Toward a Typology of European Languages 9783110863178, 9783110121087

Citation preview

Toward a Typology of European Languages

Empirical Approaches to Language Typology 8 Editors Georg Bossong Bernard Comrie

Mouton de Gruyter Berlin · New York

Toward a Typology of European Languages

edited by

Johannes Bechert Giuliano Bernini Claude Buridant

Mouton de Gruyter Berlin · New York

1990

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

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

Library of Congress Cataloging in Publication Data

Toward a typology of European languages / edited by Johannes Bechert, Giuliano Bernini, Claude Buridant. p. cm. — (Empirical approaches to language typology ; 8) Includes bibliographical references (p. ) and index. ISBN 0-89925-588-4 (alk. paper) 1. Europe —Languages. 2. Typology (Linguistics) I. Bechert, Johannes, 1931— . II. Bernini, Guiliano. III. Buridant, Claude. IV. Series. P380.T68 1990 415-dc20 90-13307

Deutsche Bibliothek Cataloging in Publication Data

Toward a typology of European languages / ed. by Johannes Bechert ... — Berlin ; New York : Mouton de Gruyter, 1990 (Empirical approaches to language typology ; 8) ISBN 3-11-012108-5 NE: Bechert, Johannes [Hrsg.]; GT

© Copyright 1990 by Walter de Gruyter & Co., D-1000 Berlin 30. 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: Arthur Collignon GmbH, Berlin. — Printing: Druckerei Gerike GmbH, Berlin. — Binding: Dieter Mikolai, Berlin. Printed in Germany.

Contents

Foreword by Bernard Pottier Introduction

vii ix

/. General problems

1

Standard Average European as an exotic language Osten Dahl

3

Typological contrasts between pidgin and creole languages in relation to their European language superstrates Suzanne Romaine

9

Area influence versus typological drift in Western Europe: the case of negation Paolo Ramat and Giuliano Bernini

25

2. Deixis

47

Deixis — a pragmatic universal? Barbara Kryk

49

Possessive adnominal modifiers Gianguido Manzelli

63

3. Morphology

113

The structure of the noun in European languages Johannes Bechert

115

L'infinitif dans les langues romanes et les langues germaniques: essai d'approche typologique Claude Buridant 141 The infinitive in south-east European languages Emanuele Banfi

165

Conjugation of the verb in modern Celtic and Basque: from inflection to periphrasis Malachy McKenna

185

vi

Contents

4. Tense, aspect and modality

193

Types of tense and aspect systems Wolfgang Raible

195

On the coding of sentential modality Dietmar Zaefferer

215

5. Actancy\voice

239

Caracteristiques actancielles de l'"europeen moyen type" Gilbert Lazard

241

Processes and actions: internal agentless impersonals in some European languages Juan Carlos Moreno

255

Questions of the investigation of the complements of adjectives in European languages Laszlo Dezsö

273

6. Complementation and subordination

305

The evolution of certain patterns in subordination in Romance and English Martin Harris

307

Usages normatifs et non normatifs dans les relatives en frangais, en espagnol et en portugais Claire Blanche-Benveniste

317

Adverbial participles, gerunds and absolute constructions in the languages of Europe Ekkehard König and Johan van der Auwera 337 Index of names

357

Index of languages

365

Index of subjects

373

Foreword

The Standing Committee for the Humanities of the European Science Foundation aims at the development of high-level research topics which require the cooperation of scientists from all European countries as well as the frequent collaboration of non-European researchers. Within the field of linguistics, a few projects on typological research have been set up, particularly at the suggestion of P. Ramat. This topic has presently received a world-wide attention as it is so closely connected with research on language universals. In Europe, the UNITYP group — supervised by H. Seller — had, in fact, already provided a noteworthy contribution in the field. After a number of preparatory meetings, the Standing Commitee decided to hold a conference in Rome to give European and non-European linguists an opportunity to talk about the results of their investigations. This volume will thus offer a wide-ranging spectrum of issues on language typology. In addition to studies on syntactic classes and semantic categories, attention is drawn to the relationship between certain types and their geographical diffusion through genetically unrelated languages (areal linguistics), as well as to the notion of "possible types" in the perspective of language universals. On the basis of the conference outcomes and of their personal experience, a committee of linguists who represented the scientific community drew up the EURO-TYP project, which was adopted by the ESF General Assembly in November 1989. Nine research groups have been organized under E. König's supervision. The research will develop from the themes discussed in the present volume and an Advisory Committee, composed of European as well as American and Russian linguists, will follow the development of the project. After a planned period of five years, the project will result in a joint publication, which will doubtless leave a mark on linguistic studies in Europe. We have, in fact, the opportunity to investigate diversified language groups, such as Finno-Ugric, Turkic, Basque, Maltese Arabic or Caucasian languages. A convergence of interests among linguists must be moreover noticed. Comparative grammar, typological research, language universal investigations, the publications of new language descriptions are restoring world languages to the place they deserve after a period when theories based on a few examples detached from social context dominated the research scene. So called "exotic" languages are included in the corpus of the languages inves-

viii

Foreword

tigated. They belong to the schemes established on the basis of languages traditionally studied, but they also bring in their specificities, which advance our knowledge. At present we are therefore witnessing an increasingly cooperative dialogue between specialists of language descriptions and theoreticians, which is resulting in a cross-fertilization beneficial for both parties. Other fields connected with linguistics will profit from this interaction, particularly the field dealing with knowledge representation — semantic or noemic models, inspired by the study of natural languages, but which have assumed the status of tertium comparationis — and the field of machine translation, such as the EUROTRA project. Typology thus appears a "coalescing theme" for present interests focussed on a better knowledge of the mental mechanisms which are revealed by the many world languages, both in a non-random fashion — cf. language universals — and as realizations which can attain great specificity (cf. linguistic variation). Bernard Pottier Former president of the Standing Committee Professor at the University of Paris-Sorbonne

Introduction

This volume contains 17 of the 20 papers read at the Workshop on Typology of Languages in Europe held at the Consiglio Nazionale delle Richerche in Rome, January 7 — 9, 1988. The workshop was organized by the Standing Committee for the Humanities of the European Science Foundation with the aim of exploring the possibilities of launching a Program in Language Typology, following a joint proposal put forward in 1985 by four European scholars since then called "The gang of four": Johannes Bechert (Bremen), Claude Buridant (Strasbourg), Martin Harris (formerly Salford, now Essex), Paolo Ramat (Pavia). This initiative actually resulted in a five-year "Program on Language Typology", started in 1990, which will be the main activity of the Humanities section of the European Science Foundation till 1994. The Program, under the "trademark" EURO-TYP, is directed by Ekkehard König (Berlin, Freie Universität) and will involve almost a hundred scholars from all over Europe (including the USSR) and the USA, organized in nine Thematic Groups ranging from pragmatics to prosody. In this respect the volume actually reflects the first attempts at finding a set of features common to all languages spoken in Europe irrespective of their genetic affiliation. The topics dealt with broadly anticipate the nine themes constituting the main concern within EURO-TYP. It is generally recognized that the languages of Europe represent a particular group — a sort of "Sprachbund" — with striking similarities shared by both Indo-European and non-Indo-European languages and reflecting a prolonged cultural contact within Europe over the centuries. This phenomenon of linguistic convergence opens a wide field of investigation not only in its own right, but also because of the long period for which records are available and the wealth of linguistic scholarship relating to the individual languages of Europe which has been amassed, specially during the last 150 years, but within a linguistic tradition going back to Ancient Greece. The methodological approach adopted by the contributors is indeed typological, linguistic typology aiming at identifying those features which characterize particular groups of languages, related or otherwise, and wherever possible at uncovering the underlying principles which give coherence to surface phenomena that may at first sight appear unrelated. Moreover, the typological principle, mainly comparative-functionalist, does not exclude the variety of points of view and of theoretical positions allowing a very rich and comprehensive analysis of different phenomena.

χ

Introduction

From the papers read in Rome the following perspectives emerge, which are in many ways of interest for general linguistics: — a preoccupation with central theoretical questions, partly from the point of view of formal logic, partly from the perspective of different linguistic traditions; — the importance of an areal (geographical) approach to language typology, centering around the notion of "continuum" in various domains; — the importance of gradation on scales in analysing the basic structures of the languages under consideration; — the significance of the sociolinguistic approach drawing on notions such as language variety, register, dialect, standard vs. non-standard language and the like; — the centrality of diachrony (language history) as a basic ingredient of the typological approach, with "drift" as a key notion. As a first step toward the definition of a cluster of features typical of European languages and the explanation of their emergence in the diachrony either through typological drift or as a result of long-lasting contacts and influences under the same roof of Greek and Latin culture, the volume already presents some features that can be tentatively considered as typical of languages of Europe, and which will be incorporated into a full-scale investigation. The volume is also an attempt at a first instantiation in linguistic terms of the notion of Standard Average European, suggested in 1939 by Benjamin Lee Whorf in order to refer to the cognitive background of Europeans as against that of American aboriginal populations (notably the Hopi) as manifested by their strongly divergent linguistic structures. The EURO-TYP Program will have attained one of its aims if at the end of its course it is possible to see Standard Average European in fact as an exotic language, as the keynote paper of the Workshop puts it, i. e. more objectively and more from the outside than now. Johannes Bechert Giuliano Bernini Claude Buridant

1. General problems

Standard Average European as an exotic language Osten Dahl

To start with, the idea of studying the typology of European languages did not make much sense to me. For many years, I have regarded typology as a method rather than as an area of study in its own right: it is one of several ways to find out about the nature of human language and from this point of view, restricting the domain to a geographical area is a rather strange thing to do. However, I had some second thoughts. One of the greatest problems that the universal study of human language has had to cope with has indeed been the European bias: most linguists have been speakers of European languages, and the other languages that they have known or had access to information about have more often than not been European. As Bell (1978) notes, even linguists who have an ambition to widen their perspective mostly end up with a European or even Indo-European bias in their data bases. This would of course not be so problematic if it were not the case that European languages are much more like each other than languages are in general. I think most of us still have a subconscious view of the "default" language as being something between English, French, German and perhaps Italian — actually, and probably not accidentally, something very much like Esperanto. One could thus turn the problem upside down: in what respects are European languages special and to what extent are the structures you find there "marked"? At this point, I recall the label "Standard Average European" coined by Benjamin Lee Whorf (1956) in his discussions of the relation between language and thinking, and it seemed to me that one might (without of course buying the ideology connected with it) discuss the problems mentioned under this heading, adding "as an exotic language" in order to suggest that it might be a good idea to get out of our ordinary European perspective and think of the 100 odd European languages as if they were, say, the languages spoken in the North West corner of New Guinea. In the rest of the paper, I shall be suggesting a list of phenomena that should be looked at, taking as my point of departure some perhaps not very well-known Swedish research. Natanael Beckman, who was Professor of Swedish in Göteborg between the two World Wars, wrote a fairly long paper in 1934 entitled (in translation) "West European Syntax — Some Innovative Constructions in the Nordic

4

Osten Dahl

and other West European Languages". In this work, Beckman notes that there are a number of common features in the grammars of modern West European languages, including the Germanic languages and in the Romance group at least French, which are not found in the "old" languages (fornspraken). Beckman's perspective is thus historical and entirely internal to Europe; still, some of the phenomena he points to are of interest also to general typology. In particular, he notes two features which appear in roughly the same area: (i) "inversion as an expression for direct interrogative sentences", i. e. the systematic marking of the distinction between declaratives and interrogatives by SVO vs. VSO word order, (ii) the use of formal and "impersonal" subjects such as German es in sentences such as Es kommt ein Mann and Es regnet. An important point that Beckman makes is that in languages where, e. g., 'It is raining' is expressed by a one-word sentence the inversion mechanism cannot be used to distinguish different sentence-types. Beckman's observation is taken up to discussion and generalized in Hammarberg & Viberg (1977), a work focussing on those typological features of Swedish that present particular difficulties for second language learners. Their "subject placeholder constraint" is roughly equivalent to what is nowadays commonly referred to as the "non-Pro-Drop" character of West European languages, i. e. the fact that the subject slot in finite clauses must normally be filled, even in cases where the referent of the subject is contextually given or where the construction is impersonal. They also note, quoting Ultan (1969), that inversion as a device for marking yes-no-questions seems to be rather infrequent outside Europe, and that word order is exploited in Continental Scandinavian and some other Germanic languages also for distinguishing between main and subordinate clauses, so that one might argue that these languages rely to an unusual extent on word order as a syntactic device, in particular for distinguishing clause types. If Beckman's conjecture is right, such a system could only arise in a non-Pro-Drop language, and given Ultan's observations that "YNQ-inversion implies a basic order type in which subject precedes verb", which, in view of the absence of rigid SOV languages with YNQ-inversion, might be strengthened to a postulation of a strong connection between YNQ-inversion and SVO order, the conditions on what a language should look like in order to allow such syntactic exploitation of word order seem rather tight. In addition, one can make a few other observations. In the discussion of the "Pro-Drop Parameter" postulated in Government and Binding theory one usually assumes that there are two main types, one as in English, where there must (practically speaking) always be either a lexical or a pronominal subject in every finite clause, and one as in Latin, where pronominal non-

Standard Average European as an exotic language

5

emphatic subjects can normally be dropped, and in addition some intermediate types. There is, however, a third type, which is not very well represented in Europe but which appears to be quite frequent in various other parts of the world. This is the kind of system where a pronominal subject is required in all sentences, irrespective of whether there is an additional lexical subject or not: in other words, among the counterparts of (1 a — d), only (c — d) are grammatical. (1) a. *Affis. b. John runs. c. He runs. d. John he runs. One problem when trying to find good examples of this type of placeholder constraint is that it is often quite difficult to see whether the morphemes that I am here referring to as "pronominal subjects" are really pronouns or, rather, agreement affixes. One interesting system within our geographical domain, viz. Genoese, is described by Vattuone (1975). He states the constraint as follows (1975: 349): "In Genoese finite 3rd person verbal forms normally must be preceded by a nominative clitic (V for m. sg., 'a' for f. sg., T for pi.). This clitic and the verb either both agree with the relational subject ... or neither agrees." For example, 'The neighbours are coming' would be (2)

/ ve^irj i venu neighbours come

(where i has the dual function of clitic and definite article). It should be clear from this description that the clitic is separate from the morphological agreement between verb and subject. What I want to claim here is that Genoese illustrates a separate type of placeholder constraint, or if we like, another possible value of the Pro-Drop parameter: "Every finite (3rd person) clause must contain a pronominal subject". In this perspective, the correct formulation of the constraint in English or Swedish would be "Every finite clause must contain a pronominal subject unless there is a lexical subject". One argument in favour of treating the constraints in Genoese and Germanic as basically of the same nature is that we find "dummy" subjects in roughly the same kinds of constructions, cf. (3)

U cöve 'It is raining'

(4)

U ve^e na dona 'There comes a woman'

Such dummy subjects are also found, interestingly enough, in another (non-European) language with a constraint similar to that of Genoese, viz.

6

Osten Dahl

Sotho (a Bantu language spoken in Lesotho and South Africa). In Sotho, every finite verb is preceded by a 'subject concord'. In impersonal constructions including, e. g., impersonal passives, the subject concord is go: (5)

Go botsididi ka title 'It is cold outside'

(6)

Go ajewa 'There is being eaten here'

As in Genoese, there is an opposition between 'normal' subject-predicate constructions with agreement and a construction corresponding to /^reinsertion cases in English, cf. (7)

Ba agile mo Basotho The Sothos live here'

(8)

Go agile Basotho mo 'There live Sothos here'

(examples from Northern Sotho [Ziervogel et al. 1979]). In both these languages, then, impersonal subject clitics show up according to rules very similar to those governing the use of dummy subjects in the West European languages. Notice also that Genoese obeys the general tendency for first and second person pronouns to be more easily droppable than third person pronouns (manifested e. g. in Finnish and Russian). A theory that treats the Genoese-Sotho type of placeholder constraint as radically different from that of West European languages thus seems to run the risk of missing significant generalizations. Notice that the Genoese-Sotho type is much less well suited to the systematic use of word order as a syntactic device, since the position of the clitic relative to the verb is constant. It is not surprising that there are no wordorder differences between declaratives and questions in Genoese (the relative order of the lexical subject and the verb is due to differences in information structure or FSP according to Vattuone). I have no statistics, but it seems that of the two kinds of placeholder constraints, the Genoese-Sotho type is more widespread among the languages of the world than the Germanic one. If this is correct, it strengthens the impression we have already formed that the West European languages are somewhat special from the syntactic point of view and that we have something that looks like a conspiracy of factors that favour the use of word order as a device for distinguishing clause types. Further possible accomplices in this conspiracy might be e. g. other "placeholder constraints" than the surface subject constraint, such as the obligatoriness of the finite verb, which seems to be directly related to the presence of an overt copula. I shall briefly mention some other properties of European languages which seem to conspire to give them their particular typological profile.

Standard Average European as an exotic language

7

It has been noted that the West European way of expressing possession, viz. using a transitive verb (have) where the owner is the subject and the possessee the direct object, is a relatively infrequent construction (see, e. g., Clark 1978). Two points can be made here. One concerns the existence of periphrastic Perfect constructions based on a possessive construction like the English / have run. Whereas there is at least one example of such a construction which is based on a different kind of possessive construction (in some North Russian dialects), the majority of the attested examples of possessive-based Perfects are of the have-type and indeed from Europe. We thus see here another example of two presumably connected traits that are quite frequent in Europe but less so in most other parts of the world. The other point relating to the European have construction is a more general one. One recent historical development in West Europe is the decrease of constructions in which the most salient animate participant is not given subject status. I am thinking of examples such as dream' which used to be constructed as 'Me dreams' in earlier dialects of Germanic. As far as I know, the West European languages are quite liberal in allowing non-agentive subjects as compared to many other languages in the world. It seems to me — or to use more idiomatic Standard Average European — I think that the use of a transitive verb in possessive constructions should be seen in the light of this general tendency. To conclude: the grammatical systems of European languages can only be properly understood if looked at in a larger typological perspective. At the same time, stressing the "exotic" features of the European or West European languages may have a positive effect on the development of linguistic theory in general in at least two ways: first, it is a necessary antidote to the pervasive European bias in all branches of linguistic research, second, it may contribute to our understanding of the ways in which "marked" grammatical constructions develop.

