Romance Objects: Transitivity in Romance Languages [Reprint 2011 ed.] 9783110919837, 9783110179606

Citation preview

Romance Objects

W DE G

Empirical Approaches to Language Typology 27

Editors Georg Bossong Bernard Comrie Yaron Matras

Mouton de Gruyter Berlin · New York

Romance Objects Transitivity in Romance Languages

edited by Giuliana Fiorentino

Mouton de Gruyter Berlin · New York

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

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

I S B N 3 11 0 1 7 9 6 0 1 Bibliographic

information

published by Die Deutsche

Bibliothek

Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at < h t t p : / / d n b . d d b . d e > .

© Copyright 2003 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Printed in Germany.

To Raffaele Simone, who organized and hosted the Conference, actively contributing to the discussion with his thought-provoking remarks

Contents

Preface

ix

Introduction

xi

Section 1: Object marking in cross-linguistic perspective What is an object in a crosslinguistic perspective? Gilbert Lazard

1

Nomina] and / or verbal marking of central actants Georg Bossong

17

(In)transitivity and object marking: some current issues Michela Cennamo

49

Section 2: Transitivity and objects in Romance languages The object clitic pronoun in Italian: a functional interpretation Alberto Nocentini

105

Prepositional objects in Neapolitan Giuliana Florentine

117

From subject to object and from object to subject: (de)personalization, floating, and reanalysis in presentative verbs Peter Koch

153

Changes of valence and their effect on objects Heidi Siller-Runggaldier

187

Transitivity and grammaticalization of obj ect. The struggle of direct and indirect object in Spanish Concepcion Company Company

217

viii Contents Non-diachrony of ditransitivity in Spanish Rosa Maria Ortiz Ciscomani

261

Argument structure of perception verbs and actance variation of the Spanish direct object Eugeen Roegiest

299

Subject Index

323

Preface

This volume brings together the papers read at the international conference on Romance Objects organized by the Linguistics Department of the Universitä Roma Tre in December 2000. It is characterized by a striking uniformity of approach, which is functional, and of methodology. The various case studies regarding the object focus on the syntax/semantics and syntax/pragmatics interfaces. The common denominator of the ten enquiries is the identification of the object category, the DO in particular, in Romance languages; 1 at the same time some of the contributors relate the specific topic to more general questions of linguistic typology. Some of the essays are based on the analysis of data from a corpus and present a diachronic picture of the evolution of the specific topic investigated. Thus this volume is addressed not only to scholars interested in the Romance languages but also all those who study the object category in a cross-linguistic perspective. The volume's existence is entirely due to the commitment of the Linguistics Department of the Universitä Roma Tre and in particular its Director, Raffaele Simone, who sustained both the organization of the conference and the publication of the volume. I wish to express my warmest thanks to Raffaele Simone not only for his support of this publishing enterprise, but also for a decade of stimulating discussions and exchanges ranging across the whole spectrum of general linguistics and the evolution of Romance syntax. I also wish to thank all those who took part in the conference, which was constantly enlivened by thought-provoking questions, interventions and discussions. The papers were enriched by the discussions that started during the two days of our encounters, and many of these reflections found their way into the essays presented here. The conference benefited from the constant attention and acute observations of Gilbert Lazard, who provided stimuli for the discussion of many of the papers. I wish to put on record here recognition for all that he has done throughout his distinguished scientific career to further the topic of our conference; many of the contributors took the results and theoretical advances of his works as their starting-points.

χ

Preface

Finally, a word of sincere thanks to Bernard Comrie, for his suggestions and comments to the whole volume and to dott. Mark Weir for his painstaking work of revising or translating the contributions to the volume. Rome, 26th June 2003

Giuliana Fiorentino

Introduction

Giuliana Fiorentino

The category of the object, together with that of the subject, constitutes one of the basic grammatical relations, with a vast specific bibliography (covering studies in theory of syntax and theory of grammatical relations).2 Yet there are still problem areas in the analysis and description of the object category, notably the definition of the category and its morphosyntactic manifestation. The problem of definition becomes particularly pressing when, in view of the variability of object formation, we try to settle on a universally valid approach, like the one evolved for typology, based on definitions which are neutral and as generic as possible, not language-specific, in order to establish linguistic categories. For the object category is not merely formal and structural, but also has functional correlates. Dryer (1986), for example, solves the problem of defining the object by considering two contrasting pairs: Subject / Object and Direct Object / Indirect Object, both relating to semantic and pragmatic feaures. With respect to the S / Ο opposition, Dryer observes that while in Subject / Object (or Nominative/Accusative) languages the opposition grammaticalizes aspects of the pragmatic function of discourse and in particular correlates with degrees of topicality (greater for S and lesser for O), in the Ergative / Absolutive languages it grammaticalizes the semantic roles, so that the ergative correlates with the Agent and the absolutive with the Patient.3 In terms of the opposition between Direct and Indirect Object, there are two types of language: either Direct Object / Indirect Object or Primary Object / Secondary Object. In the former, there is the same marking for the single object in monotransitive structures and the inanimate object in ditransitive structures, while a different case is used for the animate object in such structures; whereas in PO / SO languages there is the same marking for the monotransitive object and the animate object in ditransitive

xii

Giuliana Florentine

structures (Primary object, secondary topic) and a different one for the inanimate object in ditransitive structures. In this case too different aspects are grammaticalized. DO / IO type languages grammaticalize the semantic case, while PO / SO ones grammaticalize pragmatic functions (PO is more topical, marking the secondary topic, while SO is less so and may even represent the non-topic) (Dryer 1986: 840-842). 4 A consideration of pragmatics and semantics was already present, although less highly organized, in the approach of Givon (1979: 143-144). For him the case marking system in a language responds to the problem of codifying simultaneously both the semantic and pragmatic roles of the participants in a communicative event activated by the verb. Different languages propose different typological solutions, and even within one and the same language there may be different solutions depending on the semantics and pragmatics activated by a particular verb. This accounts for the cross-linguistic and intra-linguistic variability of the DO marking. According to Givon (1979 and 1984), the function of the object is to mark the secondary topic (1979: 206; 1984: 170), so that the pragmatic value of the object is that of Secondary topic. 5 He motivates this affirmation by verifying the marking of a nominal in ditransitive constructions. His test shows clearly how the choice between marking the animate actant as DO or IO depends on the degree of its topicality (greater in DO and lesser in IO). Consider the following examples (Givon 1984):

(1) (2) (3) (4) (5) (6)

who did Mary give the book to? (PAT-topic, DAT-focus) she gave the book to Bill (PAT is DO) ? she gave Bill the book (DAT is DO) what did Mary give to Bill? (DAT-topic, PAT-focus) she gave him a book (DAT is DO) ? she gave a book to him (PAT is DO)

(2) and (5) are respectively the most acceptable answers to the questions put in (1) and (4) because the topic is marked as the DO. A further indication of the topicality of the DO is seen in the fact that when a language does not distinguish between the semantic marking of the Patient and the pragmatic marking of the secondary topic (because both come under the morphological case of the DO), and when moreover there is

Introduction

xiii

more than one object in the same clause, the most topical precedes the others. In quantitative terms the association between secondary topic and the semantic role of the Patient is clearly highest, so that the DO has come to be identified with the semantic case of the Patient and the pragmatic case of the secondary topic (in the same way as the semantic case of the Agent generally coincides with the pragmatic case of the primary topic and the morphological case of the SUBJ). However, the sequence Patient = Secondary topic = DO does have its exceptions: verbs whose semantics involve a Patient can also codify it as a subject: e.g. to suffer, to fall, to grow up (defined as objectless verbs in Givon 1984: 94). The second problem, closely connected with the first and no less crucial to the discussion of the Object category, concerns its morphosyntactic formation. For many years the object was viewed as the syntactic manifestation of the semantic case of the Patient. At a certain point the standard marking of the Patient as object was called into question by at least two other types of evidence, intra-linguistic or cross-linguistic. I am referring to the non-canonical marking of objects on the one hand and the "discovery" of ergativity on the other.6 The "non-canonical marking" of objects indicates the phenomenon by which, in languages in which the object is normally marked by case, in some contexts objects are marked by a different case (cf. in Russian the Accusative-Genitive alternance; for the non-canonical marking of S and Ο see the recent Aikhenvald et al 2001; a lot of data on alternance in object marking is reported in Moravcsik 1978). Ergativity, on the other hand, called into question the classic understanding of case systems, based above all on languages of the Nominative/Accusative type, and showed the plausibility of a different 'association' between semantic roles and grammatical relations (A vs. Ο and S rather than A and S vs. O).7 A third area of the object category requiring investigation is the richness of phenomena, of various kinds, found in this grammatical area. Suffice it to mention so-called Differential object marking, clitic doubling and incorporation; various kinds of overlapping of the object with different semantic roles (object-Goal raggiungere la riva [reach the shore], objectLocative habiter Paris [live in Paris]); and conflicts between various types of object (double object). All go to confirm the invitation to think in terms of an internally articulated "object zone" (Lazard 1993, 1994).

xiv

Giuliana Florentine

Finally we must not forget that the functional analysis of the object has owed a lot to work on transitivity. Hopper/Thompson 1980 (a true milestone in the progress of work on transitivity) showed the abundance of syntactic, semantic and pragmatic properties which play a part in the formation of the fundamental grammatical relations of subject and object within the transitive construct. 8 Consequently the analysis of the object has to cope with notions which are themselves still being defined, such as prototypical transitivity and prototypical transitive construction. The theoretical debate on the problems I have outlined (definition of the object category, description of morphosyntactic variability at the intralinguistic and cross-linguistic levels, complexity of the "object zone") is the background to the essays published in this volume. The first contribution deals with the definition of the object, and the subsequent ones go into the complexity of its morphosyntactic formation. Some of them look at Differential object marking, the relation between nominal and clitic DOs, the diachronic exchange between S and DO over time, and the interaction between DO and IO in ditransitive constructions. Others tackle questions concerning the relationship between verbal semantics and DO (DO and verbs with valency changes which entail alternance between transitive and intransitive constructions, verbs of perception and DO). The volume is divided into two sections. The first groups the more wide-ranging essays which do not necessarily limit their investigation to the Romance languages. The second contains studies of particular aspects of the formation of the object in Romance languages. In his contribution Lazard proposes a methodology for arriving at a cross-linguistic definition of the object as a linguistic universal. 9 He starts from the assumption that whoever adopts a typological approach needs clear-cut definitions of the linguistic categories which, even more importantly, can be applied to languages which may be remote and very different. To achieve this objective Lazard begins from problem cases, the first of which is the object category in Persian. This language admits constructions with two or even three objects, only one of which is marked by a postposition -rä, and this functions only partially like analogous cases of Differential object marking. Persian gives us a glimpse of a hierarchy of objects. He goes on to consider problem cases in other languages, including Romance languages (differential object marking, differential object agreement, cognate objects, measure complements, H-actant, two-objects, second absolutive in ergative constructions, object with intransitive verbs). All these phenomena, subjected to a brief review by the author, show that at

Introduction

xv

the cross-linguistic level there are at least two types of object: close and distant objects. Once this distinction has been established, it becomes apparent that the "problem cases" are merely a manifestation of this duality: this gives rise to the invitation to consider the object category as a multi-dimensional "object zone". In drawing his conclusions Lazard returns to the initial concern to formulate a definition of the object which will be valid in cross-linguistics. He comes up with one which is both formal and semantic, by means of a rigorous methodology based on three successive "steps": a) identify in the language in question the prototypical action, i.e. an action carried out by an agent on a patient; b) identify the syntactic construction which expresses that prototypical action, necessarily a bi-actant construction; c) consider as object the second actant of the 'major' bi-actant construction, whatever the process expressed by the verb. The definition of prototypical object, which coincides with the "close object", is thus arrived at with rigour and in a form available to cross-linguistics. Bossong's essay too deals with a problem of fundamental interest: why does the marking of the fundamental grammatical relations (S and O) show asymmetry in world languages, so that S is marked in the verb and Ο in the noun. This question is investigated through a combination of theoretical reflections and empirical data taken from a range of languages. Romance languages are central to his considerations but by no means the only ones treated. The author succeeds in giving a concise and thorough account of a lot of data, some of it concerning the historical evolution of the marking system of S and Ο in Romance languages, showing how the typological approach he advocates accounts satisfactorily for the empirical data discussed. Asymmetry is explained in terms of language's logical properties: the two basic grammatical relations have opposing characteristics, and in predication S is fundamental so that the predicative act is assertive while Ο is not. Thus if a language codifies only one actant on the predicate, this is bound to be the first actant, S. The essay by Cennamo offers an overview of current views on transitivity and the object based on a broad series of cross-linguistic examples of the formation of the object. The various phenomena of noncanonical marking of objects and the explanations for this show how an event-structure type approach is particularly appropriate for arriving at a description of the object category.

xvi

Giuliana Fiorentino

The first of the studies dealing with single aspects of the formation of the object in Romance languages is by Nocentini, looking at the degree of grammaticalization of Italian object clitics when in the same sentence they are co-referential with a full NP. The analysis concerns the grammaticalization of clitics in cases of dislocation (both leftwards and rightwards) in modern Italian, and covers syntactic, intonational and pragmatic considerations. Having excluded the possibility of the presence of the clitic constituting a powerful grammaticalization, Nocentini shows how the two types of dislocation fulfil a pragmatic function which is fundamental in that it marks the non-prototypical (or non-subject) topic. However, dislocation alone does not account for the presence of clitics: this depends on discourse strategies which control word order and the system of references. Finally the author considers another level of activation of the clitic: in leftwards dislocation the presence of the clitic is controlled at sentence level by syntax, but in rightwards dislocation by factors such as the communicative aim, shared premises and so on. Thus Italian (and Romance) clitics represent an intermediate class between pronouns and affixes, governed by the pragmatic category of the topic, which explains why they are still only weakly grammaticalized. The essay by Fiorentino looks at objects in a dialectal variety of Italian, investigating the non canonical object marking in Neapolitan. This is a case of what Bossong calls differential object marking. The study covers an extensive corpus of texts and illustrates the diachronic evolution of the prepositional object in Neapolitan. Some hypotheses concerning the formation of the prepositional object are discussed and to some extent rejected, and a typological solution is proposed (the development of the marker corresponds to the typological evolution of the Romance varieties which tend to manifest the Nominative/Accusative type). The essay goes on to compare the results obtained from Neapolitan with the two languages invariably considered as the classic cases in the Romance area of differential object marking (Spanish and Rumanian). This comparison brings out both affinities and differences in the formation of Differential Object Marking in the three cases, and for Neapolitan shows that differential object marking tends to respect the animacy and definiteness hierarchies. The contribution of Roegiest introduces a handful of studies focusing on Spanish. He looks at the non canonical marking of the object as a manifestation of the assimilation of the DO with the 10. When it depends on verbs of perception, a non prototypical (non Patient) DO is marked with

Introduction xvii a and considered as an 10. Drawing on analytical data, the author shows how the alternation does not merely depend on semantic features or the internal structure of the SN (as has traditionally been asserted), but also on the semantics of verbs of perception and on the structure of the complex phrase in which the DO occurs (see the discussion of the verbs of perception followed by a subordinate in the infinitive or gerund). He also gives a clear analysis of perception, distinguishing between hearing and seeing and between direct and indirect. In particular he confirms assertions made elsewhere that perception through hearing seems to be less direct. This explains why the object in such cases is perceived as a source: it is accorded greater autonomy and figures as 10 more frequently than in cases of direct, visual perception. The essay by Siller also investigates verbal semantics, with an abundance of data. It discusses valency change and in particular the passage from transitive to intransitive in Italian verbs, to detect polysemy. Differences in meaning seem to derive from the different degrees of transitivity manifested by the verbs. The author subdivides the verbs into categories sharing systematic patterns of change of meaning associated with changes in argument structure. Following a thorough and acute analysis she observes that the changes in valence in some cases require the projection of a different frame, while in others remain within the same frame. She then sets out to verify whether in each case the change in meaning is a case of polysemy. She adopts the notion of semantic continuum proposed by Blake 1993, which includes polysemy. She draws two important conclusions about the changes of valence investigated: a) they are systematic and paradigmatic, not occasional or contextual; b) they depend on the verb, can be predicted and involve the autonomous lexical meaning and semantic traits from the sphere of transitivity. Koch's contribution starts from some impersonal constructs in order to investigate the diachronic "exchange" between subject and object. He focuses on the Italian presentative construct c e and the French verb falloir. The actant projected by these verbs, defined following Lazard as Actant H, occurs in an intermediate position between the Prototypical Subject and the Prototypical Object. The process of exchange between subject and object is interpreted as a case of syntactic reanalysis and is also connected with the depersonalization of the subject. The author compares the situation found in Italian and French with that of other Romance and Germanic languages and also mentions cases in which the opposite change occurs, from object to subject, with personalization. He concludes that exchanges between subject

xviii Giuliana

Fiorentino

and object cover a range of situations: true reanalyses, syntactic floating and reanalysis linked to metonymic lexical change. The essay by Company-Company investigates a series of problems connected with the treatment of transitivity in Spanish, highlighting their correlation and complex interdependence. Working from an extensive data base, she looks at phenomena which in some cases are quite familiar but which had never previously been considered together. In so doing she highlights a general problem in Spanish, the lack of equilibrium between the two objects (DO and 10): the DO, which ought to be the primary object, is characterized by simple case-marking while the 10, which should be secondary, shows multiple case-marking. The essay clarifies the causes and the diachronic dynamic of this lack of equilibrium and also the division of the structural space of the object between DO and IO. As the author points out, the question is crucial, because the extension of the marker a to the DO in fact reflects the strengthening of the 10 at the expense of the DO and a more general transformation of Spanish from a DO/IO to a PO/SO language (for the definition of Primary Object / Secondary Object, cf. Dryer 1986). It thus tends to become a language which prefers to treat the IO in ditransitive constructs as the DO of a (mono)transitive construct and give a different form to the DO in ditransitive constructs. (In practice a is the DO marker in monotransitive constructions and the IO marker in ditransitive constructions, while an Ο without a preposition marks the DO in a ditransitive construction). The various changes analysed are motivated by semantic and pragmatic considerations; while in themselves merely micro-phenomena, taken together they signal a profound reorganization of transitivity in Spanish. The relationship between DO and 10 in Spanish is also investigated by Ortiz Ciscomani, who focuses on the ditransitive constructs in this language (constructions with DO and 10 with dar 'to give' as prototype). She starts from two premises: a) ditransitive constructs have been largely unaffected by differential object marking; b) indeed over time they show considerable stability and resistance to change, unlike the DO in (mono)transitive constructs. This study is based on a broad corpus of texts spanning several centuries. There are two possible findings: the ditransitive construct marks only the IO with the preposition a, leaving the DO without a preposition, or else it marks both DO and IO with the preposition a. The latter case was found in only 1% of the corpus, involving a mere 10 verbs, but it does occur in the whole arc of the 13th to 20th centuries. It is possible when both

Introduction

xix

DO and IO are animate (persons). After analysing the ditransitive construct with an abundance of details, the author illustrates its syntactic and semantic parameters and sets out the reasons both for its rarity and for its persistence over time. Thus this volume has a contribution to make to discussion of the object category at both the theoretical and descriptive levels. The definition of object proposed here by Lazard is undoubtedly rigorous and eminently applicable. Moreover it places the object category within a complex "multidimensional object zone". Bossong's typological considerations on the asymmetry of marking between subjects and objects throw an interesting light on the general theory of case-marking systems. Then Cennamo's illustration of the relationship between transitivity and object marking shows just how productive the semantic approach to syntax analysis can be. Specific aspects of the formation of the object which are currently the source of debate among scholars (DOM, the double object in ditransitive constructs and changes in valence involving certain classes of predicates) are each touched on in at least one of the essays herein. They are described in an accurate and original fashion, confirming the complexity of the syntactic category of the object, suggesting solutions to specific problems, drawing attention to questions that remain open and offering insights and data for further research.10

Notes 1.

2.

The preponderance of attention dedicated to the Romance languages distinguishes this volume from others which deal with the object category (cf. Plank 1984, Aikhenwald / Dixon / Onishi 2001, Davidse / Lamiroy 2002). For another work dealing with objects in Romance languages, albeit in the wider context of Standard Average European, see Haspelmath 2001. A clear picture of Romance transitivity viewed in terms of its Latin origins is to be found in Herslund 2002. The object category can even receive an ample discussion in a general work on syntax. See for example Givön 1984, where chapter 4 presents many of the phenomena involving the formation of the object: verbs with a locative direct object (she entered the house)·, verbs with a recipient or locative direct object and implied patient {he fed the cows)', verbs with a reciprocal/associative object (he met Sylvia); verbs with cognate objects (he gave an interesting

xx

3.

4.

5.

Giuliana Fiorentino talk), in which the product of the action is presented as object; and object suppression / object incorporation (1984: 98-109). An obvious consequence of this different marking can be seen in the fact that Nominative/ Accusative (for Dryer Subject / Object) languages can have subjects which are either Patient or Agent: the door opened vs. John opened the door. In diachrony the PO/SO opposition can be linked to the human / non-human, but when grammaticalization occurs this will be in terms of PO/SO reanalysis (Dryer 1986: 842). This aspect has come to be generally accepted. Croft (1991) for example observes that the relationship S < Ο < Obi, originally indicating a hierarchy of accessibility for relativization, should also be understood as indicating a hierarchy of topicality. In slightly different terms, but still dealing with the influence of pragmatics on the syntactic formation of topics, see Dixon (1994: 6-8).

6.

It is not possible to give even an outline of the vast bibliography on this subject: I can merely mention Dixon 1976 and 1994, and, in the former, Silverstein 1976. 7. The symbols S, A, Ο for the three basic grammatical relations (semantic primitives), respectively Subject of an intransitive verb, Subject of a transitive verb and Object of a transitive verb, are in more or less universal use. As for terminology, this is not necessarily uniform throughout all essays in this volume, and it has deliberately not been made uniform. I would point out in particular that some authors use the symbols S, A, Ρ (as proposed by Comrie), preferring P(atient) to O(bject). As is well known, in his writings Lazard advocates Ζ for S, X for A and Y for Ο or P. 8. The differential object marking of the fundamental grammatical relations is easily matched to the parameters for transitivity identified by Hopper/Thompson 1980. This in fact helps to recognize a continuous rather than discrete application (Herslund 2002: 36). 9. The quest for linguistic universals and a methodology for defining them has become part and parcel of a rich reflection on language, above all in the domain of typological studies. Lazard has returned to the matter on several occasions; see at least Lazard 1991, 1998. 10. The works of Davidse / Lamiroy (2002: 12), see introduction, and Herslund (2002: 34) refer to the great complexity of formal object marking in Romance languages (as opposed to the relative simplicity of subject marking).

Introduction

xxi

References Aikhenvald, Alexandra Y., Robert M.W. Dixon and Masayuki Onishi (eds.) 2001 Non-canonical Marking of Subjects and Objects. AmsterdamPhiladelphia: Benjamins. Croft, William 1991 Syntactic categories and grammatical relations. Chicago and London: The University of Chicago Press. Davidse, Kristin and Bdatrice Lamiroy 2002 Introduction. In: Kristin Davidse and Beatrice Lamiroy, (eds.), 1-14. Davidse, Kristin and Beatrice Lamiroy (eds.) 2002 The Nominative & Accusative and their counterparts. AmsterdamPhiladelphia: Benjamins. Dixon, Robert M.W. (ed.) 1976 Grammatical Categories in Australian Languages. New Jersey USA: Humanities Press Inc. (Linguistic Series 22). Dixon, Robert M.W. 1994 Ergativity. Cambridge: CUP. Dryer, Matthew S. 1986 Primary Objects, Secondary Objects, and antidative. Language 62, 4: 808-845. Givon, Talmy 1979 On Understanding Grammar. New York: Academic Press. Givon, Talmy 1984 Syntax. A functional-typological introduction. AmsterdamPhiladelphia: Benjamins. Haspelmath, Martin 2001 Non-canonical marking of core arguments in European languages. In: Alexandra Aikhenvald et alii (eds.), 53-83. Herslund, Michael 2002 Romance transitivity. In: Kristin Davidse and Beatrice Lamiroy (eds.), 15-39. Hopper, Paul J. and Thompson, Sandra A. 1980 Transitivity in Grammar and Discourse. Language 56: 251-299. Lazard, Gilbert 1981 La quete des universaux semantiques en linguistique. Actes semantiques. Bulletin IV, 19: 26-37. Lazard, Gilbert 1993 La zone objectale. Actances VII: 15-34. Lazard, Gilbert 1994 L'actance. Paris: PUF.

xxii

Giuliana Florentine

Lazard, Gilbert 1998 L'approche typologique. La linguistique 34,1: 3-17. Moravcsik, Edith A. 1978 On the Case Marking of Objects. In: , Joseph H. Greenberg (ed.), Universals of Human Language. Syntax, vol. IV. Stanford: Stanford University Press. Plank, Frans 1984. Objects. Towards a Theory of Grammatical Relations. London: Academic Press. Silverstein, Michael 1976 Hierarchy of features and ergativity. In: Robert M.W. Dixon (ed.): 112-171.

What is an object in a crosslinguistic perspective ?

Gilbert Lazard

1. Introduction What is an object? The question seems to be rarely asked. Linguists commonly refer to "direct objects", "indirect objects", as if these notions were not problematic. In describing individual languages they appear to find no difficulty in assigning these labels to certain actants in the clause. There are languages in which direct objects are indeed actants which are "direct", i.e., not preceded by a preposition or followed by a postposition, in contradistinction to other actants. In other languages, they are in the accusative case, in contradistinction to complements in oblique cases. In ergative languages,1 they are in the absolutive and they may control verb agreement. In fact, in exotic languages the actants regarded as direct objects are those which are the translation equivalents of genuine direct objects in English or other modern European languages, or of noun phrases in the accusative in classical languages. Although this may be deemed sufficient for descriptive linguistics, provided the descriptive linguist takes into account the peculiarities of the language he describes, it is obviously unsatisfactory for typologists, who deal with many languages of diverse types and need well-defined crosslinguistic notions, in order to have a firm basis for establishing typologies and looking for invariant relationships. Before tackling the question of how to define the object2 in a crosslinguistic perspective, it is useful to have a look at problematic cases in different languages. Indeed, besides "regular" objects there are in many languages, including the most familiar ones, various kinds of problematic actants which look more or less like regular objects, but are somehow to be distinguished from them, and embarrass grammarians. I shall begin with the case of the object in Persian, which is problematic in several respects

2

Gilbert Lazard

and therefore especially interesting. I shall then consider other cases in different languages. Most of them are well known and have been abundantly discussed; however, they have not often been considered together and in the cross-language perspective. In conclusion, I propose a crosslinguistic definition of the object.

2. Objects in Persian Persian, an Indo-European language with a simple and mostly regular morphology, at first sight seems unproblematic. However, its syntax presents a number of peculiarities, especially with respect to the object (see Lazard 1982; Samvelian 2001 and forthcoming a).

2.1. Differential object marking Persian is a good example of differential object marking (Lazard 1984; 1994a: 192-199, 228-232; 1998: 184-191, 219-222; 2001; Bossong 1985; 1989). The object noun phrase may or may not be followed by the enclitic postposition -rä, as in (1). (1)

a. ketäb-rä / an ketäb-rä xänd-am book-PSTP that book-PSTP read-lSG Ί read the book / that book' b. ketäb-i-rä xändam ke book-ART-PSTP read-lSG that Ί read a (certain) book which...' c. ketäb-i xändam book-ART read-lSG Ί read a book' d. ketäb xändam book read-lSG Ί read a book/books /1 did book-reading'

The postposition marks the object when it is definite (either anaphorically or marked as definite by a demonstrative, as in (la), or a possessive determiner), also when it is specific, as in (lb) (-/ is an indefinite article). It is lacking when the object is indefinite non-specific, as in (lc), (Id). Other

What is an object in a crosslinguistic perspective

3

factors interact with definiteness: animacy, thematicity (topicality), semantic relationship to the verb, etc. (see Lazard 1982). There are thus two kinds of objects, one with postposition, the other without it. One may wonder whether one of them is the "real" object, and which one, or whether they can be regarded as objects to the same degree.

2.2. Two objects Some clauses have two objects, one without postposition, the other one with or without it (Lazard 1994a: 89-96; 1998: 86-92), as in (2).3 (2)

a. Maryam arus-rä äräyes kard-0 PN bride-PSTP make-up do-3SG 'Maryam made up the bride's face' b. Maryam arus äräyes hard 'Maryam made up a bride's face / brides' faces'

In such cases the two objects are not on the same level: only the first can take -rä, and their order is fixed. One might argue that arayes kardan (infinitive) 'to make up' is a lexical phrase, and therefore its nominal component {äräyes) is not a real object. However, this is not borne out by tests: it can be shown that, although äräyes kardan actually is a lexical phrase, its nominal component syntactically behaves in the same way as unquestionable objects, like ketäb in (Id) (Samvelian forthcoming a). In particular, it can receive determiners, as in (3)· 4

(3)

Maryam arus-rä

äräyes-e

qaliz-I

kard

PN

make-up-LK

thick-ART

do-3sG

bride-PSTP

'Maryam exaggeratedly made up the bride's face' (lit. did the bride an exaggerated make-up) If we symbolize an object with postposition as NP-ra, an object without postposition as NP-0, and the verb as V, we can summarize the constructions as in (4). (4)

NP-0 V NP-ra V

4

Gilbert Lazard N P - 0 N P - 0 V *NP-0 NP-ra V NP-raNP-0 V *NP-ra NP-ra V

The constructions of (2) and (3) may be analysed as comprising two sequences composed of object + verb at different levels, first äräyes 'makeup' + kardan 'do', and second arus 'bride' + äräyes kardan 'make up'. At the higher level, the sequence äräyes kardan, which is a lexical phrase and is usually described as a "compound verb", functions as a syntactic transitive verbal unit, which in its turn takes an object. However, this analysis does not apply to such clauses as (5), which conform to the formula NP-ra N P - 0 Verb, but in which the sequence following NP-ra can hardly be regarded as a (transitive) lexical phrase. (5)

(6)

pul-es-rä sirini mixor-ad money-his-PSTP sweets eat:PRES-3SG 'He spends his money in eating sweets' (lit. he eats sweets his money) pul-e änhä-rä pärcehä-ye goldär-e money-LK

those-PSTP

cloths-LK

flowered-LK

qasang xarid-0 nice buy-3SG 'He spent the money earned by selling those things in buying nice flowered cloths' (lit. he bought nice flowered cloths the money of those) It thus appears that, under certain conditions (which remain to be investigated), a noun phrase followed by -ra may function as a higher object dependent on a sequence which is not a lexical phrase and which is composed of a verb preceded by a lower, possibly complex, object.

2.3. Three objects Certain clauses may be analysed as comprising three objects, as in (7). (7)

a. pädäs däd-and reward give-3PL 'They gave a reward / rewards'

What is an object in a crosslinguistic perspective

b. u-rä

pädäs

5

dädand

him/her-PSTP reward give-3PL 'They recompensed him/her'

c. ketäb-i

pädäs

dädand

book-ART reward give-3PL 'They gave a book as a reward'

d. u-rä him/her-PSTP

ketäb-i

pädäs

dädand

book-ART

reward

give-3PL

'They recompensed him/her with a book' / 'They gave him/her a book as a reward' Pädäs dädan is a lexical phrase, which may take an object designating either the person recompensed (7b) or the object given (7c); it may even take both, resulting in a three-object clause (7d). These three objects are not on the same level, but on three different levels: only the first one can take -rä.

2.4. A hierarchy of objects There are thus in Persian at least two different types of objects, one with no particular marking, the other with the postposition -rä. They may alternate, in correlation with properties of the object noun phrase: definiteness, animacy, semantic relationship to the verb, thematicity, etc. On the other hand, they behave differently when the clause includes two objects. The higher one, which is placed further away from the verb, may be NP-ra or NP-0; the lower one, which is placed closer to the verb, is always NP-0. A clause can include only one object marked by -rä. Moreover, the two (or three) objects do not have the same properties: they are not equally accessible to manipulations. Besides the two types of objects, there still exist two other kinds of noun phrases accompanied by the postposition -rä (see Lazard 1982), namely: a) time complements, as in (8): (8)

emsab-rä manzel mimän-im to-night-PSTP at.home stay:PRES-lPL 'To-night, we'll stay at home'

6

Gilbert Lazard

b) thematic terms (topics) in initial position, followed by a pause and cross-referenced by a pronoun, as in (9). (9)

in doxtar-rä bä mädar-es sohbat this girl-PSTP with mother-her talk 'As for this girl, I talked to her mother'

kard-am do-lSG

Though they look like objects, these noun phrases with -rä are better not analysed as another kind of object, for they appear to behave differently from objects proper, as shown by Samvelian (forthcoming). In particular, they can co-occur with -rä objects in the same clause, as in (10) and (11), whereas there cannot be two ra-objects in one clause. (10)

(11)

in

daf'e-rä

estebäh-et-rä

this

time-PSTP

error-your-PSTP

mibaxs-am forgive:PRES-lSG

'For this one time, I forgive your error' mäsin-rä dar-es-rä bast-am car-PSTP

door-its-PSTP

close-lSG

'As for the car, I closed its door' However, the very fact that they take the marker -rä like objects suggests that they possess a functional feature in common with objects proper: they ought not to be entirely ignored in investigations concerning the object.

3. Problematic objects in other languages In this section I briefly survey different constructions involving actants which have been or may be analysed as objects, although they differ in some way from those which are traditionally regarded as "regular" or "real" objects. For the more exotic constructions, I reproduce a few examples.

3.1. Differential object marking Like Persian, many languages, including some Romance languages (for Spanish, see Company-Company and Ortiz Ciscomani in this volume), have differential object marking. Although the precise rules that govern the

What is an object in a crosslinguistic perspective

1

marking in different languages are different, the general trend is the same in all languages. In a given language the main factor may be definiteness or animacy or thematicity, but the conditioning is always shown in the same way: the object higher in definiteness or animacy or in thematic position is marked, and the object lower in definiteness or animacy or in rhematic position is unmarked. Differential object marking is not necessarily associated with the possibility of two objects in the same clause, as it is in Persian. On the other hand, it is worth mentioning that there is often, though not always, some kind of affinity between the marked object and the dative: for example, in Spanish and other western Romance dialects, the object marker is the preposition a, which also has dative and directional meanings.

3.2. Differential object agreement In some languages, the verb may or may not agree with the object, as in (12) in Tswana (Lazard 1996: 153, from Siewierska and Creissels). (12)

a. Thabo ό-e-biditse PN 3SG(AG)-3SG(OBJ)-call 'Thabo called the dog' b. Thabo ό-biditse ntsä PN 3SG(AG)-call dog 'Thabo called a dog'

ntsä dog

This variation correlates with the same kinds of semantic and pragmatic factors as differential object marking, and in the same way. However, there is an important difference between the object which is cross-referenced in the verb form and the object which is not: the latter must be placed in the vicinity of the verb, while the former may be more freely moved away from it. When there is no agreement, the object is thus somehow tied up to the verb; when there is agreement, it is more autonomous.

3.3 Cognate objects In many languages, cognate objects differ from those objects that are regarded as 'regular' by showing special properties. In particular, e.g. in

8

Gilbert Lazard

Arabic, they may co-occur with regular objects even with verbs which otherwise do not allow two objects in the same clause, as in (13) (Monteil 1960: 274); they may also appear with intransitive verbs, cf. below, § 3.8. (13) yqfhamu l-adab-a fahm-an he.understands ART-literature-ACC understanding-ACC 'amiq-an deep-ACC 'He has a profound understanding of literature' (lit. he understands literature a profound understanding)

3.4. Measure complements Like cognate objects, measure complements have been discussed many times. Smith (1992) examines measure complements in Romance languages in detail. He shows that they both partake of some properties of "objecthood" as it is usually conceived of, and are to be distinguished from ordinary objects by other properties. Their ambiguous nature is illustrated by cases like (14), in French, where the orthography indicates hesitation between analysing the noun phrase in question as an object and considering it as a circumstancial complement. (14)

les efforts que ce travail m 'a coüte / coütes 'the efforts that this work cost me'

The French Academy in 1878 preferred coüti (measure complement) and later prescribed coütes (object) (Grevisse 1964: 717).

3.5. The "H-actant" Another problematic case is that of what I called the "H-actant" (Lazard 1994b), i.e., the NP in the so-called presentative construction. This purely conventional label refers to noun phrases which seem semantically to be subjects, but are placed after the verb, as in there appeared a stranger or, in French, il arrive des voyageurs 'there arrive some passengers'. The grammatical status of such actants is not exactly the same in different languages. Although they have been analysed as a special kind of subject,

What is an object in a crosslinguistic perspective

9

because such sentences have approximately the same meaning as "regular" sentences in which the same noun phrases are subjects (des voyageurs arrivent 'some passengers arrive'), they undoubtedly possess some object properties, at least in certain languages.

3.6. Two objects Constructions with two objects like g/ve-sentences in English are not rare. In English they include two direct objects, 5 and, in the classical languages, they include two noun phrases in the accusative. It can easily be shown that these two objects are not to be put on the same level. The clearest test is passivization, as in the Latin example (15) (Fugier 1991, cf. Lazard 1994a: 90; 1998: 87) (15)

a. consul-0 roga-t senator-em sententia-m consul-NOM ask-3SG senator-ACC opinion-ACC 'The consul asks the senator his opinion' b. senator-0 roga-tur sententiam senator-NOM ask-PASS:3SG opinion-ACC 'The senator is asked (to give) his opinion' c. sententia-0 rogatur a senator-e opinion-NOM ask-PASS:3SG from senator-ABL (or senator-i) (senator-DAT) 'An/The opinion is sought of the senator'

As sententiam, in (15b), remains in the accusative, it is, in (15a), clearly (grammatically) closer to the verb than senatorem, which in (15c) becomes an oblique complement.

3.7. Second absolutive in ergative construction In three-actant clauses, in ergative languages, it may happen that the recipient actant has the same grammatical treatment as the object in twoactant clauses, i.e. absolutive case and, as the case may be, verb agreement, while the actant designating the thing given is also in the absolutive, as in

10

Gilbert Lazard

(16b) in Hayu, a Tibeto-Burman language (Michailovsky 1988: 138, cf. Lazard 1994: 88;1998: 86)). (16) a. ga

thuNnomi

me:ERG I.lead.you

gon you:ABS

Ί lead you' b. ga gon pipira me:ERG you:ABS bread Ί give you bread'

mUmnom I.give.you

In (16a), a two-actant clause, the object pronoun 'you' is in the absolutive and cross-referenced by a verbal marker. In the three-actant clause (16b), the pronoun 'you', representing the recipient, is treated in exactly the same way: it is the regular grammatical object. On the other hand, pipira 'bread' is also in the absolutive, but with no cross-reference: it is a kind of 'second object'.

3.8. Object with intransitive verb There are, in languages of different types, examples of noun phrases used as a kind of direct complement of an intransitive verb. I reproduce here in (17) an example in Indonesian, an accusative language (cf. Lazard 1994: 98; 1998: 94). (17)

Ali bertanam kopi PN plant coffee 'Ali is a coffee grower'

Verbs with prefix ber- have normally no object. On the other hand such a noun phrase as 'coffee' in this position is subject to constraints: it has no determiner and it conveys a generic meaning. It is better regarded as a sort of qualifier of the verb predicate than as an object. The same construction is described, for instance, in Dulong/Rawang, a Tibeto-Burman language, by LaPolla (2000: 291-292), with reflexive/middle (intransitive) verbs and names of bodyparts or noun phrases whose "referent will be understood as something that has some sort of strong connection with the referent of the A [ = agent, GL] argument".

What is an object in a crosslinguistic perspective

11

It is found also in ergative languages in similar conditions, e.g. in Wargamay, an Australian language , as in (18b), reproduced from Dixon (1981; cf. Lazard 1994: 100; 1998: 95). (18) a. nad'a me-ERG

wagun

ganda-nu

wood:ABS

burn:PF:TR

Ί burnt some wood' b. nayba mala I:ABS

hand:ABS

ganda-gi burn:PF:INTR

Ί burnt my hand' (18a) is the regular transitive ergative construction, with the agent actant in the ergative and the object in the absolutive. In (18b), the verb has an intransitive marker, the pronoun mapping the agent is in the absolutive as in one-actant clauses, but there is a second absolutive, whose function is somehow to qualify the verbal predicate.

4. What is an object 4.1. The object zone The case of two objects in the same clause, in Persian as well as in Latin and other languages (above, ex. (2), (3), (5), (6) and (15)), makes clear a distinction between two kinds of objects with respect to their relationship to the verb. One object is grammatically closer to the verb: in Latin, etc. it remains in the accusative (sententiam) when the verb is turned into the passive; in association with the verb form it constitutes what is predicated of the subject {senator)·, in Persian, an SOV language, it is a "lower" object, obligatorily placed in the vicinity of the verb form and after the "higher" object. The other object has the opposite qualities. We call the former "close object", the latter "distant object". The same analysis applies also to other cases. In differential object agreement (§ 3.2), the object with no agreement, which, in the clause, is closely linked to the verb form, may legitimately be considered as a close object, in contradistinction to the object cross-referenced by an object marker within the verb form: the latter, which is more autonomous and can be more freely moved, is a distant object. Cognate objects (§ 3.3), "second" absolutives in three-actant ergative constructions (§ 3.7), and quasi-

12

Gilbert Lazard

objects qualifying the predicate (§ 3.8) may also be analysed as close objects in contradistinction to "regular" objects, "first absolutives" (representing recipients), and objects of transitive verbs respectively. Similarly, all 0-objects in Persian, even in one-object clauses, may be regarded as grammatically closer to the verb than ra-objects: it is not by chance that, in two-object (and three-object) clauses, only the first one can take -rä and that, in clauses of the type illustrated by ex. (5) and (6), the first object in the chain (the higher one) is always marked by -rä. We are thus led, in cross-linguistic perspective, to posit, rather than a single object position, an "object zone", which may include two (or more) objects at different (grammatical) distances from the verb. Some languages allow clauses with more than one object, either two, a close one and a distant one, or, more rarely, three, with an intermediate one. As for special cases, like measure complements and H-actants, it seems difficult to assign them a position on the same line as the preceding cases. However, they do appear to qualify for being located somewhere in (or close to) the object zone. If we conceive of the object zone as a multidimensional space, they might be placed in other dimensions than that of objects proper.

