. It may thus suggest the entry into the event, e.g., Russian opersja na stol ‘leant over the table’. For the realizations mentioned, terms such as “resultative”, “egressive”, “punctual”, “ingressive”, and “inceptive” should be avoided, since they easily lead to confusion with modes of action (4.2). 10.4.2. *AD  [ t]

Adterminality does not operate on nontransformatives: *AD  [ t]. This is an intelligible constraint on the freedom of aspect choice, since there is no use for items signalling the attainment of a limit not present in the actional content. Thus, [ t, dyn] actional phrases only combine with AD. Russian [ t, dyn] verbs, whose lexical meanings do not imply clearly delimited extensions, are imperfectiva tantum, for example, prinadležat’ ‘belong’, sostojat’ ‘consist’, znat’ ‘know’. However, this is no reason to take [ t, dyn] actional phrases to be “monoaspectual” in principle, in the sense of IPFV being the only option. It has already been noted that INTRA  [ t, dyn] is a possible combination. The situation may seem different with [ t, dyn] actional phrases. Russian perfectives can express past events such as prožil tri goda v Moskve ‘spent three years in Moscow’. The possibility of choosing prožil here instead of žil is often regarded as evidence for mutual independence of aspect and actionality. However, this option requires a delimitative or perdurative item that provides the actional content with a crucial limit indicating the minimal- maximal extension. Perduratives such as prožil combine with mensural expressions specifying the length of the whole period of time. Such modes of action constitute [tf] actional phrases, which are naturally compatible with AD. The resulting combinations sum up the event as a totality

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

166

166

Lars Johanson

reaching the final crucial limit. Adterminality is thus by no means incompatible with expressions of duration as such. However, there is no option *AD  [ t, dyn]. This fact sharply distinguishes AD from PFV items of the INTRA type, which suggest integral readings with [ t, dyn] actional phrases. 10.4.3.

AD

[

t]

AD Nonadterminality envisages the event without highlighting a crucial limit. items exclude limit-orientation and suggest a cursus-orientation which, however, is much weaker than with INTRA items. In its textual behaviour, a PAST ( AD) item such as Russian pisal ‘wrote’ thus corresponds both to the intraterminals was writing, écrivait and to the nonintraterminals wrote, écrivit. Nonadterminality has a natural affinity with nontransformativity. If the global event referred to is expressed with a [ t] actional phrase, AD is normally required. Applied to [ t, dyn] phrases, AD presents the event as a static stretch of action without any salient natural limits, e.g., Russian visel ‘hung’. The combination is typically expressed by imperfectiva tantum. This does not mean that static nontransformatives are “aspectually indifferent” in general, as claimed by Bache (1982: 69). The claim that they are only referred to by IPFV items is not valid for INTRA languages, which use them freely with INTRA items. Combinations of AD with [ t, dyn] actional phrases present the event as a stretch of action, without suggesting totality or attainment of any limit. On the other hand, they do not yield the ‘progressive’ readings possible with IPFV items of the INTRA type. In narrative discourse, AD  [ t, dyn] does not suggest any ‘entry’ or ‘exit’, but merely ‘occurrence (for a period of time)’. It often simply denotes the lexical content, notably a content with which AD is rather infertile. Since AD implies the involvement of a crucial limit, it is natural to interpret AD items as implying no such limit. The actional content of Russian pisal (‘wrote’) pis’mo ‘letter’ is thus prima facie interpreted as [ t], without a limiting object referred to as an undivided entity. But AD is also used where the mere occurrence of the event, not its full performance or completion, is important. The event is referred to in a generalized manner without highlighting any phase. Such readings often occur in contexts where the event is conceived of as taking place outside a particular setting. Such an unqualified, “characterizing” manner of presentation is what might be expected from the unmarked member of a AD opposition. In similar cases, INTRA languages use unqualified INTRA pasts, and languages with T-marking dispense with this device. Compare, for example, Russian PAST ( AD)  [ t, dyn] Ty pisal ej? with Turkish ( INTRA)  [ t, dyn] Ona yazdın mı? ‘Have you written to her?’. As [t] is incompatible with ‘for X time’ adverbials, any actional phrase is identified as [ t] by occurring with them. Russian AD  [ t] žil tri goda v Moskve ‘lived for three years in Moscow’ has a ‘for X time’ adverbial but no limiting object.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

167

Viewpoint operators in European languages

167

The event is characterized as simply having occurred for a certain period of time. In cˇ ital knigi ot dvux do cˇ etyrex cˇ asov ‘read books from two to four o’clock’ (cf. Thelin 1978: 33), the action denoted by the [ t] verb cˇ itat’ is not limited to [tf] by the object. The interactional meaning PAST ( AD)  [ t, dyn] is ‘a stretch of action took place’ without an idea of a crucial limit implied. The temporal extension is specified by the ‘for X time’ adverbial ot dvux do cˇ etyrex cˇ asov ‘from two to four o’clock’. With [ t, dyn] verbs, both AD and AD are possible and their viewpoint values realized in a particularly clear way. If both AD  [ t, dyn] and AD  [ t, dyn] can be used to refer to a given situation, e.g., skazal and govoril ‘said’, the difference only lies in the view presented. It would be absurd to claim that the opposition is “neutralized” here or that AD is used “instead of” AD. Since AD does not highlight any limit, a AD  [ t, dyn] case such as Russian pisal pis’mo may certainly be used when the speaker is concerned with the internal structure of the event. It may textually even correspond to English ‘was writing a / the letter’ or German ‘schrieb an einem / an dem Brief’. But it does not present the event as a continuous, unfolding process of gradual achievement of the result (Maslov 1948: 313). AD items do not, as INTRA items do, describe the event with specific attention to its internal structure. Their way of presenting events is not particularly “unfolding”. On the contrary, it is essential for their use that they do not suggest any intraterminal meaning. Because the viewpoint is not explicitly intra terminos, AD readily combines with ‘for X time’ statements, e.g., pisal dva cˇ asa ‘wrote for two hours’, carstvoval tridcat’ let ‘ruled for thirty years’. This is unusual with INTRA items, since the idea of introspection is incompatible with the indication of the total temporal extension. When AD operates on [ser] actional phrases denoting pluri-occasionality, the reading [ser] may be overtly marked or not. The global event is presented by AD regardless of the actional properties of the subevents. If the latter are telic and basically represented by [t] actional phrases, we get the combination AD  [t]  [ser] [ t], e.g., Russian cˇ ital po knige v den’ ‘read a book a day’, proˇcityval knigu ‘repeatedly read through a / the book’. The subevents may also be atelic, e.g., Russian PAST ( AD)  [ t]  [ser] [ t] cˇ ital ètu knigu každyj den’ ‘read in this book every day’ or poˇcityval knigu ‘repeatedly read a little in a / the book’. 10.4.4.

AD

 [t]

While naturally operating on nontransformatives, AD can also apply to transformatives. The reason is that it expresses a perspective that neither considers nor negates the limits of the event. It does not signal non-completion, but may suggest this as a negative value. Even so, the combination AD  [t] is perfectly logical, since heading for a crucial limit is not equal to reaching it.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

168

168

Lars Johanson

Whereas the Russian-type perfectives are portmanteau AD- and T-markers, the corresponding imperfectives are neither AD- nor T-markers and thus less distinct. Besides standing for AD  [ t], they may also express a nonadterminal view of a non-momentaneous telic event. Our model includes the combination AD  [t], where the actional phrase is left T-unmarked, though it denotes a telic event. This means that the combination PAST ( AD)  [t] may also be used for fully accomplished events. It neither highlights the transformation itself, nor excludes the unabridged achievement of the event. For example, Russian pisal pis’mo ‘write a / the letter’ may not only be interpreted as PAST ( AD)  [ t], a process excluding the attainment of an inner goal, but also as PAST ( AD)  [t], an event including the total involvement of a limiting object. In INTRA languages, the latter reading is typical of PAST ( INTRA) items. Though AD is often claimed to be obligatory for telic events, the combination AD  [t, mom] is perfectly possible. With heterogeneous actional contents that imply both a statal and a dynamic phase, the Russian-type imperfectives get statal interpretations. With initiotransformatives, perfectives highlight the initium, e.g., opersja ‘leant’, whereas imperfectives may give a similar prominence to the resulting state, e.g., opiralsja ‘leant, was leaning’. These cases represent a contrast AD  [ti] vs. AD  [ti]. As for finitransformatives, AD highlights the finis, whereas AD  [tf] may, unless the event is conceived of as momentaneous, give a certain prominence to the preceding state. AD does not combine with [tf, mom] actional contents, as they only consist of the crucial limit that is to be attained. If AD is applied to verbs that are basically [tf, mom], the nontransformativizing feature [ser] is added. The typical interactional meaning ‘repetition’ is an effect of this recategorization [tf]  [ser] = [ t]. Russian imperfectives such as naxodit’ ‘find’ and vkljuˇcat’ ‘switch on’ imply repetition if the basic event is conceived of as momentaneous. Thus, the aspect distinction has important actional implications with [tf, mom] actional phrases. Several Russian [tf, mom] verbs are unpaired perfectives and offer no AD option, e.g., zaplakat’ ‘start to cry’. Russian PAST ( AD) items can be used to refer to telic events the result of which has subsequently been cancelled. Thus, otkryval dver’ ‘(had) opened the door’ may be interpreted as a so-called ‘twoway action’ with the implication that ‘the door was closed again’. Such readings are also possible with momentaneous telic events, which demonstrates a certain affinity with [ser] readings, but the common feature is simply that no crucial limit is highlighted. The claim that such imperfectives are in reality perfectives (Ferrand 1982) seems to be based on the equation of ‘perfective’ with ‘telic’. Some possible cases of AD  [t] are problematic. It seems that AD  [tf, mom] may get processual and preliminary readings suggesting a process that precedes the transformation. These readings are typical of Russian imperfectives such as brat’ ‘take’, davat’ ‘give’, padat’ ‘fall’, posylat’ ‘send’. Such verbs might be

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

169

Viewpoint operators in European languages

169

claimed to differ from their perfective counterparts, vzjat’ ‘take’ etc., by being nontransformative. However, they express a nonadterminal view of telic events, and their actional content might be taken to be transformative as well, the only difference from the perfectives residing in the viewpoint. Forms such as bral ‘took’ would thus not only represent AD  [ t], but also the case AD  [t], though without T-marking. By contrast with their perfective counterparts, vzjal etc., forms such as bral ‘took’ get preliminary readings that do not include the attainment of the point of transformation of ‘taking’, ‘giving’, ‘falling’, ‘sending’, etc., but only a stretch of action heading for the transformation. They may even express conation, the attempt to perform the telic action and to reach the crucial limit, e.g., ubeždal ‘tried to convince’; cf. Polish doganiał go, ale nie dogonił ‘tried to pass him, but did not succeed’ (Koschmieder 1934). These imperfectives do not express “nonevents”, but just events excluding the crucial limit from the range of view. However, they are unusual both in meaning and shape. Their behaviour is similar to INTRA items, and they differ from their perfective counterparts by stem-internal markers rather than by the absence of preverbs. It is possible that they should rightly be considered marked members of limited INTRA oppositions within an otherwise AD-dominated system. In that case, umiral would be a PAST (INTRA)  [tf] item meaning ‘was in the process of dying’ as opposed to PAST ( INTRA)  [tf] item umer ‘died (ceased to live)’.

11. Main aspectotemporal types 11.1. Types discussed Our discussions have so far focused on items marked for INTRA, POST and AD and their respective opposition partners. If we disregard differences with respect to high and low focality, the above analyses yield a rather limited number of aspectotemporal types, which are listed here together with more traditional – and partly misleading – designations:



PAST ( INTRAÆ )



PAST ( POST)

PAST ( AD) Imperfective Present

Present

Perfect

PAST (INTRA)

PAST (

Imperfect

Preterite

Perfective Present

PAST (POSTÆ)

PAST ( AD) Imperfective Past

PAST (

INTRA)

Aorist

Pluperfect

POST)



PAST ( AD)

PAST (AD) Perfective Past

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

170

170

Lars Johanson

This list does not exhaust the number of main aspectotemporal types to be discerned in European languages. Some more complex ones, combining positive and negative viewpoint values, have already been touched upon above in a general way. The major additional types will be listed below.

11.2.

PAST (

POST

(INTRA))

Besides the intraterminal type PAST (INTRA), there is a PAST ( POST (INTRA)) type that also takes part in a postterminal opposition, e.g., English was writing, opposed to PAST ( POST ( INTRA)) wrote and PAST (POST ( INTRA)) has written. PAST (INTRA) items are semantically more general, since they are also used for past events that a POST language would express postterminally.

11.3.

PAST

(POST (

INTRA))

Besides the postterminal type PAST (POST), there is a PAST (POST ( INTRA)) type that also takes part in an intraterminal opposition, e.g., English has written, Armenian grel e¯ , High (Hill) Mari siren. PAST (POST) items are semantically more general, being used for past events that a INTRA language would express intraterminally. Some PAST (POST ( INTRA)) items are semantically more general than others, for example, Estonian on teinud ‘has done’, since the competing PAST ( POST (INTRA)) item is less strongly grammaticalized in the sense of having a rather limited use, e.g., Estonian oli tegemas ‘was doing’. Some PAST (POST ( INTRA)) items have a relatively restricted use themselves, e.g., Irish tá tar éis leabhar a scríobh ‘has (just) written a book’ as opposed to the PAST ( POST ( INTRA)) item scríobh sé leabhar ‘wrote / has written a book’.

11.4.

PAST

(POST (INTRA))

In addition to PAST (INTRA) and PAST ( POST (INTRA)), there is a third intraterminal type, PAST (POST (INTRA)), taking part in both a postterminal and an intraterminal opposition. Thus, the English so-called Continuative Perfect has been writing is opposed to PAST (POST ( INTRA)) has written and PAST ( POST (INTRA)) was writing. The complex construction serves to express a particular kind of interaction in which a POST notion operates upon a INTRA notion to express the Os relevance of an intraterminally viewed event. The English type combines a progressive periphrasis with postterminal morphology, using a POST distinction of the ‘be’ auxiliary: PAST (POST) has been vs. PAST ( POST) was.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

171

Viewpoint operators in European languages

171

Note that this type, although taken to be a “ ” item by Kuryłowicz (1956: 26), is by no means a high-focal in the sense of PAST (POSTHF ). Thus, has been dying is not equal to Classical Greek téthn¯eke ‘is dead’. PAST (POST (INTRA)) is a special kind of postterminal, signalling that the point transgressed at Os is the O2 of the intraterminal perspective. The event is viewed diagnostically and as having continuing relevance to the primary deictic centre Os . The intraterminal view highlights an internal portion of the event at an O2 consisting of one or more intervals of unspecified localization. O2 , the vantage interval of INTRA, is situated within the limits of the event, and Os , the vantage point of POST, after the beginning of this interval. Since the localization is unspecified, this item, like other focal postterminals, does not readily combine with specific time expressions. The interpretation of the postterminal element as PFV and of the intraterminal element as IPFV can only lead to the conclusion that PAST (POST (INTRA)) items are aspectually self-contradictory or at least ambivalent. The kind of retrospective intraterminality conveyed here suggests that the event, which has begun at some time in the past, continues right up to Os and that the event itself, its effects, or its concomitants overlap this point. A sentence such as has been examining the case easily suggests, since it only refers to the involvement, that, despite all efforts in the past, the final limit of the examination has not yet been reached at Os . But has examined the case may also imply Os -inclusion, since a POST operator only requires the initial limit to be transgressed unless the actional phrase is unequivocally finitransformative. However, PAST (POST (INTRA)) items tend more strongly towards incomplete readings. The choice of an item that only highlights internal parts of a given event may, though the main function is mere introspection, get the pragmatic implicature that only part of the event has taken place, so that it is still unfinished at Os . Similar contrasts are found in Irish. A PAST (POST ( INTRA)) construction such as in tá tar éis leabhar a scríobh ‘has (just) written a book’ presents the whole event without isolating or highighting any part of it. The choice of PAST (POST (INTRA)) tá tar éis a bheith ag scríobh leabhair ‘has been writing a book’ adds an extra dimension and focuses on the intervening periods of writing (Ó Baoill 1994: 209). While PAST (POST (INTRA)) items operate on all kinds of actional phrases, nontransformatives and initiotransformatives are preferred. Restriction to the latter would mean consistent Os -overlapping and the possiblity of describing Os as the orientation point for both the POST and the INTRA view.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

172

172

Lars Johanson

11.5. Pluperfects Besides simple PAST (POSTo) items, e.g., German hatte geschrieben ‘had written’, postterminal items involved in an intraterminal opposition may occur in the pre-past stratum, e.g., English PAST (POSTo (INTRA)) had been writing and PAST (POSTo ( INTRA)) had written. Distinctions between non-dynamic and dynamic items in the pre-past stratum have been mentioned above (8.3).

11.6. Historical pasts European languages exhibit different types of historical pasts, event-oriented, propulsive preterites that may refer to events conceived of as occurring on specific past occasions. Some of them are commonly considered PFV items, whereas others are called “simple Pasts”. However, they are of rather different kinds. Most items are complex ones, deriving their values from several oppositions. Only one type is a general past in the sense of being applicable to any past event. AD, INTRA, and POST are obviously appropriate values for constituting historical pasts. AD makes an item propulsive by highlighting the crucial limit and thus changing the situation. POST makes an item particularly event-oriented and suited to refer to events conceived of as occurring on specific past occasions, in particular settings, pinpointed at specific intervals. INTRA makes it particularly suited to present the event in an integral way. An item interacting with both a PAST (INTRA) and a PAST (POST) item is thus ideally suited to envisage an event directly and in its totality. Qualified historical items of this type often cooccur with definite past time adverbials. They are typically incompatible with markers of indefinite time and of unrestricted or very long periods, as well as with adverbials of the types ‘already’ (‘the relevant limit is not later than O’) and ‘not yet’ (‘the relevant limit is later than O’). However, clear-cut AD, INTRA or POST values are not required for the historical functions. Many items are suited as narrative pasts, for narrating sequences of discrete past events under the general conditions of linear successivity, without being “perfectives” in any qualified sense. The mere absence of intraterminality allows presentation of the event in an integral way. Several events presented one after another in a nonintraterminal past tense are most naturally interpreted as a sequence in the sense of ‘did x, and [then] did y’. Nonfocal postterminals and other generalized items are potentially event-oriented and applicable to specific past occasions. Note that a past that does not compete with a postterminal can be both diagnostic and historical, covering all the focality degrees of a postterminal. On the other hand, a past that does not compete with an intraterminal can cover the focality degrees of an intraterminal. Thus, PAST ( POST) Danish skrev, PAST ( AD) Russian

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

173

Viewpoint operators in European languages

173

pisal, and Hungarian PAST írt may all stand for a limited or extended “nunc” (‘was writing, used to write, wrote’). The sources of nonpostterminals, nonintraterminals, and more general items (without any INTRA or POST commitment) will not be discussed here in detail. Many items have long developments behind them, being expressed by bound elements with reduced shapes typical of old forms representing late stages of grammaticalization. In some cases, the sources are unidentifiable. Indo-European past items are marked by an augment, originally an independent word ( adverb), proclitically added to the verbal form. Turkic and Mongolian historical items in -di and in -ba[i] are old forms with highly eroded material shapes that are difficult to etymologize. If the item in -di goes back to a verbal noun provided with a possessive suffix, it might ultimately be traced back to a possessive high-focal postterminal. Similar possessive pasts are found in Finno-Ugrian and Tungus languages (Benzing 1988: 48–49). But defocalization as such does not naturally lead to the creation of POST items. There is no evidence that European PAST ( POST) items that do not interact with INTRA pasts have developed from postterminals (cf. the discussion in Bybee et al. 1994: 81–85). Nonfocal postterminals can only develop into POST items, if new focal postterminals are created to interact with them. Nor is there evidence that European nonpostterminals go back to the same sources as PFV items of the adterminal kind, i.e. to completives or constructions with auxiliaries derived from directional and movement verbs.

PAST ( INTRA) Aorists of the PAST ( INTRA) type, e.g., colloquial French a écrit ‘wrote, has written’, only interact with a INTRA category and are indifferent to AD and POST. 11.6.1.

They may present past events both in a historical and a diagnostic way, covering the domains of English PAST (POST) and POST ( PAST) items, e.g., Turkish yazdı or Maltese kiteb ‘has written, wrote’. Some Pasts of this type are, however, opposed to a Remote Past, e.g., the Kabardian item in -a (vs. the Remote Past in -a- a-). In a system involving a INTRA distinction, a PAST (POST) item may lose its focality and generalize into a PAST (POSTo ( INTRA)) item, taking over a function of the old superseded PAST ( INTRA) item. The INTRA opposition is thus continued after a formal substitution of the unmarked member. For example, the Latin Perfect cantavit, serving to express both ‘has sung’ and ‘sang’, developed into a nonfocal postterminal and a nonintraterminal, covering the whole range of past uses except for cases that motivate an intraterminal view (cantabat). Similarly, most Romance reflexes of an originally high-focal postterminal type habet cantatum, e.g., North Italian ha cantato, have generalized to nonfocals, taking over the INTRA values from old PAST ( POST ( INTRA)) items, e.g., cantò and other reflexes of cantavit. Certain varieties, however, have preserved the passé simple, passato

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

174

174

Lars Johanson

remoto, perfectul simplu, etc., as the INTRA member of the opposition. In many non-Romance languages, the INTRA opposition similarly consists of an Imperfect and a PAST (POSTo ( INTRA)) item, e.g., in Italo-Croatian, where the old Aorist has disappeared and the former Perfect has lost its focality. It has already been noted that, if a POST opposition is given up, it is not always the former PAST (POST) item that generalizes. The former PAST ( POST) item may also develop into a more general item referring to past events both historically and diagnostically. In some South Italian dialects, the use of the former PAST ( POST) item (Simple Past) has generalized at the expense of the former PAST (POST) item (Perfect), e.g., Calabrian fu ‘became, has become’ (also instead of è stato ‘has become’; Bosco 1924: 13). While the European Spanish type PAST (POST) ha hablado is encroaching on the functional territory of PAST ( POST) habló, the opposite tendency is observed in varieties of South American Spanish, e.g., in Chilean Spanish, where PAST (POST) (pretérito perfecto compuesto) is being replaced by PAST ( POST) (pretérito indefinido). Similar tendencies are also met with in American English, where the Simple Past seems to be gaining ground from the Perfect. However, the Perfect is still present in these varieties. In Turkish, on the other hand, the item in -di, e.g., yazdı ‘wrote’, has almost generalized as a nonintraterminal past, whereas the finite item in -mi¸s mainly fulfills indirective functions, e.g., yazmı¸s ‘has [apparently] written’ and the item in -mi¸stir is, at least in the spoken language, a presumptive, e.g., yazmı¸stır ‘has [presumably] written’ (Section 8.7). A PAST ( INTRA) item that does not compete with a postterminal can be both historical and diagnostic, covering all the focality degrees of a postterminal without being marked for them. Turkish items such as bitti or kalmadı, operating on transformatives, do not only mean ‘ended, has ended’ and ‘did not remain, has not remained’, respectively, but may also refer to the posttransformational state without regard to the event: bitti ‘is finished, over’, kalmadı ‘is all gone, used up’. Similar examples are Latin PAST ( INTRA) items (“Perfects”) such as n¯ovit and c¯onsu¯evit, which may mean ‘knows’ and ‘is accustomed to’, respectively. 11.6.2.

PAST (

POST)

PAST ( POST) items interact with a POST category but are indifferent to AD and INTRA. This plain type is present in a few Germanic and Finno-Ugrian languages, e.g., Norwegian skrev ‘wrote’, Finnish kirjoitti ‘wrote’, Latvian rakst¯ıja ‘wrote’, Dutch werd geschreven ‘was written’, Swedish skrevs ‘was written’. Similar items are found in Slavic varieties that have preserved the old Aorist but lost the old Imperfect, e.g., Serbian pisa ‘wrote’. One of the two Basque Past tenses, formed from synthetical verbs only, is indifferent to INTRA and thus belongs to this category; e.g., ba-nengoen ‘I stayed’. It can always be used when the analytical

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

175

Viewpoint operators in European languages

175

constructions would necessitate a choice between PAST ( POST (INTRA)) and PAST ( POST ( INTRA)) items (Haase 1994: 284). 11.6.3.

PAST (

POST ( INTRA))

Nonpostterminals frequently compete with a INTRA category as well, which produces a PAST ( POST ( INTRA)) type of nonpostterminal and nonintraterminal items marked for anteriority. Thus, English wrote is opposed to the items PAST ( POST (INTRA)) was writing and PAST (POST ( INTRA)) has written. Other examples of PAST ( POST ( INTRA)) items are Modern Greek éghrapse ‘wrote, has written’, French écrivit, Portuguese morreu ‘died, has died’, Spanish entró ‘entered’, Catalan perdé ‘lost’, Romanian cânt˘a ‘sang’, Albanian vrau ‘killed’, Bulgarian vleze ‘entered’, Sami båradii ‘ate, has eaten’, High (Hill) Mari sirïš ‘wrote’, Tatar bardï ‘went’, Karachai aldï ‘took’, Kalmyk kelv ‘spoke’, Tati baft ‘wove’, Kirmanji hat ‘came’, Armenian grec, Lezgian k˘xena ‘wrote’. ª In several languages, e.g., French, Italian, Romanian, Serbian, Albanian, the type is limited to certain varieties (8.10.2). The Basque item (“Perfective Past”) is restricted to analytical verbs (Haase 1994: 282). The languages in question thus possess three items covering the simple past stratum: a PAST (INTRA) Imperfect, a PAST (POST) Perfect, and a PAST ( POST ( INTRA)) Aorist or Simple Past. This is also a frequent pattern in IndoEuropean languages bordering on the European area, e.g., Persian PAST (INTRA) m¯ıkard ‘was doing, did’, PAST (POST) karda ast ‘has done’ and PAST ( POST ( INTRA)) kard ‘did’. Creole systems, e.g., Romanian ones, possess non-temporalized “neutral” POST ( INTRA) items implying anteriority with transformatives and nonanteriority with nontransformatives. 11.6.4.

PAST

A plain, generalized type with the simple value PAST, not interacting with any other viewpoint category and signalling anteriority only, is found in some languages such as Hungarian, South German, Yiddish, and Ossetic, e.g., South German hat geschrieben ‘has written’. The Hungarian and Ossetic items are even general pasts, e.g., írt ‘has written, wrote, had written’, Ossetic dzïrdta ‘spoke, has spoken, had spoken’. These PAST items void of INTRA, POST and AD values cover all kinds of past events. They combine with transformatives to the effect that undivided reference implies completion. If the referent of a direct object is totally involved in the actional content, it is limiting, i.e. understood to be totally affected with the realization of the event, for example, Hungarian kiolvasta a könyvet ‘read the (whole) book’ (De Groot 1984). In the diagnostic dimension, they cover different focality

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

176

176

Lars Johanson

degrees of postterminals. Items operating on transformatives may often be used in adjective-like ways, denoting the posttransformational state, e.g., Hungarian fáradt PAST fáradt ‘has endeavoured’ ‘is tired’. Hungarian szokott ‘is accustomed’ is a lexicalized former high-focal postterminal, now functioning as a special expression of habituality. Considered diachronically, these pasts are PAST ( POSTo) items. It was noted above that PAST (POST) items may lose their focality, generalize and oust former PAST ( POST) items. Several unqualified PAST items have once been PAST ( POST ( INTRA)) items, before the POST and the INTRA oppositions in question were lost. The defocalized Romanian Perfect seems to be developing this way, taking over functions of the vanishing Imperfect (Haase 1995: 142). On the other hand, plain pasts may of course develop into PAST ( INTRA), PAST ( POST), or PAST ( POST ( INTRA)) items, if new INTRA or POST distinctions are created. 11.6.5.

PAST (

AD)

Most Slavic languages have lost their INTRA oppositions, and their former Perfects have turned into nonfocals without competing INTRA or POST items. Some of these Perfects have served as the basis for the creation of adterminality oppositions. Russian pisal ‘wrote, was writing’ and Czech spal ‘slept, was sleeping’ are basically postterminals that have lost their focality, generalized PAST (POSTo) = PAST items, which have acquired the structure PAST ( AD). They do not take part in any other viewpoint opposition. If AD items are the only past items in a system, they are both diagnostic and historical. Though European nonadterminals lack linguistically relevant focality degrees, they do cover the various degrees specified by postterminals. Thus, PAST ( AD) items are often used for what is expressed by PAST (POSTLF) in POST languages. This is natural, since in both cases the very attainment of the relevant limit is unessential: the postterminal conveys the view after the limit has been reached, and the nonadterminal denotes a past event without reference to a limit. A past that does not compete with a INTRA item can also cover all the focality degrees of an intraterminal. Thus, PAST ( AD) Russian pisal may stand for a more or less limited or extended presentness in the past: ‘was writing’, ‘wrote’, ‘used to write’.

PAST (AD) PAST (AD) items only signal adterminality and do not interact with other view-

11.6.6.

point categories. An item such as Polish napisał or Czech napsal ‘wrote’ is basically a postterminal that has lost its focality but is marked for AD. It is both historical and diagnostic, covering all the degrees of focal postterminals. Thus, Russian PAST

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

177

Viewpoint operators in European languages

177

(AD) ustal does not only mean ‘(has / had) got tired’ but may also denote the posttransformational state without referring to the event: ‘is tired’.

11.7. Relations to PF, IPFV, PFV Let us finally add a few summarizing words on how the main aspectotemporal types relate to crosslinguistic types such as PF, IPFV and PFV. Even if PF may be subsumed under the POST domain, it only represents one sector of it. Many postterminals do not fit into the PF type. On the other hand, many items that appear in typical PF uses, e.g., PAST ( INTRA) or PAST items, are not postterminals. Postterminality has a broader crosslinguistic occurrence and higher degrees of independence, stability and combinability. PF items tend to occur in limited context types. On the other hand, they are constantly threatened by defocalization. It is not easy to establish PF as a universal type, particularly since ‘current relevance’ does not seem to be a sufficient criterion for all language-specific categories claimed to belong to it. Given its vagueness, it might not have acquired the status of a cardinal crosslinguistic type, had it not been for the English Perfect, which shows a remarkable correspondence to it. PF is sometimes taken to constitute a special case of PFV to the effect that PFV signals completion of the event at Os , whereas PF adds the notion of current relevance. However, not all PFV items satiate the criterion ‘completion at Os ’, and the only common temporal property of PFV and PF seems to be: ‘at least E* is reached at Os ’. At any rate, it should be clear that postterminality cannot be considered a subspecies of PFV. The relationship between the main aspectotemporal types and the IPFV vs. PFV dualism can be summarized as follows. IPFV roughly corresponds to two kinds of items with the values INTRA and AD, respectively. Their common property, manifested in similar textual behaviour, is a view of the event without regard to boundaries and with more attention to the internal structure. However, the two IPFV types differ from each other in essential respects. Most IPFV representatives are marked INTRA categories; very few are unmarked AD categories. Whereas INTRA items view the event within the limits, AD items attract attention to the internal structure by negating or ignoring the notion of an attained limit. The inclusive semantic relationship of IPFV and PROG only applies to the INTRA type. The AD type does not include any high-focal or progressive subcategory. PFV roughly corresponds to two kinds of items with the values AD and INTRA, respectively. The unifying property, manifested in similar textual behaviour, is the view of the event as a totality including its beginning and its end, without attention to its internal structure. Both types are holistic, capable of presenting an integral

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

178

178

Lars Johanson

view of the event, and thus propulsive. Despite the analogies, however, the two types differ from each other in many respects. AD pasts function as PFV items by virtue of a positive value implying a transformation, INTRA pasts by negating the introspective meaning. Despite all similarities, the general distribution of the two types of IPFV vs. PFV oppositions tends to be different (Johanson 1971: 88–100). As we have noted, AD items have much in common with INTRA items, both being suited for the mere statement of the occurrence of the event in a ‘general factual’ or ‘simple denotative’ sense. The conflation of several “perfective” types has a long history, beginning with Curtius’ confusing equation of the Classical Greek Aorist with Slavic perfectives (1846, cf. Ruipérez 1954: 34). However, the semantic differences between the two opposition types are clearly motivated by their different values and cannot be explained as marginal Slavic idiosyncrasies (Maslov 1973: 82; Bondarko 1983: 117). They are certainly not satisfactorily accounted for by reference to the different morphological (inflectional vs. derivational) make-up. Nor is it sufficient as a synchronic definition to refer to “source determination”, pointing out that the specific grammatical meaning of AD perfectives “follows directly from their richer lexical meanings” (Bybee et alii 1994: 89). What we observe here are substantial differences, based on different linguistic values, rather than minor details of realization. Moreover, as we have seen, several other Past types in European languages are capable of suggesting a direct and integral view of the event and are thus used for propulsive narration of sequences of past events. Such similar roles should not lead us to equate the linguistic values. Only PAST (AD) is marked, the other categories (“unqualified terminal pasts”) taking up the space left over by different competing marked categories in their systems. Some are POST and INTRA items opposed to POST and INTRA items. Others have no POST or INTRA items to contrast with, and are thus indifferent to post- and intraterminality respectively. If “perfectivity” is defined in this unqualified sense, the AD type cannot be claimed to be typical of it. The different types discerned above may be summarized as follows, listed in descending order according to their degree of explicitness with respect to direct and integral presentation: PAST (AD), PAST ( POST ( INTRA)), PAST ( POST), PAST ( INTRA), PAST.

11.8. System types, areal distribution The viewpoint operator systems dealt with above are of varying complexity. Very often, older stages of development are relatively rich, whereas a reduction is observed in modern ones. Thus, the Proto-Slavic aspectotemporal system was comprehensive, including INTRA, POST, and PAST oppositions that yielded a PAST (INTRAo) Present, a PAST (INTRA) Imperfect, a PAST ( POST ( INTRA))

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

179

Viewpoint operators in European languages

179

Aorist, a PAST (POST) Perfect, and a PAST (POSTo) Pluperfect. There was also an actional T-marking perfective vs. imperfective opposition, rather freely combinable with the members of the aspectotemporal oppositions. While most Slavic languages have lost this richness, Bulgarian and Macedonian have preserved and developed it. Italo-Croatian has a INTRA opposition Imperfect vs. Aorist ( Perfect) and a T-marking system, whose values combine to form perfective Imperfects INTRA  [t], imperfective Imperfects INTRA  [ t], perfective Aorists INTRA  [t], and imperfective Aorists INTRA  [ t]. As for the Romance languages, Latin displays a richer system than its modern daughters. The rich Classical Greek system has been transformed into a not less rich Modern Greek system. Germanic systems are usually restricted to POST oppositions and lack systematic T-marking. Western Finno-Ugrian systems are rather similar to them, but have also developed different additional devices. As we have seen, this is even true of the considerably reduced and “apparently extremely simple Hungarian system” (Comrie 1994: 299), which has, for example, developed a systematic T-marking. Numerous modern languages of the Caucasian region have elaborate systems of distinctions in the POST and INTRA dimensions. On the other hand, peculiarly scanty systems are found in some languages, e.g., within the Daghestanian Andi group (Godoberi, Karata, etc.). Among the Iranian languages, Ossetic has a strikingly simple system, comprising one PAST (INTRAo ) and one PAST item. Turkic systems are, right from the Old Turkic period, highly well-equipped, including PAST, INTRA, and POST oppositions. Many more recent Turkic languages have also developed a rather systematic actional T- and T-marking, which also affects members of aspectotemporal oppositions. Questions of areal distribution, though occasionally touched upon above, will not be dealt with systematically in the present contribution. A comparison of the individual areas of reasonably grammaticalized INTRA, POST, and AD categories yields isoglosses that do not constitute coherent geographic areas. Both INTRA and POST are predominantly found in languages of peripheral areas, Icelandic, English, Irish, Ibero-Romanian, Bulgarian, Macedonian, Modern Greek, eastern FinnoUgrian, Turkic, Caucasian, Mongolian, etc. Languages void of well grammaticalized INTRA, POST, or AD categories (German, Hungarian, etc.) occupy a small, rather central area. Systems essentially restricted to POST (Scandinavian, Faroese, Frisian, Dutch, Baltic Finnic, etc.) are found farther north than systems essentially limited to INTRA (French, Italian, Romanian, Maltese, Turkish, etc.). The Slavic languages void of other categories than AD (mainly East and West Slavic) are distributed over a huge area in the eastern and central parts of Europe. The distribution is not very strongly correlated to genetic groups. Though there are highly stable systems, e.g., within the Turkic group, contact-induced changes have played an important part. The languages of Western Europe exhibit certain structural similarities that can only go back to long contact. Some are obviously due

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

180

180

Lars Johanson

to influence of dominant colonial languages, in particular Latin. There are several typical spread zones, e.g., north, south, and west of the Black Sea as well as in vast parts of Eastern Europe, areas of long-standing contact between Indo-European, Turkic, Finno-Ugrian, and other elements. Some contact-induced changes imply copying (some kind of “borrowing”) of new items, whose effects depend on what items are already present in the recipient language. The Mari system is clearly influenced by the Turkic languages of the Volga region, and there is a considerable Azeri impact on Talysh and on the Daghestanian Lezgian group. The Upper Sorbian system is rather similar to the German one. Irish has a INTRA, POST system that is strongly reminiscent of the English system. However, no evidence can be found for the alleged Romance influence on the Maltese system, which is very close to those of other Arabic dialects (Ebert, this volume). Many changes start with frequential copying, i.e. contact-induced decrease or increase in the use of a given item (Johanson 1992: 182–183). For example, the increase in the use of Latvian PAST ( POST) items instead of compound PAST (POST) items may be due to influence from Russian, where PAST (AD) items cover all past events (Mathiassen 1996: 15). A system may also change so strongly that it comes very close to the model system. The Basque system has changed in a way that mirrors the oppositions – and in part the formal structure – of the systems of neighbouring Romance languages (Comrie 1994: 299). There are also long chains of contact-induced changes, e.g., the widespread loss of PAST (POST) and PAST (POST ( INTRA)) items and their replacement by PAST, PAST ( INTRA) and PAST (AD) items in Eastern and Central Europe. Despite all differences in details, the overall similarities of the systems of viewpoint operators in European languages are rather striking. Occasionally, the available data of certain non-Indo-European languages, in particular Basque and some Caucasian languages spoken in residual zones, give an impression of underlying old deviant structures. However, it is by no means the task of the present contribution to try to trace documented structures back to prehistorical ones.

References Aerts, Willem J. 1965 Periphrastica. An investigation into the use of é˜ınai and ékhein as auxiliaries or pseudoauxiliaries in Greek from Homer up to the present day. Amsterdam: Hakkert. Anderson, Lloyd B. 1982 “The ‘perfect’ as a universal and as a language-specific category”, in: P. J. Hopper (ed.), Tense – aspect: between semantics and pragmatics. (Typological Studies in Language 1.) Amsterdam-Philadelphia: John Benjamins, 227–264. Antinucci, Francesco & Ruth Miller 1976) “How children talk about what happened”, in: Journal of Child Language 3: 167–189.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

181

Viewpoint operators in European languages

181

Aronson, Howard I. 1967 “The grammatical categories of the indicative in the contemporary Bulgarian literary language” in: To honor Roman Jakobson, I. (Janua Linguarum, Series Maior 31.) The Hague–Paris: Mouton, 82–98. Bache, Carl 1982 “Aspect and Aktionsart: Towards a semantic distinction”, Journal of Linguistics 18: 57–72. Barentsen, Adriaan A. 1985 ‘Tijd’, ‘aspect’ en de conjunctie poka. Over betekenis en gebruik van enkele vormen in het moderne Russisch. Amsterdam. [No indication of publisher]. Benzing, Johannes 1985 Kalmückische Grammatik zum Nachschlagen. (Turcologica 1.) Wiesbaden: Harrassowitz. 1988 Kritische Beiträge zur Altaistik und Turkologie. (Turcologica 3.) Wiesbaden: Harrassowitz. Bertinetto, Pier Marco 1987 “Structure and origin of the ‘narrative’ imperfect”, in: A. Giacalone Ramat, O. Carruba & G. Bernini (eds.), Papers from the 7th International Conference on Historical Linguistics. Amsterdam, Philadelphia: Benjamins, 71–85. this volume “The progressive in Romance, as compared with English”. Bertinetto, Pier Marco & Denis Delfitto this volume “Aspect vs. actionality. Some reasons for keeping them apart”. Bertinetto, Pier Marco, Karen Ebert, Casper de Groot this volume “The progressive in Europe”. Bickerton, Derek 1975 Dynamics of a creole system. New York: Cambridge University Press. Bondarko, Aleksandr V. 1983) Principy funkcional’noj grammatiki i voprosy aspektologii. Leningrad: Nauka. Borg, Albert J. 1981 A study of aspect in Maltese. (Linguistica Extranea 15.) Ann Arbor: Karoma. Bosco, Umberto 1924 Esercizi di traduzione dai dialetti della Calabria. (Dal dialetto alla lingua.) Torino. Bossong, Georg 1993 “Innovative Tendenzen im sardischen Vokalsystem”, in: Johannes Kramer & Guntram A. Plangg (eds.), Verbum Romanicum. Festschrift für Maria Iliesen. Hamburg: Helmut Buske, 337-352. Breu, Walter 1985 “Handlungsgrenzen als Grundlage der Verbklassifikation”, in: Werner Lehfeldt (ed.), Slavistische Linguistik 1984. München: Otto Sagner, 9–34. 1991 “Das italoalbanische ‘Perfekt’ in sprachvergleichender Sicht”, in: Francesco Altimari et alii (eds.), Atti del Congresso Internazionale di Studi sulla Lingua, la Storia e la Cultura degli Albanesi d’Italia. Celuc-Rende, 51–66. 1992 “Das italokroatische Verbsystem zwichen slawischem Erbe und kontaktbedingter Entwicklung”, in: Tilmann Reuther (ed.), Slavistische Linguistik 1991. Referate des XVII. Konstanzer Slavistischen Arbeitstreffens Klagenfurt-St. Georg/Längsee 10.–14.9.1991. (Slavistische Beiträge 292.) München: Otto Sagner. 93–122. Browning, Robert 1983 Medieval and Modern Greek. (2nd edition.) Cambridge: Cambridge University Press. Brugmann, Karl 1900 Griechische Grammatik. Lautlehre, Stammbildungs- und Flexionslehre und Syntax. (3rd edition.) München: Beck.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

182

182

Lars Johanson

1904

Kurze vergleichende Grammatik der indogermanischen Sprachen. Strassburg: Karl J. Trübner. Buchholz, Oda & Wilfried Fiedler 1987 Albanische Grammatik. Leipzig: VEB Verlag Enzyklopädie. Bybee, Joan L. & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13/1: 51–103. Bybee, Joan L. & Revere Perkins & William Pagliuca 1994 The evolution of grammar. Chicago & London: The University of Chicago Press. Comrie, Bernard 1976 Aspect. An introduction to the study of verbal aspect and related problems. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press. 1994 Review of R. Thieroff & J. Ballweg (eds.), Linguistics 30: 298. Csató, Éva Ágnes 1992 “On some theoretical and methodological problems of the typological study of tenseaspect categories”, in: E UROTYP Working Papers 6/1: 29–36. 1994 “Tense and actionality in Hungarian”, in: R. Thieroff & J. Ballweg (eds.), 231–246. 2000 “Zur Phasenstruktur ungarischer Aktionalphrasen”, in Walter Breu (ed.), Probleme der Interaktion von Lexik und Aspekt (ILA). Tübingen. Max Niemeyer. 75–89. Curtius, Georg 1846 Die Bildung der Tempora und Modi im Griechischen und Lateinischen, sprachvergleichend dargestellt. Berlin: W. Besser. Dahl, Östen 1985 Tense and aspect systems. Oxford: Basil Blackwell. Dahl, Östen & Eva Hedin this volume “Current relevance and event reference”. Deeters, Gerhard 1930 Das khartwelische Verbum. Vergleichende Darstellung des Verbalbaus der südkaukasischen Sprachen. Leipzig: Markert & Petters. Dik, Simon 1989 The theory of functional grammar. Part 1: The structure of the clause. (Functional Grammar series 9.) Dordrecht: Foris. Dietrich, Wolf 1973 Der periphrastische Verbalaspekt in den romanischen Sprachen. Tübingen: Narr. Doerfer, Gerhard 1993 “Das türkische Suffix -mIš als Lehnelement”, in: Jens Peter Laut & Klaus Röhrborn (eds.), Sprach- und Kulturkontakte der türkischen Völker. Wiesbaden: Harrassowitz, 87–92. Dowty, David R. 1972 Studies in the logic of verb aspect and time reference in English. (Studies in Linguistics 1.) Austin: Department of Linguistics, University of Texas. 1977 “Toward a semantic analysis of verb aspect and the English ’imperfective’ progressive”, Linguistics and Philosophy 1: 45–78. Ebert, Karen H. this volume “Progressives in Germanic languages”. Faßke, Helmut 1981 Grammatik der obersorbischen Schriftsprache der Gegenwart. Morphologie. Verfaßt von H. Faßke unter Mitarbeit von Siegfried Michalk. Bautzen: Domowina. Ferrand, Marcel 1982) “Les prétérits russes de l’aller et retour unique (xodil, prixodil, bral, etc.) sont-ils perféctifs?”, Revue des Études Slaves 54: 455–475.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

183

Viewpoint operators in European languages Forsyth, John 1970

183

A grammar of aspect. Usage and meaning in the Russian verb. Cambridge: Cambridge University Press. Friedman, Victor A. 1977 The grammatical categories of the Macedonian indicative. Columbus, Ohio: Slavica Publishers. Galton, Herbert 1962 Aorist und Aspekt im Slavischen. Eine Studie zur funktionellen und historischen Syntax. Wiesbaden: Harrassowitz. Givón, Talmy 1982 “Tense-aspect-modality: the creole prototype and beyond”, in: Paul J. Hopper (ed.), Tense–aspect: Between semantics and pragmatics. Amsterdam: John Benjamins, 115– 163. Graves, Nina this volume “Macedonian – a language with three perfects?”. Grønbech, Kaare & John R. Krueger 1955 An introduction to classical (literary) Mongolian. Wiesbaden: Harrassowitz. Groot, Casper de 1984 “Totally affected. Aspect and three-place predicates in Hungarian”, in: C. de Groot & H. Tommola (eds.), 133–151. this volume “The absentive”. Groot, Casper de & Hannu Tommola (eds.) 1984 Aspect bound. A voyage into the realm of Germanic, Slavonic and Finno-Ugrian aspectology. Dordrecht: Foris. Gvozdanovi´c, Jadranka 1995 “Western South Slavic tenses in a typological perspective”, in: R. Thieroff (ed.), 181– 194. Haase, Martin 1994 “Tense and aspect in Basque”, in: R. Thieroff & J. Ballweg (eds.), 279–292. 1995 “Tense, aspect and mood in Romanian”, in: R. Thieroff (ed.), 135–152. Harris, Martin 1982 “The ‘past simple’ and the ‘present perfect’ in Romance”, in: Nigel Vincent & Martin Harris (eds.), Studies in the Romance verb. London: Croom Helm, 42–70. 1988 “French”, in: Martin Harris & Nigel Vincent (eds.), The Romance languages. London & Sydney: Croom Helm, 209–245. Haspelmath, Martin 1994 “The tense system of Lezgian”, in: R. Thieroff & J. Ballweg (eds.), 267–277. Haugen, Einar 1976 The Scandinavian languages. An introduction to their history. London: Faber & Faber. Hedin, Eva 1987 On the use of the Perfect and the Pluperfect in Modern Greek. (Acta Universitatis Stockholmiensis: Studia Graeca Stockholmiensia VI.) Stockholm: Almqvist & Wiksell. this volume “The type-referring function of the Imperfective.” Hermann, Eduard 1927 “Objektive und subjektive Aktionsart”. Indogermanische Forschungen 45: 207–228. Holzinger, Daniel 1993 Das Rómanes. Grammatik und Diskursanalyse der Sprache der Sinte. (Innsbrucker Beiträge zur Kulturwissenschaft, Sonderheft 85.) Innsbruck: Institut für Sprachwissenschaft.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

184

184

Lars Johanson

Hopper, Paul J. 1979 “Foregrounding and aspect in discourse”, in: Talmy Givón (ed.), Syntax and semantics 12. Discourse and syntax. New York: Academic Press, 213–241. Ivanˇcev, Svetomir 1971 Problemi na aspektualnostta v slavjanskite ezici. Sofija: Izdatelstvo na BAN. Jakobson, Roman 1932 “Zur Struktur des russischen Verbums”, in Charisteria Guilelmo Mathesio quinquagenario a discipulis et Circuli Linguistici Pragensis sodalibus oblata, Pragae: Cercle linguistique de Prague. 74–84. 1957 Shifters, verbal categories and the Russian verb. Cambridge, Mass.: Department of Slavic Languages and Literatures, Harvard University. Janakiev, Miroslav 1962 “Za gramemite, nariˇcani v b˘algarskata gramartika ‘segašno vreme’ i ‘b˘adešte vreme”’, Izvestija na instituta za b˘algarski ezik 8: 419–432. Johanson, Lars 1971 Aspekt im Türkischen. Vorstudien zu einer Beschreibung des türkeitürkischen Aspektsystems. (Acta Universitatis Upsaliensis. Studia Turcica Upsaliensia 1.) Uppsala: Almqvist & Wiksell. 1974a “Sprachbau und Inhaltssyntax am Beispiel des Türkischen”, Orientalia Suecana (Uppsala) 22: 82–106. 1974b Review of Michael Weiers: Die Sprache der Moghol der Provinz Herat in Afghanistan, Acta Orientalia (Kopenhagen) 36: 459–472. 1975 “Aktionalphrase und Verlaufsordnung”, Studia Neophilologica 47: 120–150. 1988 “Iranian elements in Azeri Turkish”, in: Ehsan Yarshater (ed.), Encyclopædia Iranica, 3. London & New York: 248b–251a. 1992 Strukturelle Faktoren in türkischen Sprachkontakten. (Sitzungsberichte der Wissenschaftlichen Gesellschaft an der J. W. Goethe-Universität Frankfurt am Main 29: 5.) Stuttgart: Steiner. 1993 “Zur Entstehung historischer Präterita im Türkischen”, Türk Dilleri Arastırmaları (Ankara) 3: 119–127. 1994 “Türkeitürkische Aspektotempora”, in: R. Thieroff & J. Ballweg (eds.), 247–266. 1995 “Mehrdeutigkeit in der türkischen Verbalkomposition”, in: Marcel Erdal & Semih Tezcan (eds.), Beläk Bitig. Sprachstudien für Gerhard Doerfer zum 75. Geburtstag. (Turcologica 23.) Wiesbaden: Harrassowitz, 81–101. 1996 “Terminality operators and their hierarchical status”, in: Betty Devriendt & Louis Goossens & Johan van der Auwera (eds.), Complex structures: A functionalist perspective (Functional Grammar Series 17.) Berlin, New York: Mouton de Gruyter, 229–258. 1998 “Zum Kontakteinfluß türkischer Indirektive”, in: Nurettin Demir & Erika Taube (eds.), Turkologie heute – Tradition und Perspektive, 141–150. Wiesbaden: Harrassowitz. Kibrik, Aleksandr E. 1977 Opyt strukturnogo opisanija arˇcinskogo jazyka, 2. Taksonomiˇceskaja grammatika. Moskva: Izdatel’stvo Moskovskogo Universiteta. 1983 “Rezul’tativ v arˇcinskom jazyke”, in: V. P. Nedjalkov (ed.), 109–118. Koller, Hermann 1951 “Praesens historicum und erzählendes Imperfekt. Beitrag zur Aktionsart der Präsensstammzeiten im Lateinischen und Griechischen”, Museum Helveticum 8, 1: 63–99. Kormušin, Igor V. 1984 Sistemy vremen glagola v altajskix jazykax. Moskva: Nauka. Koschmieder, Erwin 1934 Nauka o aspektach czasownika polskiego w zarysie. Próba syntezy. (Rozprawy i materiały wydziału I Towarzystwa Przyjaciół Nauk w Wilnie, 5: 2) Wilno.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

185

Viewpoint operators in European languages

185

Kozinceva, Natalija A. 1983 “Rezul’tativ, passiv i perfekt v armjanskom jazyke”, in: V.P. Nedjalkov (ed.), 204–216. Kozintseva, Natalia 1995 “The tense system of Modern Eastern Armenian”, in: R. Thieroff (ed.), 277–297. Kuryłowicz, Jerzy 1956 L’apophonie en indo-européen. (Polska Akademia Nauk. Komitet je¬ zykoznawczy. Prace je¬ zykoznawcze 9.) Wrocław. Lindstedt, Jouko 1985 On the semantics of tense and aspect in Bulgarian. (Slavica Helsingiensia 4.) Helsinki: University of Helsinki. Lyons, John 1977 Semantics I–II. Cambridge: Cambridge University Press. Macaulay, Ronald K. S. 1978 Review of Comrie (1976), Language 54: 416–420. Maslov, Jurij S. 1948 “Vid i leksiˇceskoe znaˇcenie glagola v sovremennom russkom literaturnom jazyke”, Izvestija Akademii Nauk SSSR, otdelenie literatury i jazyka, 1948: 7, 4: 303–316. 1959 “Glagol’nyj vid v sovremennom bolgarskom literaturnom jazyke (znaˇcenie i upotreblenie)”, in: S. B. Bernštejn (ed.), Voprosy grammatiki bolgarskogo literaturnogo jazyka, Moskva: Akademija nauk SSSR. Institut slavjanovedenija, 157–312. 1973 “Universal’nye semantiˇceskie komponenty v soderžanii grammatiˇceskoj kategorii soveršennogo / nesoveršennogo vida”, Sovetskoe slavjanovedenie 1973, 4: 73–83. 1980 “Struktura povestvovatel’nogo teksta i tipologija slavjanskix vido-vremennyx sistem”, Svantevit (Århus) 6: 43–70. 1981 Grammatika bolgarskogo jazyka dlja studentov filologiˇceskix fakul’tetov universitetov. Moskva: Vysšaja škola. 1985 “An outline of contrastive aspectology”, in: Jurij S. Maslov (ed.), Contrastive studies in verbal aspect in Russian, English, French and German. (Studies in descriptive linguistics 14.) Heidelberg: Julius Groos, 1–44. 1988 “Resultative, perfect, and aspect”, in: V. P. Nedjalkov (ed.), Typology of resultative constructions. Amsterdam: Benjamins, 63–85. McCawley, James D. 1971 “Tense and time reference in English”, in: Charles J. Fillmore & D. Terence Langendoen (eds.), Studies in linguistic semantics. New York: Holt, Rinehart, and Winston, 97–113. Mathiassen, Terje 1996 Tense, mood and aspect in Lithuanian and Latvian. (Meddelelser, Universitetet i Oslo, Slavisk-baltisk avdeling 75.) Oslo. Metslang, Helle & Hannu Tommola 1995 “Zum Tempussystem des Estnischen”, in: R. Thieroff (ed.), 299–326. Miller, Boris V. 1953 Talyškij jazyk. Moskva: Izdatel’stvo Akademii Nauk SSSR. Miller, Vsevolod F. 1962 Jazyk osetin. Moskva & Leningrad: Nauka. Mourek, V. E. 1895 “Review of R. Wustmann: Verba perfectiva, namentlich im Heliand”, Anzeiger für deutsches Altertum und deutsche Literatur (Berlin) 21: 195–204. Nedjalkov, Vladimir P. (ed.) 1983 Tipologija rezul’tativnyx konstrukcij (rezul’tativ, stativ, passiv, perfekt). Leningrad: Nauka. Ó Baoill, Donall P. 1994 “Tense and aspect in Modern Irish”, in: R. Thieroff & J. Ballweg (eds.), 201–216.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

186

186

Lars Johanson

Oliveira, Fátima & Ana Lopes 1995 “Tense and aspect in Portuguese”, in: R. Thieroff (ed.), 95–115. Panzer, Baldur 1984 “Parallel internal development or interference? Post-positive article and possessive perfect in North Germanic and northern Russian”, in: P. Sture Ureland & Ian Clarkson (eds.), Scandinavian language contacts. Cambridge University Press: Cambridge, 111– 132. Pohlenz, Max 1959 Die Stoa. Geschichte einer geistigen Bewegung. (2nd edition.) Göttingen: Vandenhoeck & Ruprecht. Pollak, Wolfgang 1960 Studien zum ‘Verbalaspekt’ im Französischen. (Österreichische Akademie der Wissenschaften, Philosophisch-historische Klasse, Sitzungsberichte 233, 5.) Wien. Reichenbach, Hans 1947 Elements of symbolic logic. New York: The Free Press & London: Collier-Macmillan. Ruipérez, Martin Sanchez 1954 Estructura del sistema de aspectos y tiempos del verbo griego antiguo: análisis funcional sincrónico. (Theses et studia philologica Salmanticensia 7.) Salamanca: Colegio Trilingüe de la Universidad. Rundgren, Frithiof 1961 Das althebräische Verbum. Abriss der Aspektlehre. Stockholm: Almqvist & Wiksell. 1963 “Erneuerung des Verbalaspekts im Semitischen. Funktionell-diachronische Studien zur semitischen Verblehre”, Acta Universitatis Upsaliensis. Acta Societatis Linguisticae Upsaliensis. Nova series 1/3: 49–108. Sasse, Hans-Jürgen 1991a “Aspect and Aktionsart: a reconciliation”, in: C. Vetters & W. Vandeweghe (eds.), Perspectives on aspect and aktionsart. (Belgian Journal of Linguistics 6.) Bruxelles: Éditions de l’Université, 31–45. 1991b “Aspekttheorie”, in: Hans-Jürgen Sasse (ed.), Aspektsysteme. Arbeitspapier 14 (Neue Folge). Köln: Institut für Sprachwissenschaft, 1–35. Saussure, Ferdinand de 1916 Cours de linguistique générale. Edited by Charles Bally & Albert Sechehaye. Lausanne, Paris: Payot. Schwenter, Scott A. 1994 “The grammaticalization of an anterior in progress: evidence from a peninsular Spanish dialect”, Studies in Language 18: 71–111. Seiler, Hansjakob 1952 L’aspect et le temps dans le verbe néo-grec. (Collection de l’Institut d’Études Byzantines et Néo-helléniques de l’Université de Paris 14.) Paris: Les Belles Lettres. 1969 Zur Problematik des Verbalaspekts. Arbeitspapier 7. Köln: Institut für Sprachwissenschaft. 1993 “L’aspect verbal en perspective dimensionelle: Typologie, conceptualisation, et le grec”, Syntactica (Saint-Etienne) 5: 19–34. Serebrennikov, Boris A. 1960 Kategorija vremeni i vida v finno-ugorskix jazykax permskoj i volžskoj grupp. Moskva: Izdatel’stvo Akademii nauk SSSR. Squartini, Mario & Pier Marco Bertinetto this volume “The Simple and Compound Past in Romance languages”. Šewc, Hinc 1968 Gramatika hornjoserbskeje rˇecˇ e 1. Fonematika i morfologija. Budyšin: Domowina.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

187

Viewpoint operators in European languages

187

Smith, Carlota 1991 The parameter of aspect. Dordrecht: Kluwer. Stankov, Valentin 1976 Konkurencija na glagolnite vidove v b˘algarskija knižoven ezik. Sofija: Izdatelstvo na B˘algarskata akademija na naukite. 1980 Glagolnijat vid v b˘algarskija knižoven ezik. Sofija: Nauka i izkustvo. Sten, Holger 1952 Les temps du verbe fini (indicatif) en français moderne. København: Det Kongelige Danske Videnskabernes Selskab. Thelin, Nils B. 1978 Towards a theory of aspect, tense and actionality in Slavic. (Acta Universitatis Upsaliensis, Studia Slavica Upsaliensia 18.) Uppsala. Thieroff, Rolf (ed.) 1995 Tense systems in European languages 2. (Linguistische Arbeiten 338.) Tübingen: Niemeyer. Thieroff, Rolf & Joachim Ballweg (eds.) 1994 Tense systems in European languages. (Linguistische Arbeiten 308.) Tübingen: Niemeyer. Tommola, Hannu 1984 “K kategorii prošedšego vremeni russkogo glagola”, Studia Slavica Finlandensia 1: 134–164. 1994 “Zum Tempus im Finnischen”, in: R. Thieroff & J. Ballweg (eds.), 219–229. this volume a “Progressive aspect in Baltic-Finnic”. this volume b “On the perfect in East and West Slavic”. Trubinskij, Valentin I. 1983 “Rezul’tativ, passiv i perfekt v nekotoryx russkix govorax”, in: V. P. Nedjalkov (ed.), 216–226. Vendler, Zeno 1967 “Verbs and times”, in: Linguistics in philosophy. Ithaca, New York: Cornell University Press, 97–121. Vincent, Nigel 1988 “Italian”, in: Martin Harris & Nigel Vincent (eds.), The Romance languages. London & Sydney: Croom Helm, 279–313. Vogt, Hans 1971 Grammaire de la langue géorgienne. (Instituttet for Sammenlignende Kulturforskning, B, 57) Oslo: Universitetsforlaget. Weinrich, Harald 1964 Tempus. Besprochene und erzählte Welt. Stuttgart: Kohlhammer

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

188

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

189

Pier Marco Bertinetto and Denis Delfitto

Aspect vs. Actionality: Why they should be kept apart

1. Introduction1 1.1. Terminological matters The aim of this paper is to show that the notions of “aspect” and “Aktionsart” (or “actionality” as we prefer to call it, following a recent tradition)2 must be carefully distinguished. As is well known to all scholars working in the field, there is widespread disagreement, not to say confusion, on this matter. For some, the above statement will be self-evident; for others, it will barely make sense. This is also reflected in terminology. Those who adhere to the latter view often refer to the different verb classes as “aspectual” classes, whereas those holding our view introduce some sort of terminological distinction (“Aktionsart classes” “actional classes” or the like). The picture is further complicated by the tradition existing within the domain of Slavic studies, where scholars often use the term “aspect” to designate the basic lexical opposition to be observed in these languages (traditionally named “Perfective/Imperfective” Russian soveršennyj/nesoversennyj), and the term “Aktionsart” to refer to further semantic groupings of verbal predicates (like: static, resultative, inchoative etc.). Obviously, we do not deny that further semantic classes exist in the Slavic languages, besides what we have called “basic opposition”; and it is quite natural that these classes belong, at least in part, to what we take to be the actional domain in the sense of Vendler (1957). However, we believe that even the basic opposition “Perfective/Imperfective” belongs to the domain of actionality rather than aspect proper, although it is intricately interrelated with the latter.3 We shall return to this in Section 4. For the moment, let us simply make it clear that we do not hold the view that Slavic languages provide the ideal prototype of an aspectual system: on the contrary, we maintain that Slavic languages represent a quite peculiar case, rarely manifested outside that language family. Before we go any further, we feel it necessary to present a short sketch of the conceptual framework that we assume as background. Agreements or disagreements on what we are going to propose may be better directed, once purely terminological misunderstandings are put aside. To provide a fairly well-known reference, our approach to the domain of tense and aspect, as also developed in Bertinetto (1986), is very close to that advocated by Comrie (1976; 1985). We posit three main subdomains:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

190

190

Pier Marco Bertinetto and Denis Delfitto

(i) Temporal reference: the localization of the event with respect to the speech point. This allows us to distinguish between past, present and future temporal reference. This is normally, but not invariably, achieved by means of grammatical devices, i.e. tenses; however, the notion “tense” does not coincide with that of temporal reference, as we shall soon verify. (ii) Aspect: the specific perspective adopted by the speaker/writer. Typically, the event may be considered from a “global” or a “partial” point of view. This is the basis for the fundamental distinction between “perfective” and “imperfective” aspects. Although there may exist intermediate cases, these two compartments are normally quite distinct from each other. To the first belong, in particular, the “aoristic” (cf. the Simple Past in Romance) and the “perfect” aspect; to the latter, the “progressive” and the “habitual”. This does not exhaust the inventory of the major aspectual distinctions, but it will suffice for our present purposes. Again, aspect is normally, but not invariably, expressed by means of grammatical devices (i.e., tenses or specific periphrases); see below for further qualifications. (iii) Actionality: the type of event, specified according to a limited number of relevant properties. To make things simpler, here we shall refer exclusively to the wellknown Vendlerian classes, although various refinements have been put forth in the literature. The basic oppositions (according to a number of authors) are those between “punctual” vs. “durative” events, “telic” vs. “atelic” and “static” vs. “dynamic”.4 This allows us to isolate the following four classes: “states” (henceforth referred to as “statives”), “activities” “accomplishments” and “achievements”. While the notions of temporal reference and aspect (although ultimately of a semantic nature) are primarily anchored to the inflectional specifications available in each language, actionality is essentially rooted in the lexicon. Thus, the last category normally lacks an overt morphological marking, but it may have one. This is typically the case with the verbal lexicon of Slavic languages, where the manifestation of the basic opposition “Perfective/Imperfective” (fundamentally ascribable, as we will show, to the category of actionality, rather than aspect proper) belongs, in most cases, to the domain of derivative morphology. On the basis of the framework sketched in (i–iii), it should become clear that, as a matter of principle, tense should not be confused with temporal reference. This is a very important distinction to be made. Indeed, all tenses (without exception) have both temporal and aspectual properties, differently manifested according to the language considered, just as (in inflectional languages) categories such as case, gender, number and declensional class may or may not fuse together in a single marker in adjectival forms.5 Thus, for instance, we may say that the Imperfect of Romance languages has past temporal reference (at least in its most typical uses) and imperfective aspect (in most of its uses); or that the German Past has past temporal reference and is neutral on aspectual grounds. Note that neutralization of a given feature does not mean that such feature is absent; indeed, the German Past may receive either a per-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

191

Aspect vs. Actionality

191

fective or an imperfective reading, depending on the context. Moreover, it is sometimes the case (e.g. in classical Arabic) that tenses have a primarily aspectual value, with temporal reference being recovered by implication through contextual information (Cohen 1989). It would be utterly meaningless, in such cases, to take “tense” as the equivalent of “temporal reference”. Indeed, the notion “tense” refers to specific morphological coalescences, which have developed in each particular language. This explains, by the way, why languages may fundamentally express the same sort of temporal and aspectual values, despite the dramatic differences in the number of morphological distinctions available in each tense system (compare a “rich” system such as that of Bulgarian with a “poor” system such as that of Classical Arabic). To close on this, we may note that the identification of tense with temporal reference would have the absurd consequence that the name of one category (the one here called “temporal reference”) would also be used to designate another independent category (i.e. aspect). The choice of “tense” as a neuter term prevents this undesirable result.6 It is important to understand that our stressing the need, on theoretical grounds, to separate actional and aspectual categories does not mean that these categories may be separated in all circumstances. Indeed, there are several instances in tense-aspect systems where one observes an intricate interplay of notions belonging to different conceptual domains. To give an obvious example, take the so-called “imperfective paradox” arising from the interaction of the actional feature [telic] and the aspectual opposition [perfective]. While he read a book brings about the basic telicity of the predicate, he was reading a book obliterates this value, for there is no necessary implication of the future attainment of the goal. Although the opinions of the various scholars vary as to the interpretation of the latter sentence, we believe it legitimate to assert that the progressive aspect detelicizes the predicate (cf. Parsons 1989 and Bertinetto 1997, ch. 4, for a proposal in this sense). Clearly, if actional and aspectual categories were orthogonal to one another, we should not find similar interactions. However, we do not believe that linguistic categories need to act independently on all occasions in order to be treated as autonomous entities. According to the functional perspective that we are assuming, it is enough that they do so in a number of relevant cases. Indeed, on typological grounds, it often happens that two features, which are expressed independently of each other in language x, are fused together (or formally neutralized) in language y. If one had to judge the situation from the viewpoint of the latter type of language, one might be inclined to say that we have to make do with a single category. But a little cross-linguistic comparison would tell us that the given features, although neutralized in certain instances (or in certain languages), exist as autonomous conceptual entities. Bertinetto (1994) claims that situations of this sort (i.e., interaction and neutralization of categories) occur very frequently in the tense-aspect systems of natural languages, indeed not surprisingly, given the dramatic typological variability to be observed in this domain. In fact, al-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

192

192

Pier Marco Bertinetto and Denis Delfitto

though there is a tendency to express the relevant semantic oppositions, it is difficult (if not impossible) to find a single language which manifests them all in its morphology. And even when such categories have independent existence in a given language, they often present some degree of interaction in specific contexts.7 One reason for the frequent difficulty of communication between scholars active in this field lies in the fact that some of the terms most commonly used may mean quite different things, sometimes even in the writings of one and the same scholar. The most obvious example (already alluded to in fn. 3) is provided by the pair “perfective/imperfective”. This is of course used to designate the basic lexical opposition, typical of Slavic languages, that opposes, e.g., predicates such as cˇ itat’/proˇcitat’ ‘read ipf./pf.’, pisat’/napisat’ ‘write ipf./pf.’ etc. However, the same terms are also employed with respect to the basic aspectual opposition that is to be observed, e.g., in Romance languages between the Simple Past and the Imperfect. As claimed above, the latter use concerns the aspectual domain proper, while the former concerns by and large the actional domain. But note that this statement, as it stands, oversimplifies the data; in Section 4 we shall qualify it in a way that, we hope, will not sound outrageous to the followers of the Slavic tradition. At this point, we would like to ask the reader to delay her/his judgement until actual linguistic data have been considered. That is indeed the real test for any theory, above and beyond terminological disputes. We believe it would be pointless, at the present stage of development of these studies, to try to persuade one group or the other (i.e., Slavicists and non-Slavicists) to renounce their own terminological habits. Accordingly, in this contribution we shall pursue a different line. We would like to propose two alternative pairs of terms to be used in the actional and the aspectual domain respectively, replacing the glorious but quite opaque terms perfective/imperfective. As to these, we suggest that they be taken as cover terms both within the aspectual domain proper and for the basic lexical oppositions to be observed in the Slavic languages. We believe that this move should be welcome to anybody, for there is evidently an intuitive sense in which both a “perfective” tense in a language like, for example, English and a Slavic “Perfective” verb converge towards a similar result. In fact, this ought to be so, once we realize that Slavic “Perfective” verbs correspond, by and large, to telic verbs in the actional classification. Nevertheless, it would be wrong to conflate the notions “perfective” and “telic” because: (a) Events may be viewed from a perfective or an imperfective point of view, regardless of whether they are telic or atelic, as shown by the following French examples:

/= perfective, atelic/ il a dormi ‘he slept (Compound Past)’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

193

Aspect vs. Actionality

193

/= imperfective, atelic/ il dormait ‘he slept (Imperfect)’ /= perfective, telic/ il a écrit sa thèse ‘he wrote (Compound Past) his thesis’ /= imperfective, telic/ il écrivait sa thèse ‘he wrote (Imperfect) his thesis’ (b) Telic predicates fulfill their inherent character only in perfective situations. Thus, although écrire sa thèse is, from the standpoint of its intrinsic lexical meaning, a telic event, il écrivait sa thèse depicts, strictly speaking, a detelicized situation, i.e., a situation in which the inherently telic predicate loses its distinctive feature (remember once more the so-called “imperfective paradox”). In fact, in such a case it would be more appropriate to qualify the event as “lexically telic but contextually atelic”. For brevity, we shall henceforth simply use the term “detelicized” (which stands for “contextually detelicized”) with respect to the actional qualification of telic events viewed imperfectively. Here then is our proposal. The terms perfective/imperfective may retain much of their usefulness as cover terms, and indeed we consider it legitimate to give special status to the term “aspectology” as having comprehensive meaning, embracing both aspect proper and actionality. However, there are also good reasons to maintain that the terms perfective/imperfective should be avoided whenever they may cause misunderstandings. In order to avoid these undesirable consequences, which often obscure the intended meaning of the writer, we suggest adopting a neat terminological distinction. We shall employ the pair “terminative/nonterminative” when referring to the aspectual domain proper, and “bounded/unbounded” when referring to the lexical oppositions available in the Slavic languages. As it happens, this distinction will turn out to be very useful in Section 4, when we compare the structure of Slavic and non-Slavic languages. We are of course aware that this solution may hide its own traps. Both bounded/unbounded and terminative/nonterminative are not new terms, and may appear somehow biased to a number of readers (the second pair, in particular, has been used with respect to telic/atelic verbs by some scholars writing in German). However, we believe it will not be difficult to grasp our intention. It should be clear, for instance, that “terminative” does not mean “telic” for any event may be viewed as terminated, irrespective of whether or not it is directed towards an inherent goal (cf., the examples under (a) above). For us, terminative is nothing else but a handy way to designate a “global” aspectual perspective whereby the event is viewed in its entirety. Obviously, one may conceive of alternative pairs of terms, and we would not insist on our own if better ones were to be proposed. What really matters to us is that a higher degree of understanding is reached among the scholars of the field. A small amount of flexibility may be an acceptable price to pay to achieve this goal.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

194

194

Pier Marco Bertinetto and Denis Delfitto

Note, finally, that we do not intend to suggest that the pair terminative/nonterminative should be used on every occasion. In most cases, when describing an individual language (or a homogeneous group of languages), there is no reason to use these terms, for it should then be quite clear what we mean by perfective/imperfective. The difficulty arises when we compare tense-aspect systems as different as the Slavic on the one hand, and the Romance or Germanic on the other; or when we describe a rich system such as that of Bulgarian, which obviously demands subtler terminological distinctions. Apart from these obvious situations, it must be taken for granted that the effort towards conceptual clarity should not be disjoint from the requirement of terminological parsimony.

1.2. Design of the analysis What we need, when faced with a theoretically intricate situation, is to collect empirical data. Temporal adverbials notoriously provide us with a highly manageable test for assessing aspectual values. Indeed, they have frequently been exploited for this purpose. From our point of view, however, they render an even more valuable service, for they enable us to differentiate clearly between aspect and actionality. In Section 2 we offer a sketch of this classificatory procedure, which has major theoretical consequences. Of course, temporal adverbials have often been exploited in the definition of verbal classes. However, it is worth stressing that, to the best of our knowledge, the particular point of view adopted here (i.e., contrasting aspect and actionality) has not been exploited before, at least not in a comparably systematic way.8 To keep the discussion as short as possible, we shall examine the behaviour (i.e., the varying degrees of compatibility) of a selection of temporal adverbials with respect to a representative subset of actional and aspectual values. In particular, we shall consider the dichotomy [ telic] as a typical actional discriminator, and the dichotomy [ terminative] as an aspectual discriminator. Our aim is to show that, in a considerable number of cases, these two oppositions behave as two completely independent variables. Needless to say, [ telic] is not the only relevant feature for actional classification. However, it is crucial in natural languages, for it discriminates two major classes: “achievements  accomplishments" vs. “activities  states”. This will suffice for our purposes, especially considering that it is precisely this opposition that is mostly focussed upon in the Slavic languages by the contrast bounded/unbounded. For the moment, however, we shall restrict the discussion to English and some Romance languages (namely Italian and Spanish), delaying the scrutiny of Slavic languages until Section 4. In order to simplify the designation of the various types of adverbial, we shall make use of the following conventions. The expression ‘X Time’ stands for any

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

195

Aspect vs. Actionality

195

quantified (conventional) unit of time; e.g., two hours, nine and a half weeks, five years, and the like. The expressions ‘tx’ and ‘ty’ stand for different instants of time. For the sake of clarity, we shall group the adverbials into four types, although we shall point to specific differences within each type. Note that the following classification is simply based on the morphological shape that these adverbials take in languages such as English or Italian; still, it is interesting to observe that, over and above their formal appearance, they tend by and large to maintain the same semantic behaviour cross-linguistically: I II III IV

“until tx”, “from tx to ty”, “since tx” “in X Time”, “for X Time”, It. “da X Tempo” already, still gradually, little by little

2. Durative temporal adverbials, and their aspectual and actional values For convenience, throughout this section the reader is invited to keep an eye on Table 1 of Section 3.1, which sums up the observations developed here. As explained above, we restrict our attention to the following oppositions: [ terminative] as an illustration of aspectual distinctions, [ telic] as an instance of actional distinctions. It must be borne in mind that the notion “nonterminative” should specifically be interpreted here in the sense of the progressive aspect, which may be regarded as the prototypical nonterminative value. In fact, for reasons that will soon become clear (cf. Section 2.1.1), the habitual value of the nonterminative aspect is less typical, in as much as it also shares some properties of the terminative aspect. As to actionality, although we disregard here the other salient features within this category, it goes without saying that the opposition [ durative] plays a major role too. Indeed, all the adverbials considered in this paper demand [durative] verbs.

2.1. Type I adverbials 2.1.1. Until tx Consider first the durative adverbial type “until tx” (It. “fino a tx”, Sp. “hasta tx”). We are going to show that this adverbial has the following inclinations: it is actionally atelic, and aspectually terminative. In fact, it is clearly incompatible with telic verbs. When accomplishments are used in conjunction with it, they lose their telic character, unless telicity is inseparable from the very nature of the verb (or, more properly, of the VP), as with exhaust, eat up, solve a problem etc.; in which case, the resulting sentence is ungrammatical (but cf. fn. 31). As to achievements, they are obviously

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

196

196

Pier Marco Bertinetto and Denis Delfitto

incompatible with “until tx” due to their basically non-durative nature; however, to make things simpler, in this paper we shall only consider accomplishment verbs as illustrations of [telic] predicates. As to the aspectual point of view, the adverbial expression examined here allows only for terminativity. To prove this, consider the following examples (cf. below for further qualifications concerning sentences (b) and (d).9

(1) a.

/= atelic, terminative/ Mary danced until midnight. b. /= atelic, nonterminative/ (??)Mary was dancing until midnight. c. /= detelicized, terminative/ Mary painted the wall until midnight. d. /= detelicized, nonterminative/ (??)Mary was painting the wall until midnight.

The question marks accompanying sentences (1b) and (1d) are within parentheses, for the English progressive allows for a “prospective” or “future time reference” reading (to be interpreted here as future-in-the-past), in which case these sentences would be grammatical. However, and not surprisingly given the inclinations of the adverbials employed, this would turn the aspectual value into terminative. Apart from this idiosyncratic detail, connected with the specific properties of the English progressive,10 the same actional and aspectual observations made above hold for Italian (and Romance languages in general), with the additional qualification that the Imperfect might be admitted in these contexts with a habitual meaning. This might seem to contradict the basically terminative nature of this adverbial. In fact, it does not do so, because one typical feature of the habitual aspect is that it admits terminative adverbials as long as they simply modify each single occurrence of the event, rather than the whole event. Indeed, the core of the habitual aspect lies in the indeterminacy of the total number of occurrences, rather than in the indeterminate duration of each occurrence. This is enough to guarantee its basic interminativity. Thus, one can find Italian utterances such as:

(2) a.

Italian /= atelic, habitual/ Ogni sabato sera, Maria ballava fino a mezzanotte. every Saturday evening Mary dance:IMPF until midnight ‘Every Saturday evening, Mary used to dance until midnight.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

197

Aspect vs. Actionality

b.

197

/= detelicized, habitual/ Ogni sabato sera, Maria puliva il giardino fino a every Saturday evening, Mary clean:IMPF the garden until mezzanotte. midnight ‘Every Saturday evening, Mary used to clean the garden until midnight.’

where it is clear that the number of occurrences constituting the whole event is undetermined, for we do not know how many Saturdays are involved. Consequently, the whole event cannot be viewed terminatively, despite the terminative characterization of the single occurrences. We can make sense of this apparent paradox by saying that a habitual situation consists of a nonterminative macroevent composed of a series of terminative microevents. The adverbial “until tx” in (2) refers of course to the single occurrences, which are thus perfectly compatible with its aspectual character. These observations extend to other adverbials, as we shall see. The basically terminative nature of this adverbial is proved by the fact that in Italian the progressive periphrasis is totally excluded in these contexts. Spanish, however, is an interesting case in this respect, for in this language we may find the Simple Past with the progressive periphrasis, as in: (3)

Spanish a.

b.

/= atelic, terminative/ María estuvo bailando hasta la media noche. Mary was:SP dancing until the middle night ‘Mary kept dancing until midnight.’ /= detelicized, terminative/ María estuvo pintando la pared hasta la media noche. Mary was:SP painting the wall until the middle night ‘Mary kept painting the wall until midnight.’

This would not be possible with the Imperfect (except, as already noted for Italian, where it has habitual meaning). This is an important feature of Spanish (and IberoRomance languages in general), to which we shall return in Section 3.2. 2.1.2. From tx to ty Next, consider the adverbial type “from tx to ty” (It. “da tx a ty” Sp. “desde tx hasta ty”). Here, Italian and Spanish essentially behave as before, while English is more flexible, and tolerates to some extent nonterminative sentences, which sound acceptable to many speakers (as suggested by the diacritic): (4)

%Mary was dancing from 10 p.m. to midnight.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

198

198

Pier Marco Bertinetto and Denis Delfitto

The insertion of further temporal specifications enhances the acceptability of these sentences: (5)

Yesterday, during the maths class, i.e. from 9 to 10, Mary and John were playing cards; in fact, they went on playing even afterwards.

Note here that the relationship between adverbial and verb is only indirect, owing to the intermediation of a temporal clause, which attenuates the impact of the competing features. Thus, it is no wonder that in such contexts the Romance Imperfect, and even to some extent the progressive periphrasis, is not excluded: (6)

Italian Ieri, dalle 2 alle 3, quando tu credevi che stesse yesterday from 2 to 3 when you thought that was:SUBJ studiando, in realtà Maria giocava / stava giocando a studying in fact Mary play:IMPF / was:IMPF playing at tennis. tennis ‘Yesterday, from 2 to 3 o’clock, when you thought that she was studying, Mary was in fact playing tennis.’

We may thus state that this adverbial, besides being atelic from the actional point of view, is tendentially terminative in Italian and Spanish, although not to the utmost degree. This is true even more in English, perhaps because of the ambiguous aspectual nature of the Simple Past (see Section 3.2 for further comments on this point). Unsurprisingly, Spanish admits, as in the previous case, the Simple Past in the progressive periphrasis. 2.1.3. Since tx Let us now consider the adverbial type “since tx” (It. “da tx” Sp. “desde tx”). In the Romance languages, this adverbial combines with both telic and atelic verbs, provided they are used nonterminatively, as we gather from: (7) a.

/= atelic, terminative/ *Maria ballò da mezzogiorno. Mary dance:SP since noon b. /= atelic, nonterminative/ Maria ballava da mezzogiorno (quando (when) Mary dance:IMPF since noon ‘Mary was dancing since noon (when )’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

)

199

199

Aspect vs. Actionality

c.

/= telic, terminative/ *Maria dipinse la parete da mezzogiorno. Mary paint:SP the wall since noon d. /= detelicized, nonterminative/ Maria dipingeva la parete da mezzogiorno (quando (when) Mary paint:IMPF the wall since noon ‘Mary was painting the wall since noon (when )’

)

English is slightly different because in (b) and (d) one prefers the Past Perfect Progressive (cf. Mary had been dancing/painting the wall since noon). However, doing so amounts to using an essentially nonterminative device, for the conclusion of the event is not necessarily envisaged. Thus, the ultimate result does not change. Spanish behaves exactly like Italian (cf. María bailaba/*bailó/pintaba la pared/*pintó la pared desde el mediodía). However, it must be remarked that in this language the progressive periphrasis with the Simple Past may be allowed here, as in the preceding cases (cf. María estuvo bailando desde el mediodía), provided there is a contextual presupposition which fixes a temporal limit, such as hasta las ocho “until 8 o’clock”. But in the latter case we would in practice obtain the adverbial type “from tx to ty” which we examined in Section 2.1.2. This shows that “desde tx” has indeed a strong nonterminative value.

2.2. Type II adverbials 2.2.1. In X Time The second series of temporal adverbials that we will examine is characterized by the presence of quantified (conventional) units of time. Let us begin with “in X Time”; e.g., in one hour, in two weeks, in six months. This demands telic verbs, and is associated with the terminative aspect only, as can be seen in:

(8) a.

/= atelic, terminative/ *Mary danced in two hours. b. /= atelic, nonterminative/ *Mary was dancing in two hours. c. /= telic, terminative/ Mary painted the wall in two hours. d. /= telic, nonterminative/ *Mary was painting the wall in two hours.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

200

200

Pier Marco Bertinetto and Denis Delfitto

2.2.2. For X Time Next, consider the adverbial type “for X Time” (It. “per X Tempo”, Sp. “durante X Tiempo”): (9) a.

/= atelic, terminative/ Mary danced for two hours. b. /= atelic, nonterminative/ % Mary was dancing for two hours. c. /= detelicized, terminative/ Mary painted the wall for two hours. d. /= detelicized, nonterminative/ %Mary was painting the wall for two hours.

From the actional point of view, this is clearly a detelicizing adverbial. Indeed, (9c) does not mean that the event is carried through to the end. However, the situation regarding aspectual values is slightly complicated in English (as to Romance languages, see below). Although this adverbial type clearly prefers terminative contexts, sentences (9b) and (9d) receive contrasting judgements by the speakers, as shown by the diacritic used.11 They appear rather clumsy, at least in isolation, to some scholars (e.g., Mittwoch 1988, Hatav 1989), but acceptable to others (e.g., Vlach 1981). It is useful to compare these sentences with a truly nonterminative situation, such as: (10)

When I came in, Mary was dancing/painting the wall.

The difference in this sentence compared to (9a) and (9c) is as follows. Sentence (10), where the temporal clause provides the evaluation time, suggests an indeterminate situation, as far as the continuation of the event beyond the relevant moment is concerned. Indeed, for all we know, the event may have gone on beyond that moment, or it may have stopped right then. Sentences (9a) and (9c), on the other hand, suggest that the (relevant portion of the) event is entirely confined within the interval considered; it may of course be resumed later on, but that should explicitly be stated in a further sentence. Thus, sentences (9a) and (9c) combine an atelic and (respectively) detelicized situation with a terminative view. As already noted with (5–6), added temporal specifications seem to improve the situation when the nonterminative aspect is employed (cf. 11–12), although this is not always the case (cf. 13): (11)

%While I was sleeping, John was running for an hour.

(12)

%Yesterday I was told that, while we were in class, John was running for an hour.

(13)

*During the class, Mary was playing cards with John for 10 minutes.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

201

Aspect vs. Actionality

201

Sentence (13) is ungrammatical, due to the fact that the “for X Time” adverbial indicates a short interval included into another relatively short one, so that we are inescapably bound to view the situation terminatively (i.e., by considering the terminal point of the event), and this is clearly incompatible with the Past Progressive. Sentences (11–12), acceptable for some speakers, might be taken as counterexamples to this claim. Note however that when we add a punctual temporal clause, as in (14b), the incompatibility with the adverbial becomes quite clear, although the same sentence without this adverbial type sounds perfectly natural (cf. 14a). Clearly, in these cases the punctual temporal clause helps us to focus on a particular moment of the event, which is thus viewed nonterminatively. Note further that the ungrammaticality of (14b) is not simply due to a possible conflict between the duration indicated by the “for X Time" expression and the punctual temporal clause, since (14c) sounds natural. However, and not surprisingly, due to the preceding punctual temporal clause, the Simple Past of the last sentence takes an ingressive meaning (a subspecies of the terminative aspect), which is alien to the Past Progressive:

(14) a. When I came in, John was running. b. *When I came in, John was running for an hour. c. When I came in, John ran for an hour.

Thus, (14c) enhances the conclusion formulated above, according to which “for X Time” adverbials refer to a basically terminative situation. The acceptability, at least for some speakers, of (9b) and (9d), as well as (11–12), is probably due to the ambiguous aspectual nature of the English Simple Past (cf. Section 3.2 for further comments). Indeed, when the context forces a nonterminative interpretation, as in (14a– b), the presence of this adverbial is unanimously rejected. Now consider Romance languages. In Italian the Imperfect of the progressive periphrasis is quite inappropriate with the adverbial “per X Tempo”. The same applies to the bare Imperfect, unless it is understood in the habitual meaning. Spanish behaves like Italian as far as the Imperfect is concerned, but unlike Italian it may exploit the Simple Past in the progressive periphrasis, which sounds perfectly appropriate, due to its terminative character; cf. María estuvo bailando durante dos horas ‘Mary was dancing for two hours’ (lit.: ‘Mary was:SP dancing.’). This is the same sort of situation we noted above for “until tx” and “from tx to ty”; and indeed this does not come as a surprise, because the aspectual and actional properties of all these types of adverbial are exactly the same. To sum up, the actional and aspectual inclinations of “for X Time” adverbials are as follows: atelic, terminative.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

202

202

Pier Marco Bertinetto and Denis Delfitto

2.2.3. It. “da X Tempo”, Sp. “desde X Tiempo”, Fr. “depuis X temps” Let us now analyse the adverbial It. “da X Tempo” (Sp. “desde X Tiempo”, Fr. “depuis X Temps”). This type is quite common in Romance languages, much less so in English. The only possible equivalent is “for X Time” apparently identical to the adverbial examined in Section 2.2.2. But, as we shall see, there are differences. The adverbial “da X Tempo” is compatible with both telic and atelic verbs, and with both terminative and nonterminative tenses, although with different meanings, and (saliently) with the exclusion of one combination of these features, namely *[ telic, terminative]. It should be noted that this adverbial is not found in Romance with the Simple Past, whereas it may easily be with the Compound Past (morphologically equivalent to the English Present Perfect, but not strictly identical from the semantic point of view). Indeed, Bertinetto (1986) regards this adverbial as a powerful diagnostic tool for discriminating, within the terminative aspect, between perfectal tenses and purely aoristic ones. Consider the following Italian examples, keeping in mind that in these contexts the Compound Past takes a strictly perfectal value: (15) a.

b.

c.

d.

Italian /= atelic, terminative/ *Maria ha ballato da due ore. Mary has:PRS danced since two hours /= atelic, nonterminative/ Maria ballava/ stava ballando da Mary dance:IMPF was:IMPF dancing since ‘M. had been dancing for two hours.’ /= telic, terminative/ Maria ha dipinto la parete da due Mary has:PRS painted the wall since two ‘M. finished painting the wall two hours ago.’ /= detelicized, nonterminative/ Maria dipingeva/ stava dipingendo la Mary paint:IMPF/ was:IMPF painting the ore. hours ‘Mary had been painting the wall for two hours.’

due ore. two hours

ore. hours

parete da due wall since two

The actional value in (15d) may properly be defined as “aspectually detelicized”, because the adverbial is in itself compatible with telic verbs, as shown by (15c). Indeed, the detelicization is entirely due here to the nonterminative aspect (remember the so-called ‘imperfective paradox’), rather than to the adverbial. Compare, on the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

203

Aspect vs. Actionality

203

other hand, (1c), where the verb is detelicized independently of the aspectual value (cf. also the comment in fn. 9). The different meaning of the adverbial, depending on the aspectual value, is quite apparent. With terminative tenses, it indicates the interval elapsed between the end of the event and the evaluation time. With nonterminative tenses (specifically, with the progressive aspect), it measures the time elapsed between the beginning of the event and the evaluation time (nothing is implied, of course, as to the conclusion of the event). This explains the exclusion of (15a). This sentence looks weak because atelic predicates are not suitable for indicating that the event referred to is directed towards a specific goal; thus, it is not easy to determine from where one should start measuring the elapsed time (unless the final moment, which obviously exists, is explicitly given, as with the periphrasis ha finito di ballare ‘s/he has finished dancing’, which is based on a telic verb). Sentence (15b), on the other hand, is perfectly natural, because durative events obviously have an initial, psychologically salient, moment. As noted above, English makes much more limited use of this type of adverbial in contexts corresponding to those exemplified in (15). It is only allowed with stative verbs and Perfect tenses. However, if these appear in the progressive form, then the actional restriction is relaxed, i.e., non-stative verbs may appear: /= stative/ Mary has/had known John for two months. b. /= stative/ (*)Mary knew John for two months. c. /= stative; detelicized/ ??Mary has/had painted the wall for two hours. d. /= stative; detelicized/ Mary has/had been painting the wall for two hours.

(16) a.

2.3. Type III adverbials 2.3.1. Already A similar case is offered by the adverb already (It. già, Sp. ya), which belongs to the next type of temporal adverbials. This combines with all possible actional and aspectual types, but again with different meanings:12 (17) a.

b.

/= atelic, terminative/ Mary already danced the polka. [i.e., some other time in the past] /= atelic, nonterminative/ Mary was already dancing the polka, when I came.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

204

204

Pier Marco Bertinetto and Denis Delfitto

c.

d.

/= telic, terminative/ Mary already painted this wall. [i.e., some other time in the past] /= detelicized, nonterminative/ Mary was already painting this wall, when I came.

As noted regarding (15d), the detelicization of the predicate occurring in (17d) does not depend on the direct contribution of the adverb, but is a mere product of the progressive aspect (remember again the ‘imperfective paradox’), because the adverb is not in itself incompatible with telic verbs, witness (17c). Note, furthermore, that although already may be used in conjunction with both aspectual values, the meaning it acquires in terminative sentences (as indicated in the comments attached to (17a) and (17c)) must be regarded as a derived one, available only in particular contexts. This is shown by sentences like the following, where the terminative tense sounds inappropriate: (18)

??During the maths class, Mary already played cards with John.

2.3.2. Still The use of the adverb still (It. ancora, Sp. todavía) is more limited. It does not combine with terminative tenses, although it is indifferent to the actional value considered here:13 (19) a.

/= atelic, terminative/ *Mary still danced the polka, before leaving. [i.e. again] b. /= atelic, nonterminative/ Mary was still dancing the polka, when I came. c. /= telic, terminative/ *Mary still painted this wall, before leaving. [i.e. again] d. /= detelicized, nonterminative/ Mary was still painting this wall when I came.

Actually, Italian and Spanish are more liberal in this respect, because they allow for a resemanticization of the adverbs ancora and todavía, which in conjunction with terminative tenses mean “again, another time”, as shown by the following examples (to be compared with 19a): (20) a.

Italian Maria balló ancora la polka, prima di andarsene. Mary dance:SP still the polka before of go:INF

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

205

Aspect vs. Actionality

b.

205

Spanish María bailó todavía la polka, antes de irse. Mary dance:SP still the polka before of go:INF ‘Mary danced the polka again, before leaving.’

2.4. Type IV adverbials The next type of adverbials, which may be called adverbials of ‘graduality’ (gradually, little by little and the like), is also relevant, as these treat actional and aspectual properties as two independent parameters. Namely, they admit both major aspectual values, but are restricted to telic predicates.14 Thus, they are unaffected by aspectual values, just as still and already are unaffected by the feature [ telic]. Consider: (21) a.

/= atelic, terminative/ *Mary danced gradually. b. /= atelic, nonterminative/ *Mary was dancing gradually. c. /= telic, terminative/ Mary painted the wall gradually. d. /= telic, nonterminative/ Mary was painting the wall gradually.

Besides presenting adverbials of graduality corresponding to the English ones (gradualmente, a poco a poco), Italian exhibits a peculiar adverbial (man mano) with more constrained properties. This is found only with a combination of telic verbs and nonterminative (specifically, progressive) aspect:

(22) a.

/= telic, terminative/ ??Maria dipinse man mano la parete. Mary paint:SP little by little the wall ‘Mary painted the wall little by little.’ b. /= telic, nonterminative/ Mentre Ugo puliva il giardino, Maria dipingeva while Ugo clean:IMPF the garden Mary paint:IMPF man mano la parete. little by little the wall ‘While Ugo was cleaning the garden, Mary was little by little painting the wall.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

206

206

Pier Marco Bertinetto and Denis Delfitto

3. Aspectual and actional properties as independent entities 3.1. Synopsis It may be useful, at this point, to recapitulate what we have observed in Section 2. For simplicity’s sake, we shall tabulate only data referring to English, with the single exception of the adverbial type It. “da X Tempo” (which also stands for similar constructions in other languages, such as Sp. “desde X Tiempo” or Fr. “depuis X Temps”). Note that the diacritic ‘’ has two different meanings: either that the given adverbial type is truly indifferent to the specific value of the diacritic, or that the adverbial is compatible with both values, but with separate readings. In the latter case, we insert the diacritic within parentheses for clarity. Furthermore, we make use of the exclamation mark to suggest that the given adverbial type shows the tendency indicated to a high degree: Table 1. Actional and aspectual properties of selected English (and Romance) temporal adverbials. Actionality Type I “until tx” “from tx to ty” “since tx” Type II “in X Time” “for X Time” It. “da X Tempo” Type III already still Type IV adverbials of graduality

telic! telic! telic!

 telic!

Aspect

 terminative!  terminative terminative!

telic!  telic

 terminative!  terminative () terminative

 telic  telic

() terminative terminative

 telic!

 terminative

It should be noted that with It. “da X Tempo”, the combination *[ telic, terminative] is excluded.15 As to the detailed differences between English on the one side, and Italian and Spanish on the other, the reader is directed to the preceding discussion. To sum up briefly: in Italian and Spanish the adverbs corresponding to still are [()terminative]; in addition, all adverbials demanding the features configuration [ telic, terminative] accept in both languages the Imperfect with habitual meaning, while in Spanish they also tolerate the progressive periphrasis with the Simple Past.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

207

Aspect vs. Actionality

207

Table 1 clearly shows that aspectual and actional values are independent entities, as claimed at the outset, and contra the opinion defended in for instance Sasse (1991). Indeed, the various adverbials behave quite differently with respect to these two semantic categories. Consider for instance still. This is neutral to the [ telic] distinction, but is clearly selective with regard to aspectual values. Conversely, adverbials of graduality are very tolerant with aspectual values, but quite selective with actional ones. These are of course the most relevant cases, because they diverge in the sharpest way; but the varying behaviour of the remaining adverbials is quite revealing. Thus, “until tx”, “from tx to ty” and “for X Time”, which are [ telic!, terminative(!)], contrast neatly with “in X Time”, which is [telic!, terminative!], and so forth. We shall return to this in Section 4.2.

3.2. Progressive periphrasis and terminative aspect One topic needing clarification is a peculiarity of Spanish alluded to above. Like English, this language has preserved the possibility (once also exhibited by Italian) of combining the progressive periphrasis with both terminative and nonterminative tenses (see Section 3.3). Unlike English, however (but like any Romance language), Spanish is morphologically endowed with a typically nonterminative tense, called “Imperfect”. This provides the Spanish progressive periphrasis with additional expressive power, in contradistinction to both Italian (which accepts only nonterminative tenses with this periphrasis) and English (which lacks a markedly nonterminative tense, like the Romance Imperfect). As observed above, this has specific consequences for all adverbials of the type [ telic, terminative], such as “until tx”, “from tx to ty” and “for X Time”: (23) a. b. c. d. e.

English Italian Italian Spanish Spanish

%Mary was dancing for two hours. /= Imperfect/ *Maria stava ballando per due ore. /= Simple Past/ *Maria stette ballando per due ore. /= Imperfect/ *María estaba bailando durante dos horas. /= Simple Past/ María estuvo bailando durante dos horas.

Here, Italian yields ungrammatical sentences (cf. 23b–c), while English behaves rather ambiguously (cf. the discussion relating to examples 9–14 above). As shown by the diacritic in (23a), the progressive does not sound entirely appropriate to a number of speakers, owing to the purely terminative character of the adverbial. Spanish, on the other hand, makes a very neat distinction in these contexts (cf. 23d–e). The progressive periphrasis with the Imperfect is unacceptable (just as in Italian), whereas with the Simple Past it sounds absolutely appropriate.16 Clearly, Spanish is able to counteract the nonterminative orientation of the progressive periphrasis by means of an explicit morphological tool. This effect, on the other hand, does not

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

208

208

Pier Marco Bertinetto and Denis Delfitto

extend to the adverbial type “in X Time” (*María estuvo pintando la pared en dos horas; cf. 8d for translation). However, as we saw above, the latter adverbial demands a strictly telic situation; and this is not ensured by the progressive periphrasis, which inevitably determines a detelicization of the predicate. The reason why English encounters some difficulties in (23a) is presumably a direct consequence of the ambiguous aspectual characterization of the Simple Past. It should indeed be noted that this tense is regularly used in contexts where Romance languages would normally employ the Imperfect: (24) a. b. (25)

Italian /= Imperfect/ Maria era bruna e aveva gli occhi azzurri. /= Simple Past/ Mary was dark-haired and had blue eyes. Italian /= Imperfect/ Il cervo corse, corse, raggiunse la tribù dei cervi che vedendolo con un uomo sulle corna un po’ lo sfuggivano, un po’ gli s’avvicinavano curiosi. (I. Calvino) ‘The deer ran, ran, reached the deer troup which, seeing it with a man on its antlers, escaped (*were escaping) or approached (*were approaching) it in turn with eagerness.’

Obviously, in (24) the progressive could not be employed because of the stative character of the verb; however, it is worth noting that it is precisely in contexts such as those that Romance languages tend to make almost exclusively use of the Imperfect, reserving the Simple Past for highly marked stylistic registers.17 As to (25), the English progressive would sound inappropriate even though the verb is not stative. Thus, the English Simple Past turns out to be appropriate both for truly terminative contexts, where Romance languages would also normally employ the Simple Past,18 and for truly nonterminative contexts, where Romance languages employ the Imperfect (as in 24–25). This suggests an explanation for the divergent reactions of English speakers, some of whom tend to reject sentences such as (23a), presumably because they regard the periphrastic construction “was  VERB-ing” as genuinely nonterminative. Those who accept such sentences, on the other hand, seem to consider this construction as a sort of syncretic tool, combining the meanings of both Sp. “estaba  Gerund” and “estuvo  Gerund”. We believe that this phenomenon has still further consequences: indeed, attempts to provide a semantic treatment for the progressive have so far seriously underestimated this fact (Bertinetto and Delfitto, 1996). One point on which Spanish and English converge is the possibility of combining the progressive periphrasis with Perfect morphology (in our view, a subspecification

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

209

Aspect vs. Actionality

209

of terminative aspect). This possibility is again denied to Contemporary Italian, as shown by (26), since the restriction against terminative tenses with the progressive periphrasis has become very tight and general. By contrast, compare (27): (26) (27) a. b.

*Maria è stata ballando per due ore. Mary has been dancing for two hours. María ha estado bailando durante dos horas.

Actually, the sentences in (27a–b) are ambiguous. The most frequent interpretation is the ‘inclusive’ one, which fully exploits the aspectually ambivalent nature of the construction. On this interpretation, the terminativity of the tense employed matches the terminative nature of the adverbial, while the basically nonterminative character of the progressive form suggests that the event is not necessarily over at the relevant reference time (coinciding here with the speech time). However, another interpretation is also available, where the intended meaning is that the event has been going on for some time in the past (not necessarily a distant past, but one with no overlapping with the speech time). When the latter interpretation is chosen, (27b) may have the same reading as (23e), suggesting a substantially terminative view of the event, which is then regarded as concluded at the end of the given interval confined in the past (for further comments on this matter, cf. example (19) in section 3.2 of Bertinetto, this volume). Indeed, in such a case, the difference between (23e) and (27b) appears not to be aspectual, but rather temporal, since the latter sentence tends to point (albeit not necessarily) to a near past, while the former is more appropriate for distant events.19

4. The intertwining of aspectual and actional values in the Slavic languages 4.1. The structure of Slavic languages As observed in the introduction, Slavic languages constitute a decisive test for any aspectological theory. In the rest of this paper we shall address this topic, trying to show that the particular structure exhibited by the verbal lexicon of the Slavic languages has mostly to do with the category of actionality, although it is also connected, in a highly intricate way, with the category of aspect. As noted in the introduction, we may view any given tense-aspect system as the result of a peculiar mixture of three fundamental components: temporal reference, aspect and actionality. Let us now compare the structure of Romance languages and the structure of a typical Slavic language (such as Russian). As is well known, Romance

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

210

210

Pier Marco Bertinetto and Denis Delfitto

languages present explicit morphological marking of temporal and aspectual values (although not in all tenses).20 Slavic languages, for the most part, instead present explicit morphological marking of temporal values, and in addition mark overt oppositions between different but (usually) derivatively connected lexical entries. These tend to create bipolar contrasts, although some pairs lack one of the two poles, and in quite a few cases one (or both) of the two poles comprises more than a single member. These opposing poles are traditionally referred to as “Perfective/Imperfective”; but (as discussed in Section 1.1) in order to avoid confusion with the non-Slavicist aspectological tradition, we shall speak here of “bounded/unbounded”, as is sometimes done in the literature. Now, given the general structure of the two language families considered, it is quite clear that Romance languages tend to relegate actional values to the background, giving prominence to temporal reference and aspect, while Slavic languages privilege temporal reference and actionality over aspectual values, for the opposition [bounded] is very much reminiscent of the distinction [telic]. (For a similar approach to this matter, cf. Johanson, this volume.) This is a fundamental point, and it is very important to understand it clearly. By saying (as we did in Section 1.1) that actionality essentially belongs to the lexical domain, we are not claiming that the opposition [bounded] observed in Slavic languages does not belong to the realm of grammar. On the contrary, the specific character of the Slavic verbal system lies in the fact that these languages have found a way to overtly mark, in a fairly regular way, one of the most salient distinctions belonging to the actional domain. To the extent that this opposition is systematic and pervasive, we are obviously entitled to consider it part of the grammar of Slavic languages, just like the process which yields causative cognates from non-causative verbs is grammaticalized in quite a few languages. There is no principled reason why a word-formation process should not be considered part of the grammar of a given language, especially if it is systematically employed. To repeat Dahl’s (1985: 89) formulation, the so-called Slavic aspectual pairs may be regarded as “grammaticalized lexical categories”.21 Despite these dramatic differences, the expressive power of the two language families (Romance and Slavic) is not significantly different. They can express quite similar meanings, though with different morphological tools. As a matter of fact, the category relegated to the background can produce its effects in the appropriate contexts. This can be seen through selective reactivity to temporal adverbials. Thus, as shown in Section 2 above, English, Italian and Spanish behave in a clearly identifiable way in conjunction with specific actional values, although these do not usually receive overt marking. As we shall now show, Slavic languages do the same with respect to aspect proper (as defined in Section 1.1), which is not overtly marked, or (more precisely) is not normally marked; note, in fact, that the situation described above for Russian does not refer to all Slavic languages. As is well known, Bulgarian and Macedonian (and to a lesser extent other languages in the Slavic domain) differ

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

211

Aspect vs. Actionality

211

Table 2. English and Romance selected temporal adverbials, grouped according to their actional and aspectual properties. Telic

Terminative



 



(i) (ii) (iii) Ø

“in X Time”, It. “da X Time” (tm) “until tx”, “for X Time” “since tx”, It. “da X Tempo” (ntm), already (ntm), still

considerably from the Russian pattern, insofar as they delicately combine the characteristics of both the Romance and the Slavic types. Leaving further details aside, Bulgarian and Macedonian present: (i) A rich system of temporal reference distinctions (consider for example a tense such as the Pluperfect, which has disappeared in Russian); (ii) Explicit aspectual oppositions implementing the distinction terminative/nonterminative (cf. the contrast between the Imperfect on the one hand, and the Aorist and Perfect on the other); (iii) A highly developed system of lexical oppositions, contrasting bounded and unbounded verbs as is typical of Slavic languages in general. This gives Bulgarian an exceptionally high degree of expressive power; so high, in fact, that its possibilities are often exploited to obtain subtle modal meanings, rather than strictly temporal or aspectual ones (Lindstedt 1985, Guentcheva 1991).22 In the following section, we will compare the behaviour of Russian and Bulgarian with the behaviour of English and Romance, trying to argue that the analogies observed appear to support the principled distinction between aspect and actionality defended in the preceding sections.

4.2. Linguistic evidence Let us take another look at Table 1 in Section 3.1 above, which sums up the actional and aspectual values compatible with the adverbials analysed in Section 2. We can present the data in a slightly modified form, by grouping the adverbials compatible with a given combination of actional and aspectual values. This is done in Table 2, where three classes of adverbials are listed. For simplicity, not all adverbials are repeated here. Note further that, for some adverbials having a different meaning depending on the particular aspectual value exhibited by the verb, we split the adverbial in two different entries: one compatible with the terminative value (tm), the other compatible with the nonterminative value (ntm).23

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

212

212

Pier Marco Bertinetto and Denis Delfitto

The first observation to be made is that there is apparently no adverbial which selects the [telic, terminative] combination. This might be due to our particular selection of adverbials, but we believe that there is some deeper semantic reason lying behind this. If an adverbial has a strong telic inclination, it also necessarily has a terminative one.24 This is consistent with what we said in Section 1.1: although aspect and actionality are independent categories, they are not altogether orthogonal to one another. The sorts of semantic primes on which they are based are, ultimately, of a very similar nature. If this were not so, it would be difficult to understand why Romance languages (and English) have developed a morphological structure so dramatically different from that exhibited by Slavic languages. Bearing Table 2 in mind, let us now consider the behaviour of Russian and Bulgarian. We might expect Bulgarian to adhere to the actional and aspectual choices exhibited (with only minor differences) by English and Romance, making use of course of its rich morphological structure to explicitly mark the relevant actional meanings. As to Russian, we might reasonably expect this language to select ‘bounded’ lexemes with class (i) adverbials, those requiring a [telic, terminative] configuration, and ‘unbounded’ lexemes with class (iii) adverbials, those requiring a [ telic, terminative] configuration. The ultimate challenge lies in the treatment of the remaining configuration: [ telic, terminative], exhibited by class (ii) adverbials. Let us see what the actual linguistic data tell us. The data reported below were elicited by one of the authors from linguists who are native speakers of Russian and Bulgarian.25 Basically, the examples used were a subset of those illustrated in Section 2 above; the only major change was that instead of the verb corresponding to dance (Rus. tancevat’), we used the verb corresponding to “write a letter” (pisat’/napisat’ pis’mo).26 This was due to the restricted choice available with the first verb. It should also be remarked that whenever we use, in what follows, the label ’nonterminative’, this must be taken once more in the progressive sense, rather than in the habitual one. As we saw in Section 2, the latter meaning is often available, with the Romance Imperfect, in contexts where the progressive periphrasis (or the progressive reading of nonterminative tenses) is not acceptable. Indeed, this is true also for Bulgarian. The above expectations are confirmed (the reader is invited to verify this in Table 3 in Section 4.3 below). Both Russian and Bulgarian select bounded lexemes with class (i) adverbials (“za X Vremja”, “X Vremja nazad”, where “X Vremja” stands for ‘X Time’).27 Moreover, in Bulgarian the morphological choice that is consistently selected here is the one corresponding to the configuration [bound, terminative] (e.g., the Aorist of bounded verbs), while the choice corresponding to the configuration [ bound, terminative] (e.g., the Aorist of unbounded verbs) is systematically avoided, in a way that is reminiscent of English and Romance (cf. 8a and 15a).28 The only difference regarding the latter languages is that Bulgarian, besides presenting overt aspectual marking (Aorist and Perfect against Imperfect),

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

213

Aspect vs. Actionality

213

also presents, like the remaining Slavic languages, overt morphological marking for the category of actionality. Equally, with class (iii) adverbials (“s tx”, “uže X Vremja”, uže (ntm) and ešˇce) only unbounded lexemes are allowed in both Russian and Bulgarian. Moreover, with the second of these adverbials (“uže X Vremja”) the feature configuration [ bound, terminative] is not allowed, just as the configuration [ telic, terminative] is avoided in Romance languages (cf. the comment on Table 1, concerning the behaviour of It. “da X Tempo”). This of course depends on the deep semantics of these constructions. Let us now consider class (ii) adverbials, those requiring the configuration [ telic, terminative] which, according to Table 2, is the only permitted configuration among the two conceivable mixed cases (i.e., those where actional and aspectual categories differ in the distribution of plus and minus signs). As it happens, in Bulgarian the adverbials “do tx”29 and “X Vremja” require by and large the feature configuration [ bound, terminative], which closely corresponds to the [ telic, terminative] configuration that we find in Romance (English, as we saw above, presents a slightly more complicated picture; cf. 9). Russian, on the other hand, allows both bounded and unbounded predicates. This is in keeping with the lack of overt morphological marking of the aspectual values to be observed in this language.30

4.3. Discussion It is useful, at this point, to tabulate the data concerning Russian and Bulgarian. The main conclusions are presented in Table 3 (where the Russian adverbials also stand for their Bulgarian cognates): It should be clear that the neat correspondence to be observed between Romance (plus English) and Bulgarian, regarding the respective values of the categories telic/ atelic and bounded/unbounded, cannot merely be due to chance. Thus, we believe that there are strong reasons to affirm that these two sets of terms essentially name one and the same category, whose manifestations are by and large similar, although we do not claim strict coincidence (see below for further comments). Russian, however, shows a partly independent behaviour, but this is easy to explain if one considers that this language has no overt marking of the category of aspect proper, as defined at the outset. Thus, while in Bulgarian the dichotomy bounded/unbounded works in essentially the same way as the dichotomy telic/atelic in Romance and English, in Russian the dichotomy bounded/unbounded partly subsumes the role played in Romance and Bulgarian by the aspectual opposition terminative/nonterminative (cf. the behaviour of class (ii) adverbials). The obvious conclusion to be drawn from the discussion above is indeed (as we have already suggested) that the boundedness category essentially belongs to the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

214

214

Pier Marco Bertinetto and Denis Delfitto

Table 3. Actional and aspectual properties of selected Russian and Bulgarian temporal adverbials. Russian bounded (i) “za X Vremja” “X Vremja nazad” (ii) “do tx” “X Vremja” (iii) “s tx” “uže X Vremja” uže (ntm) ešˇce

(“in X Time”) (It. “da X Tempo” tm)

 

(“until tx”) (“for X Time”)

 

Bulgarian bounded terminative

 

   

(“since tx”) (It. “da X Tempo” ntm) (already ntm) (still)

domain of actionality, especially in the case of Bulgarian, where the marking of [telic] values has retained the opportunity to exist independent of aspect. This is hardly a matter of surprise, considering that the formal expression of boundedness is basically a matter of derivational morphology, notoriously connected with lexical meaning, rather than of inflectional morphology, as is the case with aspect proper. Indeed, although in languages such as Romance and English (and indeed in the vast majority of languages) actionality is usually not overtly marked by derivational processes, it clearly deals with lexical meaning. Thus, it is no wonder that [ telic] and [ bounded] behave very similarly. However, it is important to understand that we are not suggesting total identity. It would simply be false to claim that the boundedness category, as implemented in languages such as Russian, coincides tout court with telicity, as implemented in English or Romance. In fact, these two categories are embodied in distinct linguistic structures, and this has obvious consequences for the overall functioning of the relevant tense-aspect systems. If this were not the case, there would be no reason to introduce the pair bounded/unbounded instead of simply telic/atelic. The reason for maintaining this difference is twofold (cf. also fn. 21). First, in the non-Slavic type, telicity tends as a rule (but cf. fn. 20) not to be overtly marked in the lexicon. Consequently, the detelicization of the predicate that is obtained in conjunction with nonterminative tenses is inferred by the language user, rather than directly exhibited by the language. In the Slavic type, by contrast, detelicization simply cannot occur. Consider a language like Russian. Whenever a bounded verb is inserted in an nonterminative context, it necessarily preserves its telic character. Alternatively, an unbounded predicate may be selected in the relevant contexts, but then the [ telic] value is explicitly exhibited.31 The situation is

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

215

Aspect vs. Actionality

215

obviously more complicated in Bulgarian, which also presents overt aspectual oppositions, manifesting the terminative/nonterminative dichotomy. In this language, the speaker does have the possibility of combining bounded verbs with nonterminative tenses, but, interestingly, this does not bring about the detelicization of the predicate; rather it suggests (in most cases) a habitual reading, as observed in fn. 24. Note that this is among the logically conceivable possibilities. Indeed, the habitual aspect implies a series of terminative microevents composing a nonterminative macroevent. By consequence, any individual microevent of a habitual situation is such that it may preserve the inherent properties of telic predicates. It is no wonder, then, that we can find examples such as the following (note that ‘bd/ubd’ stand for ‘bound/unbound’): (28)

Bulgarian Štom napišeše pismo, toj ti as_soon_as write:BD:IMPF:2SG letter, he you:DAT otgovarjaše. answer:UBD:IMPF ‘As soon as you wrote a letter, he would answer you’.

Second, and most important, the actual telic value of the situation described is not always mirrored in the selection of a bounded predicate. All aspectological descriptions of Russian (and related languages) highlight this fact regarding some typical instances, such as habitual and experiential contexts. This has recently also been emphasized by Lindstedt (1995) in a contribution that was written as a reaction to a previous version of the present paper. This author had previously proposed (cf. Lindstedt 1985) for Bulgarian a model in which the features here called “bounded” and “terminative” (named “material bound” and “temporal bound” in Lindstedt 1995)32 interact in an intricate way, so that the peculiar meaning of specific verbal constructions may be interpreted in terms of the scope of one feature over the other. Even if the details of the interpretation do not correspond to the view presented here, there is reason to believe that Lindstedt’s approach is basically compatible with our own. To give an example, consider the following Russian sentences, which respectively instantiate a habitual and an experiential context: (29) a.

Russian Vypit’ ne xotite li? Net, spasibo, ja uže drink:BD:INF not want:PRS:2PL Q no, thank_you, I already pil. drink:UBD:PST ‘Don’t you want to drink? No, thanks, I have already drunk.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

216

216

Pier Marco Bertinetto and Denis Delfitto

b.

Ty možeš’ ob"jasnit’ mne etu zadaˇcu? you can:PRS:2SG explain:BD:INF me:DAT this problem Poprobuju. Kogda-to davno ja rešal ee. try:BD:PRS:1SG sometime long_ago I solve:UBD:PST it:FEM ‘Can you explain this problem to me? I’ll try. A long time ago I solved it.’

In both cases, an unbounded predicate is used instead of a bounded one, although there is no doubt as to the telic character of these two events. Yet, (29a) is perfectly acceptable if the speaker refers to a habitual event, which occurs more or less regularly. Although that given event of drinking may be regarded as completed, the whole series of drinking episodes is not yet over at the time of speaking. Equally, (29b), which is suggested by Lindstedt (1995), is grammatical, and indeed the only possible choice in the given context. Although the speaker did solve the same problem a long time before, he is not certain that he will now reproduce the solution. Lindstedt (1995) suggests that this happens only in dialogues, rather than in narratives, and proposes the following explanation. In a narrative, a telic event provides a prominent temporal reference for the ensuing development of the textual plot. In dialogues such as (29b), on the contrary, the past occurrence of the event can only be viewed as relevant to the present moment; for that matter, the event could have occurred at an unspecified time, with no immediate bearing on anything that immediately ensued from it. Thus, an unbounded verb is selected in both sentences in (29). We do not intend to discuss here the merits of Lindstedt’s proposal.33 Suffice it to say that, whatever the ultimate explanation of the facts presented here is, these are uncontroversial linguistic data that any aspectological model should take into account. What these examples ultimately tell us is that there is not a strict correspondence between the terms “unbounded” and “atelic”. Unbounded verbs may be used, in particular cases, as a sort of neuter or unmarked form, referring (so to say) to a potentially telic event taken in a generic sense, rather than as denoting a truly atelic event. In fact, these are instances of what has sometimes been called “generalfactive meaning” or “simple denotation function”. Even the Bulgarian translations of the Russian examples in (29) present this feature: (30) a.

b.

Bulgarian Ne, blagodarja. Veˇce pix / no thanks already drink:UBD:AOR:1SG / ‘No, thanks. I already drank.’ (iterative situation) Šte opitam. Predi vreme s˘am ja FUT try:BD:PRS:1SG previous time am I rešavax solve:AOR ‘I shall try. Time ago I solved it.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

s˘am pil am drink:PP rešaval / solve:UBD:PP /

217

Aspect vs. Actionality

217

The only difference is that here the terminative tenses make explicit what the Russian sentences leave only implicit: namely, the terminative character of the situation. And note that the Aorist of the unbounded predicate in (30b) (which in this particular case differs from the Imperfect only in stress placement) suggests a quasi-iterative reading (‘I solved it at least once’).34 Nevertheless, we would like to insist that the view defended here has some advantage over the opposing interpretation. Namely, we maintain that the boundedness parameter to be found in the Slavic languages essentially belongs to the domain of actionality rather than aspect proper, despite the existence of cases like those discussed above. We believe that this view permits a better understanding of the observable data at the typological level. Note, in fact, that if boundedness were a purely aspectual phenomenon, it would be hard to understand how this category could ever develop in such a systematic way in Bulgarian without endangering the grammatical status of the independent (and abundant) aspectual devices that this language has at its disposal.35 The alternative view defended here yields a straightforward explanation. The rise of the boundedness parameter provided the Slavic languages with a convenient tool for expressing the telic/atelic opposition that in most other languages is only sporadically captured at the morphological level. Once this system became fully exploited, the natural and inevitable interaction between boundedness and terminativity made it possible, for most Slavic languages, to dispense entirely with the overt manifestation of aspect, concomitantly enlarging the coverage of the boundedness category. This has to some extent also happened in Bulgarian; witness the examples in (30). However, the evidence gathered in Table 3 suggests that the difference between the purely aspectual values incorporated in the tense oppositions and the basically actional values expressed by the boundedness category (specifically implementing the telic/atelic distinction) is still strongly preserved in this language.36

Notes 1. We wish to thank our colleagues of E UROTYP Group 6 (in particular Jouko Lindstedt) for useful comments. We also want to thank Walter Breu for his detailed counterarguments, as well as Georgi Jetchev for his help with the Bulgarian data (cf. also fn. 25). The paper was jointly developed by the two authors. For academic purposes, however, PMB bears responsibility for Sections 1, 3 and 4, while DD bears responsibility for Section 2. 2. Other terms that have been used are ‘character’ and ‘intrinsic meaning’ of the verb. 3. Note that the terms “Perfective/Imperfective” appear here (as in (iii) below) with capitalized initials, to suggest that they represent traditional grammatical labels in the Slavic languages. By contrast, in (ii) below no capital is used, for the same terms represent purely theoretical terms. The clarification of the difference between the truly aspectual and the actional meaning of these terms is one of the fundamental aims of this paper. However, to avoid misunderstandings, let us say at the outset that we do not intend to

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

218

218

4.

5.

6.

7.

8. 9.

10.

11.

12.

Pier Marco Bertinetto and Denis Delfitto claim that the opposition to be observed in the Slavic languages is exclusively actional, rather than aspectual. We would like to ask the reader to wait until Section 4 before jumping to any conclusion. For a different view on the topic addressed here, cf. Klein (1995). For a full-fledged actional classification, cf. for instance Bertinetto (1986). Note that telic and perfective (or “terminative”, as we shall propose to designate this notion) should not be confused. The first term belongs to the actional subdomain, i.e., ultimately to the lexical one, whereas the latter belongs to the aspectual subdomain. Except for the partial convergence to be observed in some languages (such as Russian, which lacks separate expressions for the categories of aspect and actionality), the notions of telic and perfective (or atelic and imperfective) are normally conveyed by distinct devices. Cf. below, this section, for further comments on this point. This entails that even the notion ‘tense-aspect system’ is not very felicitous: ‘tense system’ would be a more appropriate denomination. However, since this term has now become standard, we shall use it in this paper. Note further that, in addition to temporal and aspectual values, tenses may also express modal meanings; but this is not the default case. In some sense, one could say that “tense” is a formal notion, inasmuch as it consists in a concrete, morphological expression, while “aspect” and “temporal reference” are semantic notions, which need a particular tense to manifest themselves. Note, however, that the correspondence is not one-to-one: each tense has both an aspectual and a temporal meaning, and very often it has in fact a range of aspectual and temporal meanings, which differ according to the context in which the given tense is embedded. Lazzeroni (1990) has suggested a similar view with respect to the notion “middle” in ancient languages, which he regards as historically connected with the idea of stativity. The class of verbs incorporating this notion is considered to have undergone progressive transformations, retraceable on the morphological level, through the intermediation of a number of parameters (non-eventivity, non-agentivity, subject-orientation), which ultimately also interact with the notion of Perfect. Thus, we have a constellation of meanings, producing a certain amount of vagueness, which is responsible for the cross-linguistic variation to be observed in Classical languages, and their historical development. See however, at least, Platzack (1979). Note that in (c–d) the event is qualified as “detelicized”, rather than “atelic”, because the predicate is telic in its basic meaning, but turns out to be detelicized in the given context. As to the reason for this, it is obviously different in the two cases: in (c) the detelicization is induced by the adverbial, while in (d) it is yielded by both the adverbial and the nonterminative aspect. This is not a unique case, even among European languages. Other European languages whose progressives may have a future-time reading interpretation are: Icelandic, Maltese, and to some extent Finnish. As noted already regarding example (1), it is also possible for these sentences to receive a prospective (here, future-in-the past) reading. This would not modify our conclusions, though, for this would clearly be a terminative reading. Apparently, the meaning associated with (17a) and (17c) is more common with speakers

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

219

Aspect vs. Actionality

13.

14. 15.

16. 17. 18. 19. 20.

21.

219

of American than British English. With (17a), the latter speakers are likely to prefer a reading of the following sort: “at that time, Mary was already able to dance the polka”. With (19a), if we ignore the temporal clause, it is marginally possible to have the following readings: “At the given time Mary was still able to dance the polka”, or “Mary still used to dance the polka”. Here we disregard these readings. Besides, still may also be used as a contrastive adverb, in which case it is obviously compatible with terminative tenses (Still [i.e. notwithstanding this] Mary danced the polka). More properly, they are restricted to accomplishment verbs, for achievements are incompatible with them, due to their non-durative character. It should be added that Fr. “depuis X Temps” does not easily combine with accomplishment verbs, but rather with achievements. In fact, even It. “da X Tempo” undergoes pragmatic restrictions when used with accomplishments. We shall not go into these details here. For a thorough analysis of this construction, cf. Squartini (1998). Dauses (1981) has baptized this particular use of the Romance Imperfect ‘absolute Imperfect’. Or the Compound Past, in as much as it subsumes, in the appropriate contexts and in the relevant languages, the specific functions of the Simple Past. Cf. Bertinetto and Squartini, this volume. Note that the view presented here is not restricted to the description of Romance languages. Indeed, these are singled out only as an example, but any language exhibiting a terminative/nonterminative opposition would behave in the same way. Similarly, the type of lexical oppositions exhibited by the Slavic languages is also to be observed, to some extent at least, in some non-Slavic languages, such as Lithuanian, Hungarian and Georgian. Actually, these lexical oppositions are occasionally manifested in many more languages. Although they are very marginal in, say, Romance, they are relatively well attested in Germanic: e.g., Eng. eat vs. eat up, Germ. jagen ‘hunt’ vs. erjagen ‘catch (in hunting)’. However, even in the latter case they are not nearly as systematic as they are in the Slavic languages. Interestingly, Hentschel (1991) has pointed out that the German particle mal may receive an aspectual (specifically, terminative) interpretation in certain contexts. However, this is far from being a systematic behaviour. And, in any case, it should not be confused with the problem discussed here. While the lexical pairs reported above exhibit actional oppositions, the particle mal is used by German speakers (at least in part) to overcome the poverty of aspectual devices offered by the language. For a different approach to the verbal system of Slavic languages, cf. Breu (1984a; 1984b; 1992). The view of this author is in a way diametrically opposed to ours, inasmuch as he claims that the so-called aspectual pairs constitute in fact one and the same lexical entry. So, for instance, according to Breu, in Russian there is one verb consisting of the joint paradigms of cˇ itat’/poˇcitat’, and another verb corresponding to cˇ itat’/proˇcitat’. Accordingly, there are two homonyms cˇ itat’, one with activity reading, the other with accomplishment reading. For us, on the contrary, cˇ itat’ may only have an activity reading, while its bounded cognates may supplement their actional value with different semantic specifications, according to the peculiarities of their meaning. In fact, besides the bounded/unbounded opposition which constitutes the basis of the Slavic verbal lexicon,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

220

220

22.

23. 24.

25. 26. 27.

28. 29.

Pier Marco Bertinetto and Denis Delfitto there are many additional semantic nuances, and often quite a lot of sheer idiosyncrasy, particularly with bounded predicates. Nevertheless, it is fair to say that the class of po-prefixed verbs, referred to by Breu, is quite peculiar, inasmuch as it expresses the idea of temporal limitation without implying telicity. Indeed, poˇcitat’ takes “for X Time” adverbials and rejects “in X Time” adverbials, thus behaving like cˇ itat’ in sharp contrast to proˇcitat’ (which shows the opposite distribution of adverbials), although on the other hand poˇcitat’ is similar to proˇcitat’ in substantially rejecting the adverb dolgo ‘for a long time’ (perfectly compatible with cˇ itat’). We would then like to say that po-prefixed verbs are rather exceptional in being bounded but not telic; something that should be borne in mind, to smooth our general position concerning the fundamental convergence of the features “telic” and “bounded”. Indeed, Bulgarian also presents explicit oppositions in the modal domain of evidentiality with respect to the so-called “testimonial/non-testimonial” tenses. On the topic of evidentiality, see Guentchéva (1996). In practice, this treatment applies to the adverb types exhibiting a ‘()’ mark in the aspect column in Table 1. This does not mean, however, that one cannot find verbal forms exhibiting this particular combination of features (i.e. [tel, terminative]), although the latter is excluded for temporal adverbials. This does indeed occur in Bulgarian in the translation of sentences (in the sense of: “ he copied entirely”), such as: Whenever he copied the article where a bounded verb would be used. Actually, this combination of features should be expected on the basis of what we observed in Section 2.1.1, where we claimed that habitual contexts (such as the one discussed here) consist of a nonterminative macroevent composed of a series of terminative microevents. We wish to thank Mrs. Kumuš Imanaleva, Mrs. Olga Obuchova and Prof. Neli Radanova Kuševa for their patient help. For simplicity, Russian words will also stand for their Bulgarian cognates. Cf. Table 3 for the translation of these adverb types. Note that “X Vremja nazad” corresponds more properly, in some of its uses, to “X Time ago”, whose semantic and syntactic properties are closer to It. “X Tempo fa” than to It. “da X Tempo”. It should obviously be noted, in this context, that temporal adverbs generally differ in very subtle ways from language to language. The correspondences are almost never perfect. Recall that we are assuming here progressive, rather than habitual, interpretation. Here again we have more proof of what we said in fn. 27. The adverb type “do tx” has both the meaning of “until tx” and the meaning of It. “entro tx” (roughly corresponding to “within tx”). Needless to say, this has relevant consequences for our purposes, because the latter adverb type takes only telic predicates. Cf.: (i)

(ii)

*Gianni dormì entro le 5. Gianni sleep:SP:3SG within the 5 [If this had a meaning, it would be: ‘Gianni finished sleeping before 5’.] Gianni consumò il pranzo entro l’ una. Gianni consume:SP:3SG the lunch within the one ‘Gianni ate up his lunch before 1 o’ clock.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

221

Aspect vs. Actionality

221

The data reported in the text refer only to the meaning corresponding to “until tx”. 30. A notable exception to this generalization is represented, in both Bulgarian and Russian, by bounded verbs inherently expressing the notion of temporal limitation, such as Russian pospat’ ‘sleep for a while’, postajat’ ‘stand for a while’. However, this is not surprising, given the strict correspondence of the notion of temporal limitation, included in the meaning of these predicates, with the semantics of class (ii) adverbials (see fn. 21). 31. Note that this may bring about possibilities that are not within the reach of non-Slavic languages. For instance, the Italian verb esaurirsi ’exhaust’ has a non-deletable telic character, to the extent that the Imperfect used in the following sentence does not detelicize the predicate, contrary to what normally happens in such cases: (i)

??Le scorte si esaurivano, ma non si sono poi the supplies RFL exhaust:IMPF but not RFL be:PRS:3PL then esaurite. exhausted ‘The supplies were on the verge of finishing, but ultimately did not finish.’

By contrast, the translation of this sentence into a Slavic language would involve, in sequence, first an unbounded and then a bounded lexeme, as in the following Bulgarian rendering (and note that Russian would behave essentially in the same way, as far as the category “boundedness” is concerned): (ii)

Zapas˘at se izˇcerpvaše, no v krajna smetka supply:DEF RFL exhaust:UBD:IMPF:3PL but in extreme count ne se izˇcerpaxa. not RFL exhaust:BD:AOR:3PL ‘The supplies were on the verge of finishing, but ultimately did not finish.’

Thus, in Slavic languages the verb corresponding to ‘exhaust’ may be conceived of as atelic in the relevant contexts. 32. Cf. also Bondarko (1987), quoted by Lindstedt (1995), who suggests “internal bound” and “external bound”. 33. We do sympathize with Lindstedt’s proposal, which captures also some significant analogies among Russian and Finnish (in the latter case regarding the choice of accusative vs. partitive). However, it seems to us that this proposal is designed to capture cases such as (29b), rather than (29a). It should therefore be complemented by an appropriate formulation concerning habitual contexts. In this respect, we would like to advance the following hypothesis. Recall what we said in Section 2.1.1: a habitual situation is characterized as a nonterminative macroevent comprising an undeterminate number of microevents, each of which may be viewed terminatively. Obviously, whenever the relevant type of predicate is involved, any terminative microevent may also by definition be telic. However, no matter what kind of predicate is employed, the general situation corresponding to the macroevent does not imply telicity, for the designated habit may not be yet terminated at the end of the reference interval. Consider now that unbounded verbs, which essentially suggest atelicity, are the only device available to suggest interminativity in languages like

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

222

222

Pier Marco Bertinetto and Denis Delfitto

Russian. If this is so, it is not surprising that, in order to insist on the interminativity of the macroevent (the real distinctive feature of habitual aspect), these languages make use, in contexts such as (29a), of a tool which, although not directly expressing interminativity, nevertheless hints at it by proximity of meaning. 34. Needless to say, it would also be possible in (30b) to stress the telicity of the event, by using a terminative tense like the Aorist with a bounded predicate (rešix). But the really interesting cases are those presenting apparently contrasting features. 35. It has been pointed out to us that this is not a valid argument, given the frequent cases of “layering” discussed in the grammaticalization literature. Indeed, it is not the case that new categories arise only to fill empty slots; otherwise, French would never have developed a new Future to replace the old one. We agree with this, but we would like to remark that, in the long run, when this happens, there is always a redistribution of the functional charge of the competing elements. These do not remain forever perfectly interchangeable. Now, since the tense system of Common Slavic was imported from IndoEuropean, and since it has retained its functions on the temporal and aspectual level in Modern Bulgarian, there is reason enough to argue that the innovation represented by the boundedness category was introduced to fulfill some other purpose, rather than simply express aspectual values. The latter is an evolution that took place to some extent in languages such as Russian, due to the loss of the previous aspectual distinctions. Obviously, something similar might happen in the future development of Bulgarian: but this is another story, yet to be told. 36. Jouko Lindstedt (p.c.) pointed out to us that the situation appears to be slightly different in Macedonian, a closely related language, which is in the process of losing the Aorist of unbounded verbs. Modern Croatian is a similar case: the Imperfect is apparently used only for unbounded verbs, while the Aorist is mostly employed with bounded verbs (Morabito 1992). The next step is represented by the two Sorbian languages, where the Imperfect of bounded verbs has been lost (cf. also Lötzsch 1995). Thus, although Sorbian still preserves the aspectual opposition Imperfect/Aorist, in practice this tends to come out in fixed combinations with respect to the boundedness parameter: bounded verbs are normally associated with terminative tenses, unbounded verbs with nonterminative ones. This is presumably the path along which Russian and languages of the same sort reached their present state, in which the aspectual oppositions have been lost entirely. Indeed, the research carried out by Petruxin (1996) on the chronicle Piskarevskij letopisec (written over a long period, extending from the XI to the XVII cent.), provides a striking confirmation of this. Especially in the first part of the text (XI–XIII cent.), where the so-called l-forms of the Past are still rarely used to the advantage of the Aorist and the Imperfect, one observes with prefixed verbs a very strong correlation between boundedness and the Aorist. More precisely, the Imperfect is rarely attested in general, and is virtually absent with bounded predicates. By contrast, with non-prefixed verbs both bounded and unbounded predicates appear in a more or less balanced proportion; however, the Imperfect is used only with unbounded predicates. These remarks demonstrate that each language in the Slavic family may show a different stage of development regarding the fusion of actional and aspectual values. A hint of this is also offered by the responses to our Progressive Questionnaire (= PROGQ; cf.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

223

Aspect vs. Actionality

223

Appendix 3), in which we collected data for Serbo-Croatian (four speakers) and Rusyn (one speaker). Obviously, there is no specialized progressive device in Slavic languages. In those languages that lack specifically nonterminative tenses (like Russian, or the two mentioned here), the unbounded verbs may express any nonterminative value (specifically, progressive and habitual). Among the two languages examined, Rusyn seems to make much wider usage of unbounded predicates. Although some caution should be taken, given the fact that we collected the responses of only one informant, the contrast with Serbo-Croatian is apparent. Rusyn uses a bounded predicate alone in only one case (PROGQ:5), while it presents it as an alternative to the unbounded predicate in one more case (PQ:74). All other examples elicited an unbounded verb from our informant. With our Serbo-Croatian informants, the situation is more varied. Here again, there is just one instance in which the bounded predicate is the only type of response provided by the speakers (PQ:5). However, in quite a number of cases we observe some vacillation in the behaviour of our informants. Interestingly, this seems to correlate to some extent with the degree of telicity. The more prominent the telic value of the sentence is, the more frequent are the bounded verbs provided by the informants. In what follows, a formula such as “x  y” indicates first the number of bounded verbs, then the number of unbounded ones (note that, although there were four informants, the total number of responses may be greater than four, because a single informant has sometimes provided a double answer). The most relevant cases are the following: PROGQ:14 (3 Kg of potatoes) = 2  3; PROGQ:15 (all the potatoes) = 2  2; PROGQ:17 (two chickens) = 2  2. Other examples of the same sort are: PROGQ:21 (leave (the house)) = 2  3; PROGQ:23 (begin to peal) = 3  1; PROGQ:30 (throw the stone) = 2  3; PROGQ:32 (give the sign) = 2  2; PROGQ:37 (the water boil) = 2  4; PROGQ:52 (forget the names) = 2  4; PROGQ:53 ((the level) increase) = 2  3; PROGQ:55 ((snow) cover the land) = 4  1. We must of course allow for a certain degree of variation, possibly also due to mere misunderstanding of the intended meaning of the sentence. However, the contrast between Rusyn and Serbo-Croatian is striking, and leads us to believe that the grammatical contrast bounded/unbounded works slightly differently in the two languages. Namely, in Serbo-Croatian this contrast seems to be used with more an actional than aspectual intention, while in Rusyn the aspectual value predominates. The figures for Serbo-Croatian show that when emphasis is put on the telicity of the event, the bounded predicate tends to be preferred over its competitor. This is obviously in accordance with the interpretation suggested in this paper: although the opposition bounded/unbounded may express in the relevant contexts a purely aspectual meaning, its original value is actional, and basically addresses the opposition telic/atelic. Now, since progressivity implies the detelicization of the predicate, there is an obvious interaction between the aspectual and the actional meaning, and it is thus no wonder that unbounded predicates are preferably chosen to convey a sense of progressivity. However, the comparison between Rusyn and Serbo-Croatian shows that these two languages occupy different stages in terms of the transformation of the original actional meaning into a purely aspectual one.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

224

224

Pier Marco Bertinetto and Denis Delfitto

References Bertinetto, Pier Marco 1986 Tempo, Aspetto e Azione nel verbo italiano. Il sistema dell’indicativo. Firenze: Accademia della Crusca. 1994 “Temporal reference, aspect and action: their neutralizations and interactions, mostly exemplified on Italian.” In: Carl Bache, Hans Basbøll, Carl-Erik Lindberg (eds.), Tense, aspect and action. Empirical and theoretical contributions to language typology, 113– 137. Berlin/New York: Mouton de Gruyter (also in Bertinetto 1997, ch. 5). 1997 Il dominio tempo-aspettuale. Demarcazioni, intersezioni, contrasti. Torino: Rosenberg & Sellier. this volume “The progressive in Romance, as compared with English.” Bertinetto, Pier Marco and Denis Delfitto 1996 “L’espressione della progressività/continuità: un confronto tripolare (italiano, inglese e spagnolo).” In: Paola Benincà, Guglielmo Cinque, Tullio De Mauro and Nigel Vincent (eds.), Italiano e dialetti nel tempo. Saggi di grammatica per Giulio C. Lepschy, 45–66. Roma: Bulzoni. Bondarko, Alexander V. 1987 Teorija funkcional’noj grammatiki: Vvedenie. Aspektual’nost’. Vremennaja lokalizovannost’. Taksis. Leningrad: Nauka. Breu, Walter 1984a “Grammatische Aspektkategorie und verbale Einheit.” In: Wolfgang Girke and Helmut Jachnow (eds.), Aspekte der Slavistik. Festschrift für Josef Schrenk, 7–25. München: Sagner. 1984b “Zur Rolle der Lexik in der Aspektologie”, Die Welt der Slaven 29: 123–148. 1992 “Zur Rolle der Präfigierung bei der Entstehung von Aspektsystemen.” In: Marguerite Guiraud-Weber and Charles Zaremba (eds.), Linguistique et slavistique. Mélanges offerts à Paul Garde, 119–135. Aix-en-Provence: Publications de l’Université de Provence, 1. Bybee, Joan L. and Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Cohen, David 1989 L’aspect verbal. Paris: PUF. Comrie, Bernard 1976 Aspect. An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. 1985 Tense. Cambridge: Cambridge University Press. Dahl, Östen 1985 Tense and aspect systems. Oxford: Blackwell. Dauses, August 1981 Das Imperfekt in den romanischen Sprachen. Seine Bedeutung im Verhältnis zum Perfekt. Wiesbaden. Guentchéva, Zlatka 1991 Temps et aspect: l’exemple du bulgare contemporain. Paris: CNRS. Guentchéva, Zlatka (ed.) 1996 L’énonciation médiatisée. Louvain: Peters. Hatav, Galia 1989 “Aspects, Aktionsarten, and the time line”, Linguistics 27: 487–516.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

225

Aspect vs. Actionality

225

Hentschel, Elke 1991 “Aspect versus particle: Contrasting German and Serbo-Croatian”, Multilingua 10: 139–149. Johanson, Lars this volume “Viewpoint operators in European languages.” Klein, Wolfgang 1995 “A time-relational analysis of Russian aspect”, Language 71: 669–695. Lazzeroni, Romano 1990 “La diatesi come categoria linguistica: studio sul medio indoeuropeo”, Studi e Saggi Linguistici 30: 1–22. Lindstedt, Jouko 1985 On the semantics of tense and aspect in Bulgarian. (Slavica Helsingiensia 4.) Helsinki. 1995 “Understanding perfectivity – Understanding bounds”, In: Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl and Mario Squartini (eds.), Temporal Reference, Aspect and Actionality: 2, Typological Perspectives, 95–103. Torino: Rosenberg & Sellier. Lötzsch, Ronald 1995 “Das sorbische Tempus-System”, In: Rolf Thieroff (ed.), Tense systems in European languages II, 167–179. Tübingen: Niemeyer. Mittwoch, Anita 1988 “Aspects of English aspect: on the interaction of Perfect, progressive and durative phrases”, Linguistics and Philosophy 11: 203-54. Morabito, Rosanna 1992 “L’incrocio aspettuale nel sistema aspetto-temporale della lingua croata”, Quaderni del Dipartimento di Linguistica, Università di Firenze, 3: 77–91. Parsons, Terence 1989 “The progressive in English: Events, states and processes”, Linguistics and Philosophy 12: 213–241. Petruxin, Pavel 1996 “Narrativnaja strategija i upotreblenie glagol’nyx vremen v russkoj letopisi XVII veka”, Voprosy Jazykoznanija 45: 62–84. Platzack, Christer 1979 The semantic interpretation of aspect and aktionsarten. A study of internal time reference in Swedish. Dordrecht: Foris. Sasse, Hans-Jürgen 1991 “Aspect and aktionsart. A reconciliation”, Belgian Journal of Linguistics 6: 31–44. Squartini, Mario 1998 Verbal Periphrases in Romance: Aspect, Actionality and Grammaticalization. Berlin: Mouton de Gruyter. Squartini, Mario and Pier Marco Bertinetto this volume “The Simple and Compound Past in Romance languages.” Vendler, Zeno 1967 “Verbs and times”, In: Zeno Vendler, Linguistics in philosophy, 97–121. Ithaca/London: Cornell University Press. Vlach, Frank 1981 “The semantics of the progressive”, In: Philip Tedeschi and Anne Zaenen (eds.), Tense and Aspect. Syntax and Semantics 14, 271–92. New York: Academic Press.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

226

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

227

Eva Hedin

The type-referring function of the Imperfective

1. Introduction This paper proposes a description of the perfective and the imperfective aspects based on the type/token distinction. It takes as its point of departure some uses of the Imperfective which have sometimes been treated as exceptions to the rules governing the choice of aspect, since they cannot be satisfactorily accounted for by a traditional description. It will be argued, that these uses of the Imperfective can be described in the same way as its more typical uses, and that their common function is to be used for reference to situations as types rather than as tokens. In Slavic aspectology there has been a great deal of discussion about the use of the Imperfective to refer to single bounded1 events, that is, in contexts typical for the Perfective, according to a traditional description of aspect. In Russian, where this use is widespread, it has been referred to as the “simple denotative” or “general factual” function of the Imperfective. The term has been used by different scholars in different ways and about various functions of the Imperfective. A typical example, however, is (1), where the Russian Imperfective corresponds to the English Perfect: (1)

Russian [The waiter to the customer:] Vy uže zakazyvali? you:PL already order:IPFV:PST:PL ‘Have you ordered?’

The use of the Imperfective “pro Perfective” is sometimes referred to as a piece of evidence for considering the Perfective to be the marked form in Russian and other Slavic languages (e.g., Comrie 1976: 113). The fact that the Imperfective may be used for a “typically perfective situation” would mean that it has a less specific (in the sense less marked) meaning than the Perfective. “The Perfective always has perfective meaning but the Imperfective may or may not have imperfective meaning” (Comrie 1976: 113). “Imperfective meaning” is to be understood here as habitual, continuative, etc. According to this view, the use of the Imperfective “pro Perfective” would be a case of neutralization of the aspect opposition. Contrary to the Russian one, the Modern Greek2 Perfective is normally considered to be the unmarked member of the opposition. It is the most frequently used aspect,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

228

228

Eva Hedin

“the natural aspect for the verb to be in unless there are clear grounds for using the imperfect (i.e., that the action is considered as durative or repeated)” (Mackridge 1985: 105–106). However, also in Modern and Classical Greek, the Imperfective may be used “pro Perfective”, that is, about single bounded events in contexts where a continuative or habitual interpretation is not possible, although in these languages, this use does not always look the same and – at least in Modern Greek – is not so widespread as in Russian (cf. Section 5 on aspect usage). In this paper, the two aspects are considered as members of an equipollent opposition, in which neither has a more specific meaning than the other. Thus, the use of the Imperfective "pro Perfective” is not considered as a neutralization of some aspect opposition, but as actually representing a basic imperfective function, common to all its uses, namely type reference.3

2. The Imperfective and the Perfective – a functional description 2.1. A description of aspect based on the type/token distinction The functional description of aspect proposed in this paper is based on the view that the Imperfective and the Perfective represent two ways to refer to situations. The Imperfective is used when reference is made to situations as types, when they are considered in a non-temporal perspective as abstractions not existing in time but corresponding to the denotative content of some verbal expression (like for instance ‘die’, denoting (a typical member of) the situation class DIE). The Perfective, on the other hand, is used when reference is made to situations as tokens, as instantiations of situations in time. It should be noted that temporality does not correspond to ‘temporal localization’ (temporal’naja lokalizacija) discussed in Slavic aspectology (see, e.g., Leinonen 1982). Even if a situation can be located at some unique point in time and could thus be referred to as a concrete situation token, the speaker may still choose to consider it in a non-temporal perspective as an abstract situation type. This is possible even if the temporal localization is explicitly made by the context. Consequently, the idea that the “imperfective” situation is considered as a type does not mean that there cannot be a unique situation that is subject to discourse, only that – even if there is – it is not referred to as an instantiation (cf. Section 2.4).

2.2. Boundedness The difference between verb phrases of the kind build as opposed to build a house, walk as opposed to walk a mile, etc., has been referred to as a difference of boundedness. A bounded situation is said to be typically perfective and a nonbounded typically imperfective (for a discussion, see Dahl 1981).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

229

The type-referring function of the Imperfective

229

In the present description, boundedness is seen as a property of situation types (transitional or not), represented by some verb(s) or verb phrase(s). It is not considered to be necessarily incompatible with the Imperfective. A situation may be referred to as representing some class or species of situations having or not having the property of boundedness or as the instantiation of such a situation occurring in time. A bounded verb phrase may be in the Imperfective or the Perfective but, depending on aspect, this element of boundedness has different functions. This may be illustrated by some examples from Modern Greek. In (2), (3) and (4) the Perfective is used. (2)

Éxtise éna spíti. build:PFV:PST INDEF house ‘He built a house.’4

(3)

Píje éna xiljómetro. go:PFV:PST one kilometre ‘He walked a kilometre.’

(4)

Péthane. die:PFV:PST ‘He died.’

Here some actual new state of affairs is referred to: the existence of a house, somebody having walked a kilometre or somebody being dead. In the corresponding Imperfective examples (5–7), on the other hand, nothing is said about any actual accomplishment of the situation. (5)

Éxtize éna spíti. build:IPFV:PST INDEF house ‘He built (was building) a house.’

(6)

Píjene éna xiljómetro. go:IPFV:PST one kilometre ‘He walked (was walking) a kilometre.’

(7)

Péthene. die:IPFV:PST ‘He died (was dying).’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

230

230

Eva Hedin

One does not consider the actual realisation of the situation types in question (the token, the instantiation) but makes some kind of “lexical description” of some situation as being of that kind, belonging to that class or representing that species of situations (‘build a house’, ‘walk a kilometre’, ‘die’). Since the situation is only described in non-temporal abstract quality terms as a type, its transition or bound is referred to only as part of its definition (as ‘house-building’, ‘kilometre-walking’ or ‘dying’), not as some realisation of a transition in time. The situation is described as being of that kind of situation that is characterised by the quality that, when it comes to an end, such and such becomes the case (for transitional phrases) – for instance, the existence of a house or somebody being dead – or has been the case (for non-transitional phrases) – for instance, somebody having walked a kilometre. Whether the situation, referred to in this way by a bounded verb phrase in the imperfective aspect, is actually accomplished or not is a matter of interpretation. Its accomplishment is neither stated nor denied. The actual instantiation in time is simply irrelevant, rather as it is irrelevant in a nominal description like housebuilding, letter-writing or dying.

2.3. The Imperfective Paradox One advantage of the description proposed above is that it offers another way to handle transitional verb phrases in the Imperfective oriented to some point of reference, that otherwise create problems of truth conditions (the so-called Imperfective Paradox, Dowty 1979: 133). Thus, according to this description, what is stated in the classical example (8) (in its Modern Greek version): (8)

O Jánis xtízi éna spíti. DEF John build(:IPFV):PRS INDEF house ‘John is building a house.’

is not that John at any time actually builds a house, in the sense that some house comes into existence at any present point of time, or that it may do so in the future, only that John is the agent of a situation type called build a house (house-building). The actual realisation of a transitional situation (like build a house) cannot, for obvious reasons – while it is ongoing – be referred to as something being true at the point of utterance. If, however, the situation is referred to as an abstract type (like a nominal description: house-building) without consideration of the token, its actual realisation in time, the proposition ‘John build a house at time t’ (or, perhaps, better represented as ‘John build-a-house at time t’) may well prove true or false at the point of reference, even if, at this point of time, the actual situation has not yet come to its end, and no house has yet come into existence. John is the agent of something

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

231

The type-referring function of the Imperfective

231

that could be referred to as build a house but not knit a sweater or write a letter. He is at this point of time a house-builder. In other words, in every instance of its realisation, a transitional situation may be referred to as a type or species (as BUILD A HOUSE, DIE, WALK A KILOMETRE) but it cannot in any instance be referred to as a realisation of this situation type (as build a house, die, walk a kilometre). In the first case, its actual accomplishment is neither stated nor denied, it is simply not considered. The irrelevance of the instantiation of the situation could be referred to as the neutral function of the Imperfective. But it is not a matter of neutralization of an aspectual opposition, of the Imperfective losing its “normal” functions. The Imperfective has its basic meaning of referring to a situation as a type, and by doing so it (by definition) neglects any instantiation of the situation occurring in time. This “neutral” function of the Imperfective could also be referred to as a more abstract way of referring to a situation.

2.4. Type reference in specified context Verbs are not always used to refer to some situation(s) occurring in time. In a generic sentence like Dogs bark the verb bark is used to describe the canine species, to qualify, not to refer to any instantiation(s) of barking. Barking (as opposed, e.g., to meowing) is a type of behaviour that is typical of dogs, like being hairy or having four legs. Something similar is the case in a habitual sentence like Modern Greek O Jánis kapnízi/kápnize ‘John smokes/smoked’ which can be used to say that ‘John is/was a smoker’ without referring to any instance of smoking at all. That the verb is not always used to refer to situation tokens is obvious in generic or habitual sentences which, by definition, do not focus on any particular instantiation of the situation type denoted by the verb phrase. But the verb may also have a similar function in a specific context. A sentence may contain reference to an identifiable situation (occurring at some specific time) but the situation must not necessarily be considered specifically. So even if it is possible to specify a situation referentially by other factors, like temporal adverbials etc., one does not have to refer to it as a situation token. Let us make a parallel with noun phrases: A, B, C, D and E are present in a room. Only A is a female. B utters something and D makes the reproach: (9)

Watch your language! There are ladies present.

It is quite obvious that the person indirectly referred to is A. This lady could be referred to as a unique instance of the type lady, but is not. It is not this particular woman’s presence that is stated but the fact that her species is represented. D might more explicitly have said:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

232

232 (10)

Eva Hedin

Watch your language! The species lady is represented here.

In some other context, however, the same lady may be referred to in another way: (11)

There is a lady waiting for you in your office.

Here, we refer to some instance of the species lady, and the generic affiliation (the lady’s being a lady) is not focused on as in (9). In (11), the lady in question is considered as a lady token sitting in the office rather than as the type lady represented in that local domain. The same kind of shift of referential focus is possible when we refer to situations. An identifiable situation may be referred to in two ways. Either the situation token itself is focused on or the situation type that it represents is. Thus, the situation may be considered as a type not only in non-specified contexts but also in a specified context, where it is possible to refer to it as a particular instantiation, as a token.

2.5. Boundedness, totality and the type-referring function of the imperfective The notion of boundedness is often mentioned as an important criterion for defining the difference between the Imperfective and the Perfective. According to Dahl (1985: 74–75) boundedness is crucial to the choice between the Perfective and the Imperfective in Russian. He refers to the following sentences from his typological questionnaire: (TMAQ: 9) [A: I went to see my brother yesterday. B: What he DO? (= What activity was he engaged in?)] He WRITE letters (TMAQ: 13) [A: What did your brother do after dinner yesterday?] He WRITE letters (TMAQ: 14) [A: What did your brother do after dinner yesterday?] He WRITE a letter Most of the languages investigated by Dahl and having a Perfective:Imperfective opposition used the Perfective in the translation of (TMAQ: 13). Modern Greek was one example. (12)

Modern Greek (TMAQ: 9) Éghrafe ghrámata. write:IPFV:PST letters

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

233

The type-referring function of the Imperfective

(13)

Modern Greek (TMAQ: 13) Éghrapse ghrámata. write:PFV:PST letters

(14)

Modern Greek (TMAQ: 14) Éghrapse (éna) ghráma. write:PFV:PST (INDEF) letter

233

The Slavic languages, however, used the Imperfective (in the case of Bulgarian, the Imperfective Aorist). The Russian translations were as follows: (15)

Russian (TMAQ: 9) On pisal pis’ma. He write:IPFV:PST letters

(16)

Russian (TMAQ: 13) On pisal pis’ma. He write:IPFV:PST letters

(17)

Russian (TMAQ: 14) On napisal pis’mo. He write:PFV:PST letter

Dahl points out that for (18), the Perfective counterpart of (TMAQ: 13), to be acceptable as an answer to (TMAQ: 13), it would have to refer to a definite set of letters. (18)

Russian On napisal pis’ma. He write:PFV:PST letters

“The reason that a Perfective verb cannot be used in (Q.13) is that ‘writing’ is an unbounded activity as long as we haven’t delimited the object in any way. If we do so, however, as in (Q.14), the Perfective napisal can be used” (Dahl, 1985: 75). For Slavic languages the borderline between (Q.13) and (Q.14) would thus be determined by the different values of boundedness of (Q.9) and (Q.13) on the one hand (representing non-boundedness) and (Q.14) on the other hand (representing boundedness). However, the fact that the Perfective can be used in (Q.14) does not say that it has to be.5 It is obviously not the case that the Imperfective cannot be used about bounded situations. Consider Comrie’s (1976: 113) description of the “simple denotative” or “general factual” Imperfective in Russian:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

234

234

Eva Hedin

it is possible in Russian to use the Imperfective, when there is no specific reference to the completeness of the event [ ]. Here the speaker is simply interested in expressing the bare fact that such and such event did take place without any further implications, and in particular without any implication of progressive or habitual meaning [ ].

Dahl argues that this description of the “simple denotative” use of the Imperfective is incompatible with the “totality view” of perfectivity as described by Comrie (1976: 16): perfectivity indicates the view of a situation as a single whole, without distinction of the various separate phases that make up that situation; while the imperfective pays essential attention to the internal structure of the situation.

He correctly points out that the description of the Imperfective as expressing “the bare fact that such and such an event did take place, without any further implications” makes it look very much like a prototypical case of viewing an event as a single whole without paying “essential attention to the internal structure of the situation”, i.e., it seems to describe “totality” (1985: 76). This illustrates the difficulties connected with a description of the type-referring function of the Imperfective, as it occurs in the simple denotative function, within the framework of a traditional description of aspect. When some effects of type reference are described without being recognized as such, the description of them easily becomes confused with a description of totality. It was noted above that, when a transitional situation is referred to as a type, its transitional phase is included as part of its definition (as for instance ‘house-building’). In some sense, this results in some kind of “wholeness-view” of the situation. This could explain why the intuitive description of the Imperfective “pro Perfective” may look very much like the description of the Perfective in totality terms. Likewise, a description of the neutral function of type reference, its non consideration of the instantiation of the situation, may look very much like the description of the non consideration of “the internal structure of the situation” of the totality view. However, it is also possible to understand Comrie’s description in the following way. If “the bare fact that such and such event did take place without any further implications” is not taken to mean “without paying attention to the internal structure” (thus looking at the situation as a whole, i.e. as total) but rather “without considering the situation in totality terms”, this description of the Imperfective is not incompatible with the totality view. What it does in that case, however, is presuppose some kind of neutralization of totality in the “simple denotative” Imperfective, which is a usual way of looking at it and is connected with the view of the Imperfective as the non-marked member of the aspect opposition (cf. the introduction). Also with the boundedness view one has to presuppose a neutralization to explain the use of the Imperfective in contexts of the kind discussed here. The view presented

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

235

The type-referring function of the Imperfective

235

in this paper, however, presupposes that every time the Imperfective is used, it is used by virtue of its imperfectivity and not of its non-markedness for aspect, that is, the Imperfective has a basic imperfective function also in those cases where its typical interpretations, such as continuity and indefinite repetition, are excluded.

3. Aspect and thematic structure In Slavic aspectology it is often pointed out (e.g., by Forsyth 1970 and Rassudova 1984) that there is a difference in thematic function connected with the two aspects. The Imperfective is often used when the verb phrase lies in the background of interest and some other constituent is focused on (time, place, agent). In the following Russian examples the situation is out of focus since it is the identity of the agent that is important. (19)

Russian (Forsyth 1970: 84) Poˇcemu zdes’ tak pyl’no? TY ubiral segodnja why here so dusty you clean:IPFV:PST today komnatu? room:ACC ‘Why is there so much dust here? Was it you who cleaned the room today?’

(20)

Russian (Rassudova 1984: 81) Krasivo ukrasili elku. Kto ukrašal? beautifully decorate:PFV:PST spruce:ACC who decorate:IPFV:PST ‘They decorated the Christmas tree beautifully. Who decorated it?’

There is, however, nothing that prevents the use of the Perfective in contexts where the verb phrase is out of focus (Rassudova 1984: 81). (21)

Russian èto on nauˇcil rebjat stroit’ Ved’ indeed it he teach:PFV:PST child:PL:ACC build:IPFV:INF ledjanye kreposti. ice:PL fortress:PL ‘Indeed, it was he who taught the children how to build ice castles.’

Likewise, in constructions with adverbials which evaluate the result of the action the Perfective is normally used, even if the agent is in focus.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

236

236

Eva Hedin

(22)

Russian (Rassudova 1984: 81) Kto tak cˇ isto ubral komnatu? who so cleanly clean:PFV:PST room:ACC ‘Who cleaned the room so well?’

The Modern Greek Imperfective may be used in a similar way. (23) a.

b.

(24)

Esí kathárizes to dhomátio símera? you clean:IPFV:PST:2SG DEF room today ‘Was it you who cleaned the room today?’ (according to some schedule =19) Pjos kathárise to dhomátio tóso kalá? who clean:PFV:PST DEF room so_much well ‘Who cleaned the room so well?’ (=22) –Me to kenúrjo aftokínito odhíjisan apó tin Athína with DEF new car drive:PFV:PST from DEF Athens os ti Thesaloníki se tris óres ke os tin Kavála se to DEF Thessaloniki in three hours and to DEF Kavalla in téseris óres. –Pjos odhighúse os ti Thesaloníki? four hours who drive:IPFV:PST to DEF Thessaloniki ‘– With the new car they drove from Athens to Thessaloniki in three hours and to Kavalla in four hours. – Who drove to Thessaloniki?’ (cf. 206)

In (19), (20), (23a) and (24) the question is not who actually cleaned the room, decorated the tree or drove to Thessaloniki but who was the performer of the actions in question, who “was cleaning/decorating/driving” in the sense “did the cleaning/decorating /driving”. The connection between the circumstances (time, place, agent) and the situation referred to by the verb is different when the situation is viewed non-temporally as a type and when it is viewed temporally as an instantiation. Somebody may for instance be referred to as the agent of some situation (referred to in the Imperfective) without any instantiation being considered: X is/was/will be the executor of (the) Verb-ing (of Y) (at t) X does/did/will do (the) Verb-ing (of Y) (at t) X is/was/will be (the) Verb-er (of Y) (at t) or as the agent involved in some instantiation of it (referred to in the Perfective): X Verb-ed / will Verb (Y) (at t)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

237

The type-referring function of the Imperfective

237

The interpretation that the situation is actually instantiated with the agent involved is of course close at hand in many contexts. From ‘He did/will do (the) room-cleaning’ (in many contexts) it follows that ‘He cleaned/will clean the room’. But this is not explicitly communicated since it is not the main message. The same goes for time qualifiers. There is a functional difference between the temporal qualifier of a situation referred to as a type and the temporal qualifying of some instantiation: At (or in) t there was / will be (a, the) Verb-ing (of Y by X) At (or in) t X Verb-ed / will Verb (Y) But here too the second may follow from the first in many contexts. From the Modern Greek example (25) which could be used, for instance, in a description of somebody’s life, and where we are told that 1945 was the year of somebody’s marriage, (26) follows, namely that he actually got married in 1945 (see below, 6.5, on type and token time). (25)

To 1945 pandrevótan. DEF marry:IPFV:PST ‘In 1945 he married.’

(26)

To 1945 pandréftike. DEF marry:PFV:PST ‘In 1945 he married / got married.’

In his description of the Imperfective Imperative in Modern Greek, Bakker (1965: 95) shows that the Imperfective is used when “the type of action is known” and all that is needed is a signal to perform it.7 (27)

–Na su kenóso to ghála su? SUBJ you:GEN pour:PFV:SUBJ:1SG DEF milk you:GEN –Óxi! –Ma eghó pinó. –Kalá lipón, na no but I be_hungry:PRS:1SG well then SUBJ su káno sindrofjá. Kénone! you:GEN make:IPFV/PFV:SUBJ:1SG company pour:IPFV:IMP ‘– Shall I pour your milk? – No! – But I’m hungry. – Well, all right then, I’ll keep you company. Go ahead! (lit. Pour!)’

The same is the case in Russian. Anikina (1963) (cited in Forsyth (1970: 199)) gives the following example. An engineer and his assistant have been preparing some new engine and the time has come to switch it on to see if it works. The assistant is prepared for the type of action to be executed and only needs to be told to ’do it’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

238

238 (28)

Eva Hedin

Russian Vkljuˇcajte! switch on:IPFV:IMP ‘Switch on!’

That focusing is relevant also for a description of Classical Greek aspect has been shown by Sicking (1991). Like Bakker (1966) he pays particular attention to the Imperative. As was mentioned above, it is obviously not just non-focused position as such that favours the Imperfective. In this kind of context the non-focused function seems to be due to the givenness of the situation, its being known from the linguistic or non-linguistic context. Now, if there is a connection between presupposition of a situation and the Imperfective, could this be explained on the basis of the view of the Imperfective as type-referring? In (27) it is clear to both speaker and hearer what type of action is to be performed. If in the same context the speaker used the Perfective Imperative as in (29): (29)

Kénose! pour:PFV:IMP ‘Pour!’

it would be as if he repeated the information of what is going to be done together with the order to perform it. Rather than giving the order ‘OK, go ahead–do it!’ he would give the order ‘Pour!’, by which he does not refer to ’pouring’ as something already present in one’s mind. Instead of ‘Execute (the) Verb-ing!’ the order would be ‘Verb!’ What is considered in orders like (27–28) is the known or presupposed situation (‘Do it!’) and what is known and presupposed is exactly the denotative content of the verb, that is, the situation type (Verb-ing) not any instantiation of it. Consequently, when the situation is considered as the presupposed situation (as ‘it’) it is naturally referred to as a type (Verb-ing). Naturally, this does not mean that the situation has to be referred to in this way when it is presupposed, only that if the situation is referred to as the presupposed situation as in (27)- (28), no actual instantiation is referred to and the Imperfective is used. If non-focusing, old information, givenness and presupposition are connected to the Imperfective, it may thus be in this indirect way via type reference.

4. Imperfectivity and negation In descriptions of Russian it has been claimed that negation is more characteristic of the Imperfective than of the Perfective. This could be compared to the tendency of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

239

The type-referring function of the Imperfective

239

the cross-linguistic gram type Experiential to occur in non-affirmative contexts, that is, questions and negated sentences (Dahl 1985: 142). If the Imperfective is used to refer to situations as types, it shares this characteristic with the Experiential and this may be what connects the two categories to negated context. The question is why the denial of a situation would favour the situation being referred to as a type rather than as a token. Why should situations be referred to as types when they don’t occur to a greater extent than when they do occur? Consider the following Russian examples: (30)

Russian pozvonil. Kto-to someone telephone:PFV:PST ‘Someone telephoned.’

(31)

Russian Nikto ne zvonil. no-one NEG telephone:IPFV:PST ‘No one telephoned.’

The parallel between aspect and case marking has been pointed out by Dahl & Karlsson (1976). They relate the predominance of the partitive (the case corresponding to the Imperfective) in negated sentences in Finnish and Baltic languages to the function of the Perfective to refer to some change in the state of the object referent. A negated sentence implies non-change, hence the use of the Imperfective (1976: 13). However, as is also pointed out, negation does not exclude perfective aspect. According to Forsyth (1970: 103), in Russian “there is a certain tendency to switch to the imperfective in negative statements in the past tense”. He claims, however, that the Perfective often occurs in negative sentences, including negated answers. Accordingly, the answer to the Russian question (32) can be either (33) or (34) (Forsyth 1970: 102): (32)

Russian Vy poluˇcili moe pis’mo? you:PL receive:PFV:PST:PL my:NEUT letter ‘Did you receive my letter?’

(33)

Russian Ne poluˇcil. NEG receive:PFV:PST ‘I did not (get your letter).’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

240

240 (34)

Eva Hedin

Russian Ne poluˇcal. NEG receive:IPFV:PST ‘I didn’t (get any letter).’

Consider also the following Modern Greek sentences. Both are negated but one has the Perfective and the other the Imperfective: (35)

Dhen ton píra tiléfono. NEG him:ACC take:PFV:PST:1SG telephone ‘I didn’t phone him.’