References Adams, Marianne 1987 "From Old French to the theory of pro-drop", Natural Language and Linguistic Theory 5: 1-32. Beckman, Natanael 1934 Västeuropeisk syntax. Nägra nybildningar i nordiska och andra västeuropeiska spräk. [West European Syntax — Some Innovative Constructions in the Nordic and other West European Languages.] Göteborg: Göteborgs högskolas ärsskrift.

8

Osten Dahl

Bell, Alan 1978 Clark, Eve 1978

"Language samples", in: Joseph H. Greenberg (ed.), Volume l, 123 — 146. "Locationals: Existential, locative, and possessive constructions", in: Joseph H. Greenberg (ed.), Volume 4, 85-126.

Greenberg, Joseph H. (ed.) 1978 Universal! of Human Language. Volumes 1—4. Stanford: Stanford University Press. Hammarberg, Björn — Ake Viberg 1977 "The place-holder constraint, language typology, and the teaching of Swedish to immigrants", Studia Linguistica 31: 106 — 131. Ultan, Russell 1978 "Some general characteristics of interrogative systems", in: Joseph H. Greenberg (ed.), Volume 4, 211-248. Vattuone, Bart 1975 "Notes on Genoese syntax. Kernel "VOS" strings and theme-rheme structures", Stttdi Italiani di Linguistica Teorica ed Applicata 4: 335 — 378. Whorf, Benjamin Lee 1956 Language, Thought, and Reality. New York: MIT Press/Wiley. Ziervogel, D. — D. P. Lombard — P. K. Mokgogong (eds.) 1979 A Handbook of the Northern Sotho Language. Pretoria: J. L. van Schaik.

Typological contrasts between pidgin and creole languages in relation to their European language superstrates Suzanne Romaine

Introduction I will argue here that the question of the typological affiliation of pidgins and Creoles in relation to their European language superstrates is of significance for the study of the typology of the European languages. In section 1 I will examine a number of features of pidgin and creole grammars in order to establish that their origin is problematic. That is, they may be due to substrate or superstrate influence or they may be attributable to universal constraints. This has some general implications for the study of typology which I will discuss in section 2. The question of the genetic and typological relationship between pidgins and Creoles and their lexifier languages has been a long-standing preoccupation of creolists. Hall (1966: 58) maintained that "all varieties of Pidgin English and Creoles that have grown out of them have an underlying identity of structure with English, and similarly for the French-based, Spanish-based pidgins and Creoles ... they still maintain a basically Indo-European pattern". He postulated a life cycle beginning with the spontaneous generation of a pidgin followed by its evolution to a creole. Baker and Corne (1982: 5), on the other hand, say that the "view that all Indian Ocean Creole French languages belong to the same semantactic tradition results on the one hand from a Eurocentric analysis of the facts, and on the other from a concept of language which appears to confuse etymology with function". Many creolists would agree with Valdman's view of Haitian Creole when he says (1986: 520) that it "can in no way be considered genetically related to its base language", i. e. French. However, the question of what the typological relationship is between a creole and its base language is still unanswered. Hall again took a conservative view on this matter. He was one of the strongest proponents of the family tree model in accounting for the historical relationships among various pidgins and Creoles. He maintained that in existing pidgins and Creoles the contribution of the superstrate was

10

Susanne Romaine

always greater than that of the various substrata. Thus, he concluded (1966: 118) that the ancestral form of any given group of related pidgins and Creoles could be reconstructed and that this reconstructed "proto-pidgin" would show a reasonable correspondence to certain features of the source language. Creolists such as Bickerton (1981) have emphasized the discontinuity between a newly emergent creole and the antecedent pidgin. This is based largely on the fact that Creoles share a great many semantactic similarities which cannot be traced to their pidgin ancestors. In Bickerton's view, these newly crated features must be the result of innate language universals contained in what he calls the bioprogram. In the kind of pidginization and creolization discussed by Bickerton, the links between lexifier language and pidgin/creole are severed early and influence from the lexifier is seen to be limited to the lexical inventory. However, in virtually all pidgin/creole languages influence from the lexifier language persists, often throughout the linguistic development of the descendant languages.

1. Some features of pidgin and creole grammars Bickerton (1981: Ch. 2) has identified twelve features which he believes to characterize creole grammars (see also Taylor 1971: 294 for twelve features, some of which are different): i. ii. iii. iv. v. vi. vii. viii. ix. x. xi. xii.

movement rules articles tense-modality-aspect systems realized and unrealized complements relativization and subject-copying negation existential and purposive copula adjectives as verbs questions question words passive equivalents

Mühlhäusler (1986 a: Ch. 5) identifies 9 features which are characteristic of pidgins: i. SVO word order ii. invariant word order for questions and statements

Typological contrasts between pidgin and Creole languages

iii. iv. v. vi. vii. viii. ix.

\\

sentence-external qualifiers lack of number in nouns pronoun systems prepositions lack of derivational depth bimorphemic wh-questions words anaphoric pronouns.

Of these, six are considered to be criterial features of Creoles by either Bickerton, Markey (1982) or others. M hlh usler (1986a) cites χ and xi (i.e. questions and question words) in Bickerton's list as characteristic features of pidgins which apparently had no input model. In other words, much the same solutions tend to recur wherever pidginization occurs regardless of the lexifier and substratum languages involved. Their presence in Creoles illustrates the problem in drawing a sharp boundary between pidgins and Creoles and also indicates that there can be some continuity in development from pidgin to Creole. That is, Creoles may have inherited these features from a prior pidgin stage, rather than re-invented them independently.

1.1. Questions and question words If we look at questions and question words in pidgins and Creoles, we can see that the problem is even more complicated than that of deciding whether these features are independent innovations or inheritances from a prior pidgin stage. We cannot rule out superstrate and substrate influences. With regard to questions, Bickerton states (1981: 70) that no Creole shows any difference in syntactic structure between questions and statements. If a Creole has special question particles, they are sentence-final and optional. Thus, in Guyanese Creole sentence (1) is not formally distinguishable as an interrogative or a declarative.1 The difference between the two sentence types is marked by intonation. (1)

/' bat dl eg dem.

Ήε bought the eggs/Did he buy the eggs?' This becomes an even more interesting and salient typological feature of Creoles when taken in conjunction with the fact that most Creoles tend to have SVO word order. Greenberg (1963: 81) proposes an implicational universal to the effect that if a language has sentence-external question particles, these tend to occur initially in prepositional languages (e. g. SVO),

12

Susanne Romaine

but sentence-finally in postpositional (e. g. SOV) languages. The Creole prototype thus violates this expectation.2 I will turn now to question words. In many Creoles, question words are bimorphemic. The first morpheme is generally derived from a superstrate word, e. g. Guyanese Creole wisaid 'where' (< 'which side'), and similarly Haitian Creole, ki kote 'where' (< qui cote 'which side'). Other forms include Cameroons Creole wetin 'what' (< 'what thing'), Guyanese Creole wa mek 'why' (< 'what makes'), and Haitian Creole lako^ ki 'why' (< la cause que 'the reason that'), ki fer 'why' (< qui faire 'what makes'). The same kinds of structures can be found in many pidgins. Tok Pisin, for example, has wanem 'what/which' (< 'what name') and husat 'who' (< 'who's that'). Pidgins based on African languages also have similar forms, e. g. Kenya Pidgin saa gani 'when' (< 'hour which'), Swahili sababu gani 'why' (< 'reason which'), and Fanagalo ipi skati 'when' (< 'where time') (see Heine 1973). Taylor (1977: 171) notes that these bimorphemic constructions are found in many African languages and thus suggests an African origin for them. However, in experimentally created pidgins (see Schumann 1986) bimorphemic expressions emerged spontaneously for question words too, where African substratum cannot have been a factor. In Farsi pidgin, for example, ehe so'αϊ 'why' (< 'what ask') and ehe vaqt 'when' (< 'what time') were created. A number of these bimorphemic constructions in pidgins and Creoles may have been taken over from the related superstrate languages. English, for example, has what time 'when', what way 'where/how'. So it would not be correct to say that there was no model in the superstrate for these constructions in Creoles. Similarly, in the case of sentence-final question words, while it is true that there were no models in some of the superstrate languages, it is not true for all. It is well within the norms of colloquial French and English for a question word to occur at the end of the sentence, as examples (2) and (3) show: (2)

Les tres longues dents du loup qui s'accrochent dans les habits dans la peau, on les appelle comment? 'These very long teeth of the wolf, that get hold of clothes and skin, they are called what?'

(3)

You're going where?

It is of course a separate issue that (2) is condemned by schoolteachers (see, e. g., Dannequin 1977: 76, from whom example [2] is taken), and that neither (2) nor (3) would be found in the written language. Posner (1983:

Typological contrasts between pidgin and Creole languages

13

201) also notes the use of wb in situ questions of this type in colloquial French, but suspects they are of recent origin. It is also well within the spoken norms of all the European languages to use prosody in conjunction with ordinary declarative word order for questions. Thus, in the case of question words and questions there are three sources for the similarities between Creoles (and also pidgins): superstrate, substrate and universals. Posner (1985: 170) makes explicit at least one of the points of my examples: namely, that non-standard or popular colloquial varieties of languages show more of the so-called bioprogram features in grammar than do their standards (on the closeness between joual and creole see Wittmann 1973, and also Chaudenson 1979: 101 —102 on franfais avance and creole). Given the nature of contact between indigenous peoples and their European colonizers, it is also likely that it was this kind of language which formed the input to pidgins and Creoles. This is all too often overlooked in comparisons between pidgins and Creoles and their superstrates. Typological comparisons focus their attention on standard written varieties. Not all varieties of a language, however, share the same typological or parametric affiliations. Thus, languages like Chinese, Japanese etc. are defined as having wh- in situ for simple interrogatives, while in English such structures are allegedly ruled out (see, e.g. Chomsky 1986: 53). Thus, it cannot be overlooked that the absence of passives in Creoles and pidgins must reflect at least in part the lack of models in some of the superstrate varieties to which speakers are exposed. It is well known that full agented passives are infrequent in colloquial English speech. Moreover, at least in the European languages concerned, the passive is characterized by more morphological complexity than the active, and complexity tends to be eliminated in pidginization. Some of the developments in Haitian Creole are only a step removed from documented Canadian and Belgian French usages (see Posner 1983: 195 — 8). Posner goes on to say, however, that in cases where Romance Creoles have substantially the same lexicon as a related patois, they can be considered members of the Romance "family", however marginally. Nevertheless, the issue is whether they are of the same "type". Her view (1985: 172) is that the two most salient criteria for distinguishing creole from patois, namely verbal inflection and noun gender, place these two varieties into different morphological types even though they remain by lexical-phonetic criteria members of the same family. Indeed Posner's view is that creolization involves typological change. I will next look at some of the sources for some features of verbal systems in Creoles and pidgins.

14

Susanne Romaine

1.2. Some features of verbal systems Bickerton's (1981) most substantial claims about bioprogram grammar concern tense, mood and aspect (see his and also Mühlhäusler's feature iii above). This is not the place for a detailed discussion of these categories in Creoles (see Romaine 1988: Ch. 7). My point here will be to illustrate that the same three possible sources can account for at least some of the properties of pidgin/creole verb systems: namely, substrate, superstrate and universals. Although Hall maintained that the contribution of the superstrate to pidgins and Creoles was always greater than that of the various substrata, his views elsewhere on Haitian Creole contradict this assumption. He comments (1966: 109) that "the entire inflectional system of the Haitian creole verb with its loss of tense and person- and number-endings and its use of aspectual prefixes, is straight African". While there is no doubt that the verbal systems of many French-based Creoles are strikingly different when compared to that of modern standard French, the latter should not be used as a yardstick of comparison, as I have already indicated. Comhaire-Sylvain (1936: 106) argues that it is impossible to account for the verbal system of Haitian Creole in terms of normal evolutive change from French; however, she then mentions a large number of periphrastic constructions with aspectual meaning found in vernacular varieties of French. Valdman (1977: 181) notes that analyticity in the expression of verbal categories and the absence of person-number inflection are not alien to certain types of overseas French (see also Posner 1983). Therefore, appeal need not be made to any special process of decreolization to account for the presence of these features in Creoles. In his analysis of the Reunion verbal system, Corne (in Baker and Corne 1982) claims that the morphosyntax and semantics derive mainly from seventeenth century varieties of French, and that the primary semantic distinctions are temporal, with aspect and modality functioning in a secondary role. Reunion has a relatively complex verbal morphology in which the copula carries markers of tense and aspect, as in French. Unlike French, however, et (re), the copula, does not function as an auxiliary. There are nevertheless some important creole features which distinguish it from French, such as the use of fin(i) plus past participle/adjective to mark the completive in stative predicates, e. g. lifinifatige 'he has become tired' (Baker and Corne 1982: 17). Corne (1982: 101 n. 3), however, raises the possibility of Bantu substratum in connection with the equivalent marker fin\m\n in Isle de France Creole since it does not seem in any sense other than an etymological one to be a

Typological contrasts between pidgin and Creole languages

\5

natural development of any variety of French. This marker and its semantic function are shared by other Indian Ocean Creoles. We can note here, too, a similarity of structural origin and function in the completive marker, pints, found, for example, in Tok Pisin (e. g. mi painim pints Ί found it'), and other English-based pidgins and Creoles. Otherwise, in Isle de France Creole, by contrast with Reunion Creole, the predicate system is mainly aspectually oriented with tense playing a secondary role. All distinctions of tense and aspect are marked by preposed particles, which, according to Bickerton (1981), is a characteristically Creole way of organizing the verb.

1.3. The comparative construction in pidgins and Creoles Bickerton (1981) does not include the comparative as part of the creole prototype. Markey (1982) has claimed that, generally speaking, creole comparatives are readily identifiable as input-specific. Nevertheless, a number of Creoles have a primary or secondary option of forming a comparative construction whose main characteristic is that the noun which serves as the standard of comparison is the direct object of a transitive verb whose meaning is 'surpass or exceed'. For example, in Cameroon Pidgin English, the comparative constructions in (4) and (5) are found: (4)

/ pas mi fo big.

'He is bigger than I.' (5)

/ big pas Bill. 'He is bigger than Bill.'

Reflexes of English pass are used in Jamaican Creole, Krio, Gullah and Sranan. Some French and Portuguese-based Creoles have a similar construction based on the lexeme meaning 'pass' (see Valkhoff 1966: 101 — 102 on the pasa comparative in the Portuguese-based Creole Principe). For example, Haitian Creole has bel\pi bel\pi bei pase tout 'beautiful/more beautiful/more beautiful passed all' (cf. standard French: beau\plus beau\le plus beau]. In Sranan the two types of comparative construction illustrated in (6) and (7) are found (see Voorhoeve 1962): (6)

Hugo can Ion moro betre leki Rudi. Hugo can run more better like Rudi 'Hugo can run better than Rudi.'

16

(7)

Susanne Romaine

A koni pasa mi. he smart surpass me 'He is smarter than I.'

The so-called exceed comparative (see Stassen 1985 for this term and discussion of the type) has been clearly modelled on the serial verb construction. Thus, it has been argued that the exceed comparative is a West African substratum feature (see, e.g., Hall 1966: 82 and Gilman 1972).3 It can be found in the West African languages Ewe, Yoruba, Twi and Igbo (cf. Twi ketwa sene me [small surpass me] 'smaller than I'). Modern Sranan has largely discarded this construction in favor of the /^/-comparative, a borrowing from the ///^-construction in English. Whatever the source of the construction, it would not be surprising to find it occurring independently in pidgins or Creoles because it represents a weakly grammaticalized and transparent means of expressing the notion of comparison. Seuren and Wekker (1986) suggest that Creole languages in general will tend to have a smaller set of secondary grammatical constructions.

1.4. Pronouns in pidgins and Creoles Some predictions about the shape of pidgin and creole grammars may fall out from the nature of principles and parameters in universal grammar as defined, for example, by Chomsky (1981). In fact, Bickerton has increasingly tended to frame the bioprogram in terms of the unmarked parametric settings (see Macedo 1986 and also Muysken 1981 on tense, mood and aspect in Creoles and predictions of markedness). This would of course involve a claim quite different from the one put forward by Posner. In Bickerton's view Creoles would reflect a kind of retrograde evolutionary movement to a maximally unmarked state, while in Posner's they would represent "advanced movement" in the same direction as so-called "advanced" varieties of noncreole varieties. "Advancement" need not necessarily complicate a language, although standardization often does (see Romaine 1984). In this connection Dahl's remarks (this volume) about Standard Average European languages serve to remind us that the standard written varieties of any language are artificial and reflect tendencies which Bickerton would attribute to cultural rather than natural grammar. (Whether this notion of evolutionary progress in typology is a coherent one cannot be discussed here, see also Ramat and Bernini [this volume] for a discussion of the notion of typological drift).

Typological contrasts between pidgin and Creole languages

17

Nevertheless, unmarked parametric settings may well account for certain properties of pidgin pronoun systems. Many jargons and stable pidgins, for instance, are pro-drop languages. Mühlhäusler (1986 a: 158) cites the example in (8) from Pacific Jargon English from 1840: (8)

Now got plenty money; no good work.

'Now I have lots of money so I do not need to work'. We would predict that if speakers were following the rules of their native language in inventing a pidgin they would follow the parameter setting in that language. Thus, in the case of speakers of pro-drop languages, we would expect that parameter to remain in force. Then the absence of pronouns in the resulting pidgin could be said to be due to substratum influence. Much the same argument applies to other cases of second language acquisition. White (1985), for example, has found that native speakers of Spanish (a prodrop language) learning English transfer this parameter. In the case of Pacific Jargon English we cannot argue superstrate influence because English is a non-pro-drop language. I do not know what the status of this parameter is in the numerous Oceanic substrate languages. A similar case is reported by Mufwene (1987) for Kituba in Zaire. Kikongo, the lexifier, is a pro-droplanguage, while Kituba is not. It is interesting, however, that in other cases of second language acquisition Meisel (1983: 202) claims that deletion of pronouns can be found irrespective of the first language backgrounds of the speakers. This suggests that prodrop constitutes the unmarked case. Hyams (1983) has argued this for firstlanguage acquisition.