4.2. The prototypical object It remains to see why grammarians have so persistently made use of the notion of object, not only in classical and modern European languages, but also in exotic languages, even in ergative ones, where what is usually called object may have no grammatical feature at all in common with the object in familiar accusative languages. Like most of the traditional grammatical notions, actant functions are badly defined. The label "direct object" seemingly refers to a purely formal feature, namely the fact that, in such languages as English or French, it is not marked by a case or a pre- or postposition. However this definition hardly applies to languages with case-marking, like Latin or Russian, or to ergative languages, where what is currently called object is grammatically more similar to the subject than to the object of Western European languages. It thus appears that the object is mainly identified by its semantic content, i.e., the idea that the noun phrase labelled object represents a patient. However, this semantic basis usually remains implicit, or, when a definition is formulated, it is often a mixture of semantic and morphosyntactic features. On the other hand, a purely semantic definition of the ob-

What is an object in a crosslinguistic perspective

13

ject is inadequate, for, in many languages, many sentences are said to include an object while not expressing an action at all, so that the so-called object cannot be considered as representing a patient. A solution to this difficulty can be found by drawing a clear distinction between semantic content and morphosyntactic form and by adopting a methodology which avoids directly resorting to meaning in the comparison of languages. The principles of this methodology have been explained elsewhere (Lazard 1999). Here let us say only that the first step is to posit the notion of the prototypical action as a real discrete action performed by an agent and affecting a patient. It may be taken for granted that such a notion can be expressed in any language. As a second step, we posit, in any language, the construction used to express the prototypical action as the "major bi-actant construction", whatever form it assumes in individual languages. This construction includes two actants, one of which represents an agent and the other a patient. Now, in many languages, this very construction may be used to express not only the prototypical action, but also a number of other processes, which may be non-prototypical actions or even non-actions: the actants then represent no longer agent and patient; however, they keep the same morphosyntactic properties. Our third step is to define the second actant in the major bi-actant construction as the "object", whatever process is expressed by this construction. We have thus arrived at a clear cross-linguistic definition of the object. This definition agrees satisfactorily with current practice. The actants which are usually considered as "regular" or "real" objects are easily subsumed under the notion of object as we have defined it. In contradistinction to the problematic "objects" surveyed above, this object may be characterized as the "prototypical object". It is located somewhere in what we have called the object zone, and it gives it a theoretical status. The object zone can then be defined as that which includes the prototypical object and may also include various kinds of actants which have some properties, but not all, in common with the prototypical object.

Notes 1. 2.

For the definition of an ergative language I am using, see Lazard 1997: 2 5 0 251; 1998: 40; 1999: 98-99; 2001a: 34-35. By "object", in this article, I mean what is usually called "direct object", leaving aside other kinds of so-called objects.

14 3.

4.

5.

Gilbert Lazard Examples (2), (3), (7), (8), (10) are taken from Samvelian (forthcoming). — Abbreviations: ABL ablative, ABS absolutive, ACC accusative, (AG) agent (actant index), ART article, DAT dative, ERG ergative, INTR intransitive, LK linker, NOM nominative, (OBJ) object (actant index), PF perfect, PL plural, PRES present, PSTP postposition, SG singular, TR transitive. In this case, the other object, which comes first, cannot be a "bare" noun (Samvelian 2001): * Maryam arus äräyes-e qaliz-i kard. In the following we disregard the difference between bare nouns and noun phrases. Although, traditionally, one of them is called "indirect object", on semantic grounds. It goes without saying that the two English objects have rather different properties. I cannot here deal with that question, which has given rise to a rich literature. For that matter, the different kinds of "objects" in various languages that are surveyed in this article are all characterized by different properties, which could be detailed for each individual language.

References Bossong, Georg 1985 Empirische Universalienforschung. Differentielle Objektmarkierung in den neuiranischen Sprachen. Tübingen: Gunter Narr. Bossong, Georg 1998 Le marquage differentiel de l'objet dans les langues de l'Europe. In: Jack Feuillet (ed.), Actance et valence dans les langues de I 'Europe, 193-258. (EALT/ EUROTYP 20-2) Berlin/New York: Mouton de Gruyter. Company-Company, Conception In this volume. Dixon, Robert Μ. W. 1981 Wargamay. In: Robert M. W. Dixon and Barry J. Blake (eds) , Handbook of Australian languages, 2, 1-144. Amsterdam: Benjamins. Fugier, Huguette 1991 Le verbe latin "incorpore"-t-il ses complements? In: Joseph Herman (ed.), Linguistic studies in Latin, 75-90. Amsterdam/Philadelphia: Benjamins. Grevisse, Maurice 1964 Le bon usage. Grammaire frangaise, 8e ed. Gembloux / Paris: Duculot / Hatier

What is an object in a crosslinguistic perspective

15

LaPolla, Randy J. 2000 Valency-changing in Dulong/Rawang. In: Robert M.W. Dixon and Alexandra Aikhenvald (eds.), Changing valency. Case studies in transitivity, 282-311. Cambridge: Cambridge University Press. Lazard, Gilbert 1982 Le morpheme rä en persan et les relations actancielles. Bulletin de la Societe de linguistique de Paris 77/1: 177—208. Lazard, Gilbert 1984 Actance variations and categories of the object. In: Frans Plank (ed.), Objects. Towards a theory of grammatical relations, 269—292. London: Academic Press. Lazard, Gilbert 1994a L 'actance. Paris: Presses Universitaires de France. Lazard, Gilbert 1994b L'actant H: sujet ou objet? Bulletin de la Societe de linguistique de Paris 89/1: 1-28. Lazard, Gilbert 1996 Fonction de l'accord verbe-actant. Faits de langue 8: 151-160. Lazard, Gilbert 1997 Ergativity (Review article of Robert M. W. Dixon, Ergativity, 1994). Linguistic Typology 1/2: 243-268. Lazard, Gilbert 1998 Actancy. Berlin/New York: Mouton de Gruyter (transl. of Lazard 1994a). Lazard, Gilbert 1999 La linguistique est-elle une science? Bulletin de la Societe de linguistique de Paris 94/1: 67-112. Lazard, Gilbert 2001a Etudes de linguistique generale: typologie grammaticale. LeuvenParis: Peeters. Lazard, Gilbert 2001b Le marquage differentiel de l'objet. In Martin Haspelmath et al. (eds.), Language typology and language universals, an international handbook, vol.2, 873-885. Berlin/New York: Walter de Gruyter. Michailovsky, Boyd 1988 La langue hayu. Paris: Editions du CNRS. Monteil, Vincent 1960 L 'arabe moderne. Paris: Klincksieck. Ortiz Ciscomani, Rosa Maria In this volume

16

Gilbert Lazard

Samvelian, Pollet 2001

Les complements d'objet "nus" en persan: incorporation ou non? Bulletin de la Societe de linguistique de Paris 96/1: 349-388. Samvelian, Pollet forthc. Une analyse non configurationnelle du morpheme rä en persan. Smith, John Charles 1992 Circumstantial complements and direct objects in the Romance languages: configuration, case, and thematic structure. In: Roca, I.M. (ed.), Thematic Structure. Its Role in Grammar, 293-316. BerlinNew York: Foris.

Nominal and/ or verbal marking of central actants

Georg Bossong

1. Introduction: the fundamental asymmetry There is a far-reaching difference between the marking of objects and of subjects. This difference is so deeply rooted in natural languages that it generally passes unnoticed. But, as Plato put it, astonishment is the beginning of philosophy. In this contribution, I would like to draw attention to a fact which merits, in my view, some astonishment. I refer to the fact that subjects are preferentially marked in the verb, whereas objects are preferentially marked in the noun. This phenomenon has probably been perceived as being too trivial for requiring a thorough analysis. In order to determine the meaning and scope of this fundamental asymmetry, it is convenient to outline some theoretical aspects of the marking of fundamental grammatical relations, in the way I have developed it over last two decades or so. Taking the perspective of a global axiomatic theory of actant marking, there is no a priori reason for such an asymmetry to exist. But the observable facts are in contradiction with what we would expect from a universalist approach. On the one hand, it is possible to establish, by means of deduction, a universal pattern of marking, directly derived from certain inherent properties of human language. On the other hand, it is necessary to observe attentively the empirical data of individual languages, to draw conclusions by means of induction. Finally the two approaches should be combined. It seems to me that certain empirical findings can only be properly evaluated in their specificity if viewed against the background of a theory conceived beforehand, without any reference to the actual linguistic variability. Conversely, theory remains empty if it is not measured against the empirical facts. Only in such a double perspective, which combines deduction and induction, can the typological approach achieve its full potential.

18

Georg Bossong

2. Towards a typology of central actant marking 2.1. Typological theory 2.1.1. A general theory of marking Let us start at the beginning. A grammatical relation between two elements can be expressed in four different ways. These four types of expression follow logically from the very nature of human language, which is, needless to insist on it, meaningful and linear. Slightly modifying a pattern expounded more in detail elsewhere (see Bossong 2001), I propose the following systematised typology of possible expressions of grammatical relations (cf. Figure 1). In the context of the present discussion, some of these terms are not relevant. I only mention in passing that the distinction of "implicit" and "explicit" refers to the fact that a relation can remain either completely unspecified, or be expressed by some overt means. In the latter case, given the linearity of human language, the relative position of the elements can be used for expressing the relationship; meaningful position is termed here as "taxeme" (from Greek taxis "order"), hence "taxemics" as the corresponding general term for this technique. Broadly speaking, implicitness and taxemics together cover the domain defined in traditional typology as "isolation". Strict implicitness is a rather marginal phenomenon in many, if not most, languages; it is found mainly in such domains as adverbials of time or place, which bear in themselves such a clear reference to the intended meaning that an explicit expression of the relationship is often treated as unnecessary (cf. examples like he'll come Monday (—• implicitness) vs. he ΊΙ come in January (—> explicitness)). This matter will not be further discussed here, since implicitness plays (scarcely) any role in the expression of subject and object relations; these must necessarily be specifiable in some way or other, even though they may remain unspecified ("implicit") in a few particular contexts. What is of interest here is the typology of the three basic ways of explicit expression. If explicitly expressed at all, a grammatical relation must necessarily be marked either by taxemics (i.e. by the meaningful positioning of the elements with respect to each other) or by grammemics (i.e. by an overt marker whose function is exactly the marking of the relationship); in the latter case, the marking can be unilateral (i.e. a

Nominal and/or verbal marking of central actants

19

grammeme is added to only one of the two related elements) or bilateral (i.e. a grammeme is added to both of them).

expression

implicitness

+ explicitness

taxemics

grammemics

bilateral

AB=BA

AB^BA

xAyB

unilateral

xAB/AyB

Figure 1. Types of expression of grammatical relations

It should be noted that in the diagram in Figure 1 the order of symbols is not relevant; putting in front of instead of putting it after is purely conventional: it has to be put somewhere given the linearity of writing. What is important for the discussion here is the presence or absence of a grammemic element as such. Likewise, the degree of boundedness of the

20

Georg Bossong

grammeme is not taken here as relevant, either; it can be loosely juxtaposed to, or fully integrated into the lexeme to which it refers, with the whole array of intermediate possibilities of boundedness. Criteria such as the relative position and the degree of boundedness of lexeme and grammeme are of course important for a fully elaborated typology, but in the present context they are not considered.

2.1.2. The marking offundamental

relation

If we apply this universal pattern to the domain of the subject and object relations, it becomes immediately evident that the related elements are the verbal predicate and its satellites, the nominal actant(s). This is exactly what I have termed the fundamental relation. The fundamental relation relates a verbal element to one or more nominal element(s), in whatever way the distinction between "nouns" and "verbs" in a given language is realized. Replacing the highly abstract symbols and in (1) by the slightly more concrete ones (noun, "nouny" element, noun phrase) and (verb, "verboid", kernel of the predicate), this diagram can be partially rewritten as follows:

Non-grammemic: -N Nominal: +N Verbal: -N Verbo-nominal: +N

η η η η

-ν -ν +ν +ν

Figure 2 Types of expression of fundamental relation

Non-grammemic marking, which is necessarily taxemic, will not be discussed here any further. The focus will be on nominal (unilateral), verbal (unilateral), and verbo-nominal (bilateral) marking.

2.1.3. The basic configurations Before reformulating more precisely the basic asymmetry stated at the beginning, it is necessary to clarify the notions of "subject" and "object". This

Nominal and/or verbal marking of central actants

11

will be done here in a way that might seem somewhat arbitrary, even dogmatic, but it does in fact represent a condensation of extensive research and reflection. We will have to treat the thorny problems of accusativity vs. ergativity and some related questions, albeit superficially. I have termed configuration the elements of this kind of typology; in accordance with many former contributions, its essential core can be resumed as follows: Configurations are the patterns according to which meaning and form of monovalent () and bivalent () predicates are related to each other. "Agent" and "Patient" refer to the meanings of the central actants, conceived as "prototypical" in the sense of prototype semantics. The choice of these terms does not imply that for every verb, in whatever context, a "subject" is necessarily agentive, or an "object" patientive. These are prototypical values in a system of semantic relationships. The prototypical approach has to be combined with the typological one, in order to obtain satisfactory results in the empirical analysis of, and inductive generalizations about, socalled "case-roles".

Accusativity

Ergativity

Duality

Agt 1

Pat 1

Agt 1

Pat 1

Agt 1

Pat 1

Agt 2

Pat 2

Agt 2

Pat 2

Agt 2

Pat 2

Figure 3. Elements of a configurational typology Among the numerous configurations which are theoretically possible (by simple permutations of the constitutive elements), only the three contained in Figure 3 are observed with reasonable frequency in the languages of the world. The third of these configurations was described for the first time by Klimov (1977), who used the term "active". His description was valid, but the term he chose is ambiguous in its English, German, or French translation (it was not so in the original Russian). Gilbert Lazard (1986, 1991) has proposed renaming it the "dual" type, a term which is more felicitous, although it is not totally free of ambiguity either; Lazard's term will be adopted in this contribution.

22

Georg Bossong

Figure 3 reveals a subdivision of basic configurations into two quite distinct subclasses. In the accusative and the ergative configuration, there is a clear difference between a dominant and a secondary formal category. In these two types, one of the two formal categories accumulates three functions, whereas the other one is limited to only one function. It is evident that the cumulative category is fundamental in the respective linguistic system, much more so than the corresponding non-cumulative category. Such a differentiation is lacking in the dual type, which owes its name exactly to this structural feature: the ergative and the accusative configuration are characterized by a hierarchy which is absent from the dual one. To describe this hierarchy, I use the terms created by Lucien Tesniere "prime actant" and "second actant", but with a different content. Tesniere used these terms simply as substitutes for the traditional Eurocentric notions of "subject" and "direct object". But I think that these terms are useful to represent the hierarchy which exists in both the accusative and the ergative configuration between a primary, hierarchically dominating category, and the opposed category which is hierarchically subordinated. The first actant is represented by the nominative case (Agt 1 + Pat 1 + Agt 2) in accusative languages, by the absolutive case (Agt 1 + Pat 1 + Pat 2) in ergative languages; the second actant corresponds to the accusative (Pat 2) and the ergative case (Agt 2) respectively. As for the dual configuration, there is no such thing as a hierarchy between the central actantial functions, hence nothing like a first or a second actant. Given this absence of a hierarchy, we shall not refer any more to the dual type in this contribution. The question of preferential marking of two hierarchically differentiated actants is not pertinent in that configuration type. In terms of frequency, languages where the accusative or the ergative configuration dominates are in the great majority, and probably account for over 90 % of the world's languages. Dual languages are not insignificant, however. Other configurations, which are not mentioned here, do occur, but they are comparatively rare; the neutral type, where none of the central actants is marked either by morphology or by position, occurs in South-East Asia, e.g. Burmese. Incidentally, accusative and ergative patterns may also be mixed in a given language system. We can conclude that the three basic configurations are subdivided into a hierarchical and a non-hierarchical subgroup, the former comprising the accusative and the ergative type, the latter the dual type. For the hierarchical subgroup, we can postulate a more concrete figure which takes the following form:

Nominal and/or verbal marking of central actants

23

Table 1. Hierarchy of actants

Accusativity Ergativity

First actant NOM ABS

Second actant ACC ERG

2.1.4. The foindamental asymmetry

reformulated

We are now able to reformulate more precisely the aforementioned asymmetry between the two fundamental actantial functions. The distinction between the first and the second actant is independent of the basic structure of the language in question; it is made in both the accusative and the ergative configuration. What counts is the primary vs. secondary character of the actants, not their specific semantics. Preferences, or preferential tendencies of marking can be represented by an implicational hierarchy. The marked form, which is comparatively less frequent and less fundamental in the respective system, implies the nonmarked form, which is more frequent and more fundamental. Symbolizing the implicational relation by < 3 > , we can write the basic asymmetry between actants as follows:

Table 2. Asymmetry of marking of central actants Accusativity Ergativity

Nominal marking N O M 3 ACC ABS 3 ERG

Verbal marking ACC 3 N O M ERG 3 ABS

This is a rigid form of description. It is perhaps more realistic to describe this relationship in a more flexible way, in terms of probability rather than exclusiveness. The probability that a first actant be marked in the verb is higher than its being marked in the noun; conversely, nouns are more likely to be marked for second actant functions than for the first. As a result, one obtains two converse gradient scales which illustrate the increasing or decreasing probability of marking:

24

Georg Bossong

First actant: +N η -V

-Nfl+V

+N η - ν

-Ν n + v

Figure 4. Actant marking in terms of probability

2.2. Empirical data from typology Formulated this way, this is still a highly abstract constellation. It is convenient to go down one step further towards concrete language structures and to illustrate these general schemas by examples taken from individual languages. First of all, cases which fulfil the general expectations will be cited; then an exception will be briefly discussed, namely the case of Old French. Such "exceptions which confirm the rule" can be aptly explained by specific historical circumstances; far from invalidating the postulates of general theory, they lead to a better understanding of notions such as "universal tendency" and "marking probability".

Nominal and/or verbal marking of central actants

25

2.2.1. Nominal marking in accusative languages 2.2.1.1. The normal patterns Let us begin with the properties of nominal marking in accusative languages. It is obvious that whenever there is a privative opposition between the marking of the basic functions, the nominative, i.e. the first actant, is the formally unmarked case category. Quechua in all its varieties is a classic case in point; in this language, the object is always marked with an accusative ending, whereas the nominative is always in the zero form: (1) (a) noqa-0 Lurinsu-ta tinku-rqa-ni nan-pi I-NOM Lorenzo-ACC meet-PRT-lSG way-LOC Ί met Lorenzo on the way' (b) taytay-0 tanta-ta fidiyus-ta arrus-ta my+father-NOM bread-ACC noodles-ACC rice-ACC ranti-rqa-n buy-PRT-3SG 'My father bought bread, noodles and rice' {Quechua of Cajamarca/ Northern Peru} Quechua corresponds perfectly to the typological expectations. The nominative is not marked in the noun, but necessarily and under all circumstances in the verb; the accusative is sometimes marked, sometimes unmarked in the verb, but it is consistently and obligatorily marked in the noun. Nominal marking of the accusative does not depend on the nature of the object, it is obligatory throughout, even with indefinite and inanimate mass-nouns as in (lb); there is no differentiality whatsoever. Hebrew, on the other hand, is a typical example of differential object marking: (2)

(a) be-re'sit bara' 'elohim et-ha-samayim, in-beginning created God ACC'-ART-heavens we- 'et-ha- 'ares and-ACC'-ART-earth 'In the beginning, God created the heaven and the earth'

26

Georg Bossong

(b) wa-tose' ha-'ares 0-dese' 0- 'esev and-produced ART-earth ACC-meadow ACC-grass we-0- 'es 'ose peri and-ACC-tree making fruit 'And the earth brought forth grass, herb, and tree yielding fruit' {Biblical Hebrew/Gen. 1,1; 1,12} The object is marked only if it is definite; this is what I call "differential marking". With or without differentiality, the basic configuration is similar in Hebrew and Quechua (two languages whose respective marking types are fairly widespread among the languages of the world): in the noun, the function of first actant is zero-marked, whereas the function of second actant is positively marked; in the verb, the inverse situation holds, the first actant being marked and the second actant being (partially) unmarked. Among Creole languages, morphological marking of central actant relations is rather exceptional. In Papia Kristang, a Portuguese based Creole spoken in Malacca and Singapore (see Baxter 1985), we observe differential object marking according to the well-known general rules, which means that the probability of morphological marking increases with the degree of definiteness and "humanness" (once again, I use here an English adaptation of the term introduced by Gilbert Lazard "humanitude", instead of the more widespread "animacy"; in other publications, I use the terms "referentiality" and "inherency" for these criteria):

(3) (a) yo sa kanyong gosta hung ake I GEN elder brother love ACC' DEM 'My elder brother loves that woman' (b) eli ja kazä kung~0 femi malayu he PRF marry A C C A C C woman Malay 'He has married a Malay woman' (c) Veronica sa pay matä 0-bichu Veronica GEN father kill ACC-bug 'Veronica's father has killed a bug' {Papia Kristang}

femi woman

Nominal and/or verbal marking of central actants

11

In comparison with such systems which present a privative opposition, systems presenting an equipollent opposition are a minority. Such systems are not in contradiction with the predictions of theory. Systems with equipollent oppositions are less numerous simply because, in general, natural languages prefer privative oppositions, probably for economical reasons. Latin (with markings of the well-known type dominu-s 'lord-NOM' vs. dominu-m 'lord-ACC') and other ancient Indo-European languages, such as Greek and Sanskrit, show this type of constellation. Here are a few textual examples: (4) acarya-h sisya- m pasyati teacher-NOM pupil-ACC sees 'The teacher sees the pupil' {Sanskrit} ono-s hippo-η emakarizen donkey-NOM horse-ACC called-lucky 'The donkey praised the horse for his luck' {Classical Greek/ Aesop 328} lupu-s arguebat vulpe-m furt-i crimin-e wolf-NOM accused fox-ACC theft-GEN crime-ABL 'The wolf accused the fox of the crime of theft' {Latin/ Phaedrus 1,10} Neither in Latin nor in the other classical languages quoted is equipollence the only principle of morphological marking; there is a great deal of privativity in the declension system in each of these languages. Another example of this constellation is Japanese: (5) tanakasan-ga shinbun-(w) yomimasu Tanaka-Mr.-NOM newspaper-ACC read 'Mr. Tanaka is reading the/a newspaper' {Modern Japanese} Finally, we may quote Georgian where in the present tense group (which is structured accusatively) nominative and accusative have their own case endings:

28

Georg Bossong

(6) ρ 'ropesor-i c 'ers c 'ign-s professor-NOM writes book-ACC 'The professor is writing the/ a book' {Modern Georgian} The distribution of privative and equipollent structures confirms the postulates of typological theory: if the nominative is marked by a specific grammeme, the accusative is marked so too, but not the other way round.

2.2.1.2. Old French, the rule-confirming exception Exceptions to this rule do in fact exist, but they are rare. In the Cushitic branch of the Afro-Asiatic family, this constellation seems to be relatively widespread (cf. Sasse 1974; 1981: 206). It is also reported to occur in the Yuman language family. A much more familiar example (at least among Romanists) is Old French (together with Old Provenfal). Indeed, it is possible, in this case, to study the effects of universal tendencies which reduce and finally eliminate the results of typologically deviant evolutions. The facts are well-known; it is not necessary to present them here in great detail. What follows is a short outline of some essential points. In some noun classes inherited from Latin, especially in the so-called second declension of masculine nouns (those ending in -us), the purely mechanical effect of phonetic erosion has ended up in a system where the first actant is marked, whereas the second actant is not. The reason is that final -s has resisted better, and therefore maintained itself longer, than final -m. As a result, the equipollent paradigm muru-s/ muru-m of Latin has developed into the privative paradigm murs/mur of Medieval French.

(7) murus —* murs murum —* mur {Old French}

In fact, the final -m was more an orthographic symbol of nasalization of the preceding vowel than a true consonant already in the times of Classical

Nominal and/or verbal marking of central actants

29

Latin. Once the feature [+ nasal] was lost, the equipollent opposition inherited from Indo-European became privative. Phonetic erosion acts blindly: its results only begin to interact with functional factors a posteriori. Later re-arrangements no longer obey the purely material rules of phonetics (of the type quoted: -s is more resistant than -m), but functional rules of marking distribution (of the type discussed here: nouns are more easily and more frequently marked for the accusative than for the nominative). As for the evolution of French, the stability of its two-case declension over a relatively long period of time is a more remarkable fact than its ultimate disappearance. A comparison with other Romance languages clearly shows the effect of typological factors in diachronic evolution, and the exceptional character of French in this respect. In the other Romance languages which have preserved the final -s, the evolution has followed from the very beginning the pathways traced by universal functional tendencies, without ever showing the effect of "blind" phonetic erosion. This can clearly be observed in the Ibero-Romance languages and in Sardinian. In these languages, final -s has been preserved until today (in French, it gradually disappeared from about 1300 on). As in all other Romance languages, the reflexes of final -m were eliminated at an early date, probably already in Late Latin. Consequently, a two-case declension of the aberrant Old French type could have arisen in these languages too; the phonetic conditions were identical. But in contrast to French, such a declension has never surfaced in any of these Romance languages. The typologically deviant structure +NOM Π -ACC, if it ever existed in some southern variety of Vulgar Latin, was eliminated well before the appearance of written documents. Such a structure is simply not attested, and I doubt that it ever existed even in a marginal transitory form. The effect of the "invisible hand" of typological rules becomes even more evident if we consider the case of Latin neuters. As is well-known, all neuters in Indo-European languages have identical forms for the nominative and the accusative. Consequently, there is no difference in the speed of phonetic erosion between these two case forms. Most Latin neuters end in -m, but some of them end in -s. It is remarkable that in these latter cases the final -s has been preserved, as it was in other cases where there was no functional reason opposed to this. Final -s was eliminated precisely in those cases where its presence was in contradiction with the universal tendencies of nominal case marking. Consider the following examples from Medieval Spanish and Modern Sardinian:

30

Georg Bossong

(8) (a) murus —• muro tempus —• tiempos murum —• muro tempus —• tiempos cantas cantas cantat —> cawta {Old Spanish} ( b ) murus —* muru murum —*• muru cantas —> cantas cantat —> cantat {Sardinian/ dialect

tempus —*• tempus ['tempuzu] tempus —> tempus ['kantaza] ['kantaöa] of Nuoro}

The loss of final -s in the singular masculine noun has nothing to do with phonetic laws. It is exclusively conditioned by typological marking rules. Wherever there is no contradiction between phonetic form and functional distribution of markers, as in the Latin neuters in -s, the final consonant was preserved for a long time (Spanish) or for the whole length of attested history (Sardinian). Where phonetic form and marking rules are in harmony, as in the Spanish verb conjugation (3rd persons tend to be less marked than 2nd persons), the final consonant has survived until now even in Spanish (at least in most of its standard dialects, but this is another question). But wherever there was a contradiction between phonetics and functionality, elimination of the aberrant forms took place, probably already in a "prehistoric" period, i.e. before the appearance of written documents. Note that in Sardinian the final -s was even phonetically reinforced by the addition of a "paragogic" vowel which echoes the vowel of the syllable immediately preceding it. Far from disappearing, the final -s of a neuter like tempus is fully vital in Northern Sardinian up to the present day. In Old French, the exceptional and typologically deviant pattern represented as ( 9 ) Nominal marking/ Old French:

+ N O M Π -ACC

(in contradiction with the implication N O M

z> A C C )

is not characteristic of the Old French noun system in its totality but only for a relatively substantial, part of it. This constellation applies neither to masculines of other declension classes nor feminines nor plurals in general. After all, it is a residual structure which survived in a niche of the sys-

Nominal and/or verbal marking of central actants

31

tem, but which in the long run was doomed to disappear under the pressure of universal typological tendencies, as happened long before in other Romance languages.

2.2.2. Nominal marking in ergative languages In the ergative configuration, the distribution follows the same fundamental pattern as in the accusative one. As for nominal marking, privative oppositions prevail almost exclusively. Almost any ergative language could be quoted as an example. Classical Tibetan and Sumerian are perfectly representative in this respect:

(10) bdag-gis sijon nespa-0 cizig byas I-ERG before sin-ABS which? make 'Which sins have I committed in a former life?' {Classical Tibetan} (Π) ensik-e barag- 0 i-gul=gul prince-ERG shrine-ABS PERF-destroy(REDUPL) 'The prince destroyed the shrines' {Sumerian} Cases of equipollent marking are rare and difficult to find. One of the few convincing examples is Limbu, a Sino-Tibetan language spoken in Nepal, Sikkim and the Indian province of Darjeeling: (12) üfh-ille syaPI-εη nandhag lottu camel-ERG jacal-ABS opposite side carried 'The camel carried the jackal to the other side of the river' {Limbu} Georgian, which is ergatively structured in its past tense group, shows two equipollent marking patterns with one common member: the ending -i is opposed as a nominative marker to the accusative -s in the present tenses, and as an absolutive marker to the ergative -ma in the past tenses. (I leave

32

Georg Bossong

aside further complications of Georgian, which has been analysed as "dual" by some linguists; cf. e.g. Hewitt 1987, Lazard 1995.) Compare the following example with example (5) quoted above:

(13)

ρ 'ropesor-ma

da-c 'er-a

c 'ign-i

professor-ERG PRV-wrote book-ABS 'The professor wrote a book' {Modern Georgian} As Bernard Comrie informs me (personal communication), Nias, an Austronesian (Western Malayo-Polynesian) language spoken on the like-named island off the coast of Sumatra, seems to be another case in point. Comrie writes (on the base of Pamela Leanne Brown's ANU doctoral dissertation): "This really seems to be a language with a marked absolutive. Noun phrases appear in two forms, unmutated and mutated, with good arguments for treating the mutated form as marked (not only is the unmutated form used in citation, but by far the more economical direction of derivation is unmutated —> mutated). Subjects/agents of transitive verbs appear in the unmutated form; intransitive subjects/single arguments and direct objects/patients of transitive verbs appear in the mutated form. Not all consonants mutate, so there is some incidence of neutral case marking, but wherever ergative and absolutive are distinct, it is the absolutive that is marked vis-ä-vis the ergative." I thank Bernard Comrie for having brought this point to my attention. Be it in a privative or in an equipollent constellation, the distribution of nominal markers always follows the general rules. To the best of my knowledge, there is no language which marks systematically the absolutive but leaves unmarked the ergative. An exception of the kind of Old French does not seem to exist in the ergative configuration, or if it exists somewhere, it is extremely rare.

2.2.3. Verbal marking in accusative languages We have now to discuss the possibilities of verbal marking of the fundamental relation. In the accusative configuration, the rules formulated above work perfectly well, apparently without exceptions. As for the ergative con-

Nominal and/or verbal marking of central actants

33

figuration, we still lack in-depth studies which would permit us to affirm for certain that these rules can claim universal (or quasi-universal) validity. Nevertheless, in spite of the absence of such studies, the overall impression prevails that for ergative languages, too, the probabilities stated above remain valid. As for the accusative configuration, there seems to be no natural human language where only the object is cross-referenced in the verb. In our familiar Indo-European languages, as in a great many languages all over the world, the subject function is marked in the verb, either exclusively, or in combination with verbal marking of the object function. To the best of my knowledge, there is no exclusive verbal marking of the object in any accusative language. Exclusive verbal marking of the subject is so widespread and familiar that just a few trivial examples from Indo-European will suffice here for illustration; subjects and objects differ in number, in order to show that only subjects, not objects are verbally marked:

(14) (a) ucitel0 vide-t ucenik-ov teacher-NOM see-3SG AGT pupil-ACC PL ( =GEN PL) 'The teacher sees the pupils' {Russian} (b) mard-0 za-än-rä mi-bin-ad man-NOM woman-PL ACC' PRS-see-3SG AGT 'The man sees the women' {Persian} d-en (c) d-ie Männ-er seh-en ART-ACC SG ART-NOM PL men-NOM PL see-3PL AGT Junge-n boy-ACC 'The men see the boy' As I have outlined elsewhere (Bossong 1998 a, b), there is no essential, only a gradual, difference between grammemes which are recognised in traditional grammar as conjugation morphemes and grammemes of the clitic type. Both are (more or less) integrated in the verb. As was pointed out at the beginning, the degree of boundedness of grammemes is not considered in this contribution. Therefore, clitics count as verbal marking, just

34

Georg Bossong

as conjugational morphemes do. Consequently, the following examples represent the constellation where verbal marking refers both to the subject and to the object: (15) (a) a Juan lo-ha-n visto ACC' John 3SG PAT-have-3PL AGT seen los professores ART professors 'The professors have seen John' {Spanish} (b) l-a-m väzut pe Ion 3SG PAT-have-lSG AGT seen ACC' John Ί have seen John' {Rumanian} (c) eel su-m-ja kniga-ta read+PST be-lSG AGT-3SG PAT book-ART Ί have read the book' {Bulgarian} Frequently, verbal marking of the second actant is differential: only objects of a certain kind are marked in the verb, especially definite, sometimes also human/ animate objects. This is the case of the well-known and frequently discussed "object conjugation" of Hungarian:

(16) olvaso-k jelentes-t I read/INDEF PAT report-ACC ~ olvaso-m a jelentis-t I read/DEF PAT ART report-ACC Ί read a report ~ I η d the report' {Hungarian} It can be observed that the nominal mark of the accusative is obligatory, whether the object is definite or not. Differentiality only appears in the verb. In certain Bantu languages, too, differentiality is exclusively expressed in the verb, but in contrast to Hungarian, the noun is never marked, either for indefinite or for definite objects:

Nominal and/or verbal marking of central actants

(17) ngi-bona umuntu ~ lSGNOM-see person ngi-ya-m-bona 1SG N0M-TMP-3SG ACC'-see Ί see a ~ the person' {Zulu}

35

umuntu person

In Hungarian, the opposition between the two conjugations is morphologically equipollent, which is in accordance with the essentially inflectional character of this language; in Zulu, the opposition is privative, in accordance with the agglutinative character of most Bantu languages. But whatever the morphological structure, in both languages the functions are accumulated when passing from mono- to bivalence: the grammemes of the second actant are functionally added to those of the first actant, exactly as our typological model predicts.

2.2.4. Verbal marking in ergative languages Ergative languages tend to have pluripersonal conjugation systems: the verb refers to more than one actant in the great majority of ergative languages. Nevertheless, there are examples of verbal marking of only one actant even in this configuration, and in these cases it is the first actant, i.e. the absolutive, which is marked. Avar, the most important Daghestanian (North-Eastern Caucasian) language, belongs to this type. Agreement with the first actant is not made according to its person, but according to its class membership. Every noun belongs to one of three classes: I (human masculine), II (human feminine), III (non-rational human, non-human). The verb agrees with these classes, regardless of the grammatical person. The following examples illustrate this constellation: (18) (a) ebel-0 y-ac-ula ~ mother-AE ABS-come-PRS ci-yass ebel-0 y-ac-ula man-ERG mother-ABS I ABS-come-PRS 'The mother is coming. ~ The man brings the mother'

36

Georg Bossong

(b) padisahXi-0-gi Ximala-cca kodo-be kingship-ABS-THEM children-ERG hands-in b-os-ana III ABS-take-PRT 'As for kingship, the children took it into their hands' {Avar} The fundamental asymmetry between first and second actant, which we have seen at work in many accusative languages, holds true for this ergative language, too. It is not the specific semantics of the second actant which counts (agentivity vs. patientivity), it is its secondary character. The absolutive is unmarked in the noun, but marked in the verb; conversely, the ergative is unmarked in the verb, but marked in the noun. Avar, as a typical representative of the ergative configuration, presents a mirror-image of the familiar structures of accusative languages, but on the basis of the same hierarchy between first and second actants. The typological predictions outlined in Table 2 and Figure 4 are fully borne out. As was already stated, among ergative languages pluripersonal conjugation systems abound. They are frequent in the languages of the Ancient Near East, the Caucasus, Siberia, the Himalayas, Australia, and North America. In such languages, two or three, or in a few cases as many as four actantial relations are expressed by verbal grammemes. Three examples will illustrate this constellation: Basque, Ubykh, and Abkhaz (see Bossong 1982, 1984). Ubykh is a language of the North-Western Caucasian language family which has recently become extinct and was residually spoken in Turkey since the 19th century; Abkhaz belongs to the same family and is still spoken and written in its original homeland. Incidentally, there are many surprising structural parallels between Basque and the North-Western Caucasian languages. Basque and Ubykh both show bilateral marking: the absolutive and ergative functions are marked in the noun as well as in the verb. On the contrary, marking is unilateral in Abkhaz; in this language, nouns are not marked for case, all actantial functions being expressed in the verb. In spite of this lack of nominal marking, the configuration is as unambiguously ergative as in Basque or in Ubykh. Note, besides, that in Ubykh and Abkhaz the pronoun is not inflected for case, in contrast to Basque. This aspect will not be discussed here. The examples are based on the same meaning for all three languages, making the structural differences and affinities even clearer:

Nominal and/or verbal marking of central actants

37

(19) (a) (ni-0) n-oa / gizon-a-0 d-oa ~ I-ABS 1SG ABS-go / man-the-ABS 3ABS-go (ni-k) gizon-a-0 d-akusa-t / I-ERG man-the-AB; 3ABS-see-lSG ERG / gizon-a-k (ni-0) n-akusa-0 man-the-ERG I-ABS 1SG ABS-see-3ERG {Basque} (b) say°a) s-k"an / a-tdt-0 a-k"an ~ I [ABS] 1SG ABS-go / the-man-ABS 3ABS-go (s3y°a) a-tdt-0 a-s-byan/ I [ERG] the-man 3ABS-1SG ERG-see / a-tat-dn (sdy°a) sd-0-byan the-man-ERG I-ABS 1SG ABS-3ERG-see {Ubykh} (c) (sara) s-cot' / a-wy°a d-cot' ~ I [ABS] 1SG ABS-go / the-man [ABS] 3ABS-go (sara) a-wy°d dd-s-bot' / I [ERG] the-man [ABS] 3ABS-1SG ERG-see / a-wy°d (sara) sd-y-bot' the-man [ERG] I [ABS] 1SG ABS-3ERG-see {Abkhaz} Ί am going./ The man is going. ~ I see the man./ The man sees me'

3. An outline of the evolution of Romance In this chapter, I shall briefly outline the essential points of the diachronic evolution of the Romance languages, starting from Classical Latin. In the perspective of fundamental relation marking, this development depends crucially on the presence or absence of nominal and verbal grammemes for marking subjects and objects. Two converse diachronic movements can be observed. - On the one hand, the nominal declension of Latin decays. As a result, nominal marking of fundamental relation disappears completely, at least in a first stage of evolution. In some Romance languages a new type of nominal marking makes its appearance, the marking of animate and/or definite objects with a preposition (pe in Rumanian, a in most other varieties, such as Portuguese, Spanish, Catalan, Languedocian, Sardinian, and

38

Georg Bossong

dialects of Southern Italy), a phenomenon which I have called "differential object marking". This leads to a second stage of evolution where second actants are marked in the noun whereas first actants are not, in complete accordance with the typological predictions. - On the other hand, the verbal conjugation of Latin has been much better preserved. As a result, most varieties have kept the original Latin grammemes (verb endings) marking the first actant. In some varieties, these original grammemes have been replaced by prefixed subject clitics, developed from original pronouns; such is the case in modern French, in RhetoRomance, and in most dialects of Northern Italy, such as Piedmontese, Lombardian, and Ligurian. In addition to this verbal subject marking, most if not all Romance languages have developed a new verbal object marking, also from original pronouns which have become cliticized. These object clitics may be more or less bound, or obligatory, depending on the individual languages, and even registers within individual languages. In general, these clitics are more frequent in spoken than in written language. In some varieties, such as Modern Spanish, Sardinian, and especially Rumanian, the degree of boundedness and obligatoriness is such that these grammemes can be considered already as a fully-fledged "object conjugation". In spoken French, too, the frequency of verbal object marking is much greater than its frequency in the literary language, where its development has been restricted by the influence of normative grammatical traditions. Compare the following utterance, quoted by Lucien Tesniere as an instance of authentic spoken French (compare Heger 1982): (20) il-la-lui-a donnee, he-3 SG-NOM-she-3 SG-ACC-he-3 SG DAT-have given 0-Jean, α-Pierre, 0-sa-moto NOM-John DAT-Peter ACC-his-motorbike 'John has given his motorbike to Peter' {Spoken French} Raymond Queneau has compared this kind of "Neo-French" with a Salish language of the Pacific coast in his well-known, half serious essay "Parlezvous Chinook?" (1965, see Langenbacher 1981); in 1979, I drew attention to the structural parallelism of such a constellation with the structure of Abhaz. In a sentence like (20), the expression of the fundamental relation has become purely verbal, the nouns being unmarked for subject and direct

Nominal and/or verbal marking of central actants

39

object. In languages with differential object marking, however, such verbal marking of the object is combined with nominal marking; see the examples of Spanish and Rumanian quoted above (15 a, b). Be it in combination with subject verb endings, as in Spanish, Rumanian, Sardinian etc., or with subject clitics, as in French, Romansh, Lombardian etc., the verbal marking of the object is always realized by more or less grammaticalised clitics. The expression of the fundamental relation in the Romance languages has become predominantly verbal. - Romance based Creoles have evolved in a different direction, but on the whole they are not as profoundly distinct from "normal" Romance as is frequently suggested. The verb system inherited from Latin has completely collapsed. As in Spoken French, subject clitics have taken over the role of verbal subject marking in most varieties. There is less verbal object marking than in "ordinary" Romance, but this phenomenon is not completely unknown in Creole languages either; Principense, a Portuguese Creole of the Gulf of Guinea, is a case in point. Nominal marking of the object function occurs only marginally; we have quoted Papia Kristang as a Creole variety with a fully-fledged differential object marking (compare example (3, a-c)), but it should be stressed that this is an exceptional case. Differential object marking has arisen in this Portuguese Creole under the influence of Bazar Malay, which has itself undergone the influence of Hokkien (Min), a Sinitic language brought to South-East Asia by immigrants from Fujian. All in all, explicit marking of fundamental relation in the Romance Creoles is less developed than in Romance languages in the narrow sense, but it is not as insignificant as is frequently assumed. Romance Creoles can confidently be included in the general typological patterns of morphosyntactic evolution in Romance. They are omitted from the following patterns because of lack of research in this domain, and also for reasons of space. The preceding discussion is visualised in the following table, which presents the essential points of fundamental relation marking in the Italic branch of Indo-European: (21) (a) Latin First actant (NOM) Second actant (ACC)

nominal marking +Ν +Ν

verbal marking +V -V

40

Georg Bossong

(b) Common Romance base nominal marking verbal marking - Ν (+ Ν*) First actant (NOM) + V -Ν Second actant (ACC) -V * valid for a part of the nominal paradigm of Old French and Old Provenfa (partial typological irregularity) (c) Romance languages without DOM (stage I) First actant (NOM) Second actant (ACC)

nominal marking -N -N

verbal marking + V -V

(d) Romance languages with DOM (stage I) First actant (NOM) Second actant (ACC)

nominal marking -N -/+ Ν

verbal marking + V -V

^e) Romance languages with grammaticised object clitics (stage II) First actant (NOM) Second actant (ACC)

nominal marking -N /+ Ν

verbal marking + V + V

With the exception of the two-case declension of medieval GalloRomance, all patterns found in the evolution of Romance are in accordance with the universal typological rules, as postulated above. The constellation (22) -Ν Π +V (first actant) / +N Π -V (second actant) {stage I languages with DOM} illustrates perfectly the fundamental asymmetry of marking which is the topic of this contribution. But neither the constellation

Nominal and/or verbal marking of central actants

41

(23) +N Π +V (first actant) / +N Π -V (second actant) {Latin} nor the constellation (24) -Ν Π +V (first actant) / +N Π +V (second actant) {stage II languages with DOM} contradicts the assumptions of typological theory. If the evolution of Romance is viewed in such a perspective, a surprising fact becomes obvious, which has never before been evidenced with such clarity; it can be visualised in the following way:

Latin first actant second actant

bilateral marking unilateral marking

Romance II with DOM —> —•

unilateral marking bilateral marking

Figure 5

This is an immediate effect of the asymmetry presented at the beginning: on the one hand, nominal marking declines in the course of evolution from Latin to Romance, while verbal marking is preserved and even develops beyond the limits of what was found in Latin. The fundamental asymmetry we are considering can be reformulated in yet another way. There is a positively preferred marking pattern, which is +V for the subject and +N for the object; as for subject nouns and object cross-referencing verbs, there is no such strong positive preference, but only an indirect condition, an implication, as formulated in Table 2. This means that both -N and +N are permitted for subjects, and both +V and -V for objects, as long as the respective implicational conditions are fulfilled. In the evolution of Romance, this has the following consequences: - Constellations disallowed by implicational rules are doomed to disappear; once eliminated, they are never replaced.