(36)

Dhen éperna tiléfono, jatí ... NEG take:IPFV:PST:1SG telephone because ‘I didn’t phone, because . . . ’ (‘didn’t make any phone call’)

It is obviously not the case that negation as such favours the Imperfective, but certain negated contexts do. Negation is always more or less connected with a presupposition (in some weaker sense of the term) (Givón 1978: 70; Leech 1974: 319). The denial of something presupposes some previous expectation of the opposite. This could be the link between the Imperfective and negated context. As was said above, presupposition is only secondarily linked to the Imperfective via type reference. Negation would then be a third step away: negation is connected with presupposition, which is connected with the Imperfective via type reference. A negated context for the Imperfective where the presupposition is even stronger could perhaps be described as the non-occurrence of an expected situation type. Consider the following example from Classical Greek: (37)

Classical Greek (Thuk. Hist. 8.80.1.3) auto i thérei metà to˜uto En dè to i in PRT DEF same:DAT summer:DAT after this:ACC epeid e hathróais euthùs hoi Peloponn
esioi, immediately DEF Peloponnesian:PL when collected:DAT:PL ... ta˜ıs nausìn . . . ouk antan
egonto DEF:DAT:PL ships:DAT:PL NEG attack:IPFV:PST ‘During the same summer, immediately after this, when the Peloponnesians, though their whole fleet had come together, failed to come out to meet the enemy . . . [they were perplexed, not knowing from what source they should get money to maintain so many ships . . . ].’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

241

The type-referring function of the Imperfective

241

The translation8 of the Imperfective antan
egonto ‘failed to come out’ expresses something else than just ’didn’t come out’ (which is, again, entailed). It rather tells us that the expected attack did not occur, that there wasn’t any attack (cf. Section 3 on “signal Imperfective” and Section 6.3 on the so-called inceptive Imperfective). A similar distinction may be made in English by the use of the adverbial never instead of plain negation. (38)

He never came/married (as expected/ natural)

(39)

He didn’t come/marry

The situation may also be known from the preceding context as in the following Classical Greek example: (40)

Classical Greek (Xenophon Anab. 7.1.7.2) [kaì ho Anaxíbios . . . hupiskhne˜ıto, ei diaba˜ıen misthoforàn ésesthai to˜ıs stratio
tais . . . Ek toútou diabaínousi pántes eis tò Buzántion hoi strati¯otai.] Kaì misthòn mèn ouk edídou ho Anaxíbios, and salary:ACC PRT NEG give:IPFV:PST DEF Anaxibios dè ... ek
eruxe proclaim:PFV:PST PRT ‘[and Anaxibios . . . promised that the soldiers would be paid a salary if they were shipped over . . . Then all the soldiers were shipped over to Byzantion.] Anaxibios didn’t pay (any) salary, but he proclaimed that . . . ’

A similar context for a negated Imperfective in Modern Greek is exemplified in (41) as contrasted with the one in (42). (41)

Dhe milúse. NEG speak:IPFV:PST ‘He did not speak.’

(42)

Dhe mílise. NEG speak:PFV:PST ‘He did not speak.’

Both sentences could be translated as ‘He didn’t say anything’, ‘He did not speak’. The Imperfective, however, somehow conveys the feeling of a negative state: He didn’t say anything = He remained silent. There was a state of silence. The Imperfective in (41) thus seems to have a stative function rather like in the positive state expressed by the verb sjopó ‘to be/become silent’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

242

242 (43)

Eva Hedin

Sjopúse. be_silent:IPFV:PST ‘He was/remained silent.’

In Bulgarian, as in Modern Greek, the Imperfective as well as the Perfective may be used in negative contexts of this kind (Lindstedt 1985: 204). (44)

Bulgarian D˘algo nikoj ne mu se obaždaše / for_a_long_time nobody NEG he:DAT RFL report:IPFV:IMPF / obadi. report:PFV:AOR ‘For a long time nobody contacted him.’

This use of the Imperfective would fit into a traditional description of it as expressing states etc. According to Rassudova, in Russian, “by using a verb with an appropriate lexical meaning in the Imperfective with negation, we can convey the complete absence of any changes”, that is “staticity” (1984: 89): (45)

Russian Passažir ne zagovarival. passenger NEG start talking:IPFV:PST ‘The passenger would not start talking.’

(46)

Russian Mašina ne ostanavlivalas’. car NEG stop:IPFV:PST:FEM ‘The car would not stop.’

According to the present description, however, this kind of negative state could be described as the non-occurrence of a situation type denoted by the verb. For instance, during some time stretch when one could expect the person to speak, he did not do so. (‘He did not execute (any) speaking’, ‘there was no speaking’.) The feeling of continuity that characterizes some expressions like this may in fact be conveyed by the temporal view of some time frame which is empty of the situation type referred to. The Perfective, on the other hand, just denies the statement that at some time there was an instantiation of speaking.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

243

The type-referring function of the Imperfective

243

5. Aspect usage If the Imperfective has the same function for instance in Slavic languages and Modern Greek, why is it not always used in the same way in the various languages? In the following English sentences, the verb phrase seems to have the function of referring to the occurrence of a situation type within a temporal frame, present (47) or past (48–49). (47)

Have you seen this film?

(48)

As a child, I once fell from that tree.9

(49)

Who opened the window (and closed it again)?

This is a typical context for the Imperfective in Russian, which would use the Imperfective in all three cases (Leinonen 1982: 184; Rassudova 1984: 71). Modern Greek, however, could use the Perfect and the Pluperfect, respectively. Bulgarian has the Imperfective (47), the Imperfective Aorist (48) and the Imperfective Perfect (49) (Lindstedt 1985: 216, 84). In his cross-linguistic study (1985: 143) Dahl observes that in his material there seems to be at least a partial overlapping between the contexts where the Experiential is used and those “where the Imperfective of some Slavic languages is used in the ‘general factual’ or ‘simple denotative’ sense”. He also notes (1985: 148–149) that the two languages in his material which seem to have a special Frame Past category use this about so-called “two-way actions”, exemplified in (49) above, a classic context for the Russian Imperfective. It should be noted that the Modern Greek alternative for this sentence, the Pluperfect, is often used as a kind of Frame Past (Hedin 1987: 23–28). Likewise, the overlap between the Imperfective and the experiential use of the Perfect is illustrated by the different ways of treating example (47) above, exhibited by the three languages referred to. The common factor making these different categories (Experiential, Frame Past and the Perfect categories) overlap with the Imperfective could be type reference, which in turn has a natural (although not necessary) connection with temporal frames.10 One reason for varying usage of the Imperfective across languages could be systemic differences of this kind. The development of other tense aspect categories (like the Perfect system) may contribute to a distributional shift within the tense aspect systems due to functional competition between the categories. Another reason for differences in aspect usage may be “aspect competition”. There are more and less “typical” contexts for the Imperfective. The more subtle the difference is between type and token reference to a situation in a particular context, the sooner this context may promote competition between the two aspects. This competition may operate within a language (both aspects may occur in a particular

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

244

244

Eva Hedin

context without any notable difference in meaning) or across languages (in a particular context, one language has one aspect and another language has the other). It was noted above (Section 2.5) that boundedness seems to be crucial for the choice of aspect in Slavic languages. However, the fact that boundedness seems to play a more important role in Slavic languages than in most others with a Perfective:Imperfective opposition does not mean that boundedness has to be the distinction reflected by the two aspects in these languages. (We have already seen in Section 2.5 that boundedness does not exclude the Imperfective.) The use of the Imperfective with non-bounded verb phrases like in Q.13 (He WRITE letters, cf. 2.5) could be an instance of a stronger tendency in Slavic languages to exploit the type and token focusing of verbal aspect to mark distinctions that may be marked by other means in other languages (by a different noun phrase structure, by the use of articles and pronouns to denote specificity of the object, etc.). An investigation of such factors might explain some differences in aspect usage and shed some light on the question why the Imperfective has a wider use in Russian and Classical Greek than in Modern Greek and why the Russian and the Classical Greek Imperfectives are more commonly used for single bounded events than their Modern Greek counterpart. Thus, the fact that languages with perfective:imperfective aspect do not always use aspect distinctions in the same way does not necessarily contradict a claim that the basic function of the different imperfective categories is the same. It is, however, an important task to try to find out what factors could contribute to the variation in distribution across languages and within a language system diachronically.

5.1. A reinterpretation of some uses of the Imperfective In the remainder of the paper, I am going to discuss some of the main uses of the Imperfective that are referred to in descriptions of the Imperfective in languages like Classical and Modern Greek, Russian and Bulgarian, proposing a reinterpretation of these functions on the basis of type reference. It should be noted that the different headings of the chapters below are not to be understood as describing different functions of the Imperfective, but rather as traditional labels of the uses described.

5.2. ‘Habitual’ – ‘continuative’ – ‘frequential’ The Imperfective is often said to have two main functions, namely the habitual and the continuative. In descriptions of the Imperfective, these two functions are often regarded as basic and are also referred to when other functions that do not really fit into this description are discussed. Thus, in many contexts where the Imperfective does not explicitly – or even by means of the most natural interpretation – refer to a process, a processual interpretation is forced on it, in order to legalize its existence.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

245

The type-referring function of the Imperfective

245

The habitual use of the Imperfective has led to the conclusion that repetition is part of the semantics of the Imperfective. It should be noted, however, that there is a difference between repeated or iterated situations as such and situations referred to as parts of habits and schedules (cf. Dahl 1985: 97, where it is pointed out that iterativity is almost irrelevant to the cross-linguistic category Habitual). Mere repetition does not call for the Imperfective. In Modern Greek, for instance, explicit reference to a definite number of times is a typical context for the Perfective. (50)

Píje ekí pénde forés. go:PFV:PST there five times ‘He went there five times.’

The Imperfective is, however, used in contexts expressing frequency. (51)

Píjene ekí pénde forés tin evdhomádha. go:IPFV:PST there five times DEF week ‘He went there five times a week.’

As was noted above (2.3–2.4), according to the view presented in this paper, the habitual as well as the continuative use would be based on the same basic function, namely type reference. As for frequency contexts, the picture is not as clear. In contexts with explicit frequency expressions (once a year, five times per week, often, regularly etc.) type reference is natural, since no instantiation may possibly be focused on and referred to as token, only some situation type occurring with a certain frequency. The Imperfective used with an explicit frequency expression as in (51) thus comes close to the habitual Imperfective as in (52). (52)

Píjene ekí mazí me ti María. go:IPFV:PST there together with DEF Mary ‘He used to go there with Mary.’

Time expressions with every, however, (every year, five times every week, every Sunday, two times every year etc.) differ from the explicit frequency expressions, since they may sometimes be ambiguous in scope. Consider example (53): (53) a. b.

He went to the cinema five times [every week]. He went to the cinema [five times every week].

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

246

246

Eva Hedin

This sentence may be comprehended in two ways, either as ‘Every week he went to the cinema five times’ or as ‘He went to the cinema as often as five times a week’. In the first (but not in the second) reading, it is possible to comprehend every week as a temporal frame, within which something happened a definite number of times, that is, as a typical context for the Perfective. Habitual context is also one where languages differ in aspect use (Dahl 1985: 79; Štunová 1993: 17). Some Slavic languages, like Russian and Bulgarian, prefer the Imperfective, others, like Czech, the Perfective. Modern Greek has the Imperfective in all habitual cases. Thus, (54) corresponds to both (53a) and (53b): (54)

Píjene sinemá pénde forés káthe evdhomádha. go:IPFV:PST cinema five times every week ‘He went to the cinema five times every week/ a week.’

5.3. ‘Conative’ Examples of the use of the Imperfective called ‘conative’ are easily found in Classical Greek, Russian and Bulgarian and other aspect languages. In grammars, verbs meaning ‘persuade’ often occur under this heading: (55)

Classical Greek (Xenophon, Cyr. 5.5.22) [Ouko˜un toútou tukho n parà so˜u oudèn  en, ei m e toútous peísaimi.] o˜un épeithon autoùs kaì Eltho n come:PFV:PART thus convince:IPFV:PST:1SG them:ACC and hoùs épeisa, toútous those_whom:ACC convince:PFV:PST:1SG those:ACC ékh¯on eporeuóm¯en ... have:IPFV:PART march:IPFV:PST:1SG ‘[Well then, when I had obtained this concession from you, it amounted to nothing, unless I could get their consent.] So I went to see if I could get their consent; and those whom I persuaded I took with me . . . on my expedition.’11

(56)

Russian (Comrie 1976: 19) On dolgo ugovarival menja, no ne he long persuade:IPFV:PST me but NEG ugovoril. persuade:PFV:PST ‘He persuaded me for a long time, but didn’t persuade me’ i.e., ‘he spent a long time persuading me but didn’t actually persuade me.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

247

The type-referring function of the Imperfective

(57)

247

Bulgarian Te ne uspjaxa da ja ubedjat they not succeed:PFV:AOR:3PL COMP her:ACC convince:PFV da zamine makar ce dalgo vreme COMP go_away:PFV:PRS:3SG although COMP long time ja ubeždavaxa. her:ACC convince:IPFV:IMPF:3SG ‘They didn’t succeed in persuading her to go away with them, although they spent a long time trying to.’

When translated to English and other languages lacking the distinction between imperfective and perfective aspect, the Imperfective of these verbs often calls for a translation with try if one wants to keep the correspondent lexeme persuade. This seems to be typical in a context like the one in (55) and (56) above, where the Imperfective is followed by a Perfective of the same verb, which consequently is translated by manage to persuade or something similar. However, this cannot be considered a special function of the imperfective aspect. According to Schwyzer (1950: 259) a translation with try or something similar, which is necessary when translating from Classical Greek to some other languages, says more than the original does. The Classical Greek is not so precise, since the right interpretation is given by the context. On this point Modern Greek seems to differ in aspect usage from the other languages. The imperfective form of the verb pítho cannot be used in this kind of context conveying the meaning ‘try to persuade’. A construction with a verb meaning ‘try’ would also be needed in Modern Greek. (58)

Prospathúse na me písi epí dhío try:IPFV:PST SUBJ me:ACC convince:PFV:SUBJ for two óres, alá dhe me épise. hours but NEG me:ACC convince:PFV:PST ‘He tried to convince me for two hours, but he didn’t (manage to) convince me.’

The so-called conative use of the Imperfective in some aspect languages is a good illustration of the neutral function of the Imperfective. The irrelevance of the instantiation of the situation means that the actual accomplishment of the transitional phase of the situation (for instance, that somebody was actually persuaded) is neither stated nor denied. If the sentence is followed by another one, where the success is denied by reference to the situation in the negated Perfective (‘but he didn’t (actually) persuade him’), the interpretation of the first verb phrase in the Imperfective as referring to an attempt to persuade is the only logically possible one. However, this interpretation is created by the context and not by the imperfective verb phrase, which only

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

248

248

Eva Hedin

tells us that ‘there was persuasion’, somebody was the agent of the situation type ‘persuading’. If one wants to explain the Imperfective in sentences like those exemplified above by some function of “viewing the situation from within” or as expressing “noncompletion” (that is, according to a more traditional view of the Imperfective), it is difficult to explain why in other contexts, the same form must in fact be interpreted in the opposite way:

(59)

Classical Greek (Xenophon, Hell. 6.5.19.5)12 [ ] épeithon m e poie˜ısthai mákh¯en, persuade:IPFV:PST:3PL NEG make:IPFV:INF fight:ACC prìn hoi Th¯eba˜ıoi paragénointo. before DEF Theban:PL be_present:PFV:OPT:3PL [The Mantineans, however, now desisted from coming forth from their city, for the Eleans . . . ] urged them not to fight a battle until the Thebans arrived.

In (59) the context tells us that the persuasion referred to in the Imperfective is successful (the Mantineans obeyed and remained quiet). In (60) and (61) the success of the persuasion is explicitly underlined by the adverbials meaning ‘readily’, ‘without difficulty’.

(60)

Classical Greek . . . eupeté¯os épeithe tòn Dare˜ıon. easily convince:IPFV:PST DEF Darius:ACC ‘. . . he easily convinced Darius.’

(61)

Classical Greek (examples from Rijksbaron 1984: 18)13 . . . kaì toùs akoúontas ou khalepo
s and DEF listen:IPFV:PRT:ACC:PL NEG with_difficulty épeithon. convince:IPFV:PST:3PL ‘. . . and they had no difficulties in persuading their hearers.’

Likewise, the negated Imperfective in example (62) does not say that he did not try to persuade but that he did not succeed in doing so.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

249

The type-referring function of the Imperfective

(62)

249

Classical Greek (Xenophon, Hell. 5.2.36.2) [Ho dè apeloge˜ıto mèn pròs pánta ta˜uta,] ou méntoi épeithé ge tò m e ou NEG however persuade:IPFV:PST PRT DEF NEG megaloprágm¯on te kaì kakoprágm¯on e˜ınai. aiming_high and mischievous be:IPFV:INF ‘[To all these charges he did indeed make a defence,] but he failed to persuade the court that he was not a man of great and evil undertakings.’

A search of the works of Xenophon in the Thesaurus Linguae Graecae for forms beginning with epeith- (which is the first part of an imperfective past form) gave 37 matches in all. Fifteen were active and 22 mediopassive forms (including two with habitual function). Of the active forms, 8 had to be interpreted as referring to unsuccessful persuasion (that is, as conative), but in at least 5 cases, the context tells us that the persuasion actually succeeded. In the medio-passive form, in fact, most of the forms had to be interpreted as referring to the result of the persuasion. Either somebody is successfully persuaded or convinced, which means that he obeys, as in (63), which cannot possibly have a conative interpretation: (63)

Classical Greek (Xenophon, Hell. 5.2.31.2) eidótes tò prãgma Hoi mèn d e DEF PRT PRT know:PART:PL DEF thing te kaì epeíthonto kaì par esán be_present:IPFV:PST:3PL and persuade:IPFV:MID/PASS:3PL and synelámbanon. seize:IPFV:PST:3PL ‘Now those who knew of the plan were of course present, obeyed the order and seized Ismenias; . . . ’

or else somebody is unsuccessfully persuaded, which means that he does not obey, as in (64). This likewise cannot be interpreted as one did not try to persuade it (namely the city to yield), only as one did not succeed in persuading it. (64)

Classical Greek (Xenophon, Hell. 3.1.7.3) epeì ouk epeítheto when NEG persuade:IPFV:PST:MID/PASS ‘when it refused to yield’

Like the conative interpretation, however, the opposite interpretation of successful persuasion is not conveyed by the imperfective form as such. According to the view presented here, when the Imperfective is used in contexts like this, it is not used by virtue of some function of expressing non-completion (or continuity) in order to

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

250

250

Eva Hedin

convey the interpretation of attempt. It is used – like in other contexts, where it is interpreted differently – by virtue of its neutral type-referring function, where the actual result of the situation is irrelevant, thus permitting different interpretations.

5.4. ‘Inceptive’ In descriptions of Classical Greek the Imperfect is sometimes said to have an inchoative, ingressive or inceptive use. Smyth (1920: 426) gives the following example:

(65)

Classical Greek (Thuk. 7.51.) kairòs  en, t ei mèn protér¯ai pròs epeid e dè when PRT time be:IPFV:PST DEF PRT first toward proséballon ... tà teíkh¯e to n Ath¯enaí¯on DEF walls DEF Athenians:GEN assault:IPFV:PST:3PL ‘When the favourable moment came, they assaulted on the first day the Athenian walls . . . ’

He explains the Imperfective as expressing the “starting of the action” and translates it as ‘when the proper time arrived, they began an (proceeded to) attack’. The context is the following: ‘Accordingly, they regularly manned their ships and practised for as many days as they thought sufficient. Then, when the favourable moment came, they assaulted on the first day the Athenian walls.’14 The expression proceed to used (within parenthesis) in Smyth’s translation is significant. There is a difference between saying that they began to attack and saying that they proceeded to attack. Just as the superficially inchoative Imperative (27) Kénone! referred to above does not really mean ‘Start pouring!’ but rather ‘Go ahead (pour)!’, where the speaker gives a signal to the listener to take action, to execute the situation type given from the context (namely pouring milk), the Imperfective in (65) does not really mean ’started to attack’ but rather that at a favourable moment they took action and ’executed the attack’ (the given situation). According to Bakker’s (1965: 96) definition of the Modern Greek Imperfective Imperative, “one of the most striking characteristics of the present (i.e. imperfective E.H.) imperative is [ ] that the action ordered has to start immediately”. The speaker sees the present situation at which he pronounces his order as “the startingpoint of the action”.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

251

The type-referring function of the Imperfective

251

(66)

[Elá fáe tóra pu íse nistikí apó to proí. Na, to avghó su xtipiméno.] Ríkse ghála ke tróje! pour:PFV:IMP milk and eat:IPFV:IMP ‘[Come, eat first now. You haven’t had any breakfast yet. Here is your beaten egg.] Pour your milk now and eat!’

(67)

Ánikse ta xartjá, ghráfe! open:PFV:IMP DEF papers write:IPFV:IMP ‘Unfold your papers and start writing!’

A similar standpoint is taken by Ruijgh (1985 and 1991) for Classical Greek: “Dans l’emploi inceptif de l’impératif présent, le locuteur admet au général qu’immédiament après ses paroles, la personne à laquelle il s’adresse va commencer à réaliser l’action en question (1991: 206).” He gives the following example. The speaker addresses the scribe telling him to read the law: (68)

Classical Greek (Demostenes 24,32) toutonì pro ton Anágn¯othi dé moi labo n take:PFV:PART this first read:PFV:IMP PRT me:DAT tòn nómon DEF:ACC law:ACC [hòs diarr
ed¯en ouk e ai nómon oudéna enantíon eisphérein, eàn dé tis eisphér¯ei, gr´phesthai keleúei]. Anagígn¯oske. read:IPFV:IMP ‘But first take and read the statute I have here [which expressly forbids the introduction of any conflicting law, and authorizes an indictment if such a law should have been introduced]. Read!’15

Sicking (1991: 160) objects to the idea that the situation should start immediately. But he agrees with Bakker’s (1966) view that the Classical Greek Imperfective Imperative may be chosen because the addressee knows what action is expected from him, which is compatible with his view that focusing plays a part in the choice of aspect. However, Bakker’s (1965) own translations and comments to his Modern Greek examples (66–67) demonstrate his intuition that these examples are not really those of an inchoative or ingressive situation. In his comment on the first example (66), he points out that instead of ’eat’ one might as well say ‘start (eating)’, since the kind of action the addressee is expected to perform (namely to eat) is already known to her. By putting ‘eating’ within parenthesis Bakker demonstrates that the element of commencement is not really connected to the verb phrase (making it ingressive),

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

252

252

Eva Hedin

but rather is due to a function similar to the signal function referred to above (in Section 3). It is more like pointing out the moment for the execution of the situation: ‘and (then) go ahead (and do it)’. Rather than ‘start eating’ it says ‘execute (the eating)’. In all these cases, however, the interpretation that the situation referred to actually should start or actually starts lies close at hand. To tell somebody to execute eating pragmatically implies that he has to start eating and to say that they executed attacking at some moment implies that they actually started attacking at that time. Very similar to Bakker’s idea of immediate inception at the time of utterance is that of Rassudova (1984: 135). She refers to a meaning of commencement connected directly with the moment of speaking in the Russian imperfective infinitives and gives the following examples: (69)

Russian Možno nalivat’ cˇ aj? possible pour:IPFV tea ‘May I (go ahead and) pour the tea?’

(70)

Russian Možno vam nalit’ cˇ aj? possible you:DAT pour:PFV:INF tea ‘May I pour you some tea?’ (Or do you prefer coffee?)

Under this heading she also mentions the use of the Imperfective with expressions that together with the Infinitive render the meaning ’it is time (to do something)’. With the expression ’it is time’ itself, she notes, both aspects are possible. But when it has its strict temporal meaning, denoting the time for something, only the Imperfective is used (cf. section 6.5 on time point expressions). (71)

Russian Pora zakanˇcivat’ rabotu. time finish:IPFV:INF work:ACC ‘It’s time to finish the work.’

Modern Greek has a parallel use of the Imperfective in the Subjunctive: (72)

Íne óra na teljónume / teljósume. be:PRS time SUBJ finish:IPFV:SUBJ:1PL / finish:PFV:SUBJ:1PL ‘It’s time to finish.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

253

The type-referring function of the Imperfective

253

5.5. Potentiality – modality It was said above (in 2.4) that in examples like O Jánis kapnízi/kápnize ‘John smokes/ smoked’, which is a typical habitual use of the Imperfective, the verb does not really refer to any instantiation of smoking at all. It is used in a qualifying sense to say that ’John is/was a smoker’. If, however, John never smoked in his life, it would be difficult to find an interpretation for the sentence. There must be some temporal domain where John exists which contains at least one instance of smoking with John as the agent for the habitual interpretation to be possible. In other cases, not even this is necessary in order to find an interpretation for the Imperfective. In the Modern Greek example (73) the knife referred to may never have been used to cut anything at all. (73)

To maxéri kóvi / ékove. DEF knife cut:(IPFV:)PRS / cut:IPFV:PST ‘The knife is/was sharp.’

This is sometimes called the potential use of the Imperfective (e.g., Mackridge 1985: 108). In fact, it comes close both to the habitual and the generic Imperfective. Consider the following Modern Greek examples: (74)

To aftokínito tréxi / étrexe polí. DEF car run:(IPFV):PRS / run:IPFV:PST much ‘The car is/was fast.’

(75)

To skilí dhangóni / dhángone. DEF dog bite:(IPFV):PRS / bite:IPFV:PST ‘The dog will/would bite.’

(76)

To skilí dhe ghavjízi / ghávjize. DEF dog NEG bark:(IPFV):PRS / bark:IPFV:PST ‘The dog does/did not bark.’ (about a dog of a particular breed, which does not bark)

In some contexts, the potential use may express different kinds of modality such as possibility, ability or obligation. (77)

To kréas dhen trójete / troghótan. DEF meat NEG eat:(IPFV):PRS:PASS / eat:IPFV:PST:PASS ‘The meat isn’t/wasn’t edible.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

254

254

Eva Hedin

(78)

Xtízi / éxtize éna spíti se dhío build:(IPFV):PRS / build:IPFV:PST INDEF house in two méres. days ‘He can/could build a house in two days.’

(79)

Plénete / plenótan sto xéri. wash:(IPFV):PRS:PASS / wash:IPFV:PST:PASS in_DEF hand ‘(This item) should be hand-washed.’

The borderlines between habitual, generic, potential and other uses of the Imperfective are not very clear. What they have in common, however, is the qualifying character. No instantiation of a situation is considered. Rather, some entity is characterized in one way or other by being connected to some situation, which is thus naturally viewed non-temporally as a situation type. In Russian, there are some similar contexts where the Perfective is used. This has been referred to as the graphic-exemplary use of the Russian Perfective. (80)

Russian (Rassudova 1984: 114) Obratites’ k moemu bratu, on vam vsegda turn to my:DAT brother:DAT he you:DAT always pomožet. help:PFV:PRS ‘Turn to my brother; he’ll always help you.’

It should be noted, however, that this use of the Perfective is limited to contexts with present-time reference. Since the Russian Perfective Present is used as a Future, (80) could be interpreted as corresponding to the use of the Future in, for example, English (cf. the translation). The overlapping of the Present and the Future in contexts of this kind is natural, since the situation referred to may be seen in a more or less general perspective, as exemplified in (81) and (82). (81)

If you ask him, he’ll help you.

(82)

He (is such a person who) helps people.

The Perfective would thus correspond to the first perspective, the Imperfective to the second.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

255

The type-referring function of the Imperfective

255

5.6. Schedule and time point expressions–type and token time The situation referred to in schedule expressions like (83) may also be described as a potential situation. (83)

To tréno févji / éfevje se pénde DEF train leave:(IPFV):PRS / leave:IPFV:PST in five leptá. minutes ‘The train leaves/was leaving in five minutes.’

It is not the actual leaving (the situation token) taking place five minutes later (if it ever takes place) that is predicted. What is communicated is that five minutes later is/was the time for departing (the situation type). The scheduling use sometimes comes close to the habitual use. This is the case in (84), which, in fact, is ambiguous. (84)

To kalokéri to proinó tréno tha févji pjo norís DEF summer DEF morning- train FUT leave:PFV more early ‘In the summer the morning train will leave/be leaving earlier.’

The train will be leaving earlier either according to some summer schedule, which will be valid, or just because the lazy conductor will manage to get up in time during the summer. As was noted in Section 2.4, a specific time point may qualify a situation although the situation itself is not referred to as a unique instantiation. This is exemplified in the schedule cases. In five minutes in (83) denotes a specific time point, but it does not denote the time point for some specific instantiation of leaving (by some specific train at the time point t), only the time point when – according to some schedule – ‘there should be departing’. When the temporal qualifier has the first function it could be said to denote “token time” (time for a situation token) and when it has the second function “type time” (time for a situation type – represented by a situation token or not). This differentiation is important in explaining examples like the following, where the situation is not a potential one as in the schedule cases but actually did occur at some time (cf. (25) repeated as (85) below):16 (85)

To 1945 pandrevótan. DEF marry:IPFV:PST ‘In 1945 he married.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

256

256

Eva Hedin

(86)

Sta íkosi mu xrónja spúdhaza vévea in_DEF twenty me:GEN years study:IPFV:PST naturally akóma, sta ikosiéna éperna to ptixío still in_DEF twenty-one take:IPFV:PST DEF diploma mu, ke sta ikosidhío éfevgha ja ti me:GEN and in_DEF twenty-two leave:IPFV:PST for DEF Mési Anatolí. Middle East ‘In my twenties I was still studying of course, when I was twenty-one I graduated and when I was twenty-two I left for the Middle East.’

(87)

Sarandapénde xrónja prin, séna nosokomío, ksepsixúse forty-five years ago in_INDEF hospital die:IPFV:PST énas Ískios – árostos ke perífovos ke mónos. INDEF Shadow sick and frightened and alone ‘In a hospital, forty-five years ago, a Shadow passed away – sick and frightened and alone.’

Although the temporal modifiers refer to some specific time, the situations denoted by the imperfective verb forms are not considered as the specific instantiations occurring at that time. It is not so much a question of what actually concretely took place at this specific time in the past (the token, the instantiation as such) as of pointing out the time for something (somebody’s marriage, graduation, departure or death). The situation referred to is thus viewed in a more abstract way as a whole (as ‘marrying’, ‘graduating’, ‘departing’ or ‘dying’; cf. the discussion in 2.5), as a non-temporal situation type. (Cf. also section 3 above on focusing on the circumstances instead of the verb.)17

5.7. Imperfective with verbs of communication In the literature on Russian and Classical Greek it has been pointed out that verbs of saying are often used with the Imperfective in contexts where one might expect the Perfective. This is to some extent also true of Modern Greek, where verbs denoting communication in some contexts occur in the Imperfective:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

257

The type-referring function of the Imperfective

257

(88)

[“Makári na katápina ti ghlósa mu!’] Aftó éleje xthes stin Peanía o X. pu that say:IPFV:PST yesterday in_DEF Peanía DEF COMP ipostírize tin proighúmeni Paraskeví óti ... claim:IPFV:PST DEF last Friday COMP [‘If only I had swallowed my tongue!’] ‘X said this yesterday in Peanía, he who last Friday claimed that . . . ’

(89)

[“Néi ánthropi, nées idhées ke thavmastá apotelésmata”,] sxóljaze xthes athlitikós parághondas, comment:IPFV:PST yesterday athletic representative [anaferómenos sto proxthesinó dérbi ton eoníon andipálon, Olimpiakú– Panathinaikú]. ‘[“New people, new ideas and wonderful results”,] an athletic representative commented yesterday, [referring to the derby the other day between the eternal antagonists Olympiakou–Panathinaikou.]’

Again, what is central is not that which actually concretely took place at the time of the utterance in question – the instantiation, somebody said something or commented on something at some point in time–but what (non-temporal situation type – saying, commenting, uttering, claiming, etc.) was instantiated (without focusing on the instantiation as such). Rather than ‘yesterday X said such and such’, what is communicated is something like ‘yesterday there was saying such and such by X’ or ‘X was the agent of making such and such an utterance’. Compare also the following sentences. In (90) the speaker wonders about the source of information (without considering any instantiation of saying), whereas (91) could be uttered by a school teacher who wants to know who in the class (actually, a few seconds ago) uttered the nasty comment he just heard when writing on the blackboard. (90)

Pjos (to) éleje? who (it:ACC) say:IPFV:PST ‘Who said (that)?’

(91)

Pjos to ípe aftó? who it:ACC say:PFV:PST that:ACC ‘Who said that?’

The fact that verbs of saying often occur in the Imperfective may partly be due to its possibility to denote two sides of the utterance, namely the concrete act of speaking or writing on one hand, and its “performative” function on the other. One example is the difference in interpretation of the verb say, which apart from its more concrete meaning of uttering, may have the meaning ‘tell’, ‘inform’, ‘express an opinion’,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

258

258

Eva Hedin

‘maintain’, ‘claim’ etc. (by speaking, signing or writing). In some contexts, where the content of the utterance and its source is what is important, not the particular instantiation of uttering, type-focusing instead of token-focusing of the situation may be a way to underline this more “abstract” meaning by shifting focus away from the “concrete” token.

5.8. Scripts and scenarios In Slavic aspectology it has been pointed out (Vinogradov 1947: 558, cited in Leinonen 1982: 195) that there is a connection between normality or usualness of a situation and the Imperfective in verb phrases like ‘to have breakfast’, ‘to have dinner’, ‘to have tea’. (92)

Russian My zavtrakali v vosem’ cˇ asov. we have_breakfast:IPFV:PST:PL at eight hour:PL:GEN ‘We had breakfast at eight.’

Leinonen (1982: 195) refers to this kind of expressions as representing a script, i.e., “a predetermined, stereotyped sequence of actions that define a well known situation” (Schank & Abelson 1977, cited by Leinonen 1982: 196). She considers the script to be a variant of the scenario structure of the kind discussed in Newton (1979), where it is argued that “the basic schematic form of one common class of multiple-event expressions” may be paraphrased ‘Every time p, then q’. He uses the term “scenario” to refer to such “statements of constant conjunctions of occasions” (1979: 139). A scenario structure may be overt as in (93) or covert as in (94). (93)

Every time he sees me, he insults me.

(94)

John always drinks coffee.

In an overt scenario expression the antecedent is specified whereas in a covert one it is only implied. The antecedent for (94) would be something like ’whenever he drinks anything (non-alcoholic)’ (Newton 1979: 140). The link between scripts and scenarios would according to Leinonen be repetition, which is a condition for stereotyping. However, the scenario structures describe “one class of multiple-event expressions”, namely the habitual use of the Imperfective, and I find it hard to see how a script expression like had breakfast in (92) could fit into this class. That a situation is frequently repeated may be a condition on the world for the situation to be comprehended as a stereotype, but repetition is not what is expressed by the stereotyped phrase.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

259

The type-referring function of the Imperfective

259

The connection between script expressions of the kind referred to above and the Imperfective could be explained by the type-referring function of the Imperfective. When referring, for instance, to somebody’s eating some food in the morning we may say either He greedily ate three portions of bacon and eggs or He had breakfast and somebody’s drinking tea in the afternoon we may refer to as He carefully sipped a hot cup of tea or He had tea speaking about the same situation. In the first case it would typically be the actual situation (the instantiation in time of eating some food or drinking some tea) that is referred to. In the second the situation is viewed as type and the breakfast and the tea is referred to as non-specific, only defining the type of situation (as ‘breakfasting’ or ‘tea-drinking’). In these examples the Russian imperfective verb sometimes corresponds to the English lexical construction with have instead of eat or drink etc. Consider the following example from Mehlig (1979: 156, cited by Leinonen 1982: 165): (95)

Russian Segodnja v sem’ cˇ asov Ivan prinimal lekarstvo. today at seven hour:PL:GEN Ivan take:IPFV:PST medicine ‘Today at seven Ivan took his medicine.’

According to Mehlig, the adverbial does not have the function of localizing the situation, rather it is part of the proposition. He means that it is predicated about Ivan that he normally takes his medicine at seven o’clock. Again, I find it hard to see that ’Ivan’s medication at seven’ is referred to as a habit. However, by referring to the situation as a stereotype, using the Imperfective, one may infer that it is part of a habit, something that is normally done (in this case by Ivan). (92) and (95) could also be given as examples of time point Imperfective discussed above, the time points giving type-time rather than token-time. (The normal) breakfasting/medicating was executed at eight/seven. Thus, the reason why the Imperfective is used for stereotype situations would be that these are referred to as situation types. In fact, it could be argued that, in some sense, stereotype situations are exactly what the Imperfective always refers to. To refer to a situation as a static, non-temporal type means to consider it as a non-temporal, invariant whole, that is, as a stereotype.

6. Summary In this paper it has been proposed that the functional difference between the Perfective and the Imperfective could be described in terms of type and token reference. With the Imperfective, situations are considered in a static, non-temporal perspective as types (neglecting any instantiation), as opposed to the Perfective, which is used

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

260

260

Eva Hedin

for reference to situations as tokens or instantiations occurring in time (in a temporal, “left-to-right” perspective). Some typical uses of the Imperfective have been described and reinterpreted against the background of its type-referring function. Not only in non-specific contexts as in the generic, habitual or potential uses (where no specific situation is focused on) may the Imperfective refer to the situation as a type, but also in contexts where it is possible to focus referentially on some situation and this situation consequently could be referred to in both ways. Thus, according to the description proposed here, in a sentence like O Jánis xtízi/éxtize éna spíti ‘John is/was building a house’ (that is, an example of the frequent continuative use of the Imperfective), the situation is referred to as a nontemporal situation type, as ‘house-building’, not as an instantiation of it in time, as some house actually coming into existence at the reference time or later. Presupposition seems to be important for the choice of the Imperfective in Modern and Classical Greek as well as in Russian for instance. When the situation is known or given by the context, what is given is exactly the type of situation, the denotative content of the verb phrase. If this situation is referred to as the presupposed situation, it is naturally the situation as type that is considered. The connection between negation and the Imperfective which has been observed may be a secondary one depending on this link between presupposition and the Imperfective. The ingressive and the signal uses of the Imperfective may be described as referring to a situation as an execution of some situation type typically given by the context (for instance, as ‘Execute (the) Verb-ing!’ rather than ‘Verb!’). This may naturally give rise to an ingressive interpretation. The neutral function of type reference of neglecting the instantiation of the situation (for instance neither stating nor denying its actual accomplishment) explains why the conative interpretation may occur with many transitional verb phrases in the Imperfective. That a conative interpretation is not the only possible reading of these imperfective verb forms, however, is shown by the fact that the same forms in other contexts may have the opposite interpretation, viz. that of completion. The schedule and time point Imperfectives are used for reference to time points as the time for some abstract situation type (referred to as type time as opposed to token time), not to the actual concrete instantiation at this time point (in case there is one). Verbs of communication sometimes occur in the Imperfective “pro Perfective”. This is connected with the more abstract function that some verbs of communication may have apart from referring to some concrete act of uttering. This abstract function would favour type-focusing reference to the act of communication, where a concrete token-focusing interpretation is avoided by using the Imperfective. Finally, the use of the Imperfective with phrases of the type ‘have tea’, ‘have breakfast’, well-known from Slavic aspectology, is discussed. It is connected to the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

261

The type-referring function of the Imperfective

261

function of the Imperfective as type-referring and thus in some sense always referring to situations as stereotypes.

Acknowledgment Financial support for the work presented has been received from the Swedish Council for Research in the Humanities and Social Sciences (HSFR).

Notes 1. For a definition of the term “bounded”, see 2.2. 2. For clarity, the term “Modern Greek” is used throughout this paper instead of just “Greek” as in the rest of the volume. 3. Forsyth (1970: 82) claims that the simple denotative function use reflects “the most basic and general function of the imperfective verb: to identify the type of action, naming it without reference to the question of its ‘perfectivity’ or otherwise”. Although the first part of this description looks very much like the one proposed in this paper, the second part makes it differ crucially from it. Thus, according to Forsyth, when the Imperfective is used with the simple denotative function, it has a function which, although basic, is distinct from its function in other contexts (where it is non-perfective). Although he considers the function of the Imperfective to “identify the type of action” as basic, he does not consider it as the function of the Imperfective as opposed to the Perfective. Perfectivity and non-perfectivity are treated like another kind of opposition, which is neutralized in the simple denotative function of the Imperfective. 4. Examples are from Modern Greek if there is no indication of language. 5. In some other (extralinguistic) context (TMAQ: 14) could get the Imperfective in the answer provided the sense of the question is not ‘What happened after dinner?’ but ‘Which activity did he choose to occupy himself with after dinner’ having e.g. earlier discussed different possible alternatives (such as wash the dishes, make a phone call or write a letter). 6. In the Greek correspondence to (20) the perfective would be used: Pjos to stólise; Who it decorate:PFV:PST It should be noted, however, that the Greek example, like the English translation of (20), contains an explicit object. This syntactic structure, which is the only natural one in Greek, would, according to my informants, allow both aspects in the Russian example. 7. According to Bakker the situation referred to also has to start immediately. According to his description the givenness of the type of action is thus one of the factors determining the aspect choice in this kind of context, immediacy another (cf. 6.3). 8. Translation by C. F. Smith (Loeb Classical Library no. 169 [1976]. W. Heinemann Ltd: London & Harvard University Press: Cambridge, Massachusetts.).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

262

262

Eva Hedin

9. The example is from Lindstedt (1985). Rassudova notes that the context with an adverb meaning ‘once’ is a very favourable context for the “general-factual” meaning (Rassudova 1984: 71) of the Imperfective in Russian. It is also a typical context for the Modern Greek Pluperfect. 10. A typical way of referring to situations as types is to state their existence within a temporal domain. Experiential interpretations, for example, tend to occur with temporal frame expressions (cf. Dahl & Hedin, this volume). 11. Translation by Walter Miller (Loeb Classical Library no. 52 [1968]. W. Heinemann Ltd: London & Harvard University Press: Cambridge, Massachusetts). This is also an example of the use of the Imperfective to refer to a given situation, since the verb is used earlier in the context. 12. Translations of Xenophon quotations are by Carleton L. Brownson (Loeb Classical Library no. 89 [1968]. W. Heinemann Ltd: London & Harvard University Press: Cambridge, Massachusetts). 13. I find the explanation of Rijksbaron (1984: 18) a bit strained, according to which, in examples of this kind, “the value [non-completed] of the imperfect serves to direct the attention to the consequences of the completion of the action”. It is hard to see why non-completion should lead to the interpretation of result. 14. Translation by C. F. Smith (Loeb Classical Library 169 [1976]. W. Heinemann Ltd: London & Harvard University Press: Cambridge, Massachusetts). 15. It should be noted that in the English translation by J.H. Vince (Loeb Classical Library no. 299 [1964]. W. Heinemann Ltd: London & Harvard University Press: Cambridge, Massachusetts.) the second (Imperfective) Imperative Anagígn¯oske is left out altogether. This way of rendering the imperfective verb form may perhaps be interpreted as illustrating the signal function of the Imperfective referred to above (cf. 27). The reader may imagine some extralinguistic behaviour – a gesture for instance – taking on the signalling function. 16. Given that the Imperfective always denotes situation types, this has to be the function of all time point expressions combined with it. The interpretation will differ. If in 1945 in (85) is changed to at that very moment the sentence would more naturally have the interpretation ‘At that very moment he was being married’ (if the context tells us that at that time he and his bride were standing in front of the priest). Also in this case, however, the time point expression would denote the type time and not refer to the time for a situation token (cf. 2.3). 17. Cf. the use of the Imperfective in similar constructions in French (cited by Pollak 1960: 145–151): Il fut nommé ministre, et deux ans après il mourait ‘He was appointed a minister, and two years later he died’ (p. 150). Et la semaine suivante, il s’embarquait à Brest ‘And the following week, he embarked at Brest’ (p. 150). Reparti de Las Palmas le . . . , B. arrivait à . . . le 23 décembre ‘Having returned from Las Palmas on . . . , B. arrived in . . . on the 23rd of December’ (p. 151). Pollak (in a footnote on p. 147) refers to Gamillscheg (1957: 407) as the only one of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

263

The type-referring function of the Imperfective

263

the cited authors who does not consider this use of the Imperfective as “pittoresque” or expressing a “lebhafte Vorstellung”. According to Gamillscheg the Imperfective is used for the following reasons: “In den Sätzen, die die Zeitbestimmung enthalten, ist diese vordringliche Mitteilung. Dadurch bekommt das im Imperfekt ausgedrückte Geschehen den Charakter eines Nebenumstandes, wenn auch logisch betrachtet eine unabhängige neue Mitteilung vorliegt (emphasis added).” A direct correspondence to the Modern Greek example (87) is the introduction to a chapter on Chateaubriand in a history of literature (cited by Pollak 1960: 148): Le 4 septembre 1768, naissait à Saint-Malo, dans la sombre rue des Juifs, le chevalier François de Chateaubriand. ‘On September 4, 1768, was born at Saint-Malo, in the dark street of the Jews, the chevalier François de Chateaubriand.’ Also in Italian the Imperfective is used in a similar way: In quello stesso anno nasceva a Firenze Dante Alighieri. ‘In the same year Dante Alighieri was born in Florence.’

References Bakker, Willem Frederik 1965 “The aspect of the imperative in modern Greek”, Neophilologus 49: 89–103. 1966 The Greek imperative. Amsterdam: Hakkert. Comrie, Bernard 1976 Aspect. An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Dahl, Östen 1981 “On the definition of the telic-atelic (bounded-nonbounded) distinction”, in: P. Tedeschi & A. Zaenen (eds.), Syntax and semantics 14: Tense and aspect. New York: Academic Press, 79–90. 1985 Tense and aspect systems. Oxford and New York: Blackwell. Dahl, Östen & Fred Karlsson 1976 “Verbal aspects and object marking: A comparison between Finnish and Russian”, International Review of Slavic Linguistics 1: 1–30. Dahl, Östen & Eva Hedin this volume “Current relevance and event reference”. Dowty, David R. 1979 Word meaning and Montague grammar. Dordrecht: D. Reidel. Forsyth, James 1970 A grammar of aspect: Usage and meaning in the Russian verb. Cambridge: Cambridge University Press. Gamillscheg, Ernst 1957 Historische Französische Syntax. Tübingen: Max Niemeyer Givón, Talmy 1978 “Negation in language: Pragmatics, function, ontology”, in: P. Tedeschi & A. Zaenen (eds.), Syntax and semantics 9: Pragmatics. New York: Academic Press, 69–112.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

264

264

Eva Hedin

Hedin, Eva 1987

On the use of the perfect and the pluperfect in modern Greek. Acta Universitatis Stockholmiensis: Studia Graeca Stockholmiensia VI. Stockholm: Almqvist & Wiksell.

Leech, Geoffrey 1974 Semantics. Harmondsworth: Penguin (5th edition, 1978). Leinonen, Marja 1982 Russian aspect, “Temporal’naja lokalizacija” and definiteness/indefiniteness. Helsinki. Lindstedt, Jouko 1985 On the semantics of tense and aspect in Bulgarian. Slavica Helsingiensia 4. University of Helsinki, Helsinki. Mackridge, Peter 1985 The modern Greek language. Oxford: Clarendon (2nd paper back edition, 1989). Mehlig, Hans Robert 1979 “Überlegungen zur Funktion und Determinierung der Aspekte im Russischen”, in: Jochen Raecke & Christian Sappok (eds.), Referate des VI. Konstanzer Slavistischen Arbeitstreffens, Tübingen 26.–29. Sept. 1978. (Slavistische Beiträge, Band 133, Slavistische Linguistik 1978). München: Otto Sagner, 151–169. Newton, Brian 1979 “Scenarios, modality, and verbal aspect in Modern Greek”, Language 5: 139–167. Pollak, Wolfgang 1960 Studien zum ‘Verbalaspekt’ im Französischen. (Österreichische Akademie der Wissenschaften, Philosophisch-historische Klasse, Sitzungsberichte 233, 5.) Wien: Rudolf M. Rohrer. Rassudova, Olga Petrovna 1984 Aspectual usage in modern Russian. Moscow: Russky Yazyk. Rijksbaron, Albert 1984 The syntax and semantics of the verb in classical Greek. An introduction. Amsterdam: J. C. Gieben. Ruijgh, C. J. 1985 “L’emploi ‘inceptif’ du thème du présent du verbe grec”, Mnemosyne XXXVIII, Fasc. 1–2: 1–61. 1991 “Les valeurs temporelles des formes verbales en grec ancien”, in: Jadranka Gvozdanovi´c & Theo A. J. M. Janssen (eds.), The function of tense in texts. Amsterdam: North-Holland, 197–217. Schank, Roger C. & Robert P. Abelson 1977 “Scripts, plans and knowledge”, in: P. N. Johnson-Laird & P. C. Wason (eds.), Thinking. Readings in Cognitive Science. Cambridge: Cambridge University Press, 421–432. Sicking, C. M. J. 1991 “The distribution of aorist and present tense stem forms in Greek, especially in the imperative”. Glotta 69: 154–170. Smyth, Herbert Weir 1920 Greek grammar. Cambridge, Massachusetts: Harvard University Press (8th edition, 1973). Štunová, Anna 1993 A contrastive study of Russian and Czech aspect: Invariance vs. discourse. Ph.D. dissertation, University of Amsterdam, Amsterdam. Schwyzer, Eduard 1975 Griechische Grammatik. Zweiter Band. Syntax und syntaktische Stilistik. (4th edition.) München: C. H. Beck.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

265

Rolf Thieroff

On the areal distribution of tense-aspect categories in Europe

1. Introduction In the present chapter an overview is given of the tense-aspect categories of 40 European languages and of their areal distribution. After the presentation of a list of the languages of the sample and a map localizing these languages in section 1, the theoretical basis for the investigation is presented in section 2. In 2.1 an attempt is made to define the notions of morphosyntactic category and categorization. In 2.2 some considerations about the principle of compositionality of morphosyntactic categories are presented, followed by a discussion of the limits of this principle (2.3). Finally, the differences between the concepts of morphosyntactic categories and gram-types are summarized, and a word on the notions of tense, aspect and mood is added (2.4). Section 3 is devoted to past time reference and the morphosyntactic categories used to refer to the past. The four major past time categories anterior, preterite, imperfect, and aorist are introduced and their areal distribution is shown (3.1), the development of present anteriors (the so-called perfects) is investigated (3.2) and different constructions of pluperfects are discussed (3.3). Section 4 provides an overview of the restrictions on the combinations of the category future, and in section 5 the areal distribution of the aspectual categories perfective and imperfective (5.1), progressive (5.2) and habitual (5.3) is shown. In the conclusion, six coherent areas with similar tense-aspect systems are identified.

1.1. The sample The languages investigated are listed in Table 1. The abbreviations of the languages are identical with the abbreviations of the E U ROTYP list of languages, with the following additions and modifications: Three languages are divided into two varieties, namely German, Italian, and Frisian. German and Italian both are divided into the Standard variety and respectively the Southern and Northern dialects.1 Frisian is divided into North Frisian and West

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

266

266

Rolf Thieroff

Table 1. The languages of the sample 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

Albanian Armenian Basque Bulgarian Belarusan Catalan Czech Danish Dutch English Estonian Finnish French North Frisian West Frisian Standard German Southern German Greek Hungarian Icelandic

Alb Arm Bsq Blg Blr Ctl Cz Dan Dut Eng Est Fin Fr NFrs WFrs StGrm SGrm Grk Hng Ice

21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

Irish Northern Italian Standard Italian Karaim Lithuanian Maltese Norwegian Polish Portuguese Romanian Russian Serbo-Croatian Slovene Lower Sorbian Upper Sorbian Spanish Swedish Turkish Ukrainian Yiddish

Ir NIt StIt Krm Lith Mlt Nor Pol Prt Rum Rus SCr Slve LSrb USrb Spn Swd Trk Ukr Yid

Frisian (Frysk). For details about the status of the varieties and dialects of Frisian, see Ebert (this volume a). The selection of these languages is motivated pragmatically: these are the languages about which the most reliable information is available. Most of the information is taken from the contributions in Thieroff & Ballweg (1994) and Thieroff (1995b), and from reference grammars. Admittedly, the sample is biased toward the western half of Europe; i.e., the nonSlavic languages spoken in the European part of the former Soviet Union are missing. This also implies that no less than 33 (about 80%) out of the 40 languages are IndoEuropean, and only seven (about 20%) are non-Indo-European languages. However, the exclusion of the languages spoken in the Caucasus can be motivated by the fact that “for areal generalizations, it appears most suitable to delimit Europe in the more traditional way where it does not include the Caucasus, since the languages spoken in that area are quite different in a number of respects” (Dahl, this volume a). In fact, Europe west of the Black Sea is a typical spread zone, whereas the Caucasus is a typical residual zone in the sense of Nichols (1992).2 Western European languages missing in the sample are the Celtic languages with the exception of Irish, the Romance languages Occitan, Rhaeto-Romance and Sardinian, Faroese, Slovak, Macedonian, and Latvian.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

267

On the areal distribution of tense-aspect categories in Europe

267

Thus the maps presented probably would look less homogeneous if more languages, especially more non-Indo-European languages, were investigated. On the other hand, there is already considerable variation within the languages of the sample.

1.2. The maps On Map 1, the languages of the sample are shown, with the varieties of German, Italian, and Frisian added. Note that in some cases, it is not quite obvious from the maps which languages actually have common borders. This is especially problematic for languages covering larger areas. A case in point is German, which borders Danish in the north, North Frisian, Dutch and French in the west, Italian in the south, Slovene and Hungarian in the south-east, and Polish and Czech in the east. A further difficulty in the case of German is Upper and Lower Sorbian, actually situated within the German-speaking area, which of course cannot be represented in the maps. Instead, Upper and Lower Sorbian are placed east of German, which also gives the wrong impression that Sorbian is situated between German and Polish.

Map 1. The languages of the sample

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

268

268

Rolf Thieroff

2. Describing tense and aspect systems 2.1. Morphosyntactic categories and categorizations Taking English as an example, the following forms and constructions are traditionally viewed as belonging to the paradigm of the verb: sing, sang, has sung, had sung, will sing, would sing, will have sung, would have sung, is singing, was singing, has been singing, had been singing, will be singing, would be singing, will have been singing, would have been singing. If we look at these forms, statements like the following impose themselves: all forms are finite, as opposed to further verbal forms like (to) sing, sung, singing. Two forms consist of the main verb only, one which we could call unmarked (sing), one which we might provisionally call “marked” (in this case by ablaut: sang). Four forms contain the elements have  past participle of the main verb; the element have appears either in a finite form – again, either unmarked (have) or marked (in this case by the allomorph -d, functionally equivalent to the ablaut: had) – or in the infinitive (in will have sung and would have sung). Finally, four forms contain the elements will  infinitive; will is always finite, twice unmarked and twice marked (ablaut  -d). The infinitive is either the infinitive of the main verb (in will/would sing) or the infinitive of have  the past participle of the main verb (in will/would have sung).3 It is obvious that these eight forms or constructions are combinations of (the presence or absence of) the three elements [marked], [have  past participle], and [will  infinitive]. To get a picture of the entire paradigm, one more element has to be added, namely be  -ing. This element can be combined with each of the eight forms, thus yielding a paradigm of a total of 16 forms. This can be depicted as in Table 2.

Table 2. Paradigm of finite verb forms in English unmarked [will + inf.] [will  inf.] non-progressive [have  PP] sings [have  PP] has sung progressive [have  PP] [have  PP]

is singing has been singing

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

“marked” [will  inf.] [will  inf.]

will sing will have sung

sang had sung

would sing would have sung

will be singing will have been singing

was singing had been singing

would be singing would have been singing

269

On the areal distribution of tense-aspect categories in Europe

markedness



have  will



will  inf.



269

be  -ing



Figure 1. Categories and categorizations of finite verb forms in English

tense present past future

future perfect past perfect perfect

aspect

progressive

non-progressive

mood

indicative

conditional

Figure 2. Traditional classification of the finite verb forms in English

What we have now is a classification of the 16 verb forms of English, assigning each form to four different formal categories. Each of these categories is a member of a class with two members. We could also say that each form has been classified or categorized with respect to four classes or categorizations. This is shown in Figure 1. Figure 1 represents the morphosyntactic categories and categorizations of the verb in English. Morphosyntactic categories thus are the result of a classification of a paradigm according to formal properties. Each verb form is classified or categorized with respect to four categorizations: markedness of the finite part, presence or absence of have  past participle, presence or absence of will  infinitive, presence or absence of be  -ing. Classifying or categorizing a form with respect to four categorizations implies that the form has four categories. For example, sing belongs to the categories [marked], [have  past participle], [will infinitive], [be  -ing]. Would sing has the categories [marked], [have past participle], [will infinitive], [be -ing] etc. If we now try to replace the category names in Table 2 by the traditional notions, we run into problems. At first sight we could replace “unmarked” by Present, “marked” by Past, have  past participle by Perfect, will  infinitive by Future and be  -ing by Progressive. However, “marked” will  infinitive is not called Past Future (analogous to Past Perfect) in grammars of English, but “Conditional”. This means that, obviously, not each form with the feature “marked” is regarded as Past, nor is each form with the feature will  infinitive regarded as Future. Furthermore, not every categorization in Figure 1 is regarded as a categorization in traditional grammars. Instead, both Past and Future and often also Perfect are subsumed under a categorization “tense” as opposed to the categorization “aspect” with the two categories progressive and non-progressive. Finally, would  infinitive is re-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

270

270

Rolf Thieroff

garded as belonging to an extra categorization, “mood”. A traditional classification of the forms thus would have to be reconstructed as in Figure 2. The reason for the rather different analyses in Figure 1 and Figure 2 is that in one analysis only the forms are considered, whereas in the other the meanings of the forms are taken into consideration as well. The crucial question now is how the analysis of the forms can be brought together with the meanings associated with these forms.

2.2. Compositionality Whereas it is obvious that the morphological make-up of the forms in Table 2 is strictly compositional, this is less clear for their semantics. In fact, some scholars have proposed systems in which the complex, analytic tense forms of English (and other languages) are treated as indivisible units for the purposes of semantic interpretation. In such systems the present perfect, the past present [sic], the present perfect progressive, and so on, are each given a special treatment. Yet other scholars have argued that the progressive, the perfect, and each of the tenses require separate treatments, so that the analysis of the present perfect progressive, for example, follows from, or consists in, that of the present plus that of the perfect plus that of the progressive. This requires that the treatment of such complex forms be at least weakly compositional, in the sense that semantic interpretation of the expression as a whole consists in semantic interpretation of its constituent parts. (Binnick 1991: 257f.)

Binnick continues by saying that “intuitively a compositional approach makes sense”, and this is the approach taken in the present chapter. This means that forms which are morphologically compositional are also regarded as semantically compositional as long as no obvious differences in the basic meaning speak against a compositional analysis. With regard to the basic meaning, following Kortmann, the following assumptions are made: [ ] we need to make a strict separation between semantics and pragmatics. A compositional account can only be concerned with what is asserted and not with what is inferred. It can only be concerned with intension, not with extension [ ], that is with meaning invariants out of context and not with different uses to which a grammatical category is put. Nevertheless [ ] the basic meanings which a compositional analysis postulates for individual categories must underlie, or at least be compatible with, the various uses to which the relevant categories can be put. Thus whatever interpretation we may want to give a present perfect or past perfect in a given sentence, the invariant meaning we postulate for the category perfect as such must be compatible with this interpretation

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

271

On the areal distribution of tense-aspect categories in Europe

271

[ ]. However, in view of the many idiomatic uses that often exist for individual categories there is widespread agreement that it suffices if this requirement is met only by the basic, non-idiomatic uses [ ]. (Kortmann 1995: 184f.)

A compositional analysis of complex forms thus does not exclude certain specific restrictions on the use of the elements of these forms in certain combinations. For example, the semantic description of the category anterior (“perfect”) as E R (point of event before point of reference) holds for the Present Perfect in English, although, in contrast to Past Perfect and Future Perfect, the Present Perfect cannot be collocated with adverbials denoting a specific time in the past (see 3.1, 3.2).4 Neither is the compositional analysis excluded by the fact that certain complex forms may have additional uses not entirely predictable from the meanings of their elements. Thus a compositional analysis has to account for certain idiosyncrasies occurring when certain categories are combined in one form. The problem of compositional vs. non-compositional analyses is also raised in Dahl (1985: 67). Dahl discusses arguments both for and against a compositional analysis. As for pluperfects, the author comes up with the solution to count them “as instances of both the cross-linguistic category PERFECT and PLUPERFECT ”. We have now reached a point where we can modify the scheme given in Figure 1. Instead of referring to the categories by writing down the morphological elements by which they are constituted (Table 2), we can give names to the categories which also give an indication of their semantic content (Figure 3). The names of categorizations are simply derived from the names of the categories. Henceforth, names of marked categories will be abbreviated, with capital letters. In figures, categorizations are printed in boldface. For a justification of the terms preterite (PRET) and anterior (ANT instead of perfect) see 3.1. The term “past” is used as a cover term for the categories PRET and IMPF. As has been mentioned, in English each verb form belongs to four different categories. However, instead of referring to the form sang as “preterite, non-anterior (i.e., unmarked for anteriority), non-future (i.e., unmarked for futurity), non-progressive (i.e., unmarked for progressivity)”, it will be sufficient to refer to this form as the preterite of sing, i.e., by mentioning the marked category, implying that sang is not marked with regard to anteriority, futurity, or progressivity. In this sense, the term ‘preterite’ is an abbreviation of [preterite, anterior, future, progressive]. Ac-

pastness

anteriority

PRET unm

ANT unm

futurity FUT

unm

progressivity PROG unm

Figure 3. Morphosyntactic categories of the finite verb in English

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

272

272

Rolf Thieroff

cordingly, with ‘preterite progressive’ we refer to the form categorized as [preterite, anterior, future, progressive], with ‘preterite future’ to the form categorized as [preterite, anterior, future, progressive] etc. In contrast, verb forms with the category PRET are all verb forms with the preterite marker (i.e., sang, was singing, had sung, would sing etc.). Note that in traditional tense names there are some incongruencies. The form with the categories [ preterite, anterior, future, progressive] is traditionally called the ‘Present Perfect’, whereas the form with the categories [ preterite anterior, future, progressive] is simply called ‘Future’, instead of ‘Present Future’, although this Future is opposed to the preterite future in the same way as is the present anterior to the preterite anterior. It should have become clear by now how the term ‘present’ is to be understood. With ‘present’ we refer to the form which is categorized [ preterite, anterior, future, progressive], i.e., ‘present’ is used as a term for the maximally unmarked form. In combinations such as ‘present anterior’, ‘present progressive’, ‘present’ has the meaning ‘non-preterite’. However, in order to avoid misunderstandings, I shall refer to the entirety of the forms with the category [ preterite] as the non-preterite forms (instead of present forms). Similarly, the nonprogressive forms are all forms belonging to the category [ progressive] etc. For the term ‘pluperfect’, see 3.3.

2.3. Non-compositional analyses Despite what has been said in the previous section, a non-compositional semantic analysis of morphological composite forms is not generally excluded. A case in point is Modern Basque, whose verbal paradigm consists of analytical verb constructions only.5 The paradigm of the forms is represented in Figure 4, rendering the forms listed in Table 3. However, the meanings of these forms are far less compositional than their morphology is. Haase indicates the tense-aspect oppositions in Table 4.

auxiliary unmarked

marked

non-finite participle

unm

inessive verbal noun

 -a(k)

 -ko Figure 4. Morphosyntactic categories of the finite verb in Basque I

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

273

On the areal distribution of tense-aspect categories in Europe

273

Table 3. Finite verb forms in Basque

inessive verbal noun participle participle + -ko participle + -a(k)

present auxiliary

past auxiliary

kantatzen du kantatu du kantatuko du kantatua du

kantatzen zuen kantatu zuen kantatuko zuen kantatua zuen

Table 4. Tenses and aspects in Basque (Haase 1994: 283)

inessive verbal noun participle participle  -ko (-en) participle  -a(k), ( -rik/-ta)

present auxiliary

past auxiliary

simple present (aspectually neutral) present perfect future resultative

imperfective past perfective past future of the past pluperfect

As can be seen from Table 4, only the future of the past is compositional in the sense that it is composed of the meaning of the participle  -ko (future) and the meaning of the past auxiliary (past). The bare participle has quite different meanings in collocation with the present and past auxiliary, and the same is true for the participle  -a(k) construction. Thus, though morphologically composite, the imperfect (Haase’s imperfective past) and the aorist (Haase’s perfective past) have to be described semantically as non-composite forms. The meaning of the pluperfect, on the other hand, can be described as being composed semantically of perfect  past, but not morphologically. Thus from a semantic point of view, the Basque system has to be described as in Figure 5. This renders the seven tense-aspect forms given again in Table 5, with the resultative missing, not being regarded as a tense-aspect form (cf. Thieroff 1994a: 29).

Table 5. Tense-aspect forms in Basque

unmarked anterior future

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

unmarked for perfectivity non-past past/imperfect

aorist

kantatzen du kantatu du kantatuko du

kantatu zuen

kantatzen zuen kantatua zuen kantatuko zuen

274

274

Rolf Thieroff perfectivity

AOR unm pastness unm

IMPF

futurity unm

FUT

anteriority unm

ANT

Figure 5. Morphosyntactic categories of the finite verb in Basque II

2.4. Categories and gram-types From what is said about grams, gram-types, and gram families in Dahl (this volume a), it will become clear that neither of these notions corresponds exactly to the notion of category as developed above. The category ANT is equally present in the forms has sung, had sung, will have sung, would have sung, has been singing, had been singing, will have been singing, and would have been singing. This is less clear for the Perfect gram in English as described by Bybee & Dahl. On the one hand, it is said that “combinations of [the six major gram-types] were found. For instance, the perfect and progressive are often combinable with a past tense to give a past perfect (or pluperfect) and a past progressive respectively” (Bybee & Dahl 1989: 55). On the other hand, in the text that follows the authors refer exclusively to the so-called present perfects, i.e. forms with the categories ANT, FUT, PRET/IMPF. For in combination with past or future or both, the category ANT normally does not indicate that a situation is relevant at a point of reference. Instead, in these combinations, ANT merely indicates anteriority with respect to a point of reference. Only in the case of the present perfect in some languages is the notion of current relevance crucial. Thus, current relevance is not a feature of the category ANT, while it is for the gramtype perfect. If we examine the description of the development of perfects, what the authors are actually describing is the development of present anteriors, not of ANT. It is not have  Past participle in general which develops into a (perfective) past in French and German, but only the equivalent of the English Present Perfect. The equivalents of had  past participle, will have  past participle, and would have  past participle either remain what they are, or they vanish altogether, giving rise to an entirely new synchronic system. So, what we actually find is the development of present anteriors into pasts, not of ANT.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

275

On the areal distribution of tense-aspect categories in Europe

275

Finally, for the notions tense and aspect, I follow the view presented in Dahl (this volume a). See also Thieroff & Budde (1995), where tense and aspect (together with mood) are regarded as purely semantic concepts, i.e., entities not belonging to the domain of morphosyntactic categories and categorizations.

3. Past time reference Bybee & Dahl conclude their paper by saying that the relevant entity for the study of grammatical meaning is the individual gram, which must be viewed as having inherent semantic substance reflecting the history of its development as much as the place it occupies in a synchronic system. (Bybee & Dahl 1989: 97)

While fully agreeing with this statement, I have the impression that in reality the place a gram “occupies in a synchronic system” has not always been studied sufficiently, first because the oppositions in the synchronic systems have not been considered sufficiently, and second because not enough attention has been paid to the compositionality of the categories as has been explained in the previous section. In this section I shall try to further develop these two points.

3.1. Anterior, preterite, imperfect, and aorist In the domain of past time reference, at least four semantic fields can be distinguished, three of which are connected with gram-types. The first semantic field is the domain of current relevance at the moment of speech time (but not at another point of reference; cf. Bybee & Dahl 1989: 55 on the perfect gram-type). For the notion of current relevance, which mainly comprises cases where the so-called resultative perfect is used and cases where perfects have experiential meaning, see Lindstedt (this volume). The second semantic field is that of perfective past. Categories which only cover this semantic field will henceforth be called aorist (AOR). The aorist in our sense is identical with the PFV as defined by Dahl: A PFV verb will typically denote a single event, seen as an unanalysed whole, with a well-defined result or end-state, located in the past. More often than not, the event will be punctual, or at least, it will be seen as a single transition from one state to its opposite, the duration of which can be disregarded. (Dahl 1985: 78)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

276

276

Rolf Thieroff

The third semantic field is that of imperfective past. Notional imperfective past can be negatively defined as those cases of past time reference which have neither current relevance nor are the cases where AOR (Dahl’s PFV) is used. The fourth semantic field to be related here covers different meanings within the area of nonpast time reference where counterfactuality is the core (as in the protasis in English If it rained now, I would stay home or in contrary-to-fact wishes as in English If only they were coming). These semantic fields are covered by different morphosyntactic categories in the languages of Europe. The first category to be mentioned here is the category ANT. As has been mentioned before, the meaning of this category can best be described as “E before R”, i.e., by using the category ANT the speaker situates the time of event prior to a point of reference. This point of reference can be constituted by other categories (by FUT in the future perfect, by PRET or IMPF in the pluperfect) or by the context. By default, the point of reference is the point of speech. Now it is precisely in these default cases, i.e., in the cases of the present anterior (which, recall, is short for ANT, PRET/IMPF, FUT), that the anterior is used to convey the meaning of current relevance, provided there is also a PRET or IMPF and AOR in the language. Thus the present anterior largely conforms to Bybee & Dahl’s perfect gram-type (see above). Note, however, that ANT is also defined in opposition to the categories with past time reference. This means that for a tense-aspect form to be categorized as ANT, there also have to be tense-aspect forms belonging to either categories 1) AOR and IMPF or 2) PRET. For the development of present anteriors, see 3.2. In the case of the present anterior, different restrictions on its use have to be observed in the European languages. A major issue is the use of the English Present Perfect which cannot be collocated with definite past time adverbials such as yesterday or last year. This restriction, which has been referred to as “the present perfect puzzle” (the title of Klein 1992), has been commented upon by a large number of linguists and it has been one of the major reasons for rejecting a compositional analysis of the Present Perfect in English, given that the time adverbials mentioned are allowed with the Past Perfect and the Future Perfect. However, in recent publications it has been argued convincingly that this restriction is no obstacle for a compositional analysis (see among others Quirk et al. 1985, Stump 1985, Fenn 1987, Binnick 1991, Klein 1992 and 1994, and Kortmann 1995 for an overview). While it is not possible here to discuss the arguments given by these authors in any detail, it should be noted that “the English present perfect, both from a comparative perspective and when contrasted with the past perfect and future perfect in English, is far from being the prototype of a perfect” (Kortmann 1995: 195) and that “it is precisely the present perfect that, for pragmatic reasons, provides the least insight into the nature of the perfect” (Stump 1985: 235). The same statements hold for the Present Anterior in Portuguese, whose use underlies very special restrictions (see the detailed descrip-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

277

    

On the areal distribution of tense-aspect categories in Europe

277

present anterior

current relevance



counterfactuality

IMPF

imperfective past

perfective past

AOR

PRET

Figure 6. Categories of past time reference

tion in Squartini & Bertinetto, this volume). Again, the set of restrictions described there “does not hold for the other compound forms, such as the Pluperfect, Compound Future, Compound Conditional, Subjunctive C[ompound] P[ast]” (Squartini & Bertinetto, this volume). The second category relevant to past time reference is the category preterite (PRET), which covers the three semantic fields of perfective past, imperfective past and counterfactuality. The third and fourth categories are AOR and IMPF, which together cover the same semantic fields as PRET. AOR as defined above covers the semantic field of perfective past, IMPF the two fields of imperfective past and of counterfactuality. Again, it has to be noted that the categories AOR, IMPF and PRET are also defined in opposition to the category ANT. In other words, these three categories imply the existence of ANT. For tense-aspect forms covering either the semantic fields of perfective past and current relevance (such as the Compound Past in Northern Italian) or these two semantic fields plus the semantic field of imperfective past (such as the Perfekt in Southern German and Yiddish) or all four semantic fields (the category GPST), see 3.2. Figure 6 shows how the categories mentioned cover the four semantic fields. A system with the categories PRET and ANT plus the categories FUT and PROG is typical for the Germanic languages. In Figure 7, the tense-aspect system of the Germanic languages is represented. The classification in Figure 7 shows that not only the finite forms but the entire paradigm including the non-finite forms has to be categorized with respect to the categorizations anteriority and progressivity, since the non-finite forms have the categories ANT and PROG as well.6 It is important to note that though the classification represented in Figure 7 holds for the majority of the Germanic languages, there are differences as to the usage of the categories. Besides the different status of the present anterior in, for example,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

278

278

Rolf Thieroff

finiteness

anteriority

non-finite

unm ANT

progressivity unm

PROG

type infinitive

participle

finite

pastness unm PRET

futurity unm

FUT

Figure 7. Tense-aspect categories in the Germanic languages

English and Standard German just mentioned, FUT and PROG differ considerably in their degree of grammaticalization and obligatoriness in the different Germanic languages (see 4 and 5.2). In Table 6, the finite forms of some Germanic languages having the system shown in Figure 7 are listed.7 This leaves aside the progressive which, with the exception of English, Icelandic and West Frisian (see 5.2), is a non-obligatory category in the Germanic languages.

Table 6. Tense-aspect forms in Germanic languages PRET FUT Eng Ice Nor Swd Dut Grm Eng Ice Nor Swd Dut Grm

ANT sings kallar kjøper köper vertrekt singt ANT

has sung hefur kallað har kjøpt har köpt is vetrokken hat gesungen

FUT

PRET FUT

FUT

will sing mun kalla skal kjøpe skall köpa zal vertrekken wird singen

sang kallaði kjøpte köpte vertrok sang

would sing mundi kalla skulle kjøpe skulle köpa zou vertrekken würde singen

will have sung mun hafa kallað skal ha kjøpt skall ha köpt zal zijn vertrokken wird gesungen haben

had sung hafði kallað hadde kjøpt hade köpt was vertrokken hatte gesungen

would have sung mundi hafa kallað skulle ha kjøpt skulle ha köpt zou zijn vertrokken würde gesungen haben

3rd person singular indicative of Eng sing, Ice kalla ‘call’, Nor kjøpe ‘buy’, Swd köpa ‘buy’, Dut vertrekken ‘leave’, Grm singen ‘sing’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

279

On the areal distribution of tense-aspect categories in Europe

finiteness non-finite

finite

type

perfectivity

infinitive

participle

anteriority

progressivity

unm

unm

ANT

279

PROG

AOR unm pastness unm

IMPF

futurity unm

FUT

Figure 8. Tense-aspect categories in the Romance languages

It is important to note that with respect to the combinability of categories, IMPF does not differ from PRET. Thus IMPF appears in the same combinations as PRET, whereas combinations of AOR with FUT are generally excluded, AOR  PROG is rare (cf. Bertinetto, this volume), and AOR  ANT is restricted to some Romance languages (French, Catalan, Spanish, Italian), though with a very marginal status (restriction to subordinate clauses in formal registers), and to Albanian. In Figure 8 the system of the Romance languages is represented. Without the categorization progressivity, Figure 8 also represents the systems of Albanian, Bulgarian, and Serbo-Croatian.8 For Basque, see 2.4. In Table 7, the finite forms of some Romance languages having the system shown in Figure 8 are listed, again without the progressive, which is not obligatory in these languages either. In addition to the forms displayed in Table 7, in Catalan there is also a synthetic aorist (cantà ‘he sang’) whose “forms are in regular spontaneous spoken use only in central Valencian [ ]. However, the synthetic forms are widely used in written Catalan, alongside the periphrastic forms, in all areas” (Wheeler 1988: 189); and there is a synthetic pluperfect in Portuguese, which is “limited to written forms of European Portuguese” (Parkinson 1988: 151). For the latter see also 3.3. Figure 8 shows “a system where a gram with the meaning ‘perfective past’ is opposed to everything else” (Bybee & Dahl 1989: 83), i.e., an instance of what Bybee & Dahl call a “tripartite tense-aspect system” (ibid.). The system is tripartite because the gram perfective (with the meaning perfective past, i.e., our AOR) is opposed to past and to non-past (both being imperfective). However, the authors then argue that “in the perfective, the aspectual properties could be seen as ‘dominant’ relative to the temporal properties” because there is considerably more variation both intra- and inter-linguistically as to how the ‘past time reference only’ restriction is manifested. It thus happens fairly frequently

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

280

280

Rolf Thieroff

that perfective categories may have non-past reference in non-indicative moods or [ ] certain non-assertive contexts, such as conditional clauses. Well-known examples are the Aorist in Classical Greek and the Perfective in Arabic. (Bybee & Dahl 1989: 84)

With regard to the European languages of our sample, considering the variation of the ‘past time reference only’ restriction, the opposite view is taken in this chapter. In the languages of the sample, Modern Greek is the only language where AOR may have non-past time reference in the contexts indicated. Indeed it is highly questionable whether the Greek form should be labelled AOR at all; see 5.1.9 In all other languages AOR is indeed restricted to 1) past time reference and 2) notional perfectivity. In contrast, in all languages, including Modern Greek, it is the IMPF which can occur both with non-past time reference (at least in counterfactuals, as in French S’il pleuvait je resterais à la maison ‘If it were raining, I would stay at home’) and with notional perfectivity, which is why IMPF is regarded as unmarked with regard to the categorization of perfectivity. The fact that AOR is “opposed to everything else” and is indeed restricted to past time reference is one of the main reasons for the strict separation of the opposition AOR:IMPF from the opposition PFV:IPFV (see 5.1). Map 2 shows three areas of past tense distinctions in Europe: the area with an ANT:PRET system in the north, the area with an ANT:AOR:IMPF system in the south, and an area with neither of these systems in the east (see below).10

Table 7. Tense-aspect forms in Romance languages non-past FUT Fr Ctl Spn Prt It

ANT

Fr ANT Ctl Spn Prt It

FUT

IMPF FUT

AOR

FUT

chante canta canta canta canta

chantera cantarà cantará cantará canterà

chantait cantava cantaba cantava cantava

chanterait cantaria cantaría cantaria canterebbe

chanta va cantar cantó cantou cantò

a chanté ha cantat ha cantado tem cantado ha cantato

aura chanté haurà cantat habrá cantado terá cantado avrà cantato

avait chanté havia cantat había cantado tinha cantado aveva cantato

aurait chanté hauria cantat habría cantado teria cantado avrebbe cantato

eut chanté hagué cantat hube cantado — ebbe cantato

3rd person singular indicative of Fr chanter, Ctl, Spn, Prt cantar, It cantare ‘sing’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

281

On the areal distribution of tense-aspect categories in Europe

281

ANT:PRET ANT:AOR:IMPF Map 2. Three areas with different past tense systems

The only southern European language not belonging to the ANT:AOR:IMPF area is Maltese, despite the fact that in traditional descriptions of Maltese the two simple finite verb forms are called “perfect” and “imperfect” or “perfective” and “imperfective’ (cf. Ebert, this volume b). Instead, it is the auxiliary kien which “is a past marker” (ibid.) or a “past tense verb” (Fabri 1995: 331); the “temporal reference is determined by the auxiliary [kien]” (Ebert, this volume b), i.e., kien is a marker of the category PRET. I also follow Ebert in as far as the so-called Imperfective is “the unmarked TA form par excellence” (ibid.), which of course implies that it is not IMPF in the sense developed above. As for the so-called Perfective, the indications are less clear in Ebert (this volume b). Ebert states that “the basic meaning of the Perfective is relative past time reference and perfective aspect”, but also that “the Perfective can designate past events with current relevance (perfect meaning)” and that (in combination with the auxiliaries kien and ikun), “the Perfective of the main verb expresses anteriority” or “an anterior event”. Finally Ebert concludes that “the Perfective functions as an anterior (past and perfect) and as a perfective in narratives”. It seems clear that anteriority

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

282

282

Rolf Thieroff

is the smallest common denominator for the uses mentioned and thus the so-called Perfective has to be categorized as ANT.

3.2. The development of present anteriors Bybee & Dahl (1989: 73–74) list three directions in which perfects may develop: (1) one involving the development of evidential functions, (2) the development into a past or perfective marker, (3) the use of perfects to express remoteness distinctions. With regard to the second development, they claim that the choice between the two end-results – past or perfective – is conditioned by the presence or absence of a separate imperfective past. While this is true for the case they quote (French), it is not a universally valid generalization. In fact, in European languages with the imperfect:aorist opposition both developments occur: the development of the perfect into an aorist, as in French, and also the development of a perfect into a preterite, as in most Slavic languages and possibly in Romanian (cf. Haase 1995). Bybee & Dahl are aware of the fact that “it would be an over-simplification to say that [the change of the perfect] is simply a weakening of the original meaning” (Bybee & Dahl 1989: 74), which is why two further conditions are indicated: Thus in order for a perfect to become a past the point of reference must be restricted to the moment of speech, and the part of its meaning that specifies that the past event is especially relevant to the current moment must be lost. (Bybee & Dahl 1989: 74)

While this is correct, the new “pasts” (in the sense of Bybee & Dahl) in French and Southern German still differ from older pasts in several respects. In fact, there are still considerable differences between the French Passé Composé and an aorist as defined above as well as between the Southern German Perfekt and a preterite. First of all, both the Passé Composé and the Perfekt not only cover the semantic field of the perfective past and the imperfective and perfective past respectively, but in addition they are also still used in current relevance contexts, even though “the part of [their] meaning that specifies that the past event is especially relevant to the current moment” is lost. In contradistinction to these forms, prototypical aorists and preterites cannot be used in a current relevance context, which, one will recall, is the domain of the present anterior. Second, in contradistinction to a preterite, the Southern German Perfekt does not have any modal uses – in counterfactual contexts the forms are used which in Standard German are categorized as preterite subjunctive.11 Thirdly, the French anterior can still be combined with the future to yield a future anterior (i.e., a form with roughly the meaning E before R and R after S), whereas aorists never can be combined with futures. Similarly, the combination of the Perfekt with the future yields a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

283

On the areal distribution of tense-aspect categories in Europe

283

future anterior, whereas the combination of preterite and future regularly is a futurein-the-past. If we now look at the development of present anteriors in other European languages, we see that the characteristics mentioned conform to specific stages of the development of present anteriors. In the development of present anteriors into more general pasts, at least four different stages can be distinguished. In the first stage the present anterior spreads into the domain of the past (PRET, AOR, or AOR and IMPF) and coexists with the still vivid past time categories. In a typical PRET or AOR context the speaker can freely choose between the two forms. This is the case in Modern Standard German (Thieroff 1992, 1994b), Standard Italian (Dardano & Trifone 1985: 243; Bertinetto 1992: 53), Upper Sorbian (Faßke 1981: 262f.) and in the South-East dialects of Gheg Albanian (Buchholz & Fiedler 1987: 130). In the second stage, PRET, AOR, or AOR and IMPF become more and more marginal and are only used in certain registers such as the written code or formal speech. This is the case in Serbo-Croatian (where AOR and IMPF are even losing ground in the written register; Corbett 1987: 403f), in Romanian with respect to the aorist (Beyrer, Bochmann & Bronsert 1987: 193; Mallinson 1988: 407; Squartini & Bertinetto, this volume), whereas the replacement of the imperfect by the present anterior is a recent development (Haase 1995), and to a lesser degree in French, where the aorist is restricted almost 100% to the written variety. In the third stage, PRET, AOR, or AOR and IMPF are completely lost, i.e., there is only one form for the semantic fields of current relevance and perfective past or current relevance and perfective and imperfective past. This stage is to be observed in Southern German, Yiddish (Kiefer 1994), Hungarian (Csató 1994), and Northern Italian, “where the S[imple] P[ast] does not even exist as a remote morphological possibility” (Squartini & Bertinetto, this volume). The development of present anteriors in the Romance languages is described in detail in Squartini & Bertinetto (this volume), who refer to this development as a process of “aoristicization” or an “aoristic drift”. The authors speak of “a single continuum, in which the various languages are disposed scalarly, from a minimum to a maximum in terms of proximity to the purely aoristic value”. The order of languages on this continuum is the same as in Table 8, going from Spanish and Catalan as “more perfectal” via the intermediate stages of Standard Italian, Standard French and Romanian to “various Northern Italian and French vernaculars” with “purely aoristic” present anteriors. It is only in stage 4 that the present anterior starts to take over other typical features of past tenses. In this stage, the present anterior can (at least in combination with another, non-indicative, marker) take over modal uses, and especially be used in counterfactuals with non-past time reference. This stage is reached in the North Slavic languages and in Slovene. The stages described are summarized in Table 8.12

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

284

284

Rolf Thieroff

Table 8. Stages of the development of present anteriors Stage status of AOR/PRET

x  FUT = fut.perf.

x  past = pluperf.

x  x modal uses languages = pluperf. of x

0

present anterior AOR/PRET

yes

yes

no

no

1

yes

yes

yes

no

yes

yes

yes

no

3

present anterior = AOR/PRET AOR/PRET marginal no AOR/PRET

yes

no

yes

no

4

no AOR/PRET

no

no

yes

yes

2

Scand., Dut, WFrs, Eng, Fin, Est, Spn, Ctl, Bsq, . . . StGrm, NFrs, StIt, USrb, Alb SCr, Rum, Fr SGrm, NIt, Yid, Hng Pol, Cz, Blr, Rus, Ukr, Slve

x = ANT of stage 0

With regard to the compositionality of the categories, in stages 1 and 2 no change of the behaviour of ANT can be observed. In both stages, the combination of ANT with past (PRET or IMPF) as well as with FUT is possible, and in both cases the regular meaning of pluperfect and future perfect results. That is, ANT adds to PRET/IMPF and FUT what it is expected to add, namely the meaning of anteriority of the event with regard to a point of reference, which in turn is situated in the past in the case of PRET/IMPF and in the future in the case of FUT. In stage 3, due to the disappearance of PRET/IMPF a combination of these with ANT is no longer possible, with the result that the language has lost its pluperfect too (e.g., in Hungarian), unless there is a new pluperfect, a so-called supercompound form, as in Southern German and in Yiddish (see below). However, if the combination of ANT  FUT is possible at all, the resulting meaning is still the meaning of a future perfect (as in Southern German, Yiddish, Northern Italian), not of a futurein-the-past, which means that the form in question still has characteristics of the category ANT. However, since the opposition with PRET/IMPF has vanished, the category now clearly has a status different from ANT. In principle it would be necessary to introduce two new category names here, one for the category covering the semantic fields of current relevance and perfective past (Northern Italian) and one for the category covering these two fields plus the semantic field of imperfective past (Southern German, Yiddish, and Hungarian). However, since in further discussion these categories will not play an important role, I leave these categories without names. On Map 3 the languages with these categories are referred to as stage 3 languages.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

285

On the areal distribution of tense-aspect categories in Europe

other languages

ANT:PRET present anterior present anterior GPST (stage 4)

285

ANT:AOR:IMPF past development, stages 1–3 past development, stage 3

Map 3. Development of present anteriors

Finally, in stage 4, represented by the north Slavic languages except Upper Sorbian and by Slovene, neither a combination of the former ANT with PRET/IMPF nor with FUT is possible any longer. The new category, which covers the four semantic fields displayed in Figure 6, is called general past (GPST). Whereas Map 3 shows the coherent area where present anteriors are currently developing into more general pasts, an area in the centre of the area under investigation, Map 4 shows in a sense the reverse picture: here those languages are highlighted which do have a present anterior which is exclusively or predominantly used in current relevance contexts. In other words, these are the languages possessing the perfect gram in the sense of Bybee & Dahl. As Lindstedt (this volume) points out, “the perfect has become a peculiar ‘maritime category’ in Modern Europe – most of the languages and dialects with a stable perfect are situated on the fringe of the continent”.13 This is confirmed by Map 4.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

286

286

Rolf Thieroff

stable present anterior (“maritime perfect”) Map 4. Stable present anteriors

3.3. Pluperfects In all languages of the sample having either the categories ANT and PRET or ANT and IMPF (which by definition implies the existence of AOR), these categories can be combined, yielding a pluperfect, i.e. a form whose meaning can roughly be described as referring to an event which took place prior to a point of reference which in turn is situated prior to the time of speech (E R & R S). As has been mentioned before, in languages with the IMPF vs. AOR distinction, it is always the IMPF that is used to construct the pluperfect. In addition, in some Romance languages, there is also a marginal AOR  ANT pluperfect (cf. Figure 8 and Table 7). In all languages having a pluperfect, the core meaning is composed of the semantics of its categories, i.e., the core meaning is anteriority (due to ANT) to a point of reference in the past (due to PRET or IMPF). In addition, some special uses of pluperfects may occur, as described in Dahl (1985: 144ff.). In addition to the PRET/IMPF  ANT pluperfects a second way of constructing pluperfects occurs in the languages of the sample, viz. the so-called supercompound forms, i.e., forms resulting from applying the perfect operator twice to the same verb. Supercompound forms are only attested in languages where the present anterior is

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

287

On the areal distribution of tense-aspect categories in Europe

287

undergoing or has undergone the development toward a more general past category. While this is not astonishing, given that in these languages the PRET/IMPF tends to be abolished, it is not possible to predict at which stage of the present anterior development supercompound forms occur, nor whether they occur at all. In the languages of the sample, supercompound forms exist in French, in some Northern Italian dialects and in some varieties of Romanian (Squartini & Bertinetto, this volume), in Standard German, Southern German, Yiddish, Albanian and in the three Slavic languages Serbo-Croatian, Slovene, and Polish. In French, Standard German, and Albanian, the supercompound forms coexist with the (older) ANTPRET/IMPF pluperfects; in French and German these forms are rare and are not accepted by normative grammars; in Albanian, the ANTANT pluperfect is again restricted to the north-east Gheg dialects (Buchholz & Fiedler 1987: 133). In Slovene and Polish14, the supercompound pluperfect is the only pluperfect (and to my knowledge normally not called supercompound). In Serbo-Croatian the supercompound form seems to be the only existing pluperfect today (Rehder 1991: 52; Gvozdanovi´c 1995).15 Since in these languages AOR and IMPF have been lost (Polish and Slovene) or almost lost (Serbo-Croatian), their pluperfects have to be described as GPSTGPST. In the remaining languages undergoing (Upper Sorbian) or having undergone (Northern Slavic, Hungarian) the present anterior past development, supercompound forms are not attested. Finally, it should not go unnoted that the Romanian pluperfect is not compositional morphologically, but rather a continuation of the Latin inflectional pluperfect. Still, semantically, the Romanian pluperfect does not differ from the pluperfects of the other languages described, i.e. it can be analyzed as being composed semantically of the features “past” and “anterior”, but not morphologically of the categories IMPF and ANT. This also applies to the synthetic pluperfect (cantara ‘he had sung’) in Portuguese, which is “grammatically equivalent” with the analytic pluperfect (tinha cantado ‘he had sung’) and “is effectively limited to written forms of European Portuguese” (Parkinson 1988: 150f; cf. also Hundertmark-Santos Martins 1982: 189). Note as well that the existence of such synthetic pluperfect forms is one of the reasons for Dahl establishing the cross-linguistic category of PLUPERFECT (cf. 2.3). On Map 5 the languages with a supercompound pluperfect are shown along with the present anterior past languages and the few languages lacking a pluperfect altogether.

4. Futures Under the heading “European future gram families” the morphology of the future categories in the European languages is described in Dahl (this volume b). It becomes clear from Dahl’s description that with the possible exception of what Dahl

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

288

288

Rolf Thieroff

present anterior past development, stages 1–3 pluperfect: supercompound forms no pluperfect Map 5. Supercompound pluperfects

calls the “futureless area”, there are future markers in all European languages considered by Dahl, to which we may add Maltese, Turkish, and Armenian, which also possess future markers.16 As for the “futureless area”, Dahl takes “the obligatory use in (main clause) prediction-based contexts as a main criterion for the grammaticalization” (Dahl, this volume b), and given that this does not hold for the Finno-Ugrian and Germanic languages except English, he claims that these languages constitute a futureless area. Indeed in these languages in general, future time reference can be referred to with the unmarked form (the present), in other words, in general the future is not obligatory in sentences with future time reference. However, in all languages there are special devices to refer to future time reference which can (optionally) be used in future contexts.17 In traditional descriptions of Finnish and Estonian, a future category is not mentioned (for Finnish see Hakulinen 1957, Olli 1958, Fromm 1982, Karlsson 1983). However, in both languages there are devices too which are (optionally) used to refer to future events (see Tommola 1992a, 1992b, 1994; Metslang & Tommola 1995). In any case there can be little doubt that in both Finnish and Estonian the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

289

On the areal distribution of tense-aspect categories in Europe

289

weakly grammaticalized FUT no FUT  past no FUT  ANT Map 6. FUT and restrictions of its combinability

future periphrases are currently developing into “real” future categories, as is stated by Metslang: Although there is no grammatical Future in Estonian nor in Finnish, some constructions are used as its equivalents in both languages. In particular, the future tense is needed to mark the futurity of imperfective situations. Two future constructions are developing in Estonian: the hakkama (‘begin’)-future that expresses a future action that has not yet begun, and the general and neutral saama (‘get, become’)-future. In Finnish, the general future with the auxiliary tulla (‘come’) is spreading. (Metslang 1994: 266)

Therefore, the languages described by Dahl as futureless are regarded here as languages with a weakly grammaticalized future. On Map 6, these languages are indicated as well as languages having restrictions on the combination of FUT with either the past (PRET or IMPF) or ANT, or both. The combination with FUT is generally excluded for the category GPST, i.e., a former present anterior having reached stage 4 in the present anterior past development (see 3.2, Table 8). In addition, due to the lack of PRET, there is no future-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

290

290

Rolf Thieroff

in-the-past in Southern German, Yiddish, and Hungarian (stage 3), nor in Romanian and Serbo-Croatian (stage 2), nor in Upper Sorbian (stage 1). Finally, the combination of FUT and PRET is lacking in Lithuanian and in the language with the least grammaticalized FUT, Estonian. Whereas the languages lacking a future-in-the-past belong to one coherent area, languages without ANTFUT are isolated in Europe. With the exception of Sorbian, all are situated at the fringes of the area investigated. Again, for Finnish and Estonian the lack of ANTFUT can be explained by the weakly grammaticalized FUT in these languages. In Irish it is the anterior which is only weakly grammaticalized, which may explain the lack of the combination there. In Basque and Armenian, however, both ANT and FUT are well-established categories, and a reason for the impossibility of combining both in one verbal form cannot be given.

5. Aspects 5.1. PFV:IPFV One of the major issues of the literature on aspect in the last few decades is the question whether the AOR:IMPF opposition discussed in 3.1 and the aspectual opposition found in, e.g., the Slavic languages are instances of the same cross-linguistic opposition perfective:imperfective, and if so, how differences in the use of the respective forms can be accounted for. Whereas Slavicists often claim that only the Slavic opposition deserves the name of aspect and only the members of this opposition should be called “perfective” and “imperfective”, recent typological research has come to the insight that it is rather the Slavic opposition which is atypical cross-linguistically. So for Dahl [t]he question that arises is whether the Perfectivity/Imperfectivity opposition in Russian, Polish, Czech and Bulgarian should be subsumed under PFV:IPFV at all. In particular for Bulgarian, where Perfective/Imperfective exists alongside the Aorist/NonAorist opposition, this would appear a natural conclusion. (Dahl 1985: 85)

Indeed, since, “although there is also a rather high correlation between the two systems in Bulgarian, one can find a sufficient number of disharmonic choices, such as Imperfective Aorists” (Bybee & Dahl 1989: 87), it seems to make little sense to claim that both oppositions belong to the same cross-linguistic categories. Similarly, Bertinetto & Delfitto argue that “if boundedness [i.e., the Slavic “aspect”; R.T.] were a purely aspectual phenomenon, it would be hard to understand how it could ever develop in Bulgarian, which has independent (and abundant) aspectual devices at its disposal” (Bertinetto & Delfitto, this volume).18 Thus, the existence of imperfective

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

291

On the areal distribution of tense-aspect categories in Europe

291

Aorists and perfective Imperfects in Bulgarian is considered a first, crucial argument for the assumption that AOR:IMPF are distinct from the Slavic style opposition, which I shall henceforth call the PFV:IPFV opposition. Some other arguments are listed in Bybee & Dahl 1989. The authors argue that the Slavic aspectual systems differ from the tripartite systems in their origins, their semantics, their means of expression and their relation to other parts of the system of verbal grams such as tense. (Bybee & Dahl 1989: 85)

As for the origins, Bybee & Dahl remark that in Slavic languages “one can see a tendency for bounders [such as up in English eat up; R.T.] to become grammaticized as aspectual markers” (p. 86), which “makes the Slavic aspectual systems much more derivational in their character than the tripartite systems which are typically inflectional” (p. 87). Another important feature of the Slavic style opposition is “that morphologically the opposition between perfective and imperfective aspect is almost wholly independent of the category of tense in Russian” (p. 87). In fact, the categories AOR and IMPF on the one hand and IPFV and PFV on the other hand are radically different also with regard to their compositionality (see 3.1).19 Finally, the authors point out that in (1)

[What did your brother do after dinner yesterday?] He wrote letters

cited from Dahl (1985), all the Slavic languages use the imperfective aspect, whereas normally a perfective form is used, which “strongly suggests that there is a systematic semantic difference between the two kinds of perfectivity distinctions” (Bybee & Dahl 1989: 88). However, somewhat surprisingly, the authors finally come to the conclusion that “the similarity between the perfective meaning evolved historically from bounders and that which evolves from periphrastic constructions (i.e., perfects) is strong evidence for the validity of universal gram-types for perfective aspect” (Bybee & Dahl 1989: 89). Another scholar discussing the similarities and differences between the two oppositions is Lindstedt, who, with regard to (2), uttered in the same context as (1), observes: “After all, Slavonic, Romance and many other languages agree in using what is called the perfective in [(2)], and sentences of this kind form a clear crosslinguistic focus of the postulated category” (Lindstedt 1995). (2)

Russian On napisal pis’mo. he write:PFV:SG letter:ACC ‘He wrote a/the letter’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

292

292

Rolf Thieroff

However, Lindstedt then proposes to make a distinction between a “material bound”, which is crucial for the Slavic perfective, and a temporal bound, crucial for Romance aspect. This distinction also underlies the analysis in Bertinetto & Delfitto (this volume) and at least the notion of material bound can be connected with Johanson’s ‘adterminality’, “envisaging the event in the attainment of its relevant limit: ad terminum” (Johanson, this volume). Thus, following Johanson, Bertinetto & Delfitto, Feuillet (1983), and re-interpreting the facts presented by Dahl, Bybee & Dahl and Lindstedt, I conclude that the Slavic style aspects PFV and IPFV are categories different from the categories AOR and IMPF of the Southern European languages. Whereas much has been written about the categories PFV and IPFV in the Slavic languages, it is less clear in which other languages these categories are found. From Europe, Bybee & Dahl cite Latvian and Lithuanian, Hungarian, and Georgian as languages with “a tendency for bounders to become grammaticized as aspectual markers” (Bybee & Dahl 1989: 86). As for Lithuanian, the existence of the PFV:IPFV opposition seems to be commonly accepted (cf. Sližien˙e 1995). With regard to Hungarian, rather different views can be found in the literature. In Bánhidi, Jókay & Szabó (1975) no aspects or aspect markers are mentioned at all. Abondolo (1987: 587) only remarks that “like preverbs in Slavonic languages, Hungarian coverbs are connected with aspect”. For Dahl, things are less evident (“it appears that we are not dealing with a systematic manifestation of PFV”; “many verbs do not seem to have any aspectually marked counterparts”; Dahl 1985: 86) and the Hungarian forms get the label “PFVd?” (Dahl 1985: 71). De Groot claims that Hungarian does have the PFV:IPFV opposition, the prefix meg- being a perfectivizer (De Groot 1989: 7). Csató claims that the Hungarian meg-forms represent an aktionsart distinction rather than perfective aspect. For a detailed discussion see Csató (1994: 232–237). Still a tendency for bounders to become grammaticalized as aspect markers cannot be denied for Hungarian. Therefore, on Map 7 (see 5.3) Hungarian is shown as belonging to the PFV:IPFV area. Whereas the difference between the AOR:IMPF opposition in the Romance languages, Bulgarian and Armenian on the one hand and the PFV:IPFV opposition in the Slavic languages on the other hand is quite clearcut, there remains one problematic case: In Modern Greek, the position labelled AOR:IMPF clearly has features of both the Romance and the Slavic-style oppositions. Whereas the perfective and imperfective past are used to a large extent like aorist and imperfect in the Romance languages, the Greek perfective can be combined with the future, but not with the anterior, which is the opposite of the situation in the Romance languages, Bulgarian and Armenian. At the same time, the Greek opposition is also rather different from the Slavic one: Whereas the Slavic perfectives are built derivationally, the Greek perfective is constructed inflectionally, and with regard to the semantics Hedin mentions that the boundedness distinction, which plays “an important role for instance

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

293

On the areal distribution of tense-aspect categories in Europe

293

in the Russian aspect system”, “cannot [ ] be crucial for aspect choice in Modern Greek” (Hedin 1995: 236). On the other hand, Mackridge notes that there is “an important group of verbs which exist only in the imperfective” (Mackridge 1985: 103), a feature which is reminiscent of Russian imperfective verbs without perfective partners, whereas I am not aware of verbs which cannot have both aorist and imperfect. Note finally that the Greek perfective past also occurs in current relevance contexts (see Dahl & Hedin, this volume), a usage which is not possible with the prototypical aorists of the other languages mentioned. For further discussion see Thieroff (1995: 19–20).

5.2. Progressives The categories progressive (PROG) and habitual (HAB) are commonly regarded as special cases of imperfectivity. Thus, for example, Comrie classifies progressive and habitual as subdivisions of the imperfective aspect (Comrie 1976: 24–40). In Johanson’s terminology, IPFV, PROG, and HAB are regarded as having in common the same “viewpoint aspect”, namely intraterminality. The difference between the three categories is described as a difference of degrees of focality: Johanson’s nonfocality corresponds roughly to our IMPF, his high focality to our PROG, and his low focality to our HAB (see Johanson, this volume). Similarly, Bertinetto & Delfitto (this volume) say that the progressive and the habitual “belong” to the imperfective aspects. As to how PROG is to be distinguished from the imperfective aspect, Dahl (1985: 92f.) mentions the following features: (i) In contradistinction to PFV:IPFV [i.e., AOR:IMPF; R.T.], which is strongly correlated with the distinction between past and non-past time reference, PROG is usually independent or almost independent of time reference – in other words, it is used both of the present, the past and the future [ ]20 (ii) PROG is quite infrequently extended to habitual meaning [ ] (iii) [ ] PROG is normally used only of dynamic – that is, nonstative – situations [ ]

Dahl also notes that “PROG and PFV:IPFV are very different in the ways in which they are marked – PROG [ ] is rather consistently marked periphrastically, whereas [ ] the marking of PFV:IPFV is usually much less straightforward” (p. 93).21 The progressive is one of the focal areas of the theme group on tense and aspect. For detailed information about the progressive in the Germanic languages, the reader is referred to Ebert (this volume a), about the evolution in the Romance languages to Bertinetto (this volume) and about the progressive constructions in Finnish and Estonian to Tommola (this volume).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

294

294

Rolf Thieroff

The progressive area is the largest aspectual area, covering the western half of Europe but reaching to the east as far as Estonian and (with reservations) Lithuanian. In a large area in the south and southwest, the AOR:IMPF languages and the PROG languages overlap: with the exception of Romanian, all Romance languages have both the AOR:IMPF opposition and PROG. Within the languages of the progressive area, at least two subgroups have to be distinguished, as is also observed by Comrie: In some languages, the distinction between progressive and nonprogressive meaning by means of progressive and nonprogressive forms is obligatory, whereas in others the use of the specifically progressive forms is optional, i.e., the nonprogressive form does not exclude progressive meaning. (Comrie 1976: 33).

English, Irish, West Frisian, Icelandic, and Maltese belong to the first (relatively small) group of languages, where the progressive is grammaticalized in a way that it has to be used obligatorily in the appropriate contexts, at least in some tenses. In English, the grammaticalization of the progressive has progressed the farthest, which can be seen from the fact that the English progressive can be combined with all tenses, including the perfect, and with the habitual periphrasis (as in John used to be writing poems; cf. Comrie 1976: 33). Next to the English progressive is the Icelandic progressive. According to Ebert (1996), “the Icelandic progressive [ ] is fully grammaticalized and underlies none of the restrictions formulated for the other Germanic languages. Similarly to the English progressive, Icelandic vera að V has developed various special functions” (cf. also Kress 1982: 159–162). In Maltese, the progressive cannot be combined with the anterior. For Irish, see 5.3. Similarly to perfects, which universally tend to become pasts (cf. 3.2), progressives, at least unless they are combined with marked tenses, tend to develop into aspectually unmarked forms, in other words, present progressives often become simple presents (or at least imperfective forms; cf. Bybee & Dahl 1989: 82f.). Such a development has occurred in English and Maltese, where the morphologically simple present has with most verbs (exceptions are static verbs) a habitual meaning and in unmarked present time contexts the progressive is obligatory for dynamic situations (for Maltese, cf. Ebert this volume b). A similar development can be observed for telic verbs in Icelandic, which have future time reference with the unmarked finite form (the Present) and present time reference with the progressive (cf. Bonner 1995). The remaining Germanic languages, the Romance languages (with the exception of Romanian), Finnish and Estonian belong to the second group. In these languages, there exist (sometimes several different) progressive periphrases, which are however optional. The situation can be compared to the futures in German, Dutch, Frisian, Icelandic, Finnish and Estonian. In general, the progressives in these languages can be combined with all tense categories (as in the languages of the first group), although there are a few restrictions, especially with ANT and with AOR, as for

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

295

On the areal distribution of tense-aspect categories in Europe

295

example in Italian, where the progressive periphrasis cannot be combined with the Passato prossimo (present anterior), nor with the Passato remoto (AOR). In Lithuanian there are forms with progressive meaning as well, but here PROG is restricted to 1) the past and 2) the non-anterior forms. Thus the situation in Lithuanian differs considerably from both the languages of the first and of the second group. Note that Lithuanian is the only language with both the PFV:IPFV opposition and PROG. Finally, there are two languages whose former progressives can no longer be regarded as instances of PROG. The development from a progressive into the unmarked category has almost come to an end in Basque, where the former progressive is the unmarked present in Modern Basque, with the exception of about sixteen verbs which still can be used with the older present form, which no longer exists for the other verbs (Haase 1994: 283ff.). This is the reason why Basque is outside the progressive area on Map 7. A similar case can be observed in Turkish, where in the present a former PROG morpheme (yor) constitutes the unmarked present today and has become a marker of IMPF in combination with past and future (see Johanson 1994: 261f.).

5.3. Habituals The last tense-aspect category to be treated in the present chapter is the category habitual (HAB), the only category not belonging to the “major gram-types” in Bybee & Dahl (1989).22 Indeed, of the languages of the sample, there are only a few languages with a commonly accepted habitual category. According to Comrie (1976: 27f.), “[t]he feature that is common to all habituals [ ] is that they describe a situation which is characteristic of an extended period of time, so extended in fact that the situation referred to is viewed not as an incidental property of the moment but, precisely, as a characteristic feature of a whole period.” Dahl (1985: 97) illustrates the properties of HAB as follows: It appears that the cases where HAB is typically used are those in which the adverb usually is possible in English [ ] These involve quantification over a set of occasions which is given explicitly or by context. For instance, in (Q.20)23, the set of occasions consists of the time intervals immediately following upon the breakfasts the speaker’s brother had last summer. The use of HAB indicates that what is expressed in the sentence took place in the majority of those occasions. Such sentences differ from generic ones by their lack of lawlikeness.

Of course, as for notional perfectivity, imperfectivity, and progressivity, it is possible to express the meaning of habituality in all languages. However, with regard to the expression of habituality, it seems to be more difficult to decide 1) whether a given expression has to be regarded as a fully grammaticalized morphosyntactic

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

296

296

Rolf Thieroff

category obligatory in the appropriate contexts, and hence as a form belonging to the verbal paradigm; 2) whether it is a weakly grammaticalized category, optional in the appropriate contexts but consisting of a construction whose meaning is not predictable from its elements (like the progressives in most Germanic and Romance languages); or 3) whether we simply are dealing with a lexical expression. For example, in Dahl (1985: 96) German pflegen  infinitive and Swedish bruka  infinitive are mentioned as instances of HAB, but at least for the German construction the traditional view that pflegen  infinitive is a lexical expression seems to be more appropriate.24 Bertinetto (1996) mentions four different habitual periphrases in Italian (solere, essere solito, esser uso, aver l’abitudine di  infinitive), but the mere existence of so many expressions is an indication that there is no grammaticalized HAB in Italian. So, in this section, though it is not excluded that there are (weakly grammaticalized) habitual constructions in other languages of the sample, fully grammaticalized habitual categories are only accounted for in six languages: Czech, Irish, Lithuanian, English,25 Yiddish, and Upper Sorbian. In the first three, HAB is expressed morphologically, i.e., by a bound marker – Czech -va- (Kuˇcera 1981: 178); Irish -dh/-nn (Ó Siadhail 1989: 178); and Lithuanian -da- (Sližien˙e 1995). In the latter, HAB is expressed periphrastically – in English by used to  infinitive; in Yiddish by fleg  infinitive (Aronson 1985: 177, Kiefer 1994); and in Upper Sorbian by a form of by‘be’  l-participle, formally identical with the conditional (Faßke 1981: 253, 266; Lötzsch 1995). Whereas the Czech infix -va- can be used with past and non-past forms, Lithuanian -da- occurs only in combination with PRET, and the English and Yiddish constructions have past time reference too.26 For Irish, the situation is not quite clear. Ó Siadhail (1989: 177) states that “habituality is only apparent when it is combined with a past tense”, but this is only true for the finite main verb. The verb bí ‘be’ has a habitual form (bím ‘I am usually’) distinct from the non-habitual (táim ‘I am’) also in the present, and since the progressive is constructed periphrastically with a finite form of bí, there is also a habitual and a non-habitual present progressive in Irish. And finally, the so-called impersonal form has a habitual form too, marked by the infix -t-.27 On Map 7, the three aspectual distinctions discussed in this section are represented together with the AOR:IMPF opposition discussed in 3.1.28 Note that, while the phenomena discussed so far are with very few exceptions found in coherent areas – there is a PRET, an AOR:IMPF and a GPST area, a PFV: past development and IPFV and a PROG area, an area of the present anterior a “stable present anterior” area – the (fully grammaticalized) habitual is the only category which is not found in one area, but seems to be randomly distributed over Europe, though there are more habituals in the east than elsewhere. It is also worth mentioning that the habitual is not correlated with genetic language groups. Of the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

297

On the areal distribution of tense-aspect categories in Europe

297

AOR:IMPF PFV:IPFV HAB PROG PROG highly grammaticalized Map 7. Aspectual areas

six languages with a fully grammaticalized habitual, one is Celtic, two are Germanic, two are Slavic, and one is Baltic.

6. Conclusion If we take the maps presented in the previous sections together, several areas of similar tense-aspect systems can be identified. Perhaps the most striking correlation is depicted on Map 8, which shows that the overwhelming majority of the languages with a stable perfect also have PROG, and that the overwhelming majority of languages with PROG also have a stable perfect. Exceptions are French (though only the spoken varieties) and varieties of German and Italian, which have PROG but belong to the area of the present anterior past development, and Basque, where the former PROG has developed into imperfective

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

298

298

Rolf Thieroff

stable present anterior (“maritime perfect”) PROG Map 8. Correlation of PROG and perfect

forms. Exceptions with a stable perfect, but lacking PROG, are the four languages in the very south-east of the area investigated, namely Greek, Bulgarian, Turkish, and Armenian. On Map 9, this latter group appears again (area 6), being the only group of languages with the opposition AOR:IMPF, a stable perfect, and lack of PROG. Three more areas with the feature of having a stable perfect can be defined. In the south-west, an area with three languages has the AOR:IMPF opposition and PROG, in the north an area with eight languages has the categories PRET and PROG and a weakly grammaticalized FUT, and in the north-west an area with four languages has PRET and a highly grammaticalized PROG. In the north-east, an area with the category GPST, the PFV:IPFV opposition and absence of PROG can be identified. The last area shown on Map 9, covering the centre of the area investigated, has a status different from the aforementioned five areas. It is the area already identified in section 3.2 (see Map 3), i.e., the area with the present anterior past development. It should be noted that this development is the only feature common to all languages of this area, which means that there are more differences between these languages than

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

299

On the areal distribution of tense-aspect categories in Europe

1 2 3 4 5 6

299

ANT:PRET, Pefect, highly grammaticalized PROG ANT:PRET, Perfect, PROG, weakly grammaticalized FUT GPST, PFV:IPFV, no PROG present anterior past development in different stages AOR:IMPF, Perfect, PROG AOR:IMPF, Perfect, no PROG

Map 9. Major areas of tense-aspect systems

the given picture suggests. Thus, French, Italian, Serbo-Croatian, Romanian, and Albanian belong to the languages with an AOR:IMPF opposition, German and Upper Sorbian have PRET, and Hungarian has neither. Upper Sorbian and Serbo-Croatian have the PFV:IPFV opposition, the others don’t, and German, French, Italian, and Yiddish have PROG, whereas the other languages lack this category. With regard to the criteria given on Map 9, only four languages cannot be assigned to one area. Basque differs from Area 5 on Map 9 in that it has no PROG and Maltese in that is has not the PFV:IPVF opposition. Lithuanian differs from Area 2 in that it has a fully grammaticalized FUT and the PFV:IPFV opposition, and Karaim29 differs from Area 3 in that it has neither the PFV:IPFV opposition, nor GPST. Note finally that, of course, the languages of each area still differ in various respects. Only with regard to the features indicated do they belong to the same area. A closer investigation of the tense-aspect categories and an inclusion of the mood

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

300

300

Rolf Thieroff

categories would possibly lead to other, probably more numerous and smaller, areas, but this is a task which has to be left to the post-EUROTYP era.

Notes 1. In Southern Germany as well as in Northern Italy a distinction has to be made between 1) the dialects (called “vernaculars” elsewhere in this volume) spoken in these regions and 2) the regional varieties of the standard language. Note that “Southern German” (SGrm) and “Northern Italian” (NIt) stand for the dialects of these regions, not for the local varieties of the standard language. 2. Cf. also Dahl (this volume a). For descriptions of languages not included in this chapter see Johanson (this volume). 3. Cf. also Dahl (1985: 67): “[ ] the English sentence He would have been swimming contains at least the following TMA markers: the auxiliary will, the suffix -d, the auxiliary have (in combination with a past participle) and the copula be (in combination with an -ing form).” 4. Similarly, Squartini & Bertinetto (this volume) describe idiosyncrasies of the present anterior in Portuguese, which do not hold for the category anterior when combined with other tense aspect categories. 5. There is also a small class of about 16 verbs with synthetic forms (Haase 1994: 283–287). These verbs are neglected in the following discussion. 6. For the inclusion of non-finite verb forms, see Thieroff (1994b: 129–131). 7. For the (non-)obligatoriness of FUT in the Germanic languages, see 4 and Dahl (this volume b). For other future constructions than the ones displayed in Table 6, see Dahl (this volume b). 8. For Bulgarian and Serbo-Croatian, a second categorization of perfectivity has to be added (see 5.1) 9. For a discussion whether the aorist with non-past time reference is to be regarded as a subjunctive, see Mackridge (1985: 102ff.). 10. Note that the northern Italian dialects differ from the remaining languages of the area by having neither the ANT:PRET nor the ANT:AOR:IMPF system; see below. 11. Since there is no preterite in Southern German, the same form probably would have to be categorized as one modal category, e.g., “conditional”. 12. Note the difference between the stages of development of present anteriors shown in Table 8 and the four stages in the Romance languages proposed by Harris (1982), referred to by Squartini & Bertinetto (this volume). Harris’ stage I lies chronologically before stage 0 in Table 8: in this stage, have/be  past participle is still a resultative and not yet a present anterior. Harris’ stage II represents a special development of this form (in Europe restricted to Portuguese), which is not part of the “aoristic drift”, as is argued by Squartini & Bertinetto. Thus it is Harris’ stage III which corresponds to stage 0 in Table 8, whereas stages 1 and 2 in Table 8 correspond to Harris’ stage IV. Stage 3 in Table 8 is not covered by Harris, since in this stage the aorist has

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

301

On the areal distribution of tense-aspect categories in Europe

13. 14. 15. 16.

17. 18. 19.

20. 21.

22. 23. 24.

25. 26.

301

completely vanished. Finally, stage 4 in Table 8 is not reached by any Romance language. In addition to the languages displayed on Map 4, Lindstedt mentions North German and South Italian dialects; these are not taken into consideration in the present chapter. In Polish, the pluperfect “is extremely rare and is found only in a formal literary style” (Stone 1987: 363). In Kotyczka (1987) the pluperfect is not even mentioned. However in Corbett (1987: 404) an anterior  past pluperfect is still mentioned. In Maltese there are two devices for the expression of future time reference, the auxiliary ikun and the particle sa, which function differently with different verb classes (cf. Ebert this volume b). Both are regarded here as markers of FUT, though ikun also has other functions. For details see Ebert, this volume b, Fabri (1995) and Thieroff (1995a: 31– 33). For a discussion of cases with an obligatory FUT in German, see Thieroff (1992: 125– 128). Cf. also Feuillet (1983: 74). This is not entirely clear in Thieroff (1994a). There, AOR and IMPF were treated as if they were composed of two categories, remoteness  perfective and remoteness  imperfective respectively. However, this is neither in conformity with the semantics of AOR and IMPF, nor with their morphology. Note that this criterion also holds for the difference between the AOR:IMPF and the PFV:IPFV oppositions. See above. Distinguishing the categories AOR and IMPF on the one hand from the categories PFV and IPFV on the other hand, we can say more about the marking of these categories: AOR and IMPF are usually marked morphologically, PFV and IPFV derivationally, by what Bybee & Dahl call “bounders” (see above). In Johanson (this volume), habitual constructions are not even regarded as belonging to “viewpoint aspects”. Question 20 of Dahl’s questionnaire: “What your brother usually DO after breakfast last summer?” The verb in capitals stands for the uninflected lexeme. Pflegen, together with drohen ‘threaten’ (Die Wand drohte einzustürzen ‘the wall threatened to collapse), scheinen ‘seem’ (Karl scheint zu schlafen ‘Karls seems to be asleep’) and versprechen ‘promise’ (Der Sommer verspricht schön zu werden ‘The summer promises to be beautiful’) belongs to a small class of verbs which syntactically behave like modal auxiliaries (dürfen ‘may’, können ‘can’, müssen ‘must’ etc.), with the only difference that they take the infinitive with the particle zu ‘to’ (Er pflegt zu kommen ‘He usually comes’ vs. Er darf kommen ‘He may come’). For an exhaustive discussion of the syntax of pflegen, drohen, scheinen, versprechen (called “Halbmodalverben”, ‘semi-modal auxiliaries’ by the author), see Eisenberg (1994: 382–385). Bertinetto (1996) comes to the conclusion that the English used to construction is not really a habitual, but his arguments cannot be discussed here. For this reason, Dahl labels the English used to construction as HABPAST, which “is used as a label for categories which are mainly used for habitual sentences with past time combined with a regular past reference and are not analysable as consisting of HAB tense” (Dahl 1985: 100), whereas the Czech iterative verb is labelled “HAB”.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

302

302

Rolf Thieroff

27. For further details, see Ó Siadhail (1989: 177f), Ó Baoill (1994), Thieroff (1994a: 17– 20). 28. Note that the Northern Italian dialects differ from the remaining languages of the AOR: IMPF-area in that they do not have the category ANT. 29. For a full description of the tense-aspect system of Karaim, see Csató (this volume).

References Abondolo, Daniel 1987 “Hungarian”, in: Bernard Comrie (ed.), 577–592. Aronson, Howard I. 1985 “On aspect in Yiddish”, General Linguistics 25: 171–188. Bánhidi, Zoltán, Zoltán Jókay & Dénes Szabó 1975 Lehrbuch der ungarischen Sprache. Munich: Hueber. Bertinetto, Pier Marco 1992 “Le strutture tempo-aspettuali dell’italano e dell’inglese a confronto”, in: L’Europa linguistica: contatti, contrasti, affinità di lingue. Società di linguistica italiana. SLI 30. Rome: Bulzoni. 1996 “Le perifrasi abituali in italiano ed in inglese”, Studi Orientali e Linguistici 6, 1995/96: 117–133. this volume “The progressive in Romance, as compared with English”. Bertinetto, Pier Marco & Denis Delfitto this volume “Aspect vs. Actionality. Why they should be kept apart”. Bertinetto, Pier Marco, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.) 1995 Temporal Reference, Aspect and Actionality. Vol. 2: Typological Perspectives. Torino: Rosenberg & Sellier. Beyrer, Arthur, Klaus Bochmann & Siegfried Bronsert 1987 Grammatik der rumänischen Sprache der Gegenwart. Leipzig: VEB Verlag Enzyklopädie. Binnick, Robert I. 1991 Time and the Verb. A Guide to Tense and Aspect. New York & Oxford: Oxford University Press. Bonner, Maria 1995 “Zum Tempussystem des Isländischen”, in: Rolf Thieroff (ed.), 37–58. Buchholz, Oda & Wilfried Fiedler 1987 Albanische Grammatik. Leipzig: Verlag Enzyklopädie. Bybee, Joan L. 1985 Morphology. A study of the relation between meaning and form. Amsterdam & Philadelphia: Benjamins. Bybee, Joan L. & Östen Dahl 1989 “The Creation of Tense and Aspect Systems in the Languages of the World”, Studies in Language 13: 51–103. Comrie, Bernard 1976 Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. Comrie, Bernard (ed.) 1987 The World’s Major Languages. London & Sydney: Croom Helm. Corbett, Greville 1987 “Serbo-Croat”, in: Bernard Comrie (ed.), 391–409.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

303

On the areal distribution of tense-aspect categories in Europe

303

Csató, Éva Ágnes 1994 “Tense and Actionality in Hungarian”, in: Rolf Thieroff & Joachim Ballweg (eds.), 231–246. this volume “Some typological features of the viewpoint aspect and tense system in spoken NorthWestern Karaim”. Dahl, Östen 1985 Tense and Aspect Systems. Oxford & New York: Basil Blackwell. this volume a “The tense-aspect systems of European languages in a typological perspective”. this volume b“The grammar of future time reference in European languages”. Dahl, Östen & Eva Hedin this volume “Current relevance and event reference”. Dardano, Maurizio & Pietro Trifone 1985 La lingua italiana. Morfologia. Sintassi. Fonologia. Formazione delle parole. Lessico. Nozioni di linguistica e sociolinguistica. Bologna: Zanichelli. Ebert, Karen H. 1996 “Progressive aspect in German and Dutch”, Interdisciplinary Journal for Germanic Linguistics and Semiotic Analysis 1: 43–62. this volume a “Progressives in Germanic languages”. this volume b “Aspect in Maltese”. Eisenberg, Peter 1994 Grundriß der deutschen Grammatik. 3rd edition, Stuttgart: Metzler. Fabri, Ray 1995 “The tense and aspect system of Maltese”, in: Rolf Thieroff (ed.), 327–343. Faßke, Helmut 1981 Grammatik der obersorbischen Schriftsprache der Gegenwart. Morphologie. Bautzen: Domowina. Fenn, Peter 1987 A semantic and pragmatic examination of the English perfect. Tübingen: Narr. Feuillet, Jack 1983 “Methodologische Probleme des Aspekts”, in: Norbert Reiter (ed.). Ziele und Wege der Balkanlinguistik. Wiesbaden: Harrassowitz. 66–80. Fromm, Hans 1982 Finnische Grammatik. Heidelberg: Winter. Groot, Casper de 1989 Predicate structure in a Functional Grammar of Hungarian. (Functional Grammar Series 11.) Dordrecht: Foris. Gvozdanovi´c, Jadranka 1995 “Western South Slavic tenses in a typological perspective”, in: Rolf Thieroff (ed.), 181– 193. Haase, Martin 1994 “Tense and aspect in Basque”, in: Rolf Thieroff & Joachim Ballweg (eds.), 279–292. 1995 “Tense, aspect and mood in Romanian”, in: Rolf Thieroff (ed.), 135–152. Hakulinen, Lauri 1957 Handbuch der finnischen Sprache. 1. Band. Wiesbaden: Harrassowitz. Harris, Martin B. 1982 “The ‘past simple’ and ‘present perfect’ in Romance”, in: Martin B. Harris & Nigel Vincent, Studies in the Romance verb. London: Croom Helm. Harris, Martin & Nigel Vincent (eds.) 1988 The Romance Languages. London: Routledge. Hedin, Eva 1995 “The tense aspect system of Modern Greek”, in: Rolf Thieroff (ed.), 233–252.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

304

304

Rolf Thieroff

Hundertmark-Santos Martins, Maria Teresa 1982 Portugiesische Grammatik. Tübingen: Niemeyer. Johanson, Lars 1994 “Türkeitürkische Aspektotempora”, in: Rolf Thieroff & Joachim Ballweg (eds.), 247– 266. this volume “Viewpoint operators in European languages”. Karlsson, Fred 1983 Finnische Grammatik. Hamburg: Buske. Kiefer, Ulrike 1994 “Die Tempusformen im Jiddischen”, in: Rolf Thieroff & Joachim Ballweg (eds.), 135– 148. Klein, Wolfgang 1992 “The present perfect puzzle”, Language 68: 525–552. 1994 Time in language. London & New York: Routledge. Kortmann, Bernd 1995 “Compositionality and the perfect”, in: Wolfgang Riehle & Hugo Keiper (eds.). Anglistentag 1994 Graz. Proceedings. Volume XVI. Tübingen: Niemeyer. 183–199. Kotyczka, Josef 1987 Kurze polnische Sprachlehre. Berlin: Volk und Wissen. Kress, Bruno 1982 Isländische Grammatik. München: Hueber. Kuˇcera, Henry 1981 “Aspect, markedness, and t0”, in: Philip Tedeschi & Annie Zaenen (eds.). Syntax and Semantics 14: Tense and Aspect. New York: Academic Press, 177–189. Lindstedt, Jouko 1995 “Understanding perfectivity – understanding bounds”, in: Bertinetto et al. (eds.), 95– 103. this volume “The perfect – aspectual, temporal and evidential”. Lötzsch, Ronald 1995 “Das sorbische Tempussystem”, in: Rolf Thieroff (ed.), 167–179. Mackridge, Peter 1985 The Modern Greek Language. A Descriptive Analysis of Standard Modern Greek. Oxford: Clarendon. Mallinson, Graham 1988 “Rumanian”, in: Martin Harris & Nigel Vincent, (eds.), 391–419. Metslang, Helle 1994 Temporal relations in the predicate and the grammatical system of Estonian and Finnish. (Oulun yliopiston Suomen ja Saamen kielen laitoksen tutkimusraportteja 39). Oulu: Oulun Yliopisto. Metslang, Helle & Hannu Tommola 1995 “Zum Tempussystem des Estnischen”, in: Rolf Thieroff (ed.), 299–326. Nichols, Johanna 1992 Linguistic diversity in space and time. Chicago: The University of Chicago Press. Ó Baoill, Donall P. 1994 “Tense and aspect in Modern Irish”, in: Rolf Thieroff & Joachim Ballweg (eds.), 201– 216. Ó Siadhail, Mícheál 1989 Modern Irish. Grammatical Structure and Dialectal Variation. Cambridge etc.: Cambridge University Press. Olli, John B. 1958 Fundamentals of Finnish Grammar. New York.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

305

On the areal distribution of tense-aspect categories in Europe

305

Parkinson, Stephen 1988 “Portuguese”, in: Martin Harris & Nigel Vincent (eds.), 131–169. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik 1985 A comprehensive grammar of the English language. London & New York: Longman. Rehder, Peter 1991 “Das Serbokroatische”, in: Peter Rehder (ed.), Einführung in die slavischen Sprachen. Darmstadt: Wissenschaftliche Buchgesellschaft, 46–60. Reichenbach, Hans 1947 Elements of symbolic logic. New York: The Free Press & London: Collier-Macmillan. Sližien˙e, Nijol˙e 1995 “The Tense System in Lithuanian”, in: Rolf Thieroff (ed.), 215–232. Squartini, Mario & Pier Marco Bertinetto this volume “The Simple and Compound Past in Romance languages”. Stone, Gerald 1987 “Polish”, in: Bernard Comrie (ed.), 348–366. Stump, Gregory 1985 The semantic variability of absolute constructions. Dordrecht: Reidel. Thieroff, Rolf 1992 Das finite Verb im Deutschen. Tempus–Modus–Distanz. (Studien zur deutschen Grammatik 40) Tübingen: Narr. 1994a “Inherent Verb Categories and Categorizations in European Languages”, in: Rolf Thieroff & Joachim Ballweg (eds.), 3–45. 1994b “Das Tempussystem des Deutschen”, in: Rolf Thieroff & Joachim Ballweg (eds.), 119– 134. 1995a “More on Inherent Verb Categories in European Languages”, in: Rolf Thieroff (ed.), 1–36. Thieroff, Rolf & Monika Budde 1995 “Are Tense and Aspect Categories?”, in: Bertinetto et al. (eds.), 47–62. Thieroff, Rolf (ed.) 1995b Tense Systems in European Languages II. (Linguistische Arbeiten 338) Tübingen: Niemeyer. Thieroff, Rolf & Joachim Ballweg (eds.) 1994 Tense Systems in European Languages. (Linguistische Arbeiten 308) Tübingen: Niemeyer. Tommola, Hannu 1992a “The Marking of Future Time Reference in Estonian”, in: Östen Dahl, Casper de Groot, & Hannu Tommola (eds.), Future Time Reference in European Languages I. Bulgarian, Estonian, German, Hungarian, Continental Scandinavian, Züritüütsch. (E UROTYP Working Papers VI:2.) Stockholm. 12–28. 1992b “The Marking of Future Time Reference in Finnish”, in: Östen Dahl, Casper de Groot, & Hannu Tommola (eds.), Future Time Reference in European Languages II. Dutch, Finnish, Modern Greek, Italian, Lezgian, East Slavic, Turkish. (E UROTYP Working Papers VI:3.) Stockholm. 12–28. 1994 “Zum Tempus im Finnischen”, in: Rolf Thieroff & Joachim Ballweg (eds.), 219–229. this volume “Progressive aspect in Baltic-Finnic”. Wheeler, Max W. 1988 “Catalan”, in: Martin Harris & Nigel Vincent (eds.), 170–208.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

306

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

307

Future Time Reference

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

308

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

309

Östen Dahl

The grammar of future time reference in European languages

“I think I go home now.” Utterance ascribed to Greta Garbo after a row with her Hollywood film director, illustrating differences among European languages in the marking of future time reference

1. Introduction As the title suggests, this chapter is about the ways in which speakers of European languages talk about the future; more specifically, the grammatical devices that are used in doing so. At the centre of the investigation, we will necessarily find the things traditionally called future tenses. Since their theoretical status has been the object of considerable controversy, and since we want to be open for other potentially interesting phenomena, the delimitation of the area of study is kept deliberately vague. Future time reference (FTR) was the first of the focal areas in the work of the E UROTYP Theme Group on Tense and Aspect. Questionnaire data were collected for about 30 languages, and on the basis of them a number of descriptions on the marking of FTR in different European languages were written and issued as working papers. Much of what will be said below is based on the empirical material in the questionnaires and the working papers. In this chapter, we shall first look at some major semantic/pragmatic distinctions relevant to the marking of future time reference and at what future grams look like in a typological perspective. Then, we shall see an example of how ongoing grammaticalization processes are reflected in the questionnaire material. Finally, we shall look at the areal distribution of future grams in Europe, with special attention to what will be called the ’futureless area’ of Northern Europe.

2. Predictions, intentions and scheduling From the epistemological point of view, the future has a rather different status from both the present and the past. We cannot perceive or remember future states of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

310

310

Östen Dahl

affairs, and it has been disputed whether statements about the future can be said to have a determinate truth value. Still, we do talk about the future, and there may be different grounds for our doing so. To start with, we may have intentions relating the future. This is by definition restricted to things that are under our control (or at least believed to be so) and prototypically shows up in sentences with a human subject (who is also the bearer of the intentions), as in the following example from Lewis Carroll: (1)

English (Alice in Wonderland) ‘[I know SOMETHING interesting is sure to happen,’ she said to herself, ‘whenever I eat or drink anything;] so I’ll just see what this bottle does.

Further, we may, on the basis of more or less well-grounded considerations, make predictions about the future. In the typical case, predictions concern courses of events that are not within human control or at least not within the control of the speaker, as in the following example. (2)

English (Alice in Wonderland) There was nothing else to do, so Alice soon began talking again. ‘Dinah’ll miss me very much to-night, I should think!’ (Dinah was the cat.)

We may thus distinguish intention-based and prediction-based future time reference. A straightforward grammatical opposition based on the distinction between intention-based and prediction-based FTR is less common than one would perhaps think in view of the apparent cognitive salience of that distinction. Its importance lies rather in the observation that markers that are originally restricted to intention-based FTR tend to develop into general future markers, which include prediction-based FTR as central cases but can in the normal case still be used for intention-based FTR. In fact, whether FTR is overtly and obligatorily marked in prediction-based sentences can be used as one of the major criteria for whether it is grammaticalized in a language or not. To illustrate the difference, consider the following two excerpts from real-life weather forecasts, the first from a British, the second from a Finnish newspaper: (3)

English Outbreaks of rain will clear on Monday to leave a mix of sunshine and showers across the country. Longer periods of rain are likely midweek, especially in the west. It will be mostly cool and windy. Cool and unsettled conditions over much of Scandinavia will extend into central and western Europe during Tuesday and Wednesday. Mediterranean coasts will remain sunny and very warm.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

311

The grammar of future time reference in European languages

(4)

311

Finnish Sää kylmenee, mutta keskiviikkona tuulee idästä ja pyryttää lunta. Lämpötila kohoaa tilapäisesti nollaan tai jopa vähän suojan puolelle. Torstain tienoilla voi olla jopa kymmenisen pakkasastetta. Viikonlopulla taas lauhtuu, pilvistyy ja alkaa sataa lunta. [(Rather literal translation:) ’The weather becomes cooler, but on Wednesday it blows from the east and there is drifting snow. The temperature rises temporarily to zero or even a little higher. By Thursday it can already be around ten degrees below zero. During the weekend it again becomes milder, overcast and begins to snow.’]

In the English text, the auxiliary will is used systematically throughout the text (except when there is another modal expression in the sentence, such as are likely). By contrast, the Finnish text is wholly in the Present tense. There are also no markers of modality (except for one occurrence of the modal voi ‘may’). The literal translation into English sounds rather deviant if future auxiliaries are not supplied. Arguably, then, future time reference is grammaticalized in English in a way it is not in Finnish. As we shall see later, this criterion divides the European languages into two large groups. A further notional category that turns out to be important is scheduling. It is well known that in a sentence such as (5), English tends to use the Present tense although the time referred to is in the future: (5)

English (FTRQ: 89) [According to the timetable] the train leaves at noon.

In fact, in many if not most languages, this kind of sentence is treated in a way that does not mark it grammatically as having non-present time reference. This seems to hold even for languages where future time reference is otherwise highly grammaticalized. However, in addition, we find that the Present tense is used in many European languages in sentences such as the following questionnaire sentence: (6)

(FTRQ: 37) [Talking about the speaker’s immediate plans:] I GO to town.

Some examples of translations of this sentence from the questionnaire material using a Present tense are: (7)

Russian (FTRQ: 37) Idu v gorod. go:IPFV:PRS:1SG to town:ACC

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

312

312

Östen Dahl

(8)

French (FTRQ: 37) Je vais en ville. I go:PRS:1SG to town

(9)

Serbian/Croatian (FTRQ: 37) Idem u grad. go:PRS:1SG to town:ACC

Notice that the verb used in the Russian example is Imperfective. In other words, the form used is one whose typical use is with present time reference, unlike the Perfective Present. However, in English, this is a typical context for a Progressive – the Simple Present is hardly acceptable here (cf. the quotation from Greta Garbo above): (10)

English (FTRQ: 37) I’m going to town.

It is possible that there is a common prototype for this use of the Present in Russian and French and the Progressive in English that involves a number of different factors. The examples cited in the literature tend to involve movement verbs1 and refer to relatively close points in the future. What is probably more relevant, though, is the element of planning, in the stronger sense of preparation. One can truly say I’m going to town when one has started to prepare oneself for the trip. There is a clear analogy between this use of progressives and presents and the one found with achievements in Vendler’s sense, that is, when He is reaching his goal is used in the sense ‘He is close to his goal’. Given the apparent naturalness of this way of talking, one might think that there would be as little interlinguistic variation here as in the case of sentences like (5). However, it turns out that some languages behave somewhat unexpectedly here. In the Scandinavian languages, marking of future time reference is relatively seldom obligatory. Still, in (TMAQ: 37), none of the five Scandinavian informants, representing at least four different varieties of Swedish and Norwegian, chose a present tense2. Even if this fact might be accidental, something seems to be going on here. Consider a prototypical context for a ‘preparatory’ use of a present or progressive: meeting a friend in a travel agency, I draw the conclusion that he is planning a trip somewhere. The natural things to say in Russian and English, respectively, would then be: (11)

Russian Kuda edeš’? whither go:IPFV:PRS:2SG

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

313

The grammar of future time reference in European languages

(12)

313

English Where are you going?

However, in Swedish, the auxiliary ska(ll) is the primary choice, the bare Present sounding somehow out of place: (13)

Swedish Vart ska du åka? whither shall you:SG go:INF

In other words, ‘preparatory’ contexts do not seem conducive to overt FTR marking in Swedish and Norwegian. ‘Preparatory’ contexts are confusing since they might be seen as a sub-type of intention-based FTR. It is evident, however, that they are treated in special ways in many languages.

3. Future grams from a typological point of view In the cross-linguistic studies reported in Bybee & Dahl (1989) (based on Bybee 1985 and Dahl 1985) and Bybee, Perkins & Pagliuca (1994), the traditional notion of a future tense finds its counterpart in the cross-linguistic gram-type future. Future grams develop out of a number of sources, well-known from traditional historical linguistics, such as verbs of movement (go and come), obligation, and volition, but also, for instance, from markers of progressive aspect. Sometimes, future grams may be “residual grams” in the sense that they develop as the left-over of an earlier imperfective which has yielded its central territory to an expanding progressive. (See Bybee et al. 1994, Chapter 7, for a general account of futures and their paths of developments.) A central issue in the controversy about the theoretical status of future grams concerns the distribution of labour between temporal, modal, and aspectual elements in their meanings and whether to subsume them under the traditional categories of tense, mood/modality, or aspect. It may be noted here that while the sources of future grams typically have exclusively non-temporal meanings, the temporal elements tend to grow stronger during the course of grammaticalization (“temporalization” in the terms of Fleischman 1983). One reason why future grams tend to exhibit a mixture of different kinds of semantic elements is that many of them retain part of their original meaning, at least in some contexts. For instance, English will still may indicate willingness, and shall sometimes retains an obligational element (as in legal documents). On the other hand, when a gram has undergone temporalization, new, non-temporal uses may develop (such as inferential uses with non-future time reference).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

314

314

Östen Dahl

It may be argued that whereas more or less developed future grams are very common in languages, full grammaticalization is less common. At least, it is noteworthy that some contexts are typically quite late in being reached by an expanding future gram. Cases in point are some types of subordinate clauses, most notably temporal and conditional clauses. As a general tendency, time reference is less systematically marked in these types of clauses, in particular conditionals. Thus, not only forms normally referring to the present (such as the Present tense in English) but also forms that would refer to the past when used in a main clause (such as the Semitic Perfective or the Japanese Past) show up here. (In addition, many languages use non-finite forms in such contexts.) These facts may be related to the non-assertiveness of the clause-types under discussion. But it is also clear that the lack of specificity in time reference plays a role here. Temporal clauses more often refer to specific points in time than do conditional clauses, and they also tend to acquire future marking first. When such marking is optional for a clause type, specificity often plays a role in determining the choice. In the Balkan area, languages at different stages of introducing future marking into temporal and conditional clauses can be found (see Hedin, this volume). Relevant to the issue of grammaticalization is also the way in which the gram is expressed or marked. The most salient parameter is boundness: in general, bound morphemes as primary markers of grams show up only at late stages of grammaticalization. Both Dahl (1985) and Bybee, Pagliuca & Perkins (1994) find that periphrastic (free) and morphological (bound) expression are more or less equally probable for future grams; in this respect this gram-type differs both from, for instance, past tenses (which are predominantly bound) and progressives (which tend to be periphrastic). However, bound and free futures differ from each other. Bybee, Pagliuca & Perkins (1991) studied the covariation of form and meaning in the grammaticalization of future grams and found significant relationships between the ways they were expressed and the stage they had reached in their semantic development. As noted in Bybee & Dahl (1989), the future grams in an expanded version of the sample used in Dahl’s earlier investigation (Dahl 1985) which were systematically used in both temporal and conditional clauses were all bound. In the expanded sample, the languages in question were the following: Alawa, Bandjalang (Australian), Oneida, Seneca (Algonquian), Hebrew (Semitic), Hindi/Urdu, Kurdish, Latvian (Indo-European), Georgian (Kartvelian). The futures in these languages are also characterized by a number of other indicators of high degree of grammaticalization: close adherence to the generalized cross-linguistic profile of the gram-type, high frequency of use and tendency to obligatory use in central cases (to the extent that all these things can be judged about from the questionnaire data). We may note a certain concentration to certain language families and areas here. Since Dahl’s sample is rather heavily biased areally and genetically, we cannot draw very strong conclusions from this. For the purposes of this chapter, however, the ab-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

315

The grammar of future time reference in European languages

315

sence of any languages from Western or Southern Europe, two areas that are rather over-represented in the sample, should be noted. The conclusion is that full grammaticalization of futures is not common in large parts of Europe. This is a point that we shall return to.

4. Grammaticalization of future time reference in progress It is common for there to be more than one gram moving along the same grammaticalization path. In such situations, the grams will have partly overlapping functions and will compete with each other. Good examples of this are found in the Western Romance languages, where the periphrastic de-andative construction (e.g., French je vais travailler) is gradually taking over the territory of the older, inflectional Romance de-obligational future (e.g., French je travaillerai). Since we have questionnaires from several speakers of French and Spanish, we are in a position to see how the competition between the two constructions shows up in the ways speakers choose between them in different contexts. As is probably typical of grammaticalization in progress, the choice seems to depend on several factors, which may be both stylistic and semantic. For French, Schlyter & Sandberg (1994) note that the (inflectional) Future tense is used more in formal and written language, and is favored by a combination “prediction + 3rd person + remoteness in time”, whereas the de-andative aller construction is characteristic of the spoken language and is favoured by the combination “intention + 1st person + immediateness”. The questionnaire data (from six speakers) that support the semantic generalizations are tabulated in Table 1. What is shown there is, for the respective sets of examples in the FTR Questionnaire, the percentage of responses where a form was chosen as the only alternative (OBL), where it was chosen as an optional alternative (OPT), and the sum of these two (TOT). As we can see, the situation is made more complicated by the existence of a third choice, the present tense (see the discussion of scheduling and preparation above). We may note, however, that there is a very pronounced cline with respect to the remoteness dimension, with the future tense at the remote end and the aller construction at the non-remote end of the scale. There is also a relatively clear difference between intention-based and prediction-based FTR, where the latter tends to favor the future tense. What is perhaps most notable with respect to this distinction, however, is the almost total exclusion of the present tense in prediction-based contexts. The significance of the distinction between 1st and 3rd person is less clear-cut. It appears that 1st person contexts favor the choice of the present rather than any of the marked forms. The situation for Spanish, which is tabulated in Table 2 (number of informants: 3), is similar, with even steeper curves for remoteness and the distinction between intention and prediction (see also the discussion in Hermerén, Schlyter & Thelin 1994).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

316

316

FUTURE TENSE OBL OPT TOT COGNITIVE BASE REMOTENESS

1 person intention 3 person intention Prediction Immediate “This evening” “Tomorrow” Distant

22 11 45 3 27 41 72

11 8 5 5 8 15 5

33 19 50 8 35 56 77

ALLER CONSTRUCTION OBL OPT TOT 39 62 35 61 45 18 10

9 11 5 10 8 6 6

48 73 40 71 53 24 16

PRESENT TENSE OBL OPT TOT 21 14 3 24 16 10 10

13 2 2 5 4 12 0

34 16 5 29 20 22 10

Table 2. The frequency of Spanish FTR devices in different sets of questionnaire examples FUTURE TENSE OBL OPT TOT COGNITIVE BASE REMOTENESS

1 person intention 3 person intention Prediction Immediate “This evening” “Tomorrow” Distant

53 33 72 13 45 79 33

0 0 2 0 0 0 0

53 33 74 13 45 79 33

IR A CONSTRUCTION OBL OPT TOT 35 62 12 76 30 0 33

0 0 2 0 0 0 0

35 62 14 76 30 0 33

PRESENT TENSE OBL OPT TOT 11 3 8 9 22 4 0

0 0 0 0 0 0 0

11 3 8 9 22 4 0

Östen Dahl

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Table 1. The frequency of French FTR devices in different sets of questionnaire examples

317

The grammar of future time reference in European languages

317

Summing up, the systems we have described here have several features that seem characteristic of grammaticalization in progress: (i) competition between an older, more grammaticalized and a younger, less grammaticalized construction; (ii) no absolute acceptability judgments; (iii) dependence on several factors, both stylistic and semantic; (iv) differences between written and spoken language. We seem to be rather far from the structuralist ideal of a system with neat oppositions and simple meaning correlates.

5. European future gram families In this section, I shall survey the different future gram families that characterize the languages spoken in Europe, or rather Europe excluding the non-Slavic-speaking parts of Russia and some other outlying parts like Malta and Turkey. The motivation for this delimitation is twofold: first, it makes sense from the areal-linguistic point of view; second, the available information above all about the Caucasian languages does not make it possible to map that region in sufficient detail. ‘Gram family’ is a somewhat vague term that I use for grams with related functions and diachronic sources that show up in genetically and/or geographically related groups of languages. To take one example, constructions formed with a verb meaning ‘to go’, with uses sometimes referred to as ‘prospective’, show up in a number of languages in Western Europe, both in the Germanic and the Romance group. Thus, the usefulness of the term ‘gram family’ is based on the tendency for grammaticalization processes to cluster areally and genetically. As we shall see, most of the grammatical devices that are used in a regular fashion to signal future time reference are parts of such clusters. What such a survey demonstrates is the areal nature of grammaticalization processes: in general, the distribution of gram families fits the Wellentheorie rather than the Stammbaumtheorie of linguistic change. Features that at first glance seem to characterize a whole language family, such as the Romance de-obligative construction, on closer inspection turn out be explainable only as a spread which started after the break-up of the parent language and which has never reached the whole territory of the language group in question. One notable fact is the relative independence of different gram families; in general, each of them has a unique distribution, suggesting that the introduction of new grams into a language to a significant extent is independent of what grams that language already has. In Figure 1, a schematic view of the distribution of the major gram families is presented.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

318

318

Östen Dahl

Figure 1. Main gram families with future time reference in Europe. Shaded areas denote inflectionally expressed grams.

5.1. Indo-European inflectional futures Whether earlier stages of Indo-European had an inflectional future is an open question. Admittedly, a number of future forms found in different branches of IndoEuropean (Greek, Indo-Aryan, Baltic) may be traceable to a common origin (most probably a desiderative suffix), but in several branches there are no future forms at all (as noted above) and other forms do not have a clear etymology. Among modern Indo-European languages in Europe with inflectional futures going back before the time of the first written sources are Irish in the Celtic branch and the two Baltic languages Latvian and Lithuanian.

5.2. Romance inflectional future The Romance inflectional future, one of the classic examples of the development of an inflectional tense-aspect gram from a periphrastic source – the Latin obligative construction Infinitive + habere ‘have’ – has spread over a large part of the Romance-speaking territory. It is thus found in at least the following modern Romance languages: Italian, French, Spanish, Portuguese, Occitan, Catalan, Romansh. It is perhaps less well-known that there are areas where it is not found or where it is used only to a restricted extent. Most importantly, it did not extend to the Romanian area. But also in Italy, it is not found in the vernaculars3 south of a line Viterbo–

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

319

The grammar of future time reference in European languages

319

Perugia–Ancona (Rohlfs 1968: 333), a fact that is probably behind the relatively large indeterminacy in the use of the future in Standard Italian. It should also be mentioned that the inflectional future seems to be significantly less frequent in Latin American Spanish than in the Spanish spoken in Spain.

5.3. Ukrainian inflectional imperfective future This formation, which is restricted to Ukrainian and only occurs with imperfective verbs, is structurally analogous to the more famous Romance inflectional future in that it derives from the merger of the infinitive of the main verb with a postposed auxiliary, which is originally the verb ‘have’ (imati), for instance pisatime ‘(he) will write’. The form occurs alongside of the copular imperfective future (see below) and there seems to be no difference in meaning (Dahl 1992). (‘Have’ was also used as a future-marking auxiliary in the other East Slavic languages at an earlier stage, but the inflectional variety is not attested.)

5.4. West European de-andative construction The gram family represented in English by the be going to construction is found in a number of Germanic and Romance languages located in a contiguous area in Western Europe: English, Dutch, French, Spanish, Portuguese. The source of the construction, involving a verb with the meaning ‘go’ followed by an infinitive, is still fairly transparent in most cases. Semantically, it tends to be used for intentionbased non-remote future time reference, but at least in some languages it is extended to non-intentional cases, especially ‘imminent’ ones such as It is going to rain. The construction appears to be gaining ground in many dialects and is undergoing formal changes characteristic of further grammaticalization (such as the reduction in English of going to gonna). In Dahl (1985), the constructions mentioned here were tentatively subsumed under a cross-linguistic gram-type ‘prospective’ (cf. also Comrie 1976 for a similar treatment). The evidence for the existence of a such a gram-type as distinct from early futures in general is somewhat shaky, however. The de-andative constructions seems to have developed relatively late. In English, it spread in the 17th century.

5.5. Germanic de-obligative construction (SHALL) Cognates of the English auxiliary shall (with the original meaning ‘to owe’, henceforth SHALL) are found in most Germanic languages and are or have been used

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

320

320

Östen Dahl

as future-marking devices over a rather extensive part of the Germanic territory, including earlier stages of High German, where sollen was used more frequently than it is today. According to OED, sceal was used already in Old English to express prophesies and the like. There are clear differences in the domain of use between the Germanic languages, however: it is only in English and Dutch (including Afrikaans) that SHALL is used for prediction-based FTR, although it should be added that in English, this use is rather restricted since shall is mainly used in the 1st person. In the Scandinavian languages, SHALL is restricted to intention-based FTR and obligational meanings closer to its original sense.

5.6. De-venitive constructions Under these headings, we treat constructions involving verbs with the meaning ‘come’. These include two gram families – one in Scandinavia and one mainly comprising a number of Romansh dialects – which exhibit striking similarities and seem to have arisen roughly at the same time but which still cannot be assumed to be related in view of the geographical distance between them. The evidence suggests that these constructions may be the result of a path of development that has not been properly described in the literature on grammaticalization. This motivates discussing them in somewhat more detail. 5.6.1. Scandinavian This gram family comprises Continental Scandinavian – Danish, Norwegian, and Swedish – and (somewhat marginally) Finnish. Its rise and spread appears to be relatively recent – the first attested examples in Swedish go back to the 17th century. The original form of the construction in Scandinavian (preserved in Norwegian and Danish and attested in older Swedish) is

(14)

kommer til(l) at(t) come:INF to INFM

full verb

In Swedish, the preposition till has been dropped, and there is a tendency in the present-day language to also drop the infinitive marker. In Finnish, the verb tulla is combined with the Illative case of the so-called 3rd Infinitive (suffix -maan). The construction has been looked upon with some suspicion as being a Swedish calque.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

321

The grammar of future time reference in European languages

321

5.6.2. Swiss In the Romansh dialects spoken in Switzerland, there exists a construction which is formally and perhaps also semantically very similar to the Scandinavian de-venitive (Ebneter 1973). The earliest attested examples are from the 16th century. The construction now seems receding but is still preserved in a number of dialects. An example from Surmiran (surmeirisch): (15)

Romansh/Surmiran ak a vé kr š i va la ražú. Té viñst you come:PRS:2SG yet to see:INF that I have DEF right ‘You will yet see that I am right’

In some dialects, this construction appears mainly in examples of the type ‘It’s going to rain’, e.g., Oberengadin: (16)

Romansh/Oberengadin veñ pl¯o vr.   it come:PRS to rain:INF ‘There is rain on the way / it will rain.’

Analogues to this are also found in Swiss German (Ebneter 1973: 242, Bickel 1992): (17)

Züritüütsch Es chunt cho rägnen. it come SP rain ‘There is rain on the way / it will rain.’

These de-venitive constructions are somewhat peculiar among future-referring grams in that they are primarily used for prediction-based rather than intention-based FTR. It is therefore of some interest to consider how they have come about, and we shall now look at the source that has been proposed in the literature (see, for example, Ebneter 1973). In many Germanic languages, there is a construction which is formally quite similar but whose semantics is distinct, as shown by the following English example: (18)

I came to hate him.

(18) could perhaps be best paraphrased as ‘By various causes I was led to hate him’. What is notable is thus that it expresses something that it is not under the control of the subject, in other words, a non-volitional process. This suggests that the construction might relatively easily be extended, when used with future time reference, to express predictions in general. Cf. (19) as an example that could serve as an intermediate step towards such an extension.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

322

322 (19)

Östen Dahl

You will come to hate him.

If this account is correct, it creates a problem for the claim made by Bybee, Pagliuca & Perkins (1994: 270) that “all modal and movement future sources begin with human agents and move from the expression of the intentions of that agent to the expression of prediction”.4 At any rate, there is no evidence to suggest that the Germanic de-venitives ever expressed intention. (In that case, we would have to assume that it has now lost this use, which would also create difficulties for the theory.) Rather, we have to conclude that we are dealing with a separate grammaticalization path, which in the attested cases ends in a gram with a focus in prediction-based FTR. Whether further developments from this point are possible is something that only coming research can decide.

5.7. North European de-volitive construction (WILL) Descendants of the Proto-Germanic willan ‘want’, henceforth WILL, are used as FTR markers in a restricted number of Germanic languages, mainly in the North Sea area: English (will), Danish and Norwegian Bokmål (vil), Faroese (vil), Frisian (wal) and Yiddish (vel). The area may earlier have also included High German. In English, the use of WILL for both intention-based and prediction-based FTR was well established already in the Old English (Anglo-Saxon) period, and has (in combination with SHALL) become what is probably the most grammaticalized future marker in the Germanic languages. In Scandinavian, WILL is much more marginal – it seems to have spread later and never reached Swedish. Yiddish is a special case in that WILL seems to have fused with werden (first person forms such as ix vel ‘I will’ are supposedly derived from willan while other forms such as er vet ‘he will’ seem to come from werden, although a general analogy to the regular verb paradigm is not excluded).

5.8. Circum-Baltic ‘become’ In Modern High German, the most common FTR device is the construction werden + Infinitive. Werden is identical to the verb for ‘become’. The details of the origin of this construction are somewhat controversial. In dialects and older stages of the standard language, there is an alternative construction, where the main verb has the form of a Present Participle, and according to a widespread theory, this is the original variety. This hypothesis is rejected in the detailed study by Saltveit (1962), where it is found that both constructions already existed in Old High German, although the one with the Infinitive did not become frequent until the end of the Middle Ages. According to Saltveit, there was a semantic difference between the two constructions,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

323

The grammar of future time reference in European languages

323

in that the participial one had an inchoative interpretation and the infinitival one a modal interpretation. The Infinitive construction has not become rooted in all dialects in the High German area – for instance, it is still felt to be alien in Swiss German (Bickel 1992). In Yiddish, on the other hand, it is one of the sources for the fused future paradigm. The Estonian saama construction, which is a slightly marginal FTR device, is regarded as a calque on the German werden future.

5.9. Slavic perfective present In the West and East Slavic languages, the non-past forms (Present tense) of perfective verbs are primarily used with present time reference. In general, they cannot be used to refer to events that take place at the moment of speech. For a discussion of the origin of this state-of-affairs, see below.

5.10. Balkan de-volitive construction Most languages in the Balkan area have a construction derived from a verb meaning ‘want’. There are two different types, however: one in which the marker is an uninflected particle, and one where it is an inflected auxiliary. The first type appears in ´ and Albanian Modern Greek (tha + subjunctive), Bulgarian (šte), Macedonian (ke) (do + subjunctive). The second is found in Romanian (voi) and Serbian/Croatian (´cu, c´ eš, c´ e). This gram family is one of the classical examples of a Sprachbund phenomenon, covering four different branches of Indo-European.

5.11. Balkan ‘have’ future In some Balkan languages – Bulgarian (ima da), Gheg Albanian (kam me + infinitive) and Romanian (am s˘a) – there is an alternative future construction based on the verb ‘have’, thus most probably an original obligational construction. Its range of uses in the different languages is not quite clear from the sources. The Bulgarian construction is different from the others syntactically in that ima does not agree with the subject. The Ukrainian inflectional imperfective future may also belong to this gram family, although it has gone further in grammaticalization and also differs aspectually from the others.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

324

324

Östen Dahl

5.12. Slavic copular constructions These are combinations of a copula and a participle or the like used as FTR devices. The clearest cases are in Slavic: (i) the imperfective futures formed with the stem bo¬ d-/bud- and an l-participle or infinitive in North (West and East) Slavic, and (ii) the Slovenian future construction bo- + l-participle. The North Slavic constructions are relatively late in origin. The infinitive construction seems to have spread from the Czech area at the end of the 13th century, shows up in Polish, Belarusan, and Ukrainian at the end of the 14th century and in Russian in the 15th. Although this rather nice chronology strongly suggests an areal development, Kˇrížková (1960), from whose monograph these dates are taken, seems rather skeptical about all hypotheses implying areal influences with regard to the rise of the North Slavic imperfective futures. She also takes a skeptical stance against linking this construction with the German werden future.

5.13. ‘Take, seize’ There are at least two cases of FTR markers which are derived from verbs with meanings like ‘seize’ and ‘take’ viz. Hungarian fog ( ‘seize’) and Romani le- ‘take’. Csató (1992) hypothesizes that the original function of Hungarian fog was to express inchoativity. Fog was earlier used also in the meaning ‘begin’. (Csató notes a similar polysemy in the Turkish tut- ‘grasp, hold, begin’.)

5.14. ‘Begin’ A relatively frequent way of marking FTR in Estonian is by a construction involving the verb hakkama ‘begin’ and the ma-infinitive of the main verb (Tommola 1992a). A similar use of the verb alkaa ‘begin’ is found in Finnish but appears to be quite marginal. In the Swedish dialects spoken in the Finnish province of Ostrobothnia böri ‘begin’ is used fairly extensively as an FTR marker. Areal influence seems likely although the connections are unclear.

5.15. Isolated grams Under this heading we briefly mention a couple of FTR devices that occur in only one language.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

325

The grammar of future time reference in European languages

325

5.15.1. Icelandic de-obligational munu This is an auxiliary with original obligational meaning whose distribution seems to be restricted to Icelandic. It seems never to have reached a more advanced degree of grammaticalization and may even have been more frequent in Old Icelandic. 5.15.2. Basque de-obligative futures Basque has two de-obligative future constructions: one consisting of the main verb suffixed by -ko (normally a genitive marker) and an auxiliary ‘have’ and one involving a verb bear ‘need’ (Bybee, Perkins & Pagliuca 1994: 259). 5.15.3. South Italian de-obligative periphrastic future In Sardinian and some Italian vernaculars spoken in the southern part of the Italian peninsula and Sicily, there is a future construction which is analogous in its build-up to the more well-known inflectional Romance future (5.2) but in which the (preposed) auxiliary avere ‘have’ has not fused with the main verb.

6. The futureless area One areal feature of future time reference in European languages can be formulated in negative terms: it tends to be left ungrammaticalized or only partly grammaticalized. Studying this tendency, we may focus on slightly different manifestations of it with somewhat different geographical distributions. One, which is relatively unproblematic to ascertain, is the absence of inflectional futures, which holds for all Germanic and Finno-Ugrian languages and for the majority of the Slavic ones; as well as for some more peripheral parts of Romance and for the non-Slavic languages in the Balkans. Another manifestation of a slightly more elusive kind is the tendency not to distinguish present and future time reference in any systematic way at all, be it inflectionally or periphrastically. In the survey reported in Dahl (1985), Finnish and Estonian came out as somewhat extreme examples of languages with no systematic marking of future time reference (although this does not imply a total absence of devices that show future time reference – cf. Tommola 1992a and 1992b). We saw above an example of the contrast between an English and a Finnish weather forecast showing the difference in predictive contexts, where English normally has obligatory marking and Finnish tends to have none. In fact, English turns out to be relatively isolated in the Germanic area in this respect. The use of present tenses for prediction-based future time reference seems widespread in all other Germanic languages. Taking the obligatory use in (main clause) prediction-based contexts as a main criterion for the grammaticalization, we may therefore claim that there is a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

326

326

Östen Dahl

“futureless” area in Northern Europe which includes at least all Finno-Ugrian and Germanic languages except English. What is interesting from the areal and diachronic point of view is that, going back in history, this area becomes more pronounced in several ways. The auxiliary constructions used in the modern Germanic and (to a lesser extent) Finno-Ugrian languages are all of relatively recent origin. In Germanic, we find the least propensity to mark the future in the oldest documented varieties (such as Gothic) and in some peripheral modern dialects (such as Swiss German and Icelandic). Similarly, there seems to be no evidence of any grammaticalized future marking in Finno-Ugrian. Furthermore, the area may also originally have included Slavic. The different FTR devices found in the Slavic languages (see above) are, like the Germanic and FinnoUgrian ones, relatively recent, with one exception: the use of the Perfective Present for future time reference, which, according to standard descriptions, goes back to Old Church Slavonic. However, there is good reason to doubt that this was a trait of Common Slavic. It is not in general found in the modern South Slavic languages, and there is some evidence that the link between the Perfective Present and future time reference was at least not fully developed in Old Church Slavonic (Kopeˇcný 1981). See also Dahl, this volume, for a discussion of a related phenomenon with a similar areal distribution, viz. the extended use of verbs of becoming to express future states. There is thus basically nothing that contradicts the postulation of a relatively large “futureless” area in Northern Europe at the time when the assumed Germanic protolanguage was spoken, that is, about two thousand years ago. Going beyond that takes us into the realm of speculation, however. If the Indo-European and FinnoUgrian languages influenced each other, it is hard to say when and how that took place. There is also considerable confusion concerning the situation in older stages of Indo-European: it is far from clear that there was a common future tense.

Notes 1. There may be a tendency to avoid combinations of be going to construction with movement verbs in English, which favors the use of the progressive with those verbs. 2. Two Norwegian informants translated (TMAQ: 37) as jeg skal til byen, literally ‘I shall to the town’, that is, using an auxiliary without a main verb. This is possible and indeed quite common in all Scandinavian languages when an adverbial indicating a goal follows. A corresponding usage is also found in Fering (North Frisian) (Ebert 1994). 3. See Squartini & Bertinetto, this volume, fn. 7, for an explanation of this use of the term ‘vernacular’. 4. Since Danish is one of the languages included in the GRAMCATS database, the Scandinavian de-venitive shows up in the accounts of that, too. In Bybee, Pagliuca & Perkins (1991), future grams are assigned a “future age” (FUTAGE) defined as a stage in the se-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

327

The grammar of future time reference in European languages

327

mantic development of futures from modalities. The Danish kommer til at construction is assigned the highest possible FUTAGE–4, which is given to those grams with future (= prediction) as a use “which also have epistemic uses (other than prediction) or speakeroriented modalities as uses”. It is not clear what specific use of the Danish construction motivates this assignment, but it certainly is not in harmony with the chronological age of the construction or with its apparent low degree of grammaticalization in other respects.

References Bickel, Balthasar 1992 “The Marking of Future Time Reference in Züritüütsch”. Future Time Reference in European Languages I. EUROTYP Working Papers II: 6. Bybee, Joan L. & William Pagliuca & Revere Perkins 1991 “Back to the future”, in: Bernd Heine & Elizabeth Traugott (eds.), Approaches to Grammaticalization. Vol. II. Amsterdam: Benjamins, 17–58. Bybee, Joan, Revere Perkins & William Pagliuca 1994 The Evolution of Grammar. Tense, Aspect, and Modality in the Languages of the World. Chicago: The University of Chicago Press. Bybee, Joan, and & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Comrie, Bernard 1976 Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. Csató, Éva 1992 “The Marking of Future Time Reference in Hungarian”, Future Time Reference in European Languages I. EUROTYP Working Papers II: 4. Dahl, Östen 1985 Tense and Aspect Systems. Oxford: Blackwell. this volume “Verbs of becoming as future copulas”. Ebneter, Theodor 1973 Das bündnerromanische Futur. Syntax der mit vegnir under habere gebildeten Futurtypen in Gegenwart und Vergangenheit. Romanica Helvetica 84. Bern: Francke Verlag. Fleischman, Suzanne 1983 “From Pragmatics to Grammar. Diachronic reflections on complex pasts and futures in Romance”, Lingua 60: 183–214. Hedin, Eva this volume “Future marking in conditional and temporal clauses in Modern Greek”. Hermerén, Ingrid, Suzanne Schlyter & Ingrid Thelin 1994 “The Marking of Future Time Reference in French”, Future Time Reference in European Languages III. EUROTYP Working Papers VI: 3. Kopeˇcný, František 1981 “Ein gemeinsamer Charakterzug des Altkirchenslavischen und gotischen Zeitwortes”, Wiener Slawistischer Almanach 8: 295–306. Kˇrížková, Helena 1960 Vývoj opisného futura v jazycích slovanských, zvláštˇe v ruštinˇe. Acta Universitatis Palackianae Olomucensis. Praha: Státní Pedagogické Nakladelství.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

328

328

Östen Dahl

Rohlfs, Gerhard 1962 Grammatica storica della lingua italiana e dei suoi dialetti. II. Morfologia. Torino: Einaudi. Saltveit, Laurits 1962 Studien zum deutschen Futur. Die Fügung ‘werden mit dem Partizip des Präsens’ und ’werden mit dem Infinitiv’ in ihrer heutigen Funktion und in ihrer geschichtlichen Entwicklung. Bergen: Norwegian University Press. Schlyter, Suzanne & Sandberg, Vesta 1994 “The Marking of Future Time Reference in Spanish”, Future Time Reference in European Languages III. E UROTYP Working Papers VI: 4. Setälä, Eemil N. 1973 Suomen kielen lauseoppi. Helsinki: Otava. Squartini, Mario & Pier Marco Bertinetto this volume “The Simple and Compound Past in Romance languages”. Thieroff, Rolf 1992 “The Marking of Future Time Reference in German”, Future Time Reference in European Languages I. E UROTYP Working Papers II: 3. Tommola, Hannu 1992a “The Marking of Future Time Reference in Estonian”, Future Time Reference in European Languages II. E UROTYP Working Papers II: 2. Tommola, Hannu 1992b “The Marking of Future Time Reference in Finnish”, Future Time Reference in European Languages III. E UROTYP Working Papers III: 2.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

329

Eva Hedin

Future marking in conditional and temporal clauses in Greek

1. Introduction It is a well-known fact that in many languages temporal and conditional clauses are often not marked for future. In an investigation made by Dahl (1985) of tense, mood and aspect in 64 languages based on a questionnaire with sentences to be translated, there were on average eight futures (of a total of 47 future grams) in subordinate clauses introduced by whatever, if, even if and when. Among these there were no periphrastic future grams, but all were inflectional. According to Bybee & Dahl (1989) conditional and temporal clauses are contexts where an inflectional gram is more likely to occur rather than a periphrastic gram since future time reference in these contexts is redundant. The theory they present considers an inflectional gram as more developed and thereby more highly grammaticalized than a periphrastic gram. As a gram becomes more grammaticalized it becomes more generalized and more frequent and thereby more easily occurs in contexts where it is redundant. In the E UROTYP questionnaire for future time reference (FTRQ) there were eight sentences with conditional clauses and seven with temporal clauses to be translated: CONDITIONAL CLAUSES

(FTRQ: 9)

If you PUT a stone into this bag, it BREAK.

(FTRQ: 10)

Even if you PUT a stone into this bag, it not BREAK.

(FTRQ: 14)

[The boy thinks that he will perhaps get a sum of money:] If the boy GET the money, he BUY a present for the girl.

(FTRQ: 23)

[Traveller to local:] If you SHOW me the way, I GIVE you money.

(FTRQ: 24)

[Mother to child:] If you not STOP PLAY with that ball, I TAKE it away.

(FTRQ: 66)

If it RAIN tomorrow, we STAY at home.

(FTRQ: 67)

If it BE COLD tomorrow, we STAY at home.

(FTRQ: 68)

If I GET the money tomorrow, I BUY a present for you.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

330

330

Eva Hedin

TEMPORAL CLAUSES

(FTRQ: 13)

[The boy is expecting a sum of money:] When the boy GET the money, he BUY a present for the girl.

(FTRQ: 17)

[Talking to someone who is leaving in a while:] When you RETURN, I WRITE this letter (= I FINISH it already at that time).

(FTRQ: 18)

[Said as an order by a teacher leaving the classroom] When I RETURN, you WRITE this assignment (= You FINISH it by then).

(FTRQ: 25)

[Said by a young man] When I GROW old, I BUY a big house.

(FTRQ: 26)

[My brother is late for dinner.] When he ARRIVE, the food BE COLD.

(FTRQ: 65)

When I GET home tonight, my mother BE HAPPY.

(FTRQ: 100)

[I heard a funny story the other day.] When you HEAR it, you LAUGH.