2. Discussion My examples have shown that at present there is considerable uncertainty about the role lexifier languages and substrate influence, as opposed to universals, play in the various phases of the grammatical development of pidgins and Creoles. Hall stressed the idea of what we might call typological neutralization. He argued that pidgin grammar represents a common core between the grammars of the languages in contact. Givon (1979) has suggested that only those substratum features which are compatible with universal grammar can be preserved, but there are instances where substratum features surface which violate this. The phonological systems of Chinook

18

Susanne Romaine

Jargon and Pitcairn/Norfolk Creole are cases in point (see Romaine 1988: 64). Once we admit that Creoles subsequently develop in the same kinds of ways and are subject to the same kinds of constraints as natural languages, then the validity of Creole as a synchronic type is called into question (see e.g. Givon 1979: 19 — 22 and Mufwene 1986). Those who argue for the distinctiveness of Creoles generally make appeal to their special history and the catastrophic nature of the changes undergone (see e.g. Green 1987). In his discussion of whether Afrikaans is a creole, Markey (1982: 170) observes that to label Creoles as contact languages is vacuous because all languages are in some sense the product of contact. Likewise, to call all languages "creoles" is equally fatuous. However, subsequent changes remove the traces of creole history and erode their linguistic distinctiveness. Kihm (1983) goes so far as to say that the term 'creole' has no meaning in linguistic typology since linguists have failed to demonstrate that Creoles develop differently from other languages. Part of the problem in all typological exercises lies in the choice of features taken to characterize the prototype, and another part lies in certain assumptions behind linguistic typology. It is important to bear in mind with regard to the latter that there are no absolutes in typology. None of the properties I have discussed here are unique to Creoles or pidgins. Static typologies are of very little use when dealing with languages which change as rapidly as pidgins and Creoles (see Mühlhäusler 1986b on adjective-noun ordering). Surface similarity of form is no guarantee for a common genetic origin, or for the semantic equivalence of systems. Neither can commonality of structure be equated with sameness of function. Sankoff (1984: 104) has observed that recourse to either substrate or universals has generally been little more than an exercise in pattern matching. Some claims for substratum influence have been motivated by what Dillard (1970) and others have referred to as "the cafeteria principle", i. e., the idea that Creoles were mixtures of various rules from different regional varieties of British English. In other words, features were randomly picked out and attributed to substratum influence without regard for how they might have been borrowed or incorporated into the pidgin or creole in question. Welmers (1973), Manessy (1977) and others have emphasized the fact that the so-called "African substratum" is typologically diverse, and that combinations of substratum languages varied from place to place. If substratum influence was at all significant in creolization, how could such diversity of origin lead to uniformity in structure? More specifically, Sankoff argues that in order for any particular syntactic structure to surface and be sustained, it must prove to function as a viable

Typological contrasts between pidgin and Creole languages

19

discourse strategy. She says therefore that discursive practices are better candidates for areal features than specific features of morphology and syntax. In many areas the typological uniformity of Creoles may be the result of syntactic convergence of optimal discursive strategies. A case in point is the development of certain strategies of relativization in some of the English and French-based Creoles (see Romaine 1984 and 1988: Ch. 6). Posner (1985: 180182), for example, has discussed the use of la to demarcate relative clauses in certain French-based Creoles. (La is the definite NP marker derived from the French locative particle la 'there'). It can be used as a postposed deictic like Tok Pisin ia. (Compare Tok Pisin dispela meri ia 'this woman', and Haitian Creole fam bläS la 'the white woman'.) la is now used as a relativizer in Tok Pisin (see Sankoff — Brown 1976). From what Posner says, it appears that French-based Creoles represent various stages in the grammaticalization of this construction. It appears to be unknown in Louisiana and the Indian Ocean Creoles. It is optional in Martinique and Guadeloupe, but general in Haitian Creole. Lefebvre (1982: 37) claims that la is obligatory in restrictive relative clauses in Haitian Creole as in (9). (9)

tab

la

[m te

aste a]

table determiner I tense buy determiner 'The table that I bought' It is interesting that this development has progressed further in Haitian since the latter has less contact with superstrate varieties of French than other French-based Creoles. It is not, therefore, surprising that some of these same traits should turn up in non-standard spoken varieties, where there is less pressure to counteract natural developmental tendencies. My own sympathies lie very much with an approach to typology which is based on grammaticalization and classifies languages/varieties according to the extent to which and the sequence in which certain construction types, syntactic parameters and discourse functions are grammaticalized cross-linguistically (see, for example, König, this volume).

Notes 1. This property of pidgins and Creoles is attributed by Kay and Sankoff (1974: 66) to the fact that pidgins are derivationally shallower than natural languages, and thus reflect universal deep structure more directly. 2. Another interesting violation is noted by Mühlhäusler (1986 a: 159) for Samoan Plantation Pidgin. It does not follow the universal tendency for languages to make fewer distinctions in marked categories like the plural than in unmarked ones.

20

Susanne Romaine

3. Oilman (1972: 178—9) points out that in seventeenth and eighteenth century French and English the verbs passer/pass were occasionally found to express comparative notions. For example, in Samuel Johnson's dictionary from 1775 we find an example from Ben Jonson quoted: But in my royal subject, I pass thee. From the Dictionnaire de FAcademie Franfaise, we find: Le Prince Jesus ... qui passait en beaute les vierges es ies anges 'The prince Jesus who surpassed in beauty the virgins and angels'.

References Baker, Philip — Chris Corne 1982 Isle de France Creole. Affinities and Origins. Ann Arbor: Karoma. Bickerton, Derek 1981 Roots of Language. Ann Arbor: Karoma. Chaudenson, Robert 1979 Les Creoles franfais. Paris: Nathan. Chomsky, Noam 1981 "Principles and parameters in syntactic theory", in: Norbert Hornstein — David Lightfoot (eds.), Explanation in Linguistics. The Logical Problem of Language Acquisition. London: Longman, 32 — 75. 1986 Barriers. Cambridge, Mass.: MIT Press. Comhaire-Sylvain, Suzanne 1936 Le Creole hai'tien: morphologie et syntaxe. Port-au-Prince and Wetteren: Imprimerie de Meester. Dannequin, Claudine 1977 Les Enfants Baillonnes. Paris: CEDIC, Diffusion Nathan. Dillard, Joe 1970

"Principles in the history of American English — Paradox, virginity and cafeteria", Florida FL Reporter 8.

Oilman, Charles 1972 The Comparative Structure in French, English and Cameroonian Pidgin English: An Exercise in Linguistic Comparison. Ph. D. dissertation, Northwestern University. Givon, Talmy 1979 "Prolegomena to any sane creology", in: Ian Hancock — Edgar Polome — Morris Goodman — Bernd Heine (eds.), Readings in Creole Studies. Ghent: E. StoryScientia PVBA, 3-35. Green, John N. 1987 "Romance Creoles", in: Martin B. Harris — Nigel B. Vincent (eds.), The Romance Languages. London: Croom Helm. Greenberg, Joseph 1963 "Some universale of grammar with particular reference to the order of meaningful elements", in: Joseph H. Greenberg (ed.), Universals of Language. Cambridge, Mass.: MIT Press, 58-90.

Typological contrasts between pidgin and Creole languages

21

Hall, Robert A. 1953 Haitian Creole: Grammar, Texts, Vocabulary. (American Folklore Society, Memoir No. 43.) Philadelphia. 1966 Pidgin and Creole Languages. Ithaca: Cornell University Press. Heine, Bernd 1973 Pidginsprachen im Bantu-Bereich. Berlin: Dietrich Reimer Verlag. Hyams, Nina 1986 Language Acquisition and the Theory of Parameters. Dordrecht: D. Reidel. Kay, Paul — Gillian Sankoff 1974 "A language-universals approach to pidgins and Creoles", in: David DeCamp — Ian F. Hancock (eds.), Pidgins and Creoles. Washington, DC: Georgetown University Press, 61 —72. Kihm, Alain 1983

"De l'interet d'etudier des Creoles, ou qu'ont-ils d'especial?" Espace creole 5: 75 — 100.

Lefebvre, Claire 1982 "L'expansion d'une categoric grammaticale", in: Claire Lefebvre — H. MagloireHolly — N. Piou (eds.), Sjntaxe de l'Ha'üien. Ann Arbor: Karoma, 21—63. Macedo, Donald P. 1986 "The role of core grammar in pidgin development", Language Learning 36: 65 — 75. Manessy, Gabriel 1977 "Processes of pidginization in African languages", in: Albert Valdman (ed.), Pidgin and Creole Linguistics. Bloomington: Indiana University Press, 129 — 154. Markey, Thomas 1982 "Afrikaans: Creole or non-creole?", Zeitschrift für Dialektologie und Linguistik 49: 169-207. Meisel, Jürgen 1983 "Transfer as a second-language strategy", Language and Communication 3: 11—46. Mufwene, Salikoko 1986 "The universalist and substrate hypotheses complement one another", in: Pieter Muysken — Norval Smith (eds.), Substrata vs. Universals in Creole Genesis. Amsterdam: John Benjamins, 129 — 162. 1987 Formal evidence of pidginization/creolization in Kituba. [Paper presented at the 18th Conference on African Linguistics. University of Montreal.] Mühlhäusler, Peter 1986 a Pidgin and Creole Linguistics. Oxford: Blackwell. 1986b "Bonnet blanc et blanc bonnet: Adjective-noun order, substratum and language universals", in: Pieter Muysken — Norval Smith (eds.), Substrata vs. Universals in Creole Genesis. Amsterdam: John Benjamins, 41—55.

22

Susanne Romaine

Muysken, Pieter 1981 "Creole tense/mood/aspect systems: the unmarked case?", in: Pieter Muysken (ed.), Generative Studies on Creole Languages. Dordrecht: Foris, 181 — 199. Posner, Rebecca 1983 "The origins and affinities of French Creoles: New Perspectives", Language and Communication 3: 191—202. 1985 "Creolization as typological change. Some examples from Romance syntax", Diachronica II: 2: 167-188. Romaine, Suzanne 1984 "Relative clauses in child language, pidgins and Creoles", Australian Journal of Linguistics 4: 257-281. 1988 Pidgin and Creole Languages. London: Longman. Sankoff, Gillian 1984 "Substrate and universals in the Tok Pisin verb phrase", in: Deborah Schiffrin (ed.), Meaning, Form and Use in Context: linguistic Applications. Washington, DC: Georgetown University Press, 104—120. Sankoff, Gillian — Penelope Brown 1976 "The origins of syntax in discourse", Language 52: 631—666. Schumann, John 1986 Experimentally created pidgins. [Presentation at the Colloquium on Pidgins and Creoles: Issues in Language Acquisition and Education. University of Hawaii at Manoa, August 1—2, 1986. Honolulu, Hawaii.] Seuren, Pieter — Herman Wekker 1986 "Semantic transparency as a factor in Creole genesis", in: Pieter Muysken — Norval Smith (eds.), Universals vs. Substrata in Creole Genesis. Amsterdam: John Benjamins, 57 — 71. Stassen, Leon 1985 Comparison and Universal Grammar. Oxford: Blackwell. Taylor, Douglas 1971 "Grammatical and lexical affinities of Creoles", in: Dell Hymes (ed.), Pidgini^ation and Creolization of Languages. Cambridge: Cambridge University Press, 293 — 299. 1977 Languages of the Weft Indies. Baltimore: The Johns Hopkins University Press. Valdman, Albert 1977 "Creolization: Elaboration in the development of Creole French dialects", in: Albert Valdman (ed.), Pidgin and Creole Linguistics. Bloomington: Indiana University Press, 155-189. 1986 "Applied linguistics and language planning: A case study", in: Joshua Fishman — Andree Tabouret-Keller — Michael Clyne — Bh. Krishnamurti — M. Abdulaziz (eds.), The Fergusonian Impact. Vol. I, Berlin: Mouton de Gruyter, 517-536. Valkhoff, Maurice F. 1966 Studies in Portuguese and Creole: With Special Reference to South Africa. Johannesburg: Witwatersrand University.

Typological contrasts between pidgin and Creole languages

23

Voorhoeve, Jan 1962 Sranan Syntax. Amsterdam. Welmers, W. E. 1973 African Language Structures. Berkeley: University of California Press. White, Lydia 1985 "The 'pro-drop' parameter in adult second language acquisition", Language Learning^: 47-63. Wittmann, H. 1973 "Le joual, c'est-tu un Creole?", La Linguistique 9.2: 83-93.

Area influence versus typological drift in Western Europe: the case of negation* Paolo Ramat — Giuliano Bernini

1. The study of the languages spoken in Europe (as defined by geography) in terms of areal typology and Sprachbünde in a frame derived from classical Balkan linguistics (since Sandfeld 1930; see also Banfi 1985 for a recent survey), and from the — perhaps more convincing — results of investigations on the Indian subcontinent (Masica 1976 among others) and on Central America (Campbell et al. 1986), necessarily involves assessing two crucial conditions of language contact. On the one hand the historical-cultural, and hence sociolinguistic likelihood of mutual influence between the languages in question; on the other hand the merely linguistic likelihood — i. e. in terms of structure compatibility — of Sprachbund formation.' This second point, in turn, relates directly to the field of language universals and linguistic typology, in particular the universals — or rather tendencies — constraining linguistic borrowing proposed by Moravcsik 1978 (cf. also Bynon 1977: 253-255; Comrie 1981 b: 202-203). Likelihood of borrowing depends on the language component involved, ranging highest for lexicon, lower for syntax, lower again for morphology and for phonology. This ordering of language components according to likelihood of borrowing turns out to be a true implicational hierarchy, providing the criteria for the evaluation of a supposed Sprachbund. Thus the various contact phenomena can be assigned a different specific weight according to the language component(s) involved. For example, spreading of a lexeme from one language to another will be of lesser importance than, say, spreading of a particular feature of phonology. In addition, the above hierarchy marks out the path that borrowed items necessarily follow when passing from source to adoptive language, i. e. from lexicon either to phonology or to syntax and morphology. As a straightforward example, we can take the borrowing of a new morpheme for number; We would like to thank dr. Richard Dury, Istituto Universitario di Bergamo, for the careful revision of a first draft of the English text. Errors and mistakes still present in it are of course the result of our stubbornness. P. Ramat has written §§ 1—4, G. Bernini §§ 5 — 8. The final § 9 has been written by both authors.

26

Paolo Ramat and Giuliano Bernini

this implies antecedent borrowing of lexical items containing it, as illustrated by the adoption of the plural morpheme -im in Yidd. doktojrtm 'doctors' (sing. doktor) only after loans of lexical items (from Hebrew) such as gibojrim 'strong men', sing, giber (cf. Weinreich 1953 [1974]: 46). Similarly, borrowing of a word formation suffix, e. g. -aggio found in It. lavaggio 'washing', ingrassaggio 'greasing', presupposes earlier French loans into Italian as viaggio 'voyage', coraggio 'courage', and also Old It. domaggio 'damage' (thirteenth century). Indeed, word formation suffixes, which never occur as free morphemes in a language and will always occur as part of lexical items, are an important area of transition from grammar to lexicon, and here the relevance of the above implicational hierarchy becomes particularly apparent. 2 2. Bearing in mind the likelihood of borrowing mentioned above and the specific weight that the proposed hierarchy allows us to assign to contact phenomena, we can now ask whether it is possible to identify, within the larger set of features resulting from language contact, a smaller set of features generally relevant for the characterization of a Sprachbund. We may also ask whether the examples of convergence processes which one can observe in different languages should all count as "tokens" on a simply statistical level or could perhaps be ascribed to some major underlying principles as their "types". In the classic case of the Balkan Sprachbund it seems that some of the typical phenomena ("balkanisms") could indeed be ascribed to a common denominator. Periphrastic future (as in Gk. tha graphö < thelei hina grapho, lit. 'it wants that I write'); substitution of the infinitive by a subordinate clause containing a finite verb (as in Rom. se preface ca plänge 'he pretends to be crying', lit. 'that he is crying'); analytical comparative of adjectives (as in Alb. me bukur 'more beautiful', shume bukur 'most beautiful'); formation of numerals from 'eleven' through 'nineteen' (as in Bulg. edin-na-deset, lit. Oneon-ten'); postposed articles partly substituting inflection (as in Rom. codru-lui Of the wood', codri-lor Of the woods')3 all show the tendency to a transition from synthetic to an analytic type (cf. Ramat 1988 a). Here it might be better to speak of "language types in contact" (or better still "typological contact") instead of "language contact". On closer inspection, however, this example could appear less convincing, since some of the "balkanisms" mentioned above are not limited to the Balkan area (periphrastic future, analytical comparison and articles are also shared by Romance and Germanic languages) and other features traditionally considered as typical of the Balkan Sprachbund seem to run counter to the proposed typological tendency, as shown, e. g., by coalescence of genitive and dative and of locative (übt) and directive (quo)

Area influence versus typological drift in Western Europe

27

and by retention of a separate vocative case (see again Banfi 1985). Nonetheless, the specificness of the Balkan convergence area is constituted by the cooccurrence of a set of features; each of them may in principle be shared by other, non-Balkan languages (see also Lazard's observations concerning the actant features of so-called Standard Average European, this volume). Indeed, all the various Sprachbünde proposed to date for Europe since Kopitar and Miklosich, followed by the Prague structuralists (Jakobson, Havranek, cf. Bahner 1986) up to the list of "europemes" supplied by Haarmann (1976 b: 105), do not seem to be characterized by particular internal consistency. This observation raises again the question posed at the beginning of this section about a possible set of features allowing the characterization of a Sprachbund. To take a simple example, is simultaneous presence of monotonicity and of opposition of palatalized and non-palatalized consonants a sufficient criterion for characterizing a "Euro-Asian Sprachbund" as claimed by Haarmann (1976 a: 28 ff.)? 4 The obvious answer to this question is that definition of a Sprachbund must center on features belonging to the components which are more borrowingproof, that is to say geographically neighboring languages are more likely to participate in a Sprachbund the more morphological and phonological features they share. However these area-defining features should also meet two additional requisites; they should not be exclusively surface features, as, e. g., a plural morph compared to the category of plural; and they should possess a certain degree of markedness — in terms of "linguistic naturalness" — that would make at least doubtful the chance of their independent development. In this respect, isoglosses mapping the presence of clicks in the Bantu languages of Southern Africa (borrowed from neighboring Khoisan languages, cf. Maingard 1934), are more powerful evidence than, say, presence of nasalized vowels resulting from previous sequences of vowel + nasal consonant. This mention of modalities of diffusion seems to point to a possible unification of both areal contacts and typological drifts. The identification of a Sprachbund, for which a formal and sufficiently rigorous definition is still lacking (how many and of what kind should the relevant convergences be?),5 cannot be based solely on the observation of convergence phenomena, but must take into account their progressive development within a certain historical-cultural setting. It is clear that under this point of view European languages constitute a particularly suitable research field since their forms, contacts and mutual influences are attested over centuries, not to say millennia (cf., in this volume, König — van der Auwera on the evolution of absolute constructions).