42

Georg Bossong

- Constellations allowed by implicational rules freely appear or disappear. - Positively preferred constellations tend to be maintained or, if they disappear for some mechanical reason, they tend to be replaced. Such replacements are strong indicators for typological tendencies. They show that certain constellations are not simply allowed, but positively preferred. With these concepts in mind, let us have another look at the evolution of Romance. Nominal subject marking is allowed as long as there is nominal object marking; consequently, there was no contradiction as long as the accusative ending existed. As soon as it was eliminated for mechanical reasons, the nominative ending also had to disappear; it was eliminated very early in languages such as Spanish and Sardinian, and considerably later in French and Provenfal. But the essential point is that it was never replaced, in no variety of Romance whatsoever. It disappeared once and for all. Verbal object marking did not exist in Latin. It came into being in a relatively late stage of Romance development, due to a process of cliticization, as frequently occurs in human languages. Such a development is quite natural, but it was not a necessary one; it was optional, and in most Romance languages which have undergone it, it is still a far cry from full grammaticalization. Since it is not in contradiction with typological implications, it can develop freely in some languages (such as Spanish and Rumanian), whereas it remains severely restricted in other languages (such as Standard Italian). Verbal subject marking is a preferred pattern according to typological theory. In the course of Romance evolution, it was preserved in many languages, such as Ibero-Romance, Provencal, Italian, Sardinian, and Rumanian. It has partially disappeared because of mechanical phonetic erosion in Modern French. It is remarkable that in this language it has been replaced by a process of cliticization, in perfect analogy to what has just been stated for object marking: je-chante/ tu-chante(s)/ il-chante has replaced cant-o/ cant-as/ cant-at, and nowadays on-chante is even replacing nons-chant-ons. In Northern Italy, subject clitics have been added to the still existing verbal endings; in Piedmontese, for instance, we have the following double marking pattern: i-cant-o/ it-cant-e/ a-cant-a. Far from being eliminated, verbal subject marking has been substantially reinforced in this group of Romance languages. This is in perfect harmony with its typologically preferred character.

Nominal and/or verbal marking of central actants

43

Nominal object marking is also a preferred pattern. Accordingly, it comes as no surprise that the lost Latin accusative ending has been replaced by a new kind of marking, which is differential and only applies to an animate and/or definite object. This kind of replacement is typologically quite natural, and consequently we find it independently in most areas of Romania, although not everywhere with equal strength. Structural parallelism in combination with formal diversity is a strong indicator of the effects of universal tendencies. The basic rules of differential object marking are constant in the Romance language family (and beyond, of course), although its formal realization may vary a great deal. Apart from the already mentioned prepositions p(r)e in Rumanian and a in Portuguese, Standard Spanish, Catalan, Occitan, Engadinian, numerous Italian dialects, and Sardinian, we find replacement forms like enta in Aragonese, onde in Peruvian Spanish, da in the Gallo-Italian dialect of Nicosia in Sicily, and the already quoted kung (from Portuguese com) in Papia Kristang. All this proves that differential object marking is a recurrent pattern in diachronic evolution, not only of the Romance family, but, as we can confidently affirm, of human language in general. Typological preferredness leads to the constant re-creation of nominal object marking; once it has disappeared for some mechanical reasons, it surfaces again in a different shape.

4. Conclusion: towards an explanation of the asymmetry As a tentative conclusion, we might ask ourselves about the possible reasons for this asymmetry. In theory, no logical possibility can be excluded a priori. One could expect that all logical possibilities which follow directly from such universal properties of human languages as linearity and meaningfulness are equally legitimate, and therefore equally widespread among the languages of mankind. In a deductive approach, no one logical possibility predominates. One could easily imagine linguistic systems where nominal and verbal marking of the fundamental relation is distributed differently from what we empirically observe. But the reality is there, in all its irreducibility, and it asks for an explanation. Empirical data always prevail over theoretical predictions. This is not the place to go into the combination of the deductive and inductive approaches to language typology and universals. I merely wish

44

Georg Bossong

to conclude by indicating possible directions in which to look for a solution. The finite verbal predicate is the place where the relationship between the sentence and extralinguistic reality is established. The assertive speech act as such, independently from particular meanings, is bound to the finite verb of the main predicate in the sentence. According to the felicitous expression coined by Lucien Tesniere, the verb is the "node of nodes" (le nceud des nceuds), the structural centre of the sentence. I would even go beyond this affirmation, which belongs to the domain of grammar, and venture a step into philosophy. According to Wittgenstein, "the world is what is the case" (die Welt ist alles, was der Fall ist, Tractatus 1). "To be the case" is expressed by the finite verbal predicate and by the relations which it establishes with the "subject", or better, with the first actant. The relation between the verbal node and primary argument constitutes the sentence as an assertive act. The relation between the predicate and the first actant is constitutive for the assertion. Typologically speaking, there are languages which do not mark any actant relation in the verb, whereas others mark all kinds of actants. But if there is one, and only one, relation to be marked on the verbal predicate, it will be the first actant, which is primary in the constitution of the sentence. The configurational semantics of the first actant varies from language to language; it can correspond to the pattern of accusativity or of ergativity, according to the system, or subsystem, of the language in question. What counts is that it is hierarchically superior. Seen from the perspective of the verb, the first actant is privileged for being related with it, since this relation constitutes the sentence as an assertive act. The typological predominance of first actant marking in the verb can logically be deduced from this universal property of language. Conversely, in the nominal domain, the first actant is the least relational of all actants. Seen from the perspective of the noun, it is opposed to the verbal predicate with the highest degree of autonomy, the second actant function being much more integrated in it. If one and only one actantial function remains unmarked in the noun, it should be the first actant. Once more, this is independent from the specific semantics of the accusative and the ergative configuration. The second actant (the "object" in our accusative languages) is in all respects the exact contrary of the first actant. The valency relation which it has with the verbal predicate is less central; it is not constitutive for the sentence as an assertive act. Consequently, cross-referencing it in the verb is

Nominal and/or verbal marking of central actants

45

felt as less urgent. Conversely, the noun itself depends much more on the verb and is more integrated in it if it happens to fulfil the second actant function. Explicit marking of this function in the noun itself is a strongly preferred tendency. With these -necessarily sketchy - remarks it is hoped that the fundamental asymmetry discussed here can find the beginnings of an explanation. As for methodology, the necessity of combining theoretical reflection with empirical research will once more have become obvious.

References Baxter, Alan 1985 A description of Papia Kristang (Malacca Creole Portuguese). Canberra: ANU. Bossong, Georg 1979 Prolegomena zu einer syntaktischen Typologie der romanischen Sprachen. Festschrift Kurt Baldinger zum 60. Geburtstag, I, 54-68. Tübingen: Niemeyer. Bossong, Georg 1980 Syntax und Semantik der Fundamentalrelation. Das Guarani als Sprache des aktiven Typus. Lingua 50: 359-379. Bossong, Georg 1982 Actance ergative et transitivite. Le cas du systeme verbal de l'oubykh. Lingua 56: 353-386. Bossong, Georg 1984 Ergativity in Basque. Linguistics 22: 341-392. Bossong, Georg 1998a Vers une typologie des indices actanciels. Les clitiques romans dans une perspective comparative. In: Paolo Ramat & Elisa Roma (eds.), Atti del XXX Congresso Internazionale della Societä di Linguistica Italiana, Pavia, 26-28 settembre 1996, 9—43. Roma: Bulzoni [Pubblicazioni della Societä di Linguistica Italiana 39], Bossong, Georg 1998b La typologie des langues romanes. In: Günter Holtus & Michael Metzeltin & Christian Schmitt (eds.), Lexikon der romanistischen Linguistik, vol. VII, 1003-1019. Tübingen: Niemeyer.

46

Georg Bossong

Bossong, Georg 2001 Ausdrucksmöglichkeiten fur grammatische Relationen. In: Martin Haspelmath & Ekkehard König & Wulf Oesterreicher & Wolfgang Raible (eds.), Sprachtypologie und sprachliche Universalienforschung. Ein internationales Handbuch, vol. 1, 1. Halbband, 657668.Berlin: de Gruyter. Charachidze, Georges 1981 Grammaire de la langue avar. Paris: Jean Favart. Dumezil, Georges 1967 Etudes abkhaz. Paris: Adrien-Maisonneuve [Documents anatoliens sur les langues et les traditions du Caucase 5]. Heger, Klaus 1982 'II la lui a donnee, ä Jean, son pere, sa moto' - Neue Überlegungen zu einem alten Beispiel. In: Festschrift für Johannes Hubschmid zum 65. Geburtstag, : 53-66. Bern: Francke. Hewitt, Brian George 1987 Georgian: ergative or active? In: Robert M.W. Dixon (ed.), Studies in ergativity, 319-340. Amsterdam: North Holland. Klimov, Georgij A. 1977 Tipologija jazykov aktivnogo stroja. Moskva: Nauka. Langenbacher, Jutta 1981 Das "neo-frangais": Sprachkonzeption und kritische Auseinandersetzung Raymond Queneaus mit dem Französischen der Gegenwart. Frankfurt: Peter Lang. Lazard, Gilbert 1986 Le type linguistique dit 'actif. Reflexions sur une typologie globale. Folia linguistica 20: 87-108. Lazard, Gilbert 1995 Le giorgien: actance duale ("active") ou ergative? Typologie des verbes anti-impersonnels. Sprachtypologie und Universalienforschung 48:275-293. Lazard, Gilbert 1998 Actancy. Berlin: Mouton de Gruyter [EALT 19]. Queneau, Raymond 1965 Connaissez-vous le chinook? In: Raymond Queneau. Batons, chiffres et lettres, 57-63. Paris: Gallimard. Quesada, Felix 1976 Gramätica quechua Cajamarca - Canaris. Lima: Instituto de Estudios Peruanos. Sasse, Hans-Jürgen 1974 Ein Subjektkasus im Agaw. Folia Orientalia 15: 55-67.

Nominal and/or verbal marking of central actants

47

Sasse, Hans-Jürgen 1981 Die kuschitischen Sprachen. In: Bernd Heine et al. (eds.), Die Sprachen Afrikas, 187-216. Hamburg: Buske. Steiner, Gerd 1979 The intransitive-passival conception of the verb in languages of the Ancient Near East. In: Plank, Frans (ed.), Ergativity. Towards a theory of grammatical relations, 185-216. London: Academic Press. Tesniere, Lucien 1959 Elements de syntaxe structurale. Paris: Klincksieck. Wittgenstein, Ludwig 1921 Tractatus logico-philosophicus. London: Routledge & Kegan [reprint 1966, Frankfurt/Main: Suhrkamp], Van Driem, George 1987 A grammar of Limbu. Berlin: Mouton de Gruyter [MGL 4],

(In)transitivity and object marking: some current issues

Michela Cennamo

1.

Introduction

In this paper we discuss the relationship between the degree of (In)transitivity of the clause and the marking of one of its nuclear arguments, the Ο participant (actant Y in Lazard's terminology (Lazard 1984, 1994, 1997, 2002).' In particular, we consider some aspects of the coding of Ο in the intermediate area between prototypical transitive and intransitive clauses, exemplified by the various types of object incorporation, antipassive/middle morphology, direct vs. oblique case patterns, and we investigate whether and to what extent they can be insightfully accounted for within an 'event structure' perspective on Transitivity, following recent work by Rappaport Hovav and Levin (1998), Levin (1999), an approach also put forward, within different frameworks, in Croft (1998), (2000), Van Valin (1993), (2002), Van Valin and La Polla (1997), among others. Building on a recent proposal by Levin (1999), we show that bringing together the insights from typological research on Transitivity and Actancy (Hopper and Thompson 1980, Tsunoda 1981, 1985, 1994, int. al.), with some current views on argument structure and linking, allows for a better understanding and a principled, unified account of some aspects of the coding of Ο cross-linguistically. The discussion is organized as follows. Section 2 illustrates the notion of (In)transitivity and the approaches to event structure put forward in Rappaport Hovav and Levin (1998), Levin (1999), Van Valin and La Polla (1997), Van Valin (2002), Croft (1998), (2000), providing the general background for the issue to be investigated. Section 3 reviews the varying encoding of Ο crosslinguistically, and the semanto-pragmatic and some syntactic factors involved in it. Section 4 places the crosslinguistic variation in the marking of Ο within the event structure perspective developed

50

Michela Cennamo

by Rappaport Hovav and Levin (1998), Levin (1999). And section 5 summarizes the conclusions.

2. The (In)transitivity continuum and event structure 2.1. Transitivity/Intransitivity as gradients The notions of Transitivity/Intransitivity are strongly intertwined, and represent opposite and at the same time complementary poles of one and the same continuum, concerning the grammar of participants and the states of affairs they are involved in (Hopper and Thompson 1982: 1-5). In our discussion we follow Hopper and Thompson's (1980) multifactorial, scalar view, developed and further refined in subsequent studies (Givon 1984, 1990, Hopper and Thompson 1982, Lazard 1984, 2002, Nocentini 1992a, Tsunoda 1994, int. al.), whereby Transitivity is a cline, determined by the interplay of a number of semantic parameters, illustrated in table 1, which are variously encoded in the morphosyntax of the clause in the languages of the world. Transitivity therefore is a characteristic of the clause, rather than the verb, unlike in its traditional formulation, and is determined not only by the effective transfer of the action from an Agent (A) ('subject') to the Patient (O) ('object'), but by a number of other semantic dimensions, often interconnected, such as the nature of the situation denoted by the verb/predicate (e.g., its telicity, modality, polarity), as well as the degree of Control, Affectedness and Individuation (table 2) of the core participants (S, A, O): Table 1. The Transitivity continuum - Hopper and Thompson (1980: 252) A. Participants B. Kinesis C. Aspect D. Punctuality E. Volitionality F. Affirmation G. Mode H. Agency I. Affectedness of 0 J. Individuation of 0

High Transitivity two participants or more (A and O) action telic punctual volitional affirmative realis A high in potency Ο totally affected Ο highly Individuated

Low Transitivity one participant non-action atelic non-punctual non-volitional negative irrealis A low in potency 0 not affected 0 non-individuated

(In)transitivity and object marking

51

Table 2. The Individuation Hierarchy -Timberlake (1977: 162) INDIVIDUATED proper human, animate concrete singular count Referential, definite

NON-INDIVIDUATED common inanimate abstract plural mass non-referential

Not all the semantic parameters pointed out in Hopper and Thompson (1980) are equally relevant in determining the morphosyntactic encoding of clauses cross-linguistically (Tsunoda 1985). As pointed out by Givon (1984), (1990), Tsunoda (1985), (1994), Lazard (1998), (2002), among others, the main semantic parameters affecting syntactic transitivity involve a) the presence of two highly individuated participants, A and Ο (i.e., of two distinct participants ranking high on the Animacy, Definiteness and Referentiality hierarchies), b) the Agentivity and Control of the A participant over the situation described by the verb, c) completeness/perfectivity and realis modality/affirmative polarity of the situation expressed by the verb, d) high degree of Affectedness of the Ο participant, that undergoes a change of state. 2 The morphosyntactic correlates of the semantic parameters illustrated in tables 1-2 involve a) the linguistic expression of the core participants of a clause, b) their coding (case-marking/adpositions), c) the occurrence of related patterns (passives, antipassives, reflexives, reciprocals), d) the presence of cross-reference markers, e) verb morphology, f) word order (Tsunoda 1994: 4671). Also among the morphosyntactic reflexes of semantic transitivity some features are more relevant and/or more frequently attested than others. For instance, whereas the number of participants and their case-frames are relevant to the morphosyntactic encoding of clauses in all languages, and constitute the nub of the traditional definition of Transitivity, the occurrence of special verb morphology and cross-reference markers are not relevant for all languages, and word order plays a marginal role, being mainly relevant in languages with so-called 'inverse' voice, such as the Algonquian languages (Givon 1994, Klaiman 1991, Tsunoda 1994: 4672). Also syntactic reflexes such as passivization, antipassivization and reflexivization are not found in all languages. When they do occur, however, they generally apply to highly transitive verbs/clauses and tests to

52

Michela Cennamo

ascertain the transitivity/intransitivity of the clause. In the Australian Aboriginal language Bandj alang, for instance, antipassive, normally applying only to two-argument verbs, also occurs with cognate object verbs with a 'highly specific understood O' (e.g., juuma 'smoke', yarrbi 'sing') (la), which can never be overtly expressed, i.e., with verbs with which the canonical transitive construction (with Ο overtly expressed and A in the ergative case) is impossible (lb) (Austin 1982: 38-39; 46), thereby revealing their transitivity despite the fact that these verbs never allow an O, just like intransitive verbs:

(1) a. ngay gala juuma-le-ela I-NOM this-NOM smoke-ANTI-PRES Ί here am smoking (a cigarette)' b. mali-yu dandaygam-bu yarrbi-ni that-ERG old man-ERG sing-PAST.DEF 'That old man sang (a song)' In Russian the passive is possible with resultatives (i.e., (causative/active) accomplishments) such as kill, build, but not with stative verbs (e.g., perception verbs), which do not have a passive participle. With such verbs only the active is possible, with topicalization of the Ο participant, which occurs in sentence initial position (Fici Giusti 1994: 43):

(2)

Ivana videli vcera Ivan-ACC see-PAST-3PL yesterday 'They saw Ivan at the cinema yesterday'

ν at

kino the-cinema

The active (3a) is also preferred to the passive (3b) when both A and Ο are high in Individuation (e.g., human, and proper names) (Comrie 1989: 8182):

(3) a. Masu ubila Tanja Masha-ACC kill-PAST-3SG Tania-NOM 'Tania killed Masha'

(In)transitivity and object marking

b. Masa byla Masha-NOM be-PAST-F.SG Tanej Tania-INSTR 'Masha was killed by Tania'

53

ubita kill-PPP-F.SG

Activity verbs, on the other hand, take a different passive form, the reflexive passive. (For instance, the activity verb Stroit' 'build' occurs with the reflexive passive (daca stroit-sja (rabocimi) -dacha-NOM build.3SG-RFL worker- (INSTR-PL)- 'The dacha is (being) built (by the workers'), whereas the active accomplishment verb postroit' 'build', takes the auxiliary byt' 'be'+perfective participle (daca by-l-a postro-en-a rabocimi dacha-NOM be-PAST-F.SG build-PPP-F.SG worker-INSTR-PL - 'The dacha was built by the workers') (Van Valin and La Polla 1997: 657, note 4; see also Keenan 1985: 254). As pointed out by Van Valin and La Polla (1997: 657, note 4), there seems to be a continuum of acceptability with the passive of activity verbs. A sentence such as (4a), with a generic subject, a specific passive agent and a durational adverbial phrase, for instance, is ungrammatical for some English speakers, whereas it is possible when no time interval is specified (4b), though only with a generic agent, that indeed seems to be the relevant parameter (4c) (example from Peter Austin, p.c.) (see also discussion in section 3): (4)

a. * spaghetti was eaten by Anna for five minutes b. spaghetti is eaten by Italians c. pork was eaten by these people/Anna (?) for years, but then they/ she gave it up

Transitivity and Intransitivity therefore are gradients, determined by the interplay of syntactic, semantic, and pragmatic parameters, encompassing the relationship between situations and the participants therein involved, the nature of the participants themselves, as well as the type of clause linkage and clause-clause linkage (e.g., dependent clauses, gerunds, participles) and the type of cross-clausal reference maintenance (e.g., zero anaphora pivot, switch reference) that in some languages affect the coding of twoparticipant clauses, which are coded as 'less transitive' (Silverstein 1976, 1993). Although our discussion deals primarily with the first two aspects, the last two parameters too (may) play a major role in transitivity, as we

54

Michela Cennamo

shall see in section 3.4 for the marking of Ο in dependent clauses (see also discussion in Silverstein 1976, 1993, Austin 1992, 1998, int. al.). One can grade clauses, and locate them along a continuum, usually conceived of as an axis with two poles, illustrated in table 3, that refers to the type of syntactic construction encoding events and their participants: Table 3. Transitivity continuum- Lazard (1997: 260, 1998: 75) semantics: prototypical action

non-prototypical action with two participants

one participant w

grammar:

active

passive or equivalent antipassive or equivalent

uniactant construction

At one pole of the transitivity gradient there occur prototypical transitive clauses (Construction Biactancielle Majeure according to Lazard 1998: 74), denoting an action carried out by a highly agentive, individuated, human participant, that strongly affects an Ο participant, most typically nonhuman, [± Animate], determining some change in it. Following Croft (1991, 1998, 2000), these patterns may be characterized as involving an asymmetric transmission of force from one entity to another entity, determining a change in it. Languages, in point of fact, tend to show a striking convergence in the encoding of this type of situation, which in morphologically accusative and ergative constructions is marked, respectively, by means of the NOM-ACC/ERG-ABS case-frame on the nuclear participants (A/O), A vs. Ο agreement, and a verb form expressing a non durative, completed, realis, affirmative state of affairs: (5) a. the hunter killed the deer/Mark opened the door b. lögreglan tok {Icelandic, Van Valin 1991: 150) the-police-NOM take-PAST.3PL Siggu fasta Sigga-ACC fast-ACC 'The police arrested Sigga' c. ji: shrestha-Φ syä-nä (Newari\ Tsunoda 1994: 4672) I-ERG Shrestha-ABS kill-PERF Ί killed Shresta'

(ln)transitivity and object marking

55

Prototypical transitive clauses, therefore, involve a highly Agentive, animate A participant, discontinuous with previous discourse in a main predication, acting volitionally onto an Ο participant, which is distinct from it, determining some change in it. The verbal process refers to a real, complete, punctual, actual and affirmative situation. Therefore prototypical transitive clauses most typically include verbs such as the English kill, break, wound, clean, melt, move. Verbs such as hit, love, see, read, seek, though usually regarded as transitive in the traditional definition of transitivity, are less transitive, in that they do not involve a change of state in the Ο participant, or not necessarily, like hit. This is shown by their different morphosyntactic behaviour cross-linguistically. In many languages, in fact, they take different case-frames, with Ο in the dative/instrumental case, or omitted/incorporated/marked by an adposition, according to the language. A verb such as hit, for instance, takes the ergative-dative case-frame in Newari (Sino-Tibetan, Nepal) (6a) (Tsunoda 1994: 4672). The same caseframe obtains in Warlpiri with verbs of 'seeking' (e.g., warri-rni 'seek, look for'), whose Ο is unaffected by the verbal activity and may also be non-referential (6b) (Simpson 1991: 326) ('intensional' objects in Simpson's terms): (6) a. ji: shrestha-yätä I-ERG Shrestha-DAT Ί hit Shrestha'

dä—yä hit-PRES

b. ngarrka-ngku ka-rla karli-ki warri-rni man-ERG PRES-3DAT boomerang-DAT look for-NPAST Ά man is looking for a boomerang' (Hale 1982: ex 44a; Simpson 1991:326) In Warlpiri verbs of impact/concussion/contact may occur in either the Ergative-Absolutive or the Ergative-Dative case-frame, the so-called 'conative' alternation (Guerssel et al. 1985), according to the degree of affectedness of Ο (Simpson 1991: 327-333):

(7) a. ngarra-ngku ka marlu luwa-rni man-ERG PRES kangaroo.ABS shoot-NPAST 'The man is shooting the kangaroo' (Hale 1982: ex. 48a; Simpson 1991: 330)

56

Michela Cennamo b. ngarrka-ngku ka-rla-jinta marluku luwa-rni man-ERG PRES-3ERG-3DAT kangaroo-DAT shoot-NPAST 'The man is shooting at the kangaroo' (Hale 1982: ex. 47a; Simpson 1999:330)

This alternation resembles in meaning the 'conative' alternation in English (Levin 1993: 41-42), where verbs such as kick, hit occur in a construction where Ο is coded as a prepositional object (i.e., as an oblique3 argument), introduced by the preposition at, in order to denote an attempted, unsuccessful, uncompleted action, with Ο only partially affected (8b): (8) a. I hit John b. I hit at John In other languages verbs of impact such as hit occur in so-called object incorporation constructions (9), which signal that Ο is non-referential (Cooreman 1994, Foley and Van Valin 1985: 341, Van Valin and La Polla 1997: 123): (9)

x-0-mak-wi (Jacaltek; Foley and Van Valin 1985: 341) ASP-3 SG-ABS-hit-ANTI ix naj woman he 'He hits women (women-hits)'

In Croft's (1991) terms, these verbs involve only an asymmetric transmission of force, but no change on O, i.e., no 'impingement' on Ο (see also Tsunoda 1985: 389, 1994). Ο therefore is not the end of the verbal profile (see discussion in section 2.2), and is often encoded as an oblique (Croft 1998: 46). In his cross-linguistic study of the case-marking of twoargument verbs, Tsunoda (1981), (1985) points out that deviations from the canonical encoding of A and Ο reflect verb classes, and increase with verbs that depart from the prototypical transitive template, realized by change of state verbs (so-called resultatives/(causative/active accomplishments), whose Os are highly affected), as with non-resultatives (activity verbs such as hit, shoot), creation, consumption verbs {write, eat) in their activity use (i.e., with an indefinite O), stative verbs (e.g., perception, pursuit, knowledge and feeling). The lower the Ο locates along the affectedness scale, the more likely it is to occur in an oblique case (cf. (6b)-(9) above).

(In)transitivity and object marking

57

At the other pole of the transitivity continuum there occur canonical intransitive patterns, i.e., patterns with only one participant, S. Whereas it is relatively straightforward to characterize prototypical transitive clauses, it is not easy to give a unitary definition of prototypical intransitive patterns. Even if we confine ourselves to only one of the functional domains of Intransitivity, namely that of eventivepredicates (Stassen 1997), it is difficult to single out syntactic, semantic and pragmatic properties which characterize prototypical intransitive clauses. They comprise, in fact, different patterns, such as meteorological verbs (Italian piove -rain.3SG — 'it rains'), that in some languages require the presence of a dummy subject/pivot holder (English it rains, German es regnet 'it rains'), or a reflexive marker on the verb (Russian sverkaetsja -daybreak.3SG-RFL- 'it daybreaks', Italian si fa buio -RFL make.3SG dark- 'it is getting dark'), as well as different types of impersonal constructions The latter may comprise either only one participant, S, occurring in the accusative and/or dative case (German es freut mich - i t please.3SG I.ACC- 'it pleases me', Icelandic mig/mer langar ad fara -I. ACC/DAT wish.3SG.IMP to go- Ί wish to go', mig kelur I.ACC freeze.3SG.IMP- Ί am freezing', mer konar- I.DAT get.3SG.IMP cold- Ί am getting cold') (Smith 1996: 22), sometimes with the presence of a dummy subject, as in German, or they may involve the presence of two participants, A and O, occurring in different case-frames (Icelandic-/wer likar pu -I.DAT like.3SG.IMP you.NOM- Ί like you' (example from J. Barödal) (see Smith 1996: 28-29, Barödal forthc.), Russian, snegom pokrylo dvor -snow.-INSTR cover-3SG.NEUT ground-.ACC - 'the snow covers the ground'), the verb reverting to the unmarked third person singular (if finite), or occurring in the unmarked ('neutral') form, according to the language, if non-finite (most typically expressed as a past/perfective participle) (North Russian dialects, u otca svoix detej zabyto -at father.GEN. his child.PL-ACC forget-PPP-NEUT 'the father has forgotten his children') (Fici Giusti 1994: 69). Intransitives also comprise various types of impersonal-passive patterns, i.e., patterns with a passive morphology and different types and degrees of defocusing of A, O, S, which are either unexpressed (Italian fu discusso a lungo -be.3SG.PAST discuss-PP.M.SG at long- 'one debated for a long time'; Sorbian bu spawe-be.3SG.PAST sleep-PP -'sleeping took place' (Fici Giusti 1994: 130)) or surface as a dummy subject (German es wurde getanzt 'it was danced'). The defocused argument may also be expressed by the reflexive morpheme, as in Italian (cf. si e pagato molto-KFL be.3SG pay-PP.M.SG much - 'one has paid a lot', si epagati molto -RFL

58

Michela Cennamo

be.3SG pay.PP.M.PL much- 'one is paid a lot'), where the difference between the A-Sa/O-So orientation of the pattern (according to whether the verb is either divalent or monovalent) is signalled, respectively, by the lack vs. presence of past participle agreement with the underlying argument signalled by the reflexive morpheme (e.g., (si e) pagato (A-orientation: [AGR]) vi. (si e) pagati (O-orientation: [+AGR]); si e camminato -RFL be.3SG walk.PP.M.SG- 'one has walked' (SA orientation: [-AGR] ) vs. si e partiti - RFL be.3SG leave.PP.M.PL- 'one has left' (S 0 orientation: [+AGR]) (Cennamo 1984, 1993: 40, 1995, 1997). The term includes split-S and fluid-S marking systems as well. Oneargument intransitive verbs, in fact, in many languages do not constitute a homogeneous class, but subdivide into two major subclasses, labelled SA/Unergative, So/Unaccusative verbs, depending on the theoretical perspective adopted, and in particular on whether the phenomenon is regarded as mainly semantic, determined by the interplay of semantic parameters such as inherent lexical aspect, the Agent/Patient-like nature of the S argument, as well as its degree of Control and affectedness (Van Valin 1990, Klaiman 1991: 124-132, Mithun 1991, inter al.), syntactic, reflecting the syntactic nature of S in a level of representation of the clause (Dsubject/initial 1 with unergative/SA verbs, D-object/initial 2 with unaccusative/So verbs (Hale and Kayser 1987 for English, Perlmutter 1989) or syntactico-semantic, i.e., 'syntactically represented but semantically determined' (Perlmutter 1978, Levin and Rappaport 1989: 316, Levin and Rappaport Hovav 1995, among others). The distinction between S arguments behaving morphologically and/or syntactically, like the Ο argument of canonical transitive verbs, and S arguments with morphological and/or syntactic properties of A arguments of canonical transitive verbs may show up in different areas of the grammar, both synchronically and diachronically: auxiliary selection, as in Italian — where it also correlates with the presence vs. lack of past participle agreement with S, so that SA/unergative verbs select the auxiliary avere 'have' and lack past participle agreement (Giovanna ha camminato molto -Jane have.3SG walk.PP.M.SG much- 'Jane has walked a lot') whereas So/unaccusative verbs select the auxiliary essere 'be' and show past participle agreement (Giovanna e partita -Jane have.3SG leave.PP.F.SG - 'Jane has left') - and other European languages (Sorace 2000, inter al.), passivization, as in English (Perlmutter 1978), applicatives, as in some Australian Aboriginal languages (Austin 2002a) as well as word order, case-marking, crossreferencing, causatives (Dixon 1994: 70-83 and references therein).

(In)transitivity and object marking

59

In some languages intransitives show 'variable marking'(so-called fluid-S systems), according to the semantics of the situation expressed by the verb in a particular use (Dixon 1994: 78-83). In Tsova-Tush, an ergative NorthCentral Caucasian language, for instance, with some intransitives (e.g., fall) the case-marking of S reflects its degree of Control; absolutive (or nominative in Caucasian linguistic tradition) if S has no Control, ergative if S has high Control (Holisky 1987: 105, Comrie 1989: 59) (So voz-en-so -I-ABS fall-AOR-1 SG-ABS Ί fell down, by accident', As vuiz-n-as -I-ERG fallAOR-1SG-ERG Ί fell down, on purpose'). The Intransitive pole therefore seems to comprise patterns which differ greatly from one another, both semantically and syntactically, belonging to different planes, ranging from structures with no participant(s) involved, to patterns involving S and A-O participants, characterized by a deviant or 'quirky' encoding. Probably, as Tsunoda (1994: 4677) points out, the best definition of prototypical Intransitivity is a negative one, taking the transitive pole as the point of departure. There would therefore exist several types of intransitive patterns, satellites of canonical transitive patterns, all operating at different levels, and with distinct (and at times overlapping) identificational properties. In our discussion we consider only some aspects of the (In)transitivity axis, namely the coding of the Ο participant in intermediate areas of (In)transitivity, such as antipassives, object incorporation, direct vs. oblique marking of O, and so-called cognate/internal objects. Whereas it is possible to show the decrease in transitivity of these patterns by means of the transitivity gradient, i.e., along a linear axis (Lazard 1994, 1997, 1998) (see table 3) when the non-canonical encoding of Ο reflects semantic and pragmatic features, some instances of 'quirky'/oblique objects are difficult to locate along the same continuum, in that they seem to belong rather to different levels. In point of fact, they show a reduced syntactic transitivity, which does not appear to stem from semantic and/or pragmatic principles, but is (morpho)syntactically determined, applying quite systematically in the languages in which they occur.

2.2. (In)transitivity and event structure Recent work on event structure and argument linking, carried out within different frameworks, has pointed out that the varying encoding(s) of Ο

60

Michela Cennamo

may be interpreted as reflecting differences in the nature of the event described by the predicate and in the licensing of arguments. In this section we illustrate the main claims of three such event structure approaches to Transitivity, namely Rappaport Hovav and Levin's (1998), Levin's (1999) distinction between objects (Os) licensed by structure and constant participants, the Role and Reference Grammar notion of Macrorole Intransitivity, and Croft's (1991), (1998), (2000) notions o f f e r e e dynamic relations between participants and of verb profiling. We argue that these models offer interesting generalizations when confronted with crosslinguistic variations in the coding of O, highlighting different aspects of the event structure, with the notions of structure and pure constant participant representing a higher level of analysis from which the other notions may be derived.

2.2.1. Structure vs. constant participants Levin (1999), building up on the theory of event structure developed in Rappaport Hovav and Levin (1998), and in line with much current research on the structure of events and the licensing of arguments (Grimshaw and Vikner 1993; van Hout 1996, int. al.), regards argument realization as reflecting event complexity, which, in turn, only partially correlates with the number of arguments of a verb. The different morphosyntactic realizations of Os cross-linguistically therefore are regarded as resulting from their different event structure status. The starting point is the assumption that a verb's meaning is bipartite, and consists of a 'structural' and an 'idiosyncratic' aspect - the former representing the grammatically relevant aspect of the verb meaning (the so-called event structure template), common to other verbs of the same semantic class (i.e., of the same ontological type) - the latter representing the verb's core meaning, the 'constant' (Rappaport Hovav and Levin 1998: 107), which differentiates it from other verbs of the same semantic and grammatically relevant type. For instance, resultative (i.e., causative change of state) verbs such as the English kill, (trans.) break, have the same event structure template, illustrated in (10a), but differ in the realization of the constant, the state (represented between angle brackets) (lOb-c) (Rappaport Hovav and Levin 1998: 107):

(In)transitivity and object marking

(10) a. b. c.

61

[[x ACT ] CAUSE [BECOME [y ]\\ [[x ACT ] CAUSE [BECOME [y ]]] [[ χ ACT ] CAUSE [BECOME [y ]]]

Arguments therefore are regarded as belonging to two types. They may reflect the grammatically relevant aspects of the verb meaning, common to other verbs of the same class, i.e., they may realize so-called structural participants, or they may reflect the verb's core meaning, the constant, socalled constant participants. Verbs furthermore may realize complex causative events (e.g., English kill, break, open) or simple, non-causative events, regardless of the number of arguments they take (English walk, swallow, sweep, eat, write). The arguments of complex causative event verbs are licensed both by the verb's event structure template and by the verb's core meaning, i.e., they realize structure participants. With twoargument simple event verbs (e.g., English sweep), only the A argument realizes a structure participant; the Ο argument instead exemplifies a pure constant participant. The different event structure status of Ο with these verbs is said to account for the different behaviour of these verbs in English. Verbs such as kill, break do not allow omission of their objects4; verbs of surface contact through motion such as sweep5, and verbs of consumption such as eat, and creation like write, instead allow unspecified object deletion. In addition, the objects of the sweep class are not fixed, i.e., these verbs have a wider range of objects than verbs of the kill, break class (see discussion in Levin 1999: 237-238, 2000: 425-426): (11) a. b. c. d.

John killed the beetle *John killed John broke the chair *John broke

(12) a. John swept the floor b. John swept c. John swept the floor clean (13) a. Mary ate buns b. Mary ate c. Mary ate two buns

62

Michela Cennamo

The event structure templates accounting for the various event types comprise various subtypes of simple event structure templates and one complex event structure template (14) (Rappaport Hovav and Levin 1998: 108; Levin 1999: 229-230): (14)

Simple event structure templates: a. [x ACT< MANNER>] (one-argument activity) {run, sleep) b. [x ACT Y ] (two-argument activity) (sweep, wipe) c. [ χ ] (state) (love) d. [BECOME [x ]] (achievement) (be born, die, (intr.) (break)

Complex event structure template: e. [x ACT ] CAUSE [BECOME]]] [y ]]] (accomplishment) (kill, (trans.) break) Verbs such as kill, break have a complex event structure template (14e). They consist of two subevents, the causing event (an activity) and the resulting change of state. Verbs such as sweep, eat have a basic simple event structure (14b) in both their occurrences (with and without an object) (12ab) (13a-b). Of the two arguments associated with them, only the subject realizes a structure participant; the object realizes instead a pure constant participant (which is underlined). Their event structure template can be augmented, leading to the complex event structure template illustrated in (14e), as in their accomplishment uses (12c), (13c). Verbs of consumption such as eat, in fact, behave like activities with an indefinite object (13a), and like accomplishments with a definite one (13c).6 The different morphosyntactic behaviour of verbs such as kill, break and sweep, hit in English, and the cross-linguistic variability in the encoding of Os, therefore, is regarded as stemming from the different licensing of the Ο argument. Structure participants are marked uniformly in English and across languages in that their source is in the event structure template, and therefore they fall within the universal linking rules. The χ and y participants in causative complex event structure templates such as the one illustrated in (14e) above, are therefore realized by arguments encoded as subject and object (in languages where these relations obtain) (taking the NOM-ACC or ERG-ABS case-frame, depending on the alignment and case-marking system of a language or of patterns of a language). Pure

(In)transitivity and object marking

63

constant participants vary in their encoding in that they originate in the constant, so they fall within language specific, 'oblique' rules (Levin 1999: 241). In the absence of such rules they fall within a default linking rule, whereby they are marked as canonical objects (see Levin and Rappaport Hovav 1995: 154). This difference is claimed to account for the different encoding of Ο in core and non core transitive verbs in English and in several other languages. It would account for instance for the oblique marking (dative/instrumental case) of Ο with verbs belonging to some semantic classes in a number of languages (Russian, North-West Caucasian, Hungarian, German, Rumanian, as well as in Australian and Polynesian languages), namely mental process verbs, verbs of communication, verbs of ruling and disposition (Blume 1998), which rank lower than change of state verbs on the transitivity scale. What these verbs share is an Ο argument which realizes a pure constant participant (Levin 1999: 240-242). Similar (event structure) approaches are developed in Role and Reference Grammar (Foley and Van Valin 1985, Van Valin 1993, 2002, Van Valin and La Polla 1997) and in Croft (1991), (1998), (2000).

2.2.2. Macrorole Intransitivity and event structure In Role and Reference Grammar, a monostratal theory of grammar which posits direct linking between the syntactic and the semantic levels, mediated by the semantic macroroles of Actor and Undergoer, acting as the interface between thematic and syntactic relations (Foley and Van Valin 1984, 2002, Van Valin and La Polla 1997), the non canonical encoding of Ο (either by means of object incorporation or an antipassive construction, or an oblique case, according to the language), is interpreted as reflecting the M(acrorole) intransitivity of the verb/clause. A verb may take two arguments and have only one macrorole (see discussion in Van Valin and La Polla 1997: 147-153, Van Valin 2002: ch. 2). Resultative verbs such as the English kill, break (causative accomplishments in the theory)7, for instance, have two macroroles, an Actor and an Undergoer, realized, respectively as subject and object. In German and in several other languages (Icelandic, Russian, Latin, among others) non-resultative verbs such as helfen 'help', take a dative Ο and no corresponding passive pattern is possible (15b). Only an impersonal

64

Michela Cennamo

passive construction with the original Ο retaining its dative case is allowed instead (15c): (15) a. b. c.

er hat seinem Freund geholfen (German) *sein Freund wurde von ihm geholfen seinem Freund wurde von ihm geholfen

The morphosyntactic behaviour of verbs such as helfen is regarded as reflecting the Macrorole intransitivity of the verb, that is a two-argument activity event, with only one Macrorole, the Actor. The Ο argument does not have a macrorole status, and is therefore encoded as an oblique, behaving syntactically as a non Macrorole argument (as shown by the ungrammaticality of a corresponding passive construction). In addition, verbs such as eat, read, build, write, i.e., verbs of consumption and creation, are regarded as having only the Actor macrorole, and no Undergoer, despite their having two arguments, if Ο is non-referential {John eats buns, John reads thrillers). This is brought out by their different behaviour with respect to passivization in English and in several other languages: (16) a. b.

the buns were eaten by John ? *buns were eaten by John

These verbs are regarded as basically activity verbs entering the activity/accomplishment alternation, a phenomenon which is very widespread cross-linguistically. Unlike causative accomplishments (kill, (trans.) break), causative achievements (trans, crack) and causative activities (trans. bounce), active accomplishments do not denote a relationship of causation between two events, but the sequence of two events (see also note 6). Unlike the other causative classes, in fact, this type of alternation is never marked with a causative suffix cross-linguistically (Van Valin and La Polla 1997: lOlff). The Ο participant of multiple argument activity verbs qualifies the verbal action rather than identifying a particular referent in the Universe of discourse. In English and in other languages it is an inherent argument, in that it can be freely omitted. In languages with noun incorporation and antipassives, it is incorporated into the verb ((17b), (17d)), and/or deleted, with the verb occurring in the antipassive voice if it is non-referential. If it is referential instead, with some verbs it may be marked as an oblique core

(In)transitivity and object marking

65

argument (17f) (Van Valin and La Polla 1997: 123-124; 150 and discussion in section 3): (17) a.

b.

c.

d.

e.

f.

wichäsa ki chk ki kaksä-he (Lakhota) man the wood the chop-CONT 'The man is chopping the wood' wichäsa ki chk-kdksa-he man the wood-chop-CONT 'The man is chopping wood' or 'The man is wood-chopping' na'e haka 'e he sianä 'a e (Tongan) PAST cook ERG DEF man ABS DEF Ika fish 'The man cooked a/the fish' na'e haka-ika 'a e sianä PAST cook-fish ABS DEF man 'The man cooked fish' pisasa-m chay-dr (West Circassian, Comrie 1978) girl-ERG cherkesska-ABS yadd 3 SG.sew.3 SG.TRANS 'The girl is sewing the cherkesska' pisasa-r chsy-'3m yada girl-ABS cherkesska-LOC 3SG.sew.INTR 'The girl is sewing away at the cherkesska' (

With resultatives which take both Macroroles (causative/active accomplishments, causative achievements in RRG), the Actor and Undergoer occur, respectively, in the NOM-ACC and ERG-ABS case-frames, depending on the alignment of core arguments in the language. With multiple argument activity verbs with only one Macrorole, the Ο argument occurs in an oblique case and/or is omitted, incorporated, according to the morphosyntactic type of the language and to the transitivity parameters involved. Dative is regarded as the default case8 for non-Macrorole Ο arguments (see Silverstein 1976, 1981, 1993), and accusative/absolutive the cases for Macrorole Ο arguments in accusative/ergative constructions, respectively (see Van Valin and La Polla 1997: 352-370, Van Valin 2002: ch. 4).