In Table 1, data about FTR marking in temporal and conditional clauses for 28 languages and language varieties are presented. About half the languages did not mark future time reference at all in these clauses. Among the rest, at least two (French and Macedonian) use future markings systematically in temporal but not in conditional clauses. The opposite possibility does not occur in the material. This, then, is compatible with the hypothesis that FTR marking in conditional clauses implies marking also in temporal clauses (but not the other way around). The material provides two main classes of counterexamples to the generalization that only inflectional future grams show up in temporal and conditional clauses. The first class concerns the periphrastic Imperfective Futures of East and West Slavic. The second class of counterexamples are found in a number of languages spoken in the Balkan area, notably Greek, Macedonian, Slovenian, and Romanian. Of these, only Slovenian displays systematic marking in both clause types. The variation among the Balkan languages, especially the rather striking difference between the closely related Bulgarian and Macedonian, may suggest ongoing grammaticalization in the area. In the paper some additional data from one of the Balkan languages, Greek, are provided and possible factors behind the variation are discussed.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

331

Future marking in conditional and temporal clauses in Greek Table 1. Future marked temporal and conditional clauses in the FTRQ Language

Temporal clauses with overt future marking

I NDO -E UROPEAN Slavic East and West Slavic Ý Czech Polish 25 Ý Russian Ý Ukrainian Ý South Slavic Bulgarian Macedonian (2 informants) all (except 100 by 1 inf) Slovenian (2 informants) all (at least as an alternative) Serbo-Croatian * Baltic Lithuanian all Celtic Irish all Germanic Dutch English Frisian German Norwegian (Bokmål and Nynorsk) Swedish (Standard and Ostrobothnian) Zurich German Romance French (6 informants) all marked (by most informants) Italian all (at least as an alternative) Portuguese all (1 as an alternative) Romanian 13, 25, 100 Spanish # Greek Greek 17, 25, 26 U RALIC Estonian Finnish Hungarian

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Conditional clauses with overt future marking

66, 67 Ý 66, 67 Ý 66, 67 Ý 66, 67 Ý

9, 10, 14, 23, 24, 66, 68 * all $

all (at least as an alternative) 9, 14, 23, 66, 67, 68 9, 10

14

67

331

332

332

Eva Hedin

Table 1. (continued) Language

Temporal clauses with overt future marking

A LTAIC Turkish (5 informants) C AUCASIAN Lezgian

Conditional clauses with overt future marking 14 (1 informant)

&

&

Numbers in italics represent periphrastic constructions. Ý (Czech, Polish, Russian, Ukrainian): Perfective Present forms used in remaining examples. * (Serbo-Croatian): The so-called Future Perfect is used in temporal and conditional clauses. $ (Irish): “Habitual present” used in conditional clauses. # (Spanish): Present Subjunctive used in temporal clauses. & (Lezgian): Non-finite forms used in temporal and conditional clauses.

2. Marking of FTR in conditional clauses in Greek To judge from the FTRQ, future marking in conditional clauses seems to be a marginal phenomenon in Greek. An investigation of clauses introduced by (e)an ‘if’ and having future time reference (henceforth: FTR) in four newspapers, one weekly magazine and dramas written by 17 different authors confirmed this. Of 335 conditional FTR-clauses found only 2 were marked for future. But future-marked conditional clauses do occur, in some contexts even obligatorily. Compare the following examples: (1)

An tha sikothís norís ávrio to proí if FUT get_up:PFV:2SG early tomorrow DEF morning pas ja ípno tóra. prépi na it_is_necessary SUBJ1 go:PFV:SUBJ:2SG for sleep now ‘If you are getting up early tomorrow morning you have to go to bed now.’

(2)

?An sikothís norís ávrio to proí if get_up:PFV:SUBJ:2SG early tomorrow DEF morning prépi na pas ja ípno tóra. it_is_necessary SUBJ go:PFV:SUBJ:2SG for sleep now

(3)

?If you get up early tomorrow morning you have to go to bed now.

In (1) the future particle can hardly be omitted. (2) does not make sense, nor does the literal English translation (3). The nonfuture form is on the other hand suitable in sentences like (4):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

333

Future marking in conditional and temporal clauses in Greek

(4)

333

An sikothís norís ávrio to proí dhe if get_up:PFV:SUBJ:2SG early tomorrow DEF morning NEG tha blextís stin kínisi. FUT entangle:PASS:PFV:2SG in:DEF traffic ‘If you get up early tomorrow morning you won’t get caught in traffic.’

Let us for the sake of the discussion make a distinction between conditional clauses “proper” as exemplified in (4) and “improper” as exemplified in (1). The relation that holds between the two events referred to in the protasis and in the apodosis, respectively, are different in the two kinds of conditionals from a logical as well as from a temporal point of view. In the first case there is a relation of cause and consequence corresponding to a temporal relation of order. The order relation between the two situations may be of a different kind depending on contextual and pragmatic factors. Compare with (4), where the two situations are in sequence (if A then after that B) the following example where the two situations could be considered simultaneous: (5)

An protimísis plájus dhrómus dhe tha if prefer:PFV:SUBJ:2SG back streets NEG FUT blextís stin kínisi. entangle:PASS:PFV:2SG in:DEF traffic ‘If you choose back streets you won’t get caught in traffic.’2

However, there is some kind of “left-to-right” relation between them, also temporally. It is by choosing back streets that one creates a certain traffic situation, not the other way around. To the extent that there exists an element of cause, result or consequence there must (for obvious natural reasons) also exist a temporal relation of sequence, although the two events may overlap to greater or lesser degree, and the event referred to in the protasis cannot be totally posterior to that of the apodosis as in (1). Not all conditional clauses express a cause-consequence relation, however. The relation between the protasis and the apodosis in (1) is of a different kind. Here the intention to get up early in the morning is the reason why one should go to bed early. It is not the getting up itself that is the cause of anything. In (1) this is the only possible interpretation but in other examples the difference is not so clearcut. Consider example (6): (6)

An tha páte sto xorió tha perásete oréa. if FUT go:PFV:2PL to:DEF village FUT pass:PFV:2PL nicely ‘If you go (are going) to the village you will have a nice time.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

334

334

Eva Hedin

Here the apodosis (having a nice time) is not anterior to the protasis (going to the village) but rather subsequent to or simultaneous, like the apodosis in example (4) or (5). Going to the village is clearly also logically a prerequisite for having a nice time. So why is it marked for future? In what way is it different from the corresponding example where the future is not marked: (7)

An páte sto xorió tha perásete oréa. if go:PFV:SUBJ:2PL to:DEF village FUT pass:PFV:2PL nicely ‘If you go to the village you will have a nice time.’

The difference between the two kinds of conditionals could perhaps be described as one of “scope”, the future-marked conditionals having a “wider scope” than the unmarked: ‘If it is the case that x’ rather than ‘If x’. That is, the condition referred to in the unmarked conditional is the (future) occurrence of an event x (‘If x (in the case of x), y will be the case’) whereas in the future-marked conditional the condition is the truth of some proposition p (‘If it is the case that p (it is also true that) q will be the case’). (We are now dealing with conditionals proper). Comrie (1982) gives the following example (from Quirk et al. 1972: 781) of future marking in English of what he calls contextually given propositions: (8)

If he won’t arrive before nine, there’s no point in ordering for him.

We could add: (9)

If he won’t arrive before nine, (then) he’ll be late for supper.

With intention, obligation and contextually given propositions the “wider scope” is rather obvious: ‘If (it is the case that) you intend to go to the village’ ‘If (it is the case that) you shall go to the village’ ‘If (it is the case that) he will be late (as you say)’ There are, however, other examples where neither intention or obligation nor any implicit reference to some contextually given proposition (like in the English examples) seems to be the reason why the Future is used: (10)

[Sta plésia lipón pu epitrépi ke prodhiaghráfi to Síndaghma tu 1975] tha liturjísun ke i néi dhikastikí kódhikes an FUT function:PFV:3PL also DEF new judicial laws if teliká tha psifistún. finally FUT vote:PASS:PFV:3PL ‘[Within the frame of the constitution of 1975] the new judicial laws will also be valid, provided they are voted for.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

335

Future marking in conditional and temporal clauses in Greek

335

That the new laws will actually be voted for is something that is presupposed in the phrase the new laws but then, as a parenthetically added afterthought, the speaker adds the condition that this is a true presupposition. The description ‘if it is the case that p’ could thus still hold relevant. In this case the proposition is a prediction and the condition that this prediction is a true one: ‘If it is true that p will be the case’. The relation between the protasis and the apodosis would thus be more indirect and independent when the former is marked for future than when it is not. This difference in dependency also seems to characterize the correspondent temporal clauses (cf. below 3.2).

3. Marking of FTR in temporal clauses in Greek According to the FTRQ, future marking in temporal clauses seems to be more than a marginal phenomenon in Greek. Below follow the translations of the relevant FTRQ sentences: (FTRQ: 13)

Ótan pári to aghóri ta xrímata, tha when get:PFV:SUBJ DEF boy DEF money FUT aghorási éna dhóro ja to korítsi. buy:PFV INDEF present for DEF girl

(FTRQ: 18)

Ótan when [aftí [this

(FTRQ: 65)

Ótan jiríso spíti to vrádhi i mitéra when return:PFV:SUBJ:1SG home DEF evening DEF mother mu tha íne eftixís. me:GEN FUT be:IPV happy

(FTRQ: 100)

(FTRQ: 17)

jiríso tha éxtete ghrápsi return:PFV:SUBJ:1SG FUT have:IPFV:2PL write:PFF3 tin erghasía]. DEF assignment]

Ótan tin akúsis tha jelásis. when it hear:PFV:SUBJ:2SG FUT laugh:PFV:2SG Ótan tha jirísis tha éxo ghrápsi when FUT return:PFV:2SG FUT have:IPFV:1SG write:PFF aftó to ghráma. this DEF letter

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

336

336

Eva Hedin

Table 2. Category occurrences in the Greek corpus

Non-future categories imperfective perfective PRS / PF PFV SUBJ Newspapers and magazine Dramas

4 6

6 44

Future categories imperfective perfective IPFV FUT / PFV FUT FUT PF 10

6 10

(FTRQ: 25)

Ótan tha jeráso tha aghoráso éna when FUT grow_old:PFV:1SG FUT buy:PFV:1SG INDEF meghálo spíti. big house

(FTRQ: 26)

Ótan tha ftási to fajitó tha íne krío. when FUT arrive:PFV DEF food FUT be:IPFV cold

An investigation of clauses introduced by ótan ‘when’ and used with future time reference (henceforth: FTR) in four newspapers, one weekly magazine and dramas written by 17 different authors offered a similar picture. There were in all 86 temporal FTR-clauses of which 26 were marked for future. The temporal-aspectual categories that occurred in this context in the material were the Present, the Perfect and the Perfective (Aorist) Subjunctive on one hand and the Imperfective Future, the Future Perfect and the Perfective Future on the other. The Perfect and the Future Perfect are here counted as imperfective categories. The total picture is shown in Table 2.

3.1. Imperfective categories Of the nonfuture-marked cases the vast majority is in the perfective aspect, i.e., in the Perfective Subjunctive. According to Papazafeiri (1992: 91) one cannot use the Present or the Perfect in a temporal clause referring to the future, i.e., the particle tha may not be deleted when the aspect is imperfective. She gives the following authentic examples of what she claims to be incorrect language usage: (11)

Tha ta púme arghótera, ótan íse kalá. FUT it speak:PFV:1PL later when be:PRS:2SG well ‘We’ll talk later when you feel better.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

337

Future marking in conditional and temporal clauses in Greek

(12)

337

Aftó tha to máthume ótan xiristés ke mixanikí that FUT it learn:PFV:1PL when pilots and mechanics éxun epistrépsi stis thésis tus. have:PRS:3PL return:PFF to:DEF places them:GEN ‘We’ll learn that when the pilots and the mechanics have returned to their places.’

The correct way would be to say: ótan tha íse kalá and ótan tha éxun epistrépsi. With the perfective aspect, however, the future particle may be used or be left out, according to the same author.

(13)

Tha ta púme ótan jínis / tha FUT it speak:1PL when become:PFV:SUBJ:2SG / FUT jínis kalá. become:PFV:2SG well ‘We’ll talk when you get well.’

(14)

Tha to máthume ótan epistrépsun / tha FUT it learn:PFV:1PL when return:PFV:SUBJ:3PL / FUT epistrépsun. return:PFV:3PL ‘We’ll learn that when they come back.’

If in the nonfuture-marked clauses, the imperfective forms Present or Perfect cannot be used, this would mean that the nonfuture-marked subordinate in this kind of context would always be morphologically marked for mood, since the Perfective used is the Perfective (Aorist) Subjunctive. The future-marked clauses (perfective or imperfective), on the contrary, must be considered indicative forms. Since there is no self evident definition of the subjunctive in Greek, it could, however, be questioned whether the Present and the Perfect should really be considered indicative if they occur in this context. According to a syntactic definition of the Subjunctive they could also be described as the imperfective and the perfect correspondences to the Perfective Subjunctive, although they cannot morphologically be identified as such. The investigated material does contain some examples of Present and Perfect used in temporal clauses with FTR (10 in all):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

338

338 (15)

Eva Hedin

[Aftá pu éjinan tin Kiriakí sto Vólo tha pápsun na anaparághonte, ótan o k. Saliarélis stamatísi tus palikarismús, ótan xiristí me métro tin elinikí ghlósa,] ótan dhen afíni to perístrofó tou na when NEG let:PRS DEF gun him:GEN SUBJ misofénete stin píso tsépi tu pandalonjú half-appear:IPFV in:DEF back pocket DEF trousers:GEN tu. . . him:GEN ‘[What happened on Sunday in Volos will stop reproducing itself, when Mr. Saliarelis stops his bullying, when he uses the Greek language with moderation,] when he doesn’t let his gun half-appear in his pocket . . . ’

(16)

Apó dho ke sto eksís, ótan akúte from here and in:DEF following when hear:PRS:2PL kápja na ti fonázun Klitemnístra, aftí tha somebody:F SUBJ her call:IPFV:3PL Clytemnestra she FUT ím’ eghó. be:IPFV:1SG I ‘From now on, when you hear that somebody is called Clytemnestra, that’ll be me.’

(17)

[– Ke tha to kratás sinéxja?] ke ótan dhen káni krío. – Ótan, dhen vréxi when NEG rain:PRS and when NEG do:PRS cold ‘[– And will you hold it (i.e. the hat in your hand) all the time?] – When it doesn’t rain . . . and when it isn’t cold.’

In some examples the future reference is debatable. In the following example the Present could be considered a “FTR-Present”, i.e., as not actually referring to the future (cf. Hedin 1992): (18)

[Stis prótes dhío psifoforíes apetíte pliopsifía dhío tríton.] I analités provlépun óti o Ozal tha DEF analysts foresee:PRS:3PL COMP DEF Özal FUT eklejí málon stin tríti psifoforía stis 31 elect:PFV:PASS probably in:DEF third election on:DEF Oktovríu ótan íne arketí mja aplí pliopsifía. October:GEN when be:PRS enough INDEF simple majority ‘In the first two votes a two-thirds majority is demanded. The analysts foresee that Özal will be elected in the third round on October 31, when a simple majority is enough.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

339

Future marking in conditional and temporal clauses in Greek

339

The Present is not used to refer to a future situation but to the content of some electoral law saying that in the third round (generally, whenever there are elections) only a simple majority is needed and thus consequently also in the third round scheduled for October 31. The Present is also used in some examples taken from stage directions in the dramas. They are not part of the same discourse as the examples occurring in the dialogue, but function as instructions to the actors. They do not explicitly refer to the future but are easily interpreted as doing so: (19)

AFENDIKO (ótan dhen fonázi, tha sizitáj me boss when NEG yell:PRS FUT discuss:IPFV with simvólus). (Ótan stamatáj tu enós i anása, tha advisers when stop:PRS DEF one:GEN DEF breath FUT sinexísi o álos.) continue:PFV DEF other ‘The boss: (when he doesn’t yell, he’ll be discussing with advisers) (When one gets short of breath, the other will continue.)’

The borderline is not clear, however. In other stage instructions the temporal clause is marked for future: (20)

Ótan tha miláj me ton ándra i jinéka tha when FUT speak:IPFV with DEF man DEF woman FUT íne sa na apusiázi. be:IPFV as if be_absent:PRS ‘When she talks with the man the (other) woman should be as if she is absent.’

If we do not consider as mistakes the use of the Present and Perfect found in the material perhaps the variation mentioned above could give us a clue as to their function. Consider, for example, the temporal clause in example (17) ótan dhen vréxi, ‘when it doesn’t rain’. The Present could be used here for similar reasons, that is, because it does not really refer to some future time when it will not be raining but rather means ’whenever it is not raining, on sunny days’ without explicit reference to future time. Obviously, the intuition differs between speakers of Greek as to whether the Imperfective may be used in the unmarked clauses or not. It is true that it is not very frequent. Even considering that the imperfective aspect is generally less frequent than the perfective, this does not seem to be a typical context for it. In future-marked clauses, however, the imperfective aspect seems to be relatively frequent. We will return to this issue below.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

340

340

Eva Hedin

3.2. Perfective categories The main perfective category occurring in temporal FTR-clauses is the Perfective Subjunctive. Future was, however, marked in 16 of 66 cases. Consider the following examples: (21)

[Aftá pu éjinan tin Kiriakí sto Vólo tha pápsun na anaparághonde, ] ótan o k. Saliarélis stamatísi tus palikarismús, when DEF Mr. Saliareris stop:PFV:SUBJ DEF bullying ótan xiristí me métro tin elinikí ghlósa, when use:PFV:SUBJ with moderation DEF Greek language [ótan dhen afíni to perístrofó tu na misofénete stin píso tsépi tu pandalonjú tu. . . ]. ‘[What happened on Sunday in Volos will stop reproducing itself,] when Mr. Saliarelis stops his bullying, when he uses the Greek language with moderation, [when he doesn’t let his gun half-appear in his pocket ].’

(22)

Pándos, ta teliká simberásmata, tha ta vgháli to anyway DEF final conclusions FUT them draw:PFV DEF vrádhi tu Savátu, ótan tha paj sti evening DEF Saturday:GEN when FUT go:PFV to:DEF Masalía, ja na kataskopéfsi tin Marséj, ston Marseilles for SUBJ observe:PFV DEF Marseilles in:DEF aghóna tu ghalikú protathlímatos, me tis Kánes. match DEF French cup:GEN with DEF Cannes ‘Anyway, he will draw the definite conclusions on Saturday evening, when he is going to Marseilles to observe Marseilles in its match in the French Cup against Cannes.’

(23)

idhikús.] [Íthelan éna festivál ja Ótan tha to petíxun dhen tha ipárxi when FUT it succeed:PFV:3PL NEG FUT exist:IPFV pléon “festivál”. any_longer festival experts.] ‘[They wanted a festival for When they (’ll) succeed in doing this there will be no “festival” anymore.’

In the first example there is no future marking whereas in the two others there is. So why is that? Example (22) may serve as a point of departure for the discussion. This example is very typical in that it has an explicit temporal adverbial preceding the subordinate clause: to vrádhi tu Savátu ‘Saturday evening’, referring to the future. In this case ótan has the function of a relative adverb meaning when = at which time.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

341

Future marking in conditional and temporal clauses in Greek

341

In this kind of context the future particle seems to be obligatory or at least difficult to leave out without changing the content. Also in Classical Greek this is one of the few cases where the construction with hóte ‘when’, and the indicative Future is used (Magnien 1912: 190–191; Blass 1970: 232): (24)

Classical Greek (Homer, Od. 18.272) stugeròs gámos Núx d’ éstai hóte d e night PRT be:FUT when PRT hateful marriage oulomén¯es eméthen . . . antibol
esei meet_by_chance:FUT accursed:GEN me:GEN ‘That night will come, when a hateful marriage shall fall to the lot of me accursed . . . ’

(25)

Classical Greek (Dem. 19.262) poie˜ın T¯enikaút’. . . hót’ oud’ ho_ti khr e at_that_time. . . when NEG whatever is_necessary do:INF:IPFV héxete. have:FUT:2PL when you will no more have the power of doing what you ‘That day ought.’

(26)

Classical Greek (Joh. 4.23) . . . érkhetai ho
ra. . . hóte hoi al¯ethinoì proskun¯etaì worshippers come:PRS time. . . when DEF true to i patrí en pneúmati kaì al
etheí¯ai. proskun
esousin worship:FUT:3PL DEF father:DAT in spirit:DAT and truth:DAT ‘. . . a time will come when the true worshippers will worship the Father in spirit and truth.’

The function of the subordinate in (22), ótan tha paj sti Masalía, is that of a nonrestrictive relative clause, modifying only the temporal adverbial in the main clause, to vrádhi tu Savátu. It is not an adverbial clause modifying the main clause as is the case with the subordinate in (21) and (23) above: ótan stamatísi tus palikarismús and ótan tha to petíxun. Even when there is no relative relation between the ‘when’-clause and some preceding correlate, as in the examples discussed above, some definite temporal element may occur in the context of a future-marked clause whose presence renders the temporal clause a less restrictive function in relation to the main clause.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

342

342

Eva Hedin

(27)

Thélo na tin ksanangaliáso ke want:PRS:1SG SUBJ her:ACC embrace_again:PFV:1SG and thélo na tin ksanavró ston álo want:PRS:1SG SUBJ her:ACC find_again:PFV:1SG in:DEF other kósmo ótan tha ftási i óra mas. world when FUT arrive:PFV DEF time us:GEN ‘I want to hold her in my arms again and meet her again in the other world when our time will come/comes.’

(28)

Alá ke me tin anáptiksi ke ekviomixánisi tu but also with DEF development and industralisation DEF Trítu Kósmu, ótan tha arxísi tha third:GEN world:GEN when FUT start:PFV FUT jínun meghalítera ke ekriktikótera ta become:PFV:3PL big:CMPR and explosive:CMPR DEF provlímata aftá. problems these ‘But also with the development of the industrialisation of the Third World, when that starts/will start, these problems will be bigger and more explosive.’

(29)

Ótan tha liturjísi i megháli evropaikí aghorá, when FUT function:PFV DEF big European market stin Evrópi tu 1993, xorís_álo tha ipárksun no_doubt FUT exist:PFV:3PL in:DEF Europe DEF:GEN nées meghalíteres dhinatótites stin Evrópi. . . new big:CMPR possibilities in:DEF Europe ‘When the big European market functions, in the Europe of 1993, there will no doubt be new better possibilites in Europe . . . ’

In (23), however, there is no explicit reference to future time “outside” the temporal clause itself (ótan tha to petíxun), no explicit “correlate” to create a relative relation. So why is its temporal clause marked for future whereas the one in (21) is not (ótan stamatísi tis palikariés)? Considering the obligatory future marking in relative clauses with an explicit temporal correlate, it is possible that the future-marked clauses could be described as some kind of “relative” clause implying such a correlate: ‘when x’ = ‘at time t, when x’ as opposed to the nonmarked clauses that would rather express the relation ‘when x’ = ‘at the time, when x’. In the first case when would function as something between a relative adverb and a subordinating conjunction and the temporal clause would have definite time reference, establishing a future reference time.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

343

Future marking in conditional and temporal clauses in Greek

343

The difference between the two temporal clauses could be described as a difference in the way they establish the temporal relation between the main and the subordinate clause. Whereas in (21) the subordinate presents the situation as such as temporally determining the situation described in the main clause, the subordinate in (23) presents the situation as (indirectly defining) the time when the content of the main clause will be true, i.e., presents the situation as occurring at some time (a future reference time) which temporally determines the situation referred to in the main clause.4 When the subordinate is marked for future it is temporally located at the same time as the situation referred to in the main clause (or, considering focus, the other way round). The relation between the two situations is indirect, they are oriented to the same reference time (functioning as an outer “clock”), thus being “independently simultaneous” with each other: ’A happens at the same time t as B’. fut t

sit A

 sit B

In the nonmarked clauses, however, the two situations are directly orientated to each other without “interference” by any reference time: ‘A happens when B happens’. sit A



sit B

The relation between the two situations in the first case is a temporal one, orienting them to the same future time, whereas the relation represented in the second figure is aspectual, expressing order (cf. section 2 above).

4. Future-marking and imperfectivity We have seen above that imperfective verb phrases are typically (or even obligatorily) future-marked in temporal FTR-clauses. This should probably be seen against the background of the function of the imperfective aspect. When an imperfective verb phrase is used to refer to a future situation, it normally needs some reference time, more or less explicit, a time point or a time frame to which it may be related. This could be why the Imperfective almost automatically gets the future-marking in the subordinate. On the other hand, it also seems to be the case that the imperfective (including FUT PF) is comparatively frequent in the main clause governing a future-marked temporal clause. See, e.g., FTRQ: 17 and 26:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

344

344

Eva Hedin

(FTRQ: 17)

Ótan tha jirísis tha éxo ghrápsi. when FUT return:PFV:2SG FUT have:IPFV:1SG write:PFF aftó to ghráma this DEF letter ‘When you return, I will have written this letter.’

(FTRQ: 26)

Ótan tha ftási to fajitó tha íne krío. when FUT arrive:PFV DEF food FUT be:IPFV cold ‘When he arrives, the food will be cold.’

In the material only 6 of 395 subordinates with a PFV SUBJ were linked to a main clause with a verb phrase in the imperfective aspect. Of 10 PFV FUT clauses 4 were subordinated to a main clause with an imperfective verb phrase. For the IPFV FUT / FUT PF clauses the figure was 7 of 10 and for PRS/PF 5 of 10. Even if the material is too limited to allow any conclusions on the basis of these figures, at least it seems that imperfective aspect in the main clause is less frequent in combination with the PFV SUBJ than with the FUT in the subordinate. Reference to a future state (in a wide sense) in the main clause thus seems to favour the future-marking of the subordinate. This, too, can be interpreted as support for the hypothesis that future-marking in the subordinate has the function of establishing some future reference time. One reason for future-marking the temporal clause when referring to a state in the main clause could be namely the need of a reference time for the state of affairs described. Consider the following example: (30)

When I come home my mother will be happy.

The most natural interpretation is that the mother will become happy as a result of the speaker’s coming home. Her happiness is directly related to the arrival of her child. Another possible interpretation, however, is that it so happens that the mother is (already) happy (for some other reason) at the time of the speaker’s arrival. In the second case, the subordinate gets the function of reference time for the state described in the main clause. Compare FTRQ: 17 and 18 where the Greek informant chose the PFV FUT in the first but not in the second. In both cases the main clause has a Future Perfect. (FTRQ: 17)

[Talking to someone who is leaving in a while:] When you RETURN, I WRITE this letter(= I FINISH it already at that time).

(FTRQ: 18)

[Said as an order by a teacher leaving the classroom:] When I RETURN, you WRITE this assignment (=You FINISH it by then).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

345

Future marking in conditional and temporal clauses in Greek

345

A possible reason for this choice could be that in the first sentence the reference time is more obvious than in the second. (Note the instruction = I FINISH it already at that time.) In the second the teacher tells the class to finish (have finished) the work before he comes back, but does not as obviously refer to the future state of the work already being done. It is interesting to note that a similar differentiation was made by one of the French informants who chose the Future in FTRQ: 17 (with the Future Perfect in the main clause), but the Present in FTRQ: 18: (FTRQ: 17)

Quand tu reviendras j’aurai fini d’écrire cette lettre.

(FTRQ: 18)

Quand je reviens, vous devez avoir fini ce travail.

5. Causal/conditional interpretations of ‘when’-clauses The semantics of the temporal conjunction meaning ‘when’ in many languages seems to allow for a causal interpretation in some contexts, particularly with perfective verbs. The degree to which one is inclined to give the temporal clause a causal meaning may well be due to different pragmatic factors. For a sentence like (31) it is not easy to find a direct causal interpretation whereas (32) is equally hard to give a purely temporal interpretation. (31)

When the sun sets we will stop working.

(32)

When the sun sets it’ll get dark.

As has already been mentioned, in some contexts as in (30) both interpretations are possible. Greek, however, seems to have the possibility of excluding a causal interpretation when the interpretation is not obvious from contextual and pragmatic factors, by future-marking the temporal clause. Consider the following example: (33)

Ótan érthi o Jánis tha ímaste dhéka when come:PFV:SUBJ DEF Jannis FUT be:IPFV:1PL ten edhó mésa. here inside ‘When Jannis comes we will be ten in here.’

One interpretation of (33) is thus: ‘When Jannis comes, we will be (become) ten persons in the room (as a consequence of his arrival)’. However, another possible interpretation is the following: ‘At Jannis’ arrival (at the time point when he arrives) we will (already) be ten persons in the room’, ‘At his arrival Jannis will find ten persons in the room’. If the ‘when’-clause is future-marked, however, it seems that the sentence more easily gets the second interpretation.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

346

346 (34)

Eva Hedin

Ótan tha érthi o Jánis tha ímaste dhéka when FUT come:PFV DEF Jannis FUT be:IPFV:1PL ten edhó mésa. here inside ‘When (by the time) Jannis comes we will be ten in here.’

Let us consider the two “types of simultaneity” characteristic of the two kinds of ‘when’-clauses discussed above. The future-marked ‘when’-clauses were said to refer to some future time functioning as reference time (A and B at t) whereas the nonmarked clauses express order ( B at A). While the first (temporal) relation could be called contingent the second (aspectual) shows some kind of necessity or implication; B is temporally determined by A itself: When A, then B (whenever that is). The interpretation of this “temporal implication” (when–then) as a “logical implication” (if–then) is natural (e.g. 35), but not necessary (e.g. 36): (35)

Ótan tin akúsis tha jelásis. when it:ACC hear:PFV:SUBJ:2SG FUT laugh:PFV:2SG ‘When you hear it you’ll laugh.’

(36)

Ótan sikóso to xéri tha arxísis, étsi? when lift:PFV:SUBJ:1SG DEF arm FUT start:PFV:2SG so ‘When I lift my arm you’ll start, okay?’

6. Modal interpretations Setatos has pointed out (1985: 180–181) that for Greek, there seems to be a modal difference between the future-marked conditional and temporal clauses and those that are not marked for future. The future-marked clauses seem to convey a feeling of stronger certainty: He describes both the temporal and the conditional clauses with the subjunctive as expressing something “possible, if he ever comes”, and the correspondent clauses with the Future as something “natural, as we expect”. For the conditional clauses the ‘if it is the case’ element of the future-marked clauses referred to above (in Section 2) may perhaps contribute to such an interpretation, i.e., the implicit reference to the truth of some prediction would make the conditional clause more assertive than it normally is. As for the temporal clauses, the difference in modal interpretation could perhaps be conveyed by the differences in temporal reference in the two cases. Reference to a future situation as occurring at some specific time (‘at time t when’) may convey a feeling of stronger certainty as for its occurring, i.e., the temporal definiteness may be understood as a modal definiteness. Consider the following two examples:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

347

Future marking in conditional and temporal clauses in Greek

(37) a.

b.

Ótan érthi o patéras when come:PFV:SUBJ DEF father po. say:PFV:1SG Ótan tha érthi o patéras when FUT come:PFV DEF father po. say:PFV:1SG ‘When Daddy comes I’ll tell him.’

347

tha tu to FUT him:GEN it:ACC

tha tu to FUT him:GEN it:ACC

In both cases we do in fact – by using the conjunction ótan ‘when’ in this context (instead of ‘if’ e.g.) – implicitly assert that we expect (presuppose) that Father will come at some more or less definite occasion. Whether we use the Perfective Subjunctive or the Perfective Future, we express some expectation that Father will come. However, only when we use the Future do we anticipate a specific future time for the arrival. With the Subjunctive the arrival is not defined temporally. But the indefiniteness is actually concerned only with the temporal reference. In conditional as well as in temporal clauses, the Future as indicative and temporal naturally gives a different modal status to the utterance compared to the nontemporal Subjunctive.

7. Conclusion As was noted in the introduction, future-marking in conditional clauses and in temporal clauses in Greek cannot be described in the same way. Temporal clauses seem to be more easily marked for future than conditional clauses. Specific temporal reference in the temporal subordinate seems to favour future-marking. In both cases, however, when the subordinate clause is marked for future there seems to be some weaker dependency between this clause and the main clause. This is most typically observed in the obligatorily future-marked cases where the ‘when’-clause is a nonrestrictive relative clause. Such cases, where the subordinate clause is less dependent and has a more assertive function, could be a possible starting point for a future gram to enter this kind of context from which it may expand further in a grammaticalization process to other temporal and conditional clauses.

Acknowledgment Financial support for the work presented has been received from the Swedish Council for Research in the Humanities and Social Sciences (HSFR).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

348

348

Eva Hedin

Notes 1. The Modern Greek Subjunctive is a category that has raised some controversy. Morphologically, it exists only in the perfective aspect (for instance, tha ghrápso), whereas the Imperfective Subjunctive has to be syntactically defined. Compare thélo na ghrápso ‘I want to write (PFV)’ and thélo na ghráfo ‘I want to write (IPFV)’, where the form ghráfo is morphologically identical with the Present Indicative ghráfo ‘I write/am writing’ and its function as a Subjunctive is marked only by the presence of the particle na (in the glosses: SUBJ). For further discussion see Veloudis, I & Philippaki-Warburton 1983. 2. This example was pointed out to me by Ioannis Veloudis, Thessaloniki, as a counterexample to my somewhat simplified view at the time of the temporal order relation discussed here as one of sequence (which is typically but not necessarily true). 3. PFF = Aparémfato or Perfect formant, which combines with the verb éxo to form the Greek Perfect periphrasis. 4. Note also that the main clause has imperfective aspect. Cf. 4 below. 5. Not all subordinates were followed by a main clause and some verbs have the same form in the perfective and the imperfective. This is why the figures do not correspond to those in Table 2 above.

References Blass, Friedrich & Albert Debrunner 1970 Grammatik des neutestamentlichen Griechisch. (13th edition.) Göttingen: Vandenhoeck & Ruprecht. Bybee, Joan L. & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Comrie, Bernard 1982 “Future time reference in the conditional protasis”, Australian Journal of Linguistics 2: 143–152. Dahl, Östen 1985 Tense and aspect systems. Oxford and New York: Blackwell. Hedin, Eva 1992 “Present with future time reference in Modern Greek”, in: E UROTYP Working Papers, Series VI, no. 1. August 1992. Magnien, Victor 1912 Le futur grèc. Paris: Champion. Papazafeíri, Ioanna 1992 Láthi sti xrísi tis ghlósas mas [Mistakes in the use of our language]. Athína: Smili. Setatos, Michail 1985 “Tropikótites tu rímatos stin kiní neoelinikí” [Modalities of the verb in Modern Greek], in: Studies in Greek Linguistics, Proceedings of the 6th annual meeting of the Department of Linguistics, Faculty of Philosophy, Aristotelian University of Thessaloniki. Thessaloniki: Kiriakidis, 175–182.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

349

Future marking in conditional and temporal clauses in Greek

349

Veloudis, I. & Philippaki-Warburton, I. 1983 “I ipotaktikí sta néa eliniká” [The Subjunctive Mood in Modern Greek], in: Studies in Greek Linguistics. Proceedings of the 4th Annual Meeting of the Department of Linguistics, Faculty of Philosophy, Aristotelian University of Thessaloniki. Thessaloniki: Kiriakidis, 151–169.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

350

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

351

Östen Dahl

Verbs of becoming as future copulas

In many languages of Northern Europe, there is a strong tendency to use the present tense of verbs meaning ‘become’ when talking of states in the future. This issue has been somewhat neglected in the literature, although scattered cursory statements are found (cf., e.g., Saltveit 1962; cf. also the brief discussion in Bybee, Perkins & Pagliuca 1994: 261). In this paper, I shall look at the extent of this phenomenon and try to elucidate its nature. Of course, it seems quite natural that the distinction between being and becoming should blur with respect to the future, since the prototypical situation involves both the state itself and the event that marks its beginning. Thus, the sentences (1a) and (1b) are equivalent in most contexts. (1) a. b.

You’ll soon be a big boy. You’ll soon become a big boy.

In Swedish, the most natural way of rendering the same content would be by simply using the present tense of bli ‘become’: (2)

Swedish Du blir snart en stor pojke. you:SG become:PRS soon a big boy ‘You will soon be a big boy.’

The use of blir in (2) might be explained by appealing to the general tendency to use the present tense of telic verbs for future time reference in Swedish (as in most other Germanic languages). Other cases are not equally easily dealt with. Consider the following Swedish example, which should be thought of as occurring in the context of a weather forecast: (3)

Swedish Det blir kallt imorgon. it become:PRS cold:NT tomorrow ‘It will be cold tomorrow.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

352

352

Östen Dahl

What is notable in this example is that there is no implication of change. (3) can be used even if it is cold when it is uttered (it sounds even better if också ‘too’ is added). Notice that in English, It will get/become cold tomorrow is not natural in such a situation, and It will get cold tomorrow too hardly makes any sense. Likewise, in Russian, (4) would unequivocally imply that the weather will change: (4)

Russian Zavtra stanet xolodno. tomorrow become:PF:PRS cold ‘It will become cold tomorrow.’

On the other hand, it is not always possible to translate will be as blir. The following is an example: (5)

Swedish Barnen kommer att vara mycket sömniga när pappa child:PL come INFM be:INF very sleepy when father kommer. come:PRS ‘The children will be very sleepy when father arrives.’

In (5), substituting blir for kommer att vara ‘will be’ changes the interpretation: it would mean that Father’s arrival makes the children sleepy, or at least that they become sleepy when he arrives, whereas (5), as it stands, is naturally interpreted as implying that the children are already sleepy when he arrives. Analogously, (6) suggests a causal relationship between our arrival in Gothenburg and the weather: (6)

Swedish Vädret blir fint när vi kommer fram till weather:DEF become:PRS fine when we come PRT to Göteborg. Gothenburg ‘The weather will be fine when we arrive in Gothenburg.’

In these examples, then, the use of blir does seem to imply that there is an event which changes a state of affairs. We must therefore look more closely at the examples which do not have this implication, to see what the conditions are. Consider the following Swedish example: (7)

Swedish Den här festen blir nog trevlig. this here party become:PRS surely pleasant ‘This party will surely be pleasant.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

353

Verbs of becoming as future copulas

353

This clearly does not mean that the party first is unpleasant and then becomes pleasant. Rather, what it means is that the party will display the quality of being pleasant when it takes place. A possible translation into English would be: (8)

The party will turn out pleasant.

(7) represents what could be seen as a prototypical example of this extended use of bli. An essential feature seems to be that the property concerned is one that the entity in question acquires when it comes into existence or develops. This makes it natural for the subject of such sentences to denote events and similar types of entities. The use of ‘become’ found in (7) is not restricted to future time reference. Cf. (9), where we are talking of a party that has already taken place. (9)

Swedish Festen blev trevlig. party:DEF become:PST pleasant ‘The party became pleasant.’

In other words, what we are here dealing with is, in principle, a special use of verbs of becoming rather than a future tense marker. But there is still a clear link to future time reference. The use of a verb of becoming in an example like (9) relates the state of affairs in question to some observer’s previous expectations.1 When talking of the future, the speaker and/or the listener fill the role of this observer. This makes the use particularly natural with future time reference. It may be mentioned here that in Finnish, a formal distinction between the core uses of becoming and the ‘turn out’ use, in that the predicate of the ‘turn out’ construction takes the nominative or partitive case rather than the translative. Examples like the following are commonly cited (Setälä 1973: 30, Almqvist 1990)2 (10)

Finnish Puuro(sta) tuli mustaa. porridge(:ELAT) become:PST black:PRTV ‘The porridge turned out black, i.e., when it was ready.’

(11)

Finnish Puuro tuli mustaksi. porridge:NOM become:PST black:TRNSL ‘The porridge became black, i.e., it changed from some other colour.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

354

354

Östen Dahl

What about the weather examples, with which we started the discussion? Most of these examples are constructed with a dummy subject, and it seems that the question of coming into existence does not make a lot of sense here. It might be argued that in saying that it will be cold tomorrow, we are really thinking of ‘the weather tomorrow’ as something that does not yet exist. Sentences expressing location present special problems. In Swedish, one cannot in general use bli in sentences like the following: (12)

Swedish *Blir du på institutionen imorgon? become:PRS you:SG at department:DEF tomorrow ‘Will you be at the department tomorrow?.’

Rather, one would use är ‘am, are, is’ or kommer att vara ‘comes to be’: (13)

Swedish Är du / be:PRS you:SG kommer du att vara på institutionen imorgon? come:PRS you:SG INFM be:INF on department:DEF tomorrow ‘Are you/Will you be at the department tomorrow?’

Likewise, in German, one could not say (14) but would have to add sein, as in (15): (14)

German Wirst du morgen im Institut? FUTAUX:2SG you tomorrow in_DEF department

(15)

German Wirst du morgen im Institut sein? FUTAUX:2SG you tomorrow in_DEF department be:INF

This is fairly natural, since Swedish bli and German werden do not in general function as inchoatives in locative constructions: a movement verb like Swedish komma ‘come’ is used instead. However, there are at least two classes of counterexamples to the claim that bli is excluded from locative constructions in Swedish. One might in fact say: (16)

Swedish Jag blir hemma imorgon. I become:PRS at home tomorrow ‘I will be at home tomorrow.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

355

Verbs of becoming as future copulas

355

A closer look reveals that bli here has its original meaning, ‘remain’. In German, where the verb for ‘become’, werden, does not have this ambiguity, it cannot be used in translating (16). Rather, bleiben ‘remain’ is used: (17)

German Ich bleibe morgen zu Hause. I become:PRS tomorrow at home ‘I will stay at home tomorrow.’

Another type of counterexample came to my attention when I saw the following headline in a German newspaper (about three years before the meeting actually took place, so it was clearly future time reference at the time): (18)

German Die Katholiken-Tagung 1994 ist in Dresden. the Catholic meeting 1994 be:PRS in Dresden ‘The 1994 Catholic meeting will be in Dresden.’

This is not possible in Swedish, given the context. A Swedish newspaper would say [what would be ungrammatical in German (Thieroff 1992)]: (19)

Swedish Katolikmötet 1994 blir i Dresden. Catholic meeting:DEF 1994 become:PRS in Dresden ‘The 1994 Catholic meeting will be in Dresden.’

We may note that the subject in (19) is an event and thus of a type that would typically be expected to occur with the extended use of ‘become’ verbs already discussed. But there also seem to be pragmatic principles at work. Using är ‘is’ instead of blir ‘becomes’ in such a sentence would be much more natural in some other contexts. The crucial feature of (19) seems to be that it reports a decision, as a piece of ‘hot news’. If we are talking about something that has been scheduled for a long time, är ‘is’ is possible: (20)

Swedish Nästa sammanträde är på torsdag. next meeting be:PRS on Thursday ‘The next meeting will be on Thursday.’

The ‘hot news’ reading is also possible in a future-in-the-past context, e.g.,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

356

356 (21)

Östen Dahl

Swedish Mötet blev den 24 maj. meeting:DEF become:PST DEF 24 May ‘(It was decided) that the meeting will/would be on May 24.’

Notice that (21) can be felicitously uttered both before and after May 24. Furthermore, it is actually ambiguous between the future-in-the past reading and the interpretation ‘The meeting took place, as it turned out, on May 24’ (cf. above) – another illustration of the complexities of the future-related uses of ‘become’ verbs. In addition to verbs meaning ‘become’, similar extended uses may also be found with verbs meaning ‘get, receive’. The following is an example from Swedish, which should be interpreted as saying that my children will have fair hair when they are born (or soon after): (22)

Swedish Mina barn kommer nog att få / får nog my child:PL come:PRS probably to get / get:PRS probably ljust hår. fair hair ‘My children will probably have fair hair.’

Among the Germanic languages, Swiss German should be mentioned as a special case where there are several verbs of change corresponding to become in English, all of which may be used to express future time reference in the ways discussed here. Compare the following examples from Züritüütsch (Bickel 1992: 78): (23)

Züritüütsch Er isch Leerer. he PRS teacher ‘He is a teacher.’

(24)

Züritüütsch Er wììrt Leerer. he FUT teacher ‘He will be a teacher.’

(25)

Züritüütsch Es isch chalt. it be:PRS cold ‘It is cold.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

(nominal predicates)

(impersonal adjectival stative predicates)

357

Verbs of becoming as future copulas

357

(26)

Züritüütsch Es wììrt chalt. it INCH cold ‘It’s getting cold.’ (also: ‘It will be cold.’)

(27)

Züritüütsch (adjectival stative predicates with a personal subject) Si hät chalt. she have:PRS cold ‘She is cold.’

(28)

Züritüütsch Si chunt chalt über. she come:PRS cold over ‘She becomes cold/will be cold.’

(29)

Züritüütsch Unen ine hät s e Baar. below inside PRS it INDEF:NT bar ‘Downstairs there is a bar.’

(30)

Züritüütsch

(existential clauses)

Unen ine git s e Baar. below inside FUT it INDEF:NT bar ‘Downstairs there will be a bar.’ (31)

Züritüütsch (nominal predicates in identity constructions) Das isch öises Huus. DEM:NT be:PRS our:NT house ‘This is our house.’

(32)

Züritüütsch Das git öises Huus DEM:NT FUT our:NT house ‘This will be our house.’

The extended uses of ‘become’ verbs seem to cover most of the Germanic (English is an exception) and Finno-Ugrian languages in Europe. It also turns out to have a venerable tradition at least in the former language family. In grammars of Gothic, the use of the verb wairþan ‘to become’ for ‘will be’ is regularly mentioned, and some of the examples found in the texts are intriguingly analogous to the ones discussed above. Compare, for example:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

358

358

Östen Dahl

(33)

Gothic (Mark 13: 18) Aþþan bidjáiþ ei ni wairþái sa þlaúhs but pray:IMP:2PL COMP not become:OPT:3SG this flight izwar wintráu. your winter:DAT ‘And pray ye that your flight be not in the winter.’

In some places, the development seems to have gone further. A language in which the paradigms for ‘be’ and ‘become’ seem partly integrated is Hungarian. In the past, there is a clear stative/inchoative difference between the stems vol- ‘be’ and len-, ‘become’, e.g.,3 (34)

Hungarian Beteg voltam. sick be:PST.1.SG ‘I was sick.’

(35)

Hungarian Beteg lettem. sick become:PST.1.SG ‘I became sick.’

When referring to the future, on the other hand (Csató Johanson 1992), forms of len- show up in contexts similar to those discussed above but also in contexts where Germanic languages would not normally use a verb of becoming, for instance: (36)

Hungarian Ötkor még ott leszek. five:TEMP still there LESZ:1SG ‘I will still be there at five o’clock.’

Grammatical descriptions of Hungarian (e.g., Károly 1972, Tompa 1972) note that len- forms are used instead of vol- for future time reference. Károly (1972: 92) gives a table that suggests that the opposition static:dynamic is neutralized in the future. There are some indications, however, that this neutralization is not complete. According to Csató Johanson (1992), (37a) is more natural than (37b) if the speaker is already at the place referred to. (37) a.

Hungarian Ötkor ott vagyok. five:TEMP there be:1SG ‘I am there at five o’clock.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

359

Verbs of becoming as future copulas

b.

359

Ötkor ott leszek. five:TEMP there LESZ:1SG ‘I will be there at five o’clock.’

Admittedly, this may be a fairly easily cancelled implicature (cf. 36). It may also be noted that according to many speakers of Hungarian, the literal translation of (5) into that language gives the same impression as the Swedish original, namely that the children become sleepy only after Father has arrived: (38)

Hungarian A gyerekek álmosak lesznek, amikor az apjuk the child:PL sleepy LESZ:3PL when the father:POSS.3PL hazajön. come_home:PRS:3SG ‘The children become/will be sleepy when father arrives.’

Thus, even in Hungarian, the treatment of ‘becomes’ as a future copula is somewhat problematic. Looking at the geographical distribution of the extended ’become’ phenomenon, we find that it coincides largely with what is called the North European “futureless” area in Dahl (this volume). What then about the third language family involved in this area, viz. Slavic? In most Slavic languages, there is a paradigm formed from the stem bo¬ d-, which functions as a future copula. It is used as an auxiliary in the periphrastic imperfective in West and East Slavic, but it is both significantly older and has a wider distribution than this construction. It was current already in Old Church Slavonic and there is some evidence that it had an inchoative interpretation there. In Havránek (1980: 114) the following example is given: (39)

Old Church Slavonic (Mark 4: 32) bole v˘ısˇex V˘uzdrastet˘ı i bo¬ det˘u grow:PRS and become:PRS great:COMP all:GEN:PL zelii. herb:GEN:PL ‘It groweth up, and becometh greater than all herbs.’

bo¬ det˘u here translates Greek gínetai ‘becomes’. (Admittedly, ‘will be’ would not be impossible in this context, so the evidence for bo¬ det˘u being a regular inchoative verb is perhaps not totally clear.). The etymology given for example in Havránek (1980: 114) for forms like bo¬ det˘u is that they are ‘a typical ingressive present with an n infix’ from the Indo-European root bh¯u4. If this is correct, it would show that



$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

360

360

Östen Dahl

the path from ‘become’ to ‘will be’ is a possible one, and that it has been manifested within the North European “futureless” area. It is thus possible that even if the Germanic and Baltic-Finnic verbs of becoming cannot be regarded as future copulas, the extensions of their use that we can observe represent the first step in such a grammaticalization path. One thing that remains to be elucidated is the precise relation between the extended ‘become’ verbs and the rise of periphrastic future constructions involving such a verb as an auxiliary, like the German werden construction (Saltveit 1962) or the North Slavic periphrastic imperfective future.

Notes 1. It may be noted that Romance languages often use the (perfective) Simple Past of the verb ‘to be’ in similar contexts. Consider the following French example from a corpus of newspaper texts: Les resultats de cette enquête furent surprenants ‘The results of this survey were surprising’ (Wiberg 1995: 195). 2. According to Almqvist, the elative case (puurosta) rather than the nominative (puuro) is normal in modern written Finnish. The generalization of the elative subject construction in Finnish is in itself an interesting grammaticalization process which is, however, outside the scope of this paper. For a discussion, see Almqvist (1990). 3. In the present, a zero copula is the normal choice in constructions; however, forms of vol- are used, e.g., in locative constructions. 4. The root bh¯u- in itself is often said to have ’become’ as one interpretation, but this does not appear directly relevant to the issue here.



References Almqvist, Ingrid 1990 “Varför det är lättare att bli femtio än att bli gammal på finska”, in: Ingrid Almqvist, Per-Erik Cederholm, Jarmo Lainio (eds.), Från Pohjolas pörten till kognitiv kontakt. Vänskrift till Erling Wande den 9 maj 1990. Stockholm: Department of Finnish, Stockholm University, 15–28. Bickel, Balthasar 1992 “The Marking of Future Time Reference in Züritüütsch”. Future Time Reference in European Languages I. E UROTYP Working Papers VI: 2. Bybee, Joan, Revere Perkins & William Pagliuca 1994 The Evolution of Grammar. Tense, Aspect, and Modality in the Languages of the World. Chicago: The University of Chicago Press. Csató Johanson, Éva 1992 “The Marking of Future Time Reference in Hungarian”, in: Future Time Reference in European Languages I. E UROTYP Working Papers VI: 2. Dahl, Östen this volume

“The grammar of future time reference in European languages”.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

361

Verbs of becoming as future copulas

361

Havránek, Bohuslav (ed.) 1980 Etymologický slovník slovanských jazyk˚u: slova gramatická a zajména. Sv. 2. Praha: Academia. Károly, Sándor 1972 “The grammatical system of Hungarian”, in: Loránd Benk˝o & Imre Samu (eds.), The Hungarian Language. The Hague: Mouton. Saltveit, Laurits 1962 Studien zum deutschen Futur. Die Fügung ‘werden mit dem Partizip des Präsens’ und ‘werden mit dem Infinitiv’ in ihrer heutigen Funktion und in ihrer geschichtlichen Entwicklung. Bergen: Norwegian Universities Press. Setälä, Eemil N. 1973 Suomen kielen lauseoppi. Helsinki: Otava. Thieroff, Rolf 1992 “The Marking of Future Time Reference in German”. Future Time Reference in European Languages. E UROTYP Working Papers III: 2. Tompa, Josef 1968 Ungarische Grammatik. The Hague: Mouton. Wiberg, Lars-Erik 1995 Le passé simple, son emploi dans le discours journalistique. Stockholm: Almqvist & Wiksell.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

362

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

363

The Perfect

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

364

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

365

Jouko Lindstedt

The perfect – aspectual, temporal and evidential

1. Introduction The perfect has found its way from grammars of Classical Greek and Latin to those of modern European languages – as a term. But it is usually described as part of language-specific tense and aspect systems; there have not been many attempts to explore its nature as a cross-linguistic category, and it is often not even asked whether the “Perfects” of languages A and B are really manifestations of the same typological feature at all, or only happen to share the same name for obscure historical reasons. The perfect and its development in various European languages was one of the focal research areas of the E UROTYP Tense and Aspect Theme Group. The point of departure was Dahl’s (1985: 129–153) important result that a cross-linguistic category of perfect can be identified empirically, without a preconceived definition of its semantics. The perfects of various languages centre on certain prototypical examples like the following (Dahl 1985: 131); the uninflected verb should be replaced with a properly inflected verb form in each language under investigation: (1)

[A says: I want to give your brother a book to read, but I don’t know which. Is there any of these books that he READ already? – B answers:] (Yes,) he READ this book.

Obviously, the English Perfect (as in He has read this book, or He’s read this one) would be a good candidate for an instance of this cross-linguistic gram type (for a definition of the term, see Dahl, this volume). Material about the perfect and related categories was collected by means of a typological questionnaire, referred to as the Perfect Questionnaire – PFQ for short – in the articles of this section (see Appendix 2). The first part contains 88 items, each consisting of one or more sentences to be translated with the help of contextual information (for translation questionnaires, see Dahl 1985, and Dahl, this volume); Part II, to be answered by a linguist, contains analytical questions. So far the questionnaire has been completed in over thirty languages,1 and for some of these, such as Dutch, Italian and Macedonian, as well as the Serbo-Croatian area,2 dialectal or individual variation has been taken into account to some extent. Maslov (1990) gives the following notional definition of the perfect: “an aspectotemporal form of the verb, expressing a present state as a result of a preceding action or change, and/or expressing a past action, event or state that is somehow important

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

366

366

Jouko Lindstedt

to the present and is considered from the present point of view, detached from other past facts” (my translation). Actually the first part of the definition rather applies to a gram type called resultative (see below); important elements in the definition of the perfect proper are (1) the relevance of a past situation from the present point of view and (2) detachment from other past facts, i.e., non-narrativity. According to Comrie (1976: 52), “the perfect indicates the continuing present relevance of a past situation”. The notion of current relevance (CR) is further explicated by Dahl and Hedin (this volume). In their recent major work on the grammaticalization of tense, aspect and modality, Bybee, Perkins & Pagliuca (1994) define the gram type anterior, which is essentially the same as the perfect discussed here. Their definition is, however, broader in that it also provides for past anteriors and future anteriors: “an anterior signals that the situation occurs prior to reference time and is relevant to the situation at reference time [ ] Anteriors may occur with past or future tense marking” (p. 54). Past perfects (or pluperfects) and future perfects (futura exacta) share several properties with present perfects, but they also have some special uses and characteristics of their own, and they will not be discussed in this chapter, although PFQ does contain several items pertaining to them. In PFQ, definitions have been operationalized: a language possesses a perfect if it has a gram, associated with the verb, that is used in most of the first seven examples–which illustrate different kinds of CR of past situations–but is not used in the following four examples, consisting of short narratives.3 This means that there is a negative criterion as well: the perfect is not a narrative tense, and therefore the Latin and Serbo-Croatian Perfects, for instance, do not belong to the cross-linguistic type “perfect” at all–the Latin Perfect is a perfective past tense, and the Serbo-Croatian Perfect is a general past tense, or preterite.4 In terms of Greenberg’s (1978: 75–76) distinction between the stability and frequency of typological features, the perfect is a gram type that is frequent, that is to say, likely to appear in different languages, but unstable, as it often tends to be lost. More often than not, it does not disappear as a form but becomes something else – a general past tense, for instance.

2. From resultative to current relevance Bybee and Dahl (1989: 67–68) list four typical diachronic sources of the perfect in the languages of the world: (i) copula + past participle of the main verb; (ii) possessive constructions involving a past participle of the main verb; (iii) main verb + particle meaning ‘already’; (iv) constructions involving verbs like ’finish’ or ’cast aside’. The two latter sources are by their semantics completive constructions, whereas the two first ones, common in European languages, are resultatives (Bybee, Perkins &

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

367

The perfect – aspectual, temporal and evidential

367

Pagliuca 1994: 53–74). A European perfect deriving from a possessive construction often involves a transitive verb meaning ‘to have’; if this is the case, it can be called a ‘have’ perfect.5 A copula-based perfect is a ‘be’ perfect. The distinction between resultatives and perfects has been established in linguistics only recently, largely owing to the important collective work edited by Nedjalkov (1988/1983). Resultatives “signal that a state exists as a result of a past action” (Bybee, Perkins & Pagliuca 1994: 54). Traditionally the resultative is often subsumed under the category of perfect as a special “statal” variant (as in Maslov’s definition quoted above), or it is simply called a “stative”. For the criteria of distinguishing resultatives from perfects, the reader is referred to Nedjalkov & Jaxontov (1988), Dahl (1985: 133–135), Bybee & Dahl (1989: 68), and Bybee, Perkins & Pagliuca (1994: 63–687), but the most important single difference should be mentioned here: only resultatives combine with adverbs of unlimited duration, such as ’still’ or ’as before’. In English, it is not possible to say *She has still gone (if still is used in its temporal meaning)–compare with the resultative construction She is still gone. In Finnish, the perfect and its resultative source coexist, but they are formally differentiated in the plural: (2)

Finnish He o-vat väsy-nee-t. they be-PRS:3PL become:tired-PP-PL:NOM ‘They have become tired.’

(3)

Finnish He o-vat väsy-ne-itä. they be-PRS:3PL become:tired-PP-PL:PRTV ‘They are tired.’

(2¼ )

Finnish *He o-vat yhä väsy-nee-t. they be-PRS:3PL still become:tired-PP-PL:NOM

(3¼ )

Finnish He o-vat yhä väsy-ne-itä. they be-PRS:3PL still become:tired-PP-PL:PRTV ‘They are still tired.’

The Finnish resultative construction, besides accepting an adverb meaning ‘still’, follows the syntactic rule of ascriptive6 sentences whereby the predicate adjective is usually in the partitive case when the subject is plural but does not refer to a pair of things; the perfect conserves a petrified older syntactic rule and puts the participle into the nominative.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

368

368

Jouko Lindstedt

The perfect is typically a periphrastic gram, being formally close to its resultative (or completive) source. An important exception seems to be the old Indo-European Perfect, as attested in Classical Greek and Old Indic. A newer inflectional perfect in statu nascendi is the active resultative construction in North Russian dialects (Trubinskij 1988: 394; Tommola, this volume): (4)

Russian (dialectal) On den’gi poluˇci-vši he money receive-PST:GER ‘He has received the money.’

This resultative gram – which has already assumed some features of a real perfect, cf. Nedjalkov & Jaxontov (1988: 42–43) – is not periphrastic for the simple reason that there is no present tense copula in Russian; in the past tense, the copula is regularly expressed in the resultative, too. It is possible that what look like inflectional perfects usually come into existence as copula-less ascriptive structures. Semantically, the change from resultative into perfect means the generalization of meaning from “current result” to “current relevance”. Lexically this is reflected in the spread of the gram from telic to atelic verbs (cf. Bybee, Perkins & Pagliuca 1994: 68–69; Dahl & Hedin, this volume). As the semantic connection between resultativity and CR is easy to grasp, the central use of the perfect is often called the “resultative perfect”. But since the “still test” and similar criteria are operational in showing whether the transition from resultative proper to perfect has taken place, I shall speak rather of the “CR perfect” instead. A CR perfect is a perfect in its most central, prototypical meaning. Is it a tense or an aspect category? In a gram-based approach this is not an essential question. Classifying the perfect among aspects or among tenses would not significantly add to our understanding of it. But if tenses express the temporal location of situations, and aspects their temporal shape, it can at least be said that the rise of a CR perfect is associated with the loosening of the aspectual properties of its resultative (or completive) source. As noted by Bybee (1985: 160), “it seems to resemble a tense more than an aspect, since it does not affect the internal temporal contours of the situation”.7 The central aspectual parameter of boundedness comes in two varieties: when telic situations reach their natural end-points, we can speak of material bounds (as in Susan built a beautiful house); when a situation, not necessarily telic, is limited temporally, a temporal bound is reached (as in Susan slept for nine hours; see Lindstedt 1995; cf. also Bertinetto & Delfitto, this volume). A material bound entails a temporal bound, but not vice versa. Resultativity presupposes a material bound, whereas CR only presupposes a temporal bound: a sentence like Someone has been here implicates that the presence of that someone has ceased, but it does not make “being here” a telic situation.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

369

The perfect – aspectual, temporal and evidential

369

3. From current relevance to indefinite past The CR perfect shades into what is usually called the experiential (or existential) perfect. In English, these two types are formally differentiated only in rare cases like the following (cf. Comrie 1976: 58–59): (5)

Mary has gone to Paris.

(6)

Mary has been to Paris.

In (5), the fact of Mary’s having gone to Paris may be relevant to the present state of affairs in various ways, but typically the sentence implicates at least that she is not present. The experiential perfect of (6) only expresses that the past situation in which Mary went to Paris is more indirectly part of the present state of affairs, most notably through Mary, who perhaps now knows what Paris is like. In its narrower definition, an experiential perfect presupposes an animate agent, since it expresses that “certain qualities or knowledge are attributable to the agent due to past experiences” (Bybee, Perkins & Pagliuca 1994: 62). In a broader definition, it only means that “a given situation has held at least once during some time in the past leading up to the present” (Comrie 1976: 58). This notion is further explicated by Dahl & Hedin (this volume), who call it “type-focusing event reference” (as opposed to “token-focusing event reference”, as in Mary went to Paris). The experiential perfect is a way of referring to a past situation without referring to a particular occasion, that is to say, it is characterized by non-specific past time reference. From this point of view, it is more tense-like than the CR perfect, being an indefinite past tense which typically occurs in questions and negated assertions with ‘ever’-type adverbials. However, CR and experientiality do not exclude each other; in examples like (1), elements of both can be discerned. This gradual transition of the perfect from the aspectual to the temporal domain means that it comes to be linked to an integrative way of viewing the past (Lindstedt 1983). The narrative tenses – the tenses of Weinrich’s (1963) erzählte Welt – primarily refer to specific occasions, to the past moments defined by the internal structure of each narrative. If all possible past occasions are spoken about, the dynamic worlds of narratives must be replaced by a single static world, Weinrich’s besprochene Welt, comprising the past, present and future. In the perfect’s world, all past situations are still present – first through their results, then because they themselves become parts of this “extended now” (cf. McCoard 1978). However, experientiality as such is neither a necessary nor a sufficient condition for a gram to be called a perfect. Experientials proper are a distinct gram-type, described by Dahl (1985: 139–144); the Japanese -ta koto ga aru is a well-known example (see also Dahl & Hedin, this volume).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

370

370

Jouko Lindstedt

In Slavic languages without a distinct perfect gram, the CR perfects of other languages are usually rendered with Perfective Past forms, whereas experiential perfects can often be translated with Imperfective Pasts. In Bulgarian, which does possess a perfect, a Perfect of a Perfective verb is interpreted as a CR perfect, whereas experiential Perfects are usually Imperfective (see Lindstedt 1985: 223–229; Lindstedt 1995: 99–100). Compare: (7)

Bulgarian A vie k˘ade ste cˇ u-l-i tova? but you:PL where be:2PL hear:PFV-PP-PL that:NT ‘Where did you [polite] hear that?’ (CR perfect)

An evasive answer would be: (8)

Bulgarian ˇ Cu-va-l-a s˘am. hear-IPFV-PP-F be:1SG ‘I’ve just heard.’ (experiential perfect)

Although the experiential meaning may become dominant in the perfect, historically it is usually secondary and derives from the CR meaning. The perfect of a particular language may well be compatible with specific past time adverbials. Sentence (9) is ungrammatical in English, to be sure: (9)

*I have woken up at 4 o’clock this morning.

However, a perfect would be possible – though not the only alternative – in Finnish and Bulgarian, for instance. This is because there is an obvious CR reading – I woke up so early that I am now tired.8 According to Dahl (1985: 137–138), Swedish seems to occupy an intermediary position: a specific time adverbial can combine with the perfect if it is in the information focus. I assume that the degree of incompatibility of specific time adverbials with the perfect in a particular language shows to what extent it has become a dominantly experiential form and, therefore, a kind of tense. A possible next stage in this development would be the total loss of the CR reading, and this is what may be taking place in American English. If a child asks his or her parent, “Can I go now?”, the counter-question (10a) strikes one as belonging to an American parent, whereas (10b) is the preferred alternative in British English (PFQ: 5): (10) a. b.

Did you do your homework? Have you done your homework?

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

371

The perfect – aspectual, temporal and evidential

371

I am not aware of any extensive studies of this difference, but its existence, as a tendency rather than a rule, is generally taken for granted.9 I assume that the experiential function has been preserved much better in American English than the CR function has; thus, both British and American speakers would normally ask, for instance, Have you (ever) met my sister?; the sentence Did you meet my sister? would refer to a specific occasion or interval. The American perfect can be considered to be more tense-like than its British counterpart. To return to Bulgarian, at closer examination the situation in this language is similar to the American English situation in that CR does not require a Perfect form to be used at all; as a matter of fact, the Aorist could well be used in translating (10): (11)

Bulgarian (PFQ: 5) Na-pisa li si domašno-to? PFV-write:AOR:2SG Q RFL:DAT homework-DEF:NT ‘Have you done your homework?’

Although the Perfect would also be possible here, it is actually difficult to show that the Modern Bulgarian Perfect has a CR meaning because, as was mentioned, what look like CR perfects are always instances of the Perfective Aspect. As the CR meaning can be expressed with a Perfective Aorist, but not with an Imperfective Perfect, it must be the Perfective, not the Perfect, that is the primary carrier of the CR meaning. On the other hand, the Perfect is obligatory in experiential contexts, a fact which is also reflected in native grammarians’ traditional name for this gram, minalo neopredeleno vreme ‘past indefinite tense’. Hannu Tommola and Nina Niissalo (p.c.) point out that the same situation is observed in Slavic as early as in Old East Slavic (“Old Russian”) texts: experientiality requires the Perfect, but in the CR meaning, both Perfective Perfects and Perfective Aorists are possible. It is thus possible that the CR meaning of the Slavic Perfect grew weaker early in the Proto-Slavic period when verbal aspect came into being.

4. The perfect becomes a narrative tense When a perfect can be used as a narrative tense (with the possible exception of evidential contexts, see below), it has ceased to be a perfect. This is what has happened in the majority of Slavic languages, as well as in South German and North Italian dialects, in Sardinian (Georg Bossong, p.c.) and spoken French. As a matter of fact, the perfect has become a peculiar “maritime category” in Modern Europe – most of the languages and dialects with a stable perfect are situated on the fringe of the continent: the Baltic Finnic languages, Scandinavian languages, North German dialects, English, Portuguese, Spanish, South Italian dialects, Greek, Albanian, Macedonian and Bulgarian. (Cf. Thieroff, this volume, Map 4.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

372

372

Jouko Lindstedt

The macroareal picture is completed by the observation that at the corners of this maritime fringe, Portuguese and Modern Greek exhibit what look like “young perfects”, with relatively strong restrictions on what counts as CR. As can be seen in the article of Squartini & Bertinetto (this volume), the Portuguese Perfect (perfeito, also known as perfeito composto) is mainly used to express situations or series of repeated situations that span from the past up to the present. As a matter of fact, PFQ material obtained from one Portuguese informant suggests that it is rather the adverb jà ‘already, now’ that may be grammaticalizing as the real perfect marker in this language.10 As for the Modern Greek Perfect, restrictions on its use are less drastic, but Dahl & Hedin (this volume; cf. also Hedin 1987: 80–86) point out that in CR contexts, the Aorist can be used – just as in Bulgarian, cf. (10) above. In those contexts the Perfect focuses more on the present state of affairs and less on the past event itself than the Aorist does. Thus, the Greek Perfect is still relatively close to its resultative source – this can also be seen in the fact that it can only be formed from the Perfective (“Aorist”) stem of the verb. The gradual development of the German Perfect from a perfect into a general past tense seems to fit what Dahl (this volume) describes as a two-dimensional grammaticalization cline: both the geographical distance from the southern centre of innovation and the functional distance from the old prototypical uses of the perfect (such as CR) bear on the propensity for Perfect use (see Latzel 1974 and Thieroff, this volume on the use of the German Perfect in different contexts). In those Romance dialects in which the perfect has become a narrative tense, it is now a perfective past, opposed to the imperfective past (Squartini & Bertinetto, this volume). This is an instance of semantic generalization, for the CR meaning presupposes boundedness. Semantic bleaching of this kind is typical of later stages of a grammaticalization process (Hopper & Traugott 1993: 87–93; the notion of bleaching was introduced by v.d. Gabelentz 1891: 241–242). The change whereby the perfect of most Slavic languages has become a general past with no aspectual value of its own must at least be partly due to the fact that aspect came to be expressed by other means in Slavic. The perfect did not become a perfective past, since perfectivity vs. imperfectivity could be expressed with verbal affixes, independently of tense marking. A still stronger assumption would be that the loss of the perfect vs. narrative past distinction itself was caused by the rise of this aspect opposition, as the new Perfective, with a completive source, was able to assume resultative and CR functions. As suggested by Dahl & Hedin (this volume), the development from a perfect into a more general past tense “may at least partly be interpreted in terms of gradual relaxation of the requirements of current relevance”. It is more difficult to say whether there is also a path from experientiality (the indefinite past function) to a general past. Such a process cannot be observed in American English, but histori-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

373

The perfect – aspectual, temporal and evidential

373

cal data from Slavic suggest that it is at least a typological possibility if the Slavic Perfect had already lost most of its CR functions before the other past tenses disappeared. At any rate, it is an interesting question why a CR form tends to displace a nonCR form even when the CR distinction is lost. It should be noted that this is not what happened to the old Indo-European inflectional resultative / perfect in Greek, Latin or Germanic; in those languages, the Perfective Past or Simple Past got the upper hand, though formally incorporating parts of the old perfect morphology (as in the reduplicated Perfective Pasts of Latin, or in the –k– Aorists of Modern Greek; for Germanic, see Bybee, Perkins & Pagliuca 1994: 77–78). Thus, the morphological type of the perfect – periphrastic vs. inflectional – may have some bearing on its propensity to displace a simple past tense; the mechanism seems to be the same whereby case forms are often replaced by adpositional phrases in various languages. In terms of grammaticalization processes, the perfect is typically a younger gram than the other (narrative) past tenses–less synthetic in its expression, more marked as to its semantic content. Its functional expansion and the concomitant semantic bleaching open a space for a new perfect to appear–an example of a grammaticalization cycle known as layering (Hopper & Traugott 1993: 124–129). Thus, in some Slavic languages where the old perfect has become a general past tense, a new perfect is coming into existence. Such a development is observed in Czech, where the verb mít ‘to have’ enters with the past passive participle into constructions like Máš výˇcištˇené zuby? ‘Have you brushed your teeth?’ (Bybee & Dahl 1989: 70), Mám úlohu napsanou ‘I have completed my task’ (Maslov 1988: 80), mít knihu rozeˇctenou ‘to have a book half-read’ (Short 1993: 487; see also Tommola, this volume, section 7). Another possible path of evolution mentioned in connection with the perfect becoming a narrative tense is the path leading through the so-called perfect of recent past. As pointed out by Dahl (1985: 136), the development from the CR meaning into the recent-past meaning is natural because “a recent event is more likely to have a persistent result than a distant one”. Sometimes a perfect acquires a hodiernal interpretation: events of the same day can be referred to by it, even in connected narrative. This is illustrated in the PFQ material from a speaker of Alicante Spanish:

(12) a.

Alicante Spanish (PFQ: 8) [Do you know what happened to me just an hour ago?] Yo estaba andando en el bosque. De pronto, he pisado una culebra. Me ha mordido en la pierna. He cogido una piedra y se la he tirado a la culebra. Se ha muerto.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

374

374

Jouko Lindstedt

b.

(PFQ: 9) [Do you know what happened to me yesterday?] Yo estaba andando en el bosque. De pronto, pisé una culebra. Me mordió en la pierna. Cogí una piedra y se la tiré a la culebra. Se murió. ‘I was walking in the forest. Suddenly I stepped on a snake. It bit me in the leg. I picked up a stone and threw it at the snake. It died.’

The italicized forms in (12a) are Perfects; in (12b), Perfective Pasts are used instead. (The questionnaire was administered by Scott Schwenter; see Schwenter 1994.) An analogous distinction between PFQ: 8 and PFQ: 9 is made in Catalan, which is in areal contact with Alicante Spanish. The recent-past function of the perfect in Spanish, Catalan, Occitan and pre-Modern French is discussed by Comrie (1976: 60–61), Dahl (1985: 125) and Bybee, Perkins & Pagliuca (1994: 101–102); Comrie (1976: 61) suggests that “[g]radual relaxation of the degree of recentness required for the use of the Perfect seems to have been a key part of the development of the Perfect in many Romance languages to oust the Simple Past completely”. However, the study of Squartini & Bertinetto (in this volume) on the Romance perfect does not suggest that recency has had such a central role in this development, except for a limited area in Southwest Europe, with Catalan as its centre. A more general Central and East European trend, based on areal contacts and the typological expansiveness of formally and semantically marked grams, may account for the Romance, Slavic and German development alike.

5. Evidentiality and South Slavic perfects The South Slavic linguistic area consists of a series of closely connected dialects, grouped nowadays into five or six standard languages.11 The Proto-Slavic Perfect, a periphrastic formation of the ‘be’ type, has been preserved as a form in all of the area except for some South Macedonian dialects, but almost everywhere it has been subject to grammatical changes. However, there are two epicentres of these developments, and the respective outcomes are far from identical (Lindstedt 1993, 1994). Starting from the Slovene territory in the northwest, the Perfect simply became a general past tense, ousting first the Imperfect, then the Aorist. These latter grams have completely disappeared in Slovene (excepting some dialects spoken mainly in Italy); in Croatian and Bosnian standard languages they have only survived as literary archaisms, apart from some expressive functions of the Aorist that have nothing to do with narration. In Serbian, the Aorist and Imperfect are more frequent, but the Perfect is already the prevalent narrative tense here, too (Savi´c 1991). The old system in which the Perfect is only a non-narrative tense is preserved in some Mon-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

375

The perfect – aspectual, temporal and evidential

375

tenegrin dialects, whose system is sometimes reflected even in modern fiction written by authors from that area (Lindstedt 1994: 39). From the other end of the South Slavic area, a grammatical change has spread that has made the Perfect an evidential form, or rather built an evidential system out of the Perfect. Evidential, to be distinguished from mood, is, according to Trask (1993 s.v.), “a grammatical category occurring in some languages by which all statements (and sometimes other sentence types) are overtly and obligatorily marked to indicate the source of the speaker’s evidence for her/his utterance” (see also Chafe & Nichols 1986, and Willett 1988). There are no well-established terms for different types of evidentials.12 I propose the term indirective for the most widespread evidential gram type, expressing that the speaker has not witnessed the situation he or she is speaking about, but knows of it from hearsay or other kinds of indirect evidence. Other grammatical and semantic terms needed are reportative, which should be confined to the hearsay case alone,13 and inferential, referring to statements made on the basis of inference, not hearsay. There are two major areas in Europe where grammaticalized evidentiality distinctions are common. One is the Baltic region, comprising the Baltic languages Lithuanian and Latvian and the Finnic languages Estonian and Livonian. The other area can be called the Black Sea area, as it consists of languages around this sea, though it stretches farther to Central Asia (Haarmann 1970; Dahl 1985: 149–153; Friedman 1986; Johanson 1992: 244–246). The indirectives of this area are generally regarded as having arisen due to Turkish influence during the Ottoman reign. However, the tendency to develop indirectives from various sources is typical not only of Turkish, but of the whole Turkic stock, and the area also extends from the Black Sea to regions where such Uralic languages as Komi and Udmurt, farther north, are spoken (Leinonen & Vilkuna, this volume). It has even been called into question whether Turkish was really the primary Turkic source of the indirectives now found in the Balkans (Johanson ms.). An interesting question is whether the two indirective areas really form a single “Eurasian isogloss” as suggested by Haarmann (1970). But as evidentiality has not been grammaticalized in East Slavic (see Tommola, this volume, section 4.6, for PFQ material), there are no data connecting the Baltic region with the east and southeast. Nothing is known about the grammatical structures of those extinct Baltic and Uralic languages of Central Russia that might have formed a bridge between the two areas. In several languages of the Black Sea area, a resultative or a perfect has been the main diachronic source of the indirective. We can assume that the inferential meaning forms the semantic link here. Drawing inferences from the visible results of a non-witnessed event is a natural extension of resultativity or CR; it is resultativity the other way round, as it were (Comrie 1976: 108–110; Bybee, Perkins & Pagliuca 1994: 95–96). The inference in sentence (13), for instance, can be expressed by a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

376

376

Jouko Lindstedt

perfect even in some European languages outside the Black Sea area, though not in English: (13)

(PFQ: 69) [Investigating a burglary, seeing footprints beneath a window:] The thief entered (must have entered) the house by this window.

Swedish is among the languages that can use the perfect: (14)

Swedish (PFQ: 69) Tjuv-en ha-r komm-it in genom det här thief-DEF have-PRS come-SUP in through DEF:NT here fönstr-et. window-DEF:NT

The Scandinavian inferential perfect is discussed by Haugen (1972) and, with abundant examples from Swedish, by Kinnander (1973). The use also extends to Finnish. As for the German Perfect in the same sentence (Durch dieses Fenster ist der Dieb [in das Haus] eingedrungen), its inferential value is more difficult to assess, for, as noted above, this tense has extended its use towards the preterite; but Weinrich (1964: 84–86) observes that the German Perfect is typically used in historical prose when past facts are commented on and explained, not only related in a sequence, and this can be seen as a natural extension of the inferential use. As a further step in grammaticalization, inferentiality can be extended to the reportative meaning and other types of indirect evidence, so that a real indirective comes into existence (Bybee, Perkins & Pagliuca 1994: 96–97, 105; cf. also Dahl 1985: 152–153, and Nedjalkov’s “typological appendix” in Plungian 1988). Another possible connection between the perfect and indirective is the fact that participles often enter into subordinate constructions expressing reported speech. Participles used as predicates in subordinate sentences seem to be the main source of indirectives in the Baltic area (Ikola 1953: 62–63; Schmalstieg 1988: 117–121), and some partially similar developments are perhaps not excluded in the Black Sea area, either. On this grammaticalization path it is the reportative meaning that forms the basis of an indirective category. Let us now have a closer look at the South Slavic indirectives. In Bulgarian, the Perfect has given rise to an evidential form that can be characterized as an Indirective Aorist. Later, other indirective tenses have developed according to the structural model of the Indirective Aorist, such as the Indirective Imperfect (also used as the Indirective Present tense) and the Indirective Future. The distinction between the Perfect and the Indirective has been one of the most difficult questions for Bulgarian grammarians since the past century, and no definitive solution has been reached yet. (The description I gave in Lindstedt 1985 is not satisfactory, either.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

377

The perfect – aspectual, temporal and evidential

377

The main descriptive problem is that the Perfect and the Indirective Aorist are not formally differentiated. The feature often mentioned in normative grammars, viz. the omission of the auxiliary in the third person of the Indirective, is not an unambiguous marker: the auxiliary is usually omitted when Indirective tenses are used in connected narrative, but this is not an absolute rule; and there are other uses of the Indirective forms that do not formally differ from the Perfect at all (Lindstedt 1993: 43–48; Lindstedt 1994: 44–47). Auxiliary drop in certain discourse types is also observed in the neighbouring Serbo-Croatian area, but it is not connected with evidentiality (Savi´c, this volume). Despite the appearance of the Indirective forms, the Bulgarian Perfect has also retained its old meanings, notably the experiential (Lindstedt 1993: 49–50; 1994: 49–50). The sentence Mary has been to Paris would be translated into Bulgarian with the Perfect, and so would the sentence I have been to Paris, where the meaning is far from indirective – the speaker has certainly witnessed his or her own visits to Paris. The developmental path of the Bulgarian Perfect has clearly bifurcated: one of its two main functions is now experiential and non-narrative, the other is indirective and narrative. The linking CR function has grown weak, as noted in section 2 above; what we have here, then, is a kind of “doughnut gram” as defined by Dahl (this volume). In Macedonian, the old ‘be’ perfect has acquired an indirective meaning, as in Bulgarian, but it is not so clearly polysemous owing to the appearance of a new ‘have’ perfect modelled on Greek and Arumanian. There is also a third, mainly resultative construction, so that ‘I have come’ can be expressed in three ways (the 1SG forms with a masculine subject are given):

sum došol imam dojdeno sum dojden

‘I am’ + past participle active, M.SG. ‘I have’ + past participle passive, NT.SG ‘I am’ + past participle passive, M.SG.

Friedman (1976; 1977: 52–99) and Graves (this volume) have investigated the areal and functional distribution of these grams. In Northern Macedonia the sum došol type dominates in both the perfect and indirective meaning; this is closest to the Bulgarian situation. In the southernmost part of the language area, only imam dojdeno is used. In between, in Ohrid dialects, the old perfect is only used as an indirective, and the new type imam dojdeno serves as an ordinary perfect. As for the sum dojden type, it is still mainly a resultative proper, used in sentences like Umren e ‘he is dead’ (lit. ‘died is’, PFQ: 3). However, in southwest dialects it is on its way to becoming a perfect, which is shown by the fact that even with certain transitive verbs the subject may be the agent of the action:14

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

378

378 (15)

Jouko Lindstedt

Macedonian Jas sum jad-en-a. I be:1SG eat-PPP-F ‘I’ve been eaten’; SW Macedonian: ‘I’ve eaten.’

The development of the three Macedonian grams can be schematically presented as follows, depicting the synchronic situation as an on-going process, with different grams covering different stretches of the grammaticalization path: Resultative

Perfect

Indirective

Loss

sum dojden imam dojdeno sum došol However, this diagram should not be interpreted as a prediction that the newer types of perfect are also bound to become indirectives in due course. Although this is of course possible, it is not the most probable development typologically; and areally it is not clear whether the Balkan area still favours the appearance of new indirectives, for the Turkic influence has almost vanished.

6. The path of the perfect Although expressing the current relevance of a past situation is the central and prototypical meaning of the perfect, I know of no perfects that only have this function. I propose the following tentative universal: If a gram has the CR meaning, it also has at least one of the following meanings: resultative; experiential (indefinite past); inferential; reportative. If the central or sole meaning of a gram is resultative or indirective (inferential and/or reportative), it is not yet a perfect, or no longer a perfect. If the meaning of a gram has strongly shifted towards indefinite past, as in the American English Perfect or in the non-evidential branch of the Bulgarian Perfect, it is perhaps a matter of definition whether the term “perfect” should still be used. Given that CR and experientiality do not logically exclude each other, and often one and the same utterance contains elements of both, I am inclined to retain the name "perfect” for these cases. It should, however, be pointed out that the two American English speakers who completed PFQ used the Perfect in only about half of the defining items.15 Bybee, Perkins & Pagliuca (1994: 79) list 30 “old anteriors” from their crosslinguistic sample that have developed various further uses in addition to CR. They

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

379

The perfect – aspectual, temporal and evidential

379

also present a table of 57 “grams with anterior as their only use (young anteriors)” (1994: 64–65), but the caption is misleading: grams with other uses besides CR have been admitted to the table, if those uses (such as completive) were considered “indicative of early rather than later stages of development” (1994: 63). Moreover, experientiality or the function of anterior continuing (I’ve been waiting for him for an hour) was not frequently mentioned at all in the reference material used (1994: 62), so one cannot conclude that they are really absent from the grams of the table. Still another table (1994: 66) presents several grams that combine the resultative and “anterior” (CR) meanings. One can reach the conclusion that current relevance is not only a function, but also a junction: grammaticalization paths from different sources, such as resultative and completive, converge here; and further paths lead from here to experiential (indefinite past), perfective past, general past, and also to indirective (cf. Bybee, Perkins & Pagliuca 1994: 105, and also Anderson 1982: 240). Every concrete gram is only a passer-by at this junction, still having its tail in the resultative, or pushing its head towards new aspectual, temporal or evidential meanings.

Notes 1. More precisely, Part I has been completed in over thirty languages by one or more informants. The descriptive second part does not exist for all of these languages. The material will be placed on an Internet file server, possibly in the public domain if resources allow. 2. The Serbo-Croat(ian) area is the Central South Slavic dialect area between Slovene, on the one hand, and Bulgarian and Macedonian, on the other. The standard languages of this area are now called Croatian, Bosnian (see Halilovi´c 1991) and Serbian; they are all based on a dialect group called Neo-Štokavian. 3. It is important to comprehend the nature of this method. The crucial items in PFQ do not represent notional categories chosen a priori, because they draw upon Dahl’s (1985) empirical results describing the actual clustering of grammatical categories across languages. 4. This fact need not prevent these tenses from being called Perfects in the grammars of the respective languages by force of tradition. This applies especially for languages such as Dutch and Croatian/Bosnian/Serbian, in which the old Perfect has become a narrative tense and, therefore, has ceased to be a perfect proper, yet the old narrative tense or tenses still survive as (stylistically marked) alternatives. 5. Thus, Romance perfects have mainly developed from Latin constructions involving habeo ‘have’ or teneo ‘hold’ ‘have’, e.g., Navem paratam habeo ‘I have a ship prepared’ (Squartini & Bertinetto in this volume; Maslov 1988: 73; cf. Bybee & Dahl 1989: 72). Similar constructions occur in other language groups but, on the whole, ‘have’ perfects are a distinctively European phenomenon because “expressing ‘having’ by a transitive verb appears to be an areal trait of Western and Southern Europe” (Bybee & Dahl 1989: 98, fn. 7).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

380

380

Jouko Lindstedt

6. For the term “ascriptive”, see Lyons (1977: 437–438). 7. But she also points out that the perfect often has distinct forms for the present and past tense (as in English). 8. Notice that specific time adverbials can sometimes combine with resultatives in various languages, e.g., Swedish Boken är skriven 1950 ‘the book was written in 1950’ (lit. “is written”). Nedjalkov & Jaxontov (1988: 54) present the German resultative Diese Stadt ist im Jahre 1794 gegründet ‘this town was founded in 1794’ (lit. “is founded”), but not all native speakers are happy with this example. Temporal adverbials used with resultatives typically specify the time of the resulting state, not that of the past event. 9. Unfortunately studies on “the English Perfect” do not always specify what variety of English is investigated. 10. The informant inserted já into all of the diagnostic sentences from PFQ:1 to PFQ:7, and into several other typical perfect contexts as well, but did not use the perfeito. 11. See note 2. 12. A source of confusion is that the term “evidential” is often used as a name of an evidential gram (“the evidential in language X”), just as if the imperative, for instance, were called “the mood”. 13. Dahl’s (1985: 149) “quotative”, though used in Amerindian studies and now adopted by Trask, is not felicitous, because to quote means ‘to repeat the exact words someone has said or written’; in reportatives, only the content is repeated. Dahl now prefers the term “reportative”, too (p.c.). 14. For this criterion, see Nedjalkov & Jaxontov (1988: 23). 15. Both American English informants used the Perfect in PFQ:1, PFQ:4 and PFQ:7, one of them also in PFQ:6 (for the items, see the Appendix). An informant of British English used the Perfect in all defining items from PFQ:1 to PFQ:7 (in PFQ:3 only as a second alternative besides the stative / resultative He is dead).

References Anderson, Lloyd B. 1982 “The ‘Perfect’ as a universal and as a language-particular category”, in Paul J. Hopper (ed.), 227–264. Bertinetto, Pier Marco & Denis Delfitto this volume “Aspect vs. actionality. Why they should be kept apart”. Bybee, Joan L. 1985 Morphology: A study of the relation between meaning and form. Amsterdam & Philadelphia: Benjamins. Bybee, Joan L. & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Bybee, Joan & Revere Perkins & William Pagliuca 1994 The evolution of grammar: Tense, aspect, and modality in the languages of the world. Chicago and London: University of Chicago Press. Chafe, Wallace & Johanna Nichols (eds.) 1986 Evidentiality: The linguistic coding of epistemology. (Advances in Discourse Processes 20.) Norwood, New Jersey: Ablex.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

381

The perfect – aspectual, temporal and evidential

381

Comrie, Bernard 1976 Aspect. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press. Comrie, Bernard & Greville G. Corbett (eds.) 1993 The Slavonic Languages. London & New York: Routledge. Dahl, Östen 1985 Tense and aspect systems. Oxford & New York: Basil Blackwell. this volume “The tense-aspect systems of European languages in a typological perspective”. Dahl, Östen & Eva Hedin this volume “Current relevance and event reference”. Friedman, Victor A. 1976 “Dialectal synchrony and diachronic syntax: The Macedonian perfect", in Sanford B. Steever & al. (eds.), Papers from the parasession on diachronic syntax, Chicago Linguistic Society, 96–104. 1977 The grammatical categories of the Macedonian indicative. Columbus, Ohio: Slavica. 1986 “Evidentiality in the Balkans: Bulgarian, Macedonian, and Albanian”, in Chafe & Nichols (eds.), 168–187. Gabelentz, Georg von der 1891 Die Sprachwissenschaft. Ihre Aufgaben, Methoden und bisherigen Ergebnisse. [Linguistics: Its tasks, methods and present findings.] Leipzig: Weigel. Graves, Nina this volume “Macedonian – a Language with Three Perfects?” Greenberg, Joseph H. 1978 “Diachrony, synchrony, and language universals”, in Greenberg (ed.), 61–91. Greenberg, Joseph H. (ed.) 1978 Universals of human language, I: Method & theory. Stanford, California: Stanford University Press. Haarmann, Harald 1970 Die indirekte Erlebnisform als grammatische Kategorie. Eine eurasische Isoglosse. [The indirect evidential form as a grammatical category: An Eurasian isogloss.] (Veröffentlichungen der Societas Uralo-Altaica 2.) Wiesbaden: Otto Harrassowitz. Halilovic, Senahid 1991 Bosanski jezik. [The Bosnian language.] (Biblioteka Bosanski krug.) Sarajevo: Biblioteka Kljuˇcanin. Haugen, Einar 1972 “The inferential perfect in Scandinavian: a problem for contrastive linguistics”, The Canadian Journal of Linguistics 17: 132–139. Hedin, Eva 1987 On the use of the perfect and the pluperfect in Modern Greek. (Acta Universitatis Stockholmiensis, Studia Graeca Stockholmiensia 6.) Stockholm: Almqvist & Wiksell. Hopper, Paul J. (ed.) 1982 Tense-aspect: Between semantics and pragmatics. Amsterdam: Benjamins. Hopper, Paul J. & Elizabeth Closs Traugott 1993 Grammaticalization. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press. Ikola, Osmo 1953 Viron ja liivin modus obliquuksen historiaa. [On the history of the modus obliquus in Estonian and Livonian.] Helsinki: Suomalaisen Kirjallisuuden Seura. (Suomi 106: 4; with a German summary: “Zur Geschichte des estnischen und livischen Modus obliquus”.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

382

382

Jouko Lindstedt

Johanson, Lars 1992

manuscript

Strukturelle Faktoren in türkischen Sprachkontakten. [Structural factors in contacts with Turkic languages.] (Sitzungsberichte der Wissenschaftlichen Gesellschaft an der Johann Wolfgang Goethe-Universität. Frankfurt am Main, XXIX, 5.) Stuttgart: Franz Steiner. “Zum Kontakteinfluß türkischer Indirektive”. [On the contact influence of the Turkic indirective.]

Kinnander, Bengt 1973 “Perfektum i ’sekundär’ användning” [On the perfect in a ‘secondary’ use.], Nysvenska studier 53: 127–172. Latzel, Sigbert 1974

“Zum Gebrauch der deutschen Vergangenheitstempora. Zwei Studien” [On the use of the German past tenses. Two studies.], in Hermann Gelhaus & Sigbert Latzel, Studien zum Tempusgebrauch in Deutschen. (Institut für deutsche Sprache, Forschungsberichte, 15.) Mannheim, 169–348. Leinonen, Marja & Maria Vilkuna this volume “Past tenses in Permian languages”. LÈS Lingvistiˇceskij ènciklopediˇceskij slovar’ [Encyclopedic dictionary of linguistics] 1990 Moskva: Sovetskaja ènciklopedija.

Lindstedt, Jouko 1983 “The past is present: notes on the perfect tense”, in Fred Karlsson (ed.), Papers from the Seventh Scandinavian Conference of Linguistics, I. (University of Helsinki, Department of General Linguistics, Publications 9.) Helsinki. 246–259. 1985 On the semantics of tense and aspect in Bulgarian. (Slavica Helsingiensia 4.) Helsinki. 1993

“Za razvitieto na južnoslavjanskija perfekt” [On the development of the South Slavic perfect], Studia Slavica Finlandensia 10: 31–54.

1994

“On the development of the South Slavonic Perfect”, in Three papers on the perfect, 32–53. (E UROTYP Working Papers VI: 5). “Understanding perfectivity – understanding bounds”, in Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.), Temporal reference, aspect and actionality, vol. 2: Typological perspectives. Torino: Rosenber & Sellier. 95–103.

1995

Lyons, John 1977

Semantics 1–2. Cambridge: Cambridge University Press.

McCoard, Robert W. 1978 The English perfect: Tense-choice and pragmatic inferences. (North-Holland Linguistic Series 38.) Amsterdam: North Holland. Maslov, Jurij S. 1988 “Resultative, perfect, and aspect”, in Nedjalkov (ed.), 63–85. Maslov, Ju. S. 1990

“Perfekt” [Perfect], in LÈS, p. 372.

Nedjalkov, Vladimir P. (ed.) 1988 Typology of resultative constructions. (Typological Studies in Language 12.) Amsterdam & Philadelphia: John Benjamins. [An enlarged translation of Tipologija rezul’tativnyx konstrukcij (rezul’tativ, stativ, passiv, perfekt). Leningrad 1983: Nauka.] Nedjalkov, Vladimir P. & Sergej Je. Jaxontov 1988 “The typology of resultative constructions”, in Nedjalkov (ed.), 3–62. Plungian, Vladimir A. 1988 “Resultative and apparent evidential in Dogon”, in Nedjalkov (ed.), 481–493. (With a “Typological appendix” by V. P. Nedjalkov.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

383

The perfect – aspectual, temporal and evidential Savi´c, Svenka 1991

383

“Pragmatiˇcni aspekti vremena u naraciji u srpskohrvatskom standardnom jeziku” [Pragmatic aspects of tenses in naratives in the Serbo-Croatian standard language], Godišnjak Filozofskog fakulteta u Novom Sadu 20: 149–155. this volume “Discourse features of the truncated perfect in spoken Serbo-Croatian”. Schmalstieg, William R. 1988 A Lithuanian historical syntax. Columbus, Ohio: Slavica. Schwenter, Scott A. 1994 “The grammaticalization of an anterior in progress: Evidence from a peninsular Spanish dialect”, Studies in Language 18: 71–111. Short, David 1993 “Czech”, in Comrie & Corbett (eds.), 455–532. Squartini, Mario & Pier Marco Bertinetto this volume “The simple and compound past in Romance languages”. Thieroff, Rolf this volume “On the areal distribution of tense-aspect categories in Europe”. Tommola, Hannu this volume “On the perfect in North Slavic”. Trask, R. L. 1993 A dictionary of grammatical terms in linguistics. London & New York: Routledge. Trubinskij, Valentin I. 1988 “Resultative, passive, and perfect in Russian dialects”, in Nedjalkov (ed.), 389–409. Weinrich, Harald 1964 Tempus: Besprochene und erzählte Welt. Stuttgart: W. Kohlhammer. Willett, Thomas 1988 “A cross-linguistic survey of the grammaticization of evidentiality”, Studies in Language 12: 51–97.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

384

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

385

Östen Dahl and Eva Hedin

Current relevance and event reference

1. Introduction The work reported in this paper started out as an investigation of the use of perfects in news items taken from newspapers in different European languages. The original goal was to see if one could identify anything like the “hot news” use of the perfect, as defined by McCawley (1971). To this end, we excerpted in a number of European languages with perfects or perfect-like categories the first sentence of a set of newspaper articles, characterizable as “news items” in the sense that their main point was to report some recent event. In the majority of the languages investigated (Swedish, Norwegian, American English, British English, Spanish, Finnish, Greek), we found that the majority of all news items were reported in the Simple Past, typically accompanied by a temporal adverb such as yesterday. The following exemplifies what could be seen as the “normal form” of a news item:1 (1)

British English (The Times) C ALCUTTA S TATION B OMBED A bomb exploded in one of Calcutta’s two main railway stations yesterday, killing one man, who was suspected to be carrying the explosives in his luggage, and wounding at least a dozen people standing nearby.

However, we did find a relatively large proportion of perfects, thus falsifying some earlier claims about the virtual non-use of perfects in these contexts (e.g., Inoue 1979). Out of 47 Swedish news items, 17 began with a sentence in the Perfect. Even in American English, there is a significant number of Perfects in news items: in 53 items from The International Herald Tribune, there were 12 Perfects, as in the following example: (2)

American English (International Herald Tribune) M ARILYN H ORNE T O S ING AT C LINTON I NAUGURATION Marilyn Horne, the American mezzo-soprano who has been called “the finest opera singer of her generation,” has been invited by President-elect Bill Clinton to sing at his swearing-in ceremony.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

386

386

Östen Dahl and Eva Hedin

In (2), one factor which conditions the possibility of using the Perfect rather than the Simple Past is that the news item was published before Clinton’s inauguration. There is thus a clear sense in which the invitation was “relevant” at the time. It therefore appears difficult to attribute any decisive role to the “hot news” status as such in determining what news items are presented in the Simple Past and the Perfect, respectively. It may of course still be true, as argued in Schwenter (1994), that the “hot news use” of the Perfect is a valid notion, but being a piece of hot news is clearly not a sufficient condition for being reported in the Perfect. Our attention was therefore directed to another of the traditional concepts used in accounts of the meaning of the perfect, viz. “current relevance". One further pattern that was obvious in the material was the strong tendency for the non-perfect examples to contain time adverbials like on Tuesday. The American material constituted an extreme case here: 40 out of 41 sentences followed this pattern. That the perfect examples did not in general contain any such adverbials was no great surprise, at least for languages such as English, where such combinations are known to be unacceptable. However, the question that arises here is: why do some news items demand a time specification and others do not? We shall argue in this paper that the issue goes beyond constraints on individual tense-aspect markings and has to do with how we refer to events in general.

2. Type-focussing vs. token-focussing Let us start with an example that does not involve event reference, viz. the following sentence pair: (3) a. b.

There are lions in the garden. There is a lion in the garden.

As various people have noted in the literature, (3a–b) are not obviously different in their truth conditions. Consider the question corresponding to (3a): (4)

Are there lions in the garden?

It appears that if there is just one lion in the garden, we would rather answer Yes (maybe most naturally in the version Yes, there is one) than No. Thus, the plural number in (3a) may not be essential to the truth-conditional interpretation of the sentence. So what is the difference? One way of expressing what intuitively seems to distinguish (a) from (b) is that in (a), we say something about the species of lions, more specifically about its occurrence or non-occurrence in the particular region referred to as the garden. The cardinality of the set of lions is irrelevant, however, as long as the set is non-empty. In many languages, of course, there would be no number marking in the translations of (3a), and the irrelevance of the cardinality of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

387

Current relevance and event reference

387

the set will be even clearer. To characterize how (3b) is typically used, on the other hand, we might say that in interpreting that sentence, we build up a “mental model” of the garden including a “discourse referent” characterized as a lion. We may well expect to learn something more about this individual later on in the discourse. In fact, it would be extremely strange to use (3a) as the first sentence of such a discourse. Another interesting point is what happens when we make questions out of the sentences in (3). Whereas the counterpart of (3a), (4), is felt to be rather neutral, the question corresponding to (3b–5) below – is more tied to particular kinds of contexts: the most natural situation for it would be one where we have already observed a lion and express surprise at this fact. (5)

Is there a lion in the garden?

The difference between the two sentences, then, does not lie in what fact they tell us but rather what purpose the presentation of this fact serves. It is not easy to render this distinction in a system of formal semantics. Even in recent theories of discourse semantics like that of Kamp (1981) the two possibilities are conflated, since any existential statement would be seen as introducing discourse referents in the same way. Still, the distinction shows up in natural languages in many ways, and in order to make it easier to talk of it, let us introduce the terms type-focussing and tokenfocussing, for (3a) and (3b), respectively.2 Let us now turn to our area of interest proper, event reference. Let us thus replace the lion in (3) by a past event, such as the one described in (6): (6)

John winked.

Sentences like (6) in English are not specific as to the number of the events involved. Potentially, then, (6) might correspond either to (3a) or to (3b). There is thus no general mechanism in English for making the distinction between type-focussing and token-focussing with respect to events. Still, there are contexts when it may become crucial. Consider the following sentence-pair: (7) a. b.

Has John winked? Did John wink?

Recall the discussion above of the difference between the two questions (4) and (5), and notice that (7a) and (7b) can be understood as being related in an analogous way. That is, (7a) is a way of asking if the event-type ‘John winks’ is instantiated by one or more tokens during a certain period of time, whereas (7b) may be uttered as a surprised reaction upon seeing John wink. In other words, the Present Perfect in English may – in certain contexts, it should be added – be understood as expressing type-focussing, whereas the simple past rather represents token-focussing. Some

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

388

388

Östen Dahl and Eva Hedin

tense-aspect categories not represented in English may be even more clearly specialized for one kind of focussing. (7a), under the interpretation described here, would be said to represent the so-called experiential reading of the English perfect. Some languages have constructions specifically designated for an experiential interpretation, such as the Japanese koto ga aru construction, as exemplified in (8). (8)

Japanese (Inoue 1975) Mike wa, Nihon de hataraita koto ga aru. Mike TOP Japan LOC work:PST fact SUBJ exist:PRS ‘Mike has worked in Japan.’

Experiential constructions are often said to obey a “repeatability” constraint, that is, it is unnatural to use them to express that a unique, non-repeatable event occurred, such as the death of a specific person. This follows if we assume that the basic function of experientials is to state that a certain event-type is instantiated during a period of time, rather than introducing an event as a new discourse referent. Another significant fact is that experientials cross-linguistically seem to occur particularly often in non-assertive contexts, that is questions, negated statements and the like (Dahl 1985: 143). In a similar way, aspect in Slavic may sometimes reflect the type-focussing:tokenfocussing distinction. Cf., e.g., the Russian translations of (7a–b): (9) a.

b.

Russian Ivan morgal? Ivan wink:IPFV:PST ‘Has Ivan winked?’ Ivan morgnul? Ivan wink:PFV:PST ‘Did Ivan wink?’

A further relevant observation about the sentences we are talking about here is that they do not combine with the same kinds of temporal adverbials. Thus, (7a) is naturally expanded to (10a), adding today, whereas this would change the interpretation quite considerably in (7b), where an adverbial like right now, as in (10b), feels more consonant with the intended reading: (10) a. b.

Has John winked today? Did John wink right now?

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

389

Current relevance and event reference

389

Experiential interpretations thus tend to combine with temporal expressions denoting extended periods of time. This is natural given their meaning: one or more occurrences of an event-type is asserted to have taken place. The restriction of the temporal domain to a point in time normally automatically reduces the number of possible events to one, thus favouring a token-focussing perspective. The use of the Present Perfect in English for experiential interpretations is further restricted to temporal domains which lead up to the point of speech. Type-focussing event sentences where the temporal domain is wholly in the past are normally in the Simple Past in English, e.g., (11)

Did John wink yesterday?

The temporal domain may also be retrieved from the context, or defined by pragmatic constraints. Typically, an experiential statement about a person is restricted to that person’s lifetime, giving rise to the much discussed unacceptability of statements about dead persons, formulated in the Present Perfect.

3. The current relevance interpretation and the meaning of the perfect For the time being, we shall not say anything more about type-focussing event reference but shall instead look closer at what we have called token-focussing. We shall thus concentrate on statements which introduce singular events in the past. As pointed out by Lyons (1968), sentences that assert the existence of some object(s) usually contain an indication of, or at least presuppose, a spatial domain. We may possibly say things like There is a God but There is an apple is strange if the context does not indicate a location where the apple is supposed to be. In an analogous way, assertions about past events are felt to be somehow deviant if they are not in any way anchored in time, and sometimes also space. This constraint is a rather subtle one, and extremely context-dependent, which makes it difficult to provide good illustrations of it. Suppose, however, that the following are presented as news items on TV: (12)

A bank has been/was robbed.

(13)

A volcano (has) erupted.

(14)

A bomb (has) exploded.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

390

390

Östen Dahl and Eva Hedin

We would certainly feel that these are not “complete messages”, as long as we are not told when and where the events took place (cf., e.g., (14) to (1) above). On the other hand, if we see the text Bank robbed as a headline or on a newspaper placard we would not react in the same way, but then the communicative function is different: we expect the relevant missing information to be supplied in the newspaper text. We have given (12–14) with the Simple Past and the Perfect as alternatives. It may be felt that in English, the versions with the Present Perfect are a little better than the ones with the Simple Past. We shall argue below that this is due to the existence of an alternative interpretation of the Present Perfect versions that does not, however, fit into the news context that we are interested in here. That the incompleteness effect depends on the function the sentences have in such a context rather than on the choice of tense-aspect form in English is corroborated by the fact that if we look at a language which has nothing that corresponds to the English distinction between Perfect and Simple Past, e.g., Russian, the translations of (12–14) seem equally strange in the news context: (15)

Russian Ograbili bank. (=12) rob:PFV:PST:PL bank

(16)

Russian Proizošlo izverženie vulkana. (=13) take_place:PFV:PST:NT eruption volcano:GEN

(17)

Russian Vzorvalas’ bomba. (=14) explode:PFV:PST:F bomb

Not all sentences used as news items need a temporal specification, however. The following sentences, used as news items, are not felt to be incomplete in the same way as (12–14), although they do not contain any indication of the time of the event: (18)

The Prime Minister has been killed.

(19)

England has declared war on Germany.

Sentences (18) and (19) differ from our previous examples in that they introduce well-defined changes in the states of definite discourse referents. Thus, we know from (18) that the Prime Minister is now dead and from (19) that England and Germany are in a state of war. In this sense, (18–19) are prototypical examples of “resultative perfects”, which are only possible with a subclass of telic verbs, viz. those which have a well-defined result-state as part of their inherent meaning.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

391

Current relevance and event reference

391

In the literature, the notion of “current relevance", which is often invoked in the discussion of the meaning of the perfect, has sometimes been identified with the continuance of the result of a past event into the present. At the same time, it is now generally agreed that a general description of the perfect cannot be based on such a narrow concept, and a number of alternative interpretations of “current relevance” have been proposed3, giving the impression that everyone knows that the perfect implies “current relevance” but nobody knows what that is supposed to mean. In McCoard’s influential book on the perfect (1978), “CR theory” or “current relevance theory” is included among the four major extant theories of the perfect. McCoard, who himself favors the “XN” or “extended now” theory, argues that "current relevance is not a fixed semantic content born by a particular verb form, but is only the name of diverse implications that may attach to sentences, based in part on the appearance of one or another tense form” (1978: 65). Although we would to a large part agree with the content of this quotation, we would still like to argue that McCoard’s account is too simplistic to be adequate, and that the role of current relevance in the interpretation of sentences with or without the perfect has to be much more complex. What we want to suggest here is that a better understanding both of the meaning of tense-aspect categories such as the perfect and of temporal reference in general can be obtained if we see “current relevance” as a graded concept, where the “continuance of a result” criterion is the strongest among a number of possible delimitations. Furthermore, the grammaticalization processes that involve the perfect may at least partly be interpreted in terms of a gradual relaxation of the requirements on current relevance. Thus, we may note that many perfect grams have their origin in resultative constructions, which have the “continuance of the inherent result of a past event” as part of their meaning. Grams like the English Perfect also seem to be connected with such an implication in many contexts, although it is somewhat hard to decide whether this should be seen as an entailment or just a conversational implicature. We shall return to this question shortly. A less restrictive notion of current relevance is called for to explain why (12–14) are indeed acceptable in some contexts. Consider the Present Perfect variety of (12), repeated here as (20). (20)

A bank has been robbed.

Suppose the sheriff of Tombstone, Arizona, is trying to get his deputies to leave the table in the saloon where they are playing cards. In this context, (20) sounds entirely natural. What is going on here? Clearly, the sheriff is not just informing his deputies about a certain interesting event. Rather, he wants them to draw the conclusion that they have to do something about it. If we apply the term “current relevance” here, it does not mean primarily that the direct result of the event is still valid, rather it means

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

392

392

Östen Dahl and Eva Hedin

that the event has repercussions of some kind for the participants of the discourse situation. In contrast to the prototypical cases of resultative perfects above, these repercussions are not directly derivable from the meaning of the verb. In many cases, one has to rely on specific knowledge about the situation or about some convention. Thus, beating a gong does not leave any lasting physical results, but a statement like (21) may be understood to mean that it is time to have dinner, or that a round in a boxing match is over. (21)

The gong has sounded.

To account for (20–21), we need a concept of current relevance which is not only or even primarily a condition on the world, as in the traditional understanding in terms of the “continuance of a result”, but also as a condition on the discourse, in that the speaker portrays the consequences of an event as somehow essential to the point of what he is saying. (This understanding of “current relevance” comes perhaps closest to the definition given of “current relevant state” by Li, Thompson & Thompson (1982), if not in formulation, at least in spirit.) We thus feel entitled to talk of a current relevance interpretation of an event-referring sentence when these conditions are fulfilled. Notice that we may do so irrespective of whether the language in question has a perfect or not, and one of the points we want to make in this paper is that the distinction between current relevance interpretations and other ways of understanding event-referring sentences may well also be relevant in perfect-less languages. It may also be noted that what the current relevance is supposed to imply in a specific context depends at least partly upon the type of verb. It seems that with verbs with an inherent result there is a strong tendency to identify that result with current relevance. Consider, for example a sentence such as: (22)

This soldier has lost his gun.

Losing one’s gun is an event the repercussions of which may continue even after getting it back: a soldier may for instance be punished for his negligence, but it seems that we are reluctant to associate (22) with that kind of interpretation. Notice that this does not mean that (22) cannot be interpreted in what is traditionally called an experiential, that is a type-referring, reading, but only that, if it is given a current relevance reading, there is a strong preference for interpreting it in terms of a continuing result. What we have just said means that the cases where we get the wider interpretation of current relevance are mainly those which involve verbs that do not usually show up in the resultative constructions that are among the historical sources of perfects. It is therefore natural to postulate a connection between the loosening of the conditions

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

393

Current relevance and event reference

393

on what current relevance means and the extension of resultative constructions to a larger group of verbs. Above, we related the notion of current relevance to the role an event plays in the discourse. In this connection, it should be noted that the English Perfect tends to show up in specific discourse patterns, characterized by specific “rhetorical relations”. Sentence (23) exemplifies one of these, where the function of the clause containing the Perfect verb is to give a causal explanation of a state-of-affairs referred to in another clause. Inoue (1979) goes so far as to state as a general claim “that a sentence in the Present Perfect conveys an explanatory sense”. For some of the examples she quotes, one has to stretch the interpretation of “explanatory”, however. Thus, one very typical pattern is that in which a statement in the Perfect is followed by another statement introduced by so, identifying the consequences of the fact related in the first statement. Inoue gives the following example: (23)

Vance has met with President Sadat. So, I think we can expect some new developments to break the deadlock.

Li, Thompson & Thompson (1982) give very similar examples from Chinese, illustrating the use of the sentence-final particle le, which, according to them, expresses a “currently relevant state”: (24)

Chinese (Li, Thompson, and Thompson 1982: 36) wˇo h¯e-le s¯an=b¯ei le! I drink-PFV three-glass CRS ‘I’ve drunk three glasses (so don’t pour me any more/let’s just talk now etc.).’

4. Current relevance and temporal specification What is interesting is that the fact that a sentence takes a current-relevance interpretation seems to influence quite strongly its need for a temporal specification. In a way, this may seem a trivial statement, at least as long as we are talking about English, since the relevant examples have been given in the Present Perfect, which is known to be constrained in its collocability with temporal adverbials. But the connection between temporal specification and the current relevance interpretation goes further than that. Let us consider the case of Russian. Russian has no separate form or construction which would be equivalent to the English Perfect; traditionally, however, the Perfective aspect is said to carry a “perfect meaning” (perfektnoe znaˇcenie) in some contexts. As noted by Tommola (1986), this term is usually used in a rather narrow sense that does not correspond to the range of meanings carried by the perfects of, for example, English and Finnish and indeed

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

394

394

Östen Dahl and Eva Hedin

means roughly what some people have understood by “current relevance”, viz. that the result of the action is valid at the moment of speech. What the conditions are for the “perfect meaning” to arise is usually not made very clear. Tommola (1986: 46) notes that this particular interpretation of the Perfective normally occurs without time adverbials. In fact, the presence or absence of such an adverbial may be crucial for the interpretation. (25)

Russian El’cin priexal v Moskvu. Yeltsin arrive:PFV:PST in Moscow:ACC ‘Yeltsin has arrived in Moscow.’

(26)

Russian El’cin priexal vˇcera v Moskvu. Yeltsin arrive:PFV:PST yesterday in Moscow:ACC ‘Yeltsin arrived yesterday in Moscow.’

(25) would be a normal way of conveying the information that Y is now in Moscow, while (26) would not.4 Why is this? As we saw above, the constraint on spatiotemporal specification holds for Russian as well as for English, meaning that speaking about an event without locating it in time is at least a bit strange. However, it appears that the fact that a sentence is intended to be interpreted as conveying current relevance licenses a breach of the spatio-temporal location constraint. In fact, the current relevance interpretation does not go very well together with temporal indicators at all. It is commonly said that adverbs such as today, this year etc., which can be used to refer to periods that go up to the moment of speech, are possible with the Present Perfect in English. What is often overlooked in this connection is that combining a verb in the Present Perfect with such an adverbial forces a typefocussing perspective, i.e., an experiential reading, and that the combination is not entirely felicitous if such an interpretation is excluded for one reason or another, as when today is added to (18): (27)

The Prime Minister has been killed today.

Similarly, adding segodnja ’today’ to (25) at once makes it more compatible with the possibility that Yeltsin left again: (28)

Russian El’cin priexal segodnja v Moskvu Yeltsin arrive:PFV:PST today in Moscow:ACC ‘Yeltsin arrived in Moscow today.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

395

Current relevance and event reference

395

That the relation between temporal specification and current relevance is at least partly independent of the meaning of tense-aspect categories such as perfects is also illustrated by Greek, which is different both from Russian on the one hand and languages like English and Swedish on the other in that it does have a perfect but with a weaker link to “current relevance”. Thus, in many situations where the use of a perfect is virtually obligatory in other languages, Greek prefers the Aorist. In particular, this is true of the kind of news items we have been talking about here. Consider, for instance, the following English examples, where the first is in the Present Perfect and is understood with the current relevance interpretation, and the second, with the Simple Past, contains a time adverbial and does not carry any current relevance implication: (29)

A man has been arrested for the Palme murder. = he is still under arrest

(30)

A man was arrested yesterday for the Palme murder. = he may have been released later

In Greek, both sentences are naturally translated with the Aorist: (31)

Greek Pjástike énas ándras ja to fóno tu arrest:PFV:PST:PASS a man for the murder DEF:GEN Pálme (=29) Palme

(32)

Greek Pjástike xtes énas ándras ja to fóno arrest:PFV:PST:PASS yesterday a man for the murder tu Pálme (=30) DEF:GEN Palme

In spite of this, (31) implicates that the man is still under arrest; (32) is consistent with him having been released already. Again, we see how a sentence without a time adverbial tends to get the current relevance interpretation. It is far from self-evident why the current relevance interpretation is so reluctantly combined with explicit specifications of time. It appears, though, that a temporal specification somehow detracts from the focussing on the result associated with that interpretation, perhaps by transferring the attention to the time of the past event, which is allowed to take over the function of temporal point of reference. The picture that is emerging from all this, in our view, would be roughly the following. There are a number of distinct ways of talking about past events. The first (type-focussing event reference) typically relates an event type to a temporal domain,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

396

396

Östen Dahl and Eva Hedin

specifying the occurrence or non-occurrence of the type within the domain. The second (token-focussing event reference) establishes an event token as a new discourse referent, normally anchoring it in time and space. The third (current relevance interpretation) presents an event as having “current relevance”. (Here, the distinction between type-focussing and token-focussing reference seems to recede into the background, the current relevance interpretation resembling in certain respects both these types.) These types all differ as to what kinds of situations they allow, what kinds of temporal specification they combine with and what tense-aspect forms can express them in different languages. In the remainder of the paper, we shall consider the relation of current relevance and specific tense-aspect forms in different languages.

5. Current relevance and the Greek Perfect Above we noted that Greek differs from Russian by having a Perfect category and from English and Swedish by a different function of this category. As a matter of fact, the Greek system could be described as a combination of the Russian one on the one hand and the English and Swedish ones on the other, since its Aorist – like the Perfective aspect in Russian – may also be used in sentences intended to be interpreted as conveying current relevance. Greek may thus choose between a Perfective (the Aorist) and a Perfect in such sentences. From this point of view, the link between the Greek Perfect and “current relevance” is weak in the sense that it is not obligatory in these contexts, in the way the Perfect is in many other languages. On the other hand, in the “current relevance” contexts where the Perfect is used, there tends to be a stronger focus on the present state than with the Aorist. (33)

Greek Írthe o thíos. come:PFV:PST DEF uncle ‘Uncle has come.’

(34)

Greek Éxi érthi o thíos. have:PRS come:PFF DEF uncle ‘Uncle has come.’

(33) and (34) are both interpreted as having current relevance, ‘Uncle has come and is here’, but in (33) his ‘having come’ is focused rather than his ‘being here’. The Aorist may be described as focussing on the transition from one state to another. The Perfect often has a wider backward perspective which on one hand often focuses on some development leading up to the transition, on the other hand establishes some distance between the event referred to and the point of utterance (Hedin 1987: 80 ff.).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

397

Current relevance and event reference

397

This second feature makes the Perfect in a sense more stative than the Aorist in the same context. The event referred to is backgrounded, the present state being in focus. So whether the Aorist or the Perfect is used in news items depends to some extent on what is to be described, a recent change in the world or some present state being a consequence of some (not necessarily recent) event. As a concrete illustration of what we have said, consider one of the few examples from our Greek newspaper material where the Perfect was indeed used: (35)

Greek K ALIERJITÉS K E V IOMIXANÍA AGHONÍZONDE NA farmers and factory fight:PRS SUBJ.PRT Z ÍSUN . . . live:PFV:SUBJ KÓRINTHO” “N EROPÓLEMOS S TIN water war in:DEF Corinth “PÓLEMO” éxun aníksi stin Kórintho, megh˘ali war has:PRS open:PFF in Corinth big viomixanía alandikón ke o aghrotikós síloghos tis perioxís. factory meat and the farmer society DEF region ‘FARMERS AND FACTORY ( ARE ) FIGHTING TO SURVIVE . . . “WATER WAR IN CORINTH” A war has been started in Corinth by a big meat factory and the local farmers’ society.’

The event referred to in (35) is not some proclamation of war that took place a few hours ago or so – in such a case, the Aorist would typically have been used, cf. (36)

Greek I Anglía kírikse pólemo (enandíon tis DEF England declare:PFV:PST war (against DEF Jermanías). Germany) ‘England has declared war (on Germany).’

The message in (35) is rather: “There’s a war going on in Corinth for some time now". This is underlined by the Present in the headline. It is the present situation in Corinth that is in focus, not the start of the war, and we expect the discourse to continue with an elaboration of the description of the situation, perhaps as a background to a narrative. Another way of putting this is to say that the statement in the Perfect (together with the place indicator stin Kórintho ‘in Corinth’) introduces a “scene” where later developments may take place. The discourse function of the Greek Perfect in (35) is thus analogous to that of the Pluperfect in many narrative texts, and

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

398

398

Östen Dahl and Eva Hedin

may well be one of the conditioning factors behind the choice of tense/aspect form in this sentence, although it is unclear to what extent this is generalizable. In this kind of news item describing a present state the Perfect may be used rather than the Aorist due to its stronger stative function referred to above. We might interpret the differences between the perfects in Greek on one hand and languages like English on the other as a difference in the delimitation of what is counted as current relevance.

6. Russian “two-way” action imperfective: Current irrelevance? We saw above that the current relevance interpretation is normally expressed by the Perfective aspect in Russian. A sentence such as (25) is normally interpreted as implying that Yeltsin is still in Moscow. On the other hand, as is pointed out in most accounts of Russian aspect, the Imperfective aspect may be used to indicate that the result of the action has been cancelled, as in: (37)

Russian El’cin priezžal v Moskvu. Yeltsin arrive:IPFV:PST in Moscow:ACC ‘Yeltsin paid a visit to Moscow.’

The question is how this kind of example relates to our previous discussion of current relevance. It is somewhat tempting to assume that Perfective sentences, such as (25), and Imperfective examples, such as (37), are totally parallel, in that the former expresses that the result state does hold at the point of reference and the latter that it does not hold. But this would be a simplification of the actual facts. Notice to start with that sentences like (37) can perfectly well be amplified with a time adverbial: (38)

Russian El’cin priezžal v Moskvu vˇcera. Yeltsin arrive:IPFV:PST in Moscow:ACC yesterday ‘Yeltsin paid a visit to Moscow yesterday.’

Furthermore, it is clear in (38) that both the arrival and departure of Yeltsin took place yesterday. In other words, this use of the Imperfective aspect not only implies the cancellation of the result at the moment of speech; it denotes a “two-way action” within the time frame indicated in the sentence. Actually, (38) is also compatible with Yeltsin’s having come back again today. In other words, the state of the world at the moment of speech is strictly speaking irrelevant for (37–38). The “two-way action” interpretation is probably best seen as a conventionalization of an implicature derived from the non-use of the Perfective aspect. Current relevance would thus be involved here only in an indirect way.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

399

Current relevance and event reference

399

7. Conclusion We have argued in this paper that the notion of current relevance has a wider application than has usually been assumed, in that it plays a role not only for the choice between tense-aspect forms such as the Perfect and the Simple Past in English but also for the interpretation of event-referring sentences in general, even in languages whose tense-aspect systems are structured in other ways. With regard to the perfect, we have – without providing an exhaustive account of its semantics – argued for an analysis which is basically compatible with the traditional treatments according to which it is a category with several distinct meanings, although the way in which we distinguish the different interpretations may not be quite the usual one. The disappearance of a current relevance constraint has earlier been argued to be one of the essential parts of the grammaticalization processes that lead from perfects to other gram types (e.g., by Fleischman 1983). As we have argued in this paper, we would like to take this further and see the diminishing significance of current relevance as a gradual process that characterizes those processes both at their initial stage – going from resultatives to perfects – and at later stages – going from perfects to perfectives and pasts. In fact, more stages may well be discernible. A case in point is modern written French. As is well known, what was originally a perfect, Passé composé, has virtually ousted the old perfective, Passé simple, in spoken French, although the latter is alive and well in written language. However, Passé composé is the standard choice for reference to past events in news items and is thus dominant in newspaper text (Wiberg 1995), whereas the corresponding sentences in, for instance, history textbooks are rather in Passé simple. It is tempting to interpret this in terms of a more generous understanding of current relevance, that is, including all kinds of “(hot) news”. Indeed, Schwenter (1994: 995) argues that “hot news uses arise later than other perfect functions, as the perfect construction gradually loses its connection to the present” and that they therefore form a bridge to further stages in the grammaticalization process, such as perfectives and hodiernal pasts. Thus, our investigation comes full circle, finishing on the notion we started with.

Notes 1. In the newspaper quotations, headlines are indicated by italics in this chapter, to make the layout more perspicuous. 2. The distinction between ‘type-focussing’ and ‘token-focussing’ reference introduced here is similar to what other people have labelled the distinction between ‘non-specific’ and ‘specific’ reference. (See, e.g., Lindstedt 1985: 101 ff. for an application to tense and aspect in Bulgarian.). We think that there may be a difference between the two pairs of concepts at least if the latter is understood in terms of what the speaker “has in mind”, but

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

400

400

Östen Dahl and Eva Hedin

it would take us too far to try and resolve this issue. Therefore, to those who prefer the term ‘specificity’, we offer our discussion as a means to make its meaning more precise. (See also Dahl 1988 for a discussion of the notion of specificity). 3. McCoard (1978: 64) lists the following as “properties offered as characteristics of” current relevance: recency present existence of the surface-subject referent; of the deep-subject referent; of a certain state of the subject referent; of a “posthumous personage”; of a belief in the subject referent or in some kind of validity; of the object referent; unspecified “connection with the present” continuance of a state into the present iterativity experientiality present possibility 4. This judgement has been contested by native speakers. We think that a case can still be made for time adverbials influencing the possibility of giving a sentence a current relevance interpretation, and a better example could probably be found, but factors having to do with real-world knowledge and similar things make it difficult to construct clear-cut minimal pairs.

References Dahl, Östen 1985 Tense and Aspect Systems. Oxford: Blackwell. 1988 “The role of deduction rules in semantics”, Journal of Semantics 6: 1–18. Fleischman, Suzanne 1983 “From pragmatics to grammar. Diachronic reflections on complex pasts and futures in Romance”, Lingua 60: 183–214. Hedin, Eva 1987 On the Use of the Perfect and the Pluperfect in Modern Greek. (Acta Universitatis Stockholmiensis: Studia Graeca Stockholmiensia VI.) Stockholm: Almqvist & Wiksell. Inoue, Kyoko 1975 Studies in the perfect. [Ph.D. dissertation. Ann Arbor: University of Michigan.] 1979 “An analysis of the English present perfect”, Linguistics 17: 561–590. Kamp, Hans 1981 “A theory of truth and semantic representation", In Groenendijk, J. A. G. et. al. (eds.). Formal Methods in the Study of Language, Part 1. Amsterdam: Mathematische Centrum, 277–322. Li, Charles N., Sandra A. Thompson & R. McMillan Thompson 1982 “The discourse motivation for the perfect aspect: The Mandarin particle LE”, In Hopper, Paul (ed.), Tense-Aspect: Between Semantics and Pragmatics. Amsterdam: Benjamins, 19–44.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

401

Current relevance and event reference

401

Lindstedt, Jouko 1985 On the Semantics of Tense and Aspect in Bulgarian. Slavica Helsingiensia 4. University of Helsinki, Helsinki. Lundberg, Lars-Johan 1984 “Perfekt? En verbforms grad av tillämplighet i olika kontexter”. [Term paper, University of Stockholm, Dept. of Linguistics.] Lyons, John 1968 Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. McCawley, James D. 1971 “Tense and time reference in English”, in: Charles Fillmore & Terence Langendoen (eds.). Studies in Linguistic Semantics. New York: Holt, Rinehart & Winston, 96–113. McCoard, Robert W. 1978 The English Perfect: Tense-Choice and Pragmatic Inferences. Amsterdam: NorthHolland. Schwenter, Scott A. 1994 “ ‘Hot news’ and the grammaticalization of perfects”, Linguistics 32: 995–1028. Tommola, Hannu 1986 Aspektual’nost’ v finskom i russkom jazykax. Helsinki: Neuvostoliittoinstituutin vuosikirja 28. Wiberg, Lars-Erik 1995 Le passé simple, son emploi dans le discours journalistique. Stockholm: Almqvist & Wiksell.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

402

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

403

Mario Squartini and Pier Marco Bertinetto

The Simple and Compound Past in Romance languages

1. Introduction1 This chapter concerns the distribution and usage of some of the Past tenses to be found in Romance languages. In practice, we limit ourselves to those expressing the notion of aorist and present perfect, while we do not consider the Imperfect and the tenses expressing the notion of past-in-the-past (i.e. the various types of Pluperfect to be found in Romance). A terminological qualification is in order here. The grammatical forms we are going to consider are named differently in the different grammatical traditions (the following list is limited to the main languages for which an established grammatical tradition exists): Cat.: Fr.: It.: Port.:

Rom.: Sp.:

Pretèrit Perfet Pretèrit Perfet Simple Passé Défini Passé Simple Passato Remoto Perfetto Semplice Pretérito Pretérito (Perfeito) Simples Perfeito Simples Aoristul Perfectul Simplu Pretérito (Indefinido) Pretérito (Perfecto) Simple Perfecto Simple

Pretèrit Indefinit Pretèrit Perfet Compost Passé Indéfini Passé Composé Passato Prossimo Perfetto Composto Perfeito Pretérito (Perfeito) Composto Perfeito Composto Perfectul Nedefinit Perfectul Compus (Pretérito) Perfecto, or Antepresente Pretérito (Perfecto) Compuesto Perfecto Compuesto

However, the two Past tenses that are to be found in virtually every Romance variety come, with very marginal exceptions, from the same sources (cf. Section 2 for some diachronic information). Thus, in order to have a unified terminology, we shall speak in most cases of Simple Past and Compound Past (henceforth SP and CP). Occasionally, however, it will be useful, for both practical and theoretical reasons, to use the terms “perfect” and “perfectal” when referring to the CP. This is the inevitable consequence of the linguistic situation. The CP started out as a true perfect, but underwent

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

404

404

Mario Squartini and Pier Marco Bertinetto

a process of gradual aoristicization (i.e. of transformation into a purely perfective past). We shall refer to this process, which covered a greater or lesser distance according to the individual language, as the “aoristic drift”. Accordingly, when we use the term “perfect” in this paper, we do not mean an actual tense, but rather a semantic function, that is, essentially, a gram type in the sense of Dahl (this volume; cf. also Lindstedt, this volume, for a discussion of the semantics of the perfect). The real challenge lies in assessing how much, in each language, the CP has departed from the original perfectal functions.

2. The origin of the Past forms in Romance The SP is in most cases the direct descendant of the Latin Perfect, a tense which, at the stage of the Classical language, had already developed into a general purpose perfective past.2 In the Post-Classical period, a series of compound tenses was formed, of which we find sporadic anticipations in the older texts. It is not entirely clear what the ultimate origin and the chronology of this innovation are. Although most scholars maintain that it stemmed from colloquial usage as an uninterrupted evolution of original Latin constructions, some suggest that it originated in the educated classes through Greek influence (on this topic see at least Pisani 1981, Ramat 1982, Pinkster 1987 and the literature quoted therein). In any case, the creation of these tenses, and in particular of the form which seems to provide the model for the Romance CP, fulfilled the purpose of reintroducing into the paradigm a true perfect. Indeed, the first examples, to be found already in Pre-Classical texts, had a clear resultative meaning: (1) a.

b.

Latin (Plautus, Trin. 347) Multa bona bene parta habemus many goods well obtained have:1PL ‘We possess many well obtained goods.’ (Plautus, Men. 801) Te auratam et vestitam bene habet you bejewelled and dressed well has ‘He keeps you bejewelled and well dressed.’

The distinctive features of this construction are the following: (a) there is no obligatory coincidence between the subject of the inflected verb and the subject of the Perfect Participle (e.g., in (1a) the person who owns the goods needs not be the same person who obtained them); (b) the Perfect Participle has a predicative function, and is a complement of the Object;

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

405

The Simple and Compound Past in Romance languages

405

(c) the inflected verb retains its lexical meaning of possession, i.e. it is not a true auxiliary. The resultative nature of this construction is made evident by the fact that the great majority of the first examples concern telic verbs. In the course of time, however, the construction was extended to the remaining verbs. The final result is a true reanalysis, which encompassed the following major changes: (a¼ ) the coincidence between the subject of the inflected verb and the subject of the Perfect Participle (which obviously constituted the default case, for purely pragmatic reasons) became obligatory; (b¼ ) the Perfect Participle became part of the verb, and manifested a strong inclination to lose the original gender and number agreement with the direct object,3 while the respective order of inflected verb and Perfect Participle became increasingly fixed, with severe restrictions with regard to the type of syntactic constituents allowed to appear in between;4 (c¼ ) finally, the inflected verb lost its lexical meaning and became a true auxiliary.5 These changes might have been facilitated by a number of converging factors, such as the following. First, the structure of the Perfectum in deponent verbs (e.g., profectus sum ‘I left’, lit.: ‘I am left’), or that of the passive Perfectum (e.g., laudatus sum ‘I was praised’, lit.: ‘I am praised’), which consists in both cases of the auxiliary esse preceded by the Perfect Participle, provided the emerging construction with a possible model. Second, it is conceivable that the new construction converged with a periphrastic form, frequently attested in Classical Latin, involving cognition verbs, as in cognitum habeo ‘I (have) learned / I know well’ and compertum habeo (or, equivalently, mihi compertum est) ‘I (have) learned / I know for sure’. Third, the general weakening of case endings might have precipitated to some extent the loss of agreement in the Perfect Participle.6 Finally, the change from SOV to SVO may have stabilized the order AUX  Participle. While these transformations were performed, a whole series of compound tenses was generated by analogy with the Present Perfect resultative. However, at that point the wheel had turned again, in the sense that the purely perfectal value of the Present Perfect had started to show signs of obliteration. But this is precisely the point where a comprehensive story of the Romance CP (and, concomitantly, of the SP) becomes impossible, for its evolution is different in each language. To this topic we revert in the following section. Let us simply observe here that in a few languages (or, more appropriately, in some local varieties of these languages) a series of supercompound forms was created. This happened in some Northern Italian varieties (Piedmontese, Lombard and Veneto vernaculars,7 cf. Rohlfs 1966–69, sect. 673 and Cornu 1953: 236–243), in Romansh, Ladin and Friulian (cf. Cornu 1953: 243–248 and Benincà 1989), in some varieties of Romanian (cf. Paiva Boléo 1936: 74 fn. 1, reporting Iordan’s data on supercompound forms in Northern Moldavia), Occitan (SchliebenLange 1971: 134–154), and most notably in French. In the last language this usage

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

406

406

Mario Squartini and Pier Marco Bertinetto

belongs to some extent to the literary language, where it is admitted by the grammarians only in a temporal clause, indicating anteriority with respect to a CP in the matrix clause (as in Quand on a eu fini nos études nous sommes revenus à St. Etienne ‘When we finished (lit. have had finished) our studies, we came back to St. Etienne’). Most typically the supercompound forms belong to some vernaculars, such as those spoken in Southern France and French-speaking Switzerland (the area influenced by Occitan and Franco-Provençal), where they occur in main clauses too (cf. Foulet 1925; Cornu 1953; Bleton 1982; Carruthers 1992, 1993). To complete the picture, note that in some Romance languages (most notably those of the Iberian area and some Southern Italian vernaculars) another participial construction developed, where the auxiliary used is the descendant of Latin tenere ‘to keep’ instead of habere/esse ‘to have/to be’. Such a form, which is sporadically attested in Latin as well (Pinkster 1987: 214–215), evolved into CP in Portuguese and Galician, gradually substituting for the form construed with haver. In Spanish the opposite distribution occurs, for the haber construction is the normal CP and tener  Past Participle has a more restricted usage. Although we consider all these constructions as different manifestations of Romance CP, there are specific properties that will be dealt with in the following sections.

3. Stages of development We find it useful to refer here to Harris (1982), which represents a well-known and influential point of reference and will provide the point of departure of our discussion. In summarizing the distribution of the CP in Romance (the Present Perfect, in his terminology), Harris (1982) proposes the following synchronic patterns, corresponding to different ways of conceiving the opposition SP / CP (the languages indicated within parentheses in the following points are Harris’s suggestions):8 S TAGE I: the CP is “restricted to present states resulting from past actions, and is not used to describe past actions themselves, however recent” (some Southern Italian vernaculars) S TAGE II: the CP occurs “only in highly specific circumstances” such as contexts “aspectually marked as durative or repetitive” parallel to English I have lived here / been living here all my life; I have often seen him at the theatre (Galician and Portuguese, many varieties of American Spanish) S TAGE III: the CP expresses “the archetypal present perfect value of past action with present relevance” (Castilian Spanish; some varieties of langue d’oil and langue d’oc) S TAGE IV: the CP also expresses the preterital or aoristic functions, while the SP is restricted to “formal registers” (Standard French, Northern Italian, Standard Romanian)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

407

The Simple and Compound Past in Romance languages

407

In what follows, we shall reconsider in greater detail the situation in several Romance languages. In Section 4, we shall point out the difficulties of interpreting these four stages as diachronic steps of grammaticalization.

3.1. Stage I Sicilian and Calabrian are presented by Harris as typical examples of a low degree of grammaticalization, admitting the CP only for current states, possibly connected to past situations. As to Sicilian more data will be presented in Section 3.2, showing that the CP is rather at stage II than stage I. As to Calabrian, in Harris’s source (Rohlfs 1966–69 Section 673) there is at least one example (aju jutu ‘I have gone (there) = I know the place’, Southern Calabrian dialect in the province of Catanzaro) that should be interpreted rather as experiential perfect, i.e. as an instance of stage II (or perhaps more appropriately stage III). This means that, although a past perfective situation is normally expressed by a SP, even with regard to recent events (cfr. Rohlfs u fici ora ‘I did it now’, scil. ‘a moment ago’), the CP in these varieties has already moved forward in the alleged process of grammaticalization. It is not restricted to current state situations, i.e. it is not a truly resultative perfect, but can also denote past situations with current experiential relevance. Clearly, more research needs to be done on this issue. Nevertheless, as far as we can see, no contemporary Romance language exhibits a CP with purely resultative value. Even Spanish tener + Past Participle, which coexists with the most productive CP construction, built by means of the verb haber + Past Participle, seems to be in a further stage of grammaticalization, as shown by Harre (1991). Its usage is not restricted to durative states, like in: Este chico tiene preocupada a su madre ‘The mother of this boy is worried because of him’ (lit.: ‘This boy has worried his mother’), or to durative states resulting from past events: Tengo pedido el libro ‘The book is now requested’ (lit.: ‘I-have requested the book’), because some speakers also accept iterative past contexts, such as: Me tiene dicho repetidas veces que no piensa casarse con él ‘She told me several times that she is not considering marrying him’ (lit.: ‘She-has told me ’), or even past punctual situations: Tengo oído que mañana no va a haber clase ‘I heard that there is no class tomorrow’ (lit.: ‘I-have heard.’). Admittedly, the usage of this construction has not yet become fully productive, and there is a great deal of variation as to extending it to other iterative contexts, such as for instance: Tengo perdida la cartera varias veces ‘I (have) lost the wallet many times’ (lit.: ‘I-have lost:OBJ the wallet several times’).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

408

408

Mario Squartini and Pier Marco Bertinetto

3.2. Stage II Stage II is represented by Portuguese, Galician and some varieties of American Spanish, where specific requirements must be fulfilled in order to use the CP. In fact, all the available descriptions of the CP in Portuguese (construed with auxiliary ter Lat. tenere ‘to keep’)9 agree in pointing out that the semantics of this form differs from the corresponding CPs in the remaining Romance languages, excepting the other varieties quoted above. The major requirement is that the CP should refer to a durative or iterative situation, starting in the past and continuing up to the Speech Time. This implements the so-called inclusive meaning of the perfect, in which the event is seen as still ongoing at the Reference Time (obviously coinciding with the Speech Time, in the case of the Present Perfect), while nothing is presupposed regarding what follows it (cf. Eng. (until now) I have worked (or: have been working) hard). In fact, the Portuguese CP has often been defined as an imperfective form, or as a perfective form with imperfective features (cf. Irmen 1966; Sten 1973 and Suter 1984); and this makes sense given the above characterization.10 The obvious consequence of this is that activities and (to some extent) contingent states are grammatical in these contexts (2), while non-durative (3) and telic situations (4) are rejected, unless they appear in iterative contexts (5–6), where they can be visualized as spanning a temporal interval including the Reference Time:11 (2)

Portuguese Tenho estudado imenso desde que decidi fazer have:1SG studied enormous since that decide:SP:1SG do:INF o exame the examination ‘I have been studying a lot since I decided to take the examination.’

(3)

Portuguese *O João tem chegado agora the John has arrived now ‘John has just arrived.’

(4)

Portuguese *Ultimamente o João tem lido um romance de Eça de recently the John has read a novel by Eça de Queiroz Queiroz ‘Recently John has read a novel by Eça de Queiroz.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

409

The Simple and Compound Past in Romance languages

(5)

Portuguese Nos últimos dias o João tem chegado tarde in_the last days the John has arrived late ‘In the last few days John arrived late.’

(6)

Portuguese Ultimamente o João tem lido muitos romances Recently the John has read many novels ‘Recently John has read many novels.’

409

Note that in Portuguese the CP cannot refer to truly past situations, not even when these are located in the recent past (3), or interpreted as experiential (7),12 as hot news (8), or as triggering a Reference Time reading of the Speech Time (9–10), or in hodiernal contexts (11): (7)

Portuguese (PFQ: 32) *Já tens estado em Austrália? already have:2SG been in Australia ‘Have you already been to Australia?’

(8)

Portuguese (PFQ: 56) *Tem chegado o rei! has arrived the king ‘The king has arrived!’

(9)

Portuguese (PFQ: 27) *Não, já se tem ido embora no already RFL has left ‘No, s/he has already left.’

(10)

Portuguese (PFQ: 28) *Não, ainda não tem voltado no yet not has returned ‘No, s/he has not come back yet.’

(11)

Portuguese (PFQ: 16) *Tenho acordado às quatro da manhã have:1SG woken up at_the four of morning [A question asked at 9 a.m.: Why do you look so tired?] ‘I woke up at four in the morning.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

410

410

Mario Squartini and Pier Marco Bertinetto

These data entail that notions such as current relevance (stage III) and anteriority cannot be at work here. Rather, the Portuguese form seems to be sensitive to the actional character of the situation. It selects only durative and non-telic situations encompassing the Reference Time, while non-durative and telic situations are only accepted if interpreted as spanning an interval that reaches up to the Reference Time (possibly by way of iteration). Note that this set of restrictions does not hold for the other compound forms, such as the Pluperfect, Compound Future, Compound Conditional, or Subjunctive CP. This state of affairs is clearly due to the competition between CP and SP, rather than being an intrinsic character of Portuguese compound forms in general. It is worth adding that there are contexts that, while not fulfilling the requirement of durativity or iterativity, do admit the CP. These are for instance stylistic usages restricted to given pragmatic or textual environments (such as the formulae tenho acabado ‘I have concluded’ or tenho dito ‘I have said’, uttered at the end of a formal speech), or modal extensions towards irrealis (Se tens continuado no Seminário, campavas ‘If you had kept on in the seminary, you would have been successful’ (lit.: ‘If you_have kept on in the seminary, you_be_successful:IMPF’, from Suter 1984: 84), or special usages relating to specific syntactic patterns, such as a relative clause o Júlio Difollowing a superlative (um dos maiores escritores que tenho lido é 13 Júlio Dinis’, from Suter 1984: nis ‘one of the greatest writers I have read is 174). The last example shows that there is some leakage towards the experiential function, although this is not generally the case. Moreover the CP may be used in the original resultative or current state contexts with the past participle agreeing with the direct object. Some of these cases can also be interpreted as referring to a past situation with current relevance, as in A mãe tem tudo preparado para irmos viver no andar de cima ‘our mother has prepared everything, so that we could move upstairs’ ’, Sten 1973: 234). (lit.: ‘the mother has everything prepared The distribution of the CP in Galician seems to be partially similar to Portuguese. A notable difference, though, is that in Galician all compound tenses (not only the CP) are quite rare, and most of their modern usages are due to Spanish or Portuguese influence. The CP is documented in resultative contexts or contexts denoting inclusive iterative situations (cf. Paiva Boléo 1936: 12–15; Santamarina 1974: 159–161; Rojo 1974: 128–132). As noted by Harris (1982 fn. 4), no example of durative (noniterative) context, similar to the Portuguese examples described above, is attested in the literature. Paiva Boléo (1936: 16–19) notes that also in Asturian and Leonese the usage of CP is very restricted, and this is confirmed by Cano González (1992: 666–667) for Asturian14 and by Millán Urdiales (1966: 174–175, quoted by Harris 1982: 53) for Leonese. In some varieties of American Spanish, the CP shows a distribution similar to the Portuguese form. For instance, the Mexican CP designates durative and iterative situations encompassing the Speech Time (cf. Lope Blanch 1961; Said 1976; Moreno

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

411

The Simple and Compound Past in Romance languages

411

de Alba 1978; Spitzová & Bayerová 1987). The following examples, both from Lope Blanch (1961 [1983]: 135), show typical durative contexts: (12)

Mexican Spanish Desde entonces sólo he sido una carga para ti since then only have:1SG been a burden to you:DAT ‘Since then I have only been a burden to you.’

(13)

Mexican Spanish Pero ¿Cómo? ¿Tú con lentes? – Pues claro; yo siempre but how you with glasses well of-course I always los he usado they:OBJ have:1SG used ‘What? You wear glasses? – Yes, of course; I have always used them.’

As an example of iterativity, Lope Blanch (1961 [1983]: 136) quotes the following distinctive pair: (14)

Mexican Spanish Eso ya lo discutimos ayer this already it:OBJ discuss:SP:1PL yesterday ‘We already discussed this matter yesterday.’

(15)

Mexican Spanish Eso ya lo hemos discutido muchas veces this already it:OBJ have:1PL discussed many times ‘We have discussed this matter many times.’

This contrast shows that with a semelfactive situation the SP is preferred. However, just as in Portuguese, when the same situation is iterated over a span of time encompassing the Speech Time, the CP is used. Lope Blanch (1961 [1983]: 137) also notes the opposite distribution of CP and SP with the adverbials ya ‘already’ and todavía no ‘not yet’, both referring to a Reference Time coinciding with the Speech Time. When todavía occurs in a negative context, the CP is used, while ya always requires the SP: (16)

Mexican Spanish Todavía no ha llegado yet not has come:PP ‘S/he has not come yet.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

412

412 (17)

Mario Squartini and Pier Marco Bertinetto

Mexican Spanish Sí, ya llegó yes already come:SP ‘Yes, s/he has already come.’

The contrast here is between an inclusivity-oriented situation (16), and a recent past with current relevance (17). The opposite distribution would be ungrammatical (Lope Blanch 1961 [1983]: 137 fn. 15): (18)

Mexican Spanish *Todavía no llegó yet not come:SP ‘S/he has not come yet.’

(19)

Mexican Spanish *Ya ha llegado already has come:PP ‘S/he has already come.’

Note that this opposition does not occur in Portuguese, since the SP would be used in both cases, although (16) is somehow oriented towards the inclusive meaning, in the sense that the non-occurrence of the event lasts at least up to the Reference Time. Thus, this Mexican peculiarity can be understood as an actional requirement: in negative contexts, the verb undergoes (in cases like (16)) an actional reclassification, i.e. it is turned into a durative predicate. However, just as in Portuguese, a recent past context does not trigger the CP. This is documented not only by (17), but also by ¿Qué dijiste? ¡Repítelo, si te atreves! ‘What did you say? Repeat it, if you dare’, from Lope Blanch 1961 [1983]: 137.15 The picture proposed by Lope Blanch is basically confirmed by Berschin (1976: 40–42), Said (1976), Moreno de Alba (1978) and Spitzová & Bayerová (1987), even though in these more recent works the CP shows a tendency to move forward in the process of grammaticalization, admitting also non-durative and non-iterative contexts referring to purely perfective situations. Some of these cases are also discussed in Lope Blanch (1961 [1983]: 142) and labelled as marked affective utterances16 (Pasó un carro rozándolo ¡Qué salto ha dado! ‘A truck/car passed by grazing What a jump he made!’ (lit.: has made); but in other cases, quoted by him Moreno de Alba 1978: 60, there is no affective meaning (Tú sabes que hace poco ‘You know that recently a palace was discovered han descubierto un palacio que that ’, lit.: ‘ they-have discovered.’). According to Moreno de Alba, these cases are extremely rare in the corpus analyzed (4.4% of the occurrences of the CP), and mostly limited to formal speech influenced by literary style. As for the other varieties of American Spanish, we only have some descriptions concerning the aspectual value of the CP. Rallides (1971: 24–31) and Berschin

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

413

The Simple and Compound Past in Romance languages

413

(1976) for Colombian, Cardona (1979) for Puerto Rico, Catalán (1966: 492–494) for Canarian and Kubarth (1992) for Buenos Aires Spanish all describe a distribution parallel to the Mexican one, where the CP denotes durative or iterative situations encompassing the Speech Time. Actually, the data quoted by Cardona for Puerto Rico shows that there seems to be some extension towards purely perfective contexts, as noted above for Mexican Spanish, while in Canarian some cases of experientiality are documented (Yo he ido a la escuela = he recibido enseñanza escolar ‘I have gone to school = I am an educated person’, from Catalán 1966: 493).17 Berschin (1976: 35–37) and Westmoreland (1988: 379–380) cite all the information available on the distribution of the CP in the other countries of Central and South America. In most of these areas, the SP seems to prevail over the CP, thus confirming the generalization given by Kany (1951 [1975]: 161–164) that the “the simple preterite [ ] is frequently used in most of Spanish America in cases where a purist insists on the Present Perfect [=CP]”, but no information is provided whether the usage of the CP is simply less frequent than in European Spanish, or whether it is semantically constrained as in Mexico. Nevertheless the distribution is not at all homogeneous at both the diaphasic and the geographic dimension, since all over Spanish America the CP occurs more frequently in formal style, under the influence of the peninsular norm. Moreover there are some areas (mostly in Peru, Bolivia, Ecuador, and Northern Argentina) in which the CP is frequently also used in the informal style, as noted by Kany and confirmed by more recent studies.18 Quite interestingly, there is another Romance variety, Sicilian vernacular, which is reported as showing the same actional restrictions as Mexican Spanish. Skubic (1973–1974, 1974–1975) notes that in Sicilian the CP does not express recent past or current relevance, but rather durative or iterative situations encompassing the Speech Time. See for instance the following examples (from Skubic 1973–74: 231) where an iterative situation expressed by the CP (20) is contrasted with a recent semelfactive situation expressed by the SP (21): (20)

Sicilian Aju manciatu tanti voti u piscispata, e m’ have:1SG eaten many times the sword-fish and me:DAT ha fattu sempri beni has done always good ‘I have eaten swordfish many times, and it has always done me well.’

(21)

Sicilian M’ u manciài oj e mi fici mali me:DAT it:OBJ eat:SP:1SG today and me:DAT do:SP bad ‘I ate it today and it made me sick.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

414

414

Mario Squartini and Pier Marco Bertinetto

Examples (from Mòcciaro 1978: 345–346) with durative activities encompassing the Speech Time (l’amu circatu tutta a matinata ‘we have been looking for him the whole morning’, lit.: ‘ have looked ’) or negative situations (aguannu un a chiuvutu ‘This year it has not rained’) are also reported, showing similarity to the Mexican distribution. In Section 4 below we shall discuss the problem of the diachronic interpretation of stage II regarding the evolution of Romance CPs.

3.3. Stage III Stage III represents an important turning point in the evolution of the CP as described by Harris, since at this point the CP extends its coverage to purely perfective situations (in Section 1 above we called this process “aoristic drift”). In characterizing this stage, Harris uses the widespread notion of current relevance. However, since the different languages indicated by Harris as belonging to stage III vary in their distribution of the CP, the notion of current relevance must be interpreted as a subjective notion, expressing some kind of psychological feeling of the speaker for what is currently relevant. In fact, Klein (1992) pointed out the unfalsifiable nature of this notion.19 Accordingly, Schwenter (1994a) proposed to interpret it from the point of view of the aoristic drift (although he does not use this term), whereby the different distribution of the CP in the various Romance languages should not to be accounted for as a difference in the conceptualization of the notion of current relevance, but as a varying degree of grammaticalization of this tense as a purely perfective past. According to this view, if an Englishman, unlike a Spaniard, does not allow a temporal location adverbial (e.g., today at three o’clock) to co-occur with a CP, this does not mean that these speakers have a different conception about what is currently relevant and what is not; it only means that the Spanish CP has reached a further stage in the aoristic drift. As a matter of fact, the European Spanish CP20 seems to be rather advanced in this process of transformation. It occurs not only in inclusive contexts (22), as in Portuguese and Mexican, but it is compatible with other typical perfectal contexts, such as hot news (23) and experiential (24), or contexts indicating anteriority with respect to a Reference Time (25–26) or persisting result of a past situation (27). Moreover, being sensitive to the temporal distance from the Speech Time, it also admits temporal adverbials of recent past (28): (22)

Spanish (PFQ: 50) He vivido aquí toda mi vida have:1SG lived here all my life ‘I have lived here all my life.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

415

The Simple and Compound Past in Romance languages

(23)

Spanish (PFQ: 56) Ha llegado el rey! has arrived the king ‘The king has arrived!’

(24)

Spanish (PFQ: 32) ¿Has estado en Australia? have:2SG been in Australia ‘Have you been to Australia?’

(25)

Spanish (PFQ: 27) No, ha salido ya no has left already ‘No, s/he has already left.’

(26)

Spanish (PFQ: 28) No, todavía no ha llegado no, yet not has come back ‘No, s/he has not come back yet.’

(27)

Spanish (PFQ: 03) No, ha muerto no has died [Is the king still alive?] ‘No, he died.’

(28)

Spanish (PFQ: 16) Hoy me he despertado a las cuatro de la today RFL:1SG have:1SG woken up at the four of the madrugada morning [A question asked at 9 A.M.: Why do you look so tired?] ‘Today I woke up at four o’clock in the morning.’

415

As to examples such as (25), note that the usage of the CP is not at all obligatory in peninsular Spanish, where it undergoes stylistic and geographical variation. In the E UROTYP Questionnaire the CP has been used in (25), but in an emphatic context such as the following, in which the speaker shows his/her surprise, a SP would be preferred even though, from a purely semantic point of view, the context is the same as (25): (29)

Spanish (PFQ: 31) ¡Oh, no! Ya se despertó oh no already RFL wake_up:SP ‘Oh, no! She has already woken up!’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

416

416

Mario Squartini and Pier Marco Bertinetto

Moreover, Skubic (1964: 89) provides examples of non-elicited conversational Spanish, in which the SP occurs with the adverbial ‘already’ without any sort of emphatic effect: (30)

Spanish (Granada) Profesor, [¿]ya entregó su ponencia? professor, already hand_in:SP your report? ‘Professor X, have you already handed in your report?’

Interestingly, Skubic (1964: 88) notes that in the negative counterpart of these contexts, containing not yet instead of already, the usage of the CP is more systematic (cf. No ha llegado todavía ‘S/he has not come yet’, Córdoba Spanish). Therefore, this usage, attested in Mexican Spanish, seems to be the most well-rooted also in European Spanish. As to the sensitivity of CP to the temporal distance from the Speech Time, Schwenter (1994a) shows that in Alicante Spanish the CP, in its purely perfective function, tends to respect the hodiernal vs. prehodiernal distinction. In fact, with hodiernal adverbials (such as esta tarde, hoy, hace una hora ‘tonight, today, an hour ago’ vs. el otro día, el lunes, hace una semana ‘the other day, on Monday, a week ago’) the CP is much more frequent than the SP.21 These recently elicited data on Alicante Spanish confirm the picture depicted by Berschin (1976: 76–80) for Madrid students.22 This author also notes that the percent of CPs increases with hodiernal adverbials, even though the percent of SPs with hodiernal adverbials is higher than in Schwenter’s data. The on-going character of the aoristic drift is manifested in Alicante Spanish by the difference between the younger and older generation, since the latter shows a higher percent of SP in hodiernal contexts as compared to the former. Moreover, the CP seems also to be extending to prehodiernal contexts and again in this respect the younger generation has moved further than the older. In fact, the extension of CP to prehodiernal contexts seems to be a general phenomenon in Spanish, for Kuttert (1982: 196–197) quotes several examples of CP with prehodiernal adverbials in written texts, and the same tendency is documented by Serrano (1994: 48) in oral Madrid texts. Schwenter also presents elicited and spontaneous data showing that the hodiernal/prehodiernal distinction is maintained even in narrative contexts. But in this respect there seems to be some geographic variation, since a Seville speaker in the E UROTYP Questionnaire uses the SP also for hodiernal narratives, while respecting the hodiernal/prehodiernal distinction in non-narrative contexts.23 The opposite tendency is also described (Serrano 1994: 47–53): Madrid speakers extend the CP to short distance prehodiernal narratives (as with the adverbial yesterday), restricting the SP to remote situations (e.g., two months ago). As to Occitan and Catalan (cf. Schlieben-Lange 1971: 127–132 and the literature quoted therein), they are at that point of stage III where SP (cf. fn. 2) and CP coexist

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

417

The Simple and Compound Past in Romance languages

417

with different functions (specifically, in terms of temporal distance), the CP being mostly used for situations closer to the Speech Time. According to Badia Margarit (1962 [1985], 1: 423), the distribution of CP and SP in Catalan is quite similar to that in Spanish, the only difference being that Catalan seems to be more sensitive to the hodiernal/prehodiernal distinction. This observation is borne out by the result of the E UROTYP Questionnaire, in which the hodiernal/prehodiernal distinction is respected also in narrative contexts by the Catalan native speaker in contrast to the Seville speaker mentioned above.24

3.4. Stage IV At this point of the aoristic drift, represented by Standard French, Standard Romanian, Northern Italian varieties, Romansh, Ladin, Friulian,25 and Sardinian,26 the CP can be used in any kind of purely perfective contexts and in some cases it is the only existing form. As opposed to stage III, a notion such as temporal distance is no longer at work, in the languages (and varieties) that obey the typical situation of stage IV. The diachronic data show that the French CP was previously sensitive to temporal distance, since the SP was first ousted in contexts referring to situations close to the Speech Time (Foulet 1920: 291–296). But apart from the actual relationship between stage III and IV, which will be discussed below, the list of the languages belonging to stage IV requires more accurate distinctions. First, a distinction should be made between the local vernaculars27 and the local varieties of the standard language. This is certainly relevant regarding Northern Italian, where the vernaculars have in most cases entirely lost the SP, even at the level of morphological possibility, whereas the local varieties of the standard language still present this form, at least to some extent (to this we revert in Section 5 below). Second, a distinction has to be made between colloquial language and written texts. In Standard French, the SP has disappeared from colloquial conversation, but it is reported as relatively frequent in newspapers (Zezula 1969, Herzog 1981, Engel 1990). Moreover, the SP is used is formal style and typically in literary texts (cf. among others Stavinohová 1978: 33–73), where it is traditionally considered to fullfil the function of propulsive tense, which advances the plot by situating the events in the narrative loom relative to one another. According to Waugh & Monville-Burston (1986) such a foregrounding function cannot be extended to newspaper usage, where the SP has rather a contrastive function, demarcating formal and logical articulations of the text, emphasizing special points, even particular details independently of foregrounding. The French pattern involving textual/stylistic restrictions for the SP has some similarities to Standard Italian (cf. Section 5) and Romanian, but it cannot be general-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

418

418

Mario Squartini and Pier Marco Bertinetto

ized as such. As for Standard Romanian, the restrictions on the usage of the SP are stronger than in French and the evolution of the CP even more advanced in the aoristic drift. The SP seems to have been virtually lost in the colloquial language, and its usage is also quite restricted in written cultivated texts, literary style and in narratives. In literary texts the SP is still admitted (cf. Canarache 1965, S˘adeanu 1972), but it is much rarer than in the literary register of the other Romance languages. Support for this conclusion comes from Savi´c (1990), who observes that in the translation of the same narrative passage, Romanian presents only CPs, whereas French and Italian show both CPs and SPs, and Portuguese has only SPs. Also C˘al˘ara¸su (1992), analyzing a contemporary Romanian (epistolary) novel, points out that the SP is completely absent, while it is used in its French translation. As for newspaper texts, Savi´c (1979: 189) finds that in his corpus the SP is completely missing. However, it has to be noted that there are dialectal varieties of Standard French and Romanian where the SP is still quite common. This is particularly prominent in Romanian (Siadbei ¸ 1930), where the SP is still currently used in some dialects of Daco-Romanian (especially in the Southwest, as in Oltenia and Southwest Muntenia, cf. Georgescu 1958, Moise 1977), in Arumanian28 and Megleno-Romanian, while in Istro-Romanian it tends to disappear as in standard Daco-Romanian. Even in France the situation is more varied than is often assumed: apart from the Occitan and Franco-Provençal areas, where the SP is more resistent, this tense is also recorded as still currently used (at least until the late 19th-early 20th century, when some of the data were collected) in the Northwest and Northeast (basically Normandy and Wallonia), and in some other residual areas (cf. Cornu 1953: 200–201, Harris 1982: 56–59 and the literature quoted therein). A special case should be made for some Romanian dialects, such as those spoken in Oltenia. Here, the distribution appears strikingly different from that of Standard Romanian or any other Romance language. These varieties are sensitive to temporal distance, but the form expressing proximity to the Speech Time is not the CP, like in other Romance languages, but the SP. The latter tense is used for more recent situations, mostly located in the same day regarding the Speech Time, while the CP refers to more distant situations. According to Pan˘a Boroianu (1982) such a specialization of the SP to indicate the most recent past, in particular in hodiernal contexts, is a recent phenomenon that is the product of a gradual evolution which can be traced back in local texts. In addition to the hodiernal requirement, the SP is reported to be used only with non-durative situations or with situations that ManoliuManea (1989: 108–109) labels as “limitées” (presumably, telic). This author notes that, with a predicate such as to drink, the SP is only possible when the verb is followed by a direct object, thus allowing a telic interpretation: b˘au un pahar cu ap˘a ‘he drank a glass of water’ vs. *b˘au ‘he drank’.29 Clearly, these data cannot be easily located in any of the four stages described above, although they also represent a peculiar form of aoristic drift. It seems reasonable to invoke here interference from

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

419

The Simple and Compound Past in Romance languages

419

the Greek Aorist (as suggested by Manoliu-Manea 1989: 109) or Serbian influence (cf. Lindstedt, p.c.).

4. The diachronic puzzle 4.1. On the proper interpretation of stage II At this point, it is worth discussing the diachronic interpretation of the four stages proposed by Harris. This author conceives of these four steps as representing the pattern of evolution of the Romance CP, from a resultative value to a purely perfective reading. Stage II is thus interpreted as an intermediate stage between resultativity and pure aoristicity. It refers to a durative or iterative situation, expanding (so to say) a present state into the past (or, more appropriately, stating the persistence of a past situation up to the Speech Time). This interpretation, even if appealing, is problematic for a number of reasons. First, diachronic data on 17th century Portuguese CP seem to show that instead of being actionally restricted, as it is nowadays, it was also used in non-durative past contexts. Several authors (Paiva Boléo 1936: 34–35; Irmen 1966; Suter 1984: 54–58; Harre 1991: 144) quote examples from 17th century texts (mostly Bible translations) where the CP has a non-durative meaning, a usage unknown in contemporary Portuguese. Unfortunately, it is not clear whether these examples are due to the cultivated influence of other languages (such as Spanish or French) on written texts, as maintained by Paiva Boléo (1936: 27), or genuinely mirror the situation of the spoken language of that time, as suggested by Irmen. The problem is open to further research.30 Second, Harris’s sequence presupposes the generalization of stage II to those Romance languages that do not show any attested past evidence of this stage in the usage of the CP. Whether stage II also occurred in French or Italian is an empirical issue that has not yet been documented.31 But what is problematic is the very conception of stage II as an intermediate step. Although the development suggested by Harris is plausible, an alternative interpretation is equally possible. According to this, stage II, corresponding to the inclusive meaning of the CP, would not be the second step of the aoristic drift, but rather a totally independent development in which actional values, or rather a peculiar interaction of actional and aspectual values, are foregrounded. As to actional restrictions, recall our discussion in Section 3.2. As to aspectual properties, what characterizes the inclusive meaning is the blending of perfective and imperfective values. It is perfective inasmuch as it implies a Reference Time (obviously simultaneous with the Speech Time), which is the distinctive feature of perfectal tenses, a subspecies of perfective tenses. It is however imperfective, inasmuch as the event is not necessarily terminated at the Reference Time, as

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

420

420

Mario Squartini and Pier Marco Bertinetto

is typical of the imperfective aspect in all of its manifestations. Thus, the inclusive meaning may be considered an aspectual hybrid (as suggested by Bertinetto 1986). Obviously, in the spirit of Harris’s proposal, one might maintain that stage II is subsumed in the following stages, rather than completely superseded by them. Indeed, the inclusive meaning corresponding to stage II is a semantic value inherently incompatible with the aoristic aspect and often attached to the perfect; quite significantly, it is still present in possibly every Romance language, where it is precisely the CP (or maybe the Present, but by no means the SP) which is used for this purpose. If this is the case, one might contend that there is a sort of aspectual escalation from the basically stative meaning of stage I (pure resultativity, with no sharp perfective orientation) to the hybrid status of stage II (inclusivity) to the decidedly perfective nature of stages III and IV.32 Nonetheless, we would venture to suggest that Portuguese might plausibly attest a situation which points towards a radically diverging orientation, as compared to the remaining Romance languages. Namely, Portuguese might have privileged the SP rather than the CP, just as Northern German has in contrast to Southern German. In other words, the Portuguese situation might simply exhibit one possible outcome of the frequent conflict arising between past tenses competing for the same (or for a too similar) semantic territory. Recall, in fact, that the restriction to the inclusive meaning does not concern the remaining Portuguese compound tenses, which have no direct competitors. If this is so, then the logic of the distribution of SP and CP in Portuguese would be totally alien to the tendency which characterizes, as a whole, the rest of the Romance area, should the notion of aoristic drift be taken in its obvious sense, i.e. as the gradual extension of the CP towards purely perfective values at the expense of the SP. Note, however, that this conclusion does not exclude, on strictly logical grounds, the possibility that stage II, as claimed by Harris, be an actual sequential step. We have no strong evidence to prove or disprove either hypothesis. Thus, what the above discussion suggests is simply that stage II might not belong to the same line of development as stages III–IV. Obviously, this matter cannot be settled on the ground of mere speculation. We need detailed data from the ancient stages of Romance languages to prove any of the above hypotheses. Unfortunately, the relevant input may no longer be available, given the relatively late emergence of vulgar scripts.33

4.2. On the proper relationship of stages III and IV Another problem that we have to consider is the exact distinction between stages III and IV. According to Harris, stage III implies some residue of the original perfectal meaning, as is apparent in the notion of current relevance. However, it is hard to assess what exactly this might mean in the case of Romance languages. For in-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

421

The Simple and Compound Past in Romance languages

421

stance, no Romance language of stages III and IV (thus excluding Portuguese and the Sicilian vernaculars) presents the restriction that is to be observed in English and other languages, whereby a definite temporal localization of the event cannot co-occur with the CP. Thus, in any such language one may say the equivalent of ‘I have left at 5 o’clock’, which is ungrammatical in English. This depletes the ultimate meaning of the notion current relevance, for the event, rather than being considered in its present consequences, is simply taken in and by itself (i.e. as an event localized at some point preceding the Speech Time). Indeed, as Klein (1992) convincingly argues, the notion of a link connecting a past event to the present is admittedly too vague to make real sense, unless there are testable consequences as in “persistent result” situations (like, for instance, the creation of a previously non-existing object or a change of state in a given object). We prefer therefore to view stages III and IV as a single continuum, in which the various languages are disposed scalarly, from a minimum to a maximum in terms of proximity to the purely aoristic value. The extreme is reached by those Romance varieties, such as various Northern Italian and French vernaculars, where the SP does not even exist as a remote morphological possibility. In these varieties, the CP has gone all the way through to becoming a general purpose perfective tense. In all other cases, including Standard Romanian, there are residual stylistic areas which are still reserved to the SP, as distinct from the CP. These areas, however, may be more or less significant according to the individual languages. Obviously the ultimate difference rests in the domain of discourse-related preferences. A glance at the E UROTYP Perfect Questionnaire may provide some useful hints regarding the relevant discourse contexts. This is especially the case with sentences (8–13) (personal narrative), (20– 21) (informal conversation), (72) and (74) (historical events with persistent result), (23) and (25) (historical events), and (61) (tales). The results are summed up in Table 1. On the face of this, it is certainly correct to state that Spanish is farther back in the aoristic drift than French, although it would be daring to attribute this to some specific semantic (as opposed to textual/stylistic) feature that is present in the Spanish CP and absent in the French. We prefer to view this as a statistical matter: although no purely deterministic principles may be identified, the number of contexts where the CP takes on a merely aoristic function is larger in French than in Spanish. This seems to imply that, if the aoristic drift carries on in the future as it has done so far, there will eventually be a point when Spanish and French coincide. A tentative scalar orientation, based on these observations, could be represented as in Figure 1. However, this is a very coarse-grained formulation. The actual picture is more diverse, for there are areas (such as Southern Italy) where the tense which has survived best is the SP, rather than the CP. We address this in the final section of the paper, where we discuss in some detail the situation of the different varieties of Italian.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

422

422

Mario Squartini and Pier Marco Bertinetto

Table 1. Textual distribution of CP/SP in Romance Spanish

Catalan

Italian

French

personal narratives

preferably SP, irrespective of temporal distance

preferably SP, but the CP prevails in hodiernal contexts

preferably CP SP for Southern speakers, preferably for long distance events

preferably CP SP for long-distance events, and only in style soutenu

informal conversation

SP preferred for non-hodiernal events

SP preferred for non-hodiernal events

preferably CP

preferably CP

historical events with persistent result

preferably SP

preferably SP

preferably SP

preferably CP

historical events

preferably SP

preferably SP

preferably SP

preferably SP

tales

preferably SP

preferably SP

preferably SP

preferably SP

Figure 1. Aoristic drift of the Romance CP Spanish

Occitan Catalan

Standard Italian

more perfectal

Standard French

Standard Romanian

various Northern Italian & French vernaculars purely aoristic

5. Italian: A case in detail The first issue we should consider is the extent to which the CP of Standard Italian (i.e. the variety originally spoken in Tuscany and now spoken by and large by educated people also outside of this area) shows some of the typical perfectal functions.34 As noted in Section 4 above, the CP in Standard Italian allows for the explicit temporal localization of the event (a typical aoristic function), but it goes without saying that in typically perfectal contexts this tense is by far the preferred (if not the only)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

423

The Simple and Compound Past in Romance languages

423

choice. This is notably the case with the following semantic functions, in descending order of relevance: inclusivity (as defined in Section 3.2), experientiality, hot news, and persistent result (cf. also Lindstedt, this volume). On the other hand, the SP occurs in narrative contexts, particularly in the case of historical narration (cf. Table 1). Thus, it is no wonder that the SP is widely used in tales and literary texts, as indeed also happens in French, where this tense fullfils the specifically aoristic function of a propulsive (or foregrounding) tense.35 Italian SP also occurs in newspaper (Burr 1993), where it is reported as more frequent than in French (Savi´c 1979: 189–191), even if still less frequent than Spanish SP (Savi´c 1979, Burr 1989). These tendencies may be further modulated by a number of variables such as presence of adverbs relating to the speaker’s nunc; temporal distance; and first vs. third person narration. But above and beyond this, there are territorial differences in language usage, for the varieties of Italian spoken in the North are close to the situation exhibited by French, whereas those spoken in the South are traditionally described as virtually lacking the CP and generalizing the use of the SP.36 While this is clearly the case with the vernaculars spoken in these two areas (recall what we said in Section 3.4 about the total absence of the SP in the Northern vernaculars), things are definitely more complicated regarding the corresponding varieties of the national language (recall that in Italy vernaculars normally do not coincide with the local variety of the Standard; this is quite a prominent feature of the Italian area, with respect to the rest of the Romance-speaking area, cf. fn. 27.). In order to shed some light on this problem, an investigation was carried out by the present authors (Bertinetto & Squartini 1996) in eleven towns: three in the North (Turin, Bergamo, Padua), three in the Centre (Pisa, Rome, Macerata), three in the South (Naples, Potenza, Lecce), plus one in Sicily (Palermo)37 and one in Sardinia (Cagliari). Since the inquiry was performed by means of a written questionnaire, the results obtained cannot directly reflect the spoken usage, although the subjects (mostly university students) were warned that they should produce as colloquial answers as possible, despite the usage of the national language instead of the vernacular. Presumably, the results reflect some sort of mental projection of the Standard language, rather than the actual linguistic behaviour of the individual speakers. Nevertheless, it is encouraging to observe that the data vary systematically from town to town, and especially from one geographical area to another, so that we are allowed to draw some reliable conclusions. Consider Table 2, which presents the results by geographical area. The figures in each cell show the percentage of CP used by the informants of a given area regarding the given semantic and textual function, indicated in the left column. The first four rows refer to typically perfectal functions, while the remaining three refer to functions typically fulfilled by a purely perfective Past. It should be remarked that the functions tested do not exhaust the list of possible values to be assigned to the SP and the CP; they simply represent the most salient cases for assessing the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

424

424

Mario Squartini and Pier Marco Bertinetto

contrast between the two Past tenses of Italian (and possibly of any Romance language). Although most labels are self-explanatory, a clarification is needed for the category “ST-oriented adverbs”, which groups together sentences containing adverbs yet or explicitly pointing, in the given context, to the Speech Time (such as: not already). As to the contrast “personal vs. impersonal narration”, this obviously rests on the contrast between first and third person subjects. As can be seen, the upper part of the table does not show dramatic differences between the various geographical areas, with the use of the CP overwhelming consistently the SP (cf. the row “perfectal functions”). Only the lower part exhibits notable divergences. The first conclusion to be drawn is therefore that the most characteristic perfectal functions are expressed by the CP, with very minor areal differences. As to the narrative functions typically associated with aoristic tenses (i.e. the last three categories, cf. the row “narrative functions”), they are attributed to the CP significantly more often than to the SP by Northern and Sardinian speakers, while in the case of Southern speakers the SP significantly prevails over the CP. With Sicilians, on the other hand, the distribution of the responses is not significant, while the situation of Central Italy’s speakers is close to significance (note that the different size of the two samples, with 10 and 30 subjects respectively, has a bearing on the statistical outcome). The variance in use of the CP is also confirmed by bivariate comparisons of geographical areas regarding the last three (typically aoristic) functions: the contrast between the North and the remaining geographical areas is significant in nearly all cases, the only exception being the comparison between the North and Sardinia. This shows that, indeed, as is often claimed, Northern speakers tend to generalise usage of the CP and extend it to specifically aoristic contexts, and this is fundamentally true also of Sardinians. Logically, this tendency emerges in particular with the category “personal narration”, where a deictic element, namely the first person subject, is involved.38 It is interesting to observe that the categories “personal narration” and “impersonal narration” on the one hand, and "historical narration” on the other hand, show diverging trends in the Centre as opposed to the South, a result which lends credence to the received idea that speakers from the former area make a subtler choice of the SP/CP opposition. This is partly true also of Northern speakers, who show a constant decrease in the use of CPs from “personal narration” to “impersonal narration” down to “historical narration”, while Southern speakers tend to maintain the same percentage throughout. Presumably, in a really spontaneous situation these tendencies would be further emphasized. Thus it is possible with Northern speakers that the percentage of CPs in the categories of “impersonal” and “historical narration” would be even higher than observed. Conversely, it is possible that in truly spontaneous situations Southern speakers would exhibit a somewhat lower percentage of CPs in the upper part of the table, at least in categories such as “persistent result” and “ST-oriented adverbs”. However, it is likely that the result would not

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

425

The Simple and Compound Past in Romance languages

425

Table 2. Number of CP responses and corresponding percentages in the questionnaire sentences. The proportion of SP responses (including also other solutions marginally chosen by the speakers) may be computed by subtracting the percentage indicated in each cell from 100. Asterisks mark the only cell values non-significant at the 0.05 level, according to the r2 test; asterisks in parenthesis indicate values close to significance. A cell may be significant not only when the percentage is very high, but also when it is very low, for this suggests that an overwhelming proportion of SPs were used. Note that significance is established with respect to the size of the sample, which is different for North, South, and Centre vs. Sicily and Sardinia. Semantic functions

North

Centre

South

Sicily

Sardinia

row total

a. inclusivity

30 100%

30 100%

30 100%

10 100%

10 100%

110 100%

b. ST-oriented adverbs

90 100%

29 96.5%

30 100%

319 97%

c. persistent result

206 98%

67 95.5

70 100%

724 94%

d. experientiality

86 95.5%

87 96.5%

83 92%

200 95%

181 86%

89 99%

82 91%

28 93.5%

28 93.5%

313 95% 1466 95%

FUNCTIONS

412 98%

406 96.5%

376 89.5%

134 95.5%

138 98.5%

(a–d) e. personal narration

209 69.5%

166(*) 55.5%

74 24.5%

53* 53%

78 78%

f. impersonal narration

66* 55%

37 31%

27 22.5%

13 32.5%

25* 62.5%

g. historical narration

7 23.5%

2 6.5%

6 20%

1 10%

2(*) 20%

NARRATIVE

282 62.5%

205(*) 45.5%

107 23.5%

67* 44.5%

105 70%

694 79.5%

518 70%

483 55.5%

201 63.5%

243 84%

PERFECTAL

FUNCTIONS

(e–g) column total

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

580 53% 168 38% 18 16.5% 766 46%

426

426

Mario Squartini and Pier Marco Bertinetto

change with “inclusivity” and “experientiality”, which seem to represent the most resistant CP functions. The general conclusion that seems to emerge from this data is the following. The communis opinio that Northern Italian speakers tend to extend the CP to all contexts turns out to be true to a very large extent, although in semi-formal situations (such as the completion of a questionnaire) the SP occurs relatively often in narrative contexts. As to Southerners, in the type of situation considered, they dramatically diverge from Northerners, but also from Central speakers. It should be noted, though, that this is merely a probabilistic tendency, rather than a sharp contrast of grammaticality. In actual usage, a fair amount of variability is to be observed, at least in the less characteristic contexts. Recall also that the spontaneous behaviour of the North and South, when heavily influenced by the respective vernaculars, would appear to be even more extreme than that elicited by our questionnaire. Finally, the two major islands occupy some kind of intermediate position, with Sardinia closer to the North and Sicily, somewhat surprisingly, closer to the Centre than to the South proper.39 The investigation carried out reveals that Italian, particularly the variety spoken in the Centre (which essentially coincides with the Standard variety), has not altogether dismissed the original perfectal meaning of the CP. However, it is equally clear that a fair amount of variability is to be observed along the geographical dimension, suggesting that the competition between SP and CP will continue to develop within an intricate, multifactorial interplay.