28

Paolo Ramat and Giuliano Bernini

3. Coming now to the problem of defining a European Sprachbund (or Standard Average European, as proposed by Decsy (1973), using only linguistic parameters — leaving aside, therefore, the ethno-anthropological considerations proposed by Whorf (1941) who originally introduced the notion; cf. also Mioni 1986: 106), one of the most interesting and useful linguistic features for both empirical investigation and theoretical speculation turns out to be postverbal and/or discontinuous expression of sentence negation. The development of this type of negative construction is well documented throughout the history of both the Romance and the Germanic languages. It is a new morpho-syntactic feature that did not originate from any kind of superlectal influence, it being totally absent from both Latin and Greek. It is, therefore, a feature most suitable for testing the opposed hypotheses of common innovation of geographically neighboring but otherwise different languages and of independent development within similar typological drifts, allowing us to base our conclusions on a solid body of philological data. By way of contrast let us consider for a while the analytic conjugation of the verb in Celtic languages and Basque examined in this volume by McKenna: the trend to shift from inflection to periphrastic forms using some kind of auxiliary (e. g. Manx niom teacbt Ί shall do coming' instead of tigim Ί will come') is by no means restricted to Celtic and Basque (cf. i(x) tue schloafe Ί sleep', lit. Ί do sleep', in the Walser dialect of Gressoney, Aosta; Old Pol. b^dt} c^ynic 'they will make', lit. 'they are able to make' (Andersen 1987: 27) etc.). It reflects a general, "natural" trend of morphology towards iconic, transparent forms, easy to be analyzed. Therefore a correspondence in this domain is less significant than a correspondence among marked constructs such as discontinuous/postverbal negations.6 4. Diffusion of discontinuous and/or postverbal negations across the languages of Europe appears as a continuum with a core area and three surrounding fringe areas.7 The core area comprises the following languages:8 (1)

High Ger. (TVX/SOV) Ich sehe das Haus nicht Ί do not see the house'.

(2)

Low Ger. (TVX/SOV) Dat is niht aliens Botter wat [...] 'It is not all butter what [...]' (Sanders 1982: 205).

Area influence versus typological drift in Western Hurope

29

(3)

Yid. (TVX/SVO) Far vos %pgt ir nist oyf prost yidis az [...] 'Why don't you say in clear Yiddish that [...]' (Hutterer 1975: 360).

(4)

Dutch (TVX/SOV) Morgen gaan we niet naar school 'Tomorrow we do not go to school'

(5)

Frisian (TVX/SOV) Jan is der net 'Jan is not here'.

(6)

Sursilvan Romansh (TSVX) Questa schlateina ven buc ad ir giu 'This family name shall not be lost' (Schwegler 1983: 310).

(7)

Lombard of Bergamo (for Gallo-Italic, TSVX) Se no la somenza la t ca mia Otherwise the seed won't grow'.

(8)

Occitan (TSVX) Jan manjo pas de peissoun 'Jan does not eat fish'.

Beside these languages, all of which show postverbal negative patterns, the core area also comprises the following languages with discontinuous negative constructions, partly giving rise to postverbal negations after loss of the first element of the original construction, as, e. g., in both colloquial French and Welsh: (9)

Sursilvan Romansh (cf. also above, TSVX) Ke co nu fatscbi britch 'That I do not do this' (Schwegler 1983: 309).

(10)

French (TSVX) Je ne sais pas (colloquial: Je sais pas) Ί do not know'

(11)

Pyrenean Catalan (TSVX) No se cap Ί do not know' (Schwegler, 1988: 24).

(12)

Aragonese (TSVX) No la tasiaras brenca ista coca 'You won't taste this cake (at all)' (Schwegler, 1988: 24).

30

(13)

Paolo Ramat and Giuliano Bernini

Breton (SVO/VSO) N'

her gouient

ket

NEG it they-knew NEG 'They did not know it' (14)

Welsh (VSO) Nid yw 'r bachgen ddim yn hoffi coffi NEG is the boy NEG in like coffee 'The boy does not like coffee'9

This core area is surrounded by three fringe areas: the first one comprises English and the Nordic languages and shows a more or less marked tendency, depending on the language involved, towards re-establishing preverbal negative forms, c£: (15)

English (SVO) I did not hear him vs. I never heard him.

(cf. also substandard: '... but she don't care', where generalized don't functions as a preverbal negative particle).10 (16)

Swedish (for all Nordic languages, TVX) Han k nde inte komma Ήε could not come' vs. Det var synd, att han inte k nde komma 'It was bad, that he could not come'.

The second one comprises Italian and Catalan and is characterized by discontinuous constructions with emphatic value alternating with the "regular" non-emphatic preverbal constructions, cf.: (17)

Italian (TSVX) Νοη/ώ (mica) freddo qui 'It is not cold (at all) here'.11

(18)

Catalan (TSVX) Joan no menja (pas) peix 'John does not eat fish'.

The third fringe area comprises all of the varieties of Arabic spoken on the coast of North Africa and on Malta, together with Palestine, Lebanon and North Yemen, which possess a discontinuous and a preverbal construction in partially complementary distribution, whereby the latter is a derivation of the former (Bernini 1987):12

Area influence versus typological drift in Western Europe

(19)

Egyptian Arabic (for all the relevant varieties, TSVX/VSO): ma -tdir -s vs. mis ha -^dar NEG was:able NEC NEC FUTURE is:able 'He was not able' 'He won't be able' (Bernini 1987)

(20)

Maltese (Arabic) (SVO) Ganni ma jiekol -x hut Ganni NEC eats NEG fish 'Ganni does not eat fish'.

31

By way of contrast, no Slavic, Finno-Ugric, or Balkan language shows this type of negation. Eastern Europe constitutes an area not affected by this negation and characterized, again in opposition to Germanic and Romance languages, by the obligatory expression of the preverbal negative particle also when another fully negative word precedes the verb, as shown, e. g., by Russ. Nikto ne govorit vs. It. Nessuno par/a (and Non par/a Nessuno). 5. The areal character of the constructions exemplified above for Western Europe and Northern Africa is supported by a number of arguments. First of all, the feature at issue is shared by languages belonging to different families (Indo-European, Afroasiatic) and to different groups within the same family (Celtic, Germanic, Romance for Indo-European; Semitic and Berber for Afroasiatic) in different degrees, and its distribution across the relevant languages of Europe and Northern Africa is not random, but covers a compact area. Even when other word order parameters (especially verb position) are equivalent, languages belonging to the same group and even varieties of the same language may possess or lack this type of negative construction. The former case is illustrated by Celtic, with Welsh and Breton on one side (cf. the examples reported above) and Irish together with Scottish Gaelic on the other, cf.: (21)

Irish (VSO) Ni itheann Sean iasc NEG eats Sean fish 'Sean does not eat fish'.

(22)

Scottish Gaelic (VSO) Cha robb mi anns an taigh NEG was I in the house Ί was not at home' (Mackinnon 1971: 17).13

32

Paolo Ramat and Gitiliano Bernini

As is well known, this isogloss cuts the Romance group into two parts. In the Middle East, this same construction separates Arabic from Hebrew, which has only lo as a preverbal negative particle (Simon 1986: 43 f., 56). The latter case is illustrated by different varieties of Sursilvan Romansh, as exemplified above, and moreover by Northern "popular' Italian as against other colloquial varieties (Molinelli 1984) and by Palestinian and Lebanese Arabic as opposed to the Arabic of Syria, Iraq and Saudi Arabia (Bernini 1987). As for Europe,u geographic distribution of discontinuous and postverbal constructions seems to match both the stages of diachronic development from preverbal to postverbal through discontinuous negation patterns, and the chronological order of their grammaticalization, proceeding, as illustrated by Table 1, from Germanic (first fringe area and core area) to Romance and Celtic. Table 1. N. Germ. 12th c. V NEC 17th c. V NEC 20th c. V NEG/NEG V

W. Germ. NEG V NEG V NEG V NEG

Occitan NEG V (NEC) (NEG) V NEG VNEG

French NEC V (NEC) NEG V NEG (NEG) V NEG

Breton NEG V NEG V (NEG) NEG V NEG

In the twelfth century Icelandic should already have reached the stage of postverbal negation, while in Western Germanic the discontinuous construction was the norm. At this time, discontinuous negation was emerging in French. On the other hand, in the seventeenth century, Occitan reaches the stage of postverbal negation while French fully grammaticalizes the discontinuous construction.15 In this same period of time, German, English and Dutch had long since grammaticalized postverbal nicht\not\niet and Breton started generalizing the use of the discontinuous construction with preverbal ne and postverbal ket, which closely resembles the French construction, e. g. (23)

n' ho pens ket sentet ouz ho mamm ne nous-avez pas obei ä votre mere 'Vous n'avez pas obei ä votre mere' (Trepos n. d.: 184).

(24)

n' her gouient ket ne le ils-savaient pas 'Us ne le savaient pas' (Trepos n. d.: 184).

As in French, ket alternates with other postverbal negative elements, e.g.:

Area influence versus typological drift in Western Europe

(25)

n' anave^an nag e vreur nag en ne je-connais ni son frere ni lui 'Je ne connais ni son frere ni lui' (Trepos n. d.: 94).

(26)

ne glev netra (ebet) anezan ne il-entend rien (du tout) de lui (inflected preposition) Ί1 n'entend rien (du tout) de lui' (Trepos n.d.: 194).

33

See also: (27)

n' em-eus amezeg ebet er ger-man ne j'ai amis du tout dans cette-ville 'Je n'ai pas amis du tout dans cette ville' (Trepos n. d.: 137).

The etymological source of ket is still unknown. Hemon (1975: 284) claims that the original meaning of ket was 'indeed', because it was used in Middle Breton to reinforce an affirmative statement, e. g. da-η re a jynys quet ho fet en pechedou 'to those who end their lives indeed in sin' (M. 1396). However a negative meaning emerged after the presposition hep 'without', e. g. hep quet mar and also hep dout quet 'without any doubt' (Hemon 1975: 285). We should note in particular the fact that Middle Breton quet\ket could be substituted by pas, poent, of evident French origin, as well as by takenn 'drop', tamm 'coin', barr 'bud', a well-known semantic process, e.g. nefaMas pouent 'He did not fail at all' (NI. n. 22; see Hemon 1975: 285). We shall discuss the Celtic languages, and particularly Welsh, in some more detail a little further below. Here, however, we may note contact and influence of the French negative construction on Breton, as evidenced by the negative particle nompas < nonpas used before an infinitive, e.g.: (28)

evid nompas dispign e dammig arc'hant 'Pour ne pas depenser son peu d'argent' (Trepos n. d.: 192).

Finally, the time-space dimension of the continuum of diffusion of the constructions under examination finds its neat end-points in Gothic and Italian. In Gothic the word waiht(s), at the base of later West Germanic postverbal negative particles such as Eng. not, was used exclusively as a negative pronoun meaning "nothing" in negative sentences, cf.: (29)

I>ata leik ni boteifr waiht (J 6,63) 'Flesh is nothing worth.'

On the other end-point of the chronological continuum, modern Italian shows spreading of postverbal constructions, whereby niente 'nothing' turns out to be the only form of postverbal negation shared by both literary and

34

Paolo Ramat and Giuliano Bernini

colloquial varieties, the latter now using (with further socio-geographical constraints) postverbal negation elements in the following order of decreasing frequency: mai 'never', piu '(no) more', nessuno 'nobody', and finally mica. 6. The areal character of the discontinuous and postverbal negative constructions found across the languages of Europe as discussed above is further supported by the fact that negation is a category that affects the semanticpragmatic organization of sentences, in particular as concerns the relationship between the presupposed and the asserted part of the sentence, only the latter falling into the scope of negation, as illustrated by Charles did not hit the ball with violence, meaning that 'Charles did actually hit the ball, but not with violence'. Different negative scopes, as is known, result in different meanings when quantifiers are involved, cf. the classic case of the pair: Not many arrows hit the target vs. Many arrows did not hit the targets. We might note here that Old and Middle French possessed a system of four different types of negation, each one corresponding to a differently ranging scope: non was used in the case of single word negation as in the sentence non d'une chose mais de toute; non pas served as phrase negation, e. g. in non pas par mestier que fen aie; ne ... pas appeared in the case of negation of the dependence relationships between the verb and other elements of the sentence, e. g. in Ne demande pas comme sire \ mais par amors; finally ne was used in the case of whole predicate negation as in the sentence Segnor, fet els, je ne puts / moustrer hostel autre qu'il est (Englebert 1985). Apart from strongly marked cases such as the one cited by Jespersen (1917: 87, e. g. All that glisters is not gold, All is not lost; cf. also Ramat 1988 b) 16 and the instances of NEG-raising as with verbs of thinking, illustrated by the pair / do not think that he came = I think that he did not come (see, among others, Orlandini 1985: 556), different scopes of negation result in different implications, as shown by the following Welsh examples (Payne 1985: 239): (30)

N id wyf yn mal yn bnyta pigod NEG am-I often in eat fish don't often eat fish'

(31)

Yn ami nid uyf yn bwyta pisgod often NEG am-I in eat fish Often I don't eat fish'

In the unmarked sentence the scope of negation is coincident with sentence focus (cf. Hajicovä 1984) and the NEG operator occupies the position immediately before the verb, which is part of the focus, since it predicates something about the topic.

Area influence versus typological drift in Western Hurope

35

Despite Jespersen's examples referred to above, verbs of thinking as well as the cases of NEG-attraction of the type / think I saw no one in the room for / don't think I saw anyone in the room (see Sheintuch — Wise 1976), show some permeability ("Durchlässigkeit", Lerner — Sternfeld 1984: 165) of the negation in relation to the semantic-pragmatic nature of the sentence. Diachronically these movements of negative elements can display tendencies of development (also from the typological point of view) in the context of a markedness hierarchy. As illustrated above, postverbal negation is the marked case; data on first and second language acquisition seem to support this, postverbal negation appearing later than preverbal negation in the acquisition process (see note 6). This non-surface character of negation is also reflected by the fact that negative morphemes are among those least likely to undergo borrowing processes. The few attested cases are fully consistent with the regularities illustrated above concerning "universals of borrowing". Borrowed negative items are always lexical items and these do not affect the core of the system of negation in the adoptive language; cf. the adoption of Pers. hie by Turkic languages in order to allow the formation of a separate set of negative pronouns otherwise lacking, e. g. Turkish Aif kimse 'nobody' (lit. 'NEG somebody') on the pattern of Pers. hie käs (Räsänen 1969: 159 f.). Compare also the adoption of the Udmurt (Uralic) preverbal particle an by Chuvash (Turkic) in negative imperative constructions (Comrie 1981 a: 48). 17 7. We can now turn to arguments pointing to a rather cautious position as regards the full acceptance of the area character of discontinuous and postverbal negations across the languages of Europe. The etymological sources of postverbal negative expressions found in the relevant languages are extremely varied, as illustrated by following representative examples: — Norw./Dan. ikke, Icel. ekki < *eint-gi(n) One + generalizing particle' — Ger. nicht, Eng. not, etc. < *ni-eo-wiht 'not a thing' (cf. wight 'person') — various Romance types (but not exclusively) deriving from Latin expressions for 'point' (Fr. point, It. punto); 'step' (Fr., Occ., Cat. pas); 'crumb' (Gallo-It. mi(g)a, It. mica, Fr. mie; also Gallo-It. brisa, Romansh britch}; 'mouth' (Romansh buk}; 'people' (Cat. ges, also Piedm. nen if not from neente(m) 'not a wight'); 'drop' (Fr. goutte and also Bret, takenn and OHG — Bret, ket < M. Bret, quet, discussed above — Welsh dim, cf. Ir. dim, glossed 'quippiam, quicquam'

36

Paolo Ram at and Giuliano Bernini

— Ar. -/ < Cl. Ar. say?(an) (indefinite accusative) 'thing'. Despite the variety of formatives, there appear to be some noteworthy correspondences, evidenced by the series mht:dim:(ne)ente(m):-s\ mie:mica; goutte:takenn:drof. In these series all the elements share the development of a negative meaning starting from an original positive one, and some of them share the process of negative incorporation: cf. Ger. nicht, Eng. not as the most straightforward examples. On the other hand, examples like Lat. quoi neque paratast gutta certi consili (Plaut., Pseud. 397) '(he) who does not have a drop of a definite project', ne punctum quidem temporis oppugnatio respiraverit (Cic., Phil. 8,20) 'the assaults would not have stopped for a minute', illustrating the starting points of the negative use of the particles listed above, show that the processes giving rise to new negative particles may repeat themselves in the course of time, following a recurrent pattern of semantic evolution (cf. Wackernagel 1957: 254). Indeed, there exist countless negative expressions of this kind (as found, e. g., in modern colloquial Italian or earlier in literary Middle High German, cf. Mensing 1936); however, the relevant fact is that only some of them have been grammaticalized, e. g. only niawibt in Middle High German and none of the other expressions popular among authors and poets of that time. Another argument which could speak against area influence is represented by the fact that the discontinuous and postverbal constructions found in Europe and in Northern Africa seem to be correlated with either topic initial and verb second word order patterns (as in the Germanic languages with the sole exception of Modern English) or with word orders largely determined by pragmatic principles (as in less formal, even if still educated, varieties of Italian, French and Catalan). Moreover they appear to occur in languages which at least passed through a stage with TVX order during their histories (e. g. Old French, also early Gallo-Italic, possibly Arabic). This correlation could in the end turn out to be very strong, if we remember that negation has the asserted part of the sentence, i. e. the focus, in its scope, and that an order T V NEG FOC should therefore be more "natural". In this respect it is worth recalling the markedness of sentences like Eng. All that glisters is not gold, already discussed above, in the interpretation "Not all that glisters ?> However, postverbal/discontinuous negations are not necessarily found in every language where pragmatics prevails or prevailed for a certain time over syntax in the order of constituents, as evidenced by North African Arabic (possessing this feature), compared with Middle East Arabic (with preverbal negation).