66

Michela Cennamo

2.2.3. A three-dimensional model of causal/aspectual structures in argument linking Croft (1998), starting from a model of argument linking based on the notions of force dynamic relations among participants (Talmy 1985b) and of verbal profile (i.e., the segment(s) of the causal (force-dynamic) chain denoted (=profiled) by the verb), hypothesizes that argument linking is consistent cross-linguistically if "there is a straightforward force-dynamic relationship between two participants in an event and events are clearly individuated" (Croft 1998: 25). Cross-linguistic and intralinguistic variation in the coding of Ο reflecting semantic parameters such as Control of the A participant or degree of affectedness of Ο is regarded as reflecting the degree of "self-containedness of the event" (Croft 1998: 47). If the event is clearly self-contained (i.e., if the whole event is profiled by the verb), the participants are encoded as canonical core arguments. If the event is not clearly self-contained (i.e., if only some segment(s) of the event is (are) profiled), Ο is coded as an oblique. In this case, in fact, Ο is not the "endpoint of the verbal profile" (Croft 1998: 46). This would account for the direct-oblique case alternation in English (e.g. Engl. John hit the wall/hit at the ball), and in several other languages, where it may be expressed not only by an oblique case-marking on O, but by the antipassive morphology as well (see discussion in section 3). Four universal linking rules are posited (18), based on the antecedence relationship between participants (X > Y) in the force-dynamic chain (see Croft 1998: 24, 2000: 53): (18)

1. The verbal profile is delimited by Subject and Object (if any) 2. Subject > Object 3. Antecedent Oblique > Object > Subsequent Oblique9 4. Subject > Incorporated Noun > Object (if any)

In more recent work (Croft 2000) the notions of force-dynamic relations among participants and of verbal profile are integrated into a 'threedimensional' model of the causal-aspectual structure of events and of the participants therein involved. A situation is viewed as organized along three dimensions, a) its unfolding over time (t), b) qualitative change/state (Δ), c) the causal structure between the subevents that can be therein identified. A sentence with a prototypical transitive verb such as break (19) is therefore represented as in figure 1 below (Croft 2000: 59):

(In)transitivity and object marking

67

(19) Jane broke the vase

•

be broken

t impact t Δ = qualitative states/changes + causal/non-causal relations between participants t = unfolding of a situation and its subevents over time

Figure 1. Three-dimensional representation of the event structure of break (Croft 2000:59) A complex predicate such as break is decomposed into its subevents, each involving one participant and its aspectual contour, denoting the 'course' of a situation in its unfolding over time, from its beginning (i.e., its 'rest state') to its end (i.e., its 'result state'), represented as a dotted line (....). Solid lines in the figure ( | ) represent the 'phase' ( Ce njsrano c 'era i contadini => *Non ce li era there be-IMPF-3SG DEF farmer-MASC-PL 'There were the farmers' Cercava dei contadini => Ne cercava search-IMPF-3SG INDEF farmer-MASC-PL 'He was searching for farmers'

In conclusion, the typological survey given by Lazard and this specific case study show that the diachronic frequency of "actant H" constitutes a continuum between prototypical subject actants and prototypical object actants.9 Moreover, in (9b) c 'e begins to become impersonal in the sense described in paragraph 2.; in other words, we recognize an incipient process of 'de-personalization' (see Koch (1994b: 10)). However, it must be noted that the "actant H" of c 'e is not yet a prototypical direct object.

4. Engl, there is and Fr. falir/falloir One might ask whether the subject actant of a presentative verb can be transformed - thanks to the features of the "actant H" - into a prototypical direct object actant.10 The example from English (lOa/b) seems to correspond to this type of evolution. In this case, in addition to the lack of agreement, one finds in certain oral varieties of English the participant PT0 expressed using an oblique form of the pronoun (10c). "[...] il ne semble pas deraisonnable de considerer ce terme comme un objet [...]" [it does not seem unreasonable to consider this term as an object [...]] (Lazard 1994b: 13-14). This gives rise to the analysis (lOd): (10) c. Engl. there's himO0 - and there's youDO and meOO (Bolinger 1977:116) d. Engl. there is lionsuo in Africa We have already seen another example of de-personalization due to the transformation of subject into object in example (5a/b). Li plusor in (5a) is a form of the (plural) cas sujet in Old French, in agreement with the verbal form falloient (the pre- or post-verbal position of the actant has no bearing

160

Peter Koch

on the function in Old French). Whereas de I'argent in (5b) clearly functions as direct object, as shown by the loss of agreement, the postverbal position (significant in Modern French) and above all substitution of the interrogative pronoun qu'est-ce que ? (5c) (Tobler 1902: 214; Damourette and Pichon 1930-71: IV, 503-511; Koch 1995: 130): (5)

c. Mod.Fr.

—>

qu 'est-ce qu' il nous fautl what-ACC it us be-necessary-PRES3SG 'What do we need?' il nous faut de I 'argent it us be-necessary-PRES-3SG MASS money 'We need money'

Here we have a case of de-personalization performed with a presentative verb which originally expressed the participant PT0 as subject (Koch 1994b: 10). In terms of the corresponding frame Fj (which we could represent as in Fig. 2), the de-personalization o f f a l l o i r does not imply any change in structure of the participants in the frame. Leaving aside the participant expressed by the indirect object {nous in (5b) and (5c): see n. 15), which is optional and not involved in the syntactic change, we can say that there must at least be a participant PT0, which will be rhematic in an unmarked case. The only element that changes in this process of depersonalization is the syntactic realization of the participant PT 0 (syntactic level in Fig.l). From being an actantial function of the subject, compatible in Old French with the post-verbal position in the rhematic subject of a presentative verb, this becomes an actantial function of the direct object, which is invariably a post-verbal and prototypically rhematic actant in Modern French:11 (12)

rhematic S > rhematic DO

To adopt and adapt a term of Lucien Tesniere, we can call this process a 'diachronic syntactic metataxis' with respect to the verbal valency and actants.12 The constant element in this change is, in fact, the informational (rhematic) value of the actant expressed by PT0, and this guarantees the semantic continuity of the construct. At the same time the change in the verb Fr. falirlfalloir contains another element of continuity, in this case a formal one. The change comes about -

From subject to object andfrom object to subject

161

at least potentially - on the basis of an identical syntactic-morphological and phonic surface. We have seen that the syntactic-functional floating observable in the Italian verb esserci (9a/b) probably derives from an uncertain syntactic structure like (9c). In the case of Old Fr. falir, too, we find uncertain structures of this type: (13) a. OFr.

il ne me faut it not to-me miss-PRES-3SG Ί have got everything I want'

riens>H thing

Since rien is in the singular, the problem of agreement is left in abeyance all the more so since this form is also ambiguous from the point of view of the case (cas sujet/cas regime) marking the opposition between subject and direct object. Moreover, the presence of a formal element with the value of subject (pronoun) il cannot fail to call into question the subject status of rien. Thus already in Old French we can recognise an incipient depersonalization of falir and floating in the syntactic realization of the participant PT0 towards the zone of the "actant H". As I have shown elsewhere (Koch in press b), three factors come into play at this point that are observable in the evolution of Old French into Modern French: (i) a morphological factor of general significance: bi-casual declension disappears definitively at the very end of the Old French period. The lack of a morphological distinction between subject and direct object, which was fortuitous in (13a), becomes systematic. (ii) a syntactic factor of general significance: in Modern French the postverbal position of the direct object becomes standard. The post-positioning of an element such as rien in (13a), originally due merely to its rhematic value, suggests an interpretation as direct object. (iii) a specific lexical factor: already in Old French the concept expressed by falir, i.e. LACK, moves towards NEED in a metonymic change. On the basis of factors (i) and (ii) we witness a genuine reanalysis of the original subject o f f a l i r / f a l l o i r as direct object in Modern French (see also (5c)):

162

Peter Koch

(13)

b. Mod.Fr.

il ne me faut It not to-me be-necessary-PRES -3SG Ί do not need anything'

rienO nothing

With respect to (13a/b), we can speak of a 'reanalysis' in the syntactic sense of the term - "change in the structure of an expression or class of expressions that does not involve any immediate or intrinsic modification of its surface manifestation" (Langacker 1977: 58) - if we are prepared to accept that this concept of 'reanalysis' covers not just rebracketing but also syntactic re-categorizations, as for example metataxes of the S > DO type (see Lang and Neumann-Holzschuh (1999b: 6); Detges (2001: cap. 7)). However, Waltereit and Detges have shown that the reanalysis of a sequence of morphemes, as a process initiated by the hearer, is not a merely formal process: 13 (a) In terms of a "principle of reference", the reanalysis of a sequence, such as (13a/b), presupposes not only a formal continuity, but also a referential identity in the sense that the sequence must remain compatible, in spite of the reanalysis, with the referent in the discourse. In fact the hearer can refer the sequence (13a/b) to the same extra-linguistic state of affairs both with rien as subject and with rien as direct object. (b) In terms of a "principle of transparency", the receiver assigns to the sequence in question a new formal analysis motivated by a conceptual interpretation corresponding either to functional regularities of the language in question or to a more general principle. Often - but not always - reanalysis occurs together with a metonymic semantic shift. In terms of point (b), the case of (13a/b) does in fact represent an example of a metonymic shift LACK NEED (see above, factor (iii)). This metonymy concerns a conceptual aspect that is independent of the structure of the participants in the frame, which we can represent as in Fig. 3a and 3b:14 Every metonymic shift involves a figure-ground effect with respect to a conceptual frame (see Blank 1997: 232-243; Koch 1999a; 2001). In the case of the concept LACK ((13a) = Fig. 3a) a real aspect R, the ABSENCE of PT 0 , is the figure and a virtual aspect V, the idea of PRESENCE of PT0, is the ground, while in the case of NEED ((13b) = Fig. 3b) V becomes the figure and R the ground. An inter-linguistic comparison then shows that it

From subject to object andfrom object to subject

163

is more natural to realize the participant PT 0 - virtual - as rhematic direct object than as subject (for more details see Koch, in press b). Thus the metonymic shift in the verb falir/falloir seems to have favoured the reanalysis rhematic S > rhematic DO (12) in the sequence (13a/b), which remains unaltered, according to the "principle of reference" (a), on its surface.

Figure 3a. Conceptual structure of LACK (corresponding to (13a))

Figure 3b. Conceptual structure of NEED (corresponding to (13b))

If the analysis of Lazard (1994b) is correct, the example of English (lOd) does indeed involve a complete de-personalizing reanalysis S > DO, but in this case without any metonymic shift: both in (10a) and in (lOb/d) the verb expresses the concept of EXISTENCE. However, this reanalysis obeys the "principle of transparency" (b), since in modern English a post-verbal noun phrase lacking a preposition is very probably a direct object. We should recall that although It. c 'e (9b) floats towards the "actant H", this does not result in a reanalysis. Similar syntactic floating can be observed in the Old Fr. falir, where however floating corresponds to a preliminary phase, albeit a favourable one, 15 in the reanalysis S > DO. The reanalysis becomes definitive as a sudden, discontinuous process initiated by the metonymic shift - equally discontinuous - in the verb.

164

Peter Koch

5. Sp. hay, Cat. hi ha, Port, ha et al. It is natural to wonder whether this de-personalization is necessarily a oneway process, for we do in fact observe shifts in the opposite direction in various Romance languages. A particularly striking instance can be found in certain diatopic or oral varieties (in the broadest sense) of the Ibero-Romance languages16 (Rosario 1979: 52; Real Academia Espanola 1989: Wesch 1994: 326; BrauerFigueiredo 1999: 404; Koch/Oesterreicher 1990: 223; Koch 1994b: 10; 1995: 129):17 (14) a. Sp.

b. Sp.

(15) a. Cat.

b. Cat.

(16) a. Port,

b. Port,

habia be-there-IMPF-3 SG 'There were many soldiers' habian be-there-IMPF-3 PL 'id.'

muchos soldadosuo many soldier-MASC-PL muchos soldadoss many soldier-MASC-PL

acasa hi havia els llibresuo at house there be-IMPF-3SG DEF book-PL del meu cosi of-DEF my cousin 'At home, there were the books of my cousin' a casa hi havien els llibress at house there be-IMPF-3PL DEF book-PL del meu cosi of-DEF my cousin 'id.' vai havendo go-PRES-3SG be-there-GER menos carrosOO less car-MASC-PL 'There are going to be less and väo havendo go-PRES-3PL be-there-GER menos carros$ less car-MASC-PL 'id.'

cada vez every time

less cars' cada vez every time

From subject to object andfrom object to subject

165

Here too we can apply the structure represented in Fig. 2: it is essential to have at least one participant PTo, which will be rhematic in an unmarked case. The only change needed is the syntactic realization of the participant PTo. In actual fact this type of syntactic metataxis is the opposite of cases (9) and (10) because the direct object which expresses the participant PT0 of a verb of EXISTENCE/LOCALIZATION - compatible with post-positioning as object and also as rhematic (14a, 15a, 16a) - becomes the subject, compatible with the post-positioning as rhematic subject of a presentative verb (14b, 15b, 16b): (17)

rhematic DO > rhematic S

Unlike case (9) at least, in (14b, 15 b, 16b) we can recognize not only floating, rhematic DO rhematic H, but also an apparently genuine syntactic reanalysis of the subject as direct object, and hence a complete personalization. Significantly, we find perfect congruence even in a periphrastic and rather marginal form such as Port, väo havendo (16b), or an example of type (18) occurring in a corpus of spontaneous spoken Spanish from Caracas (Rosenblat 1979: 313): (18)

Sp.

"iNo hay presiones not exist-PRES-3SG pressure-PL ni nada de eso?" nor nothing of this "De ningun tipo" of no type 'Is there no pressure nor anything like that? - None at all' - " ι Antes habian?" -"Tampoco" before exist-IMPF-3PL neither - 'Before, there was? - Neither'

Irrespective of the fact that in this variety of Spanish the impersonal use of haber (hay with direct object presiones) still coexists with the personal use {habian), the latter form demonstrates a fairly thorough personalization of the verb, which does not rule out a highly thematic participant PT 0 (anaphoric reiteration of presiones by means of the personal desinence of habian)}* As reanalysis, these processes of personalization obey the "principle of transparency" (3.2., (b); see Detges 2001: 417). In fact the type of

166

Peter Koch

unmarked V-S sentence with presentative verb and rhematic subject (see 3.1.) is well rooted in the Ibero-Romance languages. However, the verbs Sp. haberlCat. Aaver/Port. haver are conditioned by their etymological provenance, which requires the realization of the participant PT 0 as direct object (see n. 18). Thus it seems only natural to reanalyse, according to formula (17), sentences of the type (14c), lacking visible agreement, as the most prevalent occurrences of presentative sentences (14d), and go on to derive from type (14d) a plural structure such as (14b): (14)

c. Sp.

d.Sp.

habia be-there-IMPF-3 SG 'There was a soldier' habia be-there-IMPF-3 SG 'id.'

un soldadoD0 INDEF soldier-MASC-SG un soldados INDEF soldier-MASC-SG

Here we have a process of syntactic reanalysis in reverse with respect to the one exemplified in (13a/b) for Fr. falirlfalloir. see (17) vs. (12). There is, however, a fundamental difference: in the case of Fr. falirlfalloir the definitive reanalysis is linked to a lexical metonymy (LACK NEED), while in the case of Sp. haber (and also Cat. haver-hi and Port, haver) there is no semantic-lexical shift. The concept expressed by these verbs 19 (EXISTENCE/LOCALIZATION) remains unchanged. It appears to be easier to reanalyse a verb of EXISTENCE/LOCALIZATION merely at the syntactic level, without any semantic change taking place. A type of reanalysis which is quite similar to (14) can be observed, in oral and/or Latin American versions of Spanish, with the verb hacer in a meteorological context (Rosario 1979: 52; Real Academia Espanola 1989: 384; Koch/Oesterreicher 1990: 223):20 (19)

a.Sp.

b.Sp.

hizo make-PERF-3 SG 'There were crisp hicieron make-PERF-3 PL 'id'

grandes heladasuo great frost-FEM-PL onsets of frost' grandes heladass great frost-FEM-PL

The meteorological verbs of the Romance languages Fr. faire, Sp. hacer, It. fare etc., accompanied by a substantive actant21 resemble verbs of

From subject to object andfrom object to subject

167

EXISTENCE in their presentative nature. Their actant - which can only be rhematic when unmarked - expresses a participant PT0 in the sense of Fig. 2. In this case too impersonal sentences are reanalysed according to formula (17), without visible agreement (19c), as occurrences of the personal type of presentative sentence (19d) deriving from type (19d) a plural structure like (19b):

(19)

c.Sp.

d.Sp.

hizo make-PERF-3 SG 'It was very hot' hizo make-PERF-3 SG 'id.'

mucho much

caloroo heat-MASC-SG

mucho much

calors heat-MASC-SG

The Spanish and Italian verbs expressing a clock's striking present another type of personalizing reanalysis associated with presentative sentences: (20) a.Sp.

b.Sp.

(21) a.It.

b.It.

da las tresOo give-PRES-3SG DEF three 'It is striking three o'clock' dan las tress give-PRES-3PL DEF three 'id.' suona le treOO sound-PRES-3SG DEF three 'It is striking three o'clock' suonano le tre§ sound-PRES-3 PL DEF three 'id.'

6. Germ, es gibt We can perhaps also associate with examples (19)—(21), and more especially (14)—(16), a process observable in certain dialects of German, where the main verb of EXISTENCE es gibt shows floating in the syntactic realization of the actant (viele Äpfel) expressing PT0 (see Paul 1968: 286; Brugmann 1917: 19, 26-27, 48 n. 2):

168 (22)

Peter Koch a. Germ.

es gibt dies Jahr nicht it give-PRES-3SG this year not viele ÄpfelO0 many apple-PL 'There are not many apples this year'

b. Germ.

es geben dies Jahr nicht it give-PRES-3PL this year not viele Äpfeln many apple-PL 'There are not many apples this year'

Since we are again dealing here with a verb of EXISTENCE, this process recalls the one observed with Ibero-Romance verbs of EXISTENCE ( 1 4 ) (16). It is likely that in this case too the syntactic uncertainty derives from a syntactic structure such as (22c), where the singular form of the verb and the syncretism between nominative and accusative cases22 of the noun leave the problem of agreement in abeyance: (22)

c. Germ.

es gibt dies Jahr nicht it give-PRES-3SG this year not genug Wasser enough water-SG 'There is not enough water this year'

When it comes to details, however, the syntactic conformation of German is different from that of Ibero-Romance languages, because German is a "verb second" language (X-V-Y) which does not allow, in affirmative sentences, a bare V-S structure as occurs in Ibero-Romance languages. It is only thanks to the "dummy subject" es that presentative verbs can be inserted into an es-V-S sentence with rhematic post-verbal subject (23b), although this structure is in competition with an S-V sentence such as (23a), where the (rhematic) subject carries the main emphasis. Both in (23a) and in (23b) - and also in the plural (23c) - Auto(s) is a genuine subject (Lazard 1994b: 16). (23)

a. Germ.

ein Autos kommt INDEF car come-PRES-3S 'There is a car coming'

From subject to object andfrom object to subject

b. Germ.

es DUMMMY 'id.'

kommt come-PRES-3 SG

c. Germ.

es kommen DUMMMY come-PRES-3PL 'There are cars coming'

169

ein Autos INDEF car

Autoss car-PL

If in (22b) we had a complete reanalysis DO > S, we could assimilate the structure of (22c) to that of (23b), and extend it to the plural by analogy with (23c). In (22a/c), however, we have an "expletive es" which constitutes a "dummy" but obligatory subject actant, as emerges in (22d) and (22e): (22) d. Germ.

(22) e. Germ.

dies Jahr gibt es nicht this year give-PRES-3SG it not viele Äpfel / many apple-PL / nicht genug WasserOO not enough water-SG 'This year, there are not many apples/there is not enough water' *dies Jahr gibt nicht viele Äpfellnicht genug WasserD0

In (23b/c), on the other hand, we have an optional "Vorfeld-es", which, when appropriate, allows itself to be replaced by pre-verbal elements (see Eisenberg 1999: 174-176; Lazard 1994b: 15). (23) d. Germ.

e. Germ.

jetzt kommt ein now come-PRES-3 SG INDEF 'Now, there is a car coming' jetzt kommen Autoss now come-PRES-3PL car-PL 'Now, there are cars coming'

Autos car

Moreover, this "Vorfeld-es" never appears in subordinate clauses which show regular S-O-V order:

170

(23)

Peter Koch

f. Germ, ich sehe, dass (* es) I see-PRES-lSG that kommen come-PRES-3PL Ί see that there are cars coming'

Autos $ car-PL

To decide whether in (22b) we have an "expletive es", as in (22a/d), or a "Vorfeld-es", as in (23b/c), we can refer to a phrase from the Urfaust (24) where a structure analogous to (22b) appears in a subordinate clause: there is no agreement between the verb {geben) and the actant expressing PT0 {mehr 'more (persons)' - necessarily plural).

(24)

Germ,

es ist it be-PRES-lSG mehr noch it more still 'He is a queer fellow like cit. Brugmann 1917: 19)

ein Kauz, wie's INDEF queer-fellow how geben. give-PRES-3PL several others' (Urfaust: v. 1175,

Since 's — es does not disappear in the subordinate clause, we are not dealing here with a "Vorfeld-es" but with an "expletive es". A subordinate version of (22b) would read, hypothetically:

(22)

f . Germ.

...weil's because it viele many PRES-3PL '... because there are

dies Jahr this year Äpfeln apple-PL

nicht not geben give-

not many apples this year'

While the agreement of geben with mehr in (24) and with viele Äpfel in (22f) invalidates their identity as direct object, the persistence of es as subject - at least on the formal level - rules out their being pure subjects. Going on the documentation we have ready to hand, we must say that viele Äpfel in (22b) has not undergone a complete reanalysis as subject, but has merely floated towards the zone of the "actant H".

From subject to object and from object to subject 171 7. Conclusions We have seen above some cases of diachronic syntactic metataxis due to both a de-personalizing reanalysis S > DO (10, 5/13) and a personalizing reanalysis DO > S (14, 19-21 and probably also 15, 16). This prompts three questions: ο is the direction of these processes indifferent? ο why in some cases do we not find a complete reanalysis, but only syntactic floating towards the zone of the "actant H" (9, 22)? ο what is the role (and is this optional?) of metonymic shift (Fig. 3a/b) in the reanalysis? Starting from the fundamental verbs of EXISTENCE such as It. c 'e (9), Sp. hay (14), Cat. hi ha (15), Port, ha (16), Engl, there is (10) and Germ, es gibt (22), we can say that they all, by definition, fall into the category of presentative verbs with a sole rhematic participant PT0. In all these languages two syntactic structures coexist: (i) verbs with post-verbal rhematic direct object: (S-)V-Orh, (ii) presentative verbs with post-verbal rhematic subject: (X-)V-Srh. This option is rather restricted in English, but it can be found, in marginal uses, especially when the "dummy" there is introduced before the verb: Here comes the sun; There arises the question whether ... (see n. 27). Thus in theory all these languages offer the possibility of a diachronic syntactic metataxis between the constellations V-Orh and V-Srh (17) or vice versa (12). Either way this would be achieved by means of a sentence type without visible agreement, i.e. with a rhematic actant in the singular, such as (9, 14, 22), which raises uncertainties of analysis between cases (i) and (ii). This is also the case for English which, starting from the rather unusual structure V-Srh, arrives at the more common form. We must, however, point out that these possibilities of diachronic metataxis exist merely in theory, and are limited and channelled according to the typological conformation of each specific language. This is clear from Tab. 1, which covers the Ibero-Romance, Italian, English and German material we have examined, and in addition the French ily a 'there is'.23 We can divide the languages into three groups in terms of serial typology: S-V-O/V-S24 (Spanish, Catalan, Portuguese, Italian), X-V-Y25 (German) and S-V-O26 (English, French). Moreover there is a distinction between languages in which the participant PT0 is originally realized as

172

Peter Koch

subject (S) of the fundamental verb of EXISTENCE and those in which it is originally realized as direct object (OD). In languages in which the original syntactic realization of the participant PT0 contradicts the expected syntactic structure, we observe a syntactic reanalysis. This is the case above all for English, where a postverbal noun phrase without preposition is, as a general rule, a direct object. But the same reasoning can also be applied, inversely, to the IberoRomance languages, where the noun phrase that follows a presentative verb is, in the unmarked case, a subject. As we saw in sections 4 and 5, these reanalyses obey the "principle of transparency", since they generalize the syntactic realization you expect to find in a given language (and, where appropriate, in a given semantic context). Table 1. Syntactic type and changes between S and DO PT0 originally realized as ... S

S-V-O/V-S language It. c'e: floating ('actant H')

X-V-Y language

DO

Sp. hay, Cat. hi ha, Port. ha: reanalysis > S (personalization)

Germ. Es gibt floating ('actant H')i OFr. (i) a: floating ('actant H')

S-V-O language Engl. There is: reanalysis > DO (depersonalization)

ModFr. Uya: DO = DO

In languages in which the original syntactic realization of the participant PT0 does not directly contradict the expected syntactic structure, we observe floating towards the zone of the "actant H", due to tension between (i) and (ii), without complete reanalysis ever being achieved. This is true not only for Italian, which, as well as the V-S structure (presentative), also has the S-V-O structure, but also, conversely, for German, featuring both S-V-0 and X-V-S. It is interesting to note that in Old French, an X-V-Y language not unlike German, the fundamental verb of EXISTENCE (i) a shows syntactic floating like that found in German: the participant PT0 is originally realized

Front subject to object and from object to subject

173

etymologically to Sp. hay, Cat. hi ha and Port, hä - but there are also occurrences in the cas sujet (25b). (25) a.OFr.

η' i out unDο nen 27 NEG there have-PERF one-CR NEG-there-of parolt speak-SUBJ-PRES-3SG 'There was no one who got to talking about that' (Voyage de Charlemagne a Jerusalem et a Constantinople: v. 812, cit. Buridant 2000: 84)

b.OFr.

n' en i out unsH NEG of there- there have-PERF one-CS28 d 'eus tot sous of them all single-CS 'There was none of them qui osast prendre ses who dare-SUBJ-IMPF take-INF POSS-PL-CR adous armour-PL-CR 'who dared to take up arms' (Beroul, Tristan: v. 137-128, cit. Buridant 2000: 84)

In actual fact, the actant in question has not undergone a definitive reanalysis, as can be seen in Modern French: (26) Mod.Fr. qu 'est-ce qu' il y a sur what-ACC it there have-PRES-3SG on la table? DEF table 'What is there on the table ?' —> sur la table il y a on DEF table it there have-PRES-3SG des miettes de pain. INDEF-PL crumb of bread 'There are bread crumbs on the table' The introduction of the interrogative form shows that the actant des miettes de pain in (26) has remained a direct object (see also Lazard 1994b: 10).

174

Peter Koch

We see from Tab. 1 that the S-V-0 nature of Modern French confirms the object identity of the post-verbal actant of il y a and has put an end to the syntactic floating of the Old French exemplified in (25b). Whereas the case of Fr. falirlfalloir is quite different. Originally a verb of EXISTENCE - or rather non-existence (5a), - falirlfalloir, as the basis for the reanalysis featured in (13) and (5b/c), expresses POSSESSION (negative).29 This concept involves at least two participants, PT0 (the POSSESSUM) and PTi (the POSSESSOR), as is seen in Fig. 3a and 3b. In principle, this conceptual relation of POSSESSION can be subject to two opposing informational orientations. For LACK (i.e. NEGATIVE POSSESSION), PT 0 can be chosen either as rheme (13a) or as theme (27). (27)

OFr. riens que j' aie, ne vos thing that I have-SUBJ-PRES-lSG not to-you faut lack-PRES-3SG 'You do not lack for anything that I have' (Chrestien de Troyes, Erec et Enide: v. 638, cit. AFW: s.v.falir, col. 1611).

Even though the constellation PT0 = rheme (13a) is more common, the constellation PT0 = theme (27) is always possible. If there is syntactic floating (S > H) in the realization of PT0 = rheme (13), the structure PT0 = theme (27) does not allow any oscillation and consolidates the status of subject of the actant which expresses PT0. As I have shown elsewhere (Koch, in press b), everything changes with the lexical metonymy LACK —> NEED described in Fig. 3a and 3b and exemplified in (13b) and (5b). In this case, it is the semantic change which initiates a definitive reanalysis S > DO. This process is probably favoured also by a certain analogy between il (me) faut qch. and il y a qch., and by the strong S-V-O tendency of Modern French, described above. However, the fact that the semantic change is decisive is confirmed by an onomasiological observation in this conceptual sphere: while Fr. falloir, after the change LACK NEED, has undergone the reanalysis S > DO, Fr. manquer, the diachronic successor to falloir in the sense of 'lack', is one of the few verbs in modern spoken French which, in spite of the strong S-V-O tendency, allows - integrated in the so-called "impersonal" construct (see n. 4) - a post-verbal actant to express PT0 = rheme - with syntactic floating, it is true, but without a definitive reanalysis:

From subject to object andfrom object to subject (28)

175

Mod.Fr. il manque de I 'eaun it lack-PRES-3SG MASS water 'We have not got any water'

Thus when considering changes we must distinguish those concerning the subject from those concerning the object: 1. syntactic floating due to tension between (i) and (ii), which, however, does not lead to a genuine reanalysis, contrary to the prevailing syntactic type: It. c'e{S > H); Germ, es gibt (DO > Η). 2. reanalyses due to the prevailing syntactic type: Sp. hay, Cat. hi ha, Port. hd (DO > S); Engl, there is (S > DO). 3. a reanalysis due primarily to a metonymic lexical change: Fr. falir/falloir (S > DO).

Notes 1.

2. 3. 4.

5.

6.

In categorizing noun phrases, I use the following abbreviations: S = subject; DO = direct object; 10 = indirect object; OBL = other oblique actant; Loc = locative actant; C = "circonstant", i.e. not depending on the valency of the verb, cf. Tesniere 1959. For the cas sujet in Old French I use CS. Analysis of Busse and Dubost (1983), s.v. parier. Construct which runs contrary to the prescriptive norm in German, but frequent in varieties that do not conform to this norm. It would be extremely interesting, for example, to analyse such grammatical shifts as Sp. se venden aceitunass > se vende aceitunasO0 / It. si vendono delle olive?, > si vende delle oliveDO 'people sell olives' (see Manoliu-Manea 1985: 85-86; Oesterreicher 1992: 248; Baciu 1993; Wehr 1995: 110-125, 185-213; Koch 1994b: 10; 1995: 132-133). Another clearly grammatical process that also goes beyond what we can deal with here is the so-called "impersonal" transformation of French sentences such as Tes elevess viennent > II vient tes eleves·, 'Your pupils are coming' (see also cap. 7., (28); see Heriau 1980; Riviere 1981; Gorzond 1984; Lazard 1994b: 5-10). For the concept of frame in cognitive linguistics, see Fillmore (1975; 1985); Barsalou (1992); Taylor (1995: 87-92); Ungerer-Schmid (1996: 205-217); Koch (1999a: 144-153). Cf. Lazard (1994a: 140-141). - This description is also valid for a language such as French, where - in oral use - the formal mark for the third person is a pre-positioned clitic (il pleut [i(l)-pl0]) and for languages like English and German, in which, in addition to a mark (suffix) for the third person, there is

176

Peter Koch

an obligatory third person pronoun (neutral). From the typological point of view, all these languages diverge from languages which allow impersonal sentences without any formal mark of subject (viz. Tahiti: see Lazard (1994a: 69, 140)). - Although I recognise that it would be more logical to call verbs of the type (8a) 'a-subjectal' (see Creissels (1991)), I retain the traditional denomination 'impersonal' which is less clumsy (also as the root for the terms 'personalization' and 'de-personalization' which I need further on). 7. See Hetzron (1975: 347-329); Contreras (1976: 52-55); Wandruszka (1982: 6-22, 52-59); Manoliu-Manea (1985: 82-83). The informational status of the verb itself in sentences with a presentative verb is controversial: some consider it thematic (Contreras; Wandruszka; Oesterreicher 1991: 324-325), while others hold that it forms, together with the actant that expresses PT0, a sentence with global rhematicity (Manoliu-Manea; Ulrich 1985). The crucial point for what follows is that the participant PT 0 is in any case rhematic in nature, or at least belongs to a rhematic sequence (see also K?sik (1991); Lazard (1994b: 4)). 8. For the status of ne in general see Lihn Jensen (1986), Koch (1994c: 181182); for the specific case of c e: Koch (1994b: 10; 1995: 129 n. 29). 9. We should emphasize that 'prototypicality' means, in this case, 'prototypicality at the level of the metalinguistic conceptualization of linguistic facts'. It is not a prototypicality at the level of conceptualization of our daily extralinguistic knowledge, as in the emblematic cases of BIRD, FRUIT, FURNITURE, etc. (Koch 1998). 10. I shall not go into the problem of the syntactic status of there, which has been interpreted as the true subject of the sentence (see Lakoff 1987: 468-469, 546-549, and the more nuanced interpretation of Lazard 1994b: 12-14). I must, however, point out that such an interpretation does not suit It. c 'e, whose structure is very similar to Eng. there is (Koch 1999b: 293-294). 11. Since in Modern French the post-verbal position of the direct object is practically obligatory, it cannot but be rhematic in an unmarked sentence such as j'ai trouve ton parapluie 'I've found your umbrella' (see Raible (1971)). This is, however, only a prototypical affinity which does not exclude other combinations of syntactic function and informational value in marked sentences, as for example ton parapluie, je l'ai trouve 'Your umbrella, I've found it'; see Oesterreicher (1991: 368-369). 12. Tesniere's concept of 'metataxis (Fr. metataxe) is synchronic-comparative and was coined to describe the syntactic divergences between two languages that co-exist in time. It seems only logical to extend it to the two diachronic stages of the same language; see Koch (1994a: 119-120; 1995: 126-133; in press a). We should observe that for Tesniere syntactic metataxes occur in the context not only of verbal valency (the subject of this paper), but of all syntactic relations. - For a more extensive concept of metataxis, taking in divergences at

From subject to object andfrom object to subject

13.

14. 15.

16.

17.

18.

19.

20.

177

the propositional and informational levels as per schema (6), see the works just mentioned. The processes of metataxis we have looked at here only involve diachronic divergences at the syntactic level (and continuity at other levels). See Waltereit (1999); Detges and Waltereit (1999); Detges (2001): cap. 7. On reanalysis in general, see Langacker (1977); Timberlake (1977); Hopper and Traugott (1993: 32-50); Harris and Campbell (1995); Haspelmath (1998); Lang and Neumann-Holzschuh (1999a and b). For PT] corresponding to nous in (5b/c) and to me in (13), see section 7. It has been widely held that reanalysis presupposes an ambiguity in the original morpho-syntactic structure (see Timberlake (1997: 142, 148); Harris and Campbell (1995: 72); Haspelmath (1998: 56-61)). However, Waltereit (1999) has shown that ambiguity is not the premise but rather the consequence of reanalysis. Detges (2001: 419-420) states that we should not confuse ambiguity (as a consequence of reanalysis) and formal uncertainties as a frequent motive for reanalysis (e.g. the syntactic floating in the zone of the actant Η for Fr. falirlfalloir). Type (14b) is found in low sociolects of peninsular oral Spanish and Latin American varieties of Spanish; types (15b) and (16b) belong to low sociolects, but also to low registers of Catalan and Portuguese. Although Sp. haber, Cat. haver and Port, haver come from Lat. habere 'to have' and in (14a, 15a, 16a) have kept the direct object etymologically required by this verb, I have annotated Sp. habia, Cat. hi havia and Port, (vai) havendo as 'be-there/there be' because in modern Ibero-Romance languages haber/haverίhaver as verb of POSSESSION has been replaced by Sp. /ewer/Cat. tenirfPort. ter and has survived only as a temporal and modal auxiliary (only modal in Portuguese) and as verb of EXISTENCE/LOCALIZATION (in Catalan haver-hi, in Spanish in the present hay). We can add that in Latin American Spanish the personalization of haber occurs also in the particular form of the present hay, blended with the archaic adverb y 'vi' Q quienes hayn adentro ? 'Who (PL) is inside ?') and even with other verbal persons: en la close habemos cuarenta estudiantes 'In our form, we are forty students' (Lapesa 1981: 589). The conceptual difference - by no means negligible - between EXISTENCE and LOCALIZATION also derives from a semantic shift (see Koch 1999b: 291-295), but one which has no relationship with the syntactic reanalysis (17) of these verbs. A similar reanalysis occurs with another use of Sp. hacer that was originally impersonal (expressing a PERIOD OF TIME): hace muchos ahos => hacen muchos ahos 'It was many years ago' (Rosario 1979: 52; Real Academia Espanola 1989: 384-385; Koch/Oesterreicher 1990: 223-224).

178

Peter Koch

21. Of course we are not concerned here with their use followed by an adjective as in Fr. il fait chaud 'It is hot', It .fa brutto 'The weather is ugly'. 22. In standard German this syncretism is found mostly in feminine and neuter nouns (such as Wasser) and in all plural nouns (like Äpfel). Only with a masculine singular noun is there no uncertainty (see Ger. es gibt dies Jahr keinen Schneeoo 'There is no snow this year': keinen = accusative). We should note, however, that in many Geran dialects there is syncretism between nominative and accusative also in the masculine singular. 23. Note that Tab. 1 only reflects the varieties of languages considered in which a shift actually takes place. 24. I.e. S-V-O languages which allow the V-S order in certain conditions (above all with presentative verbs). 25. I.e. "verb-second" languages. 26. I.e. S-V-O languages which allow the V-S order only in very special conditions (above all residual from the diachronic point of view). 27. CR = cas regime. 28. Cf. fh. 5. 29. Typically associated with POSSESSION (ALIENABLE) are other concepts such as INALIENABLE POSSESSION, PART-WHOLE etc. (Jacob, in press), which I shall not deal with here.

References AFW = Adolf Tobler and Erhard Lommatzsch 1925ss Altfranzösisches Wörterbuch. Berlin: Weidman (vol. 1-2) / Steiner (vol. 3ss). Baciu, Ion 1993 SE habla - SI parla - fale-SE - SE vorbe§te/On parle. In: Gerold Hilty (ed.), Actes du X)C Congres International de Linguistique et Philologie Romanes. Universite de Zurich (6—11 avril 1992), III, 17—23. Tübingen/Basel: Francke. Barsalou, Lawrence W. 1992 Frames, concepts, and conceptual fields. In: Adrienne Lehrer and Eva Feder Kittay (eds.), Frames, Fields, and Contrasts. New Essays in Semantic and Lexical Organization, 21-74. Hillsdale/London: Erbaum Berruto, Gaetano 1985 Per una caratterizzazione del parlato: l'italiano parlato ha un'altra grammatica? In: Günter Holtus and Edgar Radtke (eds.), Gesprochenes Italienisch in Geschichte und Gegenwart, 120150. Tübingen: Narr.

From subject to object andfrom object to subject

179

Blank, Andreas 1997 Prinzipien des lexikalischen Bedeutungswandels am Beispiel der romanischen Sprachen. (Beihefte zur Zeitschrift für romanische Philologie 285.)· Tübingen: Niemeyer. Bolinger, Dwight 1977 Meaning and Form. London/New York: Longman. Brauer-Figueiredo, Maria de Fätima Viegas 1999 Gesprochenes Portugiesisch. Frankfurt am Main: Teo Ferrer de Mesquita. Brugmann, Karl 1917 Der Ursprung des Scheinsubjekts 'es' in den germanischen und romanischen Sprachen. Berichte über die Verhandlungen der Königlich-Sächsischen Gesellschaft der Wissenschaften. Philologisch-historische Klasse 69/5 : 1—45. Buridant, Claude 2000 Grammaire nouvelle de l'ancien frangais. Paris: Sedes. Busse, Winfried and Jean-Pierre Dubost 19832 Französisches Verblexikon. Die Konstruktion der Verben im Französischen. Stuttgart: Klett. Chang, I-Lang 1994 Les constructions dites 'impersonnelles' en chinois. L 'information grammaticale 62 : 36-38. Contreras, Heles 1976 A Theory of Word Order with Special Reference to Spanish. Amsterdam etc.: North-Holland. Creissels, Denis 1991 Approche des constructions asubjectales, improprement designees comme 'impersonnelles'. In: Maillard 1991: 47-57. Damourette, Jacques and Edouard Pichon 1930-71 Des Mots a la Pensee. Essai de Grammaire de la Langue Frangaise. 7 vol. + Complements. Paris: d'Artrey. DaneS, Frantiäek 1964 A three-level approach to syntax. Travaux Linguistiques de Prague 1 :225-240. Detges, Ulrich and Richard Waltereit 1999 Grammaticalization vs. reanalysis: a semantic-pragmatic account of functional change in grammar, manuscript. Detges, Ulrich 2001 Grammatikalisierung. Eine kognitiv-pragmatische Theorie am Beispiel romanischer und anderer Sprachen. Tübingen: Habilitationsschrift.

180

Peter Koch

Dik, Simon C. 19792 Functional Grammar. Amsterdam etc.: North-Holland. Eisenberg, Peter 1999 Grundriß der deutschen Grammatik. II: Der Satz. Metzler: Stuttgart/Weimar. Faarlund, Jan Terje 1998 Sym&rie et dissymetrie des actants centraux. In: Feuillet 1998: 147-192. Feuillet, Jack 1996 Les types de fonctions. In: Gertrud Greciano and Helmut Schumacher (eds.), Luden Tesniere — Syntaxe structurale et operations mentales. Akten des deutsch-französischen Kolloquiums anläßlich der 100. Wiederkehr seines Geburtstages Strasbourg 1993, 129-136. (Linguistische Arbeiten 348.) Tübingen: Niemeyer. Feuillet, Jack (ed.) 1998 Actance et Valence dans les Langues de I 'Europe. (Empirical Approaches to Language Typology. EUROTYP 20.2.) Berlin/New York: Mouton de Gruyter. Fillmore, Charles J. 1975 An alternative to checklist theories of meaning. In: Proceedings of the Annual Meeting of the Berkeley Linguistics Society 1: 123— 131. Fillmore, Charles J. 1985 Frames and the semantics of understanding. Quaderni di semantica 4: 217-240. Gorzond, Ira 1984 Die Linguistik der unpersönlichen Ausdrücke unter besonderer Berücksichtigung des Französischen. (Tübinger Beiträge zur Linguistik 241.) Tübingen, Narr. Graal = Albert Pauphilet (ed.) 19802 La Queste del Saint Graal. Paris: Champion. Halliday, Michael Alexander Kirkwood 1985 An Introduction to Functional Grammar. London: Arnold. Harris, Alice C. and Lyle Campbell 1995 Historical Syntax in Cross-linguistic Perspective. (Cambridge Studies in Linguistics 74.) Cambridge: Cambridge University Press. Haspelmath, Martin 1998 Does grammaticalization need reanalysis? Studies in Language 22: 315-321.