Notes 1. This paper was jointly conceived and written by the two authors. For academic purposes, though, PMB bears responsibility for Sections 2, 4 and 5, while MS is responsible for Sections 1 and 3. The authors are grateful to the members of E UROTYP Group 6, and in particular to Jouko Lindstedt, for useful comments. 2. Although this is the general pattern, there are exceptions. The most notable one is the colloquial Catalan SP, which is formed by means of the auxiliary anar ‘go’ followed by the Infinitive. Even here, though, there exists a synthetic literary variant. It might look misleading to adopt the label SP to speak of a periphrastic construction, but we shall nevertheless use this term for general conformity. 3. The various Romance varieties differ with respect to this parameter. Some languages, like Spanish, have lost the Perfect Participle agreement altogether; others, like French or (even more so) Italian, have retained it in particular circumstances. Contemporary Italian, for instance, shows agreement when the controller is a direct Object clitic (cf. (a)), but not when the Object consists of a lexical nominal or a relative pronoun (cf. (f– g)). Agreement is also exhibited relative to the internal argument of unaccusative verbs (cf. (b)), the notional object of passives (cf. (c)), the clitic object of a causative verb (cf. (d)), and the subjects of reflexive verbs (cf. (e)). The theoretical reasons for unifying all

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

427

The Simple and Compound Past in Romance languages

427

these apparently disparate cases are spelled out by La Fauci (1989) within a Relational Grammar framework. Consider: a.

ha visti (*visto) (Quanto ai bambini) Giovanni li As to the children, Giovanni them has seen:M:PL (*seen:M:SG) ‘(As to the children) Giovanni has seen them.’

b.

(*arrivato) Maria è arrivata Maria is arrived:F:SG (*arrived:M:SG) ‘Maria has arrived.’

c.

da Giovanni (*visto) Maria fu vista Maria was seen:F:SG by Giovanni (*seen:M:SG) ‘Maria was seen by Giovanni.’

d.

piangere (Quanto ai bambini) Giovanni li ha fatti As to the children, Giovanni them has made:M:PL cry (*fatto) (*made:M:SG) ‘As to the children, Giovanni has made them cry.’

e.

le scarpe (*messo) Maria si è messa Maria she:RFL is put:F:SG DEF:F:PL shoes (*put:M:SG) ‘Maria has put on her shoes.’

f.

Maria (*vista) Giovanni ha visto Giovanni has seen:M:SG Maria (*seen:F:SG) ‘Giovanni saw Maria.’

g.

Queste sono le notizie che Maria ha ricevuto These are DEF:F:PL news:PL that Maria has received:M:SG (*ricevute) (*received:F:PL) ‘This is the news that Maria has received.’

French, by contrast, shows agreement only in (a–c, g). For a full-fledged analysis, both descriptively rich and theoretically thorough, of the Perfect Participle agreement in the Romance languages, see Loporcaro (1998, 1995a). It is noteworthy that Romance languages have undergone diachronic change in this domain. For instance, Old Italian (and some varieties of Contemporary Italian) admitted the agreement also in (f). 4. A remnant of the original freedom is still to be observed, perhaps in all Romance varieties, in predicative constructions such as Port. tenho uma carta escrita ‘I have a written letter’, as opposed to tenho escrito uma carta ‘I have written a letter’ (note that agreement is only required in the first case). The emergence of these predicative (and resultative) constructions correlates with the possibility of an adjectival reading of the Perfect Participle. When neither the agreement rule nor the order of the constituents provides a cue to their interpretation, sentences may be truly ambiguous, as in:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

428

428

Mario Squartini and Pier Marco Bertinetto a.

Italian (la collana) Maria l’ ha appesa al collo (the necklace) Mary it:OBJ has hung at-the neck ‘CP’ = ‘M. has hung it around her neck’ predicative-resultative construction = ‘M. has it hanging around her neck’

5. The most commonly used auxiliaries in Romance languages are the descendants of esse ‘be’ and habere ‘have’. As to the distribution of esse and habere, they vary from language to language (cf. Vincent 1982). In Spanish and Catalan habere has been generalized to all verbs. In French, Occitan, Italian and Sardinian by contrast, both esse and habere are used, although the respective distribution varies. Especially interesting is the case of Italian, where esse is only used with unaccusative verbs. However, in Central and Southern Italian vernaculars (see fn. 7 below) the situation is quite diverse (Tuttle 1986; Loporcaro 1998). As to Romanian, habere has been generalized to all verbs for the CP, while esse is used with Future, Conditional and Subjunctive Perfects. The case of Portuguese and Galician, where another auxiliary is selected, will be presented below. It has been observed that the reduction of the auxiliary choice to habere (as in Spanish and Catalan) correlates with the loss of agreement of the Perfect Participle (cf. also fn. 3) 6. Note, however, that the received idea that full grammaticalization of the CP is reached only at the stage when agreement is lost cannot be maintained, due to the existence of Romance varieties with highly grammaticalized compound tenses, where agreement is preserved (Loporcaro 1998, 1995a; cf. also fn. 3). 7. We use the term “vernacular”, for lack of a better denomination, to refer to what in the Italian linguistic tradition is called “dialetto”, i.e. a local language which developed independently of the national language, although (in most cases) connected to it from a diasystematic point of view. “Vernacular” should not be confused with “local variety of the standard language”, a notion which would rather correspond to the most prominent meaning of English “dialect”. See further Bertinetto, this volume a, fn. 4. 8. These patterns are very similar to the diachronic steps proposed by Alarcos Llorach (1947: 136) for the evolution of Spanish CP, and a slightly modified version is also presented in Fleischman (1983: 195) and Schwenter (1994a: 77); for a similar proposal, cf. also Schlieben-Lange (1971: 128 fn. 13). 9. Paiva Boléo (1936); de Castilho (1966, 1967); Irmen (1966); Wandruzska (1966: 9– 15); Brinkmann (1970); Sten (1973: 233–259); Algeo (1976); Cella (1978: 310–391); Travaglia (1981); Suter (1984). 10. Bertinetto (1986) considers the inclusive meaning of perfectal tenses an instance of aspectual blending. One might wonder whether the restriction on this particular interpretation of the Portuguese CP is due to the presence of the auxiliary ter instead of haver. This is a reasonable hypothesis in itself; but note that some varieties of Latin American Spanish, discussed below, seem to behave in the same way, despite the presence of the auxiliary haber. 11. It is a general Romance feature that with stative and activity verbs, in inclusive contexts, it is also possible to find the Present:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

429

The Simple and Compound Past in Romance languages a.

12. 13. 14.

15.

16.

429

French J’ habite ici depuis longtemps I live here since long-time ‘I have lived here for a long time.’

Interestingly, as the translation suggests, English makes use of the Present Perfect also in these cases. Cf. by contrast the Calabrian example cited in Section 3.1, where the CP is employed in exactly this situation. According to Sten (1973: 251), this case could also be considered an example of iterativity, since there is a comparison between different occurrences of the same situation. As for Asturian, Harre (1991: 155–159, 165–166) carried out specific research on Oviedo Spanish, confirming that the CP (of the Spanish type haber  Past Participle) tends to be avoided and is substituted by the SP. But Oviedo Spaniards show quite a productive usage of the construction with tener  Past Participle, which is much more extensive not only with respect to the corresponding Spanish construction but also with respect to Portuguese. As in Portuguese, the Asturian form occurs for durative and iterative situations including the Speech Time, but, unlike Portuguese it can also occur in experiential contexts (Ayer pesqué una trucha que medía – Pues eso no es nada. Yo tengo pescado una ‘Yesterday I caught a trout measuring – Well, that’s nothing. I caught que medía one measuring’, lit.: ‘ have caught ’) and also for semelfative punctual situations with current relevance reading (Tengo roto la pierna en esos días que llovía y estaban las calles resbaladizas ‘I broke my leg in those days when it was raining and the streets were slippery’, lit.: ‘ have broken ’). Moreover in Asturian, unlike Portuguese, the situation is not required to include the Speech Time (Tiene perdido cinco kilos pero después engordó diez ‘S/he lost five kilos but then s/he gained ten kilos’, lit.: ‘ has lost five kilos ’), but significantly the invariant Past participle form is preferred in these cases, while when the Speech Time is included the agreeing form occurs (Tiene perdidos cinco kilos y espera perder cinco más ‘S/he (has) lost five kilos and hopes to lose five more’, lit.: ‘ has lost:OBJ ’). A constrained usage of the CP is also documented in contemporary Judeo-Spanish, as it is spoken and written in Israel (cf. Malinowski 1984). The auxiliary used for constructing the CP is tener, which occurs mostly in negative contexts, as in No lo tengo visto hasta agora entre los klientes de mi tante ‘Until now I had not seen him among my aunt’s ’). Notwithstanding the tendency to use the CP in customers’ (lit.: ‘ not have seen negative contexts, similar to Mexican rather than to Portuguese, Judeo-Spanish maintains the SP with the adverbial not yet (Ainda no me kazi ‘I have not married yet’, lit.: ‘I did not marry yet’), while in Mexican the CP is also used in such a case. Besides, Judeo-Spanish seems to admit experiential contexts (Tiene sintido esta palavra? ‘Have you ever heard this word?’), thus showing a further degree of grammaticalization. Schwenter (1994b: 1014–1019) considers these marked affective utterances mentioned by Lope Blanch as hot news perfects, which, according to Schwenter’s data, do occur in Mexican television newscasts, even if their frequency is much more restricted with respect to European Spanish television broadcasting. Schwenter considers hot news per-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

430

430

17.

18.

19. 20. 21.

22.

Mario Squartini and Pier Marco Bertinetto fects as an important turning point in the grammaticalization of perfects, being the first instance of purely past perfective situations. Among the special usages of the Mexican CP, Lope Blanch also mentions cases of modal extension towards irrealis, as in Si ha salido un poquito antes, no lo hubiera recibido a tiempo ‘if s/he had left a little before, s/he would not have received him/it in time’ (lit.: has left ). More recently, Almeida (1987–88) and Herrera Santana & Medina López (1991) have confirmed that in the Canaries the SP prevails over the CP and the latter is most frequent in negative contexts or when denoting durative/iterative situations encompassing the Speech Time. Nevertheless they have also shown that the CP can also be used when referring to semelfactive situations; the examples quoted often contain verbs of perception or verbs of saying (Ya he dicho antes que soy maestra ‘I have already told you that I am a teacher’, from Almeida 1987–88: 73), but other verbs can be found as well (He vuelto otra vez a jugar al fútbol ‘I started playing football again’, lit.: ‘have started’, from Herrera & Medina 1991: 236). Sociolinguistic data showing the gradual extension of the CP in the Canaries, possibly due to the influence of the peninsular norm, are presented in Herrera Santana & Medina López (1991) and Serrano (1995). As for Argentina, Donni de Mirande (1977: 46–49, 1992: 666–668) reports that in the area of Rosario (Santa Fe) all the compound forms (not only the CP) tend to be rarely used in informal style. A similar pattern, whereby the SP prevails over the CP, can be found in most of the country, including Buenos Aires (where Kubarth’s 1992 data suggest a distribution of CP similar to that observed in Mexico). However, the pattern changes in the North of the country, where it is the CP that occurs most frequently, while in some areas of the Centre the two forms alternate. The prevalence of CP over SP is also reported by Schumacher de Peña (1980) in the Southern Peruvian highlands (Cuzco, in the Quechua area and Puno, in the Quechua and Aymara area), even though this distribution is not common to the whole Peruvian area, since in the North (Huaraz, department of Ancash, in the Quechua area) it is the SP that prevails. According to Schumacher, such a distribution can be traced back to the fact that Quechua lacks a distinction corresponding to Spanish CP/SP, while the Quechua evidential form is rendered by the Spanish Pluperfect. In some other areas (Ecuador Spanish), it is the CP that assumes a modal meaning (admirative or evidential) due to Quechua influence, as for instance in: el año que viene ha sido bisiesto ‘next year will be a leap year’ (expressing surprise, lit.: ‘next year has been a leap year’; cf. Toscano Mateus 1953: 260 and Bustamante 1991). But many others, such as Dahl & Hedin (this volume), still exploit this notion, which is useful at the descriptive level, despite the theoretical problems it poses. As for a basic literature on Spanish SP/CP see Alarcos-Llorach (1947); Barrera-Vidal (1972); Berschin (1976); Kuttert (1982). On the frequency of the CP with adverbials denoting an interval close the the Speech Time or encompassing it, see de Kock (1984), who analyzes a corpus of written Spanish texts. In a later work, de Kock (1986) presents statistical data on the usage of the CP in the same corpus, and the specific temporal perspective conveyed by it. Actually, it is not clear whether the subjects were Madrileños or generally Spaniards.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

431

The Simple and Compound Past in Romance languages

431

23. Skubic (1964: 87 fn. 2) reports that children in Southern Spain use the SP also in hodiernal narratives. 24. That Catalan has proceded further and faster in the aoristic drift (cf. Section 4) is also demonstrated by the diachronic data presented in Eberenz (1977), who compares the usage of SP and CP in a 15th century Catalan text with its Spanish translation, published at the beginning of the 16th century. It appears that the Catalan CP has assumed the function of a purely perfective past earlier. See also Morales (1993) on the gradual extension of CP in a Catalan dialect (the valenciano as spoken in Vall d’Uxó), where CP prevails when there is no temporal adverbial localizing the Event Time, but it may also occur with these adverbials or denote long-distance situations. Note, moreover, that in the dialect of Catalan spoken in Roussillon (France) the CP has proceeded even further in extending its functions, for it is the only form used, thus adopting the spoken French distribution (Östen Dahl, p.c.). 25. Nevertheless Benincà (1989: 576) reports that in some areas of Friulian the SP is still currently used. 26. In most varieties of Sardinian the SP has completely disappeared, and, as a result of this process, two main patterns can be depicted: in the Northern varieties the SP has been replaced by the CP, as in the other Romance languages at stage IV, while in the Southern area of Sardinia (i.e. Campidano), the Pluperfect and even the Imperfect are reported as substitutive forms instead of the SP (cf. Loi Corvetto 1982: 144–147; Blasco Ferrer 1984: 30–33, 1986: 147–152; Bossong 1993). 27. Recall, that in the Italian area, more than in any other Romance language, the vernacular does not always coincide with the local variety of the Standard. This should also be kept in mind when considering the data in Section 5, where what has been studied is the local variety of Standard Italian rather than the vernacular. 28. The usage of SP in Arumanian is confirmed by Savi´c (1991), who notes that in a recent Arumanian Bible translation the SP is quite frequent, while it is completely missing in the corresponding Romanian version. 29. However, Mateica-Igelmann (1989: 46–49) notes that other situation types (in particular non-telic) admit the SP (Dormi¸si? Dormii toat˘a ziua ‘Did you sleep? I slept the whole day’). 30. Analogous considerations can be formulated regarding the Sicilian vernacular, which is reported nowadays as being at stage II. Historical data from literary texts show that the CP used also to occur in other contexts, in particular those expressing current relevance (cf. Ambrosini 1969 and Skubic 1973–1975). Skubic interprets these data as the result of literary influence from other Romance languages. 31. Referovskaja (1949, quoted by Schogt 1964: 10) maintains that, when in Old French texts the CP is used for referring to purely past situations rather than to the current result of a past situation, the first verbs to occur are accomplishments or achievements. As noted by Fleischman (1983: 207 fn. 27), these data contrast with the distribution of stage II, in which the CP is favored in non-telic situations. Further research should clarify whether the French CP underwent a different process of grammaticalization with respect to the Portuguese form, inasmuch as actional restrictions are concerned (On the diachronic evolution of French CP cf. also Saettele 1971, Blumenthal 1986).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

432

432

Mario Squartini and Pier Marco Bertinetto

32. In fact, such an evolution seems to be confirmed by diachronic data on European Spanish presented by Alarcos Llorach (1947), who claims that the CP has passed through a stage in which it expressed a durative or iterative situation leading up to the Speech Time. He notes that while in The Cid and in 13th century texts the CP mainly has a resultative value (although in some cases its usage is simply because of metrical reasons or to variatio), in the 14th century it is already used for expressing mere durativity and iterativity, and later evolves into a past perfective tense. However, there are data contradicting Alarcos’ account. Consider, for instance, the following example quoted by Harre (1991: 114) from Gonzalo de Berceo (13th century), where the CP refers to a semelfactive past action: Yo nunca te tollí valía de un grano e tú hasme tollido a mí un capellano ‘I never took the slightest thing from you / and you have taken a priest from me’. Clearly, in this case the CP, alternating with the SP, is used for stylistic variatio; nevertheless it is undeniable that it could also occur with a mere semelfactive value. Also in Company’s (1983) data on medieval Spanish the CP is to be observed in semelfactive perfective contexts. 33. Blasco Ferrer (1984) notes that there are similarities among some peripheral languages of the Romance domain, such as Portuguese, Spanish, Sicilian and dialectal Romanian. This might suggest an areal interpretation, with obvious implications as to the situation of Vulgar Latin. However, among these peripheral languages, differences are sometimes more prominent than similarities. Take, for instance, the Romanian dialect of Oltenia, which exhibits a very different and peculiar evolution, as compared to Portuguese, Spanish and Sicilian. We would like to reformulate Blasco Ferrer’s proposal by claiming that all we can say at the present time is that there are clear semantic similarities among Sicilian, Portuguese, and some varieties of Latin American Spanish (not Spanish in general). 34. On the semantic properties of Italian SP/CP see Bertinetto (1986) and Lepschy & Lepschy (1992). 35. In some cases the textual distribution of SP and CP is even more complicated: Centineo (1991) has shown that even in a Southern variety, namely the variety of Italian spoken in Sicily, SP and CP alternate in oral narratives, and both forms show a pure perfective function. Centineo claims that the alternation of the two forms in the same text is due to textual strategy purposes, signalling the switch from the sequential narration to the evaluative section, without any restriction as to which form can appear in the narrative sequence and or in the evaluation. 36. Solarino (1991) and Lo Duca & Solarino (1992) have studied the relationship between geographical variety and textual restrictions of SP/CP. They show that in speakers from a Northern variety the SP is restricted to the oral telling of fairy tales, while it is absent in autobiographical stories and in movie plots; but even in fairy tales the percent of SP is lower than in speakers from a Southern variety, who, on the contrary, do use the SP also for autobiographical stories. Some suggestions on the distribution of SP in spoken Italian can be found in Gambarara (1994), who reports data from LIP (Lessico dell’Italiano Parlato, ‘Spoken Italian Lexicon’). First, an inspection of the 200 most frequent verbs shows that the SP is used comparatively more often with irregular than with regular verbs, despite the fact that the SP of irregular verbs must be learned by rote. As to the different types of communicative style, the percentage of SP reaches the minimum in TV and radio talks, and in face-to-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

433

The Simple and Compound Past in Romance languages

433

face conversations (between 0.06% and 0.4% of all verbal forms), while it rises slightly in telephone conversations, monologues (such as lectures) and formal discussions (between 0.4% and 0.8%). Unfortunately, no indication is given as to the frequency of usage in the four towns where the recordings were made (Florence, Milan, Naples and Rome). One piece of data relevant for our discussion is the following: among the observed data, the SP is virtually absent in the first and second persons of the verb, and is almost exclusively concentrated in the third person (sg. and pl.), a restriction applying in French too. Apparently, the use of this tense in the spoken language mirrors the Benvenistian opposition between discours and histoire. This seems to agree perfectly with the data presented in Table 2. 37. Note that the variety under scrutiny is the variety of Italian spoken by Sicilians rather than the local vernacular, where (as shown in Section 3.4) the CP is restricted to inclusive contexts. 38. Obviously, this deictic element suggests the possible current relevance of the narrated event, while the third person marks the distance of the narrator from her/his topic. 39. One should not forget, however, that the insular data come from only two points of observation, Palermo and Cagliari. Thus, one should be cautious before generalizing these results. This is actually also necessary with reference to the other major areas, for slight divergences were often gathered from town to town. Furthermore, in the Southern town of Cosenza (Northern Calabria, not included in the data reported here) our questionnaire elicited a very extreme behaviour, yielding virtually no SP in the responses of the subjects: something that did not happen in any of the Northern locations investigated. Although this is clearly a deviant phenomenon, not representative of the overall behaviour of Southern speakers, it tells us that the observed regularities should not be extended to each individual location. Interestingly, although the total disappearance of the SP is reported to be a recent phenomenon in Cosenza and the surrounding area in general (Loporcaro 1995b: 550), the linguistic behaviour of its inhabitants is much more extreme than that of our subjects from Cagliari, where the disappearance of the SP from Sardinian vernaculars (not to be confused with the local variety of Italian spoken there, as scrutinized in our questionnaire) started already by the end of the 17th century and is nowadays complete, as claimed by Blasco Ferrer (1984: 30).

References Alarcos Llorach, Emilio 1947 “Perfecto simple y compuesto en español”, Revista de Filología Española 31: 108–139. Algeo, James E. 1976 “The Portuguese present perfect", Luso-Brazilian Review 13: 194–208. Almeida, Manuel 1987–88 “Perfecto simple y perfecto compuesto en el español de Canarias”, Revista de Filología (Universidad de la Laguna) 6–7: 69–77. Ambrosini, Riccardo 1969 “Usi e funzioni dei tempi storici nel siciliano antico”, Bollettino del Centro di Studi Filologici e Linguistici Siciliani 10: 141–178.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

434

434

Mario Squartini and Pier Marco Bertinetto

Badia Margarit, Antonio M. 1962 Gramática catalana. Madrid. Gredos. (Reprinted 1985). Barrera-Vidal, Albert 1972 Parfait simple et parfait composé en castillan moderne. München: Hueber. Benincà, Paola 1989 “Friaulisch: Interne Sprachgeschichte I. Grammatik”, in: Günter Holtus, Michael Metzeltin, Christian Schmitt (eds.), Lexikon der Romanistischen Linguistik. Tübingen: Niemeyer 3: 563–585. Berschin, Helmut 1976 Präteritum- und Perfektgebrauch im heutigen Spanisch. Tübingen: Niemeyer. Bertinetto, Pier Marco 1986 Tempo, Aspetto e Azione nel verbo italiano. Il sistema dell’indicativo. Firenze: Accademia della Crusca. Bertinetto, Pier Marco & Mario Squartini 1996 “La distribuzione del Perfetto Semplice e del Perfetto Composto nelle diverse varietà di italiano”, Romance Philology 49: 383–419. Blasco-Ferrer, Eduardo 1984 “Rumeno dialettale ajunse acum, portoghese chegou agora, spagnolo llegó ahora, siciliano arrivau ora = ‘è arrivato adesso’, Posdeterminazione arcaica contro predeterminazione innovativa nell’espressione romanza del perfetto latino”, Annali della Facoltà di Magistero dell’Università di Cagliari, n.s. 8/2: 5–63. Blasco-Ferrer, Eduardo 1986 La lingua sarda contemporanea. Grammatica del logudorese e del campidanese. Cagliari: Edizioni Della Torre. Bleton, Paul 1982 “La surcomposition dans le verbe français”, Canadian Journal of Linguistics 27: 31–40. Blumenthal, Peter 1986 Vergangenheitstempora, Textstrukturierung und Zeitverständnis in der französischen Sprachgeschichte. Stuttgart: Steiner. Bossong, Georg 1993 “Innovative Tendenzen im sardischen Verbalsystem”, in: Johannes Kramer & Guntram A. Plangg (eds.), Verbum Romanicum – Festschrift für Maria Iliescu. Hamburg: Buske, 337–352. Brinkmann, Margarete 1970 Das portugiesische Verbum im Übersetzungsvergleich Portugiesisch–Spanisch–Italienisch–Französisch: Die semantische Verteilung von ‘perfeito simples’ und ‘perfeito composto’. [Doctoral Dissertation Universität Tübingen.] Burr, Elisabeth 1989 “Il sistema verbale italiano e spagnolo. Indagine sulla lingua dei quotidiani”, in: Fabio Foresti, Elena Rizzi, Paola Benedini (eds.), L’italiano tra le lingue romanze, Atti del XX Congresso SLI (27), 253–276. 1993 Verb und Varietät. Ein Beitrag zur Bestimmung der sprachlichen Variation am Beispiel der italienischen Zeitungsprache. Hildesheim: Georg Olms. Bustamante, Isabel 1991 “El presente perfecto o pretérito perfecto compuesto en el español quiteño”, Lexis 15: 195–231. C˘al˘ara¸su, Cristina 1992 “Quelques significations des temps verbaux roumains en perspective romane (avec applications aux langues roumaine et française)”, Revue Roumaine de Linguistique 37: 137–143.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

435

The Simple and Compound Past in Romance languages

435

Canarache, Ana 1965 “Perfectul compus în texte narative”, Limba Român˘a 14: 689–695. Cano González, Ana María 1992 “Asturianisch: Interne Sprachgeschichte”, in: Günter Holtus, Michael Metzeltin, Christian Schmitt (eds.), Lexikon der Romanistischen Linguistik. Tübingen, Niemeyer 6.1: 653–680. Cardona, Julia 1979 “Pretérito simple y pretérito compuesto: presencia del tiempo/aspecto en el habla culta de San Juan”, Boletín de la Academia Puertorriqueña de la lengua española 7/1: 93– 110. Carruthers, Janice 1992 “Une étude sociolinguistique des formes surcomposées en français moderne", in: Ramón Lorenzo (ed.), Actas do XIX Congreso Internacional de Lingüística e Filoloxía Románicas 3: 145–162. 1993 “Passé composé, passé surcomposé: marqueurs de l’antériorité en français parlé”, in: Gerold Hilty (ed.), Actes du XX Congrès International de Linguistique et Philologie Romanes 1: 109–122. Catalán, Diego 1966 “El español en Tenerife. Problemas metodológicos”, Zeitschrift für romanische Philologie 82: 467–506. Cella, Claudino 1978 Système verbal français et système verbal portugais. Etude comparative. Documents linguistiques du Centre d’Analyse Syntaxique de l’Université de Metz. Centineo, Giulia 1991 “Tense switching in Italian: the alternation between passato prossimo and passato remoto in oral narratives”, in: Suzanne Fleischman & Linda R. Waugh (eds.), DiscoursePragmatics and the verb. The evidence from Romance. London & New York: Routledge, 55–85. Company, Concepción 1983 “Sintaxis y valores de los tiempos compuestos en español medieval”, Nueva Revista de Filología hispánica 32: 235–257. Cornu, Maurice 1953 Les formes surcomposées en français. (Romanica Helvetica 42.) Berne: Francke. de Castilho, Ataliba T. 1966 “A sintaxe do verbo e os tempos do passado em português”, Alfa (Faculdade de Filosofia, Ciéncias e Letras de Marília, Departamento de Letras) 9: 105–153. 1967 “Introdução ao estudo do aspecto verbal na língua portuguesa”, Alfa (Faculdade de Filosofia, Ciéncias e Letras de Marília, Departamento de Letras) 12: 7–135. de Kock, Josse 1984 “El pretérito perfecto compuesto y las indicaciones de tiempo extraverbales”, Revista de Filología Románica 2: 103–112. 1986 “Del pretérito perfecto compuesto o de la importancia del contexto y de la cuantificación”, Revista de Filología Española 66: 185–236. Donni de Mirande, Nélida Esther 1977 El español hablado en el litoral argentino. Formas personales del verbo. Consejo de Investigaciones: Universidad Nacional de Rosario, Argentina. 1992 “El sistema verbal en el español de la Argentina: rasgos de unidad y de diferenciación dialectal”, Revista de Filología Española 72: 655–670. Eberenz, Rolf 1977 “Zur Entwicklung der Opposition perfecto simple vs. perfecto compuesto im Spanischen

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

436

436

Mario Squartini and Pier Marco Bertinetto und Katalanischen der Renaissance”, Zeitschrift für romanische Philologie 93: 518– 527.

Engel, Dulcie M. 1990 Tense and Text. A study of French past tenses. London & New York: Routledge. Fleischman, Suzanne 1983 “From Pragmatics to Grammar. Diachronic reflections on complex pasts and futures in Romance”, Lingua 60: 183–214. Foulet, Lucien 1920 “La disparition du prétérit”, Romania 46: 271–313. 1925 “Le développement des formes surcomposées”, Romania 51: 203– 252. Gambarara, Daniele 1994 “Il passato remoto nell’italiano parlato”, in: Tullio De Mauro (ed.), Come parlano gli italiani. Firenze: La Nuova Italia, 183–194. Georgescu, Alexandru 1958 “Perfectul simplu în dialectul dacoromîn. Observa¸tii asupra r˘aspîndirii geografice a perfectului simplu în dacoromîn˘a”, in: Omagiu lui Iorgu Iordan. Bucharest: Editura Academiei Republicii Populare Romîne, 317–324. Harre, Catherine E. 1991 Tener  Past Participle. A case study in linguistic description. London & New York: Routledge. Harris, Martin 1982 “The ‘past simple’ and the ‘present perfect’ in Romance”, in: Nigel Vincent & Martin Harris (eds.), Studies in the Romance verb. London & Canberra: Croom Helm 42–70. Herrera Santana, Juana & Javier Medina López 1991 “Perfecto simple/perfecto compuesto: análisis sociolingüístico”, Revista de Filología (Universidad de la Laguna) 10: 227–239. Herzog, Christian 1981 Le passé simple dans les journaux du XXe siècle. Bern: Francke. Irmen, Friedrich 1966 “O pretérito composto em português”, Revista de Portugal, Série A: Língua Portuguesa 31: 222–238. Kany, Charles E. 1951 American-Spanish syntax. Chicago & London: The University of Chicago Press. (2nd edition). Reprinted Midway (1975). Klein, Wolfgang 1992 “The present perfect puzzle”, Language 68: 525–552. Kubarth, Hugo 1992 “Perfecto compuesto y perfecto simple en el habla de Buenos Aires”, in: Ramón Lorenzo (ed.), Actas do XIX Congreso Internacional de Lingüística e Filoloxía Románicas, Santiago de Compostela 3: 505–516. Kuttert, Rainer 1982 Syntaktische und semantische Differenzierung der spanischen Tempusformen der Vergangenheit perfecto simple, perfecto compuesto und imperfecto. Frankfurt am Main & Bern: Peter Lang. La Fauci, Nunzio 1989 “Ausiliari perfettivi e accordo del participio passato in italiano e in francese”, in: Fabio Foresti, Elena Rizzi, Paola Benedini (eds.), L’italiano tra le lingue romanze, Atti del XX Congresso SLI (27), 213–242. Lepschy, Anna Laura & Giulio Lepschy 1992 “I tempi del passato”, Linguistica 32: 75–88.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

437

The Simple and Compound Past in Romance languages

437

Lo Duca, Maria Giuseppina & Rosaria Solarino 1992 “Contributo ad una grammatica del parlato: testi narrativi e marche temporali”, in: Luciana Brasca & Maria Luisa Zambelli (eds.), Grammatica del parlare e dell’ascoltare a scuola. Quaderni del GISCEL 13. Firenze: La Nuova Italia. 33–49. Loi Corvetto, Ines 1982 L’italiano regionale di Sardegna. Bologna: Zanichelli. Lope Blanch, Juan M. 1961 “Sobre el uso del pretérito en el español de México”, in: Studia Philologica, Homenaje ofrecido a Dámaso Alonso. Madrid, 2: 373–385 (reprint in: J. M. Lope Blanch, Estudios sobre el español de México. Universidad Nacional Autónoma de México: México, 1983, 2nd ed. revised, 131–143). Loporcaro, Michele 1995a “Grammaticalizzazione delle perifrasi verbali perfettive romanze e accordo del participio passato”, Archivio Glottologico Italiano 80: 144- 167. 1995b

“Raddoppiamento fonosintattico dopo III persone plurali del verbo nei dialetti di Conflenti (CZ) e di San Giovanni in Fiore (CS)”, Atti della Accademia Nazionale dei Lincei, Classe di Scienze Morali, Storiche e Filologiche, Rendiconti, Serie IX, 6: 543–553.

1998

Sintassi comparata dell’accordo participiale romanzo. Torino: Rosenberg & Sellier.

Malinowski, Arlene C. 1984 [1989] “Distribution and function of the auxiliaries tener and aver in Judeo-Spanish”, Orbis 33: 211–221. Manoliu-Manea, Maria 1989 “Rumänisch: Morphosyntax”, in: Günter Holtus, Michael Metzeltin, Christian Schmitt (eds.), Lexikon der Romanistischen Linguistik. Tübingen: Niemeyer 3: 101–114. Mateica-Igelmann, Michaela 1989 Moyens d’exprimer les aspects de la phrase verbale en roumain contemporain. Bochum: Brockmeyer. Millán Urdiales, José 1966 El habla de Villacidayo. León. Madrid: Anejos del Boletín de la Real Academia Española 13. Mòcciaro, Antonia G. 1978 “Passato prossimo e passato remoto in siciliano. I risultati di una inchiesta”, in: Manlio Cortelazzo (ed.), La ricerca dialettale 11: 343–349. Pisa: Pacini. Moise, Ion 1977

“Aria de r˘aspîndire a perfectului simplu în Oltenia s¸i Muntenia”, Limba Român˘a 26/1: 91–93.

Morales, Montserrat 1993 “Transfert linguistique et évolution du passé défini dans les langues romanes: le cas valencien”, Revue de Linguistique Romane 57: 79–92. Moreno de Alba, José G. 1978 Valores de las formas verbales en el español de México. Universidad Nacional Autónoma de México: México. Paiva Boléo, Manuel 1936 O Perfeito e o Pretérito em português em confronto com as outras línguas românicas. Coimbra: Biblioteca da Universidade. Pan˘a-Boroianu, Ruxandra 1982 “Remarques sur l’emploi du passé simple dans les textes non littéraires d’Olténie”, Revue Roumaine de Linguistique 27: 423–434.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

438

438

Mario Squartini and Pier Marco Bertinetto

Pinkster, Harm 1987

Pisani, Vittore 1981

“The strategy and chronology of the development of Future and Perfect Tense Auxiliaries in Latin”, in: Martin Harris & Paolo Ramat (eds.), Historical development of auxiliaries. Berlin: Mouton de Gruyter, 193–223. “Origini e fortuna del Passato Prossimo”, in: Horst Geckeler, Brigitte Schlieben-Lange, Jürgen Trabant, Harald Weydt (eds.), Logos semantikos. Studia linguistica in honorem E. Coseriu. Berlin & New York: de Gruyter / Madrid: Gredos 4 (Christian Rohrer, ed.): 435–441.

Rallides, Charles 1971 The tense aspect system of the Spanish verb as used in cultivated Bogotá Spanish. The Hague & Paris: Mouton. Ramat, Paolo 1982 “Ein Beispiel von ‘Reanalysis’ typologisch betrachtet”, Folia Linguistica 16: 365–383. Referovskaja E. A. 1949 “K voprosu o kategorii vida v jazyke francuzskogo narodnogo eposa”, Uˇc Zap., L.G.U. 97, Ser. Fil. Nauk 14: 140–159. Rohlfs, Gerhard 1966–69 Grammatica storica della lingua italiana e dei suoi dialetti. Torino: Einaudi. Rojo, Guillermo 1974 Perífrasis verbales en el gallego actual. Verba, Anejo 2. S˘adeanu, Floren¸ta 1972 “Perfectul simplu s¸i perfectul compus. Compara¸tie între spaniol˘a s¸i român˘a”, Studii s¸i Cercet˘ari Lingvistice 23: 615–626. Saettele, Hans 1971 Das französische Passé composé. Funktionsveränderung eines Tempus. Zürich: Juris. Said, Sally E. S. 1976 Variation in usage of the present perfect tense in the spoken Spanish of Mexico City. Ph.D. dissertation, University of Texas at Austin. Santamarina, Antonio 1974 El verbo gallego. Estudio basado en el habla del valle del Suarna. Verba, Anejo 4. Savi´c, Momˇcilo D. 1979 “L’uso dei tempi passati nei quotidiani pubblicati nelle lingue romanze con particolare riguardo all’italiano”, Linguistica 19: 171–197. 1990 “L’uso dei preteriti nello stile biblico italiano (nel quadro delle lingue europee)”, Italica Belgradensia 3: 9–18. 1991 “Une différence fondamentale dans la langue biblique entre le roumain et l’aroumain”, Linguistica 31: 107–119. Schlieben-Lange, Brigitte 1971 Okzitanische und katalanische Verbprobleme. Ein Beitrag zur funktionellen synchronischen Untersuchung des Verbalsystems der beiden Sprachen (Tempus und Aspekt). Tübingen: Niemeyer. Schogt Henry G. 1964 “L’aspect verbal en français et l’élimination du passé simple”, Word 20: 1–17. Schumacher de Peña, Gertrud 1980 “El pasado en el español andino de Puno/Perú”, in: Hans-Dieter Bork, Artur Greive, Dieter Woll (eds.), Romanica Europaea et Americana. Festschrift für Harri Maier. Bonn: Bouvier Verlag Herbert Grundmann, 553–558. Schwenter, Scott A. 1994a “The grammaticalization of an anterior in progress: Evidence from a Peninsular Spanish dialect”, Studies in Language 18: 71–111.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

439

The Simple and Compound Past in Romance languages

439

1994b “«Hot news» and the grammaticalization of perfects”, Linguistics 32: 995–1028. Serrano, María José 1994 “Del pretérito indefinido al pretérito perfecto: un caso de cambio y grammaticalización en el español de Canarias y Madrid”. Lingüística española actual 16, 1: 37–57. 1995 “Sobre el uso del pretérito perfecto y pretérito indefinido en el español de Canarias: pragmática y variación”. Boletín de Filología de la Universidad de Chile 35: 527–560. Siadbei, ¸ Ion 1930 “Le sort du prétérit roumain”, Romania 56: 331–360. Skubic, Mitja 1964 “Pretérito simple y compuesto en el español hablado”, Linguistica 6: 87–90 1973–1975 “Le due forme del preterito nell’area siciliana”, Atti della Accademia di Scienze, Lettere e Arti di Palermo, serie IV, 33 parte II (Lettere): 225–293 and 34 parte II (Lettere):353– 427. Solarino, Rosaria 1991 “Cambia il tempo?”, Italiano & oltre 6: 141–146. Spitzová, Eva & Marcela Bayerová 1987 “Posición del perfecto compuesto en el sistema temporal del verbo en el español de México”, Études Romanes de Brno 18: 37–50. Stavinohová, Zdeˇnka 1978 Les temps passés de l’indicatif dans le français contemporain. Brno: Univerzita J. E. Purkynˇe. Sten, Holger 1973 L’emploi des temps en portugais moderne. Copenhagen: Munksgaard. Suter, Alfred 1984 Das portugiesische Pretérito Perfeito composto. Bern: Francke. Toscano Mateus, Humberto 1953 El español en el Ecuador. Madrid: Revista de Filología Española, Anejo 61. Travaglia, Luiz Carlos 1981 O aspecto verbal no português. A categoria e sua expressão. Uberlândia: Universidade Federal de Uberlândia. Tuttle, Edward F. 1986 “The spread of ESSE as universal auxiliary in Central Italo-Romance”, Medioevo Romanzo 11: 229–287. Vincent, Nigel 1982 “The development of the auxiliaries habere and esse in Romance”, in: Nigel Vincent & Martin Harris (eds.), Studies in the Romance verb. London & Canberra: Croom Helm 71–96. Wandruszka, Mario 1966 “Les temps du passé en français et dans quelques langues voisines”, Le Français moderne 34: 3–18. Waugh, Linda R. & Monique Monville-Burston 1986 “Aspect and discourse function: the French Simple Past in newspaper usage”, Language 62: 846–877. Westmoreland, Maurice 1988 “The distribution and the use of the Present Perfect and Past Perfect forms in American Spanish”, Hispania 71: 379–384. Zezula, Jaroslav 1969 “Le passé simple dans la langue de la presse française d’aujourd’hui”, Beiträge zur romanischen Philologie 8: 336–345.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

440

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

441

Hannu Tommola

On the perfect in North Slavic

1. Introduction It might seem a hopeless effort to describe a grammatical category in languages where this particular category does not exist. However, the absence of grammatical forms to express, for instance, perfective and imperfective aspect has not prevented functional research into aspect semantics in various languages where aspect is not acknowledged. Even if no verbal aspect, in a strict sense, can be found, findings in ways of expressing similar meanings may turn out to be important. A case in point is the aspectual value of the direct object case opposition in Baltic Finnic (Dahl 1985: 69, 72, 89; Tommola 1986). It is well known that in East and West Slavic the perfect “does not exist”. One line of argument followed in this paper is to show on the basis of empirical material – mainly supplied by the responses to the E UROTYP Perfect Questionnaire (PFQ) – by which means the cross-linguistically defined perfect meaning(s) is/are rendered. Another question that will be discussed, though without going into details, is what the meaning of the Perfect was in Proto-Slavic. In all modern East and West Slavic languages, with the exception of Upper Sorbian and literary Lower Sorbian, there is only one morphologically marked past tense, which is derived from both imperfective and perfective stems. The old Slavic compound Perfect form has in all these languages lost its specific function in the tense-aspect (TA) system due to the disappearance of the simple past tense forms (Imperfect and Aorist). The ancient perfect construction, deprived in most cases – either totally or in the nonthird person forms – of its original auxiliary verb, has taken over the functions of both, while it continues at the same time to include the perfect meaning.1 It is commonly felt that there was a perfect in Proto-Slavic, i.e., a compound tense form that is called Perfect, for example, in Old Church Slavonic grammar. What makes its decline difficult to follow is that it never disappeared. This paradox could lead to a situation characterized by Östen Dahl as a “doughnut category”: an aggressive gram eats up the functions of other grammatical forms expanding in use, leaving the kernel domain – the expression of the specific meaning that it was originally created for – uncovered. We have a doughnut category in North Slavic as soon as the old perfects (Imperfective and Perfective Pasts) can no longer be used to express typical perfect situations. So far, only in some Russian dialects has this kind of situation been discovered (see section 5 and conclusions).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

442

442

Hannu Tommola

Until recently, there was little evidence of the existence of a universal perfect category. A prototypical perfect gram-type was posited by Dahl (1985: 129), and more evidence was provided by Bybee et al. (1994: 51–87), who call the same gramtype “anterior”. Consider the definition (Bybee et al. 1994: 54): “[ – ] an anterior signals that the situation occurs prior to reference time and is relevant to the situation at reference time.” Most perfects are periphrastic (in Dahl’s sample 85 per cent). Among the grams found by Dahl there is a descendant of the ancient Slavic Perfect, the Bulgarian one. It is likely that this is what the perfect also was in those Slavic languages that have not preserved it as a functional category.2

2. Aspect morphology Aspect in North Slavic languages is provided on the lexical level, i.e., the imperfective/perfective opposition functions not only in the past, but also in the (present/) future and in the repertoire of nonfinite forms. In principle, aspect derivation is identical in all North Slavic languages. For example, in Modern Russian there is only one finite past tense form, which is derived from the infinitive stem with the suffix -l. As the form is an old participle, it is inflected in gender and number (Masc. -l, Fem. -la, Ntr. -lo, and Pl. -li). Because a lexical meaning is represented, as a rule, both by an imperfective (IPFV) and by a perfective (PFV) verb, there are two grammatically different past tense forms, Imperfective Past and Perfective Past, e.g.: (1)

a. Imperfective verb:

b. Perfective verb:

Infinitive: pisa-t’ (write-INF) ‘to write’ Present: piš-u, -eš’, -et, -em, -ete, -ut (write-PRS) ‘I write/am writing’ etc. Future: bud-u, -eš’, -et, -em, -ete, -ut pisa-t’ (AUX.FUT:PRS write-INF) ‘I’m going to write / I’ll be writing’ etc. Past: pisa-l (write-PST) ‘wrote, was writing, have/has/had written’

Infinitive: na-pisa-t’ (on-write-INF) ‘to write (up, ready)’ Present = Future: na-piš-u, -eš’ etc. (on-write-PRS) ‘I’ll write’ etc.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Past: na-pisa-l (on-write-PST) ‘wrote, have/has/had written’

443

On the perfect in North Slavic

443

Thus, we have past forms that can be analyzed as in (1¼ ): (1¼ ) a. b.

pisa-l na-pisa-l

/ (0=IPFV:write-PST) (on=PFV-write-PST)

i.e., the zero prefix in pisal can be taken as an imperfective marker compared with the prefixed napisal. However, imperfective verbs can be marked morphologically. Consider the following aspect pairs: (2) a.

Perfective pro-ˇcita-l (through-read-PST) ‘(have, has / had) read (through)’

b.

Imperfective pro-ˇcit-yva-l (through-read-IPFV-PST)3 ‘(have, has / had) read through / used to read through’

(3) a.

Perfective reš-i-l (solve-PFV-PST) ‘(have/has/had) solved’

b.

Imperfective reš-a-l (solve-IPFV-PST) ‘was, were solving / (have, has / had) solved / tried to solve’

In this paper nonprefixed verbs are glossed for aspect only when perfective (PFV), the default being imperfective (IPFV). This is also the case with (nonproductive) suppletive pairs where – at least synchronically – neither of the two verbs can be regarded as derived (neither a prefix nor a suffix can be distinguished); e.g., vzjat’ (take:PFV) will be marked for aspect, whereas the imperfective brat’ (take) is left unmarked. Perfective verbs derived from a nonperfective with the help of a momentative suffix could be glossed as “semelfactive”, e.g., pryg-nu-l (jump-SMLF-PST) as opposed to the unmarked imperfective pryga-l (jump-PST). However, they will be glossed simply as perfectives (pryg-nu-l (jump-PFV-PST)).4

3. Formal development from a compound perfect towards a simple past There are differences in the Past tense (i.e., the old Perfect) morphology regarding the use of the auxiliary, the extreme cases being the East Slavic languages (where it never occurs; see Belarusan below) on one hand and Sorbian on the other. The auxiliary is always used in both Sorbian languages, where this tense is called the Perfect, e.g.:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

444

444

Hannu Tommola

(4) a.

b.

Lower Sorbian Som/Smy choj´zi-ł /chojzi-l-i po gol-i. COP:1SG/1PL walk-PP.SG /-PL along forest-DAT ‘I/We were walking / walked in the forest.’ Upper Sorbian Šule su lˇepše by-l-i haˇc nˇetko. school:PL COP:3PL better COP-PP-PL than now ‘The schools were better than nowadays.’

In Czech, the auxiliary verb is used in the 1st and 2nd person forms, while the personal pronoun is dropped (5a,b), unless it is focused (5d). Also, the third person pronoun may be omitted (5c) (however, this is not a compound past specialty, because subject personal pronouns are normally dropped in Czech). The second person form of the auxiliary jsi (COP:PRS) is often cliticized and attached as an -s to the personal (5d) or reflexive pronoun (5b), to the main verb or other words, such as conjunctions and interrogatives. (5) a.

b.

c.

d.

Czech Šl-i jsme les-em. go:PP-PL COP:1PL forest-INST ‘We were walking / walked in the forest.’ U-dˇela-l si-s úkol? PFV-do-PP RFL-COP:2SG assignment ‘Did you do your homework?’ Tuhle knih-u do-ˇcet-l-a. this:ACC.F book-ACC PFV-read-PP-F ‘She has finished this book.’ Ty-s otvíra-l okno? PRON.2SG-COP:2SG open:IPFV-PP window ‘Did you open the window?’

In Polish the first and second person forms of the auxiliary are regular clitics that are combined with the main verb form ending in -l.5 Thus there is a new inflectional past paradigm with bound personal morphemes. This also means that Polish deviates from the other Slavic languages in having full person agreement also in the Past, and in that there is no gender agreement in the nonthird person forms: (6) a. b. c. d.

Polish Pozna-ł-em ja.˛ Odrobi-ł-e´s ju˙z lekcje? On(-a) czyta-ł(-a) t˛e ksia˙ ˛zk˛e. Chodzi-l-i-´smy po lesie.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

‘I’ve met her.’ ‘Have you done the homework?’ ‘He/She has read this book.’ ‘We walked in the forest.’

445

On the perfect in North Slavic

e. f.

Chodzi-l-i-´scie po lesie. Szkoły by-ł-y lepsze ni˙z teraz.

445

‘You walked in the forest.’ ‘The schools were better than now.’

In Rusyn,6 the auxiliary of the 1st and 2nd person seems to be in complementary distribution with the corresponding personal pronouns – cf. (7a) with (7b) – except that the 3rd person pronoun can be omitted, although there is no auxiliary (7c), similar to Czech: (7) a.

b.

c.

Rusyn Šepta-l-y zme še po lješ-e. walk-PP-PL COP:1PL RFL Prep forest-DAT ‘We walked in the forest.’ Ja kupe-l rezervn-y cˇ asc-y. I buy:PFV-PP spare-PL part-PL ‘I have bought spare parts.’ Ja duma-m že ho toteraz za-konˇce-l. I think-1SG that PRON:ACC by_now PFV-finish-PP ‘I think he has finished it by now.’

Finally, as an example of the East Slavic situation, consider Belarusan with a unified system, deprived of any analytic forms: (8) a.

b.

Belarusan Ja/Ty wžo atryma-l-a zamežny pašpart. PRON.1/2SG already get:PFV-PST-F foreign passport ‘I/You (have) got my/your passport already.’ My/Vy/Jany pra-gljadze-l-i kambinacy-ju. PRON.1/2/3PL through-look-PST-PL combination-ACC ‘We/You/They (have) overlooked the manoeuvre.’

4. Perfect meanings in the EUROTYP Perfect Questionnaire data 4.1. Resultative As a rule, all North Slavic languages have the Perfective Past in the items that can be labelled as contexts for a perfect of result, i.e., where current relevance of bounded situations or completed actions is referred to. In most of the languages of the PFQ sample in items PFQ: 27–29 the Perfect can be used, but in PFQ: 28 resultative or stative constructions are preferred.7 Also in North Slavic languages resultative (Russian) or stative (Belarusan, Ukrainian, Czech, Polish, Rusyn, Upper Sorbian) constructions are quite frequent (10). Bulgarian and some variations of Macedonian tend to use the Aorist in PFQ: 27 (cf. also the English Past in 9).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

446

446

Hannu Tommola

(9) a. b. c.

(10) a. b. c.

‘No, she’s already left. / No, she already left.’ (PFQ: 27) Upper Sorbian Nˇe, wona je hižo wotešła. (PFV:PF) Rusyn Nje, vona už pošla (PFV:PST). Czech Ne, už je pryˇc. (COP:3SG away) / Ne, už tu není. (NEG:COP) ‘No, she’s still gone. / No, she is still away.’ (PFQ: 28) Czech Ne, ještˇe se nevrátila. (PFV:PST) Belarusan Ne, jana vsë jašˇcè w ad’ezdze. (PREP-departure) Russian (dial.) Net, ona ešˇcë ušedš-aja. (leave:PFV:ACT.PP-F)

The Czech example in PFQ: 30 (11) is a stative construction that corresponds, for example, to the Finnish one, and can be seen in contrast to the Perfective Past in PFQ: 31 (12), where it denotes a resultative event in immediate past.8 (11) a. b. (12)

‘He is already awake. / He has already woken up.’ (PFQ: 30) Czech Už je vzhuru. ˚ (already COP awake) Finnish Se on jo hereillä. (PRON.3SG COP already awake) Czech (PFQ: 31)

Už se probudila! (PFV:PST) ‘Oh no! She has already woken up.’

Many languages use the Perfect in all the items PFQ: 42–45 (Armenian, Dutch, British English, Modern Greek, Lower Sorbian, Italian, Norwegian, Swedish, etc.), whereas there are languages that allow resultative constructions in all the other items, but not in PFQ: 45 (14). Possessive constructions would not be expected here as the lexical meaning of the main verb is ‘lose’. Nevertheless, five Macedonian informants use the imam Perfect here, which is quite good evidence of its grammaticalization as a perfect (see Graves, this volume). (13)

Czech (PFQ: 42) Má-m na-psá-n-o/na-psa-n-ých 50 stran. have-1SG PFV-write-PPP-NT/GEN.PL 50 page:GEN.PL ‘I have written fifty pages.’

(14)

Russian (PFQ: 45) V ètom godu ja poterjal (PFV:PST) pjat’ zontikov.9 ‘This year I have lost five umbrellas.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

447

On the perfect in North Slavic

447

4.2. Anterior continuing In most Slavic languages the (Imperfective) Present tense is used in the function that has been called “perfect of persistent situation” (Comrie 1976: 60), “inclusive perfect” (Tommola 1986: 48–49, 51, 58 n. 35, 1993: 137 fn. 6), and, from a crosslinguistic point of view, more adequately “anterior continuing” (Bybee et al. 1994: 62). There are, however, minor differences in treating the situation denoted by PFQ: 47 (15) with reference to several occasions and that denoted by PFQ: 48 (16), where an activity is going on at the reference time. In the latter case, most languages choose the present (simple or progressive), and the perfect (in English the Perfect Progressive) is represented only in the typical “perfect languages”, such as Norwegian, Swedish, Finnish, and Estonian.10 In Italian both the Perfect and the Present are possible. Lower Sorbian follows the perfect line, too. On the contrary, in PFQ: 47 the use of the present clearly occurs in the minority of the languages (Hungarian, Turkish), but some of the Slavic languages (Polish and Russian) make use of it. (15)

‘I haven’t slept for three days.’ (PFQ: 47) a. b. c. d.

(16)

Lower Sorbian Russian (dialectal) Rusyn Polish

Njejsom spał p´sez t´si nocy. (IPFV:PF) Ja tri dnja ne spamši. (IPFV:GER.PST) Nje spal som try nocy. (IPFV:PST) Od trzech dni nie sypiam (nie s´pi˛e). (IPFV:PRS)

‘She’s watched / been watching it (TV) for three hours.’ (PFQ: 48) a. b.

Lower Sorbian Belarusan11

Wona jo glˇedała t´si go´ziny zasobu. (IPFV:PF) Jana hljadzic’ jaho vos’ užo try hadziny. (IPFV:PRS)

The situation is similar in PFQ: 49, where, in North Slavic, the present dominates (see, however, the Lower Sorbian and Ukrainian informants’ choices in (17a) and (17b)). Still more languages use the perfect (in North Slavic, the Past), when the continuing situation comprises all the individual past of the speaker (18): in Italian there is no longer a choice, and German, Bulgarian and Sorbian also use the perfect. A resultative construction was recorded in dialectal Russian, and a Perfective Past in Ukrainian. All the other North Slavic languages follow the pattern exemplified by the Ukrainian variant in (18d). (17) a. b. c.

‘I’ve lived / been living here for seven years.’ (PFQ: 49) Lower Sorbian (Ja) Som był žywy tudy sedym lˇet. (PF) Ukrainian Ja prožyv tut (uže) sim rokiv. (PFV:PST) Upper Sorbian Ja bydlu tu hižo sydom lˇet. (IPFV:PRS)

a.

‘I’ve lived / been living here all my life.’ (PFQ: 50) Lower Sorbian Som był žywy tudy p´sez mojo cełe žywjenje. (PF)

(18)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

448

448

Hannu Tommola

b. c. d.

Russian (dial.) Ukrainian Ukrainian

Ja vsju žizn’ zdes’ proživši. (PFV:GER.PST) Ja prožyv tut vse svoje žyttja. (PFV:PST) Ja žyvu tut vse svoje žyttja. (IPFV:PRS)

4.3. Experiential vs. habitual and sequenced past PFQ: 51 (19) is contrasted to PFQ: 50 (18) in that a habitual/experiential meaning is involved, and the situation does not reach up to the time of speech. Therefore presents are excluded, but pasts or specific habitual/experiential constructions or forms can be used instead of perfects.12 (19) a.

b.

(Imperfective) Perfect (PFQ: 51) Norwegian Jeg har bodd her, så jeg kjenner hver gate. Upper Sorbian Ja sym tu jónu bydlił, tohodla znaju tu kóždu dróhu. (Imperfective) Past (PFQ: 51) Icelandic Ég átti hér heima, svo ég ekki hverja einustu ögtu. Polish Mieszkałam tu, wi˛ec znam tu ka˙zda˛ ulic˛e. ‘I lived / used to live here so I know every street here.’

An event that is specified as to the time it happened appears in an overt contrast to experiential. Among the languages where the choice between the perfect and the past seems to be made on the basis of nonspecific vs. specific time reference are the Scandinavian languages, Baltic Finnic, English, Modern Greek, and Bulgarian (i.e., “perfect languages”). Compare with (20a) and (20b): (20) a.

b.

Perfect: Icelandic (PFQ: 33b) angað. Já, ég hef far-ið yes I have:1SG travel-PP there ‘Yes, I’ve been there.’ Past: Icelandic (PFQ: 33d) angað í janúar 1987. Já, ég fór in January 1987 yes I travel:PST there ‘Yeah, I was there in January 1987.’

This kind of a clear distinction between the expressions of an experiential and a specific event meaning was not found in any North Slavic language. In Czech, Polish, Lower Sorbian, Rusyn and Belarusan there is no variation. There is, however, a problem with the copula verb. Other verbs could render a difference, as ‘meet’ in Russian (see below, 50). Here the variation found in Ukrainian, Russian and Upper Sorbian is optional and very vague. Anyhow, consider (21) with at least different preference for the verbs employed.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

449

On the perfect in North Slavic

(21) a.

b.

449

‘Yes, I’ve been there.’ (PFQ: 33b) Ukrainian Da, ja buvav / buv tam. (IPFV:HAB.PST/IPFV:PST) Russian Da, ja byval / byl tam. (IPFV:HAB.PST/IPFV:PST) Upper Sorbian Haj, sym tam hižo był(a) / po-był(a). (IPFV/PFV:PF) ‘Yeah, I was there in January 1987.’ (PFQ: 33d) Ukrainian Da, ja buv tam v siˇcni 1987 roku. (IPFV:PST) Russian Da, ja byl tam v janvare 1987 g. (IPFV:PST) Upper Sorbian Haj, sym tam w januarje 1987 (po)był(a) / Bˇech tam w januarje 1987. (PFV/IPFV:PF / PST)

4.4. Past perfect The lack of Perfect tenses may cause obvious ambiguity, notably with past and future time reference, as is seen in the Rusyn examples (22) and (30). (22) a.

b.

Rusyn (PFQ: 75) Ked som pryšol domu vˇcera, moja šestra napysala (PFV:PST) dva pysma. ‘When/After I came home yesterday, my sister had written / wrote two letters.’ Ked som pryšol domu vˇcera, moja šestra pysala (IPFV:PST) dva pysma. ‘When I came home yesterday, my sister had written / been writing two letters.’

In other Slavic languages there are various strategies to tackle the problem. A stative (possessive) construction is found in Czech (23a), accompanied with the relational adverb ‘already’, which also appears in Russian, or the temporal clause is embedded in a temporal adverbial (23b). The utterance is also disambiguated by a temporal conjunction (‘before’), various adverbs, or by adding a completive verb (23c). (23)

(PFQ: 75) a.

b.

Czech Když jsem vˇcera pˇrišla domu, moje sestra mˇela už napsané dva dopisy. lit. ‘When I came home yesterday, my sister had already two letters written.’ Russian Do moego prixoda moja sestra napisala dva pis’ma. lit. ‘Before my arrival my sister (had) written / wrote two letters.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

450

450

Hannu Tommola

c.

Polish Gdy przyszedłem do domu, moja siostra sko´nczyła wła´snie pisa´c dwa listy. lit. ‘When I came home, my sister (had) just finished writing two letters.’

In Upper Sorbian there is a real Pluperfect (called “Plusquampräteritum” by Faßke 1981: 252, 265–266) that involves the auxiliary in the Imperfect form of the copula. Besides, there is also a narrative compound tense (used also as a conditional) called “Iterativpräteritum” (Faßke 1981: 253, 266–267) that is built with the copula in the Aorist form. Both are illustrated in (26): (24)

Upper Sorbian (PFQ: 75) Hdyž sym ja wˇcera domoj pˇrišoł, bˇe moja sotra runje dwaj listaj napisała. ‘When I came home yesterday, my sister had just written two letters.’

(25)

Upper Sorbian (PFQ: 76) Ja njejsym ju namakał/nadešol. Wona bˇe wotešła. ‘I didn’t find her. She had left.’

(26)

Upper Sorbian (PFQ: 79) Pˇreco hdyž by-chmoj so tehdy zetka-ł-oj, always when COP:AOR.1DU RFL then meet:PFV-PP-DU by mi wón wo film-je powˇeda-ł, kotryž COP:AOR I:DAT he PREP film:LOC tell:PFV-PP REL bˇe runje wid´za-ł. COP:IMPF just see:IPFV-PP ‘Every time I met him in those days, he told me about the film he had just seen.’

In some of the PFQ items, the lexical verb that is chosen by the informant may affect the tense and aspect use. In PFQ: 77, for example, in German kennenlernen ‘get to know’ is used for ‘meet’. Thus, in (27) the Bulgarian alternative (27a) is an experiential, and (27b) a resultative Pluperfect that corresponds to the German usage:13 (27) a. b.

(PFQ: 77) Bulgarian Bulgarian

c.

German

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Da, bjax ja sreštal. (meet:IPFV:PLUP; ‘had met’) Da, bjax se za-poznal s neja. (PFV-know:PLUP; ‘had got to know’) Ja, ich hatte sie schon kennengelernt. (knowlearn:PLUP; ‘had got to know’) ‘Yes, I had met her.’

451

On the perfect in North Slavic

451

The pluperfect is difficult to get in PFQ: 81 in contradistinction to PFQ: 80, where the present perfect is supposed to be used, because many languages, like English, would not use the Perfect tenses at all with a specific result. One can actually make a Russian informant differentiate the uses in (28) and (29) and choose different aspect forms, but it is hardly done spontaneously:14 (28)

(PFQ: 80) [Looking at a house] a. b.

Czech Russian

a. b.

(PFQ: 81) [Looking at a picture of a house which has been torn down] Czech Kdo ten d˚um postavil (PFV:PST)? Russian Kto stroil (IPFV:PST) ètot dom? ‘Who [had] built this house?’

(29)

Kdo postavil (PFV:PST) ten d˚um? Kto postroil / stroil (PFV/IPFV:PST) ètot dom? ‘Who built this house?’

4.5. Future perfect The anterior future (future perfect) situation is similar to the anterior past (pluperfect). The Rusyn example (30a) is ambiguous in the same way as (22) above. In Russian uže ‘already’ is used in (30b) as a disambiguator: (30) a.

b.

(PFQ: 84) Rusyn [Ked se vracys narok,] preda-m svo-ju xyž-u. sell:PFV-PRS.1SG POSS-ACC.F house-ACC Russian [Kogda ty vernëš’sja domoj,] ja uže proda-m moj dom. I already sell:PFV-1SG my house ‘[When you come back next year,] I will have sold my house.’

It is noteworthy that possessive or stative constructions that were used in Czech, Polish, Russian and Upper Sorbian do not suffice to express the difference, but in all the languages ‘already’ was added (see 31). (31)

Polish (PFQ: 84) [Kiedy wrócisz w przyszłym roku,] mój dom b˛edzie ju˙z my house COP.FUT already sprzeda-ny. sell-PPP lit. ‘[When you come back next year,] I will have my house sold already.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

452

452

Hannu Tommola

The Sorbian languages can use here the Perfect: (32)

Upper Sorbian (PFQ: 84) Hdyž ty klˇetu zaso pˇri´nd´zeš, sym swoju chˇezˇu hižo pˇredal. ‘When you come back (next year), I will have sold my house (already).’

(33)

Lower Sorbian (PFQ: 82) Ja njewupłasim tebje twojo myto, nježli sy doko´ncował twoj ceły nadawk. lit. ‘I won’t pay you until you have finished the entire job.’

The meanings can also be disambiguated by proper conjunctions, like Polish dopóki ‘until’ in (34) (similarly in Russian and Czech, in negated sentence), but in (35) the anteriority is equal to immediate consequence, and a neutral temporal conjunction can be used: (34)

Polish (PFQ: 82) Nie zapłac˛e ci, dopóki nie sko´nczysz pracy. ‘I won’t pay you until you finish the entire job.’

(35)

Polish (PFQ: 83) Zapłac˛e ci jak sko´nczysz prac˛e. ‘I will pay you when you have finished the entire job.’

The persistent situation in future is expressed by the usual present Perfect as in Lower Sorbian (36a), by the Present tense (“anterior continuing posterior”) as in Upper Sorbian (36b), by the imperfective Future as in Czech (36c), or by a typical Slavic construction COP:FUT  ‘a period of time’  CONJ (‘as’) + PRS (illustrated by the Rusyn example in (37); similarly in Czech and Russian): (36)

‘By June this year I will have worked here for 30 years.’ (PFQ: 85) a. b. c.

(37)

Lower Sorbian Upper Sorbian Czech

Lˇetosa we juniju som z´eˇ łał how p´sez t´si´zas´ca lˇet. Lˇetsa w juniju d´zeˇ łam ja tu hižo tˇrice´ci lˇet. V cˇ ervnu tady budu pracovat 30 let.

Rusyn (PFQ: 85) U junu toho roku budze trycec roky jak robym tu. lit. ‘In June this year it will be 30 years that/how I work here.’

In Lower Sorbian the present Perfect is also used to express a resultative event in future, whereas in other languages usually the perfective Future is used (notice, however, the idiomatic Imperfective Future in Czech in PFQ: 86):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

453

On the perfect in North Slavic

(38) a. b. c. (39) a. b.

453

‘If I get my wages tomorrow, I will buy you a beer.’ (PFQ: 86) Lower Sorbian Gdyž som wit´se dostał (PFV:PF) swojo myto, ga kupim tebje piwo. Upper Sorbian Jeli jutˇre swoju mzdu dóstanu (PRS:PFV), wudam c´ i piwo. Czech Jestli budu zítra brát, pozvu tˇe na pivo. (IPFV:FUT) ‘The day I get my wages, I will buy you a beer.’ (PFQ: 87) Lower Sorbian Ten z´ e´n, na kotaryž som dostał (PFV:PF) swojo myto, kupim tebje piwo. Czech Až dostanu (PFV:PRS) plat, pozvu tˇe na pivo.

4.6. Evidentiality In some parts of South Slavic, out of the Perfect an indirective evidential form has developed (see Lindstedt, this volume). As mentioned in section 5.1, the old Perfect seems to mark indirect evidentiality in opposition to the narrative pasts in Upper Sorbian (PFQ: 60). However, where the perfect has become a universal past, it cannot have this function. In all North Slavic languages different lexical means are used instead.15 There are also two occurrences of future or modals with an inferential meaning. The Upper Sorbian zmˇeje in (40) is a Perfective Present, hence future (Faßke 1981: 268), though with present time reference, and the Czech modal auxiliary ‘must’ in Past tense has epistemic meaning in (41): (40)

Upper Sorbian (PFQ: 70) Nˇe, wón drje z-mˇe-je je hižo hotow-e. no he sure PFV-have-3SG PRON:ACC.PL already ready-PL ‘No, he’ll have finished it [them] already.’

(41)

Czech (PFQ: 71) To muse-l-o být velk-é mˇesto. this must-PST-NT COP:INF big-NT town ‘This was [must have been] a huge city.’

5. North Slavic data in the EUROTYP Perfect Questionnaire In all North Slavic languages the following grammatical verb forms were used in PFQ:16

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

454

454

Hannu Tommola

1. Perfective Past, roughly speaking, to denote a bounded situation (including resultative contexts, e.g., items PFQ: 03, 05, 16, etc.); additionally, it is used in narrative contexts to denote completed events (e.g., PFQ: 08–13). 2. Imperfective Past to denote nonbounded past situations (including experiential contexts, e.g. PFQ: 01, 32–34, 51, etc.); additionally, in narrative contexts to denote ongoing activities or to describe the background (e.g., the first sentences in PFQ: 08–11, and the clause expressing the setting in PFQ: 12–13). 3. Perfective Present to denote events in future (including anterior future, e.g., in PFQ: 82–84, 86–88); 4. Imperfective Present to denote an activity or state going on at reference time (including anterior continuing contexts, e.g., in PFQ: 48–50).17 Thus, in the prototypical perfect sentences PFQ: 01–PFQ: 07 in Russian, Ukrainian, Belarusan, Czech, Polish, and Rusyn either the Imperfective or the Perfective Past tense is used. They are also used in a narrative context (PFQ: 08–13), hence they are clearly not perfect grams. In most North Slavic languages the following types of periphrastic constructions also occurred: 5. stative constructions with COP + adverb (e.g., ‘away’, ‘awake’, etc.); 6. state passive constructions with (or without) COP + passive past participle; 7. possessive constructions a) with the auxiliary ‘have’ + passive past participle, or b) with COP + passive past participle and a prepositional expression indicating the subject (in Russian, PFQ: 42–44); 8. (COP +) active past participle construction (as a dialectal alternative in Russian); 9. (COP +) past gerund (as a dialectal alternative in Russian).

5.1. Narrative context and the question of the Sorbian Perfect While in the other North Slavic languages the Past (historically a perfect) is the only past tense, in grammars of both Sorbian languages the old perfect form is still called Perfect, because simple past tenses also exist. However, in Lower Sorbian data no simple past form was found (the Perfect also being used in PFQ: 08–13). Evidently, of these descendants of the Slavonic Perfect only the Upper Sorbian compound form could possibly be a perfect gram. But it does not seem to be the case because it is not out of question in a narrative context – at least in spoken communication. In Upper Sorbian the simple past forms (Aorist and Imperfect) are preferred, but not compulsory, in narrative (42). In PFQ: 10 (when the story is not about the speaker him/herself) the Perfect is at least equally likely (43), and in indirective function (PFQ: 60) the only choice, whereas the simple past forms seem to be obligatory in the fairy tale item (44 PFQ: 61).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

455

On the perfect in North Slavic

455

(42)

Upper Sorbian (PFQ: 08) Pˇre-chod´z-ow-ach so po lˇesu. Nadobo Prep-walk-IPFV-IMPF.1SG RFL along forest-DAT suddenly stupi-ch na had-a. Tón mje do noh-i step:PFV-AOR.1SG on snake-GEN PRON I-DAT to foot-GEN kus-ny. Ja wza-ch kamje´n a bite-PFV:AOR.3SG I take:PFV-AOR.1SG stone and c´ is-ny-ch jón na had-a. A tón throw-PFV-AOR.1SG PRON on snake-GEN and PRON slak-ny. die-PFV:AOR.3SG ‘I was walking in the forest. Suddenly I stepped on a snake. It bit me in the leg. I took a stone and threw it at the snake. It died.’

(43)

Upper Sorbian (PFQ: 10) Smój (COP:1DU) so po lˇesu pˇrechod´zowa-ł-oj (IPFV-PP-DU). Nadobo je (COP) wón na hada stupi-l (PFV-PP). Tón je (COP) jeho do nohi kusny-ł (PFV-PP). Tuž je (COP) wón kamje´n wza-ł (PFV-PP) a jón na hada c´ isny-ł (PFV-PP), tón je (COP) slakny-ł (PFV-PP). ‘We were walking in the forest. Suddenly he stepped on a snake. It bit him in the leg. He took a stone and threw it at the snake. It died.’

(44)

Upper Sorbian (PFQ: 61) Bešˇe (COP:IMPF.3SG) pak nˇehdy muž. Tón wuchod´zowa-še (IPFV-IMPF. 3SG) so po lešu. Nadobo stupi (PFV:AOR) wón na hada. Tón kusny (PFV:AOR) jeho do nohi. Wón wza (PFV:AOR) kamje´n a c´ isny (PFV: AOR) jón na hada. A tón slakny (PFV:AOR). ‘Once upon a time there was a man. He was walking in the forest. Suddenly he stepped on a snake. It bit him on the leg. He took a stone and threw it at the snake. It died.’

The simple past forms seem to be out of place in indirect narration (PFQ: 60), where they are replaced by the Perfect. But the Perfect has not only taken over the function of an indirective narrative tense. As the Perfect can be used in direct narration (for example in 43), it is not possible to regard it as a perfect gram any more. It has become a universal past, while the simple past tenses are rather a stylistic device that is required in certain literary genres (44). In the analysis of the Sorbian Aorist and Imperfect, some terminological problems arise. The traditional (historical) analysis differentiating tense and aspect does

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

456

456

Hannu Tommola

not have much sense any more, because the Aorist endings (Sg.: -/e/ch, -0/e; Pl.: /e/chmy, -/e/š´ce, -/e/chu; Dual: -/e/chmoj, - /e/štaj/-ej) are consistently used with Perfective verbs only, and the Imperfect endings (Sg.: -/e/ch/-ach, -/e/še; Pl.: -/e/chmy/achmy, -/e/š´ce, -/e/chu/-achu; Dual: -/e/chmoj/-achmoj, -/e/cštaj/-ej) with the Imperfective verbs only (Šewc 1968: 171–172). That is, the forms are unambiguous only in the 2nd and 3rd Sg. (and even there they are automatically determined by the aspect). In fact, the -a- of the Imperfect endings of some verb classes in the 1st Sg. and Pl., and 3rd Pl. can be reinterpreted as belonging to the suffix of secondary Imperfectives. Thus the forms analysed in (42) above as AOR and IMPF can result in an alternative analysis as in (45): (45)

pˇre-chod´z-owa-ch stupi-ch kusny-0/ wza-ch c´ isny-ch slakny-0/

PREP-walk-IPFV-PST.1SG step:PFV-PST.1SG bite:PFV-PST.3SG take:PFV-PST.1SG throw:PFV-PST.1SG die:PFV-PST.3SG

Indeed, the Aorist and the Imperfect are described together in Šewc (1968: 178) as synthetic past forms. Faßke (1981: 251–252, 261–265 etc.) does not use these terms at all and speaks instead of the Past (“Präteritum”). As can be seen from the responses to the PFQ, in Lower Sorbian the simple past forms have practically disappeared in everyday communication. Janaš (1976: 321) states that in colloquial speech the Perfect can replace the past tenses (“Präterita”) in all functions that they have in the literary language. Upper Sorbian seems, if not to retain better the semantic distinction between the Perfect and the simple forms, to preserve the simple forms as a narrative technique and not to use them instead of the Perfect, although the reverse is common in colloquial speech (Faßke 1981: 262). The Pluperfect use is still consistent (see above), and it is worthwhile to note that the adverb hižo ‘already’, unlike the perfect contexts, is not required then (see below).

5.2. Imperfective Past – experiential? The distribution of the North Slavic past grams in the prototypical perfect sentences is roughly such that Imperfective Past is used in PFQ: 01 (= 46) and PFQ: 07, and Perfective Past in PFQ: 02 through PFQ: 06 (see 47–49). Since PFQ: 01 and PFQ: 07 are experiential items, one could tentatively maintain that experientiality is among the functions of the Imperfective Past in the East and West Slavic languages. This is not quite the case, because the experiential need not co-occur with the Imperfective aspect, although it often does. In fact, the North Slavic languages choose different

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

457

On the perfect in North Slavic

457

aspects in the items where experiential meaning is denoted, and the reasons are often lexically motivated. Besides, in West Slavic the Perfective Past does not exclude an iterative reading, and is therefore suitable in experiential contexts, unlike the Perfective Past in East Slavic. The difference between PFQ: 01 and PFQ: 02 is not one between an experiential and a resultative perfect, because both refer to experiential situation; the latter simply emphasizes the completedness of a particular activity.18 As the only North Slavic language Rusyn uses the Perfective Past in both items: (46) a. b. c. (47)

‘Yes, she has read this book.’ (PFQ: 01) Icelandic Já, hún hefur lesið essa bók. (PF) Polish Tak, ona czytała t˛e ksia˙ ˛zk˛e. (IPFV:PST) Rusyn Tej, vona preˇcytala totu knïžku. (PFV:PST) ‘She has read this one (all the way through).’ (PFQ: 02)

a. b. c.

Icelandic Hún er búin að lesa essa bók. American English She actually read this book. Belarusan Jana praˇcytala vos’ hètuju knihu. (PFV:PST)

East Slavic differs from the West Slavic languages in that Imperfective Past is also used in PFQ: 04 and PFQ: 06.19 PFQ: 04 (48) is also one of the most typical perfect contexts, and in PFQ: 06 (49) only American English differs from the pattern. (48) a. b.

‘Have you met my sister?’ (PFQ: 04) Belarusan Vy sustrakalisja z maëj sjastroj? (IPFV:PST) Czech Setkal ses s mou sestrou? (PFV:PST)

a. b.

‘Yes, I’ve met her.’ (PFQ: 06) Russian Da, ja vstreˇcal eë. (IPFV:PST) Rusyn Upoznal som ju. (PFV:PST)

(49)

In PFQ: 35 the experiential vs. specific event contrast may be rendered by the aspect opposition, as in Russian and Ukrainian. In (50b), in Russian, both aspects are possible – exactly as in Finnish, with both the Perfect and the Past. In the other North Slavic languages there is little or no variance here. The lexical verb (‘meet’) also seems to be treated differently in different languages (see 27 and note 10). In Polish and Czech as well as in Upper Sorbian, the verb used here is invariantly Perfective,20 in Lower Sorbian the mention of a single event (‘once’) seems to trigger the Perfective aspect. In Belarusan, where the verb used in all sentences is said to be imperfective, one might ask whether it has an aspectual value at all, rather than just being inherently resultative.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

458

458

Hannu Tommola

(50) a.

b.

c.

‘No. I’ve never met her.’ (PFQ: 35a) Polish Nie nigdy jej nie spotkałem. (PFV:PST) Ukrainian Ni, ja nikoly ne zustriˇcav jiji. (IPFV:PST) ‘Yes, I met her once.’ (PFQ: 35b) Russian 1 Da, ja vstreˇcal (IPFV:PST) eë odin raz. Russian 2 Da, ja odnaždy vstretilsja (PFV:PST) s nej. ‘Yes, I actually met her in January of 1987.’ (PFQ: 35c) Belarusan Tak, ja sustrakawsja (IPFV:PST) z ëj u studzeni 1987 h. Ukrainian Da, ja zustriv (PFV:PST) jiji v siˇcni 1987 roku.

In PFQ: 52 and PFQ: 53 a past period is specified as to its length. The difference between these two contexts is crucial for the perfect use: in the former item (51) some “perfect languages” (for example, Baltic Finnic) can choose either the Perfect or the Past,21 whereas in the latter (52) the past period is sequenced together with a following event that concludes the past situation, and the use of the Perfect is excluded.22 The Perfective Past is used in all East Slavic languages (52c). (51) a. b.

(52)

‘I lived here for seven years, so I know every street here.’ (PFQ: 52) Lower Sorbian Som how žywy był nˇekak sedym lˇet; znajom how kuždu wulicu. (PF) Russian Ja (pro)žil (PFV/IPFV:PST) zdes’ sem’ let; znaju každuju ulicu. ‘I lived here for seven years, but then I had to move away.’ (PFQ: 53)

a.

Lower Sorbian

b.

Polish

c.

Belarusan

Som how žywy był sedym lˇet, ale potom som musał (to mešto) wopuš´ci´s. (PF) Mieszkałam (IPFV:PST) tu przez siedem lat, ale musiałam si˛e przeprowadzi´c. Ja pražyla (PFV:PST) tut sem hod, ale potym pavinna byla peraexac’ u druhoe mesca.

6. On the meaning of the Perfect in Old East Slavic Russian linguists often maintain that, while there are no overt perfect forms in Russian, there is a “perfect meaning” (Rus. perfektnoe znaˇcenie), but the conditions of the appearance of this meaning remain quite vague. Usually the lexical semantics of the verb is seen as an important factor; thus Past tense forms from such Perfective verbs as potolstet’ ‘get fat’, poxudet’ ‘get meagre’, postaret’ ‘grow old’, pomolodet’ ‘grow young(er) (again)’ etc., are said to have exclusively a perfect meaning, whereas the imperfective verbs should not have this meaning, in general. Only

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

459

On the perfect in North Slavic

459

some cognitive imperfective verbs that can be conceived of as resultative when used in the past tense are accepted as having, potentially, the perfect meaning. Current examples are on mnogoe videl ‘he has seen a lot’, ja slyšal ‘I have heard’ etc. (see, for example, Bondarko 1980: 612). Thus, what is identified as the perfect meaning is the resultative proper. However, it can be shown that, firstly, the resultative perfect reading is not necessary for the perfective verbs mentioned above. Secondly, this particular resultative meaning is, of course, not restricted to verbs that denote gradual change. Thirdly, current relevance of a past event or situation at the reference time does not exclude nonresultative situations. Consequently, imperfective verbs should be possible in this meaning without the lexical limitations referred to above. The conception of perfect as essentially resultative has affected the treatment of the Old East Slavic (OES, usually called Old Russian) tense system. Frequently only perfects of result were regarded as correct use of the Perfect form in OES texts. The use of nonresultative verbs is simply taken as a sign of degeneration of the Perfect category. In a chapter devoted to a comparison of temporal systems in Russian and Finnish in Tommola (1986: 35–39) some remarks were made about the so-called perfect meaning, and this rather skewed interpretation of the facts in OES texts was criticized (see also Tommola 1993: 139). Examples (53) and (54) are from Ivanov (1982: 101–102), used there to illustrate the decline of the Perfect in OES. Ivanov argues that in (53) “the old man’s adolescence is totally in the past”, and that (54) simply is “a statement of a past fact”. He does not recognize the current relevance of the situations in experiential past (53), and in recent past (54) as perfect meanings, obviously because no proper “result” can be pointed out. (53)

Old East Slavic star’ muž’ unoš-eju by-l’ est’, unoša že ne vˇest’ old man youth-INST COP-PP COP youth but NEG know ašˇce do-id-et’ starost-i if to-go-3SG old_age-GEN ‘An old man has been young, but a young man doesn’t know if he will reach old age.’

(54)

Old East Slavic cˇ ast’ post-a preminu-l-a est’ part fast-GEN pass:PFV-PP-F COP ‘A part of the fast-time has passed.’

The process of a reorganization of the tense system must have been going on already in early OES, because the Imperfect and the Aorist disappeared by the beginning of the 14th century. Nevertheless, these kinds of examples do not prove that the old Perfect form had lost anything of its original meaning. They are manifestations of a regular meaning of perfects.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

460

460

Hannu Tommola

Niissalo (1994) has found some evidence for a different status of the Perfect forms with and without the copula auxiliary in the Kievan period, which had been noted earlier by Istrina (1919–1921) and van Schooneveld (1959). In a recent paper Klenin (1993) challenges the traditional analysis of the OES Perfect in the sense I have been suggesting. What she shows is that the seemingly nonproper Perfect uses in the Laurentian Manuscript of 1377 can be explained as being consistent, if the traditional concept of a unified resultative meaning of this form is rejected. Both the resultative and a nonresultative use of the Perfect appear to be systematic (Klenin 1993: 333). Jakobson (1948) and van Schooneveld (1959) already “pointed out that the perfect can also express the temporal-aspectual relationship of an event to a chronological plane in the distant past” (Klenin 1993: 331). Nonresultative Perfects seem to overlap with the functions of the Conditional and the Pluperfect, as in (55) provided by Klenin, “in ways still poorly understood” (Klenin 1993: 331): (55)

Old East Slavic (Laurentian Ms., 04a09–10) Ini že ne svˇedušˇce rek-oša jako Kii est’ others but NEG knowing say-AOR.3PL as Kij COP perevoznik by-l” ferryman COP-PP ‘Others who did not know said that Kij has been a ferryman.’

As Klenin states, neither the traditional treatment nor a more recent theory (that Klenin calls revisionist”) has acknowledged this use. The first step on the way to revising the traditional theory was probably Sobolevskij’s (1907/1962) observation that the OES tenses are distributed differently in religious and secular texts. Out of this grew the view that there never existed simple tense forms in spoken OES. This opinion is held by Isaˇcenko (1980–83), Gorškova & Xaburgaev (1981), Uspenskij (1987), and Remneva (1988). For example, Uspenskij (1987: 144) and Remneva (1988: 9) argue, that the Aorist and the Imperfect were never used actively. According to this view, the occurences found in OES texts are artificial, literal archaisms deriving from Old Church Slavonic, and there is no consistency in the use of these tenses. It is curious that the Aorist and, to a lesser extent, the Imperfect were, nevertheless, used in narrative texts as late as in the beginning of the 18th century (Tommola 1993: 139). Of course, this does not substantiate the living use of these forms but shows that they must still have been understood by people so many centuries after their “disappearance”. For a very long time there was, apparently, a situation reminiscent of the current French one, when people read and understand passé simple but never use the form in their speech. The rise of Modern Russian literature and the codification of the literary language ended the kind of diglossia that characterized Russia up to the 18th century.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

461

On the perfect in North Slavic

461

Each of the past tenses seems really to have had a function of its own in the OES of the Kievan period (van Schooneveld 1959, Klenin 1993, Niissalo 1994). What is crucial to this claim is that the Perfect tense must be interpreted as having both resultative and experiential use.23 The Perfect becomes a narrative tense in this period, but its rise in this function can be observed, for example, in the Laurentian Manuscript, and as Klenin shows, there are motivations for the different uses of the Perfects. Thus, the replacement of the original narrative simple tenses by the Perfect is not arbitrary (Klenin 1993: 342).

7. Third round The old Perfect in Slavic was a periphrastic construction consisting of the copula and an old participle form. There are new periphrastic constructions that use the copula or other auxiliary verbs in North Slavic languages and dialects. The complement employed can be a passive past participle form, or (in Russian) a past gerund form. These constructions are potential new perfects although so far grammaticalized only in dialects. Maslov (1983: 52–54) distinguishes two structural types of resultative constructions that he discusses under the title of a “third round” of perfect constructions (the “first round” being the old Indo-European synthetic perfect): possessive and nonpossessive. Possessive constructions are presented in the PFQ material of Russian, Czech and Upper Sorbian. Data of active nonpossessive constructions are, in fact, only available from dialectal Russian.24

7.1. Resultative vs. perfect Past participles are often used to form resultative constructions. If the copula verb (‘be’) is used as the perfect auxiliary, the corresponding resultative construction cannot be distinguished from the perfect. In Baltic Finnic languages that not only differentiate active and passive participles in the past (and, consequently, in the perfect tense forms) but also use the copula verb as the perfect auxiliary, there is ambiguity (consistent in the singular) between the resultative and the (actional) perfect reading both in active (subject resultative) and in passive (object resultative). In Russian, the corresponding participle forms exist, but in (56b), for example, instead of the active past participle vospalivšajasja the passive participle of the transitive verb is used, because the so-called short (predicative) forms are not built from reflexive verbs. Cf.: (56) a.

Finnish Haava on tulehtu-nut (eilen). wound COP inflame-PP (yesterday)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

462

462

Hannu Tommola

b.

(57) a.

b.

Russian (Knjazev 1983: 149) Rana vospal-en-a (?vˇcera). wound inflame:PFV-PPP-F (yesterday) ‘The wound is / has (become) inflamed (became inflamed yesterday).’ Finnish Katto on maala-ttu (viime kesä-nä). roof COP paint-PPP (last summer-ESS) Russian (Knjazev 1983: 149) Kryša po-kraše-n-a (prošl-ym let-om). roof PFV-paint-PPP-F (last-INST summer-INST) ‘The roof is / has been painted (was painted last summer).’

A possible disambiguating test is to add a past time adverbial to the construction. The Russian (56b) seems to be a real subject resultative, since if one adds the time adverbial, it turns into a passive construction proper that implies an agent. In (57) the time adverbial is possible both in Finnish and in Russian. However, the perfect meaning probably gets lost simultaneously: most Russians then do not feel any difference between this construction and the same construction with a past copula (byla pokrašena), the default reading of which is simply a passive past. Another criterion of resultatives is their compatibility with temporal adverbs meaning ‘still’ (Nedjalkov & Jaxontov 1983: 12, Dahl 1985: 134; Bybee et al. 1994: 54). In (56) they would be normal, but in (57) at least strange, which may be another sign of the tendency of the resultative constructions “to be highly lexically restricted” (Dahl 1985: 134).

7.2. Passive constructions The analysis of passive perfects involves serious problems. If they are perfects of result, they tend, simultaneously, be resultatives, e.g.:25 (58) a.

b.

Latin Alea iac-ta est. die cast-PPP COP Russian Žrebij broš-en. die cast-PPP ‘The die is / has been cast.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

463

On the perfect in North Slavic

(59)

463

Russian Magazin by-l otkry-t (v vosem’). shop COP-PST open:PFV-PPP (PREP eight) ‘The shop was open / opened (at eight o’clock).’

Sentences like (58) and (59) have two readings: stative and dynamic. Sentence (59) remains ambiguous even if a specific time adverbial is added. It is ambiguous not only in terms of stative vs. dynamic, but in the case of dynamic reading, also as to the past (narrative) or pluperfect reading. The different readings of (58) between a perfect (dynamic) and a present resultative (stative) reading are not really an instance of ambiguity. If (59) can only be disambiguated by the context, (58) cannot be disambiguated at all. Some Russian linguists follow the tradition begun by Šaxmatov (1941: 486) in distinguishing alongside with the Present, Past and Future, a specific (passive) “perfect” tense in the Russian tense system (for instance, Bulanin 1995: 122). This “šaxmatovskij perfekt”26 – a copula construction where the Passive Past Participle is employed – is a construction that can be used both as a state passive and a dynamic passive. It is ambiguous, exactly like the corresponding English construction, which was illustrated by Jespersen (1924: 274) – and cited by Thieroff (1994: 11) – with the sentence translated into Russian in (60):

(60)

Russian Kogda ja prišël v pjat’ cˇ as-ov, dver’ when I come:PST PREP five hour-GEN.PL door by-l-a zakry-t-a, no ja ne zna-ju, kogda COP-PST-F shut:PFV-PPP-F but I NEG know-1SG when on-a by-l-a zakry-t-a. PRON-F COP-PST-F shut:PFV-PPP-F ‘When I came at five, the door was shut, but I don’t know when it was shut.’

7.3. Possessive constructions Another type of participle construction in Slavic is represented by the passive past participle involved in a construction with verbs for ‘having’ or some other type of possessive marking. The following utterances are not rare in colloquial Russian (61a) or in Russian dialects (61b):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

464

464

Hannu Tommola

(61) a.

b.

Russian U žen-y stol’ko uže pro vas PREP wife-GEN so_much already about PRON.2PL:ACC na-slyš-an-o. PFV-hear-PPP-NT ‘My wife has heard so much of you already.’ Russian (non-standard) U syn-a institut za-konˇc-en/-o. PREP son-GEN institute PFV-finish-PPP:M/-NT ‘My son has higher education [has finished the institute].’

Possessive and other resultatives tend to be emphatic, something that Bybee et al. (1994: 54) specifically attach to “completives”. Therefore it is interesting that verbs with a completive meaning are frequent in resultative constructions.27 In the standard North Slavic languages possessive constructions employ either a verb for ‘having’, or (in Russian) a copula construction with a preposition phrase indicating the “possessor”.28 The occurrences found in the PFQ responses can be divided into two, or even three, categories: a) those using adjective complements (62), and b) those having a verbal (participle) form as the complement (63). The adjective constructions remain statives and can hardly grammaticalize. The participle constructions can show agreement of the participle with the object in number, case and gender (with numerals, the participle form may be governed by it; see 13), or the form of the participle is not inflected. In the latter case the development towards a perfect gram has become possible. Possessive constructions were found in the Czech (13, 23, and 62a), Russian (63) and Upper Sorbian (62b) PFQ responses:29

(62) a.

b.

(63)

Czech (PFQ: 05) Má-š hotový úkol? have-2SG ready assignment ‘Have you done your homework?’ Upper Sorbian (PFQ: 70) Nˇe, wón drje z-mˇe-je je hižo hotow-e. no he sure PFV-have-3SG PRON:ACC.PL already ready-PL ‘No, he’ll have finished it [them] already.’

Russian (PFQ: 44) U menja sobra-n-o uže 200 kukol. PREP I:GEN gather:PFV-PPP-NT already 200 doll:GEN.PL ‘I have collected some two hundred dolls so far.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

465

On the perfect in North Slavic

465

7.4. Gerund construction In Northern dialects of Russian the copula  Present Gerund construction has been regarded as a potential new perfect.30 In the standard (spoken) language it appears, however, only with some typical verbs with a resultative state meaning, and these expressions have remained stylistically marked. To some extent this construction is a systemic feature of the tense-aspect grammar of the speakers of certain dialects. In the language of dialect speakers Imperfective verbs also occur (64b), while Perfectives are found in the standard language, often with just a few typical verbs, like the one in (64a). (64) a.

Russian On by-l vy-pi-vši. he COP-PST out-drink-GER.PST ‘He was drunk’ Russian (non-standard) My davno ne spa-mši, ne e-mši. we long NEG sleep-GER.PST NEG eat-GER.PST ‘We have neither slept nor eaten for a long time.’

In some Russian dialects the gerund construction is used as an object resultative, in spite of the actual active semantics of the gerund itself (see Trubinskij 1984: 156– 186): (65)

Russian (non-standard) Pol po-my-vši. floor PFV-wash-GER ‘The floor is / has been washed.’

7.5. Other strategies31 In a context where the object is topicalized and the verb is focused, the experiential perfect reading seems to be pragmatically preferred to the narrative reading. (66)

Russian (PFQ: 02) Net, èt-u knig-u ona pro-ˇcita-l-a. no this-ACC.F book-ACC she PFV-read-PST-F ‘She has read this one (all the way through).’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

466

466

Hannu Tommola

Leinonen (1994: 143) reiterates Crystal’s (1966: 8, fn. 2) observation about the role of intonation in the interpretation of the English Perfect in the sentence I’ve been to the Old Vic!. The reading is experiential, if the verb form (been) is stressed, but the stress on the Old Vic refers to a recent situation. It is well known that the Russian Imperfective Past has a so-called obšˇcefaktiˇceskoe znaˇcenie, general factual meaning, or a simple denotative use that occurs if just the existence of a past action is stated. Then it is the verb that is stressed. The theme-rheme structure cannot in written text be rendered by intonation, but the same function is partly taken over by the word order. The experiential meaning does not belong to the imperfective aspect domain alone (see above, 46–47). In questions as in (67a) (experiential) and (67b) (resultative), the verb has to be stressed; otherwise they may not have the perfect meaning: (67) a.

b.

Russian (PFQ: 04) Vy kogda-nibud’ vstreˇca-l-i mo-ju sestr-u? PRON.2PL ever meet-PST-PL my-ACC.F sister-ACC ‘Have you met my sister?’ Russian (PFQ: 05) A ty s-dela-l urok-i? and/but PRON.2SG PFV-do-PST lesson-PL ‘Have you done your homework?’

Some informants tended to specify the context, and with this pragmatic addition, example (68), for instance, is an unambiguous (nonnarrative) context in Russian. (68) a.

b.

Finnish (PFQ: 16) nous-sut / Nous-i-n tänään neljä-ltä. Ole-n COP-1SG stand_up-PP / -PST-1SG today four-ABL Russian (PFQ: 16) Delo v tom, cto ja prosnulsja v 4 matter PREP that:LOC that I wake_up:PST PREP 4 cas-a. hour-GEN ‘[It’s because] I woke up at 4 o’clock today.’

7.5.1. A Russian experiential In Tommola (1986: 47, 48, 58) the function of the Russian modal auxiliary prixodilos’ – Past of the Imperfective prixodit’sja ‘have to’ – was observed in uses that do not imply any modal element, but rather an experiential meaning. If in (69a) some modal nuance might be included, in (69b) prixodilos’ is absolutely a pure experiential auxiliary:32

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

467

On the perfect in North Slavic

(69) a.

b.

467

Russian (Tommola 1986: 48) Tebe prixodilos’ cˇ elovek-a PRON.2SG:DAT happen:IPFV:PST:NT (hu)man-ACC ubi-va-t’? kill-IPFV-INF ‘Have you ever killed a man (a human being)?’ Ob èt-om nam uže prixodilos’ PREP it-LOC we:DAT already happen:IPFV:PST:NT pisa-t’: ... write-INF. . . ‘I have already written about this [as follows]: . . . ’

Some prerequisites for grammaticalization are there – the lexical meaning of the auxiliary has totally disappeared – but some are lacking; the construction is optional. The same content can be communicated by using the simple Imperfective Past (Ty ubival kogda-nibud’ cˇ eloveka, and Ob ètom my uže pisali). It is noteworthy, however, that the experiential meaning tends to require adding the adverbial kogdanibud’ ‘ever’ in (69a), if the auxiliary is deleted. In (69b) ‘already’ can be taken as a sufficient experiential marker.33 7.5.2. ‘Already’ In the sample of Dahl (1985: 129–130), there is at least one language (Yoruba) that has grammaticalized a particle (ti) having the basic meaning ‘already’ as a perfect marker. The data recorded by Bybee et al. (1994: 64) add to these languages Inuit and Buli, in which “young anteriors” ( perfects) are of this same origin. Relational adverbs meaning ‘already’ and ‘just’ are typically found both in progressive and current state contexts, when “all is ready”, i.e., an activity ‘is already going on’, or the (beginning of a new) state ‘is (has been) already achieved’. If relative tense forms (perfect, pluperfect, future perfect) are not available, in temporal sentences ‘already’ is often needed in the main clause to indicate the order of the actions, since temporal connectors like ‘when’ tend to be neutral as to the mutual anteriority, simultaneity or posteriority of the actions denoted by the finite verbs (see above, 22–23, 30–31). As was mentioned above, in some languages ‘already’ also occurs frequently in experiential perfect contexts. It may be a later development, and the obligatoriness of this use remains to be explored. But at any rate ‘already’ is a potential perfect marker, rather than a resultative marker, for it is attested to in all the different perfect uses.34 Nevertheless, it cannot, for example, in Russian be regarded as compulsory, even though it is logically necessary to disambiguate temporal sentences.35 On the other hand, in Upper Sorbian the results of PFQ show that hizo ‘already’ is extensively used manifestly in experiential meaning. The use of ‘already’ could here be

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

468

468

Hannu Tommola

regarded as being due to German influence, for the German schon is typically used in experiential meaning (also, if not primarily, with the simple Past form). It may also appear paradoxical that of all the North Slavic languages, it is precisely Upper Sorbian, which has retained the perfect/simple past formal opposition in the tense system, that makes the largest use of ‘already’ as a perfect marker. As a matter of fact, neither the Sorbian nor the German Perfect are typical perfect grams, and the Upper Sorbian simple past tense forms are more or less literary narrative devices, whose main function is stylistic. In the sentences to be translated in PFQ, ‘already’ figured – in a kind of English metalanguage – four times (in the context descriptions additionally two times). In the English responses it was used four to five times, in the Scandinavian languages only three times, in Finnish five times, and so on. Therefore the considerable overuse of ‘already’ in some languages that do not have a tense qualifying as a perfect gram is important. Most striking is the amount of items with ‘already’ in Upper Sorbian (the word for ‘already’ used 24 times), but it is remarkable also in Hungarian (18), German (14), and Czech (14). Ukrainian (with 9 to 14 occurrences), Russian (9 to 12), Lower Sorbian (7 to 11) and Fering (10) are less obvious cases, but the difference in the basic PFQ sentences (and in the frequencies in “perfect languages”) is still considerable.36

8. Conclusions The Slavic periphrastic perfect has in East Slavic become a synthetic (bound) form, while in West Slavic it has with some variation retained the auxiliary as an alternative means to indicate the subject in the nonthird person forms. The different stages that the individual languages have undergone in the development of the formal expression have, however, very little importance as far as the meaning of these forms is concerned. In all North Slavic languages the old perfect has ceased to be a perfect and become a universal past (as a matter of fact, also in Sorbian, even though simple past tenses exist). In the light of the cross-linguistic data about grammatical TA categories (Bybee et al. 1994: 105) this evolution is not unexpected: resultative anterior ( perfect) past. The only problem is the view maintained indirectly by some scholars that, for instance, in OES the perfect was only resultative, thus suggesting that the development proceeded directly from resultative to past. In section 6, an attempt was made to reject this claim – a claim which was, in fact, never formulated in modern terms of cross-linguistic research. Because of another change in the TA system that presumably took place simultaneously with the loss of the perfect/past distinction, there are two pasts, a Perfective and an Imperfective. The very complicated process that still awaits explanation cannot be discussed here. It is possible that the perfective grams

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

469

On the perfect in North Slavic

469

are a further development of what Bybee et al. (1994: 54) call completives. The imperfectivization and the rise of the binary aspect system is the most mysterious question in the development in Slavic. Iteratives are possible foregoers of the Imperfective verbs. Attempts to discover signs of new developments in the material of the modern North Slavic languages have not been very fruitful. However, imperfective pasts seem to expand into the domain of the experiential (perfect). Some signs of lexical rendering of this meaning were also found. Resultative constructions are not rare in North Slavic languages. They typically coincide with passive expression. Other than passive resultatives were found in the Russian, Czech and Upper Sorbian PFQ data. Trubinskij (1983, 1984; see also Kuz’mina 1971) has shown that there are different tense systems in Northern Russian dialects. First, there are dialects, where a Gerund perfect is used (pol pomyvši ‘the floor is / has been washed’). This is not a “general” perfect (anterior), but rather a resultative. Second, there are dialects that use the Passive Past Participle to denote the same type of resultative states. Third, a most interesting group of dialects shows a possessive construction with the Passive Past Participle that occurs in clear opposition to the “aoristic” use of the Perfective Past; i.e., there is a tense system as presented in (70): (70) a. b. c. d.

Perfective Past Perfective Perfect Imperfective Past Imperfective Perfect

on uexal u nego uexano on exal u nego exano

‘he left’ ‘he has left’ ‘he travelled’ ‘he has travelled’

In those dialects where this kind of system prevails, the old perfect (the Past forms of the standard language) has really become a “doughnut” category. It no longer expresses the perfect meaning that was once its original and only meaning. One observation that may turn out to be significant is that several languages use the relational temporal adverb ‘already’ strikingly often. The extreme case is Upper Sorbian, and from the languages of the sample, the second in frequency of use of ‘already’ is a non-Slavic language, Hungarian, which also lacks a perfect tense or aspect. There is also a very high frequency in German and Czech, and, indeed, it is tempting to assume here an areal phenomenon.

9. Addendum: The case of Kashubian (and Slovincian) After I finished my work on this article Östen Dahl made me aware of the descriptions of past tense in Kashubian, one of the languages (spoken by ca. 150.000 persons) forming the Lechitic group of West Slavic (the others being Polish, and the extinct Polabian). Kashubian is sometimes considered a dialect of Polish. Reasons

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

470

470

Hannu Tommola

exist, however, to assume a separate linguistic entity, which is supported by the interest some linguists, notably Friedrich Lorentz, have shown to this language from the 19th century onwards. Indeed, the linguistic identity of Slovincian dialects, separate from Kashubian, has been discussed (Lorentz 1903, Mikkola 1897). A chapter on Kashubian (Cassubian) is included in Comrie’s and Corbett’s volume on Slavic languages, while, for example, they choose to discuss Rusyn (Ruthenian) only together with Ukrainian (Comrie & Corbett (eds.) 1993: 996). There are three past tense forms in Kashubian: one synthetic, and two periphrastic constructions. In the absence of questionnaire data on Kashubian the actual usage of these forms remains somewhat unclear. Anyhow, two of the forms are regular exponents of the ProtoSlavic Perfect, one with and one without the auxiliary ‘be’ (cf. Sorbian and Russian, respectively, in section 3 above). They are both general past tense forms, the one with the auxiliary mostly used by older people. The third one is a periphrastic construction that involves the verb miec ‘to have’ and the passive past participle. Interestingly, neither Stone (1993: 777), nor Breza & Treder (1981: 133) discuss their possible specific semantic function. In fact, all the examples provided attest a perfect meaning. What is not commented on in these sources, either, although mentioned and evident from the examples given, is that, unlike all the other Slavic languages, this auxiliary combines also with the active past l-participle. Somewhat unexpectedly, Breza & Treder (1981: 133) remark about the Russian-type past (without the auxiliary) that it does not need to be a borrowing from German, whereas they do not say that the ‘have’-construction apparently is a German calque. This is, instead, explicitly noticed for Slovincian by Lorentz (1903: 11), while also he ignores the idiosyncrasy of the construction combining ‘have’ and the l-participle. Lorentz speaks explicitly of the Perfect and Pluperfect considering these constructions both in Kashubian and in Slovincian (Lorentz 1925: 192; 1903: 302–303). But, without discussing the functional relationship between the different tenses, he also points out that the (standard North Slavic type) past tense forms of Kashubian are used with the perfect meaning as one of their functions (“fungiert als Perf. Präsens”; Lorentz 1925: 192). Thus, it seems to be the case that there are specific perfect forms in Kashubian, although they are not totally grammaticalized. Another type of Perfect, which is also a transparent German calque, is mentioned by Stone (1993: 777): In the case of intransitive verbs of motion, this tense is formed with the auxiliary bëc ‘to be’ (instead of miec ‘to have’). The participle then agrees in gender and number with the subject: Ta białka je precz jidzonô. ‘The (or that) woman has gone away.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

471

On the perfect in North Slavic

471

Acknowledgements I am indebted to Alexandr and Ljudmyla Duliˇcenko, Natalija Kozinceva, Jakub Lapatka, Helena Leheˇcková, Helle Metslang, Fejsa Mihajlo, Madlena Norberg, Erlingur Sigurðsson, Sonja Wölke, and Joanna Zach-Bło´nska for their kind help with their respective native languages. I also express my gratitude to Östen Dahl and Jouko Lindstedt for many useful remarks on earlier versions of the paper.

Notes 1. To some extent, the aspect distinction between the simple past forms is replaced by a new aspect opposition that is not restricted to past tenses. The whole TA-system has changed radically, which has had considerable influence, for example, on the expression of future time reference. In the new verb system the perfect/nonperfect distinction lacks systematic marking, and the original Imperfect/Aorist opposition is not quite rendered by the new Imperfective Past and Perfective Past opposition either. 2. Problems have arisen as researchers – especially of Old Russian – have used a set of semantic features to describe “the Perfect proper” that do not, as a matter of fact, comprise anything other than the resultative perfect (see section 6). 3. The interpretation suggested here is that the perfectivizing effect of the prefix (pro‘through’ = PFV) is cancelled by the involvement of a secondary imperfective suffix (-yva- = IPFV), while the lexical meaning of the prefix is retained. 4. A verb form is glossed Perfective in three different ways: (i) the prefix is marked as Perfective (as in s-dela-t’ PFV-do-INF); (ii) the (semelfactive) suffix is marked as Perfective (either genuine momentary verbs like pryg-nu-t’ jump-PFV-INF, or Perfectives like otdox-nu-t’ rest-PFV-INF); (iii) the lexeme as a whole is marked Perfective (e.g., prefixed verbs as vy-jt-i outgo:PFV-INF, or simplexes like skaza-t’ say:PFV-INF). Imperfective simplexes are not explicitly glossed as such, whereas secondary Imperfectives are always marked out (as the suffix in vy-bras-yva-t’ out-throw-IPFV-INF, or the stem in vy-xodi-t’ out-go:IPFV-INF). Otherwise lacking Perfective gloss is sufficient indication of imperfectivity. 5. The 1st and 2nd person clitics, though, are not necessarily attached to the main verb. Thus, at least my´smy/wy´scie robili ‘we/you did/have done’ are possible instead of the more frequent robili´smy/robili´scie, whereas analoguous forms in the 2nd person singular (ty´s robił) are rare, and the 1st person singular (jam robił) is obsolete (Joanna ZachBło´nska, p.c.). 6. Whether the variant of Rusyn (also called Ruthenian) that our PFQ material reflects has to be regarded as a West Slavic (because of similarities, for example, with Slovak), or an East Slavic language (Ukrainian), or perhaps something else, is an interesting question of its own that will not be addressed here. Obviously, it is suitable to treat it as a North Slavic language. The material for PFQ was supplied by three speakers of the dialect

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

472

472

Hannu Tommola

spoken in Vojvodina, Yugoslavia, and made available by Svenka Savi´c. There are more, at least twenty or thirty thousand, Rusyn speakers in Ukraine. 7. This pattern occurs, for example, in Armenian, Fering, German, Italian, Norwegian, Swedish, and possibly in Mari (Cheremis). It is possible in English and Finnish, too. A notable exception is Modern Greek, where the Perfect is also used in PFQ:28. There is no variation in Lower Sorbian, either, and three Macedonian informants also use only the Perfect. In Icelandic a resultative construction is used consistently, not only in PFQ:27– 29, but also in PFQ:30–31. PFQ:29 was rendered in Icelandic: Nei, hún er ekki enná kom-in. come-PP.F no she COP NEG yet ‘No, she hasn’t / didn’t [isn’t] come back yet.’ 8. In many languages a stative construction is used in both cases (Dutch), or is an alternative to the Perfect, which is used in both (Swedish). However, in Greek and in Armenian there is a Perfect in PFQ:30, and an Aorist in PFQ:31. Some Macedonian informants preferred the same distribution. 9. There are Russians that accept U menja poterjano pjat’ zontikov in colloquial speech (Valentin Trubinskij, personal communication), while others reject it categorically. 10. The fact that in Icelandic the vera búin(n) construction is employed in PFQ:48 (Hún er búin að horfa á að í rjá tíma) suggests that it has features of a completive in the sense of Bybee et al. (1994: 54). 11. Similarly in Russian, Czech, Polish, Upper Sorbian, and Rusyn. 12. In Russian there are habitual forms, specifically, verbs, whose use is restricted to the Past tense. They are quite similar in meaning to the English used to construction, but never grammaticalized and are in the process of disappearing from the standard language. The following example could be archaic for PFQ:51: ?Ja ži-val zdes’; zna-ju každ-uju ulic-u. I live-HAB:PST here know-1SG every-ACC.F street-ACC In Czech this kind of habitual is more frequent. The example provided in Havránek & Jedlicka (1981: 253) corresponds almost exactly to the situation in items PFQ:71 and PFQ:73: Na tom míst-ˇe stá-va-l pevný hrad. PREP this:LOC place-LOC stand-HAB-PST firm castle ‘There was [used to stand] a fortified castle on this place.’ 13. Another Bulgarian informant had used the Imperfect (Poznavaxme se) here, and, indeed, the logical consequences of ‘having met’ go through ‘getting to know’ to ‘knowing’, and all stages are represented in the Bulgarian responses to this item: 1) bjax ja sreštal 2) bjax se zapoznal s neja 3) poznavaxme se.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

473

On the perfect in North Slavic

473

14. The difference is rather one pertaining to aspect proper. According to Vinogradov (1986: 454) we could interpret the examples as follows: the Perfective is preferred in the question, if the speaker wants to know who “is to be blamed or praised” for the result (i.e., the initiative or decision), while the Imperfective asks for the persons who actually carried out the project (architects, designers). 15. Like the adverbs ‘probably’ (Polish pewnie, Upper Sorbian najskerje), ‘certainly’ (Upper Sorbian drje) in inferential meaning, and ‘allegedly’ (Czech prý, Upper Sorbian pjeˇca and Polish podobno) in reportative meaning. A more skeptical attitude to the facts reported is expressed by Russian vrode by, kak budto by and Ukrainian niby(to). A neutral attitude is also simply rendered by lexical verbs of saying and thinking (Russian govorjat ‘they say’, Czech tvrdí, Polish utrzymuje ‘claims’, and Lower Sorbian jo wobtwarzil ‘has claimed’, Rusyn hvarela ‘told’, Rusyn dumam, Czech myslým, and Polish mysl˛e ’I think’). 16. In both Sorbian languages, the corresponding gram labels are Imperfective Perfect and Perfective Perfect instead of Imperfective Past and Perfective Past. Additionally, in Upper Sorbian the (Perfective) Aorist and the (Imperfective) Imperfect were used (see section 5.1). 17. In many languages praesens historicum is used in narrative contexts (PFQ:08–11; in the fairy tale item, PFQ:61, it seems to be obligatory, for example, in Greek). Consider the following Polish examples of Present forms used in reporting what happened recently (PFQ:12), and Past forms used when the episode is in remote past (PFQ:13): (PFQ:12) Siedz˛e sobie pod drzewem, a˙z tu jabłko spada mi na głow˛e. ‘I was sitting under a tree, when an apple fell on my head.’

18.

19.

20. 21. 22.

23.

(PFQ:13) Kiedy siedziałem sobie pod drzewem, jabłko upadlo mi na głowe. ‘While I was sitting under a tree, an apple fell on my head’. Icelandic seems to have a similar opposition; therefore Icelandic examples are given. The hafa  Supine (see 46) corresponds to habeo perfects (ég hefi talað ‘I have spoken’), whereas the construction vera búin(n) að + Infinitive (47) clearly has perfective (or completive) features. There is an Imperfective Perfect in PFQ:04 also in Lower Sorbian, and it is possible in Upper Sorbian, too, but I am very skeptical where aspect forms are concerned in Sorbian in general, and in Lower Sorbian in particular. In colloquial Sorbian, for instance, the periphrastic future is formed not only from the Imperfective verbs as in the other Slavic languages but also from the Perfective stems. This is not in agreement with the norms of the literary language, however (Šewc 1968: 176). In some West Slavic languages, for example in Czech, Perfective verbs can be used to denote frequentative and habitual situations, if the single events are completed. Note the possible Perfective Past (a so-called Perdurative Aktionsart) in Russian, similarly in Ukrainian. Symptomatic of the loss of the genuine perfect meaning is the use of the German and Italian Perfect in PFQ:53, e.g., Italian: Ho vissuto qui per 7 anni, poi ho dovuto andarmene ‘I [have] lived here for seven years, but then I [have] had to move away.’ I will not discuss here at any length the possible functional differentiation of the perfects with and without auxiliary suggested by Istrina, van Schooneveld and Niissalo. Lavrentij

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

474

474

24. 25. 26.

27.

28.

Hannu Tommola Zizanij in his “Slavonic” grammar of 1596 does not include perfect forms as a separate tense, but in the 2nd person sing. compound forms are cited as replacing both Aorist and Imperfect: jav-i-x (1SG), jav-i-l” esi and jav-i (2SG), jav-i (3SG); and jav-lja-x (1SG), jav-lja-l” esi and jav-lja-se (2SG), jav-lja-še (3SG) from jav-iti, jav-lja-ti ‘show, display’. Also in Meletij Smotrickij’s grammar of 1619 the participle forms replace these tense forms in the 2nd person, but they appear without the auxiliary, whereas a separate (perfect) tense with the auxiliary forms is presented. (Smotrickij’s book was the last and most authoritative Slavonic grammar before the appearance of the first Russian grammars.) Maslov (1983: 53) also quotes (after Havránek 1937: 77, 80) Czech (dial.) Vseˇcko bylo vymˇrete ‘everything was dead’ and Polish (dial.) On był jechany ‘he had ridden/driven’. Lindstedt (1994: 33) remarks after having quoted Maslov’s (1990: 372) definition of the perfect that its first part applies to the meaning of resultative. Šaxmatov was not the first to detect the perfect meaning of the passive past participle, cf. Šafranov (1866: 130): “Perfectum vsego toˇcnee možet byt’ peredano nastojašˇcim stradatel’nogo zaloga” (‘Perfectum is most properly conveyed by the Present tense of the Passive voice’). The Russian verb in (61) is a completive (naslyšat’ ‘hear very much, enough’), and examples provided by Helena Leheˇcková (p.c.) from Czech seem to be so to a great extent, too, e.g.: Mám nachozeno 100 km ‘I have walked 100 km’, Jsem nasezená dost’ ‘I’ve been sitting enough’, Mám na-/odpracováno ‘I’ve worked enough’, Mám nalyžováno na 2 roky dopˇredu ‘I’ve skied (enough) for two years ahead’. In Baltic Finnic languages possessive constructions also occur. Consider the Estonian one used in (PFQ:70): Ei, küllap ta-l on juba õpi-tud. no PRAGM PRON.3SG-ADESS COP already study-PPP ‘No, I think he finished already.’

29. The Czech habeo-construction mít + Passive Past Participle was found in Dahl’s (1985: 130) investigation in 6 items of the questionnaire, which did not entitle it to a perfect gram status, but still indicates a potential development ( PFCT). 30. Leinonen (1994) observed the statement by Vinogradov (1947: 568; in the 3rd ed. 1986: 462), according to whom the prediction was formulated by Žitomirskij (1915: 5–6). 31. After having finished this paper I came across an expression in Slavic that resembles the Hiberno-English use of after as a perfect marker (see, for example, Kallen 1991). In reply to a suggestion about having lunch, a Czech person uttered, in Russian: Ja uže posle obeda ‘I (am) already after lunch’. I consulted Helena Leheˇcková, who supplied the following examples from Czech: (i) (ii)

(iii)

Jsem už po snídani/obˇedˇe/veˇceˇri ‘(I) am already after breakfeast/lunch/supper’ a. Dítˇe je po nemoci. ‘(The) child is after illness’ b. Dˇedeˇcek je po infarktu. ‘Grandfather is after heart attack’ c. Babiˇcka je už po smrti. ‘Grandmother is already after death’ a. Karel je po flámu. ‘Karel is after a party’ b. Jana je po promoci. ‘Jana is after the graduation’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

475

On the perfect in North Slavic

32.

33.

34. 35.

36.

475

It has not been possible to gather more evidence for the purposes of this article, but it seems reasonable to assume that the construction also occurs in other North Slavic languages. (Cf. Bertinetto (in this volume, fn. 3) about the periphrasis être après  INF, which has the function of a progressive in some varieties of French, and Johanson’s (in this volume) examples from Welsh and Irish with ‘after’ having perfect meaning.) Cf. the etymological similarity of Rus. prixodit’ ‘come’ with Fin. tuli käytyä and Swed. jag kom att besöka ‘[It so happened that] I visited’ (or ‘I came to visit’), expressions that employ the verb ‘come’ as an auxiliary. In Nedjalkov & Jaxontov (1983: 12) the readings of the English John has sung are explained in Russian by means of a) uže ‘already’, and b) a periphrastic construction with sluˇcalos’ ‘happened’ (observed by Tommola 1986: 47, 58 n. 31). For example in an inclusive perfect context in (16), and frequently in experiential contexts, especially in Upper Sorbian. Vinogradov [1986: 459] remarks that translating the perfect into Russian, uže ‘already’, teper’ ‘now’, nastojašˇcee vremja ‘at present’ etc. are often added “dlja jasnosti” (’in order to make it clear’). For some reason, Belarusan (six occurrences) and Rusyn (3 to 4) do not seem to employ this type of marker. The same applies to Dutch (6 to 7) and especially to Italian (3). But on the whole, the fifteen “perfect languages” of the sample show an average of 4.9 occurrences of ‘already’, while the other fifteen languages (without a functional perfect category) resulted at an average of 10.1.

References Bertinetto, Pier Marco this volume “The progressive in Romance, as compared with English”. Bondarko, A[leksandr] V[ladimiroviˇc 1980 “Upotreblenie vidov”, in: N.Ju. Svedova (ed.), Russkaja grammatika I. Moskva: Nauka, 604–612. Bondarko, A[leksandr] V[ladimiroviˇc (ed.) 1987 Teorija funkcional’noj grammatiki. Vvedenie. Aspektual’nost’. Vremennaja lokalizovannost’. Taksis. Leningrad: Nauka. Breza, Edward & Jerzy Treder 1981 Gramatyka kaszubska. Zarys popularny. Gda´nsk. Bulanin, L[ev] L[’voviˇc 1995 “K obosnovaniju sistemy vremeni v sovremennom russkom jazyke”, in: Meždunarodnaja jubilejnaja sessija, posvjašˇcënnaja 100-letiju V.V. Vinogradova. Tezisy dokladov. Moskva, 122 - 123. Bybee, Joan, Revere Perkins & William Pagliuca 1994 The Evolution of Grammar. Tense, Aspect, and Modality in the Languages of the World. Chicago and London: The University of Chicago Press. Comrie, Bernard 1976 Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. Comrie, Bernard & Greville G. Corbett (eds.) 1993 The Slavonic languages. London and New York: Routledge.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

476

476

Hannu Tommola

Crystal, David 1966 Dahl, Östen 1985 Faßke, Helmut 1981

“Specification and English tenses”, Journal of Linguistics 2: 1–34. Tense and Aspect Systems. Oxford: Basil Blackwell.

Grammatik der obersorbischen Schriftsprache der Gegenwart. Morphologie. Verfaßt von Helmut Faßke unter Mitarbeit von Siegfried Michalk. Budyšin: Domowina. Gorškova, K[lavdija] V[asil’evna] & G[eorgij] A[leksandroviˇc] Xaburgaev 1981 Istoriˇceskaja grammatika russkogo jazyka. Moskva: Vyšsaja škola. Graves, Nina this volume “Macedonian – a language with three perfects”. Havránek, Bohuslav & Alois Jedliˇcka ˇ 1981 Ceská mluvnice. Praha: Státní pedagogické nakladatelství. 4. vydání, pˇrepracované (1st ed. 1959). Issatschenko [Isaˇcenko], Alexander 1980–83 Geschichte der russischen Sprache, 1–2. Heidelberg: Winter. Istrina, Evgenija Samsonova 1919–21 “Sintaksiˇceskie javlenija Sinodal’nogo spiska Novgorodskoj letopisi”, Izvestija Otdelenija Russkogo Jazyka i Slovesnosti 24: 2 [1919/1923], 1–172; 26 [1921/1923]: 207– 239. Ivanov, Valerij Vasil’eviˇc 1982 “Istorija vremennyx form glagola”, in: Ruben Ivanoviˇc Avanesov & Valerij Vasil’eviˇc Ivanov (eds.), Istoriˇceskaja grammatika russkogo jazyka. Morfologija. Glagol. Moskva: Nauka, 25–131. Jakobson, Roman 1948 “Quelques remarques sur l’édition critique du Slovo, sur sa traduction en langues modernes et sur la reconstruction du texte primitif”, in: Henri Gregoire & Roman Jakobson & Marc Szeftel (eds.), La Geste du Prince Igor’. Épopée russe du douzième siecle [= Annuaire de l’Institut de Philologie et d’Histoire Orientales et Slaves, 8 (1945–1947)], 5–37. Janas, Pˇetr 1976 Niedersorbische Grammatik (für den Gebrauch der Sorbischen Erweiterten Oberschule). Budyšin: Domowina. Jespersen, Otto 1924 The philosophy of grammar. New York: Allen & Unwin. Johanson, Lars this volume “Viewpoint operators in European languages”. Kallen, Jeffrey L. 1991 “Sociolinguistic variation and methodology: after as a Dublin variable”, in: Cheshire, Jenny (ed.), English around the world. Sociolinguistic perspectives. Cambridge: Cambridge University Press. 51–74. Klenin, Emily 1993 “The Perfect Tense in the Laurentian Manuscript of 1377”, in: Robert A. Maguire & Alan Timberlake (eds.), American Contributions to the Eleventh International Congress of Slavists. Bratislava, August–September 1993. Literature. Linguistics. Poetics. Columbus, Ohio: Slavica Publishers, 330–343. Knjazev, J[urij] P[avloviˇc] 1983 “Rezul’tativ, passiv i perfekt v russkom jazyke”, in: V. P. Nedjalkov (ed.), 149–160. Kuryłowicz, Jerzy 1965 Indoeuropejskie “perfectum” w słowia´nskim. Studia z filologii polskiej i słowia´nskiej, 5. Warszawa: PAN.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

477

On the perfect in North Slavic

477

Kuz’mina, I[rina] B[orisovna 1971 “Predikativnoe upotreblenie priˇcastnyx form”, in: I[rina] B[orisovna Kuz’mina & Elena Vasil’evna Nemˇcenko (eds.), Sintaksis priˇcastnyx form v russkix govorax. Moskva, 16– 223. Leinonen, Marja 1994 “Interpreting the perfect: the past as explanation”, in: Susanna Shore & Maria Vilkuna (eds.), SKY 1994 [= 1994 Yearbook of the Finnish Linguistic Society]. Helsinki: Suomen kielitieteellinen yhdistys, 135–156. Lindstedt, Jouko 1994 “On the Development of the South Slavonic Perfect”, in: Three Papers on the Perfect, 32–53 (E UROTYP Working Papers VI: 5). this volume “The perfect – aspectual, temporal and evidential”. Lorentz, Friedrich 1903 Slovinzische Grammatik. St. Petersburg: Buchdruckerei der Kaiserlichen Akademie der Wissenschaften. 1925 Geschichte der pomoranischen (kaschubischen) Sprache. Grundriß der slavischen Philologie und Kulturgeschichte [1]. Berlin & Leipzig: Walter de Gruyter & Co. Maslov, Jurij Sergeeviˇc 1949 “K voprosu o proisxoždenii posessivnogo perfekta”, in: Uˇcënye zapiski LGU, No 97, serija filologiˇceskix nauk, vyp. 14: 76–104 (also in Maslov 1984: 224–248). 1983 “Rezul’tativ, perfekt i glagol’nyj vid”, in: V.P. Nedjalkov (ed.), 41–54. 1984 Oˇcerki po aspektologii. Leningrad: Izdatel’stvo Leningradskogo universiteta. 1987 “Perfektnost’ ”, in: A.V. Bondarko (ed.), 195–209. 1990 “Perfekt”, in: Victorija Nikolaevna Jarceva (ed.), Lingvistiˇceskij ènciklopediˇceskij slovar’. Moskva: Sovetskaja ènciklopedija, 372. Mikkola, J[ooseppi] J[ulius] 1897 “K izuˇceniju kašubskix govorov. I. Neškol’ko zamˇetok po kašubskim govoram v sˇeverovostoˇcnoj Pomeranii”. Izvˇestija Otdˇelenija russkago jazyka i slovesnosti Imp. Akad. Nauk, II, kn. 2. Sankt-Peterburg. 400–428. Nedjalkov, V[ladimir] P[etroviˇc] (ed.) 1983 Tipologija rezul’tativnyx konstrukcij (rezul’tativ, stativ, passiv, perfekt). Leningrad: Nauka. Nedjalkov, V[ladimir] P[etroviˇc] & Sergej E. Jaxontov 1983 “Tipologija rezul’tativnyx konstrukcij”, in: V.P. Nedjalkov (ed.), 5–41. Niissalo, Nina 1994 Upotreblenie prošedšix vremennyx form v drevnerusskom jazyke Kievskogo perioda. Unpublished Master’s thesis. University of Tampere. Slavonic Philology. Remneva, Marina Leont’evna 1988 Literaturnyj jazyk Drevnej Rusi. Moskva: Izdatel’stvo Moskovskogo Universiteta. Šafranov, S. 1866 Russkij sintaksis. Riga: Izdanie knigoprodavca Kimmelja. Šaxmatov, Aleksej Aleksandroviˇc 1941 Sintaksis russkogo jazyka. Leningrad: Uˇcpedgiz. Schooneveld, Cornelis H. van 1959 A semantic analysis of the Old Russian finite preterite system. The Hague: Mouton & Co. Šewc, Hinc 1968 Gramatika hornjoserbskeje rˇecˇ e. 1. zwjazk. Fonematika i morfologija. Budyšin: Domowina.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

478

478

Hannu Tommola

Sobolevskij, Aleksandr Ivanoviˇc 1962 Lekcii po istorii russkogo jazyka. Izd. 4e, repr. S’-Gravenhage. [1st pub., Moskva 1907]. Smotrickij, Meletij 1619/1979 Grammatiki slavenskija pravil’noe sintagma. Faksim. ed. by V. V. Nimˇcuk & V. M. Rusanovskij. Kiev: Nauka dumka. Stone, Gerald 1993 “Cassubian”, in Bernard Comrie & Greville G. Corbett (eds.), The Slavonic languages. London and New York: Routledge, 759–794. Thieroff, Rolf 1994 “Passives, Perfects, Resultatives, and Statives”, in: Three Papers on the Perfect, 1–20 (E UROTYP Working Papers VI: 5). Tommola, Hannu 1981 “On the semantics of ‘situations’ and ‘events”’, in: Terminologie et traduction. Tome B [= Vaasan korkeakoulun julkaisuja. Tutkimuksia 80, Philologie 7]. Vaasa: University of Vaasa, 80–119. 1986 Aspektual’nost’ v finskom i russkom jazykax [= Neuvostoliittoinstituutin vuosikirja 28]. Helsinki: Neuvostoliittoinstituutti. 1993 “’Perfektnoe znaˇcenie’ v russkom jazyke”, in: Jouko Lindstedt & Pekka Pesonen (eds.), Studia Slavica Finlandensia in Congressu XI Slavistarum internationali Bratislavae anno MCMXCIII oblata [= Studia Slavica Finlandensia, Tomus X]. Helsinki: Venäjän ja Itä-Euroopan instituutti, 134–141. Trubinskij, Valentin Ivanoviˇc 1979 “O russkom razgovornom posessivnom perfekte”, Severnorusskie govory, vyp. 3. Leningrad, 154–173. 1983 “Rezul’tativ, passiv i perfekt v nekotoryx russkix govorax”, in: V.P. Nedjalkov (ed.), 216–226. 1984 Oˇcerki russkogo dialektnogo sintaksisa. Leningrad: Izdatel’stvo Leningradskogo universiteta. Uspenskij, Boris Andreeviˇc 1987 Istorija russkogo literaturnogo jazyka (XI–XVII vv.). Sagners Slavistische Sammlung 12. München: Verlag Otto Sagner. Vinogradov, Viktor Vladimiroviˇc 1986 Russkij jazyk (Grammaticeskoe ucenie o slove). Moskva: Vyssaja skola. 3rd ed. (1st ed. 1947, Moskva: Ucpedgiz). Žitomirskij, K. G. 1915 Molox XX veka. [Pravopisanie]. Moskva: Trud.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

479

Nina Graves

Macedonian – a language with three perfects?

1. Introduction Macedonian is a relatively small South Slavic, Balkan language which became a literary language only after the Second World War. For a long time it was considered to be a dialect or western variant of Bulgarian. The Macedonian literary language is based on the central dialects (Bitola – Veles – Prilep), but it has also liberally adopted forms from other dialects. It seems that the southwestern dialects are those most influenced by other Balkan languages, and thus are the gateway for accepting new forms. From these dialects forms spread to other dialects. The focus of this paper is on the perfect tenses of Macedonian and the fact that Macedonian grammars describe three different ways to form a perfect: 1. 2. 3.

The ‘be’ Perfect, formed with the auxiliary ‘to be’  the Past Active Participle. The ‘have’ Perfect, formed with the auxiliary ‘to have’  the Past Passive Participle. The Third Perfect, formed with the auxiliary ‘to be’  the Past Passive Participle.

The aim of this article is to describe the characteristic use of each form in various contexts. The origin of these perfects will be briefly explained. Functional similarities and differences will be pointed out. The descriptions draw from both normative grammar and dialectal variation. Special attention is given to the southwestern dialect of Ohrid, since in this dialect the use of these three perfects is the most well-defined, with each one of them having its own characteristic semantic value. For this article I had seven informants, three of whom were assistants at the Department of Macedonian Language at the University of Kiril and Metodi in Skopje. All three are from Skopje and speak the a northern central dialect, referred to hereafter as a northern dialect. On the map they are indicated by the numbers 1, 2 and 3. The other four informants are students of English language at the same university. They come from various dialectal areas: Kavadarci (number 4 on the map), Sveti Nikole (5), Prilep (6) and Ohrid (7).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

480

480

Nina Graves

Map 1. Geographical distribution of the informants

2. The origin of different perfects in Macedonian 2.1. The ‘be’ Perfect The ‘be’ Perfect is formed with the auxiliary sum ‘to be’  the Past Active Participle. This form is descended from Proto-Slavic, where the perfect was formed by the present of the auxiliary ‘to be’ (byti)  the past resultative participle, whose formation is possible only from the Aorist stem. There are two main formal differences between the perfect in Proto-Slavic (as evidenced in Old Church Slavonic) and present-day Macedonian. The first is that the past participle can nowadays be formed not only from the Aorist stem but also from the Imperfect stem. The second difference is that in current usage the auxiliary is lost in 3rd person singular and plural. There are three clear stages in the process of auxiliary loss: 1. 2. 3.

The auxiliary occurs in all persons in both singular and plural. Some kind of confusion arises in the use of the auxiliary in the 3rd person. The auxiliary disappears completely from the 3rd person forms.

The loss of the auxiliary first occurred in western dialects of Macedonian. From there it spread to all the other dialects until it recently became grammaticalized. Even nowadays, however, there are some dialects (e.g., Tikveš) that are still in the second stage, and one can hear confusion in the use of the auxiliary:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

481

Macedonian – a language with three perfects?

(1) a.

b.

481

on e vikal he be:3SG shout:PPA:M ‘He has shouted.’ (Koneski 1986: 197–198) oni e vikale they be:3SG shout:PPA:PL ‘They have shouted.’

The standard language uses no auxiliary in such 3rd person constructions; if it did, however, a plural auxiliary would be expected: oni se vikale. The ‘be’ Perfect is most widely used in northern and eastern dialects, which are most influenced by the neighboring language Serbian, where the ‘be’ Perfect is the only type of perfect. The ‘be’ Perfect occurs over most of the Macedonian linguistic territory, though it has entirely lost its use as a perfect at least in the dialect of Bobošica and some dialects of Kostur. There seems to be a similar tendency in such southwestern dialects as Ohrid. In these dialects the ‘be’ Perfect has been replaced by the ‘have’ Perfect (Koneski 1986: 197). This will be discussed below. The use of the ‘be’ Perfect in literature, especially in poetry, is quite limited. In prose, though, it is the most often used of the three perfect forms occurring in Macedonian. In a study of the various possibilities of translating the English Present Perfect into Macedonian (Arsova-Nikoli´c 1973: 157), it was shown that, in the majority of cases, the Present Perfect in English is translated into Macedonian by sum ‘to be’  l-participle. The author considers this tendency to be due to the semantic and formal similarity of these two perfects, even though the English Present Perfect is in the form of a ’have’ Perfect.

2.2. The ‘have’ Perfect The ‘have’ Perfect is formed with the auxiliary ima ‘to have’  the Past Passive Participle, which is always neuter. It can be formed from both Imperfective and Perfective verbs. It is difficult to determine when this form of the perfect first appeared, but one of the earliest texts in which a form similar to the ‘have’ Perfect exists is a manuscript from the monastery of Krnino (now in the Kicevo region), written in 1706: (2)

(Koneski 1986: 201) imam go aforesan i proklet have:1SG he:ACC excommunicate:PPP:M and curse:PPP:M ‘I have excommunicated and cursed him.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

482

482

Nina Graves

The ‘have’ Perfect in Macedonian did not develop for internal structural reasons only, but most likely arose under the influence of another Balkan language, possibly Greek, where the perfect is regularly formed with the auxiliary ‘have’, and possibly also Albanian or Arumanian. Speculation about the origin of the ‘have’ Perfect is not of relevance to this paper, since the further development and the specific present-day character of the Macedonian ’have’ Perfect differ greatly from those of similar forms in other Balkan languages. (Friedman 1977: 96, 98; Koneski 1986: 200; Velkovska 1987: 304). The ‘have’ Perfect (or, to be more precise, the construction ‘to have’  a verbal adjective) was first adopted by southwestern dialects, where it has now replaced the older ‘be’ Perfect. This development will be examined later in this paper. The use of the ‘have’ Perfect in literature is very rare. There are two reasons for this phenomenon: first, the ‘have’ Perfect is a colloquial, dialectal form, a form that in most contexts is foreign and unacceptable to speakers from some areas, including Skopje, the country’s capital and its cultural and scientific centre. Second, in poetry the ‘have’ Perfect is considered clumsy (Koneski 1987: 503). Interestingly, in the study of the possible translations of the English Present Perfect into Macedonian, Arsova-Nikoli´c (1973: 172) found that the ‘be’ Perfect was used most often and that the ‘have’ Perfect was hardly used at all. As the matter of fact, the ‘have’ Perfect occurred in only 14 out of 516 cases.

2.3. The Third Perfect The third type of perfect occurring in Macedonian can with reason be called “a mixed Perfect”, since structurally it seems to be a combination of both the ‘be’ and ‘have’ Perfects. Specifically, the Third Perfect is formed with the auxiliary ‘be’  a Past Passive Participle. As with the ‘have’ Perfect, this mixed-type Third Perfect is a relatively new phenomenon. Again it seems to have entered the language through the influence of another non-Slavic Balkan language (either Arumanian or Greek), with southwestern Macedonian dialects providing the open door for this form. The use of this perfect has not yet spread to all parts of the Macedonian linguistic territory (Koneski 1986: 199).

3. The use of different perfects, as described in normative grammar In this section, we shall look at the use of the different perfects as described in normative grammars. It should be borne in mind that the actual use of the perfects –

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

483

Macedonian – a language with three perfects?

483

especially the ‘have’ and Third Perfects – is much more diverse than the rules of the normative grammars suggest. The ‘be’ Perfect describes an event that has been completed in the past, with no clear reference to a particular moment in the past. It is most often used as an unmarked (indefinite) past, but it can also be used with definite past time reference (Koneski 1987: 461–464). In describing actions completed in the past and relevant for the present, both the ‘be’ and ‘have’ Perfects can be used interchangeably. Thus: (3) a.

b.

(Velkova 1987: 313) [Koga k’e dojdiš kaj carot, ništo da ne mu sakaš od kolku edno pile što ] ‘have’: ima zatvoreno v kavez. have:3SG close:PPP:NT in cage ‘be’: go zatvoril v kavez. he:ACC close:PPA:M in cage ‘[When you come to the tsar, you should not ask for anything but a chicken that] he has locked in a cage.’

But, in contexts where the ‘be’ Perfect can be used with a specific time reference, it cannot be changed to a ‘have’ Perfect: e.g., (4)

Sum stanal nok’eska vo eden. be:1SG get_up:PPA:M last_night at one ‘I got up last night at one o’clock.’

It is ungrammatical to replace the sentence above with: (4¼ )

*Imam stanato nok’eska vo eden. have:1SG get_up:PPP:NT last_night at one

The principal difference is that when the ‘be’ Perfect is used, the main attention is given to the action itself, whereas the ‘have’ Perfect emphasizes the result of that action (Velkova 1987: 313). (5)

Toj ima dojdeno. he have:3SG arrive:PPP:NT ‘He has arrived.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

484

484

Nina Graves

Here, toj is the personal pronoun ‘he’ and dojdeno is an invariant (neuter) form of the Past Passive Participle. Besides the meaning ‘he has arrived’ the sentence carries the implicature of ‘he is here/there’. The participle of the ‘have’ Perfect does not agree with the direct object but is an invariant form, and it can be formed from both transitive and intransitive verbs (Friedman 1976: 99). As for the Third Perfect, according to normative grammar it can be formed only from intransitive verbs, since it acquires a passive meaning when formed from transitive verbs. The Third Perfect has a strong resultative meaning. Compare: (6) a.

b.

(intransitive) sum vraten be:1SG return:PPP:M ‘I have returned.’ (transitive) sum viden be:1SG see:PPP:M ‘I have been seen.’

4. Evidentiality Koneski’s (1987) grammar of the Macedonian literary language describes the evidentially indirect character shared by the ‘be’ and ‘have’ Perfects; namely, both of these perfects can carry the implicature of an action unwitnessed by the speaker, i.e., an action that the speaker has not experienced himself but is reporting as secondhand information. Evidentiality is a characteristic feature of several Balkan languages, such as Bulgarian, Turkish, and Albanian. Grammatical evidentiality distinctions are among the most important qualities that distinguish the perfect from the Aorist/Imperfect. Compare: (7) a.

b.

Aorist Toj padna he fall_down:AOR:3SG ‘[I saw when] he fell down.’ Perfect Toj padnal he fall_down:PPA:M ‘[I was told that] he has fallen down.’

So, the use of perfect conveys that the speaker did not see the situation him/herself, and so cannot confirm it. (Koneski 1986: 173).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

485

Macedonian – a language with three perfects?

485

The perfect does not always have to indicate a reported, unwitnessed action, and can also be used in direct speech. But the Aorist and the Imperfect always have to denote a witnessed or experienced action or an action that has become common knowledge and is widely accepted as truth (Koneski 1986: 174; 1987: 463–465). With the ‘be’ Perfect the witnessed action is principally the domain of the first person singular and plural, because use of the first person in itself indicates presence in the action or event being described. With the other two persons the use of the perfect in the description of witnessed action is rather rare (Koneski 1987: 467). Compare: (8)

1st singular Nikoga ne sum te molela kako sega za ova! never not be:1SG you:ACC beg:PPA:F as now for this ‘I have never begged you as I do now for this!’

(9)

3rd singular I najposle dojde glas: (toj) zaginal po and finally come:AOR:3SG voice (he) die:PPA:M during patot. journey:DEF ‘And finally comes the message: he died on the way.’ (Koneski 1987: 465, 479)

Evidentiality can also be expressed with the ‘have’ Perfect, by forming the l-participle of the auxiliary ima, resulting in a kind of double perfect: (10)

Si kazal grexovite sto imal storeno. he:DAT say:PPA:M sin:DEF:PL that have:PPA:M do:PPP:T ‘He told himself about the sins he had committed.’

However, this is not used as widely as the expression of evidentiality formed with the ‘be’ Perfect.

5. The use of different perfects in dialects In contexts where one of the typical elements for using a perfect tense (such as resultativity of a past event relevant to the present, or action completed in the past before the time of reference) was present, but where the action itself was also witnessed, informants of the southwestern dialect used the ‘have’ Perfect exclusively, if using a perfect at all.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

486

486

Nina Graves

With informants from the capital, Skopje (northern dialect), both the ‘have’ and ‘be’ Perfects were accepted. On the other hand, the informants from the eastern dialect used the ‘be’ Perfect exclusively. (11) a.

b.

SW/N (PFQ: 7) [Is it possible for anybody to swim in this lake?] Da, barem jas se imam plivano yes at:least I RFL have:1SG swim:PPP:NT vo nego. in it E/N (PFQ: 7) [Is it possible for anybody to swim in this lake?] Da, barem jas sum plivala nekolku yes at:least I be:1SG swim:PPA:F some ‘Yes, at least I have swum in it several times.’

nekolku pati some time:PL

pati vo nego. time:PL in it

The significance of the reflexive/nonreflexive contrast between examples (11a) and (11b), above has not been determined and requires further study. When the perfect is formed from a transitive verb, and thus has a direct object, it seems to be more appealing for speakers of all dialects to use the ‘have’ Perfect. (12)

E/N/SW (PFQ: 1) [Are there any of these books that your sister has READ already?] Da, taa ja ima procitano ovaa kniga yes she it:ACC have:3SG read:through:PPP:NT this book ‘Yes, she has read this book.’

When the perfect is formed with a transitive verb and the sentence has a possessive meaning, i.e., the subject possesses the result of the past action, all dialects, when using a perfect, choose the ‘have’ Perfect exclusively. An Aorist is also possible, but then the sentence emphasizes the past action, not the result of that action, and definitely not the possessivity of that result. (13) a.

E/N/SW (PFQ: 43) [I was told you collect dolls. You COLLECT many of them?] Dosega imam sobrano 200 kukli. up:to:now have:1SG collect:PPP:NT 200 doll:PL ‘I have collected 200 dolls by now.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

487

Macedonian – a language with three perfects?

487

This construction is quite similar to the verbal adjective. Even Serbian and Bulgarian, where the ‘have’ Perfect is a very marginal feature, form the sentence with the auxiliary ‘have’, though in those languages the participle has not yet become an invariant form and still agrees with the direct object. Note:

(13) b.

c.

Bulgarian (PFQ: 43) [I was told you collect dolls. You COLLECT many of them?] Imam sabrani okolo 200 kukli. have:1SG collect:PPP:PL about 200 doll:PL Serbian (PFQ: 43) [I was told you collect dolls. You COLLECT many of them?] Imam dosad 200 lutaka sakupljeni. have:1SG up:to:now 200 doll:GEN:PL collect:PPP:PL

In narrative contexts with definite time reference, all the informants used the Aorist and Imperfect rather than the perfect. The Macedonian perfect is thus not a principal narrative tense, though it is, of course, a narrative tense in (nonfirst person) unwitnessed narration like example (16).

(14)

N/E/SW (PFQ: 8, 9, 11) [This happened to me (a) an hour ago; (b) yesterday; (c) when I was a child.] Šetav vo šumata. Odednaš zgaznav na walk:IMPF:1SG in forest:DEF suddenly step:AOR:1SG on zmija. Taa me kasna vo nogata. Zedov snake she I:DAT bite:AOR:3SG in leg:DEF take:AOR:1SG kamen i go frliv kon zmijata. Taa stone and it:ACC throw:AOR:1SG at snake:DEF she umre. die:AOR:3SG ‘I was walking in the forest. Suddenly I stepped on a snake. It bit me in the leg. I took a stone and threw it at the snake. It died.’

When the perfect is used with a temporal adverbial, all the informants used the ‘be’ Perfect, even the informant with the southwestern dialect. Use of the ‘have’ Perfect in this context would have been ungrammatical, since it can occur only as an indefinite past.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

488

488 (15)

Nina Graves

N/E/SW (PFQ: 14) [It is morning. A wakes up, looks out of the window and sees that the courtyard is wet.] Vrnelo nok’eska. rain:PPA:NT last_night ‘It has rained overnight.’

The most interesting phenomenon in the use of different perfects in Macedonian dialects is the way that they occur in evidential contexts. An unwitnessed event was always expressed with the ‘be’ Perfect by all informants, even by an informant of the southwestern dialect. Thus: (16)

N/E/SW (PFQ: 60) [This happened to my brother yesterday (I did not see it, but he told me.] Si odel niz šumata. Naednaš zgaznal na he:DAT walk:PPA:M in forest:DEF suddenly step:PPA:M on zmija. Go kasnala za noga. Zel kamen i snake he:ACC bite:PPA:F in leg take:PPA:M stone and go frlil po nea. Ja ubil. it:ACC throw:PPA:M at she:ACC she:ACC kill:PPA:M ‘He was walking in the forest. Suddenly he stepped on a snake. It bit him in the leg. He took a stone and threw it at the snake. It died.’

Another example: (17)

N/E/SW (PFQ: 62) [A tells what she has heard from her father. Nothing shows that she would not believe it.] Koga tatko mi bil dete školite bile when father I:DAT be:PPA:M child school:DEF:PL be:PPA:PL podobri od segašnive. better than present:one:DEF:PL ‘When my father was a child, schools were better than nowadays.’

The Third Perfect expresses a strong resultativity as explained in the grammar section. The following sentence is a good example of a context where all the informants accepted the use of the Third Perfect. For most of the informants it was the primary choice.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

489

Macedonian – a language with three perfects?

(18)

489

N/E/SW (PFQ: 3) [Is the king still alive?] Ne, umren e. No die:PPP:M be:3SG ‘No, he is dead.’

In the southwestern dialects the Third Perfect can also be formed from transitive verbs and the construction still has an active meaning. (19)

SW (Koneski 1986: 199; 1987: 443–444). Jadi ti, sinko, jas sum jadena i tamo. eat you son I be:1SG eat:PPP:F and there ‘Just you eat, son; I have already eaten elsewhere.’

6. The distinctive use of perfects in the dialect of Ohrid The previous section described the uses of, and the main differences among, the three forms of the perfect tense in various Macedonian dialects. This section focuses on the most interesting of the dialects – that of Ohrid. The Ohrid dialect falls within the southwestern dialect area and is, for the purposes of this paper, the most important dialect in that area, at least among those in the Republic of Macedonia. (In the previous sections the informant for the examples from the southwestern dialect is a speaker of the Ohrid dialect.) In the Ohrid dialect each of these three different perfects has its own characteristic meaning, even though some overlap exists, at least in theory, between the ‘have’ Perfect and the Third Perfect. One interesting possibility, which shows the stage of development of both the ‘have’ and Third Perfects as used in the Ohrid dialect, is the formation of these perfects from the auxiliaries, as in the following examples of the ‘have’ Perfect and the Third Perfect, respectively: (20) a.

b.

Ohrid dialect Imam bideno have:1SG be:PPP:NT ‘I have been’ Imam imano have:1SG have:PPP:NT ‘I have had’

as well as in the following example of the Third Perfect:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

490

490

Nina Graves

c.

Sum biden be:1SG be:PPP:M ‘I have been’

The most characteristic context for using the Third Perfect in the Ohrid dialect arises when one wants to emphasize a strong resultative meaning in a sentence. And, as explained earlier, a peculiarity of this dialect is that one can form the Third Perfect even from transitive verbs without the sentence having a passive meaning: (21)

Ohrid dialect (Koneski 1987: 443; 1986: 199) Jadi ti, sinko, jas sum jadena i tamo. eat:IMP you son I be:1SG eat:PPP:F and there ‘Just you eat, son; I have already eaten elsewhere.’

The ‘have’ Perfect is by far the most common perfect used in the Ohrid dialect. It carries with it all the characteristic features of the perfect tense: resultativity, result of a past action relevant to the present, and action completed all in the past before the time of reference. Furthermore, not only is it used in a wider variety of contexts than either of the two competing perfect types, but it is usually the primary if not the only perfect form chosen by the informant of the Ohrid dialect. Below are some examples of typical perfect contexts: a) Resultativity: (22)

Ohrid dialect (PFQ: 37) [It is cold in the room. The window is closed.] Go imaš otvoreno prozorecot? it:ACC have:2SG open:PPP:NT window:DEF ‘Have you opened the window?’

b) A resultative and possessive context, where the perfect is being formed with a transitive verb: (23)

Ohrid dialect (PFQ: 42) [I was told you are writing a book. How many pages you WRITE by now?] Imam napišano pedeset stranici. have:1SG write:PPP:NT fifty page:PL ‘I have written fifty pages.’

c) The result of a past action relevant to the present:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

491

Macedonian – a language with three perfects?

(24)

491

Ohrid dialect (PFQ: 6) [Do you know my sister?] Da, se imam zapoznaeno so nea. Yes, RFL have:1SG get:acquainted:PPP:NT with she:ACC ‘Yes, I have met her.’

So, in the Ohrid dialect both the Third and ‘have’ Perfects are used in contexts typical of the perfect tense. The Third Perfect is by its nature more strongly resultative. It can also be used in some experiential contexts, although that is not one of its characteristic uses. The informant of the Ohrid dialect provided these examples: (25)

Ohrid dialect (PFQ: 32) Bidena si vo Avstralia? be:PPP:F be:2SG in Australia ‘Have you been to Australia?’

(26)

Ohrid dialect (PFQ: 33) Ne, nikogaš ne sum bidena tamu. No never not be:1SG be:PPP:F there ‘No, I have never been there.’

The ‘have’ Perfect is less marked in meaning, and is thus more commonly used. This leaves only the old ‘be’ Perfect unaccounted for. The old ‘be’ Perfect with the l-participle can be used in some situations where the use of the ‘have’ Perfect would not be possible, as in the already familiar example: (27)

Ohrid dialect (PFQ: 14) [It is morning. A wakes up, looks out of the window and sees that the courtyard is wet.] Vrnelo nok’eska. rain:PPA:NT last_night. ‘It has rained overnight.’

Here the ‘be’ Perfect was preferred over the ‘have’ Perfect because the temporal adverbial is inconsistent with the indefinite past nature of the ‘have’ Perfect and because the example conveys evidentially indirect information. In the Ohrid dialect the old ‘be’ Perfect tends to occur so often in such evidentially indirect contexts that its current use in this dialect can be said to be limited to expressing reported events that the speaker has not experienced himself. The form thus carries the implicature that the speaker is just passing on information that he/she has received from elsewhere. In the Perfect Questionnaire there were fifteen sentences

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

492

492

Nina Graves

expressing various kinds of evidentially indirect information; in the Ohrid dialect all of them were expressed by the l-participle, i.e., by the old ‘be’ Perfect. Compare: (28)

Ohrid dialect (PFQ: 68) [Said by a person who has just heard about the event but has not seen it:] Sestra mi mi reˇce deka došol kralot. sister I:DAT I:DAT tell:AOR:3SG that arrive:PPA:M king:DEF ‘My sister told me that the king has arrived.’

Another example: (29)

(PFQ: 73) [A guide, showing ruins to tourists:] Ovoj bil golem grad. this be:PPA:M big town ‘This was a big town.’

In some situations a sentence formed with the l-participle, besides expressing evidentially indirect information, can also reveal the speaker’s own attitude about the sentence; that is, the sentence can carry a dubitative implicature. Thus: (30)

Ohrid dialect (PFQ: 65) [A doubts what her father has told her.] Tatko mi veli deka školite bile podobri father I:DAT say:3SG that school:DEF:PL be:PPA:PL better koga toj bil dete. when he be:PPA:M child ‘My father claims the schools were better when he was a child.’

Friedman has an excellent example of the use of the old ‘be’ Perfect in a dubitative context, and it would certainly be accepted by speakers of the southwestern Ohrid dialect: (31)

(Friedman 1981: 15) Toj povek’e od tebe znae za boksiranje. – Toj he more of you know:3SG about boxing. he povek’e znael! more know:PPA:M ‘He knows more about boxing than you do. – He knows more, indeed!’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

493

Macedonian – a language with three perfects?

493

There is a very interesting example, where the ‘be’ Perfect was used as the primary choice, but where the informant provided as a secondary choice a ‘be’ form made from the ‘have’ construction, resulting in a double perfect, which in meaning approaches the pluperfect. Nevertheless, this type of ‘have’ form is based on the ‘be’ form and thus similarly implies nonconfirmative, reported speech: (32) a.

b.

Ohrid dialect (PFQ: 81) [Looking at a picture of a house which has been torn down:] a typical ‘be’ form: Koj ja izgradil ovaa kuk’a? who it:ACC build:PPA:M this house a ‘be’ form from the ‘have’ construction: Koj ja imal izgradeno ovaa kuk’a? who it:ACC have:PPA:M build:PPP:NT this house ‘Who has built this house?’

7. Conclusions It seems that in most dialects of Macedonian the use of a particular perfect in a specific context has not yet stabilized. The ‘be’ Perfect is most often used in evidentially indirect contexts, the ‘have’ Perfect in possessive and resultative contexts and the Third Perfect in strongly resultative and sometimes experiential contexts. But in most ordinary perfect tense contexts, the ‘be’ and the ‘have’ Perfects (and with some more restrictions the Third Perfect) are equally possible. This is the situation in the majority of dialects. One of the extreme dialects is the northern dialect, where, most likely due to its close contact with Serbian, the use of the ‘have’ Perfect is practically limited to the previously mentioned possessive context and the Third Perfect is limited to extreme resultative contexts, such as death (see examples). The ‘be’ Perfect was virtually the only perfect tense used in this dialect, with little challenge from the ‘have’ and Third Perfects. The other extreme is the southwestern dialect, which freely allows the ‘have’ Perfect to be formed also from auxiliaries and allows the Third Perfect to be formed from transitive verbs without this form signifying a trace of the passive meaning that the grammars describe. The ‘be’ Perfect is used only in contexts with a specific time reference and, of course, in evidentially indirect contexts, showing the information to have been unwitnessed by the speaker. In the Perfect Questionnaire there were fifteen sentences expressing various kinds of evidentially indirect information. One would have expected the informant with the southwestern dialect to prefer a ‘have’ construction here, since these constructions are a speciality of that dialect. Instead

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

494

494

Nina Graves

she used the ‘be’ counterpart in every sentence. Given this, one can seriously question whether the ‘be’ form, in this dialect, is a tense at all. The context of evidential information suggests it is a mood (an evidential) rather than a tense, and its use in a context with a specific time reference is not a feature characterizing the perfect tense. It remains to be seen whether or not the ‘be’ and ‘have’ Perfects in other dialects of Macedonian will eventually follow the southwestern dialect by starting to use the old ‘be’ Perfect as a mood and by transferring the meaning of the perfect tense onto the ‘have’ Perfect. The tendency in most dialects seems to be that the ‘have’ Perfect is gaining ground at the expense of the ‘be’ Perfect.

Acknowledgements I would like to use this opportunity to thank Prof. Jouko Lindstedt of the University of Helsinki for his help and guidance in my study for this paper. Most of the examples given here are from the Perfect Questionnaire, designed by Prof. Lindstedt for E UROTYP’s Tense and Aspect Theme Group. I would also like to express my sincere thanks for the kindness I encountered at the University of Kiril and Metodi in Skopje where, on very short notice and during the busiest exam period in May, staff of the Department of Macedonian Language, particularly Elena Petroska, Tomislav Trenevski, and Katerina Veljanovska, as well as students at the Department of English Language, including Larisa Boškova, Dejan Georgievski, Katica Koteska, and Biljana Mitreva, managed to find time to fill out the extensive Perfect Questionnaire.

References Arsova-Nikoli´c, L. 1973 “Makedonski ekvivalenti na Angliskiot segasen perfekt” [Macedonian equivalents for the English present perfect], Makedonski jazik 1973: 157–172. Friedman, Victor 1987 “The Macedonian perfect”, in: Steever, Walker & Mufwene (eds.), 96–104. 1981 “Admirativity and confirmativity”, Zeitschrift für Balkanologie 17/1: 12–28. Koneski, Blaže 1986 Istorija na Makedonskiot literaturen jazik. [The history of the Macedonian standard language.] Skopje: Kultura. 1987 Gramatika na Makedonskiot literaturen jazik. [Grammar of the Macedonian standard language.] Skopje: Kultura. Steever, Sanford B., Carol A. Walker & Salikoko S. Mufwene 1987 Papers from the parasession on diachronic syntax. Chicago: Chicago Linguistic Society. Velkovska, Suzana 1987 “Ima-konstrukciite vo tekstovite na Marko K. Cepenkov” [Ima constructions in the texts of Marko K. Cepenkov] , Makedonski Jazik 1987: 301–314.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

495

Marja Leinonen and Maria Vilkuna

Past tenses in Permic languages

1. Introduction

Komi and Udmurt (formerly called Zyrian and Votyak, respectively) form the Permic branch of the Uralic, or Finno-Ugrian, language family. The languages are spoken in the northeasternmost part of Europe. The approximately 250,000 Komi speakers inhabit a large area between the River Severnaya Dvina and the Ural Mountains, and considerable numbers live outside this area. The Udmurt, who number about 500,000, live to the south of the Komi, in the area of the Rivers Vyatka and Kama. A southeastern dialectal variant of Komi, Permyak, has its own standardized literary language; Permyak is not treated in this paper, because its tense system does not differ from Standard Komi. Komi and Udmurt are transparently related, but not mutually intelligible. The differences between Komi and Udmurt are to a considerable extent due to the more marked Turkic influence on the lexicon and structure of the latter. The Permic languages belong to the large area where an evidentiality-based opposition exists between two past tenses (Haarmann 1970, Dahl 1985: 152). This paper is a short survey of Permic past tenses with special reference to this phenomenon. The data for this purpose come from modern fiction, dialect samples and, in the case of Udmurt, from a sizable corpus of present-day fiction1. In addition, native speakers have answered the questions presented in the EUROTYP Perfect Questionnaire (PFQ). The Permic verb is morphologically marked for mood, tense and person, and has a number of derivational Aktionsarten. The simple (non-compound) tenses are the Present, the Future (coinciding with the Present in 1st and 2nd persons in Komi), and two past tenses. The opposition of these two is generally characterized in terms of evidentiality, or “witnessing”, in the literature. The first past, henceforth PST1, is called the “witnessed” past, implying that the speaker has first-hand knowledge of the event. Of the two pasts, it is the one with a wider applicability and can perhaps be characterized as the default choice for the expression of past events. The second past, PST2, is generally characterized as the “unwitnessed past”, sometimes as perfect. The basic paradigms of the two simple pasts of Komi s’etny and Udmurt s’otyny ‘to give’ are given below:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

496

496

Marja Leinonen and Maria Vilkuna

PST1 KOMI

SG PL UDMURT SG PL PST2 KOMI SG PL UDMURT SG PL

1ST s’eti s’etim s’oti s’otim(y) (-?) (-?) s’otis’kem s’otis’kemmy

2ST s’etin s’etinnyd s’otid s’otidy s’etömyd s’etömnyd s’otem(ed) s’otil’l’am(dy)

3RD s’etis s’eti s’etisny s’otiz s’otizy s’etöma s’etöm s’etömaös’ s’etömny s’otem(ez) s’otil’l’am(zy)

PST2 is based on a past participle form with the ending -öm in Komi, -(e)m in Udmurt. The same form sometimes occurs as a deverbal noun, so that words such as Komi kulöm and Udmurt kulem can mean both ‘death’ and ‘die:(PAST)PART’. The past participle has an important role in Udmurt sentential complementation, and it is widely used as a noun modifier in both languages, as illustrated in the following examples. The participial suffix can be added to both transitive and intransitive verbs. Thus the inherent lexical meaning of the verb and the surrounding context dictate its interpretation as “active” or “passive”. (1) a.

b.