Area influence versus typological drift in Western Europe

37

8. Having now discussed in some detail relevant aspects of the discontinuous/postverbal negative constructions found in the European-Mediterranean continuum, it would be methodologically opportune to check whether any other linguistic features show an areal distribution similar to, or, in the extreme case, overlapping with that found for the negative constructions under consideration. Körner (1987: 180 f.) groups the two types of Romance negation, i.e. (in his terms) "adverbial negation" (tie ... pas) and "adjectival negation" (non), with seven correlations which he claims distinguish two general linguistic types within Romania. His theory, however, cannot be said to account successfully for the lingusitic facts, nor does it provide any plausible explanation for the co-occurrence of the discussed features. Although, for the reasons stated above, cases such as negative constructions certainly count as stronger evidence than, for example, the spread of idioms like crocodile tears or iron curtain (Peruzzi 1985), which point to long-lasting and profound cultural rather than strictly linguistic contacts, it is obvious that a Sprachbund cannot be convincingly identified on the basis of just one single case. With this in mind, in order to link certain seemingly disparate facts and demonstrate that they ultimately obey the same underlying principles and strategies, it would be useful here to identify some link between the present case and some other typological developments. Now, at least in the case of the Germanic and Romance languages, which represent the core of the spreading discontinuous (and subsequently postverbal) negative constructions, the development of this type of negation seems to be concomitant with the typological change SOV —» SVO. Indeed, closer inspection reveals that the postverbal negation element is actually a former postverbal direct object with transitive verbs, e. g. non bibo guttam "I don't drink a drop", or accusative of measure with intransitive verbs, e. g. non vado passum don't go a (foot)step" (cf. Molinelli - Bernini - Ramat 1987a,b). As a test-case we might examine the situation in Welsh. Now, since Welsh has VSO as its basic order, a negation originating from a former object should show up after the subject, as is actually the case, e. g. (32)

Nid jy'r bachgen (ddim) yn hoffi coffi NEG is the boy N EG in like coffee 'The boy doesn't like coffee' (Payne 1985: 225).

This results in the pattern NEG V S NEG [

t— oo c\ c\ o ^ \Q Ό SO Γ^·

l

£P

lG .2G.s-S

rt

ϋΐ -22

G

α

is S ·Η g ^ 3 ·£ o

£ 3 < Λ «

G 'S £

W

_

I I -3 ~

u

c

^ )g

ίθ

11 ΐ ΐ ι ΐ ΐBl| - ϋ | ι |ι|1 sll'h II ill l Il|δ δεΐr§1l|s s| ||ιϊϊ &1 l «jj^piSZaj^WDSKtLlSE^^JS^ \O ^-

t^^-

OO GN ^ ^-

jZF

O O ^—' C1^ CNJ r

•sl J2 N J2

C/2 C/3

.3S

r^ rt ^

LO \O fO fO

ΛC Λb

-s ° § ^ S - S

§'8S

§ - §· "B G

iζ ill il 11 1 ii 33£«£ί*££3θ

^ - ' C N C O f O ^ J ' L O X O t ^ - O O ^

iO LO Ό O

t^(O

OO CN f^l rO

O ^"

«—' ^^

' ™H

U

» cS C

^ »Ϊ » ^ -S « -S ί

G u

^~ \O

g rt

(Λ

rt

rt

c New irish mo theach 'id.'; cf. also Gaulish mo gnäte (my son) 'son!' (Martinet 1986: 203), even though in a clearly marked position (cf. Latin mi gnäte 'id.'). The Scandinavian languages have had possessives placed after (also before) nouns since the earliest attestations, i. e. Runic inscriptions, but in Swedish the possessive adjectives usually precede nouns and in Danish they never follow nouns. All the Western Germanic languages now always place possessives before nouns, but word order was once quite free, cf., for instance, Old High German thiu qmna minu (this/the woman my) 'my wife' (Otfrid von Weissenburg, ninth century A.D.), see Behaghel (1923: 119); cf. also OHG mm fater / fater mm / der mm fater / der fater mm 'my father', four possible combinations mentioned by Jolivet —Mosse (1972: 167) and discussed by Ramat (1986: 110); Middle Dutch in den banden dijn 'in your hands' (Braunmüller 1982: 239); Old English sunu min (son my) 'son!' (Mitchell 1985: 120). Apart from the question of marked and unmarked forms, what is relevant is the fact that modern languages no longer allow such constructions even when expressing emphasis (cf. French). Constructions such as Danish en ven af mig 'a friend of mine' are another problem and surely not a matter of markedness. Slavonic languages present a change of pattern similar to that of Romance languages; see Table 21. The Old Church Slavonic pattern (Xaburgaev 1986: 241) was better preserved in Old Russian cf., for instance, "a clmi blagoslovilü tebja oteci moi" 'and how did my father (New Russian moj otec) bless you?' (Gramota Knja^a Dmitrija, 1388 A. D., in Janovic 1986: 164). The development of syntactic convergence is illustrated by biblical translations, even though linguistic inferences from translated texts require careful handling (see Table 22). In the end, it is interesting to notice that the Creoles derived from Romance and Germanic languages present both prenominal and postnominal possessives, (see Table 23, based mainly on D'jackov 1987: 12 —IS; see also Green 1988 b: 446—448), but, obviously, Creoles cannot be grouped with their lexifier languages from a typological point of view.

13. Balkan Sprachbund vs. Standard Average European Greek, which seems to have influenced the Balkan area, has followed a pattern that contrasts it with Western Romance, Celtic, Eastern Scandinavian, Western Germanic and Slavonic (except Bulgarian and Macedonian). If we were to

Possessive adnominal modifiers

75

examine the development of Greek since Homer's time to the present day, we would be forced to provide a more detailed discussion than is possible here (see Schwytzer-Debrunner 1966: 189, 201-202, etc.).

14. Conclusion The linguistic nature of possessive modifiers is not considered a crucial point here. Thus, I do not wonder, as Jacqueline Brunei (1980: 2) does in her Italian grammar, whether "L'adjectif possessif est-il un determinant?" [Is the possessive adjective a determinant?]. On this issue a passage by Vincent, though restricted to the Romance languages, is of special interest: A third category of determiner in a number of languages is the possessive, hence the absence of, say, Fr. *le mon cheval 'the my horse'. The Romance languages are not, however, united in this respect. The equivalent It. /'/ mio cavallo, Port, ο meu cavalo, OProv. ίο mieus cavals, Cat. el meu cavall are impeccable, as are forms such as It. questo mio cavallo, literally 'this my horse', un mio cavallo One of my horses'. For these languages, then, the possessive must be treated as either a straightforward adjective or as occupying a separate syntactic position between articles and adjectives (Vincent 1988: 54). This is a topic I could only touch on, as already seen above (Tables 18 and 19). Certainly a thorough investigation of noun-phrase structures in European languages, as conducted by Johannes Bechert with regard to the structure of the noun (see his paper in this same book), should yield interesting results. If we turn back to Table 16, which contains a wide sample of European languages, the figures linked to the placing of unemphatic possessive modifiers before or after nouns give a clear picture of the situation (see Table 24). Map 2 shows the geographical continuity of some of the above-mentioned syntactic features. In my opinion the unemphatic placing of possessive modifiers before nouns in European languages has a predominance too striking to be fortuitous. Such a syntactic feature has a good chance of being a characteristic of the posited Standard Average European.

76

Gianguido M anheilt

Table 1. Predicative English French Hungarian Finnish Greek

Substantival

this book is mine, ce livre-ci est a moi, ez a könyv a% enyem, tämä kirja on minim, auto to biblio einai diko mou,

jours is there le tien est la a tied ott van omasi on tuolla to diko sou einai ekei

Table 2. Russian

moja mo + ja

I

kniga knig+a

1

I

possessor morpheme (1st person singular)

ego

Russian

'my book'

case/gender/number agreement possessed morpheme (nominative feminine singular) kniga

'his book'

e+go

I case/gender/number agreement possessor morpheme (genitive masculine singular)

possessor morpheme (3rd person [singular])

Table 3.

Noun + suffix Prep/Postp + suffix Verb + suffix

Maltese

Hungarian

ihobbo/fc

hizad alatW szereta/

'your house' 'under you' 'he loves you' / 'you love him'

Possessive adnominal modifiers Table 4. 'my house" Squamish (Salishan, Coast Salish) Menomini (Algonquian) Navaho (Na-Dene, Athapaskan) Classical Nahuatl (Azteco-Tanoan, Uto-Aztecan) Jacaltec (Penutian, Mayan) Hixkaryana (Ge-Pano-Carib, Carib) Guarani (Equatorial, Tupi)

?«-la'm? »i'k //-kin »o-cal »>-atut ro-mtn f

Squamish: see Kuipers (1967: 87). Menomini: see Bloomfield (1962: 4). Navaho: see Hoijer (1969: 155). Classical Nahuatl: see Andrews (1975: 194). Jacaltec: see Craig (1977: 109). Hixkaryana: see Derbyshire (1979: 179). Guarani: see Gregores — Suärez (1967: 107).

Table 5. Abkhaz (NW Caucasian) Nivkh (Gilyak) Ket (Imbat dialect) Burushaski Dongola (Nilo-Saharian)

j--ab »'-ryf b-am a'u am-bab

'my 'my 'my 'my 'my

father' house' mother' father' father'

Abkha^: see Hewitt (1979: 262). Nivkb: see Panfilov (1968: 411) and Comrie (1981: 267). Ket: see Krejnovic (1968: 457). Burushaski: see Lorimer (1935: 127). Dongola: see Zavadovskij — Smagina (1986: 56).

Table 6. c

_ r_

__

Separate words e. g. Italian mio padre 'my father'

Clitics Bulgarian bastä mi 'my father'

Suffixes Hungarian apa;w 'my father'

77

78

Gianpuido M anbellt

Table 7. Gender agreement 'my mother' Separate words - nhy / ['i] ma 'i\-me mitt hus mein Haus mojate. küsta az en \aa.am minu maja minim talo»/ mü dallo

(man1) kudow evz« piirtem

mon' kudos' benim evim manän pürt/w

German: cf. Substandard German ihm sein Haus (him his house) 'his house', but *mir mein Haus or *dir dein Haus are not allowed (Seiler 1983: 71).

Table 14. Vocative forms

Italian Catalan Spanish Portuguese Russian Slovak Bulgarian

[— vocative] 'my son'

[+ vocative] '(my) son!'

mio figlio mon fill mi hi jo meu filho moj syn my sin sinüt mi

figlio mio\ fill meu\ jhijo mio\ filho meu\ syn moj\ sin moj\ moj sin!

Possessive adnominal modifiers

81

Table 15. Redundancy French Romanian Macedonian Greek

ma petite maman ä moi frate-j» lui majka mi moja emena ho gios mou

my mummy 'his brother' 'my mummy' 'my son'

Romanian: see Asan (1963: 158). Macedonian: see Koneski (1954: 100). Greek: see Stanitsas (n. d.: 59). Table 16. [See p. 84] Table 17. Some dead languages Romance languages Old Southern Dalmatic (Ragusa/Dubrovnik) (14th c.) Northern Dalmatic (Veglia/Krk) (+ 1898) Celtic languages Coidelic languages Manx (+ 1974) Brythonic languages Cornish (+ 1777) Slavonic languages Western Slavonic languages Polabian (+ 1798?) Baltic languages Old Prussian (+ 1677?)

meu fiiol la sua litera el mi tuota el tuta-JA la maja kuoza

'my son' 'his letter' 'my father' 'their father' 'my house'

my hie yn thie aym

'my house' 'my house'

ow thas

'my father'

müh lohlja (\rnuj 1'ol'a/)

'my father'

mats kermens

'my body'

Old Southern Dalmatic: see Bartoli (1906b: 262). Northern Dalmatic: see Bartoli (1906b: 9, 13, 53); enclitic forms are attested only with kinship terms, cf. el tuta-su (the fahter-their) 'their father', to be compared with Popular Romanian tatuso (father-the his) 'his father' (Asan 1963: 158). Manx: see Thomson (1984: 308);j« thie aym (the house at-me) 'my house', aym is the preposition ec 'at' + 1st singular person affixed pronoun. Cornish: see Pedersen (1913: 82); Thomas (1984: 285); cf., in modern spelling, ow thäs 'my father' (fas 'father') (Smith-Hooper 1972: 22). According to Lyon (1984: 24), "In Colloquial Cornish, the possessive pronoun itself ... is often left out and the little suffix [i.e. the postpositive personal pronoun used to emphasize ownership] only used ... Note! These latter construction will only be found in colloquial Cornish"; cf., for instance, cby-vy (house-I = my) 'my house', tas-hy 'her father', etc. Polabian: see Schleicher (1871: 251); Suprun (1987: 85). Old Prussian: see Endzelin (1944: 118); mats 'my' was a possessive adjective — see also Senn (1966: 32).

82

Gianguido Man^elli

Table 18. Possessive (Poss) and Definite article (Art) compatibility 1. Poss + Art + N 2. Poss + N + Art 3. N + Poss + Art 4. Art + Poss + N

5. Art + N + Poss 6. N + Art + Poss

Schiavonisco (Bulgarian and Macedonian [+ emphatic]) Basque (Erzya Mordvinian [+ emphatic]) (not attested) Italian, Milanese, Corsican, Friulian, Catalan, Aragonese, Asturian, Galician, Portuguese (Occitan [+ emphatic]) Greek, Southern Italian dialects, Sardinian, Hungarian, Maltese (Spanish [+ emphatic]) Moldavian, Romanian, Bulgarian, Macedonian, Albanian, Icelandic, Norwegian (Nynorsk and Bokmal)

Schiavonisco: Schiavonisco was the Italian language spoken by Serbo-Croats in the sixteenth century, cf. Schiavonisco mio-lpadre (my-the father) 'my father' (Bartoli 1906a: 286).

Table 19.

friend of mine' French

(Standard) un des mes amis einer meiner Freunde (Standard)

(Colloquial) un ami ä mot ein Freund von mir (Standard)

'

German

1 (Literary/humorous) un mien ami ein meiniger Freund (Bavarian) ·

Bavarian German: ein meiniger Freund (a my friend) 'a friend of mine' is a structure based on the Bavarian dialect, cf. ä meiniga Freindd 'id.' (Merkle 1975: 141).

Table 20. T

domus mea domus paterna patris domus

ma maison la maison paternelle la maison du pere

'my house' 'the paternal house' 'the father's house'

Table 21. Old Church Slavonic

Russian

domü moi domü oft ft domü otica

moj dom otcovskij dom dom otca

'my house' 'the paternal house' 'the father's house'

Possessive adnominal modifiers

83

Table 22. The Lord's Prayer (Matthew VI, 9 ff.; Luke X, 2 ff.) New Testament Greek Latin Italian French

Pater hemdn ... hagiastheto to onoma sou Pater noster ... sanctificetur nomen tuum Padre nostro ... sia santificato il tuo nome Notre pere ... que ton nom soil sanctifie

OHG (S. Rhine Franconian) (9th c.) Middle High German (14th c.)

Fater unser ... giuuihlt si namo thin Vater unser ... geheiligt werde din name

Old English New English

(1611 A. D.)

Faeder ure ... si pin nama gehälgöd Our father ... hallowed be thy name

Old Church Slavonic New Bulgarian

(9th c.) (1924 A. D.)

Otlce nast ... da sv^titu s? ime tvoje Otce nasit ... da se sveti (voeto ime

Table 23.

French

Portuguese

English

Creoles Seychellois

mo lakaz

Haitian Caboverdiano

kay mn/e mi kaza

ma maison

a minha casa

my house

Principense ka§i me Krio mi hos (Sierra Leone) Tok Pisin haus bilong mi (Papua New Guinea)

Table 24. Languages (a) with both prenominal and postnominal and ambipositional possessives (b) with both prenominal and postnominal possessives (c) with exclusively postnominal possessives (d) with exclusively prenominal possessives (e) with (not exclusively) prenominal possessives

Number (out of 92)

Percentage

6 11 15 48 69

6.5 11.9 16.3 52.2 75.0

84

Gianguido Man^elli

'S' υ

3 0

5$

-ί*

..„

-S

υ

«

·«

•B, w 'S

(Λ

Ic

-S!

t>

:U a ·«

-S, o, ·»

'S

_rt to c3 υ

^ "^ §

S

ΙΛ

'3 3

«

C 0 (Λ

ε Ι S? j

c

rt

ί

Ζ

ΑΪ* υ S is r rt

2

rt

ΙΛ

W

U

0

85

"u U

1 1 rt α

Λ '3

-S -S ΡΛ sS

t^. sl

VH

»H

, ,

-Ό Ό ^ ·«·*·«·* t> ^>

^i

1 11 «o 1 4-1 l) (Λ

4-t U (Λ

1 s -0 1 0

.R

'ΪΛ

-3

JS

*3

(ή (Λ e/i 3 3 3

3

rt

G & •a;

^

86

Gtanguido Man^elli

"u CO

ni

^ _S

1

O O

Cr tdr ty- t y- t - j >Mu -^i - i) i^ GOO

^ 1 -S

& ·1ί

·!3

|)§

1

.

1 -s

II

ε o

8

_0

:c

U

89

';Q~

« . "fc ^ »^ > β 1 ·— 81

Χι

υ

4-«

Ο

SC

„cc -ss 'S„c ·£

ε .„ «ι 'cε

'·?