From subject to object andfrom object to subject

181

Heriau, Michel 1980 Le verbe impersonnel en frangais moderne. 2 vol. Lille/Paris: Champion. Hetzron, Robert 1975 The presentative movement or Why the ideal word order is V.S.O.P. In: Charles N. Li (ed.), Word Order and Word Order Change, 345-388. Austin/London: University of Texas Press. Hopper, Paul J. and Elizabeth C. Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Jacob, Daniel in press 'Possession' zwischen Semasiologie und Onomasiologie. In: Andreas Blank and Peter Koch (eds.), Kognitive romanische Onomasiologie und Semasiologie. Tübingen: Niemeyer. K^sik, Marek 1991 Impersonnel et progression du texte. In: Maillard 1991: 183-192. Koch, Peter 1981 Verb • Valenz • Verfügung. Zur Satzsemantik und Valenz französischer Verben am Beispiel der Verfügungs-Verben. Heidelberg: Winter. Koch, Peter 1994a Verbvalenz und Metataxe im Sprachvergleich. In: Werner Thielemann and Klaus Welke (eds.), Valenztheorie - Werden und Wirkung. Wilhelm Bondzio zum 65. Geburtstag, 109— 124.Münster: Nodus. Koch, Peter Depersonalisation (et repersonnalisation). A propos de la 1994b diachronie des verbes impersonnels. L 'information grammatical 62:9-11. Koch, Peter 1994c L'italiano va verso una coniugazione oggettiva? In: Günter Holtus and Edgar Radtke (eds.), Sprachprognostik und das 'italiano di domani'. Prospettive per una linguistica 'prognostica', 175-194. (Tübinger Beiträge zur Linguistik 384.) Tübingen: Narr. Koch, Peter 1995 Aktantielle 'Metataxe' und Informationsstruktur in der romanischen Verblexik (Französisch, Italienisch, Spanisch im Vergleich). In: Wolfgang Dahmen et al. (eds.), Konvergenz und Divergenz in den romanischen Sprachen. Romanistisches Kolloquium VIII, 115-137. (Tübinger Beiträge zur Linguistik 396.) Tübingen: Niemeyer.

182

Peter Koch

Koch, Peter 1998

Prototypikalität: konzeptuell - grammatisch - linguistisch. In: Udo L. Figge, Franz-Josef Klein, and Anette Martinez Moreno (eds.) Grammatische Strukturen und grammatischer Wandel im Französischen. Festschrift für Klaus Hunnius zum 65. Geburtstag, 281-308. (Abhandlungen zur Sprache und Literatur 117.) Bonn: Romanistischer Verlag.

Koch, Peter 1999a Frame and contiguity: On the cognitive basis of metonymy and certain types of word formation. In: Günter Radden and KlausUwe Panther (eds.), Metonymy in Language and Thought, 139— 167. (Human Cognitive Processing 4.) Amsterdam/Philadelphia: Benjamins. Koch, Peter 1999b Cognitive aspects of semantic change and polysemy: the semantic space HAVE/BE. In: Andreas Blank and Peter Koch (eds.), Historical Semantics and Cognition, 279-305. (Cognitive Linguistics Research 13.) Berlin/New York: Mouton de Gruyter. Koch, Peter Metonymy: unity in diversity. Journal of Historical Pragmatics 2001 2/2: 201-244. Koch, Peter in press a Metataxe bei Lucien Tesniere. In: Vilmos Agel et al. (eds.) Dependenz und Valenz. Ein internationales Handbuch zeitgenössischer Forschung, Article 16. (Handbücher zur Sprachund Kommunikationswissenschaft) Berlin/New York: Mouton de Gruyter. Koch, Peter in press b II ne me faut plus nule rien - changement semantique, metataxe et ^analyse. Syntaxe & Semantique 4. Koch, Peter and Wulf Oesterreicher 1990 Gesprochene Sprache in der Romania: Französisch, Italienisch, Spanisch. (Romanistische Arbeitshefte 31.) Tübingen: Niemeyer. Lakoff, George 1987 Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. Chicago/London: University of Chicago Press. Lambert, Pierre-Yves 1998 L'impersonnel. In: Feuillet 1998: 295-345. Lang, Jürgen and Ingrid Neumann-Holzschuh (eds.) 1999a Reanalyse und Grammatikalisierung in den romanischen Sprachen. (Linguistische Arbeiten 410.) Tübingen: Niemeyer.

From subject to object and from object to subject

183

Lang, Jürgen and Ingrid Neumann-Holzschuh 1999b Reanalyse und Grammatikalisierung. Zur Einfuhrung in diesen Band. In: Lang and Neumann-Holzschuh 1999: 1-17. Langacker, Ronald W. 1977 Syntactic reanalysis. In: Li 1977: 57-139. Lapesa, Rafael 19819 Historia de la lengua espanola. (Biblioteca Romänica Hispänica III, 45.) Madrid: Gredos. Lazard, Gilbert 1981 Les structures de la phrase. In: Jacques Barbizet (ed.), Actes du colloque international et multidisciplinaire sur la Comprehension du langage, Cr6teil 25-27 septembre 1980, 4 3 45. (Collection 'Linguistique' 12.) Paris: Didier. Lazard, Gilbert 1994a L 'actance. Paris: Presses Universitaires de France. Lazard, Gilbert 1994b L'actant H: sujet ou objet? Bulletin de la Societe de Linguistique de Paris 89: 1-28. Lazard, Gilbert 1998 Definition des actants dans les langues europöennes. In: Feuillet 1998: 11-146. Li, Charles N. (ed.) 1977 Mechanisms of Syntactic Change. Austin: University of Texas Press. Lihn Jensen, Bente 1986 Clitico di soggetto, il clitico 'ne' in italiano? Memoires de la Societe Neophilologique de Helsinki 44: 147—157. Maillard, Michel (ed.) 1991 L'impersonnel. Mecanismes linguistiques et fonctionnements litteraires. Grenoble: Ceditel. Manoliu-Manea, Maria 1985 Tipologla e historia. Elementos de sintaxis comparada romänica. (Biblioteca Romänica Hispänica II, 337.) Madrid: Gredos. Oesterreicher, Wulf 1991 Verbvalenz und Informationsstruktur. In: Peter Koch and Thomas Krefeld (eds.), Connexiones Romanicae. Dependenz und Valenz in romanischen Sprachen, 349-384. (Linguistische Arbeiten 268.) Tübingen: Niemeyer. Oesterreicher, Wulf 1992 SE im Spanischen. Pseudoreflexivität, Diathese und Prototypikalität von semantischen Rollen. Romanistisches Jahrbuch 43: 237-260.

184

Peter Koch

Paul, Hermann 19688 Prinzipien der Sprachgeschichte. Tübingen: Niemeyer. Raible, Wolfgang 1971 'Thema' und 'Rhema' im französischen Satz". Zeitschrift für französische Sprache und Literatur 81: 208-224. Real Academia Espanola 198912 Esbozo de una nueva gramätica de la lengua espanola. Madrid: Espansa-Calpe. Riviere, Nicole 1981 La construction impersonnelle en franqais contemporain. (Documents de linguistique quantitative 41.) Paris: Jean-Favard. Rosario, Rubin del 19792 El espanol de America, Rio Piedras (Puerto Rico): Campos. Rosenblat, Ängel (ed.) 1979 El habla culta de Caracas. Materiales para su estudio, Caracas: Universidad de Caracas. Taylor, John R. 19952 Linguistic Categorization. Prototypes in Linguistic Theory. Oxford: Clarendon. Tesniere, Lucien 19591 Elements de syntaxe structurale. Paris: Klincksieck. Timberlake, Alan 1977 Reanalysis and actualization in syntactic change. In: Li 1977: 141-177. Tobler Adolf 19022 Subjektloses faut und fait. In: id., Vermischte Beiträge zur französischen Grammatik. Vol. 1, 213-218. Leipzig: Hirzel. Ulrich, Mioriia 1985 Thetisch und Kategorisch. Funktionen der Anordnung von Satzkonstituenten am Beispiel des Rumänischen und anderer Sprachen. (Romanica Monacensia 24.) Tübingen: Narr. Ungerer, Friedrich and Hans-Jörg Schmid 1996 An Introduction to Cognitive Linguistics. London/New York: Longman. Waltereit, Richard 1998 Metonymie und Grammatik. Kontiguitätsphänomene in der französischen Satzsemantik. (Linguistische Arbeiten 385.) Tübingen: Niemeyer. Waltereit, Richard 1999 Reanalyse als metonymischer Prozeß. In: Lang and NeumannHolzschuh 1999: 19-29.

From subject to object andfrom object to subject

185

Wandruszka, Ulrich 1982 Studien zur italienischen Wortstellung. Wortstellung — Semantik - Informationsstruktur. (Tübinger Beiträge zur Linguistik 193.) Tübingen: Narr. Wehr, Barbara 1995 SE-Diathese im Italienischen. (Romanica Monacensia 37.) Tübingen: Narr. Wesch, Andreas 1994 Elemente des gesprochenen Katalanisch. In: Axel Schönberger and Klaus Zimmermann (eds.), De orbis Hispani Unguis litteris historia moribus. Festschrift für Dietrich Briesemeister, I 309— 332. Frankfurt am Main: Domus Editoria Europea.

Changes of valence and their effect on objects1

Heidi Siller-Runggaldier

1. Preliminary considerations One feature of Italian is the large number of verbs which have a variable valence with respect either to the number of arguments, so that the same verb may occur with a variable number of constituents, or to the syntactic function of the arguments, meaning that the same verb may occur with the same number of constituents but with different syntactic roles. The two forms of valency alternation occur in various subtypes, which are distinguished above all by different degrees of semantic transitivity of the respective verbs. These different degrees of transitivity can, in turn, be identified on the basis of the parameters of transitivity established by Hopper and Thompson (1980). Thus this study does not include verbs whose change in valence is accompanied by a change of meaning (i.e. of their denotative value), so that there is no, or only a remote, semantic link between the various meanings. Thus for example the verb assistere, when used transitively, means 'be with sb. to help him; look after, help sb.' {assistere un ammalato 'nurse a sick person') or else 'aid sb.' (assistere il chirurgo 'assist the surgeon'); while when used intransitively it means 'be present at st.; intervene, participate' (assistere a uno spettacolo 'attend a show'). Nor does this study consider verbs whose changes of valence are due to contextual variants of their meaning, as is seen for example in the various uses of the verb andare (i bambini vanno a casa 'the children go home' I 'orologio non va piu 'the clock isn't working any more' - e una cosa che non va 'it's something which won't do') or the verbprendere (hopreso due valigie Ί took two cases' - il rampicante trapiantato non ha preso 'the transplanted climber didn't take' - e una colla che prende bene 'it's a glue that holds well'). These are verbs whose semantic variations are, as we shall see later, highly idiosyncratic, and thus not systematic or attributable to specific semantic classes.

18 8

Heidi Siller-Runggaldier

The valency variable considered in this paper may involve all three types of object, viz. direct objects, dative objects, recognised as such because they can be made into pronouns using dative clitics, and objectoids, which in the terminology I proposed in (1996a) represent the class of non-dative prepositional objects. Valency variability affects not only objects but also adverbials, and is enriched by the possible absolute use of a considerable quantity of verbs, and hence a wide range of functions. I begin by describing the various types of valency change affecting the three types of object. I shall then seek to answer the question as to whether the various changes of valence I have identified and described are to be attributed to a particular form of polysemy of these verbs and hence to semantic/lexical factors intrinsic to them. This requires a detailed analysis of the variants of the respective verbs based on their semantic decomposition, making it possible to identify the common semantic traits and other, specific traits. In this respect the arguments implicit in the various valences of the verbs are seen to be crucial, for they are responsible for specifying, at surface level, the various semantic values activated by the verb. It is not only the adverbials and absolute uses of the verbs, but also the objects which serve as indicators of the alternative meanings of our verbs. Thus we shall interpret objects not merely as the result of projections arising from the verb's meaning (viz. the so-called projectionist theories), but as elements carrying additional meaning and explicitation (viz. the constructional approach). I shall not consider the various forms in which these arguments can be constituted at surface level (pronoun, noun with or without determiners [sentire freddo, sentire ilfreddo 'feel cold, feel the cold'], nominal phrase, prepositional phrase, clause). I shall also say little, and only when necessary, about their semantic structures and the conditions regulating the way in which they combine with the respective verbs. Thus I shall not, for example, go into the possible semantic fillers for the direct object and objectoid in combination with verbs such as acquistare, applaudire, discutere etc., nor, in this respect, what semantic restrictions are placed on them (due among other things to such semantic traits as 'animate'/'nonanimate', 'abstract'/'concrete', 'particular'/'generic', 'singular'/'plural', 'referential'/'non-referentiaP, 'countable'/'non-countable' etc.). In short, I shall limit myself to describing the semantic traits concerning merely the syntactic category.

Changes of valence and their effect on objects Five types of analyse: (i) (ii) (iii) (iv) (v)

189

valency change have been identified, so that we shall

Verbs with transitive and unergative use Verbs with transitive and unaccusative use Verbs with transitive use and a variable number of arguments Verbs with unergative or unaccusative use and a variable number of arguments Verbs with the same arguments, made up of syntactically distinct constituents in "cross-over" syntactic constructions.2

2. Verbs with transitive and unergative use In this group we find verbs which, if used transitively, occur with a direct object, and if used unergatively, appear ο in the absolute (bere [bere acqua - suo marito beve 'drink water - her husband drinks'], brontolare [brontolare insulti - non fa che brontolare 'mutter insults - he does nothing but grumble'], contare [contare i soldi - cosi piccolo sa giä contare 'count the money - so young and he/she can already count'], covare [la chioccia cova i pulcini - la chioccia non vuole covare 'the hen broods the chicks: the hen refuses to brood'], cucinare [cucinare la came - non so cucinare 'cook the meat - I can't cook'], cucire [cucire una gonna - non so cucire 'to sew a skirt - I can't sew'], dipingere [Van Gogh ha dipinto moltipaesaggi campestri - Mario dipinge 'Van Gogh painted many outdoor scenes - Mario paints'], gocciolare [la grondaia gocciola acqua - i rubinetti gocciolano 'the gutter is dripping water - the taps drip'], guadagnare [guadagnare un sacco di soldi -finalmente guadagno 'earn a packet at last I'm earning']), ο with an objectoid (acquistare [i ragazzi hanno acquistato fama - i ragazzi hanno acquistato in simpatia 'the youths acquired fame - the youths gained in popularity'], applaudire [applaudire la cantante', applaudire I 'intervento del ministro - applaudire alprogramma proposto: applaud

190

Heidi Siller-Runggaldier the singer, 'applaud the minister's intervention - speak well of the proposed programme'], azzardare [azzardare un affare — non mi piace azzardare negli affari 'risk a deal - I don't like taking risks in business'], badare [badare il bestiame — badare aifigli, alia casa 'tend the livestock - look after the children, the house'], contrattare [contrattare ilprezzo di un immobile - non ho intenzione di contrattare sul prezzo 'negotiate the price of a building - 1 am not going to haggle over the price'], convenire [abbiamo convenuto il prezzo - tutti i presenti hanno convenuto sulla necessitä di rinviare la riunione 'we have agreed the price - all present are agreed on the need to adjourn the meeting'], votare [votare una delibera, una legge - votare contr ο l'abolizione di qc. 'vote a resolution, a law - vote against the abolition of St.']) ο with an adverbial (abitare [questi animali abitano le zone fredde — i nostri amici abitano in campagna, con i genitori, presso amici 'these animals inhabit cold regions - our friends live in the country, with their parents, with friends'], accelerare [il cuore accelera il battito — I 'autista accelerö di colpo; il treno accelera dopo la galleria 'the heart accelerates its beat - the driver accelerated suddenly, the train accelerates after the tunnel'], alloggiare [questa signora alloggia studenti - alloggiamo presso amici, in una vecchia casa 'this lady takes in students - we are staying with friends, in an old house'], aprire [aprire una porta - la banca apre alle nove 'open a door - the bank opens at nine'], chiudere [chiudere la casa - i negozi hanno gia chiuso 'shut up the house - the shops have already closed'], combattere [combattere il nemico, I 'ignoranza - combattere contro il nemico, la fame 'fight the enemy, ignorance - fight against the enemy, hunger'], condurre [condurre 1'automobile - questa strada conduce in cittä 'drive the car - this road leads into town'], girare [girare la polenta la Terra gira attorno al Sole·, la strada gira a sinistra·, girare per il paese 'stir the gruel - the earth goes round the sun; the road turns left; to go round the village'], gridare [gridare aiuto, vendetta - gridare a gran voce, gridare contro qcu. - 'shout for help, cry out for revenge — shout out loud, cry out against sb.'], pascolare [pascolare le mucche le mucche pascolano sui prati 'graze the cows - the cows are grazing in the fields'], piegare [piegare un tubo, le gambe - la strada piega bruscamente\ il sentiero piega a destra, verso la valle 'bend a tube,

Changes of valence and their effect on objects

191

your legs - the road bends sharply; the path bends to the right, turns downhill']). The different uses of these verbs derive from the fact that, when used transitively and hence accompanied by a direct object, the verb specifies the theme, i.e. the object affected by eventuality,3 which in turn is implicit as a variable in the verb's meaning. When used unergatively, the verb highlights various perspectives of a given extra-linguistic scenario, according to which variant is being employed. Used in the absolute, the verb does not project externally the argument of theme intrinsic to it, leaving this indeterminate. It merely expresses eventuality, so that, according to the context, the action may be repetitive and habitual {Gianni fuma, beve:4 'Gianni smokes, drinks') or correspond to a vocation (Mario dipinge, canta 'Mario paints, sings'), ability, disposition, bent, quality or characteristic (/'/ bimbo sa giä scrivere/leggere/contare\ e una malattia che uccide; il sole abbaglia 'the child can already write/read/count; it's a disease that kills; the sun dazzles'). Here eventuality is rendered as deriving from an agentive force belonging to the entity expressed by the subject and is thus less transitive, since the parameter of affection is not activated. Moreover there is a difference with respect to the aspect of eventuality: the variant with the direct object tends to be telic and punctual (it is obviously the presence of the direct object which determines this aspect), while the absolute variant is atelic and imperfective. The result is a different degree of volition and control on the part of the agent rendered with the subject in terms of the type of eventuality expressed. For it is clear that a punctual eventuality which affects a well defined patient implies a volitive and more controlled force than an eventuality in which neither the patient nor the end point are made explicit and for which the subject can even be a non-animate entity. In pragmatic terms, we can deduce that the absolute uses of these verbs are particularly apt for rendering the background of a representation, while the transitive uses with a direct object tend to be foregrounded. The unergative variants with an objectoid can be interpreted in most cases as the reflection, at surface level, of a different conceptualization of a given extra-linguistic entity. While a holistic semantic component occurs in transitive use, it is the locative component that prevails in unergative use. The former is inherent to the verb itself and can be compared to the meaning of the prefix be- in German verbs of the type bewachen 'watch over, guard', beraten 'give advice', bedienen 'serve' etc. The latter, on the

192

Heidi Siller-Runggaldier

other hand, is projected externally, onto the preposition, where it takes on semantic values and nuances which vary according to the concrete preposition required by the verb. Referents, expressed by means of both types of object, are affected to different degrees, although they can be identical in both cases. Thus the referents expressed by the direct object are much more intensely affected by eventuality than those expressed by the objectoid. Moreover, when rendered with the direct object, they help to limit the scope of the verb, specifying the theme or content of the eventuality expressed and reducing its impact to just one entity. On the other hand, the referent expressed with the objectoid serves to place the verbal process within a certain context. The arguments of the objectoid always provide the idea of limits, trajectories or configurations such as "origin-path-goal". Thus for example the objectoid in the syntagm acquistare in simpatia 'gain in popularity' indicates the context within which the eventuality expressed by the verb is valid; in the syntagm applaudire a un 'iniziativa 'speak well of an initiative' it specifies the goal of the verbal eventuality, while in the syntagm votare contro I 'abolizione di qc. 'vote against the abolition of st.' it evokes an entity towards which there is an attitude of resistance or disapproval. Thus we see that the differentiated rendering of the arguments of the verb by means of the two types of object provides a sort of semantics of clarification which activates and realizes one or other of the potential readings inherent in the verbs in question. From what has been said so far, it can be seen that a transitive reading of the verbs shows a greater degree of transitivity than an unergative reading with the objectoid. The unergative variants with an adverbial specify or classify various circumstances which may be locative (alloggiare in una vecchia casa 'stay in an old house'), modal (gridare a gran voce 'shout out loud'), final {combattere per la libertä 'fight for freedom'), temporal (combattere tutto il giorno 'fight the whole day') and so on. There is to some extent a conceptual difference between the transitive variant with a direct object and the respective unergative variant with an adverbial. Thus for example abitare, used transitively (questi animali abitano le zone fredde 'these animals inhabit cold regions') evokes the place expressed with the direct object as inhabited in its entirety, presenting it as an entity affected holistically. Used unergatively and with the adverbial (questi animali abitano nelle zone fredde 'these animals live in cold regions'), the same verb suggests an area which is only inhabited in various points of its entire extension.

Changes of valence and their effect on objects

193

The difference between the two variants has something to do with the semantic trait of causativity. Thus the verbs condurre, girare, pascolare and piegare are causative; with them the subject expresses an agent which in some way acts upon the entity expressed by the direct object, altering or modifying it. The corresponding unergative variants, on the other hand, indicate with the subject an operating entity, so to speak, on the basis of an intrinsic force, whether animate or inanimate. Various circumstances for this eventuality can be clarified by means of the adverbial. Finally the two variants can serve to highlight two different aspects of the same extra-linguistic scenario, i.e. of the same frame. The verb ricamare, for example, with the direct object denotes the entity undergoing the act of embroidering (ricamare un fazzoletto 'embroider a handkerchief), and with the adverbial the way in which this activity is carried out {ricamare a memo 'embroider by hand'). The conjecture of two different valences is suggested by two readings of the verb: while the variant with direct object is telic and punctual, because the eventuality expressed implies a clearly determined and holistically affected patient, the variant with adverbial is atelic and imperfective. The adverbial merely contributes to the specification of a subcategory of the eventuality expressed, without limiting it in temporal terms. The possibility of making the adverbial and direct object co-occur in the same syntactic structure {ricamare a mano un fazzoletto 'embroider a handkerchief by hand') does not call into question this interpretation. For the adverbial merely adds information concerning the single eventuality of the embroidery of a handkerchief, and thus acts as an additional element and not as a valency element. The two variants in question are seen to reflect two different valences of ricamare and not a complex valence which makes it possible to activate at will one or other of the two arguments.

3. Verbs with transitive and unaccusative use Used transitively, the verbs in this group occur with a direct object and are causative, but when used unaccusatively they are anticausative and employed ο in the absolute {annegare [la padrona ha annegato i cuccioli - i cuccioli sono annegati 'the owner drowned her puppies - the puppies have drowned'], arricchire [arricchire la famiglia - il banchiere e

194

Heidi Siller-Runggaldier

arricchito 'make the family rich - the banker has got rich'], arrugginire [la salsedine arrugginisce le lamiere - il cancello arrugginisce 'the salt rusts the sheet metal - the gate is rusting'], aumentare [la nuova legge aumenta i tabacchi - le spese sono aumentate 'the new law puts up tobacco prices - expenses have gone up'], cambiare [tante brutte esperienze lo hanno cambiato', dopo I 'ammonizione il giocatore ha cambiato atteggiamento - mio padre e cambiato\ la situazione cambiö all'improvviso 'many nasty experiences changed him; after the caution the player changed his tune - my father is changed; the situation changed unexpectedly'], crescere [crescere i figli; il governo ha cresciuto le tasse - il bambino sta crescendo·, il caldo e cresciuto 'bring up children; the government has increased taxes - the child is growing up; the heat has increased']) ο with an objectoid (accrescere [accrescere la produzione — accrescere in fama 'increase production - acquire greater fame'], ingelosire [// suo comportamento ingelosisce il marito - ingelosire della persona amata 'her behaviour makes her husband jealous become jealous of the loved one'], mutare [mutare idea·, la malattia I'ha mutato - mutare in meglio\ mutare d'opinione 'change one's mind; the illness changed him - change for the better; change one's opinion'], stupire [la sua forza mi stupisce - non stupisco piit di nulla 'his/her strength amazes me - 1 don't marvel at anything any more']) ο with an adverbial (impaurire [/ film gialli mi impauriscono — i bambini impauriscono per il buio 'thrillers make me frightened — children are frightened by the dark'], insospettire [il tuo comportamento mi insospettisce - insospetti sentendo strani rumori 'your behaviour makes me suspicious - he got suspicious hearing strange noises'], intimidire [quella sua aria severa intimidisce tutti intimidire davanti a estranei 'that severe manner of his/hers intimidates everybody - become shy in front of strangers'], invecchiare [questa pettinatura ti invecchia\ invecchiare un vino - in questi ultimi anni e invecchiato rapidamente; invecchiare con serenitä; questo vino non e invecchiato abbastanza 'this hairstyle ages you; age a wine - over the last few years he has aged fast; grow old serenely; this wine hasn't aged sufficiently'])

Changes of valence and their effect on objects

195

The difference in the use of the verbs in this group is due to the presence of the semantic trait of causativity in the transitive variants and its absence in the unaccusative variants. This trait endows the transitive variants with a high degree of transitivity, for it implies an active agent in the role of subject, although this is not necessarily animate. Moreover in the role of direct object it implies an entity which, as patient, undergoes the effect caused by the agent, which in turn is specified by the verb. Although this effect can be prolonged in time, the eventuality expressed is in any case telic, because it includes its terminal point. Since it lacks the causative trait, the unaccusative variant excludes mention of the agent, which in terms of relational grammar means that it is rendered chömeur. This is why the patient can advance to the position of subject and take on its status. Its change of state is no longer interpreted as the result of the incidence of a definite agent, but rather as the result of an effect due to an intrinsic or extrinsic force which is not made explicit.5 The eventuality expressed thus shows a reduced transitivity, even though it continues to be telic. In this way the two variants make it possible to give different perspectives to the extra-linguistic eventuality expressed, which is interesting above all from the pragmatic point of view. Here at a lexicalsemantic level we find a dichotomy that recalls the dichotomy between active and passive diathesis at the morpho-syntactic level, even though there is still a substantial difference: while the anticausative variant of our verbs completely rules out mention of an agent, the passive variant allows this, endowing the agent, if it is specified, with a high degree of rhematicity. If these verbs have an objectoid, they not only indicate a change in state of the patient, which advances to the position of subject, but also highlight the entity affected by the change. Thus, compared to the absolute variants, those with an objectoid are more transitive. We can note one further difference: while the use in the transitive and absolute unaccusative constitutes a dichotomy enabling the same eventuality to be illustrated from two different perspectives, unaccusative verbs with an objectoid clarify aspects of a given frame that are different from those indicated by the respective transitive verbs with a direct object. In the variants with an adverbial, the latter adds specifications concerning circumstances of a temporal, local, causal, modal nature, etc., and in many cases is in fact necessary to permit an unaccusative use of these verbs. This can be seen, for example, in the following verbs:

196

Heidi Siller-Runggaldier

rotolare (rotolare un barile - il masso rotolava a valle; rotolare giü per le scale 'roll a barrel - the boulder rolled downhill; roll down the stairs'), sporgere (sporgere la testa - il cornicione sporge troppo dal muro; la terrazza sporge sulla scogliera; la cornice sporge troppo 'thrust one's head out - the cornice juts out too far from the wall; the terrace juts out over the clifftop; the frame sticks out too far'), terminare (terminare un lavoro - il film terminerä verso mezzanotte; il sentiero termina in una radura; il coltello termina a punta; le parole italiane perlopiü terminano in vocale 'finish a job - the film will finish around midnight; the path ends in a clearing; the knife tapers to a point; Italian words mostly end in a vowel'), torcere (torcere i pannv, torcere gli occhi — la strada torce in un 'altra direzione, a destra 'wring out washing; roll one's eyes - the road bends, to the right'). With these verbs it is particularly clear how the meaning of the phrases does not merely depend on the sum of the meanings of the single elements, but also on the function of their syntagmatic conditions. Thus in addition to symbolic processes, it is also the processes of connection which specify the meaning transmitted by these verbs. This group can also include verbs which are distinguished from those presented merely because when used transitively they are not causative and in unaccusative use they are not anti-causative. When used unaccusatively these verbs can appear ο in the absolute (sfogare [sfogare I 'ira - il raffreddore non passa se prima non sfoga 'vent one's anger - the cold won't go away until it has run its course']), ο with an objectoid (seguire [seguire una macchina, un esempio; la 'b' segue la 'a' - alia conferenza e seguito un dibattito\ da certi discorsi seguono certe decisioni 'follow a car, an example; 'b' comes after 'a' - the lecture was followed by a discussion; certain affirmations lead on to certain decisions'], sfuggire [sfuggire un 'arnica pettegola sfuggire a un attentato, a una promessa 'flee from a tell-tale woman friend - escape an attempt on one's life, get out of a promise']),

Changes of valence and their effect on objects

197

ο with an adverbial (fuggire [fuggire gli amici; fuggire un pericolo — fuggire di casa, dalla prigione, all'estero, verso la campagna, in montagna; fuggire all'impazzata 'run away from friends; escape a danger - run away from home, escape from prison, abroad, take to the countryside, the mountains; flee headlong'], penetrare [I'umidita penetra i muri - il proiettile penetra nella carne; i ladri sono penetrati nella gioielleria\ molte parole inglesi sono penetrate nella nostra lingua 'the damp seeps into the walls - the bullet lodges in the flesh; the thieves got into the jeweller's; many English words have found their way into our language'], schizzare [questi animali schizzano un liquido nero - il dentifricio e schizzato fuori dal tubetto 'these animals squirt out a black liquid - the toothpaste squirted out of the tube']).

4. Verbs with transitive use and a variable number of arguments The verbs in this group are all syntactically transitive but differ in terms of the number of arguments. The following verbs belong to this group: interessare (questo libro non lo interessa - interessare gli studenti alia lettura 'this book does not interest him - give students an interest in reading'), invogliare {questo viaggio mi invoglia - invogliare gli amici al viaggio 'this trip tempts me - get friends interested in the trip'), leggere {leggere una lettera - leggere una favola al bambino 'read a letter - read a fairy story to the child'), nutrire {nutrire un bambino - nutrire affetto per il proprio fratello; ho sempre nutrito sospetti verso di lui, nei suoi confronti 'feed a child - feel affection for one's brother; I have always been suspicious of him, had my suspicions about him'), sommare (sommare delle cifre - al totale devi sommare la spesa dell'agenzia 'add up figures to the total you must add on the agency's charges'), spazzare (spazzare il cortile - spazzare la cittä dai ladri 'sweep the yard - clear the city of thieves'), strappare (strappare un fazzoletto - strappare il figlio alia madre; strappare le erbacce dal prato 'rip a handkerchief - tear the son from his mother; pull up the weeds in the lawn'), voltare (voltare una moneta, pagina — voltare la testa all 'indietro; voltare I 'automobile verso la campagna·, voltare le spalle a qlcu., voltare le armi contro qlcu. 'turn over a coin, the page - turn round; turn the car towards the country; turn one's back on sb., turn a weapon on sb.')

198

Heidi

Siller-Runggaldier

The variants with several actants have more semantic traits than those with only a direct object. This is particularly clear in the case of the verbs interessare and invogliare which, used with direct object and objectoid, have a causative and directional value: the direct object represents the human entity which is driven by the agent (made explicit by the subject) towards the entity expressed by the objectoid. However, causativity does not imply that the intended effect of the agent on the patient necessarily comes about. In fact the aspect of these verbs does not imply achievement of the established objective. Variants which only take a direct object show a weak degree of causativity. In this structure causativity is attributed to an entity presented as an agent able to spark off interest or desire in the person rendered with the direct object. The variants of the verbs nutrire, sommare, spazzare are characterised by the presence or absence of the locative trait of directionality. If this trait is activated, they require an objectoid to indicate the entity towards which the action is driven, and a direct object to express the entity which undergoes this action (nutrire odio per qcu., sospetti verso qcu. \ sommare una spesa al totale·, spazzare la citta dai ladri 'nurse hatred for sb., be suspicious of sb.; add an expense to the total; clear the city of thieves'); while without this trait, they merely specify the theme of the activity expressed. Directionality is the additional trait also in the variants of the verbs strappare and voltare, when, in addition to the direct object, they require locative adverbials to specify the type of change of place undergone by the entity expressed by means of the direct object. The semantic trait of directionality is also present in the verbs leggere, strappare, voltare if, in addition to the direct object, they occur with a dative object. For with the latter they express the destinee or beneficiary of the action which, in turn, involves the entity rendered with the direct object as patient, or simply as theme.

Changes of valence and their effect on objects

199

5. Verbs with unergative or unaccusative use and a variable number of arguments This group of verbs can be used either with a single argument in the role of dative object, objectoid or adverbial, or simply in the absolute. unergative verbs: (confinare [ΓItalia confina con la Francia - I'Italia e la Francia confinano 'Italy borders on France - Italy and France are adjoining'], credere [credere in Dio, agli spiriti - rispetto le persone che credono 'believe in God, in ghosts - I respect people who believe'], esagerare [tu esageri sempre nei complimenti; forse esagera un po' col vino - vi ho permesso diparlare tra voi, ma adesso state esagerando 'you always overdo it when you praise people; perhaps he/she does overdo it with the wine - I said you could talk among yourselves, but now you're going too far'], quadrare [le entrate non quadrano con le uscite; il tuo amico non mi quadra - questo conto non quadra 'income does not match outgoings; your friend doesn't convince me - this bill doesn't add up'], parlare [devo parlare con il direttore; parlare del licenziamento con i colleghi; tutti i giornali parlano del tuo libro; parlare ai fedeli — non sa ancora parlare', il colpevole finalmente ha parlato Ί must speak to the manager; talk about the dismissal with colleagues; all the papers are talking about your book; speak to the faithful - he/she isn't talking yet; at last the guilty party has spoken up'], ragionare [ragionare di politica — non mi interrompere\ sto ragionando 'discuss politics - don't disturb me: I'm thinking'], ubbidire [ubbidire alle leggi — e un bimbo che ubbidisce 'obey laws - he's a child who will obey']) unaccusative verbs: (costare [I'abito costa ottantamila lire — oggi le case costano! 'the suit costs 80.000 lire - today houses cost a fortune!'], crescere [il bambino e cresciuto di peso, in altezza — il bambino sta crescendo 'the child has put on weight, grown taller - the child is growing'], esplodere [esplodere in una risata; il caldo e esploso improvvisamente; in Italia e esplosa la mania delle vacanze all 'estero - e esplosa una bomba; se miprovocano ancora una volta, esplodo 'burst out laughing; the heat came unexpectedly; a mania for holidaying abroad has broken out in Italy - a bomb has exploded; if they goad me once more I'll explode'], evadere [tre detenuti sono evasi dal carcere — stanotte il prigioniero e evaso 'three convicts escaped from prison - last night the prisoner escaped'], variare [variare di colore; le opinioni variano da persona a persona; la temperatura varia a seconda deU'altitudine - il

200

Heidi Siller-Runggaldier

prezzo puö variare 'vary in colour; opinions vary from one person to the next; temperature varies according to altitude - the price can vary']) With these verbs the dative objects can indicate the beneficiary (ubbidire ai genitori 'obey one's parents'), the experiencer (il tuo amico non mi quadra 'your friend doesn't convince me') or the destinee (parlare aifedeli 'speak to the faithful') of the eventuality rendered by the verb. The objectoids give information about the context covered by the eventuality which the verb expresses (crescere di peso; variare di colore 'put on weight; vary in colour'). As well as being atelic, the eventualities expressed with these verbs can also be telic, especially if the context indicated by the objectoid represents the goal of the eventuality expressed (esplodere in una risata 'burst into laughter'). Objectoids introduced by the preposition con 'with' have a particular character: endowed with a comitative value, they show reciprocity. For the entity rendered with an objectoid can also be codified as a subject, if it is connected in a relationship of coordination with the first subject in the phrase (I'Italia confina con la Francia - 1'Italia e la Francia confinano 'Italy borders on France - Italy and France are adjoining'). However, there is a difference between the two complementations: the entity expressed with the objectoid appears less agentive or central, and less intensely involved in the eventuality than the one expressed with the subject. The perspective given to the respective representation is not the same. The adverbials offer information about circumstances which may be modal (il caldo e esploso improvvisamente 'the heat came unexpectedly'), local (tre detenuti sono evasi dal carcere 'three convicts escaped from prison'), quantitative (la casa e costata poco 'the house cost little'), etc. They specify factors which are coexistent as variables in a given frame. The absolute variants of these verbs focus on the eventuality as such, endowing it in most cases with an atelic character. Thus they are particularly suited for indicating states of things (i nostri poderi confinano, oggi le macchine costano 'our farms are adjoining, today cars cost a fortune'), psychic and intellectual potential (il bambino non sa ancora parlare 'the child isn't talking yet'), attitude (Paola non crede 'Paola doesn't believe'), behaviour (e un bimbo che ubbidisce 'he's a child who will obey'), mental processes (non mi interrompere: sto ragionando 'don't disturb me: I'm thinking'), disposition (il bimbo sta crescendo: the child is growing up), etc. Not all the uses, however, are atelic. Among the absolute uses of unaccusative verbs the events expressed can also be punctual and telic (e esplosa una bomba 'a bomb has exploded').

Changes of valence and their effect on objects

201

6. Verbs with the same arguments, made up of syntactically distinct constituents in "cross-over" syntactic constructions We can mention the following examples: abbondare (il suo compito abbonda di errori - nel suo compito gli errori abbondano 'his test is full of mistakes - in his test there are no end of mistakes'), accoppiarsi (Anna si e accoppiata a Gianni-Anna e Gianni si sono accoppiati 'Anna has paired off with Gianni - Anna and Gianni have paired o f f ) , brillare {il cielo brilla di stelle - le stelle brillano nel cielo 'the sky glistens with stars - the stars twinkle in the sky'), calare (il lago cala di livello - il livello del lago cala 'the lake is shrinking - the level of the lake is falling'), caricare (hanno caricato di sassi il camion - hanno caricato sassi sul camion 'they loaded the lorry with rocks - they loaded rocks onto the lorry'), mancare (Anna manca di rispetto - a Anna manca il rispetto 'Anna shows a lack of respect - Anna lacks respect'), spalmare (Max spalma di burro il pane - Max spalma burro sul pane 'Max spreads the bread with butter - Max spreads butter on the bread'), sprizzare (i suoi occhi sprizzano gioia - la gioia gli sprizza dagli occhi 'his/her eyes shine with happiness - happiness shines out of his/her eyes'), sprofondare (il peso eccessivo ha sprofondato il pavimento - il pavimento e sprofondato per il peso eccessivo 'the excessive weight caused the floor to collapse the floor collapsed under the excessive weight'), troncare (hanno troncato I'albero della cima - hanno troncato la cima all'albero 'they polled the tree - they lopped the top off the tree') The cross-over structure of the syntactic variants available with these verbs makes it possible to represent the same extra-linguistic scenario from two different perspectives. This recalls active and passive diathesis, for they too enable us to vary the thematic or rhematic character of the arguments. It thus produces analogous pragmatic effects. Unlike the two diathesis, however, the variability in question is not connected to the discursive strategies which are activated deliberately during communication, but is an integral part of the verb itself. It makes two readings available thanks to its particular lexical/semantic nature, which can be activated alternately. The variability is closely linked to the different complementation of the arguments and consists in the promotion of one argument and, at the same time, demotion of another. Thus the objectoid with the verb abbondare can be promoted to the role of subject, but this entails the demotion of the original subject to an adverbial of place (il suo compito abbonda di errori - nel suo compito gli errori abbondano 'his test

202

Heidi Siller-Runggaldier

is full of mistakes - in his test there are no end of mistakes'), the objectoid promoted to subject can bring about the demotion of the original subject to an attribute (il lago cala di livello - il livello del lago cala 'the lake is shrinking - the level of the lake is falling'), and so on. However, there are also verbs for which the promotion of one constituent does not necessarily condition the demotion of another constituent. This is the case, for example, of comitative objectoids introduced by the preposition con (le forze dell'ordine collaborano con la giustizia - le forze dell'ordine e la giustizia collaborano 'the police collaborate with the courts - the police and the courts collaborate') and dative objects in connection with verbs such as accoppiarsi, associarsi {Anna si e accoppiata a Gianni - Anna e Gianni si sono accoppiati 'Anna has paired off with Gianni - Anna and Gianni have paired o f f ) which, when they are promoted, achieve the rank of subject; at this point they enter into a relationship of coordination with the subjects already present. In all the cases we have presented, the different kinds of syntactic rendering of the arguments reflect a particular view of the underlying argumental structure. We only have to recall the locative alternation enacted with verbs like caricare di sassi un camion - caricare sassi su un camion 'they loaded the lorry with rocks - they loaded rocks onto the lorry', which has been widely discussed in the literature.6 In the first variant the verb functions as a "verb of covering" and expresses a causative change of state. The camion as direct object is presented as a patient affected holistically by the eventuality expressed by the verb. The sassi, rendered with the objectoid, indicate the entity involved in the loading of the camion. This variant displays a high grade of transitivity, also because the eventuality is characterised as telic. For it includes its terminal point, inasmuch as it will be complete at the moment in which the camion is completely loaded. The second variant is less transitive than the first, because the verb is used as a "verb of putting" to express a causative change of location. A crucial feature is thus the change of place undergone by the sassi in their role as direct object. The camion, expressed with an adverbial of place, is represented as the space within which the sassi are deposited, so that it is less affected than as the direct object in the first variant. Furthermore, this second variant is not telic. It does not imply that the camion is completely loaded: the action can be interrupted even before the camion is full. Thus its terminal point is of no interest.

Changes of valence and their effect on objects

203

7. Changes of valence: an effect due to polysemy? The question now arises as to whether the semantic differences existing between the variants in meaning of the individual verbs can be attributed to polysemy. To attempt an answer, we must be clear about what concept of polysemy we are referring to. Blank has been particularly illuminating on this subject, in (1993) and (in press). He identifies five causes as crucial in the origin of a polysemy: ο

metaphor as a relation of similarity or contrast between concepts

ο

metonymy as a relation of contiguity between concepts, established on the basis of an associative principle (metaphor and metonymy concern relations between concepts regarding extra-linguistic entities)

ο

hypero-fhyponymy and hence also semantic extension and semantic restriction, co-hyponymy, antonymy - (in the latter Blank (1993: 39) no longer includes cases of auto-converse change of the kind we find in verbs like noleggiare 'rent (out)' or qffittare 'rent'. This permits a single event to be represented from different perspectives, highlighting the contiguity rather than the contrast between the semantic roles of the arguments and constituting merely a "weak" form of antonymy) - and the weakening or strengthening of meaning (for example Fr. tuer has come to mean 'kill' following an intermediate phase in which it was used in this sense as a euphemism).