Komi2 kulöm mort die:PART person ‘a dead person’ puöm jaj cook:PART meat ‘cooked meat’ Udmurt dun’n’ejez addz’em murt world:ACC see:PART person ‘an experienced (widely travelled) person’ addz’em murt see:PART person ‘a person one has seen [but does not remember the name of]’

In addition to the -öm/-(e)m participial suffix, the PST2 paradigm contains markers of predicativity, number and person. The Komi -a is presumably associated with an adjectival derivative suffix, and -ös’ is the normal plural predicative agreement marker for adjectives. The corresponding Udmurt agreement suffix -es’ does not combine with the participle. The Udmurt PST2 paradigm consists of both simple participial forms with possessive suffixes and forms with additional derivational material (-is’k-, -il’l’a-). In both languages, the PST2 person endings (-yd, -nyd in Komi and -ed, -ez, -my -dy and -zy in Udmurt) are identical to possessive suffixes. Comparing the two

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

497

Past tenses in Permic languages

497

paradigms shows that the PST1 person endings are close to the possessive suffixes as well, but the relation is more transparent in the PST2 paradigm. In Udmurt, the PST2 person suffixes are not obligatory except in 1st person plural. Note that even though PST2 is participle-based, neither language uses a copula with this form. For present tense, only a locative-existential copula exists. The past tense copulas combine freely with different tensed verb forms yielding compound tenses, some of which will be briefly discussed later in this paper. The PST1 and PST2 forms of the copulas, which will be called AUX1 and AUX2 respectively, are as follows: Komi Udmurt

COP:PST1 = AUX1 völi val

COP:PST2 = AUX2 völöm vylem

In what follows, we discuss the role of evidentiality and resultativity in the Komi and Udmurt PST2 and address the question whether PST2 can appropriately be called perfect in the sense of the working definition used in PFQ, that is, as a gram used in crucial contexts in the questionnaire. One important difference between the two languages under consideration is the use of PST2 in narration; this is common in Udmurt but not in Komi. The languages also differ in the way resultativity is expressed. The Komi PST2 forms are able to convey both evidential and resultative senses, while Udmurt resorts to a specific resultative construction, viz. the inessive form of the participle, to express resultativity. On the other hand, even Komi has a way of differentiating between the two senses, as resultative sentences are often rendered with an impersonal (possessive) construction. The third difference is the existence in Udmurt of a special optional construction for the so-called experiential perfect.

2. Evidentiality The traditional characterization of PST2 as an “unwitnessed tense” indicates that the speaker did not witness the event personally but reports it on the basis of indirect evidence. This evidence may consist of either hearsay or inference from evidence subsequent to the event, such as some result of the event at the time of speech. To conform to the usage in this volume, we will use the term ‘indirective’ instead of ‘unwitnessed’, but use ‘evidentiality’ as a cover term. In general, both Komi and Udmurt informants seem to be guided in their choice of forms by considering whether or not the events were witnessed by the speaker. Still, in the case of second-hand or inferred information, both past tenses are possible, depending on whether or not the speaker wishes to express confidence in the reliability of his information. Examples:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

498

498

Marja Leinonen and Maria Vilkuna

(2) a.

b.

Komi (PFQ: 26) [What do you know about this book?] Sijös gižis Grem Grin. it:ACC write:PST1 Graham Greene ‘Graham Greene wrote it.’ (‘I am convinced of it’) Sijös Grem Grin gižöma. it:ACC Graham Greene write:PST2 ‘Graham Greene wrote it.’ (‘according to what I heard’)

Thus, the actual reliability of the information is not the issue, but rather whether the speaker presents the information as his own “responsibility”. Note that variations in word order do not affect the interpretations. The following context illustrates possible grounds for the selection between PST1 and PST2 and the role of inference. The question presents a supposition based on the present state, and PST2 is normally used, as in (3a). PST1 can be used in the question if the action is presupposed and the question focuses on the actor, as in (3b). The answer is, however, in PST1, as the speaker normally has direct evidence of his own actions. (3) a.

b.

c.

Komi (PFQ: 37) [It is cold in the room. The window is closed.] Te vos’tlömyd öšin’sö? you open:PST2 window:ACC ‘Did you open the window [and close it again]?’ (‘perhaps, in my absence’) Te vos’tlin öšin’sö? you open:PST1 window:ACC ‘Did you open the window [and close it again]?’ (‘Was it you who ?’) Da, me vos’tli sijös. yes I open:PST1 it:ACC ‘Yes, I opened it.’ (‘I did and was conscious of doing it.’)

In the following example, the narrator, watching a man sawing wood, draws a conclusion about his past actions. Without the frequentative suffix, PST2 would simply mean that the action seems to be or must be completed: the frequentative affix invokes, as a pragmatic implication, an “experiential” meaning of the past tense. (4)

Komi (V. Beznosikov/Vojvyv kodzuv 5/1963) Tödts’ö: mortys pil’itts’yvlöma n’in, i, tydalö, n’e is_felt man:3SG saw:FREQ:PST2 already and it_seems not ötts’ydys’. once ‘It is obvious that the man has sawed before, and more than once.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

499

Past tenses in Permic languages

499

If the present state shows a result that is the only evidence of a past event, PST2 is used even with temporal adverbials: (5)

Komi (PFQ: 14) [It is morning. A wakes up, looks out of the window and sees that the courtyard (or the street) is wet.] Vojnas zeröma. night:INESS.3SG rain:PST2 ‘It has rained at night.’

An equally unambiguous PST2 context is PFQ: 69, where one encounters a broken window and says, ‘The thief has entered by this window’. Perhaps a typical example of how the past tenses work in Udmurt texts can be seen in the following sample from a story about a schoolboy called Mat’i. The main storyline, here represented by (6a), (d) and (g), is rendered with PST1, although “historical” present is common as well, as in (6c). We will later turn to the resultative form in (6b). Of immediate interest at this point are (6e) and (f), where the author has Mat’i infer a past event, the falling of snow, from its results, and uses PST2. The example also shows how PST2-type time reference can relate to a previous event as well as to a present one. No pluperfect is used here, as it would be in Finnish. (6) a. b. c. d. e. f. g.

Udmurt (corpus) sajkaz. Tšukaz’ejaz vaz’ ik next_morning early PRT wake_up:PST1:3SG Anajez koškemyn in’i. mother:3SG leave:PART:INESS already Suzerez iz’e na. sister:3SG sleep.PRS still Mat’i ukno dory myniz: kytše kuaz’? M window to go:PST1:3SG what_like weather Ujin lymy us’em, at_night snow fall:PST2 kud-og intyosy l’ukjos puks’il’l’am. some places:ILL snowdrift:PL sit:PST2.PL Dis’as’kysa pijaš kuasen urame potiz. dress:GER boy ski:INST street:ILL go_out:PST1:3SG ‘[Mat’i] woke up early the next morning. His mother is already gone. His sister is still sleeping. Mat’i went to the window: what was the weather like? It had snowed during the night, snowdrifts had formed in some places. The boy dressed and went out to the street on his skis.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

500

500

Marja Leinonen and Maria Vilkuna

The same is true for Komi: PST2 is inserted into the past event series just as easily as into the historical present:

(7)

Komi (Žilina & Sorvaˇceva 1971: 53) Ses’a ji olöm mys’t’i bara loktis joma-baba. Sie then PRT life after again come:PST1 witch she vetlöma kuz’n’its’aö. go:PST2 smithy:ILL ‘After some time, the witch came again, (it appears) she had been to the smithy.’

Serebrennikov (1960) calls this use the ‘impressive perfect’, and notes that in some other languages it corresponds to the pluperfect. That is, the PST2 event precedes the past event currently in focus, leaving a resulting state that is in force at the stage that the speaker is referring to. In a narration cast in PST1, then, PST2 typically implies a “break” in the sequence of events. Things are different when PST2 is used in narration. This does not happen much in modern fiction, but it was the norm in the 1912 Udmurt New Testament (Svjatoe evangelie)3 and occurs quite often in Udmurt dialect samples, where elderly speakers are interviewed by students, and the main interests are tradition, past events and tales the interviewee has heard people tell during his life. (This is why Udmurt speakers sometimes characterize PST2 as pertaining to “old” events, even though the tense is not a remote past.) Personal experiences are rendered with PST1 (with frequent present tense and compound tenses). An example:

(8) a. b. c. d. e.

Udmurt (Kel’makov & Saarinen 1994: 238)4 odik kyšnomurtlen kartez vojne byrem. one woman:GEN husband:3SG war:ILL die:PST2.3SG solen bydes jurt nylpiosys kyl’il’l’am. s/he:GEN whole house child:PL:3SG remain:PST2.3PL okpol džytadz’e t’at’azy bertem. once evening father:3PL return:PST2.3SG Van’zy s’is’kyny puks’il’l’am. all:3PL eat:INF sit:PST2.3PL ogezlen vilkez us’em. one:3SG:GEN fork:3SG fall:PST2.3PL

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

501

Past tenses in Permic languages

f.

501

džök ule mykyrkkem no, otyn, pe, iskal pydjos.5 table under bend_down:PST2 and there PRT cow foot:PL ‘One woman’s husband died in the war. She was left with a house full of children. One evening their father returned. They all sat down to eat. One of them dropped his/her fork. S/he bent down under the table and there, they say, were cow’s hooves.’

The Komi PST2 is less common in narration, although fairytales and other past sequenced events can be presented as hearsay. In actual practice, PST1, historical present or future tenses are used, and PST2 indicates a break in the main line of narration. The Permic PST2 also covers such extensions of the notion of indirectivity as sudden revelations or reinterpretations of one’s experience or action: something, surprisingly, turns out to be something else. Having supposed something to have been the case, one is distanced from one’s conviction by direct perception or other people’s opinions. The inferential of surprise is easiest to separate from resultative and non-witnessed meanings in atelic verbs in 1st and 2nd persons. An example:

(9)

Komi (Juškov 1970: 97) Mis’a, gaškö, te munin n’in. A tani na I_think perhaps you go:PST1:2SG already but here still völömyd. be:PST2:2SG ‘I thought you had already gone, but here you still are.’

Analogous Udmurt examples can be found in tales where the characters’ honesty is tested (e.g., Kel’makov & Saarinen 1994: 276). In the following example, the main character of a folk tale, a rabbit, realizes his methods of fighting against his enemies have been misguided and says:

(10)

Udmurt (corpus) Eh, mon šuz’i vylis’kem. oh I fool COP:PST2.1SG ‘Oh, I have been foolish.’

Sudden revelation need not be evaluative; in the following, its object is the speaker’s own unconscious action:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

502

502 (11)

Marja Leinonen and Maria Vilkuna

Udmurt (Kel’makov & Saarinen 1994: 194) pöjasal ke, pöjasal, dyr, no, pöjal’l’as’kon trick:COND if trick:COND sure but trick_making pujyme gurte kel’tis’kem bag:ACC.1SG village:ILL leave:PST2.1SG ‘I would indeed trick you, but I left my trick bag at home. [Said by trickster to a passer-by who has asked to see one of his tricks.]’

Recent research on evidentiality provides us with similar conceptual categories based on concrete linguistic forms. Thus, connected with indirect evidence is also the idea of the ‘unprepared mind’ (surprise, doubt) presented for evidentials in Turkish (Aksu-Koç & Slobin 1986: 1673) with an apt characterization: Normal experience is characterized by premonitory consciousness of the contents of coming moments [ ] When a mind is unprepared, however, events cannot be assimilated at once. The speaker stands back, saying, in effect, by use of the [ ] form: ‘It seems that I am experiencing such-and-such’ or ‘It seems that such-and-such must have taken place.’ [ ] The event has become apparent through its consequences, or through verbal report; or the experienced event is radically different from the consciousness that preceded the experience. (Aksu-Koç & Slobin 1986: 164)

‘Surprise’ does not equal unexpectedness per se, but rather the ‘unprepared mind’, the speaker stating a fact that goes against possible preconceptions of an opposite state of affairs. Having noted that the ‘sudden revelation’ reading of PST2 is often most explicit in sentences with one of the speech act participants as subject, we should now take a closer look at 1st person PST2 in general, something that is perhaps too easily ruled out if one only considers the most concrete variants of indirectivity. Evidentiality in its “witnessed” sense is obviously not frequently associated with 1st person, as the speaker can usually be assumed to be conscious of his or her actions. There is indeed some controversy concerning the acceptability of PST2 1st person forms in Komi; they do exist, but only in dialects (Cypanov 1992). In form, they are identical with the 3rd person verb forms s’etöma, s’etömaös’, while Udmurt has developed a specific morphology in 1st persons. According to Batalova (1982: 141), the form is not found in Permyak. In Udmurt, the existence of the 1st person forms of PST2 is not in doubt, although they are infrequent. Examples were previously met in connection with “sudden revelation”, but a speaker can also use 1st singular PST2 when reporting incidences during less than full consciousness, such as (12):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

503

Past tenses in Permic languages

(12)

Udmurt Mon ujbyrtis’kem I sleepwalk:PST2.1SG ‘I have walked in my sleep.’

503

ujbyrtysa vetlis’kem. sleepwalk:GER go:PST2.1SG

A rather revealing narrative with “non-witnessed” PST2 can be found in Kel’makov and Saarinen (1994: 212). The speaker and his friend Igor had been teasing a ram. The story goes on: “The ram ran (PST1) around and probably got tired (PST1). Our ram was (PST2) wiser than us6: it ran (PST1) towards me and hit (PST1) its horns against my forehead.” From this point on, the speaker goes on using PST2 for events during which he was unconscious: “I fell down immediately, but Igor ran away, frightened. I lay there a long time. My mother came from work and saw: I’m lying (PRS) in front of the shed. She quickly splashed me with water and took me into the house. But I still have (PRS) a mark on my forehead.” Similar cases occur in Komi dialects. Baker (1983) points out that PST2 occurs if the speaker is somehow “distanced” from the event–does not remember it, perhaps did not take it into account. The examples are analogous to the Udmurt (10)–(12). The usual Russian translation resorts to the predicate okazyvaetsja ‘it turns out’. The narrator’s non-involvement is emphasized by Baker in the dialectal example Me vojnas ts’ets’yllöma da pats’lö lontyllöma ‘I must have got up in the night and fired the stove’ (Baker 1983: 73–80). Lack of control of action on one hand, the state meaning on the other, in the 1st person forms in the dialects are pointed out in Bartens (1982). These nuances of meaning can clearly be accommodated under the heading of indirectivity. However, the Komi first person PST2 apparently need not be indirective, at least not in the spoken varieties. In the following section, we turn to the other facet of PST2, its resultativity, which distinguishes Komi from Udmurt.

3. Resultativity Cypanov (1992) reports further examples of the 1st person form from prose writers whose language reflects their dialectal background. The speaker merely reports on his present state, the result of a past action which he certainly must have been conscious of: (13)

Komi (Cypanov 1992: 26/B, Šahov, Vojvyv kodzuv 1990/8: 33) Me Tol’a, te talun stolövöjad ötnad vetly Tol’a you today dining-room:ILL:2SG alone go:IMP I s’ojöma-juöma. eat:PST2.1SG-drink:PST2.1SG ‘Tolya, go to the diningroom alone today. I have eaten (and drunk).’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

504

504

Marja Leinonen and Maria Vilkuna

In fact, the use of 1st person PST2 forms of certain verbs in certain contexts seems to be fairly normal in conversation. The speaker offers explanations, conclusions, motivations, as in the following example elicited from an informant: (14)

Komi uz’i n’in, tyrmas. Me uz’öma sleep:PST1 already be_enough:FUT.3SG I sleep:PST2 ‘I have already slept enough.’

The range of Komi verbs that can be used with the 1st person forms without indirective meaning seems to be restricted. According to our non-conclusive elicitations, they signify either everyday activities–in addition to the above examples reflexives me myss’öma, pyvs’öma, pas’tas’öma ‘I have washed myself, been to the sauna, got dressed’, in the ‘already, enough’ context–or a few other resultative states, traditionally seen as participles, e.g., me ranitts’öma ‘I am wounded’. Thus, it appears that the Komi PST2 has a use that should be characterized as Resultative in the sense of Nedjalkov & Jaxontov (1988): it signifies a state resulting from a past action, not past reference in a strict meaning. In principle, the construction is strictly intransitive, its subject representing the subject of an intransitive verb or the object of a transitive verb. In the latter case, the meaning is passive. Recall that the Permic past participle is neutral as to active or passive orientation (for this term, see Haspelmath 1994). We may compare the occasional 1st person “non-indirective” usage with Russian, where dialectal gerundial forms of generally perfective non-transitive verbs with the suffix -ši are used as predicates in situations comparable to examples (13–14). Although the basic dialectal area where the forms are extensively used is rather far from Komi in the West, certain verb forms are widespread everywhere in Russia and are used in substandard urban speech as well. The most frequent verb forms, e.g., vypivši ‘drunk’, odevši ‘clad’, (ne) evši ‘(not) eaten’, (ne) pivši ‘(not) drunk’, (ne) spavši ‘(not) slept’ (Kuz’mina 1993: 142–146), are nearly lexicalized to signify a state – in the dialects where the form is used widely it has a clearly resultative meaning (cf. Trubinskij 1988). It would seem that in both languages, the exceptional verbs are either reflexive or transitive “active-oriented”, i.e., the agent is saliently affected by the action (Haspelmath 1994: 161). Although the resultative meaning is based on the past participle form in both Komi and Udmurt, it is a point of difference between the languages. Udmurt does not resort to PST2 in cases like (13–14) but uses a specific resultative construction, whose predicate consists of the inessive form of the past participle. This form in itself denotes a present state and can be turned into a past by using one of the copular auxiliaries.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

505

Past tenses in Permic languages

505

The two types of the resultative are illustrated below. In (15), it is formed from a transitive verb and takes on a prototypical passive meaning, as it is the patient argument whose state is predicated upon and therefore appears as the subject. In (16), however, the verb is lexically intransitive, and the only argument is the subject of the construction. The examples in (16) are translations of those Komi cases where the resultative meaning is at its clearest. (15) a.

b.

(16) a.

b.

c.

d.

Udmurt (corpus) Ukno us’temyn. window open:PART:INESS ‘The window is open.’ Ved’ ta as’me udmurt kylyn gožtemyn?! PRT this own:1PL Udmurt language:INST write:PART:INESS ‘This is written in our own Udmurt language.’ Udmurt Mon s’iis’kemyn. Köt tyremyn. I eat:PART:INESS stomach fill:PART:INESS ‘I have eaten. My stomach is full.’ iz’emyn n’i. Mon dyšetskemyn sleep:PART:INESS already I study:PART:INESS ‘I have already studied slept (enough).’ Mon mis’tis’kemyn, munts’oje vetlemyn. I wash(INTR):PART:INESS sauna:ILL go:PART:INESS ‘I have washed, been to the sauna.’ Mon ug kynmis’ky, mon šunyt dis’as’kemyn. I NEG:PRS catch_cold I warm dress(INTR):PART:INESS ‘I won’t catch a cold, I’m warmly dressed.’

The essential intransitivity of the Udmurt resultative construction is displayed by the fact that the corresponding form for Komi (13) cannot be formed from the usual transitive verb s’iyny but from its intransitivized counterpart s’is’kyny ‘to eat: to have a meal’, as in (16a). As is to be expected, the Udmurt resultative requires that a change of state or affectedness can somehow be attributed to the subject of the construction. Thus, no examples of the form with the verb ‘to sneeze’ could be constructed. On the other hand, the affectedness interpretation is not always immediately obvious. Cases in point are valamyn ‘understood’, todemyn ‘known’, or jaratemyn ‘loved’: that thing has been correctly understood, the thing is commonly known, or the person is much loved. By way of summary, it is instructive to consider the three following ways of saying one has caught a cold in Udmurt, using the verb kynmyny ‘to get cold’:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

506

506

Marja Leinonen and Maria Vilkuna

(17) a. b. c.

Udmurt PST1: Mon kynmi. PST2: Mon kynmis’kem. Resultative: Mon kynmemyn.

The PST1 variant places the event in a sequence experienced by the speaker: she is aware of the occasion when she caught the cold: ‘I caught a cold’. The PST2 in (17b), with its 1st person subject, could mean ‘I have caught a cold’ with the implication ‘I don’t know when or where it happened’. Finally, (17c) is the normal way of reporting one’s physical state: ‘I have a cold’. At this point, it should be added that even Komi has a particular construction for the expression of resultativity. This is the impersonal form in (18) with the subject as a Possessor (Genitive case, sometimes also called Adessive); it appears to be more frequent than the simple PST2 construction in (14). For indirectivity or nonintentionality in the 1st person, the reflexive impersonal construction illustrated in (19) is used: (18)

Komi Menam uz’öma n’in. I:GEN sleep:PST2 already ‘I have already slept [no need for more].’

(19)

Komi (Juškov 1970: 175) Menam pöz’avs’öma že. I:GEN sweat:RFL:PST2_3SG PRT ‘I have become sweaty.’

Unlike the Udmurt resultative, however, the possessive construction also accepts transitive verbs. The patient may be marked as the surface subject as in (20b), or as the object, using the Accusative case as in (20c): (20) a. b. c.

Komi (PFQ: 02) Sijö lydd’is n’in vot tajö knigasö. She read:PST1 already PRT that book:ACC So tajö knigays sylön lydd’öma n’in. PRT that book:3SG she:GEN read:PST2 already So tajö knigasö sylön lydd’öma n’in. PRT that book:ACC she:GEN read:PST2 already ‘[It seems that your sister never finishes books. – That is not quite true. ] She has read through this book (already).’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

507

Past tenses in Permic languages

507

In Udmurt, this context systematically elicits PST1. Note that the Komi impersonal possessive construction is conspicuously similar to the North Russian dialectal construction with Accusative or Nominative objects, unvarying non-congruent past participle and Possessor (cf. Bartens 1982, Trubinskij 1988). Whether this similarity is a result of language contact remains unclear. In summary, the Komi PST2 names the result of a past event and, like the perfect of most Indo-European languages, refers predominantly to the moment of speech. It is resultative and/or indirective in specific contexts, though the lexical content of the verb predisposes the interpretation: telic verbs (2nd or 3rd person) produce resulting states, and one-argument situations with salient states do not leave much need for suppositions:

(21)

Komi (Leonik Palkin/Vojvyv kodzuv 5/1963: 36) Ton’a akan’ kod’: s’ödov jurs’isö mits’aa synalöma, Tonya doll like dark hair:ACC prettily comb:PST2.3SG kösa kyöma. plait plait:PST2.3SG ‘Tonya is like a doll; her hair is prettily combed and plaited.’

Atelic verbs (notably ‘to be’), on the other hand, produce no stable subsequent state, and thus bring forth the indirective interpretation, as in the following example (a). The same applies to telic verbs with agents which do not “carry” the subsequent state (b):

(22) a.

b.

Komi (Juškov 1970: 192) Germanly ošjys’ömyd n’in. German:DAT boast:PST2:2SG already ‘You have already been boasting to German (it turns out, as I hear etc.).’ (Juškov 1970: 40) Vera vajöma. Talun asyvnas mamys völi Vera bring:PST2:3SG. Today morning:INESS mother:3SG AUX1 pöžalö, pyrali na ordö da. bake:PRS.3SG drop_in:PST1.1SG they to when ‘[The heroine has brought the hero a pie to eat, he refuses because he already has some pies] Vera has brought (them). This morning her mother was baking, when I dropped in to visit them.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

508

508

Marja Leinonen and Maria Vilkuna

4. What about Perfect? Of the past tense forms in the Permic languages, it is obviously PST2 that is a candidate for the status of perfect. Being based on a past participle, PST2 contains some of the ingredients of the perfect, as it includes an anterior event resulting in a subsequent state (resultative) or a state of affairs inferred (indirective). The first seven questions in PFQ have been used as an operational test for the prototypical perfect (Lindstedt, this volume; see also Dahl 1985). Some confusion arises from the fact that so many of these are formulated in the first person, which rules out PST2 in practice. PST2 is always possible in these contexts, but at the cost of an indirective interpretation, which is not natural without further motivation. The first answer to the questions was always PST1, which supports the idea of PST1 as a default past tense. In Komi and Udmurt, the dimension of evidentiality thus clearly overrides the “present relevance” meaning characteristic of the prototypical perfect. In Udmurt, there is a possibility of solving the ambiguity presented by the Komi PST2, that is, indirective vs. resultative meaning, by choosing either PST2 or the resultative construction. Consider once more example (6). In (6b), the absence of the mother is expressed with the resultative, but in (6e, f), the appearance of snow is rendered with PST2. The difference between the two in this context is very small, but the resultative, which is nominal in form, provides a time-stable situation, while PST2 still refers to an event, a verbal concept. This is why speakers tend to experience the resultative variant as more remote or irreversible than the other one. One of the ingredients of perfect not treated so far is the so-called experiential meaning. It is of some interest that Udmurt has a special construction for this purpose, which takes on some of the load of the prototypical perfect. This construction is not obligatory, though; questionnaire items like the following elicited PST1 answers from informants, but when the Experiential construction was offered as an alternative, it was accepted. (23) a.

b.

Udmurt (PFQ: 04) Ti mynes’tym apajme kuke no you(PL) I:ABL older_sister:ACC.1SG sometime pumital’l’ady-a? meet:PST1-Q Pumital’l’amdy van’-a mynes’tym apajme? meet:PART:2PL COP.PRS-Q I:ABL older_sister:ACC.1SG ‘Have you met my sister [at any time in your life up to now]?’

The Udmurt Experiential Perfect consists of the present or past tense copula and the past participle form of the main verb in a possessive construction; there is a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

509

Past tenses in Permic languages

509

possessive suffix on the participle, and an optional genitive possessor-subject. Below, (24a) is an instance of the Experiential construction and (24b), of the simple ‘have’ construction. This is one of the many cases where some type of perfectivity is a grammaticalized ‘have’ construction. Metaphorically, at least, the connection is clear: ‘You have a horse’, and ‘You have in your experience a past event of hearing about the revolutionaries.’ (24) a.

b.

Udmurt (corpus) Tynad kylemed van’-a revol’utsion’erjos s’arys’? you:GEN hear:PART:2SG COP.PRS-Q revolutionary:PL about ‘Have you heard about the revolutionaries?’ Tynad valed van’-a? you:GEN horse:2SG COP.PRS-Q ‘Do you have a horse?’

It should be made clear that, although this construction is quite similar to the Komi impersonal resultative, there is a difference that points to a different origin: in Komi, the existential copula is not used in Komi cases like (18). Moreover, Udmurt has other constructions consisting of the genitive, the past participle, and a “light” verb. For example, substituting the future copula luyny for the copula in (24a) yields the modal meaning ‘can, is possible’. As the reader may have noted, the particle we have translated as ‘already’ (Komi n’in, Udmurt (i)n’i) is fairly frequent in our examples. As ‘already’ is known to be a common source of the perfect (Bybee and Dahl 1989: 67–68), the question should be raised whether a similar grammaticalization process is going on or is to be expected in the Permic languages or whether the extensive use of the particle is due to Russian influence in a more superficial sense. However, its use is so wide and variable both in Komi and Udmurt that we must refrain from further discussion on this occasion.

5. Compound tenses Recall the two copular auxiliaries, völi/val and völöm/vylem. In Udmurt, and to a lesser degree in Komi, essentially all the tense and mood forms combine with the two copular auxiliaries, forming compound tenses with various interpretations, which are for the most part outside the scope of the present paper.7 However, we would like to conclude with a short survey of the remaining ways of referring to past events. A word on the syntactic structure of the compound tenses is in order here. An important feature of the Permic copulas is that they do not inflect for person or number like normal verbs, with the potential exception of AUX2 in simple predicative clauses – that is, outside the compound-tense system. An example of the copula in the 1st

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

510

510

Marja Leinonen and Maria Vilkuna

singular PST2 form was seen in (10) above. The Permic compound tenses are therefore structurally rather different from the Finnish and Germanic constructions, as can be seen by looking at (25a). It is the main verb that inflects – kalgis’ko is marked as both present and 1PL. The auxiliary remains the same when person/number changes. The following examples illustrate the difference between PST1 and the combination of present (PRS) and AUX in Udmurt. PST1 gives descriptions of situations and shows the events in sequence, and, as we saw, this can also be done with PST2. This contrasts with PRSAUX, which here indicates a state functioning as the background of the subsequent action. (25) a.

b.

Udmurt (PFQ: 08) [Do you know what happened to me just an hour ago?] Mon n’uleskyn kalgis’ko val. I forest:INESS walk:PRS.1SG AUX1 ‘I was walking in the forest.’ Šödtek šorys’ kyj vyle l’ogis’kyny šed’ti. unexpectedly snake upon step:INF happen:PST1.1SG ‘Suddenly I stepped on a snake.’

Varying the auxiliary adds the evidential dimension. As an extension of background states, PRS/FUTAUX has a habitual use. This is common in the dialect texts when the interviewees describe the habits of earlier times. (Another, less frequent compound tense used for generic past is future with AUX, which is at home also in habitual uses.) PRSAUX is also used for interrupted action: ‘to be about to do something’. The habitual and “interrupted” uses are disambiguated by the context, as in the following: (26) a.

b.

Udmurt Jegit dyrjam, avtobus dugde val korkamy young time:ILL.1SG bus stop:PRS.3SG AUX1 house:1PL dory. at ‘When I was young, the bus used to stop at our house.’ Avtobus dugde val korkamy dory no, bus stop:PRS.3SG AUX1 house:1PL at and koškiz. leave:PST1:3SG ‘The bus was about to stop at our house, but it went on.’

All these uses of PRSAUX are found in Komi (KRS 1961, see völi; Serebrennikov 1963: 270–273), but actual occurrences are apparently less frequent. Serebrennikov

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

511

Past tenses in Permic languages

511

(1960) associates the Komi and Udmurt PRSAUX pattern with the English past continuous, calling it ‘durative, descriptive past’. Furthermore, Permic languages have a compound tense called “pluperfect” by Western scholars. It consists of PST2 and an auxiliary indicating the evidentiality dimension. The following Komi example shows that even the 1st person is natural here. The interpretation is purely resultative, but embedded in the past. (27)

Komi Kor sijö loktis, seki me völi sad’möma. when s/he come:PST1 then I be:PST1 wake_up:PST2 ‘When she came, I was/had woken up.’

In Udmurt, there are two choices: either PST2 (without apparent indirective meaning) or the resultative form may be used with the auxiliary. (28)

a. b.

Udmurt (PFQ: 76) [A’s sister was not at home when A arrived. Question: Did you find your sister at home? A answers: No, I did not (find her).] vylem in’i. So koškem val AUX2 already s/he leave:PST2.3SG AUX1 vylem in’i. So koškemyn val AUX2 already s/he leave:PART:INESS AUX1 ‘She had left (was gone).’

As a further complication, there is the combination of PST1 and AUX, which comes close to PST2AUX in meaning, but is not identical to it. We will not discuss the meanings of the compound tenses any further, but one thing is clear: they are not narrative. The presence of the auxiliary always seems to signal a break in the narrative sequence, but how exactly this happens is a matter for future research.

6. Conclusions The central topic of this paper is the distinction between the two basic past tenses in the Permic languages. We have shown how two central semantic components of the perfect gram, viz. indirectivity and resultativity, can combine without giving rise to a perfect proper. This happens in Komi, whereas Udmurt makes a grammatical distinction between the two components. The clearest situation for the distinction is, naturally, when the speaker is the subject – the speaker knows best about his experience. It is for this reason that the 1st person of PST2 has been considered to be non-existent in Komi. However, as the Udmurt examples amply prove, indirective meaning is far from excluded from the 1st person, and the resultative interpretation of

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

512

512

Marja Leinonen and Maria Vilkuna

PST2 makes 1st person subjects even more natural. An essential difference between Komi and Udmurt is that the indirective undertones of PST2 cannot be escaped in the latter. Both languages have developed specific morphosyntactic means for the expression of resultativity. Beside the very productive resultative form, Udmurt also has an experiential construction and appears to resort to compound tenses more often than Komi.

Acknowledgements We are indebted to Jevgeni Cypanov, Valentina Ludykova and Sergei Gabov (Komi), Tatjana Krasnova, Aleksandr Škl’ajev and especially Bibinur Zaguljajeva (Udmurt) for sharing their native knowledge with us. The interpretations of the data remain our responsibility.

Notes 1. The Udmurt corpus was compiled by Pirkko Suihkonen and Bibinur Zaguljajeva. Approximately 9,000 orthographic sentences of prose (20th century fiction) are included, situated in the corpus server of the Department of General Linguistics, University of Helsinki. Examples from this corpus are marked “corpus”. 2. Examples with no indication of source come from speaker interviews. 3. Predictably, the choice of past tense is a source of some debate in the ongoing Bible translation work (Marja Kartano, p.c.). 4. Some minor adjustments of dialect texts to the current transliteration system of the standard language have been made. 5. Note the particle pe indicating quotation or hearsay. This word and its Komi counterpart pö combine with all tenses. 6. Note the “sudden revelation” use of PST2 here. 7. For example, the combination of imperative and AUX1 may be a polite request or a ?’. counterfactual advice of the type ‘why didn’t you

References Aksu-Koç, Ayhan A. & Dan I. Slobin 1986 “A psychological account of the development and use of evidentials in Turkish”, in: Wallace Chafe & Johanna Nichols (eds.) Evidentiality: The linguistic coding of epistemology. (Advances in Discourse Processes Vol. XX.) Norwood: Ablex Publishing Corporation, 159–167. Baker, Robin W. 1983 “Komi Zyryan’s second past tense”, Finnisch-Ugrische Forschungen Band XLV, Heft 1–3 : 69–81.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

513

Past tenses in Permic languages Bartens, Raija 1982

513

“Die Dialektmonographien für das Komi-Syrjänische”, Finnisch-ugrische Forschungen 44: 150–164.

Batalova, R. M. 1982 Areal’nye issledovanija po vostoˇcnym finno-ugorskim jazykam. [Areal investigations in Eastern Finno-Ugrian languages.] Moskva: Nauka. Bybee, Joan L. & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Cypanov, Jevgenij 1992 “Dopolnenija k paradigme II prošedšego vremeni v komi jazyke” [Additions to the paradigm of the second past tense in Komi], Linguistica Uralica XXVIII:l: 24–31. Dahl, Östen 1985

Tense and aspect systems. Oxford: Basil Blackwell.

Haarmann, Harald 1970 Die indirekte Erlebnisform als grammatische Kategorie. Eine eurasische Isoglosse. Veröffentlichungen der Societas Uralo-Altaica, Band 2. Wiesbaden: Otto Harrassowitz. Haspelmath, Martin 1994 “Passive participles across languages”, in: Barbara Fox & Paul J. Hopper (eds.) Voice: Form and function. (Typological Studies in Language 27) Amsterdam/Philadelphia: John Benjamins Publishing Company, 151–178. KRS 1961

= Komi–russkij slovar’. [Komi–Russian dictionary.] Ed. V.I. Lytkin. Moskva: Gosudarstvennoe izdatel’stvo inostrannyx i nacional’nyx slovarej.

Kuz’mina, I. B. 1993 Sintaksis russkix govorov v lingvogeografiˇceskom aspekte. [The syntax of Russian dialects from the point of view of linguistic geography.] Moskva: Nauka. Lindstedt, Jouko this volume “The Perfect – aspectual and temporal and evidential.” Nedjalkov, Vladimir P. & Sergej Je. Jaxontov 1988 “The typology of resultative constructions”, in: V.P. Nedjalkov (ed.), 3–62. Nedjalkov, Vladimir P. (ed.) 1988 Typology of resultative constructions. (Typological studies in language 12). Amsterdam/Philadelphia: John Benjamins Serebrennikov B.A. 1960 Kategorii vremeni i vida v finno-ugorskix jazykax permskoj i volžskoj grupp. [The categories of tense and aspect in the Permic and Volgaic groups of the Finno-Ugrian languages.] Moskva: Izdatel’stvo Akademii Nauk SSR. 1963

Istoriˇceskaja morfologija permskix jazykov. [Historical morphology of the Permic languages.] Moskva: Izdatel’stvo Akademii Nauk SSR.

Svjatoe evangelie ot Matfeja, Marka, Luki i Ioanna na udmurtskom jazyke 1912 [The Holy Gospel, according to Matthew, Mark, Luke and John in Udmurt.] Facsimile edition. Kazan’: Central’naja tipografija. Trubinskij V.I. 1988

“Resultative, passive, and perfect in Russian dialects”, in: V. P. Nedjalkov (ed.), 389– 410.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

514

514

Marja Leinonen and Maria Vilkuna

Sources of examples Juškov, Gennadij 1970 Kuim teš. [Three comedies.] Syktyvkar: Komi knižnoe izdatel’stvo. Kel’makov, Valentin and Sirkka Saarinen 1994 Udmurtin murteet [Udmurt dialects]. Publications of the Department of Finnish and General Linguistics of the University of Turku No. 47 and Udmurt State University, Department of General and Finno-Ugrian Linguistics No. 2. Turku-Iževsk. Vojvyv kodzuv [Northern Star.] No. 5, 1963 A literary journal. Syktyvkar. Žilina, T. I. & V. A. Sorvaˇceva 1971 Obrazcy komi-zyrjanskoj reˇci. [Samples of Komi–Zyrian speech.] Syktyvkar: Komi knižnoe izdatel’stvo.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

515

The Progressive

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

516

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

517

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

The progressive in Europe

1. Progressive: questionnaire and sample1 1.1. Method The progressive aspect is often identified with the imperfective aspect, rather than being treated as an aspect in its own right. When looking at the progressive aspect in the languages of Europe, one is inclined to take English as a starting point because English seems to be the preeminent language illustrating a fully grammaticalized progressive. On the basis of the use of the progressive in English, one could infer rules and relevant properties for describing “the" progressive in other languages. We decided to do otherwise. On the basis of a typological questionnaire (PROGQ), we established whether a language has a special form (which we call PROG) and whether such a form may be used in a number of specific cases. The informants for the different languages were asked to give the translation of the sentences after considering the contexts and extra information specified between slashes and square brackets. The first sentences were aimed at establishing whether a language has a dedicated PROG form or not. Consider the following sentences: (1)

(PROGQ: 1) [Somebody on the phone wants to know about Ann; the answer is: – Ann is next to me] She WORK [right now].

(2)

(PROGQ: 2) [A: What does Ann do every Saturday morning?] B: She CLEAN THE HOUSE/READ

The obvious difference between (1) and (2) is that in PROGQ: 1 the event is viewed as still in progress at the relevant moment (in this case, the speech moment), whereas in PROGQ: 2 no single moment is focussed on (the sentence indicates a habitual situation). In order to say that in a language there is a specific PROG form, it should be possible: (i) either to use a different form in these two sentences, or (ii) at least to have available in the first sentence an alternative form that is not available in the second.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

518

518

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

A straightforward translation of sentences (1–2) into Dutch gives the following picture: (3) a.

b.

Dutch (PROGQ: 1) Ze zit te werken. she sits to work ‘She is working.’ (PROGQ: 2) Dan maakt ze het huis schoon. then makes she the house clean ‘She is cleaning the house then.’

A construction as in (3b) can also be used to answer the question implicitly present in (1), whereas a construction as in (3a), based on a postural verb, would not be appropriate as an answer to the question that provides the context in (2). Consider: (4)

Dutch a. (PROGQ: 1) Ze maakt het huis schoon. she makes the house clean ‘She is cleaning the house.’ b. (PROGQ: 2) *Dan loopt ze het huis schoon te maken. then walks she the house clean to make ‘She is then cleaning the house.’

On the basis of these observations one could consider Dutch to have a specific PROG form. Languages may also present special constructions, named “absentive” (ABSV) in De Groot (this volume), which are used to suggest that a given agent is remote from the deictic centre, performing a certain action. There are subtle analogies, as well as differences, between PROG and ABSV (cf. Section 3 below). In order to avoid the use of an ABSV instead of a PROG in the translation of sentences like PROGQ: 1, the context was specified with the locational phrase next to me. This type of locational phrase allows PROG but is incompatible with ABSV. As is well-known, languages may have a progressive/non-progressive opposition in other tenses than the Present, sometimes to the exclusion of the latter (cf. Section 2.8 below). For that reason, sentences similar to (1–2) were also provided with past time reference:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

519

The progressive in Europe

519

(5)

(PROGQ: 3) [Last night at 8 o’ clock,] when John came, Ann still WORK.

(6)

(PROGQ: 4) Last year we [usually] CLEAN THE HOUSE on Saturdays [now we do it on Thursdays].

(7)

(PROGQ: 5) Last summer, John VISIT us three times.

Sentence PROGQ: 3 could contain PROG, whereas the other two sentences should not. PROGQ: 4 is a habitual sentence, while PROGQ: 5, by virtue of the numerical specification, is only compatible with a perfective viewpoint. As to PROGQ: 3, the following two assumptions were made: (i) If a language uses a PROG form in the present tense, it also allows the use of this form in the past tense. (ii) If a language does not use a PROG form in the present, the language may use it in the past tense. In order to test the usage of the forms elicited, in fact to obtain some indication as to their degree of grammaticalization, a number of sentences were concerned with the possible combination of particular grammatical categories, such as tense, mood, passive, causative etc. Section 2 briefly presents some of these topics. For a detailed report, cf. the chapters by Bertinetto, Ebert and Tommola in Part IV of this volume.

1.2. The sample The potential sample of languages was the list of 150 languages of Europe, as established in the E UROTYP Guidelines. Since the investigation of the progressive started rather late in our project, it was not possible to get information on as many languages as we hoped. Questionnaires were only returned for the following languages: Albanian, Basque, Catalan, Danish, Dutch, English, Estonian, Fering, Finnish, Frash, Frysk, German, Greek, Hungarian, Icelandic, Irish, Italian, Kalmyk, Karachai, Maltese, Portuguese, Romanian, Russian, Rusyn, Serbo-Croatian, Spanish, Turkish, Swedish, and Züritüütsch.

1.3. The data For some languages, there was more than one informant. The answers of the different informants for one language sometimes showed considerable variation. This is due to the fact that most of the languages do not have a grammaticalized PROG. The

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

520

520

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

variation consists in the use of the Simple (imperfective) tenses in contrast to specific PROG forms on the one hand, and in the use of different PROG candidates on the other. Obviously, it would always be preferable to have a rather large number of informants for each language. The PROGQ consists of two parts (cf. Appendix). Part I contains 83 sentences to be translated. Part II, which is meant for linguists with a good knowledge of the language investigated, contains several questions related to grammar.

2. The morphosyntactic expression of PROG 2.1. Introduction Progressivity may be specified in several ways. See table 1 for a synoptic view of the morphosyntactic devices used in four major language families of Europe. Note that in this chapter we shall only consider morphosyntactic tools that are specialized for progressivity, or that have it as their main reading. We shall therefore leave out the progressive uses of simple tenses, although these are in some cases (e.g., in the Baltic Finnic domain) the devices most frequently employed to convey the idea of progressivity. One further possibility could consist in a specification outside the verb by means of an adverbial (such as at this moment). However, we did not take this kind of device into consideration, for it is quite unspecific.2 We only focussed on expressions related to the verb or the verb phrase. Thus, whenever we use the abbreviation PROG, this must be intended as referring to a specialized morphosyntactic device, rather than to the broad semantic notion of progressivity. In his typological study of the progressive, Blansitt (1975) proposed the following morphosyntactic classification:3 a. Affixal progressive markers b. Complex verb phrases as progressive signals i. verb phrases with a copula as auxiliary ii. verb phrases with a motion or postural verb as auxiliary iii. verb phrases with a pro-predicate (do-type) as auxiliary iv. verb phrases with a special progressive auxiliary verb If we apply this classification to our material, the languages of Europe mainly exhibit examples of complex verb phrases, the types (i), (ii) and (iv). In this section we will shortly discuss the morphosyntactic situation. A more detailed discussion of the data is given in the following chapters, dealing with subsets of the languages of Europe (Germanic, Romance, Baltic Finnic), as well as in the chapter on Maltese by Ebert.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

521

The progressive in Europe

521

2.2. Affixal progressive Turkish is the only language of Europe which expresses PROG by means of an affix, the suffix -yor, as in: (8)

Turkish Çalı¸sı-yor-du work-YOR:PST ‘He was working.’

Some varieties of Arabic could be considered to have a prefixal progressive marker. Maltese, the only variety which falls inside the domain of the languages of Europe, uses an analytical form (cf. Ebert, this volume b).

2.3. Verb phrases with a copula as auxiliary One type of expression using a copula is the one where a copula combines with a Gerund. Examples of this type are Catalan, English, Italian, Brazilian Portuguese, Spanish, Kalmyk and Karachai. (9) a. b.

c.

English Peter is writing a letter. Spanish Estaba hablando con una chica. was:1SG talk:GER with a girl ‘I was talking with a girl.’ Italian Lei sta lavorando. she is work:GER ‘She is working.’

We consider It. stare in expressions such as (9c) to be a copula, although this verb can be used as an independent lexical verb with the meaning of ‘stand, be (in a certain location)’. In fact, in examples such as (9c) stare seems to have lost its lexical meaning. Nevertheless, it is fair to say that, except for some Central and Southern varieties of spoken Italian, this verb has not reached the full copula function typical of Spanish estar, which can also be used with adjectival predicates. A second type combines a copula with an Infinitive or a related form, differently denominated in the various grammatical traditions (such as Supine or the like). Languages using this type of expression are Estonian, Finnish, Karelian, Sami, Livonian, and Vepsian. As can be seen from the list of languages, this type is limited to Finnic

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

522

522

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

languages. Consider for instance Finnish, which uses the so-called “third Infinitive” form with the inessive case ending: (10)

Finnish Minä olen myy-mä-ssä lippuja. I am sell-3INF-INESS tickets.PRTV ‘I am selling tickets.’

It is interesting to note that constructions with a copula followed by the Infinitive in Dutch, German, and Hungarian denote ABSV rather than PROG (cf. De Groot, this volume). Finnish (but not the other Finnic languages) is a special case, for the infinitival construction shown in (10) serves both purposes (cf. Sections 2.6 and 3–4 below; for Faroese, cf. Ebert, this volume a, ex. 43). A third type combines a copula with a prepositional phrase containing a nonfinite form of the verb. Examples of this are found in Breton (Hewitt 1985/86), Danish, Dutch, all varieties of Frisian, German, Icelandic, Italian, Portuguese, and Züritüütsch. Most of the prepositional phrases contain an Infinitive. They come in different forms (cf. table 1 for some details): (i) preposition  Infinitive (e.g. Breton, Icelandic, Italian) (ii) preposition  article  Infinitive (e.g. Dutch). (11) a.

b.

c.

Icelandic Hún er að vinna. she is at/to work:INF ‘She is working.’ Italian Lei sta a lavorare. she is at work:INF ‘She is working.’ Dutch Peter is aan het zwemmen. Peter is at the swim:INF ‘Peter is swimming.’

Finally, the copula may also combine with lexical expressions of the type busy, be at work, be after, be under way etc. Examples are found, e.g., in Basque, Danish, Dutch, French, all varieties of Frisian, German, Swedish, and Züritüütsch, but they are sometimes reported as marginal types also in other languages (cf. table 1): (12)

Dutch Wim was bezig de stofzuiger te maken. Bill was busy the vacuum cleaner to repair ‘Bill was busy repairing the vacuum cleaner.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

523

The progressive in Europe

523

2.4. Verb phrases with a motion or postural verb as auxiliary Expressions with a motion verb (‘come’ and ‘go’) as an auxiliary may sometimes be associated with progressive meaning. Note, however, that in a number of cases this type of construction indicates other meanings, such as ingressivity or near future (e.g. Dutch), future (French, Portuguese, Spanish), or past (Catalan). And even when it constitutes a true PROG device, it is often not interchangeable in the same contexts with the copular constructions. As a matter of fact, these constructions take on the “durative” rather than the “focalized” meaning of the progressive (cf. Section 3 for further qualifications). Examples of this are to be found in the Romance languages: Catalan, Italian, Portuguese and Spanish (as well as French, where it is, however, quite rare). (13) a.

b.

Spanish El calor venía durando demasiado the heat came last:GER too much ‘The heat had been lasting too long.’ Italian (Sardinian variety) (PROGQ: 52) Va dimenticando. goes forget:GER ‘He is forgetting (names).’

Another type of motion verb (‘walk’, ‘go around’) is used in some Germanic languages (Danish, Dutch, and all varieties of Frisian). Consider: (14)

Dutch Marie loopt bloemen uit te venten Mary walks flowers out to hawk:INF ‘Mary is hawking flowers.’

This type of expression is sometimes indistinguishable from the postural verb construction (cf. section 5 and Ebert, this volume a, section 2.5). The majority of the Germanic languages allow postural verbs as an auxiliary. Danish, Icelandic, Norwegian, and Swedish coordinate the postural verb with the finite form of the main verb. Consider: (15)

Swedish Han sitter och läser tidningen. he sits and reads newspaper-the ‘He is reading a newspaper.’

Dutch and all varieties of Frisian use a postural verb  infinitival marker  Infinitive:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

524

524

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

(16)

Frysk Hy sit in krante te lêzen. he sits in newspaper to read:INF ‘He is reading the newspaper.’

2.5. Verb phrases with a special progressive auxiliary verb There are very few languages which use a dedicated verb as an auxiliary. Swedish and Yiddish are two examples. They express PROG with the help of a verb meaning ‘hold’: (17)

Swedish En ny kyrka håller på att byggas. a new church keeps on to build:PASS ‘A new church is being built.’

2.6. Other types We have found other types of expressions which can be associated with progressive meaning. The first type is based on the use of a particle. A clear example of a language using a particle is Albanian. One way of expressing PROG in Albanian consists of the use of the emphatic particle po.4 Compare: (18) a.

b.

Albanian Laj. wash:1SG ‘I wash.’ Po laj. PTCL wash:1SG ‘I am washing.’

We also consider Basque to use a particle, of the form ari. This particle, however, is originally the Infinitive form of a verb meaning ‘keep doing something’. The use of this form requires the auxiliary izan which agrees with the subject, as well as the Imperfective Participle. For that reason, the auxiliary together with the form ari could also be considered a compound verb (cf. King 1994: 383).5 (19)

Basque Zer idazten ari zara? what write:IPFV PTCL be:2SG ‘What are you writing?’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

525

The progressive in Europe

525

The other device we found is a syntactic device. In Finnish and Hungarian word order plays a role in the interpretation of PROG. In Finnish the use of the copula together with the “third Infinitive” with the inessive case ending usually expresses PROG. The same construction may also express the ABSV. When an object is present, the two readings can be disambiguated by means of word order. Compare: (20) a.

b.

Finnish Minä olen myy-mä-ssä lippuja. I am sell-3INF-INESS tickets:PRTV ‘I am selling tickets.’ (progressive) Minä olen lippuja myy-mä-ssä. I am tickets:PRTV sell:3INF:INESS ‘I am off selling tickets.’ (absentive)

Hungarian lacks any morphological marking for progressive. However, word order together with a specific intonation contour of the clause allows for a progressive interpretation (cf. Hetzron 1982, Kiefer 1994). Compare the following two examples, indicates focal stress and  neutral stress: where (21) a.

b.

Hungarian a bort, amikor csengetnek. Mari le-vitte Mary down-carried the wine when ring:3SG ‘Mary carried down the wine when the doorbell rang.’ le a bort, amikor csengetnek. Mari vitte Mary carried down the wine when ring:3SG ‘Mary was carrying down the wine when the doorbell rang.’

2.7. Progressive and other “aspectual” distinctions PROG does not seem to be compatible with the so-called Perfective aspect, as present in the Slavic languages or Hungarian.6 In the sentences of the PROGQ where some form of PROG could be used, the Slavic languages tend to use the Imperfective (cf. fn. 36 in Bertinetto & Delfitto, this volume, for some detail). Hungarian might be considered an exception. In fact, the construction illustrated in (21b), allowing for the progressive interpretation, is based on the Perfective verb form, but the sentence does not express perfective aspect. The point is that this example lacks an element with focal stress to the left of the verb, which is a requirement for the perfective aspect to apply in Hungarian (the word order in (21a), on the other hand, necessarily involves a focalized element). Note, however, that Hungarian could hardly be considered to be a language with a genuine PROG construction.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

526

526

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

The case of the Ibero-Romance languages is clearly different. The auxiliary estar has an imperfective and a perfective conjugation: both the imperfective and the perfective Past can be used with PROG (cf. Bertinetto, this volume; Olbertz 1998). Consider for instance: (22)

Spanish Estuve deambulando por las calles. was:SP:1SG stroll:GER about the streets ‘I was strolling about the streets (i.e., for some time).’

However, these are aspectual distinctions of a different sort, which have little to do with the distinctions to be observed in Hungarian or in Slavic languages (cf. again fn. 5). Viewed from this perspective, example (22) does not represent an exception to the generalization above. Note further that Perfect tenses (generally expressed in European languages by the compound tenses), which constitute a particular subdomain of the perfective domain, may be associated with PROG devices in quite a number of languages (cf. below Sections 3 and 4, and Bertinetto, this volume, Ebert, this volume a, Tommola, this volume). Summing up, the progressive aspect is in principle compatible with both perfective and imperfective tenses, although it occurs most often with the imperfective ones. As to the languages presenting the distinction Perfective/Imperfective, PROG clearly favours the Imperfective predicates.

2.8. Progressive and tense distinctions Our material supports the assumption made by Blansitt (1975: 30) that there are never more tense distinctions in progressive than in non-progressive. On a broad typological scale, several languages allow a progressive/non-progressive opposition in the past tenses, but not in the present. For instance, the Hungarian construction presented in (21b) can only be found in the past. Another example comes from Lithuanian. While the Simple Present is ambiguously progressive/non-progressive, among the past and future tenses the simple forms of the copula b¯uti combines with the Present Active Participle to form PROG (cf. Blansitt 1975: 20).

3. Typological observations 3.1. Individuating the main types The research conducted within E UROTYP made it possible to delineate, with a good amount of detail, a comprehensive picture of the typology of progressive constructions in European languages. However, since most of our data concern Western

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

527

The progressive in Europe

527

Europe, in this section we are going to restrict our observations to the Baltic Finnic, Germanic and Romance languages, where (as shown in Thieroff, this volume) the presence of specialized PROG devices is particularly well attested (but cf. Section 5 for a glance at Eastern Europe). Recall that, as stated in Section 2.1 above, the abbreviation PROG must be intended as referring to specialized morphosyntactic devices, rather than to the semantic notion of progressivity, for which the single languages may have at their disposal other grammatical tools. The different status of PROG in the various languages has obvious typological consequences. In languages like English, Irish, Icelandic or Maltese, it is quite likely that the respective PROG devices have reached a status of complete grammaticalization, considering that these are the only tools available to express the notion of progressivity. In other languages, however, the situation looks more complex. On the one hand, PROG constructions are not without competitors in that specific function, and not only in relation to simple imperfective tenses, but also with respect to the existence of more than one PROG device in one and the same language. On the other hand, these constructions may appear altogether to have a fairly low frequency in spontaneous linguistic usage. And here again the situation may vary. In Baltic Finnic, PROG seems to be used more in formal than in informal styles, whereas in the Romance and most of the Germanic area the reverse is true (at least as far as the most standard devices are considered).7 Table 1 is a synoptic presentation of the main morphological types to be found in the three groups considered here (cf. Bertinetto, this volume, Ebert, this volume a, and Tommola, this volume, for further qualifications concerning these data). Despite the multiplicity of meanings conveyed by these constructions in the various languages, three main types (or functions) may be isolated: (i.) “Focalized” progressive constructions (henceforth Foc-PROG), i.e. those expressing the notion of an event viewed as going on at a single point in time, here called “focalization point”. The focalization point may be overtly expressed in the sentence, or else it may be recovered through the context, being the object of a presupposition. Needless to say, the focalization point does not exhaustively localize the event; it simply indicates a point in time overlapping the progressive event, while the actual duration of the latter remains indeterminate. (ii.) “Durative” progressive constructions (henceforth Dur-PROG), i.e. those that are evaluated relative to a larger interval of time. Here again, however, the actual duration of the event remains indeterminate. Even when a durative temporal adverbial is present, this does not delimit the event but merely yields a vantage point from which the situation is observed. (iii.) “Absentive” constructions (ABSV), i.e., those conveying the meaning of an event occurring in a place (characteristically reserved for a given purpose) displaced from the deictic centre.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

528

528

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Table 1. Morphosyntactic tools for expressing progressivity in four major language families of Europe A. GERMANIC languages, with the exception of English (from Ebert, this volume, with modifications)

Icelandic: Swedish Danish: Norwegian: Frasch (Wiidingh.): FeringÖömrang: Frysk: Dutch: German: Züritüütsch:

Prepositional type

Postural type

Marginal types

er að  INF hålla på att  INF er ved at  INF

sit og  V sitter och  V sidder og  V

äs oont  INF

står og/ligg(er)og  V sät to/ sät än  INF

? ? er i gang/i fœrd med at  INF driv(er) og  V äs bai to/än  INF

as uun’t  INF

sat tu  INF

as bi tu  INF

is oan’t  INF is aan het  INF ist am/(beim)  INF isch am  INF

siet te  INF zit te  INF

is oan ‘e gong mei  INF is bezig te  INF ist dabei zu  INF isch draa ze  INF

B. ROMANCE languages, plus English (from Bertinetto, this volume, with modifications)

Catalan: French: Italian:

copular AUX type

motion AUX type

Marginal types

estar  GER

anar  GER aller  GER andare/venire  GER

? être en train de  INF essere dietro a  INF

vir  GER

?

stare  GER stare a  INF Portuguese: estar a  INF estar  GER Romanian: Spanish: estar  GER English: be  V-ing

ir/andar/venir  GER?

a fi în curs de  INF

C. BALTIC FINNIC languages Estonian Finnish

olema  V-ma s be  SUPINE  INESS olla  V-mA ssA be  3rd-INF  INESS

NB: The same construction also exists in Karelian, Livonian, Vepsian (Eastern Baltic Finnic languages) and in Sami.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

529

The progressive in Europe

529

D. ALTAIC languages sim. converb type Kalmyk Karachay Tatar

-jˇ bää-/yov-; CONV-SIM  ‘be’/‘move’ -a tur-a torCONV-SIM  ‘be, stand’

participial type -jˇa- -a ( bää-) PCPL:IPFV ( ‘be’)

NB: The same constructions as in Karachay and Tatar exist in most other Turkic languages, though functions vary somewhat from language to language.

The distinction between Foc-PROG and Dur-PROG is illustrated in particular in Bertinetto, this volume, with reference to Romance languages (cf. also Johanson, this volume, for the notion of focality), but it was already implicitly hinted at by Blansitt (1975).8 As to the notion ABSV, cf. De Groot, this volume. Suffice it to say that ABSV constructions exist in no fewer than eight European languages. Actually, in most cases ABSV devices differ morphologically from specifically PROG ones. However, in at least two languages (Finnish and Faroese) the two constructions coincide. It is therefore legitimate to examine this type in conjunction with PROG. In the final part of Section 4 we shall tackle the question of the proper relation between PROG and ABSV. To illustrate the issue, consider the following Finnish sentences, exemplifying the three main types, in the order given above (for Faroese, cf. Ebert, this volume a, ex. 43): (23)

Finnish a.

b.

Foc-PROG Kun lamppu putosi pöydälle, Hannu oli syömässä when lamp fell table:ALL Hannu was eat:INESS puuroa-nsa. porridge:PRTV-POSS:3SG ‘When the lamp fell on the table, Hannu was eating his porridge.’ Dur-PROG Viime tiistaina Nina oli pitämässä esitelmän Helsingissä. last tuesday Nina was keep:INESS talk Helsinki:INESS ‘Last Tuesday, Nina gave a talk in Helsinki.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

530

530

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

c.

ABSV Jouko oli lippuja myymässä. Jouko was tickets sell:INESS ‘Jouko was off, selling tickets.’

Finnish is a somewhat exceptional language, for (as noted above) the same construction can be used in all three meanings. It is interesting, however, that this case exists because, as we shall see, there are a number of analogies between ABSV and PROG. Although we would not wish to claim that they are, in a broad perspective, typologically related, it is probably justified to put them side by side at least for comparative purposes. Indeed, it is often the case that both PROG and ABSV may occur in the same context, as in the following examples: (24) a.

b.

Italian PROG Aldo non è qui. Si sta allenando. Aldo not is here. RFL is training ‘Aldo is not here. He is training.’ ABSV Aldo non è qui. E’ ad allenarsi. Aldo not is here. is at training:RFL ‘Aldo is not here. He is off training.’

although, strictly speaking, they ultimately answer different questions (“What is X doing?” vs. “Where is X?”). By contrast, the opposition Foc-PROG vs. Dur-PROG does not always correspond to a morphological difference. The same constructions may often express both meanings. However, this is not the case with all PROG devices. For instance, the standard PROG constructions to be found in Standard Italian and French seem by and large restricted to the focalized interpretation. Conversely, some of the constructions to be found in certain Germanic languages (namely, those based on postural verbs) are most naturally interpreted in the durative meaning; and this is also true, in fact, of the Italian PROG variant built with the Infinitive instead of the Gerund (cf. Table 1). There is some evidence suggesting that the first examples of PROG constructions, in languages for which we have old testimonies, could be used in contexts which exclude a purely focalized reading. Consider the following texts, quoted from Dietrich (1973) and Bybee et al. (1994) respectively: (25) a.

Latin (Vulgata, John 10.40) erat Johannis prius baptizans [ ] [ ] ubi where was John before baptizing ‘[ ] where John had previously been baptizing people [

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

]’

531

The progressive in Europe

b.

531

Old English (Ælfric, Lives, I, 11.52–55) Sume syndan creopende on eorða mid eallum lichoman, swa some are creeping on earth with whole body, as swa wurmas doð. Sume gað on twam fotum, sume on as worms do. some go on two feet, some on feower fotum, sume fleoð mid fyðerum. four feet, some fly with feathers ‘Some (animals) creep on the earth with their body, just as worms do. Some walk on two feet, some on four, some fly with their wings.’

Of special interest is example (25b), for the PROG construction used there makes up (in Carlsson’s 1978 terms) an “individual-level” predicate, i.e., a predicate exhibiting a “permanent” stative meaning, namely: ‘some animals have the permanent property of creeping on the ground’. This suggests that even though these old constructions could appear in contexts, such as (26), which are compatible with a focalized reading, their overall semantic interpretation must have been different from the one that is available to their modern equivalents (note, however, that the focalized interpretation of (26), although very likely, is not necessarily the meaning intended by the writer):

(26)

Latin (Vetus Latina, Coloss.,1.6) et est fructificans et crescens sicut in vobis ex qua and is fructifying and growing as in you since that die audistis et cognivistis gratiam dei in day hear:2PL:PST and learn:2PL:PST grace god:GEN in veritate.” truth ‘ and it is giving fruit and growing in you since the day when you heard it and became acquainted with God’s grace.’

Nevertheless, the existence of potentially focalized contexts indicates an easy line of development, which was taken by virtually all PROG devices, sometimes up to the extreme consequence of specializing as a purely focalized periphrasis, as in the case of the Italian Gerund PROG. By this, we do not want to suggest that the evolution undergone by the latter device shows the ultimate stage to be reached by these constructions. As is well known, there may be further stages (possibly reached at the end of an alternative developmental path), like the stage consisting of the complete reinterpretation of PROG as a general-purpose imperfective tool. See the following section for further comments on this.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

532

532

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

3.2. Semantic and morphosyntactic classification Let us now have a close look at the data, with the aim of finding the analogies and the differences that concern the three main types listed above as (i–iii). Table 2 is an attempt at listing the behaviour of these three types with respect to a number of relevant features. It should be borne in mind that the labels “Foc-PROG” and “Dur-PROG” stand for broad semantic categories, rather than for independent morphosyntactic constructions. Indeed, as we are going to show in Section 4, one and the same PROG construction may often correspond to more than one semantic category. As to the correspondence between these broad semantic categories and the concrete manifestations of PROG, further qualifications will be provided in Section 4. The rest of this section will be a commentary on Table 2. The first cell of Table 2 relates to the locative content of the three types considered. As is well known, PROG constructions include, in one way or another, a locative morpheme. This may consist for instance of an auxiliary verb indicating existence or position (as in virtually every European PROG device), of an explicit marker of locativity (like the inessive case in Finnish PROG), or of a combination of more than one such morphemes (as again in Finnish PROG, which combines both of the above features). However, although the morphological structure of these constructions is based on a locative morpheme of some kind, the degree to which this meaning component persists in each construction varies from case to case. It is obviously prominent in ABSV devices, with the addition of an implication of remoteness/invisibility of the agent. This is indeed the distinctive feature of this construction. As to FocPROG, it should be observed that no residue of the original locative meaning persists in this type, as is proven by the possibility of using motion verbs, like in: Fred is going home. Note that motion verbs are admitted with the ABSV only when they indicate the specific activity that the agent is performing in the particular location in which s/he finds her/himself. For instance, German Hans ist laufen ‘Hans is (away and is) engaged in the activity of running’ could be said of somebody who regularly visits the running track and is currently away for that purpose. By contrast, verbs of directed motion, designating the locomotion towards the remote site, cannot appear with the ABSV. This restriction is absent with Foc-PROG, and this distinguishes it not only from the ABSV but, above all, from Dur-PROG, which is hardly compatible with all sorts of motion verbs. This is notably the case with Germanic PROG constructions based on postural auxiliaries, which preserve to some extent their etymological meaning (Ebert this volume a, Section 3.1.). But even with Germanic constructions of the prepositional type (cf. Table 1), the availability of motion verbs is, generally speaking, quite weak (Ebert this volume a, Section 3.1.). As to Finnish PROG, which has only recently been grammaticalized in the focalized meaning, it rejects directed movement verbs altogether, and employs instead nominal constructions (Tommola, this volume).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

533

533

The progressive in Europe

Table 2. Semantic properties (cells A–F) and morphological compatibilities (cell G) of the three types of periphrases considered in section 3: focalized progressive, durative progressive, absentive. The parentheses indicate that the given diacritic suggests a mere tendency, rather that a strong characterization.

A. persistence of locative meaning compatibility with directed motion verbs remoteness / invisibility B. focalization point reading compatibility with non durative verbs C. temporariness availability of determinate duration adverbials pragmatically defined duration D. compatibility with perfective tenses possible inclusive value of ‘perfect’ tenses E. compatibility with the habitual reading availability of interpretative meaning F. agentivity requirement compatibility with negation compatibility with stative verbs G. compatibility with the Imperative compatibility with the passive possible occurrence with the Infinitive deontic modals’ government

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

FOC-PROG

DUR-PROG

ABSENTIVE



() ( )

 

  0/

  

 ( )

  



( )





( )





( ) () ( )

( ) ( ) ( )



() ( )





?

534

534

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Cell B of Table 2 refers to the features that specifically make up the focalized meaning (cf. point (i) of section 3.1 for an illustration of this notion). As shown in (23a) above, this reading presupposes that the focalization point be “familiar” to the speaker and the addressee. This may be obtained by means of a punctual temporal adverb (such as at 2 o’ clock), or by a temporal clause of the appropriate kind which instantiates the so-called “incidential schema”9, as in the example cited. This is a crucial property, for it shows that Foc-PROG has a purely imperfective aspectual value. Indeed, Foc-PROG is only compatible with imperfective tenses, such as the Imperfect in languages which exhibit an overt aspectual opposition among the past tenses. Dur-PROG and ABSV, by contrast, may easily be employed with perfective tenses, such as the Simple or Compound Past in Romance languages (cf. below for further comments on this). The impossibility of using perfective tenses with the Italian, French or Albanian PROG (the latter, more specifically, in the variant based on the particle po) is a decisive argument for ascribing these tools to the focalized type. It should be noted, in this connection, that the type of syntactic frame to which the incidential schema belongs enables a focalized reading even in languages where PROG is but weakly grammaticalized, like the Baltic Finnic languages (Metslang 1995, Heinämäki 1995, Tommola, this volume) or those, among the Germanic languages, that are less advanced in the grammaticalization process. Another typical textual situation which forces the focalized reading is “reportive” discourse, i.e., the kind of description that radio and television commentators make about live events. In such cases, the focalization point coincides (either explicitly or implicitly) with the speech time. Unsurprisingly, only Foc-PROG is compatible with non-durative verbs, which are instead ruled out in durative contexts such as those implied by Dur-PROG. In particular, achievement verbs (which constitute the bulk of non-durative predicates) may develop an “imminential” reading with Foc-PROG, suggesting that the event will soon occur although is not yet occurring at the focalization point. Consider wellknown examples such as: The grandfather was dying, or The train is leaving. A special class of achievements is constituted by “phasal” verbs such as begin, finish and the like, which again allow only for Foc-PROG (cf. PROGQ: 23–26). Interestingly, with achievement verbs Italian often tends to use an explicit PROG construction, although the Simple (imperfective) tenses may usually express the progressive reading by themselves (Delfitto & Bertinetto 1995). Apparently, in these cases the speaker needs to emphasize that the intended interpretation of the sentence is indeed progressive, despite the non-durative character of the verb. Conversely, not all Germanic languages may freely employ PROG in such contexts, and this is notably the case with the variant based on postural verbs (cf. Table 1). As to the ABSV, nondurative verbs may be employed, although they do not constitute the most common choice. An ABSV situation typically includes “adjacent” actions, besides the one specifically referred to. The structure of the event is in fact as follows: “go to remote

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

535

The progressive in Europe

535

location–perform the action–return to source location”. Thus, even a momentary verb like cut the ribbon may be used (cf. Italian Il sindaco è a tagliare il nastro ‘The Mayor is off cutting the ribbon’). Cell C concerns the duration of the event. Several authors have claimed temporariness to be a relevant feature of the progressive. Indeed, it is even the case that in English some stative verbs (mainly of the postural/locational type) may be used with PROG precisely when the notion of temporariness is involved. Consider cases such as: (27)

(PROGQ: 59) The statue is standing in the garden. [i.e., for a limited period]

Note, however, that this restriction does not concern Foc-PROG, which can also be used in cases such as PROGQ: 60: The earth is turning around the sun, despite the permanent nature of the condition referred to. As to ABSV, temporariness is, of course, one of its defining features, as shown by De Groot, this volume. As a matter of fact, the ABSV implies the idea of a conventionally defined duration, tied to the expected time-window involved in performing the event. As to compatibility with the adverbials indicating “determinate duration”, this is excluded with Foc-PROG for obvious reasons, but admitted by Dur-PROG, although the situation varies from language to language, and from adverbial to adverbial. It is a fact, however, that Dur-PROG may admit adverbials such as “from tx to ty”, “for X Time”, “until tx”, “for the whole duration of ”, and the like. For some Germanic languages this is actually the preferred context for the use of the constructions based on postural verbs (Ebert, this volume, Section 3.2.; and this is equally true for the Romance constructions based on motion verbs when perfective tenses are employed (Bertinetto, this volume). As to ABSV, however, the presence of “determinate duration” adverbials seems to be generally (but not necessarily) avoided. Apparently, since the duration of the event is pragmatically defined and easily recoverable, these periphrases tend to leave in the background the explicit indication of the time boundaries. Cell D is, from the point of view of the aspectual characterization, somehow symmetric to cell B. Just as a strictly imperfective interpretation is required for FocPROG, Dur-PROG and ABSV are easily accessible to perfective tenses.10 Consider a sentence such as: (28)

Spanish estuvo leyendo todo el día. was:SP:3SG read:GER all the day ‘S/he spent the whole day reading.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

536

536

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

A special case, among perfective tenses, is represented by those that express the aspectual value of “perfect”, which in European languages often assume the form of compound tenses. When this aspectual value is instantiated, Dur-PROG may convey (with activity verbs) an “inclusive” meaning, suggesting that the event has been going on up to the reference time, and may possibly continue beyond that point (as in: I have been reading the whole day, /i.e., until the present moment/).11 Significantly, although the Standard Italian Foc-PROG, based on the auxiliary verb stare, cannot be used with compound tenses, the corresponding Dur-PROG constructions (those based on a motion verb followed by Gerund, or on stare a followed by the Infinitive) can. Cell E deals once more with the possible aspectual interpretation of the three types of constructions under discussion. Foc-PROG is generally unavailable to the habitual reading; it may tolerate it only in correlative structures of the type: whenever PERFECTIVE NON-DURATIVE, then PROGRESSIVE. This is no wonder, for these syntactic frames, besides making explicit the habitual meaning of the sentence, enable us to isolate a series of punctual temporal locations, each of which can work as a focalization point for the progressive event. By contrast, Dur-PROG and ABSV are often available to a habitual interpretation, even without the facilitation of an appropriate syntactic structure. The licensing contexts seem to be the same in both cases. Consider, as an illustration, examples (28–29) in Bertinetto, this volume, from which it also emerges that not all languages have equally easy access to habitual contexts. Somehow related to habituality, although distinct from it, is the so-called “interpretative” use of PROG (König & Lutzeier 1973; König 1995; cf. also Bertinetto this volume, Section 6.3.2). Although this use is quite marginal in most languages, it is fairly frequent, for instance, in English. Here is an example: (29)

If we selected the best described languages, we would also be selecting the languages with the largest number of speakers.

The reason for the analogy with habituality is that this use may be fostered by correlative frames such as: by doing X, you are (implicitly) doing Y, which are somewhat reminiscent of the correlative habitual frames cited above. However, since in interpretative contexts the first clause does not contain a non-durative perfective event, it cannot provide any sort of focalization point. Hence, the interpretative use is only possible with a durative reading. Cell F focuses on agentivity. In the languages where PROG may readily be used with a focalized meaning, this requirement plays no role. Consider: (30)

When I came, the sun was shining gloriously.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

537

The progressive in Europe

537

Although this particular sentence may require an appropriate contextualization, it is perfectly acceptable in the relevant situations. By contrast, both ABSV and DurPROG are likely to be sensitive to agentivity, at least to some extent, although (as far as the latter is concerned) the strength of this restriction varies from case to case. It is specially strong in Baltic Finnic languages (Heinämäki 1995), whereas Germanic languages show a differentiated behaviour.12 Obviously, negation is inversely correlated with the notion of agentivity. Thus, it is no wonder that in Baltic Finnic languages and in most Germanic ones, i.e. in the languages where PROG fairly often exhibits a durative interpretation, there are restrictions on the usage of negation in progressive contexts. As to ABSV, the restriction concerning negation is fairly strong for pragmatic reasons but is not absolute. For instance, negation may present a contrastive value (cf. German Rolf ist nicht laufen, er ist einkaufen ‘Rolf is away, his purpose being running rather than shopping’). As to stative verbs, they should be ruled out altogether. Indeed, this is often regarded as one of the most important defining features of progressivity. However, it turns out that stative verbs may sometimes combine with PROG, although the situation varies from language to language (Bertinetto 1994). But note that things differ for Foc-PROG and Dur-PROG. With the former, the stative verbs that appear in progressive sentences normally take on a non-stative meaning, with the marginal exception of some postural and locational English verbs (such as the verb stand discussed in (27) above), which preserve their stative character. Consider, for example, the copular predicates of PROGQ: 42–43, where the use of PROG (available to most Ibero-Romance languages as well as English) forces the activity interpretation in these intrinsically ambiguous predicates. To illustrate, John is being clever hints at John’s temporary behaviour, rather than to a permanent characterization of his. Consequently, these are not true counterexamples to the regularity stated above. By contrast, Dur-PROG does not necessarily exclude stative verbs. This is shown in particular by languages where the relevant construction is at a very early stage of grammaticalization, such as Estonian (Metslang 1995; Tommola, this volume). And this, as observed above in relation to example (25), seems to be a prominent fact in the diachronic evolution of PROG. Finally, cell G reports data concerning the compatibility of PROG and ABSV constructions with some relevant morphological categories. To start with, consider their compatibility with the Imperative, which is often quite weak. The languages which allow this use tend to be those that present a fully grammaticalized PROG device, like English and Catalan, although this feature may surprisingly appear even in Romanian, a language virtually without PROG (Bertinetto, this volume). Our data suggest that the Imperative is most probable in durative contexts, although it may also appear in focalized ones, such as PROGQ: 73 (Be working when the boss returns!). On the other hand, the Imperative is definitely ruled out with the ABSV. Compatibility with the passive is also a clue to an advanced stage of grammaticalization. Apparently,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

538

538

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

this possibility is again excluded with the ABSV.13 Essentially the same applies to the use of the Infinitive with PROG. To be exact, we refer here to contexts where PROG takes the Infinitive as a consequence of being syntactically governed by a predicate governing this mood.14 Note that compatibility with the Infinitive seems to arise before compatibility with the Imperative and the passive. For example, the Infinitive is at least marginally possible in Italian, where the passive and the Imperative are not at all available. Once more it should be noted that this morphological feature is mostly to be expected in durative contexts. One peculiar case of Infinitive PROG is constituted by sentences where this construction is governed by modal verbs, as in PROGQ: 79–80. The interesting fact here is that only Dur-PROG admits the deontic interpretation, although these cases seem to exist only in informal registers. Normally, the only reading available with Foc-PROG is the epistemic one, as in: Anne must be feeding the animals, I guess (Bertinetto, this volume, Section 3.1).

4. A diachronic sketch 4.1. Stages of development Let us now consider the diachronic issue. We shall first examine the problem of Foc-PROG and Dur-PROG, delaying the question of the ABSV (namely of its relationship to PROG constructions) until the end of this section. As to Foc-PROG and Dur-PROG, one could in principle defend the proposal that both types of periphrases can constitute the original nucleus, depending on the language. However, there is some evidence that Dur-PROG represents the original stage. Consider first the case of Italian. We know for sure that this language underwent a dramatic diachronic development. The PROG construction based on the auxiliary stare is now used exclusively (or almost exclusively) in focalized contexts, but in the early stages it could also be used in durative contexts (Bertinetto, this volume, Section 2). This invites the conclusion that Dur-PROG instantiates the original meaning of this construction, and this seems to be true in other cases. The PROG constructions of Ibero-Romance languages and English may cover both the focalized and the durative meaning, but we have good evidence that the forerunners of these periphrases, in Latin and Old English, were readily available in contexts which presented a purely durative, or even stative character (as shown by examples (25) above). It is thus reasonable to suppose that the use in focalized contexts was the result of an expansion of the original durative interpretation. As to Germanic languages other than English, one can say with good approximation that the PROG constructions based on postural verbs seem to be especially appropriate for durative contexts; and even the prepositional constructions, which are more flexible in use, often undergo restrictions in focalized contexts (Ebert, this volume a). Putting this together,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

539

The progressive in Europe

539

it appears that in all these cases the original meaning of PROG constructions seems to have been the durative one. This is not true in every case, though. Among the European languages for which we have reliable data, the only ones, besides Italian, that present a PROG construction exclusively restricted to the focalized meaning are French and Albanian. Now, the French periphrasis “être en train de  INF” started out as an intentional construction, and only towards the beginning of the 19th century was it eventually specialized in its current meaning (Gougenheim 1929). There is no evidence that this periphrasis, in its progressive interpretation, ever played the role of a durative device. The same applies to the Albanian construction based on the particle po, whose meaning is close to German gerade.15 However, it seems rather implausible that the French and Albanian constructions, due to their inherent constitution, will ever expand their use to typically durative contexts. Therefore, the exceptions provided by these two languages only prove that some PROG constructions can directly appear as focalized devices. Until we find clear evidence of a Dur-PROG construction that started out as a purely Foc-PROG device, we are justified in assuming that the diachronic development, when there is one, goes in the direction indicated above. This invites the hypothesis that, putting aside the exceptions represented by French and Albanian, the original meaning of most PROG devices must have been that of a stative construction, expressing the idea of “being (i.e., finding oneself/itself) in a state”, as is especially clear with the forerunners of Romance and English PROG constructions, as noted above in relation to examples (25). The purely dynamic (or processual) meaning, which is particularly salient in the focalized type, is in most cases a later development, attained at the end of a rather lengthy grammaticalization process. In other words, it appears that most PROG constructions started out as “actional” periphrases, rather than truly “aspectual” ones. The complete attainment of the latter status corresponds to the stage of full grammaticalization. We may thus hypothesize that, as far as their meaning is concerned, the most typical PROG constructions possibly underwent the stages of development shown in Table 3. Since the historical data are not equally clear for all the languages belonging to the three groups considered in the preceding section (Romance, Germanic and Baltic Finnic), the observations that follow will deal mostly with Romance, with just a few hints to the other groups. The relevant data derive from Bertinetto, this volume. Stage (i) is linked to the inherent meaning of the morphemes employed. At this stage, the non-finite form of the verb accompanying the locative or postural or motion verb presents a purely “con-verbal” meaning.16 Stage (ii) corresponds to the initial stage of grammaticalization, in which the locative (or postural or motion) verb begins to turn into an auxiliary, while the non-finite form (or the verbal noun) is gradually promoted to the status of head of the construction. At this stage, the semantic bleaching of the auxiliary may begin, although this process is completed only at stage (iii).17 Stage (iv) shows a further development: the context must provide the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

540

540

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Table 3. Diachronic development of progressive constructions in Romance (from actionality to aspect) (i)

pure locativity

(ii)

progressivity I

(iii)

progressivity II

(iv)

progressivity III

(v)

pure imperfectivity

stative, durative (ex.: the meaning to be observed in some Latin examples) residually locative, durative, accessible to perfective aspect (ex.: PROG periphrases based on the verb ‘come’, which preserve some kind of deictic orientation) durative, accessible to perfective aspect (ex.: PROG periphrases based on the verb ‘go’) focalized, strictly imperfective (ex.: Modern Italian “stare  Gerund”) loss of the progressive character (ex.: possibly to be observed on some non-standard varieties of Latin American Spanish)

explicit (or presupposed) indication of a single focalization point, which gives PROG its prototypical character. At this stage, the availability of perfective tenses is totally lost: the transformation from an actional to an aspectual device is completed. Interestingly, French PROG (whose origin is quite different from that of most PROG constructions) entered directly at this level, bypassing all previous stages. Finally, stage (v) is a possible future development, not yet attained by any Romance language. This step, which has been taken by many European and non-European languages, consists in the eventual reduction of PROG constructions to a purely imperfective form; i.e., a form not restricted to progressive contexts, but appearing also in habitual ones, thus behaving like a typically general-purpose imperfective tense such as the Romance Imperfect. To the European examples quoted by Johanson, this volume (section 7.8), one may at least add Igbo and Yoruba (Comrie 1976: 99–101), as well as Punjabi and Urdu (Dahl 1985). Obviously, at this final stage the restriction concerning stative verbs (cf. cell G of Table 2) is dropped altogether. The diachronic reconstruction sketched in Table 3 is not intended to suggest that each PROG device to be found in a given language should correspond, at any given moment, to a single and definite stage. This may be true in particular cases, some of which have been mentioned above (recall e.g. French), but it is easy to point out PROG constructions whose usage covers more than one stage of Table 3. Consider for instance English PROG, or the PROG periphrases based on copular verbs in Ibero-Romance languages, or the prepositional type available in a number of Germanic languages (cf. Table 1). Although the degree of grammaticalization varies (being highest in English and Ibero-Romance languages), all these constructions embrace stages (iii) and (iv) at the same time, i.e. they have steadily reached the focal-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

541

The progressive in Europe

541

ized stage, without losing ground as durative devices. This may be seen, for instance, in the following English sentence, which may be read in the durative meaning: (31)

[A scout, pointing to a series of paw prints, says:] A grizzly was walking here!

The reading that is relevant to our present purpose is the one whereby the scout intends to convey the idea that a grizzly trampled the ground in that place for a certain period of time, as indicated by the traces on the ground. According to this interpretation, there is no focalization point whatsoever. The same result may be obtained in Ibero-Romance languages by combining PROG with a perfective Past (e.g., Spanish “estuvo  Gerund”). By contrast, Italian gerundial PROG would be absolutely out of place in a similar context.18 In fact, the span of meaning of each PROG device may be even larger than two stages (cf. again Table 3). This is, for instance, the case with Estonian PROG, which is but weakly grammaticalized (Metslang 1995; Tommola, this volume). On the one hand, this construction may at times suggest the persistence of a locative meaning (stage ii); on the other hand, it seems to appear more and more often in focalized contexts (stage iv), possibly under the external influence of other languages such as Finnish (and perhaps English).

4.2. Progressive vs. absentive The inspection of Table 2 may suggest to us some plausible conclusion concerning the position of ABSV in comparison with PROG. In general, there seems to be little doubt that the ABSV resembles PROG on semantic grounds, as is also proven by the interchangeability of these devices in certain contexts (cf. again example (24)). However, this does not necessarily involve the evolutionary domain. The situation is in fact fairly complex. As may be seen, cell A of Table 2 indicates some kind of solidarity between Dur-PROG and ABSV in contradistinction to Foc-PROG, but cell G presents more of a solidarity between Foc-PROG and ABSV, and the remaining cells exhibit a rather variable situation, where the behaviour of the ABSV approaches one type or the other of PROG, depending on the feature considered. Actually, considering the general characterization of the ABSV as provided in De Groot, this volume, we should expect the relationship to be tighter between Dur-PROG and ABSV than between Foc-PROG and ABSV. However, the fact that Finnish and Faroese are, among the languages known to us, the only ones where PROG and ABSV can be expressed by the same morphosyntactic construction invites the conclusion that an evolutionary link between these grammatical devices may indeed exist, but is unlikely to be the rule.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

542

542

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Let us consider more closely the case of Finnish. As a PROG device, this periphrasis seems to be at a more advanced stage of grammaticalization than its Estonian equivalent, yet it basically spans the same variety of uses. The interesting question is whether this construction started out as ABSV or as PROG. Only a thorough historical investigation could settle the matter. What is clear, though, is that this construction presents some residue of locativity, for it rejects motion verbs such as tulla ‘come’ or mennä ‘go’, which have to be replaced by the corresponding nouns (tullo, menno). Now, we observed in the discussion of cell A of Table 2 that locativity is a prominent feature of ABSV constructions. Another prominent feature is agentivity. It is then interesting to note that Finnish PROG, even when used as a true progressive construction, obeys an agentivity constraint, or at least a constraint of intentional control of the event (Heinämäki 1995).19 For example, a sentence such as Kynttilät on palamassa ‘(the) candles are burning’ may only be uttered when one wants to suggest that somebody has lit the candles, with implicit recovering of the hidden agent. Thus, the connections between the two main functions of this periphrasis (PROG and ABSV) appear to be rather evident. But this does not solve the diachronic puzzle. The only thing that may plausibly be said is that the Finnish periphrasis must have entered the developmental path sketched in Table 3 no later than stage (ii). In fact, there is even ground to hypothesize that it actually entered at stage (i), for it may be employed with the prototypical stative verb olla ‘be’ to convey the meaning of existence, as in: On olemassa toinen mahdollisuus ‘there exists another possibility’ (lit.: is be-INF-INESS other possibility). Interestingly, the morphologically identical Estonian periphrasis may be used with an even larger group of statives (Metslang 1995; Tommola, this volume). In any case, one should not forget that the situation of Finnish and Faroese is rather exceptional. In all other languages in which an ABSV construction has been described there is no morphological coincidence of ABSV and PROG. There is even ground to believe that in some languages, like German and Italian, the ABSV has arisen out of the mere deletion of the Participle, as in: Er ist einkaufen gegangen Er ist einkaufen, È andato a comprare È a comprare ‘he is shopping’ (lit.: he is gone (to) buy).20 Thus, in most cases the semantic proximity of ABSV and PROG, as it emerges in Table 2, although not accidental, may be the effect of the overall similarity of the contexts in which these grammatical devices appear, rather than the result of a true evolutionary convergence.

5. A glance towards Eastern Europe: Altaic languages The investigation of progressives, which was started rather late in the Tense-Aspect Group, concentrated on the better known European languages. As progressives have not been much investigated for European languages except English, the inclusion

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

543

The progressive in Europe

543

of lesser known languages depended on the availability of some expertise. Besides for the languages treated in the individual articles, we received questionnaires for Basque, Armenian, Kalmyk and Karachai. We have nothing new to say about the first two, but we do want to present some data from Kalmyk and Karachai21 for the following reasons: Kalmyk uses four different forms (besides a general imperfective and a habitual), where English has one and other Germanic languages have two forms. Karachai is one of the languages for which one form (-ib tur-) has been claimed to have both progressive and perfect meaning, which gave rise to a corresponding question in the theoretical part of the PROGQ (II, 1j).

5.1. Kalmyk As little is known about the function of Kalmyk TA-forms and as the terms used in the available descriptions are of little help, the questionnaire data were difficult to evaluate. The most useful source was Bläsing (1984); without his numerous text examples the interpretation of the questionnaire data would not have been possible. The forms to be briefly discussed here are (for a more detailed analysis see Ebert, to appear): our term IPFV PROG I PROG II Continuative Durative

traditional name (Benzing 1985) Präsens I -na -jˇa- (contraction of PROG II?) Präs. durativum I22 Präs. durativum I -jˇ bää-/ jov- (simultaneous converb  ‘be’/‘move’) Präsens III -a (bää-) (ipfv. participle  ‘be’) Präs. durativum II -ad bää-/ jov-/ suu-/ (anterior converb  ‘be’/‘move’/‘sit’ )

The form used most often in sentences where we expected a progressive form is PROG I (with the Imperfective suffix in present time contexts). It occurs, for instance, in PROGQ: 1, 6–17, 19–27 (cf. Table 4): – with agentive verbs: ter ködl-jˇana ‘she is working’ (PROGQ: 1) – with non-agentive verbs: usn busl-jˇana ‘the water is boiling’ (PROGQ: 37) – with motion verbs: har-ˇcana ‘is going out’ (PROGQ: 21) – with phasal verbs: küce-jˇänä ‘is finishing’ (PROGQ: 26) – with temporary states: üüdn xoornd zogs-jˇana ‘is standing by the door’ (PROGQ: 58) It was not used with stative verbs or with the stative component of initio-transformative verbs (med-ne ‘knows’ – PROGQ: 39), and not with limiting temporal adverbials (PROGQ: 48). -jˇa combines with finite and nonfinite TAM suffixes, e.g.,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

544

544

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

(32) a.

b.

Kalmyk (PROGQ: 70) Martin sad-t naad-jˇa-la Martin garden-LOC play-PROG-ANT ‘ Martin was playing in the courtyard’ (PROGQ: 30) Pjotr cˇ olu terz-d haj-jˇa-sn cag-t P. Peter window-LOC throw-PROG-PART time-LOC ‘ when Peter was throwing the stone at the window’

The form -jˇ bää- is seldom used and does not figure in the PROGQ. However, if motion is involved, a PROG II periphrasis with the verb jov- ‘go’ is preferred: orad kür-ˇc jovna ‘is reaching the top’ (PROGQ: 31). (The verb for ‘go around, be in motion’ is commonly used as a postural verb in periphrastic constructions; cf. also Germanic (Ebert, this volume a)). Combinations with other verbs (umš-jˇ suuna ‘sits reading’, unt-jˇ kevtna ‘lies sleeping’) are extremely rare. There is not a single occurrence in the questionnaire or in the text examples of the various sources. Both -jˇana and -jˇ jovna can be found in generic or habitual contexts (Bläsing 1984: 19). The imperfective participle in -a (-ha after vowels) followed by a form of bää- in non-present contexts, was used in three sentences of the PROGQ:23 narn gerlt-ä ‘the sun is shining’ (PROGQ: 36), ködl-ä bilä ‘was still working’ (PROGQ: 3), and bi xotan ke-hä bääxv ‘I will still be cooking’ (PROGQ: 83). Two of the questionnaire sentences contained the adverb “still”, which signals that the situation has been going on for some time. This is confirmed by other examples in the literature; cf. ter oda cˇ ign surhul’ sur-a ‘er studiert noch’ (Benzing 1985: 128), xuur saak kevtän or-a ‘rain is/ keeps falling’ (Todaeva 1968: 46). The use of the a-form resembles the ‘continuative’ in Lezgian (Haspelmath 1993: 145). We tentatively adopt the term here, although an explicit notion of continuativity is not always implied; PROGQ: 83 is not equivalent to ‘ I will go on cooking’. The Continuative is compatible with focalizing contexts such as PROGQ: 3 and 83, but also with habituality.24 We therefore do not consider it to be a Foc-PROG. The fourth form to be considered, -ad bää-, specifies actionality. We have little explicit information on the actionality of Kalmyk verbs, but if the Kalmyk-Russian dictionary gives a Perfective-Imperfective pair for a Kalmyk lexeme, we can be quite sure that we are dealing with a two-phase verb;25 e.g., suux ‘sit; sit down’, ääx ‘be afraid; become afraid’, šatx ‘burn, sparkle; start to burn’. A form like šatad bääspecifies the non-transformative meaning component ‘burn, be sparkling’. But -ad bää- serves not only to build new lexemes, as Bläsing claims. It can indicate various types of durative situations, like continuativity, iterativity, graduality,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

545

The progressive in Europe

545

Table 4. Forms used in the PROGQquestionnaire. The numbers on the left refer to the questionnaire sentences, as specified in the Appendix. Kalmyk -jˇaPROG

Karachay -na -a-ad bää- -a turIPFV CONT DUR PROG

a) focalized  1  6–17 44,45,47   33–35  37–38 MOT 46 MOT 31 b) habitual  2 4 c) duration /(iter.) 3, 83 36 18 62 d) temporal limits 48 49 50 60  70b e) inclusive reading 81–82 f) imperative, negation 73 * 76 77 g) stative verbs  39  40 42 h) postural verbs  29,59  58  28 i) telic verbs  54  55 57 [Æ ]

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

      ()

-a-ib tur-ibIPFV DUR/RES RES ()

   

V just now V just now V just now insult, admit, dream boil, rot shop reach top/climb

   ()   MOT





sit  MOT 



    

 

 work/cook still shine write dissertation bees humming . . .

  

play for 2 h. talk during (and go on) talk during (and stop) earth turn while . . . play while . . .

MOT 

-n bää

bake/work all x

?  () 

be working! not work not rain

     

() () ()

know like now be kind now

   MOT MOT MOT

 

hang/stand stand (tempor.) sit (tempor.) improve grad. cover grad. die

546

546

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Kalmyk -jˇaPROG

Karachay -na -a-ad bää- -a turIPFV CONT DUR PROG

j) motion verbs  21 MOT 22  66–67 k) phasal verbs  23/25  24/26

MOT 

-a-ib tur-ibIPFV DUR/RES RES

MOT MOT



go out just now fly to go tomorrow begin finish

* ködl-jˇ uga bää-jˇ Æ öngr-n gi-jˇä-lä (die-CONV SAY-SIM-ANT) ‘was about to die’ () forms judged less good by the informant

or simply extended duration. All postural verbs are possible in this form, which often corresponds to the postural verb construction in Germanic languages (d). It is compatible with the Perfective (e) and even with the Progressive (f): (33) a. b. c. d.

e. f.

Kalmyk naad-ad bää-nä ‘continues playing’ (Todaeva 1976: 154) end-tendän xälä-häd jov-na ‘is looking here and there’ (Bläsing 1984: 25) ükrmüd arhd idš-äd jov-la ‘the cows were grazing in the meadows’ (PROGQ: 62) ter dissertatsan biˇc-äd suu-na ‘she is [sits] writing her dissertation’ (PROGQ: 18) casn narna kücnd xääl-äd bää-v ‘the snow melted slowly in the force of the sun’ (Bläsing 1984: 25) Oˇcrig Kitd or-ad jov-jˇana ‘Ocr was travelling in China ’ (Bläsing 1984: 27)

Unlike the Germanic postural verb forms, the Durative was not used in focalized contexts (cf. Table 3). This and the combinability with Progressive show, that it is an actional periphrasis and does not qualify as a Dur-PROG.

5.2. Karachai Our material for Karachai is very limited and the interpretation of the data must therefore remain rather tentative. The great advantage of the Karachai questionnaire

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

547

The progressive in Europe

547

is that the informant gives all possible translations – sometimes up to five different forms – and often comments on their degree of acceptability. The forms to be considered here are: Imperfective Progressive Resultative I Resultative II Durative

-a-( -y-) -a-( -y-) tur-/bar-ib-ib tur-ib tur-/bar-

(simultaneous converb) (simultaneous converb  ‘be’/‘go’) (anterior converb) (anterior converb  ‘be’) (anterior converb  ‘be’/’go’ )

In the typical progressive contexts (e.g. PROGQ: 6–20), the informant used the PROG marked by -a tur-, usually besides an Imperfective; e.g., with – agentive and nonagentive verbs: suu k’ayna-y tur-a-dï (boil-SIM be-IPFV-3:SG) ‘the water is boiling’ (PROGQ: 37), – motion verbs: cˇ ig-a tura-dï / bara-dï ‘is just going out’ (PROGQ: 21), – phasal verbs: e˙ t-ib boša-y tura-dï ‘is finishing teaching’ (PROGQ: 26), – temporary states: bildir-e tura-sïz ‘you (PL) are being nice’ (PROGQ: 42). It was not used with postural verbs denoting a temporary state (PROGQ: 28, 58), and not with true states (bil-e-di ‘knows’). In contrast to Kalmyk -ad bää-, the parallel Karachai form with the anterior converb in -ib was used almost as often as PROG in the questionnaire. This form is sometimes claimed to have resultative, perfect or progressive/actual present function (Nedjalkov & Nedjalkov 1987: 114).26 In most occurrences of the PROGQ, the form in -ib tur- is comparable to Kalmyk -ad bää-, i.e., it serves as an actional specifier. Often it is listed together with the simple form as a lexical unit in the dictionary; e.g., Russ. sidet’ ‘sit (ipfv)’: oltur-urg’a, oltur-ub tur-urg’a; Russ. spat’ ‘sleep (ipfv)’: džuk’la-rg’a, džukl’a-b tur-urg’a; Russ. deržat’ ‘hold (ipfv)’: tut-arg’a, tut-ub tur-urg’a. (34)

Karachai (PROGQ: 3) Džon kel-gen-in-de, Anna alk’ïn džukla-b John come-PF-POSS-LOC Anna still sleep-ANT.CONV tur-a-dï / džukl-a-dï be-IPFV-3SG / sleep-IPFV-3SG ‘When John came, Anna was still sleeping.’

Postural verbs in actual present contexts were translated with three different forms: (35)

Karachai oltur-a-dï / oltur-ub tur-a-dï / oltur-ub-dï ‘is sitting’ (PROGQ: 28) süel-e-di / süel-ib tur-a-dï / süel-ib-di ‘is standing’ (PROGQ: 59) IPFV IPFV/RES II RES I

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

548

548

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

The Imperfective is formed from the stative meaning component of the initio-transformative verbs oltur- ‘sit down, sit’, süel- ‘stand up, stand’, or from the composite verbs olturub tur-, süelib tur-. The Resultative is formed from the transformative meaning component. The second form is structurally ambiguous, although the denotative content is identical: it is either a RES II27 of the verbs oltur-, süel-, or an IPFV of the verbs olturub tur-, süelib tur-. There is no progressive and no perfect meaning involved in this ambiguity. The Durative is used with all types of verbs except momentaneous ones, and exhibits the same range of meanings as the Kalmyk Durative. With accomplishment verbs the informant prefers the auxiliary bar- ‘go, move’, probably to avoid the ambiguity which -ib tur- can create with these verbs. (36) a.

b.

Karachai (PROGQ: 57)28 k’art öl-üb bar-a e˙ di. old_man die-ANT.CONV go-SIM.CONV was ‘The old man was dying.’ (PROGQ: 55) k’ar uak’-uak’ džer-ni džab-ïb snow slowly-slowly ground-ACC cover-ANT.CONV bar-a e˙ di. go-SIM.CONV was ‘ snow was gradually covering the ground.’

The Durative is not excluded in habitual contexts; e.g., ol kitab ok’u-y-dï (IPFV)/ ok’ub turadï (DUR) ‘she reads [every Saturday]’ (PROGQ: 2), and it can be combined with a PROG. (37)

Karachai (PROGQ: 27) Ol xapar-nï ayt-ïb bar-a tur-a-dï. s/he story-ACC tell-ANT.CONV go-SIM.CONV be-IPFV-3SG ‘She is continuing telling a story.’

In the PFQ the -ib tur- form was used only with telic predicates. Still, not all (nondurative) -ib tur- forms can be interpreted as resultatives. Nedjalkov & Nedjalkov (1987) found that the test adverb alk’ïn ‘still’ can not be inserted before -ib tur- forms from active transitive verbs,29 i.e., there seem to be no possessive resultatives. (38) has perfect meaning only:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

549

The progressive in Europe

(38)

549

Karachai (Nedjalkov & Nedjalkov 1987: 116) Ata ešik (*alk’ïn) aˇc-ïb tur-a-dï. father door still close-CONV be-IPFV-3:SG ‘Father has closed the door.’

alk’ïn was also rejected with some intransitive verbs, e.g., ol (*alk’ïn ) ketib turadï ‘he has come’. The resultative -ib tur- form thus goes the common way towards a perfect meaning (the perfect in -gan develops towards a preterite). Which verbs allow a resultative interpretation is not predictable. Nedjalkov & Nedjalkov list 15 verbs (out of 400 they tested) for which they claim actual present meaning with -ib tura-, among them the postural verbs, but also ‘sew’, ‘plough’, ‘graze (trans.)’, ‘fish’. We do not see any reason to interprete the -ib tur- form in ol kiyim tigib turadï ‘she sews / is sewing dresses’ different from the durative forms.

6. Conclusions Of the four forms used in the Kalmyk PROGQ, two (V  -jˇana and V  -jˇ bää-/jov-) can be classified as Foc-PROG, although they show some signs of defocalization. Karachai has only one Foc-PROG form (-a tur-). The Kalmyk imperfective participle form in -a is a rare subtype of the imperfective dimension which expresses that a situation holds at a point of reference and for some time before. It could probably be located somewhere between Foc-PROG and Dur-PROG. In both languages the form made up of the anterior converb  postural verb marks various types of durative actionality. The Karachai form in -ib tur- (and no other auxiliary) is also a resultative, often with perfect meaning. There is no evidence in our data that the durative forms can have progressive function. Unlike the Germanic postural verb constructions, which were classified as instances of Dur-PROG, the Kalmyk and Karachai duratives can be combined with a progressive and are not suitable to mark focalization. The situation is similar in other Turkic languages (Schöning 1984: 324, Johanson 1971, 1995). In Tatar the combination of the converb in -p  postural verb is an actional periphrasis specifying the phasal meaning of fini-transformative and initiotransformative verbs, while other auxiliaries specify the transformational meaning component, e.g., (39)

Tatar ül-ep kit-te die-CONV go_away-PT:3 ‘he died’ vs.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

550

550

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

ül-ep yat-tï die-CONV lie-PT:3 ‘he was dying’ kür-ep al-dï see-CONV take-PT:3 ‘he saw (discovered)’ vs. kür-ep tor-dï see-CONV be/stand-PT:3 ‘he saw’ The simultaneous converb in -a together with tor- ‘be, stand’ has been grammaticalized as a progressive, ruling out combinations of -a with other postural verbs (Schöning 1984: 277). Several authors mention that constructions with -p tor- and -p utïr(‘sit’) can also mark an actual present and thus come close in meaning to the progressive. The form also expresses perfect meaning: uqïp tordï ‘has read’. Kumyk has actional periphrases with actionally ambiguous, but also with activity verbs: yuxlap tur- ‘sleep’, ašap tur- ‘eat’, oxup- tur- ‘read’. The Imperfective from these complex lexemes is formally identical to the Resultative: e.g., oxup tura ‘reads’, gelip tura ‘is/has come’ (cf. Johanson 1995 and to appear). The Altaic forms are not unlike those in Western European languages. Simultaneous converbs or participles  auxiliary ( postural verb) yield progressives, anterior converbs or participles  auxiliaries yield resultatives and perfects. Postural verbs are used in durative periphrases. There are, however, some important differences: combinations of converb  auxiliary have been lexicalized to a considerable degree in Altaic languages. In Western European languages such lexicalizations are rare and mainly restricted to non-agentive verbs (e.g., Engl. be tired, Germ. umgeben sein von). The durative periphrases (at least in Kalmyk and Karachai) are not used in focalization and are therefore no instances of Dur-PROG. A further difference results from the weak notion of anteriority associated with the converbs -ad, -ib, which allows them to be used much like simultaneous forms. This leads to a further characteristic of some Turkic languages, namely that one form can have resultative/perfect and durative function.

Notes 1. This work was jointly developed by the three authors. However, CdG bears the main responsibility for Sections 1 and 2, PMB for Sections 3 and 4, and KE for Section 5. 2. An apparent counterexample could be the Albanian construction with the particle po (cf. fn. 3). However, the meaning of this particle is not strictly temporal; thus, it represents a different type.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

551

The progressive in Europe

551

3. For our purposes, the classification proposed by Blansitt turned out to be more insightful than, for instance, the classification by Bybee et al. (1994: 128–129), which on the one hand refers to morphological devices not used by European languages (such as “reduplication”), and on the other hand is mostly oriented towards identifying the possible sources of the progressive (a preoccupation which does not concern us here). 4. The particle po is considered to be an intensifier by Duchet (1995). Indeed, besides forming PROG in combination with the Present or the Imperfect, it may have a purely intensifying function when used with other tenses or other grammatical forms (like pronouns). However, its service as a PROG marker is well established. Albanian has another way of expressing progressivity, namely the construction ‘be’  Gerund (Camaj 1984): (i)

Albanian Zogitë janë duke sjellë kandrra në çerdhe. birds are GER bring insects to nest ‘The birds are bringing insects to their nest.’

In Geg, the Gerund-building form is tue rather than duke. Note however that in Northern Geg PROG is preferably expressed by the construction ‘be’  kah  inflected verb (where kah is a preposition meaning ‘towards’). Apparently, the latter construction is both a PROG marker and a marker of imminentiality (Alexander Murzaku, p.c.). In Arbëresh (a Tosk variety spoken by the Albanian minority in Italy), PROG takes the form of ‘be’  e  inflected verb (where e is the conjunction) or ‘be’  që  inflected verb (where që is a complementizer analogous to English ‘that’, its actual realization varying from dialect to dialect) (Francesco Altimari, p.c.). In some Arbëresh varieties one finds also the periphrasis vete (‘go’)  Gerund (Breu 1982), which apparently has the same functions as the analogous constructions to be found in several Romance languages (cf. the “Motion auxiliary type” in Table 1). 5. It is also possible to use other verbs as auxiliaries, such as egon ‘stay’, ibili ‘walk, function/work’, ihardun ‘keep doing something’, again combined with the Imperfective Participle. However, the use of these constructions is dialectally bound. The inessive case of nominal forms substitutes for the Imperfective Participle in certain cases (PROGQ:1, 35: ametsetan ‘dreaming’, lit. ‘in dreams’). Sometimes, both possibilities (verbal and nominal) coexist: e.g., hitzegiten ari da or hizketan ari da ‘s/he is talking’; hitzegiten ‘talk’ (lit. do talk (Participle)), hizketan ‘in the talk’ (cf. PROGQ:44, 53). Still another possibility (cf. PROGQ:51) is offered by nominal forms with the instrumental suffix combined with ibili ‘walk’ or joan ‘go’ as conjugated forms (note that no sense of motion is implied in these cases). A few verbs do not have the analytic construction in ari, but present synthetic forms that are used with the Present and the Preterite: izan ‘be’, egon ‘stay’, joan ‘go’, etorri ‘come’, ibili ‘walk’, eduki ‘have’, ekarri ‘bring’, eraman ‘take (somewhere)’, erabili ‘use’, jakin ‘know’, ihardun ‘keep doing something’, and some other relatively unproductive ones. These synthetic forms can take the progressive meaning in the appropriate contexts. However, with some of these verbs the analytic construction is gaining ground, as in jakiten ari naiz (lit.: I am knowing) vs. dakit ‘I know’. (Data from Miren Lourdes Oñederra, p.c.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

552

552

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

6. Since this is a potential source of misunderstanding, we would like to clarify the conventions that we are going to adopt in this chapter. When referring to the distinction Perfective/Imperfective, as used in the grammatical descriptions of Slavic languages, we shall use capital initials. When this convention is not used, these terms should be intended as they usually are in the grammatical descriptions of Germanic, Romance etc. We refer the reader to Bertinetto & Delfitto (this volume), for a detailed discussion of this topic, and a possible terminological distinction, whereby the terms ‘bounded/unbounded’ are used in the former sense, and ‘terminative/nonterminative’ in the latter. However, as the authors make clear, the need for this disambiguation arises only in particular situations. When not strictly necessary (i.e., when there is no real danger of misunderstanding), the traditional terminology can be retained. Cf. also Johanson (this volume), for a thorough clarification of the aspectual matter. Still with respect to terminology, the reader should be aware that we use the word ‘actional’ as the adjective corresponding to ‘actionality’ (i.e. Aktionsart). 7. As observed in Bertinetto, this volume, the Italian PROG device based on motion verbs (cf. Table 1) is almost uniquely used in formal styles. 8. Blansitt (1975:3) refers to Chafe (1970) for the notion of “generic progressive” (exemplified by John is playing golf once a week), but also quotes Bee (1973) for the distinction between the two meanings of He is eating, which may indicate either a currently occurring event, or the fact that someone is again able to eat after a severe illness. In addition, Blansitt (p. 4) notes that there is an “intensive durative” construction, somehow related to the progressive, as in He is reading away, which may correspond to Spanish Está lee y lee. This type resembles the hyperbolic uses of PROG, exemplified by Eng. He is always reading or Spanish Está siempre leyendo (cf. examples (11) in Bertinetto, this volume). 9. A situation where an event overlaps with a simultaneously ongoing process was called “Inzidenzschema” by Pollak (1960: 129); cf. “a relation between a dynamic situation and a point in time” in Dahl (1985: 91), and Comrie (1976: 3), where this kind of situation is used to illustrate imperfective (progressive) as opposed to perfective aspect in certain languages. 10. In Hungarian, Perfective verbs cannot be used with the ABSV (de Groot 1995), although, as shown in (21b), they appear in progressive contexts (cf. Section 2.6 for our remarks). But the contradiction is only apparent; one should not confuse Perfective with perfective (cf. again fn. 6 for our notational conventions). The incompatibility of Perfective verbs with the ABSV is an obvious consequence of the fact that this construction depicts a situation as occurring at a particular reference interval, and thus concerns an event that cannot be viewed as completed. 11. As to the contrast between Sp. estuvo leyendo/ha estado leyendo todo el día on the one side, and Eng. he has been studying the whole day, cf. Bertinetto & Delfitto (this volume, Section 3.2). 12. As shown by Ebert, this volume a, in some Germanic languages this constraint takes the form of an “intentionality” requirement, or in some cases of a “dynamicity” requirement. In particular, prepositional periphrases (cf. Table 1) are preferred with higher dynamicity (cf. table 3 in Ebert, this volume a), and are in general more readily available with a focalized interpretation. Among the Germanic languages considered, only Icelandic seems

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

553

The progressive in Europe

13.

14.

15.

16.

17.

553

immune from this component; but significantly, the PROG construction of Icelandic is, together with that of English, the most grammaticalized one within the Germanic group. As to the variable strength of the agentive/intentional requirement with the ABSV, cf. De Groot (this volume). An apparent exception is represented by the Swedish ABSV construction, which may be based on the inherently passive auxiliary blir (De Groot, this volume). But one may contend that in this construction the original passive meaning of the auxiliary is bleached. Needless to say, our observations do not refer to those PROG or ABSV periphrases that are normally constructed with an Infinitive governed by the copula with the possible help of a preposition (for the relevant morphological data, see Table 1 here and De Groot, this volume). Whether or not a PROG construction based on the Infinitive in the main verb may also admit the Infinitive in the auxiliary verb depends on the given language. For instance, when PROG is governed by a modal verb, Standard Portuguese presents the Infinitive in both the copula and the main verb. Interestingly, Germ. gerade is also indicated by some scholars as a PROG device. But see the contrary opinion expressed by Ebert, this volume a. In fact, one should note that gerade can be employed in stative, i.e. clearly non-progressive, contexts (such as: Er ist gerade da), whereas Alb. po is rejected in the same contexts. As to the expression of PROG in Albanian, cf. fn. 4. Note, however, that the permanent stative (or individual-level) interpretation of example (25b) presupposes an equative, rather than a locative meaning of the copula. Thus, there are grounds to believe that there are in fact two possible sources for progressive periphrases, incorporating an existential-locative meaning or an existential-equative meaning respectively, with the latter converging with the former at some later stage. Another hypothesis that one could plausibly put forth is that stage (i) constitutes an entirely independent evolutionary path. Accordingly, one could suppose that the periphrases of stage (ii) exploited an already existing morphosyntactic structure, which had become available for a different usage. However, this hypothesis is weakened by the fact that the same development has made its appearance in two fairly heterogeneous languages, Latin and Old English. Note finally that PROG constructions based on motion verbs presumably enter the evolutionary path directly at stage (ii), for the purely stative meaning typical of stage (i) is alien to the inherent semantic import of these verbs. Examples of PROG devices implementing the initial phase of stage (ii) may still be found in Modern Romance languages. This is to be observed for instance in: (i)

Italian Che ci stai a fare? What there stand:2SG:PRS at do ‘What are you doing here/there?’

(ii)

Spanish Todos los días estás una hora escribiendo. all the days be:PRS:2SG one hour writing ‘You write one hour every day.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

554

554

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot The periphrasis in (i) is similar to the Italian infinitival PROG (cf. Table 1 and Bertinetto, this volume), but differs from it in that it includes an explicitly locative morpheme. As to (ii), it may easily be confused with Spanish copular PROG, but its syntactic behaviour is different, as shown by negation (Luis García Fernández, p.c.): (iii)

Spanish a.

b.

c.

d.

Está escribiendo. be:3SG:PRS writing ‘S/he is writing.’ escribiendo. / *Está sin escribir. No está not be:3SG:PRS writing be:3SG:PRS without write ‘S/he is not writing.’ Está una hora escribiendo. be:3SG:PRS one hour writing ‘S/he spends one hour writing.’ No está una hora escribiendo. / Está una hora not be:3SG:PRS one hour writing be:3SG:PRS one hour sin escribir. without write ‘S/he does not spend one hour writing / S/he remains for one hour without writing.’

In (d), the presence of the temporal adverbial allows for the negation with sin, which is not admitted in (b) with progressive meaning. What is particularly remarkable is that the auxiliary in (c–d) preserves by and large its original, locative meaning (thus, it is not a true auxiliary). The same applies to the Italian construction “starsene a  Infinitive”, which exhibits a verb with an explicit locative meaning preceding the prepositional Infinitive. In both cases, there is the implication that the event takes place in a specific place. 18. Even in Catalan there is a tendency to avoid PROG in a context like this. But the usage of PROG is in general slightly more restricted in Catalan than in the other Ibero-Romance languages (Bertinetto, this volume). 19. The same remark is put forth for Breton PROG by Hewitt (1985/86). 20. Note that this is not always the case. In Dutch, for instance, when the adverb heen is present, the Participle may not be added: (i)

Dutch Ze zijn heen aardappels rooien (*gegaan). they are away potatoes dig:INF gone ‘They are off digging up potatoes.’

21. We thank Igor and Vladimir Nedjalkov, who translated the questionnaire into Russian, as well as B. A. Biˇceev and A. I. Xasanov, both trained linguists and native speakers of Kalmyk and Karachay respectively, for their collaboration.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

555

The progressive in Europe

555

22. The term “Präsens” applies to forms with the suffix -na: -jˇana Präsens durativum I, -ad bäänä Präsens durativum II. The combination of -jˇa with the past/anterior marker -la is called Imperfectum III (p. 39), with FUT -ax Präsens II (-jˇax does not occur in my data); -ad bää-  PFV -v is called Imperfectum I. 23. This form is rare also in texts; cf. Todaeva (1976: 159), Bläsing (1984: 75). 24. Cf. Badm traktorist ködl-ä ‘Badm works as a tractor driver’ (Benzing 1985: 128). 25. This does not hold vice-versa. For fini-transformatives the dictionary is very careful to give only imperfective translations. Apparently Russian linguists expect telic/perfective verbs to be marked morphologically, and consequently they describe only verbs marked with telecizers like -ˇck as “perfective”. 26. A dual progressive-perfect function has been claimed also for Khalkha Mongolian -aad bai-, but the only example found in articles and texts with a perfect meaning is ir-eed baina ‘has come’. Otherwise -aad bai- is a durative periphrasis, like Kalmyk -ad bää(cf. Ebert 1995: 196). 27. There seems to be no big semantic difference between RES I and RES II; whenever the informant used RES I, he also gives a RES II form as an alternative, though not vice versa. RES I seems to be rare. In the PFQ it was given only as one of several possibilities in: ol uyan-ïb-dï (RES I) / uyan-ïb turadï (RES II) ‘he woke up already’ (PFQ: 30–31). 28. The progressives öle tura e˙ di and džaba tura e˙ di were considered “worse” in those examples. 29. The authors claim that resultatives can be formed from a few transitive active verbs, but the only example given, ol kitab-nï alk’ïn al-ïb turadï ‘s/he is still holding the book’ is most probably an actional periphrasis of the initio-transformative verb al- ‘take, hold’.

References Bee, Darlene L. 1973 Neo-tagmemics: An integrated approach to linguistic analysis and description. Ukarumpa: Summer Institue of Linguistics. Benzing, Johannes 1985 Kalmückische Grammatik zum Nachschlagen. Wiesbaden: Harassowitz. Bertinetto, Pier Marco 1994 “Statives, progressives and habituals: Analogies and divergences”, Linguistics 32: 391– 423. this volume “The progressive in Romance, as compared with English”. Bertinetto, Pier Marco, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.) 1995 Temporal reference, aspect and actionality. Typological perspectives. Torino: Rosenberg & Sellier. Bläsing, Uwe 1984 Die finit indikativischen Verbalformen im Kalmückischen. Wiesbaden: Steiner. Blansitt, Edward L. Jr. 1975 “Progressive aspect”, Working Papers on Language Universals 18: 1–34. Breu, Walter 1982 “Forme verbali perifrastiche arbërisht”, in: Guzzetta, Antonino (ed.), Etnia albanese e minoranze linguistiche in Italia. Palermo: Istituto di Lingua e Letteratura Albanese, Università di Palermo, 313–333.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

556

556

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Bybee, Joan L., Revere Perkins & William Pagliuca 1994 The evolution of grammar. Tense, aspect, and modality in the languages of the world. Chicago/London: The University of Chicago Press. Bybee, Joan & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Camaj, Martin 1984 Albanian Grammar with exercises, chrestomathy and glossaries. Wiesbaden: Harrassowitz. Carlson, Gregory 1978 References to kinds in English. Bloomington: Indiana University Linguistics Club. Chafe, William L. 1970 Meaning and the structure of language. Chicago: The University of Chicago Press. Comrie, Bernard 1976 Aspect. An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Dahl, Östen 1985 Tense and aspect systems. Oxford: Blackwell. De Groot, Casper 1995 “The absentive in Hungarian”, in: I. Kenesei (ed.), Levels and structures (Approaches to Hungarian 5). Szeged: JATE, 45–61. De Groot, Casper this volume “The absentive”. Delfitto, Denis & Pier Marco Bertinetto 1995 “A case study in the interaction of aspect and actionality: The Imperfect in Italian”, in: Pier Marco Bertinetto, Valentina Bianchi, James Higginbotham, Mario Squartini (eds.), Temporal reference, aspect and actionality: Semantic and syntactic perspectives. Torino: Rosenberg & Sellier: 125–142. Dietrich, Wolf 1973 Das periphrastische Verbalaspekt in den romanischen Sprachen. Tübingen: Narr. Duchet, Jean-Louis 1995 “The Albanian tense system”, in: Rolf Thieroff (ed.), 253–275. Ebert, Karen H. 1995 “Ambiguous perfect-progressive forms across languages”, in: Bertinetto, Pier Marco, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.), 185–203. 1996 Kodava (Kodagu). Lincom Europa: Languages of the World Materials No. 104. ª ª to appear “Focality degrees in Kalmyk imperfectives”, in: Leonid Kulikov & Heinz Vater (eds.), Typology of verbal categories. Papers presented to Vladimir Nedjalkov on the occassion of his 70th birthday. (Vol. 2). Tübingen: Niemeyer. this volume a “Progressive markers in Germanic languages”. this volume b “Aspect in Maltese”. Gougenheim, Georges 1929 Étude sur les périphrases verbales de la langue française. Paris: Nizet. Haase, Martin 1994 “Tense and aspect in Basque”, in: Rolf Thieroff & Joachim Ballweg (eds.), 279–292. Haspelmath, Martin 1993 A Grammar of Lezgian. Berlin: Mouton de Gruyter. Heinämäki, Orvokki 1995 “The progressive in Finnish: Pragmatic constraints”, in: Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl, Mario Squartini (eds.), Temporal reference, aspect, and actionality. Typological perspectives. Torino: Rosenberg & Sellier: 143–153.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

557

The progressive in Europe

557

Hetzron, Robert 1982 “Non-applicability as a test for category definitions”, in: F. Kiefer (ed.), Hungarian Linguistics. Amsterdam: Benjamins, 131–183. Hewitt, Steve 1985/86 “Le progressif en breton à la lumière du progressif anglais”, La Bretagne Linguistique 2. Johanson, Lars 1971 Aspekt im Türkischen. Vorstudien zu einer Beschreibung des türkeitürkischen Aspektsystems. Uppsala: Acta Universitas Upsaliensis. 1995 Mehrdeutigkeit der türkischen Verbalkomposition. In: M. Erdahl & S. Tezcan (Hrsg.), Beläk Bitig. Sprachstudien für Gerhard Doerfer zum 75. Geburtstag (Turcologica 23). Wiesbaden, 81–101. to appear “Grenzbezogenheit in Aspekt und Lexik.” Paper read at the workshop “Interaktion zwischen Lexik und Aspekt”. May 1995, Universität Konstanz. this volume “Viewpoint operators in European languages”. Kiefer, Ferenc 1994 “Aspect and syntactic structure", in: Ferenc Kiefer & Katalin É. Kiss (eds.), The syntactic structure of Hungarian. Syntax and Semantics vol. 27. San Diego: Academic Press, 415–464. King, Alan R. 1994 The Basque Language. A practical introduction. Reno: University of Nevada Press. König, Ekkehard 1995 “He is being obscure: Non-verbal predication and the progressive”, in: Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl, Mario Squartini (eds.), Temporal reference, aspect, and actionality. Typological perspectives. Torino: Rosenberg & Sellier, 155– 167. König, Ekkehard & P. Lutzeier 1973 “Bedeutung und Verwendung der Progressivform im heutigen Englisch”, Lingua 32: 277–308. Kozintseva, Natalia 1995 “The tense system of Modern Eastern Armenian”, in: R. Thieroff (ed.), 277–297. Metslang, Helle 1995 “The progressive in Estonian”, in: Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl, Mario Squartini (eds.). Temporal reference, aspect, and actionality. Typological perspectives. Torino: Rosenberg & Sellier, 169–183. Munieva, B.D. 1977 Kalmycko-Russkij slovar / Xal’mg-Ors Tol’. Moskva: Izdatelstvo “Russkij Jazyk”. Nedjalkov, Igor V. & Vladimir P. Nedjalkov 1987 “Karaˇcaevo-balkarskaja glagol’naja forma na -b/-p tur-a- co znaˇcenijami nastojašˇcego i prošedšego vremeni (v sravnenii s formami na -b tur-a/tur-ib- v uzbekskom jazyke”, Funkcional’no-semantiˇceskie aspekty grammatiki. Moskva: Nauka. 13–21. Poppe, Nicholas 1968 Tatar manual. Descriptive grammar and texts with a Tatar-English glossary. [2d rev. ed.]. (Indiana University Publications. Uralic and Altaic series, v. 25). Bloomington: Indiana University. Olbertz, Hella 1998 Verbal periphrases in a functional grammar of Spanish. Berlin: Mouton de Gruyter. Schönig, Claus 1984 Hilfsverben im Tatarischen. Wiesbaden: Steiner. Thieroff, Rolf (ed.) 1995 Tense systems in European languages II. Tübingen: Niemeyer.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

558

558

Pier Marco Bertinetto, Karen H. Ebert, Casper de Groot

Thieroff, Rolf & Joachim Ballweg (eds.) 1994 Tense systems in European languages. Tübingen: Niemeyer. Thieroff, Rolf this volume “On the areal distribution of tense-aspect categories in Europe”. Todaeva, Buljaš Chojˇcievna 1968 “Kalmyckij jazyk”, in: Skorik, P. Ja. (ed.), Jazyki Narodov SSSR. Leningrad: Nauka, 35–52. 1976 Opyt lingvistiˇceskogo issledovanija èposa “Dzangar”. Èlista: Kalmyckoe knižnoe izdatel’stvo. Tommola, Hannu this volume “The progressive in Baltic Finnic”.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

559

Pier Marco Bertinetto

The progressive in Romance, as compared with English

1. Prolegomena1 This chapter collapses two apparently disparate entities: the Romance languages and a single language belonging to the Germanic group. This might be considered inappropriate from a strictly genetic point of view. But the typological perspective that we are assuming in this series of contributions justifies such a move. In fact, this solution should not even appear particularly surprising, given the fact that English is, among the Germanic languages, the one that has most dramatically departed from the other languages of the group as a result of the protracted contact with French in a crucial phase of its history. But what matters most is that, regarding the particular phenomenon we are going to discuss here (i.e. the progressive), the languages under consideration present deep affinities. It can easily be shown that Germanic languages other than English exhibit quite different features, as illustrated in the companion chapter by Karen Ebert (this volume). It has even been claimed that the English progressive was shaped by the Romance model. However, the alternative view, according to which this construction represents an autonomous development, is equally defendable (Scheffer 1975). This issue of course needs to be addressed in proper terms, namely through the comparative investigation of ancient texts (to the extent that they provide evidence for the crucial period). Here I shall disregard it, and merely concentrate on the observable synchronic situation, characterized by strong similarities between Romance and English, be it a matter of common origin or of mere convergence. A clarification is in order at the outset. We have to distinguish between “progressive” as a semantic notion and as a formal manifestation (i.e. a morphosyntactic device). To convey the latter sense, I shall use the abbreviation PROG (except when I cite the traditional grammatical denomination of a given tense, such as, e.g., the English “Present Progressive”). Although in many cases there is a reciprocal implication, this is not always so. Consider Romance languages, where the progressive aspect is not necessarily conveyed by specialized morphosyntactic devices (namely periphrases), for the mere usage of imperfective tenses is in most cases perfectly adequate. And this is not the only complication. In fact, on the one hand we find prototypical contexts where the notion “progressive” is necessarily present indepen-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

560

560

Pier Marco Bertinetto

dent of the particular device employed, be it a general purpose imperfective tense (of the sort we can find in Romance) or a specific periphrasis (like in English, as well as in Romance). On the other hand, specialized PROG devices may also appear in contexts which have little to do with the aspectual notion “progressive”. To quote an obvious example, consider English PROG with future-time reference, as in: I am leaving tomorrow. Although the development of this meaning must originally have been licensed by some specific property possessed by what we might call the “prototypical” progressive aspect (as is shown by the fact that English is not the only language showing this particular development; cf. Section 6.3.3), it is clear that this usage of PROG does not convey any progressive meaning, in the proper sense of this term. Thus, the correspondence of form and meaning is not always perfect. Nevertheless, it is fair to say that whenever a specialized device exists, it is quite likely that we find at least some contexts where this is considered by the speakers as a quite natural choice (or maybe as the only option available). In this chapter I shall mainly be concerned with the morphosyntactic device PROG, rather than with the progressive aspect in the strict sense, and its possible manifestations. The data I shall discuss here are drawn in part from the questionnaire that was prepared and distributed by EUROTYP Group 6 (henceforth PROGQ; cf. Appendix 3), but it will also be complemented by a survey of the available literature. The languages for which we collected data, among those relevant here, are the following:2 Catalan: English: French: Italian: Portuguese: Romanian: Spanish:

1 subject (Standard Central Catalan) 1 British subject (no declared dialectal background) 2 subjects (no declared dialectal background) 4 subjects (1 from the North, 1 from the Centre, 2 from Sardinia; all speaking varieties of Standard Italian) 3 European subjects (no declared dialectal background) 3 subjects (no declared dialectal background). 2 European subjects (Standard Spanish).

As to the morphosyntactic devices employed, Table 1 lists the most relevant ones. The label ‘St-PROG’ indicates periphrases based on auxiliary verbs approximately meaning “be, stand”. ‘Mot-PROG’ stands for periphrases based on auxiliary verbs meaning “go, come”. The third type is a miscellaneous category comprising the residual devices. As can be seen, in French and Romanian the third type is virtually the only one existing (considering that “aller  GER” in Modern French is extremely rare). However, even with these languages it is correct to call this type “Marginal” because of its relatively infrequent usage, which is quite remarkable in Romanian. This appears also in PROGQ, where PROG is very seldom employed by the Romanian informants. Consequently, in what follows I shall have very little to say about this language, except for some narrowly targeted observations (cf. Section 6.1).3

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

561

The progressive in Romance, as compared with English

561

Table 1. Survey of the morphological manifestations of PROG devices in Romance and English. State-PROG English Catalan French Italian Portuguese Romanian Spanish

be  V-ing estar  GER stare  GER stare a  INF estar a  INF estar  GER estar  GER

Motion-PROG

Marginal types

anar  GER (aller  GER) andare/venire  GER

? ? être en train de  INF (essere dietro a  INF)

ir/vir  GER

?

ir/andar/venir  GER

a fi în curs de a  INF ?

St-PROG divides further into a gerundial and an infinitival type, as shown in Table 1. For clarity, I shall designate them St-PROG-GER and St-PROG-INF respectively. However, I shall avoid providing these additional qualifications when I refer to the general type, or whenever it is sufficiently clear which is the subtype I am referring to. St-PROG-INF is the standard device in European Portuguese, although the gerundial type is equally present, especially in the written language. Curiously, in Brazilian Portuguese the situation is reversed, with St-PROG-GER commonly used, at the expense of its competitor. As shown in Table 1, St-PROG-INF also exists in Italian. This device is now confined to the colloquial usage of the speakers of Central Italy (most typically in Rome, where the rival construction is much less frequent; cf. D’Achille and Giovanardi 1998), but it is also occasionally to be found in literary texts, including early ones. This shows that, besides standard varieties, one should also consider the situation of the local varieties, which in some known cases seem to behave quite differently. However, very little information is available on this, and even less is known on the vernaculars, which in some regions (like Italy and Romania) often present remarkable differences, at all structural levels, relative to the national standards.4 Thus, the picture I am going to present here is far from exhaustive.

2. On the evolution of PROG in Romance In Romance languages, the copula of both types of St-PROG is a descendant of Latin STARE, rather than ESSE. However, this was not necessarily the case in the ancient stages of these languages, where both types of copulae could be encountered. As to the origin of these periphrases, cf. in particular Dietrich (1973),5 who reports abundant evidence from Late Latin, particularly from Christian texts, i.e., from a type

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

562

562

Pier Marco Bertinetto

of literature relatively close to the spoken language (suggesting that these devices must have been fairly common in the actual usage of the speakers). In Latin, we find four types of construction, which are mirrored almost exactly by the early stages of Italian (Bertinetto 1986: 134–136; the Latin examples are from New Testament versions (Vetus Latina, Vulgata) and late Latin authors; the Italian ones are authors of the 13th and 14th century): (a)

“esse  Imperfective Participle” [ [ ] gemens et tremens eris moaning and trembling be:2SG:FUT ‘[ ] you will be moaning and trembling [ ]’

“essere  Imperfective Participle” La Misericordia è parlante [ the Mercy is speaking ‘God’s Mercy says [ ]’ (b)

Italian ]

Late Latin “esse  Ablative Gerundive” [ ] erat Darius vociferando et congregando multitudinem was Darius shouting and gathering crowd:ACC ‘[ ] Darius was shouting and gathering the people.’ “essere  Gerundive” Le mani me son lavando [ the hands 1SG:RFL am washing [ ‘I am washing my hands [ ]’

(c)

(d)

Late Latin ]

Italian ] ]

Late Latin “stare  Imperfective Participle” autem [ ] scribae constanter accusantes [ ] stabant were:3PL thus scribes constantly accusing eum [ ] him ‘Thus [ ] the scribes were constantly accusing him [ ]’ “stare  Ablative Gerundive” stetit dux diu cunctando [ ] was chief long_time hesitating ‘The chief hesitated for a long while [ ]’

“stare  Gerundive” molt’anni libertà sognando [ ] stetti was:1SG many years freedom dreaming ‘[ ] for many years I dreamed of freedom.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Late Latin

Italian

563

The progressive in Romance, as compared with English

563

Of these constructions, the first was the most frequently attested in Latin, whereas the last became the standard device in Modern Italian.6 Although this development is also basically shared by Spanish and Catalan (and to some extent Portuguese) PROG, the evolution of Italian St-PROG is almost unique, in the sense that it underwent a radical reinterpretation. The Latin precursors, as well as the early Italian attestations, show that this device could easily be used to indicate a purely durative (static) situation, rather than a true progressive one. In fact, the verbal noun often fulfilled a purely adjectival function, and as such could combine with a habitual (1–2a) or an imperative (2b) meaning, as in the following examples (taken from Dietrich 1973 and Durante 1981, respectively), in which the event is not viewed with respect to a focalized point in time, but rather relative to an unrestricted interval:

(1)

Latin (Vulgata) Erat autem docens in synagoga eorum sabbatis. be:IMPF:3SG thus teaching in synagogue they:GEN Saturdays ‘Thus he taught in their synagogue on Saturdays.’

(2) a.

b.

Early Italian (Elegia Giudeo-Cristiana, 12–13th cent.) La notti e la die sta plorando. the night and the day be:3SG crying ‘He cries night and day.’ (C. Bascapé, 16th cent.) sta più tosto Però lascia i piaceri [ ] e and be:IMP rather therefore leave:IMP the pleasures pregando. praying ‘Therefore, abandon the amusements and rather spend your time in prayer.’

Interestingly, something fairly similar occurred in Old English, where we find sentences such as example (25b) of Bertinetto, Ebert and De Groot, this volume.7 As suggested in Section 4.1 of the quoted chapter, it is likely that this is the preliminary stage in the evolutionary path followed by PROG in most languages. And there is certainly little doubt that Italian St-PROG-GER could be used in the past (up until the beginning of the 19th century, as far as the literary language is concerned) to indicate a purely durative situation, as is proven by examples such as:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

564

564

Pier Marco Bertinetto

(3) a.

b.

Italian (G. Galilei, beginning of 17th cent.) domani vi starò attendendo amendue per [ ] e and tomorrow you be:1SG:FUT waiting both to continuare i discorsi cominciati. continue the speech:PL begun ‘[ ] and tomorrow I shall be here, waiting for you both to continue our conversation.’ (A. Manzoni, 19th cent.) Renzo lo stava guardando con un’ attenzione estatica, Renzo him be:IMPF:3SG watching with an attention ecstatic come un materialone sta sulla piazza guardando al like a guy stands on the square watching at_the giocator di bussolotti. player of dice ‘Renzo was looking at him with ecstatic attention, like a guy standing on a square and staring a player of dice.’

Example (3a) may be compared to English sentences such as: While you stay here, I’ll be going home, which typically present a durative, non-focalized situation. As to (3b), consider in particular the second occurrence of PROG. In Modern Italian, these possibilities were lost (or at least severely restricted in the case of the habitual meaning; cf. Section 6.3.2). This is proven in particular by the fact that the combinability with perfective tenses, such as the Simple Past or any of the Compound tenses, has been entirely lost (Bertinetto 1986). It may be said that Italian St-PROGGER has gone all the way along what could be called “PROG imperfective drift” (cf. Bertinetto, Ebert & De Groot, this volume, Section 4.1). Indeed, it is easy to show that Italian St-PROG-GER may now be employed (with very few exceptions) only in cases of strict focalization, as in the typical ‘incidential schema’8 , where the speaker is only concerned with what is going on at a particular point in time. Italian St-PROG-GER is thus a typical instance of what in the just quoted chapter is called “focalized” PROG. Consider, as an illustration, the following examples taken from PROGQ (cf. also PROGQ:30,76,82): (4)

PROGQ:3: [Last night at 8 o’clock] when John came, Ann still WORK. quan en Joan va venir [SP], l’Anna encara estava treballant. Cat: [IMPF:PROG] quand Jean est arrivé [CP], Anne travaillait [IMPF] encore . Fr: quando Gianni è arrivato [CP], Anna stava ancora lavorando. It: [IMPF:PROG]

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

565

The progressive in Romance, as compared with English

565

quando o João chegou [SP], a Ana ainda estava a trabalhar. [IMPF:PROG] cînd a venit [CP] Jon, Ana înc˘a lucra. [IMPF] Rom: cuando Juan llegó [SP], Ana todavía estaba trabajando. Spn: [IMPF:PROG] when John came, Ann was still working. [PST:PROG] Eng: Prt:

(5)

PROGQ:32: /The pardon arrived/ just when the captain GIVE the sign /to the firing squad/. justament mentre el capità estava fent [IMPF:PROG] el senyal Cat: Fr: It:

justement au moment où le capitain donnait [IMPF] le signal proprio mentre il capitano stava dando [IMPF:PROG] il segnale

Prt:

exactamente enquanto o capitão estava a dar [IMPF:PROG] o sinal Rom: exact în timp ce c˘apitanul d˘adea [IMPF] semnalul Spn: justo cuando el capitan estaba dando [IMPF:PROG] la orden Eng: just while the captain was giving [PST:PROG] the sign As may be seen, PROG is readily used in all the languages considered, with the exception of French and Romanian. However, the situation of these two languages is not identical: in PROGQ:76, i.e., in a context very similar to PROGQ:3, one of our French informants provided an Imperfect Progressive as an alternative to the bare Imperfect, whereas PROG appeared only very marginally in the responses of our Romanian informants and, significantly enough, never in the instances that we might regard as prototypical. This shows that while PROG has a perfectly recognizable grammatical status in French, it barely exists in Romanian. What makes French different with respect to the other languages listed in (4–5) is the relatively low frequency of usage of this device, as compared with its cognates in English, Italian, and the Ibero-Romance languages. But, as I said, English differs from Italian and the Ibero-Romance languages because PROG is the only device available to express a focalized aspectual view at any temporal location (Past, Present or Future), while all Romance languages, even those where PROG appears to be fairly frequently employed, may freely resort, in examples like (4–5), to the Present, the Imperfect or the Simple Future, depending on temporal location. In fact, our informants often provided these responses as a possible option.9 To the extent that PROG is employed in contexts like the ones above, this usage may rightly be considered prototypical, for this is precisely the type of context that is most often referred to in the literature. But in several Romance languages, as well as in English, PROG is not restricted to a purely focalized interpretation. This can be observed, most notably, in sentences containing durative adverbials, which

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

566

566

Pier Marco Bertinetto

necessarily rule out the focalized interpretation, for the event must be conceived of as developing over a stretch of time, rather than at a particular instant. Accordingly, in such cases it is appropriate to speak of “durative” PROG, as is done in Bertinetto, Ebert and De Groot, this volume. Consider the following example (cf. also PROGQ:48–50): (6)

PROGQ:51: [Moment by moment] the policeman TAKE NOTES of what the speaker said. el policia estava prenent nota [IMPF:PROG] del que deia Cat: l’orador. le policier notait [IMPF] ce que l’orateur disait. Fr: il poliziotto prendeva nota [IMPF] di ciò che diceva l’oratore. It: o polícia estava a tomar notas [IMPF:PROG] do que o falante Prt: disse. poli¸tistul nota [IMPF] ce spunea vorbitorul. Rom: el policía anotaba [IMPF] lo que decía el que hablaba. Spn: the policeman was taking notes [PST:PROG] of what the speaker Eng: said.

In Catalan, English and Portuguese our informants overwhelmingly yielded, at least as an alternative, a PROG response. This could in principle have happened also in Spanish, but the behaviour of our informants suggests that Spanish speakers are more cautious with using PROG in these contexts.10 As to the remaining languages, our informants behaved just as expected, given the presence of the particular adverbial employed (moment by moment), which clearly prevents focalization. Considering that Romanian makes very limited use of PROG devices, the really interesting cases here are Italian and French. Let us consider the situation of Italian (as to French, cf. Section 4 below). We saw in (2) above that in the early stages of the language St-PROG could also appear in contexts that presuppose a purely durative situation (and even, in the most extreme cases, a purely stative situation). Yet in Modern Italian, this possibility is completely lost with St-PROG-GER (as to St-PROG-INF, cf. below). Although it is not possible to state when exactly this evolution started, there is now good evidence concerning the recent development of St-PROG-GER and Mot-PROG in Italian (Bertinetto 1996). The literary prose of the last two centuries shows that the use of St-PROGGER with perfective tenses, which is only compatible with a durative view, was still possible at the beginning of the 19th century. However, at that time it was already restricted to a very limited set of verbs, such as guardare ‘look’ or aspettare ‘wait’ (cf. stette guardando/aspettando ‘be:SP:3SG looking/waiting’). Obviously, severe lexical specializations normally indicate that the device considered is not yet fully grammaticalized, or is undergoing a process of degrammaticalization (or, possibly,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

567

The progressive in Romance, as compared with English

567

a functional recategorization). There is thus reason to hypothesize that St-PROGGER completed its metamorphosis into a purely focalized device right around that period, and probably somewhat before in the spoken language, considering that the literary language of that time was fairly conservative. Another interesting observation emerging from the same piece of research is that, although the frequency of StPROG-GER and Mot-PROG was never very high (at least in comparison with what we observe in Ibero-Romance texts), it was nevertheless much higher in the first half of the 19th century than in the period immediately following, where the frequency of both periphrases dropped dramatically. However, in recent times (especially after World War II), St-PROG-GER gained ground again, both in written and spoken language. It has been claimed (Durante 1981) that this is due to English influence, although no real evidence has been gathered. As to Mot-PROG, it also recovered a good deal of lost ground, but is still mostly confined to the written language, being seen as a hallmark of formal style. (Andrea Villarini (personal communication) points out to me that in Lessico dell’Italiano Parlato there are 14 occurrences of MotPROG against 640 occurrences of St-PROG-GER.) Finally, there is evidence that the recent recovery of St-PROG-GER was accompanied by a considerable growth in the use of achievement verbs, which tended to be avoided with this device in the previous stages, where a striking prevalence of activities was observed.11 We can thus conclude that, due to the profound transformation that occurred in the semantic interpretation of St-PROG-GER, Italian has radically restructured this subdomain of the grammar, as compared with the Ibero-Romance languages. When durativity, rather than focalized progressivity, is imposed by the context, Modern Italian must revert to Mot-PROG (cf. Section 5 for further comments). Thus, Italian St-PROG-GER is by and large in complementary distribution with Mot-PROG regarding the features ‘focalized’ vs. ‘durative’. Another device available in Italian to express durative progressivity is St-PROG-INF, at least for the varieties having access to it (cf. above). However, although its usage is fairly common in some spoken varieties, its presence in modern literature is negligible (Bertinetto 1996). St-PROGINF is a much more flexible tool compared to Mot-PROG, since it is also available for focalized contexts, whereas Mot-PROG is restricted to durative contexts. But despite this difference, the similarity of St-PROG-INF and Mot-PROG in durative contexts is quite remarkable, especially considering that both periphrases, as opposed to St-PROG-GER, are compatible with perfective tenses. St-PROG-GER, on the other hand, is compatible with achievement verbs, which are rejected or severely constrained by both St-PROG-INF and Mot-PROG.12

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

568

568

Pier Marco Bertinetto

3. Durative (non-focalized) PROG 3.1. Durativity in Spanish State-PROG As suggested in the preceding section, Spanish St-PROG has preserved a number of possibilities that are now precluded in its Italian cognate, and this is generally true of St-PROG in Ibero-Romance languages. As a matter of fact, Portuguese would provide an even better illustration of this issue. However, for convenience the following examples will be taken from Spanish, which may be said to occupy an intermediate position between Portuguese and Catalan with respect to the use of PROG. To start with, note that Spanish St-PROG admits perfective tenses. As observed above, this is a clear indication that this device is not restricted to focalized contexts. This fact may typically be seen in sentences where St-PROG is used in conjunction with durative adverbials, such as durante dos horas or desde las tres hasta las cinco: (7)

Spanish Pedro estuvo leyendo en la cama durante dos horas / desde Pedro was reading in the bed for two hours / from las tres hasta las cinco. the 3 to the 5 ‘Pedro read in bed for two hours / from 3 to 5.’

In these cases, the event is presented as ongoing during a stretch of time of definite duration. Other types of evidence for durativity in Spanish St-PROG are provided by Squartini (1998), in research based on an extensive survey of the specialized literature and of recent corpora of actual linguistic usage. I owe to this work most of the examples reported in this section. For instance, St-PROG is allowed in conjunction with modal verbs, a possibility that is only marginally observed in Italian, and almost invariably with an epistemic interpretation, while the Spanish equivalents may retain the deontic value: (8)

Italian (PROGQ:80) Anna deve stare facendo lezione adesso (suppongo). Ann must be making lesson now guess:PRS:1SG ‘Ann should be teaching now.’

(9)

Spanish (Caracas, spoken corpus) [ ] entonces todo individuo está forzado, debe estar then each individual is forced, must be constantemente produciendo. constantly producing ‘Thus, each individual is forced to produce, must constantly produce.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

569

The progressive in Romance, as compared with English

569

Sentence (8) was provided, as one of two alternatives, by only one of the Italian informants, and is certainly considered marginal by many speakers, who tend to reject PROG in these contexts. Besides, it is clear that this sentence must be taken in the epistemic meaning, according to which the speaker suggests a supposition about the situation developing at the speech time. Consequently, the situation is conceived of as focalized. By contrast, sentence (9) was produced with a clearly deontic intention, and suggests that the event referred to must occur during a certain interval of time, also due to the adverb employed.13 Futhermore, Spanish St-PROG may also appear in two coordinated sentences depicting durative events: (10)

Spanish (Colombian, spoken corpus) conversando está escribiendo [ [ ] mientras estamos while be:PRS:1PL talking is:PRS writing ‘While we talk, he writes.’

]

Although none of our Spanish informants exploited this solution in PROGQ:70 (Yesterday, while Ann READ in her room, Martin PLAY in the courtyard), this possibility is frequently exploited in spoken Spanish. It is quite understandable that this occurs in Spanish rather than in Italian, because of the obvious durative character of the situation envisaged. This feature is even emphasized, because of the adverb employed, in hyperbolic contexts such as:14 (11)

Spanish ¡Siempre te estás quejando! always you:RFL are complaining ‘You are always complaining!’

In these contexts, Italian speakers would rather use, as an alternative to the simple Present, a periphrasis like non fare altro che  INF (lit. ‘do nothing but  INF’).15 Alternatively, some speakers would use St-PROG-INF: ti stai sempre a lamentare ‘you are always complaining’. No wonder, then, that Spanish St-PROG may also colloquially appear in habitual contexts, such as: (12)

Spanish a.

No me digas que la echas de menos, porqué la Not me tell that her miss because her estás viendo todos los días. be:PRS:2SG seeing all the days ‘Do not tell me that you miss her, because you see her every day.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

570

570

Pier Marco Bertinetto

b.

Estoy yendo al centro cada tres días. be:PRS:1SG going to_the centre every three days ‘I am going downtown every three days.’

Obviously, there may be focalized habitual contexts, such as: Whenever I arrive, he is writing. In such contexts, St-PROG would be allowed in any language possessing this device (cf. Section 6.3.2). But this is clearly not like sentences (12). It is of special interest, in this connection, to understand the meaning of constructions such as Spn Pedro estuvo dormiendo todo el día (lit.: ‘P. was sleeping [SP-PROG] the whole day’), based on the use of a perfective tense. Since these sentences are grammatical only in conjunction with adverbials expressing a delimited duration, it is clear that St-PROG retains here the perfective meaning inherent to the tense employed. On the other hand, there is good evidence that even in these cases St-PROG detelicizes any basically telic predicate. Consider the following examples: (13)

Spanish En Mérida estuvieron reconstruyendo el puente *en / durante in Merida be:SP:3PL rebuilding the bridge in / for dos años. two years ‘In Merida they kept rebuilding the bridge for two years’ (ungrammatical: ‘in two years.’)

(14)

Spanish a. ??Ayer Pilar llegó a su casa, estuvo leyendo la yesterday P. arrived to her home, be:SP:3SG reading the carta, estuvo preparando su ponencia, estuvo comiendo letter be:SP:3SG preparing her paper be:SP:3SG eating y se fue a la cama. and her:RFL went to the bed ‘Yesterday, P. came home, spent some time reading the letter, preparing her paper and eating, and finally went to bed.’ bailando, b. Aquel día nos lo pasamos muy bien: estuvimos that day we it spent very well: were:SP:1PL dancing estuvimos charlando, estuvimos comiendo. were:SP:1PL chatting be:SP:1PL eating ‘That was a very nice day: we spent some time dancing, chatting, and eating.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

571

The progressive in Romance, as compared with English

571

From (13) we learn that, with potentially telic predicates, adverbials such as en dos años (which insist on the completion of the event) cannot be employed, in contrast to adverbials such as durante dos años, which, although only compatible with perfective tenses, require an atelic situation (Bertinetto and Delfitto, this volume). From (14) we learn that a series of “estuvo  Gerund” forms cannot be used to suggest a strict temporal sequence of events, such that the end of one coincides with the beginning of the next, as in example (a). We can only use them as in example (b) to express a temporally unordered series of events, where each event localization may be interpreted independent of others, allowing for iterations and overlappings. In conclusion, Spanish St-PROG with perfective tenses expresses a perfective, durative and (in most cases) atelic situation. However, some caution is in order concerning telicity, because (as pointed out by Squartini, p.c.) one can find examples such as Estuvimos investigando todos los locales de Valladolid ‘We kept searching all the bars of Valladolid’, which seems to suggest that the search was conducted till the end. However, I am inclined to think that this is simply a pragmatic implicature, rather than a true semantic effect.

3.2. Durativity in Catalan, Portuguese and English St-PROG Admittedly, some of the uses reported in the preceding section appear to be strictly colloquial, like those illustrated in (9–11), but others are fairly common, not only in Spanish, but also in the remaining Ibero-Romance languages. Support for this also comes from PROGQ, as may be gathered from the following examples, in which I also list the responses of the English, Italian and French subjects for comparison: (15)

PROGQ:48: [Yesterday, during my sleep] Ann PLAY for two hours all by herself. l’Anna va estar jugant [SP:PROG] tota sola durant dues hore. Cat: Anne a joué [CP] pendant deux heures toute seule. Fr: Anna ha giocato [CP] per due ore tutta sola. It: A Ana esteve a jogar [SP:PROG] sozinha durante dos horas. Prt: Ana estuvo jugando [SP:PROG] dos horas ella sola. Spn: Ann was playing [PST:PROG] for two hours all by herself. Eng:

(16)

PROGQ:72: [What did Martin do yesterday evening?] He STUDY from 2 to 6, he READ the paper from 6 to 7, he EAT from 7 to 8, and then he GO to bed. Cat: our informant alternates the Simple Past and the Simple Past Progressive with the first two verbs, while the last two trigger the Simple Past only: va estar estudiant, va estar llegint, va menjar, se’n va anar al llit.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

572

572

Pier Marco Bertinetto

Fr: It:

both informants use the Compound Past only. three out of four informants use nothing but the Compound Past; only one (significantly, from Sardinia) alternates this tense, again with the first two verbs, with St-PROG-INF: è stato a studiare, è stato a leggere, ha mangiato, è andato a letto. Prt: all three informants alternate the Simple Past and the Simple Past Progressive with the first two verbs; only one of them extends this usage to the third verb, as in: esteve a estudar, esteve a ler, esteve a comer / comeu, foi para a cama. Spn: one of the two informants uses the Simple Past Progressive with the first two verbs: estuvo estudiando, estuvo leyendo, cenó, se fue a la cama. Several observations are in order here. First, as can be seen in (16), Spanish is not the most liberal among the Ibero-Romance languages concerning the usage of StPROG with perfective tenses, as is shown by the behaviour of one of the Portuguese speakers.16 Second, the general allergy of Italian speakers to the usage of PROG with perfective tenses does not concern all local varieties of the language. Sardinia is one of the areas (together with large areas in Southern Italy) where St-PROGGER is used most freely. This could be one of the linguistic features left behind by the long period of Spanish domination of the island, but of course this is little more than speculation. Finally, it is interesting to note that, among the four events of (16), the ones which elicit most of the PROG responses are the first two. It will not go unnoted that precisely the first two events correspond to atelic verbs, while the last two correspond to telic ones (with the possible exception of the verb EAT, although it may easily be taken in the sense of ‘complete one’s meal’). This peculiar distribution of the responses suggests that the durative interpretation of PROG bears clear evidence of its link with the early stages of development of this periphrasis, when it presumably worked like an actional operator much more than an aspectual one, i.e., like a device conveying the idea of the continuous development of the event over a given interval of time (cf. Bertinetto, Ebert and De Groot, this volume, Section 4.1, and Squartini 1998, who presents convincing evidence related to the interaction of actional and aspectual values within the category of progressive). English does not significantly differ from the Ibero-Romance languages. In both cases, we can find St-PROG with perfective tenses, as in the following examples, which exhibit more specifically perfectal tenses: (17)

(PROGQ:81) ‘[I am so tired:] I BAKE all day since I got up this morning.’ a.

English I have been baking all day since I got up this morning.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

573

The progressive in Romance, as compared with English

b.

c.

d.

(18)

a. b.

c.

d.

573

Catalan He estat fent pa tot el dia des que m’ have:1SG been making bread all the day since RFL he llevat aquest matí have:1SG got_up this morning Portuguese Tenho estado a cozinhar desde que me levantei have:1SG been at cooking since 1SG:RFL got_up Spanish He estado cocinando todo el día desde que me have:1SG been cooking all the day since RFL levanté. got_up (PROGQ:82) ‘[When John came home yesterday,] he was very tired because he WORK hard all week.’ English he was very tired because he had been working hard all week. Catalan estava molt cansat perqué havia estat treballant be:IMPF:SG very tired because had:3SG been working molt tota la setmana. much all the week Portuguese estava muito cansado porque tinha estado a be:IMPF:3SG very tired because had:SG been at trabalhar muito toda a semana working much all the week Spanish estaba muy cansado porque había trabajado toda la was very tired because had:SG worked all the semana. week

In these contexts, PROG carries an ‘inclusive’ meaning: it suggests that the event considered has been going on for some time up to (and including) the reference time, which in (17) coincides with the speech time. English has a marked preference for this solution in contexts of inclusivity: this is indeed the most typical function of PROG with compound tenses in this language. But in other languages this may not be the only way to convey this interpretation, as can also be gathered

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

574

574

Pier Marco Bertinetto

from our informants. The non-progressive Compound tenses are a viable alternative in the Ibero-Romance languages, provided the requirements for the usage of these tenses are met. Indeed, this consideration is simply obvious in the case of Portuguese Compound Past, which conveys almost uniquely an inclusive meaning (cf. Squartini and Bertinetto, this volume). Note that a non-progressive tense is the only response given by our Spanish informants in (18) above, although PROG is not altogether excluded in that type of sentence.17 But of course the crucial case is (17), for in some Romance languages there could in principle be competition between Simple and Compound Past. The behaviour of Spanish in this case is particularly instructive, given the broader freedom with which Spanish speakers use the Simple and the Compound Pasts. (Note however that the distribution of these tenses is to some extent, i.e., not for all speakers, regulated by the criterion of proximity, in the sense of “hodiernal/prehodiernal”; cf. Squartini and Bertinetto, this volume). Consider: (19)

Spanish Esta mañana, Pedro estuvo /ha estado this morning Pedro be:SP:3SG /have:PRS:3SG been estudiando durante cinco horas. studying for five hours ‘This morning, Pedro studied (was busy studying) for five hours.

Here, as opposed to (17–18), there is no inclusive meaning (a possibility that would anyway be precluded to the Simple Past Progressive). The event referred to is entirely confined in the past, i.e., in an interval of time which may even be quite remote from the Speech Time when the Simple Past Progressive is selected. Note, however, that a similar usage is also available to English. Bégin (1996) cites quite a number of examples like: I’ve been talking to a clever woman friend of mine this afternoon, or What are you doing up at this hour?–Bertie had a headache I have been giving him an aspirin. Here again, the event is clearly confined in the past. Bégin explains this usage with the notion of “partly conditioned outcome”, whereby the impression of a persistent effect of the event at Reference Time (coinciding here with the Speech Time) is due to some “spin-off or accompanying effect of the activity”. For instance, in the first case, the event of talking is presented as though its effects were still active at Reference Time, although the event itself was necessarily completed before. Obviously, this type of English sentences tend to suggest that the event is situated in the recent past, given the well-known restrictions impinging on the English Present Perfect; but this is also true with the Spanish Compound Past, although for different reasons (cf. above). It is perhaps appropriate in these cases to invoque the notion of “aspectual metaphor”: the event is presented as though it were still going on, although the context makes clear that this is not the case. Note, in fact, that the (non progressive) Present Perfect is often ruled out completely in these contexts. For

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

575

The progressive in Romance, as compared with English

575

comparison, consider that Italian and French informants used the Compound Past in (17) and, preferably, the Pluperfect in (18).18 A good indication of the fact that English PROG may have a durative meaning is provided by the fact that in cases like (10–11), the speakers would normally employ PROG. For further support of this, consider now the following example, in which English and Italian are contrasted: (20) a. b.

c.

English Nero was fiddling, when Rome burned. Italian Nerone *stava suonando/ suonava la cetra, mentre Roma Nero was playing/ play:IMPF the lyre while Rome bruciava. burn:IMPF Italian Nerone stava suonando la cetra, quando Roma Nero was:3SG playing the lyre when Rome bruciò. burn:SP:3SG

In (a), the dependent clause may receive two interpretations: a simultaneous reading, in which the fiddling and the burning supposedly have (nearly) the same duration, and an inceptive reading, according to which the beginning of the burning took place at a point in time when Nero was already engaged in his musical activity. These two readings are clearly distinguished in Italian, as shown by (b–c), exhibiting the simultaneous and the inceptive reading respectively, due to the imperfective vs. perfective tense employed in the dependent clause. Crucially, PROG is allowed in (c) and is ruled out in (b).

4. A comprehensive picture From what we saw above, we can draw some conclusions as to the present state of evolution of the Romance and English St-PROG devices. Consider again Table 2 in Section 4.1 in Bertinetto, Ebert & De Groot, this volume, repeated here for convenience, which exhibits what may be called “PROG imperfective drift”. Ibero-Romance St-PROG and English PROG embrace stages (iii) and (iv), for they may occur both in prototypically focalized contexts (cf. (4–5) above), and in durative contexts such as those exemplified in (6) and in Section 3. This shows that, in principle, St-PROG is not restricted to truly imperfective situations, although it shows a striking predilection for such contexts. In fact, as suggested by Squartini (1998), in the early stages of development the component which is most involved

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

576

576

Pier Marco Bertinetto

Table 2. Diachronic development of progressive constructions in Romance (from actionality to aspect) (i)

pure locativity

(ii)

progressivity I

(iii)

progressivity II

(iv)

progressivity III

(v)

pure imperfectivity

stative, durative (ex.: the meaning to be observed in some Latin examples) residually locative, durative, accessible to perfective aspect (ex.: PROG periphrases based on the verb ‘come’, which preserve some kind of deictic orientation) durative, accessible to perfective aspect (ex.: PROG periphrases based on the verb ‘go’) focalized, strictly imperfective (ex.: Modern Italian “stare  Gerund”) loss of the progressive character (ex.: possibly to be observed on some non-standard varieties of Latin American Spanish)

appears to be actionality, rather than aspect proper. This also transpires through the preference for atelic predicates (with the obvious restriction concerning stative verbs) in sentences such as (16) where, in the relevant languages, St-PROG is likely to be used to express a purely durative situation. The extent to which these properties are manifested varies from language to language. As we observed, some varieties of spoken (particularly Latin American) Spanish allow for an extreme behaviour, but on the whole Standard Spanish does not seem to be as liberal as English and Portuguese (or as some colloquial varieties of Italian). Obviously, more research is needed to ascertain this. What is already clear, however, is that Italian St-PROG-GER behaves in a very different way. This periphrasis has steadily reached stage (iv), being now a purely focalized (thus, strictly imperfective) device, although it previously behaved exactly like the English and Ibero-Romance types. On the other hand, Italian St-PROG-INF has retained its ancient status, and is thus strictly comparable to its nearest cognate, namely Portuguese St-PROG-INF, which is the standard PROG device in European Portuguese. As to present day French PROG, its status is fairly similar to that of Italian StPROG-GER, although the story is quite different. The original Old French PROG periphrases were morphologically identical to the ones exhibited by Italian and the Ibero-Romance languages. However, their usage declined in the course of time, so that by the end of the 16th century they had virtually disappeared (Gougenheim 1929; Werner 1980). The “être en train de  INF” periphrasis, which in Table 1 is listed under the label “marginal type”, was registered by the grammarians in its current progressive meaning only at the beginning of the 19th century, replacing the original modal (namely intentional) meaning. According to Gougenheim, in the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

577

The progressive in Romance, as compared with English

577

17th and 18th centuries “être en train de  INF” had more or less the sense of: “être en humeur / en disposition de” ‘to be in the mood to / to be inclined to’. Thus, apparently, this device entered directly at stage (iv), bypassing all previous stages. If this claim is correct, the French case is interesting both in itself, and for what it tells us about the general evolutionary picture. Although it is easy to reconstruct a locative meaning in French PROG, it is possible that this feature did not play the same role as with the other PROG devices we are considering here. It certainly was not conducive to the purely durative stage (iii). What is particularly remarkable is that, to my knowledge, the only other example in European languages of a PROG device exclusively actualising stage (iv), besides Italian St-PROG-GER and French “être en train de  INF”, is the (colloquial) Albanian PROG periphrasis based on the particle po (cf. Section 2.6 of Bertinetto, Ebert and De Groot, this volume). It is thus a fairly rare phenomenon. As to the other, less prominent, Romance languages, it is worth observing that in Galician (Rojo 1974) and Occitan (Schlieben-Lange 1971) PROG is used with both focalized and durative meaning, just like in the Ibero-Romance languages analysed above (Squartini 1998). From the morphological point of view, the Occitan construction (“estre a  INF”) is identical to Italian St-PROG-INF, while in Galician there is a larger variety of forms: “estar  GER”, “estar a  INF”, and even “ser a  INF”, the first being the most frequently employed.

5. Mot-PROG As observed in Section 2, Mot-PROG may only carry a durative meaning. But there is more to say about this device. To start with, it is worth observing that it presents a multifarious morphology, as shown in Table 1. In most languages where it appears (Catalan is a notable exception) there is the choice between ‘go’ and ‘come’ as auxiliaries. This was to be observed already in the early stages of Romance languages. It is perhaps daring to make a general statement about the difference between these two auxiliaries. I shall limit myself to noting that in Italian there is clear evidence that ‘come’ still implies some kind of deictic orientation, possibly at a metaphorical level (Bertinetto 1991).19 What is certainly common to all Romance languages exhibiting Mot-PROG is that the form with ‘come’ is definitely less frequent than the form with ‘go’, which has reached a relatively high degree of grammaticalization, as witnessed by the semantic bleaching of the auxiliary verb.20 Although Mot-PROG was characterized above as a device conveying durativity, it is interesting to observe that in a language like Spanish, where all the morphological possibilities are fully exploited, one may contrast St-PROG and Mot-PROG in order to obtain subtle semantic differences. The following example was suggested by Ignacio Bosque:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

578

578

Pier Marco Bertinetto

(21) a. b.

Spanish Juan estuvo colocando libros de 3 a 5. Juan was placing books from 3 to 5 Juan fue colocando libros de 3 a 5. Juan went placing books from 3 to 5 ‘Juan kept putting back books from 3 to 5.’

While (a) simply depicts a durative situation, (b) adds to it the idea that the event be conceived of as a sequence of identical gestures which follow each other, as though the sentence said something like: “Juan kept putting back one book after the other ” To put it differently, (a) describes a static scenario, while (b) presents a dynamic one. To render this contrast in a sufficiently plastic way, one could legitimately claim that while (a) is a mere case of durativity, (b) is (so to say) an instance of “plurifocalization”, i.e., of a situation in which every instant of the given interval is conceived of as a possible vantage point for the evaluation of the event. In the rest of this section I shall concentrate mainly on the actional restrictions that impinge on Mot-PROG. In fact, this device presents striking properties from this point of view, which differ from language to language. As to Italian, there is a notable preference for accomplishment verbs; however, activities may often be accommodated, by means of the appropriate adverbials. Compare the following examples: (22) a.

b.

c.

d.

Italian *Luca andava ballando la mazurka. Luca go:IMPF:3SG dancing the mazurka ‘Luca was busy dancing the mazurka.’ Luca andava ballando la mazurka con un crescendo Luca go:IMPF:3SG dancing the mazurka with a crescendo di trepidazione e di rapimento. of trepidation and of ecstasy ‘Luca was dancing the mazurka with increasing trepidation and ecstasy.’ *Luca andava gradualmente / a poco a poco ballando la Luca go:IMPF:3SG gradually / little by little dancing the mazurka. mazurka Luca andava gradualmente / a poco a poco scoprendo Luca go:IMPF:3SG gradually / little by little discovering la verità. the truth. ‘Luca was gradually / little by little discovering the truth.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

579

The progressive in Romance, as compared with English

579

In (a) we have an activity verb, and this creates an unacceptable sentence, whereas in (b) the same verb yields a grammatical sentence, due to the presence of a “manner” adverbial of graduality. Note, however, that in (c) the “temporal” adverb of graduality does not guarantee the desired result. The latter type of adverbial is instead perfectly compatible with the achievement verb of (d), where it plays a reinforcing role, for the sentence would be acceptable even without the adverbial. One might wonder how it is possible to employ achievements, as in (d), given that Mot-PROG necessarily requires durative situations. The answer is that whenever this periphrasis may felicitously be applied to an achievement, the verb is inevitably durativized, possibly via an iterative interpretation (but the latter requirement is not even necessary in (d) above). More precisely, when employed with Mot-PROG, achievements take on the reading that is typical of “gradual completion verbs” (such as increase, get fatter etc.; cf. Bertinetto and Squartini 1995); i.e., they suggest that the completion of the event may be reached at the end of a gradual (and, by implicature, slow) process. Although this is the general trend, one significant exception should be pointed out. There is in fact a class of Italian activity verbs which are easily combinable with MotPROG, namely those that I propose to call “inherently intensified verbs”. Among these we may for instance count salterellare ‘hop about’, scribacchiare ‘scribble’ and the like, i.e., verbs expressing the idea of an event which is carried out through actions (in most cases frequently iterated ones), similar in nature to the more neuter action to which they are related (e.g., ‘jump’, ’write’), but specifically connotated regarding the peculiar way in which the action is performed. In this class we may also include verbs such as scrutare ‘look very carefully’, meditare ‘meditate’ etc., which once again involve a particular insistence or intensity in the performance of the event. For instance, ‘meditate’ indicates more than a simple event of thinking, for it means something like ‘think over and over’. Interestingly, these activity verbs may easily accept Mot-PROG without the help of any sort of intensifying adverbial, as in: (23) a.

b.

Italian Lucia andava scribacchiando sul quaderno. Lucia go:IMPF:3SG scribbling on_the exercise book ‘Lucia was scribbling on her exercise book.’ Teresa andava scrutando l’ orizzonte. Teresa go:IMPF:3SG scanning the horizon ‘Teresa was scanning the horizon.’

To sum up, in Italian Mot-PROG shows the following actional restrictions (Bertinetto 1997: ch. 7; cf. also Brianti 1992, Giacalone Ramat 1995a, Squartini 1998). It is preferably combinable with durative telics (although not without exceptions),

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

580

580

Pier Marco Bertinetto

and accepts achievements only insofar as they are contextually turned into “gradual completion verbs”. Activities are normally rejected, unless (a) they belong to the class of inherently intensified verbs, or (b) they are reinforced by means of the appropriate expressions of intensification (like the “manner” adverbials of graduality exemplified in (22b) above). Note, however, that despite the telic orientation of Mot-PROG, there are clear indications that this periphrasis yields the detelicization of telic verbs, as is evidenced by: (24)

Italian a. Filippo risolse il problema in due Filippo solve:SP the problem in two ‘Filippo solved the problem in two days.’ b. *Filippo andò risolvendo il problema Filippo go:SP solving the problem

giorni. days in due giorni. in two days

In (a), the event described fulfills its telic character, due to the aspectual value of the tense employed (a perfective past). The same should happen in (b), for the tense of the auxiliary is the same; yet the sentence is not acceptable. This is clearly an effect of the periphrasis. The in-adverbial strongly requires perfectivity and telicity (Bertinetto and Delfitto, this volume); since perfectivity is guaranteed by the tense employed, the obvious conclusion to be drawn is that the periphrasis as such has a detelicizing effect, this being the only difference between the two examples presented. This may look like a paradoxical situation: apparently, one and the same device is telic-oriented, but induces detelicization. However, we should not confuse the passive actional restrictions that Mot-PROG undergoes in the selection of the appropriate predicates with the active consequences that this periphrasis induces in the actional character of the verb employed. Interestingly, Squartini (1998) shows that in Spanish and Portuguese Mot-PROG exhibits different restrictions. Recall that in these languages the auxiliary ‘go’ can have two translations: ir and andar. The former indicates a goal-directed motion, while the latter preserves by and large its original meaning (present also in Old Italian) of an undirected motion. This accounts for the divergent meaning of the two constructions. In Spanish, “ir  GER”, is preferably used with telic verbs and, when used with activities or statives, suggests an inceptive reading, while “andar  GER” prefers activities and may easily be used with statives (Luis todavía anda queriendo comprar el coche ‘Luis still wants to buy the car’). As often claimed in grammatical descriptions, the latter form of this periphrasis is often felt a possible alternative to St-PROG, to which it adds something like a hyperbolic meaning, not unlike the semantic nuance added by siempre in (11). The only contexts in which the form with ir and the form with andar appear to be basically interchangeable are those expressing

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

581

The progressive in Romance, as compared with English

581

iterativity (i.e., with achievements reinterpreted iteratively). As to Portuguese, the situation is essentially like in Spanish, the only relevant difference being that “andar  GER” accepts far more accomplishments and (durativized) achievements than its Spanish counterpart. If one takes this as the decisive factor, one may conceivably claim that Portuguese “andar  GER” is more advanced in the process of grammaticalization than any of its direct competitors because it undergoes virtually no actional restrictions.21 As to the remaining Romance languages, it should be observed that Mot-PROG was originally quite widespread even in languages like French, Catalan and Occitan, where it has now become extremely rare. Although “aller  GER” was still used by French novelists of the last century (cf. the following example, taken from Flaubert’s Madame Bovary: elle alla [ ] montant et s’en détachant, lit.: ‘she rose-Mot-PROG and detached-Mot-PROG herself’), and although it may occasionally be met even in contemporary journalistic prose, the decline of this construction began as early as the 17th century. Significantly, in each of the three languages mentioned above the evolution seems to have been the same as in Italian. From the early stage in which activity verbs were frequently employed, the periphrasis has evolved into a condition in which telic predicates tend to be preferred. PROGQ was not specifically devised to investigate all possible semantic subtleties of Mot-PROG. Yet this construction appeared in a number of instances. Specifically, it was used by one Portuguese informant in PROGQ:18 (anda a escrever), by one Spanish informant in PROGQ:51 (iba tomando notas) and PROGQ:54 (iba mejorando), by one Italian (from Sardinia) in PROGQ:52 (va dimenticando), and by the Catalan informant in PROGQ:55 (anava cobrint). It is notable that Mot-PROG made its appearance in two sentences containing “temporal” adverbials of graduality (PROGQ:54–55). In fact, although quite compatible with St-PROG, these adverbials show a high statistical co-occurrence with Mot-PROG (Squartini 1990; Bertinetto, to appear).