W

If It llll

D Ο

'§

-s

»u

C Λ

'>o

^

r

'a .2 1 'i

^ *^

G rt

^ *-)

1 « 1 ;«

^S> -S § rt

C

H **»

f} *^

π _^

sH c Ί c .| rt «^ rt *^

1 § 1 'Ss ^.2 '8s

s ^

S»

·ί^

S;

*5

rt

s § § „ s _·%>

rt

1 ·»* 8

n

'o'

c

2

^ 0

3 «

J3 (n

JS c/i

C

3

4-*

Ο υ

\ο τ—
Modern Greek pou) is modeled, and passes from a consecutive function (cf. toso kakos, pou 'so bad, that ...') to a more general completive function (cf. kserö pou eimai kalos know, I am good'). From Medieval Greek, either by half-educated mediation, or popular mediation, the Greek form is likely to have been accepted by the southern Slavic (Bulgarian, Macedonian) environment. (30) a. Bulgarian se prestoril

ce (sto) umrel

REFL pretended-3SG that 'he pretended to be dead' b. Macedonian mislam

died-3SG

deka Selman-Aga imal

sobrano

arnauti

think-lSG that had-3SG gathered PPL Albanians think that Selma-Aga had gathered the Albanians'. 11.2. On the other hand, Roman, cä + indicative present, and Alb. qe + indicative present, represent the continuation of Romance forms based on

The infinitive in south-east European languages

179

Lat. QUOD (cf.: Ital. ehe, Fr. que, Span, que, etc.). This is the normal, general Romance evolution. 12. The synchronical, diachronical and geo-linguistic analysis of the data allows the establishment of an interpretative model of the dynamics [INF + ] vs [INF — ], typical of south-east European languages. It is clear that this model must be very articulated. The crisis of the infinitive within south-east European languages began from the Byzantine-Medieval Greek environment, as demonstrated, very correctly, by Sandfeld (1930: 173-179). However, this innovation spread in two main ways: i) Through both educated and half-educated influences — because of the contact between the Byzantine environment and the Macedonian, Bulgarian, Serbian church environment — the reduction of the infinitive, according to the Byzantine MT, spread within the old Slavic scriptae. Therefore, through southern Slavic mediation (Serbian or Bulgarian), the MT was accepted by Daco-Romanian. ii) Through popular influence — because of the contact between northern Modern Greek dialects and the multilingual environment of the Balkan Kerngebiet (in which the Vlahi, the Albanians, the Macedonians, the western Bulgarians, and the Serbians all took part) — the innovation spread into the popular levels of those linguistic systems. However, this innovation is likely to have consolidated itself between the ninth and the thirteenth centuries, when the Balkan ethno-linguistic situation was solidifying itself, before the Turks came. Vice versa, the crisis of the infinitive did not reach the north-western part of the south-east European environment (Dalmatia, Croatia, Bosnia, Hercegovina, Vojvodina, the Geg area of Albania, the north-eastern part of the Daco-Romanian area: Transylvania, Bukovina, Moldavia). In fact, those regions, oriented towards both West-Latin and Germanic-Latin culture, were only superficially touched by Byzantine and Medieval Greek culture. This explains the continuation of the old infinitive, as the result of its normal, general evolution within the Romance, Slavic and Albanian linguistic systems.

Notes 1. Sandfeld (1930: 173-180) and Schaller (1975: 156) consider the partial reduction of the infinitive in Balkan languages as a "primary balkanism". See Solta (1980: 210). 2. On the contrary, Rohlfs (1958: 733 — 744) considers the linguistic data pertaining to the reduction of the infinitive, in the light of a few Albanian, Modern Greek, Bulgarian,

180

3.

4.

5. 6. 7.

8.

9. 10. 11. 12.

13.

14.

Emanuele Banfi Macedonian, and Serbian dialectal forms. He also considers some parallel dialectal forms from both the Romance and Greek dialects spoken in the South of Italy. Among Balkan languages, only Greek has a diachronical uninterrupted tradition. Southern Slavic languages are only attested from the twelfth century; Romanian from the fifteenth century; and Albanian from the sixteenth to seventeenth centuries. On the one hand, in the case of the first records of Bulgarian, Serbian, and Macedonian, we have to consider very particular linguistic data (translations from Byzantine-Greek texts); on the other hand, the first records of Slovenian and Croatian are translations based on the Latin model. Finally, in the case of the first records of Romanian, texts are presented which are simply translations of Byzantine-Greek models, through a slavic mediation. It is clear that these texts are strongly "standardized", and it is very difficult to analyze them. For a discussion of the problematic see Ramat (1984). The research on Balkan dialectology appears very different, depending on the different regions. For instance, as regards Modern Greek dialectology, satisfactory descriptions pertaining to several areas of the Greek environment are lacking. In Triandaphullides (1938: 62 — 74), Browning (1983: 119 — 137), and in Newton (1972), one can find some interesting linguistic data. More satisfactory, as regards the Albanian environment, are the descriptions of Gjinari (1966: 31-50), Gjinari (1970), Desnitskaja (1968), Pellegrini (1977: 12-17) and Gabinksij (1971). The description of Romanian dialects is excellent, due to the fact that one can consult the rich material of the Noul Atlas Lingvistic Roman pe regiunei (NALR: 1967 fol.). As regards the Serbo-Croatian dialects, one can consult Popovic (1960: 401—455); for the Bulgarian dialects, see Mladenov (1929) and Holiolcev — Kostov — Mladenov (1977: 65 — 71); for the Macedonian dialects, see Koneski (1966). An excellent monography on the diachronic evolution of the infinitive in Balkan languages is the work by Joseph (1983), rich with evidence. A synthesis of the different points of view is to be found in Banfi (1985: 58 — 64). On one hand, the Slavic -// infinitive is thriving in eastern Hercegovina, in Sumadija Vojvodina, in Kosovo, in the Smederevo-Vrsac area, and in the Ikavo-Istrian area. On the other hand, the -/' infinitive is used in both Croatia and Slovenia. Elsewhere, both infinitives are found. The phenomenon distinguishing the two Bulgarian dialect areas is the treatment of the old Slavic vowel /e/ > West Bulgarian [e] vs East Bulgarian [ja]: cf. mleko 'milk' > [mleko] in the West Bulgarian dialects, vs [mljako] in the East Bulgarian dialects. For this particular form of infinitive (with me < Lat. MODO), cf. Pellegrini (1977: 184). Iliescu (1968: 115-118), on the basis of NALR data. For the periphrastic future in the Balkan area, cf. Banfi (1985: 54 — 55). Lunt (1974: 129 and 141) discusses the origin of the -ft' Slavic infinitive. According to him, there is a partial coincidence between the slavic -it infinitive and the Indo-European -tu supine form, both deriving from deverbative nouns. See also Joseph (1983: 101 — 105). Joseph (1983: 37 fol.): the first evidence of the reduction of the infinitive is already found in Thukidides 7, 21.3. For further examples pertaining to post-classical Greek, cf. Mandilaras (1973). In fact, according to Schaller (1975: 96 — 108), one can find all the so called 'primary balkanisms' in a relatively small region of South East Europe. They are present in Macedonia, Serbia, West Bulgaria, South Albania, and South Romania.

The infinitive in south-east European languages

181

References Banfi, Emanuele 1985 Linguistica balcanica. Bologna: Zanichelli. Browning, Robert 1983 Medieval and Modern Greek. Cambridge: Cambridge University Press. Burguiere, Paul 1960 Historic de l'infinitif grec. Paris: Klincksieck. Desnitskaja, Agnia V. 1968 Albanskijja^yk i ego dialekty [Albanian and its dialects]. Leningrad: Nauka. Gabinskij, Mark 1971 proisxozdenii starogo albanskogo infinitiva' [On the Origin of the old Albanian Infinitive], Zeitschrift für Balkanologie 8: 32 — 44. Gjinari, Jorgji 1966 "Essai d'une demarcation dialectale de la langue albanaise", Studia albanica 3, 2: 31-50. 1970 Dialektologja Shqipitare [Albanian Dialectology]. Prishtina: Enti i Teksteve dhe imjeteve Mesimore i Krahines Socialiste Autonome te Kosoves. Hendriks, Peter 1976 The Rado^da-Vevcani Dialect of Macedonian. Structure, Texts, Lexikon, (PdR Lisse: Ridder. Press Publications on Macedonian 1). Hesseling, Dirk-Christian 1892 "Essai historique sur l'infinitif grec" in: Psuchäres, löännes (ed.), Etudes de Philologie Neogrecque. Paris: Bouillon, 1 —43. Holiolcev, Hristo — Kiril Rostov — Maksim SI. Mladenov 1977 „Fragen der Zusammenstellung eines Atlas Linguarum Paeninsulae Balcanicae", Balkansko E^iko^nanie 20, 1—2: 65 — 71. Iliescu, Maria 1968 „Encore sur la perte de l'infinitif en roumain", Actes du premier Congres international des etudes balkaniques et sud-est europeennes, 6. Sofija [no indication of publisher], 115-118. Joseph, Brian D. 1983 The synchrony and diachrony of the Balkan Infinitive. A study in areal, general and historical linguistics. Cambridge: Cambridge University Press. Koneski, Blaze 1966 Istorija makedonskog je^ika [A History of the Macedonian language]. Beograd: Prosveta. Ljungvik, Herman 1932 Beiträge %ur Syntax der Spätgriechischen Votksssprache. Leipzig: Harrassowitz.

182

Emanuele Banfi

Lunt, Horace 1952 A Grammar of the Macedonian literary language. Skopje: Drzavno Knigoizdatelstvo. 1974 Old Church Slavonic Grammar. The Hague: Mouton. MacRobert, C. M. 1980 The Decline of the Infinitive in Bulgarian. Oxford: Oxford University Ph. D. Dissertations. Mandilaras, Basil 1973 The Verb in the Greek Non-Literary Papyri. Athenai: Hellenic Ministry of Culture and Science. Meillet, Antoine — Vaillant, Andre 1952 Grammaire de la langue serbo-croate. Paris: Champion. Mirambel, Andre 1961 "Participe et gerondif en grec medieval et moderne", Bulletin de la Societe de Linguistique de Paris, 56: 46 — 79. Mladenov, Stefan 1929 Geschichte der bulgarischen Sprache. Berlin: de Gruyter. Newton, Brian 1972 The generative interpretation of dialect. A study of Modern Greek phonology, Cambridge: Cambridge University Press. Pellegrini, Giovan Battista 1977 Introdu^ione allo studio delta lingua albanese. Padova: Seminario di Filologia Balcanica. Popovic, Ivan 1960 Geschichte der serbokroatischen Sprache. Wiesbaden: Harrassowitz. Ramat, Paolo 1984 Linguistica tipologica, Bologna: II Mulino. Reichenkron, Günter 1962 „Der Typus der Balkansprachen", Zeitschrift für Balkanologie, 1: 91 — 122. Rohlfs, Gerhard 1958 "La perdita deH'infinito nelle lingue balcaniche" in: Omagiu lui lorgu Jordan, Bucure§ti [no indication of publisher], 733 — 744. Sandfeld, Kristian 1930 Linguistique balkanique. Problems et resultats. Paris: Klincksieck. Schaller, Helmut Wilhelm 1975 Die Balkansprachen. Eine Einführung in die Balkanphilologie. Heidelberg: Winter. Solta, Renatus 1980 Einführung in die Balkanlinguistik mit besonderer Berücksichtigung des Substrats und des Balkanlateinischen. Darmstadt: Wissenschaftliche Buchgesellschaft. Togeby, Knud 1962 "L'infinitif dans les langues balkaniques", Romance Philology 15: 221—233.

The infinitive in south-east European languages

183

Triantaphullidcs, Manoles 1938 Neoellenike Grammatike. I (Historike Eisagöge) [A Grammar of Modern Greek. Historical Introduction]. Athenai: Oikos tou syggrapheös. Weigand, Gustav 1925 "Programm des Balkan-Archivs", Balkan Archiv 1: 5 — 15.

Conjugation of the verb in modern Celtic and Basque: from inflection to periphrasis Malachy McKenna

Introduction Lewy (1942: 15 —117) suggested a five-fold geographical classification for a number of languages in Europe; the classes are as follows: 1) Atlantic, 2) Central, 3) Balkan, 4) Eastern, 5) Arctic. The Atlantic group consists of Basque, Spanish, Italian, French, Irish, English, and Swedish, and Lewy identified Flexionsisolierung as a major feature which all of these had, to a greater or lesser degree, in common. It is against this background that it has been decided to examine Basque and Irish — along with the other modern Celtic dialects — from the viewpoint of the loss of verbal inflection and its replacement by periphrastic constructions, and it is hoped that in so doing some light will be shed on the convergence of languages which are areally related but genetically unrelated. Considerations of time do not permit an examination here of the impersonal inflections for tenses and moods, nor of the verbal adjective, though these are all subject to loss of inflection and replacement by periphrasis. For each of the languages in question I will ask two questions: in a given source for each language, a) which form categories are productive and b) which form categories have been replaced by periphrasis.' If one postulates a linear continuum with an ideal inflected type at one end and an ideal periphrastic type at the other, then the answers to the above questions should provide data to establish the approximate positions on the continuum for each of the languages. It is important to note that while it is regular in dialects of Celtic for certain form categories to merge, this does not necessarily give rise to a new periphrastic construction as a substitute. In Scottish Gaelic, for example, what was historically the conditional has merged formally with the imperfect, but this latter then functions both as imperfect and conditional. Thus chuirfinn Ί would put' falls together with chuirinn Ί used to put', and chuirinn then signifies both Ί used to put' and Ί would put'.

186

Malachy McKenna

Irish: system as in Lucas (1979) Lexical verbs have six form categories, viz. present, preterite, future, conditional (which also signifies past subjunctive), present subjunctive and imperative. Apart from the third person of the imperative, these form categories are all productive. The third person of the imperative is replaced by periphrasis in the following ways: (1)

spread/) se X 'let him spread X' > cead aige X a spreamh 'he has permission to spread X'

(2)

teadh siad X 'let them heat X' > lig dobhtha X a theamh 'let them heat X'

Historically Irish had an inflected past habitual tense but, according to Lucas, it has merged with the conditional; a periphrastic construction is however replacing this form category when it signifies the past habitual: (3) dhiolfadh se X 'he used to sell X' > ba ghnach leis X a dhiol 'it was customary for him to sell X' Some verbs are inflected for tense and mood only rarely or not at all, their only productive form being the verbnoun. For these, periphrasis is the rule, the auxiliaries being hi 'be' and dean 'do': (4) a. bhi me ag dull Ί expected' b. beidh se ag srannfach 'he will snore' (5) a. rinn an t-easbog e a choinnealbhadhadh 'the bishop excommunicated him' b. dheanfadh an madadh tafaint 'the dog would bark' Some auxiliaries other than 'be', 'do', are 'give', 'put', 'take', 'go', etc.: (6)

thug se tarrthail uirthi 'he saved her' (literally 'he gave saving on her')

(7)

chuir se bacail arm 'he hindered me' (literally 'he put hindering on me')

Scottish Gaelic: system as in Holmer (1938) Lexical verbs have four form categories, viz. future, preterite, imperfect (which also signifies the conditional) and the imperative. These are productive. Holmer cites no certain evidence for the replacement of form categories by periphrastic constructions.

Conjugation of the verb in modern Celtic and Basque

187

Manx: system as in Broderick (1984) As far as lexical verbs are concerned, form categories have become unproductive 2 and have been replaced by periphrastic constructions consisting of an inflected auxiliary (generally dean 'do') plus verbnoun of the lexical verb: (8)

tigim Ί will come' > mm teacht literally Ί shall do coming'

(9)

thog se 'he built' > rinn se togail literally 'he did building'

(10)

dheanfadh se X > dheanfadh se deanamh X literally 'he would do (the) doing (of) X'

(11)

na seasl 'don't stand' > na dean seasamh literally 'don't do standing'

Welsh: system as in Jones (1977) For lexical verbs three form categories are productive, viz. future, preterite and imperative. What is termed here "future" was historically the present, but it no longer functions productively as a present. The role of the present has been taken over by a periphrastic construction, type inflected auxiliary (bod 'be') plus jn 'in, at' plus verbnoun of the lexical verb: (12)

darllenaf Ί read' > rydw iyn darllen

Historically Welsh had an inflected imperfect and an inflected pluperfect but these have been replaced by constructions of a type similar to (12): (13)

gwydwn Ί knew' > roeddivn iyn givybod

(14)

darllenasai 'she had read' > roedd hi wedi darllen

A form category of gwneud 'do' plus verbnoun constitutes another type of periphrasis: (15)

agorwch 'you will open' > newcb cbi agor3

Cornish: system as in Jenner (1904) Lexical verbs had six form categories, viz. present indicative (which also signifies the future), imperfect, preterite, pluperfect (also signifies conditional), subjunctive and imperative. All of these have become unproductive 4 in lexical

188

Malachy McKenna

verbs and are replaced by a periphrasis which consists of an inflected auxiliary (generally gwil 'do') in combination with the verbnoun of the lexical verb: (16)

carav Ί love' > gvrav cara5

(17)

kerys Ί loved' > grwigav cara

(18)

carsen Ί had loved' > gwressen cara

(19)

carev Ί may love' > gwrellev cara

(20)

car\ 'love!' > gwra cara

The inflected imperfect is replaced by periphrasis with the imperfect of bos 'be' plus ow 'at' plus verbnoun of the lexical verb: (21)

caren Ί used to love' > therav ow cara

The inflected present when signifying future time is replaced by periphrasis with the present indicative of menny 'will, wish' plus verbnoun of the lexical verb: (22)

carav Ί will love' > mennav cara

Breton: system as in Denez (1972) In this variety of Breton, lexical verbs have five form categories, viz. present indicative, future, imperfect, conditional and imperative. All of these are productive and are, in fact, obligatory subject to certain conditions which will be outlined below. However in the case of almost all lexical verbs, all form categories, with the exception of the imperative, can be expressed by a periphrasis consisting of verbnoun plus particle a or e plus inflected auxiliary (generally ober 'do').6 When the negative particle precedes a lexical verb, then a form category of the latter must be used: (23) a. ne lennan ket Ί do not read' Furthermore certain types of word order require the use of form categories of lexical verbs; in a cleft sentence, for example, when the focus is the subject, the object or an adverbial, an inflected form of the lexical verb must be used: (23) b. me a lenn al levr bemde^ '(It is) I who read the book everyday' c. al levr a lennan bemde^ '(It is) the book that I read everyday' d. bemde^ e lennan al levr '(It is) everyday that I read the book'

Conjugation of the verb in modern Celtic and Basque

\ 89

However when the lexical verb is the focus, a periphrasis must be used which consists of verbnoun of the lexical verb plus a plus inflected auxiliary: (24)

lenn al levr a ran bemde^ '(It is) the reading of the book that I do everyday'

Basque: system as in Lafitte (1944) There are three form categories, viz. present, imperfect and imperative though they are productive for only a small number of lexical verbs (approximately five). 7 They have been replaced by periphrastic constructions consisting of a form of the verbnoun of the lexical verb plus an inflected auxiliary verb. For transitive verbs the principal auxiliary is ukan 'have' and for intransitive verbs, 'be': (25)

dagal 'he loses it' > galten du

(26)

^egalen 'he lost it' > galt^en

(27)

begal 'let him lose it' > gal be%a

By way of summary an indication may be made of the position each language holds on the continuum mentioned at the beginning of this paper; the model in Figure 1 may be suggested for the varieties we have looked at. I Inflected Type Irish ^\^ Scottish Gaelic

1 Welsh v^

.^ ^^

Periphrastic Type Manx Cornish Basque

Breton ^ Figure 1. Inflection/Periphrasis Continuum

Conclusion Historically, dialects of Celtic showed a high degree of verbal inflection (in the Goidelic group, for example, there were eight form categories) though periphrasis was not unknown in them. 8 This situation has been radically

190

Malachy McKenna

altered in some dialects in that for lexical verbs inflections have been set aside to a great degree and have been replaced by periphrasis in which generally only the main auxiliaries retain form categories of the original system; and it is in this respect that Basque shows a remarkable parallel with Manx and Cornish. We have here, then, a clear instance of the convergence of languages which are related areally but not genetically. In the discussion which followed this paper at the Rome workshop it was suggested that the rise of periphrasis in Cornish and Manx might be linked to simplification processes which are in turn associated with bilingualism and language "loss"; it was also suggested that borrowed verbs might be more prone to periphrasis than native verbs of long-standing. It would not be appropriate here to go into these matters as they lie outside the general scope of the present study. While explanations for the rise of periphrasis are undoubtedly called for, it would be hazardous to look for them without detailed reference to the interplay of periphrasis and inflection in other languages in Europe;9 this theme must await further analysis.