All these cases feature relations of similarity or contrast between meanings {signifies) and thus relations of a taxonomic nature inherent in the language itself. ο

popular etymology inasmuch as it can lead to a similarity of words (signifiants) that bear no relationship. Blank (1993: 41) maintains that, rather than a true mechanism of semantic change, they should be considered as a mechanism able to initiate a process resulting in the polysemy of a signifiant.

204

Heidi Siller-Runggaldier ο

ellipsis if it is the result of a process through which a word, by frequent use in a syntagm, takes on the meaning of the whole syntagm.

Thus for Blank (1993: 43) polysemy is due to a state in which distinct meanings of a word are linked by relations which derive from situations or processes of the kind we have listed above.7 Whereas in cases for which it is impossible to establish a semantic distance corresponding to one of these relations, and where the meaning of a word undergoes modifications on the basis of a use that stands more or less as prototype, Blank (1993: 43) talks of contextual variants of a meaning.8 These variants are ascribed to a vagueness of meaning, lexicalised and thus proper to the word itself, which makes it possible to emphasise, according to context, different aspects of this meaning without thereby implying the activation of all the other semantic components. Thus the contextual variants of a meaning do not point to referents of different extensional classes, but to referents of the same category, albeit distinguished by a different degree of prototypicality. According to Blank (1993: 44), polysemy can be located with precision between the contextual variants of a meaning and homonymy. It is not a single, hard and fast concept, for it embraces the whole range of minimal nuances, from those concerning the conceptualization of extra-linguistic entities to those involved with the language's internal organization. Thus if the whole 'semantic continuum' is included between the two poles, polysemy cannot be determined or predicted in absolute terms, but it does enable us to identify the processes and models typical of polysemic relationships. In addition to the types of polysemy due to the causes listed above, Blank (in press) distinguishes a type of polysemy at the synchronic level described as "rule-based polysemy". This follows from his perception that, although there are general or universal conceptual processes guided by a certain understanding of the world (such as cultural knowledge), languages lexicalise these concepts in different ways, if at all. As an example Blank (in press) mentions the phenomenon of transitivization which, in spite of being a general conceptual process, is in fact highly idiosyncratic from one language to another. Thus in French the verbs which indicate movement in a certain direction (sortir, entrer, monter, descendre etc.) developed a causative sense (less commonly used with entrer), but this is not the case for the corresponding verbs in English and German. In addition, it does not appear that even within one and the same language a general rule

Changes of valence and their effect on objects

205

necessarily obtains for every case which is possible in theory. There would seem to be specific restrictions in each language concerning the formulation of general concepts, which are not valid for other languages. Thus for example conceptual metonymies such as CONTAINER CONTENTS {glass - drink a glass of wine), INSTITUTION - BUILDING {school - restructure a school), AUTHOR - WORKS {Dante - read Dante), ANIMAL - MEAT OF THIS ANIMAL {fish - eat fish) etc. do not necessarily hold good from one language to another. This suggests that the fillers which complete the universal slots of these concepts are idiosyncratic and language specific. It may be that a language performs lexicalizations also on the basis of specific rules proper to itself rather than on the basis of general concepts; in this case, the lexicalizations in question would prevail over a general cognitive background. To get back to the verbs we have identified as subject to valency change, can we recognise in their different behaviour at the syntactic level an effect due to polysemy? When we analyse them we find that the differences in meaning are largely attributable to different degrees of transitivity, due in turn to the presence or absence of traits such as 'causativity', 'anti-causativity', 'telicity', 'atelicity', 'completeness, 'imperfectivity', 'punctuality', 'iterativity', 'agentivity', 'volition', 'control', 'affection', 'localization', 'directionality', 'holisticity', 'partitivity' etc. This leads to different syntactic configurations, which are reflected in a different number and type of phrasal constituents or alternatively in the same number of constituents, but with a different syntactic function. They are, however, based on systematised models and schemes, as we have seen, and thus may be available for future alternations;9 they reflect different conceptualizations of a given extralinguistic frame: for in fact there are identified and lexicalised characteristics of the same eventuality, which differ in terms of quality and/or number, but also characteristics of different eventualities, if linked in a relation of similarity. The representation of extra-linguistic facts by means of these verbs, or more exactly by their variants, thus reflects different cognitive concepts, which are in fact common to the whole Italophone linguistic community and hence systematic. On several occasions I have used the term frame. It is particularly useful in my investigations, because it embraces the whole network of relations between concepts concerning an extra-linguistic scenario evoked by the semantic potential of a verb. Furthermore it delineates the boundaries within which one can observe the meanings and various uses of the words

206

Heidi

Siller-Runggaldier

which indicate entities belonging to them (cfr. Hüning 1996: 219). In the present case the aspects highlighted by the various uses of the same verb either fall within the same frame and are linked together by relations of contiguity, or else they come under different frames and are linked by relations of similarity. The latter thus derive from a discrepancy between the concepts being matched, and this necessarily implies that there is a certain contrast between them (cfr. Blank 1993: 32). If the variants evoke the same frame, they may reflect relations deriving from the addition or subtraction of semantic traits such as 'causativity', 'agentivity' etc. (qcu. cuoce la came - la came cuoce 'sb. is cooking meat - the meat is cooking/being cooked'). But they can also indicate different details of the same eventuality (scrivere una lettera a qcu. - scrivere a mano 'write a letter to sb. - write by hand') or highlight a deviant verbal aspect, so that, for example, the transitive variant is telic and the intransitive variant atelic. This characteristic is also reflected in the different degrees of agentivity, volition and control in the entity expressed by the subject {odorare ifiori, il basilico, il caffe - le lenzuola odorano di spigo e di lavanda; questa came odora di marcio 'smell the flowers, basil, coffee - the sheets smell of corn and lavender; this meat smells putrid'). Finally the variants can give different perspectives to the same details in the frame (acquistare simpatia - acquistare in simpatia 'acquire popularity gain in popularity'). In all these cases the relation of contiguity established by the two variants cannot be assimilated to the classic relation between base word and the metonymic uses of this base, which tend to point to different referents of the same frame. Its peculiarity actually consists in making possible different conceptualizations of the same frame. The variants that derive from this can thus be considered as the lexicalizations of the divergent conceptual relations which a linguistic community identifies and establishes within one frame. Given their systematic character, they can be adopted as alternatives by the users, according to what they intend to highlight in their communication. It must however be pointed out that, although they have a unitary and common semantic basis, part of the variants in meaning of our verbs rule out inter-changeability of their contexts, for semantic reasons, ontological or cultural incompatibility, or for conventional uses; in this case the single variants prove to be specialised only for certain types of context. However, one might ask whether in the long run these restrictions are not due to an incompatibility between some property of the verb and the

Changes of valence and their effect on objects

207

specific valency type adopted. As an example we can consider the verb fuggire which, when used transitively, only allows an abstract referent in the role of direct object (fuggire le tentazioni, il pericolo 'flee temptations, danger'), and when used unaccusatively, only allows a concrete referent in the role of adverbial (fuggire dalla casa in flamme 'flee from the burning house'). To take another example: the verb terminare in the transitive use implies a subject indicating a human agent (terminare un lavoro 'finish a job'), and in the unaccusative use a subject indicating an inanimate entity (il film termina a mezzanotte 'the film finishes at midnight'). Thus these examples show the difference which exists, in Coseriu's (1975) terms, between the grammaticality of the type of valency change as a phenomenon of langue on the one hand, and the acceptability of its concrete realizations as a phenomenon of norm on the other. In cases when the frames indicated by the same verb are different, the two variants make it possible to conceptualise different aspects of these frames (sporgere la testa dal finestrino - il cornicione sporge troppo dal muro 'put one's head out of the window - the cornice juts out too far from the wall'; immettere il gas nelle tubature - la scorciatoia immette nella strada principale 'feed the gas into the pipes - the short cut comes out in the main road'; strappare un foglio - strappare ilfiglio alia madre 'tear off a leaf - tear the son from his mother'). The respective verbs, in their different valences, should be considered as sub-lemmas with inter-related meanings of a lemma which is formally identical. These sub-lemmas thus share the valency characteristics with the verbs of the respective valency class which, unlike them, merely have a single valence. Thus the frames evoked by our verbs constitute a matrix within which the various conceptualizations of an eventuality can be formulated on the basis of a relation of contiguity or similarity. There are two types of conceptualization, contributing either to indicating different aspects of the same frame or various frames, activating a different number and type of arguments, or to highlighting the same aspects of the same frame, but from different perspectives. In this case the number and type of arguments remain the same, but the phrasal constituents used to express them change. So can we speak of polysemy in relation to our verbs? The variants which evoke different frames do not pose any difficulty for interpretation: given the diversity of the state of things and events to which they refer and the relation of similarity existing between them, the verbs can be considered polysemic. It is more difficult to evaluate the variants which evoke the same frame. If we take polysemy to mean a semantic

208

Heidi

Siller-Runggaldier

configuration with a prototypical nucleus and a network of meanings, which make it possible through extensions of the prototype and specific elaborations of the latter to interpret the word itself as the result of the lexicalization of concepts related to extra-linguistic referents different one from the other, the variants cannot be considered polysemic. If on the other hand we extend the concept of polysemy to include lexicalizations in the same word of different conceptualizations of one and the same frame taken as a whole, we could also consider verbs with variants of the latter type as polysemic. This would however be a weak form of polysemy and not prototypical, with affinities to the form deriving from metonymy. Unlike the latter, which canonically concerns relations established on the basis of a conceptual contiguity between concepts dealing with different extralinguistic entities within the same frame10, or between these concepts and the frame itself or two inter-related frames (see Blank 1999), the form in question concerns the relations existing between different conceptualizations of the same frame. The concept proposed by Blank (1993: 44) of semantic continuum within polysemy proves at this juncture to be very useful, for it allows us to integrate this last type of plurisemanticity too, albeit in a rather peripherica! position and thus scarcely prototypical. It is nonetheless a polysemy based on general and universal rules, so that we are not dealing with an idiosyncratic polysemy but, in Blank's terms, with a "rule-based polysemy" which in Italian corresponds to the types listed and discussed above. Thus it would be wrong to consider the valency variants of our verbs in the same light as the phenomenon of the contextual variants of a meaning (in the sense proposed by Blank, see above), for the latter, being neither systematic nor paradigmatic, are highly idiosyncratic and incidental. Our verbs do not exhibit variations of this type, connected with the different ways in which they are used in practice. Their meaning does not vary randomly as the context varies, although, as we saw above, it is only through context, and depending on this, that one can clarify the type of lexicalization and hence of conceptualization to be activated.11 However, the context is predetermined by the verb itself and always conforms to models which are highly systematic, recurrent and predictable. Thus it is always the verb which establishes the syntactic configuration and indicates the collocations that can be associated with it, in other words which decides the collocators and their collocational classes (for example human, animate, non-animate, abstract, concrete, particular, generic, singular, plural, referential, non-referential, countable, non-countable entities etc.), which as

Changes of valence and their effect on objects

209

arguments are rendered as actants in the various constituents. At the same time this marked sub-categorization also affects the three types of object and the arguments that depend on them. In conclusion we can affirm that valency variability cannot be attributed to a particular profile of use of the respective verbs, but rather to their autonomous lexical meaning. As I have shown, this can vary according to the presence or absence, or indeed the modification, of semantic traits belonging for the most part to the sphere of transitivity. This gives rise to different syntactic projections, in which the arguments, for their part, fulfil the important role of specifiers and integrators of the various meanings activated by the verb. An analysis of our verbs thus requires both a projectionist approach to identify the argumental and syntactic configuration and hence the sub-categorization enacted by each separate valence of the verb, and a constructional approach to identify the semantic input from the context for the semantic clarification and saturation of the respective valency variant.

Notes 1.

2.

3.

A first, longer version of this paper, published in Italienische Studien 2000, 21: 233-268 with the title Fra semantica e formazione delle parole: cambiamenti di valenza verbale, has a different slant: it sets out to establish whether changes of valence of one verb, and thus changes in its argumental structure, are merely a semantic phenomenon or figure in the ambit of word formation. Term used by Elia (Renzi and Elia 1997: 124) to designate syntactic structures which paraphrase a standard construction, inverting or "crossing over" the arguments which comprise it: Le stelle brillano nel cielo - II cielo brilla di stelle: 'the sky glistens with stars - the stars twinkle in the sky'. (See also Boons: 1987). Following Chierchia (1997) I use the term 'eventuality' as hyperonym to indicate indiscriminately the main groups of verbal action classes 'states', perceived as a plurality of instantaneous states which exist, and 'occurrences', subdivided in turn into 'events', which are telic in having an intrinsic point of culmination, and 'processes/actions', which are atelic because they lack this but have a set of homogeneous culminations; the latter are identical to the result of the process and take place in time, so that the processes they constitute can be interrupted at any moment.

210 4.

Heidi Siller-Runggaldier

In the case of beve the absolute use of the verb has, as Salvi rightly points out (1991: 34), a specific interpretation in the sense of beve troppi alcolici 'he /she drinks too much alcohol'. 5. Wunderlich (1996: 178) interprets the respective verbs thus: "[...] the causal factor may be illustrated by a stative condition that, given the appropriate circumstances, induces the process that causes the transition expressed by the sentence. However, the whole causal factor remains unspecified in the sentence itself'. According to this interpretation, in a phrase like sto gelando Ί am freezing', the cold or analogous factors can be considered the stative condition that initiates the process of transition from a state of wellbeing to a state of hypothermia; but they are not expressed as causative factors. 6. We should mention, among others, for Italian the work of Renzi and Elia (1997), for English Levin (1993), Levin and Rappaport (1996), for French Boons (1974), (1985), (1987), and Postal (1982), for German Brinkmann (1993), Sauerland (1994), and Wunderlich (1987), and for various languages Dixon (1989). 7. In Halevy's opinion (1996: 224-225), it is the shape of a 'periphery' of a word and not the shape of a 'core' which constitutes the permeable membrane allowing semantic changes of the type mentioned. 8. Referred to as "allosemy" by Deane (Polysemy and Cognition, in Lingua 1988, 75: 345) (cit. in Blank 1993: 43). 9. I cannot go so far as to speak of productive models at this juncture, because my work is based exclusively on lexicalised material taken from DISC 1997. This prevents me from formulating previsions on the basis of the number of verbs collected for which the possibility of change of valence is memorised as a "usual" occurrence (viz. Rainer 1993: 31) and regularised. I have no grounds for establishing whether the regularity of changes of valence I have registered can ensure the possibility and probability (cfr. Rainer 1993: 32-34) of future changes. In this respect Rainer says significantly (1989: 4) that "the most serious handicap of using exclusively material taken from dictionaries is [...] that there is no logically valid method enabling one to deduce the possible from the given [...]". 10. Bierwisch (1983) in this context speaks of "konzeptuelle Verschiebung" and explains the concept with the word Schule 'school', usually used to denote the educational activities associated with the institution, but which in metonymic uses can also mean 'the building', 'the people involved', 'the morning activity' etc. 11. Following Weigand we can note that "different predicating concepts do not always require different lexical items, they may result from different uses" (Weigand 1996: 157), and the following: "[...] it is evident that it is not isolated words, neither lexical nor grammatical ones, that can be correlated

Changes of valence and their effect on objects

211

with cognitive concepts, but rather the ways of use of whole phrases" (Weigand 1996: 175).

References Bierwisch, Manfred 1983 Semantische und konzeptuelle Repräsentation lexikalischer Einheiten. In: Ruzicka, Rudolf and Mötsch, Wolfgang (eds.) Untersuchungen zur Semantik, 61-99. (=Studia Grammatica XXIX.) Berlin: Akademie-Verlag. Bittner, Maria and Ken Hale 1996 Ergativity: Toward a Theory of a Heterogeneous Class. Linguistic Inquiry 27: 531-604. Blank, Andreas 1993 Polysemie und semantische Relationen im Lexikon. In: Wolfgang Börner and Karl Vogel (eds.), Wortschatz und Fremdsprachenerwerb, 22-56. Bochum: AKS-Verlag. Blank, Andreas 1999 Co-presence and Succession. A Cognitive Typology of Metonymy. In: Karl-Uwe Panther and Günter Radden (eds.), Metonymy in Language and Thought, 169-191. Amsterdam: Benjamins. Blank, Andreas In press Polysemy in the lexicon. In: Brigitte Neriich et al. (eds.), Polysemy, Amsterdam: Benjamins. Blumenthal, Peter 1996 Polysemie im italienischen Valenzlexikon. In: Gertrud Greciano and Helmut Schumacher (eds.), Luden Tesniere - Syntaxe structurale et operations mentales. Akten des deutsch-französischen Kolloquiums anlässlich der 100. Wiederkehr seines Geburtstages, Strasbourg 1993, 263-279, Tübingen: Niemeyer. Blumenthal, Peter 1999 Variation valencielle et polys6mie en italien. Studi Italiani di Linguististica Teorica e Applicata 28: 225-247. Boons, Jean-Paul 1974 Acceptability, interpretation and knowledge of the world: Remarks on the verb planter (to plant). Cognition 2: 183-211. Boons, Jean-Paul 1984 Sceller un piton dans le mur, desceller un piton du mur. Pour une syntaxe de la prefixation negative. Langue frangaise 62: 95-128.

212

Heidi Siller-Runggaldier

Boons, Jean-Paul 1985 Preliminaires ä la classification des verbes locatifs: Les compliments de lieu, leurs criteres, leurs valeurs aspectuelles. Lingvisticae Investigationes 9: 195-267. Boons, Jean-Paul 1987 La notion sömantique de deplacement dans une classification syntaxique des verbes locatifs. Langue franqaise 76: 5-40. Brinkmann, Ursula 1993 Nonindividuation versus affectedness: What Licenses the Promotion of the Prepositional Object? In: Eve E. Clark (ed.), Proceedings of the Twenty-Fifth Annual Child Language Research Forum, 158-170. Stanford, CA: Stanford University. Cennamo, Michela 1998 Transitivitä e inaccusativitä in testi antichi abruzzesi e napoletani. In: Paolo Ramat and Elisa Roma (eds.), Sintassi Storica. Atti del XXX Congresso Internazionale della Societä di Linguistica Italiana (Pavia, 26-28.9.1996), 197-213. Roma: Bulzoni. Chierchia, Gennaro 1997 Semantica. Bologna: II Mulino. Comrie, Bernard 1981 Language Universals and Linguistic Typology. Oxford: Blackwell. Comrie, Bernard 1985 Causative verb formation and other verb-deriving morphology. In: Timothy Shopen (ed.): Language Typology and Syntactic Description, III: Grammatical Categories and the Lexicon, 309-348. Cambridge: Cambridge University Press. Comrie, Bernard 1997 The typology of predicate case marking. In: Joan Bybee, John Haiman, and Sandra A. Thompson (eds.): Essays on Language Function and Language Type, 39-50. Amsterdam: John Benjamins. Coseriu, Eugenio 1975 System, Norm und Rede. In: Coseriu Eugenio: Sprachtheorie und allgemeine Sprachwissenschaft. 5 Studien, 11—101. München: Fink. DISC 1997 Dizionario Italiano Sabatini Coletti. Firenze: Giunti. Dixon, Robert M.W. 1989 Subject and Object in Universal Grammar. In: D. Arnold, M. Atkinson, J. Durand, C. Grover and L. Sadler (eds.), Essays on Grammatical Theory and Universal Grammar, 91-118. Oxford: Oxford University Press.

Changes of valence and their effect on objects

213

Dixon, Robert M.W. and Alexandra Y. Aikhenvald (eds.) 2000 Changing Valency: Case Studies in Transitivity. Cambridge: Cambridge University Press. Fanselow, Gisbert and Peter Staudacher 1991 Wortsemantik. In: Arnim v. Stechow and Dieter Wunderlich (eds.), Semantik/Semantics. Ein Internationales Handbuch der Zeitgenössischen Forschung, 53-71. Berlin/New York: de Gruyter. Geisler, Hans and Daniel Jacob (eds.) 1998 Transitivität und Diathese in Romanischen Sprachen. Tübingen: Niemeyer. Halevy, Rivka 1996 Contextual Modulation of Lexical Meaning. In: Edda Weigand and Franz Hundsnurscher (eds.), Lexical Structures and Language Use. Proceedings of the International Conference on Lexicology and Lexical Semantics (Münster, 13.-15.9.1994), Volume 1, 223-231. Tübingen: Niemeyer. Hopper, Paul and Sandra Thompson 1980 Transitivity in grammar and discourse. Language 56: 251-300. Hundsnurscher, Franz 1995 Das Gebrauchsprofil der Wörter. Überlegungen zur Methodologie der wortsemantischen Beschreibung. In: Ulrich Hoinkes (ed.), Panorama der Lexikalischen Semantik. Thematische Festschrift aus Anlaß des 60. Geburtstages von Horst Geckeier, 347-360. Tübingen: Narr. Hundsnurscher, Franz 1996 Wortsemantik aus der Sicht einer Satzsemantik. In: Edda Weigand and Franz Hundsnurscher (eds.), Lexical Structures and Language Use. Proceedings of the International Conference on Lexicology and Lexical Semantics (Münster, 13.-15.9.1994), Volume 1, 39-51. Tübingen: Niemeyer. Ikegami, Yoshihiko 1996 'Motion' and 'Action' as Basic Schemata for Linguistic Encoding. In: Edda Weigand and Franz Hundsnurscher (eds.), Lexical Structures and Language Use. Proceedings of the International Conference on Lexicology and Lexical Semantics (Münster, 1 3 15.9.1994), Volume 1, 233-242. Tübingen: Niemeyer. Jackendoff, Ray 1996 The Proper Treatment of Measuring Out, Telicity, and Perhaps Even Quantification in English. Natural Language and Linguistic Theory 14:305-354.

214

Heidi Siller-Runggaldier

Kaufmann, I. 1995 O- and D-Predicates: A Semantic Approach to the UnaccusativeUnergative Distinction. Journal of Semantics 12: 377-421. Koch, Peter 1999

Frame and Contiguity. On the Cognitive Bases of Metonymy and Certain Types of Word Formation. In: Klaus-Uwe Panther and Günter Radden (eds.), Metonymy in Language and Thought, 139— 167. Amsterdam: Benjamins. Krifka, Manfred 1989 Nominalreferenz und Zeitkonstitution. Zur Semantik von Massentermen, Pluraltermen und Aspektklassen. München: Fink. Krifka, Manfred 1998 The Origin of Telicity. In: Susan Rothstein (ed.) Events and Grammar, 197-235. (Studies in Linguistics and Philosophy 70.) Dordrecht: Kluwer Academic Press. Levin, Beth 1993 English Verb Classes and Alternations: A Preliminary Investigation. Chicago, IL: University of Chicago Press. Levin, Beth and Malka Rappaport Hovav 1996 From Lexical Semantics to Argument Realization, unpublished ms., Northwestern University and Bar Ilan University, Evanston, IL and Ramat Gan, Israel. Levin, Beth and Malka Rappaport Hovav Forthc. Two Structures for Compositionally Derived Events. Proceedings of SALT 9, Cornell Linguistics Circle Publications, Cornell University, Ithaca, NY. (25 pages, postscript) Martin, Robert 1983 Pour une logique du sens, Paris: Presses Universitaires de France. Parsons, Terence 1995 Thematic Relations and Arguments. Linguistic Inquiry 26: 635-662. Postal, Paul M. 1982 Some arc pair grammar descriptions. In: Pauline Jacobson and Geoffrey K. Pullum (eds.), The Nature of Syntactic Representation, 341425. Dordrecht: Reidel. Rainer, Franz 1989 INomi di Qualitä nell'Italiano Contemporaneo. (Wiener Romanistische Arbeiten 16.) Wien: Braumüller. Rainer, Franz 1993 Spanische Wortbildungslehre. Tübingen: Niemeyer.

Changes of valence and their effect on objects

215

Renzi, Lorenzo and Elia Annibale 1997 Per un vocabolario delle reggenze. In: Tullio De Mauro and Vincenzo Lo Cascio (eds.), Lessico e Grammatica. Teorie Linguistiche e Applicazioni Lessicografiche. Atti del Convegno Internazionale della SLI (Madrid, 21-25 febbraio 1995), 113-129. Roma: Bulzoni. Salvi, Giampaolo 3 1991 La frase semplice. In: Lorenzo Renzi, Giampaolo Salvi and Anna Cardinaletti (eds.), Grande Grammatica di Consultazione, Volume 1, La Frase. I Sintagmi Nominale e Preposizionale, 29—113. Bologna: II Mulino. Sauerland, Uli 1994 German Diathesis and Verb Morphology. In: A. Jones Douglas et al. (eds.), A. I. Memo 1517. Verb Classes and Alternations in Bangla, German, English, and Korean, 50-69. Cambridge, MA: Artificial Intelligence Laboratory, MIT. Siller-Runggaldier, Heidi 1996a Das Objektoid. Eine neue syntaktisch-funktionale Kategorie, aufgezeigt anhand des Italienischen, (pro lingua 27.) Wilhelmsfeld: Egert. Siller-Runggaldier, Heidi 1996b Kasus, syntaktische Funktionen und semantische Rollen im Italienischen. Romanistik in Geschichte und Gegenwart 2: 189—204. Siller-Runggaldier, Heidi 1998 Verben mit variablem Rektionsverhalten (Subjekt + direktes Objekt/Objektoid) im Italienischen und Französischen. Romanistik in Geschichte und Gegenwart 4: 139—151. Siller-Runggaldier, Heidi 2000 Fra semantica e formazione delle parole: i cambiamenti di valenza verbale. Italienische Studien 21: 233-268. Tenny, Carol 1994 Aspectual Roles and the Syntax-Semantic Interface. Dordrecht: Kluwer Academic Publishers. Verkuyl, Henk 1993 A Theory of Aspectuality. (Cambridge Studies in Linguistics 64.) Cambridge: Cambridge University Press. Waltereit, Richard 1999 Grammatical Constraints on Metonymy. On the Role of the Direct Object. In: Klaus-Uwe Panther and Günter Radden (eds.), Metonymy in Language and Thought, 233-253. Amsterdam: Benjamins.

216

Heidi Siller-Runggaldier

Weigand, Edda 1996

Words and their Role in Language Use. In: Edda Weigand and Franz Hundsnurscher (eds.), Lexical Structures and Language Use. Proceedings of the International Conference on Lexicology and Lexical Semantics (Münster, 13.-15.9.1994), Volume 1, 151-167. Tübingen: Niemeyer. Wunderlich, Dieter 1987 An Investigation of Lexical Composition: The Case of German beVerbs. Linguistics 25: 283-331. Wunderlich, Dieter 1996 Models of Lexical Decomposition. In: Edda Weigand and Franz Hundsnurscher (eds.), Lexical Structures and Language Use. Proceedings of the International Conference on Lexicology and Lexical Semantics (Münster, 13.-15.9.1994), Volume 1, 169-183. Tübingen: Niemeyer. Wunderlich, Dieter 1996 Cause and the Structure of Verbs. Linguistic Inquiry 28: 27-68.

Transitivity and Grammaticalization of Object. The diachronic struggle of direct and indirect object in Spanish

Concepcion Company Company

1. Introduction In this paper we look at how transitivity works in Spanish: what the syntactic object in this language is and how it is determined. We can begin by summarizing four well studied facts of the diachronic /synchronic transitivity system in Spanish. 1) The first fact is that the direct object (DO, ACC) has extended the prepositional accusative marker a beyond the original contexts of the personal and animate sphere. Ever more frequently, the inanimate abstract or concrete DO appears in Spanish with a prepositional marker a, as shown in examples (1) (for the corpus on which the analysis is based, see the last paragraph of this Introduction, and the Bibliography, Part A). Such an extension results in the loss of the preposition's status as a differential objectmarking mechanism which it had until recently: with a, it marked human or animated, definite DOs, or DOs with a special discourse relevance; without the a, DOs that lack those features (Bossong 1998; Garcia ms.; Laca in press). Consequently, the prepositional a marking is losing its old semantic restrictions. It is no longer a kind of classifier associated with the lexical class of DO, nor is it a "personal as Bello ([1847] 1978) defined it. It is becoming a true grammatical marker of accusative case (Company 2002). 1 (l)a.

despues de conocer mucho a la vida, after of to know much to the life-DO ya no me interesa tanto el teatro no longer me it-interest so much the theatre (Proceso, May 1999) 'After knowing much of life, theatre does not interest me anymore'

218

b.

c.

d.

e.

Conception Company Company

pues si, son un problema las lluvias, well yes, are a problem the rains, afectan al träfico enormemente they-affect to the traffic-DO enormously 'Well, yes, rain is a problem; it affects traffic tremendously' los japoneses han invertido todos the Japanese they-have invested all sus esfuerzos para construer a la torre their efforts for to construct to the tower mds alta del mundo most high of the world-DO 'The Japanese have invested all their efforts in constructing the world's highest tower' ama mds a este pais de Iο que yo loves more to this country-DO than what I amo a una cerveza fria. love to a beer cold-DO 'He/she loves this country more than I love a cold beer' para que no nos peleemos, puse a la so not we fight, I-put to the silla en medio chair-DO in between Ί put the chair in between, so we won't fight'

2) A second fact is the origin and diffusion of this prepositional casemarking. It is an analogical extension of the meaning of the Latin preposition ad, whose original etymological meaning of direction toward a locative goal is extended to mark an animate entity which in some way is reached by the action of the verb: the dative or indirect object (DAT, IO), being the recipient or final target of transitivity. In turn, this DAT-marking was generalized and analogically extended to mark an entity which is affected by the verbal action, the accusative or DO, patient or theme (Company in press a). The process began in those DOs which, because of their humanity, individuation and topicality, were next to the IO (Pensado 1984, 1985). The result of the change was, as is known, the creation in Spanish of a grammatical marker for objects, through the reanalysis of the locative preposition, ad > a, as a grammatical case-marker. The consequence was that in Spanish the preposition a is more polysemous than its Latin etymon ad, since it

Transitivity and grammaticalization

of object

219

added the grammatical object-marker meaning to the earlier locative etymological meaning. The global change is summarized in (2). (2)

1: directive preposition indicating movement toward a locative goal > 2: analogical extension to mark the goal or destination of the verbal action in the recipient 10 > 3: analogical extension to mark the entity affected by the verbal action in the patient DO.

3) A third fact is that the foregoing analogical extension of the preposition eroded the differences between the DO and the 10, since, as is shown in (3), both the DO and the 10 display the same preposition a as case-marker. Such a lack of formal distinction has led more than one scholar to say that it is not possible to speak of two different functional categories, DO and IO, in Spanish (Alarcos 1994:280; Blansitt 1984:144; Bourciez 1967:449; Company 1998, 2001; Delbecque and Lamiroy 1996:109; Echarte 1996:105; Faltz 1978:85 footnote 4; Roegiest 1990:242, 254, 1998, 1999:8). (3) a. DO =

b.

envio a su hija al (s)he sent to his/her daughter-DO to the colegio school 'He/she sent his/her daughter to school' IO = envio un regalo a su hija (s)he sent a present to his/her daughter-IO 'He/she sent a present to his/her daughter' DO = se representan dos de las them-REFL they-represent two of the cuatro piedras que adornaban al four stones that they-embellished to the antiguo templo (Alzate, 1793,11.399) old temple-DO 'Two of the four stones embellishing the old temple are represented' IO = dono sus bienes al antiguo (s)he donated his/her properties to the old templo. temple-IO 'He/she donated his / her properties to the old temple'

220

Conception Company Company

The lack of distinction is not merely formal but seems to be much deeper, since the two categories behave similarly in several areas of Spanish grammar. Both objects allow two traditional tests for objectivity: i) adjacency to the verb in immediate post-verbal order, and ii) passivization. i) Under certain syntactic and semantic conditions, the patient-DO, as well as the recipient-ΙΟ can be adjacent to the verb, occurring immediately after it, with both noun phrases (NP) showing very similar phonological weights, as the examples in (4a) and (4b) show; these examples form strict minimal pairs, each sentence having either the DO or the 10 adjacent to the verb. ii) Either DO or IO can appear as the subject of a passive sentence, the other object not being passivized but remaining as an oblique prepositional phrase (4c). Adjacency to the verb means that the verb governs the NP, and, consequently, that the NP has object status in many of the world's languages, according to a "Linear precedence principle" of constituents (Alsina 1993:420-422). Likewise, the possibility of passivization is strong syntactic proof of the object status of a nominal (Chung 1976; Givon 1984a; Hudson 1992:260; Goldberg 1992:38^10; Hyman and Duranti 1982:223-25; Newman 1996:28-29). (4) a. ya han dado el tercio al licenciado already they-have given the third-DO to the attorney-IO 'They have already given a third [part] to the attorney' (Santa Teresa, III.4b) mande v. senoria escribir quien ha de dar order Your Lordship to write who has of to give a Miranda estos tercios (Santa Teresa, III.4b) to Miranda-ΙΟ these thirds-DO 'Your Lordship should send word as to who should give Miranda these thirds' b. y para que si quisieren hacerlo and for that if they-wanted to do it-ACC enbio el dicho oro a v. merged (DLNE, 16th, 48.186) I-send the said gold-DO to Your Grace 'And so, in case they want to do it, I send the gold to Your Grace' con los quales dicho Juan Gallego y Rrebolo enbio with which said PN I-send

Transitivity and grammaticalization

of object

221

a v. merged 300pesos (DLNE, 16th, 33.156) to Your Grace-IO 300 pesos-DO 'With which the said Juan Gallego y Rrebolo I send Your Grace 300 pesos' c. ACT. = advert! el peligro mi a padre I-cautioned the danger-PAT-DO to my father-REC-IO Ί warned my father of the danger' pecaria yo de presumido e would sin I of presumptuous and insolente si advirtiese a mi padre insolent if I-were to warn to my father-REC-IO del peligro of the danger Ί would be presumptuous and insolent if I were to warn my father of the danger' (Pepita Jimenez, 246) PASS = el peligro fue advertido a mi padre the danger-PAT-SUBJ was warned to my father-IO 'My father was warned of the danger' mi padre fue advertido del peligro my father-REC-SUBJ was warned of the danger-OBL 'My father was warned of the danger' The formal and functional closeness of DO and IO in Spanish seems to favour a dative clitic which express the direct object function; that is, dative for both masculine and feminine accusative. This is the well-known phenomenon of leismo, exemplified in (5a). The accusative clitic may also function as an indirect object, that is, accusative for dative. This is known either as laismo or loismo, exemplified in (5b). However, I will try to show that the lack of distinction between DO and IO in Spanish is in fact only a syntactic illusion. (5) a. DAT-DO = el telefono ya cuelgale the telephone now hang it-DAT-DO up 'The telephone, hang it up' maestra / maestro, lie ayudo? teacher-MALE/FEMALE, you-DAT-DO I-help 'Teacher, may I help you?'

222

Conception Company Company

b.

ACC-OI =

la dio un beso her-ACC-IO (s)he gave a kiss '(s)he gave her a kiss' lo dio una bofetada him-ACC-IO (s)he gave a slap '(S)he gave him a slap'

4) A final fact is that the 10 is manifested in present-day Spanish by at least four formal devices shown in (6): i) with an obligatory prepositional marker a (6a);2 ii) with categorial duplication of the lexical IO by a coreferential clitic in the same verb phrase; any lexical type of IO may be doubled, as the examples in (6b) show; iii) with an invariable singular DAT-clitic le affixed to the verb, to refer to plural IO, which behaves as an object-verb agreement marker (6c); and iv) with systematic "anomalous" direct object marking, when in the same ditransitive sentence an IO-clitic se < Lat. illi(s) and a singular DO-clitic, lo or la, co-occur; in these cases the ACC clitic is anomalously marked with an extraneous number-marker -s, as in the first example in (6d), and/or an extraneous gender-marker, as in the second example in (6d). leon (Calila, 158) (6) a. el camello que se ofrecio al lion-IO the camel that was offered to the 'The camel that was offered to the lion' que et preguntava α los vezinos what and asked to the neighbours-IO podria ser... (Lazarillo, 33.1) could be 'And asked the neighbours what could be' sangre a b. le he de beber la to him-DAT I-have of to drink the blood ese roto (Bandidos, 99) that dandy-IO Ί shall drink the blood of that dandy' al ve r a Buenaventura bien sport le at the to see to PN very sport it-DAT dio un tiron a su corbata (El Financiero, March 2000) he-gave a pull to his tie-IO 'On seeing Buenaventura in sports attire, he pulled his tie'

Transitivity and grammaticalization of object

c.

d.

223

en vez de fatiga sentia fiebre que les instead of fatigue he-felt fever that them-DAT daba vigor de acero a las piernas (Regenta, 104) gave vigour of steel to the legs-IO 'Instead of fatigue, he felt a fever that gave his legs the vigour of steel' el problema es bajarle los humos the problem is to decrease-it-DAT-SG the smokes a las grandes ciudades en todo el mundo (Proceso, May 1999) to the large cities-IO-PL in all the world 'The problem is to decrease smoke in large cities throughout the world' tengo muchas [anecdotas], per ο esa me I-have many [anecdotes], but that one to me parece que es la mäs divertida para seems that is the most fun to contärselas a ustedes tell-you-DAT-PL=it-ACC-SG=PL to you (Habla Culta de Buenos Aires, 173) Ί have many anecdotes, but that one would be the most fun to tell you' si ellas me quieren comprar el if they me-DAT want to buy the caballo, yo se las horse, I them- DAT- PL-FEM it-ACC-SG-MALE= PL-FEM vendere (Lope Blanch 1953) will sell 'If they want to buy the horse from me, I will sell it to them'

The identical prepositional marking in the DO and the 10 certainly generates a lack of distinction between the two objects, but, as I have already said, it is only a superficial lack of distinction, since in Spanish grammar there is actually a lack of equilibrium of object-marking devices. The DO has two object-markings, in that it can be constructed with or without a. This case-marking device consisting of the presence-absence of only one form could be labelled simple case-marking. On the contrary, the 10 has multiple case-markings, all comprising some formal mechanism, and these multiple devices may accumulate in the same IO phrase: the preposition a,

224

Conception Company Company

plus duplication, plus object-verb agreement marking, plus anomalous concordance. In summary, we have one formal case marking for the DO vs. multiple marking for the 10, as (7) shows: (7)

DO = 10 =

primary object? = secondary object? =

simple case-marking multiple case-marking

Such a lack of equilibrium in marking devices creates a rather paradoxical situation in transitivity relations in Spanish: the 10, a secondary object, near a locative oblique - a s defined in most studies and Spanish reference grammars (Butt and Benjamin 1994:144; Cano 1981:335; De la Pena 1985:338345; Folgar 1993:98ff; Gili Gaya 1970; RAE 1973:372-75; Sanchez Marquez 1982:129-130) - displays many more case-marking devices than the DO, which seems to be, at least at first sight, the true object and is considered as such in the aforementioned studies. Besides the four well-known aspects of transitivity, three other features should be investigated: i) the effects of this outward lack of distinction; ii) the diachronic causes of the paradoxical grammatical imbalance; iii) the dynamic relationship between the DO and the 10 in the structural space of 'object'. We shall investigate these three features. The paper is organized in six sections. Section 2 formulates the theoretical proposal and the hypothesis that guides the analysis of a set of changes affecting the DO and the 10 in Spanish. Sections 3 and 4 show the quantitative and qualitative contexts, as well as the syntactic changes in which the diachronic struggle of these two arguments for object status can be clearly detected. In section 5,1 present other changes that show how the DAT has reinforced its case-marking as primary object. Most of these changes have taken place at the expense of weakening the accusative marking. Section 6 briefly presents the possible reasons for this diachronic competition. Finally, the conclusions are presented in section 7 with some brief theoretical reflections on the balance between synchrony and diachrony in the construction of grammar. The corpus on which the research is based consists of prose texts from the 13th to the 20th centuries, with diverse chronological cuts in accordance with the change under analysis, although for all the changes I focus on the 16th century texts, since this is a period, as we shall see, of important microirregularities in the Spanish transitivity system. For the 20th century, prose texts as well as transcriptions of spoken language and spontaneous speech are culled. The sources are listed in section 1 of the Bibliography, under

Transitivity and grammaticalization of object

225

Corpus. The analysis will focus on those aspects of change that make the diachronic struggle between the DO and IO evident, although, for reasons of space, I will not enter into the specific details and causes of each change. The examples without reference between parentheses correspond to spontaneous speech in Mexican Spanish.

2. The hypothesis The aim of this paper is to analyse the diachronic behaviour of the 'object' category in Spanish and to show the diachronic causes - o r at least some of them - that lead to the paradoxical imbalance mentioned in (7) above. We shall examine several changes that have affected the Spanish transitivity system - o n e concerning the DO, and seven related to the IO. They all go to show that in the syntactic development of this language there is a diachronic competition between the DO and the IO, or rather between the accusative and the dative, to occupy the functional, structural slot o f ' o b j e c t ' , a struggle expressed by the following hypothesis (8). (8) Each time in the history of Spanish DO-marking by the preposition a advances significantly, the IO strengthens its status as object by reinforcing its marking devices or weakening the DO, or by both processes at once. The consequence of such a diachronic struggle was that in Spanish the dative category enormously expanded its syntactic scope with respect to its Latin etymon, the dative case, and invaded functional spaces that it did not cover originally. Hence, the dative has become ever more argumental in Spanish and has invaded the structural space of primary object which the accusative DO originally had, weakening the accusative object in different ways. The general consequence is that Spanish has progressively acquired the features of a Primary Object-Secondary Object language (PO-SO) in several areas of the transitivity system (Dryer 1986; Shibatani 1996),3 sliding from a DO-IO language toward a PO-SO language. That is, Spanish has moved from being a language with two object markings: ACC and DAT, to being a language with just one object marking: DAT. As a consequence of this sliding, Spanish has undergone a deep change in its general type and today has features of a PO-SO language (Company 2001), with a complex, mixed typological configuration: it is a PO-SO language in several areas of

226

Conception Company Company

transitivity, it is a DO-IO language in others, and it is still of difficult classification in other contexts. The changes I shall analyse have already been studied for the most part, some of them as classical case studies in Spanish historical syntax, but they have usually been analysed as isolated phenomena, quite distinct from each other. Nonetheless, if we look carefully, we will realise that all show the same underlying grammatical pattern. My claim is that all the changes are a manifestation of a unique global shift and that they evidence the same diachronic structural pattern: a trend toward incorporating the dative as an argument or valence of the verb, the primary object of the sentence structure, and, even, the sole object of its sentence. The mechanism in all the changes was that either a dative marker appeared where an accusative marker was expected (that is, the dative marking displaced that of the accusative) or the dative marking was reinforced, changing from an optional to a mandatory status. The set of changes can be summarized as progressive and multiple grammaticalization in several dimensions of transitivity. One change affected the accusative: the acquisition of the prepositional case-marker a that brought about the approximation of the accusative to the dative. Several changes involved the dative and had three grammatical repercussions: the functional approximation of the dative to the accusative, the structural weakening of the accusative as a syntactic object, and/or the reinforcement of the dative object in Spanish. The changes may be analysed as a global, multiple and unique grammaticalization that yielded a new linguistic type in Spanish. All the grammaticalizations reflect the same structural situation: two participants, dative and accusative, in transitive or ditransitive sentences show a syntactic behaviour of frank rivalry for object-marking, and consequently, for the status as object. In all cases, the dative wins the syntactic space. Each of the changes I analyse constitutes a micro-breakage or microirregularity in the developmental continuum of Spanish, which, by itself, did not affect the basic structure of the linguistic system, but together they produced a macro-breakage, a qualitative leap in the configuration of this language, that I believe may be classified as an elementary catastrophe in the sense of Catastrophe Theory (Thom 1983, Wildgen 1987); a catastrophe that has deeply altered the linguistic typology of Spanish. By grammaticalization I understand a constant, never-ending dynamic process of codification and organization of grammar, a sense quite close to that of Hopper's (1987) 'emergent grammar'. Grammar is the crystalliza-

Transitivity and grammaticalization of object

227

tion of language use. Grammaticalization has its source or originates in language use, in real discourse and it comprises different kinds of dynamic processes, which do not necessarily lead to a semantic bleaching or the loss of freedom of forms. Often the only indication that grammaticalization is going on is the different frequency of use of a form or construction in one environment rather than another. Changes of this sort do not necessarily imply an external, formal change of the entities involved (Kuryiowicz 1965). Grammaticalization has two main effects: it produces synchronic variation, since the earlier and the latter stages of a change, that is, innovative and conservative forms coexist in a given stage of language even for centuries; and it produces conflictive analyses since the same form can admit two or three alternative analyses. At the same time, the synchronic variation is a symptom of, and a prerequisite for, grammaticalization to occur; that is, grammaticalization and synchronic variation determine each other, and constantly coexist in each language stage, so that the boundaries between synchrony and diachrony are weakened or eroded. The changes I shall analyse in the following sections are: 1) diachronic advance of the prepositional case-marker a on the DO; 2) loss of the prepositional case-marker of the human DO in ditransitive sentences; 3) semantic flexibilization of the dative; 4) generalization of 10 duplication; 5) depronominalization of the dative clitic; 6) anomalous marking of the dative over an accusative clitic; 7) generalized use of leismo; and 8) progressive invasion of the dative as the causee subject in causative constructions.