6. Some properties of Romance and English PROG In the remainder of this chapter, I shall review the most notable morphological, semantic and syntactic restrictions impinging on PROG in Romance and English. Except when explicitly stated, in this section I shall not consider Mot-PROG. Of course, this survey is by no means intended to be exhaustive. Note that in Romanian the usage of PROG is very limited; thus, I shall quote this language only when relevant.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

582

582

Pier Marco Bertinetto

6.1. Morphological restrictions One morphological restriction which is often cited in the literature is that concerning the Imperative mood (Hirtle 1967; Scheffer 1975). The relevant input comes here from PROGQ:73 (/For goodness sake,/ WORK when the boss comes back!). Although English Imperative Progressive is reported in the literature, our informant did not use it, thus confirming that this is but a marginal possibility. The Imperative Progressive was however employed by the Catalan informant (estigues treballant), and by two Portuguese ones (você esteja a trabalhar (SG) / estejam a trabalhar (PL)). Surprisingly enough, two Romanian subjects offered here what seems to be a genuine PROG construction (s˘a fii în curs de a lucra lit. ‘be-Subj. in the course of work-INF’). The fact that our Romanian informants employ PROG devices only in this case and (to some extent) in relation to future-time reference (cf. Section 6.3.3), both admittedly non-prototypical circumstances, demonstrates the very low degree of grammaticalization reached by these constructions.22 As to passive PROG (cf. PROGQ:75), the only two languages which appear to use it rather freely are Portuguese (está a ser servido) and English (is being served). The other informants produced the passive form of the Simple Present or, in order to preserve PROG, transformed the passive sentence into an active one (by means of an impersonal construction in one case, and by means of an unspecified 3PL. subject in another). The latter solutions were adopted by the two Spanish informants. This is notable because passive PROG is reported to appear in Spanish. Squartini (1998) reports the following example from a linguistic textbook: El corpus de los diccionarios españoles está siendo publicado ‘The corpus of Spanish dictionaries is being published’.23 PROG with the Infinitive is also considered to be fairly marginal, although it is possible in English. Indeed, our English informant used it in PROGQ:79–80 (Tom must FEED the animals /I guess/; Ann should TEACH now /I guess/), and this solution was also adopted by all the French and Portuguese speakers, as well as by one Italian informant (from Central Italy). The Catalan informant employed it only in PROGQ:80. As to Spanish, example (9) attests that this form is allowed in some colloquial varieties. The reason why the speakers tend to avoid PROG with the Infinitive must be due to the relative clumsiness of the construction, rather than to any morphological restriction proper.

6.1.1. Syntactic restrictions Negation has no effect on the use of PROG in the languages considered (cf. PROGQ: 76–78). The situation looks more varied with causative constructions (PROGQ:19– 20), for the various languages seem to differ as to the readiness with which they

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

583

The progressive in Romance, as compared with English

583

allow PROG in these cases. In fact, our Portuguese and Romanian informants did not use PROG in any of the quoted sentences.24 One of the major syntactic functions of PROG is of course backgrounding. However, when a series of events considered from a progressive viewpoint are juxtaposed in a text, in order to compose a complex situation in which different activities occur simultaneously (as in PROGQ:62),25 not all languages appear to be equally ready to repeat PROG in adjacent sentences. The reason for this may be purely stylistic: it may have to do with the relative heaviness of the construction, which induces the speakers to make use of possible alternatives, whenever they exist. Thus, most Romance languages tend to avoid the repeated use of PROG, resorting to simple tenses or alternating the two options. Obviously, since this possibility is not available in English, it is no wonder that a sequence of several PROG is not uncommon in this language. But it is notable that two Portuguese informants, and one Italian (from Sardinia), offered this as a possible option. This is further evidence that Portuguese, as well as some varieties of other major Romance languages, tends to expand the usage of PROG considerably. This is also confirmed by the use of PROG in two adjacent coordinated clauses (as in PROGQ:70). Once more, English has no choice but to employ PROG in both clauses, while Portuguese proves to be the most flexible among the Romance languages. The remaining languages, in fact, present PROG in only one clause, and a simple tense in the others (a solution also proposed as an alternative by the Portuguese informants). However, this is not an inviolable constraint: one of our Sardinian subjects adopted the same options as the Portuguese informants; and cf. example (10) in Section 3, showing that some Spanish colloquial varieties behave in the same way.

6.2. Semantic restrictions 6.2.1. Actional restrictions In section 5 above, the actional restrictions that impinge on Mot-PROG were described. I will now outline the situation of St-PROG (cf. also fn. 10). The most obvious restriction concerns stative verbs. Indeed, the unavailability of PROG is often considered to be a diagnostic criterion for the individuation of stative verbs in languages like English or Romance. However, this problem is rather complex, for in quite a number of cases the same lexical entry may or may not be compatible with PROG, depending on interpretation. When this happens, there are grounds to believe that this lexical entry is in fact ambiguous between a stative and a non-stative meaning (Bertinetto 1994). Compare:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

584

584

Pier Marco Bertinetto

(25) a. b.

John resembles his father. John is resembling his father more and more.

(26) a. b.

The mountains surround the lake. The army was surrounding the enemies.

While the (a) sentences depict a static situation (and indeed no informant used PROG in PROGQ:41, repeated here as (26a), the (b) sentences suggest a process of transformation, to the effect that the predicate of, e.g., (25a) may be paraphrased by means of expressions such as: ‘is becoming more and more similar to’. Not all languages are equally prone to exhibiting this kind of duplicity in their lexical storage. This is particularly evident with copular verbs, such as be silly, be kind, be rude, be clever etc. (but also non copular ones, such as have a headache), which in English may instantiate both individual-level and stage-level predicates, to repeat Carlson’s (1978) terminology. This is not possible in Italian, French and Catalan. However, this is not a distinctive feature of English. In PROGQ:43 (You BE RUDE this evening), not only the English but also the Portuguese informants presented PROG as an option; and in PROGQ:42 (you BE very KIND, now!) the Spanish informants also presented this solution in addition to the ones already quoted. In fact, Squartini (1998) cites a number of Spanish examples of this type. Brazilian Portuguese is claimed to be even more flexible from this point of view, for even non-copular stative verbs like saber ‘know’ or poder ‘can’ may be treated in this way, suggesting a permanent or a temporary situation, respectively (Schmitz 1982). In fact, it was precisely the verb KNOW that elicited PROG as a possible alternative, with one Portuguese informant in PROGQ:39 (/Now, unexpectedly/ Peter KNOW the answer). The verb LIKE in PROGQ:40 (/Now, unexpectedly/Tess LIKE the music) yielded this option not only with all the Portuguese informants, but with one Sardinian subject and (with a question mark added) our English Subject. Thus, in general, English does not seem to be the most liberal language from this point of view.26 English, on the other hand, is unique among the languages considered here, for it exploits this possibility with a small set of (mostly) postural verbs, like stand, lie, sit, wear, and a few others. Indeed, in PROGQ:58–59 (Ann STAND in the doorway /right now/; The statue STAND in the garden /for the summer/), our English informant is the only one who employs PROG. These examples are especially relevant, because with these verbs it is not the case that PROG instantiates a non-stative meaning. Rather, it suggests the idea of the temporary validity of the (inherently stative) situation. In other words, in this particular case English does not exploit this grammatical device in order to destativize the predicate, but only to impose on it a temporal limitation.27 Connected with stativity is the notion of “non-agentivity”, of which the last example was an instance. Among the test sentences, PROGQ:36 (The sun SHINE)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

585

The progressive in Romance, as compared with English

585

and PROGQ:37 (The water BOIL) were less effective, possibly for idiosyncratic reasons, in eliciting PROG as compared to PROGQ:35 (He DREAM of his girlfriend) and PROGQ:38 (The apples ROT on the tree). As to the various languages, Portuguese and English presented PROG in each sentence, while some Italian, Spanish and Catalan informants rejected it in a few cases, and only one of the French informants used it in sentences PROGQ:35 and PROGQ:38. Thus, although the single languages differ, non-agentivity as such is no hindrance to the use of PROG. This also applies to non-intentionality (PROGQ: 33–34),28 which did not prevent PROG in any of the relevant languages, although again the only languages where this was the sole solution proposed were Portuguese and English. Non-durative verbs like THROW, REACH, GIVE (PROGQ:30–32), on the other hand, seem to create serious difficulties for French speakers compared to the others, who used it to a greater (Portuguese and English) or lesser extent. This situation is also confirmed by the usage of the verbs LEAVE and DIE in PROGQ:56–57, two sentences devised with the purpose of checking a specific effect of PROG as applied to non-durative verbs, i.e., the possible emergence of an imminential meaning.29 Further inquiries with native speakers made it clear that the restriction concerning nondurative verbs is not absolute in French, but rather corresponds to a strong tendency. One special case of non-durative verbs is represented by the so-called phasal verbs like begin and FINISH etc. (PROGQ:23–26). Here again, only French informants avoid PROG.30 Finally, it should be noted that in principle no restriction is to be observed concerning verbs of motion like GO OUT and FLY (PROGQ:21–22), which even in French may be associated with PROG in the appropriate context. This is also true of the verbs GO and COME that provide the auxiliaries of Mot-PROG. On the other hand, postural verbs like SIT and HANG in PROGQ:28–29 never elicit PROG in Romance (unlike English), nor do they constitute, as in some Germanic languages, the basis for a PROG construction (cf. Ebert, this volume). 6.2.2. More on aspectual restrictions In Sections 4 and 5, I observed that St-PROG and Mot-PROG are in principle compatible with both the perfective and the imperfective aspect. However, the situations of the two constructions are not completely identical. The latter appears to be intrinsically available to any kind of aspectual value, whereas the former is liable to be caught in what I named above “PROG imperfective drift”, which indeed explains the development of the purely focalized meaning of Italian St-PROG-GER, as well as the transformation of PROG into a general purpose imperfective form in quite a few languages (cf. Section 4.1 of Bertinetto, Ebert and De Groot, this volume). It is appropriate to add here a few more observations about the habitual aspect and its combinability with PROG devices. As we saw in (1–2) above, in their very

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

586

586

Pier Marco Bertinetto

early stages these periphrases had easy access to this aspectual value. However, this possibility was completely lost in the course of time with Italian St-PROG-GER, and severely constrained in the remaining Romance languages and in English. As a matter of fact, none of our informants used PROG in PROGQ:2 (A: What does Ann do every Saturday morning? B: She CLEAN the house) and PROGQ:4 (Last year we usually CLEAN the house on Saturdays). All speakers used a simple tense (in PROGQ:4, our Romance informants used the Imperfect or, as one Italian and one Spanish speaker did, a habitual periphrasis with the Imperfect).31 Nevertheless, the co-existence of PROG with the habitual interpretation can frequently be observed in any of the languages considered, provided correlative connectives such as whenever appear: (27)

Eng: Fr: It: Spn:

Whenever I checked, he was working. Chaque fois que j’ai contrôlé, il était en train de travailler. Ogni volta che ho controllato, stava lavorando. Cada vez que yo controlé, él estaba trabajando.

These sentences do not constitute any problem whatsoever, for they even allow a focalized interpretation of PROG (in contrast to example (12) above). The habitual event is viewed as occurring at some particular points in time, which repeat themselves more or less regularly. But if we depart from this kind of correlative contexts, we may find that not all languages are equally ready to accept PROG in truly habitual situations, where the recurring event is not focalized by means of an incidential event which isolates single instants as evaluation points. Indeed, our English informant was the only one to use PROG in PROGQ:63 (At that time, he GO to dance every Saturday). And this is substantially confirmed by the following example (or similar ones, inspired by Hirtle 1967 and Leech 1971), despite the presence of an incidential clause:32 (28)

Eng: Peter is often smoking a cigar when he comes in. Cat: ?En Pere está sovint fumant un cigar quan arriba. Fr: *Pierre est souvent en train de fumer un cigare quand il vient. It: *Pietro sta spesso fumando un sigaro quando viene. Prt: O Pedro está frequentemente a fumar um charuto quando chega. Spn:?Pedro frecuentemente está fumando un cigarro cuando llega.

One type of contexts rather close to habituality, although presenting peculiar characters of its own, is that exemplified in (11) (cf. also fn. 15). The hyperbolic meaning of such sentences suggests that the event tends to repeat itself with rather unusual frequency. Besides English and Portuguese, whose behaviour is predictable, Catalan and Spanish also have free access to this usage (at least in the colloquial variety), while in French and Standard Italian this possibility is ruled out altogether, due to the focalization requirement that characterizes PROG in these languages:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

587

The progressive in Romance, as compared with English

(29)

587

Cat: En Pere está sempre fumant un cigar. Spn: Pedro está siempre fumando un cigarro. ‘Peter is always smoking a cigar.’

However, in colloquial Italian one may hear sentences like the following, which do not seem very far from the hyperbolic contexts illustrated above:33 (30)

Italian E’ un mese che me lo sta dicendo. is one month that me:DAT it:ACC is saying ‘He’s been telling me for a month now.’

Even more common are sentences like: Sta piovendo molto, quest’anno ‘It is raining a lot, this year’, which express again a meaning of insistence, or (in Blansitt’s 1975 terms) a “generic” meaning close to habituality. Also somewhat related to habituality are the “interpretative” uses of PROG, which are not uncommon in English. What is typical of these sentences is that a given action performed by somebody runs parallel to an equivalent action, to which a positive or negative value (possibly a merely explicative value) is assigned (König 1995). Consider PROGQ:64: If you insist in calling me Fred, you INTRUDE in my private life. Given the correlative structure of these sentences, they bear a resemblance to examples such as (27). PROG is consistently used here by the English, Portuguese and Spanish informants, while avoided by the others. However, in the related sentence PROGQ:65 (As soon as you start asking what is the use of education, you ABANDON the basic assumptions of any true culture), two of the Italian informants also offered this alternative, together with the preceding subjects. Once more, we find English and Portuguese in the lead, with French far behind.34 6.2.3. Temporal restrictions There are no restrictions concerning temporal reference in the languages considered. PROG may occur with past, present or future localization of the event. In particular, PROGQ:83 (If you come at 8 o’ clock, I still COOK) was specifically devised in order to test the availability of PROG with the Future tense, which in some languages is claimed to be subject to restrictions. With the exception of French, Future PROG (or something close to it) was used, or at least offered as an alternative, by virtually all informants. One surprising finding is represented here by two Romanian infors˘a m˘a g˘ase¸sti înc˘a mants, who proposed the following quasi-PROG construction: g˘atind ‘you will find me still cooking’. These data suggest that, except for French and to some extent Romanian (two languages known for the limited exploitation of PROG in general), there is no constraint at work, provided the context is felicitous

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

588

588

Pier Marco Bertinetto

enough. The observation concerning the restricted use of PROG with future temporal reference has thus more to do with pragmatics, namely with the relative rarity of such contexts in actual communication, than with morphology or semantics.35 One notable fact is the use of PROG in contexts corresponding to a sort of extended present (but similar examples may be built with past temporal reference). PROGQ:61 is an example (The boss TYPE his own letters, while the secretary is ill). The only informants who use PROG in this case are, once more, the English and Portuguese ones. This possibility seems to be totally excluded in French, while in the remaining languages, choices may vary depending on register. In the standard varieties, speakers tend to avoid this usage. Perhaps the most striking feature of English PROG, as compared to all the remaining languages considered here, is the possibility of expressing future-time reference, as in sentences like: Ann is leaving tomorrow (cf. PROGQ:66; cf. also PROGQ:67– 69).36 Note that in these cases the future-time reference is conveyed by the Present Progressive, rather than by the Future Progressive, as in sentences like: Tomorrow I’ll be leaving (cf. also the comment to sentence (3a) above). This property of PROG, although rather rare, may be found in other European (as well as non-European) languages. This is notably the case in Icelandic (Ebert, this volume) and to some extent also in Finnish (Tommola, this volume), but also in Judeo-Spanish, and in some colloquial varieties of Latin American Spanish (Squartini 1998), as well as in some colloquial varieties of Southern Italian (like in Naples; cf. Gliela sto passando subito ‘I am going to pass her [on the phone] right now’). It is not easy to understand how this use may have arisen. The hypothesis that most obviously comes to mind is that it is somehow related to the imminential meaning often expressed by achievement predicates under PROG (cf. Section 6.3.1), which conveys something close to a futural sense. The data gathered from languages other than English are too scanty to allow us to put forth a hypothesis as to the actual meaning of PROG with future-time reference. As to English, Haegeman (1981) suggests, among others, the following facts. First, the Simple Future differs from both the Present and the Present Progressive because it is not oriented towards the speech time: (31) a. ??I will already meet John for lunch and Ann for dinner; I cannot have any other appointment. b. I already meet John for lunch and Ann for dinner; I cannot have any other appointment. c. I am already meeting John for lunch and Ann for dinner; I cannot have any other appointment. Here, the relevance of the speech time is ensured by the adverb already, and the contrast of (a) with (b–c) is quite evident. Moreover, the Simple Future is characterized

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

589

The progressive in Romance, as compared with English

589

by what may be called “subjective speaker-commitment”, while the two remaining tenses are rather characterized by some kind of “objective factual commitment”. This confirms again the same distribution of grammaticality judgements among the three tenses: (32) a. ??The Queen, who will open Parliament tomorrow, may be detained at the airport. In this case, Prince Charles will take her place. b. The Queen, who opens Parliament tomorrow, may be detained at the airport. In this case, Prince Charles will take her place. c. The Queen, who is opening Parliament tomorrow, may be detained at the airport. In this case, Prince Charles will take her place. The event designated in (32) may be regarded as fairly objective, inasmuch as it is a scheduled one. The two Present tenses seem to be highly preferable in these contexts. On the other hand, in order to have a perfectly felicitous use of the Present Progressive, as opposed to the Simple Present, the scheduled event should best be conceived of as fairly exceptional, rather than predictable and routine: (33) a. Who is being Santa Claus at the party tomorrow? b. ?Who is being captain of the team tomorrow? (34) a. ?The train is leaving tomorrow at 5. b. The train leaves tomorrow at 5. The difference in (33a–b) is due to the fact that the presence of a Santa Claus must be considered as a fairly uncommon event, while there always ought to be a captain of a team. As to (34a–b), these sentences should be regarded as statements uttered while consulting the timetable, rather than statements concerning a non-customary event. If the train were to leave at 5 for some exceptional reason, then the Present Progressive would be perfectly acceptable.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

590

590

Pier Marco Bertinetto

Appendix. Responses to the Progressive Questionnaire by Romance and English informants Legenda: A slash separates alternative responses provided by the informant. A blank indicates a missing or irrelevant response. Sometimes, the informant provided an irrelevant response in addition to a relevant one; in these cases, only the latter was reported (e.g., in sentence 53 a number of informants provided a perfectal tense, which obviously is not the intended answer, although it sounds plausible in the given context).

 Inf Mot % ? *

0/  adv (-caus) cont hab imm imps main mod modep N rec Sard. sb seguir

standard PROG device Infinitival State-PROG construction (cf. Table 1) Motion-PROG construction (cf. Table 1) PROG device used for some, not all verbs in the sentence (mostly depending on actionality) the informant expressed some reservations on her/his own response possibly a (not yet grammaticalized) PROG device no PROG device was used sentence not presented to the informant an appropriate temporal adverb was added the intended causative construction was avoided by the informant continuous periphrasis (It. “non far altro che  Infinitive”, Fr. “ne pas s’arreter de / ne pas cesser de  Infinitive”) habitual periphrasis imminential periphrasis impersonal construction PROG device in main clause only modal verb construction modal verb construction with epistemic meaning nominal construction ‘recent past’ construction (Fr. “venir de  Infinitive”) Sardinian informant subjunctive with iussive meaning continuative periphrasis “seguir  Gerund”

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

591

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Informant

Catalan 1

French 1

2

0/ 0/ 0/ 0/ 0/

0/ 0/ / 0N 0/ 0/

Italian 1 2

3 Sard.

4 Sard.

/0/ 0/ /0/



Portuguese 1 2

3

 0/ 

 0/ 

 0/ 

            

            

         

Romanian 1 2

3

Spanish 1 2

English 1

0/ 0/ 0/

 0/ /0/

 0/ /seguir

 0/ 

 /0/             /0/    

             ( (      

                            /0/  

Sentence

 /0/ 0/ 0/

/0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ ( /0/ /0/ /0/  /0/  0/ 0/ 0/

/0/ /0/ /0/ /0/ /0/ 

caus)

0/ idiom 0/

 

0/ idiom 0/

            0/  

0/ adv 0/

0/ 0/

0/ adv 0/ rec 0/  adv 0/ 0/ 0/

0/

0/ imm

0/ 0/

  /0/ 0/

 0/ 0/

0/ 0/

/0/ /0/ /0/ /0/ /0/ /0/ /0/

0/ 0/

/0/ /0/ /0/ /0/ /0/ /0/ 0/

0/ 0/

/0/ /0/ /0/              /0/ adv  

0/ 0/ 0/

/0/    /0/

0/ 0/ 0/ hab 0/

               /0/ N      0/ 0/ 0/

/0/   /0/

0/ 0/

 0/ N    

0/ 0/ 0/

  /0/    

0/ 0/

0/

 0/ N    

0/ 0/ 0/

/0/  /0/    

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/

 (      

0/ 0/ 0/ 0/

 /0/    

caus)

0/ 0/ 0/ 0/ 0/ / 0/ imm 0/ / 0/ N 0/ 0/ / 0/ imm 0/ 0/ / 0/ imm 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ N

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ / 0/ imm 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

/ 0/ hab 0/ 0/

0/ 0/ 0/ 0/

/0/ /0/ /0/ /0/ /0/

0/ 0/

0/ 0/ 0/ 0/

/0/

0/

  

caus) caus)

0/ 0/

591

0/

 0/  0/    0/  0/  

/0/ /0/ 0/ /0/ /0/ 0/ /0/ hab 0/ 0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/ /0/  /0/ /0/ /0/ /0/ /0/ /0/ /0/  /0/  /0/  /0/  /0/ /0/ /0/ /0/ /0/ /0/ /0(/ caus)  /0/ adv  /0/ adv /0/ /0/  /0/  /0/  /0/ 0/ /0/ 0/

The progressive in Romance, as compared with English

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

592

French 1

/0/ 



2

Italian 1

2

3 Sard.

4 Sard.

/0/ /0/

/0/ /0/

/0/  0/ /0/

 

Portuguese 1 2

3

Romanian 1 2 3

Spanish 1

2

English 1

Sentence 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72

0/ 0/ 0/ 0/ 0/

      0/ /0/ 0/ /0/  Mot 0/  0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

/0/ % /0/ % /0/

0/

0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ cont 0/ 0/ 0/

0/ 0/ 0/ cont 0/ cont 0/ 0/

0/ 0/ 0/ 0/ idiom 0/ 0/ /0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ idiom 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ idiom 0/ 0/



0/ 0/ 0/ 0/ 0/

/0/ /0/  /0/

0/ / 0/ cont 0/ 0/ 0/ 0/ /0/ /0/ /0/ /0/



0/ 0/

/0/

0/ 0/ 0/ 0/

/0/

0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ 0/ 0/

/0/ /0/ /0/ /0/ 0/ 0/ 0/ 0/ 0/

/0/ /0/ /0/ /0/ /0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

/0/ 0/ (?) 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/

     Inf//0 Inf//0 / Inf /0/ /Inf /0/ Mot/0/      0/

 0/ /0/

0/ 0/ 0/ 0/ 0/ 0/ 0/

main/0/ 0/ Inf%//0

0/ 0/ 0/ 0/ 0/

   

0/ 0/



0/ 0/

  

0/ imm 0/



0/ 0/ 0/

0/



0/ 0/ 0/ 0/ 0/

  0/ /0/ 0/ /0/ /0/

  0/ /0/ 0/ /0/ /0/

/0/  /0/ /0/ /0/  /0/ /0/ /0/ imm /0/ imm

/0/   /0/ /0/  /0/ /0/ /0/ imm /0/ imm

/0/ /0/ /0/ 0/  /0/

/0/ /0/ /0/ 0/  /0/

   

0/ 0/

0/ 0/ 0/ 0/

/0/ 0/ % /0/

   

0/ 0/

0/ 0/ 0/ 0/

/0/ 0/ % /0/

  /0/ /0/ 0/ /0/ /0/    /0/    0/   /0/    0/ /0/ /0/ 0/ 0/

 

0/ 0/ 0/ 0/ 0/ 0/

% /0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

/0/ 0/ 0/

/0/   0/  0/ 0/ 0/ 0/ 0/ 0/ 0/



0/ 0/ 0/

/0/ 0/ 0/ 0/

  0/ 0/ 0/ 0/ 0/ 0/

% /0/

  0/ 0/ 0/

/0/ 0/   0/     Mot 0/  Mot    0/ 0/ 0/ 0/ 0/ 0/

 

0/ 0/ 0/ 0/ 0/ 0/ 0/

  0/ ? //0 0/ /0/ /0/     /0/ /0/ /0/ /0/ /0/       /0/    /0/ /0/ /0/ /0/ /0/ /0/ /0/  % /0/ % /0/

Pier Marco Bertinetto

Catalan 1

592

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Informant

593

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Informant

Catalan 1

French 1

2

Italian 1 2

3 Sard.

4 Sard.

0/ 0/ 0/

0/ 0/ 0/

0/ 0/ 0/

Portuguese 1 2

3

Romanian 1

2

3

Spanish 1

2

0/ 0/

0/ 0/

English 1

Sentence

 0/ imps//0  /0/ 0/

   

/ 0mod 0/ 0/

/ 0N 0/ 0/ / N 0/ 0/

 

 

/0/

0/ cont 0/ 0/

0/ 0/ / 0N

/0/ /0/ /0/ 0/ 0/ 0/ 0/



0/ 0/ 0/

/0/  /0/ /0/ 0/ 0/ /0/ 0/ /0/ 0/ 0/ Inf//0 0/ 0/  

 /0/ 0/ 0/ 0/ 0/



0/ 0/

   0/    /0/ 

/0/ 0/ / /0N   0/     

sb//0mod 0/    0/    0/ 

* //0

0/ 0/ 0/ 0/ 0/ / 0modep / 0/ / 0modep / 0/ 0/ 0/ */0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ 0/

*

* //0

0/ 0/ 0/ 0/ 0/ / 0modep 0// / 0modep 0// 0/ 0/ 0/

imps /0/ /0/

0/

0/ 0/

0/ 0/

 

0/

imps 

0/



0/

         

The progressive in Romance, as compared with English

73 74 75 76 77 78 79 80 81 82 83

593

594

594

Pier Marco Bertinetto

Notes

1. I wish to thank for useful comments the members of EUROTYP Group 6 (in particular Karen Ebert) and Mario Squartini, whose thesis (Squartini 1998) provided the basis for many of the observations in Section 3.1. 2. I would like to thank the people who acted as informants (following the order given in the text): Maria-Rosa Lloret, Greville Corbett, Nicole Rivière, Pierre Jalenques, myself (with no special gratitude), Mario Squartini, Cristina Lavinio, Sabrina Salis, Maria Helena Abreu, Rita Benamor, Fatima Oliveira, Lauren¸tia Dasc˘alu Jinga, Sorin Stati, Laura Vasiliu, Ignacio Bosque, Manuel Carrera Díaz. The generous gift of their time is gratefully acknowledged. 3. The question marks in Table 1 are motivated by the fact that marginal types may also exist in the languages for which I have not collected enough information in this respect. Indeed, even French has further PROG devices, such as the periphrasis “être après  INF” (as in il est après lire ‘he is reading’, lit.: he is after read), which is mainly used in the Loire Valley and in parts of the South of France. A limited geographical distribution has also to be ascribed to the morphologically almost equivalent Italian marginal type listed in the Table, which can be found only in some areas of the North-East. Cf. also fn. 4. 4. As explained in Squartini & Bertinetto, fn. 7, I use the term “vernacular” to refer to what in the Italian linguistic tradition is called “dialetto”, i.e., a local language which developed independent of the national language, although (in most cases) connected to it from a diasystematic point of view. Very little attention has so far been given to tense-aspect problems in the study of Italian vernaculars (see however Cordin 1997). Rohlfs devotes only three sections of his monograph (§§ 739–741) to this. With specific regard to PROG, he points out a number of morphological variants. In the North-East (Verona, Trieste) the form “stare  INF” is to be found. In the South-East (Salento) we find the form što ffazz ‘I am doing’, repeating the type “STARE AC  VERB”, where the auxiliary and the main verb agree in tense and person (in most cases the descendant of the Latin conjunction ac is merely inferable through the doubling of the following consonant). In some localities, the auxiliary presents the undeclinable form šta, which in a restricted area has even been grammaticalized as an obligatory marker of imperfectivity. In one part of the Centre (Abruzzo), PROG may manifest itself as “tenere a (‘keep at’)  INF”, with possible absorption of the preposition and concomitant doubling of the following consonant. A somewhat similar periphrasis (“TENERE  Past Participle”) was used in Piedmontese vernacular until the end of the 19th century (Ricca 1998). Some remarks concerning Sardinian and the Italian variety of Sardinian are provided by Loi Corvetto (1983), while Amenta (1994–95) is mostly concerned with the variety spoken in Palermo. Both authors attest that these varieties are much more liberal than Standard Italian with respect to the usage of PROG. 5. Cf. also Škerlj (1926), Spaulding (1926), Gougenheim (1929), Lyer (1934), Werner (1980), Dietrich (1985). As to general information concerning verbal periphrases in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

595

The progressive in Romance, as compared with English

595

Romance languages, see the bibliography at the end of this chapter, and the references quoted therein (in particular Squartini 1998). 6. The first construction survives only in Sardinian, where it is extremely frequent and can also occur with states and habituals (Loi Corvetto 1982): (i)

Sardinian esti kreendi de essi no sciu kini is believing of be not know:1SG who ‘s/he is believing her/himself I don’t know who’

(ii)

Sardinian su mesi passau fiara proendi d’ ognia dominigu the month past was raining of every Sunday ‘last month it rained every Sunday’

As pointed out by Jones (1993), Sardinian admits not only the auxiliary essere, but also istare; this fills the gap of the third type, for which no Old Italian examples are reported in my sources. The two Sardinian constructions show different degrees of grammaticalization, for only the former is compatible with states. As in Ibero-Romance languages, and unlike Italian (see below), in Sardinian PROG can be combined with perfective morphology: (iii)

7. 8. 9.

10.

So istatu travallande am been working ‘I have been working’

As I shall also point out below, the extensive usage of PROG in the vernacular percolates into regional Italian spoken by Sardinians, so that they use the Italian PROG construction in contexts which would not be admitted in the standard language. For the history of English PROG, cf. at least Mossé (1938), Visser (1969/1973), Kisbye (1971) and Scheffer (1975), Vezzosi (1996). See Bertinetto, Ebert & De Groot, this volume, fn. 9 for an explanation of this term. Note however that in Ibero-Romance languages and in Italian PROG is often preferred over simple tenses with achievement verbs (cf. fn. 11). Thus, this statement should be taken with caution. In PROGQ:3, one Spanish informant provided as an alternative, the continuative periphrasis “seguir  GER” ‘keep V-ing’ (seguía trabajando). This is not to be considered a PROG device in the proper sense, for it expresses the idea of continuation, but it certainly bears some resemblance to Mot-PROG. One of the Spanish informants also provided, as an alternative, an instance of Mot-PROG: iba anotando. This is remarkable, because the specific function of this periphrasis is precisely that of expressing durativity; cf. Section 5. Obviously, PROGQ:51 could also be rendered by means of a purely perfective tense; indeed, some of our informants provided this alternative. Interestingly, the Portuguese subjects also yielded the Simple Past Progressive esteve a tomar notas; as to the use of PROG with perfective tenses, cf. Section 3.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

596

596

Pier Marco Bertinetto

11. The last observation corroborates previous data found by Squartini (1990), who also gathered data concerning the non-literary language. On the other hand, it is worth noting that, whenever a PROG device has reached a sufficient degree of grammaticalization, it is precisely the presence of achievement verbs that makes the appearance of the relevant construction almost compulsory, even in languages which have the alternative of non-periphrastic imperfective tenses. The reason for this is that the use of an overt morphological device is strong enough to force a progressive reading with predicates which are not easily amenable to such an interpretation. In fact, this is to be observed even in Estonian (Metslang 1995), a language where PROG is only weakly grammaticalized. As to the contribution of PROG with achievements in Italian, as opposed to non-periphrastic tenses, cf. Delfitto & Bertinetto (1995). 12. Indeed, St-PROG-INF is employed in PROGQ, as a possible alternative, by one Italian informant speaking the Sardinian variety. Cf. fn. 13 and 17 for further comments. 13. Actually, the possibility of a deontic use of modal verbs with PROG is not entirely ruled out even in Italian. Mario Squartini has drawn my attention to the following colloquial sentence: (i)

Italian Per aver diritto al ritardo del servizio militare, devi star for have right to delay of_the service military must-2SG be facendo il dottorato. doing the doctorate ‘In order to be entitled to postpone military service, you must be working on your PhD.’

An even stronger example is quoted by Amenta (1994–95) from the Lessico dell’Italiano Parlato, a corpus of Contemporary Spoken Italian (this sentence was presumably uttered by a Southern speaker): Non stanno potendo piú cambiare nulla ‘They cannot change anything anymore’ (lit.: ‘they can-PROG not’). However, these should be regarded as fairly exceptional cases, as evidenced by the following contrast between St-PROG-GER and St-PROG-INF: (ii)

Italian *Maria deve sempre stare lamentandosi. Maria must:3SG always be complain:GER:RFL Maria deve sempre stare a lamentarsi. Maria must:3SG always be at complaining:INF:RFL ‘Maria is always complaining about something.’

This sentence asserts that what is typical of the intended person is her readiness to complain, as though she had to obey some sort of inner compulsion. 14. In PROGQ, there was one sentence which tested this particular interpretation. Only the Portuguese and English subjects used PROG in such a sentence, proving that these languages are the most liberal in this respect (as in many other respects concerning the use of PROG):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

597

The progressive in Romance, as compared with English (i)

15.

16.

17.

18.

19.

597

PROGQ:52: He continually FORGET people’s names. Port.: Está sempre a esquecer [Prs-PROG] os nomes das pessoas.

However, similar sentences are quite possible in Spanish (witness example (29)), and are not impossible even in colloquial Italian, as shown by the following sentence reported by Amenta (1994–95) from a corpus of Contemporary Spoken Italian (the sentence was presumably uttered by a Southern speaker): mi sta chiedendo sempre cose sugli autori ‘he is always asking things about the authors’. Periphrases of this sort were indeed used by some Italian and French informants in PROGQ. Namely, we find them with one Italian subject in PROGQ:49, while both French informants used “ne pas cesser / s’arrêter de  INF” (lit.: ‘do not stop doing’) in the same sentence, and one of them employed it in PROGQ:50 and in PROGQ:81. Obviously, all these constructions insist on the duration of the event. Although they cannot be considered PROG devices in the proper sense, they bear some relationship to durative PROG devices. Intriguingly, the Catalan informant is the only one who used PROG in PROGQ:71 (/What did Martin do yesterday evening?/ He STUDY, he READ the paper, he EAT, and then he GO to bed). This pattern of responses is exactly the same as the one just illustrated. This sentence differs from PROGQ:72 only because of the lack of durative adverbials. Although the two situations are apparently identical regarding the problem of the sequentiality of events, the absence of explicit indications of duration strongly suggests viewing every single event as directly anchored to the preceding and following ones, whereas in PROGQ:72 it is easy to view every single event in and by itself, i.e., relative to selfsufficient temporal coordinates. Another alternative that is given by two Portuguese informants for (17) is PROG with the Present tense. It is important to observe that PROG seems to be the most natural response in this class of examples. Note further that the same Sardinian informant who allows for St-PROG-INF in PROGQ:72, as shown in (16), admits this possibility in (17): sono stata ad arrostire (lit.: I have been-(femin.) at roasting). This informant also used this option in PROGQ:48–51, i.e., in examples crucially testing the effect of durative adverbials on PROG. It should be remarked that some informants (namely, one French, one Italian, two Romanian, one Spanish) produced the Compound Past instead of the Pluperfect in (18). Evidently, these speakers interpreted the clause because he WORK hard all week independent, so to say, of the preceding context, as though it expressed mere anteriority to the speech time. As an example, consider the following sentences:

(i) a.

Italian La barca si andava / veniva avvicinando a riva. the boat RFL go / come:3SG:IMPF approaching to shore ‘The boat was approaching the shore.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

598

598

Pier Marco Bertinetto b.

Italian La barca si andava / ?? veniva allontanando da riva. the boat RFL go / come:3SG:IMPF getting farther from shore ‘The boat was getting farther from the shore.’

As may be seen, andare is independent of deictic orientation, whereas venire seems to be sensitive to this parameter, thus showing that the bleaching of its meaning is far from complete. Even in Spanish “venir  Gerundive” is the preferred option in case of an event “continuing up to the present or to a past moment of reference” (Blansitt 1975: 26), as in: (ii)

Spanish Vienen estudiando ese problema desde hace tres años. come:3PL studying this problem since three years ‘They’ve been studying this problem for three years.’

20. Obviously, the level of grammaticalization varies from language to language. In Spanish and Portuguese it is certainly higher than in Italian, as is shown by the virtual lack of actional restrictions (cf. below). According to Giacalone Ramat (1995a), Mot-PROG in Italian is an instance of what she calls “interrupted grammaticalization”. As to the acquisition of PROG in Italian as L2, cf. Giacalone Ramat (1995b). Needless to say, the semantic bleaching shown in most cases by the auxiliary “go” does not prevent it from preserving its original meaning in specific contexts, where this periphrasis plays the role of a “perambulative” construction. This is the case, for instance, in: (i)

Italian Il mendicante andava bussando di porta in porta. the beggar go:3SG:IMPF knocking from door to door ‘The beggar was walking around, knocking at every door.’

Apparently, the persistence of a perambulative meaning in Spanish Mot-PROG is more frequent with andar than with ir, judging from examples such as: Anda revisando las puertas ‘He is (moving about) checking the doors’ (Blansitt 1975: 25). 21. Portuguese also presents the much less frequent variants “ir/vir a  INF”, which are shaped like the St-PROG variant most commonly used in Standard Portuguese. 22. The informants who did not use PROG in PROGQ:73 proposed either one of the following solutions: bare Imperative or some form of a modal verb meaning ‘must’ followed by the Infinitive. 23. In Italian the passive is admitted in the form of the so-called “process” passive, with the dynamic auxiliary venire instead of essere, as in the following sentence heard by the author: Il pacco ti sta venendo (*essendo) mandato al tuo indirizzo ‘the parcel is being sent to your address’. Note further that the marginal types (cf. Table 1) admit more freedom. Thus, in the Northern vernacular spoken in Pavia the periphrasis corresponding to “essere dietro a  INF” may be passivized, as in: l’e dre ves mangià ‘it is being eaten’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

599

The progressive in Romance, as compared with English

599

24. It should be observed that sentence PROGQ:19 (He HAVE his hair CUT /right now/) was not correctly interpreted by some informants, who used a Compound Preterite. This sentence was also infelicitous because the causative construction employed elicited in Portuguese an idiomatic expression. 25. Here is the example: It was a bright sunny day. The bees HUM, the birds SING, the cows GRAZE in the greenfield. Suddenly, the earth opened and the devil came out. 26. The constraint concerning stative verbs may sometimes be circumvented even in Italian, especially in some colloquial varieties. Amenta (1994–95) presents a few spontaneous examples gathered in Palermo. But even a Northern newspaper like La Stampa may occasionally present sentences such as the following: Non credo che ci sia un maggior narcisismo. Anche se una certa generazione di sacerdoti quarantenni si sta vestendo un po’ alla ‘monsignore’ ‘I do not believe there to be more of a narcisistic attitude, even though a certain generation of priests is (currently) dressing so to say à la monsignor’. This case is remarkable because the usage of PROG does not destativize the predicate, as normally happens in such cases. Rather, it merely introduces the idea of the temporary validity of the statement. We shall find further examples of this in the English postural verbs quoted below. Another example leading to the same conclusions is the following, heard on the radio in a commentary about the economic situation: I profitti quindi stanno rimanendo all’estero ‘the profits are thus kept abroad for the time being’ (lit.: are remaining). 27. Apparently, this could be considered a quite natural application of the intrinsic semantic value of the progressive, which is obviously related to the notion of temporariness. Indeed, also in John is being kind there is a clear implication that this is a temporary situation. However, it must be understood that this is just a strong preference, rather than an inviolable constraint, as we may gather from PROGQ:60, a sentence depicting a permanent (non-stative) situation (Think, while we are talking about our matters, the earth TURN around the sun), which elicited PROG in almost all the languages considered (excepting Catalan and of course Romanian). Thus, temporariness is not a necessary feature of PROG. 28. Here are the examples: John inadvertently INSULT his neighbour with his silly questions; Philip unconsciously ADMIT the guilt. 29. Actually, PROGQ:56 (The train LEAVE) was an unfortunate choice in the case of French because the word train could not possibly appear as the subject of a periphrasis based on the same word. As to imminentiality, it should be noted that a few subjects (one Italian, two Portuguese) also provided, as an alternative to PROG, explicit imminential periphrases in sentences PROGQ:56 and PROGQ:57, as did one French informant in PROGQ:32 by means of the construction “aller  INF”. By contrast, the future-time reference meaning (cf. Section 6.3.3) focussed upon in PROGQ:66–69 did not elicit a single instance of an imminential periphrasis. This type of construction was instead used by one Romanian informant in PROGQ:24 and PROGQ:26, containing the phasal verb FINISH. 30. PROGQ:27, containing the durative phasal verb CONTINUE, elicited quite a number PROG responses, including those of three Italian subjects, of the English subject, and even of one French subject. Curiously, no Ibero-Romance speaker used PROG in this

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

600

600

31.

32.

33.

34.

35.

36.

Pier Marco Bertinetto context; however, one Spanish informant and the Catalan informant made use of the continuative periphrasis (“seguir  GER” in Spanish), which is morphologically close to PROG, and bears some semantic resemblance to it. Respectively: eravamo soliti pulire, soléamos limpiar (‘we used to clean’). These are not the only habitual constructions available in the languages under investigation; and indeed “used to  Inf” could have been employed in PROGQ:4. For more information, cf. Bertinetto (1991; 1997: ch. 9), Gougenheim (1929), Werner (1980). The problem, for the relevant languages, seems to derive in part from the adjacency of PROG to an adverb such as often, and in part from the respective order of main and subordinate clause, which is reversed in comparison with (27). Although the data collected suggest that Spanish is less prone to using PROG in these contexts as compared to Portuguese, it should be noted that in several South American varieties of Spanish this usage is widespread, as reported by Squartini (1998). It might be claimed that here there is some sort of focalization point, analogous to what is to be observed in: Stava mangiando da un’ora ‘s/he had been eating for an hour’ (lit.: (at the given point) s/he was eating since an hour). However, the verb dire ‘say’ in (30) is compatible with a hyperbolic, rather than with a truly progressive meaning (i.e., it suggests an idea of speaker-insisted iteration). In fact, the event of saying is not necessarily occurring when the sentence is uttered. In the following example, found in a scientific paper, the interpretative use of PROG is combined with passive morphology: This suggests that the relative clause tense is being interpreted indexically (meaning: ‘if the given condition occurs, then ’) Portuguese presents a choice between synthetic and analytic Future. Both allow PROG in cases such as PROGQ:83: estarei a cozinhar, vou estar a cozinhar. In Spanish too, there is an analytic future construction (“ir a  INF”), but it may not be used with PROG. Our English informant also used PROG in PROGQ:74, which presents a negative Imperative: [Mother to daughter, whom she wants to punish:] You NOT GO to that party! The use of PROG in this sentence suggests that the Imperative has been replaced by a tense expressing a future-like meaning. In fact, the Simple Future is employed by most other informants.

References Amenta, Luisa 1994–95 Bausch, Karl 1964

La perifrasi aspettuale stare  Gerundio. [Dissertation, University of Palermo.] Verbum und verbale Periphrase im Französischen und ihre Transposition im Englischen, Deutschen und Spanischen. Ph.D. thesis, Universität Tübingen.

Bégin, Claude 1996

“Characterizing the type of outcome evoked by the Perfect Simple and the Perfect Progressive in English”, Revue Québecoise de Linguistique 24: 39–52. Bertinetto, Pier Marco 1986 Tempo, Aspetto e Azione. Il sistema dell’indicativo. Firenze: Accademia della Crusca. 1991 “Il verbo”, in: Lorenzo Renzi & Giampaolo Salvi (eds.), Grande grammatica italiana di consultazione. Vol. II, Bologna: Il Mulino: 13–161.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

601

The progressive in Romance, as compared with English 1994

601

“Statives, progressives and habituals: Analogies and divergences”, Linguistics 32: 391– 423. 1996 “Le perifrasi progressiva e continua nella prosa dell’Otto e Novecento”, in: Lucio Lugnani, Marco Santagata & Alfredo Stussi (eds.). Studi offerti a Luigi Blasucci dai colleghi e dagli allievi pisani. Lucca: Pacini Fazzi: 77-100. 1997 Il dominio tempo-aspettuale. Demarcazioni, intersezioni, contrasti. Torino: Rosenberg & Sellier. in press “Sui connotati azionali ed aspettuali della perifrasi continua (“andare / venire  Gerundio”)”, in: Giuliano Bernini, Pierluigi Cuzzolin & Piera Molinelli (eds.), Ars linguistica. Studi offerti da colleghi ed allievi a Paolo Ramat in occasione del suo 60Æ compleanno. Bulzoni: Roma. to appear “Verso una definizione della perifrasi ‘continua’ (“andare / venire  Gerundio”)”, in: Luciano Agostiniani, Maria Giovanna Arcamone, Onofrio Carruba, Fiorella Imparati e Riccardo Rizza (eds.). do-ra-qe pe-re. Studi in onore di Adriana Quattordio Moreschini. Pisa: Giardini. Bertinetto, Pier Marco & Mario Squartini 1995 “An attempt at defining the class of ‘Gradual Completion Verbs’ ”, in: Pier Marco Bertinetto, Valentina Bianchi, James Higginbotham & Mario Squartini (eds.), Temporal reference, aspect, and actionality. Semantic and syntactic perspectives. Torino: Rosenberg & Sellier: 11–26. Blansitt, Edward L. 1975 “Progressive aspect”, Working Papers on Language Universals 18: 1–34. Blücher, Kolbjörn 1973 “Considerazioni sui costrutti del tipo stare cantando, andare cantando, venire cantando”, Revue Romane 8: 13–20. Brianti, Giovanna 1992 Périphrases aspectuelles de l’italien. Le cas de andare, venire et stare  gérondif. Berne etc.: Lang. Brinton, Laurel J. 1981 The historical development of aspectual periphrases in English. PhD. Dissertation, University of California. Böckle, Klaus 1979 “Zur aspektuellen Verbalperiphrasen im Französischen, Portugiesischen und Italienischen”, in: Richard Baum, Franz-Joseph Hausmann & Irene Monreal-Wickert (eds.), Sprache in Unterricht und Forschung. Schwerpunkt Romanistik. Tübingen: Narr: 195– 216. 1980 “Zum Diasystem der portugiesischen Verbalperiphrasen mit dem ’gerundialen Infinitif’ ”, Zeitschrift für Romanische Philologie 96: 333–354. Carlson, Gregory 1978 Reference to kinds in English. Indiana University Linguistics Club. Cordin, Patrizia 1997 “Tense, mood and aspect in the verb”. In: Martin Maiden & Mair Paerry (eds.), The Dialects of Italy. Routledge: London. 87–98. D’Achille, Paolo & Claudio Giovanardi 1998 “Dal romanesco del Belli al romanaccio contemporaneo: conservazione e innovazione nella sintassi verbale”, in: Paolo Ramat & Elena Roma (eds.), Sintassi storica – Atti del 30Æ Congresso della Società di Linguistica Italiana. Roma: Bulzoni: 469–494. Delfitto, Denis & Pier Marco Bertinetto 1995 “A case study in the interaction of aspect and actionality: The Imperfect in Italian”, in: Pier Marco Bertinetto, Valentina Bianchi, James Higginbotham & Mario Squartini

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

602

602

Pier Marco Bertinetto

(eds.), Temporal reference, aspect and actionality. Semantic and syntactic perspectives. Torino: Rosenberg & Sellier: 125–142. Dias da Costa, Albano 1976 “Periphrastic verbal expressions in Portuguese”, in: Jürgen Schmidt-Radefeldt (ed.), Readings in Portuguese linguistics. Amsterdam: North-Holland: 187–243. Dietrich, Wolf 1973 Der periphrastische Verbalaspekt in den romanischen Sprachen. Tübingen: Narr. 1984 “Zum Typ estar cantando und estar a cantar im Portugiesischen", in: Günter Holtus (ed.), Umgangsprache in der Iberoromania. Festschrift für Heinz Kröll. Tübingen: Narr: 137–145. 1985 “Die Entwicklung der aspektuellen Verbalperiphrasen im Italienischen und Spanischen”, Romanische Forschungen 97: 197–225. Dominicy, Marc 1977 “Les périphrases verbales des langues romanes”, Revue Belge de Philologie et d’Histoire 55: 914–934. Durante, Marcello 1981 Dal latino all’italiano moderno. Saggio di storia linguistica e culturale. Bologna: Zanichelli. Fente Gómez, Rafael, Jesús Fernández Alvarez & Lope G. Feijóo 1972 Perífrasis verbales. Madrid: S.G.E.L. Fernández de Castro, Félix 1990 Las perífrasis verbales en español. Comportamiento sintáctico e historia de su caracterización. Oviedo: Departamento de Filología Española. García González, J. 1992 Perífrasis verbales. Madrid: S.G.E.L. Giacalone Ramat, Anna 1995a “Sulla grammaticalizzazione di verbi di movimento: andare e venire  gerundio", Archivio Glottologico Italiano 80: 168–203. 1995b “Progressive periphrases, markedness, and second language data”, in: Stig Eliasson & E. H. Jahr (eds.), Memorial volume for Einar Haugen, Berlin: Mouton de Gruyter. Gougenheim, Georges 1929 Etude sur les périphrases verbales de la langue française. Paris: Nizet. Haegeman, Liliane 1982 “The futurate progressive in Present-Day English”, Journal of Linguistic Research 2: 13–19. Hirtle, W. H. 1967 The simple and the progressive forms: An analytical approach. Québec: Les Presses de l’Université Laval. Iglesias Bango, Manuel 1988 “Sobre perífrasis verbales”, Contextos 12: 75–112. Jones, Michael Allan 1993 Sardinian syntax. London/New York: Routledge. King, Larry D. & Margarita Suñer 1980 “The meaning of the progressive in Spanish and Portuguese”, The Bilingual Review / La Revista Bilingüe 7: 222–238. Kisbye, Torben 1971 An Historical Outline of English Syntax. Aarhus: Akademisk Boghandel. König, Ekkehard 1995 “He is being obscure: Non-verbal predication and the progressive”, in: Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.), Temporal reference,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

603

The progressive in Romance, as compared with English

603

aspect, and actionality. Typological approaches. Torino: Rosenberg & Sellier: 155– 168. Leech, Geoffrey 1971 Meaning and the English verb. London: Longmans. Loi Corvetto, Ines 1982 L’italiano regionale di Sardegna. Bologna: Zanichelli. Lyer, Stanislav 1934 Syntaxe du gérondif et du participe présent dans les langues romanes. Paris: Droz. Markiˇc, Jasmina 1990 “Sobre las perífrasis verbales en español”, Linguistica 30: 169–206. Metslang, Helle 1995 “The progressive in Estonian”, in: Bertinetto Pier Marco, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.), Temporal reference, aspect, and actionality. Vol. 2: Typological approaches. Torino: Rosenberg & Sellier: 169–183. Meyer-Hermann, Reinhard 1978 “Zu den ‘Verbalperiphrasen’ im heutigen Portugiesisch”, Portugiesische Forschungen des Görresgesellschaft, Erste Reihe, Aufsätze zur portugiesischen Kulturgeschichte 15: 204–226. Mossé, Fernand 1938 Histoire de la forme périphrastique être  participe présent en germanique. Première partie: Introduction, ancien germanique – vieil anglais. Paris: Klincksieck. Nakaona, Shoji 1978 “Notas sobre las perífrasis de gerundio de sentido durativo”, Linguística Hispánica 1: 161–191. Reese, Susanne 1991 Gerundialkonstruktionen im Spanischen. Tübingen: Narr. Ricca, Davide 1998 “Una perifrasi continua nei testi piemontesi dal Cinquecento all’Ottocento: TENERE  participio passato”. in: Paolo Ramat & Elena Roma (eds.), Sintassi storica – Atti del 30Æ Congresso della Società di Linguistica Italiana. Roma: Bulzoni: 345–368. Roca Pons, J. 1958 “Estudio sobre perífrasis verbales del español”, Revista de Filología Española, anejo 67. Rohlfs, Gerhard 1969 Grammatica storica della lingua italiana e dei suoi dialetti. Torino: Einaudi (orig. ed. 1954). Rohrer, Christian 1977 “Die Beschreibung einiger spanischen Verbalperiphrasen im Rahmen eines zeitlogischen Systems”, in: Christian Rohrer (ed.), On the Logical Analysis of Tense and Aspect. Tübingen: Narr: 99–129. Rojo, Guillermo 1974 Perífrasis verbales en el gallego actual. Verba, Anejo 2. Scheffer, Johannes 1975 The progressive in English. North-Holland, Amsterdam. Schemann, Hans 1983 Die portugiesischen Verbalperiphrasen. Tübingen: Niemeyer. Schlieben-Lange, Brigitte 1971 Okzitanische und katalanische Verbprobleme. Tübingen: Niemeyer. Schmitz, John R. 1982) “The progressive construction and stativeness in Brazilian Portuguese", in: James P.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

604

604

Pier Marco Bertinetto Lantolf & Gregory B. Stone (eds.), Current Research in Romance Languages. IULC: 145–156.

Schnerr, Walter J. 1954 “The progressive tenses in Brasilian Portuguese”, Hispanic Review 22: 282–305. Škerlj, S. 1926 Syntaxe du participe présent et du gérondif en vieil italien, avec une introduction sur l’emploi du participe présent et de l’ablatif du gérondif en latin. Paris: Bibliothèque de l’Ecole des Hautes Etudes, Sciences Historiques et Philologiques. Spaulding, Robert K. 1926 “History and Syntax of the Progressive Constructions in Spanish”, University of California Publications in Modern Philology 13: 229–284. Squartini, Mario 1990 “Contributo per la caratterizzazione aspettuale delle perifrasi italiane andare  gerundio, stare  gerundio, venire  gerundio. Uno studio diacronico”, Studi e Saggi Linguistici 30: 117–212. Squartini, Mario 1998 Verbal periphrases in Romance: Aspect, actionality and grammaticalization. Berlin: Mouton de Gruyter. Vezzosi, Letizia 1996 “La costruzione participio presente + verbo essere in anglosassone: un antesignano della forma progressive o qualcosa di diverso?”, Studi e saggi Linguistici 36: 157–210. Werner, Edeltraud 1980 Die Verbalperiphrase im Mittelfranzösischen. Eine semantisch-syntaktische Analyse. Frankfurt a. M.: Lang. Visser, Fredericus Th. 1969/1973 An historical syntax of the English language. Part III. Leiden: Brill.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

605

Karen H. Ebert

Progressive markers in Germanic languages

1. Introduction

1.1. The state of the art

Progressive markers have never been treated systematically in any grammar of a Germanic language other than English. (The English progressive was not part of the present investigation, as it is dealt with by Bertinetto (this volume)). The reason for the neglect may lie in the fact that the progressive is nowhere grammaticalized to the same extent as in English. This is reflected on the one hand in the optionality of the markers, on the other in a variety of alternative forms. The progressive constructions are, however, mentioned in various subchapters of grammars, sometimes in a chapter on participles (Koefoed 1969: 188) or on the use of the preterite (Jones & Gade 1981: 119), sometimes only in a footnote (Duden 1984: 94, Erben 1972: 75). In the German tradition they are subsumed under ‘durative aspect’ or ‘aktionsart’; in grammars with an Anglo-Saxon orientation they are often presented as optional possibilities for translating the English progressive. Kress (1982: 159f) has a whole chapter on the Icelandic progressive formed with að, but refers to the postural verb construction only in a footnote. Hansen (1967: 30– 31) classifies the Danish postural verb construction as a durative aktionsart/aspect, whereas være ved at / være i færd med is said to indicate non-completed activities. The Dutch grammar by Geerts et al. (1984) offers the most explicit discussion, devoting several pages to both the postural verb construction and aan het V zijn. Both are, together with continuative, subsumed under ‘duratief aspect’. Of the few recent articles, Andersson (1989) deserves special mention. Andersson describes the use of the German am-form in the Rhineland dialect, but he also gives a brief summary of the Standard German progressive markers which gives more information than any of the grammars. Boogaart (1991), van der Hauwe (1992) and Bhatt & Schmidt (1993) are attempts to explain the Dutch and German progressives in theoretical models. I have earlier published two articles on Fering and on German and Dutch progressives (Ebert 1989, 1996), based mainly on native speaker intuition and/or written prose.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

606

606

Karen H. Ebert

1.2. The data The investigation of tense and aspect in E UROTYP was based on questionnaires. Answers to the Progressive Questionnaire (PROGQ, see p. ??? in this volume) were provided for the following languages: Icelandic: Swedish: Danish: North Frisian1: Frysk: Dutch: German: Züritüütsch:

1 7 3 6 (4 Fering-Öömrang, 1 Wiidinghiird, 1 Mooring) 1 (based on several native speakers’ judgments) 2 8 (5 standard colloquial2, 1 Rhineland, 2 mixed with Rhineland influence) 4

We collected no data for Faroese, Norwegian, Low German, and Yiddish, which will be mentioned briefly in part 5. The different forms used in the answers to PROGQ are listed in Table 4 at the end of this article. Eight questionnaires have been neglected in this Table: three Swedish ones which do not contain a single progressive construction; the questionnaire from the Mooring dialect of North Frisian, as practically only IIb forms (see Table 1) are used; two German questionnaires which have progressive markers only in the incidential schema3 and in PROGQ: 1. These results are, of course, interesting in themselves, but they need not be listed. Two of the Züritüütsch questionnaires, which did not yield anything new, were neglected mainly for reasons of space. Evaluation of the PROGQs was problematic for several reasons. Some consultants tried to list all possibilities, whereas others gave only the translation that came to mind first. Some checked and discussed with several native speakers, so that their questionnaire already represents a broader consensus (e.g., Frysk). The informants did not always understand what the intended meaning was. If the misunderstanding is obvious, I have left a blank in Table 4. Very few informants answered the theoretical questions.4 In some of the languages the progressive constructions belong to an informal style. They are frequent in the spoken language, but avoided in writing. In a few cases I had the possibility to conduct an oral interview following the completion of the questionnaire; the results were sometimes rather different, especially for German, where many dialectal and sociolectal differences exist. A systematic comparison of the written and the elicited data on one hand with spoken discourse on the other would be interesting, but was beyond the aims and possibilities of the project.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

607

Progressive markers in Germanic languages

607

1.3. Inventory of progressive markers Most Germanic languages have at least two types of constructions used in the typical progressive contexts: I. Postural verb constructions: (POS) II. Prepositional constructions: (PREP) III. ‘hold’ constructions: (HOLD)

“sit”  to  INF (Frisian, Dutch) “sit”  and  V (Scandinavian) be + in/at + the + INF (Dutch, Frisian, German) be + at + to + INF (Danish) be + to + INF (Icelandic) hold on/in (Swedish, Norwegian, Yiddish)

The postural verbs include the verbs for ‘sit’, ‘stand’, ‘lie’, ‘hang’ and ‘go (around)’. The verb ‘go’ in a POS-construction always indicates undirected motion. German and Züritüütsch lack the POS-construction. The PREP- and the HOLD-constructions are used in roughly the same way, and languages have either one or the other. I have distinguished a subgroup PREP IIb for Frisian and German forms with the preposition bi, bei ‘at, near’, which is restricted to agentive verbs. The expressions listed under IV in Table 1 all emphasize the dynamic, active character of the event, comparable to French être en train de or English be busy doing. All are marginal; only a few forms were used in the questionnaires (see Appendix).

Table 1. Progressive markers (3rd person sing. forms) I POS

IIa PREP

Ice Swd Dan Wiid FerÖöm Frysk Dut Grm Züri

sittur og sitter och sidder og sät to/än sat tu sit te zit te

er að

Far Nor Yid LowGrm

sit og står og

er og

sit un

is an’t

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

er ved at äs oon’t as uun’t is oan ’t is aan het ist am isch am

IIb ‘bei’

III HOLD

IV BUSY

hålla på och/att

er i färd med att er i gang med

äs bai to/än as bi tu/an is oan ’e gong mei is bezig te ist dabei zu isch draa z’

ist beim

er fáast við at holde på (med) halt in is bii un

608

608

Karen H. Ebert

2. Use of the progressive markers 2.1. Test cases The PREP and POS-constructions were used in typical progressive contexts, e.g., as answers to the question ‘What is X doing right now?’ (PROGQ: 1, 6–18) and in the incidential schema (see (3)). (As the structure of all Germanic languages is very similar to that of English, I have chosen to provide word-by-word rather than morphemic glosses, which would complicate the reading unnecessarily. I also gloss sentences with parallel constructions together for reasons of space. Icelandic að corresponds to both ‘at’ and ‘to’). (1)

(PROGQ: 1)5 Ice Hún Dan Hun Wiid Jü she Swd

FerÖöm Frysk Dutch Grm Züri

(2)

er er äs is

að at to to

ved bai AT

vinna. arbejde. äprüüten. work / clean_up

Hon håller på att arbeta / håller på och arbetar. she holds on to work holds on and works köögin. Hat as uun t iten-sieden. Hja is oan t Ze is aan het koken. Sie ist am Kochen. Si isch am choche. she is IN/AT the cooking6 ‘She is working / cleaning up / cooking.’

(PROGQ: 7) Swd Han Dan Han Wiid Hi he

sitter sidder sät sits

uun t in_the

läser læser löst reads

tidningen. avisen. et bläär. the newspaper

bleed newspaper

tu to

FerÖöm

Hi he

Frysk Dut

Hy sit in krante te Hij zit de krant te he sits a/the newspaper to ‘He is reading a/the newspaper.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

sat sits

och og än and

leezen. read lêzen. lezen. read

609

Progressive markers in Germanic languages

(3) a.

b.

609

(PROGQ: 3) egar Jón kom, da John kom . . . Ice . . . var Anna enn að vinna. Dan . . . var Anne stadigvæk ved at arbejde . . . was A. still AT work:INF Swd . . . höll Ann fortfarande på att arbeta held Ann still on to work:INF leezen FerÖöm . . . wiar Anne noch uun t lêzen Frysk . . . wie Anne noch oan t Dut . . . was Anne nog aan het lezen Grm . . . war Anna noch am Lesen. was A. still IN/AT_the reading (lacking in Icelandic and German) Dan . . . sad A. og arbejdede. Swd . . . satt A. och arbetade. sat A. and worked FerÖöm . . . seed A. noch tu leezen. Frysk . . . siet A noch te lêzen. Dut . . . zat A. nog te lezen. sat A. still to read ‘[When John came], Anne was still working / reading.’

As a rule, progressives were not used in the negative test sentences PROGQ: 2, 4, 5, 71 exemplifying habitual (but cf. (33)), repeated or sequential actions. No progressives were used with stative verbs in any of the languages (PROGQ: 39–43). Unlike in English, it is not possible to use a progressive to express temporary states. Thus none of the other Germanic languages can have a progressive marker in the translations of: She is sitting in the kitchen (PROGQ: 28), The shirt is hanging on the nail (PROGQ: 29), You are being rude this evening (PROGQ: 42). Progressives generally do not combine with adverbs that specify a limited duration. No PREP- or HOLD-constructions were used in PROGQ: 48. The fact that a number of informants in various languages chose a POS here (cf. (29)) suggests a different status for this construction.

2.2. Transitive verbs and incorporation The North Frisian and Standard German PREP and POS do not combine with a direct object. In order to use a progressive, the verb phrase has to be intransitivized by way of incorporation. Incorporation is common, also in Frysk and Dutch7, if the combination verb  noun designates a typical activity. This characterization is nec-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

610

610

Karen H. Ebert

essarily vague, as no well-defined rules exist and there is much individual variation. Thus peeling potatoes and catching butterflies are obviously ‘typical’ activities in Germanic societies, whereas catching elephants and writing a thesis are not.8 If the object cannot be incorporated, North Frisian and German speakers have to resort to some other construction, as in (5b).9 (4)

(PROGQ: 12) Frysk Hja is oan ‘t jirpel-skilen/ is j. Dutch Ze is aan het aardappel-schillen/ is a. FerÖöm Hat as uun’t eerdaapler-skelin/ *as e. Grm Sie ist am Kartoffel-schälen/ *ist K. she is IN/AT_the potato(es)-peeling is p. ‘She is peeling potatoes.’

(5) a.

b.

(PROGQ: 14) Frysk Hja is trije kilo jirpels she is 3 kilo potatoes Dutch Ze zit drie kilo aardappelen she sits 3 kilo potatoes FerÖöm Hat as diarbi, 6 pünj eerdaapler Grm Sie ist dabei, 6 Pfund Kartoffeln she is there_AT 6 pound potatoes ‘She is peeling 6 pounds of potatoes.’

oan t skilen. aan het schillen. uun t skelin. am Schälen. IN/AT_the peeling

skilen oan t AT_the peeling te schillen. to peel tu skelin. zu schälen. to peel

There is an alternative construction which transforms the direct object into a prepositional phrase, thereby detelecizing it (cf. also FerÖöm (2)). (6)

(PROGQ: 18) FerÖöm Hat skraft bi sin doktorarbeit. Grm Sie schreibt an ihrer Doktorarbeit. she writes at her dissertation Dutch Ze zit aan haar proefschrift te werken.10 she sits at her dissertation to write/work ‘She is writing her dissertation.’

In the Rhineland dialect, am is used with transitive verbs just as naturally as with intransitives (cf. Andersson 1989; Bhatt & Schmidt 1993).11 Some Züritüütsch speakers incorporate definite noun phrases and even locative complements.12 (The Züritüütsch examples are rendered in the transcription used in the questionnaires.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

611

Progressive markers in Germanic languages

(7)

(PROGQ: 13) GrmRhi Sie ist die Kartoffeln am Schälen. she is the potatoes AT_the peeling Züri Si isch (grad) am t’ herdöpfel schele. she is just AT the potatoes peel ‘She is peeling the potatoes.’

(8)

(PROGQ: 16) GrmRhi Sie ist die Hühner am Raus-jagen. she is the chickens AT_the out-chasing Züri Si isch am t’ hüener us em huus jage. she is AT the chickens out the house chase ‘She is chasing the chickens out of the house.’

611

Locative phrases can be incorporated together with objects to some degree also in Standard German and Frisian. Object, locative and verb together are then quasilexicalized. Note, however, that definite objects are totally unacceptable. (9) a.

FerÖöm Grm

b.

Frysk FerÖöm

Jo san uun’t (*dön) köölen-deel-uun-kääler-dreegen. Sie sind am (*die) Kohlen-in-den-Keller-tragen. they are AT_the the coals-(down)in-the-cellar-carrying ‘They are carrying coals down into the basement.’ Hja binne oan’t hea yn ’e skuorre bringen. Jo san uun’t fooder iin uun skini keeren. they are in_the hay into in the barn taking ‘They are taking hay into the barn.’

2.3. Combinability with tense, voice and modal verbs There are in principle no restrictions for the combination of progressives with tenses. As marking of future time reference is not common in any of the languages, it is not surprising that we found only present progressive forms in future contexts (e.g. PROGQ: 83). Only the Icelandic future marker cannot be combined with the progressive (Kress 1982: 159). The Dutch future marker is more acceptable with aan’t than with POS (De Groot 1992: 7). In Fering-Öömring, inserting wal or skal with a progressive marker invariably yields a modal interpretation (cf. Ebert 1994a). (10) a.

(FTRQ: 2, Dahl 1992: 64) Swd Han kommer att hålla på att skriva ett brev. he comes to hold on to write a letter

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

612

612

Karen H. Ebert

Dutch

b.

Grm

Hij zal brieven aan het schrijven zijn /?zitten te schrijven. he shall letters AT the writing be sit to write ‘He will be writing a letter/letters.’ Wenn du nachhause kommst, werde ich am when you home come FUT I AT_the Briefeschreiben sein. letter-writing be ‘When you come home, I will be writing letters.’

In PROGQ: 81 (=11a) with a present perfect both progressive constructions were used, though POS is somewhat more natural in Dutch and the Frisian languages.13

(11) a.

b.

(PROGQ: 8l) Ice ég er búinn að vera að baka í allan dag. I PF be AT baking in all day Swd Jag har hållit på att baka hela dagen. I have held:PP on to bake whole day:DEF FerÖöm Ik san di hiale daai uun ‘t baagen weezen. Frysk Ik ben de hiele dei oan ‘t bakken west. Grm Ich bin den ganzen Tag am Backen gewesen. I am the whole day AT_the baking been FerÖöm Ik haa di hiale daai stänen tu baagen. Frysk Ik ha de hiele dei stien te bakken. Dutch Ik heb de hele dag staan bakken. I have the whole day stood:PP to bake ‘I have been baking the whole day.’

Icelandic is the only language where a past perfect was used in PROGQ: 82. Combinations of past perfect with progressive markers seem somewhat odd in the other languages; I have only found a single natural occurrence with a POS. The example translates naturally into Fering with a POS, but not with PREP.

(11) c.

Dan

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

[På politigården i Malmö bed Per Månsson tandstikkeren over] som han havde siddet og tygget på, . . . which he had sat:PP and chewed:PP on (Sjöwall &Wahlöö, cited from Haberland 1978: 65)

613

Progressive markers in Germanic languages

FerÖöm

613

[Üüb a politsei uun Malmö beed Per M. det swaawelstook troch,] huar hi üüb tu kauin seeden hed / where he on to chew sat:PP had ?huar hi uun’t kauin üüb weesen wiar ‘[In the Malmö police station Per Månsson bit through the match] he had been chewing on.’

The combination of a PREP with a modal verb in PROGQ: 79, 80 yields only an epistemic reading (12a). The two sentences in question do not invite a POS, but this construction combines freely with modal verbs in the deontic meaning. In sentences (12b, c) PREP is excluded. (12) a.

b.

c.

(PROGQ: 79) Swd Tom måste hålla på och mata djuren. T. must hold on and feed animals:DEF Dutch Tom moet de dieren aan het voederen zijn. FerÖöm Tom mut uun’t fulrin weez. T. must the animals AT_the feeding be ‘Tom must be feeding the animals.’ FerÖöm Skääl dü imer sat tu liaren? must you always sit to learning ‘Do you have to sit and learn all the time?’ FerÖöm Wi maad äi linger üüb a bus stun tu teewen. we liked not longer on the bus stand to waiting ‘We did not want to wait for the bus any longer.’

The progressive is excluded with ‘be’-passives and with imperatives except in Icelandic. It is possible with the Scandinavian passives formed with bli/blive ‘become’ and with some passives in -s. (13) a.

b.

c.

(PROGQ: 75) að bera matinn á borð. Icelandic að er verið it is become:PP to bring meal:DEF to table ‘Dinner is being served.’ (PROGQ: 3e) Swd Hur länge har den här bron hållit på att bygga-s? how long has this here bridge:DEF held on to build-PASS ‘How long has this bridge been being built?’ (PROGQ: 19) Swd Han håller på och blir klippt. he holds on and becomes cut:PP

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

614

614

Karen H. Ebert

Dan

Han er ved at blive klippet. he is at to become cut:PP ‘He is getting a haircut.’

2.4. Combinability with types of verbs The progressive constructions are excluded with stative verbs. They are common with dynamic verbs, both atelic (activities) and telic (accomplishments). The restriction of PREP and POS to intransitive constructions in North Frisian and German is syntactic rather than semantic. Telic intransitives do not pose a problem with uun’t/am, although they are somewhat odd with POS (cf. part 3.1). The progressive is even obligatory in some cases, for example in (14a), if the people were later saved from starvation. Replacing the progressive by a simple form in (14b) would describe the situation as much less dramatic. (14) a.

FerÖöm Grm

b.

Grm

c.

FerÖöm Grm

Jo wiar uun’t ferhongrin, jo hed wegen niks rochts tu Sie waren am Verhungern, sie hatten Wochen nichts they were AT_the starving . . . iidjen fingen. Rechtes gegessen. ‘They were starving, they had not had proper food in weeks.’ Alle Varianten des Nordfriesischen sind am Aussterben. all varieties of North Frisian are AT_the out_dying ‘All varieties of North Frisian are dying out.’ Hi as uun’t promoviirin. Er ist am Promovieren. he is AT_the dissertation_writing ‘He is writing his dissertation.’

If a progressive is used with a verb that is normally conceived of as momentaneous, it refers to a short pretransformative phase14, or it has the meaning of ‘almost’. No POS is possible here. (15) a.

FerÖöm Grm

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Ik wiar uun’t /*laai tu tusliapen, iar at telefon Ich war am Einschlafen, als das Telefon I was at_the / lay to falling_asleep when the phone klingert. klingelte. rang ‘I was falling asleep when the phone rang.’

615

Progressive markers in Germanic languages

b.

Dan

615

Jeg var ved at falde i søvn. I was at to fall in sleep ‘I almost fell asleep.’ (no progressive reading possible)

The progressive is generally excluded with directed motion verbs. Cases with momentaneous motion verbs like (16) are interpreted in the same way as (15a), i.e., referring to a short time span before the transition point. (16) a.

b.

(PROGQ: 21) Grm Sie ist (gerade) am Rausgehen. Züri Si isch (grad) am usegaa. ‘She is going out right now.’ (PROGQ: 56) Swd Tåget håller på att gå. Dan Toget er ved at køre. ‘The train is about to leave.’

No progressives were used with phasal verbs in Dutch, standard colloquial German, North Frisian or Frysk. The few cases where a progressive was combined with ‘start’ or ‘finish’ (PROGQ: 23–26) have an imminential meaning. The first Swedish informant (see Appendix) uses hålla på att, which generally can express imminential or progressive in these cases, whereas he has hålla på och with progressive meaning.15 Agentivity plays a major role in the choice between PREP and POS. With nonagentive verbs POS is clearly preferred in the languages which have it. The distribution of PREP vs. POS will be dealt with in section 3. The ‘busy’-constructions listed under IV are explicit means of indicating that a person is in the process of doing something. They combine only with agentive verbs, as do the PREP-constructions listed under IIb. (17)

Dan Dutch FerÖöm Grm

*Han er i gang med at sove. *Hij is bezig te slapen. *Hi as diarbi tu sliapen / *bi tu sliapen. *Er ist dabei zu schlafen / *beim Schlafen. ‘He is sleeping.’

2.5. Motion progressive The theoretical part of the PROGQ contained a question about a ‘motion progressive’ (Part II, 1e) and a ‘locomotive’ (Part II, 2c). Of the informants who answered the theoretical part, all but one stated that there were no such constructions in their language. Nevertheless the following sentences were given in the translations of the questionnaire sentences:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

616

616

Karen H. Ebert

(18) a.

b.

c.

(19)

Danish (PROGQ: 1) Hun går rundt og laver mad. she goes around and makes food ‘She is preparing dinner. / She is running around preparing dinner.’ (PROGQ: 52) Han går rundt og glemmer folks navne. he goes around and forgets people’s names ‘He keeps forgetting people’s names.’ (PROGQ: 44) Hun går og danser ved siden af. she goes and dances at side:DEF off ‘She is dancing in the next room.’ Wiidinghiird (PROGQ: 33) Hi lapt steeriwäch än fernärmed sän nääber mä sin he walks continuously and insults his neighbor with his dääsie fraage. stupid questions ‘He goes around insulting his neighbor with his stupid questions.’

The Danish consultant explains: One would normally say Hun står og laver mad (‘stands and’); with går rundt one gets the impression that the person is moving around “like a wild hen”. A negative connotation of this construction is also mentioned in the Dutch grammar (Geerts et al.: 538). A negative emotion can, however, also be expressed by (other) postural verbs (cf. (21a), (26)), and a motion verb construction can have a neutral interpretation, parallel to that of POS-constructions, as in: (20) a.

b.

Danish (Jones & Gade 1981: 119) Stephen gik og glædede sig til sin fødselsdag. S. went and looked_forward to his birthday ‘Stephen was looking forward to his birthday.’ Dutch Casper was ziek en liep te hoesten. FerÖöm Kasper wiar kraank an lep tu hoostin. was sick and walked to cough C. ‘Casper was ill and was (going around) coughing.’

A negative interpretation seems to be invited by the particle ‘around’, which has the same effect with ‘sit’ or ‘stand’ (21a), but in some languages ‘around’ does not exclude a neutral reading. Inserting ambi in Fering (21b) would imply a negative

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

617

Progressive markers in Germanic languages

617

evaluation of the activity, whereas the Swedish sentence with runt does not necessarily have this reading. (21c) on the other hand, given that the sentence implies a negative judgement, would be odd without ’around’ in Fering and Danish, though not in Dutch. It is totally impossible without herum in German, which does not have a POS-construction. (21) a.

b.

c.

FerÖöm Grm

Hi sat ambi tu süpen. Er sitzt rum und säuft. he sits around to booze / and boozes ‘He sits around boozing.’ Swd Hon går (runt) och samlar pengar för Jugoslavien. she goes around and collects money for J. FerÖöm Hat leept (*ambi) tu saamlin. she walks around to collect:INF ‘She is collecting (for Yugoslavia).’ (PROGQ: 2) FerÖöm Hi leept ambi tu pochin. Dutch Hij loopt te pochen. Dan Han går rundt og praler. Grm Er läuft herum und gibt an. he walks around to brag / and brags ‘He goes around bragging (all the time).’

There is thus no clear evidence for distinguishing a motion progressive from postural verb constructions, nor for distinguishing a motion progressive from a ‘locomotive’16 of the type ‘go around’. German is the only language that has no POS, and it does not allow läuft und V. It does have the attitudinal construction läuft herum und V, but also sitzt/steht herum und V. In the other languages there is no clear distinction between constructions with ‘go’ and ‘go around’. The motion progressive can therefore be included in the POS-constructions, ‘going (around)’ being a type of constant body position.17 A POS-construction with a verb meaning ‘go, walk’ is common in Danish, Dutch and Frysk, but very restricted in North Frisian. For Icelandic and Swedish more detailed information is necessary.

3. Prepositional vs. postural verb constructions 3.1. Semantic and pragmatic conditions What factors determine the choice between PREP/HOLD and POS? Although the PROGQ was not specially designed to test this opposition, some general rules and tendencies became apparent. In Table 2 I have arranged the forms used in the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

618

618

Karen H. Ebert

Table 2. POS vs PREP with types of verbs Swedish

agentive telic PROGQ: 8 10 14 17 telic 1 7 9 12 44 45 agentive telic 38? telic 35 37 77 non-intentional 33 34

agentive telic 8 10 14 17 telic 1 7 9 12 44 45 agentive telic 35 37 38 77 intentional 33 34

Danish

/ 0,h* / 0,h h h

/ 0,h 0/ 0/ 0/

/ 0,h h / 0,h / 0,h

/ 0,h 0/ / 0,h 0/

0/ 0/ ved ved

(POS),ved POS,ved POS,ved POS,ved

i gang ved i gang ved

/ h 0, / 0,POS / 0,POS / 0,POS 0/ 0/

/ 0,h 0/ 0/ / 0,h / 0,h 0/

/ 0,h / 0,POS 0/ / 0,h / 0,h POS

/ 0,POS,h / 0,POS 0/ / 0,h 0/ 0/

0/ 0/ 0/ 0/ 0/

POS POS,ved POS POS,ved / 0.POS / 0,POS

/ gang 0,i ved POS / i gang 0, 0/ 0/

h

0/

/ 0,h

0/

/ 0,POS

/ 0,POS

POS

/ 0,h / 0,POS 0/

0/ 0/ 0/

0/ 0/ 0/

POS 0/ 0/

0/ / 0,POS 0/

/ 0,POS 0/

POS 0/ 0/

0/ 0/

0/ 0/

/ 0,h / 0,POS

0/ 0/

0/ 0/

0/ 0/

0/ 0/

North Frisian

Frysk

Dutch

0/ 0/

bai 0/ bai 0/

bi 0/ 0/ bi

uun 0/ / 0,bi 0/

uun 0/ 0/ 0/

bi 0/ / 0,POS

oan oan,POS oan,POS oan

bai POS POS POS bai bai

POS,uun POS uun POS,uun uun uun

uun POS uun uun uun uun

uun POS uun POS uun, bi POS,uun

POS,uun 0/ POS,uun bi,POS / 0,uun POS,uun

POS oan,POS oan oan,POS oan oan

aan 0/

POS 0/ 0/ 0/

POS / 0,POS / 0,POS 0/

0/ 0/ 0/ 0/

POS uun POS uun

0/ 0/ / 0,POS 0/

/ 0,POS / 0,POS / 0,POS 0/

POS POS 0/ 0/

POS 0/ POS 0/

POS POS

0/ 0/

0/ 0/

0/ 0/

0/

0/ 0/

POS POS

0/ 0/

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

0/ 0/ aan 0/

bezig 0/ POS bezig POS POS bezig POS

619

Progressive markers in Germanic languages

agentive, telic PROGQ:8 10? 14 17

build a shed give a present peel 3 kilos of potatoes chase 2 chickens out of the house

agentive, telic 38?

intentional 34

619

telic PROGQ:1 7 9 12 44 45

work read the newspaper sing (a song) peel potatoes dance play cards

rot (Scandin.)

38 37 35 77

rot (Frs) boil dream rain

admit one’s guilt

33

insult the neighbor

* h stands for hålla på; the other small letters abbreviate various types or PREP forms, which should speak for themselves. Cf. Table 1.

PROGQ according to verb types. This arrangement reveals certain patterns that are lost in the general list in Table 4 (Appendix). Agentive verbs combine in principle with both PREP/HOLD and POS. Since the postural verbs still carry some of their original meaning, they are in general restricted to events that take place with the protagonist in a constant position. POS was used in PROGQ: 7 (=2) and PROGQ: 70 for ‘read’ in all the Frisian and in most Scandinavian questionnaires, in PROGQ: 12 for ‘peel potatoes’ in most Frisian and in some Scandinavian questionnaires, i.e., with activities that are typically carried out while sitting (or standing in the case of Scandinavians peeling potatoes). But POS is always possible if the activity is carried out, or imagined as being carried out, with a certain position of the body. Thus informants have people ‘sit’ or ‘stand’ cleaning a gun (PROGQ: 6) or singing a song (PROGQ: 9), ‘sit’, ‘stand’ or ‘go around’ admitting their guilt (PROGQ: 34), ‘sit’ playing cards (PROGQ: 45) and ‘go around’ dancing (PROGQ: 44). The choice between PREP, POS and simple form seems to be partly dependent on personal preferences. The second Danish consultant translated practically all agentive verbs with both POS and PREP; the other two Danes used no or one POS in the examples of Table 2. The POS-constructions of the two Dutch informants are practically in complementary distribution: one has nearly all his POS forms with agentive verbs, the other has POS only with non-agentive verbs.18 What is most striking in Table 2 is the rarity of PREP forms with non-agentive verbs. The Danish, Frysk and Dutch informants did not use a single PREP here, and only two PREP forms occurred in one of the five North Frisian questionnaires. Whether a postural verb was used or not with non-agentive verbs depends again on the general condition mentioned above, namely the constant position of the protago-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

620

620

Karen H. Ebert

nist or object. There was a high number of POS in PROGQ: 35 (=22). One usually dreams while lying down, and although it is also possible to sit, stand or go around dreaming only the verb ‘lie’ was used in the questionnaires. (22)

(PROGQ: 35) Dan Han ligger og drømmer om sin pige. Wiid Hi läit än driimt fuon sin fründin. he lies and dreams about/of his girl(friend) FerÖöm Hi läit faan sin foomen tu dremen. Frysk Hy leit fan syn faam te dreamen. Dutch Hij ligt van zijn vriendin te dromen. he lies of his girl(friend) to dream ‘He is dreaming of his girlfriend.’

The verb ‘dream’ is odd with a PREP/HOLD form. Two consultants wrote that this combination would imply that the person is not sleeping, or that he is absorbed in his dreaming; i.e., a more active interpretation results. Table 2 suggests that agentivity is the crucial factor and non-agentive verbs rule out PREP. That this is not generally true was clear already from Fering and German (14a, b). No clear picture emerged for the role of telicity in the choice of the progressive form. PREP is sometimes preferred with telic verbs (cf. (23)), but this did not become clear from the questionnaires. Östen Dahl (PROGQ comments) thinks that “hålla på is more often used with telic verbs, and in general with goal-directed activities. With the postural verbs, it is rather the other way round”. This is only partly confirmed in Table 2. There is not a single POS with telic verbs in Swedish, but there are many HOLD forms with atelic verbs. Dahl himself does not make a clear difference between atelic PROGQ: 12 (peel potatoes) and telic PROGQ: 13–15 (peel the / all the / 3 kilos potatoes). Though he uses only POS in the former and only HOLD in PROG 14–15, he gives both a POS and a HOLD form for telic PROGQ: 13 (peel the potatoes). As Table 4 in the Appendix shows, most informants marked all four sentences in the same way. The exceptions are of course North Frisian and German, where incorporation is possible only with the indefinite object. Atelic PROGQ: 16 and telic PROGQ: 17 (chase chickens / two chickens out of the house) were marked identically by all Scandinavian informants, and there are no relevant differences in the other questionnaires. A candidate for a telic non-agentive verb in the questionnaire is ‘rot’ (PROGQ: 38), but the actionality of this verb in the individual languages is far from clear. Intransitive telic verbs often have ambiguous actionality. For instance apples can (in English) rot in a week or for a week, towels can dry in two hours or for two hours. But the actionality of the translational equivalents of these verbs need not be the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

621

Progressive markers in Germanic languages

621

same. Whereas Fering drügi has the same actionality as the English dry there are two different verbs for ‘rot’. The atelic röödi combines preferably with POS, the telic ferröödi with PREP (23). With the verb drügi ‘dry’ both forms are possible (24), but I could not say that uun’t drügin implies the idea that eventually the towels will become dry and hingi tu drügin does not. The Swedish verb ruttna seems to be telic. Dahl finds hålla på “natural since the process is not yet complete”. HOLD was used in two of the Swedish questionnaires. Haberland (p.c) excludes a progressive interpretation for er ved at rødne (23), but not for er ved at tørre (24). (23)

(PROGQ: 38) FerÖöm A aapler hingi üüb a buum tu röödin /* tu ferröödin. the apples hang on the tree to rotting A aapler san uun’t ferröödin / ?uun’t röödin. the apples are IN_the rotting Dan Æblerne hænger og rådner på træet. / ??er ved at rødne. apples:DEF hang and rot on tree:DEF Swd Äpplena håller på och ruttnar på träden. apples:DEF HOLD on and rot on tree:DEF ‘The apples are rotting (on the tree).’

(24)

Dan FerÖöm Grm

Tørreklæderne er ved at tørre. A hoonskütjen san uun’t drügin /hingi noch tu drügin. Die Handtücher sind am Trocknen. the towels are AT_the drying hang still to drying ‘The towels are drying.’

Whether there is a different actionality involved or whether the difference is just idiosyncratic needs to be clarified. Generally more research into the actionality of the verbs in the individual languages is necessary. With ‘boil’ (PROGQ: 37) Swedish HOLD and Danish PREP have only an imminential reading. This suggests that the verb is ingressive-phasal in those languages, meaning ‘come to the boil, boil’.19 With HOLD/PREP the ingressive component is selected, with POS the phasal component. In Frisian and Dutch the verb is atelic. (25)

(PROGQ: 37) Swd Vattnet står och kokar. Dan Vandet står og koger. water:DEF stands and boils

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Vattnet håller på att koka. Vandet er ved at koge.

622

622

Karen H. Ebert

FerÖöm Frysk Dutch

At weeder stäänt tu köögin. = ?At weeder as uun’t köögin. It wetter stiet te sieden. *It wetter is oan’ t sieden Het water staat te koken. = ?Het water is aan het koken. the water stands to boil ‘The water is boiling.’

It is not clear whether the impossibility or oddity of a PREP-construction with ‘boil’ in Frisian and Dutch is due to non-agentivity or to non-telicity. In Fering telicity seems to matter, as shown in (23) and in (14). According to the Frysk expert oan’t is not possible with non-agentive verbs, but sentences like de see wie oan ‘t bearen ‘the sea was roaring’ are clearly acceptable. This could be a quasi-agentive use, but there could also be a further factor involved, namely dynamicity.20 Highly dynamic verbs favor PREP, whereas verbs of low dynamicity hardly occur with PREP even if they are agentive; e.g., in Germanic languages people generally ‘sit/stand to wait’. Non-intentional events are treated like non-agentive ones. Table 2 shows no PREP forms and only one Swedish HOLD. The POS-constructions chosen esp. by the Wiidinghiird and one Dutch informant seem to imply a negative evaluation, cf. (26)

(PROGQ: 34) Swd Han står och erkänner skulden oavsiktligt. he stands and admits guilt:DEF unintentionally Wiid Soner dat ’r ’t wiitj lapt ’r je än säit dat ’r without that he it knows goes he PART and says that he skili äs. guilty is Dutch Hij zit onbewust zijn schuld toe te geven. he sits unconsciously his guilt PREP to give (toegeven – admit) ‘He is (standing/sitting/going around) unintentionally admitting his guilt.’

Two contributors remark that the postural verb in (25) implies duration. Haberland writes (PROGQ): “Vandet står og koger (og har gjort det et stykke tid)” [the water stands and boils (and has done this for a while)]. I think POS always implies some noticeable duration (cf. Ebert 1989), and this explains why it is incompatible with momentaneous verbs, where we only find PREP (cf. (15a)). That the remark is linked to this example is probably due to the fact that one would not normally let the water boil for a long time. In oral interviews Fering speakers sometimes inserted a temporal adverb indicating duration in postural verb sentences:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

623

Progressive markers in Germanic languages

(27)

FerÖöm –Saaist dü: At say you the –Je, at weeder yes the water

weeder water stäänt stands

stäänt stands al already

623

tu köögin? to boil tjiin minüten tu köögin. ten minutes to boil

Another somewhat vague factor has to do with goal-directedness or the importance attributed to the activity (cf. citation from Dahl above). This criterion has been elaborated somewhat in Ebert (1989) for Fering, where uun’t is used if an activity belongs to the work that has to be done in a farming society and POS is preferred with leisure activities. Thus one can only say hat as uun’t moolkin ‘she is milking the cows’, although I have never seen anyone milk a Frisian cow other than in a sitting position. The verb is atelic, but the activity is goal-oriented. Mending socks is also goal-oriented, but it is most normal to have a postural verb here: Hat sat tu höözen-stoopin. Mending socks, knitting sweaters, and writing letters counted as a leisure activity in traditional Frisian society, done when the serious work was finished. All activities in the living quarters of the house allow POS; cf. also stun tu baagen (11b), stun tu köögin, stun tu eerdbäären-iinmaagin (stand to bake, to cook, to strawberries-preserving), sat tu eerten-ütjpülin, sat tu höözenpreglin (sit to peashelling, to stockings-knitting). The choice between PREP and POS here has to do with the degree of importance or urgency attributed to the activity. This factor seems to be relevant only in Fering. De Groot (p.c.) confirms that the factors a)–e) of Table 3 are also crucial for the choice between PREP and POS in Dutch, where one can, however, say hij zit te melken (he sits to milk). Apart from some variation between languages or idiolects, the use of POS is remarkably stable across languages and apparently also across time.21 Compare the following combinations with POS from Dutch and Swedish22 prose with examples from English sources.

Table 3. Factors determining the choice between PREP and POS

a) b) c) d) e) f)

PREP

POS

constant position agentive telic

constant position agentive

high dynamicity shorter duration goal-directed, important

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

telic low dynamicity longer duration not goal-directed, less important

624

624

Karen H. Ebert

(28)

Dutch ZAT TE SAT TO

STOND TE STOOD TO

LAG TE LAY TO LIEP TE WENT TO Swd SATT OCH SAT AND STOD OCH STOOD AND MiEngl SAT AND STOD AND LAY AND ModEngl SAT AND

lezen, schrijven, kijken, breien, bedenken, read, write, look, knit, think, schommelen, appels schillen rock to and fro, peel apples wachten, dromen, kijken, bekijken, vertellen, wait, dream, look, look at, tell, pulken, stampen picking (one’s nose), stamp (one’s foot) wachten, dromen, slapen, rommelen wait, dream, sleep, mess around/fuck leuren peddle/hawk läste, pratade, vaktade, stirrade read, talked, waited, stared vacklade vred på , höll ögat intill , rubbed, held the eyes against, shook lokede, thoughte, playd, behelde, song, dinede, loghe, lokede, thoughte, beheld, logh, wondrede, prechede, lokede, plaid, thoughte,

talked, gazed, looked, smiled, listened, thought, mused, worked, STOOD AND talked, gazed, looked, smiled, watched, stared, LAY AND gazed, watched,

(and similar for Old English; see van der Gaaf 1934, Visser 1973: 1401–1408, 1902– 1916). Earlier periods of English also had the constructions sit V-ande, sit V-ing and – with decreasing frequency – sit to V. Note that all the text examples have human subjects and none has a telic verb.

3.2. Temporal conditions The POS-construction occurs in some contexts that are usually incompatible with progressives, e.g. with adverbs that indicate a temporal limit. In PROGQ: 48 (=29) no PREP, but various POS forms were used.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

625

Progressive markers in Germanic languages

(29)

625

(PROGQ: 48) Dan Anne sad to timer og legede helt alene. A. sat 2 hours and played all alone FerÖöm Anne seed tau stünj alian tu spelin. Dutch Anne zat twee uur alleen te spelen. A. sat 2 hours alone to play ‘Anne played (?was playing) for two hours by herself.’

Although in some languages limiting adverbs can be combined with progressives (cf. Bertinetto & Delfitto, this volume), they are excluded with PREP-constructions in most Germanic languages. Dutch and the Rhineland dialect are an exception. (30) a.

b.

Dan FerÖöm Grm

Anne Anne Anna A. Dutch (?)Anne GrmRhi Anne A.

var (*2 timer/*fra 2 til 4) ved at lege. wiar (*2 stünj/*faan 2 tu 4) uun’t spelin. war (*2 Stunden /*von 2 bis 4) am Spielen. was (2 hours/from 2 to 4) AT_the play:INF/VN was twee uur alleen aan het spelen. war zwei Stunden alleine am Spielen. was 2 hours alone AT_the playing

The PREP-constructions usually imply a focussing perspective: at a certain moment such-and-such event is in progress. This explains why they do not combine with adverbs that indicate a temporal limit. No such perspective is involved with POS. Here the event can be located in relation to a time interval with a specified duration or temporal endpoint. The temporal structure for the contexts in question can be symbolized as follows: (A)

(B)

E1 E2

E1 E2

¯ R

¯ Ti

¯ Tj

(A) symbolizes the focussing perspective, which is usually taken as a test case for progressives (cf. examples in section 2.1): E1 holds at reference point R, which can be a second event (E2) (B) symbolizes the durative constellation, i.e., E1 holds over a time interval: E1 while E2 / from Ti to Tj23

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

626

626

Karen H. Ebert

In situation A all Germanic languages use a progressive construction, PREP and POS often being equally adequate; in situation B most languages allow only POS. In the focussing perspective the termination of E1 may roughly coincide with an E2 serving as R. Sentences like (3) ‘When John came, Ann was still working’ with a PREP-form in the subordinate clause can in all languages continue: ‘but she got up immediately to mix a drink’. PREP is also compatible with adverbials which delimit a stretch of time including R, like ‘since’ (31a). The crucial factor is that the situation holds at R and leaves the endpoint unspecified. PREP is not possible if there is no temporal overlap, e.g., with adverbials like ‘after 5 / after John left’, ‘before 5 / before John left’. (31) a.

FerÖöm Grm

b.

FerÖöm

Anne as sant 2 stünj / sant klook 4 uun’t spelin. Anna ist seit 2 Stunden / seit 4 Uhr am Spielen. A. is since 2 hours / since 4 o’clock AT/IN_the playing ‘Ann has been playing for (since) 2 hours / since 4 o’clock.’ [Efter a klook 7 / biföör John kaam,] haa wi seeden tu koordin. / *wiar wi uun’t have we sat to card_playing were we IN_the koordin card_playing ‘[After 7 o’clock / before John came] we were playing cards.’

For the POS-construction no limitation and no temporal overlap are necessary. Adverbials with somewhat vague limits, like ‘during’ and ‘while’ clauses, are bad with PREP (though apparently not with Swedish HOLD). Most informants used a POS in PROGQ: 49 and 50. Only one Swedish informant made a difference and used progressive markers in the unlimited context of PROGQ: 49 (=32a), but not in PROGQ: 50, which suggested that the talking stopped when the class was finished. Most Scandinavian and Frisian informants used POS in the ‘while’ clause of PROGQ: 70 (=32b), but not in the main clause. The clauses can easily be reversed, as in (32c), with the POS remaining with the verb ‘sit’. If temporal limits are pragmatically not relevant, the use of POS is conditioned mainly by the type of activity. If temporal limits do matter, both POS and PREP are excluded with co-extensive events, as is shown by the oddity a POS or PREP would create with (32d). (32) a.

Swedish (PROGQ: 49) . . . höll Ann på och pratade / satt Ann och pratade med held A. on and talked sat A. and talked with sin granne. her neighbor ‘[During the whole class] Ann was talking to her neighbor.’ [and she carried on even afterwards].

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

627

Progressive markers in Germanic languages

b.

c.

d.

627

Danish (PROGQ: 70) Mens Ann sad og læste på sit værelse, legede Martin i while A. sat and read on her room played M. in gården. garden:DEF ‘While Anne was reading in her room, Martin was playing in the garden.’ Danish Mens Martin legede i gården, sad Ann og læste på sit værelse. ‘While Martin was playing in the garden, Ann was reading in her room.’ Dutch (Boogaart 1991: 6) [Oom hurkte, nam de arm van de man en telde diens polsslagen,] terwijl hij op zijn horloge keek/*zat te kijken/*aan het kijken was. while he on his watch looked ‘[Uncle squatted down, took the arm of the man and counted his pulse] while he looked /*was looking at his watch.’

Due to the lack of a focussing perspective the POS-constructions combine easily with habitual contexts ((33a) and analogous for the other languages). But then PREP seems to be compatible with a habitual interpretation, too (33b). (33) a.

b.

FerÖöm At lidj sat an injem bluat noch tu fernseen. the people sit in_the evenings only PART to TV_watching of tu leesen. Iar seed s tu koordin earlier sat_they to card_playing or to reading ‘Nowadays people only watch TV in the evenings. Earlier they used to play cards or read.’ FerÖöm Hat as imer uun t werkin. Dut Zij is altidj aan het werken. Grm Sie ist immer am Arbeiten. she is always AT_the working ‘She is always working.’

To the question ‘What does Ann do on Saturdays’ (PROGQ: 2) one can answer with (c), but hardly with (d).24 (33) c.

d.

FerÖöm Grm

An saninjem as hat imer uun t renskin. Samstags ist sie immer am Putzen. on-Saturdays is she always IN/AT_the cleaning ‘On Saturdays she is always cleaning.’ FerÖöm *An saninjem as hat uun t renskin. Grm *Samstags ist sie am Putzen.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

628

628

Karen H. Ebert

The adverb ‘always’ is crucial in those sentences, as it invokes plurifocalization: E1 holds at every R.

3.3. Grammaticization and stylistic variation The Germanic progressives correspond to typologically frequent patterns; locative expressions and postural verbs are among the most frequent sources for progressive markers (Bybee & Dahl 1989, Bybee et al. 1994: 127ff., Heine et al. 1991: 117f.). Grammaticization has proceeded to different degrees. The PREP-constructions are all desemanticized. Syntactically they do not behave like ‘be’ + locative complement any longer. Only the German and North Frisian progressives do not allow a direct object, the PREP-construction still being analogous to a locational phrase; cf. Grm. sie ist am Schreibtisch25 ‘she is at the desk’, sie ist am Schreiben ‘she is writing’. But the Rhineland forms are spreading into other areas. Most German speakers accept a sentence like Das sind wir noch am Diskutieren ‘We are still discussing that’ with a pronominalized and preposed object, even if they otherwise reject objects with am. The POS-constructions of North Frisian underlie the same syntactic restrictions as the PREP forms. Syntactically the postural verbs behave like auxiliaries. In FeringÖömrang the perfect auxiliary with motion verbs is ‘be’, but in a POS-construction the perfect takes the auxiliary ‘have’.26 In Dutch auxiliaries appear in a past participle position without the prefix ge-, and te is or can be omitted after a non-finite POS (cf. Dutch (12a)). The meaning of the postural auxiliaries is bleached, as mentioned by several authors (Geerts et al. (1984: 538), Hansen (1967: 30)). Desemantisation has proceeded furthest in Danish (cf. (18b), (20a)). Generally the postural verbs can not replace each other, but in emotive use they are desemanticized in all languages (cf. also (19), (26)). (34)

Dutch Loop / lig / zit niet te zeuren. walk / lie / sit not to whine ‘Stop whining.’

Postural verb periphrases indicating durative actionality are found in many languages. They may develop into progressives, but then it is usually only one verb that is generalized as a grammatical marker.27 Often the actional and the aspectual functions exist side by side, and this seems to be the case in the Germanic languages. There are some loose semantic restrictions on both progressive constructions, as described in part 3.1. These restrictions are also indicators of incomplete grammaticization. As a corollary of grammaticization a construction becomes obligatory, often ruling out alternative markers. This happened with English at V-ing, which replaced

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

629

Progressive markers in Germanic languages

629

the postural verb constructions that were common in earlier periods. In most Germanic languages marking of progressive aspect is not obligatory. For Frysk, Hoekstra (PROGQ) writes that “Progressive marking is nearly obligatory in West Frisian. The use of a simple verb instead of a progressive construction is perhaps not fully ungrammatical, but marked to say the least [ ] There is no relevant difference with respect to style and register. Both constructions are very frequent both in the spoken language and the written (standard) language.” In traditional Fering progressive marking is strongly preferred. Some speakers find a simple present inadequate to describe what is going on at the moment (Ebert 1989: 308). That the constructions are seldom in the written materials is due to the fact that written Fering is influenced by German even more than the spoken language (cf. Ebert 1994b). According to van der Hauwe (1992: 9) a progressive marker is necessary in Dutch in some contexts, for example in the incidential schema (cf. (3)) a simple form would be infelicitous.28 This can hardly be correct for the written language, as in some texts not a single progressive marker was found. In German the progressive is quasi-obligatory at least in the Rhineland and adjacent areas, but also in Züritüütsch. Many speakers in the northern and northwestern part of Germany regularly use the am-form in conversation. In spite of being banned by school teachers, the am-form is spreading into all the German speaking areas and also into more formal registers (see Ebert (1996) for examples). Thieroff, although he claims that there is no grammatical aspect in German, admits “dass der Gebrauch der sog. ‘rheinischen Verlaufsform’ (er war am Essen) in weiten Teilen des deutschen Sprachgebietes ständig zunimmt” (1992: 70). The am-form seems to become indispensable in the incidential schema; two informants who otherwise do not use the progressive (in the PROGQ) felt it to be necessary here. The Danish POS-construction is very common in the spoken language. In Rifbjerg (1957), written in an extremely colloquial style, I found several occurrences of POS on each page (see Ebert (1989: 315–316) for examples), but practically no PREP. In other prose texts I found fewer, sometimes not a single occurrence in a whole short story (but cf. (35a) and Hansen (1967: 30) for examples from older literature). In a Swedish novel POS and HOLD occurred with approximately equal frequency, but half of the hålla på att examples were imminentials. Hålla på och belongs to a lower style and is not written (except in the continuative meaning). In approximately 100 pages of Dutch short stories written in a colloquial style, I found numerous POSconstructions (cf. (28), (35b)), but only two occurrences of aan het. POS is thus on the whole more frequent than PREP in the written languages.29 In order to illustrate the natural occurrence of series of POS, let me cite two examples from literature: (35) a.

Danish (Jepsen 1964: 15) Nu stod de og råbte op om skramlet inde i vognporten, og han følte, hvor det rev i hans bryst. Det var jo en af hans tanker, de stod og tog fra ham.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

630

630

Karen H. Ebert

b.

Skramlet i vognporten, som han havde tænkt så ofte på, at det var blevet hans indvendige ejendom. Det er mit. Det er mit, ville han sige. Skal i da gå og ødelægge alt mit? ‘Now they were crying out (stood and cried out) the trash in the coachhouse, and he felt how it cut in his breast. It was after all one of his thoughts they were taking (stood and took) from him. The trash in the coachhouse, which he had thought of so often that it had become his inner property. It is mine. It is mine, he wanted to shout. Do you have to go around destroying everything that is mine?’ Dutch. (Broos n.d.: 30/ E29) Mantinea and Tegea liggen te slapen, hun lijven tegen elkaar aangedrukt om warm te blijven. De hulpverleenster staat met een setje condooms in haar hand naar de twee meisjes te kijken ‘Mantinea and Tegea are sleeping (lie to sleep), their bodies pressed against each other to keep warm. Holding a set of condoms in her hand, the social worker stands looking at the two girls.’

4. Other constructions used in the questionnaires According to prescriptive grammars (e.g., Duden 1984: 94), German beim and im should be used instead of am in the written language. Apart from the fact that the accepted forms are just as scarce in written texts as am is, the instruction is rather misleading, especially as no restrictions on the use of the two forms are mentioned. Beim and im as progressive markers are far more specific than am is. Beim combines only with agentive verbs, and im is used only in a few idiomatic expressions.30 It never occurs with agentive verbs, so that beim and im exclude each other, whereas both can be replaced by am. (36) a.

b.

German Der Mond ist am / im / *beim Abnehmen. ‘The moon is decreasing.’ Die Wunde ist am / *im / *beim Heilen. ‘The wound is healing.’ Sie ist am / beim / *im Einkaufen. ‘She is shopping.’ (PROGQ: 45/47) Sie ist am Kartenspielen / Sie ist beim Kartenspielen. (in the club, as usual) ‘She is playing cards.’ (in the next room) /

The am-sentence of (36b) is used if Anna (=she) is actually playing cards at reference time; beim implies that she is in a certain place where she usually plays cards (she need not be playing right at the moment). The beim-phrase is – at least in the northern part of Germany – associated with some locative meaning. Beim V, though not am V, is a possible answer to the question ‘where is Anne?’31 A third (‘absentive’)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

631

Progressive markers in Germanic languages

631

construction Anna ist Kartenspielen indicates that Anna has gone to the place where she usually plays cards; she may be on her way – an interpretation that is excluded for am and beim – or at that place (cf. De Groot, this volume, Ebert 1996). The absentive construction appears with some regularity in PROGQ: 46,47 in all languages except Icelandic and Züritüütsch. In Züritüütsch the construction is totally unknown, and students found, e.g., a note on the door saying bin Einkaufen ‘am shopping’, ben eten ‘am eating’, a typical use in German and Dutch, utterly funny. In absentive contexts Germanic languages often use a prepositional phrase with a derived noun. Thus PROGQ: 63 was translated with an absentive (a), with the verb ‘go’ (b), or with a noun (c). (37) a. b.

c.

(PROGQ: 63) Grm Damals war er jeden Samstag tanzen. at_that_time was he every Saturday dance:INF Frysk Doedetiids gie er alle sneonen te dûnsjen. at_that_time went he every Saturday to dance:INF Dutch Hij ging toen iedere zaterdag dansen. Grm Er ging damals jeden Samstag tanzen. he went at_that_time every Saturday dance:INF FerÖöm Dojütidj wiar hi arken saninj tu daans. Grm Damals war er jeden Samstag zum Tanz. at_that_time was he every saturday to(the) dance:N ‘At that time he went dancing every Saturday.’

A nominal construction is also often preferred with motion verbs, e.g., Swd hon är på väg ut, Isl hún er á leiðini ut (instead of intended: she is going out = PROGQ: 21), Swd hon är på väg till New York (instead of: she is flying to N.Y. = PROGQ: 22), and with ‘work’, e.g., Grm Sie ist zur Arbeit, FerÖöm Hat as tu werk ‘she is working; she is at work’ (= PROGQ: 1). German tun ‘do’ occurred with some frequency in the Rhineland PROGQ and in one of the Züritüütsch PROGQs and is not restricted to agentive verbs (‘doings’). As the verb ‘do’ is attested as a progressive marker in a few languages, the German verb tun seemed a possible candidate.32 However, tun is not restricted to progressive contexts, but appears just as naturally in habitual sentences, with stative verbs and in the imperative, e.g., GrmRhi Sie tut putzen, Züri Si tut putze (lit. ‘she does clean [every Saturday]’ = PROGQ: 2), GrmRhi Peter tut die Antwort wissen (lit. ‘P. does the answer know’ = PROGQ: 39). We can therefore exclude that tun is a progressive marker. Two informants list Swedish just or German gerade as progressive markers. As this particle has also been mentioned in the literature, I have listed it in Table 4 in the Appendix. Although just/gerade may sometimes disambiguate a sentence, it is by no

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

632

632

Karen H. Ebert

means a progressive marker. In the PROGQ sentences just/gerade as well as North Frisian jüst, Danish lige translate ‘right now’, but not the progressive. The particles combine with all tenses and aspects, including the progressive.

(38) a.

b.

FerÖöm

Ik san jüst uun’t teeapjiten / haa jüst tee apjööden / wal jüst tee apjit. Grm Ich bin gerade am Teekochen / habe gerade Tee gekocht / will gerade Tee kochen. ‘I am just preparing tea / have just prepared tea / am just about to prepare tea.’ (PROGQ: 73) Swd Se till att ni (just) håller på att arbeta när see_to_it that you:PL just hold on to work when chefen kommer. boss-DEF comes Grm Seht zu, dass ihr (gerade) am Arbeiten seid, see_to_it that you:PL just AT_the working are wenn der Chef kommt. when the boss comes ‘See to it that you are (just) working when the boss comes.’

The particle is rather odd with expected events and impossible with events of longer duration. Inserting it in (38b) has a comic effect, as it suggests that the periods in which the addressees work are rather short and that they happened to be working when the boss came in.

5. Other Germanic languages For the other Germanic languages we did not get any questionnaires, and I can only give a summary of the information I found in the literature and on the basis of my knowledge. Yiddish has a progressive periphrasis formed by haltn in ‘hold’ (Aronson’s “aspekt fun gedoyerikayt”, 1985: 175). The same verb with in eyn has continuative meaning (cf. Swd hålla på och), with baym imminential meaning (cf. Swd hålla på att). There are no restrictions for combining the progressive with tenses nor with the habitual marker fleg (Aronson 1985: 177).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

633

Progressive markers in Germanic languages

(39) a.

b.

Yiddish ikh halt in shraybn I hold in write:INF ikh halt in eyn shraybn ikh halt baym shraybn ikh vel haltn in oysgrabn ikh hob gehat gehaltn in oysgrabn ikh fleg haltn in oysgrobn (Weinreich 1952: 100)

633

‘I am writing’ ‘I keep on writing, I continually write’ ‘I am about to write’ ‘I will be digging’ ‘I had been digging’ ‘I used to be digging’

Low German has a progressive marker an’t, corresponding to German am and Dutch aan het. This form is reported by Anderson (1989) for the dialect of Lower Saxony. Slesvig Low German uses a loan translation33 from Danish is bii un which can be combined with objects. At least in this dialect a postural verb periphrasis is also common, usually with ‘and’  finite verb (as in Danish). All constructions exist also in the variety of (High) German spoken in Slesvig and in the mainland North Frisian dialects (see Ebert & Hoekstra (1996)). (6¼ )

(PROGQ: 12) LoGrm Se is an’t kartüfelschelen. she is AT_the potato-peeling ‘She is peeling potatoes.’

(9¼ )

(PROGQ: 16) SlesLoGrmSe is bii un jaagen de hööner ut huus. SlesGrm Sie is bei und jagen die Hühner aus dem Haus. Moor Jü as bai an jååg e håne üt et hüs. Wiid Jü äs bai än jaag e hoane üt. she is AT and chase the chickens out the house ‘She is chasing the chickens out of the house.’

The Slesvig German sentences of (40a, b) were intended as translations into Standard German by the author.34 (40) a.

(Bock 1933: 99) LoGrm he lichd un sløbd SlesGrm er liegt zu schlafen he lies and sleeps / to sleep ‘He is sleeping.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

634

634

Karen H. Ebert

b.

(Bock 1933: 99) LoGrm vad sdais du doa un chobsd? SlesGrm was stehst du da zu gaffen? what stand you there and/to stare ‘What are you staring at?’

Pennsylvania German has the same construction as Standard German: copula  am ( incorporated noun)  INF. It is interesting that the variety spoken by conservative religious groups allows the incorporation of definite objects, like Züritüütsch, but the preposition is then not contracted with the article (cf. Van Ness (1994: 435), phonetic symbols neglected): (41) a.

b.

Pennsylvania German ar is am brif shraiwe he is at_the letter write:INF ‘he is writing a letter’ (conservative speakers) ar is an sai bugi fikse he is at his buggy fix:INF ‘he is fixing his buggy’

Norwegian has, according to Fabricius-Hansen (1994: 54), the same ‘imperfectivizing’ forms as Danish, which she relegates – without giving criteria – to the realm of Aktionsarten. Marm & Sommerfelt (1967: 49) mention only two constructions of the ‘hold on’ type: holde på (med), drive på med. Both are translated ‘keep on with’ by the authors, but the examples given are progressives and not continuatives. Askedal (1994) mentions the postural verbs gå, stå, ligge (but not sitte) and the periphrasis drive på med, but gives only a Nynorsk example for drive og. (42) a.

b.

c.

Norwegian Bokmål (Fabricius-Hansen 1994: 54) Hans er ved å rydde opp. Hans is at to clean up ‘Hans is cleaning up.’ Bokmål Da jeg kom hjem sto Hans og lagde mat. when I came home stood Hans and made food ‘When I came home, Hans was preparing dinner.’ Bokmål (Marm & Sommerfelt) jeg holder på (med) å skrive et brev. I hold on with to write a letter ‘I am writing a letter.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

635

Progressive markers in Germanic languages

d.

e.

635

Bokmål (Askedal 1994: 246) hun sto og tenkte she stood and thought ‘she stood there thinking’ Nynorsk (Askedal 1994: 246) han dreiv og las he carried_on and read ‘he was reading’

Both sources for Faroese, Lockwood (1964) and Barnes & Weyhe (1994), mention the postural verb construction as a possible translation of the English progressive: (43) a.

b.

Faroese (Lockwood 1964: 140) vit sótu og prátaðu we sat and talked ‘we sat talking, we were talking’ teir ganga og mala they go and drift ‘they are drifting around, doing nothing’

Lockwood gives “be and V” as a second possibility for expressing progressive meaning. This construction is also used as an absentive: (43) c.

d.

meðan teir vóru og drógu lunda while they were and drew puffin ‘while they were ‘drawing’ puffin’ Eri og fái mær millum-mála. Verði skjótur am and get:PRS:1SG me between-meal will:FUT:1SG return aftur. later ‘Gone to get myself a snack. Will be back soon.’

Barnes & Weyhe (1994: 211) do not mention the latter construction, but suggest an “incipient progressive” of the ‘busy’-type. The example they give is a complement to the verb ‘see’, which is not a progressive context, but it can be changed to the following sentence: (43) e.

Menninar var fáast við at seta gørn á vatninum. man:DEF:PL were busy with to set nets in water:DEF:DAT ‘The men were busy putting nets into the water.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

636

636

Karen H. Ebert

5.1. Areal distribution The POS-constructions can be separated into two areas according to form: POS  ‘to’  INF is used in Dutch, Frysk and North Frisian, POS  ‘and’ in the Scandinavian languages (map 1). In Mainland North Frisian the Scandinavian construction exists besides the Frisian form. The form PREP  DEF  VN is spread over the Dutch, Frisian and German area. The Scandinavian languages have no common PREP-construction: Danish has PREP  ‘to’  INF, Icelandic ‘to’  INF; Swedish (and Norwegian) share a ‘hold’-periphrasis with Yiddish (map 2). Certain features are found in adjacent areas across language boundaries. Mainland North Frisian uses not only the Danish ‘sit and’ construction, but bai (corresponding in form to Danish ved) is the most frequent progressive marker here. The German am-form is constantly gaining ground. It has quite recently reached Swabia and Berlin, while it is apparently unknown further east (Thieroff, p.c.). The possibility of combining am with definite objects in the Rhineland dialect can possibly be attributed to the Dutch neighborhood (but the transitive construction is found also in Züritüütsch). From the dialect the transitive am-construction leaked into the colloquial standard in the Rhineland and is now spreading in the German speaking area. Some probable recent areal influences are tentatively presented in map 3. Needless to say, much more detailed research is necessary both into the use and the spread of the Germanic progressive constructions.

Ice

Far

Nor

Swd

2

Dan////// NFrs Frysk ///// Dutch /////

Yid

1

Grm Züri

1 2 ////

POS+ ‘to’ + INF POS+ ‘and’ + V motion PROG common

Map 2. Distribution of POS-constructions

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

637

Progressive markers in Germanic languages

Nor

2

Ice

Swd

4

Yid

Dan

Far 3

NFrs

1

Frysk Grm

Dutch

Züri 1. 2. 3. 4.

‘be’ + PREP + DEF + VN ‘be’ (+PREP) + ‘to’ + INF ‘be’ + ‘and’ + V HOLD

Map 3. Distribution of PREP-constructions

Dan Wiid SlesvGrm LoGrm

FerÖöm

Dut

?

RhineGrm SouthGrm

Map 4. Some recent areal influences

Special abbreviations ABS AT FTQR PROGQ POS PREP SlesvGrm Wiid

absentive grammaticized preposition ‘at, in’ E UROTYP future questionnaire E UROTYP progressive questionnaire postural verb (construction) prepositional construction Slesvig High German Wiidinghiird Frisian

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

EastGrm

637

638

638

Karen H. Ebert

Acknowledgements I would like to thank all friends, colleagues and students who went through the trouble to translate the questionnaire sentences: Susanne Anschütz, Antje Arfsten, Klas Danielson, Östen Dahl, Jens Edlund, Karin Gunkel, Hartmut Haberland, Ingeline Hamann, Jarich Hoekstra, Bjarne le Ferre Jacobsen, Christa König, Ulrika Kvist, Jens Jessen, Annakarin Magnusson, Dani Marmur, Johannes Merz, Mrs. and Mr. Metzger, Jos van der Molen, Hans Frede Nielsen, Adeline Petersen, Roberto Rizzi, Therese Rutishauser, Christina Tadsen, Rolf Thieroff, Gunnel Tottie, Alastair Walker, Paul Zehnder, Petúr Helgasson and an anonymous student from Berlin. Special thanks are due to Jarich Hoekstra, Casper de Groot, Hartmut Haberland and Östen Dahl for their thorough comments provided for Frysk, Dutch, Danish and Swedish respectively, and to Jarich Hoekstra, Lars Johanson, and Christa König for comments on earlier versions of this paper.

Appendix

Table 4. Complete list of markers used in PROGQ

Abbreviations for all languages 0/ no progr. marker used (blank) not translated or with different meaning POS postural verb ABS absentive N construction with noun for Swedish & Danish h+o håller på och  V h+a håller på att  INF h+m håller på med att  INF ved er ved at  INF i gang er i gang med at + INF

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

(HOLD) (HOLD) (HOLD) (PREP) (BUSY)

639

Progressive markers in Germanic languages

639

for Frisian and Dutch uun as uun’t  INF (PREP) oan is oan ‘t  INF (PREP) aan is aan het  INF (PREP) bi/bai as bi tu  INF, as bai tu / an  V (PREP IIb) bezig is bezig te  INF (BUSY) jüst ‘just’ N nominal construction for German am beim dabei G tun go

ist am  INF (PREP) ist beim  VN (PREP IIb) ist dabei zu  INF (BUSY) gerade = ‘just’ ‘do’ periphrasis ‘gone’

Icelandic does not fit in with the other Scandinavian languages, as its PREP has a quite different distribution: að: / 0: / að or 0: POSV: N:

1, 3, 6–20, 23–26, 35, 44–47, 49–50, 56–57, 64, 70a/b, 73, 75–76, 79–83 2, 4–5, 27–29, 32, 36–37, 39–43, 48, 51, 53–55, 58–63, 65, 69, 71–72, 74, 77 30–31, 33–34, 38, 52, 66–68 21,22

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

640

KD

UK

GT

Danish BJF

HH

HN

1 2 3 4–5 6 7 8 9 10 11 12 13 14–15 16–17 18 19 20 21 22 23 24–26 27 28–31 32–33 34 35 36 37 38 39–43 44 45

/ 0,h+o

/ 0,h+a

/ 0,h+a

/ 0,POS,h+o

0/

POS

/ i gang 0,

/ 0,h+o,POS

/ 0,h+a

/ 0,h+a

0/

0/

POS

0/

/ 0,h+o, POS / 0,POS / 0,h+o / 0,POS / 0,h+o / 0,h+o,POS / 0,POS / 0,h+o,POS h+o h+o h+o ?h+o h+o N N

/ 0,h+a 0/ just,h+a 0/ just 0/ / 0,h+a 0/ just just 0/ ABS just just (just)

/ 0,h+o / 0,POS / 0,h+o 0/ 0/ 0/ / 0,h+o / 0,h+o / 0,h+o 0/ 0/ / 0,h+o 0/ N N 0/ 0/ 0/

0/ 0/ 0/ 0/ 0/ 0/ 0/ ved ved ved / ved 0, ved 0/ lige+ved

POS POS,ved ved POS POS,ved POS,ved POS,ved POS,ved POS,ved POS,ved POS

ved ved i gang POS ved ved / i gang 0, i gang i gang lige+ved lige+ved ABS 0/ N N

0/ 0/

h+just+a 0/

(just)

/ 0,h+a / 0,POS / 0,h+a 0/ h+a / 0,h+a / 0,h+a / 0,h+a / 0,h+a / 0,h+a / POS 0, ABS, h+a h+a N N 0/ h+a h+m?

0/ 0/ / 0,h+o

0/ 0/ 0/

/ 0,h+a / 0,POS 0/

0/ 0/ POS

0/ 0/ 0/

0/ 0/

0/ 0/ POS

/ 0,POS h+o

0/ 0/

0/ / 0,h+a

0/ 0/

/ POS 0, / POS 0,

/ POS 0, / POS 0,

0/ POS

0/ 0/

/ 0,h+a 0/

/ 0,h+a POS

0/ 0/

0/ ABS

/ POS 0, / POS 0,

0/

0/ 0/

0/ 0/

0/ 0/ 0/

ved ved N ved 0/

lige+ved

Karen H. Ebert

Swedish OD

640

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

1. inform.

641

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

0/

0/

0/ 0/ 0/

0/

ABS / 0,POS / 0,h+o,POS 0/

ABS 0/ / 0,POS / 0,POS

ABS 0/ / 0,h+a / 0,h+a

ABS 0/ 0/ 0/

ABS 0/ 0/ 0/

0/ 0/ / 0,POS

0/ 0/ h+a h+a

/ h+a 0, 0/ 0/ h+a

0/ 0/ / 0,h+a / 0,h+a

0/ h+a 0/ 0/

0/ 0/ 0/ 0/

0/ 0/ ved

ABS POS POS POS (52:POS) 0/ 0/ 0/ ved

0/ 0/ POS

0/

/ 0,h+a

0/

0/

0/

0/

POS

N

(went)

(went)

ABS

0/ 0/

POS

POS

POSV

0/

POS

POS

/ h+o 0,

0/

0/

0/

0/

POS

0/

/ 0,h+o 0/ h+o

ABS ABS 0/

/ 0,h+a / 0,h+a h+a

0/ 0/ h+a

0/ 0/ 0/

0/

0/ i gang / POS 0,

/ 0,h+o

h+m

h+a

h+o

0/

0/

0/

Progressive markers in Germanic languages

46–47 48 49 50 51–53 54 55 56 57 58–59 60 61–62 63 64–69 70, a b 71–75 76 77–78 79 80 81 82 83

641

642

1 2 3 4–5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23–29 30 31–32 33 34 35 36 37 38 39–43

0/

0/

Fering KE

bai 0/ oont

uun,POS 0/ uun

0/ POS bai POS 0/ POS POS 0/ bai 0/ bai 0/ 0/

Dutch

JJ

JH

JM

CG

uun,POS 0/ uun,POS,bi

POS 0/ oan, POS

0/ 0/

POS 0/ POS

jüst(+bi),POS jüst jüst(+bi) jüst,uun,POS

oan, POS oan, POS oan oan, POS oan, POS oan, POS oan, POS oan, POS oan, POS oan, POS oan oan oan, POS

0/ aan 0/ aan 0/ aan

POS POS bezig POS 0/ POS POS POS POS POS 0/ bezig bezig

0/ 0/ , oan

aan 0/

0/ 0/ N

uun’t

0/

0/

0/

0/ 0/ POS

0/

0/ 0/ / POS 0,

POS POS POS

0/ 0/ POS

uun POS

0/ / 0,POS

/ POS 0, / POS 0,

POS 0/

0/ POS

AA uun uun, uun

uun uun,POS uun

bi POS bi uun 0/ 0/ uun,POS 0/ 0/ 0/ bi bi POS+bi

uun POS uun uun 0/ bi uun bi / bi 0, POS uun 0/ bi

0/ POS uun uun 0/ uun POS 0/ 0/ POS uun 0/ POS

0/ , bai oont

0/ jüst N

uun 0/ N

POS jüst

bai

0/

0/

/ bi 0,

POS POS POS

0/ 0/ POS

0/ 0/ 0/

0/ 0/ 0/

/ 0,POS / 0,POS

0/ 0/

0/ 0/

0/

Frysk Öömrang CT

jüst jüst(+bi),POS jüst(+bi) jüst bi

uun,bi,POS

aan aan 0/ 0/ aan

Karen H. Ebert

inform.

North Frisian Wiidingh. AP

642

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

2.

643

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

0/

0/ 0/

0/

bai bai ABS ABS 0/

uun uun ABS ABS / POS 0, / POS 0,

uun uun ABS ABS 0/ 0/

uun, bi POS,uun ABS POS POS POS?

/ 0,uun / 0,uun,POS ABS ABS / 0,POS / 0,uun,POS

oan oan ABS, oan ABS, oan POS POS

aan 0/ aan 0/ POS POS

aan POS ABS ABS POS POS

0/ 0/ 0/ (läi oont) 0/ (went) 0/ POS 0/

0/ , uun 0/

0/ 0/

0/ ,uun 0/

0/ 0/

0/ 0/

0/ aan

0/ 0/

(laai uun)

uun

(laai uun)

(laai uun)

(lei yn’t)

aan,POS

POS

N

0/

N

N

0/

0/

(went)

POS 0/

POS 0/

POS 0/

uun 0/

oan,POS oan,0/

aan aan,0/

POS 0/

0/

0/

0/

uun

0/

0/

0/

POS 0/ 0/ 0/ 0/ 0/

0/ , uun 0/ 0/ uun / uun 0, POS

uun 0/ 0/ 0/ 0/ uun

/ uun 0, uun 0/ / ABS 0, uun 0/

uun 0/ 0/ 0/ bi, POS

0/ 0/ oan oan oan

0/ aan 0/ 0/ POS

POS 0/ 0/ aan aan POS

bai

uun

uun

bi

bi

oan

aan

POS

Progressive markers in Germanic languages

44 45 46 47 48 49–50 51–52 53 54–55 56 57 58–62 63 64–69 70, a b 71–72 73 74–75 76 77 78 79 80 81 82 83

643

644

1 2 3 4–5 6 7 8 9 10 11 12 13 14 15 16-17 18 19–20 21 22 23 24 25 26 27 28–29 30 31 32 33–34 35 36

German Standard German KE KG

SA

CK

Rhinel. RT

Züritüütsch MM

PZ

RR

am

G+am

am,beim

am

G(+am)

am

am

am

/ 0,am

am,tun tun am

/ 0,am

am

/ 0,am tun am

/ 0,dabei am 0/ am 0/ 0/ am / dabei 0, / dabei 0, / dabei 0, / dabei 0, 0/

G G G G+am G G G+am G G

0/ am,beim 0/ am 0/ 0/ am am,beim 0/

am am am am 0/ 0/ am 0/ 0/

G(+dabei) G(+am) G G G G(+am) G(+am)

dabei ,tun am dabei am,tun

am,tun am,tun G,am / 0,G,tun tun G(+am),tun / 0,am 0/ G

/ 0,am / 0,am / 0,am 0/ tun 0/ / 0,am / 0,am 0/

G G

0/ beim

0/ 0/

/ 0,am,tun / 0,am,tun

0/ / 0,am

am N

G+am G G

am,im 0/

beim N

0/

0/

0/ 0/

0/ 0/

0/ beim,im 0/

0/ 0/ / ?0,am 0/ 0/ am

0/ 0/

0/ 0/ 0/ 0/

0/

0/ 0/ 0/ 0/ 0/ 0/ / 0,am 0/

0/ G+dabei G G G G G

am 0/

G+dabei G(+am,dabei) G(+dabei) G G G+am N G G+dabei G G+dabei G

/ 0,am am am tun tun dabei,tun dabei,tun am G

/ 0,am

G(+am) G(+tun) G(+)am

dabei,tun

G(+am)

0/

0/ G 0/ 0/ / 0,tun 0/

0/ / 0,G 0/

0/

/ 0,(am) 0/

0/ am,tun

0/

Karen H. Ebert

inform.

644

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

3.

645

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

/ 0,am am

0/ am

0/ 0/

am 0/

/ 0,am / 0,am

am tun

/ 0,am,tun / 0,tun

/ 0,am 0/

am am ABS ABS,beim

G+am G+am ABS ABS

am am beim beim

am beim,am beim 0/

G(+am) G(+am) ABS ABS

am,tun am,tun ABS ABS,tun

am,tun / am, 0,tun go go,tun

/ 0,am / 0,am go,am am

0/

0/

0/

0/

0/

am

0/

0/

0/

0/

/ 0,am

0/

tun

/ 0,am 0/ 0/ am (lag im)

0/ 0/ 0/ am am

0/ 0/ 0/ 0/ am

0/ 0/ dabei / 0,am (lag im)

/ 0,am 0/ 0/

am tun / 0,tun

0/ 0/ N

0/

0/ 0/ beim

0/ am 0/

0/ 0/ (went)

0/ 0/

G+am 0/

0/ am,beim

0/ am

/ 0,am 0/

0/ 0/

/ 0,am am am

0/

0/ 0/ 0/

0/ 0/ 0/

0/

am

am,tun am

am

am 0/ (went)

/ 0,tun 0/ go,am

go

0/ 0/

/ 0,am / 0,am

/ 0,am 0/

am 0/

am am

am 0/ (73:tun) 0/ 0/

am 0/

0/ am

/ 0,dabei 0/

/ 0,G am(PRS)

am

am am,tun

/ 0,am

am,beim

/ 0,am

beim,am

am

am,tun

am

0/

0/

am 0/

Progressive markers in Germanic languages

37 38 39–43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58–59 0/ 60–61 62 63 64–69 70, a b 71–75 76 77 78 79–80 81 82 83

645

646

646

Karen H. Ebert

Notes 1. Frisian is usually regarded as one language (cf. also EUROTYP guidelines), but the distance between Frysk (‘West Frisian’) and the North Frisian dialects is greater than that between for example Danish and Swedish. Fering and Öömrang, the dialects of the islands Föhr and Amrum, show only minimal dialectal variation and are treated together here. The distance between these island dialects and the mainland dialects or Sölring (spoken on the island Sylt) is comparable to that between Danish and Swedish. 2. Standard colloquial German is not codified, but it represents a style that is acceptable everywhere in half-formal speech, as will be heard for example in TV interviews or in seminars. The five questionnaires are from Slesvig-Holstein, Kassel, Frankfurt, Heidelberg, and Berlin. 3. See Bertinetto, Ebert, and De Groot, this volume, fn. 8 for an explanation of this term. 4. As the progressives have not been described as yet, answering the theoretical questions presupposed a detailed linguistic analysis of the facts, which is beyond the scope of lay people. (One consultant wrote that she would need six months’ work to answer the questions.) 5. The verb of the test sentence PROGQ: 1 (also PROGQ: 3) was an unhappy choice, as ‘be working’ is more often referred to by a nominal construction of the type Grm sie ist zur Arbeit, Dutch ze is aan het werk, Fering hat as tu werk ‘she is at work’. Otherwise one would specify the type of work, and some informants replaced the verb in PROGQ: 1 and 3. 6. The infinitive is used as a verbal noun in all languages considered here. Frisian has a long and a short form, e.g., FerÖöm köögi, tu köögin ‘cook, to cook’, luup, tu luupen ‘walk, to walk’. I have glossed nominal forms with an English gerund. The infinitive following am in German is treated as a noun in the standard orthography, though it hardly has any nominal properties; e.g., it can not take a genitive complement or an adjective, cf. sie war beim/*am Korrigieren der Seminararbeiten . . . she was AT_the grading DEF:PL:GEN seminar-works ‘she was grading papers’ Bhatt & Schmidt (1993) consequently omit the capital letters with the infinitive after am; cf. their examples in fn. 10. 7. Incorporation is not necessary in Frysk and Dutch, as the progressives can be combined with direct objects. Object incorporation is excluded with te in Dutch (except in the Groningen dialect; cf. Schuurman 1987) and therefore with the POS-construction; in Frysk it is optional. Dutch Frysk

Ze *zit te aardappel-schillen/ zit aardappelen te schillen Hja sit te jirpel-skilen/ sit jirpels te skilen she sits to potato-peeling/ sits potatoes to peel ‘She is peeling potatoes.’ (= PROGQ: 12)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

647

Progressive markers in Germanic languages

647

8. Cf. also the discussion in Geerts et al. (1984: 590), who find is aan het vlinders vangen ‘is catching butterflies’ fully acceptable, but ?is aan het olifanten vangen ‘is catching elephants’ doubtful. (Different from other sources, the authors write the incorporated noun as a separate word.) 9. German and Fering-Öömrang speakers resort to these constructions only if am is impossible for syntactic reasons, as in (5b). There is, however, much variation between individual speakers and much insecurity in the judgement of such sentences. Andersson (1989: 95) cites ich war dabei, einzuschlafen as Standard German, but many speakers find dabei impossible with naked verbs and especially with activities: *sie war dabei zu schreiben ( sie war dabei, Briefe zu schreiben). 10. In German this would be expressed without the verb ‘write’: Sie sitzt an ihrer Doktorarbeit (*zu schreiben). No such constructions exist for the other postural verbs. 11. Bhatt & Schmidt (1993: 74ff) give a number of examples as belonging to standard colloquial German, including progressives with definite objects and with postural verbs. None of the following examples, which they cite as ‘standard’ German, was accepted by my informants: a) der Pilot ist den Airbus am wegfliegen b) er ist die Kartoffeln roh am essen c) wir sind die Kinder am spielen lassen d) Noch ist das Bild am hängen, aber gleich fällt es herunter 12. Sentences like Züritüütsch (7) and (8) were rejected by some speakers. Cf. also Bickel (1992: 75), who states that “if objects have an article, they cannot be incorporated. Therefore, speakers have to resort to a two-nexus construction with draa sii ” 13. Most informants left out the ‘since’ phrase as it is odd to combine it with ‘the whole day’. As German seit and North Frisian sant ‘since’ are used with the present and not with the perfect, some of the progressives listed for PROGQ: 81 in Table 3 are present progressives. Fer Grm

Ik san sant jimaaren uun’t baagen. Ich bin seit heute morgen am Backen I am since this morning AT_the baking’ ‘I have been baking since morning’

14. There are few strictly momentaneous verbs (cf. also Comrie 1976: 47 on this point and the endless discussions of the verb ‘die’ in the literature). There is no consensus among German speakers whether einschlafen or sterben can refer to a period before one ‘really’ falls asleep or dies, or whether it refers only to the transition point. 15. Hålla på att has progressive or imminential meaning, hålla på och progressive or continuative. Holmes & Hinchcliffe (1994: 278) give the example: Hon höll på att gråta / Hon höll på och grät hela vägen. ‘She kept crying all the way.’ But only the second construction has the intended continuative reading according to my Swedish consultants. (There may be some confusion, because att and och are both pronounced [], but the constructions are clearly distinct, as the latter combines with an inflected verb.)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

648

648

Karen H. Ebert

16. I tentatively used the term ‘locomotive’ in a paper presented at the Amsterdam meeting of the EUROTYP Tense-Aspect Group, adducing examples mainly from languages outside of Europe (Ebert 1993 ms). I now call the grammaticized motion progressive ‘ambulative’ (cf. Ebert 1994c: 35ff for examples from Kiranti languages). Fortescue (1992) uses the term ‘perambulative’ for a similar construction in Koyukon. 17. This should of course not surprise us, as verbs of undirected motion function like ‘sit, stand’ etc. in the marking of durative and/or progressive in a variety of languages, e.g., Turkic, Mongolian, Kiranti languages; cf. also It., Span. andare  Gerundio. 18. HH and CG strongly preferred POS: HH has 19 POS vs. 11 PREP in Danish, CG has 19 POS vs. 2 PREP in Dutch. Interestingly, CG does not use POS in 3 cases, where the other Dutch informant, who has only 8 POS altogether, uses it: in PROGQ: 33, 34 with an emotive undertone, and in PROGQ: 37, where most languages have ‘stand to boil’. The Frysk expert, who listed all possible forms that came to mind, has 18 POS and 25 PREP-forms. Generally there is no confirmation that the PREP-form is the most common way to express progressive aspect (as claimed, e.g., in Donaldson 1981: 165f). 19. The Danish consultant confirmed this interpretation. Ingressive-phasal verbs are rare in European languages, but otherwise widespread (cf. Ebert 1995, where this type of verb was called inchoative-dynamic). 20. The factor dynamicity probably accounts for the high degree of variation with certain verbs, like ‘sleep’, ‘rain’. Progressives were sometimes accepted with ‘rain’ (PROGQ: 77), especially when not negated. Dan *Det er nu ikke ved at regne. FerÖöm *Det as nü ei uun’t riinen. / ??Det as uun’t riinen. Grm ??Es ist jetzt nicht am Regnen./ ?Es ist am Regnen. GrmRhi Es ist jetzt nicht am Regnen. it is now not AT_the/to raining ‘It is not raining now.’ / ‘It is raining.’ 21. I have a larger body of examples only for older stages of English, but cf. Gothic sat du aihtronds (lit.: ‘sat to begging’), OHGrm ir stantet zi betonne (lit.: ‘you stood to pray’). 22. I thank Östen Dahl for providing the examples from Karin Boye’s Kallocain. 23. Cf. the distinction between Bertinetto’s (in this vol., 1989–90) focalized and durative progressive, which are distinguished in the Italian stare  Gerundio vs. andare  Gerundio. However, the semantics of andare  Gerundio differs from that of the Germanic POS construction in that it does not normally combine with activity verbs. 24. Again tolerance for sentences like (33d) varies. Two FerÖöm informants used uun’t without imer in PROGQ: 2. 25. Heine (1994) postulates an erosion of the form am ( an dem) in the progressive as a result of grammaticization. However, the contracted form is most often also obligatory in locative phrases, as in the given example. 26. Compare A hünj hee a hiale daai lepen tu blafin. the dog has the whole day run:PP to barking ‘The dog has been running around barking the whole day.’ But: hi as lepen ‘he has walked’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

649

Progressive markers in Germanic languages

649

27. A detailed description for a European language is Schönig (1984): Tatar uses all kinds of postural verbs to specify durative actionality (as well as other verbs to specify telicity), but only the verb tur- ‘stand’ is used as a progressive marker. No Germanic language has generalized one verb, although van der Hauwe (1992: 13 fn) reports a tendency to generalize liggen in Dutch if there is no clear indication for using any of the other postural verbs. However, Dutch children sometimes generalize zitten. 28. Van der Hauwe reports that four Dutch students, who answered Dahl’s (1985) TMAquestionnaire, did not use a single simple form in the typical progressive contexts. All sentences contained the verb phrase ‘write a letter’. 29. Note that the Italian periphrasis andare  Gerundio, which corresponds to the Germanic POS-constructions in respect to durativity, is preferred in the written style. In former times it was more frequent in literary texts than stare  Gerundio (Bertinetto 1989–90: 38). 30. In the questionnaires there is one single occurrence of im by an informant from southern Germany, who also has an unusually high percentage of beim forms. All other persons I asked found her sentence totally unacceptable: ??Er war im Erreichen des Berggipfels. ‘He was reaching the top of the mountain.’ 31. I do not agree with Bybee et al. (1994: 133) that the English progressive is per se an appropriate anwer to a location inquiry. The given examples like “Where’s Lou? – He’s taking a bath (having a nap, etc.)” work only because the activities are associated with certain locations. 32. This has been suggested by Heine (1994, fn. 27). However, the habitual meaning may even be the more basic one. In Pennsylvania German and Yiddish ‘do’ is a marker of habitual present (as opposed to als and fleg in the past habitual).

PennGrm

Yid

si dut shtrige ‘she knits’ (habitually, for a living) ar hat si als gekent ‘he used to know her’ (Van Ness 1994: 435) ikh tu shraybn ‘I write’ ikh fleg shraybn ‘I used to write’ (Aronson 1985)

A progressive marker originating in ‘do’ is reported for Navajo (Blansitt 1975) and Syrian Arabic (Ebert, in this volume). 33. Danish ved at is here understood as ved og (PREP  ‘and’) due to the identical pronunciation of at and og as [], but is still followed by the infinitive. For the partial collapse of at and og, cf. Jespersen (1895), Haberland (1978). 34. The construction POS  to  INF was apparently more widespread in earlier German; cf. also Was steht ihr zu horchen? (lit.: what stand you to hark?) in Sanders’ “Handwörterbuch der deutschen Sprache” (cited from van der Gaaf 1934: 81).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

650

650

Karen H. Ebert

References Abraham, Werner & Theo Janssen (eds.) (1989 Tempus – Aspekt – Modus. Die lexikalischen und grammatischen Formen in den germanischen Sprachen. Tübingen: Niemeyer. Abraham, Werner & Josef Bayer (eds.) 1993 Dialektsyntax. Linguistische Berichte. Sonderheft 5. Andersson, Sven-Gunnar 1989 “On the Generalization of Progressive Constructions. ‘Ich bin (das Buch) am Lesen’ – status and usage in three varieties of German”, in: Lars-Gunnar Larsson (ed.), 95–106. Aronson, Howard I. 1985 “On aspect in Yiddish”, General Linguistics 25: 171–188. Askedal, John Ole 1994 “Norwegian”, in: Ekkehard König & Johan van der Auwera (eds.), 219–270. Barnes, Michael P. with Eivind Weyhe 1994 “Faroese”, in: Ekkehard König & Johan van der Auwera (eds.), 190–218. Bertinetto, Pier Marco 1989–90 “Le perifrasi verbali italiane: saggio di analisi descrittiva e contrastiva”, Quaderni Patavini Linguistica 8–9. Università di Padova. 27–64. this volume “The progressive in Romance, as compared with English”. Bertinetto, Pier Marco & Denis Delfitto this volume “Aspect vs. Actionality: Why they should be kept apart”. Bertinetto, Pier Marco, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.) 1995 Temporal reference, aspect and actionality. Vol. 2: Typological perspectives. Torino: Rosenberg & Sellier. Beukema, F. & P. Coopmans (eds.) 1987 Linguistics in the Netherlands. Dordrecht: Foris. Bhatt, Christa & Claudia Maria Schmidt 1993 “Die am  Infinitiv-Konstruktion im Kölnischen und im umgangssprachlichen Standarddeutschen als Aspekt-Phrasen”, in: Werner Abraham & Josef Bayer (eds.), 71–98. Bickel, Balthasar 1992 “The marking of future-time-reference in Züritüütsch”, in: Östen Dahl et al. (eds.), 73– 84. Blansitt, Edward L. 1975 “Progressive aspect”, Working Papers on Language Universals 18. Stanford University, 1–34. Blijkoningen, Frank & Ans van der Kemenade (eds.) 1991 Linguistics in the Netherlands 1991. Amsterdam: Benjamins. Bock, Karl Nielsen 1933 Niederdeutsch auf dänischem Substrat. Deutsche Dialektgeographie, Heft XXXIV. Boogaart, Ronny 1991 “Progressive Aspect in Dutch”, in: Frank Blijkoningen & Ans van der Kemenade (eds.), 1–9. Bybee, Joan & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13.1. 51–103. Bybee, Joan, Revere Perkins & William Pagliuca 1994 The evolution of Grammar. Tense, aspect, and modality in the languages of the world. Chicago: The University of Chicago Press. Comrie, Bernard 1976 Aspect. Cambridge University Press.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

651

Progressive markers in Germanic languages Dahl, Östen 1985 1992

651

Tense and aspect systems. Oxford: Blackwell. “The marking of future time reference in Continental Scandinavian”, in: Östen Dahl et al. (eds.), 60–72. Dahl, Östen, Casper de Groot & Hannu Tommola (eds.) 1992 Future time reference in European languages I. (E UROTYP Working Papers VI: 2). 1994 Future time reference in European languages II. (E UROTYP Working Papers VI: 3). Donaldson, Bruce C. 1981 Dutch reference grammar. The Hague: Martinus Nijhoff. Duden Grammatik Mannheim etc. 19844 Ebert, Karen H. 1989 “Aspektmarkierung im Fering (Nordfriesisch) und verwandten Sprachen”, in: Werner Abraham & Theo Janssen (eds.), 293–322. 1993 “Perfect and Progressives in Kiranti languages.” Paper read at the Amsterdam meeting of the E UROTYP Group 6. 1994a “Future time reference in Fering”, in: Östen Dahl et al. (eds.), 1–11. 1994b “Fering – eine todkranke Sprache?” In: Philologica Frisica Anno 1993. Ljouwert: Fryske Akademy. 9–28. 1994c The structure of Kiranti languages. Zürich: ASAS. 1995 “Ambiguous perfect-progressive forms across languages.” In: Pier Marco Bertinetto et al. (eds.), 185–203. 1996 “Progressive aspect in German and Dutch”, Journal of Germanic Languages and Literature 1.1. 41–62. this volume “Aspect in Maltese”. Ebert, Karen H. & Jarich Hoekstra 1996 “The progressive in West Frisian and North Frisian – similarities and areal differences”, in: Adeline Petersen & H. F. Nielsen (eds.), A Frisian and Germanic Miscellany. Odense & Bredstedt: Odense Universitets Forlag & Nordfriisk Instituut. 81–101. Erben, Johannes Deutsche Grammatik. München: Hueber. 197211 Fabricius-Hansen, Cathrine 1994 “Das dänische und norwegische Tempussystem im Vergleich mit dem deutschen”, in: Rolf Thieroff & Joachim Ballweg (eds.), 49–68. Fortescue, Michael 1992 “Aspect and Superaspect in Koyukon: An application of the Functional Grammar model to a polysynthetic language”, in: Michael Fortescue et al. (eds.), 99–141. Fortescue, Michael, Peter Harder & Lars Kristoffersen (eds.) 1992 Layered structure and reference in a functional perspective. Amsterdam: Benjamins. Gaaf, W. van der 1934 “The connection between verbs of rest (lie, sit, and stand) and another verb, viewed historically”, English Studies XVI: 81–99. Geerts, G., W. Haeseryn, J. de Rooij, M. C. van den Toorn 1984 Algemene Nederlandse Spraakkunst. Groningen: Wolters-Noordhoff. Groot, Casper de 1992 “The marking of future time reference in Dutch”, in: Östen Dahl et al. (eds.), 4–11. this volume “The Absentive”. Haberland, Hartmut 1978 “Eine Anmerkung über ‘Identität sprachlicher Einheiten’ ”, in: Dietrich Hartmann et al. (eds.), 57–67.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

652

652

Karen H. Ebert

Hansen, Aage 1967 Moderne Dansk III. Sprogbeskrivelse. København: Grafisk Forlag. Hartmann, Dietrich, Hansjürgen Linke & Otto Ludwig (eds.) 1978 Sprache in Gegenwart und Geschichte. Festschrift für Heinrich Matthias Heinrichs zum 65. Geburtstag. Köln & Wien: Böhlau Verlag. Heine, Bernd 1994 “Grammaticalization as an explanatory parameter”, in: William Pagliuca (ed.), 255– 287. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer 1991 Grammaticalization. A conceptual framework. Chicago: The University of Chicago Press. Holmes, Philip & Ian Hinchcliffe 1994 Swedish: A comprehensive grammar. London etc.: Routledge. Houwe, Jo van der 1992 “Progressive markers in a Functional Grammar of Dutch”, Working Papers in Functional Grammar, no. 8. Jespersen, Otto 1895 “En sproglig værdiforskydning. Og = at”, in: Dania 3: 145–182. Jones, W. Glyn & Kirstan Gade 1981 Danish. A Grammar. København: Gyldendal. Koefoed, H.A. 1969 Danish. London: Teach Yourself. König, Ekkehard & Johan van der Auwera (eds.) 1994 The Germanic Languages. London etc.: Routledge. Kress, Bruno 1982 Isländische Grammatik. München: Hueber. Larsson, Lars-Gunnar (ed.) 1989 Proceedings of the Second Scandinavian Symposium on Aspectology. Uppsala. Lockwood, W.B. 1964 An introduction to Modern Faroese. København: Munksgård. Marm, Ingvald & Alf Sommerfelt 1967 Norwegian. London: Teach Yourself. Pagliuca, William (ed.) 1994 Perspectives on grammaticalization. Amsterdam: Benjamins. Schönig, Claus 1984 Hilfsverben im Tatarischen. Wiesbaden: Steiner. Schuurman, Ineke 1987 “Incorporation in the Groningen dialect”, in: F. Beukema & P. Coopmans (eds.), 339– 350. Thieroff, Rolf 1992 Das finite Verb im Deutschen. Tempus, Modus, Distanz. Tübingen: Narr. Thieroff, Rolf, & Joachim Ballweg (eds.) 1994 Tense systems in European languages. Tübingen: Niemeyer. Tiersma, Pieter Meijes 1985 Frisian reference grammar. Dordrecht: Foris. Van Ness, Silke 1994 “Pennsylvania German”, in: Ekkehard König & Johan van der Auwera (eds.), 420–438. Visser, F. 1973 Historical syntax of English. Part III. Leiden. Weinreich, Uriel 1952 “Tsurik tsu aspektn”, Yidishe shprakh 12: 97–103.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

653

Progressive markers in Germanic languages

653

Text sources Bril, Martin & Dirk van Weelden 1991 Arbeidsvitaminen. Amsterdam: De Bezige Bij. Broos, Ton (ed.) n.d. Een Half Dozijn Nederlands: Zes Writers-in-Residence aan Amerikaanse Universiteiten. Ann Arbor: Q.E.D. Press. Jepsen, Hans Lyngby (ed.) 1964 Ny dansk prosa. Fra Tove Ditlevsen til Christian Kampmann. København: Stig Vendelkærs Forlag. Rifbjerg, Klaus 1957 Den kroniske uskyld. København: Gyldendal.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

654

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

655

Hannu Tommola

Progressive aspect in Baltic Finnic

1. Introduction The aim of this paper is to give a description of the expression of the progressive aspect in Baltic Finnic languages. Section 2 is devoted to a state of the art report on research into the meaning of a construction that is commonly regarded as progressive. An inventory of all devices is presented in Section 3. In Section 4 the empirical facts are put into a larger framework and compared with data from Germanic languages. In Section 5 restrictions on the use of progressive constructions are discussed and idiosyncratic and expanding tendencies sketched. In the concluding Section 6 some essential questions of grammaticalization are considered anew, and an attempt is made to define the status of the progressive forms in Estonian and Finnish.

1.1. Identification of the progressive meaning The kind of diagnostics as in (1) is often used to test grammatical aspect. The event denoted by the subordinate clause either overlaps with the activity denoted by the main clause or it does not. If this kind of aspect is grammaticalized, the overlap must be indicated, in English by the Progressive form, in Russian by using an Imperfective verb. (1)

Russian Kogda ja vo-šël, on cˇ isti-l/ o/po-ˇcisti-l revol’ver. when I in-go:PST he clean-PST PFV-clean-PST revolver ‘When I came in he was cleaning / cleaned the gun.’

In Finnish, the simple Past form can be used in both instances, and the difference in meaning is marked by the direct object case (2b). The same applies, as a rule, to Estonian. However, the Partitive and Genitive endings may coincide, as in (2a), and that is why a perfectivizing particle (here ära ‘away’) is frequently compulsory. Since the object case marking is not available in intransitive clauses, but the Estonian type particle is, the reason why progressive constructions seem to be more frequent in Finnish than in Estonian might be sought in the fact that Estonian more consistently marks perfectives. Moreover, sometimes transitive clauses with simple tense forms

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

656

656

Hannu Tommola

are also ambiguous, because the Partitive can be triggered by something other than aspectual functions, and the progressive (ongoing activity) meaning is unambiguously expressed only by the 3rd Infinitive Inessive construction as in (3), although it is not compulsory. (2) a.

b.

(3)

Estonian Kui ma tuppa tul-i-n, puhasta-s ta when I room:ILL come-PST-1SG clean-PST PRON:3SG püssi. / püssi ära. gun:PRTV / gun:GEN away Finnish Kun tul-i-n sisään, hän puhdist-i pyssy-ä / when come-PST-1SG in PRON:3SG clean-PST gun-PRTV / pyssy-n. gun-GEN ‘When I came in he was cleaning / cleaned the gun.’ Finnish (PROGQ: 83) [Jos tulet kahdeksalta,] laita-n / ole-n vielä make-1SG COP-1SG still laitta-ma-ssa ruoka-a. make-3INF-INESS food-PRTV ‘[If you come at 8 o’clock,] I will still be cooking.’

Of course, the fact that the English Progressive and the Russian Imperfective being used in (1) coincide with ongoing activity meaning does not prove that they represent the same gram-type. In order to confirm the specifically progressive meaning, another test that differentiates the ongoing meaning from habitual or generic meanings (e.g., 4) is needed. While it is well known that in Slavic there is no grammatical progressive, at least a common pragmatic interpretation of the determinate motion verbs is an ongoing process, as shown in (4a). In Finnish (4b) the 3rd Infinitive Inessive construction would be possible only if a goal (e.g., ‘to the South’) is mentioned. (4) a.

b.

Russian Ptic-y leta-jut. / Let-jat žuravl-i! bird-PL fly-3PL fly-3PL crane-PL Finnish Linnu-t lentä-vät. / Kurje-t lentävät! bird-PL fly-3PL crane-PL fly-3PL Kurje-t ?o-vat lentä-mä-ssä (etelä-än)! crane-PL COP-3PL fly-3INF-INESS (south-ILL) ‘Birds fly.’ ‘Cranes are flying!’ (‘to the South’)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

657

Progressive aspect in Baltic Finnic

657

1.2. Preliminaries: From locative to imminential The Baltic Finnic candidate for a progressive construction used in (3) fits nicely into the worldwide tendency of progressive markers to be expressed periphrastically, and to originate from locative sources (Comrie 1976: 98–103, Dahl 1985: 90–91, Bybee et al. 1994: 128–133). In a copula clause denoting location the infinitive can take the position of a noun in the same case form. Cf.: (5)

Estonian Maret on söökla-s / söö-ma-s. Mary COP café-INESS eat-mINF-INESS ‘Mary is in the café / eating (= having her lunch).’

In this use the construction does not actually refer to ongoing action (while the ongoing reading is possible), but to an activity that the subject is supposed to be engaged in and that takes place in a location that may somehow be pragmatically predictable. On the contrary, the ongoing reading is excluded in the imminential meaning, which progressives tend to have as well, when the verb phrase involved in the construction is an accomplishment or an achievement. Synchronically, there is no evidence of a connection with locative semantics or adverbials of place used in the same syntactic position: (6)

Estonian Pomm on kabineti-s. / plahvata-ma-s. bomb COP office-INESS explode-mINF-INESS ‘The bomb is in the office / (on the verge of) exploding.’

Even though this construction is not fully grammaticalized as a progressive gram, as seen in (3–4), it is regarded here as the main Progressive construction in Baltic Finnic and called in this paper INFPROG. It is regular and productive in uses like (5) and (6), i.e., in locative (‘to be somewhere at reference time [in order to V]’) and imminential (‘to be on the verge of V-ing at reference time’) meanings.1 In Dahl’s (1985: 91) description the Finnish 3rd Infinitive Inessive construction was characterized as “ PROG”, i.e., with several occurrences in relevant contexts it could be labelled as a candidate for a “PROG” gram, while it did not occur in the items that were regarded as prototypical (see section 4). Estonian data did not justify any ranking at all. Apparently, in the Baltic Finnic languages the use of INFPROG, due to its unstabilized grammatical status, will vary from individual to individual, and generating reliable statistics would require a much larger group of informants for each language (and hard work in designing and organizing the test so as to ensure the comparability of the responses). The situation seems to be similar in most Germanic languages as well.2

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

658

658

Hannu Tommola

The locative use of the Baltic Finnic INFPROG in (5) is an instance of a separate absentive gram-type, introduced by De Groot (this volume). The notion of absentive is of great importance for the discussion of the progressive, because absentive turns out to overlap with progressive semantics. It is not surprising that – unless absentive is grammaticalized – a progressive construction can denote absentive situations (e.g., in English and Icelandic). In Baltic Finnic languages the connection is obvious. However, INFPROG allows an absentive interpretation but does not require absence of the subject from the deictic centre (see below).

2. Research into the Progressive in Baltic Finnic Both in Finnish and in Estonian INFPROG consists of the copula olla ‘be’ as the finite verb and an infinitive stem of the main verb in the Inessive case. A morphologically identical construction also exists in the eastern Baltic Finnic languages (Karelian, Livonian, Vepsian) and in Sami (Tauli 1966: 78, Serebrennikov 1963). The construction was used in Old Literary Finnish from the middle of the 16th century (Häkkinen 1994: 311), but, unfortunately, it is hard to say anything about its functions in older periods.

2.1. Finnish In general works on Finnish the copula  3rd Infinitive Inessive has been regarded as a construction that is used to express ongoing activity (Heinämäki 1995 refers to Setälä 1973: 113, Siro 1975: 32–33, Hakulinen & Karlsson 1979: 381–382). Häkkinen (1994: 395) speaks of “a kind of continuous form” occurring in the oldest literary period. In her examples from Agricola – the translator of the first Finnish bible – locative adverbials play a role, except in the sentence given in (7). (7)

Finnish (16th century) coska hen yxinens ol-i rucole-ma-s when he alone COP-PST pray-3INF-INESS ‘when he was praying alone.’

Here semantic elements that I will call purposive are involved.3 According to Häkkinen (1994: 396), the 3rd Infinitive Inessive becomes more frequent in texts of the 19th century, and she also records a qualitative expansion: the imminential meaning, as well as an adverbial use, which she does not characterize in semantic terms. The latter use is clearly locative and/or purposive, and its possible novelty in the literary language does not necessarily mean that it was new in the spoken language.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

659

Progressive aspect in Baltic Finnic

659

Some investigators take the locative meaning as basic (e.g., Kangasmaa-Minn 1978: 286–287 and Markkanen 1979: 65–67). Hyvärinen (1982) expressly attaches progressivity even to the locative use.4 It has also been pointed out that the construction performs an emphatic function (Penttilä 1963: 384), especially when it is used in the Perfect or Pluperfect (Serebrennikov 1963). Metslang (1993: 332) has done research into the use of INFPROG in Finnish dialects and found that it is not very frequent (below one per cent of the predicates), but regular.5 Most of the occurrences attest to a locative interpretation, but there are also clear cases of the progressive meaning. The statistics from Modern Finnish presented in Hyvönen & Jämsä (1978) reflects the usage twenty to thirty years ago. The average frequency remains at a 0.2 per cent level, while the data emphasize a rather marginal type of use (6.2 percent of all finite forms in captions). Of a similar type is the frequent use of the Progressive in newspaper headlines, as a rule lacking the copula (cf. Tommola 1984: 115, 131 n. 17, and (30) below). Heinämäki (1995) concludes her recent paper by emphasizing two types of conditions for the use of INFPROG. The first condition is the dependence on the aspectual type of the verb. This is common for the progressive grams in general. The other central idea of Heinämäki seems to be language-specific: the requirement that the activity be purposeful. However, it applies very well to all absentives and thus may be common for the young locative-based progressives. In Tommola (1986: 225–232) I tried to show that the locative functions of the 3rd Infinitive in Inessive, Elative and Illative cases have exact parallels in aspectualtemporal use: the Inessive construction (INFPROG) can denote ongoing activity or process, while the 3rd Infinitive in Elative and Illative is used as a complement of the verbs of ‘stopping’ and ‘beginning’, respectively.6 In some contexts a locative element is still present in the semantics of the progressive grams (Bybee and al. 1994: 133; cf. “persistence of meaning” in Hopper & Traugott 1993: 3). Nevertheless, the characterization ‘being in an activity’ must be taken as a metaphor. For INFPROG to be included in the tense-aspect system, it is crucial that the locative interpretation can be proved impossible (cf. Tommola 1986: 229–230). While both readings may coincide, the progressive meaning can only be verified, if the locative reading is excluded. In (8) and (9) from (Heinämäki 1981: 383) and (Tommola 1986: 231) no locative interpretation is needed:

(8)

Finnish Ol-i-n ajattele-ma-ssa asia-a. COP-PST-1SG think-3INF-INESS thing-PRTV ‘I was thinking of it / the matter.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

660

660 (9)

Hannu Tommola

Finnish [En kuitenkaan puutu tähän periaatteelliseen puoleen, koska] si-tä o-vat jo muu-t pohti-ma-ssa. it-PRTV COP-3PL already other-PL ponder-3INF-INESS ‘[However, I’ll not consider it as a matter of principle, because] there are others already discussing it.’

2.2. Estonian Less scholarly interest has been devoted to the Estonian mas-Infinitive construction. Erelt (1985, 1987) and Metslang (1993) are the only linguists to have seriously tackled the question. Metslang has undertaken a thorough investigation of diverse factors that may influence the use of the mas-form. Her list of felicitous contexts (Metslang 1993: 474) includes: 1) perfective7 verbs (imminential meaning), 2) verbs denoting gradual change, 3) unmarked verbal categories (active voice, indicative mood etc.), 4) existential sentences and rhematic subject, 5) nonfocused position of the construction, 6) iterative or distributive context, 7) coordination, parallelism, dialogue, 8) polarization with local adverbials, 9) sentences expressing a temporal frame, 10) backgrounding in narrative, and 11) reportage. One of Metslang’s four nonfavourable factors is reference to “imperfective dynamic situations, notably agentive”8. It clearly does not hold for Finnish, nor does it fit into the picture we have of the prototypical progressives in general, and of the development of the English Progressive (Dahl 1985: 91–92, Bybee et al. 1994: 132–137). According to Erelt (1987: 45–46) the mas-form has to be analyzed depending on the verb semantics: 1) as a locative adverb when it denotes durative activity (ta on söömas = ta on sööma läinud ‘s/he’s gone to eat’; 2) as expressing progressive aspect when it denotes a durative process (kellad on helisemas = kellad helisevad parajasti ‘the bells are ringing right now’); and 3) as expressing immediate future when it denotes a “momentaneous process” (kontsert on algamas = kontsert algab kohe ‘the concert will begin soon’). These are the three types of reading the construction will have both in Finnish and in Estonian: locative (absentive/purposive), progressive, and imminential. Even if the use of INFPROG cannot be delimited so neatly to comprise only (nonactional) processes, the Estonian mas-construction differs from many progressive constructions in allowing agentless processes with progressive reading.

3. The expression of the progressive and related meanings If we do not count time adverbials like ‘right now’ and some frequentative verbs in Finnish, eight different constructions in Finnish and five in Estonian were used, or could have been used in the PROGQ sentences. The devices are distributed among

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

661

Progressive aspect in Baltic Finnic

661

the meanings they denote as follows: the standard copula  infinitive Inessive (= INFPROG) construction is 1) an all-round marker that can denote all the meanings relevant here. The rest can be divided into 2) devices denoting both the progressive and the imminential meaning (copula  noun in a locative case = NOUNPROG; time adverbials), 3) devices of imminence (the Finnish copula  infinitive constructions; the Estonian ‘begin’ construction), 4) progressive and locative markers (postural verb constructions), and 5) progressive/continuative markers (just marginally: the Estonian serial constructions, motion verb  infinitive Inessive construction). Additionally there is 6) a construction with locative interpretation (‘go’  infinitive Inessive). See Table 1 for an overview of the meanings of the most important markers.

3.1. Copula  infinitive in the Inessive (= INFPROG) The forms in which the main verb shows up in the INFPROG construction in Finnish and Estonian are treated differently in traditional grammars. In Finnish, the form in question is called the 3rd Infinitive (suffix -ma-/-mä-) and the Inessive case ending is -ssa/-ssä. In Estonian grammar the corresponding periphrastic form is usually called the mas-construction, where -s is the Inessive marker.9 The 3rd Infinitive Inessive construction is, in most cases, not obligatory in Finnish. In PROGQ it was the only choice in 4 items: PROGQ: 44, PROGQ: 47 (= 27), PROGQ: 79 (= 21) and PROGQ: 80. It is important to observe that these are all absentives. In the ongoing meaning INFPROG is not compulsory even in the “incidential schema”10, although possible, and at least preferred in PROGQ: 75 (= 11); and also in PROGQ: 32 (= 41) with an achievement verb. In the meaning of gradual change (PROGQ: 54 = 19 and PROGQ: 55 = 42), it is also difficult to choose anything other than INFPROG. It – or the semantically equivalent construction NOUNPROG described in 3.2 – is preferable in a total of eighteen items (22 percent), and possible in 47 items (57 percent). The Estonian mas-construction is compulsory in PROGQ: 55 with a gradual process meaning, and in PROGQ: 57 (= 20) with an imminential meaning.11 The other items where the construction can be used without hesitation are PROGQ: 23–25 with phasal verbs, PROGQ: 56 (= 24), PROGQ: 67, PROGQ: 31 and PROGQ: 32 (= 41) with achievement verbs in the imminential meaning, and PROGQ: 53 as well as PROGQ: 54 (= 19) with the gradual process meaning.

3.2. Copula  noun in a locative case (= NOUNPROG) In both Finnish and Estonian a noun with a verbal stem can replace the Infinitive form in the Progressive construction. In Finnish the noun normally takes the Inessive just like the 3rd Infinitive in INFPROG, whereas in Estonian the noun is in the Adessive

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

662

662

Hannu Tommola

case. In Finnish the construction (= NOUNPROG) is regular, replacing, as a rule, INFPROG from certain verbs. In Estonian it is more sporadic. (10)

Finnish (PROGQ: 21) Hän on meno-ssa / lähdö-ssä ulos. PRON:3SG COP going-INESS departure-INESS out ‘She is going out [right now; do you want me to hold her back?]’

(11)

Estonian (PROGQ: 75) Toit on just serveerimise-l. food COP just serving-ADESS ‘[Come in, please!] The meal is being served [right now].’

The NOUNPROG construction was used in seven items of the PROGQ in Finnish. Its distribution is complementary to INFPROG, as it has gained ground and superseded this with the common most motion verbs. The relevant nouns are: tulo ‘coming’ tulla ‘come’; meno ‘going’ mennä ‘go’; lähtö ‘departure, leaving’ lähteä ‘leave (for), depart’ (see 10, 24–26, 30).12 Infinitives denoting gradual change are also often replaced by nouns in the Progressive construction (lasku laskea ‘drop, fall, decrease’, nousu nousta ‘rise, increase’; PROGQ: 53). Because of the lexical semantics involved the NOUNPROG construction has no locative reading. If INFPROG has evolved from a locative expression and is still used with absentive/purposive meaning NOUNPROG has never been ambiguous in this sense. Another reason why the construction is so frequent is simply that these verbs (and the corresponding nouns) belong to the core lexicon.13 Otherwise, there is apparently no semantic difference between INFPROG and NOUNPROG, no matter whether they are used to mark 1) ongoing process or motion (towards a goal), 2) imminence, or 3) event with future time reference (Tommola 1992b: 16). In this paper, the term “Progressive construction” is used to refer not only to INFPROG, but to NOUNPROG as well, unless it is necessary to keep them apart.14 In Estonian, an Adessive nominal expression corresponds in progressive/imminential meaning to the Finnish construction with the noun in the Inessive, e.g., mineku-l/ tuleku-l olema ‘be going, leaving / coming’ (see Metslang 1993: 333). It was used in the PROGQ two times and, not being restricted to the nouns from motion verbs, seems to some extent productive. It also performs a complementary function, for the mas-construction is avoided in marked forms (Metslang 1993: 474), as in (indefinitepersonal) Passive (11). As mentioned above, NOUNPROG is impossible in the absentive meaning. From NOUNPROG with nomina actionis15 are to be distinguished formally identical lexical expressions employing nomina acti (results of the action etc.) and other nouns.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

663

Progressive aspect in Baltic Finnic

663

In absentive contexts adverbials with a noun in the Adessive (or Inessive) case combined with the copula are common, in Finnish sometimes with the noun in the plural (12).16 (12)

Finnish (PROGQ: 46) Hän on ostoks-i-lla. PRON:3SG COP purchase-PL-ADESS ‘She is shopping.’