Notes 1. My choice of sources is an arbitrary one which was dictated chiefly by their availability and by the shortness of time at my disposal in drafting this paper. I am grateful to colleagues who read the work and suggested improvements: W. Gillies, F. Kelly, Β. Ο Buachalla, C. Ο Cleirigh, M. O Murchu, O. Padel, H. Wagner, A. Watkins. The paper is dedicated to the memory of Prof. Heinrich Wagner who died some ten months after having encouraged me to undertake the present study. 2. On this matter Broderick (1984: 86) comments, "Inflected forms of verbs in Spoken Manx are not common and those that are attested are largely confined to the irregular verbs ... The occurrences of inflected forms of regular verbs are scarce". In this, Broderick is in close agreement with Thomson (1984: 310) who writes, "There is thus no compulsion to utilize the inflected tenses of any verb except the two auxiliaries, and the preference for this analytical form increases steadily to an almost exclusive use of it in late Manx". Note that in the Manx examples I cite I have normalized the spelling using the conventions of modern Irish orthography. 3. For examples of other form categories (pluperfect and preterite) of gwneud in use as auxiliary cf. Fynes-Clinton (1913: 155). I am grateful to Prof. A. Watkins for pointing out to me that periphrasticisation of the Welsh verbal system (particularly in the Gwynedd region) has gone much further than Jones (1977) notes for "Living Welsh". Cf. also Thomas (1967). 4. Jenner (1904: 135) outlines the situation as follows: "Most tenses have at least two forms, the simple verb ... and the compound, or verb with auxiliaries. In late Cornish the compound is by far the more usual in almost every tense". I note however that form categories are predominant in responsives in Jenner (1904: 159, 161) e. g, 'Ellough why cowsa Kernuafc? Gellam (or) mi ellam 'Can you speak Cornish? I Can.'

Conjugation of the verb in modern Celtic and Basque

191

5. While the auxiliary in this example and all the other examples cited here is inflected for person (-v = T), Jenner frequently includes an independent subject pronoun (= vi in the first person singular) immediately after the inflected verb, e. g. givrav vi cam; for the sake of clarity I have omitted this pronoun in the examples. 6. For further remarks on periphrasis in a variety of spoken Breton cf. McKenna (1988: 150— 151). 7. Lafitte (1944: 250, 290) describes this situation as follows, "Les regies de construction que nous avons exposees au chapitre precedent permettraient theoriquement de conjuguer beaucoup de verbes ... En fait, il n'y a ä suivre la conjugaison forte, actuellement, que de cinq verbes intransitifs; egon rester; joan aller; ibili marcher; i%an etre; edin se trouver. Cinq autres se conjuguent tres partiellement" and "On pourrait, en appliquant les regies du chapitre precedent, conjuguer un grand nombre de verbes transitifs. En fait, actuellement, peu de verbes ont une conjugaison forte. Aucun n'a une conjugaison forte complete. Les moins incomplets sont les auxiliaires ukan et i%an, avoir, dont l'ensemble constitue un Systeme verbal imposant. Apres eux viennent eraman empörter; eduki tenir; erabili mouvoir; ekarri porter. Tout le reste est nettement defectiP'. According to Tovar (1957: 94) form categories were in use for around sixty verbs in sixteenth-century written Basque. 8. Cf. Wagner (1981: 56) for the periphrastic progressive present in Old Irish and also for a detailed discussion of periphrasis in Celtic and other languages, especially those of North Africa, in the context of substratum theory. On periphrasis with dean 'do' as auxiliary from an early period in Irish cf. O'Rahilly (1950: 318). 9. Cf. in particular Vincent (1987: 237 — 256). Two papers from the Rome workshop that come to mind as being relevant to the present study are "The development of cases and adpositions in Latin" (H. Pinkster) and "Area influence vs. typological drift in W. Europe" (Ramat — Bernini), the latter in this volume.

References Broderick, George 1984 A Handbook of Late Spoken Manx. (2 vols.) (Buchreihe der Zeitschrift für celtische Philologie 3). Tübingen: Niemeyer. Denez, Per 1972

Bre^honeg ... buan hag aes. Paris: Editions-Disques Omni vox.

Fynes-Clinton, O. H. 1913 The Welsh vocabulary of the Bangor District. London: Oxford University Press. Harris, Martin — Paolo Ramat (eds.) 1987 Historical Development of Auxiliaries. Berlin: Mouton de Gruyter. Holmer, Nils 1938 Studies on Argyllshire Gaelic. (Skrifter utgivna av. K.-Humanistiska VetenskapSamfundet i Uppsala, 31). Uppsala: Almqvist & Wiksell. Jenner, Henry 1904 A Handbook of the Cornish Language. London: David Nutt.

192

Malachy McKenna

Jones, T. J. Rhys 1977 Living Welsh. London: Hodder and Stoughton. Lafitte, Pierre 1944 Grammaire Basque. Bayonne: Librairie "Le Livre". Lewy, Ernst 1942

"Der Bau der europäischen Sprachen", in: Proceedings of the Royal Irish Academy 48: 15-117. Dublin: Hodges, Figgis and Co.

Lucas, Leslie 1979 Grammar of Ros Goill Irish, Co. Donegal. (Studies in Irish Language and Literature 5). Belfast: Queen's University. McKenna, Malachy 1988 A Handbook of Modern Spoken Breton. (Buchreihe der Zeitschrift für celtische Philologie 6). Tübingen: Niemeyer. O'Driscoll, Robert (ed.) 1981 The Celtic Consciousness. Canada: McClelland and Stuart. O'Rahilly, Thomas F. 1950 "Do-nim, deanaim, proceed, go'", Celtica 1: 318. Dublin: The Dublin Institute for Advanced Studies. Thomas, Alan R. 1967 "Constituents of the periphrastic verbal phrase in Welsh", Stadia Celtica 2: 147— 170. Cardiff: University of Wales Press. Thomson, R. L. 1984 "Manx", in: Peter Trudgill (ed.), 306-317. Tovar, Antonio 1957 The Basque Language (translated from Spanish original, La Lengua Vasca, second edition, by Herbert Pierrepont Houghton). Philadelphia: University of Pennsylvania Press. Trudgill, Peter (ed.) 1984 Language in the British Isles. Cambridge: Cambridge University Press. Vincent, Nigel 1987 "The interaction of periphrasis and inflection: some Romance examples", in: Martin Harris — Paulo Ramat (eds.), 237—256. Wagner, Heinrich 1981 "Near Eastern and African connections with the Celtic World", in: Robert O'Driscoll (ed.), 51-67.

4. Tense, aspect and modality

Types of tense and aspect systems Wolfgang Raible

1. A major cause of the difficulties we have to cope with when analyzing verbal systems is interference between categories of the respective system. For this reason the following reflections are aimed at the interrelation between aspect, Aktionsart, tense, and negation in certain types of verbal systems. In my view, "aspects" and "tenses" are deictic, i. e. they refer to the speaker, whereas Aktionsarten are understood as definitory categories.1 Tenses refer to the temporal relation holding between the speech event and the represented or — as Roman Jakobson put it — "narrated event". With Erwin Koschmieder we would say that they affect the relation between " Vorgangs^eit" and "Akt^eii". There is a fundamental opposition between the time of the speech event on the one hand and future/past on the other. With regard to aspects, I refer to the difference between the imperfective and the perfective aspect as defined by Eduard Hermann (1933). The "imperfective" aspect means that the viewpoint of the speaker is inside the action whereas the perfective aspect is an action seen from the outside. What is seen from the outside is seen as a whole, as a closed unit — this is meant by expressions such as "complexive" or "perfective". In order to emphasize the difference: tense locates the narrated event with respect to the speech event, aspect locates the speech event with respect to the narrated event. Karl Bühler explained the difference with one of the tales of Nasrettin Hoca. Tense brings the mountain to the prophet whereas in aspect the prophet moves to the mountain. The following example should make it clear what may be understood by the definitory category of Aktionsart. Let us take the verb to be in flower, German blühen. When I say the amaryllis is in flower, I refer to a continuous event. The amaryllis was in flower before, it is in flower now, il will be in flower later on: there is no change. According to Klaus Heger, this would be [FTP], where T is a temporal functor relating two predicators. The square brackets (added by the author) show that we are dealing with the definitory category of Aktionsart. At some time the amaryllis will have begun to bloom, at some other time, it will fade. In German we would say die Amaryllis erblüht, die Amaryllis blüht auf, 'it bursts into bloom'; we could say die Amaryllis verblüht, 'it fades'. A verb such as erblühen would have the structure [nPTP]

196

Wolfgang Ra'ible

("the amaryllis was not in flower before, afterwards it is in flower"), where the position at the left and at the right side of T expresses anteriority and posteriority. The verb verblühen would then have the internal structure [PTnP]. If we look for the common denominator of erblühenj'aufblühen and verblühen on the one hand and blühen on the other, we might use terms such as: transformative cyclic resultative telic

vs. vs. vs. vs.

non-transformative (Martin Sanchez Ruiperez) non-cyclic (William E. Bull) non-resultative non-telic (a-telic)

In studies on Aktionsarten we find numerous terms highlighting either the transformative type to burst into bloom or the transformative type to fade: in one case the verbs are "incohative", "ingressive", "initial", "non-conclusive", etc.; in the other case the verbs are "definitive", "desinent", "effective", "egressive", "final", "conclusive", "preteritive", "terminative", etc. As regards the type to be in flower, there are designations such as "durative", "extensive", "medial", "perdurative", "permanent", "continuative", etc.2 Appendices I and II show some of the possibilities to be considered for the classification of Aktionsarten. In Appendix II there is a further subdivision of the class of non-transformative Aktionsart into "state" and "activity". At the same time, the transformative modes of action corresponding to the initial phase in Appendix I are absent from the scheme in Appendix II. As regards the "durative" verbs, further subcategorizations are possible: there may be slow processes of augmentation or diminution concerning especially the subclass of non-transformative verbs which is called "activity" in Appendix II. In Appendix III, five noematic layers are distinguished. The scheme is a visual representation of their interlocking, their mutual relationships and/or affinities as well as their affinity to aspect categories, and shows clearly why a hierarchical representation or a deduction of other layers/categories from only one layer is not possible. These transformative and non-transformative modes of action can be combined with another kind of Aktionsart — more precisely: with "Repetition" (layer [d] in Appendix III) — emphasizing the frequentative, distributive, iterative character of the event (verbs such as to tremble, to flap one's wings, to flicker, to bounce etc.). As Dressier (1968) puts it, there are cases of "verbal plurality". 2. Tense systems and aspect systems may be homogeneous or combined with each other. There are four major possibilities:

Types of tense and aspect systems

(la) a pure aspect system (2a) an aspect system by priority, combined with a tense system

197

(Ib) a pure tense system (2b) a tense system by priority, cornbined with an aspect system

To make things more complex, we also have to take into consideration the modes of action. There seem to be two extreme possibilities. In the first one, the definitory modes of action are already realized in the verb stem itself or by means of special grammemes. The other extreme are systems where the mode of action only emerges from the combination of a neutral verbal lexeme and tense/aspect markers. Thus I should subdivide type (la) into (laa) (a pure aspect system based on verbs expressing by themselves modes of action) and (lab) (a pure aspect system based on verb lexemes which are neutral with respect to Aktionsarten), and so on. A good candidate for a system of type (laa) is, e.g., the Hopi language as described in Stahlschmidt's (1982) doctoral thesis. On the other hand, the varieties of Arabic, e. g. Egyptian Arabic, seem to be good candidates for the subtypes (lab) or (2ab). In order to avoid too great complexity I will be content, in this paper, to examine the four major possibilities enumerated above, i.e. (la), (Ib), (2a) and (2b). In the case of aspect systems, there should be two other possibilities: in one kind of system the perfective form(s) would be the — formally as well as semantically — unmarked pole, whereas the imperfective forms would be the marked term of the opposition. This would be the case in classical Arabic and in a great majority of Creole languages. The other possibility would be a system where the imperfective form is (formally and semantically) unmarked. This seems to be the case in Mandarin Chinese, Russian, etc.3 We should thus have the four major types with different subtypes, as in Table 1. The interrelations and the interferences forming the object of the following considerations concern one or more of the four types of systems found in Table 1 (la, Ib, 2a, 2b). 3. Particular forms of verbs may be perfective or imperfective, i. e. the point of view of the speaker is inside the represented process in the case of imperfectivity, and outside when there is perfectivity. In the perspective of perfectivity the event is seen from the outside, i. e. as a whole. What is true for particular verbs and the sentences they are contained in can apply to more complex signs, e. g. to entire passages of a text as well. In this context we distinguish two types of narration which I have called "narration with episodesignals" and "narration with iteration-signals" (Raible 1971). A passage beginning with one day or on May 18th, 1888 etc. is, by and large, a singular event. A passage of this kind constitutes a "frame of temporal uniqueness" for the story as a whole. We could say as well: the entire passage is "per-

198

Wolfgang Raible

Table 1. Types of tense and aspect systems (la) a pure aspect system (laa) based on verb lexemes expressing by themselves Aktionsarten (laaa) perfective forms being morphologically and semantically unmarked (laab) imperfective forms being morphologically and semantically unmarked (lab) based on verb lexemes which are neutral with respect to Aktionsarten (laba) perfective forms being morphologically and semantically unmarked (labb) imperfective forms being morphologically and semantically unmarked

(Ib) a pure tense system (Iba) based on verb lexemes expressing by themselves Aktionsarten

(2a) an aspect system by priority, combined with a tense system (2aa) based on verb lexemes expressing by themselves Aktionsarten etc. (2ab) based on verb lexemes which are neutral with respect to Aktionsarten etc.