3. The diachronic struggle of the DO and the IO: the quantitative analysis There are several interesting chronological parallelisms in the changes undergone by the two objects in Spanish. Both DO and IO appear to make quantitative breakages and progressions during the same chronological periods, as can be seen in tables 1-5. Table 1 shows the advance of prepositional a case-marking in the DO for different classes of nominals. It can be seen that the individuated, highly definite, human nominals, such as personal pronouns and proper nouns, necessarily, or in a high percentage of cases, take the preposition from very early times, whereas the prepositional case-marking in the inanimate DO, last line, is practically non-existent in Medieval Spanish:4 it begins to crop up in the 15th century (3%), more than doubles in the 16th century to 8%, and continues a sustained progression

228

Conception Company Company

through the 20th century, to 17% (average frequency of spoken and written texts). Tables 2 and 3 below show a change in the DAT category that consists in the semantic flexibilization5 undergone by the dative in the history of Spanish; this category extended beyond its etymological human sphere to include inanimate entities in a greater frequency, as examples from the 19th century in (9) show. Table 1. Diachronic advance of the prepositional a direct object case-marking 13th

14th

15th

16th

20th

PERS. PRON. PROP. NOUN HUMAN

100% (53/53) 99% (124/125) 42% (243/574)

100% (46/46) 99% (170/172) 35% (224/631)

99% (67/68) 96% (129/134) 35% (181/518)

99% (182/183) 88% (124/147) 50% (541/1086 )

100% (55/55) 100% (32/32) 57% (81/141)

ANIMAT Ε INANIM ATE

3% (4/155) 1% (2/300)

3% (2/64) 0% (1/300)

6% (2/34) 3% (8/300)

7% (11/168) 8% (54/641)

(9)

17% (64/373)

Da ser α las creaciones de la give to be to the creations-ΙΟ of the voluntad (Pepita Jimenez, 189) will 'It gives being to creations of the will' un periodico ha faltado a la verdad. (Bandidos, 6) a newspaper has faulted to the truth-IO Ά newspaper has fallen short of the truth'

The change, although faint, is significant: from 5% in the 16th century to 16% in the 19th century (see table 2). The fact reflected in the last column of table 3 is even more interesting: it was lexical IOs and not clitic IOs that led this semantic flexibility. Lexical IOs importantly increased their capacity to refer to non-human entities: from 13% and 12% in the 13th and 16th

Transitivity and grammaticalization of object

229

centuries, respectively, to 37% in the 19th century,6 whereas clitic IOs maintained their lexical features practically unchanged (Company in press b). The change can be interpreted as an approximation of the IO to the inanimation typical of the DO. Furthermore, in these two tables it can be seen that the quantitative breakage occurred after the 16th century, as was the case with the advance of prepositions in inanimate DOs (table 1 above). The X2 of 65.5 for table 2, and 55.8 for table 3 (> 3.84, with a degree of probability of 0.05 and a gl of 1) indicate that the increases from 5% to 16% and from 12% to 37% are significant. Thus, the null hypothesis of no association can be rejected. Table 2. Semantic flexibilization of the IO clitic and NP th

13 16th 19th

+HUMAN 94% (1032/1094) 95% (1176/1236) 84% (1001/1197)

-HUMAN 6% (62/1094) 5% (60/1236) 16% (196/1197)

Table 4 reports on the advance of IO duplication, considering only nonpronominal lexical IOs. It can be seen that the first period of real generalization of the IO doubling occurred after the 16th century. In fact, between the 16th and the 18th century, the relative frequency of use of the duplicate construction increased fourfold: from 10% in the 16th century, it jumped to 40% in the 18th century, whereas between the 13th and the 16th centuries, an even longer period of time, the progression of duplication was much slower. That is, the diffusion of the change was much more active between the 16th and the 18th centuries than during the previous three hundred years. Table 3. Semantic flexibilization of the IO clitic vs. the NP

13,h

16th 19th

CLITIC +HUM 98% (735/751) 99% (801/808) 97% (709/733)

NP -HUM 2% (16/751) 1% (7/808) 3% (24/733)

+HUM 87% (297/343) 88% (375/428) 63% (292/464)

-HUM 13% (46/343) 12% (53/428) 37% (172/464)

230

Conception Company Company

Table 4. Progression of 10 uplication -DUPLICATION 94% (321/343) 13th th 16 93% (311/336) 18th 60% (258/430) th 20 17% (104/609)

+DUPLICATION 6% (22/343) 10% (34/336) 40% (172/430) 83% (505/609)

Lastly, table 5 shows the first documentation of the depronominalization of the DAT clitic, exemplified in (6c) above; it took place precisely in the analysed corpus in the 16th century, 7 and therefore, chronologically coincides with the quantitative breakages observed in the foregoing changes. Table 5. Depronominalization of the plural dative clitic 13th 16th 20th

+AGREEMENT 100% (9/9) 88% (22/25) 43% (144/333)

-AGREEMENT 0% (0/9) 12% (3/25) 57% (189/333)

I believe the common feature of all of these changes is that the most significant quantitative advances in the five phenomena occurred starting in the 16th century. In fact, the preposition a with an inanimate DO jumps from 3% in the 15th century to 8% in the 16th. Semantic flexibilization of the 10 to express inanimate referents increases notably after the 16th century: from 12% in that century to 37% in the 19th. 10 duplication also increases from 10% in the 16th century to 40% in the 18th, and the first manifestation of an ΙΟ-verb agreement marker is attested in the 16th, after which it progressively increases to 57% in the 20th century. We can see that the diffusion of all the changes seems to begin from the 16th century. This is no coincidence: it is due to the close grammatical affinity of the five changes, all of which, although apparently unconnected, reflect the diachronic competition of the DO and the 10 in the development of Spanish. When prepositional a marking began to impact interestingly on the inanimate DO, that is, the DO prototype, it can be said that the actualization of the reanalysis (Timberlake 1977) preposition > classifier > case-marker took place. The preposition a ceases to be a 'personal a ' and extends beyond the sphere of animate human objects. It stops being a type of classifier for human or animate objects and the second reanalysis, classifier > case-

Transitivity and grammaticalization of object

231

marker, takes place, the preposition becoming a true case-marker. At that time, a begins to lose its value of differential mechanism for marking internal lexical classes of DOs, and the formal and partly functional indistinction between the two objects occurs, DO = IO. The progression of a in the DO provoked a reaction in the IO, causing several changes in the dative category, whose most important initial quantitative manifestations coincide with the actualization of the reanalysis of the preposition α as a case-marker for all types of DO. All of this seems to happen from the 16th century. The multiple changes in the IO, on the one hand, allowed this category to become closer to the prototypical inanimation of the DO (tables 2 and 3), and on the other hand, allowed the IO to reinforce itself through the generalization of duplications, via a coreferential dative clitic in the same verb phrase (table 4), and with the generation of an object-verb agreement marker affixed to the verb through the depronominalization of the dative clitic (table 5).8 If we integrate the relevant categories from the preceding tables into just one table, 6, putting together those changes that seem to establish a causally-linked relationship, that is, a) inanimate DO, b) inanimate IO, b) duplicated IO and c) plural IO with no agreement with its clitic, and if we focus on the periods of quantitative breakage in each change, it is possible to observe that all innovations are triggered in the 16th century or thereafter. 9 In fact, the 16th century seems to be the chronological turning point for all the changes, the great moment of the developmental breakthrough, and it is in that period after the 16th century that the IO undergoes important quantitative leaps, increasing the relative frequency of reference to inanimate entities and generalizing the use of the other dative constructions that are analysed here, yielding a three- or even four-fold increase in the quantitative manifestations those phenomena had had before the 16th century. The chronological coincidence of these four changes, summarized in table 6, is shown in another form in figure 1 below. In this graph, it can be seen that although the four changes converge in the 16th century with similar frequencies, during the period immediately following, the diffusion of the respective changes is quite different, since the various innovations involving the IO had a significantly greater increase than the only innovation concerning the inanimate DO. The figure also shows that the 16th century constitutes not only the take-off point for the diffusion of the four phenomena, but also that this period was the time of concentration of micro-irregularities or micro-breakages in the Spanish transitivity system, which all together caused a very deep readjustment, an elementary catas-

232

Conception Company Company

all together caused a very deep readjustment, an elementary catastrophe in the typological configuration of Spanish. Table 6. Chronological parallelisms of DO and 10 in the diffusion of changes BEFORE 16™

16™

AFTER 16™

INANIMATE DO

l%-3%

8%

17%

INANIMATE IO-NP

13%

12%

37%

10+ DUPLICATE

6%

10%

40-83%

AGREEMENTIO

0

12%

57%

%

INANIMATED D O

- • - I N A N I M A T E D 10-NP

D U P L I C A T E D 10

N O A G R E E M E N T 10

Figure I. Chronological convergence of changes in transitivity

Transitivity and grammaticalization of object

233

4. The diachronic struggle of DO and IO: the qualitative analysis Is there a qualitative analysis corresponding to the breakages and the quantitative chronological convergence we have just seen in table 6 and figure 1 ? Is there a qualitative analysis that gives support to our hypothesis of the diachronic struggle of DO vs. IO for the functional 'syntactic object' status, with the result that Spanish has been moving toward a PO-SO language for several centuries? If our hypothesis is correct, it is to be expected that the changes become extremely active when the DO and the IO equate in terms of their formal and lexical manifestation. That is, once a true formal and semantic indistinction of the two objects occurs, the changes analysed above have to be more active. In fact, the struggle between the two objects becomes especially noticeable in two areas of transitivity: 1) when the DO is close to the IO prototype, as regards human and definite features, and 2) when the IO gets closer to the DO prototype as regards inanimacy features. Let us examine these two changes.

4.1. Loss of the human prepositional DO case-marker in ditransitive sentences When the accusative-DO approximates the dative-ΙΟ and both objects are equal in terms of referential lexical features, both objects show very different behaviour as regards the prepositional case-marking. That is, when both objects in the same ditransitive sentence refer to human beings (10), the DO behaves as a degraded object in that it regularly loses the prepositional a case-marker. The 'personal a' that characterizes human DOs does not appear, whereas the IO always retains its prepositional marker (10a), and sentences such as those of (10b), in which a nominal DO maintains its prepositional case-marking in the presence of an IO, are grammatically rare and unusual (Meyer-Lübke 1890-1906, 3:385; RAE 1973:374). If the DO is a proper noun, that is, very close to the IO prototype, as regards human, individuation and definiteness features, then it is easier for the DO to retain the preposition and both objects can appear with a prepositional case-marker (10c), but even those DO proper nouns lose their case-marking in the presence of an IO, as is shown in (lOd) (both DO and IO are emphasized in bold type).

234 (10) a.

b.

c.

Conception Company Company el cacique entrego 0 su hija the headman gave/delivered 0 his daughter-ACC α los conquistadores to the conquerors-DAT 'The headman gave his daughter to the conquerors' el maestro presento 0 su mujer a the teacher introduced 0 his wife-ACC to sus alumnos his pupils-DAT 'The teacher introduced his wife to his pupils' para ofrecer 0 su hija al fuego de to offer 0 his daughter-ACC to the fire of Dios (Roegiest 1998:475) God-DAT 'To offer his/her daughter to the fire of God' ?? el cacique entrego a su hija ?? the headman delivered to his daughter-ACC a los conquistadores to the conquerors-DAT 'The headman gave his daughter to the conquerors' ?? el maestro presento a su mujer ?? the teacher introduced to his wife-ACC a sus alumnus to his pupils-DAT 'The teacher introduced his wife to his pupils' * el cacique dio a su hija a los * the headman gave to his daughter-ACC to the conquistadores conquerors-DAT 'The headman gave his daughter to the conquerors' el traidor Judas vendio a Jesus a los the traitor Judas sold to Jesus-ACC to the sacerdotes y fariseos (Bello 1978:269) priests-DAT and Pharisees-DAT 'The traitor Judas sold Jesus to the priests and Pharisees' di a Diana a don Sancho (Tirso de Molina, Gili Gaya 1970:209) I-gave to PN-ACC to PN-DAT Ί gave Diana to don Sancho'

Transitivity and grammaticalization of object

d.

23 5

alii se dar ία ordert de llevar there it would give-PASS order of to take aDorotea a sus padres (Cervantes, Gili Gaya 1970: 209) to Dorotea-ACC to her parents-DAT 'There, the order would be given to take Dorotea to her parents' prefiero 0 Luisa α Petra (Gili Gayal970:209) I-prefer 0 Luisa-ACC to Petra-DAT Ί prefer Luisa to Petra' haber preferido 0 este Eugenio Colorado having preferred 0 this PN coloured grueso y Tosco a dona Maria y thick and unrefmed-ACC to PN-DAT and al abuelo Martin (Roegiest 1998:484) to the grandfather PN-DAT 'Having preferred a coloured, thick and unrefined Eugenio over dona Maria and grandfather Martin'

Prepositional marking on both objects is a possibility of the system as some of the examples in (10) above show. However, it will be seen that in actual language use, such double case-marking is almost non-existent in the history of Spanish.10 The personal DO takes a prepositional case-marker in monotransitive sentences and lacks the preposition in ditransitive ones. In other words, the personal DO lacks a case-marker in the presence of an IO, and takes case-marking only in the absence of an IO. Being the dative recipient, the nominal that controls the case-marking in ditransitive constructions is typical behaviour for PO-SO languages (Aissen 1983, Chung 1976). In this respect, Spanish seems to operate as a clear PO-SO language. We find a change where the dative and the accusative compete in an identical grammatical semantic space and the dative wins the struggle, in that the dative systematically and consistently retains its prepositional case-marker, indicating therewith its higher hierarchical status, while the accusative loses its case-marker. It is worth noting that the prepositional case-marking on both objects improves noticeably when the IO is duplicated with a coreferential clitic (11). This means that the double prepositional marking of objects is allowed in Spanish if the IO reinforces its presence in the sentence structure with double DAT marking.

236 (Π)

Concepciön Company Company

el cacique le(s)\ dio a su hija the headman them-DAT gave to his daughter-ACC α los conquistadoresi to the conquerors-DAT 'The headman gave his daughter to the conquerors' el maestro le(s)t presento a su mujer the teacher them-DAT introduced to his wife-ACC α los alumnosi to the pupiis-DAT 'The teacher introduced his wife to the pupils'

Quantitative diachronic data are enlightening with regard to the unequal struggle of the DO and IO for the grammatical object status. In an exhaustive reading of a broad corpus formed by twenty-four texts in prose from the 13th to the 20th century, with 3075 ditransitive sentences with both objects in the form of an NP, Ortiz (2001) only documents 33 sentences having both objects marked with a preposition, which represents 1.07% (33/3075) of the total corpus; that is, the sentences of (10c) above with both DO and IO marked with preposition are very rare in the transitivity system of Spanish. The standard case-marking of ditransitivity for the whole period is 0-ACC-DO + Λ-DAT-IO (98.9%, 3042/3075). The author points out that those DOs retaining the prepositional marker in ditransitive sentences have a special pragmatic status in the discourse in which they appear. In support of our hypothesis, it is also interesting to emphasize that when comparing the earlier texts with later ones in the corpus analysed by Ortiz, Spanish has in fact decreased the frequency of the double prepositional marking. From 3.06% (17/554) during the 14th and 15th century, it decreases to half of that in the 20th century: 1.50% (4/265).11 These quantitative data indicate that while prepositional case-marking of the DO advanced in Spanish in monotransitive sentences, it was inhibited in double object constructions, which means that in this area, Spanish practically behaves like a PO language: the case-marker falls on the patient in monotransitive sentences, but it falls on the recipient in ditransitive sentences.

4.2. Depronominalization of the DAT clitic with an inanimate DO This syntactic change is closely related to and is a consequence of the generalization of the IO duplication seen in table 4. In Spanish, as we have al-

Transitivity and grammaticalization of object

23 7

ready seen, a plural IO can be duplicated with a singular clitic pronoun (12). The productivity of this change at the end of the 20th century, in a broad learned oral and written corpus, can be seen in table 7 (Huerta 2000). The lack of agreement indicates that the dative clitic is losing or has already lost its status as anaphoric pronoun to become a true object-verb agreement marker, a grammatical marker, almost like a verbal affix, signalling the presence of a prominent argument, the IO.12 The process giving rise to this was depronominalization involving reanalysis: unstressed anaphoric pronoun > object-verb agreement marker. In this regard, Spanish has another of the characteristics of a PO-SO language, that the dative recipient, not the accusative patient, controls the object-verb agreement marking. Table 7. Depronominalization of the plural dative clitic IO-V V-IO

(12) a.

b.

+AGREEMENT 96% (140/146) 43% (144/333)

- AGREEMENT 4%(6/146) 57% (189/333)

esclame usted con Eneas avisandole exclaim you with PN notifying-them-DAT-SG a sus companeros (Alzate, 1789,1.437) to his companions-DAT-PL 'Exclaim together with Aeneas notifying his companions' no hay que darle tanta no it-has that to give-them-DAT-SG so much importancia a las apariencias (spontaneous speech) importance to the appearances-DAT-PL 'Not so much importance should be given to appearances' su club de admiradores piensa organizar un his club of fans thinks to organize an acto en su memoria, aunque a muchos act in his memory, although to many-DAT-PL podria sonarle macabro (El Financiero, June 2000) it-could to sound-them-DAT-SG macabre 'His fan club plans to organize a commemorative act, although to many this could sound macabre'

238

Conception Company Company

α las to the

telenovelas soap operas-DAT-PL

le them-DAT-SG

hemos

agregado

no

un public

traditional

(Proceso, March 1998) we-have added an audience not traditional 'We have added a non-traditional audience to soap operas' The depronominalization process, as can be appreciated in table 7, is much more active in non-marked V-IO order. Nonetheless, it seems to be extending to all positions and it is already possible to document it with the 10 in a pre-verbal position, as is exemplified in (12b) above, attaining a nonnegligible 4% in the IO-V order. That is, the grammaticalization of objectverb agreement in the IO is also generalizing in word orders that are marked for the IO (Company in press a). Such a depronominalization has never occurred with accusative clitic pronouns in the history of Spanish. Accusative clitics continue to be anaphors and agree in gender and number with their referents. On the other hand, this change shows that Spanish is typologically consistent with the generalized feature that pronominal arguments incorporated into the verb usually have less phonological content that the respective independent pronouns (Bresnan and Mchombo 1987:765). The most important point that sustains the hypothesis presented here is that the loss of plural agreement of the dative clitic noticeably increases its frequency of use when the IO is inanimate. Table 8 (Huerta 2000) shows that the frequency of depronominalization increases from 52% with human IOs (57% considering all types of IO, table 7), to 82%, when only inanimate IOs are considered. That is, when the IO is similar to the DO as regards the prototypical lexical feature of the DO, inanimation, the depronominalization of the dative clitic increases and the object-verb agreement marker (an invariable le&) appears many more times; the IO reinforces its presence in the verb, indicating therewith that it continues to be the primary object of its sentence. Table 8. Depronominalization with inanimate IO

HUMAN INANIMATE

+ AGREEMENT 48% (135/282) 18% (9/51)

- AGREEMENT 52% (147/282) 82% (42/51)

Transitivity and grammaticalization of object

239

In terms of the hypothesis of this study, the two changes we have just analysed (loss of the preposition a in human DOs in ditransitive sentences, and increase of depronominalization with inanimate IOs) are similar, since both express the struggle of the DO and the IO for the functional space of object and for the corresponding grammatical object-marking. When the two objects are equal as regards lexical features -whether the DO gets close to the prototype of the IO via human DO or the IO comes close to the DO prototype via an inanimate IO - the dative IO always obtains the prominent grammatical status in the space of transitivity, either by retaining its own marking and weakening that of the accusative, the first change we analysed in this section, or by reinforcing its participation in the sentence structure with the generation of an object-verb agreement marker, the change we have just analysed. The lexical feature of animacy -or more specifically the feature of humanity, given the almost total absence of reference to animals in the diachronic analysed corpus - is the axis that builds the structural space of 'object' in Spanish and contrasts the IO with the DO. In the face of the typical animation of the IO, the most frequent and non-marked DO in Spanish is mostly inanimate.13 Therefore, the prototypes of both objects are located at opposite ends of the animacy hierarchy: human > other animate > inanimate force > inanimate (Givon 1995:46, 92). The IO naturally occupies the highest position, the DO, the lowest one. In the light of these contrasting positions as regards animacy, the two changes we have just analysed are very significant for the struggle of the DO vs. the IO: when the IO and the DO are located in the same zone of the animacy hierarchy, the IO always retains and reinforces its own marking, while the other object is always weakened or loses its marking.

5. Other changes to the DAT in its constitution as primary object Three further changes experienced by the dative category in the development of Spanish also show the status of the IO as a prominent or primary object, and are proof of the diachronic struggle of DO vs. IO, a struggle which, as I have been illustrating, caused Spanish slowly and progressively to acquire the features of a PO-SO language. These three changes, just like the changes analysed in the previous sections, indicate that the dative invaded functional spaces that were originally the accusative's privilege. These three changes are: anomalous dative marking on an accusative clitic;

240

Conception Company Company

generalization of leismo, and progressive invasion of the DAT in causative constructions. I will analyse them in that order. 5.1. Anomalous DAT marking on an ACC clitic In standard Spanish, the dative and accusative clitics must agree in number and person with their antecedents, and the accusative clitic, but not the dative, must also agree in gender. When both clitics occur in the same ditransitive sentence, le(s)-la, le(s)-lo, the dative clitic le(s), due to an old morphophonemic rule, becomes an invariable pronoun se < illi-illis, which is completely opaque as regards number, gender and person: se for el-ellosusted-ustedes 'he/she-they-you sg.-you pi.'. 14 The invariable result of the dative illi(s) > se when co-occurring with the accusative clitic is exemplified in (13), where a sentence such as Juan se lo/la compro derives from at least four ditransitive sentences (13a-d) (both objects are marked in bold type).15 (13) a. Juan compro John bought

unlibro, asuhijoj abook-ACC-SG to his son-DAT-SG

/ a usted)

b.

/ to you-DAT-SG *Juan let lOj compro John him/you-DAT-SG it-ACC-SG-MS bought > Juan sej lOj compro John him/you-DAT-SG it-ACC-SG-MS bought 'John bought a book for his son/you. John bought his son/you a book' Juan compro una bicicleta, a su hijOj John bought abike-ACC-SG to his son-DAT-SG / a ustedj

c.

/ to you-DAT-SG *Juan lej /a, compro John him/you-DAT-SG it-ACC-SG-FEM bought > Juan sej la, compro John him/you-DAT-SG it-ACC-SG-FEM bought 'John bought a bike for his son/you. John bought his son/you a bike' Juan compro un libro, a su hijoSj John bought a book-ACC-SG to his sons-DAT-PL

Transitivity and grammaticalization

d.

of object

241

/a ustedesι / to you-DAT-PL *J uan lej compro lo, John them/you-DAT-PL it-ACC-SG· MS bought > Juan sey compro toi John them/you-DAT-PL it-ACC-SG· MS bought 'John bought a book for his sons/you. John bought his sons/you a book' hijosj Juan compro una bicicletaj a su John bought a bike-ACC-SG to his sons-DAT-PL /a ustedesj / to you-DAT-SG *Juan lej compro la, John them/you-DAT-PL it-ACC-SG-FEM bought > Juan se) compro to, John them/you-DAT-PL it-ACC-SG-FEM bought 'John bought a bike for his son/you. John bought his son/you a bike'

Intermediate solutions between the normative solutions in (13) and the innovative solutions in (15) below, are attested in Old Spanish (14) with maintenance of the two sonorants: le lo (14), but they are very uncommon. (14)

todoj le lOj debe entregar all him-DAT it-ACC must to give (Fuero Juzgado, 13th c., apud Company 1998) 'He must give it all to him'

In American and Canarian Spanish, that is, in the dialects known as Atlantic Spanish, and in some areas of Peninsular Spanish, therefore in most Spanish dialects, an "anomalous" innovative pronominalization has become generalized in ditransitive sentences (15), when the two objects manifest as clitics and when the dative clitic has a plural referent and the accusative a singular one, as in (13c) and (13d) above. In this innovative pronominalization the DAT se < illi-illis projects its referential properties on the accusative clitic that follows it. The accusative behaves here as a mere phonological host that exhibits the morphological features of the dative (Company 1998), displaying either an anomalous plural -s marker, that is the number of the IO (15a) or the gender of the IO in addition to its number (15b), or as

242

Conception Company Company

of the IO (15a) or the gender of the IO in addition to its number (15b), or as in (15c) a complete dative clitic les which fully displaces the accusative clitic.16 That is, the accusative can totally give up its morphological properties in favour of the dative. (15) a.

b.

c.

el poco tiempo que me queda estoy the little time that me-DAT remains I-am con mis hijos y se los with my children and them-DAT it-ACC-SG=PL entrego totalmente a ellos I-give entirely to them-DAT 'In the little time I have left, I am with my children and I dedicate it entirely to them' si ellas me quieren comprar el caballo, if they me want to buy the horse, yo se las venderi I them-DAT it-ACC-SG-MS=PL-FEM will sell (Lope Blanch 1953) 'If they want to buy the horse from me, I will sell it to them' el cesto se les the basket them-DAT-PL it-ACC-SG=DAT-PL he regalado a unos chicos. (Gili Gaya 1970:234) have given to some boys Ί have given the basket to some boys'

The innovative pronominalization se los, se las, se les indicates that the dative behaves as the sole object in these ditransitive constructions and that the accusative remains inert. Data from Rivarola (1985), De Mello (1992:171) and Company (1998) show that the new cliticization behaves like a new non-analysable form that is frequently written as a single word, selos, selas, seles, in which the speakers do not recognize two objects, but a simple structure with only one object, the DAT-ΙΟ; they no longer identify the DO. The immobilization or weakening of the accusative in double object constructions is one of the main characteristics of PO-SO languages (Comrie 1982:97, Vazquez 1996:549-550, Company 2001). This innovative cliticization displaying only a dative object is undoubtedly a strong argument to assert that many Spanish dialects behave in this area like a true PO-SO language.

Transitivity and grammaticalization of object

243

5.2. Generalized leismo This change deals with the extension of the dative case to mark the patient of a monotransitive sentence. It may be considered a quite natural change in a language that is acquiring primary object features. As is well-known, it consists in using a dative clitic le(sj to express the DO function, as examples in (5a) above showed, repeated here as (16a). This is a greatlyextended innovation, present in numerous Spanish dialects, including those traditionally classified as more conservative, such as the American ones. It is perfectly well, and even systematically, accepted from very early times, and it exhibits broad syntactic distribution, since it appears with a large number of transitive verbs and a varied lexical gamut of referred nominals. An apparently similar change, that at first glance seems to constitute a counterweight to leismo, the use of an accusative clitic to express the 10 function, laismo (16b) and loismo (16c), has not succeeded in becoming systematic in practically any dialect of Spanish. It is limited to a few dialects in Central and Northern Spain (Klein-Andreu 1992:171), continues being stigmatized and restricted to certain lexical types of verbs and is completely unknown in the conservative varieties of Spanish, as for example American Spanish. (16) a.

b.

c.

el telefono ya cuelgale the telephone already hang it-DAT-DO up 'The telephone, hang it up' maestra / maestro, £le ayudo? teacher-MS/FEM, him/her-DAT-DO I-help 'Teacher, may I help you?' la dio un beso her-ACC-IO (s)he gave a kiss '(S)he gave her a kiss / (S)he gave a kiss to her' lo dio una bofetada him-ACC-IO (s)he gave a slap '(S)he gave him a slap / (S)he gave a slap to him'

Although at first sight leismo, laismo and loismo seem complementary changes and seem to weaken our hypothesis, the three phenomena have a very different status since there are many dialects with leismo but that lack laismo and loismo, and there are no dialects that only have laismo - loismo, and lack leismo. That is, dative for accusative is accepted in all dialects, but

244

Conception Company Company

accusative instead of dative is usually rejected. In this area of transitivity a situation similar to that of the previous change is found: the dative can easily take away accusative marking, but the opposite case, accusative instead of dative, presents multiple restrictions and is conflictual or unknown, which means that in an implicational hierarchy, leismo, dative for accusative, occupies a higher hierarchical position than laismo and loismo. Leismo may be classified, like the foregoing change, as a certain kind of dative promotion, but laismo and loismo are usually rejected17 because they suppose a demotion or weakening of the prominence of the dative in Spanish. In a certain measure, they violate the structural pattern of Spanish we have been analysing, by which datives displace accusatives from the status of object, and reinforce their own marking at the expense of weakening the accusative, but never in the history of Spanish has the accusative case displaced the dative case. Quantitative diachronic data of leismo, laismo and loismo are very enlightening as regards the privileged status of the dative as the primary syntactic object. In the period between the 13th and the 19th centuries, the global frequency of leismo for a large prose corpus in Peninsular and American Spanish is 68%, vs. 28% for laismo and only 4% for loismo. By the end of the 18th century, leismo in Castillian texts reaches 99% and loismo represents only 9% (Flores in press). In the 20th century, the most innovative Spanish dialects, Central and Northern Castille, have only 15% of loismo, in comparison to 90%-100% of leismo (Fernandez Ordonez 1993:92; 1999). This means that an unbalanced situation exists in this area of transitivity which favours dative marking, and disfavours accusative marking. That is, the same state of affairs exists as in the previous change: the dative can easily take over the accusative distribution in almost any (di)transitive construction, but the opposite process has never been successful in the grammar of Spanish.

5.3. Progressive invasion of the DAT in causative constructions The use of a dative clitic as the causee subject in causative constructions with a subordinate clause in the infinitive is a Romance innovation with respect to the Latin construction, which required an accusative as the causee subject of the lower clause, a construction known as acusativus cum infinitivo (17). What shows that this change is part of the same general structural evolutionary pattern by which Spanish is acquiring the features of a PO-SO

Transitivity and grammaticalization

of object

245

language is that this language led the generalization of the dative for expressing the causee subject to extremes unknown in other Romance languages. As has been pointed out in Lazard (1998), causativity is a test area for showing the existence of a participant dative object. The change I shall present now is solid evidence of this. (17) a.

qui nati coram me who-PL born-PL-PTC before me-ACC cernere letum fecisti (Aeneid 2.539, Alfonso 1998:199) to see death-ACC you-made [me-ACC-causee subject; cerraere-subordinate infinitive; fecisticausative verb] 'You that made me look at my son's death face to face'

In general, Romance languages, like most of the world's languages (Comrie 1976, Alfonso 1998), select a dative or an accusative as the causee subject of the subordinate clause, according to the transitivity or intransitivity of the subordinate infinitive: if it is transitive, the causee subject appears in the dative case (18a); and if intransitive, the causee subject appears in the accusative (18b). (18) a.

b.

j 'ai fait manger la pomme a Claude > je lui ai fait manger la pomme. /MI-DAT causee; manger-subordinate transitive infinitive Ί made Claude eat the apple' faccio mangiare la mela al bimbo > Glifaccio mangiare la mela. gli-DAT causee; manger-subordinate transitive infinitive Ί make the boy eat the apple' j 'aifait aller Claude > je l'ai fie ai) fait aller. le-ACC causee; a//er-subordinate intransitive infinitive Ί made Claude go' Gianni fa venire Paolo > Gianni lo fa venire. lo-ACC causee; vew're-subordinate intransitive infinitive 'John makes Paul come [home]'

Spanish is peculiar in that the dative case for coding the causee subject has advanced further than in other Romance languages. In fact, from very early times, it is possible to document numerous intransitive infinitives with the causee subject in dative (19a). That structural scheme constitutes a con-

246

Conception Company Company

struction exclusive to Spanish, not shared by other Romance languages. The opposite marking, that is, the accusative as the causee of a transitive infinitive, is also possible in Spanish (19b), but invasion in this second case did not come anywhere near the advance of the dative case in the other environment. The frequencies of use of the respective invasions are quite different, and those differences are consistent with the global diachronic movement we have been analysing, because they corroborate the extension of the dative and the weakening or displacement of the accusative. In Medieval Spanish, from the 13th to the 15th centuries, Alfonso (1998) documents only 8% (15/184) of accusative causees with transitive infinitives, whereas the invasion of dative causees to intransitive verbs increases to 43% (60/138). (19) a.

b.

e descendieron a el e besaronle las manos and they got off to him and kissed-him the hands e fizieronle cavalgar en aquel and made-him-DAT-causee to ride-INTR on that palafren (14th c., Alfonso 1998) tame horse 'They got off and kissed his hands and made him ride on that tame horse' Fizo los a todos he-made them-ACC-causee to all yurar quel ayudassen a deffender to-swear-TR that-him help to to-defend Roma( 13th, Alfonso 1998) Rome 'He made them all swear that they would help him to defend Rome'

This change shows a double reinforcement of the marking in Spanish. One was an early one shared with the other Romance languages, consisting of the emergence of the dative for marking the causee subject. Another later reinforcement, exclusive to Spanish and having greater weight for our hypothesis, consisted of the progressive displacement of the accusative in favour of the dative as the causee subject of the lower clause, whether the infinitive was transitive or intransitive.

Transitivity and grammaticalization of object

247

6. Causes of the diachronic DO-IO struggle Each of the changes analysed above undoubtedly has its own history and causation. Nonetheless, in my opinion, there are two general underlying reasons of a theoretical nature at work, common to all the changes. On the one hand, there is the very different prototypical semantics inherent to accusative and dative, and on the other the hierarchical relations sustained by these cases. Both phenomena brought about a deep restructuring of transitivity in Spanish in the functional 'object' slot. Most known languages are sensitive to the semantics of cases and make grammatical distinctions based on the degree of animacy, definiteness, relative topicality and degree of affectedness or relative salience of each of the two objects. Spanish is also sensitive to these semantic features and structures a good part of its transitivity system by exploiting the lexical and discoursive semantics of the accusative and dative cases. An almost universal contrast that organizes transitivity in most languages is the distinction between agent-like cases and patient-like cases (Dowty 1991, Tenny 1992, Van Valin and La Polla 1997: cap.3, Van Valin and Wilkins 1996). The dative belongs to the former, the accusative, to the latter one. That is, the two cases are defined in two different and contrasting conceptual domains, and such a contrast is reflected in the syntactic structure of transitivity. The dative prototypically refers to specific, individuated human or animate entities with a high degree of topicality and a certain degree of volition and agentivity. On the contrary, the accusative is prototypically inanimate, lacks energy or volition, regularly refers to things or abstract concepts that are the affected patient, and usually experiences a change of state because of the transmission of energy entailed in the action of the verb. Given these important semantic differences in the prototypical configuration of each of the two objects, each of them is located at opposite ends of the animacy and topicality hierarchies; thus, the dative occupies a very high position, the highest after the nominative-agent, and the accusative occupies the lowest one (Dowty 1991:578; Givon 1984a:156, 1984b: 133-134, 169). The dative has greater semantic prominence than the accusative and, consequently, appears coded in the syntax of many languages above the accusative. Spanish sharply fits such case-coding distinctions. Different studies on (di)transitivity have shown the core role that the semantics of objects plays in defining what object becomes the primary syntactic object. There seems to be a general pragmatically-driven tendency

248

Conception Company Company

within which cases that inherently have a greater semantic and pragmatic salience, and that occupy higher hierarchical positions in animacy and topicality scales are coded as primary objects more easily than those of lesser prominence and hierarchy (Bresnan and Moshi 1990:151; Hawkinson and Hyman 1974:149; Hyman and Duranti 1982:224; Newman 1996:28-29). There are two consequences of this general typological behaviour: on the one hand, the human vs. non-human distinction is usually an important diachronic source of primary objectivity (Dryer 1986:841), and, on the other, there is migratory grammatical behaviour in which cases in higher positions, such as the dative, may easily extend their functions and marking to overtake the role of low animacy and topicality nominals, like the accusative, whereas low animacy nominals hardly ever take the place of those high in animacy (Rodrigues Aristar 1997:353-355). The changes we have just analysed show that Spanish is consistent with this general semanticpragmatic configuration, as well as with typological migratory behaviour, and the human vs. non-human distinction, or more generally, animate vs. inanimate, is a fundamental axis for organizing transitivity in this language. 18 A recent distinction in functional linguistics between two cognitive patterns or models of transitivity is closely related to the selection of the accusative patient or the dative recipient as the primary syntactic object. This distinction is the one between the manipulation model and the interaction model (Tuggy 1997, Taylor 1997). In my opinion, this distinction provides strong support for the grammatical distinction between DO-IO vs. PO-SO languages. The manipulation pattern emphasizes the patient and results in a DO-IO language. In this pattern, an agent, prototypical ly human, manipulates a patient, prototypical ly inanimate, with the consequence that in DOIO languages the accusative patient is the nominal that is emphasized or profiled in the transitivity relations, and it is the patient the nominal that has the status of syntactic object in the argument structure and controls the tests for objectivity. On the contrary, the interaction pattern highlights the final goal of the transitivity, the dative recipient, and the result is a PO-SO language. In this second model of transitivity, an agent, prototypically human, interacts with the recipient, also prototypically human, that is, a human interacts with another human, consequently the dative recipient is the object that is highly profiled in transitivity and which, therefore, displays the properties of the syntactic object. Languages can grammatically exploit both models or can select one of them. Spanish would originally have been among the former languages, but diachronically it has been moving toward

Transitivity and grammaticalization of object

249

the interaction model with an explicit and multiple coding of the dative participant, and a sustained hiding and/or weakening of the accusative participant. Nowadays Spanish has a mixed status, somewhere between a DO-IO and PO-SO language.

7. Conclusions We have seen that eight apparently non-related syntactic changes in the development of Spanish constitute a unique underlying diachronic pattern that yields a deep rearrangement in the transitivity system. The origin of the changes was the old diachronic struggle between the dative recipient and the accusative patient to attain the status of object in transitivity from the language's beginnings. We have also seen that this is an unequal struggle. The consequence of such unbalanced competition was that the dative always reinforced its formal coding in the argument structure and progressively took over the grammatical functions that were originally exclusive to the accusative, while the accusative progressively weakened its marking and lost structural spaces within (di)transitivity. The changes may be interpreted as a multiple grammaticalization of the dative participant. The very different semantic and pragmatic meanings of the two objects involved in the changes were decisive in causing this multiple grammaticalization of dative marking and the weakening of accusative marking. The changes are micro-irregularities, but together constitute a crucial global shift in the transitivity system of Spanish that may be interpreted as an elementary catastrophe in terms of Catastrophe Theory. We have also seen that all the changes caused Spanish to exhibit clear characteristics of a PO-SO language in several zones of (di)transitivity, the dative being the prominent, and sometimes the only, syntactic object of the argument structure in the sentence. This innovative typological configuration that Spanish has been acquiring for over a thousand years causes the distinction between the dative and the accusative, or between the D O and the 10, to be quite evanescent and very difficult to apprehend in this language. Finally, I consider that this paper gives support to the well-known fact that historical linguistics is an explanatory discipline, since it may inform controversial synchronic analysis, and shed light on grammatical areas which are difficult to characterize. Dativity in Spanish is undoubtly a case in point: one of those complex and heterogeneous synchronic areas in

250

Conception Company Company

which a diachronic view contributes new perspectives of analysis and a b e t ter understanding of such an intricate dynamic category.

Notes 1.

2.

3.

4.

5.

6.