(13)

Finnish (PROGQ: 22) Hän on matka-lla New York-iin. PRON:3SG COP trip-ADESS New York-ILL ‘She is on her way to New York.’

However, sentences (12) and (13) can be used not only to answer the question ‘Where is (s)he right now?’ but also ‘What is (s)he doing here?’, uttered, for example, in a department store or simply on the street (12) or at an airport (13). This fact suggests that absentive is a special case of purposive, rather than vice versa.

3.3. Imminential constructions (14)

Finnish (PROGQ: 57) Vanha mies ol-i kuole-maisi-lla-an. old man COP-PST die-NR:PL-ADESS-POSS ‘The old man was dying [but they finally found the right medicine].’

(15)

Finnish Ol-i-n kuol-la nauru-un. COP-PST-1SG die-INF laughter-ILL ‘I was dying of laughter.’

In Finnish, the construction olla  -maisi-lla-an / -mäisi-llä-än (‘to be on the verge of’) with the copula and the so-called 5th Infinitive (which can be analysed as a diminutive verbal noun in the Adessive Plural, always with a Possessive suffix) is a specific imminential marker. With past time reference it normally denotes a situation where the event presented as imminent did not take place (14).17 In Estonian the standard INFPROG is used (see 20). Another Finnish imminential construction, olla (vähällä / lähellä) ‘be (near)’  1st Infinitive, is used primarily in Past tense, with implicit reference to negation of the event imminent at the point of reference (see Tommola 1992b: 18). In metaphoric use a progressive reading is sometimes possible (15).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

664

664

Hannu Tommola

The construction hakkama ‘begin’  ma-Infinitive in Illative is the most frequent explicit device to express future time reference in Estonian (Tommola 1992a: 12–13, 20). It also seems to compete with INFPROG as an imminence marker (PROGQ: 21 and 78).18

3.4. Postural and motion verb constructions: serial and infinitive Postural and motion verbs can be used in an infinitive construction similar to INFPROG where they replace the copula as auxiliary. However, they always retain their lexical meaning. They are also occasionally used in the same way as in some Germanic languages to build a serial construction (16a). In Finnish, motion verbs are seldom found in the serial construction, and a sentence corresponding to the Estonian one would hardly be possible in Finnish without an overt locative expression. The construction with the 3rd Infinitive is preferred instead, and a negative attitude of the speaker to the activity may be implied (16b).19 (16) a.

b.

Estonian Ta jookse-b ja räägi-b tühja juttu. PRON:3SG run-3SG and talk-3SG empty:PRTV talk:PRTV Finnish Hän juokse-e (kylä-llä) kerto-ma-ssa PRON:3SG run-3SG (village-ADESS) tell-3INF-INESS juoru-j-a. gossip-PL-PRTV ‘He/She is running around (the village) spreading gossip.’

3.5. Käydä / käima ‘visit’  infinitive Inessive (17)

Finnish Käv-i-n eilen pelaa-ma-ssa kössi-ä. go-PST-1SG yesterday play-3INF-INESS squash-PRTV ‘Yesterday I went to play squash.’

The motion verbs Fin. käydä, Est. käima combine with the infinitive Inessive form in their meaning ‘go and come back, come and go away, visit’. However, because ‘going’ and ‘coming back’ cannot take place simultaneously, this construction is not progressive in meaning. Moreover, it can be regarded as a dynamic counterpart of INFPROG used in the locative meaning or as a device rendering a situation denoted by the corresponding copula construction (INFPROG) bounded. The construction occurs once in PROGQ (18), and it is noteworthy that if INFPROG is used here, it has the same – in this case, habitual and absentive – reading (Heinämäki 1995).20

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

665

Progressive aspect in Baltic Finnic

(18) a.

665

Estonian (PROGQ: 63) Tollal kä-is/ol-i ta iga-l laupäeva-l then go/COP-PST PRON:3SG every-ADESS Saturday-ADESS tantsi-ma-s. dance-mINF-INESS ‘At that time, he went to dance / was away dancing every Saturday.’

3.6. Other devices In Finnish, the simple forms of the verbs with a Frequentative suffix may get a progressive/imminential interpretation. The Frequentatives from the phasals lopet-ella lopettaa ‘finish’ (see 59) and aloit-ella aloittaa ‘begin’ seem to have primarily this meaning, rather than a frequentative one.21 Typical adverbs that co-occur with the progressive meaning are ‘just’, ‘right now/ then’, ‘already’, and ‘still’. As long as the progressive meaning is not grammaticalized, the adverbs Fin. juuri (nyt/silloin), parhaillaan, parastaikaa and Est. just, praegu, parajasti ‘just, right (now/then)’ are sufficient markers of the ongoing process meaning.22 An overview of the meanings that are marked by the different devices is presented in Table 1.23

4. The uses of the Baltic Finnic Progressive constructions From Table 1 it can be seen that the most important progressive marker in Finnish and Estonian is, indeed, INFPROG. Its semantic scope comprises a large set of meanings allowing generalization that is necessary for grammaticalization. This Section is devoted to an establishment of the status of INFPROG (including, marginally, also that of NOUNPROG). A comparison of the Finnish data from PROGQ and from Dahl (1985) shows that for a frequency analysis of the distribution of INFPROG, the number of informants should be increased and/or large corpus investigations undertaken. In Dahl’s data most of INFPROG occurrences are absentives. There is only one item (1402: ‘[When I came home yesterday,] he was writing two letters.’) where it is to be regarded as primarily progressive. One might wonder why the informant did not use the same construction in, for example, items 91 and 101 (“he WRITE letters / a letter”), which are similar situations, and are higher in the ranking list for the PROG (Dahl 1985: 92). I think that it can be explained by the fact that in both these latter items the situation is conceived of as a homogenous process (the activity the subject is engaged in is writing, and it does not matter if it is one or more letters he is going to write), whereas in 1402 mentioning the number of letters distributes the activity in time:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

666

666

Hannu Tommola

Table 1. Markers of the progressive and related meanings in Finnish and Estonian

Progressive Processive Continuous Stative

  ( ) ( )

  ()  

Absentive Presentive

   

Functions (N)

6

4

Imminential FTR

( ) ( )



  ( ) ()



1

2



SERIAL ( ) ()

( ) ()



 

2

2

( )

 

IMM

NOUNPROG

INFPROG

FREQ

Estonian

LOC

IMM1

NOUNPROG

INFPROG

Finnish

  (1)

5

(1)

Devices INFPROG NOUNPROG IMM1 LOC FREQ

Finnish olla V-mAssA olla N-ssA olla V-mAisillAolla N-llA/ssA VFREQ

INFPROG NOUNPROG IMM SERIAL

Estonian olema V-mas olema N-l hakkama V-ma VFINpostural/motion ja VFIN

most likely the first letter should be finished before the second one is written. It can be taken for granted that speakers of Finnish use the simple Past here, too, but then they have to be prepared to answer the ironic question: “Really? Simultaneously?”. We see that even here, strictly speaking, the Finnish INFPROG does not only refer to the activity going on just at the point of reference. Moreover, it expresses (purposive) prospectivity (‘He was writing in order to finish two letters.’).24

4.1. A contrastive Northern Germanic–Baltic Finnic look According to Dahl’s (1985: 90–91) analysis the Swedish Progressive construction hålla på att can be regarded as quite a typical PROG gram, whereas the Finnish 3rd Infinitive Inessive was not found to qualify for such a status. It is well known that the English Progressive has too wide a range to qualify as the most typical progressive

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

667

Progressive aspect in Baltic Finnic

667

construction (e.g., Comrie 1976: 33, 36–38, Dahl 1985: 93–94). The plain frequency was, of course, not decisive in the evaluation of the gram status, so English with 39 occurrences received a lower ranking than Swedish with sixteen occurrences. What would perhaps be expected is that Swedish resembles English in marking progressivity more so than Finnish, which yielded only seven occurrences. Considering the PROGQ sentences we can state that the English Progressive was given as the only alternative in roughly 70 percent of the items (and was considered possible in more than 90 percent), the Icelandic vera að in almost 50 percent (possible in nearly 60 percent). Finnish and Swedish lag far behind, with about twelve per cent each, and in Estonian INFPROG was the only choice in just three items (3.5 percent). Consequently, Finnish and Swedish seem to use their Progressive constructions, roughly, equally often. But we get a strikingly different picture if we look at the distribution of the constructions among the questionnaire items. The Swedish construction hålla på (att/och) seems to be preferred in PROGQ: 14 through 20, and is at least common in (PROGQ: 79 = ex. 21, and PROGQ: 81 = 57). In Finnish, non-Progressive forms are preferred in (PROGQ: 14–20), whereas INFPROG is the only choice in (PROGQ: 80) with epistemic modals, but is not used in (PROGQ: 81). In Estonian the only occurrence among these items was in PROGQ: 19 (= 28), as an alternative to the simple form. On the other hand, a Progressive construction was often used in contexts where Swedish has no Progressive marker. While these languages share the Progressive use in more than thirty items, there are approximately twenty instances where the Progressive use does not coincide in Swedish and Finnish. What are the conditions for the use in both languages? To begin with, let us look at some of the contexts where English, Icelandic, Swedish, Finnish, and Estonian all use the Progressive (PROGQ: 19, 24–26, 54, 56, 57, 79), e.g.:25 (19) a.

b.

(20) a.

(PROGQ: 54) Swedish Situation-en höll på att bli gradvis bättre. situation-DEF keep:PST Prep INFM become gradually better Estonian Olukord ol-i juba vähehaaval parane-ma-s. situation COP-PST already gradually improve-mINF-INESS ‘[When I arrived] the situation was already improving little by little.’ (PROGQ: 57) Swedish Den gaml-e man-nen höll på att dö. DEF old-DEF man-DEF keep:PST Prep INFM die

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

668

668

Hannu Tommola

b.

(21) a.

b.

Estonian Vana mees ol-i sure-ma-s [kuid lõpuks leiti õige rohi]. old man COP-PST die-3INF-INESS ‘The old man was dying [but they finally found the right medicine].’ (PROGQ: 79) Swedish Tom måste hålla på och mata djur-en. Tom must keep Prep and feed animal-PL:DEF Estonian Tom pea-ks praegu loom-i sööt-ma-s Tom must-COND right_now animal-PL:PRTV feed-mINF-INESS ole-ma. COP-mINF ‘Tom must be feeding the animals [I guess].’

In (19) a gradual change is reported as ‘already’ being in the process. In (20) a typical imminential meaning is demonstrated, whereas (21) is an absentive. Now, consider the following examples of the use of the Swedish serial construction with postural or motion verbs (22–23), which display situations not typical of the Finnic Progressive: (22)

Swedish (PROGQ: 48) Ann lek-te / sprang och lek-te i två timm-ar Ann play-PST ran:PST and play-PST Prep two hour-PL alldeles för sig själv. entirely for RFL self ‘[Yesterday, during my sleep] Ann played / was playing two hours all by herself.’

(23)

Swedish (PROGQ: 49) Ann prata-de / satt och prata-de med sin granne. Ann talk-PST sit:PST and talk-PST with POSS neighbour ‘[During the whole time of prayer] Ann talked / was talking to her neighbour.’

What seems to prevent the use of the Finnic construction is that there is an element of duration in these examples (observe that the Progressive is not compulsory in English, either). The continuous reading has been regarded as typical of both Progressive devices in Swedish, the hålla på (att/och) construction and the serial construction. However, there are two different hålla på constructions, each with its own semantics (see Ebert, this volume). I will take this issue up later (section 6.2).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

669

Progressive aspect in Baltic Finnic

669

Let us next take a look at those contexts where Finnish – and possibly Estonian – but not Swedish Progressive constructions are used. While Progressives are possible both in the Baltic Finnic and in the Scandinavian languages in imminential meaning (24), they are allowed in a future meaning (with reference to ‘tomorrow’) only in Finnish and in Icelandic (25–26): (24) a.

b.

c.

d.

(25) a.

b.

(26) a.

b.

(PROGQ: 56) Swedish Tåg-et håll-er på att train-DEF keep-PRS Prep INFM Icelandic Lest-in er að fara. train-DEF COP at drive Estonian Rong välju-b kohe. / Rong train leave-3SG at_once train Finnish Juna on lähdö-ssä. train COP departure-INESS ‘[Hurry up!] The train is leaving.’

gå. go

on välju-ma-s. COP leave-mINF-INESS

(PROGQ: 66) Icelandic Anna er að fara á morgun. Ann COP at drive tomorrow Finnish Anne lähte-e / on lähdö-ssä huomenna. Anne leave-3SG COP departure-INESS tomorrow ‘Ann will leave / is leaving tomorrow.’ (PROGQ: 68) Icelandic Jón er að gifta sig á morgun. John COP at marry RFL tomorrow Finnish Jussi mene-e / on meno-ssa naimisiin huomenna. John go-3SG COP going-INESS marrying:ILL tomorrow ‘John’ll get married / John is getting married tomorrow.’

In (21) we already had examples of Progressives used in absentive meaning (with an epistemic reading). Sentences (27) and (28) are further examples of such a use,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

670

670

Hannu Tommola

which is common in English, Icelandic, Finnish and Estonian, whereas in Continental Scandinavian and in the other Germanic languages a separate absentive construction tends to be used. (27) a.

b.

(28) a.

b.

(PROGQ: 47) Finnish Hän on kerho-lla pelaa-ma-ssa kortti-a. PRON:3SG COP club-ADESS play-3INF-INESS card-PRTV Swedish Hon är och spel-ar kort på klubb-en. she COP and play-PRS card on club-DEF ‘She is playing cards in the club [as usual].’ (PROGQ: 19) Estonian Ta on juukse-id lõika-ma-s. PRON:3SG COP hair-PL:PRTV cut-mINF-INESS Swedish Hon är och bli-r klipp-t. she COP and become-PRS cut-PART:PF ‘She is having her hair cut [right now].’

But, consider also following example, which does not necessarily have an absence reading: (29) a.

b.

Finnish Vesi on kiehu-ma-ssa. water COP boil-3INF-INESS Estonian Vesi on kee-ma-s. water COP boil-mINF-INESS ‘The water is being heated [to make it boil].’

4.2. Preliminary conclusions To summarize the above, the first set of examples (19–21) suggests three main functions that the Progressive constructions of all these languages share: progressive = gradual process (19), imminential (20), and absentive (21). The further four sets of examples (22–23); (24–26); (27–28); (29) can be claimed to justify following conclusions: 1. The Baltic Finnic Progressive does not typically combine with expressions of duration (22–23), unless an absentive reading is allowed;

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

671

Progressive aspect in Baltic Finnic

671

2. It typically denotes either a dynamic situation which is developing towards some kind of change (19) or a situation where a change (event) is expected (20, 24–26); 3. It is largely used in an absentive function (21) and (27–28); 4. It is also used to refer to situations where the process denoted by the verb is not going on, although they cannot be interpreted as imminential or absentive either. This is the purposive reading of (29). 5. In spite of its limited (optional) use in prototypical progressive contexts, it is nevertheless a Progressive, sharing extended uses with, for instance, the Icelandic and English Progressives. It differs from those in its purposive semantics, which can be seen as the single main obstacle to its grammaticalization.

5. Tendencies in the Baltic Finnic Progressive The Baltic Finnic INFPROG has turned out to be essentially something other than a prototypical progressive. The available data suggest that the construction is obligatory in imminential and absentive (purposive) meanings. It is impossible to tell where the limits are for the progressive use, because the system is in a state of change. One can simply try to show what the motivations are that trigger the seemingly idiosyncratic uses. In Finnish newspaper headlines, occurrences of INFPROG and NOUNPROG constructions with omitted auxiliary (finite) verbs are extremely numerous. The usual reading, then, is future time reference (30): (30)

Finnish Aittoniemi lähdö-ssä Smp:stä Keskusta-an Aittoniemi departure-INESS Smp-ELAT Centre-ILL ‘Aittoniemi [is] leaving the SMP [Finland’s Countryside Party] for the [Party of the Finnish] Centre.’

Example (30) conveys that the subject intends to leave the party, probably very soon (intention-based uses are also given in (25) and (26)). But nonintentional actions are also referred to by the Progressive construction. In (31a) with a performative verb the motivation of INFPROG may be seen as suggesting unintentionality: indirect ‘admitting’ is seen as a (result of the) process that the speaker is observing, while the subject does not realize it.26 (31) a.

(PROGQ: 34) Finnish Philip-hän on myöntä-mä-ssä syyllisyyte-nsä! Philip-PRAGM COP admit-3INF-INESS guilt-GEN:POSS

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

672

672

Hannu Tommola

b.

Swedish (PROGQ: 34) Philip håll-er på att erkänna sin skuld! Philip keep-PRS Prep INFM admit POSS guilt ‘Philip is admitting his guilt!’27

But (31) is, at the same time, imminential; that is why the Swedish construction in (31b) is possible.28 The difference between a “normal” (imminential) and a “marked” (interpretational) use may easily vanish altogether.

5.1. The role of emphasis (extended uses) Even if some examples with INFPROG are regarded as “odd” by most speakers, those speakers that produced them must have felt some sort of need to use the construction. Having not yet grammaticalized, a periphrastic form like INFPROG is necessarily more emphatic, expressing something more than a simple tense form, i.e., pragmatic interpretation or attitude of the speaker to the event or state described. Let us call this first type of emphatic use exemplified by (31) interpretative. This is essentially the explanation of interpretative uses of the English Progressive that Metslang (1993: 471) presents (with reference to König 1980): the speaker interprets an activity that the subject is involved in in a particular way, emphasizing its supposed significance, effects or certain consequences (“if you do that, you are, in fact, doing this”). The following examples are of the same type:

(32)

Estonian (PROGQ: 65) [Nagu sa hakkad küsima, mis kasu on haridusest, nii] ole-d sa loobu-ma-s tõelise COP-2SG PRON:2SG abandon-mINF:INESS real:GEN kultuur-i põhieeldus-te-st. culture-GEN basepresumption-PL-ELAT ‘[As soon as you start asking what is the use of education,] you abandon / are abandoning the basic assumptions of any true culture.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

673

Progressive aspect in Baltic Finnic

(33)

673

Finnish [Maailman hätä on mittaamaton, sadat miljoonat näkevät nälkää, pakolaisten olot sodan jaloissa ovat hirvittävät.] E-n ole si-tä kiistä-mä-ssä e-n-kä NEG-1SG COP it-PRTV deny-3INF-INESS NEG-1SG-and si-ltä silm-i-ä-ni sulke-ma-ssa. it-ELAT eye-PL-PRTV-POSS:1SG close-3INF-INESS ‘[The world’s misery is immeasurable, hundreds of millions are starving, refugees ravaged by war are living in horrible circumstances.] I am not denying it, nor closing my eyes in view of it.’

Another type of marked use of INFPROG is where it conveys negative attitude (see 3.4). It is obviously not a property of the construction itself, but being a marked form, INFPROG lends itself to various expressive functions to be specified pragmatically by the context. This meaning is supported by lexical units like aina ‘always’ and taas ‘again’, as in (34) in Estonian (cf. Ebert, this volume on ‘around’ with motion verbs in Germanic). (34)

Estonian (PROGQ: 33) Jüri on jälle rumala-te küsimus-tega oma George COP again stupid-GEN:PL question-COM:PL own solva-ma-s. naabri-t neighbour-PRTV insult-mINF-INESS ‘[Look, there he goes again!] George is [inadvertently] insulting his neighbour with his silly questions.’

Some Finnish people, when interviewed, for example, on TV (notably politicians and officials speaking publicly), tend to “overuse” INFPROG in ways that astonish most ordinary speakers. This use could perhaps be motivated by a desire to express guarded opinion. (35)

Finnish Kyllä me poliisi-ssa olem-me kannatta-ma-ssa PRAGM we police-INESS COP-1PL support-3INF-INESS sitä, että . . . it:PRTV that ‘We in the police force support (are supporting) [the idea] that . . . ’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

674

674 (36)

Hannu Tommola

Finnish Ole-n näke-mä-ssä, että realistinen ajankohta timepoint COP-1SG see-3INF-INESS that realistic ol-isi lokakuu-ssa. COP-COND October-INESS ‘As I see it, the realistic time would be in October.’

There may also be a connection between this kind of “tendentive” use and the function of INFPROG to express future time reference. In (37) the use of INFPROG can hardly be explained other than as referring to the uncertainty in calculations being discussed.

(37)

Finnish Tä-hän ei vielä ole sisält-y-mä-ssä this-ILL NEG still COP contain-ACAUS-3INF-INESS tuki. siirtymäaja-n transitiontime-GEN support ‘This doesn’t yet include (is not including) the support for the transition time.’

5.2. Morphological, syntactic, and lexical properties of INFPROG As pointed out by Metslang (1993: 474), marked verbal categories may constrain the use of INFPROG. This restriction is valid in Estonian, but not to such a degree in Finnish. If the Finnish INFPROG is acceptable in the not fully grammaticalized future construction, it is odd with the saama-Future, and impossible with the hakkama-Future in Estonian (Tommola 1992a, b). INFPROG of causative verbs is used in both languages (PROGQ: 20), but Passive is not typical of the Estonian construction (occasionally, NOUNPROG can be used instead, see 11). Negation does not seem to influence the use of INFPROG or NOUNPROG. Reference to a single moment in time may be so important that INFPROG can be motivated in the incidential schema even as an exhortation, although the construction is otherwise odd in the Imperative. The pragmatic function of the Progressive Imperative (38a) is the same as in (38b), where a special “pretending” construction is used in Finnish.29 Negated Imperative (or prohibitive speech act as in PROGQ: 74), though, is not possible.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

675

Progressive aspect in Baltic Finnic

(38) a.

b.

675

(PROGQ: 73) Estonian Ol-ge tööta-ma-s, kui ülemus naase-b! COP-IMP:2PL work-mINF-INESS when boss approach-3SG ‘[For goodness sake,] work when the boss comes back (is approaching)!’ Finnish Ole ihmee-ssä teke-vi-nä-si työ-tä, COP:IMP wonder-INESS do-PART:PL-ESS-POSS:2SG work-PRTV [kun pomo tulee takaisin]! ‘For goodness sake, pretend to work [when the boss comes back]!’

When postural verbs appear in INFPROG in Baltic Finnic, which is not typical, they are usually combined with an explicit locative adverbial, as in Häkkinen’s example (39) from Agricola. But temporal contrast by means of ‘still’ or ‘already’ also makes it acceptable (40), while the purposeful activity meaning (’what they are supposed to be doing’) is involved as well – in fact, it may be decisive here. (39)

Finnish (16th century, Häkkinen 1994: 395) [(sangen ialo Tulispä) teutti caiken Honen,] cussa he ol-i-t istu-ma-s. where they COP-PST-3PL sit-3INF-INESS ‘[(A very noble blast of wind) filled up the whole room] where they were sitting.’

(40)

Finnish Kaikki muu-t o-vat jo/vielä istu-ma-ssa. all other-PL COP-3PL already/still sit-3INF-INESS ‘All the others are already/still sitting.’

5.3. Imminentiality vs. process meaning Phasal verbs being interpreted as processes is a classic example of the difficulties in deciding whether we are referring to an ongoing process or to an imminent event when we use INFPROG. Since the verbs for ‘beginning’ and ‘finishing’, strictly speaking, refer in their semantics to a single point, they imply future time reference in the Present tense, and imminence in the Progressive (see section 3).30 When an event is “located” in the middle of an activity or process, the focus is easily transferred to the temporal simultaneity, thus not necessarily depending on the agent’s active engagement in the activity. A nondurative event being presented as a state gives rise to a paradox, as in (41).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

676

676

Hannu Tommola

(41) a.

b.

c.

(PROGQ: 32) Estonian [Armuandmisteade jõudis kohale just siis, kui kapten] and-is / ol-i and-ma-s käsklus-t give-PST COP give-mINF-INESS command-PRTV hukkamiskomando-le. firing_squad-ALL ‘[The pardon arrived just when the captain] was giving the sign to the firing squad.’ Finnish1 [Armahdus tuli juuri, kun kapteeni] ol-i anta-ma-ssa tulikäsky-ä. COP-PST give-3INF-INESS firecommand-PRTV ‘[The pardon arrived just when the captain] was giving the sign to the firing squad.’ Finnish2 ... ol-i COP-PST give-3INF-INESS/5INF-ADESS-POSS:3 anta-ma-ssa/anta-maisi-lla-an tulikäsky-n. firecommand-GEN ‘ was about to give the sign to the firing squad.’

Nondurative verbs in INFPROG normally imply imminentiality but can still remain ambiguous as to the imminential vs. incidential reading. In Finnish, a difference is seen in transitive sentences where the object case can be either Partitive (41b) or Accusative (41c) with INFPROG, while only Accusative is possible with the 5th Infinitive, an unambiguous imminential construction (see section 3.3). The constructions in (41c) are synonymous, whereas in (41b) ‘giving the sign’ acquires some sort of durative interpretation, hence it will not be known whether the execution ever took place. We can say that (41b) is a proper progressive (although imminential in a broad sense) in meaning because the clause is, strictly speaking, marked for ‘process’ by the Partitive, whereas (41c) is marked for ‘event’ by the Accusative, and therefore imminential (in a narrow sense). The tendency of verbs denoting gradually developing processes to occur in INFPROG has frequently been observed. In these cases the locative and the purposeful activity explanations do not work. ‘Gradual change’ means exactly that there is some ‘progress’ over a period of time, change is both going on and being expected. Bybee and al. (1994: 135) assume that the “application of the progressive to states that are developing by degrees [ ] is probably a later development”. If this is true, we can assume that the situations becoming possible in the process of grammaticaliza-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

677

Progressive aspect in Baltic Finnic

677

“processive” continuous stative (or: tion follow the pattern: progressive activities processes “durations” states). The Finnish verb for ‘cover’ used in (42) belongs to the group of accomplishment and achievement verbs having two readings, actional and stative. In metaphorical use they acquire a sense of the resulting state.31 The role of INFPROG with such verbs consists of discarding the stative (resultative) reading, while the object case opposition stands for the specification of the process vs. imminence meaning. Thus, (42) is a process. (42)

Finnish (PROGQ: 55) Lumi ol-i hiljalleen peittä-mä-ssä maa-ta. snow COP-PST little_by_little cover-3INF-INESS earth-PRTV ‘[When I arrived] the snow was gradually covering the land.’

5.4. Trouble with Estonian In some prototypical progressive sentences of the PROGQ a Progressive occurs in Finnish, but not in Estonian, e.g., in (43). In (44) it is optional in both languages. (43)

Finnish (PROGQ: 06) Hän puhdista-a / on puhdista-ma-ssa pyssy-ä. PRON:3SG clean-3SG COP clean-3INF:INESS gun-PRTV ‘He is cleaning the gun.’

(44)

Estonian (PROGQ: 20) Ta on las-t putru söö-ma PRON:3SG COP child-PRTV porridge:PRTV eat-mINF sundi-ma-s. persuade-mINF-INESS ‘He is making the child eat the porridge.’

The impression from the basic data for the PROGQ, supported by Dahl’s (1985) results and Metslang’s (1993) opinion, suggests that the Finnish INFPROG is considerably more frequent than the corresponding Estonian one. However, the greater the number of informants consulted, the more various examples will be collected of the Estonian Progressive. After all, there are no real constraints that would explain why the Estonian Progressive could not be used, for example in (43). Therefore the frequency should perhaps not be paid too much attention, especially as there is no corpus-based data. In Finnish, statives and nonagentive predicates do not allow INFPROG in the current reference meaning. In Estonian it is sometimes possible. Consider (45) and (46) from Metslang (1993: 327, 330):

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

678

678

Hannu Tommola

(45)

Estonian Ümberringi ol-i must maa, halli-de-st pilve-de-st all around COP-PST black earth grey-PL-ELAT ol-i sada-ma-s lõputu-t peenikes-t cloud-PL-ELAT COP-PST rain-mINF-INESS endless-PRTV vihma. tiny-PRTV rain:PRTV ‘[All around the earth was dark] and light rain was falling from the grey clouds.’

(46)

Estonian Torm on puid murd-ma-s. storm COP tree:PL:PRTV break-mINF-INESS ‘The storm is breaking trees.’

Metslang (1993: 474) notes that INFPROG from stative and postural verbs is largely used in existential sentences. But also in nonexistential intransitive sentences continuous (47), generic (48) and atelic (49) situations seem to be marked more frequently with it in Estonian than in Finnish. Metslang underlines that this use is rapidly spreading in the language of the media (radio, TV), notably in reportages such as sportcasts (49). (47)

Estonian (PROGQ: 27) Ta on oma lugu jätka-ma-s. PRON:3SG COP own story:PRTV continue-mINF-INESS ‘He is continuing his story-telling [right now].’

(48)

Estonian (PROGQ: 60) . . . on maa pöörle-ma-s ümber päikese. COP earth turn-mINF-INESS around sun:GEN ‘[Think! While we are here talking about our matters] the earth is turning around the sun.’

(49)

Estonian Mei-e võistleja on praegu jooks-ma-s väga we-GEN competitor COP right_now run-mINF-INESS very hea-s tempo-s. good-INESS tempo-INESS ‘Our competitor is running at a very good speed now.’

Metslang (1993: 474) also points out the function of the mas-construction in clauses providing a temporal frame and background in narrative.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

679

Progressive aspect in Baltic Finnic

(50)

679

Estonian (PROGQ: 62) [Oli päikseline suvepäev. Õhus] ol-i-d sumise-ma-s mesilase-d ja COP-PST-3PL hum-mINF-INESS bee-PL and laul-ma-s linnu-d, sing-mINF-INESS bird-PL [karjamaal sõid lehmad rohtu. Äkitselt avanes maa ja sealt tuli välja kurat.] ‘[It was a bright summer day.] Bees were humming and birds singing [in the air], and [cows were grazing in the green field. Suddenly, the earth opened and the devil came out.]’

It is remarkable that these and similar Estonian examples do not add anything to the reading of the sentences if compared with nonprogressive predicates, while in cases discussed in section 5.2 it was possible to discover a pragmatic motivation for the use of INFPROG, and thus a meaning different from the simple tense form.

6. Conclusion: criteria and degree of grammaticalization The problems in the choice between an imminential and a progressive interpretation of INFPROG were mentioned before. We often have similar difficulties in trying to distinguish the locative (absentive/purposive) implications of the meaning expressed by INFPROG. These must be the instances that represent the very core of the grammaticalization of progressive constructions: the reinterpretation of a nondurative action as already being in process, on the one hand, and the loss of the locative meaning, on the other. It is feasible to think that the latter process is the crucial one for the originally locative constructions. The resistance of INFPROG to atelic predicates and sentences other than those denoting purposeful activity can be taken as evidence of the first step being from locative to purposive (‘he is writing his thesis’). Once the locative meaning recedes, the temporal one (‘at reference time’; cf. Bybee et al. 1994: 136) comes into focus, and the purposive element decreases in importance (‘I’m listening to the news’). At this stage nonagentive (or personified) subjects become possible in the incidential situation (‘the train is nearing Helsinki’), and nonduratives get an imminential interpretation (‘the bomb is exploding’). From here on the progressive constructions may develop future meanings (‘they’re getting married next month’), and/or expand to comprise continuative or stative semantics (‘the sun is shining’). Languages – synchronically observed – seem to differ in classifying states and dynamic situations (Comrie 1976: 35; cf. Bybee et al. 1994: 151–152). In what follows, an attempt is made to shed some more light on the evolution of progressives. The original source of progressive constructions is not without signifi-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

680

680

Hannu Tommola

cance in their further development, as will be exemplified by once more contrasting devices from Swedish and Baltic Finnic, and comparing these to English and Icelandic Progressives. The question of the real nature of absentive/purposive will also be taken up again, because of the paradoxical claim that constructions that derive from a locative meaning are found to have this same meaning as an “extended use”.

6.1. Absentive vs. purposive To express absentive meaning, INFPROG is sometimes the only real choice in Baltic Finnic (21, 28, 51). The telic verb käydä ‘go and come back, visit’ in the simple form would provide future meaning in (51), while it in INFPROG has an absentive reading. Here the absentive meaning is clear from the locative adverbial, and INFPROG merely presents the absence as temporary, which is one of the pragmatic functions of the progressive (Comrie 1976: 37, Tommola 1981: 87, Heinämäki 1995). (51)

Finnish Hän on käy-mä-ssä posti-ssa. PRON:3SG COP go-3INF-INESS post-INESS ‘He went to the post office [and hasn’t come back yet].’ [or, more literally: ‘He is (away) to the post office’]

To say that absentive meaning has grammaticalized in a language implies that there is a particular verbal form (periphrastic construction) which the predicate takes when it denotes something that happens somewhere else, i.e., not at the deictic centre (see De Groot, this volume). Not all absentive constructions fulfill this requirement. For example, the Baltic Finnic INFPROG is a standard device to render this kind of situation, but it does not necessarily refer to absence from the deictic centre, as repeatedly stated above (see also Heinämäki 1995). The spatial element in this use is vague: ‘he is having lunch’ does not specify location, although it may be pragmatically known. The loss of the explicit locative meaning can be seen as a necessary first step on the path of grammaticalization: the infinitive form is no longer a parenthetic adjoined element of the clause expressing location, but a part of the predicate clause. On the other hand, the simple form does not convey the same information. Vauva nukkuu (PRS) ‘the baby is sleeping’ need not have the interpretation: ‘the baby has been put to bed’ (on purpose), whereas vauva on nukkumassa (INFPROG) ‘the baby is in the bed’ means exactly that, and does not need to imply: ‘the baby is sleeping’ (see also 29). I maintain that absentives are purposives, and that the purposive element continues to play a role in the progressive use of the Baltic Finnic INFPROG. In some varieties of Swedish spoken in Finland, the special absentive construction vara och VFIN can have “presentive” (purposive) meaning, e.g., in the following conversations:32

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

681

Progressive aspect in Baltic Finnic

(52)

Swedish in Finland (Korsholm, Ostrobothnian) [Hej, Britta! Ji dy jieer? – ] Jå, ja ji å dansar. yes I COP and dance ‘[Hi, Britta! Are you here? – ] Yes, (I’m) dancing.’

(53)

Swedish in Finland (Korsholm, Ostrobothnian) [Va(d) jää(r) dy jieer? – ] Ja ji å grillar korv. I COP and grill sausage ‘[What are you doing here? – ] I’m grilling sausages.’

681

6.2. Continuative vs. imminential In Section 4 a comparison of progressive constructions derived from locative vs. postural or motion verb expressions was undertaken. The data from Scandinavian and Baltic Finnic languages suggest that while both may grammaticalize as progressives, the scope of “extended” (or “remaining”) meanings that they additionally convey is different, reflecting their origin. A crucial test to prove the (beginning of) grammaticalization of the locative constructions, which was mentioned above, consists of exclusion of the locative meaning (section 2). To establish the grammatical status of the serial constructions employing postural or motion verbs, evidence is needed of uses where these verbs cannot have their lexical meaning. Such evidence is not even available from Estonian, where the serial construction is more frequent than in Finnish. In Swedish such examples can be found. A typical one is (54) where ‘the water stands boiling’. (54) a. b. c.

Swedish (PROGQ: 37) ‘The water is boiling. [Shall I make tea?]’ Vattn-et kok-ar. water-DEF boil-PRS Vattn-et stå-r och kok-ar. water-DEF stand-PRS and boil-PRS Vattn-et håll-er på och kok-ar. water-DEF keep-PRS Prep and boil-PRS

The serial construction tends to have continuative meaning. But the hålla på construction can also be used in (54)33, and it seems natural that it conveys this meaning, as it is derived from a continuative verb (’to keep’). But this construction appears in two main variations – hålla på ((med) att/å)  VINF, hålla på och  VFIN – which are not freely interchangable (see Ebert, this volume, n. 14). The construction where the preposition is followed by the Infinitive (marked by att, often phonetically identical with och) cannot be replaced in all contexts with the construction where

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

682

682

Hannu Tommola

the main verb in the second component is in finite form, and vice versa. The Infinitive construction behaves very much like the locative-based Baltic Finnic INFPROG (except the absentive/purposive meaning): (54) receives an imminential interpretation (‘the water’ll be boiling soon’) with hålla på att koka, exactly as with the Baltic Finnic INFPROG, and in contrast to the serial construction. The finite verb variant is, in fact, a serial construction, and a continuative, which does not convey the imminential meaning. On the other hand, the Infinitive variant is not used in continuous (durative) contexts.34 Thus, there is a difference in meaning between (55) and (56 = 20a): (55)

Swedish Den gamle mannen höll på å [= och] dog. ‘The old man was dying [and died].’

(56)

Swedish Den gamle mannen höll på å [= att] dö. ‘The old man was dying [but they finally found the right medicine].’

Like (23) above with the serial verb construction, (57) is continuous (durative) in meaning, and impossible with the Finnish and Estonian INFPROG (except with an absentive reading). The hålla på construction appears in (57), of course, with the serial – finite verb form – structure. (57)

Swedish (PROGQ: 81) Jag ha-r håll-it på och baka-t / *att I have-PRS keep-PART:PF Prep and bake-PART:PF INFM baka ända sedan i morse. bake all since this_morning ‘[I am so tired:] I have been baking ever since I got up this morning.’

As mentioned before, processes developing by degrees tend to neutralize the imminential/progressive opposition, as the process itself is continuous, but simultaneously there is “a piece of change” to be expected all the time. (58) a.

b.

(PROGQ: 38) Estonian Õuna-d on puu otsa-s mädane-ma-s. apple-PL COP tree:GEN top-INESS rot-mINF-INESS Finnish1 Omena-t (puu-ssa) o-vat mätäne-mä-ssä. apple-PL tree-INESS COP-3PL rot-3INF-INESS

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

683

Progressive aspect in Baltic Finnic

c.

d.

(59) a.

b.

c.

683

Finnish2 Omena-t o-vat mätäne-mä-ssä puu-hun/*?puu-ssa. apple-PL COP-3PL rot-3INF-INESS tree-ILL/INESS Swedish Äpple-n-a håller på och ruttn-ar på träd-en. apple-PL-DEF keep Prep and rot-PRS Prep tree-PL:DEF ‘[Look, what a shame!] The apples are rotting on the tree.’ (PROGQ: 26) Swedish1 Han håller just på att avsluta en språkövning. he keep just Prep INFM finish:INF indef languageexercise Swedish2 Han håller just på å avslut-ar en språkövning. he keep just Prep and finish-PRS indef languageexercise Finnish Hän lopett-ele-e juuri ääntämisharjoitus-ta. PRON:3SG finish-FREQ-3SG just pronunciationexercise-PRTV ‘He is finishing a language drill [right now].’

In (58) the Finnish INFPROG is problematic, because as soon as the locative adverbial ‘on the tree’ is added, the sentence acquires a purposive reading (58c with ‘tree’ in Inessive), which is odd in this context. A thematic position of the adverbial helps a lot, and (58b) conveys exactly the same meaning as the Swedish (58d). A directional adverbial ‘onto the tree’ (58c) again makes the sentence unambiguously imminential, referring to the result in future.35 In (59) both Swedish hålla på-constructions can be used, something that confirms the assumption that gradual processes are intermediary between progressives and imminentials. Further evidence from Finnish suggests that processes are increasingly conceived of as progressives, rather than imminentials, for the object is marked with the Partitive, no matter whether the predicate is a Frequentative in the simple Present (as in 59c), or INFPROG.

6.3. Recapitulation I have attempted to argue that absentives are purposives, and that progressives derived from locative sources develop through the stage of purposive to arrive at a tense-aspect function involving both ongoing and imminential meaning. It seems plausible to assume that the locative element has to disappear first, while the purposive element then remains. This purposive meaning is well suited for submitting to

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

684

684

Hannu Tommola

Table 2. English, Icelandic and Baltic Finnic Progressives

Purposive Imminential Progressive Continuous Stative

English

Icelandic

Baltic Finnic

    

   

  

the activity goal-oriented dynamics which brings about the imminential/future reading. Then it is just a question of increasing usage of the construction with different kinds of verbs and subjects, whereby atelic activities and nonagentive processes appear. What remains is the temporal connection with the reference time. The behaviour of serial progressive constructions is different from the Baltic Finnic INFPROG that originates from a locative construction. The English Progressive has expanded to the continuous domain, but it is not compulsory there: cf. (22–23), where English allows both the simple and the Progressive Past.36 In Table 2 the information from Table 1 (see section 3) on the meanings of the Finnish and Estonian devices is complemented with Swedish constructions, and the Icelandic and English Progressives have been added for comparison. From Table 2 we can see how the functions of the progressive constructions are distributed: the English Progressive can be used in all eight functions, the Icelandic in seven (not used for states as in ‘it’s raining’), the Finnish INFPROG in six (additionally deprived of the continuative function as in ‘the water is boiling’), the Estonian INFPROG in five (minus the explicit future as in ‘he’s getting married next week’), the Swedish hålla på-constructions in three (minus both purposive uses), and so on. To conclude, the meanings of the English, Icelandic and Baltic Finnic progressive constructions are presented in Table 3. The Baltic Finnic INFPROG has at present reached the stage where purposive and imminential are grammaticalized, and all the uses from the locative to the gradual process meaning are quite “normal”, whereas occasional continuative and stative uses are “odd”. The position of imminential in this chain is somewhat problematic, because it seems to be the case that imminential, progressive and “processive” meanings coexist in languages at the same time, and the imminential meaning would be easy to explain as developing from the “processive” – which is intermediary between stative and future. The Progressive constructions in the two Baltic Finnic languages are developing differently in that the Estonian INFPROG, to some extent, favours nonagentive and stative situations, while the Finnish one is more frequent in indicating dynamic and controlled situations. It is difficult to say what the reasons might be,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

685

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Table 2. Markers of the progressive and related meanings English Icelandic Eng Ice

S1a S1b S2a

  

Imminential FTR Absentive Presentive

       

   

( ) ( )

( ) ( )

( ) ( )

() ( )

() ( )

Functions (N)

8

7

3

3

2

2

3

Progressive Processive Continuous Stative

 

  

Swedish S2b S3

 

S4

  

Swedish: S1a S1b S2a S2b S3 S4

  ( )

() ()

  

() 2

6

4



hålla på (att/å) VINF hålla på (och/å) VFIN VFINpostural och VFIN VFINmotion och VFIN vara på väg vara och VFIN

F5

1

Estonian E2 E3

   ( ) ()

E4 ( ) ()

( ) ( )

     

( )

  2

E1

0

5

2

2

0 Progressive aspect in Baltic Finnic

be V-ing vera að VINF

 

Finnish F3 F4

   

Devices English (Eng) Icelandic (Ice)

F2

( ) ( )

( )



F1

685

686

686

Hannu Tommola

but in both languages the step from a locative/purposive adverbial to a progressive predicate form has been made.

Acknowledgements Many of the insights of this paper are due to inspiration from Orvokki Heinämäki’s work and words. My cooperation with Helle Metslang should be mentioned as a necessary condition for the appearance of this paper. I also express my gratitude to Peeter Torop for Estonian data, and to Marianne Nordman for Finland Swedish intuitions. Useful remarks on earlier versions of my paper were made by Karen Ebert and Östen Dahl.

Notes 1. The simple forms if used in (5) and (6): Maret sööb and Pomm plahvatab, respectively, do not convey a natural utterance, unless taken as generic characteristics. 2. For example, the data from the E UROTYP Questionnaire on the Progressive Aspect (= PROGQ) provided by Swedish informants are very heterogenous. While three of the informants did not use any progressive constructions at all, one did 39 times. The hålla på (att/och)-construction was used by at least one informant in 30 items, the serial construction with postural verbs in eleven, and the absentive construction in four (see Ebert in this volume). 3. I hesitate to label this use “purposive”, because the term has been used in slightly different meanings, e.g., in Bybee et al. (1994: 222–225) referring to a grammatical morpheme in Gugu-Yalanji; however, in Heine & Traugott (1993: 2–3, 83–84) it refers to purpose clauses and is quite close to the meaning I am aiming at. 4. This is plausible only in the primitive sense that both progressives and locative expressions have a time reference. The difference is that the progressive proper refers to an activity the subject is said to be engaged in at the reference time, whereas locative/purposive refers to the activity the subject is said to be supposed to be engaged in. 5. Metslang investigated materials from the Syntax Archive at the University of Turku. 6. Cf. the periphrastic constructions in French: (i) être en train de V, (ii) venir de V, and (iii) aller V used to express progressive, conclusive and future meanings, respectively. In Finnish there is a gap where a completed action should be denoted, while an Inessive Progressive and an Illative Future are developing towards grammaticalization. 7. “Perfektiivsed ja perfektiivsusele kalduvad verbid”, i.e., achievement and accomplishment verbs. 8. The other three are: marked forms of verbal categories (such as Passive, Conditional, etc.), rhematic predicates, and foregrounded sentences in narrative (Metslang 1993: 474). 9. In Erelt et al. (1993: 251–260) the -ma-Infinitive is called supiin. Estonian grammars in general distinguish two Infinitives: the -ma-Infinitive (= the Finnish 3rd Infinitive Illative)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

687

Progressive aspect in Baltic Finnic

10. 11. 12.

13.

14.

15.

16.

17.

687

and the -da-Infinitive (= the so-called short form of the 1st Infinitive in Finnish grammar). In Estonian the -ma-Infinitive is regarded as basic and used as the entry form, whereas in Finnish grammar the 1st Infinitive is the basic form. In this paper, in the glosses, the -ma-Infinitive is marked “mINF”, the Illative being the default form; the other cases of the -ma-Infinitive will be marked out, e.g., “mINF:INESS” = the -ma-Infinitive in the Inessive case. The -da-Infinitive is marked simply “INF”. See Bertinetto, Ebert & De Groot (this volume, fn. 8) for an explanation of the term “incidential schema”. Erelt (1987:45–46) has pointed out that the Estonian mas-construction has grammaticalized in the imminential meaning (“nondurative perfective situation”). In the Bible translation confirmed for use in the Lutheran church in 1938 these motion verbs were still used in the 3rd Infinitive form, e.g., in an imminential meaning as in (John 5: 7): “[ ] ja kun minä olen menemässä, astuu toinen sinne ennen minua” ‘[The sick man answered, “Sir, I don’t have anyone here to put me in the pool when the water is stirred up;] while I’m trying to get in, somebody else gets there first” ’ (The English translation is from Good News For Modern Man: The New Testament. Today’s English Version, New York: American Bible Society, 1966). In the new translation from 1992 NOUNPROG is often also found where the old translation had a simple form, e.g., (John 11: 31): [ ] koska arvelivat hänen olevan menossa haudalle itkemään ‘[ ] They thought that she was going to the grave, to weep there’. The nouns for ‘going’ and ‘coming’ seem to be primarily used only in this construction. Out of nineteen occurrences of meno in the New Testament 1992 translation, eleven were instances of NOUNPROG, and for tulo the corresponding figures were 23 out of 38. At least in Finnish there are reasons to consider NOUNPROG as lexicalizations. The nouns involved are most frequent in this particular construction (see n. 12), and motion verbs in many languages form a group that behaves in a nonconformist way. Compare with Russian motion verbs of definite and indefinite direction (see Section 1) and the tendency of motion verbs to be used in progressive or present with future time reference (Dahl this volume). The use of the noun käynti in NOUNPROG and the standard INFPROG from the polysemous motion verb Fin. käydä is differentiated: olla käynniss (NOUNPROG) means ‘to go on, to function, be in motion’, whereas the meaning of kydä employed in the INFPROG construction is ‘go and come back, visit’, thus olla käymässä ‘be visiting, (be) pay(ing) a visit, be somewhere temporarily’ (similarly Est. käigus olema vs. käimas olema). The derivational type of the Finnish men-o (similarly Est. mine-k) is not typical of nomina actionis; the productive suffix -minen is not used in NOUNPROG, while the corresponding Estonian -mine is possible (as seen in 11). Consider also: Hän on (ulkona) tupakalla (tobacco:ADESS) / (baarissa) kahvilla (coffee:ADESS) ‘She/He is (outside) smoking / (in the bar) drinking coffee’ and He ovat mustikassa (blueberry:INESS) / sienessä (mushroom:INESS) ‘They are (away) picking blueberries/mushrooms’ etc. The meaning provided by these Finnish constructions – imminential in a narrow sense – was called by Hakulinen (1968: 210, 211) “propinquative” (cf. Latin propinquus ‘near’).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

688

688

Hannu Tommola

18. What is interesting but not surprising – in view of an inherent kinship between the resultative perfect meaning and the progressive meaning – is that it occurred three times in PROGQ in the Perfect form (PROGQ: 40, and PROGQ: 53–54) in sentences denoting a current state or process. 19. According to Helle Metslang, the construction itself does not provide any emotional nuance in Estonian. 20. It is therefore problematic whether this construction is a “perfectivizer” or not. For example, in (16) it corresponds rather to the “general factual” (“simple denotative”) use of the Russian imperfective aspect. Fin. käydä, Est. käima are similar to Russ. xodit’ in the “two-way action” meaning and use. They cannot refer to a single moment, hence they cannot express actual present time reference nor, consequently, build a progressive marker. Moreover, in certain cases, they may be “progressivized” themselves by the standard device (see 51). 21. The simple transitive phasals (‘begin’, ‘finish’) are not quite normal with a direct object in the Partitive, whereas INFPROG and Frequentatives derived from them are. With an Accusative object the simple phasals and their INFPROG forms get a futurate or imminential interpretation, respectively. There is thus a real reinterpretation of phasal progressives as ongoing processes ‘to V in the very beginning/concluding phase of V’ in the combinations V:INFPROG/V:FREQ  PRTV. Toivainen (1990: 19, 22, 26, 45, 79–86) regards the Frequentatives as belonging to aspectual flexion in Finnish children language (cf. Tommola 1986: 242–254, 267–274). 22. Note that in Dahl (1985: 90) gerade is presented as PROG in German. This is, however, an “artifact of the methodology” (Östen Dahl, p.c.). In English, according to the Collins Cobuild Dictionary, just is used with the “continuous form” primarily in the imminential meaning. 23. The meanings are grouped in three sets. The first set comprises types of situations with increasingly extended reference time allowed: activity (“Progressive”), gradual process (“Processive”), durative activity or process (“Continuous”), and non-controlled states like it’s raining (“Stative”); the second set includes futurate meanings: event in very near future (“Imminential”) and explicit future time reference (“FTR”); the third set: purposive meanings with absentive (“Absentive”) and nonabsentive (“Presentive”) readings. 24. There is another interesting detail in Dahl’s TMAQ Finnish material which can be called the “remoteness factor”. INFPROG was used in the introductory sentences of narrative where what happened either “yesterday” or “once upon a time” is reported. The only version of this story that does not begin with INFPROG (unless there is an error in the data?) is one where the speaker is “right back from a walk in the forest”. One occurrence alone would not be significant, but this fact is corroborated by intuitions concerning the prototypical cases of PROG. In all seven items exemplifying prototypical use in Dahl (1985: 92) INFPROG could have been used (although the simple forms are perhaps more usual). To formulate this carefully: I should judge the Progressive a) not exactly odd, but less motivated in those contexts where the utterance is made by “someone who can see” the person, and b) if not preferable, then at least as normal as the simple form in those contexts where the knowledge of the speaker is based on communication by telephone. The use of a specific form (with more emphasis) to express a state of affairs at a specific

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

689

Progressive aspect in Baltic Finnic

25. 26.

27. 28.

29.

30.

31.

32.

689

point of time is better motivated when the situation referred to is not obvious, locally (‘here’) or temporally (‘now’), but absent from the deictic centre. In Icelandic no Progressive was used in PROGQ: 54. In general, performative verbs are not performatives any more, if they are used in a progressive form. If I say ‘I thank you’, I perform a thanking act, but if I say ‘I am (just) thanking you’, I explain what I am doing. Cf. also the following quotation from the 1938 Bible where INFPROG from a verb of saying (in the 1992 translation replaced with a simple Present form) suggests that the activity is not controlled by the subject: (Mark 13: 11) Sillä ette te ole puhumassa, vaan Pyhä Henki ‘For the words you speak will not be yours; they will come from the Holy Spirit’. In English the Progressive is normal, the simple Present (admits) “just possible” (Grev Corbett, p.c.). According to Wide (1994: 59, 71, 97–100, 107) the Swedish constructions hålla på and vara på väg (att) are not optional in this use – contrary to the genuine progressive use (cf. also Andersson 1977: 110–111). Wide’s examples confirm what she does not make explicit, namely that this applies to the hålla på (att/och) construction only with VINF, not with VFIN. PROGQ: 73 is the only item where Finnish and Estonian (marginally) allow INF-PROG, while English does not. The Progressive was also used in Icelandic (see Ebert in this volume), where it seems not to be rare at all; cf. ex. from Wide (1994: 103) with negated Imperative: Vertu ekki að trufla Íðu ‘Don’t disturb Ida’, Vertu ekki að eyða orku í að telja mér hughvarf ‘Don’t waste (your) energy on persuading me’. The phasal verb for ‘continue’ in INFPROG can only have an imminential reading: ‘to be about to continue’ (‘start again after a pause’). Also Russian prodolžit’, the Perfective counterpart of the Imperfective prodolžat’ ‘continue’, can only mean ‘start again’. The Estonian INFPROG seems to allow the process meaning, e.g., in (47). In Finnish, such verbs – or such use of verbs – have been called “quasiresultative” (Itkonen 1976: 174–185), because the object can take the Accusative case in spite of a clearly imperfective (stative) meaning of the sentence (see Tommola 1986: 140–159). The construction is used also with other meanings in various Swedish dialects in Finland. The uses in (i) can be said to be variations of the progressive, whereas (ii) is neither progressive nor absentive, but rather perfective, with special emphasis on the unexpectedness of the event. While the latter use is common in Ostrobothnian dialects, too, the progressive meaning proper seems to be restricted to ambiguous cases, i.e. contexts where it cannot be distinguished from the absentive (iii). Notice that this is exactly the problem we have in analyzing the Finnish and Estonian INFPROG construction that merges both meanings in one and the same form. (i)

Swedish in Nyland, Finland [Lundström 1939: 131] a.

[Deä so mong land som] ä o jäs-er po varann nu. COP and swell-PRS Prep each_other now ‘[There are so many countries that] are bearing a grudge against one another’.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

690

690

Hannu Tommola b.

c.

(ii)

[Ja, dom där Lunnbärs, dom] ä o klag-ar, dom. COP and complain-PRS they ‘[Well, those Lundbergs, they] are (always) complaining’. Un va jyst o sku ti dö. she be:PST just and shall:PST Prep die ‘She was dying [right then]’ Swedish in Nyland, Finland [Lundström 1939: 131] Senn va gumma-n hans o fikk en livsarvinge. then be:PST old_woman-DEF his and get:PST Indef heir ‘Then it happened that his wife got an heir’.

(iii)

Swedish in Finland (Korsholm, Ostrobothnian) Hör-ru, ja va o föll igoor. hear-2SG I be:PST and fall:PST yesterday ‘[By the way,] I fell yesterday’.

33. Again, at least, in the Swedish spoken in the Finnish province of Ostrobothnia. 34. It is, thus, not the choice between the infinitive marker att ‘to’ and the connector och ‘and’ that makes the difference: they are often pronounced identically (see Ebert in this volume) – and att is occasionally also written å, e.g.: Morfar, morfar, mamma håller på å dö (Icelandic translation: Afi, afi, mamma er að deyja; Wide 1994: 98) ‘Uncle, uncle, mother is dying’. 35. These problems, apparently, do not bother Estonians. 36. The aspectual opposition of perfective and imperfective may be neutralized, because the situation can either be conceived of as temporally bounded, or as having internal structure (cf., for example, perdurative perfectives competing with imperfectives and/or imperfective Aorists in Slavic languages).

References Andersson, Erik 1977 Verbfrasens struktur i svenskan. En studie i aspekt, tempus, tidsadverbial och semantisk räckvidd. Meddelanden från stiftelsens för Åbo Akademi forskningsinstitut, nr 18. Åbo, 564. Bertinetto, Pier Marco & Karen Ebert & Casper de Groot this volume “The progressive in Europe”. Bybee, Joan & Revere Perkins & William Pagliuca 1994 The evolution of grammar. Tense, aspect, and modality in the languages of the world. Chicago and London: The University of Chicago Press. Collins cobuild English language dictionary 1987 London & Glasgow: Collins. Comrie, Bernard 1976 Aspect. An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

691

Progressive aspect in Baltic Finnic Dahl, Östen 1985 this volume Ebert, Karen H. this volume Erelt, Mati 1985

691

Tense and Aspect Systems. Oxford: Blackwell. “The grammar of future time reference in European languages”. “Progressives in Germanic languages”.

“ma-, mas- ja mast-infinitiivist eesti keeles”, in: Ars Grammatica. Tallinn: Valgus, 4–22. 1987 Sekundaartarindid eesti keeles. Preprint KKI-50. Tallinn. Erelt, Mati & Reet Kasik & Helle Metslang & Henno Rajandi & Kristiina Ross & Henn Saari & Kaja Tael & Silvi Vare 1993 Eesti keele grammatika II. Süntaks. Lisa: kiri. Tallinn: Eesti Teaduste Akadeemia Keele ja Kirjanduse Instituut. Groot, Casper de this volume “The Absentive”. Groot, Casper de & Hannu Tommola (eds.) 1984 Aspect Bound. A voyage into the realm of Germanic, Slavonic and Finno-Ugrian aspectology. Dordrecht: Foris Publications. Hakulinen, Auli & Fred Karlsson 1979 Nykysuomen lauseoppia. SKS 350. Jyväskylä: Suomalaisen Kirjallisuuden seura. Hakulinen, Lauri 1968 Suomen kielen rakenne ja kehitys. 3rd impression. Helsinki (4th impr. 1979). Häkkinen, Kaisa 1994 Agricolasta nykykieleen. Suomen kirjakielen historia. Helsinki: WSOY. Heinämäki, Orvokki 1981 “On the meaning of olla  3rd inf.iness. in Finnish”, in: Ikola 1981: 383–388. 1995 “The progressive in Finnish: pragmatic constraints”, in: Pier Marco Bertinetto & Valentina Bianchi & Östen Dahl & Mario Squartini (eds.), Temporal reference, aspect and actionality. Vol. 2: Typological perspectives. Torino: Rosenberg & Sellier, 143–153. Hopper Paul J. & Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press. Hyvärinen, Irma 1982 “Suomen kolmannen infinitiivin verbisidonnaisten inessiivin, elatiivin ja illatiivin lauseenjäsenfunktioista ja niiden saksalaisista vastineista valenssiteorian näkökulmasta”, in: Lauseenjäsennyksen perusteet. Seminaari Seilissä 9.–10.9.1982. Suomen kielitieteellisen yhdistyksen julkaisuja 9. Turku, 59–89. Hyvönen, Tuula & Tuomo Jämsä 1978 Tempukset 1960-luvun suomen lehti- ja yleis- puhekielessä. Oulun yliopiston suomen ja saamen kielen laitoksen tutkimusraportteja 13. Oulu. Ikola, Osmo 1981 Congressus Quintus Internationalis Fenno-Ugristarum. Turku 20.–27.VIII.1980. Pars VI. Turku: Suomen Kielen Seura. Itkonen, Terho 1976 “Erään sijamuodon ongelmia”, in: Suomalainen Tiedeakatemia: esitelmät ja pöytäkirjat 1974. Helsinki, 173–217. König, Ekkehard 1980 “On the context-dependency of the progressive in English”, in: Christian Rohrer (ed.), Time, tense and quantifiers: proceedings of the Stuttgart conference on the Logic of tense and quantification. Linguistische Arbeiten, 83. Tübingen: Niemeyer. 269–291.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

692

692

Hannu Tommola

Lundström, Gudrun 1939 Studier i nyländsk syntax. Stockholm: Norstedt & Söner. Metslang, Helle 1993 “Kas eesti keeles on olemas progressiiv?”, Keel ja Kirjandus 6: 326–334; 7: 416–422; 8: 468–476. Penttilä, Aarni 1963 Suomen kielioppi. 2., tarkistettu painos. Hki: WSOY. Pollak W[olfgang] 1960 Studien zum ‘Verbalaspekt’ im Französischen. Österreichische Akademie der Wissenschaften, Philosophisch-historische Klasse. Sitzungsberichte 233,5. Wien. Serebrennikov, Boris Aleksandroviˇc 1963 “Kategorija vremeni v pribaltijsko–finskix jazykax”, in: Eesti keele süntaksi küsimusi. Keele ja Kirjanduse Instituudi uurimused VIII. Tallinn. 426–511. Setälä, E[mil] N[estor] 1891 Suomen kielen lauseoppi. Kolmas, muutettu painos. Helsinki: Otava. (16th edition 1973) Siro, Paavo 1977 Sijakielioppi. 2., korjattu painos. Helsinki: Gaudeamus. (1st edition 1975) Tauli, Valter 1966 Structural tendencies in Uralic Languages. Indiana University Publications. Uralic and Altaic Series, vol. 17. London/The Hague/Paris: Mouton. Toivainen, Jorma 1990 Acquisition of Finnish as a First Language: General and Particular Themes. Publications of the Department of Finnish and General Linguistics of the University of Turku. Turku. Tommola, Hannu 1981 “On the semantics of ‘situations’ and ‘events’ ”, Terminologie et traduction. Tome B. Vaasan korkeakoulun julkaisuja. Tutkimuksia 80, Philologie 7. Vaasa. 80–119. 1984 “On the aspectual significance of the ‘phase meanings’ ”, in: De Groot & Tommola 1984: 111–132. 1986 Aspektual’nost’ v finskom i russkom jazykax. Helsinki: Neuvosto-liittoinstituutti. 1992a “The Marking of Future Time Reference in Estonian”, in: Östen Dahl & Casper de Groot & Hannu Tommola (eds.), Future Time Reference in European Languages I. (EUROTYP Working Papers VI:2). Stockholm. August 1992. 12–28. 1992b “The Marking of Future Time Reference in Finnish”, in: Östen Dahl & Casper de Groot & Hannu Tommola (eds.), Future Time Reference in European Languages II. (EUROTYP Working Papers VI: 3). Stockholm. December 1992. 12–28. Wide, Camilla 1994 Konstruktionen vera að  infinitiv i isländskan och dess motsvarigheter i svenskan. Unpublished Master’s thesis. University of Helsinki. Department of Scandinavian Philology.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

693

Casper de Groot

The absentive

1. Introduction 1.1. What is the absentive? This chapter concerns the grammatical expression of absence as illustrated in the following examples from Hungarian, Italian, and Norwegian: (1) a.

b.

c.

Hungarian János level-et volt föladni. John letter-ACC was post:INF ‘John was off posting a letter.’ Italian Gianni è a mangiare. John is at eat:INF ‘John is off having lunch.’ Norwegian (Nynorsk) Jan er og handlar. John is and shop:PRS ‘John is off shopping.’

The examples in (1) contain the following types of information: (2) i. John is not present; ii. John is involved in an activity indicated by the lexical verb; iii. Based on pragmatic knowledge, it is predictable how long John will be away, or there is an assumption about the period of time that John will be away; iv. John will return after a period of time. This section further specifies the absentive, in particular the way in which the absentive is expressed in eight languages of Europe, together with alternative expressions of the absentive. Section 2 discusses properties which the absentive has in all eight languages. Section 3 deals with the relation between the absentive and aspect, i.e. progressivity and the imperfective/perfective opposition. Section 4 concerns the class of verbs which can be used in the absentive, and Section 5 the boundedness

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

694

694

Casper de Groot

of the activity. The last section summarizes the general findings and the typological situation.

1.2. The motivation for the investigation of the absentive The absentive is basically a spatial deictic construction. The construction is, however, often felt to be a kind of progressive. Moreover, in Finnish there is a construction which can have both the progressive and the absentive interpretation. Similarities between progressive and absentive on the one hand, and differences between them on the other, justify a discussion of the absentive in the section on the progressive of this volume. Furthermore, as will be shown later, the absentive has a number of properties relevant to the study of ‘Aktionsart’.

1.3. Defining the absentive Languages may have several types of lexical means to express absence, i.e., a way of expressing that somebody is not present at a certain place. A tentative list of possible expressions includes the following: (3) a. b. c. d. e. f.

Peter is absent. Peter is not here. Peter is walking to the railway station. Peter is shopping at the market. Peter is away shopping. Peter went out for lunch.

(predicate) (predicate) (adverbial phrase) (adverbial phrase) (adverbial phrase) (predicate)

Examples (3a) and (3b) contain only the information of absence. Examples (3c) and (3d) may, but do not necessarily, denote absence. The last two examples, (3e) and (3f), contain the information of Peter’s absence and also the activity he is, or will be involved in. The examples in (3) give one or more types of information with respect to the absence of Peter. These types of information are restricted to ‘being absent’, ‘being somewhere else’, ‘being on one’s way to some place, or ‘being out (doing something)’. They do not (necessarily) contain information about the period of time that Peter will be absent, or that Peter will come back after some time. This study is not concerned with lexical expressions of absence, but only with grammatical expressions such as the examples in (1). Van ‘be’  Infinitive in Hungarian can be considered a grammatical device because it always denotes that the person specified by the subject is absent. Note that Hungarian does not need any kind of lexical material to support this meaning. Italian and Norwegian use devices which are slightly different from the one Hungarian uses. What they have in common is that they do not contain any lexical material which indicates absence.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

695

The absentive

695

One context in which the absentive is most naturally used is the following questionanswer pair (on the phone/at the door): (4)

Hungarian A: Péterrel szeretnék beszélni. Peter:COM like:COND:1SG speak:INF B: De hiszen úszni van. well swim:INF is ‘I would like to speak with Peter.’ ‘Well, he is off swimming.’

Another situation in which the construction is often used is in messages people leave on the door, for instance: (5)

Dutch We zijn lunchen. we are have_lunch:INF ‘We are off having lunch.’

The absentive can be defined as the grammatical expression of absence. It contains four types of information about the referent of the subject (Subj): (i) Subj is absent, (ii) Subj is involved in an activity, (iii) it is predictable how long Subj will be absent, and (iv) Subj will return after a period of time.

1.4. Method and sample With the help of a questionnaire (De Groot 1993) based on a description of the absentive in Dutch, linguists with knowledge of various European languages were asked to answer 22 questions about the grammatical expression of absence and the distribution of a number of its semantic and pragmatic properties.1 The occurrence of a grammatical absentive seems to be limited to the following languages of Europe: Dutch2, Frisian3, German4, Hungarian5, Italian, Norwegian6 and Swedish. There is one other language which I include. In Finnish there is a form with the function of absentive. This form, however, is also used to express a progressive meaning (cf. section 3.1 below). The eight languages exhibit five types of expressing devices. The expressions are: (i) (6) a.

Copula  infinitive Dutch Jan is boksen. John is box:INF

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

696

696

Casper de Groot

b.

c.

German Jan ist boxen. John is box:INF Hungarian János boxolni van. John box:INF is Copula  to  infinitive

(ii) (7)

Fering Jan as tu boksin. John is to box:INF Copula  and  finite verb

(iii) (8) a.

b.

Norwegian Jan er og John is and Swedish John är och John is and

boksar. box:PRS boxas. box:PRS

(iv)

Copula  at  infinitive

(9)

Italian Gianni è a boxare. John is at box:INF

(v) (10)

Copula  infinitive-inessive Finnish Jussi on nykkeile-mä-ssä. John is box-3INF-INESS

There is an alternative expression for the absentive in Dutch and German which is less specific.7 An alternative expression of (6a) may be: (11)

Dutch Jan is naar de sportschool. John is to the fitness center ‘John went to the fitness center.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

697

The absentive

697

Constructions such as (11) are ambiguous in the sense that they can also have the restricted meaning ‘on his way’, i.e., (11) can be paraphrased as ‘John is on his way to the fitness center’. It is possible to disambiguate sentence (11) by adding the absentive by means of the infinitive as a kind of apposition (12a) or by adding a phrase of the type ‘on his way’ (12b). Phrases of the type ‘on his way’ are not compatible with the absentive, which holds for all languages. That is why (12a) and (12b) cannot be combined into one clause (12c). Consider: (12)

Dutch a. (absentive) Jan is naar de sportschool, boksen. John is to the fitness center box:INF ‘John is off to the fitness center, boxing.’ b. (directional) Jan is onderweg naar de sportschool. John is on his way to the fitness center ‘John is on his way to the fitness center.’ c. *Jan is onderweg naar de sportschool, boksen. John is on his way to the fitness center box:INF ‘John is on his way to the fitness center, boxing.’

Expressions with ‘be’  infinitive and also expressions with the infinitive as a type of apposition as in (12a) may have a purposive meaning. In that case they do not denote absence, as can be seen from the following example. Note the use of the adverb ‘here’, which, of course, is incompatible with the absentive reading. (13)

Hungarian Péter aludni van itt. Peter sleep:INF is here ‘Peter is here in order to rest.’

2. Invariable properties of the absentive 2.1. Absence 2.1.1. Deictic centre The absentive expresses that the person referred to by the subject is not present at what we shall call the deictic centre.8 If the deictic centre is specified in a clause, the use of the absentive entails the dislocation of the event from the deictic centre, as is illustrated in example (14a). Collocation of the deictic centre and an event is

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

698

698

Casper de Groot

possible by not using the absentive, but rather using, for instance, the progressive as in (14b). Consider: (14) a.

b.

Dutch Toen Peter de kamer binnenkwam was Marie lunchen. when Peter the room came_in was Mary have_lunch:INF ‘When Peter entered the room, Mary was off having lunch.’ Toen Peter de kamer binnenkwam zat Marie (daar) te when Peter the room came_in sat Mary there to lunchen. have_lunch:INF ‘When Peter entered the room, Mary was having lunch (there).’

Example (14b) is appropriate in the context where Mary was having her lunch in the room which Peter entered. The other example (14a) cannot be used in this context. Example (14a) can only be understood in the sense that Mary was not present in the room. The deictic centre is not restricted to the place of the Speaker (S). It may be anywhere, e.g., home, office, beach, etc. The following examples illustrate that the deictic centre may be distinct from the place where S utters the message: (15) a.

b.

Hungarian Nem érdemes odamenni, mert Imre ilyenkor úszni not worth there-go because Imre at that time swim:INF van. is ‘It is not worth going there, because Imre will be off swimming then.’ Fering Üüz ik jister uunrep, wiar Jan tu boksin. when I yesterday called, was John to box:INF ‘When I called yesterday, John was off boxing.’

2.1.2. Remoteness Absence also has to do with remoteness, i.e., it is not natural to use the absentive when the person referred to is in the direct neighbourhood of the Speaker. Consider for instance the following example from Swedish: (16)

Swedish John är och duschar. John is and have_a_shower:PRS ‘John is off having a shower.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

699

The absentive

699

When sentence (16) is used, the Addressee (A) knows or assumes that there is no shower in John’s house or in the hotel room where John is staying, and that John went somewhere else to have a shower. The absentive cannot be combined with a clause of the type ‘X is sitting next to me’, because such phrases are, of course, incompatible with the absentive. Consider for instance the following example:

(17)

Swedish *John sitter bredvid mig. Han är och skriver ett brev. John sit:PRS next_to me he is and write:PRS a letter ‘John is sitting next to me. He is off writing a letter.’

Another point which has to do with the property of absence is that the person who is absent should not be visible to S. For that reason, examples such as (18) do not occur, because S has a direct perception of the person absent:

(18)

Hungarian *Látom Pétert az ablakból, futballozni van. see:1SG Peter:ACC the window:ELAT play_football:INF is ‘I see Peter from the window, he is off playing football.’

This condition also holds in the following, somewhat artificial situation. S is walking outside while speaking with someone on a portable telephone. At the same time Imre happens to walk along with S. When the other person on the phone inquires about where Imre is, S cannot use the absentive to indicate that Imre is off shopping (cf. 19a). The appropriate expression in this context would be (19b):

(19)

Hungarian a. *Imre itt megy mellettem, vásárolni van. Imre here go:3SG next:1SG shop:INF is ‘Imre is walking next to me, he is off doing shopping.’ b. Imre itt megy mellettem, vásárol. Imre here go:3SG next:1SG shop:3SG ‘Imre is walking next to me, he is doing shopping.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

700

700

Casper de Groot

2.2. Duration When the absentive is used, there is an assumption about the duration of the absence of the Subject. Both Speaker and Addressee should hold this assumption. The following dialogue illustrates this observation: (20)

Hungarian A: Péterrel szeretnék beszélni. Peter:COM like:COND:1SG speak:INF B: De hiszen, futballozni van. well play_football:INF is A: Ja persze. Nyolckor visszahívom indeed at_eight back_call:1SG ‘I would like to speak with Peter.’ ‘Well, he is off playing football.’ ‘Indeed. I’ll call back at eight ’o clock.’

Participant B could have answered the initial question in a different way had he assumed A did not know about the activities of Peter. In that case he might have used something equivalent to “Peter is not here at the moment. You may call him after eight ’o clock.” One may, of course, also find the following type of dialogue: (21)

Hungarian A: Péterrel szeretnék Peter:COM like:COND:1SG B: De hiszen, futballozni well play_football:INF A: Nem is tudtam, not even know:PAST:1SG ‘I would like to speak with Peter.’ didn’t know that he plays football.’

beszélni. speak:INF van. is hogy jár futballozni. that go:3SG play_football:INF ‘Well, he is off playing football.’ ‘I

I do not, however, consider (21) a counter-example to the observation that there must be an assumption about the duration of absence. In the case of (21), participant B made the wrong assumption as to the pragmatic knowledge of A. By using the absentive, B wishes to inform A about the predictable period of time of the activity Peter is engaged in, but not about the fact that Peter is a football player. The period of time someone is engaged comprises all activities which are connected to the activity specified by the lexical verb. The number and type of activities are not determined lexically but pragmatically. The routine of Peter may for instance be:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

701

The absentive

(22)

701

a. leave the house at 2 p.m. b. go to the football field c. change clothes d. do warm up exercises e. play a football match f. take a shower g. get dressed h. go to a bar i. have a beer j. go home k. arrive home at 8 p.m.

2.3. Absentive is stative The basic information expressed by the absentive is that somebody is not present but somewhere else. That is why the absentive can be used as an answer to a question of the type Where is X?, and therefore the absentive must be considered a stative construction. For instance: (23)

German A: Wo ist der Hans? where is the Hans B: Er ist schwimmen. he is swim:INF ‘Where is Hans?’ ‘He is off swimming.’

By using the absentive B leaves open several possibilities. Hans may be on his way to the swimming pool, in the swimming pool, or on his way back home. In many languages, stative constructions cannot be used with deontic modality, with the imperative or as the complement of the verb want, because in all these cases an Agent is required. It is possible to say Close the door!, but not *Be 68 years old! It seems that the absentive cannot be used with deontic modality (cf. 24a), with the imperative or as the complement of the verb ‘want’. The absentive can – like all stative expressions – be used with the optative (cf. 24b).9 (24)

Dutch a. *Je moet vanmiddag zwemmen zijn! you must this afternoon swim:INF be:INF ‘You must be off swimming this afternoon!’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

702

702

Casper de Groot

b.

Ik wou dat ze zwemmen waren. I would that they swim:INF were ‘I wish they were off swimming.’

2.4. Direction The absentive encompasses two directional movements. First there is the movement away from the deictic centre and then the movement towards the deictic centre. Verbs which specify just one direction, such as ‘go to’ or ‘come from’, are therefore incompatible with the absentive. The use of adverbial phrases which specify a direction of the type ‘towards’ or ‘from’ are, for the same reason, not compatible, because the absentive involves both opposite directions. Consider: (25)

Hungarian *Feri sétálni van a pályaudvarra. Feri walk:INF is the railway_station:SUBL ‘Feri is off walking to the railway station.’

3. Aspect 3.1. Progressive versus absentive None of the eight languages which has an absentive has a fully grammaticalized progressive. The languages have different types of constructions which express progressivity. For instance, Dutch has three ways of expressing progressivity. The most common way to express progressivity is (26a). Constructions with postural verbs such as (26b) are also quite common, but have more restrictions. Construction (26c) has many restrictions. Compare the three types of progressive with the absentive (27): (26) a.

b.

c.

Dutch (progressive) Jan is een brief aan het schrijven. John is a letter at the write:INF ‘John is writing a letter.’ Jan zit een brief te schrijven. John sits a letter to write:INF ‘Jan is writing a letter.’ Jan is bezig een brief te schrijven. John is occupied a letter to write:INF ‘John is writing a letter.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

703

The absentive

(27)

703

Dutch (absentive) Jan is een brief schrijven. John is a letter write:INF ‘John is off writing a letter.’

The absentive in Dutch differs from the three progressive constructions in that they denote that John is writing a letter at the moment of speech. With sentence (27) John may be on his way to the place where he usually writes letters, e.g., his office, writing a letter, or on his way back home. Moreover, constructions such as (26) may be coordinated. It is not possible to coordinate the absentive construction with one of the progressive constructions. The languages with an absentive do not allow the combination of the absentive and a phrase of the type being busy or being occupied. The illustration for this comes from Hungarian, in which it is possible to combine the phrase el van foglalva ‘s/he is busy’ with a nominalization expressing the activity (28a), but not the absentive (28b). (28) a.

b.

Hungarian El van foglalva levélírással. PREV is occupied letter-writing:INSTR ‘S/he is busy writing letters.’ El van foglalva. *Levelet van írni. PREV is occupied. letter:ACC is write:INF ‘S/he is busy. S/he is off writing letters.’

The absentive and a progressive form may syntactically be very similar, as for instance in Fering (Ebert 1989) and Norwegian. The absentive uses copula ‘be’, and the progressive construction uses a postural verb such as ‘sit’ or ‘stand’. Compare: (29) a.

b.

(30) a.

Fering Hat as tu säien. she is to sew:INF ‘She is off sewing.’ Hat seed tu säien. she sit to sew:INF ‘She is sewing.’ Norwegian Jan er og skrivar eit brev. John is and write:PRS a letter ‘John is off writing a letter.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

704

704

Casper de Groot

b.

Jan sit og skrivar eit brev. John sit and write:PRS a letter ‘John is writing a letter.’

The situation of Finnish is of particular interest here. The absentive form in Finnish also allows for the progressive reading. Depending on contextual differences, sentences with the copula and the third infinitive marked by the inessive case may have two interpretations: (i) the absentive when answering the question ‘where are you / will you be?’, and (ii) the progressive answering ‘what are you doing?’. Word order may also differentiate between the absentive and the progressive reading. Consider the following two examples and note the different order of constituents: (31)

Finnish a.

(absentive) Minä olen lippuja myy-mä-ssä I am tickets:PRTV sell-3INF-INESS ‘I am off selling tickets.’ b. (progressive) Minä olen myy-mä-ssä lippuja. I am sell-3INF-INESS tickets:PRTV ‘I am selling tickets.’

The Finnish expression with the 3rd infinitive  inessive case using a directional verb can only have the progressive interpretation and not the absentive (see section 2.4 above). Consider: (32)

Finnish Hän on ui-ma-ssa rantaan. he is swim-3INF-INESS shore:ILL ‘He is swimming to the shore.’ (progressive) *‘He is off swimming to the shore.’ (absentive)

3.2. Imperfective/perfective aspect in Hungarian From aspectual pairs in Hungarian, such as imperfective ír / perfective megír ‘write’, only the imperfective form can be used in the absentive. Compare: (33) a.

Hungarian Zsuzsa a leckét írni van. Zsuzsa the lesson:ACC write:INF is ‘Zsuzsa is off doing her homework.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

705

The absentive

705

b. *Zsuzsa a leckét van megírni. Zsuzsa the lesson:ACC is PFV:write:INF ‘Zsuzsa is off doing her homework.’ When Hungarian uses perfective megír it is not possible to refer to a point of time within the state of affairs, because with perfective aspect the state of affairs is presented as an indivisible unit with beginning, middle and end rolled into one (cf. De Groot 1995b). Therefore the perfective cannot be used in the absentive, because when using the absentive, reference is made to a point of time within the state of affairs designated by the verb. Perfective forms such as fölad ‘post’ in (1a), in which the preverb has some meaning of its own or contributes to the meaning of the verb, may be used. Note, however, that these forms are less perfective than, for instance, megír, where the pure aspectual marker meg- neither has any meaning of its own, nor contributes to the meaning of the verb.

4. Variable properties of the absentive The absentive imposes selection restrictions on the class of verbs which can be used. This section discusses several types of restrictions. It also addresses the passive, as well as the expression of the object and time phrases of the type ‘for an hour’.

4.1. Class of verbs 4.1.1. Agentive verbs It seems that the class of verbs which can be used with the absentive is almost entirely limited to those verbs which are agentive. Verbs which do not have an Agent, i.e., a controller of the action designated by the verb, cannot be used in the absentive in Dutch, German, Hungarian, and Italian. For that reason, the following examples are ungrammatical: (34) a.

German *Hans ist Geschenke kriegen. Hans is presents receive:INF ‘Hans is off getting presents.’ b. Italian *Il pullman è a girare. the bus is at turn:INF ‘The bus is off turning.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

706

706

Casper de Groot

All northern languages (Fering, Finnish, Norwegian, and Swedish), however, marginally allow non-agentive Subjects in the absentive. In those cases there must be an interpretation available under which the absentive is explained as part of intentional activity. For instance (i) the human Subject goes some place to be the undergoer of some activity (cf. 35a), or (ii) somebody has removed the Subject in order to do something with it (cf. 35b–c). Still, in both cases there should be an assumption about the period of time the activity will last. Consider:

(35) a.

b.

c.

Norwegian Jan er og får presanger. John is and get presents ‘John is off getting presents.’ Finnish Veturi on käänty-mä-ssä. locomotive is turn-3INF-INESS ‘The train engine is (off) to be turned.’ Fering A hingst as tu bislauen.10 the horse is to shoe:INF ‘The horse is off being shoed.’

The crucial parameter here seems to be ‘controllability’. Note the following points. Firstly, the absentive is used to give information about the absence of persons rather than that of things. Secondly, the absentive assumes a number of adjacent activities which will be performed by the person who is absent, among others ‘going away’ and ‘coming back’. Thirdly, the Subject of the verb should have control over the duration of the absence, because the duration of absence is a relevant property of the absentive. Fourthly, in most languages the absentive does not allow the syntactic passive (see section 4.2 below). We can now formulate two types of conditions on the selection of classes of verbs that can be used in the absentive:

(36) i.

ii.

The strong condition: there is a controller of both the activity and the situation specified by the Subject. Therefore only agentive verbs can be used (Dutch, German, Hungarian, and Italian); The weak condition: there is a controller of only the situation. The controller may be specified by the agent, but may also not be specified at all. Therefore both agentive and non-agentive verbs can be used (Fering, Finnish, Norwegian, and Swedish).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

707

The absentive

707

4.1.2. Causative verbs Causatives, which are agentive too, occur in the absentive as well. Some examples are (37): (37) a.

b.

Hungarian Mari a kutyát van sétáltatni. Mary the dog:ACC is walk:CAUS:INF ‘Mary is off walking the dog.’ Italian Gianni è a farsi tagliare i capelli John is at have:RFL cut:INF the hair:PL ‘John is off having his hair cut.’

In Dutch, Fering, German, Hungarian, and Italian, it is, however, not possible to overtly express the Causee. Consider example (38c), which is contrasted with (38a), which demonstrates the causative with overt expression of both the Causer and the Causee in a non-absentive construction, and (38b), which is the absentive without an overt Causee: (38)

Hungarian a. Imre a szerel˝ovel javíttatja a rádiót. Imre the mechanic:CAUSEE repair:CAUS:3SG the radio:ACC ‘Imre is having the mechanic repair the radio.’ b. Imre a rádiót van javíttatni. Imre the radio:ACC is repair:CAUS:INF ‘Imre is off having the radio fixed.’ c. *Imre a rádiót van javíttatni a Imre the radio:ACC is repair:CAUS:INF the szerel˝ovel. mechanic:CAUSEE ‘Imre is off having the mechanic repair the radio.’

Finnish, Norwegian and Swedish behave differently in the sense that they do allow the overt expression of the Causee. Consider: (39) a.

Norwegian Jan er og får bilmekanikaren til å reparere John is and get:PRS car-mechanic:DEF to to repair:INF bilen hans. car:DEF his ‘John is off having the mechanic have his car fixed.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

708

708

Casper de Groot

b.

Finnish Jussi on korja-u-tta-ma-ssa autoaan John is repair:CAUS:PASS:3INF:INESS car:PRTV:POSS3SG naapurin pojalla. neighbour:GEN boy:ADESS ‘John is off having the neighbour’s boy fix his car.’

We may relate the difference between the two groups of languages to the two types of conditions formulated above. Because of the prominent role of the Agent in the absentive as the controller of both the event and the duration of the absence, there may not be room for a second Agent, i.e., the Causee, who is the controller of the embedded event. The strong condition rules out the possibility of an overt Causee, whereas the weak condition does not. 4.1.3. Dynamicity All languages with the exception of Fering and Norwegian allow non-dynamic verbs such as ‘sleep’ or ‘sit’ in the absentive. Some languages need a phrase of the type ‘(for) an hour’ in order to produce a grammatical sentence (Dutch, German, Swedish). Finnish and Italian do not allow the extension and in Hungarian the extension is optional. Compare:

(40) a.

b.

c.

German *(eine Stunde) hinlegen. Paul ist sich Paul is himself one hour lie ‘Paul is off having a nap for an hour.’ Italian Gianni è a riposare (*un’ora). John is at_ rest:INF an hour ‘Paul is off having a rest.’ Hungarian Pali aludni van (egy órát). Paul sleep:INF is one hour:ACC ‘Paul is off having a nap (for an hour).’

Although the lexical verbs in (40) do not designate dynamic states of affairs, the verbs can be used because the absentive construction refers to situations which comprise several activities including the activities “going some place” and “coming back”.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

709

The absentive

709

4.2. Passive In Dutch, Fering, German, Italian, and Norwegian, the absentive is incompatible with the syntactic passive.11 Consider for instance the following examples from Dutch. Example (41a) shows the passive construction in a non-absentive clause, whereas (41b) shows the ungrammaticality of the passive in the absentive. No difference is made between passives with a human and non-human Subject. They are both ungrammatical. (41)

Dutch a.

Marie wordt (door Alex) geholpen. Mary PASS.AUX (by Alex) help:PRT ‘Mary is being helped (by Alex).’ b. *Marie is geholpen worden. Mary is help:PRT PASS.AUX:INF ‘Mary is off being helped.’ The incompatibility of the passive and the absentive may be related to the fact that the passive puts more emphasis on the Patient and reduces the relevance of the Agent. Note that the Agent plays an important role in the use of the absentive (cf. section 4.1.1 above). Some of the languages allow for non-agentive subjects. One of them, Swedish, also allows for the passive and the absentive. Consider: (42)

Swedish Han är och blir fotograferad he is and becomes photographed ‘He is off getting photographed.’

4.3. Boundedness 4.3.1. Complete activity One way of indicating the duration of the absence is the specification of a complete action, e.g., ‘post a/the letter’, ‘buy a bunch of flowers’, or ‘build a house in France’. The first two activities will not last very long, whereas the last one may take several months or more than a year. Compare the following two examples: (43) a.

Dutch Adam is een brief posten. Adam is a letter post:INF ‘Adam is off posting a letter.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

710

710

Casper de Groot

b.

Hungarian Mihály házat építeni van Franciaországban. Michael house:ACC build:INF is France:INESS ‘Michael is off building a house in France.’

The use of (43a) supposes for instance that Adam leaves the house, walks to a mailbox, puts the letter in the mailbox, and walks back home. If the answer to the question ‘Could I speak with Michael, please?’ were (43b), one would not expect Michael to be home soon. One expects him to return after he has finished building the house. The specification of location in (43b) is relevant. If the person referred to by the subject were occupied with building a house in the neighbourhood, an answer such as (44) would be appropriate: (44)

Italian Gianni è a restaurare una casa nel centro storico. John is at restore:INF a house in centre historical ‘John is off restoring a house in the historical centre.’

With respect to the examples such as (44), Hungarian and Finnish follow the Italian pattern. The other five languages, however, do not. They do not allow an equivalent of (44). Compare: (45) a.

Dutch *Michael is hier in het dorp een huis bouwen. Michael is here in the village a house build:INF ‘Michael is off building a house here in the village.’ b. German *Hans ist in einem anderen Stadtteil ein Haus Hans is in an other part_of_the_city a house bauen. build:INF ‘Hans is off building a house in another part of the city.’

One explanation for the fact that Finnish, Hungarian and Italian allow examples such as (44) is that in these languages “to build a house” can be used as an atelic activity, whereas this is not possible in the other languages. For that reason example (46) can have only one interpretation in Dutch: Michael will return after he has finished building the house. Consider: (46)

Dutch Michael is een huis bouwen. Michael is a house build:INF ‘Michael is off building a house.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

711

The absentive

711

Languages behave differently with respect to transitive verbs in the absentive. Fering only allows intransitive verbs. Transitive verbs can be used after ‘object incorporation’. Other languages allow both intransitive and transitive verbs in the absentive. As for the overt expression of the object of transitive verbs, the languages may have different restrictions.In some cases the object can be left unexpressed; in other cases there must be overt expression of the object. Consider for instance the following examples from Dutch: (47)

Dutch a.

Marie is koffie drinken. (object incorporation) Mary is coffee drink:INF ‘Mary is off drinking coffee.’ b. Marie is een kopje koffie drinken. (with object) Mary is a cup coffee drink:INF ‘Mary is off drinking a cup of coffee.’ c. *Marie is drinken. (no object) Mary is drink:INF ‘Mary is off drinking.’ Another type of complete activity based on the use of directional verbs cannot occur in the absentive. This type of construction is discussed in section 2.4 above. 4.3.2. Time phrase A second way of indicating the duration of the absence is a specification of time, such as ‘two hours’, ‘yesterday’, or ‘a month’. For instance: (48) a.

b.

Hungarian Jen˝o tegnap biciklizni volt. Eugene yesterday cycle:INF was ‘Eugene was off cycling (all day) yesterday.’ Italian Gianni è a studiare i gorilla per un mese. John is at study:INF the gorilla for a month ‘John is off studying gorillas for a month.’

The languages may have (sometimes very subtle) restrictions on the addition of time phrases. When using a non-dynamic verb in the absentive, some languages need a time phrase (see section 4.1.3 above).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

712

712

Casper de Groot

4.3.3. No specification In those cases in which there is no indication of the duration in the sense of a complete activity (section 4.2.1) or by means of a time phrase (section 4.2.2), there is the expectation of what is normally the case, because the activity expresses the occupation, hobby, personal habit, or stretch of time a particular activity usually takes. Consider (49), where the specification of time indicates the normal period of time in which the person is involved in the action: (49) a.

b.

c.

Hungarian Klára dolgozni van. (9 a.m. – 5 p.m.) Klara work:INF is ‘Klara is off working.’ Zsolt úszni van. (7 a.m. – 8 a.m.) Zsolt swim:INF is ‘Zsolt is off swimming.’ Pista ebédelni van. (12. a.m. - 2 p.m.) Steve have_lunch:INF is ‘Steve is off having lunch.’

It seems that there is the following type of pragmatic constraint on the use of the absentive in which there is no specification of the duration of the activity. If a verb cannot be associated with an activity people do for a living or a hobby, or regularly for some other reason, the verb will not be used in the absentive construction.

4.4. Directional adverbs with the absentive Some languages allow a directional adverb in combination with the absentive. The behavior of the adverb is different between most languages. First consider Dutch. (50)

Dutch Jan is uit spelevaren. John is out boat:INF ‘John is out boating.’

The use of uit in Dutch is a bit old-fashioned. It is, however, interesting to note that the adverb is used only with verbs associated with ‘pleasant things’. The adverb does not combine with the verb boksen, even if it expresses someone’s hobby. There may be a semantic difference between the construction with and without uit. Compare the following two expressions in Dutch. Sentence (51a) can be used to tell that John is having his lunch break. Sentence (51b) cannot be used in this context. It indicates that John is out to dinner (for pleasure).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

713

The absentive

(51) a.

b.

713

Dutch Jan is eten. John is eat:INF ‘John is off having lunch/dinner.’ Jan is uit eten. John is out eat:INF ‘John is out having dinner.’

The use of aus ‘out’ in German is much more limited, but there is the same distinction as in Dutch. The other languages do not have this property. Hungarian and Finnish do not allow any kind of locational or directional adverb in the absentive. Fering prefers to use an adverb because S and A know the topography of the island and the places where the activities are typically carried out, for instance: (52)

Fering Jan as deel/ap/auer/ütj/aam tu boksin. John is down/up/over/out/around to box:INF ‘John is down/up/over/out/around [the corner] boxing.’

The use of fuori ‘out’ with the absentive in Italian indicates that the person referred to by the subject is outside the building. Compare: (53) a.

b.

Italian Gianni è a fare fotocopie John is at make:INF photocopies ‘John is off making photocopies [probably inside the building].’ Gianni è fuori a fare fotocopie John is out at make:INF photocopies ‘John is off making photocopies [somewhere outside the building].’

The use of via ‘away’ with the absentive in Italian indicates that the activity takes a rather long time, for instance because the place where the activity indicated by the infinitive will be performed is not very near the deictic centre. Compare for instance the following two examples, where the use of via in (54b) indicates a longer duration of absence.12 (54) a.

b.

Italian Gianni è a pescare. John is at fish:INF ‘John is off fishing.’ Gianni è via a pescare. John is away at fish:INF ‘John is off fishing.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

714

714

Casper de Groot

5. The typological situation 5.1. Spatial and temporal deixis Languages may have grammatical devices to specify spatial and temporal deixis. Spatial deixis prototypically correlates with objects and temporal deixis with events or situations. Objects can be located in space, for instance by means of case markers or adpositions, whereas events can be located in time, for instance by means of tense inflection. The following typology seems to hold for the languages in the world: (55)

deixis spatial temporal

category noun verb

There are, however, some exceptions to this typology in the sense that languages sometimes also have grammatical distinctions for the location of events in space.13 The absentive would then constitute another example of an exception to the general typology. Because the absentive does not fit in with the general typology of deixis in natural languages, it cannot be expected that the absentive will be a widely spread grammatical category. The absentive has so far only been found in some languages of Europe.

5.2. Invariable properties of the absentive The absentives in the eight languages have a number of properties in common. These properties are closely related to the defining properties of the absentive. They can be summarized in the following way. (i)

Remoteness The absentive implies remoteness of a person (sometimes an object) from a deictic centre. That is why the absentive does not combine with an adverbial phrase of the type next to me.

(ii)

Distance When using the absentive the distance between the person absent and the deictic centre should not be (too) short. The person absent must be invisible. For that reason the absentive cannot be used with verbs denoting direct (visual) perception.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

715

The absentive

715

(iii)

Temporal boundedness When the absentive is used, there is an assumption about the duration of the absence of a person. Because the absentive encompasses both directions “going away” and “coming back” directional verbs cannot be used, because they specify just one direction and exclude the second one.

(iv)

Dynamicity The absentive is a (non-agentive) stative expression. For that reason the absentive cannot be used with deontic modality and imperative illocution.

1. 2. 3. 4. 5. 6. 7. 8.

remoteness next-to-me phrase distance direct perception directional verbs stative deontic modality imperative

Dut

Fer

Fin

Ger

Hun

Ita

Nor

Swe

















































5.3. Variable properties of the absentive With respect to the use of the absentive in the eight languages there is some variation. The basic parameters are the following. (i)

Controllability This parameter relates to the requirements imposed on the classes of verbs which can be used in the absentive (agentive or non-agentive verbs), the possibility of the overt expression of the causee in causative absentive constructions, and the possibility of using the absentive in the passive.

Dut 9. 10. 11.

non-agentive verbs overt causee passive

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Fer

Fin



 

Ger

Hun

Ita

Nor

Swe

 

  

716

716 (ii)

12. 13. 14. 15.

Casper de Groot

Telicity This parameter relates to the use of telic and atelic events in the absentive, the necessity of specifying the duration of an activity, or the overt expression of the object of transitive verbs.

telic events atelic events non-dynamic events with an hour

Dut

Fer

Fin

Ger

Hun

Ita

Nor

Swe





  



   

  











Those languages which allow atelic events in the absentive (“John is off restoring a house in the centre.”) also allow atelic non-dynamic events (“John is off sleeping.”) as shown by rows 13 and 14. Row 15 shows that some other languages allow nondynamic events only if specified for duration (“John is off sleeping for an hour.”). (iii)

16. 17. 18.

(iv)

The use of adverbs This parameter relates to the possibility of using an adverb similar to ‘out’ and the semantic modification of ‘pleasure’. It also relates to the possibility of using different adverbs, i.e., adverbs other than ‘out’.

adverb out/away pleasure/non-pleasure other adverbs

Dut

Fer

 

 

Ger

Hun

Ita

Nor

Swe

 













?

Alternative expressions Some languages have alternative, less specific expressions of absence. Dut

19.

Fin

alternative forms



Fer

Fin

Ger





Hun

Ita

Nor

Swe

5.4. Progressivity For typological differences between progressive and absentive, I refer the reader to the introductory chapter to this part of the volume by Bertinetto, Ebert and De Groot.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

717

The absentive

717

6. The absentive and other languages The strict criterion used in this study to consider a language to have an absentive is that a language must have grammatical means to express absence. For that reason Danish has been excluded, because the Danish expression of absence requires an adverb such as ude ‘out’ or henne ‘away’. The construction is very similar to the absentive found in Frisian and Swedish. Consider the following example: (56)

Danish Jens er ude at bokse. John is out to box:INF ‘John is off boxing.’

Because of the obligatory use of the adverb Danish was not included in the group of languages with an absentive. However, speakers of Danish do not seem to conceive the adverbs as real deictic adverbs, but rather as elements without any particular meaning. Moreover, speakers of Danish hesitate between two possible spellings of the element preceding the infinitive. The correct alternative seems to be at ‘to’, whereas the other alternative og ‘and’ is also accepted. Note that in this case there is no difference in pronunciation between at and og in Danish. With og Danish strongly resembles the absentive expression in Swedish and Norwegian. Danish also resembles these languages in the way non-agentive verbs and even the passive can be used. Compare: (57) a.

b.

Danish Bussen er henne at vende. bus:DEF is away to turn ‘The bus is off turning.’ Stolen er ude at blive malet. chair:DEF is out to become painted ‘The chair is off being painted.’

The examples with an absentive in this chapter have all been rendered into English using off and a gerund. This type of expression seems to be close to the absentive. Some native speakers of English feel a difference between the use of off opposed to away in the sense that the expression with off could be considered the absentive. The expression with away would be more lexical and would fail to have the pragmatics of the absentive. Compare: (58) a. b.

John is off posting a letter. John is away posting a letter.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

718

718

Casper de Groot

Since the absentive is a “newly discovered” grammatical category, one does not find examples or discussions of phenomena related to the grammatical expression of absence in the grammars of languages. Each language could be investigated on the possible existence of an absentive or on constructions which may have an absentive reading.

Notes 1. I would like to thank the following people for investigating languages on the basis of the questionnaire: S. Anschütz, C. Bache, P.M. Bertinetto, A. Corda, Ö. Dahl, K. Ebert, J.T. Faarlund, P. Harder, A. King, L.I. Komlósi, J. Ladefoged, I. Nedjalkov, J. Orts Molines, K. Polgárdi, R. Thieroff, and H. Tommola. 2. The absentive is not used in some parts in the south of the Netherlands and in Belgium. A detailed description of the construction in Dutch is given in De Groot (1995a). 3. The absentive is used in all varieties of Frisian. Examples in this chapter are all taken from Fering. 4. Swiss German does not have an absentive. 5. A description of the absentive in Hungarian is given in De Groot (1995b). 6. The Norwegian examples are all from the Nynorsk variety. 7. There is one more alternative which is only used in the southern part of the Netherlands and Belgium. This expression uses the verb gaan ‘go’: Jan is gaan boksen. (lit. John is go box). I do not take this type of construction into consideration, because (i) there is the lexical verb gaan ‘go’, and (ii) the construction rather seems to express that somebody has left and is on its way to some place. 8. Cf. Comrie (1985) who uses the notion deictic centre in relation to tense distinctions. Van Werkgem (1994) applies the notion to spatial distinctions. 9. German, however, rather prefers the construction with gehen ‘go’ here: schwimmen gegangen ‘went to swim’. 10. In Fering it is sometimes difficult to distinguish between the construction as the absentive or as a local adverb with a verbal noun, because the infinitive has the form of the verbal noun. For instance, the form bislauen could be ‘to shoe’ or ‘the shoeing’. 11. Finnish and Hungarian do not have a passive which corresponds to the syntactic passive in the Germanic languages. 12. This observation is not shared by all informants of Italian. 13. See for instance the locative verbal expressions in Chadic (Frajzyngier 1987). Lexical expressions of the spatial location of events are not problematic, e.g., John is playing in the garden.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

719

The absentive

719

References Comrie, Bernard 1985 Tense. Cambridge: Cambridge University Press. Ebert, Karen 1989 “Aspektmarkierung im Fering (Nordfriesisch) und verwandten Sprachen”, in: W. Abraham & Th. Janssen (eds.), Tempus – Aspekt – Modus. Die lexikalischen und grammatischen Formen in der germanischen Sprachen, 293–322. Tübingen: Niemeyer. Frajzyngier, Zygmunt 1987 “Ventive and Centrifugal in Chadic”, Afrika und Übersee 70: 31–47. Groot, Casper de 1993 The Absentive. Typological questionnaire E UROTYP . 1995a “De absentief in het Nederlands: een grammaticale categorie” [The absentive in Dutch: A grammatical category], Forum der Letteren 36: 1–18. 1995b “The absentive in Hungarian”, in: I. Kenesei (ed.), Levels and Structures (Approaches to Hungarian, Vol. 5). Szeged: JATE, 45–61. Werkgem, Fienie G. van 1994 “Hic et nunc”, in: Dubbel Nederlands. 23 opstellen voor Simon C. Dik, 71–74. Amsterdam: IFOTT.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

720

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

721

Case Studies

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

722

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

723

Éva Ágnes Csató

Some typological features of the viewpoint and tense system in spoken North-Western Karaim

1. Abstract The present paper discusses typological properties of the system of viewpoint and tense categories in the North-Western dialect of Karaim, an endangered KipchakTurkic language spoken in a non-Turkic linguistic area. Recently recorded material of the dialect spoken in Lithuania and the findings of E UROTYP questionnaires are evaluated. After a presentation of the Karaim language, some typical Turkic viewpoint oppositions are illustrated on the basis of Turkish data, followed by a characterization of some common typological properties of the contact languages. Subsequently, the Karaim system is described, and some hypotheses concerning its typological properties are formulated.

2. The Karaim language Karaim [Kar. kaRAYˇc ]1 is a Turkic language spoken today mainly in Lithuania and Poland by a small group of speakers whose ancestors are said to have migrated from Crimea about six hundred years ago during the reign of Vytautas (1350–1430), Grand Duke of Lithuania. Their Kipchak-Turkic language is closely related to other Kipchak linguistic cognates, for instance, the now obsolete language of the Codex Cumanicus, Crimean Tatar, Karachai-Balkar, and Nogai. In their present linguistic surroundings in North-Central Europe, the Karaims have become bi- or multilingual, also speaking the majority languages of this area, such as Polish, Ukrainian, Belarusan, Russian and Lithuanian. Their religious education has also included the study of Hebrew, since the Karaims are followers of the old and prominent religious community of the Karaites, recognizing the Old Testament as the only source of divinely inspired legislation. An artificial but useful terminological distinction is sometimes made between the Karaites in general and the Turkic-speaking Karaites in particular by referring to the latter as Karaims. This religious identity of the Karaims has played an important role in the maintenance of their Turkic mother tongue.2 Although Karaim scholars wrote their

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

724

724

Éva Ágnes Csató

theological works mostly in Hebrew, Karaim was also used in liturgy and religious practice. Karaim translations of the Bible and a number of prayers and songs witness a long written tradition. There is also a secular literature which began to be printed in the first part of this century. The present survey aims at a partial description of this unique language still used as a native tongue by a few speakers. It should also serve as a reminder of the place Karaim occupies on the European language map. Moreover, Karaim is interesting from a purely typological point of view. A better understanding of its development may give some insights into how a typical Turkic system behaves under the longstanding influence of foreign areal features. The findings could also be relevant for the understanding of processes of typological change in other Turkic languages strongly influenced by Slavic, such as Gagauz spoken in Moldova.

3. A typological confrontation 3.1. Some characteristics of Turkic viewpoint and tense systems The strongly synthetic Turkic languages have an overwhelmingly rich verbal morphology, in which a great number of viewpoint and tense notions are grammaticalized. A short characterization of the most typical categories should serve here as a point of departure for the typological study of the Karaim viewpoint and tense system. For a more detailed account of the terminology used here, see Johanson (this volume). Viewpoint categories such as intraterminality and postterminality encode notions of how a speaker may characterize an event with respect to its natural ‘terminal’ properties. Speakers of a Turkic language have the possibility to express that they are looking at a given event as it is going on, disregarding its limits, which are the beginning and the end of the event. Compare examples (1a) and (1b).

(1) a.

b.

Turkish (PROGQ: 49) Ay¸se, ders boyunca arkada¸sı ile konu¸suyordu. Ay¸se class throughout friend:POSS.3 with talk:IYORDU.PST ‘Throughout class Ay¸se was talking to her friend.’ [In fact she carried on even after class.] Turkish (PROGQ: 50) Ay¸se, ders boyunca arkada¸sı ile konu¸stu. Ay¸se class throughout friend:POSS.3 with talk:DI.PST ‘Throughout class Ay¸se talked to her friend.’ [But as soon as class was over, she suddenly became very silent.]

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

725

The viewpoint and tense system in spoken North-Western Karaim

725

The verbal expression in (1a) containing the intraterminal Past Tense marker -iyordu does not say anything about the beginning or the end of the event. Ay¸se’s talking to her friend might have started and continued after class. A speaker who wants to say that Ay¸se talked to her friend throughout class but stopped when class was over would prefer to use (1b), because the Past Tense formed with -di does not indicate intraterminality and is usually interpreted as referring to an event in its entirety including its beginning and end. A linguistic expression which is marked for an intraterminal viewpoint may also be further specified with respect to focality, that is, it may be ‘focal’, designating that the speaker envisages the event as actually going on, or ‘non-focal’, indicating that the event is regarded as generally, habitually on-going or potentially relevant at the given time. Thus the difference between example (1a) and (1c) is normally interpreted to the effect that (1c) reports on the habitual behaviour of Ay¸se and not on a situation in which Ay¸se was, as a matter of fact, chatting with her friend. Both (1a) and (1c) are marked for intraterminality, though (1c) is ‘non-focal’.

(1) c.

Turkish Ay¸se, ders boyunca arkada¸sı ile konu¸surdu. Ay¸se class throughout friend:POSS.3 with talk:RDI.PST ‘Throughout class Ay¸se used to talk to her friend.’

Postterminal categories express the idea that the speaker views an event after its crucial limit has been transgressed. The crucial limit is defined by the actional content of the given predication. It might either be the beginning or the end of the action. A verb such as Turkish sat- ‘sell’ defines the crucial limit as being the end of the action. A Past Tense form of this verb, sattı, means ‘the subject sold the object’, implying that the subject is not selling the object any more. The verb otur- ‘sit’, on the other hand, means both ‘sit down’ and ‘be seated’. Thus, the crucial limit is defined as the beginning of the action. Consequently, oturdu means: ‘he (has) sat down and might still be sitting’. Using a postterminal form, the speaker indicates that he does not refer directly to a given event but rather looks at it in a ‘diagnostic’ way, after the crucial limit has been transgressed. Such forms have, therefore, inferential shades of meaning in many Turkic languages, as in (2a), expressing that the speaker himself did not witness the event of coming, but concludes from indirect evidence that it has taken place. Postterminal categories may differ with respect to what degree of focality they designate. A postterminal category with a high degree of focality designates a state that has emerged from the event, as in (2b), which does not imply any inferentiality.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

726

726

Éva Ágnes Csató

(2) a.

b.

Turkish Gelmi¸s. come:MIS.PST ¸ ‘He has (apparently) come.’ Ölmü¸s bulunuyor. die:MIS.PART ¸ find:RFL:IYOR.NONPST ‘He is dead.’ (lit. ‘He finds himself in the state of having died.’)

Intraterminality, postterminality and focality are notions which can be used in typological comparisons of the basic viewpoint notions grammaticalized in different languages and, thus, also in Karaim. Modality categories in Turkic languages typically include optative, imperative, conditional, and potential. Terminality categories may also get modal interpretations. A non-focal, non-past, intraterminal category, as, for instance, the Turkish R-form gelir ‘comes’, may be used to express different modal shades of meaning. The usual temporal opposition in Turkic is one of non-anterior versus anterior. Pluperfect categories designating anteriority in the anteriority are also used, mostly as a combination of a postterminality marker with a past tense suffix. The number of conjugational forms in Turkic verbal paradigms is very high, due to the combinability of the categories grammatical person, viewpoint, tense and mood. Actionality is, as a rule, morphologically unmarked in simple Turkic verb forms. Actional modifications may be designated by periphrastic expressions including a gerundial form of the lexical verb construed with an auxiliary verb expressing an actionality notion such as durativity. In example (3), the gerundial form of the Turkish verb et- ‘do’, edip ‘doing’, is combined with the finite form of the auxiliary verb dur‘stand’.

(3)

Turkish (PROGQ: 35) Sevgilisini hayal edip duruyor. darling:POSS.3:ACC dream do:IP.GER stand:IYOR.NONPST ‘He keeps on dreaming of his darling.’

Turkic languages also have a rich system of non-finite verb forms, such as gerunds, participles and infinitives, functioning as predicates in embedded clauses. Some of these forms are marked for particular viewpoint-aspectual and modal distinctions, sometimes with a suffixed subject agreement element. In the left-branching Turkic syntax, subordinating of clauses containing main-clause verbal predicates is not typical.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

727

The viewpoint and tense system in spoken North-Western Karaim

727

3.2. Viewpoint and tense systems in the contact languages The major Slavic languages spoken in North-Central Europe, in the neighbourhood of Western Karaim, are Polish, Ukrainian, Belarusan and Russian. The basic grammaticalized viewpoint distinction in these contact languages is the one holding between ‘imperfectivity’ and ‘perfectivity’, that is, between a ‘non-adterminal’ and an ‘adterminal’ category. Johanson (this volume) defines adterminality as a viewpoint category envisaging the event in the attainment of its crucial limit. This distinction is the result of a historical development of a former actionality opposition between classes of lexical verbs. Today, in most cases, there “exist two parallel sets of verb forms carrying identical lexical meanings” (Forsyth 1970: 1), but distinguished with respect to adterminality. The distinction is often morphologically marked by a prefix, as in the Russian verbs proˇcitat’ and cˇ itat’, both meaning ‘read’, but with a difference that the former is marked for adterminality. In other verbs, a corresponding opposition can be expressed by a suffix as, for instance, in the verbs opisyvat’ and opisat’ meaning ‘describe’, the former being the non-adterminal form. Slavic tense systems are relatively poor. In Polish, for instance, there are two simple tenses, a Non-Past and a Past, whereas the Pluperfect is rarely used in colloquial style. The Future Tense is formed periphrastically. Predicates of embedded clauses are typically expressed by finite verbs forms. The Lithuanian system is in several respects different from the Slavic ones. It contains, though in a less systematic way, formations similar to the Slavic aspect forms. The simple tenses include a Present Tense, dìrb-u ‘I work’, and a Past Tense dìrb-au ‘I worked’. There is also a category designating ‘frequentativity’ used as a simple form only in the past, e.g., rašydavo ‘he wrote several times’.3 Furthermore, Lithuanian has a synthetic Future Tense, e.g., dìrb-siu ‘I shall work’. The auxiliary verb b¯uti ‘be’ is used to form periphrastic expressions with different kinds of participles. A comprehensive study of the possible influence the contact languages have exerted on Karaim would presuppose more knowledge about the Karaim speakers’ linguistic habits in a historical perspective. The contact situation has been very complicated. Speakers who were born in Trakai about sixty years ago spoke Karaim at home, standard Polish and later Lithuanian at school, and learned to write and read Karaim and Hebrew in the religious school. They heard the peasants in the neighbourhood speak Belarusan and applied a dialect of Polish when communicating with the local people. After the Second World War, they had to learn Russian, the new official language. Unfortunately, very little is known about how Karaim speakers actually used these languages. Consequently, the present study confines itself to giving a first evaluation of the typological status of the Karaim system and formulating some tentative thoughts about its relation to some typological features of the contact languages.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

728

728

Éva Ágnes Csató

4. The North-Western dialect The Eastern Karaim dialect once spoken on Crimea is most probably extinct today. Western Karaim has two branches, one of them being the South-Western dialect of Halich and Luck. Most of the speakers of this dialect emigrated or were deported from their home regions after the Second World War. To the best of my knowledge, there are today not more than eight Karaim speakers still living in Halich. The present paper confines itself to describing the second, North-Western, dialect which is the only functioning one, spoken in today’s Lithuania, predominantly in Trakai, Pan˙evežys and Vilnius. It is difficult to estimate the actual number of speakers. According to the Soviet census of 1989, 2,600 persons identified themselves as Karaims (280 of them lived in Lithuania). Although 503 of them stated Karaim to be their mother tongue, only 52 claimed to have a good command of the language. For the present study, spoken material recorded recently (1994) in Vilnius has been analysed. All my informants speak the North-Western dialect, and their average age is above 50. The findings of an evaluation of the E UROTYP questionnaires – on Perfect, Future Time Reference, and Progressive (referred to henceforth as PFQ, FTRQ and PROGQ respectively) – will also be taken into consideration.

5. A typological metamorphosis? Karaim is often regarded to be a Turkic language which has, to a great extent, lost its original Turkic typological character. A closer look at its phonological and grammatical properties, however, gives arguments for claiming that some basic Turkic typological features have been preserved. Morphology has remained mainly synthetic and agglutinative. Some prefixes have been copied from contact languages (such as the Polish po- ‘after’ in Karaim potanda ‘the day after tomorrow’; cf. the corresponding Polish word pojutrze), but these prefixes have not become productive in Karaim. On the other hand, the syntax and especially the basic constituent order have, as summarized below, undergone a real metamorphosis. The main question of the present study is to what degree the system of semantic notions grammaticalized as viewpoint categories correspond to the Turkic type.

5.1. Phonological properties Spoken North-Western Karaim sounds very much like a Slavic language, due to strong palatalization of consonants in front syllables. Thus, the pronoun ‘I’ is pronounced [mj enj ]. This first impression should, however, not blur the fact that the principles underlying the phonological structure are genuinely Turkic. The system

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

729

The viewpoint and tense system in spoken North-Western Karaim

729

Table 1. Basic constituent orders in Turkish and Karaim Verb – Object

OV

VO

Relative clause – Noun Possessive construction Adjective attribute – Noun Adposition – Noun

Srel N Ngen Nposs Adj N N Postposition

N Srel N(poss) Ngen , Ngen N(poss) Adj N, (N Adj) N Postposition, (Preposition N)

of vowels is maintained, including the ö and ü sounds, which are not present in any of the contact languages. The typologically important principle of Turkic syllabic harmony is observed consistently.4 According to the principle of syllabic harmony, syllables are classified as either front (palatal) or back (velar). The North-Western dialect of Karaim differs from other Turkic languages only with respect to the phonetic realization of this principle. Front syllables are pronounced with strongly palatalized consonants. In the environment of palatalized consonants, the articulation of ü and ö may be centralized. Thus, when ü and ö are not in absolute word initial position, they will be rendered as u˙ and o˙ in the transcription of the Karaim examples. There are several variants of e-sounds; one of them is a closed [e] pronounced mostly in first syllables of stems like the verb b’er- ‘give’. Neutralized vowels occur in some suffixes; see for instance -ˇce, in [karayˇc ] ‘in Karaim’ or ‘Karaim (language)’. As this example illustrates, [ ] can stand both in front and in back syllables. There is also a very open [æ] sound, pronounced relatively long [æ] in stressed syllables. Both of these sounds will be written here as a˙ .5 In the transcription used here, both vowel quality and palatalization of consonants will be marked in order to reflect pronunciation, although this is, in most cases, redundant. Palatalization of consonants will be designated with ’. Karaim orthographies, which are based on the Polish, Russian and recently Lithuanian writing systems, indicate the palatal articulation of consonants, so that, for instance, a word written by me as s’˙oz’l’˙ar ‘words’ is rendered in the Lithuanian orthography as sio´zliar.6

5.2. Constituent order The basic constituent order of Karaim is dominantly SVO. Table 1 compares Karaim and Turkish with respect to the basic constituent order. See also Csató (1994). See, for instance, example (4), observing the neutral constituent order.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

730

730 (4)

Éva Ágnes Csató

Bu b’er’˙at’ maya astrï this give:A.NONPST:3SG I:DAT very [SUBJECT] [VERB] [INDIRECT OBJECT] [DIRECT k’˙op b’iy˙an’ˇc. much pleasure OBJECT] ‘This gives me much pleasure.’

Example (5) illustrates a construction in which the two complement clauses (‘that they say that ’ and ‘that they will go to the children’) follow their predicates (‘I have heard’, ‘they say’) and contain main-clause predicate forms. The usual Turkic construction would be a left-branching one, in which the subordinated predicates would be formed with participles or infinitives. See the construction tuydum k’i [aytadlar k’i [barïrlar al’ ulanlara]] in example (5): (5)

al’ V’il’n’ius’d’˙a bun’d’i k’en’esa k’˙or’k’l’˙u nu to Vot kenesa beautiful now this look now Vilnius:LOC such tuydum k’i aytadlar k’i barïrlar hear:DI.PST:1SG that say:A.NONPST:3PL that go:R.NONPST:3PL al’ ulanlara ulanlar da anda ür’˙an’ir’l’˙ar now child:PL:DAT child:PL also there learn:R.NONPST:3PL karayˇc s’˙oz’l’˙am’˙a. Karaim speak:INF ‘Look, this kenesa in Vilnius is so beautiful now, and I have heard that they say that they will now go to the children and the children will learn to speak Karaim there.’

Karaim postnominal relative clauses are introduced by a relative pronoun. Their predicates are formed as main-clause predicates. Such relative constructions have also developed in other Turkic languages as a result of contact with non-Turkic-type languages. See the relative clause construction, yïlïsuvda kayda astrï yaxšï yuvundux, in example (6). (6)

Ed’ik Yusufta, yïlïsuvda kayda astrï yaxšï COP:DI.PST:1PL Yusuf:LOC sauna:LOC where very good yuvundux. wash:RFL:DI.PST:1PL ‘We were at Yusuf’s (house), in the sauna, where we had such a nice bath.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

731

The viewpoint and tense system in spoken North-Western Karaim

731

6. Viewpoint and tense system of the North-Western dialect 6.1. Finite forms All the suffixes mentioned below in the description of the Karaim viewpoint and tense system have a number of morphophonological variants. The most important principle determining their use is the one of syllabic harmony. Thus, for instance, the first person singular Present Tense ending is -˙am in k’el’˙am ‘I come’ and -am in alam ‘I take’, depending on whether the stem is front or back. Standardized forms of the suffixes will be used here for all the variants. Table 2 presents the Karaim system of viewpoint and tense categories used productively in the spoken language. The arrows show the main oppositions. Each category carries a name including the standardized form of a suffix and a functional label. 6.1.1. Non-anterior categories There are two non-anterior forms, (i) the A-nonpast, which is the Karaim Present Tense, and (ii) the R-nonpast, which is also called the ‘Aorist’ in Turcology. See the paradigms of the verb al- ‘take’ in the appendix. The A-nonpast consists of the stem plus the suffix A – originally a gerundial suffix – and pronominal personal suffixes. In the third person singular and plural, a suffixed form of the auxiliary verb tur- ‘stand’ is standardized as the personal ending. In older varieties, this auxiliary was used in all persons, e.g., Crimean Karaim bir-äy-dir-män ‘I give’, literally ‘I stand giving’ (Pritsak 1959: 321).

         

Table 2. The system of viewpoint aspect and tense oppositions in North-Western Karaim Non-anterior

Anterior



Anteriority in anteriority

R-Past

intraterminal

R-Nonpast

 

A-Nonpast

 

intraterminal DI-Past postterminal   GAN-perfect

postterminal

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

GANPluperfect

732

732

Éva Ágnes Csató

The R-form is originally a participle, the so-called ‘Aorist Participle’. There is no terminality opposition between the A-nonpast and the R-nonpast. The distinction which can be expressed by using the R-form instead of the A-form was originally a difference with respect to focality as, for instance, in Turkish, see above §2.1. In present-day usage, however, non-anterior R-forms in Karaim are mostly used to express that the event will take place in the future. Consequently, the A-nonpast is used as a general ‘present tense’, designating all kinds of non-anterior events, those which are actually going on, but also habits in the present, general truths, etc. See, for instance, examples (7a), (7b), and (7c). The R-nonpast, on the other hand, is used to refer to events that are going to take place in the future and events with modal shades of meaning; see examples (8a) and (8b). The old Future formed with the suffix -asï is now only found in lexicalized forms as, for example, k’el’˙as’i ‘future’. (7) a.

b.

c.

(8) a.

b.

Kïzïmïz k’˙or’k’l’˙u karayˇc s’˙oz’l’eyt’. daughter:POSS.1PL beautiful Karaim speak:A.NONPST:3SG ‘Our daughter speaks beautiful Karaim.’ M’en’ D’ianab A al’ n’e koduy iš’l’eys’? do:A.NONPST:2SG I Diana:WITH and now what you yazam b’it’ik. write:A.NONPST:1SG letter ‘And what are you doing now? I am writing a letter together with Diana.’ Yaš ulan astrï t’er’k ür’˙an’˙at’. young child very quick learn:A.NONPST:3SG ‘Young children learn very quickly.’ Tanda uzax yuklarm. tomorrow long sleep:R.NONPST:1SG ‘I will sleep longer tomorrow.’ Ür’˙an’s’˙ak, b’il’ib’iz’. learn:COND:1PL know:R.NONPST:1PL ‘If we learn it, we’ll know it.’

6.1.2. Anterior categories The following four categories designating anteriority are used in spoken Karaim: (i) a simple anterior formed with the suffix DI (DI-past); (ii) the periphrastic form based on the R-participle and the DI-forms of the copula, bearing subject agreement morphology (R-past); (iii) the GAN-perfect; and (iv) the GAN-pluperfect, which is also a periphrastic form consisting of the past participle and the DI-forms of the copula with agreement morphology.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

733

The viewpoint and tense system in spoken North-Western Karaim

733

The opposition between the simple DI-past and the R-past is a viewpoint distinction. Whereas the DI-past is unmarked as regards viewpoint distinctions, the R-past designates intraterminality. Thus, the DI-past may be used to refer to past events in their entirety; see examples (9a), (9b) and (9c). It is also used to refer to events which are denoted with an English Present Perfect. In such expressions, the adverb už’e ‘already’ (copied from Slavic) is often used to stress present relevance, see example (10). (9) a.

b.

c.

(10)

Tuvdum Pon’˙av’˙až’d’˙a. be born:DI.PST:1SG Panev˙ežys:LOC ‘I was born in Panev˙ežys.’ M’en’ anda yumuš et’t’im I there work do:DI.PST:1SG ‘I worked there for four years.’ yet’m’iš yïl Kal’ maya už’e now I:DAT already seventy year ‘I have now completed my 70th year.’

d’˙or’t’ yïl. four year toldu. fulfill:DI.PST

(PFQ: 1) Bunu už’e oxudu. this:ACC already read:DI.PST ‘She has already read this.’

The DI-past is used in the description of a sequence of events in example (11a). The presence of an intraterminal R-past in this chain of DI-past forms would interrupt sequentiality. However, a sequence of events which was habitual or repeated in the past can be described by expressing each event with an intraterminal R-past; see example (11b). In this case, the R-past designates that the speaker envisages the whole global event in an intraterminal perspective. (11) a.

(PFQ: 11) Ür’˙ud’˙um ormanda. K’ep k’en’et’˙a ïlan walk:DI.PST:1SG forest:LOC suddenly snake üs’t’˙un’˙a bastïm. T’iš’l’˙ad’i surface:POSS.3SG:DAT step:DI.PST:1SG bite:DI.PST Aldïm taš da ayaïma. leg:POSS.1SG:DAT take:DI.PST:1SG stone and saldïm yïlanaa. Tïndï. throw:DI.PST:1SG snake:DAT die:DI.PST ‘I walked in the forest. Suddenly I stepped on a snake. It bit me in the leg. I took a stone and threw it at the snake. It died.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

734

734

Éva Ágnes Csató

b.

Utrulašïr ed’ik karïndašïmb , k’et’˙ar meet:R COP:DI.PST:1PL brother:POSS.1SG:WITH go:R da cˇ omunur ed’ik ed’ik ormana COP:DI.PST:1PL forest:DAT and bathe:R COP:DI.PST:1PL g’˙ol’d’˙a. lake:LOC ‘I used to meet my brother, go to the forest, and bathe in the lake.’

The R-past, which I have characterized as [intraterminal], is often used to describe events in the past which are characteristic of a period or are regarded as usual/habitual events; see examples (12a) and (12b). (12) a.

b.

Da ür’˙at’˙uv’ˇc’˙ul’˙ar aytïr ed’l’˙ar n’in’d’i yaxšï and teacher:PL say:R COP:DI.PST:3PL what good karaylar. Karaim:PL ‘And the teachers used to say that the Karaims are so good.’ Öz’˙um astrï k’iˇc’iˇc’ek ed’im da korxar self:POSS.1SG very small:DIM COP:DI.PST:1SG and be afraid:R ed’im barma oram ašïra. COP:DI.PST:1SG go:INF street over ‘I was very small myself and used to be afraid of crossing the street.’

The R-past is used to mark intraterminality in the translation of the following two examples in the Progressive Questionnaire; see (13a) and (13b). Analogous usages occur in my recorded text; see example (14). (13) a.

b.

(PROGQ: 3) T’˙un’˙ag’˙un’ k’eˇc’kurun n’eˇc’ik Adam kayttï y˙uv’g’˙a, yesterday evening when Adam return:DI.PST home:DAT Anna barˇc anuz iš’l’˙ar ed’i. still work:R COP:DI.PST Anna all ‘Yesterday evening, when Adam came home, Anna was still working.’ (PROGQ: 30) K’˙or’d’˙um Adamnï n’eˇc’ik ol salïr ed’i see:DI.PST:1SG Adam:ACC when he throw:R COP:DI.PST t’er’˙až’˙ag’˙a. tašb stone:WITH window:DAT ‘I saw Adam, when he was throwing stones at the window.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

735

The viewpoint and tense system in spoken North-Western Karaim

(14)

735

B’iz’ bunda b’iz’d’˙a baxˇcada t’˙oz’˙ar ed’ik we here we:LOC garden:LOC wait:R COP:DI.PST:1PL b’iz’ alarnï utrulama. K’ir’d’il’˙ar alar baxˇca a, they:ACC meet:INF enter:DI.PST:3PL they garden:DAT we da cˇ ïybalb . alarnï sïyladïx xïyarb they:ACC offer:DI.PST:1PL cucumber:WITH and honey:WITH ‘We were waiting here at home in the garden to meet them. They entered the garden; we offered them cucumber and honey.’

Since a Karaim R-past, as for instance s’˙oz’l’˙ar ed’i ‘she was talking’, can be used in contexts requiring ‘progressive’ descriptions, its use in such cases corresponds to that of the Turkish intraterminal past form konu¸suyordu ‘she was talking’, as in example (1a). There is, however, in Karaim no opposition between focal and non-focal intraterminal past and thus the R-past is used neutrally with respect to focality. The form s’˙oz’l’˙ar ed’i ‘she was talking’ can, therefore, be used as the Karaim equivalent of both Turkish forms konu¸suyordu ‘she was talking’ in (1a) and konu¸surdu ‘she used to talk’ in (1c). Consequently, the Karaim translation of (PROGQ: 49), (see example 15) is ambiguous between the readings of the Turkish examples (1a) ‘was talking’ and (1c) ‘used to talk’. In the translation of (PROGQ: 50), both Turkish and Karaim use the DI-past; see (1b) and (16) respectively. (15)

(PROGQ: 49) B’˙ut’˙un’ vaxtnï s’˙oz’l’˙ar’ ed’i konšub . whole time:ACC talk:R COP:DI.PST neighbour:WITH ‘She was talking to (her) neighbour all the time.’ or ‘Throughout class Ay¸se used to talk to her friend.’

(16)

(PROGQ: 50) B’˙ut’˙un vaxtnï s’˙oz’l’˙aš’t’i konšub ? whole time:ACC talk:RECP:DI.PST neighbour:WITH ‘She talked to (her) neighbour all the time.’ [But as soon as class was over, she suddenly became very silent.]

The simple GAN-perfect designates a postterminal viewpoint. It is highly ‘focal’ and is used to describe a state resulting from an event after the transgression of its crucial limit, as in the following utterances taken from my recorded data; see examples (17a) and (17b). The use of this form is restricted to verbs which are semantically suited to express postterminal states. Its meaning is similar to the meaning of English expressions of the type is gone, is written. (17) a.

Iš’l’˙ag’˙an’ yanï, yïl okrom anar. year only it:DAT make:GAN.PART new ‘It is newly made, only a year ago.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

736

736

Éva Ágnes Csató

b.

(PFQ: 3) B’iy t’ir’il’˙at’m’ anuz? Yo, ol öl’g’˙an’. still no he die:GAN.PART king live:A.NONPST:3SG Q ‘Is the king still alive? No, he is dead.’

The GAN-perfect without a copula can also be used to express inferentiality, as in the following example (18). This inferential meaning is, however, not confirmed by all of my informants. Normally, the DI-past, and not the GAN-perfect, is used in Karaim in situations where the speaker infers the event from the state resulting from it; see example (19). (18)

Ol tuymaan. he hear:NEG:GAN.PART ‘He has apparently not heard it.’

(19)

(PFQ: 14) yamur yavdï. K’eˇc’˙ab’ night:WITH rain rain:DI.PST [It is morning. A wakes up, looks out of the window and sees that the courtyard (or the street) is wet.:] ‘It has rained during the night.’

The GAN-participle can be combined with different forms of the copula bol- ‘be, become’. A combination of the A-nonpast form of bol-, bolat ‘is’, and the GANparticiple expresses habituality which is due to the specific meaning of the A-nonpast form of bol-.7 When the R-nonpast of the copula, bolur, is chosen, the periphrastic expression designates a prospective postterminal state. See examples (20a) and (20b). (20) a.

b.

buzlaan bolat. G’˙ol’ barïb lake completely freeze:GAN.PART be:A.NONPST:3SG ‘The lake is usually completely frozen over.’ Ax barïnda, kar Martta, barï white everywhere snow March:LOC all:POSS.3SG bolur buzlaan. be:R.NONPST freeze:GAN.PART ‘It is white everywhere, there is snow in March, everything will be frozen.’

A particular case is when the participle is used in a construction analogous to a possessive construction, as for instance in example (19). The finite verb of the construction is the copula bart ‘is’ and the participle bears a possessive suffix. (21)

Bart tuyanïm. existent:3SG hear:GAN.PART:POSS.1SG ‘I have heard it.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

737

The viewpoint and tense system in spoken North-Western Karaim

737

The periphrastic GAN-pluperfect is formed with the past participle and the DI-past of the copula e- ‘be’. Since it is the only form designating anteriority in the anteriority, it does not mark a specific viewpoint. See examples (22) and (23).

(22)

n’eˇc’ik Da ot bu ed’i e’l’i ek’in’ˇc’i yïlda, and see! this COP:DI.PST fifty second year:LOC when da maya ed’i eg’ir’m’i to uz yïl, anuz atadan nine year yet father:ABL and I:DAT COP:DI.PST twenty mamadan yïrax yox ed’im k’et’k’˙an’, mother:ABL away not COP:DI.PST:1SG travel:GAN.PART k’er’t’i už’e ata ed’i öl’g’˙an’. true:POSS.3 already father COP:DI.PST die:GAN.PART ‘And you see, this was in ‘52, when I was 29 years old; I had not yet been away from father and mother; father was actually already dead.’

(23)

Kaˇcan maya yomaxladï bu yomaxlarnï, m’en’ alarnï when I:DAT tell:DI.PST this story:PL:ACC I they:ACC yox ed’im. tuy an hear:GAN.PART not existent COP:DI.PST:1SG ‘When he told me these stories, I had not yet heard them.’

6.1.3. Modal categories The following modal categories are used in spoken Karaim: (i) an imperative; (ii) a non-past conditional formed with -sa; (iii) a past conditional marked by -sa  the DI-past of the copula e-; (iv) a non-past optative formed with -gey; (v) a past optative formed with -gey  the DI-past of the copula e-; (vi) a possibility form marked by the suffix -al or its negated form -alma, which designates impossibility; (vii) a periphrastic form expressing possibility, formed with the potential form of the verb bol- ‘be, become’, bolal-, and the infinitive of the lexical verb; (viii) the auxiliary mog- ‘be able’, which has been copied from a Slavic contact language; (ix) the verb yara- ‘be suitable, may’ and the infinitive of another verb. The suffix -sa is used both in conditional and temporal clauses; compare examples (24a) and (b).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

738

738

Éva Ágnes Csató

(24) a.

b.

Kïsxa b’it’ik maya yazsey, astrï short letter I:DAT write:COND:2SG very b’iy˙ani’m. be glad:R.NONPST:1SG ‘If you write me a short letter, I will be very glad.’ (PFQ: 84) Kaytsey k’el’˙as’i yïlda, y’˙uv bolur return:COND:2SG coming year:LOC house be:R.NONPST satxan. sell:GAN.PART ‘When you come back next year, the house will be sold.’

The R-nonpast can also be found in conditional clauses; see example (25), in which the conditional clause is introduced by the conjunction eg’er ‘if’. (25)

(FTRQ: 9) Eg’er bu kapˇcuxka taš koyarsïn, if this bag:DAT stone put:R.NONPST:2SG yïrtïlïr. break:PASS:R.NONPST ‘If you put a stone into this bag, it will break.’

Optative forms can be used to express a wish, both in main clauses and in complement clauses. See examples (26) and (27), respectively. (26)

T’en’r’i b’er’g’ey savlux! God give:OPT health ‘God give (you) health!’

(27)

K’l’eyb’iz’ k’i bol ey. want:A.NONPST:1PL that be:OPT ‘We want it to be so.’

The past form of the optative can also express greater tentativeness or politeness. In the recorded material, it is frequently used with the verb kl’˙a- ‘want’; see examples (28) and (29). (28)

Astrï k’l’˙ag’eyd’im bunu iš’l’˙am’˙a. very want:OPT:DI.PST:1SG this:ACC do:INF ‘I would very much like to do this.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

739

The viewpoint and tense system in spoken North-Western Karaim

(29)

739

M’en’ astrï k’l’˙ag’eyd’im k’i bol ey I very want:OPT:DI.PST:1SG that be:OPT xuppa Vil’n’˙an’in’ k’en’esada. wedding-ceremony Vilnius:GEN kenesa:LOC ‘I would like very much that the wedding ceremony take place in the kenesa in Vilnius.’

No clear semantic difference has been observed between the three modal forms (vi), (vii) and (viii). Each of them can express a range of different modal meanings such as potentiality, possibility, ability; see the following examples (30) and (31). (30)

(31)

(PFQ: 82) Alalïm -m get:AL.POT:A.NONPST:1SG Q axˇcamnï? wage:POSS.1SG:ACC ‘May I get my weekly wage?’

al’

now

aftalïx weekly

mog m aytma k’i Kal’ t’ir’il’˙am, now live:A.NONPST:1SG can:A.NONPST:1SG say:INF that t’ir’l’ik xor t’˙uv’˙ul’. life bad not ‘Now I live and I can say that life is not bad.’

Note that the modal verb mog- ‘be able’, copied from Slavic, is used here with Karaim verbal endings. The conjugated form is stressed on the first syllable, as in (31) [MOg m]. The modal auxiliary bolal- expresses permission in example (32) and ability in (33) and (34). (32)

(PFQ: 5) Ulan sorat: Bolalamm barma? son ask:A.NONPST:3SG may:A.NONPST:1SG Q go:INF iš’iyn’i? Anasï: Iš’l’˙ad’iym’ y˙uv house work:POSS.2SG:ACC mother:POSS.3 do:DI.PST:2SG Q ‘The boy asks: May I go? His mother: Have you done your homework?’

(33)

(FTRQ: 83) Karïndašïm bolalat k’˙ot’˙ur’m’˙a bu brother:POSS.1SG can:A.NONPST:3SG lift:INF this tašnï. stone:ACC ‘My brother can lift this stone.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

740

740 (34)

Éva Ágnes Csató

(FTRQ: 84) Karïndašïm bolalat s’˙oz’l’˙am’˙a b’eš brother:POSS.1SG can:A.NONPST:3SG speak:INF five t’il’d’˙a. language:LOC ‘My brother can speak five languages.’

The verb yara- ‘be suitable, be possible, may’ is often used in the spoken language to express permission or objective possibility; see example (35). (35)

Yareyt k’ir’m’˙a? may:A.NONPST:3SG enter:INF ‘Is it possible to enter?’ or ‘May I enter?’

6.2. Non-finite forms Turkic languages usually have a great number of gerunds, participles and verbal nouns. Some of these non-finite forms may also designate viewpoints. In spoken Karaim, however, the use of gerunds is rather restricted. The following forms are still used: (i) the present participle formed with the suffix -ado on can also function adverbially as a gerund; (ii) a gerund formed with the suffix -ip; (iii) a gerund meaning ‘not doing’ or ‘not having done’ formed with the suffix -mayïn; (iv) other gerunds formed from a participle or a verbal noun plus a case suffix. The gerunds formed with the suffixes -ado on and -ip originally indicated different viewpoints. Traces of these meanings can still be observed. The gerund formed with -ado on seems to indicate intraterminality as, for example, in ba ïnado on alïn a ‘looking forward’ in example (36). The meaning of this Karaim form can be compared to that of the gerund formed with the suffix -arak in Turkish; see Johanson (1971: 261). In examples (37) and (38), the gerunds formed with -ip, k’˙us’˙an’ip ‘having longed for’ and bašlap ‘having started’, have non-intraterminal meanings. Since ip-forms are rare in the spoken language, example (37) is taken from a written literary text. (36)

alïn a k’˙or’m’˙am vaxt Ba ïnado on consider:ADOKON.GER forward see:NEG:R.NONPST:1SG time k’i bol ey bu iš’k’˙a. that be:OPT this work:DAT ‘Looking forward, I do not see that there is any time (left) for this work.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

741

The viewpoint and tense system in spoken North-Western Karaim

741

(37)

K’˙us’˙an’ip tuvmuš yer’n’i, ištïrdïm n’eˇc’˙a long for:IP.GER birth place:ACC collect:DI.PST:1SG some k’˙or’k’l’˙u osmak da k’el’d’im Troxka. nice coin and come:DI.PST:1SG Trakai:DAT ‘Having longed for my place of birth, I collected some nice coins and came to Trakai.’

(38)

(PROGQ: 53) T’˙un’˙ag’˙un’d’˙an’ bašlap suv k’˙ot’˙ur’˙ul’d’˙u. yesterday:ABL begin:IP.GER water increase:PASS:DI.PST ‘From yesterday on, (the level of) the water increased.’

The ado on-form, cï ado onnu ‘leaving:ACC’, is used as a verbal noun indicating intraterminality in the following example (39). (39)

(PROGQ: 21) cˇ ï ado onnu K’˙or’˙am Annanï see:A.NONPST:1SG Anna:ACC leave:ADOKON.PART:ACC iˇc’k’ir’id’˙an’. room:ABL ‘I see Anna leaving the room.’

The following sentence (40) illustrates the use of the gerund MAYIN. (40)

(FTRQ: 22) T’˙un’˙ag’˙un’ t’˙oz’m’˙ayin’ karïndašïmnï yesterday wait:MAYIN.GER brother:POSS.1SG:ACC yattïm. go to bed:DI.PST:1SG ‘Yesterday I went to bed without waiting up for my brother.’

There are some gerundial forms based on a participle with case endings, as in (41) and (42), or with both case ending and a postposition, as in (43). Verbal nouns with postpositions can also be used adverbially, as in example (44). These forms do not, however, designate any viewpoint. (41)

(PFQ: 54) B’il’˙as’m’ n’e kïlïndï what make:PASS:DI.PST know:A.NONPST:2SG Q ulan bolanïmda? m’en’im’b’ I:GEN:WITH child be:GAN.PART:POSS.1SG:LOC ‘Do you know what happened to me when I was a child?’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

742

742

Éva Ágnes Csató

(42)

(PFQ: 12) a aˇc t’˙ub’˙un’˙a, alma Oltur anïmda sit:GAN.PART:POSS.1SG:LOC tree bottom:POSS.3SG:DAT apple t’˙uš’t’˙u bašïm üs’t’˙un’˙a. fall:DI.PST head:POSS:1SG surface:POSS.3SG:DAT ‘When I was sitting under a tree, an apple fell on my head.’ Literally: ‘Being in the state of having sat down under the tree ’

(43)

Kaytxanïmdan burun t’˙ug’˙at’ return:GAN.PART:POSS.1SG:ABL before finish:IMP iš’l’˙ar’iyn’i. assignment:PL:POSS.2SG:ACC ‘Finish your assignments before I return.’

(44)

(FTRQ: 99) N’e iš’l’eyt’ karïndašïy ašamaxtan son? what do:A.NONPST:3SG brother:POSS.2SG eat:DER:ABL after ‘What does your brother do after having eaten?’

Participles may retain their viewpoint meanings also when used attributively. Compare the intraterminal meaning of the participle formed with -ado on in (45a) with the postterminal meaning of the GAN-participle in (45b). This opposition is, however, limited to expressions in which a choice between the two participles is possible.

(45) a.

b.

tarlavlar sar arado on turn yellow:ADOKON.PART field:PL ‘fields which are turning yellow’ tarlavlar sar ar an turn yellow:GAN.PART field:PL ‘fields which have turned yellow’

The attributive use of participles is, however, less frequent in spoken Karaim than in other Turkic languages. Alternative expressions formed with a postnominal relative clause (fil’mi kaysïn k’˙or’d’˙u ‘the films which he had seen’ in example (46b)) are more common than the attributive expression in k’˙or’g’˙an’ fil’mi ‘seen film’ in example (46a), especially when the relative expression is more complex. Note that the verb k’˙or’g’˙an’ in (46a) is not in the passive.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

743

The viewpoint and tense system in spoken North-Western Karaim

(46) a.

b.

743

(PFQ: 54) Kar for n’eˇc’ik utruleym anï, ol maya each time when meet:A.NONPST:1SG he:ACC he I:DAT yomaxleyt k’˙or’g’˙an’ fil’m’i. tell:A.NONPST:3SG see:GAN.PART film:ACC ‘Each time when I meet him, he tells me about the films which he has seen.’ (PFQ: 79) utrulašïr ed’ik ol Kar for n’eˇc’ik anïnb each time when he:GEN:WITH meet:R COP:DI.PST:1PL that yïlarda yomaxlar ed’i maya fil’m’i, year:PL:LOC tell:R COP:DI.PST I:DAT film:ACC kaysïn k’˙or’d’˙u. which:POSS.3:ACC see:DI.PST ‘Each time when I met him in those years, he used to tell me about the films he had seen.’

7. The typological status of Karaim The main typological difference between the Turkic and the Slavic viewpoint and tense types is that the grammar of Slavic languages requires, as a rule, that the speaker choose between an ‘adterminal’ and a ‘non-adterminal’ expression when speaking about an event. As mentioned before, the Slavic aspect distinction has developed from distinctions of actionality. Karaim has no corresponding devices to designate such aspect distinctions systematically, although there are some Turkic methods of indicating actional modifications which could have been used for such purposes. The most common way of marking actional modifications in Turkic is using a periphrastic construction containing a gerund and an auxiliary verb. Such constructions are also present in Karaim, though they are less often used. Expressions designating ‘to be about to do something’ are, for instance, constructed with a gerund plus the auxiliary verb yez- ‘miss’, which is no longer used as a lexical verb in Karaim. (47)

(FTRQ: 85) Yuklayez’d’im. fall asleep:A.GER:YEZ:DI.PST:1SG ‘I was about to fall asleep.’ Literally ‘I missed falling asleep.’

There is also a suffix -xala denoting frequentative or iterative events; see the verb aˇcxala ‘open several times’ in example (48). Considering the fact that Slavic iteratives have developed into “imperfectives”, this Karaim suffix might also have been a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

744

744

Éva Ágnes Csató

candidate for becoming a non-adterminality marker, but such a development has not taken place. The suffix -xala designates a pure actional modification of the lexical meaning of the verb. (48)

Aˇcxaladïym t’er’˙až’˙an’i? open:XALA:PST:2SG Q window:ACC ‘Have you opened the window (several times)?’

Speakers of Karaim have, of course, been influenced by their multilingual linguistic habits. Many examples could be cited here to illustrate that Karaim forms are used as translations of expressions in the contact languages. There is actually a remarkable affinity between the usages of, for instance, Lithuanian and Karaim verbal categories, which helps the speaker to find functional correspondence between the forms. The Lithuanian Present Tense, dìrb-u ‘I work’ often corresponds to the Karaim Anonpast form iš’l’eym ‘I work’, the Lithuanian Past Tense dìrb-au ‘I worked’ to the Karaim DI-past iš’l’˙ad’im ‘I worked’, and the Lithuanian synthetic future dìrb-siu ‘I will work’ to the Karaim R-nonpast iš’l’˙ar’m ‘I will work’. The Lithuanian frequentative category dirb-dav-au ‘I used to work/I worked several times’ is often, but not always, a translation equivalent of the Karaim intraterminal R-past iš’l’˙ar ed’im ‘I used to work/I was working’. Participle constructions also show clear correspondence. Nevertheless, as I have tried to argue in this paper, the system of underlying semantic oppositions in Karaim can be better understood if we do not deduce them from these functional correspondences, but rather define them on the basis of the language-specific system of viewpoint and tense oppositions. There are, of course, several phenomena that might be interpreted as induced by contact. Let me present one of them here. The Karaim R-nonpast is more frequently used with ‘pure’ future time reference than, for instance, the corresponding Turkish non-past and non-focal R-form. The use of the Karaim R-nonpast corresponds to the use of the Lithuanian synthetic Future Tense (see 2.2), which is frequently used in expressions containing an adverbial element with future time reference: “Im Gegensatz zum Deutschen, wo das Futur seltener ist, muß im Litauischen jede zukünftige (d.h. in der Zukunft eintretende oder erwartete) Handlung mit dem Futur ausgedrückt werden” [In contrast to German, where the Future is less frequent, in Lithuanian, every future action (i.e., an action that will occur or is expected to occur in the future) must be expressed with the Future Tense] (Senn 1966: 52). See, for example, the R-form k’et’˙ar’b’iz’ ‘we shall travel’, designating ‘pure’ future time reference in (49). (49)

Tanda k’et’˙ar’b’iz’ Troxka. tomorrow travel:R.NONPST:1PL Trakai:DAT ‘We will travel to Trakai tomorrow.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

745

The viewpoint and tense system in spoken North-Western Karaim

745

The opposition between a focal and a non-focal form, as observed in Turkish (see examples 50a and 50b) does not seem to be maintained in Karaim. The meanings of the two Turkish forms are slightly different. While the focal form -iyor in (50a) designates that the event ‘I do’ is scheduled, the non-focal R-form only designates that the event is likely to occur. Thus, while (50a) is a relatively certain promise, (50b) is usually interpreted as a vague one. (50) a.

b.

Turkish Yarın yapıyorum. tomorrow do:IYOR.NONPST:1SG ‘I am going to do it tomorrow.’ Yarın yaparım. tomorrow do:R.NONPST:1SG ‘I could do it tomorrow.’

In Karaim, as claimed above, the R-nonpast is more void of modal shades of meaning than the corresponding form in Turkish and its use in expressions of future time reference seems to be rather consistent. The expression (51b) with the adverb tanda ‘tomorrow’ and an A-nonpast form is characterized by several speakers as ‘not good Karaim’. See the Karaim examples (51a) and (51b). (51) a.

Tanda iš’l’˙ar’m. tomorrow do:R.NONPST:1SG ‘I will do it tomorrow.’ b. ?Tanda iš’l’eym. tomorrow do:A.NONPST:1SG ‘I am going to do it tomorrow.’

However, there are reasons to be cautious when ascribing this use of the R-nonpast exclusively to the influence of the Lithuanian synthetic Future Tense, since the influence of Lithuanian has only lately become important. Moreover, the use of the R-nonpast form for designating future time reference might be an internal Karaim development, since similar tendencies can also be observed in other Turkic languages which are not in contact with Lithuanian.

8. Summary The following hypotheses might be formulated about the typological characteristics of the Karaim viewpoint and tense system: (i) The basic viewpoint oppositions in the Karaim system are still intraterminality and postterminality, grammaticalized in a way that is very similar to those found in other Turkic languages.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

746

746

Éva Ágnes Csató

(ii) The viewpoint system of Karaim is less rich than that of more typical Turkic languages. This may be the result of a process of simplification, which is partly due to the fact that Karaim has ceased to be used as a full-fledged vernacular. Focality oppositions which are so central in other Turkic languages are lost both in non- anteriority and anteriority. Inferentiality does not seem to be grammaticalized, as it is in other Turkic languages. (iii) An important typological difference between Karaim and the contact systems is maintained in that Karaim has not developed any opposition of adterminality. At the same time, the typically Turkic ways of expressing actionality modifications are scarcely exploited. (iv) Syntactic changes have led to a reduction of the inventory of non-finite forms. The use of the Optative in certain complement clauses is also characteristic of other Turkic languages that have developed a dominantly right-branching syntax.

Appendix. The paradigm of the verb al- ‘take’. Finite and nonfinite forms8 The tables contain only the forms preferred in spoken language.

Table 3. A-nonpast

1 2 3

Affirmative Singular

Plural

Negative Singular

Plural

aL-A-M aL-A-S aL-A-T

aL-A-bïz aL-A-sïz aL-A-D-lar

AL-m-ï-m AL-mï-s AL-mï-t

AL-mï-bïz AL-mï-sïz AL-mï-d-lar

Affirmative Singular

Plural

Negative Singular

Plural

aL-Ï-M aL- Ï-S aL-ÏR

aL-Ï-bïz aL-Ï-sïz al-ïr-LAR

al-MA-M al-MA-S al-MAS-T

al-MA-bïz al-MA-sïz al-MAS-T-lar

Table 4. R-nonpast

1 2 3

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

747

The viewpoint and tense system in spoken North-Western Karaim

747

Table 5. DI-past

1 2 3

Affirmative Singular

Plural

Negative Singular

Plural

al-DÏ-M al-DÏ-Y al-DÏ

al-DÏ-X al-dï-YÏZ al-dï-LAR

AL-ma-dï-m AL-ma-dï-y AL-ma-dï

AL-ma-dï-x AL-ma-dï-yïz AL-ma-d-lar

Affirmative Singular

Plural

Negative Singular

Plural

aL-ÏR e-d’i-m aL-ÏR e-d’i-y aL-ÏR e-d’i

aL-ÏR e-d’i-k aL-ÏR e-d’i-yiz’ aL-ÏR e-d’-l’˙ar

al-MAS e-d’i-m al-MAS e-d’i-y al-MAS e-d’i

al-MAS e-d’i-k al-MAS e-d’i-yiz’ al-MAS e-d’-l’˙ar

Plural

Table 6. R-past

1 2 3

Table 7. GAN-pluperfect

1 2 3

Affirmative Singular

Plural

Negative Singular

al-KAN e-d’i-m al-KAN e-d’i-y al-KAN e-d’i

al-KAN e-d’i-k al-KAN e-d’i-yiz’ al-KAN e-d’-l’˙ar

al-KAN et’-m’˙a-d’i-m al-KAN et’-m’˙a-d’i-k al-KAN et’-m’˙a-d’i-y al-KAN et’-m’˙a-d’i-yiz’ al-KAN et’-m’˙a-d’i al-KAN et’-m’˙a-d’-l’˙ar

Affirmative Singular

Plural

Negative Singular

Plural

al-KEY-m al-KEY-s al-KEY

al-KEY-bïz al-KEY-sïz al- ey-LAR

AL-ma- ey-m ALma- ey-s AL-ma- ey

AL-ma- ey-bïz AL-ma- ey-sïz AL-ma- ey-lar

Table 8. Optative

1 2 3

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

748

748

Éva Ágnes Csató

Table 9. Optative past

1 2 3

Affirmative Singular

Plural

Negative Singular

Plural

al-KEY-dï-m al-KEY-dï-y al-KEY-T

al-KEY-dï-x al-KEY-dï-yïz al-KEY-D-lar

AL-ma- ey-dï-m AL-ma- ey-dï-y AL-ma- ey-t

AL-ma- ey-dï-x AL-ma- ey-dï-yïz AL-ma- ey-d-lar

Affirmative Singular

Plural

Negative Singular

Plural

al-SA-M al-SE-Y al-SA

al-SA-X al-se-YÏZ al-sa-LAR

AL-ma-sa-m AL-ma-se-y AL-ma-sa

AL-ma-sa-x AL-ma-se-yïz AL-ma-sa-lar

Table 10. Conditional

1 2 3

Table 11. Conditional past

1 2 3

Affirmative Singular

Plural

Negative Singular

Plural

al-SA e-d’i-m al-SE-Y-dï-m al-SA e-d’i-y al-SE-Y-dï-y al-SA e-d’i al-SE-Y-T

al-SA e-d’i-k al-SE-Y-dï-x al-SA e-d’i-yiz’ al-SE-Y-dï-yïz al-SA e- d’-l’˙ar al-SE-Y-D-lar

AL-ma-sa e-d’i-m AL-ma-se-y-dï-m AL-ma-sa e-d’i-y AL-ma-se-y-dï-y AL-ma-sa e-d’i AL-ma-se-y-t

AL-ma-sa e-d’i-k AL-ma-se-y-dï-x AL-ma-sa e-d’i-yiz’ AL-ma-se-y-dï-yïz AL-ma-sa e-d’-l’˙ar AL-ma-se-y-d-lar

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

749

The viewpoint and tense system in spoken North-Western Karaim

749

Table 12. Possibility  A-nonpast Forms of possibility Singular Plural

Forms of impossibility Singular Plural

1

al-aL-A-M

al-aL-A-bïz

aL-AL-mï-m

2

al-aL-A-S

al-aL-A-SÏZ

aL-AL-mï-s

3

al-aL-A-T

al-aL-A-D-lar

aL-AL-mï-t

Plural

Negative Singular

aL-AL-mï-y-bïz aL-AL-mï-bïz aL-AL-mï-y-sïz aL-AL-mï-sïz aL-AL-mï-y-dïr-lar aL-AL-mï-d-lar

Table 13. Imperative Affirmative Singular 1 2 3

al-aYÏM al AL- ïn al-SÏN

al-aYÏX Al-ïyïz al-sïn-LAR

AL-ma-yïm AL-ma AL-ma- ïn AL-ma-sïn

Plural AL-ma-yïx AL-ma-yïz AL-ma-sïn-lar

Table 14. Selected non-finite forms Non-finite

Affirmative

Infinitive Present Participle / Gerund Past Participle Gerund

al-MA aL-Ado on al-KAN aL-ÏP

Acknowledgements The field research in Poland and Lithuania the results of which are reported in this paper was carried out in the framework of a project financed by the Deutsche Forschungsgemeinschaft. I thank Prof. Dr. Hans-Jürgen Sasse of the University of Cologne for his friendly support and willingness to supervise the project. I am most grateful to many members of E UROTYP, especially Lars Johanson, for discussions and comments on this paper. I am deeply indebted to Dr. Aleksander Dubi´nski for translating the E UROTYP questionnaires and for his untiring support

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

750

750

Éva Ágnes Csató

and valuable advice. I also thank my Karaim friends, first of all Halina Kobeckait˙e, Mykolas Firkoviˇcius and Karina Firkavici¯ut˙e, for teaching me their Karaim mother tongue.

Notes 1. The syllable written with capital letters is stressed. Stressed syllables are always highpitched in Karaim. In other Turkic languages, the two features, stress and high pitch, may be realized on different syllables. 2. Another Turkic minority of Lithuania, the Tatars, who did not use their Turkic language in religious contexts, have abandoned their Tatar idiom. 3. Mathiassen (1996: 9) argues that the habitual meaning of this Lithuanian form is only contextual, while the iterative meaning is systematic. 4. Distinctions have to be made between different types of syllabic harmony, such as intrasyllabic and intersyllabic harmony, harmony in stems and harmony in suffixes. See more about the principles underlying Turkic syllabic harmony in Johanson (1991). 5. Note that [˙u], [˙o] and [˙a] are phonologically still classified as front vowels and are also phonetically different from the back vowels [u], [o] and [a]. Since the orthography does not reflect this fact, previous descriptions of the rules of Karaim ‘vowel harmony’ regard these vowels to be phonetically and phonologically back (see, among others, Comrie 1981: 63–64). 6. For more detailed accounts of the phonological system, see Kowalski (1929), Pritsak (1959) and Dubi´nski (1978). 7. The habitual meaning of the Karaim verb bolat is presumably a copy of the meaning of the Polish verb bywa (from bywa´c ‘to frequent, to be (go) often, to happen’). 8. See the presentations of verbal paradigms in Kowalski (1929), Pritsak (1959) and Musaev (1964).

References Comrie, Bernard 1981 The languages of the Soviet Union. (Cambridge Language Surveys.) Cambridge: University Press. Csató, Éva Ágnes 1994 “On word order differences between Turkish and Karaim”, Dilbilim Ara¸stırmaları 1994: 54–61. Deny, Jean & Kaare Grønbech & Helmut Scheel & Zeki Velidi Togan (eds.) 1959 Philologiae Turcicae Fundamenta. Vol. 1. Wiesbaden: Steiner. Dubi´nski, Aleksander 1978 “Phonetische Merkmale des Łuck-Halicz Dialektes der karaimischen Sprache” [Phonetic features in the Luck-Halich dialect of Karaim], Rocznik Orientalistyczny 49: 33– 44, also in Dubi´nski (1994). 1994

Caraimica. Prace karaimoznawcze. [Caraimica. Karaim studies.] Warszawa: Dialog.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

751

The viewpoint and tense system in spoken North-Western Karaim

751

Firkoviˇcius, Mykolas 1996 Mie´n karajˇce ürianiam. [I learn Karaim.] Vilnius: Danelius. Forsyth, James 1970 A grammar of aspect. Usage and meaning in the Russian verb. Cambridge: University Press. Johanson, Lars 1971 Aspekt im Türkischen. Vorstudien zu einer Beschreibung des türkeitürkischen Aspektsystems. [Aspect in Turkish. Preliminary studies on the description of the Turkish aspectual system.] (Studia Turcica Upsaliensia, 1), Uppsala: Almqvist & Wiksell. 1991 “On syllabic frontness oppositions in Turkic”, in: Varia Eurasiatica. Festschrift für Professor Róna-Tas. Szeged, 77–94. 1994 “Türkeitürkische Aspektotempora” [Turkish aspect and tense categories], in: R. Thieroff and J. Ballweg (eds.), 247–266. this volume “Viewpoint operators in European languages”. Kowalski, Tadeusz 1929 Karaimische Texte im Dialekt von Troki. [Karaim texts in the dialect of Troki.] (Mémoires de la Commission Orientale de l’Académie Polonaise des Sciences et des Lettres 11.) Kraków: Académie Polonaise des Sciences et des Lettres. Mathiassen, Terje 1996 Tense, mood, and aspect in Lithuanian and Latvian: Tense, aspect, modality. (Meddelelser, 75), Oslo: Universitetet i Oslo, Slavisk-baltisk avdeling. Musaev, Kenesbaj Musaeviˇc 1964 Grammatika karaimskogo jazyka. Fonetika i morfologija. [Karaim grammar. Phonetics and morphology.] Moskva: Nauka. Pritsak, Omeljan 1959 “Das Karaimische” [The Karaim language], in: Jean Deny et al. (eds.), 318–340. Senn, Alfred 1966 Handbuch der litauischen Sprache. [Lithuan manual.] Heidelberg: Carl Winter. Thieroff, Rolf & Joachim Ballweg (eds.) 1994 Tense and aspect in European languages. (Linguistische Arbeiten, 308.) Tübingen: Niemeyer.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

752

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

753

Karen H. Ebert

Aspect in Maltese

1. Purpose of the study1 My venture into Maltese started with the E UROTYP Progressive Questionnaire (PROGQ). When I tried to evaluate the data and looked for information on the Maltese tense-aspect (TA) system I found very little. The sources available to me then (Aquilina 1965, Borg 1981, Comrie 1991) give only a few constructed sentences. Borg (1981) is an investigation in the framework of localist theory; aspect is investigated mainly in examples of the type ‘walk from x to y’. The first aim of this study therefore was to give a more comprehensive overview of the Maltese TA system. This was possible only after the appearance of Vanhove (1993), which offers a rich amount of data both from the spoken and the written language.2 Special attention will be given here to the use of the Active Participle (AP) as finite verb (section 6). Other authors either do not mention the AP at all, or they describe only the progressive, but not the resultative function, which Maltese shares with other varieties of Arabic. Further, I shall discuss some problems of actionality and its relevance for the interpretation of tense-aspect forms (section 7). An incentive to look more closely at Maltese also came from Dahl’s (1985) typological investigation of tense and aspect systems. Somewhat surprisingly, Maltese together with Tunisian Arabic turned out to have the most prototypical perfectives in his sample (hit ratio 0.92, 0.93), followed by Italian and Latin (both 0.91), whereas Standard Arabic figures much lower with 0.79. This raises the interesting question whether Maltese and Tunisian have changed under areal influence, as suggested by Myhill (1992: 270). This possibility will be discussed at the end of the paper.

2. Overview of Maltese tense-aspect forms Maltese has two simple finite verb forms, one characterized by personal suffixes, the other by personal prefixes (cf. Table 1). In Semitic studies the forms are therefore often referred to as ‘suffix-conjugation’ and ‘prefix-conjugation’ (e.g., Vanhove 1993; for Semitic in general: Rössler 1950, Cohen 1984). In descriptions of Maltese most authors use the terms ‘perfect’ and ‘imperfect’ (Aquilina 1965, Borg 1981, Borg and Comrie 1984, Comrie 1985) or ‘perfective’ and ‘imperfective’ (Dahl 1985, Fabri 1993, 1995), which I will adopt here. The forms are combined tense-aspect

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

754

754

Karen H. Ebert

Table 1. Paradigm for kiteb ‘write’ and kien ‘be’ PFV

IPFV

PFV

IPFV

SG 3M 3F 2 1

kiteb kitb-et ktib-t ktib-t

j-ikteb t-ikteb t-ikteb n-ikteb

kien kien-et kon-t kon-t

i-kun t-kun t-kun n-kun

PL 3 2 1

kitb-u ktib-tu ktib-na

j-iktb-u t-iktb-u n-iktb-u

kien-u kon-tu kon-na

i-kun-u t-kun-u n-kun-u

forms. The characterization Comrie (1976: 78–81) gives for Standard Arabic also holds for Maltese: the basic meaning of the Perfective is relative past time reference and perfective aspect, that of the Imperfective is relative nonpast time reference and imperfective aspect. The direct object is cross-referenced as a suffix on transitive verbs. An indirect object or a human definite object is indicated by lil (li, l) ‘to’  personal object suffix (for details see Borg & Comrie 1984, Fabri 1993: Ch. 5). The suffixes indicate possession when added to a noun. The subject markers are agreement markers, while the object markers serve as pronouns; cf. (1) a.

b.

Anna kitb-et l-ittra lil missier-ha. A. write-3F:PFV DEF-letter to father-her ‘Anna wrote the letter to her father.’ Anna kitb-it-hie-l-u. A. write-3F:PFV-her-to-him ‘Anna wrote it (F) to him.’

The Active Participle, which is formed from only a small set of verbs in Maltese, is used as a finite verb. It is characterized by the vocalization -ie-V- ( class. Arabic -aa-i-) in the singular. The Active Participle agrees with the subject in gender (in the singular) and number. (2)

liebes (M) liebs-a (F) lebs-in (PL)

‘he is dressed’ ‘she is dressed’ ‘they are dressed’

Maltese has a progressive formed with qieg ed (qieg da F, qeg din PL), the Active Participle of the verb qag ad 3 ‘stay, sit, be located’, usually contracted to invariable qed, and a future marked by sejjer (sejra F, sejrin PL), the Active Participle of ‘go’, or invariable ser/ se /sa 4. Both combine exclusively with the Imperfective.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

755

Aspect in Maltese

(3)

qed / qieg ed jikteb qed / qieg da tikteb qed / qeg din niktbu

‘he is writing’ ‘she is writing’ ‘we are writing’

sa / se(r) / sejjer nikteb sa / se(r) / sejra tikteb sa / se(r) / sejrin jiktbu

‘I am going to write’ ‘she is going to write’ ‘they are going to write’

755

The simple forms as well as the progressive and future combine with kien (the Perfective of the verb ‘be’), and with ikun (the Imperfective of ‘be’) to form complex TA forms.

3. Imperfective and Perfective With dynamic verbs the Imperfective is restricted to generic or habitual sentences. With lexical imperfectives and stative verbs it has actual present time reference (cf. also t-ista’ ‘you can’ (6a), n-iftakar ‘I remember’ (12a); for actional verb classes see section 7). (4) a.

b.

c.

afna. Pawlu j-orqod P. 3:IPFV-sleep lot ‘Pawlu sleeps a lot.’ (PROGQ:2) T-naddaf id-dar. 3F:IPFV-clean DEF-house [What does she do on Saturdays?] ‘She cleans the house.’ (Borg 1988: 31) g alkemm j-af b-l-Ingli˙z i- oss li G ax because although 3:IPFV-know to-DEF-English 3:IPFV-feel that mag -hom. ma jkun-x j-ista’ j-la aq NEG FUT-NEG 3:IPFV-be_able 3:IPFV-compete with-them ‘Because even though he knows English well, he feels that he will not be able to compete with them.’

The Imperfective of motion verbs can refer to future events. With other verbs an explicit time adverbial or some other disambiguating context is needed. The simple Imperfective is regularly used in temporal and conditional clauses with nonpast time reference (cf. (37a, c), (39b, d)). (5) a.

Immur wa d-i. 1:IPFV:go one-1SG:POSS ‘I shall go on my own.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

756

756

Karen H. Ebert

b.

(Vanhove 200 = Vanhove 1993: 200) il-bieb hemm fuq tkun t-ista’ X’ in n-ift -u when 1:IPFV-open-PL DEF-door there up FUT 2:IPFV-be_able ritratti a jar. t-ie u 2:IPFV-take photo:PL better ‘When we open the door up there, you will be able to take better pictures.’

As Maltese has no infinitive, the Imperfective functions as a default form in dependent position, e.g., with modal, phasal or motion verbs. (6) a.

b.

c.

T-ista’ t-mur. 2:IPFV-can 2:IPFV-go ‘You can go.’ j-id ak. Da al come_in:3:PFV 3:IPFV-laugh ‘He came in laughing.’ (Borg 1981: 165) Marija t-mur t-ixtri l-Belt. M. 3F:IPFV-go 3F:IPFV-buy DEF-city ‘Maria goes shopping in Valletta.’

The simple Perfective is restricted to past time reference. It is used in narrative texts to indicate the succession of events, whereas the Past Imperfective (constructed with the auxiliary kien) serves as a backgrounding form. (7)

(Vanhove 476, text 4) hi din: illi f’dawk j-g id-u Il-le˙gg˙ enda li DEF-legend REL 3:IPFV-tell-PL she this that in-that:PL kien hemm xi familji jew familja illi kienu l-in awi DEF-region was there some family or family:PL REL PT daqsxejn a˙zina, permissiva z˙ -˙zejjed. U j-g ix-u 3:IPFV-live-PL little evil permissive DEF-excess and illi darba minn-hom g amel terremot u j-g id-u 3:IPFV-tell-PL that once of-them make:3:PFV earthquake and sfronda-t l-art bi-hom. give_way-3F:PFV DEF-earth with-them ‘The legend they tell is this: In that area there was some family, or some families who lived (PT:IPFV) in a bad way, too permissively. And they say that once there occurred (PFV) an earthquake and that the earth gave way (PFV) with them.’

The Perfective can also have perfect and resultative meaning.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

757

Aspect in Maltese

(8) a.

b.