(2b) a tense system by priority, combined with an aspect system (2ba) based on verb lexemes expressing by themselves Aktionsarten

(Ibb) based on verb lexemes which are neutral with respect to Aktionsarten

(2bb) based on verb lexemes which are neutral with respect to Aktionsarten

fective". On the other hand, the passage could be introduced by signals of iteration (every Sunday, sometimes, by and now, whenever my uncle Peter came home etc.). In such a case the entire passage, i. e. the complex sign consisting of a whole series of sentences, is "imperfective", "iterative", "repetitive", etc.4 An example will show the difference. The following passage is from the beginning of Marcel Proust's A. la recherche du temps perdu (Paris: Gallimard, 1954). The same event is narrated first with iteration-signals and some pages later with episode-signals and the corresponding tenses (the italics are mine): Quelquefois quand, apres m'avoir embrasse, [ma mere] ouvrait ma porte pour partir .... Mais ces soirs-/a, ou maman en somme restait si peu de temps dans ma chambre, etaient doux encore en comparaison de ceux ou il y avait du monde a diner et ou, ä cause de cela, eile ne montait pas me dire bonsoir. Le monde se bornait habituellement a M. Swann, qui, en

Types of tense and aspect systems

199

dehors de quelques etrangers de passage, etait a peu pres la seule personne qui vmt chez nous a Combray, quelquefois pour diner en voisin, (plus rarement depuis qu'il avait fait ce mauvais mariage, parce que mes parents ne voulaient pas recevoir sa femme), quelquefois apres le diner, a 1'improviste. Les soirs ou, assis devant la rnaison sous le grand marronnier, autour de la table de fer, nous entendions au bout du jardin, non pas le grelot profus et criard qui arrosait, qui etourdissait au passage de son bruit ferrugineux, intarissable et glace, toute personne de la maison qui le declenchait en entrant "sans sonner", mais le double tintement timide, ovale et dore de la clochette pour les etrangers, tout le monde aussitot se demandait: "Une visite, qui cela peut-il etre?" mais on savait bien que cela ne pouvait etre que M. S wann; ma grand'tante parlant a haute voix, pour precher d'exemple, sur un ton qu'elle s'efforcait de rendre naturel, disait de ne pas chuchoter ainsi; que rien n'est plus desobligeant pour une personne qui arrive et ä qui cela fait croire qu'on est en train de dire des choses qu'elle ne doit pas entendre; et on envoyait en eclaireur ma grand'mere ... (p. 13/14). Now the corresponding passage with episode-signals: Mais une fois, mon grand-pere lut dans un journal .... [Ma grand'tante] blama le projet qu'avait mon grand-pere d'interroger Swann, le soir prochain ou il devait venir diner .... Nous etions tous au jardin quand retentirent les deux coups hesitants de la clochette. On savait que c'etait Swann; neanmoins tout le monde se regarda et on envoya ma grand'mere en reconnaissance. ... "Ne commencez pas a chuchoter", dit ma grand'tante. (p. 21 / 22) Perfectivity and imperfectivity of text passages mean that every language should be able to express these aspect categories on the level of texts or textpassages. 4. I now turn to the particular cases of interrelation or interference between the categories in question. The first case is an interrelation between Aktionsarten on the one hand and tense/aspect categories on the other, in a system of the type (2b). There is a natural affinity between the imperfective aspect and verbs of a non-transformative Aktionsart (e. g. to be in flower). There is, on the other hand, a natural affinity between the perfective aspect and verbs of transformative Aktionsart (to shut the door, to open the window).5 The French verb chasser is a verb of the non-transformative type. The French verb attraper should be transformative as regards its Aktionsart. If we combine chasser in the past

200

Wolfgang Raible

tense with the imperfective aspect, and if we continue to symbolize Aktionsarten with square brackets, we would have the following structure for // chassait: [FTP] T [FTP]... This is to say: the event is going on, afterwards, it is as non-transformative as it was before. No "end", no transformation is apparent. The other "natural" combination would be the combination of the transformative verbs with the perfective aspect, for instance the verb form // attrapa. The event symbolized by [nPTP] will be transformed into /[nPTP]/ (the slashes before and behind the bracketed structure symbolize the fact that the event is seen from the outside, i. e. that it is perfective). Seen from the content-side: in the very moment Achilles reaches the turtle, the event has come to its — successful — end. What will happen to the unusual combinations? Speakers of natural languages profit from the possibilities of combination offered by grammar; therefore we will have non-transformative verbs combined with a perfective aspect as well as transformative verbs in an imperfective form. A first example will be the non-transformative verb chasser in a perfective form, i. e. the form il chassa. Thereby the Aktionsart symbolized by [FTP] becomes /[FTP]/. In other words: the monotonous event of hunting, not interrupted by a transformative change, is represented as closed; the hunting comes or has come to an end — and this may change the familiar meaning of chasser in cases like // chassa, 'he was on the hunt', or il chassa I'ennemi, 'he chased the enemy', where the meaning becomes transformative. Speakers of French very often make use of this possibility, saying for instance // se taisait as opposed to /'/ se tut ('he was silent' vs. 'he fell silent'). Another example would be Lorsque f appris que ma voisine avait une compagne, je fus la voir (Bernardin de Saint-Pierre) as opposed toj'efais che^ eile or Puis eile mit la chienne a terre, se leva et fut s'accouder au balcon (Roger Martin du Gard, Les Thibaulf) as opposed to eile etait au balcon. In the same sense we would say in Spanish Juan se fue a la ventana 'John went to the window'. In the perspective of pluperfect, we would have L'un d'eux, qui avait ete jusqu'a la maison, cria aux autres; J'ignorais qu'il avait ete les rejoindre; Elle avait ete s'asseoir, machinalement, sur ce canape-lit; La garde avait ete prendre le stethoscope; etc. (all examples from Martin du Gard).

Types of tense and aspect systems

201

The other case would be the transformative verb used in an imperfective form, for instance // attrapait le papillon. If we attribute to attraper the Aktionsart structure [nPTP], we would have [ηΡΤΡ] Τ [ηΡΤΡ]... This would give us different possibilities of interpretation. There could be someone chasing the butterfly without reaching it; in other contexts there would be somebody catching it again and again. The last possibility is self-evident in cases in which the text containing such a sentence constitutes a frame of iterativity. If we describe the events happening every day in a certain Gallic village, we could say perfectly well: Chaque jour, Qbelix attrapait deux sangliers. It is clear that this kind of interference is not only possible in the Romance languages; it should, rather, be possible in all languages of type (2b) which do not, like Russian, dispose as often of two different verb forms for transformative and non-transformative events. The same procedure functions for instance in Finnish. The verb puristaa 'to press' means in its imperfective version 'to keep (firmly)', 'to keep against oneself, whereas, in its perfective version, it can mean 'to take somebody in one's arms'. Therefore, ban puristi Kristiinaa (partitive) syliss n (inessive) means 'he held Kristina in his arms', 'he embraced her', whereas the perfective version ban puristi Kristiinan (accusative) syliinsa (illative) means 'he took Kristina in his arms' (examples from Raible 1976: 26 ff.). 6 5. A second type of interference concerns the relation between aspect and assertion, i. e. affirmation and negation. Let us assume a transformative event seen from the perfective point of view, for instance Achilles chasing the turtle and just catching it, so that we can say: // I'attrapa. Let us assume, now, that this transformative event has to be negated, i. e. that it did not take place. In this case, in French, we would say: Acbille n'attrapa pas la tortue — which is perfectly good French. Nevertheless, there is a certain — statistical — relation between aspectual perfectivity on the one hand and affirmation/negation on the other. Negated instances of perfective aspect are relatively rare in Romance languages. The same relation seems to hold between the perfective aspect and negation in Russian, although this language, too, makes use of the combination between negation and perfective aspect. In this case the gap between trial and result is highlighted.7 On the other hand it is not possible in Finnish to combine negation with the perfective aspect. The "logical" reason is evident. If we express the

202

Wolfgang Ruble

transformative content of a verb with perfective aspect, this means that the event has come to an end. It is precisely this which is denied in the case of negation. A further example of the same kind seems to be a "parier vehiculaire" in the Chad region described by Hagege (1973). In this case the verb may be either negated or aspectually marked, but not both at the same time.8 In this context the scope of negation is an important factor. It seems to vary from language to language. Therefore negation can be combined with perfectivity in some languages, while in others this combination is not possible. Nevertheless, it would be worthwhile to examine the relation between aspectual perfectivity and the possibility of negation in individual languages, not only in languages belonging to type (2b), but also in languages representing the types (2a) and (la). 9 6. A further phenomenon of interference concerns only languages of the type (2b), i. e., languages with a primary tense system and a secondary aspect system. These languages allow not only the "normal" combination of aspectual perfectivity and temporal non-present as well as the combination of the present tense and aspectual imperfectivity, but also the "unusual" combinations of present tense and perfectivity as well as between non-present and aspectual imperfectivity. Among these combinations the one between nonpresent and aspectual imperfectivity is not problematic. Far more interesting is the combination of present tense and aspectual perfectivity. If we try to imagine a — transformative or non-transformative — event in the present tense combined with aspectual perfectivity, we shall presume that this combination would not be possible, anyway. If the event is perfective, it has come to an end, and when it has come to an end it can hardly be a present event. An example will help us. In Finnish, I have the choice between luen ('to read', present tense) kirjaa ('book', partitive), English Ί am reading the book', on the one hand, and luen kirjan (accusative) on the other hand. Luen kirjan would be literally Ί read the book' + perfective aspect. The perfective meaning of 'to read' is 'to come to an end with the process of reading'. Now, whenever I express by means of a verb in the present tense that I have completed the process of reading a book (which has not been read as yet but is being read by me at the moment), it expresses a promise or statement which will only become true in the future. This is why in Finnish the combination of the present tense and perfective aspect is the equivalent of a future tense — there is no "true", morphological future

Types of tense and aspect systems

203

in the Finnish verbal system. Future meaning is expressed either, if there is an object in the sentence, by means of the perfective aspect or, in the other cases, by means of the verb 'to be' + present participle, for instance ban on tuleva 'he will come'.10 The same applies to Russian: ja procitaju (perfective aspect + present) knigu means Ί will read the book'; with the imperfective version of the verb 'to read' the sentence has a present meaning: ja citaju knigu.

This also means that Russian has to have special forms for the future of imperfective verbs: ja budu citat' knigu. n

The forms chosen to express future tense seem to be a good indicator of the existence of a primary tense or a primary aspect system. Primary tense systems have the possibility of expressing future tense by the combination of present tense and perfective aspect. 12 Hence, in our analysis, the Russian system is a primary tense system combined with a secondary aspect system — a system in which nevertheless, peculiarly enough, the category of aspects is lexicalized in a very systematic way.13 That is to say that verbs have two — often entirely different — forms, one with perfective, the other with imperfective meaning. That is why speakers of Russian have to have two ways of expressing future tense, one with perfective verbs and one with imperfective verbs. If aspect were not to be, so to speak, lexicalized in Russian, the combination between present tense and a perfective marker would be sufficient for all verbs. In another way, the same is true of Finnish. Since the morphological expression of perfectivity is restricted to sentences with an object, a separate solution for the expression of future tense in intransitive sentences has to be found. 7. The last type of interference I shall deal with in this paper refers especially to languages of the type (2a), i. e. languages with a primary aspect system and a secondary tense system. Amongst these are many Creole languages. Creole languages, as other languages too, can have verbs presenting the processes of the events they express as transformative or non-transformative. In the case of Creole languages, special interest is due, not to the borderline between transformative and non-transformative verbs, but rather to the opposition between static and dynamic verbs (see Appendix II). The device

204

Wolfgang Raible

I shall describe presumably forms the basis for phenomena which have been observed particularly by Bickerton (1981). In Creole languages with a primary aspect system, the verb form which is morphologically least marked is, as a rule, perfective. If we choose a verb which could be transformative, for instance the equivalent of French arriver, in Martinique Creole or in Guadeloupe Creole, we would use rive. In the third person singular we would thus say / rive. This may be illustrated with a short passage from Sylviane Telchid's Ti-Chica (Paris: Editions Caribeennes, n.d.: 79): Ijenn eve Ijeni te fo lekol toubolman: Ijeni Eugene et Eugenie etaient forts a 1'ecole extremement: Eugenie te toujou premye, Ijenn te toujou de^yem. etait toujours premiere, Eugene etait toujours deuxieme. Lew-gade, onjou Ijenn pote premye plas, Puis, un jour Eugene remporta la premiere place, Ijeni pote decent plas. Le tifi-la ba Eugenie remporta la deuxieme. Quand la petite fille donna ä Misye Louiva papye a—y, papa li di konsa: ... Monsieur le Roi papier son, son papa lui dit ceci: ... In order to aid comprehension, I have added a French interlinear version. Te is a past marker. The unmarked form of the non-stative verbs has a past meaning by itself (the instances are marked with italics). If we understand the verb rive — which, by virtue of its lexical meaning or Aktionsart, should be transformative — as aspectually perfective in its morphologically unmarked form, it would mean in our — tense-dominated — Central European view that the event of arriving has come to an end, since it is perfective. Consequently, we tend to interpret verb forms such as i rive as instances of past time. If, in our Central European view, this verb has to have a present meaning, we would have to use an imperfective marker which, in the case of Martinique Creole, would be the particle ka. Thus we would have / ka rive, meaning 'he/she is arriving'. Whereas in a language of the type (2b) (with a primary tense and a secondary aspect system) future time reference becomes accessible by means of a combination of the — existing — present tense and perfective aspect, in languages of the type (2a) (and, presumably, also languages of the type [la]) future tense should be expressed by a special form or a special particle. (Nonetheless, exceptions should be possible in such cases in which not the perfective, but the imper-

Types of tense and aspect systems

205

fective aspect form is the unmarked one [types laab, 2aab]. It should even be possible that future time reference is in the scope of every imperfective form which is not marked with respect to tense. The imperfective form could also assume the function of future.) With regard to Martinique Creole, the future has the form / kal rive. In this case, kal is probably a combination of the imperfective particle ka with a form whose French etymon is aller. Let us look now, instead of at the verb rive with its transformative Aktionsart, at a verb which is characteristically non-transformative (or, in this case, also Stative), for instance the verb tint 'to have'. We could, for instance, say / tini 50 a 'his age is 50'. With regard to the Aktionsart, the person was fifty years old before and he/she is still fifty years old after I have made this statement. If this non-transformative event becomes perfective by virtue of the morphologically least marked form tini, it might be considered as complete or closed, but, in contrast to / rive, it would, so to speak, be closed only once. Rive is closed already by virtue of its transformative character and it would be closed once more by the aspectually perfective form, whereas / tini 50 ä would be closed only "grammatically" by the perfective aspect. Thus it is understandable that, viewed with the European eye, the form / tini would, on account of its Aktionsart, have, rather, a present meaning ('she/he is 50 years old'). This would mean that the non-transformative verbs of such languages should occur less frequently with the imperfective particle which eliminates the past value of transformative verbs used in the perfective zeroform. (Nevertheless, this argumentation is problematic: Is the transformative or non-transformative character of a verb always the property of the verb itself or is it not, rather, produced by the context?) This line of thought would make a further phenomenon clear. In the combination with a transformative verb, the past particle te — which can be combined with all verbs — means that, in European understanding, this verb carries the meaning of a plusquamperfectum whereas the same past particle te, used with a stative verb, would pass on the information of "simple past". To state an example: the distinction between / te rive and / te tini 50 a would be a distinction between 'he had arrived' and 'he was 50 years old', and not 'he had been 50 years old'. The interference between the transformative or non-transformative character of a verb and aspect categories does not play the same role in all Creole languages. The analysis of Nubi, an Arabic-based Creole, shows that there is no interference of this kind. This means that, as has already been mentioned at the beginning of section 2, in some languages single verbs represent special types of modes of action whereas in others the Aktionsart-meznmg only

206

Wolfgang Raible

results from the linguistic context, the verbs being neutral. (Nubi is a Creole language evolving from type [laba] into [2aba].) The considerations which concern languages of type (2a) (primary aspect system combined with a tense system) should be completed by the perspective mentioned briefly in section 3. It has to be taken into account that, on the one hand, a text passage which, as a whole, is perfective (frame of temporal uniqueness) and, on the other, a text passage which is, as a whole, imperfective (marks of iteration), bring about parameters which can modify the meaning of particular verb forms. For Creole languages, it should additionally be taken into account that an interference between perfectivity and negation as well as a reduction of the tense-aspect possibilities in subordination are possible. 8. The objects of my consideration of the interference between aspect, tense, negation, and Aktionsarten above have all been of the types (2a) and (2b), type (2b) being very common among the languages of Europe. These considerations should be completed by interferences in systems like (la) and (Ib) and, in the case of (la), in the subsystems (laaa), (laab), (laba), and (labb) (see Table 1). Nevertheless, it seems to me that these preliminary thoughts already explain numerous phenomena by means of what we might call the "inner logic" of such systems. On the other hand, when analyzing particular languages we should be able, by observing certain kinds of phenomena, to draw (careful) conclusions regarding the attribution to one of the major types (la), (Ib), (2a) or (2b). If, for instance, we notice that certain verb forms are rarely or even never combined with negation, there could be an aspect system. If we discover a future meaning of seemingly non-future forms, it could be an indication for the existence of a system of the type (2b). If, for instance, with regard to the Creole Nubi mentioned before, the analysing linguist observes that the morphologically unmarked zero-form or basic form might mean past as well as present (but never future) tense, this could be an indication for a system of type (la) or type (2a) (in these types the zero-form is not, per se, present tense; it therefore cannot acquire a future meaning if it were combined with a perfective aspect). It might also be the case that certain types of tense/ aspect systems (especially the types [la] and [2a]) favour the evolution of serial-verb constructions.14 9. It goes without saying that a more thorough analysis of particular languages could in some way blur this still relatively clearcut typology. As in many languages we find so-called "split ergativity", split tense/aspect systems could, for instance, exist. The great majority of such "blurring"

Types of tense and aspect systems

207

effects would possibly result from diachronic and ontogenetic transformation of language systems. From the fundamental investigations made by the Piaget school in the 1970s, it should be clear that tense is a relatively late human acquisition and that children (until the age of seven or eight) use the verb forms of the adult language predominantly as aspect forms. Thus, even children learning languages which are primary tense systems first become acquainted with an aspect system. With regard to diachrony, it is not impossible that there is a periodic change between tense-dominated and aspectdominated systems just as there is, for instance, a regular change between tense-dominated and modally-dominated future forms. But even change and transitory states are more easily comprehensible if we have an insight into the inner logic of such systems. And it should be clear that this inner logic has to do with tense, aspect, Aktionsarten, even negation, and, perhaps, some further major constituents.

208

Wolfgang Raible N

Ό .'S "u *-*^ (Λ

Λ

Ο U

s et

G U ι u >^ [i, Λ 0 W U ^ t> G G ''a Jf C . u «n

"

-G

Q-

N

"w

Ja U

G a

|

1o

|

s

GO

,Ξ

1

(Λ

es

o OH

Ό υ

υ (Λ

IH

a2 (ή

rt f!

ο-

α.

α 'S H

C/J

υ

OJ

|| g o ι

Ί

*^

*H

£

trt

G ^

ΐ* —'

T

(L) b. > .£

ΙΛ

•a c -r· > .«o .ti 1>

's 3 3 H S C

V

c c t 3 S 2

' "u i 'S

(U

OJ >

ϊ-ι

υ

1 2

υ

> 1 - 'i 1 1

•a S .2 3 2 .S g S -ο Ό ε « *J

S

u

in

IH

C

3 ^ »-M

—u^

>

'4-*

i*i t-l

u

U OH J

H

>, 'S

(J 4-»

15 · O «

J„.

G

~tT (Λ

dj «

C

rt

ΰ

" >, ^ -

Iflfu t |

^C

•^ cli, o 33 ^ « f> -S 'S -O "^

-o u "H.

-
> dJ ur o -S ? -i"S a § .S 2 ^ 1

iL)

£ H « ΰN

jrt u

4-1

«

^

Duration

u

T

Movemen

S -3

S

" · " ^ S S

'u 3 a o s cs

O

θ

OH (Λ rt

l·!

^

S-

~ΰ a

Λ

0

4J

u ω

U

j-^u°^l sS ^^ o w

M 3 rt

υ Ά υ

aj

(LJ

00

55

3

v,

H t_i o

•S -¥ 5.

3

'S ^ S S ΰ u

C/5

! §

S

>, SS il -^ ^ - S, >> ° ^ ""

'ΪΓ >

'C

ς«

c

2 -SH> («

υ

(Λ >-.

c

S ^ ^^ -0 4-. Λ S « G Λ

«

O tu

ε.

'So

c/T

ω

Z

£>

I

«s 4-1

G

UL,

C^