757

(Vanhove 49) illi sraq-t-hom. Jekk inti sraq-t-hom g id if you steal-2:PFV-them say:IMP that steal-2:PFV-them ‘If you have stolen them, say that you have stolen them.’ Xbaj-t. become_satiated-1SG:PFV ‘I am satiated.’

With telic predicates the Perfective signals completion of the event, but with atelic verbs like those in (9) there is only an arbitrary temporal limit. A series of verbs in the Perfective denotes a series of events, where one activity is finished before the next begins. (9) a.

b.

(PROGQ: 48) sag t-ejn wa ed-ha. Anna lag b-et play-3F:PFV hour-DU one-3F:POSS A. ‘Anna played for two hours all by herself.’ (PROGQ: 71) Studja, qara l-gazzetta, kiel u mbag ad study:3:PFV read:3:PFV DEF-newspaper eat:3:PFV and after mar fi-s-sodda. go:3:PFV in-DEF-bed ‘[What did Martin do yesterday evening?] He studied, he read the newspaper, he ate and then he went to bed.’

The Perfective thus denotes a specific event which occurred before ‘now’ or before reference time and which is presented “as a single unanalyzable whole” (cf. Comrie 1976: 3). With telic expressions it implies the attainment of the inherent limit; with atelic expressions it simply indicates that the event occurred. The Imperfective is restricted to inactual contexts with dynamic verbs; with stative verbs it also has present meaning.

4. Auxiliaries 4.1. kien The auxiliary kien (PFV of ‘be’, see Table 1) is a past marker. With lexical imperfectives, which have no Perfective form, and with stativizing forms (Active Participle, Progressive) the auxiliary indicates past time reference (10a, b). With the Perfective of a dynamic verb it expresses anterior past (10c,d); the temporal reference is determined by the auxiliary, while the Perfective of the main verb expresses anteriority (cf. also Comrie 1985: 77).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

758

758

Karen H. Ebert

(10) a.

b.

c.

d.

(Vanhove 218) j-ismu t-tifel? X’kien what-PT5 3:IPFV-be_called DEF-child ‘What was the child called?’ (PROGQ: 3) ˙ Meta Ganni g˙ ie Anna kienet g ad-ha qed still-she PROG when G. come:3:PFV A. PT t-a dem. 3F:IPFV-work ‘When Gianni came, Anna was still working.’ (Vanhove 226) g˙ ie s-surmast il-kbir u konna di˙gà come:3:PFV DEF-master the-big and PT already tg allim-na l-alfabet. learn-1PL:PFV DEF-alphabet ‘The headmaster came when we had already learned the alphabet.’ (Vanhove 68) il-wa da tfajla Maltija u U dana kien ataf and that PT catch:3:PFV to-one:F girl Maltese and bie-ha g˙ o dan l-g ar. hide:3:PFV-her inside that DEF-cave ‘And that one had caught a young Maltese girl and hidden her in that cave.’

The last example shows that the tense marker need not be repeated in reporting a series of anterior events; indeed, often a repetition would sound odd (Fabri 1995). The tense marker is also omitted if different TA-forms are combined in a sentence as in (11), where the first verb form is a past of a Progressive (a) or a past of a stative verb (b), the second (unmarked for tense) a past anterior. All of Vanhove’s examples for simple Perfectives allegedly functioning as anterior pasts are of this type. (11) a.

b.

(Vanhove 48) kienu qed i-fittx-u-ha dawn il-pirati, G ax because PT PROG 3:IPFV-look_for-PL-her these DEF-pirates ra-w-ha fi-l-wied. g ax because see:3:PFV-PL-her in-DEF-valley ‘Because they were looking for her, those pirates, because they had seen her in the valley.’ (Vanhove 48) Ma kinu-x j-ifhm-u kif g˙ ie-t. NEG PT-NEG 3:IPFV-understand-PL how come-3F:PFV ‘They did not understand how she had come.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

759

Aspect in Maltese

759

Kien followed by an Imperfective of a dynamic verb refers to non-specific situations in the past. It regularly appears in the apodosis of counterfactual conditionals. (12) a.

b.

(Vanhove 434, text 1) Jien n-iftakar illi f’dak i˙z-˙zmien in-nies kienu I 1:IPFV-remember that in-that DEF-time DEF-people PT il- obz huma. j-morr-u l-forn j-ag g˙ n-u 3:IPFV-go-PL DEF-oven 3:IPFV-knead-PL DEF-bread they ‘I remember that at that time the people used to go to the bakery to knead the bread themselves.’ (Fabri 1995) rba -na l-lotterija konna n-ixtr-u dar Kieku6 if_it_were win-1PL:PFV DEF-lottery PT 1:IPFV-buy-PL house Malta. M. ‘If we had won the lottery, we would have bought a house in Malta.’

4.2. ikun Ikun is a future marker with stative verbs and with stativizing forms, which do not combine with the future marker sejjer/ se/ sa. Cf. also ma jkun-x j-ista’ ‘he will not be able’ in (4c), tkun liebes ‘you will be dressed’ in (28b). (13) a.

b.

(Vanhove 202) rendikont. Imma mbag ad ikunu j-rid-u after FUT 3:IPFV-want-PL report but ‘But afterwards they will want to have a report. (Fabri 1995) Ray ikun qed j-iekol. R. FUT PROG 3:IPFV-eat ‘Ray will be eating.’

With dynamic verbs ikun  IPFV has non-specific readings only, in accordance with Vanhove’s observation that the form does not occur together with an adverb of definite temporal reference.7 (14) a.

(Vanhove 197) li l-iswed ikun j-ixraq-l-i T-a seb 2:IPFV-think that DEF-black IPFV 3:IPFV-suit-to-me ‘Do you think that black would suit me?’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

760

760

Karen H. Ebert

b.

c.

(Vanhove 198) mar c˙ kejken u saqaj-k ikunu i-miss-u Sib and feet-your IPFV 3:IPFV-touch-PL find donkey small ma-l-art. with-DEF-ground ‘Find a small donkey and your feet will touch the ground.’ (Borg 1988: 4, cited from Vanhove) Il-lingwa t-irrifletti t-tqassim spazjali ta-l-kuntest DEF-language 3F:IPFV-reflect DEF-part spatial of-DEF-context li fi-h ikun j-in-sab il-kelliem. that in-her IPFV 3:IPFV-RFL-find DEF-speaker ‘The language reflects whichever part of the spatial context the speaker finds himself in.’

The combination ikun  PFV refers to anterior events (expressed by PFV) in inactual contexts (expressed by the auxiliary).8 (15) a.

b.

c.

(Vanhove 235) fuq il-bejt Dan b al meta wie ed j-itla’ that like when one:M 3:IPFV-climb on DEF-house bi-s-sieq is-sellum li jkun tela’ j-ag ti 3:IPFV-give with-DEF-foot DEF-ladder which IPFV climb:3:PFV minn-u. from-him ‘That is like when someone climbs on the (terrace of the) house and overturns with his foot the ladder which he has climbed up.’ (Vanhove 442) g ax tkun g oli-et u mli-et il-lembi because IPFV put_up-3F:PFV and fill-3F:PFV DEF-basin koll-u. all-it ‘ because she had/would have taken up and filled all the basin.’ (From the text “La fabrication du pain à la maison dans les années vingt”) (Fabri 1995) Pawlu dejjem ikun kiel meta n-asal jien. P. always IPFV eat:3:PFV when 1:IPFV-arrive I ‘Pawlu has always finished eating when I arrive.’

Ikun  PFV can express both an anterior future and a perfective future, although the latter use is rare. In temporal and conditional clauses perfective future meaning is expressed by the simple Imperfective (cf. section 8). (16a) for example can not be understood as ‘if he signs the contract’ (Fabri, p.c.). Vanhove (1993: 233) translates

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

761

Aspect in Maltese

761

all her examples for an ‘accompli futur’ by a French anterior future, and most of her examples could be interpreted either way (e.g., (16b)). (16) a.

b.

c.

(Vanhove 234) Ma j-ista-x mil-l-kelma j-er˙ga’ NEG 3:IPFV-can-NEG of-DEF-word 3:IPFV-come_back lura hu la jkun iffirma fuq il-kuntratt. afterwards he if IPFV sign:3:PFV on the contract ‘He won’t be able to go back on his word later, if he has signed the contract.’ (Vanhove 233) jumejn o ra, u tkun re˙gg -et N-ag ti-ha 1:IPFV-give-her day:DU other and IPFV return-3F:PFV id-dar. DEF-house ‘I give her two more days, and she will return / will have returned home.’ (“elle sera revenue ”) (Vanhove 233) Dak koll-u li tkun sib-t g˙ ib-u-l-i. that all-him REL IPFV find-2:PFV bring-him-to-me ‘Bring me everything you find.’ (“tout ce que tu auras trouvé ”)

4.3. kien ikun In some idiolects the past marker kien can precede all ikun-forms. Fabri (1995) gives examples for all combinations, but he states that these constructions are rarely used. His examples, reproduced here as (17a–d), are ordered according to acceptability. (17) a.

b.

c.

d.

il-futbol. Pawlu kien dejjem ikun j-ilg ab P. PT always IPFV 3:IPFV-play DEF-football ‘Pawlu always used to play football.’ Pawlu kien dejjem ikun qed j-ilg ab il-futbol. IPFV PROG ‘Pawlu always used to be playing football.’ Pawlu kien dejjem ikun sa j-ilg ab il-futbol. IPFV FUT ‘Pawlu always used to be about to play football.’ il-futbol. Pawlu kien dejjem ikun lag ab IPFV play:3:PFV ‘Pawlu always used to have played football.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

762

762

Karen H. Ebert

According to Vanhove the combination kien ikun  IPFV is obligatory to express a habitual past of lexical imperfectives (18a), for which kien  IPFV is a past. With other verbs, including other statives (18c), the form is exchangeable with kien  IPFV. In Vanhove’s data it is attested only in a few utterances from older people in Gozo. She adds that the construction is more widespread, but that it is felt to be archaic. There is one example for a past habitual progressive (18d – the past tense marker has scope over the relative clause). (18) a.

b.

c.

d.

(Vanhove 218) In-nies kif kienet tkun t-af li hemm fuq DEF-people how PT IPFV 3F:IPFV-know that there up a ji-t an? ta-l-mit na of-DEF-mill FUT 3:IPFV-grind ‘How would people know that the one up there at the mill was going to grind?’ (cf.: in-nies kienet taf ‘the people knew (then)’) (Vanhove 219) kienu jkunu j-i˙g-u jew F-l-img oddi dawk li in-DEF-past those that PT IPFV 3:IPFV-come-PL or dawn kienu il-pirati ta-l-ba ar jew il- allelin DEF-pirates of-DEF-sea or DEF-thieves these PT dejjem it-torok. g˙ eneralment j-g id-u 3:IPFV-say-PL always DEF-Turk:PL generally ‘In the past, whoever came, either the pirates from the sea or the robbers, they generally called them “Turks”.’ Gozo dialect (Vanhove 219) kellu x’j-aqsam ma’ kien hemm wie ed li PT there one:M REL have:PT:3 what-3:IPFV-share with kienu jkunu j-rid-u l-istess xi ie or forsi IPFV 3:IPFV-want-PL DEF-same what other maybe PT xebba. girl. ‘There was someone who had a quarrel with someone else, maybe they wanted the same girl.’ (Vanhove 119) U kienet t-kun reazzjoni afna drabi g al dak li and PT 3F:IPFV-be reaction many times for that REL jkun qed j-i˙gri fi-l-mument. IPFV PROG 3:IPFV-happen at-DEF-moment ‘And that often used to be the reaction to what was going on at the moment.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

763

Aspect in Maltese

763

4.4. se with auxiliaries The future marked by se (ser, sa, sejjer ‘going’) is sometimes characterized as an immediate or definite future (Aquilina 1965: 222, Vanhove 1993: 175). It translates most often literally into English ‘is going to’: se jiktib ‘he is going to write’. Its original intentional meaning probably explains why it does not combine with stative verbs, which form the future with ikun (13). According to Vanhove (1993: 205) se jkun expresses a definite future with lexical imperfectives and a few stative verbs (19a). In the speech of two intellectuals from Valletta she found se jkun also with non-stative verbs; there is no difference between se and se jkun in examples like (19b). Fabri (p.c.) finds se jkun unacceptable except with nonverbal predicates (e.g., se jkun id-dar ‘he will be at home’).

(19) a.

b.

(Vanhove 205) Hemm c˙ erti poe˙ziji f’din il-kollezzjoni l-˙gdida li there_is certain poetry in-this DEF-collection DEF-new that se jkun j-isim-ha Xibkatulis. FUT IPFV 3:IPFV-be_called-she X. ‘There are certain poems in this collection which will be called Xibkatulis.’ (Borg 1988: xv, cited from Vanhove 206) nkunu biss B’dan il-pro˙cediment f’ afna ka˙zijiet se with-this DEF-procedure in-many case:PL FUT IPFV only n-oborx-u l-wi˙cc˙ minflok ma n- affr-u 1:IPFV-scratch-PL DEF-face instead NEG 1:IPFV-dig-PL fi-l-fond. in-DEF-depth ‘With this procedure we will scratch only the surface instead of digging into depth.’

According to Comrie (1991: 8) future perfect should be expressed by se jkun  PFV. Such a form does not occur in Vanhove’s data, and I have not come across it elsewhere. The combination ikun se yields an imminential meaning. There is a regular past formation kien ikun se (see also (17c)), but sometimes the posterior past form kien se is used instead.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

764

764

Karen H. Ebert

(20) a.

b.

c.

d.

(Vanhove 206) izda t-er˙ga’ lura. Tkun se t-o ro˙g, IPFV FUT 3:F:IPFV-go_out but 3:F:IPFV-return back ‘She is about to go out, but she comes back.’ (From a stage description, which seems to be in the Imperfective throughout; cf. other examples in Vanhove 206f) (Fabri 1995) Pawlu dejjem ikun sa j-iekol meta n-asl-u. P. always IPFV FUT 3:IPFV-eat when 1:IPFV-arrive-PL ‘Pawlu is always about to eat when we arrive.’ (Vanhove 207) Kont inkun ser in-we˙gg˙ a’. PT IPFV FUT 1:IPFV:RFL-hurt ‘I was about to hurt myself.’ (Vanhove 194) u qajjim-t-ni Ja asra kont ser n-orqod INTJ PT FUT 1:IPFV-fall_asleep and wake-2:PFV-me b’dak il-vers. with-that DEF-verse ‘I was about to fall asleep, and you woke me up with this verse.’

The posterior past kien se  IPFV (Comrie’s ‘conditional’, 1985: 77) is sometimes found in the apodosis of counterfactual conditionals (instead of kien  IPFV, cf. (12b), (38b)). No conditional meaning is involved in (21a,b). (21) a.

b.

c.

(Vanhove 193) Kont se n-saqsi-k. PT FUT 1:IPFV-ask-you ‘I was going to ask you.’ (Vanhove 193) Insomma ma kinu-x se i-werwr-u-h. after_all NEG PT-NEG FUT 3-terrorize-PL-him ‘After all, they were not going to terrorize him.’ (Vanhove 193) Kieku n-af fejn kell-ha xi flus kont se if_it_were 1:IPFV-know where have-her what money PT FUT n-ie u dawk l-erba’ c˙ raret n-ikkuntenta ru -i 1:IPFV-content soul-my 1:IPFV-take those DEF-four scarf:PL ma mu˙g-in? dirty-PL ‘If I had known where she had the money, would I have contented myself with taking the four dirty scarfs?’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

765

Aspect in Maltese

765

5. The qed-form The progressive markers qieg ed (AP of qag ad ‘stay, sit’) and the short form qed are interchangeable (Vanhove 1993: 112, Borg 1981: 147). In the Progressive Questionnaire the form qed was used in the overwhelming majority of cases. I shall therefore refer simply to the qed-form. The past progressive is formed with kien, the future with ikun (cf. (10b), (13b)). The qed-form was used in all our test cases. (22) a.

b.

c.

(PROGQ: 1) Qed t-a dem. PROG 3F:IPFV-work ‘[What is Anne doing right now?] – She is working.’ (PROGQ: 83) Jekk t-i˙gi fi-t-tmienja nkun g ad-ni qed if 2:IPFV-come at-DEF-eight FUT still-my PROG in-sajjar. 1SG:IPFV-cook ‘If you come at 8 o’clock, I will still be cooking.’ (PROGQ: 76) ˙ Il-boss kien i-rrabjat g ax Ganni ma kien-x DEF-boss PT 3:IPFV-be_angry because G. not PT-NEG meta da al. qed j-a dem PROG 3:IPFV-work when enter:3:PFV ‘The boss was angry, because Gianni was not working when he came in.’

Stative verbs do not normally occur in the Progressive (*qed jaf *he is knowing, *qed irid *he is wanting), but as in English it is often possible to use a Progressive for temporally limited states. (23) a.

b.

c.

(Vanhove 116) Kemm qed in-kun g˙ eneruz mag -kom. how_much PROG 1:IPFV-be generous with-you:PL ‘Look how generous I am (being) with you!’ il-bard? Qed t- oss PROG 2:IPFV-feel DEF-cold ‘Are you feeling cold?’ (Borg 1981: 149) ˙ il-Belt. Ganni qed j-oqg od G. PROG 3:IPFV-stay DEF-B. ‘Gianni lives / is living in Valletta.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

766

766

Karen H. Ebert

d.

(Aquilina 1965: 148) Qed n-ifhem il-gazzetta; ma jiniex PROG 1:IPFV-understand DEF-newspaper NEG I:be:NEG tqil. n-ifhem il-ktieb g ax 1:IPFV-understand DEF-book because difficult ‘I understand the newspaper, but I don’t understand the book, because it is difficult.’

The qed-form can also express a temporally limited habit (cf. also (31), (35b)).

(24)

(Vanhove 139) ˙ Guzè qed j-orqod afna da-l-a ar. G. PROG 3:IPFV-sleep lot this-DEF-late ‘Giuseppe sleeps a lot lately.’

It does not combine with adverbs that set a temporal limit (25a), but adverbials like ‘during x’ or ‘while’-clauses are possible.

(25) a.

b.

c.

(PROGQ: 48) Anna lag b-et sag tejn wa id-ha. A. play-3F:PFV hours:DU one-3F:POSS ‘Anna played two hours by herself.’ (PROGQ: 49/50) Anna kienet qed t-itkellem ma’ ta’ dej-ha. A. PT PROG 3F:IPFV-talk with of near-her ‘Anna was talking to her neighbor [during the whole class].’ (PROGQ: 70) waqt li Anna kienet qed t-aqra Il-biera DEF-day_before while_ A. PT PROG 3F:IPFV-read f’kamart-ha Martin kien qed j-ilg ab fi-l-bit a. in-room-her M. PT PROG 3:IPFV-play in-DEF-court ‘Yesterday, while Anna was reading in her room, Martin was playing in the courtyard.’

The Maltese qed-form is a fully grammaticized broad progressive, which is not restricted to a focalizing viewpoint as defined in Bertinetto (this volume).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

767

Aspect in Maltese

767

6. The Active Participle 6.1. Inventory The Active Participle has a marginal status in Maltese compared to other varieties of Arabic. It is formed from only a few verbs. Vanhove (p. 134) gives a supposedly exhaustive list of “motion verbs” (in which she includes ‘sleep’): (26)

sejjer nie˙zel die el miexi tielaq tiela’

‘going’ ‘descending’ ‘entering’ ‘walking’ ‘leaving’ ‘mounting’

g˙ ej iere˙g riesaq rieqed g addej

‘coming’ ‘going out’ ‘approaching’ ‘sleeping, asleep’ ‘passing’ (form of agentive noun)

to which we can add: wieqaf qieg ed liebes riekeb nieqes

‘standing’ ‘sitting, staying’ ‘dressed’ ‘riding’ ‘lacking’

eles ‘set free’), Many APs have become frozen as adjectives, e.g., ieles ‘free’ ( fierag ‘empty’ ( farag ‘ebb’), biered ‘cold’ ( bired ‘become cold’). The AP is a stative form; the auxiliaries are used as with stative verbs, kien indicating past time reference, ikun future time reference.

6.2. AP as resultative All authors agree on the point that the AP expresses progressive meaning (Borg 1981, Vanhove 1993, Fabri 1995). This seems to be due to the fact that they have the motion verbs in mind, and that the AP of ‘sleep’ is best translated into English by a progressive. (27) a.

(Aquilina 1965: 148) fi-d-dar il-qadima. Karlu rieqed wa d-u K. fall_asleep:AP:M one-3M:POSS in-DEF-house DEF-old ‘Karlu is sleeping alone in the old house.’ (Aquilina’s translation) = ‘Karlu is (fallen) asleep, and therefore sleeping.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

768

768

Karen H. Ebert

b.

(Aquilina 1965: 148) Il-mara liebs-a libsa g˙ did-a u t-tfal DEF-woman dress:AP-F dress new-F and DEF-child:PL azin. lebs-in dress:AP-PL bad ‘The woman is wearing a new dress and the children are dressed shabbily.’

In (27b) the first AP is rendered as a progressive by Aquilina, the second as a resultative. However, both liebsa and lebsin refer to the posttransformative phase of the initio-transformative9 verb libes ‘put on, get dressed; wear’. In the first clause we are dealing with a possessive resultative ‘the woman has a new dress put on’, in the second with a subjective resultative ‘the children are dressed’ (cf. Nedjalkov & Jaxontov (1988) for a typology of resultatives). The resultative meaning of the AP of libes is also attested in Vanhove’s texts:

(28) a.

b.

(Vanhove 463, text 2) ˙ an li lil Ga suppost kien liebes Sej -u-l-u call:3:PFV-PL-to-him to G. that supposed PT dress:AP:M ta’ mara. Huma ra-w xkupa liebs-a ta’ mara, as woman. they see:3:PFV-PL broom dress:AP-F as woman kien-et mara tassew. imma min-g ali-hom li who-for-them that be-3F:PFV woman real but ˙ an, who supposedly was dressed as a woman. They saw ‘They called Ga the broom dressed as a woman, but for them it was a real woman.’ (Vanhove 470, text 3) imma ma t-rid-x tkun “G ada, ejja tomorrow IMP:come but NEG 2:IPFV-want-NEG IPFV missier-u ta-[h] parir u liebes” father-his give:3:PFV[-him] advice and dress:AP:M qal-l-u: “ilbes xibka ta-s-sajjieda u tell:3:PFV-to-him IMP:dress net of-DEF-fisherman:PL and in b’hekk tkun liebes u f’l-istess with-such IPFV dress:AP:M and in-DEF-same time m’inti-x liebes.” NEG-you-NEG dress:AP:M ‘ “Come tomorrow, but you must not be dressed.” [the sultan said to the His father gave him the advice, he said: “Put on a young farmer]. fishing net, and so you will be dressed and at the same time you will not be dressed.” ’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

769

Aspect in Maltese

769

The APs of rikeb ‘mount; ride’, waqaf ‘stand up; stand’10 also apply to the first lexical meaning component and refer to the posttransformative phase of the event (cf. section 7). (29) a.

b.

(Vanhove 472, text 3) ejja riekeb imma ma Mela g ada NEG now tomorrow IMP:come mount:AP:M but t-rid-x tkun riekeb. 2:IPFV-want-NEG IPFV mount:AP:M ‘Come tomorrow on horseback, but you must not be on horseback.’ [sic!, the order is intentionally paradoxical in the story, cf. (28b)] (PROGQ: 58) Anna wieqf-a fi-l-bieb. A. stand:AP-F in-DEF-door. ‘Anne is standing at the door.’ (= is stood)

6.3. AP as progressive? The AP of motion verbs apparently corresponds to the qed-form of other verbs. The qed-form can only have a restricted habitual interpretation with motion verbs. (30) a.

b.

c.

(PROGQ: 77) Mhux nie˙zl-a x-xita issa. NEG-it-NEG descend:AP-F DEF-rain now ‘It is not raining now.’ (PROGQ: 66) Anna sejr-a g ada. A. go:AP-F tomorrow ‘Anna is leaving tomorrow.’ L-ajruplan tielaq issa. DEF-airplane leave:AP:M now. ‘The airplane is leaving now.’

but: (31) a.

L-ajruplan qed j-itlaq kmieni da-˙z-˙zmien. DEF-airplane PROG 3:IPFV-leave early this-DEF-time ‘The airplane is leaving early these days’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

770

770

Karen H. Ebert

b.

(PROGQ: 63) Dak i˙z-˙zmien kien qed i-mur j-i˙zfen kull nhar that DEF-time PT PROG 3:IPFV-go 3:IPFV-dance all day ta’ Sibt. of Saturday ‘At that time he used to go dancing every Saturday.’

This is in accordance with Borg’s (1981: 144f) findings that the Imperfective typically encodes unrestricted habituality, the qed-form restricted habituality, and the AP progressivity (this holds of course only for motion verbs); cf. his examples:11 (32) a.

b.

c.

˙ Ganni miexi mi-d-dar sa l-iskola b alissa. G. walk:AP:M from-DEF-house to DEF-school just_now ‘Gianni is walking from home to school just now.’ ˙ Ganni qed j-imxi mi-d-dar sa l-iskola G. PROG 3:IPFV-walk from-DEF-house to DEF-school da-˙z-˙zmien. this-DEF-time ‘Gianni walks from home to school these days.’ ˙ Ganni j-imxi sa l-iskola. G. 3:IPFV-walk to DEF-school ‘Gianni walks to school.’

Vanhove reports the same distinction from her young informant from Valletta: (33) a.

b.

(Vanhove 140) b alissa. G addejj-in afna karrozz-i minn di-t-triq pass:AP-PL lot car-PL from this-DEF-street just_now ‘A lot of cars are passing through this street right now.’ afna karrozz-i minn di-t-triq Qed j-g add-u car-PL from this-DEF-street PROG 3:IPFV-pass-PL lot da-l-a ar. this-DEF-late ‘A lot of cars pass through this street these days.’

However, a translation equivalent does not necessarily prove that we are dealing with a progressive. There are other contexts where the AP of motion verbs corresponds to an Imperfective of other verbs.12 (34) a.

(Vanhove 147) l-mewt riesqa dejjem. Iz-zmien g addej u DEF-time pass:AP and DEF-death approach:AP:F always ‘Time goes by and death comes always closer.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

771

Aspect in Maltese

b.

771

(Vanhove 149) . . . beda miexi c˙ eklem c˙ eklem lejn id-dar. start:PFV walk:AP:M slowly slowly towards DEF-house ‘He started to walk slowly slowly towards the house.’

The AP of motion verbs is therefore an imperfective rather than a progressive. There are two independent criteria to distinguish the two types of APs in Maltese: a) only the resultative AP can stand as an adjective (Fabri, p.c.): ir-ra˙gel il-wieqaf ‘the standing man”, but: *ir-ra˙gel il-miexi ‘the walking man’; b) only the motion verb AP can have future time reference (30b).

7. Aspectual verb classes and grammatical aspect Vanhove (1993: 239), in spite of the wealth of her data, arrives at the following simple schema of the Maltese verbal system, optional markers not taken into account.13 Table 2. The Maltese verbal system according to Vanhove (1993: 239)

past present future

‘inaccompli’

‘accompli’

IPFV IPFV IPFV

PFV PFV ikun  PFV

This presentation is meant to prove that the two basic forms are not tenses. I agree on this point, but I find the presentation inadequate nevertheless: – The PFV is not adequately described by the term ‘accompli’. – The simple PFV does not have present time reference. – The IPFV has present reference only with stative verbs. – A past ‘inaccompli’ is never referred to by an IPFV form alone; with stative verbs a simple PFV is used. It is not possible to describe the Maltese tense-aspect system without taking into account the auxiliaries and preverbal particles and without distinguishing aspectual verb classes. There is very little information available on the actional character of Maltese verbs.14 Authors usually mention three stative verbs which do not form a Perfective: jaf ‘knows’, jismu ‘is named’, jixba ‘resembles’. Borg (1981: 4) adds (for his idiolect) ifu ‘smells good’, itul ‘grows tall’, i uf ‘wanders around’, isus ‘follows persistently’, which suggest that stativity is not a common characteristic of this group of lexical imperfectives. On the other hand, most stative verbs do form a Perfective. Their Imperfective refers to a present situation, whereas the Perfective refers to a

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

772

772

Karen H. Ebert

Table 3. TA-forms of different verb classes

 kien a) STATIVE VERBS (3 RD SG . PFV rid-t she wanted IPFV

 ikun

 kien ikun

tkun rid-t she will have wanted tkun t-rid she will want

kienet tkun t-rid she used to want

tkun t-af she will know

kienet tkun t-af she used to know

FEM .)

t-rid she wants

PROG DEF.FUT

kienet rid-t she had wanted kienet t-rid she used to want [se tkun t-rid] she will certainly want

b) LEXICAL IMPERFECTIVES (3 RD SG . FEM ) PFV IPFV t-af kienet t-af she knows she knew PROG DEF.FUT

[se tkun t-af ] she will certainly know

c) DYNAMIC NON - MOTION VERBS (3 RD SG . MASC .) PFV libes kien libes ikun libes he put on he had put on he will have put on, /usually has put on IPFV jilbes kien jilbes ikun jilbes he puts on he used to put on he would put on PROG qed jilbes kien qed jilbes ikun qed jilbes he is putting on FUT

AP

se jilbes

he was putting on kien se jilbes

he will be putting on /is usually . . . ikun se jilbes

he is going to put on liebes he is dressed

he was going to put on kien liebes he was dressed

he is about to put on ikun liebes he will be dressed

[ ] marginal forms

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

[kien ikun libes] he used to have put on kien ikun jilbes he used to put on [kien ikun qed jilbes] he used to be putting on [kien ikun se jilbes] he was about to put on

773

Aspect in Maltese

773

past situation. Like the lexical imperfectives they do not combine with se15 and qed, but like dynamic verbs they form an inactual past with kien  IPFV. Some verbs of this group are: tama / jittama ‘hope’; tie˙g / je tie˙g ‘need’; aseb / ja seb ‘think’, stenna / jistenna ‘wait’, qag ad / joqg od ‘stay, sit’. Table 3a represents the system of stative verbs, 3b that of lexical imperfectives. The full system for dynamic nonmotion verbs is given in 3c. (The English translations of the isolated forms are of course only tentative.) I have chosen the verb libes / jilbes ‘get dressed, put on; be dressed, wear’ in its first meaning component (cf. discussion of initio-transformative verbs below) in order to get a paradigm including the Active Participle. The system under 3c is derived from the data in Fabri (1995) and Vanhove (1993). Forms in square brackets are marginal according to Fabri and do not occur in Vanhove’s material. We have to include the se-forms with the basic TA forms if we want to get a symmetric arrangement. Interestingly, Fabri reaches the same conclusion; he regards se as a marker of prospective aspect. Maltese has some verbs of a type that notoriously lead to confusion in the literature.16 Let us take raqad as an example: (35) a.

b.

c.

d.

e.

afna. Pawlu j-orqod P. 3:IPFV-sleep lot ‘Pawlu sleeps a lot.’ (habitual) afna da-˙z-˙zmien. Pawlu qed j-orqod this-DEF-time P. PROG 3:IPFV-sleep lot ‘Pawlu sleeps a lot these days.’ (restricted habitual) Pawlu kien qed j-orqod quddiem it-television meta P. PT PROG 3:IPFV-fall_asleep in_front DEF-TV when c˙ empl-et il-qanpiena. ring-3F:PFV DEF-bell ‘Pawlu was falling asleep in front of the TV, when the doorbell rang.’ (progressive) Pawlu raqad il-˙gurnata koll-ha. Il-biera DEF-yesterday P. sleep:3:PFV DEF-day all-her ‘Yesterday Pawlu slept the whole day.’ Pawlu kien rieqed quddiem it-television meta P. PT fall_asleep:AP:M in_front DEF-TV when wasal-t jien. arrive-1:PFV I ‘Pawlu was sleeping in front of the TV when I arrived.’

From these examples one can infer that raqad is an initio-transformative verb, i.e., that its lexical meaning comprises both an ingressive phase ‘fall asleep’ and the resulting situation ‘sleep’. Both meaning components combine with all tense-aspect forms except that the AP is restricted to the verb in its transformative meaning. A

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

774

774

Karen H. Ebert

rough parallel can be found in the English verb hide; he is hiding can refer both to the process of bringing oneself into a position and keeping oneself in this position (cf. Ebert (1995) for more examples from other languages). lexical meaning: 1. ‘fall asleep’ raqad1 ‘fell asleep’ ‘falls asleep’ (HABIT) jorqod1 qed jorqod1 ‘is falling asleep’ rieqed ‘is asleep, is sleeping’

2. ‘sleep’ raqad2 jorqod2 qed jorqod2

‘slept’ ‘sleeps’ (HABIT) ‘sleeps’ (restr. HABIT)

However, according to the intuition of native speakers raqad basically means ‘fall asleep’. This is the default interpretation for most forms which should be ambiguous if the verb had two equivalent meaning components. A sentence like (36) has only the ingressive interpretation for raqad. (36)

Pawlu raqad quddiem it-television meta wasal-t P. sleep:3:PFV in_front DEF-TV when arrive-1:PFV jien. I ‘Pawlu fell asleep in front of the TV when I came.’

The PFV form raqad means ‘slept’ only in special contexts like (35d), which exclude the ingressive reading. Also, a regular Progressive and the AP can only be formed from raqad 1, whereas the qed-form of raqad 2 has the reading of a restricted habitual (35b). The lexical meaning of raqad should therefore rather be given as ‘1. fall asleep; (2. sleep)’. Let us now return to motion verbs. For some of them the dictionary available to me (Bugeja 1982) gives two meanings, e.g., mexa ‘set out; walk’. If we assume that ‘set out’ once was the basic meaning, the AP miexi referred to the situation of having set out, which by inference means ‘is on his way, is walking’. That the form is used in situations where English and other languages use a progressive does not necessarily mean that it is a progressive. (Here a serious disadvantage of our questionnaire method, which was based on meaning equivalents, shows up.) Indeed, the AP occurs in typical imperfective contexts, as shown in (34a, b). The Progressive form qed jimxi has a restricted habitual interpretation only (32b) and cannot mean ‘is setting out’. This suggests that the transformative meaning component mexa1 ‘set out’ is conceived as momentaneous – in contrast to transformative raqad 1, which is an accomplishment. Probably all verbs with an AP (listed in (26)) belong to the group of initio-transformatives; i.e., Maltese has preserved APs only with this type of verb, though not with all of them. From dictionary entries and examples I conclude that there are other initio-transformative verbs, like qabad ‘catch; hold’; beza’ ‘become frightened; be

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

775

Aspect in Maltese

775

Table 4. Interaction of TA-forms and actionality lex. ipfv. PFV IPFV kien  PFV ikun  PFV kien  IPFV ikun  IPFV sa  IPFV qed  IPFV AP

stative

dynamic durative

past

past (anter,pfv) habit., generic

present

past future

initio-transf. with AP transf. posttr.

past anterior fut. anterior past inactual

(temporary)

inactual future progr.,restr.habit.

progr. result.; ipfv

restr.habit.

inactual: habitual, hypothetical

afraid’; fehem ‘come to understand; understand’, but for each of them a detailed investigation is necessary to determine their exact actionality. They do not form APs; a present situation is referred to by the Imperfective or the Progressive, e.g., nifhem ‘I understand’, qed nifhem ‘I am understanding’ (cf. (23d)). Table 4 is a very tentative account of the interaction of tense-aspect forms with different verb classes. The list is not exhaustive. Momentaneous, ingressive and inchoative verbs are not represented due to the lack of adequate data. Note that the distinction between telic or rather transformative and atelic is relevant only in the group with APs.

8. The question of areal influence We can now come back to the question whether the Maltese system has undergone areal influence, as the results of Dahl’s investigation suggest. First we can state that the Maltese Perfective resembles the French passé simple and the Italian passato remoto in that (different from the Russian Perfective) it combines also with atelic verbs. But the Perfective of other varieties of Arabic does, too.17 The overall organization of Maltese TA forms differs considerably from that of Romance languages, which have an aspectual opposition only in the past. Further, Italian and French have no grammaticized progressives. The basic system of Romance can therefore be laid out as in Table 5.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

776

776

Karen H. Ebert

Table 5. Italian basic TA forms PFV PRS PT FUT

IPFV scrive

scrisse

scriveva scriverà

If we add the composite tenses of Italian, the picture does not change. On the other hand the parallels between Maltese and other varieties of Arabic (Table 6) are obvious; cf., e.g., Maltese kiteb and jikteb with Standard Arabic kataba and yaktubu, Maltese kien kiteb and kien jikteb with Standard Arabic kaana kataba and kaana yaktubu, Maltese se/sa jikteb with Standard sa yaktubu. The varieties of Arabic differ in the development of new imperfectives and progressives and in the combinability of TA markers. Egyptian and Syrian Arabic have developed a new Imperfective marked by bi-, which expresses habitual and generic, and in Egyptian also progressive meaning. Syrian has a fully grammaticized progressive marked by Èam ( ‘do’), and a prospective future derived from the verb rah ‘go’, whereas Egyptian marks future ª (without bi-) is relegated to with the prefix ha- (cf. fn. 4). The simple Imperfective ª subordinate and imperative functions in both languages and does not figure in Table 6. In both languages the future is best grouped with the basic TA forms. Cairene has a fourth column with ha-ykuun, but most of these forms are doubtful (data from Eisele 1990a; the questionª marks are his). No such forms could be elicited for Syrian Arabic, but the informant produced some forms with biykuun kaan, corresponding to Maltese kien ikun (with different order of the auxiliaries). Comparison with the inventory of TA forms in other varieties of Arabic shows that the Maltese TA system corresponds closely to these varieties and differs profoundly from Romance. As noted earlier, Comrie (1985, 1991) gives Maltese forms that do not occur in the other sources: se jkun kiteb ‘he will have written’, se jkun qed jikteb ‘he will be writing’, kien se jkun qed jikteb ‘he was going to be writing’.18 The combined future-imperfective marker se jkun corresponds to Cairene ha-ykuun and MSA sa ª yakuunu. Se jkun was found mainly in the speech of a few intellectuals. One of the features that made Maltese appear different from Standard Arabic in Dahl’s evaluation is the impossibility of having Perfective in subordinate clauses with future time reference (this is indeed the only feature in which Maltese differs from Modern Standard Arabic in Table 3.2 of Dahl (1985: 71)). The crucial sentence in Dahl’s questionnaire was (37a), where the Standard Arabic informant uses a Perfective in the subordinate clause. In Maltese only the Imperfective is possible. Here again Maltese conforms with the eastern dialects, as Syrian (37b) shows. Different from Standard Arabic, both Syrian and Maltese have an Imperfective also in the pro-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

777

Aspect in Maltese

777

Table 6. TA forms in other varieties of Arabic MODERN STANDARD ARABIC

 kaana

 (sa) yakuunu

PFV

kaana kataba he had written kaana yaktubu he used to write /was writing kaana sa yaktubu he was going to write

(sa) yakuunu kataba he will have written

 kaan

 biykuun

 ha-ykuun ha-ykuun katab ª a-ykuun ??h ª bi-yiktib ?ha-ykuun ª ha-yiktib ª ?ha-ykuun kaatib

IPFV

FUT

kataba he wrote yaktubu he writes /is writing sa yaktubu he will write

CAIRENE

PFV bi-IPFV

katab bi-yiktib

kaan katab kaan bi-yiktib

biykuun katab biykuun bi-yiktib

FUT

ha-yiktib

kaan ha-yiktib

?biykuun ha-yiktib

ª

AP

kaatib

ª

kaan kaatib

ª

biykuun kaatib

ª

ª

SYRIAN

 kaan

 biykuun

 biykuun kaan

PFV

katab

kaan katab

biykuun katab

??biykuun kaan katab

bi-IPFV PROG

byiktib Éam yiktib

kaan byiktib kaan Éam yiktib

biykuun Éam yiktib

biykuun kaan Éam yiktib

FUT AP

rah yiktib ª kaatib

kaan rah yiktib ª kaan kaatib

biykuun rah yiktib ª biykuun kaatib

tasis of conditional clauses with present or future time reference, though Syrian can also have a Perfective here (37d). (37) a.

Maltese Meta it-tifel j-ir˙cievi l-flus, ser j-ixtri when DEF-boy 3:IPFV-receive DEF-money FUT 3:IPFV-buy rigal li-t-tifla. present for-DEF-girl ‘When the boy gets the money, he’ll buy a present for the girl.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

778

778

Karen H. Ebert

b.

c.

d.

Syrian l-walad l-ma aaš rah yi-štara Wa t b-yi-Èbad  ª DEF-boy DEF-salary FUT 3:IPFV-buy when b-3:IPFV-receive hadiiye li-sadii t-u.  present for-girlfriend-his ‘When the boy gets his salary, he’ll buy a present for the girl’. Maltese Jekk it-tifel j-ir˙cievi il-flus, ... if DEF-boy IPFV-3-get DEF-money ‘If the boy gets the money, ’ Syrian / b-yi-Èbad l-walad l-ma aaš ... Iza Èabad ª ª b-3:IPFV-receive DEF-boy DEF-salary if get-3:PFV ‘If the boy gets his salary, ’

The Perfective is obligatory only in the protasis of conditionals with past time reference.19 (38) a.

b.

Syrian l-walad l-ma aaš Èabad kaan yi-štara Law ª if:IRLS get:3:PFV DEF-boy DEF-salary PT 3:IPFV-buy hadiiye li-sadii t-u.  present for-girlfriend-his ‘If the boy had gotten his salary, he would have bought a present for his girlfriend.’ Maltese (Vanhove 222) Kieku bqaj-t hemm kont immut bi-d-daq. if_it_were stay-1SG:PFV there PT 1:IPFV:die in-DEF-laughter ‘If I had been there I would have died of laughter.’

Here Maltese (together with Syrian) differs from the North African dialects, which have Perfectives in all types of conditional clauses (Vanhove 1993: 66). (39) a.

b.

c.

Moroccan n-mši-u l-l-bhar. iila žaa-t g˙ dda  if come-3F:PFV tomorrow 1:IPFV-go-PL to-DEF-sea Maltese im-morr-u l-baar. jekk t-i˙gi gada if 3F:IPFV-come- tomorrow 1:IPFV-go-PL to-DEF-sea ‘If she comes tomorrow, we shall go to the sea.’ Moroccan iila žaa-t amiina daaba n-mši-u l-l-bhar.   if come-3F:PFV right_now 1:IPFV-go-PL to-DEF-sea

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

779

Aspect in Maltese

d.

779

Maltese kieku t-i˙gi issa im-morr-u l-bahar. if_it_were 3F:IPFV-come right_now 1:IPFV-go-PL to-DEF-sea ‘If she came right now, we would go to the sea.’

The only difference between Maltese and other varieties of Arabic with respect to the verb form in subordinate clauses is that in conditional clauses with nonpast time reference the Perfective is obligatory in the Standard and some other dialects, possible in Syrian, and not possible in Maltese. The most striking difference in the TA system of Maltese and other Arabic dialects is that Maltese makes little use of the Active Participle. In other varieties of Arabic the Active Participle can be formed from practically every verb (for a few exceptions in Cairene see Eisele (1990b: Appendix A)). It has progressive or imperfective meaning with motion verbs, resultative or perfect meaning with other verbs. Concerning the AP of motion verbs Maltese corresponds to Arabic in general, but the resultative function has been reduced to a few verbs and the perfect meaning is totally absent. Another point where Maltese sometimes differs from other varieties is the actionality of the verb. As information on aspectual character is scarce for Maltese, not very reliable and often contradictory for the other dialects of Arabic, I can only point out some obvious parallels and differences here. The Maltese initio-transformative verbs which form APs have parallels in Standard, Syrian and Cairene Arabic. The defective paradigm of the posttransformative meaning component has led linguists to different classifications. Cowell (1964) and McCarus (1976) analyse those verbs as two-phasal (their ‘inceptive’) in Syrian and Modern Standard Arabic respectively; Eisele (1990b) classifies the corresponding Cairene verbs as ‘interval inchoative’, and Woidich (1975) as punctual (which can hardly be maintained in the light of (41a)). (40)

Syrian naam (PFV) waam naam (PFV) seeteen biy-naam (IPFV) naayim (AP)

(41) a.

b.

‘he fell asleep right away’ ‘he slept for two hours’ ‘he always falls asleep’ ‘he is asleep’

Cairene (Mitchell 1978: 104; his translations) Bi-yi-rkab il-husaan. bi-3:IPFV-mount DEF-horse ‘He is mounting the horse.’ Huwwa raakib il-husaan. he mount:AP DEF-horse ‘He is riding (i.e. has mounted) the horse.’

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

780

780

Karen H. Ebert

(42) a.

b.

c.

Modern Standard Arabic (McCarus 1976: 21–22; his translations) irkab l-tayyaarata ª mount:IMPER DEF-plane ‘Get on the plane!’ rakib-naa min bayruuta Èilaa l-saami. to DEF-Damascus ride-1PL:PFV from B. ‘We rode from Beirut to Damascus.’ raakib-an Éalaa hisaan-in. wasala ª ª ª arrive:3:PFV mount:AP-ACC on horse-ACC ‘He arrived riding a horse/ mounted on a horse.’

For some other initio-transformative verbs Maltese diverges from the usual Arabic pattern. The verb arafa means ‘get to know, know’ in Standard Arabic, Syrian and Egyptian. The verb in the Perfective usually activizes the transition ‘got to know’, the Imperfective the stative meaning ‘knows’. The state can also be referred to by the Active Participle aarif. A past state is expressed by kaana  IPFV or kaana  AP. Maltese has split this verb up into garaf ‘recognize, realize’ and the lexical imperfective jaf ‘know’. The ingressive meaning ‘get to know’ can be expressed with the help of sar ‘become’. This change in the actionality of the Maltese verb can hardly be due to Romance influence, as the behavior of Italian and French ‘know’ corresponds more closely to the older Arabic pattern. Table 7. The verb ‘know’ in Arabic and Romance

MSA Syr

Malt

It Fr

Transition

Present state

Past state

‘came to know’ PFV

‘knows’ IPFV

AP

‘knew’ PT:IPFV / PT:AP

arafa araf

ya rifu biy arif

aarif aarif

kaana ya rifu / kaana aarif kaan biy arif / kaan aarif

‘became’  IPFV

IPFV

PT:IPFV

sar jaf

jaf

kien jaf

PT:PFV

PRS

PT:IPFV

seppe sut

sa sait

sapeva savait

In other dialects of Arabic, f-h-m ‘understand’ is a transformative verb only, so that ‘do you understand?’ is expressed by a Perfective: fahim-t? The corresponding Maltese verb can also be used as an activity verb, cf. the progressive qed nifhem in

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

781

Aspect in Maltese

781

(23d). The sources on Syrian and Egyptian list numerous ‘inchoative’ verbs. Although it remains to be investigated whether all of them really are inchoative (or ingressive?),20 the Maltese counterparts rarely are. I thus come to the conclusion that there is no evidence for Romance influence in the organization of the Maltese aspect system, but that it corresponds to the patterns found in other Arabic dialects. It is also unlikely that Romance should have influenced the TA forms used in subordinate clauses; Romance languages have past or conditional forms in conditional clauses. An influence from European languages is, however, possible in the domains where Maltese differs most from other Arabic dialects; i.e., the reduced inventory and use of Active Participles and a different actionality of many verbs. Whether this is due to Romance influence is hard to determine. Sometimes the result is more similar to English, e.g., the stative verb ‘know’. The qed-Progressive resembles the Iberian type in its use (cf. Bertinetto, this volume), but a direct influence from Iberian languages is unlikely, given the fact that Maltese has been heavily influenced by Italian (especially in the vocabulary). It is just as likely that the changes in Maltese are independent developments. ‘Know’ is a stative verb in many languages, and the development from progressive towards a more general present or imperfective is a universal tendency, as is the relegation of the imperfective to the inactual domain.

Notes 1. This article owes a lot to Ray Fabri, who not only translated the sentences of the EUROTYP Progressive Questionnaire (PROGQ), but also answered numerous questions following it. Our e-mail discussion helped a lot to clear my ideas about Maltese. Thanks are also due to Daniela Triantafyllidou for eliciting some relevant data from her Syrian informant. 2. The problem with Vanhove’s data is not only that her examples and texts are given in a phonetic transcription, but also that she doesn’t make clear which data are standard Maltese and which are dialectal. As my presentation is restricted to the system of standard Maltese, I have left out all dialectal and unclear examples. Sentences cited from Vanhove are rendered in the standard orthography here. 3. As Maltese has no infinitive the 3rd person masculine Perfective form is used for citation. 4. Ser and se are most probably abbreviated from sejjer, while sa is also found in Standard Arabic. Maltese has a further future marker a-, which is avoided in educated speech. Its function seems otherwise not to differ from sa. 5. I leave the auxiliary unanalyzed and gloss it as PT (instead of, e.g., kien be:3:PFV); for the forms of ‘be’ see Table 1. 6. kieku ( kien ikun) marks counterfactual conditionals 7. Fabri marginally accepts the combination ikun  IPFV with the adverb ‘tomorrow’, although he states that it would be more common to have the progressive marker qed in the following sentence:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

782

782

Karen H. Ebert

nkun n-ilg ab fi-l-g nien meta t-i˙gi. Jien g ada tomorrow IPFV 1:IPFV-play in-DEF-garden when 2:IPFV-come I ‘I will be playing in the garden when you come tomorrow.’ (Fabri 1995) (better: nkun qed n-ilg ab) 8. Schabert (1976: 132) mentions two sentences where ikun  PFV seems to have present perfect meaning, cf. (dialectal): yk¯unu abdu šaaˇga (IPFV catch:3PFV:PL something) ‘They have caught something.’ However, without context we cannot be sure that this is not a habitual sentence. (Schabert collected his texts by interviewing fishermen about their work; the two texts printed in Schabert (1976) are habitual throughout.) 9. I prefer Johanson’s term “initio-transformative” (cf. Johanson, this volume, section 7) to the misleading term “inchoative-stative” (cf. Breu 1985, Sasse 1991), which I adopted earlier (Ebert 1995). Most initio-transformative verbs are not inchoative in the traditional sense, and with many the situation resulting from the initial transformation is not necessarily stative. 10. For ‘sit’ the construction qieged bi-l-qiegda is used; obviously the meaning of the verb has been bleached too much (as a progressive marker), so that a stronger form was needed. Cf. fi-l-k˙cina. Qieg d-a bi-l-qieg da stay:AP-F at-DEF-sit:NML in-DEF-kitchen ‘She is sitting in the kitchen.’ (PROGQ: 28) 11. I shall not go into Borg’s idiolect, in which the AP of motion verbs is exchangeable with the qed-form. Borg also accepts the following sentences, judged ungrammatical by other speakers (Borg 1981: 153). a.

b.

˙ Ganni qed j-imxi mi-d-darsa l-iskola b alissa. G. PROG 3:IPFV-walk from-DEF-house to DEF-school just_now ‘Gianni is walking from home to school just now.’ ˙ Ganni miexi mi-d-dar sa l-iskola da-˙z-˙zmien, G. walk:AP:M from-DEF-house to DEF-school this-DEF-time il-karozzi m-hu-x qed j-a hm-u. g ax because DEF-bus NEG-he-NEG PROG 3:IPFV-work-PL ‘Gianni walks from home to school these days, because the buses are not running.’

12. This is also true of resultative APs, e.g., imissek tkun rieqed ‘you should be asleep’. 13. There are other tables in Vanhove (1993), which include the Progressive and also a ‘durative aspect’. The latter is marked by qagad ‘stay, sit’. This construction is a durative actional phrase of the type sits reading, Dutch zit te lezen. Unlike in Dutch and other Germanic languages (cf. Ebert, this volume) there is no overlap with the progressive in Maltese. 14. Borg has a short chapter on the interaction of aspect and aspectual character (1981: 150– 154). His examples are:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

783

Aspect in Maltese mexa telaq wasal

walk leave arrive

783

– an extended journey – a border crossing followed by an extended journey – an extended journey followed by a border crossing.

This is not a description of the actionality of the verbs. Both telaq and wasal designate only transformations (‘border crossings’), and not the journey that follows or precedes. 15. The verb ‘be’ is an exception: its future is formed with se (probably to avoid the sequence *ikun ikun); cf. Min sa jkun il-kaptan tat-tijm g ada? ‘Who will be the captain of the team tomorrow?’ (PROGQ: 69) 16. I know of no treatment of those verbs in Maltese, but for other varieties of Arabic cf. Blohm (1981), Cowell (1964), McCarus (1976), Eisele (1990b). 17. I will not discuss which of the two types of perfectives is more prototypical. The two types are different systems of organizing actionality and viewpoint, and it is unfortunate that they have been designated by the same name, leading to considerable confusion in the literature. For an excellent clarification of the different features involved see Johanson (this volume). 18. A similarly unclear situation holds in Modern Standard Arabic. McCarus (1976: 8) states: “ the imperfect yakuunu, which normally occurs in MSA with the predictive particle sa-, makes a prediction”; e.g., sa-ykuunu qad jaaÈa ‘he will have come’. Kropfitsch (1980: 127) deals with the perfect periphrasis yakuunu qad  PFV and gives examples for present, past and future perfect without mentioning the possibility of the particle sa; e.g., wa-bi-d¯alika yak¯unu qad tamma tanfidu l-insiha¯ b ‘and with that the retreat will ª be completed’. 19. In Maltese the lexical imperfectives have to appear in the IPFV also in past counterfactual clauses, cf. kieku n-af ‘if I had known’ in (21c). This is a morphological restriction rather than a semantic one. Other stative verbs require the PFV, as kieku bqajt hemm ‘if I had been there’ in (38b) shows. 20. Eisele (1990b: 227ff) lists verbs like Cairene šaaf ‘see’, saaÈ ‘drive’, xaaf ‘fear’, naam ‘fall asleep’ with the ‘inchoatives’.

References Aquilina, J. 1965 Teach yourself Maltese. London: The English Universities Press. Bertinetto, Pier Marco this volume “The progressive in Romance and in English”. Blohm, Dieter 1981 “Zur Funktion des Partizip Aktiv im modernen Hocharabisch”, ZPSK 34: 143–151. Borg, Albert J. 1981 A study of aspect in Maltese. Ann Arbor: Karoma. 1988 Ilsienna. Studju grammatikali. Malta. Borg, Albert J. & Bernard Comrie 1984 “Object diffuseness in Maltese”, in: Frans Plank (ed.), Objects. Towards a theory of grammatical relations. London etc.: Academic Press, 109–126.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

784

784

Karen H. Ebert

Breu, Walter 1985

“Handlungsgrenzen als Grundlage der Verbklassifikation”, in: W. Lehfeldt (ed.), Slavistische Linguistik 1984. München: Sagner. 9–34. Bugeja, Kaptan Pawlu 1982 Kelmet il-Malti. Dizzjunarju Malti–Ingliz, Ingliz–Malti. Malta: Gulf Publishing. Bybee, Joan & Östen Dahl 1989 “The creation of tense and aspect systems in the languages of the world”, Studies in Language 13: 51–103. Cohen, David 1984 La phrase nominale et l’évolution du système verbal en sémitique. Études de syntaxe historique. Leuven-Paris: Éditions Peeters. Comrie, Bernard 1976 Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge University Press. 1985 Tense. Cambridge University Press. 1991 “On the importance of Arabic for general linguistic theory”, in: Bernard Comrie & Mushira Eid (eds.), Perspectives on Arabic Linguistics III. Papers from the Third Annual Symposium on Arabic linguistics. Amsterdam: Benjamins. 3–30. Cowell, Mark W. 1964 A reference grammar of Syrian Arabic. Washington D.C.: Georgetown University Press. Dahl, Östen 1985 Tense and aspect systems. Oxford: Blackwell. Ebert, Karen 1995 “Ambiguous perfect-progressive forms across languages", in: Bertinetto, Pier Marco, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.), Temporal reference, aspect and actionality. Vol. 2: Typological perspectives. Torino: Rosenberg & Sellier. 185– 203. this volume “Progressive markers in Germanic languages”. Eisele, John 1990a “Time reference, tense and formal aspect in Cairene Arabic”, in: Mushira Eid (ed.), Perspectives on Arabic Linguistics I. Papers from the First Annual Symposium on Arabic Linguistics. Amsterdam: Benjamins. 173–233. 1990b “Aspectual classification of verbs in Cairene Arabic”, in: Mushira Eid & John McCarthy (eds.), Perspectives on Arabic Linguistics II. Papers from the Second Annual Symposium on Arabic Linguistics. Amsterdam: Benjamins. 192–233. Fabri, Ray 1993 Kongruenz und die Grammatik des Maltesischen. Tübingen: Niemeyer. 1995 “The tense and aspect system of Maltese”, in: Rolf Thieroff (ed.), Tense systems in European languages II. Tübingen: Niemeyer. 327–343. Johanson, Lars this volume “Viewpoint operators in European languages”. Kropfitsch, L. 1980 “Semantische Tendenzen im Neuhocharabischen”, Zeitschrift für Arabische Linguistik 5. McCarus, Ernest N. 1976 “A Semantic analysis of Arabic verbs”, in: Louis L. Orsin (ed.), Michigan Oriental Studies in honor of George Cameron. Ann Arbor. Mitchell, T. F. 1978/1956 An introduction to Egyptian Colloquial Arabic. Oxford: Clarendon Press. Myhill, John 1992 Typological discourse analysis. Oxford: Blackwell.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

785

Aspect in Maltese

785

Nedjalkov, Vladimir P. & Sergej Je. Jaxontov 1988 “The typology of resultative constructions”, in: V. P. Nedjalkov (ed.), Typology of resultative constructions. Amsterdam: Benjamins. 3–62. Rössler, Otto 1950 “Verbalbau und Verbalflexion in den Semitohamitischen Sprachen: Vorstudien zu einer vergleichenden Semitohamitischen Grammatik”, Zeitschrift der Deutschen Morgenländischen Gesellschaft 100, 2: 461–514. Sasse, Hans-Jürgen 1991 “Aspekttheorie”, in: H.-J. Sasse (ed.), Aspektsysteme. Institut für Sprachwissenschaft, Universität Köln. ANF Nr. 16. 1–35. Schabert, Peter 1976 Laut- und Formenlehre des Maltesischen anhand zweier Mundarten. Erlangen: Palm & Enke. Vanhove, Martine 1993 La langue maltaise. Études syntaxiques d’un dialecte arabe “périphérique”. Wiesbaden: Harrassowitz. Woidich, Manfred 1975 “Zur Funktion des aktiven Partizips im Kairenisch-Arabischen”, Zeitschrift der Deutschen Morgenländischen Gesellschaft 125: 273–293.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

786

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

787

Appendices

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

788

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

789

1 The Future Time Reference Questionnaire Part 0 [contained mainly practical instructions, omitted here]

Part I 1. (TMAQ:15) Q: What your brother DO if you don’t go to see him today, do you think? A: He WRITE a letter (to me) 2. (TMAQ:16) Q: What your brother DO when we arrive, do you think? (=What activity will he be engaged in?) He WRITE letters (=He MANAGE his correspondence) 3. (TMAQ:17) Q: What your brother DO when we arrive, do you think? (=What activity will he be engaged in?) He WRITE a letter 4. (TMAQ:22) Q: What are you planning to do right now? A: I WRITE letters (= I MANAGE my correspondence) 5. (TMAQ:23) Q: What are you planning to do right now? A: I WRITE a letter 6. (TMAQ:24) Neither A nor B can see B’s brother. A: What he DO right now, do you think? (=What activity is he engaged in?) He WRITE letters (I think so because he does that every day at this time) 7. (TMAQ:27) A: My brother has got a new job. He’ll start tomorrow. B: What kind of work he DO there? He WRITE letters 8. (TMAQ:36) It’s no use trying to swim in the lake tomorrow The water BE COLD (then) 9. (TMAQ:79) If you PUT a stone into this bag, it BREAK 10. (TMAQ:80) Even if you PUT a stone into this bag, it not BREAK 11. (TMAQ:81) Q: What HAPPEN if I eat this mushroom? You DIE 12. (TMAQ:82) (According to the contract) we not WORK tomorrow 13. (TMAQ:103) The boy is expecting a sum of money

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

790

790

Appendices

When the boy GET the money, he BUY a present for the girl 14. (TMAQ:104) The boy thinks that he will perhaps get a sum of money If the boy GET the money, he BUY a present for the girl 15. (TMAQ:105) the speaker knows the boy was expecting money, but he doesn’t know if he got it If the boy GET the money (yesterday), he BUY a present for the girl 16. (TMAQ:106) the speaker knows the boy was expecting money and that he did not get it If the boy GET the money (yesterday), he BUY a present for the girl 17. (TMAQ:107) Talking to someone who is leaving in a while When you RETURN, I WRITE this letter(=I FINISH it already at that time) 18. (TMAQ:108) Said as an order by a teacher leaving the classroom When I RETURN, you WRITE this assignment (=You FINISH it by then) 19. (TMAQ:124) Of the water in a lake which is not visible to the speaker and the hearer My brother HOPE (right now) that the water BE COLD 20. (TMAQ:125) Uttered as a promise I PROMISE to COME to you tomorrow 21. (TMAQ:131) You MUST GO to bed before you GET tired (today) 22. (TMAQ:132) (Yesterday evening) I GO to bed before my brother COME home 23. (TMAQ:145) Traveller to local: If you SHOW me the way, I GIVE you money 24. (TMAQ:146) Mother to child: If you not STOP PLAY with that ball, I TAKE it away 25. (TMAQ:152) Said by a young man When I GROW old, I BUY a big house 26. My brother is late for dinner. When he ARRIVE, the food BE cold. 27. Mother to children: We EAT (alt. HAVE dinner) now! 28. Do you intend to stay here? No, I LIVE in X-place next year. 29. Talking about the speaker’s plans for the evening: I STAY at home. 30. Talking about the speaker’s plans for the evening: I WORK in the garden. 31. Talking about the speaker’s plans for the evening: I WRITE a letter. 32. Talking about the speaker’s plans for the evening:

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

791

The Future Time Reference Questionnaire

I GO to town. 33. Talking about the speaker’s plans for the evening: I GO to bed early. 34. Talking about the speaker’s immediate plans: I STAY at home. 35. Talking about the speaker’s immediate plans: I WORK in the garden. 36. Talking about the speaker’s immediate plans: I WRITE a letter. 37. Talking about the speaker’s immediate plans: I GO to town. 38. Talking about the speaker’s immediate plans: I GO to bed. 39. Talking about the speaker’s plans for tomorrow: I STAY at home. 40. Talking about the speaker’s plans for tomorrow: I WORK in the garden. 41. Talking about the speaker’s plans for tomorrow: I WRITE a letter. 42. Talking about the speaker’s plans for tomorrow: I GO to town. 43. Talking about the speaker’s plans for tomorrow: I GO to bed early. 44. My brother is tired. He WAKE UP late tomorrow. 45. My brother is tired. He FALL ASLEEP early in the evening. 46. There are black clouds in the sky. It RAIN in the evening. 47. There are black clouds in the sky. It RAIN (very soon). 48. There are black clouds in the sky. It RAIN in a few minutes. 49. The weather is changing. It RAIN tomorrow. 50. The weather is changing. Maybe it RAIN tomorrow. 51. The weather is changing. It be COLD in the evening. 52. The weather is changing. It be COLD tomorrow.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

791

792

792

Appendices

53. The sun RISE at six o’clock tomorrow. 54. Does your brother intend to stay here? No, he LIVE in X-place next year. 55. Talking about a third person’s plans for the evening: He STAY at home. 56. Talking about a third person’s plans for the evening: He WORK in the garden. 57. Talking about a third person’s plans for the evening: He WRITE a letter. 58. Talking about a third person’s plans for the evening: He GO to town. 59. Talking about a third person’s plans for the evening: He GO to bed early. 60. Talking about a third person’s immediate plans: He STAY at home. 61. Talking about a third person’s immediate plans: He WORK in the garden. 62. Talking about a third person’s immediate plans: He WRITE a letter. 63. Talking about a third person’s immediate plans: He GO to town. 64. Talking about a third person’s immediate plans: He GO to bed. 65. When I GET home in the evening, my mother BE HAPPY. 66. If it RAIN tomorrow, we STAY at home. 67. If it BE COLD tomorrow, we STAY at home. 68. If I GET the money tomorrow, I BUY a present for you. 69. If I GET the money today, I BUY a present for you. 70. My brother SAY yesterday that he COME here today. 71. My brother HOPE yesterday that you COME here today. 72. My brother SAY yesterday that he COME here next week. 73. My brother SAY yesterday that it RAIN today. 74. My brother SAY yesterday that it BE COLD today. 75. My brother HOPE yesterday that it BE COLD today. 76. My brother WANT (now) to buy a house. 77. This screw does not want to turn. 78. My brother WANT (now) to be a farmer. 79. My brother INTEND to buy a house. 80. I HAVE TO thank my brother for helping me. 81. My brother MUST sell his house now. 82. My brother MUST sell his house soon.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

793

The Future Time Reference Questionnaire

83. 84. 85. 86. 87. 88. 89.

90. 91. 92. 93. 94. 95.

96.

97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107.

793

My brother CAN lift this stone. My brother CAN read and write. I BE ABOUT TO fall asleep. Yesterday, I BE ON THE VERGE OF being run over by a car. What you DO when you COME home in the evening? I WRITE a letter, (then) I DRINK some tea and (then) I GO to bed. How I GET to your brother’s house? You TURN left at the crossing. How I GET to your brother’s house? You TURN left at the crossing, (then) WALK for ten minutes and (then) TURN left again. (According to the schedule) the train LEAVE at noon. My brother SAY (now) that he GO to town tomorrow. My brother SAY (now) that it RAIN tomorrow. My brother BELIEVE (now) that it RAIN tomorrow. My brother HOPE (now) that it RAIN tomorrow. What you DO right now? I WRITE a letter to my brother in order that he KNOW that I COME to see him. What you DO yesterday? I WRITE a letter to my brother in order that he KNOW that I COME to see him. A: You promised to make some food for me. When it BE ready? B: It BE READY in five minutes. (TMAQ:73) What kind of sound do cats make? They MEOW (TMAQ:18) What your brother usually DO after breakfast? He WRITE letters I heard a funny story the other day. When you HEAR it, you LAUGH. A: I have a headache. B: Take this medicine. It MAKE you feel better. I not LIKE this person (now) and I not LIKE him (in the future). I not KNOW where he BE. (an order:) OPEN the door! (let us) OPEN the door! (a prohibition): (do) not OPEN the door! (a warning): (Look out, do) not STEP in the mud!

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

794

794

Appendices

108. (wishing someone good health:) (may) you always BE HEALTHY! 109. (Uttered at eight o’clock - the speaker’s brother left at six and has not returned yet): He RETURN at seven o’clock 110. (Added in June 1991 version) Yesterday when I woke up in the morning, there were dark clouds in the sky. I took my umbrella, because it RAIN in a few minutes. 111. (Added in June 1991 version) I met your brother a few days ago. He was very worried, because he GO to the dentist next day.

Part II 1. FTR devices in the questionnaire sentences Which grammatical devices are involved in expressing future time reference in the sentences in Part I? For each device D, give the following information: (a) Does D have bound (morphological) or free (periphrastic, syntactic) expression or is it unmarked? (b) If the answer to a) is “bound”, then specify the kind of morphological processes involved and list the morphemes (or their main variants) used as markers (c) If the answer to b) is “free”, describe the syntactic and lexical make-up of the construction (d) What is the historical source (etymology) of D (if known)? (e) If D has developed out of a known historical source, are there uses of D (or homonymous expressions) that reflect that source? (Explanatory example: if a future has developed out of a verb meaning ‘to want’, can it still be used to express volition without any implication of the desired state-of-affairs coming true?) (f) Is the use of D restricted to future time reference? If not, what are the other main uses? (g) Is the use of D in any way dependent on style or register? Among the ‘FTR devices’, it may be practical to include also ‘lack of marking’, as when a present tense or a bare verb stem is used for FTR.

2. Additional FTR devices Are there any grammatical devices that do not occur in the translations of Part I but which you know are regularly used for future time reference?

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

795

The Future Time Reference Questionnaire

795

If the answer is yes, (a) describe them in the same way as the devices under 1) (b) try to characterize the main uses with illustrative examples

3. Intention vs. prediction Background: References to future time differ in many ways, one of the most important being whether they involve an element of INTENTION or PLANNING by a conscious being. A proposition that describes an intention or a plan does not by itself necessarily imply anything about the future; “John intends to fly to the moon” may be true even if we know that he’ll never go there. Thus, for a sentence to as a case of future time reference, we also require it to contain an element of PREDICMON, that is, a claim that the state-of-affairs described will come true at some point in the future. A major distinction will thus be between intentional and pure-prediction FTR sentences. Typical examples in Part I of the two kinds are: intentional: 4, 5, 29–43 pure prediction: 8, 11, 26, 44–52 Question: What are the major grammatical devices that characterize (a) intentional FTR? (b) pure-prediction FTR?

4. Scheduling Background: The ontological status of the future has been a controversial issue since Aristotle – is the future determined in the same way as the present or the past or does it only exist as a potentiality? One might argue that natural language leaves the question open in that speakers may choose between different ways of viewing future states-of-affairs, either as determined or as undetermined. This may be difficult to distinguish from the question of certainty: do we know or do we just make guesses about the future? In any case, it has been observed that languages may refrain from marking FTR grammatically (i.e., use forms that normally refer to the present) for future statesof-affairs that are somehow regarded as predetermined, e.g., Tomorrow is Sunday, and also for things that are ‘scheduled’ in a more or less direct sense, e.g., The train leaves at 6. Examples of such sentences in Part I are 12, 90.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

796

796

Appendices

Question: (a) Do sentences expressing ’predetermined’ or ‘scheduled’ future differ in any way from other FTR sentences? (b) Does ‘degree of certainty’ influence the choice of FTR devices?

5. Differences between grammatical persons Are there, for any of the devices treated above, systematic differences between 1st, 2nd and 3rd person? Note: this does not refer to ordinary inflectional differences also found in other forms but rather patterns like the alternation between shall for 1st person and will for 2nd and 3rd person in English.

6. FTR and aspect Are there any grammaticalized aspectual distinctions in FTR sentences? (a) Is there any systematic difference between imperfective and perfective sentences (e.g. between the following groups in Part I: 2, 3, 7, 8, 12 (imperfective) vs. 1, 4, 5, 11, 13 (perfective)? If the answer is yes, what is the relation to the expression of the distinction imperfectivity/perfectivity in other contexts? (b) Is progressivity marked – optionally or obligatorily – in FTR sentences (cf. sentences 2–3 in Part I)? In other words, are progressive constructions used in FTR sentences and do they combine with any of the devices listed above? (Note the difference between this question and No. 8) (c) Is habituality marked – optionally or obligatorily – in FTR sentences (cf. sentence 7 in Part 1)? Is there interaction with other markers?

7. Progressives as FTR devices To what extent are constructions that normally have progressive meaning used in FTR sentences without that meaning? What are the restrictions, if any?

8. FTR in subordinate clauses How is FTR normally marked in the following kinds of subordinate clauses when the matrix clause has non-past time reference? (a) temporal clauses (ex. 17, 18, 25, 26, 65) (b) conditional clauses (ex. 14, 15, 16, 23, 24, 66–69) (c) complements of verbs of saying, believing etc. (91–93)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

797

The Future Time Reference Questionnaire

797

(d) complements of verbs of wishing, hoping etc. (94) (e) purpose clauses (95) (f) ‘before’-clauses (21)

9. “Future in the past” How is “future in the past”, i.e. reference to points in time that are future relative to some point of reference in the past, marked in the following kinds of clauses? (a) complements of verbs of saying, believing etc. (70, 72–74) (b) complements of verbs of wishing, hoping etc. (71, 75) (c) purpose clauses (96) (d) ‘before’-clauses (22) (e) in main clauses in contexts like (109)

10. FTR in marked speech act types Are there any peculiarities in the expression of FTR in the following kinds of speech acts? (a) promises with first person subjects (17,20) (b) promises with non-first person subjects (97) (c) threats (24)

11. Additional uses of FTR devices Which of the devices used to express FTR may also be used (alone or in combination with others) for non-future time reference with the following kinds of interpretation? (a) guesses, suppositions etc. (24) (b) generic and habitual statements (98, 99) (c) counterfactual conditions (16) (d) complements of negated cognitive verbs such as ‘know’(103) (e) to express ‘politeness’, e.g., in requests

12. Imperatives/optatives Which of the following may be expressed with devices that are otherwise used for ‘normal’ FTR? (a) 2nd person imperatives (104) (b) 1st person imperatives (105) (cf. also 11e) (c) optatives (108)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

798

798

Appendices

(d) prohibitions (106) (e) warnings (107)

13. Prospectivity Background: One factor that may influence expression of FTR is what (following Bernard Comrie) can be labelled ‘prospectivity’, that is, the question whether a future event is related directly to a present state, as when something ‘is about’ to happen or seen as necessary by the way things are at present. Examples in Part I would be 46, 47. The distinction in English between be going to and will has been claimed to related to prospectivity. Question: Is prospectivity as defined here relevant for the choice between different ways of expressing FTP,?

14. Remoteness distinctions in the future Background: In many languages, different devices are used for expressing FTR depending on the distance between the point of speech and the time to which reference is being made. When such distinctions are made systematically, the cut-off point is often between what will happen today and what will happen tomorrow or later, but other distinctions are also possible. Even in languages where no such grammaticalized cutoff points are found, it happens relatively often that some FTR devices are claimed to be more commonly used, e.g., for events that are close in time. Sometimes this seems to be related to notions such as ‘intention’ and ‘prospectivity’. Questions: (a) Can any systematic differences be found in the use of FTR devices that can be related to remoteness distinctions? (cf., e.g., 46 vs. 47 vs. 48) (b) If the answer to (a) is ‘yes’, are there identifiable cut-off points, e.g., ‘today : tomorrow and later’, ‘this year : next year’ etc.?

15. Sequences of events Are there any peculiarities in the expression of sequences of events in the future (as compared to isolated events, cf. 87 and 89 vs. 88)? If the answer is yes, how do these peculiarities relate to the expression of sequential events in the past (e.g., in narratives)?

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

799

The Future Time Reference Questionnaire

799

16. Perfectivity/telicity/inchoativity as markers of FTR Background: Certain types of verb phrases tend not to be interpreted as referring to the moment of speech. This concerns generally verb phrases denoting punctual events (such as ‘die’) and in certain languages, verbs grammatically marked as being perfective aspect. As a consequence, verbs not marked as having past time reference will easily be understood as referring to the future, something which may or may not be exploited systematically in the language. Question: Is grammatical aspect (or similar devices) exploited to mark FTR?

17. Inchoativity and FTR Background: To the verb phrases mentioned in 17 which tend to have FTR belong those denoting changes of state. In addition, a future change of state event implies a future state – if I become rich in the future, I will be rich. This opens up the possibility for secondary stative interpretations of such verb phrases. Question: Do inchoative verb phrases (verb phrases denoting changes of state) have stative interpretations with FTR (e.g., would one say ‘I become rich’ meaning ‘I will be rich’)?

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

800

2 The Perfect Questionnaire Instructions The aim of this questionnaire is to investigate the form, meaning and use of the perfect and similar or related grammatical categories – whether inflectional, periphrastic or otherwise marked – in various languages. It can also be applied to a language in which there is no category resembling to the perfect. For a successful completion of the questionnaire, two persons are usually needed: a professional linguist who is at least superficially acquainted with the structure of the language under investigation (referred to as “L”), and an informant who is a native speaker of L (meaning ‘a person who used L actively and daily while growing up’). Several native speakers may be used so as to enhance the reliability of the results. Linguists who are native speakers themselves may do without an informant, but they should check their intuition with “non-professionals” to avoid obtaining results they are predisposed to obtain. There are two parts in this questionnaire: Part I (“Examples”) consists of sentences-with-contexts which are to be translated into L. It is essential to translate each example as a whole, not the individual words, aiming at a maximally natural rendering in L. Individual words (such as temporal adverbs or culturally biased lexical items) can be deleted, changed or added, if this makes the examples more natural. Also pay attention to the position of temporal adverbials, for instance: the most natural position is not always the same in other languages as it is in English. In order not to influence the translation, the finite verb of each English sentence is usually left uninflected (and written in CAPS); the meaning intended should be clear from the context. Each translated example should also be provided with word-by-word glosses, showing not only the lexical meanings but also the inflectional categories. Part II (“Description”) contains questions about a perfect-like grammatical category which is referred to as "P", and identified with the help of the first examples. In order to answer them, the investigator usually has to elicit more sentence examples than there are in this questionnaire, or perhaps consult reference grammars (if available) or survey some printed texts. Use separate sheet(s) to answer the theoretical questions; make reference to the translated sentences when necessary. In the sentences to be translated, the investigator must strictly avoid causing P to be used more than is natural. However, it is sometimes asked in Part II whether P would have been possible in the examples of Part I where it was not used, and if so, how it would have changed the meaning or other properties of those sentences. These alternative translations must not be presented to the informant before all of Part I has been translated lest the subsequent responses become biased.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

801

The Perfect Questionnaire

801

Part I: EXAMPLES 1. [A: I want to give your sister a book to read, but I don’t know which one. Are there any of these books that she READ already?] B: Yes, she READ this book. 2. [A: It seems that your sister never finishes books.] B: (That is not quite true.) She READ this book (= all of it). 3. [Question: Is the king still alive?] No, he DIE. 4. Question: You MEET my sister (at any time in your life up to now)? 5. [A child asks: Can I go now?] Mother: You DO your homework? 6. [Question: Do you know my sister?] Answer: Yes, I MEET her (so I know her). 7. [Question: Can you swim in this lake? (=Is it possible for anybody to swim in this lake?)] Answer: Yes, at least I SWIM in it several times. 8. [Do you know what happened to me just an hour ago?] I WALK in the forest. Suddenly I STEP on a snake. It BITE me in the leg. I TAKE a stone and THROW (it) at the snake. It DIE. 9. [Do you know what happened to me yesterday?] I WALK in the forest. Suddenly I STEP on a snake. It BITE me in the leg. I TAKE a stone and THROW (it) at the snake. It DIE. 10. [Do you know what happened to my brother yesterday? I saw it myself.] We WALK in the forest. Suddenly he STEP on a snake. It BITE him in the leg. He TAKE a stone and THROW (it) at the snake. It DIE. 11. [Do you know what happened to me once when I was a child? (Note: The speaker was, however, old enough to remember the incident.)] I WALK in the forest. Suddenly I STEP on a snake. It BITE me in the leg. I TAKE a stone and THROW (it) at the snake. It DIE. 12. [This happened to me just an hour ago.] I SIT under a tree, when an apple FALL on my head. (Or, if more natural: While I SIT under a tree, an apple FALL on my head.) 13. [Do you know what happened to me once when I was a child? (Note: The speaker was, however, old enough to remember the incident.)] I SIT under a tree, when an apple FALL on my head. (Or, if more natural: When I SIT under a tree, an apple FALL on my head.) 14. [It is morning. A wakes up, looks out of the window and sees that the courtyard (or the street) is wet.] A: It RAIN during the night. 15. [Question: You MEET my sister (at any time in your life up to now)?]

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

802

802

Appendices

Answer: Yes, I MEET her several times. 16. [A question asked at 9 o’clock A.M.: Why do you look so tired?] Answer: I WAKE UP at 4 o’clock this morning (or: today). 17. [A question asked at 3 o’clock P.M.: Why do you look so tired?] Answer: I WAKE UP at 4 o’clock today. 18. [A question asked at 9 o’clock A.M.: Why do you look so tired?] Answer: I NOT SLEEP well during the night. 19. [A question asked at 3 o’clock P.M.: Why do you look so tired?] Answer: I NOT SLEEP well during the night. 20. [A has got his wages and says:] I GET my wages today, so I can now BUY you a beer. 21. I GET my wages yesterday, so I can now BUY you a beer. 22. [Note: These sentences do not necessarily imply the passive voice though BE BORN happens to be formally a passive in English. Treat it as a single lexical unit.] A: When you BE BORN? – B: I BE BORN on the first of June 1950. 23. [A guide presenting his home town to tourists. Note: This sentence does not necessarily imply the passive voice, unless it really is the most natural way of expressing this sentence in L.] Our town BE FOUNDED in 1550. 24. [Question: Do you know what remarkable event TAKE PLACE in 1550? Note: as in 23.] Answer: In that year, our town BE FOUNDED. 25. [Question: When Columbus ARRIVE at America for the first time?] Answer: He ARRIVE at America in 1492. 26. [Question: What do you know about this novel? Note: This sentence does not necessarily imply the active voice or the word order given here if it is not natural in L.] Answer: Graham Greene WRITE it. 27. [Question: Your sister still BE at home?] Answer: No, she already GO AWAY. 28. [B’s sister is known to have gone to another town. Question: A: Your sister COME BACK? (Note: a free translation may be needed for B’s answer.)] B: No, she still GO AWAY. 29. [As in 28. Question: Your sister COME BACK?] Answer: No, she NOT COME BACK yet. 30. [A: Don’t talk so loud! You’ll wake the baby.] B: He WAKE UP already. 31. [The baby wakes up one hour earlier than expected and starts screaming. Mother (in another room):]

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

803

The Perfect Questionnaire

803

Oh no! He WAKE UP already! 32. [Note: use BE or VISIT, or some other predicate, according to what sounds the most natural in L.] You BE to (VISIT) Australia (ever in your life)? 33. [These are alternative answers to 32. They should all be translated.] No, I never BE (VISIT) there. – Yes, I BE (VISIT) there. – Yes, I BE (VISIT) there several times. – Yes, I BE (VISIT) there in January 1987. 34. [A has been talking about the way of life in Australia. Note: the sentence construction may have to be changed – even in English.] B: You BE to (VISIT) Australia as you know all that? – A: Yes, I BE (VISIT) there, so I know. 35. [Question: You MEET my sister (at any time in your life up to now)? Note: All these alternative answers should be translated.] No, I never MEET her. – Yes, I MEET her once. – Yes, I MEET her in January 1987. 36. [A has been talking to B about C’s personal tastes.Note: the sentence construction may have to be changed – even in English.] B: You MEET her (sometime) as you know all that? – A: Yes, I MEET her, so I know. 37. [It is cold in the room. The window is closed.] Question: You OPEN the window (and closed it again)? 38. [This is an answer to 37.] Yes, I OPEN it. 39. [This is an answer to 37.] No, I NOT OPEN it. 40. [The window is open but A has not noticed that. A asks B: why is it so cold in the room?] B: I OPEN the window. 41. [Question: Is your sister still abroad?] Answer: No, she COME BACK and is now staying with us. 42. [Question: I was told you are writing a book. How many pages you WRITE by now?] Answer: I WRITE fifty pages. 43. [Question: I was told you collect dolls. You COLLECT many of them?] Answer: I COLLECT some two hundred dolls by now. 44. [Question: I was told you intend to collect 300 different dolls. How many you already COLLECT?] Answer: I COLLECT some two hundred dolls by now. 45. [Question: I was told you always forget your umbrella somewhere. Is it true?] Answer: Yes, this year I LOSE five umbrellas. 46. [A is setting out on a long journey in an old car. B asks: What if something

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

804

804

47.

48. 49. 50.

51.

52. 53. 54.

55.

56. 57. 58.

59.

60.

Appendices

goes wrong with your car on the way?] A: I BUY spare parts and tools in case something happens ( = I have got them now). [Question: Why do you look so tired? (Note: you may replace “three days” by “three nights” or whatever seems most natural.)] Answer: I NOT SLEEP for three days. [She is still watching television! How long she DO that?] Answer: She WATCH (it) for three hours. [A is still living in this town.] A: I LIVE here for seven years. [A is still living in this town. As in 49, the intended meaning of LIVE is ‘to dwell somewhere’, not ‘to spend one’s life’.] A: I LIVE here all my life. [A is visiting a town she used to live in several years ago; now she lives somewhere else.] A: I LIVE here, so I know every street here. [As in 51. A now lives somewhere else!] A: I LIVE here for seven years, so I know every street here. [As in 51 and 52.] A: I LIVE here for seven years, but then I had to move away. [The speaker meets his friend about once a week; “the film” refers to a different film each time:] Every time I MEET him, he TELL me about the film he (just) SEE. [A has just seen the king arrive and reports it to B, who knows that the king has been expected to visit their town but does not know that he has now actually arrived.] A: The king ARRIVE! [A has just seen the king arrive. The event is totally unexpected.] A: The king ARRIVE! [Telling what a baby just DO. “N” should be replaced with a girl’s name.] N just SAY her first word! [A comes from the kitchen very agitated and tells B what he has just seen happen:] A: The dog EAT our cake! [A comes from the kitchen where he has just seen the sad remains of the cake. He tells B what he assumes to have happened:] A: The dog EAT our cake! [Do you know what happened to my brother yesterday? I did not see it, but he told me.] He WALK in the forest. Suddenly he STEP on a snake. It BITE him in the leg. He TAKE a stone and THROW (it) at the snake. It DIE.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

805

The Perfect Questionnaire

805

61. [This is the beginning of a story (tale). “Once upon a time” should be replaced with the formula stories typically begin with in L.] Once upon a time there was a man. He WALK in the forest. Suddenly he STEP on a snake. It BITE him in the leg. He TAKE a stone and THROW (it) at the snake. It DIE. 62. [A tells what she has heard from her father. Nothing shows that she would not believe it.] A: When my father BE a child, schools BE better than nowadays. 63. [A tells what she has heard from her father. Nothing shows that she would not believe it.] A: My father TELL me that when he BE a child, schools BE better than nowadays. 64. [A tells what she has heard people saying. Nothing shows that she would not believe it, but she does not present this as her own opinion. Add words if needed!] A: Sixty years ago schools BE better than nowadays. 65. [A doubts what her father has told her.] My father CLAIM that when he BE a child, schools BE better than nowadays. 66. [A does not believe what she has heard from her father; she only reports what he has told her.] A: When my father BE a child, schools BE better than nowadays. 67. [Said by a person who has just heard about the event but has not seen it.] The king ARRIVE! 68. [As in 67.] My sister just TELL me that the king ARRIVE. 69. [Investigating a burglary, seeing footprints beneath a window:] The thief ENTER the house by this window. 70. [A and B are not in the room in which B’s son has been doing his homework. Question: A: Is your son still doing his homework?] B: No, (I think) he FINISH (it) by now (or: already). 71. [An archaeologist, having investigated an excavation site, says:] This BE a huge city. 72. [An archaeologist, having investigated an excavation site, says:] This city BE DESTROYED about three thousands years ago. 73. [A guide, showing ruins to tourists:] This BE a huge city. 74. [A guide, showing ruins to tourists:] This city BE DESTROYED about three thousands years ago. 75. [A’s sister finished writing two letters just before A came home. A tells:] When I COME home yesterday, my sister WRITE two letters. 76. [A’s sister was not at home when A arrived. Question: Did you find your sister

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

806

806

77.

78.

79.

80. 81. 82. 83. 84.

85.

86. 87. 88.

Appendices

at home? A answers:] No, I did not (find her). She LEAVE. [A meets B’s sister. Later A moves to the town where B and B’s sister live. Still later, B asks A: When you came to this town a year ago, did you know my sister? A answers:] Yes, I MEET her. [Question: Why did you believe what she told you about Paris? Note: use BE or VISIT or whatever is most natural in L.] Answer: I BELIEVE her, because she BE to (VISIT) Paris. [The speaker used to meet his friend once a week, but nowadays he does not see him at all. “The film” refers to a different film each time:] Every time I MEET him in those years, he TELL me about the film he just SEE. [Looking at a house.] Who BUILD this house? [Looking at a picture of a house which has been torn down.] Who BUILD this house? [Question: Can I get my wages now?] Answer: I NOT PAY you your wages before you FINISH the entire job. [As in 82 above.] I PAY you your wages after you FINISH the entire job. [B is setting out on a journey. A intends to sell her own house while B is away. A tells B about this:] A: When you COME BACK next year, I SELL my house. [A began working here in June for almost thirty years ago. It is April and A tells that the anniversary is approaching:] A: In June this year I WORK here for thirty years. If I GET my wages tomorrow, I BUY you a beer. [The speaker has not received his wages yet:] The day I GET my wages I BUY you a beer. Those who GET their wages tomorrow certainly GO to have beer.

Part II: DESCRIPTION Tentative identification of P P is that gram (grammatical category) of L which is common to most of sentences (E01)–(E07) and has something to do with the relationship, temporal or not, between the present state of affairs and the past event referred to. This is a working definition which does not presuppose P is actually a perfect. If several grams co-occur in the

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

807

The Perfect Questionnaire

807

sentences, choose the one which seems to indicate more than mere “pastness” – things such as “result”, “relevance” or “completion” (provided that there exists such a gram). If there is no unique P identifiable – if there are alternative translations of single sentences, or if no gram has a clear majority among those used in (E01)–(E07) The questions about P then apply – you may choose more than one gram P1, P2, to all of these separately. Since P is a grammatical category, it must be marked either inflectionally or with a free morpheme, such as an auxiliary or an adverb-like element devoid of (full) lexical meaning. Semantic categories without a formal marking do not qualify! If you cannot find a P in L at all, you can merely translate the remaining example sentences and leave those theoretical questions that do not apply unanswered. (Q01) Did you find a P (or P1, P2, )? What is it (or what are they) called in reference grammars of L? How is its meaning (are their meanings) traditionally described? Note: all subsequent questions about P will also apply to P1, P2, ! (Q02) Describe how P is marked in L. Pay attention to the possible differences between different persons, or between main and subordinate clauses. (Q03) Describe the historical origin of P (if known). (Q04) Is there regional or social variation in the use of P, or variation between different styles and registers? If yes, what variety of L do the examples in Part I represent?

Narratives (Q05) In those sentences of texts (E08)–(E11) where P was not used, would it have been grammatically possible? How would it have changed the meaning or other properties of those sentences? (For a warning about questions like this, refer to the Instructions section on page 2!) (Q06) In those sentences of texts (E08)–(E11) where P was used, could it have been replaced with a different grammatical category? (Q07) Is P used in connected narratives in contemporary fiction written in L? Has the situation changed in this respect during the last hundred years or so? If the gram tentatively labelled P was used in all or most of the sentences of texts (E08)–(E11) and could not have been replaced with another gram without making the texts less typical as narratives, P may possibly not be a perfect at all but, say, a general past tense. In this case, part of the remaining questions may not be applicable.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

808

808

Appendices

Boundedness tests (Q08) In connected texts (E08)–(E11) above, was there any difference in the grammaticality of P in the first sentences (with the verb WALK), as opposed to the remaining sentences of each text? (Q09) If P was not used in (E12)–(E13), could it have been used? If it was used, would it have had alternatives? Are there any differences between the SIT sentence and the FALL sentence in either example?

Temporal adverbials (Q10) If P was not used in (E14)–(E26), could it have been used? How would it have changed the meaning or other properties of the sentences? (Q11) If P was used in (E27), was it obligatory? What could it have been replaced with?

Possessive & resultative (Q12) Is there in L any construction, as distinct from P, that combines a possessive construction with a transitive verb, such as the English “I have fifty pages written (by now)”, i.e. ‘I have got fifty pages which I have written’? If it was not used in (E42)–(E46), could it have been used in any of them?

Past & present (Q13) The “temporal frame” of (E48)–(E50) combines the past and the present: I lived in this town, and still do. Are there different translations of these sentences that would differently emphasize the “past” and the “present” aspect? If yes, are they equally natural? How would the temporal “mirror image” of (E50) be translated, i.e.: I LIVE here till the end of my life. (Q14) If P was used in (E47) or in the SEE clause of (E54), was it obligatory?

Evidentiality (Q15) Examples (E59)–(E74) express various kinds of evidentially indirect information, i.e., events which the speakers have not seen themselves but only know them from other people’s accounts or from the results these events have left. Describe how such indirect evidentially is grammatically marked

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

809

The Perfect Questionnaire

809

in L – if it is. Are there some grammatical means for this that are not covered by the examples mentioned? (Q16) Are there special grammatical categories or rules in L that obligatorily apply to reported speech (oratio obliqua), such as non-indicative moods or rules for consecutio temporum? Are they main-clause or subordinate-clause phenomena? (Q17) Is P used in newspapers reports? If yes, would it be used in a news story reporting on (i) a major event in the capital of the country, with lots of participants and eye-witnesses, (ii) a minor incident in a small village, with only few eye-witnesses?

P in the past (Q18) Do (E75)–(E79) and (E81), or some of them, contain any gram that could be considered a past counterpart of P – at least semantically or functionally, if not formally? If yes, apply questions (Q01)–(Q04) to this gram as well!

P in the future (Q19) Do (E82)–(E88), or some of them, contain any gram that could be considered a future counterpart of P – at least semantically functionally, if not formally? If yes, apply questions (Q01)–(Q04) to this gram as well!

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

810

3 Questionnaire on the Progressive Aspect Part I Tentative definition S01- /Somebody on the phone wants to know about Ann; the answer is: – Ann is / She WORK [right now] near me S02- A: What does Ann do every Saturday morning? B: She CLEAN THE HOUSE / READ S03- [Last night at 8 o’ clock,] when John came, Ann still WORK S04- Last year we [usually] CLEAN THE HOUSE on Saturdays [now we do it on Thursdays] S05- Last summer, John VISIT us three times

Transitive verbs and valency === /Somebody on the phone wants to know about John; the answer is: – John is / near me He CLEAN a gun S06He READ a newspaper S07He BUILD a shelter [for the sheep] S08He SING a song S09He GIVE a present to his sister S10He TELL a story to his sister S11-

Object incorporation === /Somebody on the phone wants to know about Ann; the answer is: – Ann is / near me She PEEL potatoes S12She PEEL the potatoes S13She PEEL 3 kilos of potatoes S13She PEEL all the potatoes S14She CHASE chickens [out of the house] S16She CHASE two chickens [out of the house] S17She WRITE her thesis [I think she will never finish] S18-

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

811

Questionnaire on the Progressive Aspect

811

Causative verbs === /Somebody on the phone wants to know about Ted; the answer is:/ S19- She HAVE his hair CUT [right now] S20- He MAKE the child EAT the porridge [right now]

Motion verbs === /Somebody on the phone wants to know about Julie; the answer is/ S21- She GO OUT [right now; do you want me to hold her back?] S22- Well, [right now] she FLY to New York / Moscow [you can call her tomorrow at her hotel]

Phasal verbs === /Somebody on the phone wants to know about Fred; the answer is: – Fred is near me, / he BEGIN to peel the potatoes [right now] S23he FINISH repairing the lamp [right now] S24he BEGIN a language drill [right now] S25he FINISH a language drill [right now] S26he CONTINUE his story-telling [right now] S27-

Postural verbs S28- /Somebody on the phone wants to know about Mary; the answer is/ [Right now] She SIT in the kitchen/yard S29- A: I need my blue shirt right now; where is it? B: It HANG on the nail

Non-durative verbs S30- I took the photo exactly while John THROW the stone against the window S31- [Right now] The climber REACH the top of the mountain S32- The pardon arrived just while the captain GIVE the sign to the firing squad

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

812

812

Appendices

Non-intentional verbs S33- [Look, there he goes again!] George inadvertently INSULT his neighbour with his silly questions. [He really cannot understand the situation]. S34- [Incredible! Listen to him now! With his words] Philip unconsciously ADMITS the guilt

Non-agentive, intransitive processes S35S36S37S38-

[Look at John, on the sofa!] He DREAM of his girlfriend [Look out of the window now!] The sun SHINE The water BOIL [shall I make tea?] [Look, what a shame!] The apples ROT on the tree

Stative verbs S39- [Now, unexpectedly,] Peter KNOW the answer S40- [Now, unexpectedly,] Tess LIKE the music S41- The mountain SURROUND the plain

Copular verbs S42- [What a wonderful present!] You BE very KIND, now! S43- /John has made a negative comment on Ann’s hair-style; Ann says with a tone of surprise/ You BE RUDE this evening

Remoteness/invisibility === S44S45=== S46S47-

/on the phone/ A: Is Ann with you right now? B: No, she DANCE [in the next room] B: No, she PLAY CARDS [in the next room] /on the phone/ A: Is Ann at home right now? – B: No, she SHOP. She left one hour ago she PLAY CARDS in the club [as usual]

Durative adverbials S48- [Yesterday, during my sleep] Ann PLAY for 2 hours all by herself

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

813

Questionnaire on the Progressive Aspect

813

S49- [During the whole time of the class/prayer] Ann TALK to her neighbour [in fact, she carried on even afterwards] S50- [During the whole time of the class/prayer] Ann TALK to her neighbour [but as soon as that was over, she suddenly became very silent] S51- [Moment by moment] The policeman TAKE NOTES of what the speaker said S52- He continually FORGET people’s names

Graduality adverbs S53- The level of the water INCREASE slightly since yesterday S54- [When I arrived] the situation already IMPROVE little by little S55- [When I arrived] the snow COVER gradually the land

Imminential meaning S56- [Hurry up!] The train LEAVE S57- The old man DIE [but finally they found the right medicine]

Temporariness S58- Ann STAND in the doorway, [right now] S59- The statue STAND in the garden [for the summer] S60- [Think! While we are here talking about our matters] the earth TURN around the sun S61- The boss TYPE his own letters, while the secretary is ill

Backgrounding S62- It was a bright summer day. The bees HUM, the birds SING, the cows GRAZE in the greenfield. Suddenly, the earth opened and the devil came out

Habitual (and quasi-habitual) S63- At that time, he GO to dance every Saturday. S64- If you insist on calling me Fred, you INTRUDE in my private life. S65- As soon as you start asking what is the use of education, you ABANDON the basic assumptions of any true culture

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

814

814

Appendices

Futurate meaning S66S67S68S69-

Ann LEAVE tomorrow Ann LEAVE in a minute John GET MARRIED tomorrow. Who BE his witness at the wedding? Who BE captain of the team tomorrow?

Sequence and coordination of events S70=== S71S72-

Yesterday, while Ann READ in her room, Martin PLAY in the courtyard /What did Martin do yesterday evening?/ He STUDY, he READ the paper, he EAT, and then he GO to bed He STUDY from 2 to 6, he READ the paper from 6 to 7, he EAT from 7 to 8, and then he GO to bed

Imperative S73- [For goodness sake,] WORK when the boss comes back! S74- /Mother to daughter, whom she wants to punish/ You NOT GO to that party!

Passive S75- [Come in, please!] The meal BE SERVED [right now]

Negation S76- The boss was angry, because John not WORK when he came in S77- [Let’s go out,] it not RAIN now S78- [This is disgusting;] it is 8.30 and the train not yet LEAVE

Modal verbs S79- Tom must FEED the animals [I guess] S80- Ann should TEACH now [I guess]

Temporal location of the event S81- [I am so tired:] I BAKE all day since I got up this morning

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

815

Questionnaire on the Progressive Aspect

815

S82- When John came home yesterday, he was very tired because he WORK hard all week S83- If you come at 8 o’ clock, I still COOK [Come a little later, please]

Part II This part contains purely theoretical questions. They should be answered after completing the first part. Please remember that when the notion Progressive is mentioned, this refers specifically to the verb form(s) which is/are used in sentences 01 and 03.

1. General characterization ab-

cd-

e-

fg-

hij-

Which devices are used to express the Progressive, as typically defined in examples 01 and 03? Are there synthetic devices besides periphrastic ones? If there is more than one device, describe them morphologically, and point to the possible differences (e.g.: Italian has “stare a  Infinitive” and “andare/venire  Gerund”, besides “stare + Gerund”, which is the most typical Progressive periphrasis). Do the ‘auxiliary’ verbs used in the various periphrastic expressions retain their usual meaning, or are they fully grammaticalized? If postural verbs are used to express the Progressive, are they all possible? Do they keep some of their original meaning? (e.g., SIT in 07, STAND in 37, LIE in 35, HANG in 38). Is there a special ‘motion’ Progressive? i.e., is a verb like GO or MOVE possible or necessary if the action is carried out while moving around? (e.g., in 44). Is the ‘motion’ Progressive possible with goal-oriented motion, or only in the sense of “moving around”? Do the alternative devices cover the same meanings? Are they used with the same type of verbs? (see also below, point 2,b) To which extent is an explicit marking of progressivity obligatory? Is it always obligatory, or only in certain contexts? (e.g., the ‘incidential’ scheme, cf. 03, 30, 32, 76, 82) Are there relevant differences with respect to style and register? How frequent is the construction in actual usage (written and spoken)? Is there an overlap between Perfect and Progressive markers? E.g., has the form used with atelic verbs in 01 a perfect interpretation with telic verbs, as in 02? Has a perfect been used in any of the sentences? If so, is this due to a special character of those verbs in your language (e.g., 28, 59)?

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

816

816

Appendices

2. Aspectual and actional (i.e., Aktionsart) properties a-

bb-

c-

How do the various devices integrate in the grammar, from the point of view of the basic aspectual oppositions? (obviously, perfective vs. imperfective; but make explicit whatever system of oppositions you think is at work here. Cf. again sentences 01–05). Is it possible to combine the Progressive with a habitual (or quasi-habitual) meaning? (cf. 63–65) How do these devices integrate in the grammar, from the point of view of the basic meaning of the predicate? Specifically, is the Progressive possible with: - intransitive activities? (cf. 01; consider also verbs such as boil, blossom, walk, cry etc.) - transitive verbs? (cf. 06–11); does the definite / indefinite nature of the object matter? Is object incorporation possible or necessary, and under what conditions? (cf. 12–18) – causative verbs? (cf. 19–20) – motion verbs? (cf. 21–22) – phasal verbs? (cf. 23–27) – ‘postural verbs’? (cf. 28–29) – non-durative verbs? (cf. 21, 30–32); do they imply imminentiality? – non-intentional verbs? (cf. 33–34) – non-agentive, intransitive processes? (cf. 35–38) – stative verbs? (cf. 39–43). ‘Locomotive’ meaning, and related matters. In some languages, like German, it is likely that specific periphrases are used in the following contexts. Needless to say, the situations presented here are not Progressive in the strict sense; it is interesting to see what happens in your language: i– He is a terrible person: he BRAG all the time (about his merits) [herumlaufen und angeben] ii– Ann INTERVIEW people (from house to house), now

3. Morphological properties a-

bc-

Is the system of oppositions the same in all tenses? (cf. Present vs. Past in English. For instance, is there an opposition in the Present between a Progressive and a (basically) generic interpretation, depending on the morphology used?) Do all finite tenses combine with the specific devices that you have specified under 1 above? Do all non-finite tenses combine with the Progressive? (cf. the Infinitive in 79–80) (Incidentally: can the Infinitive in these sentence have other meanings than the epistemic one?).

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

817

Questionnaire on the Progressive Aspect

de-

fg-

h-

817

Is the imperative allowed with the Progressive? (cf. sentence 73–74) Is the passive allowed with the Progressive? (cf. sentence 75; if the answer is yes, to what extent is this possible? can you say something like: How long has this bridge been being built?) Is the copula allowed with the Progressive? If the answer is yes, what do such constructions mean? (cf. 42–43) Are all persons of the verb possible with the Progressive? Try to use them with sentence 01 and 03, or any other sentence that you consider prototypical, and point out any peculiarity that you note. Do you know of any other relevant morphological property?

4. Syntactic properties a-

bc-

d-

May the elements of the Progressive (periphrastic) constructions be separated, and by what sort of materials? (e.g.: he is already working; consider also still / presently / precisely / at this moment etc.) May a portion of the construction be marginalized in a parenthetic clause? (e.g.: Jean, en train de se préparer pour le voyage, en fut beaucoup surpris) Is it possible to build constructions which may be interpreted as the coordination of a locative expression and a Progressive, in which the copula and/or the coordinative conjunction are deleted? (e.g.: Ann is inside, (and) working hard; John is in bed, (and) sleeping soundly) Can Progressive and non-Progressive be coordinated? (as in the following sentence, where only one verb must be in the Progressive form): i– John WRITE and Ann PAINT the windows. Are there different ways to coordinate Progressive and non-Progressive forms?

5. Temporal reference abcd-

ef-

Are the Progressive devices possible with any location of the event with respect to speech time? (cf. 03, 81–83) May durative adverbials be used with the Progressive? (cf. 48–52) Does the Progressive imply a ‘temporariness’ meaning? (cf. 58–61) Is any of the Progressive forms restricted to ‘unbroken’ activity? i.e. is the actor necessarily engaged in the activity at the moment referred to? Consider the following sentence: i– John is playing tennis very often, this summer. Is there a merely ‘interpretative’ use of the Progressive? (e.g.: Those on the roll who do not vote will in fact be voting ‘no’) Can the Progressive express a futurate meaning? (cf. 66–69)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

818

818 g-

Appendices

Can the Progressive be used to express a strictly coordinated sequence of events? (cf. 70–72)

6. Any additional remark is welcome. Please add your comments, using as much paper as you wish.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

819

4 List of abbreviations used in interlinear glosses Abbreviation 1 2 3 A.GER A.NONPST ABL ABS ACC ADOKON.PART AL.POT ALL AND ANT AOR AP AT AUX BD BOLAL CMPR COMP COND CONV COP DAT DEF DEM DI.PST DI.PST DIM DU ELAT F

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Explanation 1st person 2nd person 3rd person in Karaim: gerund formed with the suffix -a in Karaim: finite nonpast category based on the gerund formed by the suffix -a ablative absentive accusative in Karaim: participle formed with the suffix -adoKon in Karaim: potentiality category formed with the auxiliary verb al allative andative anterior aorist active participle preposition with meaning ‘at’ auxiliary bounded in Karaim: potentiality category formed with the auxiliary verb bolal comparative complementizer conditional converb copula dative definite demonstrative in Turkish: past category formed with the suffix -di in Karaim: past category formed with the suffix -di in Karaim: diminutive dual elative feminine

820

820

Appendices

FREQ FUT FUTAUX FUTIPFV GAN.PART GEN GER ILL IMP IMPF IN INCH INDEF INESS INF INFM INST INTJ IP.GER IP.GER IPFV IRLS INTR IYOR.NONPST IYORDU.PST LESZ LOC M MAYIN.GER MID MIS.PART NEG NOM NONPST NT OBJ OPT PART PASS PF

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

frequentative future future auxiliary future imperfective in Karaim: participle formed with the suffix -gan genitive gerund illative imperative imperfect preposition with meaning ‘in’ inchoative indefinite inessive infinitive infinitive marker instrumental interjection in Karaim: gerund formed with the suffix -ip in Turkish: gerund formed with the suffix -ip imperfective irrealis intransitive in Turkish: nonpast category formed with the suffix -iyor in Turkish: past category formed with the compunded suffix -iyordu in Hungarian: auxiliary lesz ‘become’ locative masculine in Karaim: gerund formed with the suffix -mayin middle voice in Turkish: participle formed with the suffix -mi¸s negative nominative non-past neuter object optative participle marker passive perfect

821

List of abbreviations used in interlinear glosses

PFF PFV PL POS POSS PP PPA PPP PREP PROG PRS PRT PRTV PST PT Q R.NONPST R.NONPST RDI.PST RECP REL RES RFL SG SIM SP SUBJ TEMP TO TRNSL VN XALA YEZ

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

821

in Greek: perfect formant (non-finite form used in forming the perfect) perfective plural postural verb possessive past participle past participle active past participle passive preposition progressive present particle partitive past preterit question morpheme in Turkish: nonpast category formed with the suffix -r in Karaim: nonpast category formed with the suffix -r in Turkish: past category formed with the compound suffix -rdi in Karaim: reciprocal relative resultative reflexive singular simultaneous in Romance: “Simple Past” subjunctive temporal preposition with meaning ‘to’ translative verbal noun in Karaim: iterative suffix in Karaim: auxiliary verb yez- used in expressions of ‘to be about to do something’

822

5 List of working papers EUROTYP WORKING PAPERS SERIES VI – TENSE AND ASPECT

NO. 1 AUGUST 1992 Pier Marco Bertinetto & Denis Delfitto, Aspect vs. Actionality: Some reasons for keeping them apart Éva Ágnes Csató, On some theoretical and methodological problems of the typological study of tense-aspect categories Eva Hedin, Present with future time reference in Modern Greek Rolf Thieroff, Tense, Aspect, and Mood Categories in European Languages

NO. 2 AUGUST 1992 Future Time Reference in European Languages I Edited by Östen Dahl, Casper de Groot, Hannu Tommola The Marking of Future Time Reference in Bulgarian (Jouko Lindstedt) The Marking of Future Time Reference in Estonian (Hannu Tommola) The Marking of Future Time Reference in German (Rolf Thieroff) The Marking of Future Time Reference in Hungarian (Éva Csató) The Marking of Future Time Reference in Continental Scandinavian (Östen Dahl) The Marking of Future Time Reference in Züritüütsch (Balthasar Bickel)

NO. 3 DECEMBER 1992 Future Time Reference in European Languages II Edited by Östen Dahl, Casper de Groot, Hannu Tommola The Marking of Future Time Reference in Dutch (Casper de Groot) The Marking of Future Time Reference in Finnish (Hannu Tommola) The Marking of Future Time Reference in Modern Greek (Eva Hedin) The Marking of Future Time Reference in Italian (Pier Marco Bertinetto) The Marking of Future Time Reference in Lezgian (Martin Haspelmath) The Marking of Future Time Reference in East Slavic (Östen Dahl) The Marking of Future Time Reference in Turkish (Éva Csató – Lars Johanson)

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

823

List of working papers

823

NO. 4 JANUARY 1993 Vladimir Nedjalkov, Tense-Aspect-Mood Forms in Chukchi

NO. 5 JULY 1994 Three papers on the Perfect Rolf Thieroff: Passives, Perfects, Resultatives, and Statives Östen Dahl and Eva Hedin: Current Relevance and Event Reference Jouko Lindstedt: On the Development of the South Slavonic Perfect

NO. 6 SEPTEMBER 1994 Future Time Reference in European Languages III The Marking of FTR in Fering (Karen Ebert) : The Marking of FTR in French Suzanne Schlyter & Vesta Sandberg Coralia Ditvall, Suzanne Schlyter & Medina Fodor: The Marking of FTR in Romanian Ingrid Hermerén, Suzanne Schlyter & Ingrid Thelin: The Marking of FTR in Spanish Barbara Moltzer: The Expression of Future Time Reference in Germanic Languages

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

824

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

825

Indices

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

826

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

827

Subject index ability, 739 ablaut, 10, 16 absentive, 527, 541–542, 630, 666, 671, 679, 685, 686, 688, 689 absentive gram-type, 658 accomplishment, 657, 677, 686 accusative, 676, 688, 689 achievement, 657, 661, 677, 686 actional, 410, 412, 419 actional character, 771 actional content, 27, 39, 53 actional modifications, 743 actional recategorization, 66 actionality, 189–217, 726, 779 activity, 408, 428 adessive, 661–663 adterminality, 29, 32, 45, 135, 727 adverbial, 9 adverbial clause, 341 adverbials locative, 657, 675, 680, 683, 686 temporal, 73, 195–207, 211–213, 462, 660, 661 agentive verbs, 619 agentivity, 536, 615, 705 Aktionsart, 189 anterior, 276, 366, 442, 469, 732 anterior continuing, 447 anteriority, 34 anteriors young, 467 aorist, 277–282, 441, 445, 450, 454, 456, 459, 460, 484, 731 aoristic drift, 404, 414–421 aoristic use, 469 aprioristic application of distinctions, 45 areal, 18, 19 areal distribution, 636 areal influence, 775 areal phenomena, 3

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

aspect, 189–217 aspectology, 193 aspectotemporal systems, 28 aspectotemporality, 27 aspectual terminality, 28 atelic, 678, 679, 684 attrition, 12 auxiliary, 441, 443, 445, 450, 453, 454, 460, 461, 470 auxiliary drop, 12 background, 678 backgrounding, 660 Balkan area, 330 besprochene Welt, 369 binary, 13 bound, 444, 468 bounded, 210, 664, 690 bounded situation, 445, 454 boundedness, 228, 232 bounders, 18 case object, 655, 676, 683 causal interpretations, 345 causative, 707 change gradual, 660, 662, 668, 676, 682 clitic, 444 cognitive categories, 3 colloquial, 417 combined (portmanteau) marker, 32 communicative motivation, 9 completed events, 16 completive meaning, 464 completive verb, 449 completives, 469 compound tenses, 509 conceptual content, 48 conditional, 737 conditional clauses, 777

828

828

Subject index

future marking in, 329–347 conditional interpretations, 345 consonant gradation, 16 constative, 117 constituent order, 729 content, 8 context-independent meanings, 49 contextual interpretations, 14 continuative, 679, 681, 684 continuous, 666, 677, 684, 685, 688 control, 706 conventional interpretations, 14 converbs, 27 copula future, 351–360 copula drop, 12 corpora, 4, 6, 19 cost, 5 creole items, 60 current relevance, 274, 276, 283, 366–368, 372, 378, 385, 445, 459 current relevance interpretation, 389 de-andative, 14, 319 de-obligative construction Germanic, 319 de-venitive constructions, 320 de-volitive construction, 323 definiteness hierarchy of, 9 defocalization, 99, 129 deictic centre, 680, 697 deixis, 714 deponent verbs, 405 derivational, 18 desemanticized, 628 diachronic developments, 52 diachronic paths of development, 3 diathetic conditions, 112 direction, 702 doughnut grams, 10 duration, 622, 668, 670, 700 durative actionality, 628 durativity, 726

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

dynamicity, 16, 64, 708 egressives, 64 emphasis, 688 emphatic function, 659 emphatic use, 672 epistemic meaning (reading), 453, 669 epistemic modals, 667 erzählte Welt, 369 Europe, 20 evidentiality, 453, 497 evidentials, 374–378 experiential, 243, 388, 448, 450, 454 proper, 369, 370 experiential auxiliary, 466 experientiality, 423, 426 extralinguistic factors, 11 finite verb forms, 27, 731 finitransformatives, 61 focality, 38, 85, 725, 745 degree of, 85, 92, 108 high, 87, 130 higher, 92 low, 87, 130, 132 lower, 92 renewal of, 91 focality oppositions, 89 focalization point, 527 focussing perspective, 625 form, 8 Frame Past, 243 free-floating gram, 46 frequency, 9 function, 10 Functional Grammar, 4 futura exacta, 366 future, 14, 18, 21, 287, 732 anterior, 451, 760 immediate, 660 immediate or definite, 763 imperfective, 336, 452 in the past, 290 perfective, 336, 347, 760 future grams

829

Subject index inflectional, 330 future marker, 759 future time reference, 309, 744 futuroids, 18 gerade as progressive marker, 631 gerund, 454, 461, 740 gram, 7, 47, 441, 454, 455 inflectional, 329 periphrastic, 329 gram family, 7, 317 gram type, 442 core, 14 crosslinguistic, 7 peripheral, 14 grammatic(al)ization, 7, 8, 628 grammatical space, 7, 14 grammaticalization clines, 14 grammaticalization paths, 14 grammaticization incomplete, 628 habeo, 21 habitual, 18, 196, 295–297, 448, 664 habituality restricted, 770 high-focal postterminality, 110 hodiernal, 18 hold constructions, 607 hot news, 423, 429 hot news use of the perfect, 385 hypergram type, 16 imminential, 666, 684, 685, 763 imperative, 537, 674 imperfect, 277–281, 441, 454, 456, 459, 460, 484 imperfective, 11, 14, 27, 44, 192–194, 227, 228, 290–293, 343, 704, 727, 755, 776 “pro Perfective”, 227 “simple denotative” fucntion of, 227 conative, 246 continuative, 244 frequential, 244

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

829

habitual, 244 inceptive, 250 lexical, 771 modal, 253 past, 19 potential, 253 with verbs of communication, 256 imperfective paradox, 230 imperfective verbs, 458 imperfectivization, 145, 469 incidential schema, 43, 534, 552, 564, 661 inclusive, 209, 408, 419, 423, 426, 428, 536 incorporation, 609 independence of relevance considerations, 9 indicative, 10 indirective, 18, 21, 375, 497 indirective postterminals, 121 individual-level, 531, 584 Indo-European inflectional futures, 318 inferential, 375, 453, 725, 736 infinitive, 538 infixes, 16 ingressive, 64, 780 ingressive phase, 773 initio-transformative, 61, 768, 773, 774, 779, 780 intention-based future time reference, 310 intention-based uses, 671 intentions, 309 internal phase structure, 58 interpretative, 536 interpretative uses, 672 intonation, 466 intraterminality, 29, 32, 45, 76, 93, 95, 99, 724, 725 intraterminality oppositions, 78 invariant meanings, 14 Inzidenzschema, see incidential schema irrealis, 430 iterative reading, 457 iteratives, 469 Karaites, 723 kind of action, 30

830

830

Subject index

language acquisition, 3 lexical semantic change, 8 linear, 10 linear successivity, 42 linguistic structuralism, 13 locative-based progressives, 659 loss of autonomy, 10 low-focal postterminality, 114 mas-construction, 661 ma-Infinitive, 664 macrolevel, 18 mas-Infinitive, 660 metaphor, 8 methodology, 4 microlevel, 18 modal categories, 737 modal interpretations, 346 modes of action, 40, 55 modification, 10 momentaneous process, 660 morpheme, 13 motion verbs determinate, 656 narrative context, 9 narrativity, 371–374 negation, 663, 674 imperfectivity and, 238 nomina acti, 662 nomina actionis, 662 non-agentive verbs, 619 non-anterior, 731 non-durative, 408 non-finite, 27, 740 non-set, 42 non-specific readings, 759 nonadterminality, 32, 45, 135, 727 nonbounded past situations, 454 nonfocal, 87, 132, 725, 745 nonfocal intraterminals, 87 nonfocal postterminality, 119 nonintraterminality, 32, 45, 76 nonlinear, 10 nonpossessive, 461

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

nonpostterminality, 32 nonprefixed verbs, 443 nonrestrictive relative clause, 341 nonterminative, 193 nontransformative, 61, 64 nontransformatives non-dynamic, 65 nontransformativization, 70, 72 North European de-volitive construction, 322 North European futureless area, 325 notional categories, 7 notional imperfectivity/perfectivity, 16 O-overlapping, 117 O-relevance, 117 object resultative, 461 obligatorization, 9 obligatory, 14 ontological classification of events, 53 opposition, 13, 16 optative, 738 optional, 14 orientation, 34 participle, 27, 742 active, 767, 774, 779 active in the past, 461 past, 496 passive, 537, 709 dynamic, 463 process, 598 state, 454, 463 passive participles in the past, 461 passive perfect, 463 past, 14, 283 anterior, 451 compound, 403, 444 general, 470 immediate, 446 indefinite, 369–371 inflectional, 444 perfective, 454 recent, 459 universal, 455

831

Subject index past marker, 757 pasts, 9, 19, 21, 282 historical, 172 perfect, 9, 18, 21, 27, 276, 282–285, 336, 339, 365–379, 385, 404, 479, 508 ‘be’, 480, 488, 491 ‘have’, 481, 489 existential, 369–371 experiential, 369–371, 465, 508 future, 336, 451 inclusive, 447 of persistent situation, 447 of recent past, 373 of result, 445, 459 third, 482, 489, 493 perfect auxiliary, 12 perfect marker, 467 Perfect Questionnaire, 365 perfective, 14, 27, 44, 192–194, 228, 290– 293, 655, 704, 755 in subordinate clauses, 776 past, 454 perfective future, 452 perfective subjunctive, 340 perfective verbs, 459 perfectivity, 21, 727 perfectivizing particle, 655 performative verb, 671 peripheral items, 97 periphrastic, 461, 468, 470 periphrastic form, 672 persistence of meaning, 659 persistent result, 423, 424 persistent situation, 452 phase structure, 145 pidgin items, 60 pluperfect, 172, 286–287, 366, 450, 451, 456, 737 plurifocalization, 578 plusquampräteritum, 450 possessive, 461 possessive constructions, 446, 451, 454 possibility, 739 poststatal markers, 57

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

831

postterminality, 29, 32, 102, 124, 127, 129, 724, 725 postterminality oppositions, 104 posttransformative phase, 769 postural, 668 postural verb constructions, 607, 617 postural verbs, 607, 628, 664, 675, 678, 681 postverbs, 55 potentiality, 739 preaspectual, 41, 96 prediction-based future time reference, 310 predictions, 309 preparatory use of a present, 312 prepositional constructions, 607 prepositional verb constructions, 617 present, 336, 338, 339 present anterior, 282–285 presentive, 666, 680, 685, 688 presentness, 142 prestatal markers, 57 prestige, 11 preterite, 277, 282 pretransformative phase, 614 preverbs, 55 primary deictic centre, 34 processes, 16 processive, 666, 677, 684, 685 PROG, 536 progressive, 11, 18, 21, 27, 293–295, 467, 517–550, 559–589, 605, 702–704, 765, 769 combinability with tense, 611 durative, 527–531, 568 focalized, 527–531 motion, 560, 577–581, 615 state, 560 with ‘be’ passives, 613 with directed motion verbs, 615 with phasal verbs, 615 with present perfect, 612 with stative verbs, 614 progressive construction serial, 684 progressive markers, 607

832

832

Subject index

Progressive Questionnaire, 606 prohibitive speech act, 674 propulsive uses, 82 prosodic patterns, 10 prospectivity, 36 prototype theory, 14 prototypes, 10 purposive, 658, 663, 671, 679, 684 purposive prospectivity, 666 quantitative reinterpretation, 70 questionnaire, 4 recategorization, 66, 73 reduction, 10, 12 reduplication, 16 relational adverbs, 467 relational temporal adverb, 469 relevance, 9 remoteness, 19, 698 reportative, 375 residual categories, 10 residual grams, 10, 313 resultative, 21, 117, 366–368, 445, 504, 767 resultative meaning, 460 Romance inflectional future, 318 scenarios, 258 scheduling, 309 scripts, 258 secondary imperfective, 456 secondary orientation, 35 semantic bleaching, 8, 9 semantic categories, 7 serial, 666 serial construction, 661, 664, 668, 681 serialization, 71 simple past, 403, 454, 456 simple tense forms, 460 Slavic aspect, 743 Slavic perfective present, 323 special marker, 32 Sprachbund, 18, 140 stage-level, 584 Standard Average European, 28

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

state, 16, 408, 675, 677, 679, 684 stative, 16, 428, 537, 666, 677–679, 684, 685, 701, 780 stative (possessive) construction, 449 stative construction, 445, 446, 451, 454 stative verbs, 771 stem alternations, 10 structuralist, 13 style colloquial, 629 stylistic variation, 628 subevents, 144 subject resultative, 461 subjunctive perfective, 336, 347 perfective (aorist), 336 subjunctives, 10 subordinate, 12 substantive properties, 13 substantive similarities, 13 supercompound, 286 syllabic harmony, 729 synchronic report, 77 synthetic, 461, 468 T-markers, 45 telic verb, 680 telicity, 192, 408, 620, 709–712 temporal clauses, future marking in, 329–347 temporal distance, 414, 416 temporal interpretations, 36 temporal limit, 624 temporal meanings, 34 temporal reference, 190 temporalization, 137, 313 terminality, 27 terminative, 193 thematic structure aspect and, 235 time reference nonspecific vs. specific, 448 token, 228 token-focussing, 386 totality, 232

833

Subject index transformativity, 59 transformativization, 67 transition, 780 translation questionnaire, 5 “two-way” action imperfective, 243, 398 type, 228 Type-focussing, 386 typological, 23 Ukrainian inflectional imperfective future, 319 umlaut, 10 unbounded, 210 unidirectionality, 11 unintentionality, 671 universal past, 468 unwitnessed past, 497

phasal, 599, 661, 665 verb classes aspectual, 771 verbs gradual completion, 579 inherently intensified, 579 motion, 661, 662, 664, 668, 673, 681, 769, 774 Verlaufsform rheinische, 629 vernacular, 428, 594 viewpoint markers, 32 viewpoint operators, 27 word order, 466 fixed, 8 written texts, 417

verb of becoming, 351–360

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

833

zero prefix, 443

834

Language index Abaza, 124 Adyghe, 94 Afrikaans, 132, 320 Agul, 110 Akhvakh, 91, 110 Akkadian, 82, 124, 129 Alawa, 314 Albanian, 29, 33, 35, 46, 79, 93, 104–107, 115, 120, 122, 123, 126, 133–135, 158, 175, 323, 371, 482, 484, 524, 539, 577 Gheg, 133, 134, 323 Altaic, 121, 542–550 Andi, 92, 95, 179 Arabic, 37, 180, 775 Classical, 191 Modern Standard, 99, 129 Syrian, 776 Archi, 91, 92, 95, 111–113, 124, 162 Armenian, 33, 56, 78, 79, 94, 100, 104, 105, 115, 116, 118, 120–122, 124, 127, 128, 130, 160, 162, 163, 170, 175, 446 East, 38, 62, 64, 83, 110, 113, 128, 153, 155, 161 Arumanian, 124, 377, 418, 431, 482 Asturian, 410, 429 Avar, 92 Aymara, 430 Azerbaijani, 90, 94, 100, 115, 119 Azeri, 180 Bagvalal, 94 Baltic, 21, 56, 68, 72, 111, 121, 123, 125, 140, 239, 318, 375 Baltic Finnic, 127, 179, 371, 441, 448, 655, 657, 658, 665, 669, 670, 675, 680, 681, 684 Baltic-Finnic, 360 Bandjalang, 314

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Basque, 78, 79, 89, 90, 94, 100, 105, 110, 114, 115, 120, 124, 126, 174, 175, 180, 272–273, 295, 325, 522, 524 Belarusan, 132, 135, 137, 324, 445, 447, 448, 454, 457, 723, 727 Bosnian, 374 Breton, 522 Bulgarian, 4, 28, 33, 36, 44, 46, 54, 55, 68, 70, 71, 75, 78–80, 101, 104, 105, 107, 112, 115, 116, 118, 123, 125, 127, 139, 140, 142–144, 146, 148, 149, 151, 156–159, 161, 175, 179, 191, 194, 210, 212, 213, 215–217, 242, 244, 246, 323, 330, 370, 371, 376, 378, 399, 442, 445, 447, 448, 450, 484, 487 Buli, 467 Calabrian, 131, 174, 407, 429 Cassubian, see Kashubian Castilian, 133 Catalan, 57, 78, 94, 95, 105, 115, 117, 118, 131, 133, 175, 279, 318, 374, 416, 422, 426, 428, 431, 521, 523, 560, 563–566, 568, 571, 573, 577, 581, 582, 584–587 Caucasian, 33, 36, 89, 91, 121, 123, 124, 131, 179, 180 Northeast, 91, 92 Central Asian, 121 Chaghatay, 94 Chechen, 55, 79, 91, 94, 105, 107, 115, 120 Chinese, 393 Church Slavonic Old, 326, 441, 460 Chuvash, 69, 79, 94, 101, 106, 115 Creole, 175 Croatian, 101, 132, 312, 323, 374 Czech, 29, 33, 55, 63, 113, 119, 125, 132, 136–138, 142, 176, 246, 296,

835

Language index 444–446, 448, 449, 451–454, 457, 461, 464, 468, 469 Daco-Romanian, 133 Daghestanian, 93, 179 Lezgian group, 180 Danish, 57, 94, 95, 108, 153, 154, 172, 320, 322, 522, 523, 605, 606, 617, 619, 621, 627, 629, 632–634, 717 Dutch, 94, 96, 107, 112, 115, 120, 124, 126, 132, 174, 179, 319, 320, 365, 446, 522, 523, 605, 606, 611, 612, 615–617, 619, 621–623, 625, 627–629, 633, 636, 695, 698, 701, 703, 705, 707, 709–711, 713 Egyptian, 776 English, 9, 10, 21, 35, 38, 41, 42, 45, 46, 50, 51, 54–56, 62–64, 68, 87, 89, 91, 94, 97, 104, 105, 108, 110, 112, 113, 115–117, 120, 121, 126, 127, 131, 132, 135, 149, 152, 153, 162–164, 167, 170, 172, 173, 175, 177, 179, 180, 194, 196–203, 205–208, 210, 212–214, 227, 241, 247, 254, 259, 268–272, 274, 294, 310–314, 319–322, 326, 332, 334, 353, 357, 365, 367, 369–372, 378, 386–391, 393–396, 398, 399, 445, 447, 448, 451, 463, 466, 468, 521, 559–589, 607, 623, 628, 655, 656, 658, 660, 666–668, 670–672, 680, 684, 717, 733, 735 American, 134, 174, 385, 386, 457 British, 103, 385, 446 Old, 94, 124, 125, 538, 624 Esperanto, 105, 125, 164 Estonian, 40, 71, 72, 80, 90, 96, 105, 107, 115, 116, 123, 128, 133, 144, 170, 288, 323–325, 375, 447, 521, 655, 657, 660, 661, 664, 665, 667, 669, 670, 672–674, 676–679, 681, 682, 684 Eurasian languages, 164

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

835

Faroese, 104, 179, 522, 529, 541, 542, 606, 635 Fering, 468, 605, 612, 616, 617, 620, 622, 629, 696, 698, 703, 706, 707, 709, 711, 713 Fering-Öömrang, 606, 611, 628 Finnic, 123, 375, 668 Finnish, 33, 70–72, 89, 96, 97, 106, 112, 115–118, 120, 124, 143, 144, 148, 174, 239, 288, 310, 311, 320, 324, 325, 353, 367, 370, 376, 385, 393, 446, 447, 457, 459, 461, 466, 468, 521, 522, 525, 529, 541, 542, 588, 655–661, 663–667, 669–671, 673, 674, 676, 678, 680–682, 684, 694, 696, 704, 706–708, 710, 713 Old Literary, 658 Finno-Ugrian, 4, 21, 33, 36, 68, 72, 89, 104, 115, 121–125, 131, 173, 174, 180, 325, 326, 357 Eastern, 33, 179 Franco-Provençal, 406, 418 French, 35, 38, 41, 54, 57, 71, 79–84, 87, 90, 95, 101, 106–108, 110, 112–114, 119–121, 126–128, 133–135, 147, 148, 157, 158, 160, 164, 173, 175, 179, 206, 282, 312, 315, 318, 319, 330, 345, 371, 374, 405, 406, 417, 418, 422, 423, 426, 428, 429, 460, 522, 523, 530, 539, 540, 559, 560, 564–566, 571, 575–577, 581, 582, 584–588, 607 Old, 134, 431 French vernaculars, 422 Frisian, 4, 179, 522, 523, 607, 612, 619, 621, 623, 626, 632, 636, 695, 717 North, 606, 614 West, see Frysk, 629 Friulian, 405, 417, 431 Frysk, 606, 615, 617, 619, 622, 629, 636 Gaelic Scots, 99 Gagauz, 79, 91, 94, 724

836

836

Language index

Galician, 133, 134, 164, 406, 410, 428, 577 Georgian, 68, 69, 78, 88, 100, 105, 112, 113, 115, 120, 126, 130, 140, 143, 145, 149–151, 156, 158, 160, 314 Modern, 68 Old, 111, 130 German, 6, 12, 36, 50, 51, 56, 57, 61, 69, 70, 72, 74, 83, 96, 108, 110, 112, 113, 119, 125, 126, 132, 135, 140, 144, 149, 152, 154, 164, 167, 172, 179, 180, 190, 282, 320, 322–324, 354, 355, 360, 371, 374, 376, 447, 450, 468–470, 522, 605–607, 614, 615, 617, 620, 628–631, 633, 636, 695, 701, 705, 707, 708, 710, 713, 744 Low, 606, 633 North, 115, 120 Pennsylvania, 634 Rhineland dialect, 605, 625, 628, 629, 631, 636 South, 38, 51, 110, 119, 132–134, 175 Swiss, see Swiss German Germanic, 4, 9, 10, 21, 33, 89, 96, 115, 131, 152, 162, 174, 179, 194, 277, 278, 317, 319, 321, 322, 325, 326, 351, 357, 360, 373, 530, 559, 585, 605, 625, 626, 628, 655, 657, 664, 666, 673 North, 104 Gheg, see Albanian, Gegh Godoberi, 179 Gothic, 56, 68, 162, 326, 357, 358 Greek, 98, 318, 329–347, 371, 373, 377, 385, 395–397, 404, 419, 482 Attic, 69 Classical, 61, 63, 64, 68, 69, 75, 76, 82, 94, 98, 99, 107, 110–113, 115, 124, 129, 132, 141, 143, 149, 158, 160–162, 171, 178, 179, 228, 238, 241, 244, 246–248, 250, 251, 256, 341, 365, 368 Hellenistic, 99 Homeric, 99 Modern, 33, 35–37, 54, 62–64, 76, 78,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

79, 81–84, 95, 97–99, 104, 105, 107, 112, 115–117, 126, 127, 138, 147, 148, 150, 151, 155–158, 164, 175, 179, 227, 229, 230, 232, 233, 236, 237, 240, 241, 243–247, 250, 252, 256, 280, 292, 323, 446, 448 Hebrew, 314, 723, 724, 727 Hindi, 314 Hittite, 78, 90, 110, 124, 126 Hungarian, 4, 18, 35–37, 56, 57, 63, 68–70, 75, 76, 95, 96, 106, 108, 119, 124, 133, 140–145, 150, 156, 159, 173, 175, 176, 179, 292, 324, 358, 359, 447, 468, 469, 522, 525, 693–695, 697–699, 702–704, 707, 708, 710, 711, 713 Ibero-Romance, 97, 179, 197, 526, 565, 568, 572, 574–577 Icelandic, 57, 89, 92, 104, 112, 120, 127, 153, 154, 179, 294, 325, 326, 448, 522, 523, 588, 605, 606, 611–613, 617, 636, 658, 667, 669–671, 680, 684 Old, 114, 125, 126, 131, 162, 325 Igbo, 540 Indic Old, 124, 132, 368 Indo-Aryan, 318 Indo-European, 36, 60, 79, 94, 110, 113, 114, 121, 122, 124–126, 131, 164, 173, 175, 180, 314, 318, 323, 326, 359, 461 Ingrian, 115 Inuit, 467 Iranian, 33, 69, 89, 90, 95, 104, 113, 115, 131, 179 Irish, 29, 36, 46, 55, 89, 91, 99, 105, 111, 120, 124, 125, 152, 170, 171, 179, 296 Istro-Romanian, 133, 418 Italian, 4, 14, 57, 78, 80, 83–85, 90, 92, 94, 105, 107, 115, 119, 125, 126, 128, 133, 150, 155, 156, 158, 160, 175, 179, 194, 196–199, 201–208, 210,

837

Language index 318, 319, 365, 371, 422, 426, 428, 432, 433, 446, 447, 521–523, 530, 560–569, 571, 575–581, 583–588, 693–696, 705, 707, 708, 710, 711, 713 North, 133, 173, 406, 417, 422 South, 174, 325 Italian vernaculars Central, 428 Southern, 406 Italo-Albanian, 122 Italo-Croatian, 75, 76, 102, 150, 152, 155, 156, 158, 174, 179 Japanese, 314, 369, 388 Judeo-Spanish, 429, 588 Kabardian, 36, 79, 173 Kalmyk, 35, 55, 56, 63, 69, 79, 81, 89, 91, 94, 95, 105, 110, 113, 115, 118, 121, 123–125, 131, 134, 158, 162, 163, 175, 521, 543–546, 549–550 Karachai, 41, 56, 79, 93, 105, 110, 120, 121, 161, 175, 521, 546–550 Karachai-Balkar, 723 Karaim, 113, 723–749 Crimean, 731 Eastern, 728 Karata, 179 Karelian, 521, 658 Kartvelian, 56, 68, 70, 79, 100, 104, 122, 140, 149, 151 Kashubian, 469, 470 Khvarshi, 93, 94 Kipchak-Turkic, 723 Kirmanji, 90, 95, 107, 113, 115, 120, 124, 126, 175 Komi, 33, 375, 495–497, 499–502, 504, 506, 508, 509, 511 Komi-Zyryan, 79, 105, 107, 115, 121 Kumyk, 79 Kurdish, 122, 314 Ladin, 405, 417

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

837

Latin, 33, 45, 68, 82, 94–97, 107, 114, 121, 124, 125, 127, 131, 132, 173, 174, 179, 180, 318, 365, 366, 373, 404, 405, 538, 561–563 Vulgar, 432 Latvian, 18, 68, 105, 118, 123–125, 140, 144, 174, 180, 314, 318, 375 Laz, 156 Lechitic, 469 Leonese, 410 Lezgian, 92, 95, 100, 101, 105, 106, 112, 120, 121, 124, 125, 153, 175, 544 Lithuanian, 18, 33, 55–57, 61, 68–72, 75, 91, 96, 97, 105, 110, 112, 113, 116, 123, 128, 140, 142–145, 149–151, 154, 156, 159, 295, 296, 318, 375, 723, 727, 744, 745 Livonian, 124, 375, 521, 658 Lombard, 405 Macedonian, 101, 104, 112, 113, 115, 122, 125, 126, 129, 131, 151, 156, 179, 210, 323, 330, 365, 371, 374, 377, 445, 446, 479 North, 120 Southwestern, 123 Maltese, 35–37, 56, 63, 82, 89, 91, 92, 94, 99, 106, 119, 125, 127, 129, 133, 156, 158, 161–163, 173, 179, 180, 281, 294, 753–781 Gozo dialect, 762 Mari, 170, 180 High (Hill), 105, 115, 175 Low (Meadow), 79, 105, 107, 115, 125 Megleno-Romanian, 418 Moghol¯ı, 113 Mongolian, 36, 55, 69, 72, 113, 122–124, 131, 173, 179 Mooring, 606 Mordvin, 79, 81 Nakh, 91 Nakh-Dagestanian, 91, 92 Nenets, 60 Nogai, 41, 63, 93, 107, 115, 723

838

838

Language index

North Frisian, 615, 617, 619, 628, 633, 636 Norwegian, 33, 38, 51, 55, 95, 104, 106, 162, 174, 312, 320, 322, 385, 446–448, 523, 606, 634, 636, 694–696, 703, 706, 707, 709, 717 Nynorsk, 634, 693 Nuristan, 123 Oberengadin, 321 Occitan, 133, 318, 374, 405, 406, 416, 418, 428, 577, 581 Occitan Catalan, 422 Oghuzic, 123 Old Slavic, 124 Oneida, 314 Ossetic, 56, 68, 69, 72, 78, 95, 140, 142, 175, 179 Persian, 63, 100, 129, 175 Piedmontese, 405 Polabian, 469 Polish, 33, 36, 119, 125, 132, 136–139, 142, 143, 165, 169, 176, 444, 445, 447, 448, 450–452, 454, 457, 469, 723, 727, 728 Old, 125 Pontic, 121 Portuguese, 82, 89, 90, 94, 95, 107, 115, 118, 125, 127, 131, 133, 134, 153, 155, 164, 175, 287, 318, 319, 371, 406, 408, 409, 418–420, 427, 428, 432, 522, 523, 560, 561, 563, 565, 566, 571–574, 576, 580–588 Brazilian, 521, 561, 584 Proto-Slavic, 141, 178, 374, 441, 470, 480 Punjabi, 540 Quechua, 430 Romance, 4, 33, 45, 54, 79, 89, 90, 92, 94–97, 101, 106, 115, 116, 119, 128, 129, 131, 133, 134, 152, 173, 179, 180, 190, 192, 194, 196, 198, 200– 202, 207–210, 212–214, 279–280,

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

283, 315, 317–319, 325, 403, 539, 559–589, 775 Romanian, 63, 78, 90, 95, 102, 107, 119, 125–127, 131, 133, 151, 175, 176, 179, 287, 323, 330, 405, 406, 417, 418, 422, 428, 432, 560, 565, 566, 581–583, 587 Romansh, 119, 133, 318, 320, 321, 405, 417 Romany, 79, 110, 113, 122, 133, 161 Russian, 18, 32, 33, 35, 40, 44–46, 51, 53–57, 60, 62–64, 68–70, 110, 112, 113, 118, 120, 132, 133, 135–145, 147, 150, 155, 156, 158, 165–168, 172, 176, 180, 189, 209, 210, 212–216, 227, 232, 233, 235, 237, 238, 244, 246, 252, 254, 256, 259, 311, 312, 324, 352, 368, 388, 390, 393–396, 398, 442, 445, 447–449, 451, 452, 454, 457, 459–461, 464–467, 469, 470, 655, 656, 723, 727 dialects, 441, 447, 448, 454, 461, 463, 465 non-standard, 464, 465 North, 125 Old, 131, 371, 459 Rusyn, 132, 445, 447–449, 451, 452, 454, 457, 470 Sami, 79, 91, 104, 124, 175, 521 Sanskrit, 122 Sardinian, 133, 325, 417, 428, 431, 572, 583, 584 Scandinavian, 115, 116, 122, 132, 179, 312, 320, 321, 371, 376, 448, 468, 619, 626, 636, 669, 681 Continental, 670 Semitic, 99, 314 Seneca, 314 Serbian, 57, 108, 132, 134, 155, 174, 175, 312, 323, 374, 419, 481, 487 Serbo-Croatian, 101, 119, 125, 133, 137, 365, 366, 377 Sicilian, 131, 407, 413, 431, 432 Slavic, 4, 12, 18, 21, 36, 45, 55, 56, 68,

839

Language index 69, 71, 72, 75, 76, 89, 95, 99, 108, 119, 120, 125, 126, 130–132, 135, 137–141, 145, 155, 174, 176, 179, 189, 192–194, 209–217, 243, 290–292, 323–325, 359, 370–374, 388, 441, 442, 444, 447, 449, 452, 454, 461, 463, 468, 470, 656, 724, 727, 733, 737, 743 Common, 326 East, 102, 108, 135, 179, 441, 443, 445, 456, 457, 468 Modern, 75 North, 131, 135, 360, 441, 442, 445, 447, 448, 453, 454, 456, 457, 461, 464, 468, 470 Old East, 371, 459, 460 South, 101, 140, 376, 453 West, 102, 108, 140, 179, 441, 456, 457, 468, 469 Western South, 108, 121, 135 Slavonic, Church, see Church Slavonic Slovene, 132, 330, 374 Slovincian, 470 Sorani, 105, 107 Sorbian, 102, 124, 125, 132, 443, 447, 452, 454, 455, 468, 470 Lower, 132, 441, 444, 446–448, 452– 454, 456, 457, 468 Upper, 81, 82, 132, 180, 441, 444, 445, 447, 448, 450–457, 461, 464, 467, 469 Spanish, 78, 85, 90, 94, 98, 105, 106, 110, 115, 117, 131, 133, 152, 175, 194, 197–199, 201, 203, 204, 206–208, 210, 315, 318, 319, 371, 373, 374, 385, 406, 407, 414–416, 422, 423, 426, 428, 430, 432, 521, 523, 560, 563, 565, 566, 568–574, 576–578, 580–582, 584–587 American, 119, 134, 406, 410, 412 Argentinian, 430 Buenos Aires, 413 Canarian, 413, 430 Chilean, 174

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

839

Colombian, 413 Ecuador, 430 European, 174 Latin American, 319, 432, 576, 588 Mexican, 411, 412 Peruvian, 430 Puerto Rico, 413 South American, 174 Surmiran, 321 Svan, 69, 70, 149, 151, 156 Swedish, 6, 12, 36, 57, 72, 96, 103, 104, 110, 114, 120, 122, 126–128, 154, 174, 312, 313, 320, 322, 324, 351– 356, 359, 370, 376, 385, 395, 396, 446, 447, 522–524, 606, 615, 617, 620, 622, 623, 626, 636, 666–668, 672, 680, 681, 683, 684, 695, 698, 706–709, 717 Ostrobothnian, 681 Swiss German, 108, 321, 323, 326 Tabasaran, 91 Talysh, 63, 79, 90, 93, 94, 113, 115, 120, 124, 162, 180 Tatar, 46, 57, 61, 63, 69, 79, 101, 125, 175, 549 Crimean, 723 Tati, 79, 90, 93, 175 Tibetan, 123 Tungus, 173 Turkic, 33, 36, 41, 47, 55, 56, 68, 69, 72, 79, 89, 90, 92–96, 99, 100, 104, 115, 122–125, 127, 129–131, 134, 152, 173, 179, 180, 375, 378, 549, 723, 724, 726, 728, 730, 740, 742, 743 Old, 113, 131, 179 Turkish, 4, 35–38, 40, 47, 48, 51, 54, 62–64, 66, 69, 72, 80–85, 87, 88, 90, 93– 95, 97, 100, 101, 106, 107, 110, 112, 114, 119, 122, 123, 127, 128, 131, 147–153, 155, 157, 158, 160, 162, 166, 173, 174, 179, 295, 324, 375, 447, 484, 502, 724–726, 729, 735, 745

840

840

Language index Modern, 113

Udmurt, 79, 105, 106, 375, 495–497, 499– 502, 504, 506, 508, 509, 511 Ukrainian, 108, 125, 132, 319, 323, 324, 445, 447, 448, 454, 457, 468, 470, 723 Uralic, 121 Urdu, 314, 540 Veneto vernaculars, 405 Vepsian, 105, 107, 125, 521, 658

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Welsh, 99 Wiidinghiird, 606, 622 Yiddish, 56, 95, 108, 119, 126, 132, 154, 175, 322, 323, 524, 606, 632, 636 Yoruba, 467, 540 Züritüütsch, 108, 321, 356, 357, 522, 606, 629, 631, 634, 636 Zan, 156

841

Author index Abelson, Robert P., 258 Agricola, Mikael, 675 Aksu-Koç, Ayhan A., 502 Alarcos Llorach, Emilio, 428, 430, 432 Algeo, James E., 428 Almeida, Manuel, 430 Almqvist, Ingrid, 353 Ambrosini, Riccardo, 431 Anderson, Lloyd B., 116, 379, 633 Andersson, Erik, 605 Anikina, A. B., 237 Antinucci, Francesco, 132 Aquilina, J., 766, 767 Aronson, Howard I., 116, 632 Arsova-Nikoli´c, L., 481, 482 Askedal, John Ole, 634, 635 Bache, Carl, 54, 87, 88, 166 Badia Margarit, Antonio M., 417 Baker, Robin W., 503 Bakker, W. F., 237, 238, 250, 251 Barentsen, Adriaan A., 136 Barnes, Michael P., 635 Barrera-Vidal, Albert, 430 Bartens, Raija, 503, 507 Batalova, R. M., 502 Bayerová, Marcela, 411, 412 Bégin, Claude, 574 Benincà, Paola, 405, 431 Benzing, Johannes, 118, 173, 543, 544 Berschin, Helmut, 412, 416, 430 Bertinetto, Pier Marco, 4, 14, 31, 43, 73, 83, 85, 189, 191, 202, 208, 368, 372, 420, 423, 428, 432, 519, 525, 526, 529, 534, 535, 537–539, 577, 579, 580, 583, 585, 605, 625, 716, 766, 781 Bhatt, Christa, 605 Bickel, Balthasar, 321, 323, 356 Bickerton, Derek, 60 Bläsing, Uwe, 543, 544, 546

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Blansitt, Edward L. Jr., 520, 526, 587, 598 Blasco Ferrer, Eduardo, 431–433 Blass, Friedrich, 341 Bleton, Paul, 406 Blumenthal, Peter, 431 Bondarko, Aleksandr V., 178, 459 Boogaart, Ronny, 605, 627 Borg, Albert J., 162, 755, 760, 763, 765, 767, 770, 771 Bosco, Umberto, 174 Bosque, Ignacio, 577 Bossong, Georg, 133, 371, 431 Breu, Walter, 58, 59, 61–64, 75, 76, 122, 150–152, 155, 158 Breza, Edward, 470 Brianti, Giovanna, 579 Brinkmann, M., 428 Browning, Robert, 132 Brugmann, Karl, 40, 82 Buchholz, Oda, 93, 106, 134 Bulanin, L.L., 463 Burr, Elisabeth, 423 Bustamante, Isabel, 430 Bybee, Joan, 4, 6, 7, 16, 19, 47, 52, 130, 173, 178, 313, 314, 322, 325, 329, 351, 366–369, 373–376, 378, 379, 442, 447, 462, 464, 509, 530, 628, 657, 660, 676, 679 C˘al˘ara¸su, Cristina, 418 Canarache, Ana, 418 Cano González, Ana María, 410 Cardona, Julia, 413 Carlson, Gregory, 584 Carlsson, Lauri, 531 Carruthers, Janice, 406 de Castilho, Ataliba T., 428 Catalán, Diego, 413 Cella, Claudino, 428 Centineo, Giulia, 432 Chafe, Wallace, 375

842

842

Author index

Chung, Sandra, 3 Claudi, Ulrike, 8 Cohen, David, 191 Company, Concepción, 432 Comrie, Bernard, 3, 7, 27, 31, 37, 44, 50, 55, 61, 86–88, 104, 116–118, 122, 134, 136, 141, 153, 179, 180, 189, 227, 233, 234, 319, 334, 366, 369, 374, 375, 447, 470, 540, 657, 667, 679, 680, 757, 764, 776 Corbett, Greville, 470 Cornu, Maurice, 405, 406, 418 Crystal, David, 466 Csató Johanson, Éva, 48, 49, 67, 141, 324, 358, 729 Curtius, Georg, 178 Cypanov, Jevgenij, 502, 503 Dahl, Östen 4, 6, 9, 16, 47, 52, 116, 130, 136, 210, 228, 232–234, 239, 243, 245, 313, 314, 319, 325, 329, 359, 365–370, 372–376, 388, 400, 404, 431, 441, 442, 462, 469, 495, 508, 509, 540, 611, 620, 621, 623, 628, 657, 660, 665, 666, 677, 775, 776 Deeters, Gerhard, 111 Delfitto, Denis, 31, 73, 208, 368, 525, 534, 580, 625 Dietrich, Wolf, 97, 530 Dik, Simon, 4, 69, 87, 133, 147, 149 Doerfer, Gerhard, 131 Donni de Mirande, Nélida Esther, 430 Dowty, David R., 58, 149, 230 Duden, 605 Eberenz, Rolf, 431 Ebert, Karen, 4, 9, 43, 96, 154, 180, 519, 520, 526, 529, 532, 535, 538, 543, 544, 577, 585, 588, 605, 611, 622, 623, 629, 631, 633, 668, 673, 681, 703, 716 Ebneter, Theodor, 321 Engel, Dulcie M., 417 Erben, Johannes, 605 Erelt, Mati, 660

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Faßke, Helmut,132, 450, 453, 456 Fabri, Ray, 759, 760, 764, 767, 771, 773 Fabricius-Hansen, Cathrine, 634 Fiedler, Wilfried, 93, 106, 134 Fleischman, Suzanne, 313, 399, 428, 431 Forsyth, James, 65, 235, 237, 239, 727 Foulet, Lucien, 406, 417 Friedman, Victor, 116, 375, 377, 482, 484, 492 Friedrich, Paul, 3 van der Gaaf, W., 624 v. d. Gabelentz, Georg, 372 Gade, Kirstan, 605, 616 Galton, Herbert, 43, 126 Gambarara, Daniele, 432 Geerts, G., 605, 616, 628 Georgescu, Alexandru, 418 Giacalone Ramat, Anna, 579 Givón, Talmy, 132, 240 Gorškova, K. V., 460 Gougenheim, Georges, 539, 576 Grœnbech, Kaare, 114, 131 Graves, Nina, 122, 129, 131, 377, 446 Greenberg, Joseph H., 366 Grimm, Jakob, 61 De Groot, Casper, 4, 43, 96, 156, 175, 518, 522, 529, 535, 541, 577, 585, 611, 623, 631, 658, 680, 695, 705, 716 Guentchéva, Zlatka, 211 Gvozdanovi´c, Jadranka, 121 Häkkinen, Kaisa, 658, 675 Hünnemeyer, Friederike, 8 Haarmann, Harald, 375, 495 Haase, Martin, 79, 90, 100, 175, 176 Haberland, Hartmut, 612, 621, 622 Haegeman, Liliane, 588 Hakulinen, Auli, 658 Hansen, Aage, 605, 628, 629 Harre, Catherine, 407, 419, 429, 432 Harris, Martin, 106, 129, 406, 407, 410, 414, 418–420

843

Author index Haspelmath, Martin, 93, 95, 100, 101, 106, 112, 153, 504, 544 Haugen, Einar, 122, 376 van der Hauwe, Jo, 605, 629 Havránek, Bohuslav, 359 Hedin, Eva, 4, 54, 80, 83, 116, 117, 150, 243, 314, 338, 366, 368, 369, 372, 396 Heinämäki, Orvokki, 534, 537, 542, 658, 659, 664, 680 Heine, Bernd, 8, 628 Hermann, E., 135 Hermerén, Ingrid, 315 Herrera Santana, Juana, 430 Herzog, Christian, 417 Hetzron, Robert, 525 Hewitt, Steve, 522 Hirtle, W. H., 582, 586 Hockett, Charles F., 13 Hoekstra, Jarich, 629, 633 Holzinger, Daniel, 113 Hopper, Paul J., 8, 11, 43, 372, 373, 659 Hyvärinen, Irma, 659 Hyvönen, Tuula, 659

843

Ikola, Osmo, 376 Inoue, Kyoko, 385, 388, 393 Irmen, Friedrich, 408, 419, 428 Isaˇcenko, Alexander, 460 Ivanˇcev, Svetomir, 158 Ivanov, V.V., 459

Kamp, Hans, 387 Kangasmaa-Minn, Eeva, 659 Kany, Charles E., 413 Karlsson, Fred, 239, 658 Károly, Sándor, 358 Kel’makov, Valentin, 502, 503 Kibrik, Aleksandr E., 111, 112 Kiefer, Ferenc, 525 King, Alan R., 524 Kinnander, Bengt, 376 Klein, 414, 421 Klenin, Wolfgang, 460, 461 Knjazev, Ju.P., 462 Koefoed, H.A., 605 de Kock, Josse, 430 König, 536, 587, 672 Koller, Hermann, 98 Koneski, Blaže 481, 482, 484, 485, 489, 490 Kopeˇcný, Frantisek, 326 Kormušin, Igor V.,130 Koschmieder, Erwin, 169 Kozinceva (Kozintseva), Natalija 83, 118, 124 Kress, Bruno, 605, 611 Krueger, John R.,114, 131 Kubarth, Hugo, 413, 430 Kuryłowicz, Jerzy, 41, 44, 78, 90, 99, 110, 129, 130, 171 Kuttert, Rainer, 416, 430 Kuz’mina, I. B., 469, 504

Jämsä, Tuomo, 659 Jakobson, Roman, 35, 136, 460 Janaš, Pˇetr, 456 Janakiev, Miroslav, 116 Jaxontov, S. E., 367, 368, 462, 504, 768 Jespersen, Otto, 463 Johanson, Lars, 4, 21, 27, 31, 33–35, 37, 38, 41–44, 50, 58, 61, 66, 69–71, 76, 77, 83, 84, 86, 89, 92, 95–97, 100, 113, 119, 123, 127–130, 134, 149, 152, 154, 156, 178, 180, 375, 529, 540, 549, 727, 740 Jones, Michael Allan, 605, 616

La Fauci, Nunzio, 427 Latzel, Sigbert, 372 Leech, Geoffrey, 240, 586 Lehmann, Christian, 8 Leinonen, Marja, 228, 243, 258, 259, 375, 466 Lepschy, Anna Laura, 432 Lepschy, Giulio, 432 Li, Charles N., 392, 393 Lindstedt, Jouko, 4, 21, 75, 148, 158, 211, 216, 242, 243, 370, 399, 404, 419, 423, 426, 453, 508 Lo Duca, Maria Giuseppina, 432

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

844

844

Author index

Lockwood, W. B., 635 Loi Corvetto, Ines, 431 Lope Blanch, Juan M., 410, 412, 429 Lopes, Ana, 82, 89, 119, 153 Loporcaro, Michele, 427, 428, 433 Lutzeier, Peter, 536 Lyons, Sir John, 40, 54, 79, 389 Macaulay, R. K. S., 55 Mackridge, Peter, 228, 253 Magnien, Victor, 341 Malinowski, Arlene C., 429 Manoliu-Manea, Maria, 418 Markkanen, Raija, 659 Marm, Ingvald, 634 Maslov, Jurij Sergeeviˇc 36, 69, 108, 116, 126, 130, 133, 151, 158, 167, 178, 365, 373, 461 Mateica-Igelmann, Michaela, 431 Mathiassen, Terje, 96, 154, 180 McCarus, Ernest, 780 McCawley, James D., 116, 385 McCoard, Robert W., 369, 391, 400 Medina López, Javier, 430 Mehlig, Hans Robert, 259 Meillet, Antoine, 8 Metslang, Helle, 72, 534, 537, 541, 542, 659, 660, 662, 672, 674, 677, 678 Millán Urdiales, José, 410 Miller, Boris V., 162 Miller, Ruth, 132 Miller, Vsevolod F., 142 Mitchell, T.F., 779 Mòcciaro, Antonia G., 414 Moise, Ion, 418 Monville-Burston, Monique, 417 Morales, Montserrat, 431 Moreno de Alba, José G., 411, 412 Mourek, V. E., 140 Nedjalkov, Igor V., 504, 547, 549 Nedjalkov, Vladimir P., 111, 367, 368, 376, 462, 504, 547, 549, 768 Van Ness, Silke, 634 Newton, Brian, 258

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Nichols, Johanna, 19, 375 Niissalo, Nina, 371, 460, 461 Ó Baoill, Dónall P., 91, 111, 171 Olbertz, Hella, 526 Oliveira, Fátima, 82, 89, 119, 153 Pagliuca, William, 6, 7, 313, 314, 322, 325, 351, 366–369, 373–376, 378, 379 Paiva Boléo, Manuel, 405, 410, 419, 428 Pan˘a Boroianu, Ruxandra, 418 Panzer, Baldur, 126 Papazafeiri, Ioanna, 336 Parsons, Terence, 191 Paul, Hermann, 61 Penttilä, Aarni, 659 Perkins, Revere, 6, 7, 313, 314, 322, 325, 351, 366–369, 373–376, 378, 379 Pinkster, Harm, 404, 406 Pisani, Vittore, 404 Plungian, Vladimir A., 376 Pohlenz, Max, 33 Pollak, Wolfgang, 43, 83 Pritsak, Omeljan, 731 Quirk, Randolph, 334 Rallides, Charles, 412 Ramat, Paolo, 404, 579 Rassudova, Olga Petrovna, 235, 242, 243, 252, 254 Referovskaja, E. A., 431 Reichenbach, Hans, 34, 103 Remneva, M. L., 460 Rijksbaron, Albert, 248 Rohlfs, Gerhard, 319, 405, 407 Rojo, Guillermo, 410, 577 Ruijgh, C. J.,251 Ruipérez, Martin Sanchez, 98, 111, 178 Rundgren, Frithiof, 46, 82, 124 Saarinen, 502, 503 S˘adeanu, Floren¸ta, 418 Saettele, Hans, 431 Said, Sally E. S., 410, 412

845

Author index Saltveit, Laurits, 322, 351, 360 Sandberg, Vesta, 315 Santamarina, Antonio, 410 Sasse, Hans, 58, 59, 61–64, 157, 207 Saussure, Ferdinand de, 13, 140 Savi´c, Momˇcilo D. D., 418, 423, 431 Savi´c, Svenka, 374, 377 Šaxmatov, A .A., 463 Schank, Roger C., 258 Scheffer, Johannes, 582 Schlieben-Lange, Brigitte, 405, 416, 428, 577 Schlyter, Suzanne, 315 Schmalstieg, William R., 376 Schmidt, Claudia Maria, 605 Schmitz, John R., 584 Schönig, Claus 549, 560 van Schooneveld, C. G., 460, 461 Schumacher de Peña, Gertrud, 430 Schwenter, Scott A., 134, 374, 386, 399, 414, 416, 428, 429 Schwyzer, Eduard, 247 Seiler, Hansjakob, 31, 63, 75, 82, 84, 98, 99 Serebrennikov, B.A., 81, 121, 500, 510, 658 Serrano, María José, 416, 430 Setälä, E.N., 353, 658 Setatos, Michail, 346 Šewc, Hinc, 141, 456 Short, David, 373 Siadbei, ¸ Ion, 418 Sicking, C. M. J., 238, 251 Siro, Paavo, 658 Skubic, Mitja, 413, 416, 431 Slobin, Dan, 502 Smith, Carlota, 28 Smyth, Herbert Weir, 250 Sobolevskij, A.I., 460 Solarino, Rosaria, 432 Sommerfelt, Alf, 634 Spitzová, Eva, 411, 412 Squartini, Mario, 14, 372, 423, 575, 577, 579, 580, 582, 584, 588 Stankov, Valentin, 151, 159 Stavinohová, Zdeˇnka, 417 Sten, Holger, 83, 408, 410, 428, 429

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

845

Stone, Gerald, 470 Štunová, Anna, 246 Suter, Alfred, 408, 410, 419, 428 Tauli, Valter, 658 Thelin, Nils B, 167, 315 Thieroff, Rolf, 4, 355, 371, 372, 463, 527, 629, 636 Thompson, Sandra A., 392, 393 Thompson, R. M., 392, 393 Timberlake, Alan, 3 Todaeva, Buljaš Chojˇcievna, 544, 546 Tommola, Hannu, 4, 72, 96, 116, 131, 141, 324, 325, 368, 371, 373, 375, 393, 441, 447, 459, 460, 466, 519, 526, 534, 537, 541, 542, 588, 659, 662, 663, 674, 680 Tompa, Josef, 358 Toscano Mateus, Humberto, 430 Trask, R. L., 375 Traugott, Elizabeth, 8, 11, 372, 373, 659 Travaglia, Luiz Carlos, 428 Treder, Jerzy, 470 Trubinskij, V. I., 125, 368, 465, 469, 504, 507 Tuttle, Edward F., 428 Uspenskij, B. A., 460 Ultan, 3 Vanhove, Martine, 756, 758–760, 762–765, 767, 768, 770, 773 Velkovska, Suzana, 482, 483 Vendler, Zeno, 58, 62, 64, 189, 312 Vilkuna, Maria, 375 Vincent, Nigel, 83, 428 Vinogradov, V. V., 258 Visser, Fredericus Th., 624 Vogt, Hans, 69, 70, 79, 143, 149, 150 Vostokov, A. X., 136 Wandruszka, Mario, 428 Waugh, Linda R., 417 Weinrich, Harald, 43, 369, 376 Werner, Edeltraud, 576 Westmoreland, Maurice, 413

846

846

Author index

Weyhe, Eivind, 635 Wiberg, Lars-Erik, 399 Willett, Thomas, 375

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

Xaburgaev, G. A., 460 Zezula, Jaroslav 417

847

Empirical Approaches to Language Typology EUROTYP Mouton de Gruyter
Berlin
New York

Constituent Order in the Languages of Europe Edited by Anna Siewierska Word order surveys, including the detailed treatment of, e.g., the Kartvelian and Dagestanian languages. Parameters of word order variation, such as: flexibility and consistency in word order patterns; relative order of recipient and patient; variation in major constituent order; word order variation in selected SVO and SOV languages. Discourse configurationality. Issues in a performance theory of word order. Presentation of twelve word order variables. Actance et Valence dans les Langues de l’Europe Edited by Jack Feuillet Traitement des problèmes de définition pour les notions de sujet et d’objets direct et indirect, surtout dans les langages périphériques. Réflexion théorique (approfondissement de l’analyse des fonctions syntaxiques centrales) et description minutieuse de tous les phénomènes liés à l’actance dans le langues de l’Europe : définition des actants, étude des divers marquages, constructions impersonelles, oppositions de diathèse, expression de la possession externe, visée communicative, structure attributive. Esquisses typologiques des groupes de langues représentés en Europe. Adverbial Constructions in the Languages of Europe Edited by Johan van der Auwera with Dónall P. Ó Baoill Descriptive study of adverbials, focusing on eight subjects including word level entities (e.g., phasal adverbs, adverbial quantifiers, sentence adverbs), phrasal entities (e.g., equative and similative constructions), and clausal constructions (e.g., concessive conditionals, adverbial subordinators, dependent versus independent adverbial clauses). Structural, semantic, functional, genetic, areal explanations of the observed phenomena. Scrutinization of, and support for, the hypothesis of the ‘Standard Average European’ Sprachbund. Noun Phrase Structure in the Languages of Europe Edited by Frans Plank The morphology and syntax of the noun phrase, in particular the nominal inflectional categories and inflectional systems, the syntax of determination, modification, and conjunction. Comprehensive, genuinely typological view of the full range of crosslinguistic variation within this structural domain.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

848

Empirical Approaches to Language Typology EUROTYP Mouton de Gruyter
Berlin
New York

Word Prosodic Systems in the Languages of Europe Edited by Harry van der Hulst Part 1: Topics. Theoretical background (basic concepts, metrical theory, accent and tone); stress domains, rhythmic organization of compounds and phrases; word prosody and information; phonetic manifestations of stress; diachrony. Part 2: Analytical case studies of word prosodic systems. Survey of word prosodic systems of European languages; word stress and word tone in Germanic languages; stress in Romanic languages; Slavic languages; Baltic languages; Greek word accent; Basque accentuation; Dagestanian languages. Clitics in the Languages of Europe Edited by Henk van Riemsdijk Descriptive part: Area studies and database with information on clitics in various European languages. Theoretical part: Feature articles with peer comments; views on the position of clitics within linguistic theory. Topical part: Clitic clusters and the Wackernagel position; functional categories and the position of clitics; clitics and scrambling; semantic features; phonological aspects. Pragmatic Organization of Discourse in the Languages of Europe Edited by Giuliano Bernini The interaction of discourse structure with sentence shaping. Discussed subjects include: the effect of the diamesic dimension on the principles of discourse organization; the tracking of discourse referents in anaphoric chains; the coding of distinct pragmatic functions, such as topic and focus, in special constructions (e.g., dislocations, clefts); ‘theticity’ and the status of constructions with marked verb–subject order; coding of definiteness and the systems of articles. Subordination and Complementation in the Languages of Europe Edited by Nigel Vincent Part 1: General theoretical and empirical overview; a reconsideration of various formal approaches and the corresponding theoretical constructs. Part 2: Presentation of data, a genetic and areal grouping of investigated languages: Germanic, Romance, Celtic, Slavonic, Caucasic, Finno-Ugric, Balkan. Part 3: Theoretical results, e.g.: grammar of non-finite forms, diachrony of complement systems, word order differences between main and subordinate clauses, functional categories in subordination, theoretical typology of C-systems.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

849

Empirical Approaches to Language Typology Edited by Georg Bossong and Bernard Comrie Mouton de Gruyter
Berlin
New York 1 Paolo Ramat, Linguistic Typology. Translated by A. P. Baldry. 1987. 2 Emma Geniusien˙e, The Typology of Reflexives. 1987. 3 Dieter Wanner, The Development of Romance Clitic Pronouns. From Latin to Old Romance. 1987. 4 Ann M. Cooreman, Transitivity and Discourse Continuity in Chamorro Narratives. 1987. 6 Armin Schwegler, Analyticity and Syntheticity. A Diachronic Perspective with Special Reference to Romance Languages. 1990. 7 Doris L. Payne, The Pragmatics of Word Order. Typological Dimensions of Verb Initial Languages. 1990. 8 Toward a Typology of European Languages. Edited by Johannes Bechert, Giuliano Bernini, and Claude Buridant. 1990. 9 Paradigms. The Economy of Inflection. Edited by Frans Plank. 1991. 10 Meaning and Grammar. Cross-Linguistic Perspectives. Edited by Michel Kefer and Johan van der Auwera. 1992. 11 Franz Müller-Gotama, Grammatical Relations. A Cross-Linguistic Perspective on their Syntax and Semantics. 1994. 12 Tense, Aspect and Action. Empirical and Theoretical Contributions to Language Typology. Edited by Carl Bache, Hans Basbøll and Carl E. Lindberg. 1994. 13 Converbs in Cross-Linguistic Perspective. Structure and Meaning of Adverbial Verb Forms – Adverbial Participles, Gerunds. Edited by Martin Haspelmath and Ekkehard König. 1995. 14 The Grammar of Inalienability. A Typological Perspective on Body Part Terms and the Part-Whole Relation. Edited by Hilary Chappell and William McGregor. 1996. 15 Elke Nowak, Transforming the Images. Ergativity and Transitivity in Inuktitut (Eskimo). 1996. 16 Giuliano Bernini and Paolo Ramat, Negative Sentences in the Languages of Europe. A Typological Approach. 1996. 17 Harrie Wetzer, The Typology of Language Predication. 1996. 18 Bernd Kortmann, Adverbial Subordination. A Typology and History of Adverbial Subordinators Based on European Languages. 1996. 19 Gilbert Lazard, Actancy. 1998.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $

850

20 E UROTYP. Nine volumes. 1997–2001. 20–1 Constituent Order in the Languages of Europe. Edited by Anna Siewierska. 1997. 20–2 Actance et Valence dans les Langues de l’Europe. Edited by Jack Feuillet. 1997. 20–3 Adverbial Constructions in the Languages of Europe. Edited by Johan van der Auwera in collaboration with Dónall P. Ó Baoill. 1998. 20–4 Word Prosodic Systems in the Languages of Europe. Edited by Harry van der Hulst. 1999. 20–5 Clitics in the Languages of Europe. Edited by Henk van Riemsdijk. 1999. 21 Mario Squartini, Verbal Periphrases in Romance. Aspect, Actionality, and Grammaticalization. 1998. 22 Lexical Anaphors and Pronouns in Selected South Asian Languages. A Principled Typology. Edited by Barbara C. Lust, Kashi Wali, James W. Gair, and K. V. Subbarao. 2000. 23 Approaches to the Typology of Word Classes. Edited by Petra M. Vogel and Bernard Comrie. 2000. 24 Evidentials. Turkic, Iranian and Neighbouring Languages. Edited by Lars Johanson and Bo Utas. 2000.

$Id: M_dahl.tex,v 1.36 2000/04/27 16:10:02 eyrich Exp $