Despite the fact that the extension of the prepositional case-marker to inanimate direct objects is a very active change in present-day Spanish, this marking continues to be classified as being for "usos excepcionales" 'exceptional uses' (Roegiest 1998:472). Sporadically the 10 can appear as a simple NP heading its sentence and with a co-referential clitic, a construction known as nominativus pendens: senorita extranjera le gustaria compartir habitation con muchacha del pais 'foreign lady would like to share a room with a native girl'. For the analysis of these constructions in Spanish diachrony and its proximity to a subject, cf. Company (in press b). In the current bibliography on primary objectivity, there are two slightly different, although complementary, ways of understanding the concept of 'primary object'. One is more formally syntactic, and the other gives greater weight to the semantics of the involved entities. In the first sense, a primaryobject language is one which deals with the dative of a ditransitive sentence in the same way as with the accusative of a monotransitive sentence. In monotransitives, the argument which controls the tests for objectivity (adjacency to the verb, control of object-verb agreement, passivization, access to relativization, etc.) is the accusative, but it is the dative which controls the tests in ditransitive sentences (Aissen 1983, Chung 1976, Comrie 1982, Dryer 1986, among many others). In the more semantic way of treating primary objectivity (Givon 1984a, Hyman and Duranti 1982, Shibatani 1996, Company 2001), a primary-object language is one which in general gives a higher hierarchy to the DAT than to the accusative, both in ditransitive sentences, and in other areas of (di)transitivity, although the dative may not systematically fulfill the objectivity tests. The counts for the 13th to the 15th centuries were done on the first three hundred inanimate direct objects that appeared in each of the texts from the corpus of that period. Semantic flexibilization went hand in hand with a formal flexibilization in the manifestation of the DAT category, since in the latter periods in the corpus, the IO appears with greater frequency in the form of an NP, although the clitic continues to be the formal basic manifestation of this category (Company, in press a). This semantic flexibilization has passed unnoticed in most studies. The reference to humans is such a key feature of the dative or IO that it is generally as-

Transitivity and grammaticalization of object

251

serted that "los complementos indirectos que refieren a cosas ya de por si son sumamente excepcionales" 'indirect objects that refer to things are extremely exceptional' (Nieuwenhuijsen 1998:39). When inanimate IO are recognized, some traditional grammars suppose a metaphoric personalization: "Si yo digo: le arranco las hojas al ärbol, hojas es complemento directo de cosa; ärbol es complemento indirecto tambiön de cosa; pero en el primer complemento hojas hay un matiz deferenciado que lo deja en objeto. En cambio, ärbol aparece digniflcado con una especie de personalizaciön gramatical" (Alonso 1964:431) 'If I say, I tear the leaves from the tree, leaves is the direct object which is a thing; tree is the indirect object which also refers to a thing; but in the first object, leaves, there is a differentiated nuance that leaves it as the object. On the contrary, tree appears to be dignified with a sort of grammatical personalization'. 7. This is a change whose real behaviour is difficult to evaluate in periods prior to the 20th century. It is very probable that style correction by editors may have constantly interfered with the process of critical editing of the texts, and that the editors have replaced a plural dative clitic on more than one occasion. Nonetheless, it is significant that the first documentation is precisely in the 16th century. 8. The multiple changes in the dative category would reflect what is pointed out in Lexical Functional Grammar, that in languages it is more important to distinguish between objects than between object and subject (Alsina 1993:453— 454). 9. As in most syntactic changes, the ones analysed here are multiply motivated. It is not by chance that the 16th century is the time of breakage: it was during this period that the noun phrase vuestra merced 'your Grace' grammaticalised in the personal pronoun usted-ustedes 'you' (singular and plural), and also that the pronoun vosotros-as, its adjective-pronoun vuestro(s)-a(s) and the clitic os, that together formed the pronominal paradigm of the second person plural, were lost in American Spanish. This loss forced the generalization of the pronoun ustedes as the only possible reference for several hearers and the extension and generalization of the dative clitic les as the sole possible clitic for that new pronoun. As regards the loss of vosotros-as and the readjustment of the pronominal system, see Lapesa (1970), and for the impact of this change on syntax, see Company (1997). 10. An infrequent solution to this conflictive type of construction is for the DO to keep a preposition and the IO to become a circumstantial complement, such as an oblique PP, which is in some way similar in its meaning to a goal of the transitivity: presento a su mujer con sus alumnos 'He introduced to his wife with his pupils'. Nonetheless, it is much more frequent to maintain the double object construction and leave the DO without the prepositional case-marking.

252

Conception Company Company

11. For the pragmatic relevance and pervasiveness of this atypical marked scheme with double preposition a, see Ortiz in this volume. 12. A similar analysis, although in terms of case discrimination, dative vs. accusative, is that of Garcia (1975:388): given that, as has already been pointed out, the animate DO as well as the 10 has the preposition a: vi al nino Ί saw to the child' - le dije al nino Ί told to the child'. Duplication with a dative clitic means that there is an 10 in the sentence containing it. 13. The relative frequencies of the DO indicate that the prototypical feature of this category is inanimate: 86% for the 16th century (Navarrete 2000) and 98% (Väzquez Rozas 1995:155) or 91% (Garcia Miguel 1995:50) for the 20th century, depending on the respective corpora. 14. The morphophonemic process was: Latin illi(s) + illum-illam-illud > Spanish le(s) + lo-la > ge lo- ge la, phonologically, /2e lo-ze la/ > se lo- se la. 15. The sequence se lo, se la also results from reflexive sentences. For the weight of the homonymy of dative se < Uli with reflexive se < se in the motivation of this change, see Company (1998). 16. This phenomenon is tagged as false leismo (Llorente 1980:23) and is not associated with the other similar constructions se los, se las, despite the fact that they show the same structural pattern as the examples in (14a) and (14b). 17. The traditional explanation is that laismo and loismo are rejected because they are stigmatized. In my opinion such an explanation is circular and inverts the cause-effect relation. Rather, the cause is that laismo and loismo do not fit the general structural pattern analysed in this paper, by which the DAT can easily usurp the primary object function, and the stigmatization of laismo and loismo is the consequence or the effect. That is, they would be rejected because they lack the structural Spanish pattern by which datives displace accusatives, and consequently are stigmatized. 18. It is worthwhile quoting Roegiest (1990:245): "El espaflol se deja guiar por parämetros semänticos, hasta tal punto que todo segmento susceptible de ser interpretado como el participante menos activo es identificable con un objeto directo y contribuye a considerar el objeto +humano como objeto indirecto y a marcarlo consecuentemente" 'Spanish is guided by semantic parameters to such a point that every segment susceptible of being interpreted as the least active participant is identifiable with a DO and contributes to the consideration of the +human object as an IO and to marking it accordingly'.

Transitivity and grammaticalization of object

253

References A) Corpus (in chronological order and by abbreviations) 13th: Calila = Anönimo, Calila e Dimna. Edited by Juan Manuel Cacho Blecua and Maria Jesus Lacarra. Madrid: Castalia, 1988. GE = Alfonso X, General estoria. Segunda parte. Edited by Antonio G. Solalinde, Lloyd A. Kasten and Victor R.B. Oelschläger. 2 vols. Madrid: CSIC, 1957. 14th: Zifar = Libro del cavallero Zifar. Edited by Joaquin Gonzalez Muela. Madrid: Castalia, 1982. Pedro I = Pero Löpez de Ayala, Cronica del rey don Pedro I. Edited by Constance L. Wilkins and Heanon Wilkins. Madison: Hispanic Seminary of Medieval Studies, 1985. 15th: Corbacho = Arcipreste de Talavera, Corbacho. Edited by Joaquin Gonzälez Muela. Madrid: Castalia, 1970. Celestina = Fernando de Rojas, La Celestina. Edited by Dorothy S. Severin. Madrid: Cätedra, 1993. 16th: Lazarillo = Anonymous, Trilinear edition of Lazarillo de Tormes of 1554. Burgos, Alcala de Henares, Amberes. Edited by Joseph V. Ricapito. Madison: Hispanic Seminary of Medieval Studies, 1987. Santa Teresa = Escritos de Santa Teresa. Afiadidos e ilustrados por don Vicente de la Fuente. Vol. 2: Cartas (1-12). Madrid: Biblioteca de Autores Espanoles (BAE), volume LV, 1862. [Reprinted in Madrid: AtlasReal Academia Espafiola, 1952], DLNE = Conception Company-Company, Documentos lingüisticos de la Nueva Espaha. Altiplano Central. Mexico: UNAM, 1994. Bernal = Bernal Diaz del Castillo. Historia verdadera de la conquista de la Nueva Espaha. Edited by C. Säenz de Santamaria. Madrid: Alianza Universidad, 1991. 18th: Alzate = Gazetas de Literatura de Mexico por don Jose Antonio Alzate Ramirez, years 1788-1793. Mexico: Fondo Reservado de la Biblioteca Nacional, 4 vols. Moratin = Leandro Fernandez de Moratin, La comedia nueva and El si de las ninas. Edited by John Dowling and Rene Andioc. Madrid: Castalia, 1975. 19th: Pepita Jimenez = Juan Valera, Pepita Jimenez. Edited by Leonardo Romero. Madrid: Cätedra, 1989. Regenta = Leopoldo Alas "Clarin", La Regenta. Edited by Gonzalo Sobejano. 2 vols. Madrid: Castalia, 1981. Bandidos = Manuel Payno. Los bandidos de Rio Frio. 2 vols. Mexico: Cläsicos de la Literatura Mexicana, 1979.

254

Conception Company Company

20th: El habla de la Ciudad de Mexico. Materiales para su estudio. Compiled by Juan M. Lope Blanch. Mexico: UNAM, 1971 El habla popular de la Ciudad de Mexico. Materiales para su estudio. Compiled by J.M. Lope Blanch. M6xico: UNAM, 1976 Different numbers of newspapers and magazines: Proceso, El Financiero and La Jornada between 1996-2000. B) References Aissen, Judith L. 1983 Indirect object advancement in Tzotzil. In: D.M. Perlmutter (ed.), Studies in relational grammar, 272-302. Chicago: The University of Chicago. Alarcos Llorach, Emilio 1994 Gramätica de la lengua espanola. Madrid: Espasa-Calpe Real Academia Espaflola. Alfonso, Milagros 1998 Construcciones causativas en el espanol medieval. Estructura y evolution. Mexico: UNAM-E1 Colegio de M6xico. Alonso, Martin 1964 Evolution sintäctica del espanol. Madrid: Aguilar. Aisina, AAlex [1993] 1995 Predicate composition: A theory of syntactic function alternations. PhD Dissertation. Stanford University. Ann Arbor, Mi: University Microfilms International. Bello, Andres [ 1847] 1978 Gramätica de la lengua castellana. Madrid: Edaf Universitaria. Blansitt Jr., Edward L. 1984 Dechticaetiative and dative. In: Frans Plank (ed.), Objects. Towards a theory of grammatical relations, 127-150. London: Academic Press. Bossong, Georg 1998 Le marquage differentiel de l'objet dans les langues d'Europe. In: Jack Feuillet (ed.), Actance et valence dans les langues de I'Europe, 193-258. Berlin-New York: Mouton de Gruyter. Bourciez, Edouard [1910] 1967 Elements de linguistique romane. Paris: Klincsieck. Bresnan, Joan and Sam A. Mchombo 1987 Topic, pronoun and agreement in Chichewa. Language 63, 4: 741782.

Transitivity and grammaticalization of object

255

Bresnan, Joan and Lioba Moshi 1990 Object asymmetries in comparative Bantu syntax. Linguistic Inquiry 21,2: 147-185. Butt, John and Carmen Benjamin 1994 A New Reference Grammar of Modern Spanish. Lincolnwood: NTC Publishing Group. Cano Aguilar, Rafael 1981 Estructuras transitivas en el espanol actual. Madrid: Gredos. Chung, Sandra 1976 An object-creating rule in Bahasa Indonesia. Linguistic Inquiry 7,1: 41-87. Company-Company, Concepciön 1997 El costo gramatical de las cortesias en el espafiol americano. Las consecuencias sintäcticas de la pördida de vosotros. Anuario de Letras 35: 167-191. Company-Company, Concepciön 1998 The interplay between form and meaning in language change. Grammaticalization of cannibalistic datives in Spanish. Studies in Language 22,3: 529-566. Company-Company, Concepciön 2001 Multiple dative-marking grammaticalization. Spanish as a special kind of primary object language. Studies in Language 25, 1: 1-47. Company-Company, Concepciön 2002 Grammaticalization and category weakness. In: G. Diewald and I. Wischer (eds.), New reflections on grammaticalization, 201-216. Amsterdam: John Benjamins. Company-Company, Concepciön in press a El objeto indirecto. In: Concepciön Company-Company (ed.), Sintaxis historica del espanol, vol.1: La frase verbal. M6xico: Fondo de Cultura Econömica and UNAM. Comrie, Bernard 1976 The syntax of causative constructions: cross-linguistic similarities and divergences. In: M. Masayoshi Shibatani (ed.), Syntax and Semantics, 6: The grammar of causative constructions, 261-312. New York: Academic Press. Comrie, Bernard 1982 Grammatical relations in Huichol. In: Paul J. Hopper and Sandra A. Thompson (eds.), Syntax and semantics, 15: Studies in transitivity, 91-115. San Diego - New York: Academic Press.

256

Conception Company Company

Delbecque, Nicole and Beatrice Lamiroy 1996 Towards a typology of the Spanish dative. In: W. Van Belle and W. Van Langendonck (eds.), The dative, vol. 1: Descriptive studies, 73117. Amsterdam: John Benjamins. De Mello, George 1992 Se los for se lo in the spoken cultured Spanish of eleven cities. Hispanic Journal 13,1: 165-179. Dowty, David 1991 Thematic proto-roles and argument selection. Language 67, 3: 547620. Dryer, Matthew .S. 1986 Primary objects, secondary objects, and antidative. Language 62, 4: 808-845. Echarte Cossio, Maria Jos6 1996 Acusativo y dativo: dinämica sincrönica del latin al castellano. Revista Espanola de Lingüistica 26, 1: 83-107. Faltz, Leonard M. 1978 On indirect objects in universal syntax. Chicago Linguistic Society 14: 76-88. Fernändez-Ordöfiez, Inds 1993 Leismo, laismo y loismo: estado de la cuestiön. In: Olga Fernandez Soriano (ed.), Los pronombres ätonos, 63-96. Madrid: Taurus Universitaria. Fernändez-Ordöfiez, Ines 1999 Leismo, laismo y loismo. In: Ignacio Bosque and Violeta Demonte (eds.), Gramätica descriptiva de la lengua espanola, vol. 1: 13 Π Ι 398. Madrid: Espasa Calpe. Flores, MMarcela In press Leismo, laismo y loismo en el espahol. Estructura y evolution. Möxico: UNAM. Folgar, Carlos 1993 Diacronia de los objetos directo e indirecto. Del latin al castellano medieval. Santiago de Compostela: Universidad de Santiago de Compostela [Verba, anexo 37]. Garcia, Erica C. 1975 The Role of Theory in Linguistic Analysis. Amsterdam: North Holland. Garcia. Erica C. 1992 Syntactic diffusion and the irreversibility of linguistic change: personal a in Old Spanish. Manuscript.

Transitivity and grammaticalization of object

257

Garcia-Miguel, Jose Maria 1995 Transitividady complementation prepositional en espanol. Santiago de Compostela: Universidad de Santiago de Compostela [Verba, Anexo 40]. Gili Gaya, Samuel [ 1961 ] 1970 Curso superior de sintaxis espanola. Barcelona: Bibliograf. Givon, Talmy 1984a Direct object and dative shifting: Semantic and pragmatic case. In: Frans Plank (ed.), Objects. Towards a theory of grammatical relations, 151-182. London: Academic Press. Givon, Talmy 1984b Syntax. A Functional-Typological Introduction. Amsterdam: John Benjamins. Givön, Talmy 1995 Functionalism and Grammar. Amsterdam: John Benjamins. Goldberg, Adele Ε. 1992 The inherent semantics of argument structure: The case of the English ditransitive construction. Cognitive Linguistics 3, 1: 37-74. Hawkinson, A. and L. Hyman 1974 Hierarchies of natural topic in Shona. Studies in African Linguistics 5: 147-170. Hopper, Paul J. 1987 Emergent grammar. Berkeley Linguistics Society 13: 139-157. Hudson, Richard 1992 So called 'double-objects' and grammatical relations. Language 68, 2: 51-276. Huerta Flores, Norohella 2000 Presencia vs. ausencia de concordancia del OI duplicado en el espanol actual. Bachelor thesis. Mexico: UNAM Hyman, L. and Alessandro Duranti 1982 On the object relation in Bantu. In: Paul J. Hopper and Sandra A. Thompson (eds.), Syntax and Semantics, 15: Studies in Transitivity, 217-239. New York: Academic Press. Klein-Andreu, Flora 1992 Understanding standards. In: G. W. Davis and G. K. Iverson (eds.), Explanation in Historical Linguistics, 167-178. Amsterdam: John Benjamins. Kurytowicz, Jerzy [1965] 1976 The evolution of grammatical categories. Esquisses linguistiques, 38-45. Munich: Fink.

258

Conception Company Company

Laca, Brenda In press El objeto directo. In: Concepciön Company-Company (ed.), Sintaxis historica del espanol, vol. 1: La frase verbal. Mexico: UNAM. Lapesa, Rarafel 1970 Personas gramaticales y tratamientos en espanol. Homenaje a Menindez Pidal, vol. IV. Revista de la Universidad de Madrid 19, 74: 141-167. Lazard, Gilbert 1998 Definitions des actants dans les langues europeennes. In: Jack Feuillet (ed.), Actance et valence dans les langues de l'Europe, 11— 146. Berlin-New York: Mouton de Gruyter. Lope Blanch, Juan M. 1953 Observaciones sobre la sintaxis del espanol hablado en Mexico. Mexico: Publicaciones del Instituto Hispano Mexicano de Investigaciones Cientificas. Llorente Maldonado, Antonio 1980 Consideraciones sobre el espaftol actual. Anuario de Letras 18: 5-61. Meyer-Lübke, Wilhelm [1890,1906]1974. Grammaire des langues romanes, vol.3: Syntaxe. GeneveMarseille: Slatkine-Lafitte Reprints. Navarrete, Maria de los Angeles 2000 El objeto directo en el siglo XVI: espanol de indigenas y espanol de espanoles. Master thesis. Mexico: UNAM. Newman, John 1996 'Give'. A cognitive linguistic study. Berlin: Mouton de Gruyter. Niewenhuijsen, Dorien 1998 Cambios en la colocacion de los pronombres ätonos en la historia el espanol. PhD Dissertation. Amsterdam: Thesis Publishers. Ortiz, Rosa Maria 2001 Diacronia de las oraciones bitransitivas en el espanol. PhD Dissertation. Mexico: UNAM. Pensado, Carmen 1984 Sobre la "-i" de algunas formas pronominales en los antiguos dialectos hispänicos. Boletin de la Real Academia Espanola 64: 231232,143-170. Pensado, Carmen 1985 La creation del objeto directo preposicional y la flexiön de los pronombres personales en las lenguas romänicas. Revue Roumaine de Linguistique 30, 2: 123-158. Real Academia Espanola (RAE) 1973 Es bozo de una nueva gramätica de la lengua espanola. Madrid: Espasa Calpe.

Transitivity and grammaticalization of object

259

Rivarola, Jose Luis 1985 Se los por se lo. Lexis 9: 239-242. Rodrigues Aristar, Anthony 1997 Marking and hierarchy. Types and the grammaticalization of casemarkers. Studies in Language 21, 2: 313-368. Roegiest, Eugeen 1990 La tipologia sintäctica del objeto transitivo en espanol. Verba 17: 239-248. Roegiest, Eugeen 1998 Variation del objeto directo espafiol y dinamicidad verbal. In: Nicole Delbecque and C. De Paepe (eds.), Estudios en honor del profesor Josse de Kock, 469-488. Leuven: Leuven University Press. Roegiest, Eugeen 1999 Object direct pr0positionnel ou objet indirect en espagnol. Verbum 21, 1: 67-80. Sänchez Märquez, Francisco 1982 Gramätica moderna del espanol. Teoria y norma. Buenos Aires: EDIAR. Shibatani, Masayoshi 1996 Applicatives and benefactives: A cognitive account. In: Masayoshi Shibatani y Sandra A. Thompson (eds.), Grammatical Constructions. Their Form and Meaning, 157-194. Oxford: Clarendon Press. Taylor, John 1997 Double object constructions in Zulu. In: J. Newman (ed.), The Linguistics of Giving, 67-96. Amsterdam: John Benjamins. Tenny, Carol 1992 The aspectual interface hypothesis. In: I. Sag y A. Szabolcsi (eds.), Lexical Matters, 1-27. Stanford: Stanford University Press. Thom, Ren6 1983 Paraboles et catastrophes. Paris: Flammarion. Timberlake, Alan 1977 Reanalysis and actualization in syntactic change. In: Charles N. Li (ed.), Mechanisms of Syntactic Change, 141-177. Austin: Texas University Press. Tuggy, David 1997 Giving in Nahuatl. In: J. Newman (ed.), The Linguistics of Giving, 35-66. Amsterdam: John Benjamins. Van Valin, Robert D. and David P. Wilkins 1996 The case for 'effector': Case roles, agents, and agency revisited. In: Msayoshi Shibatani and Sandra A. Thompson (eds.), Grammatical Constructions. Their Form and Meaning, 289-322. Oxford: Clarendon Press.

260

Conception Company Company

Van Valin, Robert D. and Randy J. La Polla 1997 Syntax (Structure, Meaning & Function). Cambridge: Cambridge University Press. Vazquez Rozas, Maria Victoria 1995 El complemento indirecto en espanol. Santiago de Compostela: Universidad, Servicio de Publicaciönes e Intercambio Cientifico. Vazquez Soto, Veronica 1996 El participante no-sujeto en cora: orden de palabras, codificaciön y marcaciön de niimero. In: Z. Estrada, M. Figueroa and G. Löpez Cruz (eds.), Tercer encuentro de lingüistica en el nordeste, vol. 1: Lenguas indigenas, 533-554. Hermosillo, Sonora: Editorial Unison. Wildgen, F. 1987 Catastrophe theory as a basic tool in theoretical linguistics. Theoretical Linguistics 14: 259-294.

Non-diachrony of ditransitivity in Spanish

Rosa Maria Ortiz Ciscomani

1. Introduction Many are the papers that deal with transitivity in languages. Particularly since Hopper and Thompson's 1980 article, which has become classic, transitivity has generated a very broad bibliography, for it has been a recurrent subject in papers adopting both a typological and theoretical perspective. But what practically has not been studied is ditransitivity. Today we know scarcely anything about what happens when the two structural boxes of a verb's objects are lexically saturated. This paper deals with ditransitive lexical constructions in Spanish, that is, those in which two concurrent objects, direct object (DO) and indirect object (10), express themselves through full nominal phrases, that is, with a noun or full pronoun as the nucleus. The study, while dealing with an unexplored subject in Spanish, provides important data for ditransitivity in other languages, as well as for synchronic and historical Spanish grammar. In diachronic perspective, the data reveal an unexpected fact: the historical resistance to change that the ditransitive construction has maintained for eight centuries. In comparison with the behaviour of the object of the monotransitive construction, in which the prepositional marking a, proper to the indirect object, has practically extended to every type of object (Company in press, Delbecque and Lamiroy 1996:109),' the marking of the direct object concurrent with the indirect object in ditransitive constructions in Spanish, present in a very low proportion, has not experienced any change whatsoever, but rather, has been preserved in almost identical conditions throughout the language's history. The order of constituents in this construction registers a similar phenomenon.

262

Rosa Maria Ortiz Ciscomani

The resistance to change in languages is a subject that has been insufficiently researched. The bibliography on the historical development of languages has focused up to now on the change experienced by languages, so that the state of knowledge about those structures that show a different diachronic behaviour is limited to hardly marginal notations (Hock 1986: 332, 272, 215-216); 2 and the reasons for such different behaviour are similarly neglected. The ditransitive construction in Spanish is, precisely, one of those structures that despite the passing of time has basically been preserved intact, so that it is highly resistant to change. What is the reason for such an overwhelming stability? This is one question the article attempts to answer. Furthermore, we find that despite the ingrained resistance of the construction to change, a certain tendency to monotransitivization does emerge.

2. Exposition Using a corpus comprising 24 texts of American and Peninsular Spanish (see section a in bibliography), written in prose, of different genres, all of which have been fully analysed, the study encompasses eight centuries of the history of Spanish, from the 13th to the 20th centuries, to characterize the behaviour of Spanish ditransitive sentences, that is, the syntactic-semantic behaviour of concurrent direct and indirect objects and their relationships with the verb. The analysis shows that two marking schemes for objects of the ditransitive construction have existed throughout this period with very different rates of frequency of usage. A first scheme with the indirect object marked by the preposition a and an unmarked direct object, shows a very high diachronic frequency:^

(l)a.

Como el rey Don Juan I envio sus cartas a as the King PN sent his letters to Mosen Aymon ά le decir algunas rezones PN to him-IO to-tell some explanations (Juan I, 76) [14th centuiy] 'As the King Don Juan I sent his letters to Mosen Aymon to give him some explanations'

Non-diachrony of ditransitivity

b.

263

4

jQue no le, vayan α dar un cheque that no her-DAT they-go to to-give a check a la portera, porque si no el senor va to the janitor because if not the man go a llamar a la policial (Suerte, 122) [20th century) to to-call to the police 'Do not give a check to the janitor, because otherwise the man will call the police'

A second scheme, with both objects marked with the preposition a, shows a very low frequency, which tends to decrease slightly during the history of this language: (2) a. Enviamos ά ellos al honrado padre Doctor en (we) sent to them to the honoured father Doctor in Decretos nuestro consejero el Obispo de Zamora Decrees our advisor the Bishop of Zamora 'We sent to them the honoured father Doctor in Decrees, our advisor, the Bishop of Zamora' (Juan I, 54) [14th century] b.

la escena bordada en el tapiz conmemoraba the scene embroidered on the tapestry commemorated la occasionen la que el Rey Sol presento a theoccasion in which the King Sun introduced to su hijo α los Grandes de Espaha (Noticias, 207) [20th century] his son to the Grandees of Spain 'the scene embroidered on the tapestry commemorated the occasion on which the Sun King introduced his son to the Spanish Grandees'

The total of ditransitive constructions identified and analysed in the corpus amounts to 3,061 sentences. Of these, only 34 (1%) admit the "atypical" marking scheme, as can be seen in Table 1; a similar ratio of both marking schemes, but with the atypical one decreasing, was documented by century, according to the data of Table 2, that also registers the only 10 verbs that have accepted this scheme in 800 years of history of the Spanish language. Table 1. Total of typical and atypical constructions in the corpus Typical marking scheme 99% (3027/3061)

Atypical marking scheme 1% (34/3061)

264

Rosa Maria Ortiz Ciscomani

Table 2. Atypical constructions and their verbs by century Verb

13th

14th

Dar 'to give' Enviar 'to send' Encomendar 'to entrust' Toller 'to take away' Echar 'to cast, to throw' Llevar 'to take' Entregar 'to deliver' Mandar 'to send' Mostrar 'to show' Presentar 'to present, to introduce' Total 34/3061

2

1

2

5

15th

16th

17th

18th

1 4

19 th 20th 1

4

1 1

1

2

1

1 1

1

1 1 1 1 1

7

6

5

6

1

2

3

4

These initial data already show that the ditransitive construction has remained practically unchanged in the history of Spanish. Consequently, we are facing a core structure of grammar that is highly resistant to change. Why this stability of the construction? Why, despite its very low frequency, is the construction with an atypical marking scheme maintained historically? This phenomenon seems to be the consequence of the interaction of multiple factors among which the most outstanding are the lexical, semantic and pragmatic properties of objects that have the particularity of co-occurring in this construction. The aim of this paper is to characterize ditransitive construction with both marking schemes in Spanish, as well as to provide the motivation of the stability, or absence of change, in the construction. It illustrates the

Non-diachrony of ditransitivity

265

grammatical and lexical-semantic properties of concurrent objects in the ditransitive structure, as well as the verbs that admit one or the other marking scheme. Therefore, to characterize the construction, several analysis parameters were adopted, of which in this paper I emphasize the lexical and grammatical nature of objects and verbal semantics. Furthermore, I present some notes on the relative order of objects. I assume that form and function are not dissociated, but that there is an isomorphic relation between them (Kirsner 1985; Givon 1985). In fact, the analysis shows the close relationship between markings of concurrent objects and the relevance of the objects' lexical value, as well as that of the individuation thereof. As to the marking system, indirect objects that are permanently marked resist yielding the marking to concurrent direct objects that are generally unmarked, a fact that contrasts with the diachronic extension of the marking in direct objects of transitive sentences (Company, in press). This behaviour of objects with respect to marking is the index of the prominence of the indirect object in Spanish construction. The analysis reveals that the lexical parameter is of great importance. The concurrent indirect object in the ditransitive construction generally refers to a human entity; but, as I will show below, with the passing of time, it makes its inherent human character more flexible and gradually admits inanimate nouns as the nucleus, thus coming to resemble the direct object that is lexically more flexible. Individuation is another relevant factor, together with the lexical one. I will show that the indirect object, while admitting a non-personal nucleus, loses individuation. In parallel fashion, the direct object argument historically increases its indetermination, which contributes to the loss of its semantic independence, to such a degree that it seems to be involved in an incorporation process to the verb. Flexibilization of the indirect object, together with the incorporation of the direct object to the verb, are features that show a slight historical increase and that have their repercussions on the monotransitivization of the construction. The paper is organized as follows: First, in Sections 2 and 3,1 present the theoretical exposition that underlies the analysis of the data. Next, in Sections 4 and 5, I present the general analysis and the chief findings in the research. In Section 6, I show the most relevant features of sentences that share the scheme of the typical marking in the eight centuries covered by the study. Here I emphasize that the human indirect object

266

Rosa Maria Ortiz Ciscomani

slightly decreases from the 16th to the 20th centuries, while at the same time, the inanimate indirect object increases its frequency in a growing proportion. Furthermore, I emphasize the lexical and grammatical flexibility of the direct object, features that facilitate the decrease of the construction's notional ditransitivity. Thereafter, in Section 7, I characterize the atypical construction with both marked objects. In the exposition of data, I pinpoint the almost null historical presence of the direct object with a preposition, which, nonetheless, is maintained due to its high pragmatic information. Hence, in Section 8, the analysis is geared to evidence, supported by quantitative data, on the ditransitive construction's resistance to change. In Section 9,1 show some data on the order of objects and lastly, in Section 10, I state my conclusions.

3. What is a ditransitive construction? I have already stated that transitivity has been a classic subject in grammatical studies. On the other hand, ditransitivity has not been dealt with as systematically in languages. Especially in the last five years, interesting papers have begun to appear, dealing with this structure, but not as the centre of the exposition, rather as the ambit that makes it possible to deal with the behaviour of the indirect object in languages, or to argue in favour of one of the objects as the true object of construction, matters which are, moreover, limited to constructions with model verbs, such as is the case of studies on the verb dar 'to give'. 5 In this paper, I assume (Hopper and Thompson 1980) that language is usage and, therefore, in the corpus, I include all the ditransitive sentences with the two concurrent objects - direct and indirect - that are manifested by a full NP, that is, constructions with both objects strictly required by the verb or subcategorized and those with objects not strictly required by the verb. At first sight, all the sentences identified in the corpus share structural or formal configuration, in the sense that all of them have the same immediate constituents: a subject-NP + verb + direct object-NP + indirect objectPP. However, if we look at the semantics of the sentences, they have slightly different meanings. In my exposition, ditransitivity is a "construction" (Goldberg 1989, 1992 and 1995), that is, a basic pattern of correspondence that matches a form

Non-diachrony of ditransitivity

267

with different, but related meanings. The basic pattern assumes the form of SUB-V-DO-IO, which is associated to a central meaning whose value is causative and of transference: NP1 does something to NP2, so that NP3 manifests a change. In short, the resemblance or similarity in coding is associated with semantic diversity of such a kind that the subtlety of different meaning can only be accounted for by a functional approach.6 Furthermore, the exposition also implies that the meaning of the construction is the result of integrating the meanings of its lexical items to the meaning - caused transference - of the basic pattern, that exists independently of uses. This basic pattern instances itself in a prototypical construction with the verb dar 'to give' whose lexical value coincides with that of the construction, favouring its status as prototype. A series of constructions that share the same syntactic pattern, but that semantically distance themselves from the dar prototype, are related to it to a certain degree, originating a case of constructional polysemy. Thus, the approach is of prototypes and considers that belonging to a determined category is a matter of degree. Also, the analysis of data uncovers the relevance - for the gradation of ditransitivity - of lexical properties and individuation of nominal nuclei of combined objects. In each of the centuries covered by the study, this basic pattern underlies structures whose semantic differences result from the properties of the nuclei of combined objects. The same verb, dar, for instance, assumes different meanings according to the lexical character of the objects with which it combines. On the other hand, not all verbs have the capacity of admitting both combinations.

4. The starting point: documented combinations of lexical objects In the eight centuries of history of the Spanish language covered by the corpus, 4 different combinations of lexical objects were documented: 4.1. INANIMATE DO (concrete) 4.2. INANIMATE DO (abstract) 4.3. INANIMATE DO 4.4. ANIMATE DO (person)

+ + + +

ANIMATE 10 (person) ANIMATE 10 INANIMATE IO ANIMATE IO (person)

These four combinations constitute the starting point of my analysis. Two parameters were key in that analysis: 1) the typical/atypical character

268

Rosa Maria Ortiz Ciscomani

of object marking and 2) the lexical semantic properties of the nucleus of those objects. Although the four combinations share the same syntactic coding scheme, SUB-V-DO-IO, the nuclei of the objects in combinations 4.1 and 4.2 (see examples 3 and 4) are lexically asymmetric - non-personal (thing, or an entity conceptualized as a thing) direct object, combined with a personal indirect object - , whereas combinations 4.3 and 4.4 involve lexically symmetric objects (examples 5 and 6): both inanimate/both personal, respectively. Combination 4.1: thing DO + 10 person (3) a. Como el rey Don Juan I envio sus cartas a as the King PN sent his letters to Mosen Avmon ά le decir algunas rezones PN to him-IO to-tell some explanations (Juan I, 76) [14th century] 'As the King Don Juan I sent his letters to Mosen Aymon to give him some explanations' b. iQue no lej vayan α dar un cheque that no her-DAT they-go to to-give a check a la porteroj porque si no el senor va to the janitor because if not the man go th a llamar a la policial (Suerte, 122) [20 century) to to-call to the police 'Do not give a check to the janitor, because otherwise the man will call the police' Combination 4.2: abstract DO + animate, person 10 (4) a. Medea, sennora ,la ventura dio a ti el poder Medea, lady, the fortune gave to you the power e el alvedrio de la nuestra salud (GEH 84) [13th ] and the free will of the our health 'Medea, lady, fortune gave you the power and free will over our health' b. El Arzobispo de Santiago tomo juramento al rev the Archbishop of Santiago took oath to the King de Portoeal έ a todos los erandes de su Resno of Portugal and to all the grandees of his Kingdom

Non-diachrony of ditransitivity

269

sobre el cuerpo de Dios en el altar (Juan, 8) [14th ] on the body of God on the altar 'The Archbishop of Santiago took the oath of the King of Portugal and of all the grandees of his Kingdom at the altar, before the body of God.' Combination 4.3 is a construction in which, although the SUB-V-DO-IO coding scheme is maintained, due to combining lexically similar objects the causative meaning fades to a certain degree, until it is lost in certain cases (5). Combination 4.3: inanimate DO + inanimate IO (5) a. La muerte del padre puso termino a sus the death of the father put end to his th estudios cläsicos (Obras IV, 30) [19 ] studies classical 'The death of his father put an end to his classical studies' b. Puso treguas al coraje con que se put truces to the anger with which them-REFL hallaban todos (Trofeo, 123) [17th ] were all 'Put an end to the anger they all felt' c. mientras el chulo retrocedia dando traspies, yo while the pimp backed up giving stumbles I le, eche mano a la navaidi (Cachito, 110) [20th] it-DAT put hand to the switchblade knife 'while the pimp backed up stumbling, I got hold of the switchblade knife' The objects of this combination, besides being lexically symmetrical since both are inanimate, exhibit specific grammatical properties that reflect the absence of individuation of the direct object, an aspect which I will talk about in more detail below. Combination 4.4 with both objects manifested as animate, a person, is the only one that admits the marking in both objects. It is a scarcely documented combination in the data which, it seems, the language tries to avoid, but in the rare cases in which it appears, that is, when the semantic valuation coincides in both objects -both are persons, specific, prominent- the marking also coincides in both. In this combination, the direct object is always a

270

Rosa Maria Ortiz Ciscomani

highly individuated person and it is always marked, that is, it has very similar properties to that of the indirect object. See examples in (6). Combination 4.4: animate, person DO + animate, person 10 (6) a. Enviamos ά ellos al honrado padre Doctor en (we) sent to them to the honoured father Doctor in Decretos nuestro consejero el Obispo de Zamora Decrees our advisor the Bishop of Zamora 'We sent to them the honoured father Doctor in Decrees, our advisor, the Bishop of Zamora' (Juan I, 54) [14th century] b.

la escena bordada en el tapiz conmemoraba la thescene embroideredon the tapestry commemorated the occasion en la que el Rey Sol presento a su hijo occasion in which the King Sun introduced to his son α los Grandes de Espana (Noticias, 207) [20th century] to the Grandees of Spain 'the scene embroidered on the tapestry commemorated the occasion on which the Sun King introduced his son to the Spanish Grandees'

The aforementioned properties, lexical value and individuation of objects, have repercussions especially on the construction's degree of ditransitivity. The typical marking scheme, that is, DO without prepositional marking and IO with prepositional marking, is that of combinations 4.1, 4.2 and 4.3. Of the identified combinations, the most typical is combination 4.1, since the ditransitive construction, as well as the transitive one itself, is characterized by the asymmetry of its participants. In terms of individuation, the two objects of this combination are highly individuated; in lexical terms, they are opposites (Hopper and Thompson 1980). Which of these combinations admits the atypical marking scheme? Only combination 4.4 with both objects manifested as animate entities, persons, that is, in cases in which the objects lose their asymmetry to such a degree that they identify themselves. At the same time, very special semantics are involved with the presence of the marking in both objects. A group of semantic and pragmatic conditions proper to the indirect object, such as individuation, animateness, specificity and prominence, extend to the direct object together with the marking.

Non-diachrony of ditransitivity

271

5. Constant features of ditransitivity I have already stated that lexical and individuation factors of objects have repercussions on the construction's degree of ditransitivity. Which factors are responsible for the non-diachrony of the ditransitive construction? The analysis of the data has uncovered that there are several and that they are intimately related: 1) lexical properties of objects: animateness/inanimateness, concreteness/abstraction 2) individuation of objects that subsume several different, but inter-related features, such as number, determination, definiteness and expansion of objects 3) grammatical properties: proper noun, common noun, pronoun 4) verbal semantics In this paper, I shall focus on lexical properties (1) and some factors that control the individuation of objects (2), constant characteristics of importance in the motivation of non-diachrony of the construction, as well as verbal semantics. What is the lexical value proper to direct objects and concurrent indirect objects in the ditransitive construction? In the development of the foregoing section, I anticipated that the direct object of the ditransitive construction is characterized by a certain lexical flexibility, that is, it admits inanimate as well as animate nouns as its nucleus. In fact, the analysis of the data shows that this direct object has been predominantly inanimate and, as can be observed in the data of table 3, it has shown a historical trend that favours abstract nouns (in bold in table 3), as opposed to concrete nouns, but it has also admitted animate nouns as its nucleus. Table 3 also shows that the ratio of animate nouns in the direct object has decreased historically (data in bold in the animate direct object column). This fact is a consequence of the asymmetry of the transitivity relations, of the language's resistance to incorporating two participants (three participants in total in ditransitive construction) with identical lexical characteristics, into the object positions. Despite the foregoing, the direct object of ditransitive constructions has admitted as nucleus nouns with the same lexical properties as those of the indirect object, that is, person, but in a very low proportion. Having experienced highs and lows during the first four centuries, this has decreased very

272

Rosa Maria Ortiz Ciscomani

noticeably in the last four centuries, as can be seen in Table 4 (data highlighted in bold). Table 3. % Lexical characteristics of inanimate (concrete/abstract) vs. animate (person and non person) DO Concrete DO

Abstract DO

Animate DO

13th

26% (81/316)

57% (181/316)

17% (54/316)

14th

25% (29/115)

30% (34/115)

45% (52/115)

15th

14% (33/239)

71% (169/239)

15% (37/239)

16th

26% (149/567)

61% (345/567)

13% ( 73/567)

17th

34% (284/249)

63% (158/249)

3%(

18th

27% ( 32/120)

67% ( 80/120)

6% ( 8/120)

19th

27% (104/381)

68% (260/381)

5%( 17/381)

20th

36% (101/282)

59% (166/282)

5% ( 15/282)

7/249)

Table 4. % person DO 13th

11% (34/316)

14th

36% (41/115)

15th

9% (22/239)

16th

10% (57/567)

17th

2% ( 4/249)

18th

4% ( 5/120)

19th

2% ( 9/381)

20th

3 % ( 8/282)

The indirect object, in contrast to the direct object, has been characterized by a certain lexical rigidity that has decreased diachronically. It has

Non-diachrony of ditransitivity

273

tended to be predominantly animate, person, over the eight centuries (See Table 5). The ratio of animate indirect objects, which was very strong from the 13th to the 16th centuries, has decreased in the last four centuries, while, at the same time, the inanimate indirect object has grown in a non-negligible proportion. The degree of individuation of objects is relevant for ditransitivity, as is the lexical class. The less the individuation of the object, the less the construction's ditransitivity. The analysis of the ditransitive construction's objects underscores that, in fact, these refer to entities that are highly differentiated among themselves, with respect to the subject entity and their own surrounding, that is, they tend to be highly individuated (Hopper & Thompson 1980: 253). Individuation assumes several different but related features. A concrete singular noun that can be counted is more individuated that an abstract plural noun that cannot be counted. The less the degree of individuation, the less the (di)transitivity. One of the categories that reveal individuation of entities in syntax is that of number. As can be seen in Table 6, the analysis shows that number is not a parameter that by itself, contributes to differentiate objects. The direct object as well as the indirect object show a trend for the singular, although the ratio of singular indirect object is just a bit higher than that of the singular direct object. The determination and definiteness of the objects are the actual relevant factors, as can be proven through the data in tables 7 and 8, which explain that throughout the eight centuries, indirect objects are more determined (Table 7) and defined (Table 8) than direct objects (see data of greater determination of the 10 as compared to the DO, highlighted in bold). The results of the analysis of the data about individuation, concreteness, singularity, high determination and definiteness of objects, as well as data about lexical properties, show that these parameters are relevant for the characterization of ditransitivity. The DO is shown to be a more flexible argument as far as individuation of the indirect object, which is highly determined and defined, is concerned, and as an inanimate argument, as opposed to the indirect object, predominantly a person. These features are essentially maintained throughout the eight centuries covered by the study.

274

Rosa Maria Ortiz Ciscomani

Table 5. % Indirect animate and inanimate objects Animate

Inanimate

13th

87% (276/316)

13% (40/316)

14th

98% (113/115)

2%

15th

96% (230/239)

4% ( 9/239)

16th

96% (546/567)

4%

17th

83% (207/249)

17% (42/249)

18th

75% ( 90/120)

25% (30/120)

19th

82% (314/381)

18% (67/381)

20th

77% (216/282)

23% (66/282)

(2/115)

(21/567)

Table 6. % Number in objects by century Singular

Plural

DO

10

DO

IO

13th

72 (266/369)

28(103/369)

18(73/369)

14th

46 (69/150)

54 (81/150)

21 (31/150)

15th

77 (269/349)

23 (80/349)

40(141/349)

16th

60(431/724)

40 (293/724)

26(186/724)

17th

77 (268/348)

80 (296/369) 79 (119/150) 60 (208/349) 74 (538/724) 75 (261/34)

23 ((80/348)

25 (87/348)

18th

65(113/173)

35 (60/173)

28 (48/173)

19th

74 (421/567)

26(146/567)

21 (120/567)

20th

73 (277/381)

27(104/381)

20 (75/381)

72 (125/173) 79 (447/567) 80 (306/381)

Non-diachrony of ditransitivity

275

Table 7. % Determiner in objects by century +determiner

-determiner

DO

IO

DO

IO

13™

83(199/241)

96(220/230)

17(42/241)

4(10/230)

14™

68(70/103)

91(105/116)

79(81/103)

9(45/116)

15™

68(130/191)

88(265/302)

32(61/191)

12(37/302)

16™

49(207/423)

84(369/440)

51(216/423)

16(71/440)

17™

66(145/221)

86(212/247)

34(76/221)

14(35/247)

18™

68(82/121)

93(131/141)

322(39/121)

7(10/141)

19™

58(213/370)

90(375/416)

42(157/370)

10(41/416)

20™

60(171/286)

93(243/261)

40(115/286)

7(18/261)

6. The typical scheme of object marking The data I now present refer to finite ditransitive constructions of the corpus, that is, only constructions with a conjugated verb. These represent 74% (2269/3061), or approximately three quarters of the entire corpus. For the eight centuries under study, the corpus documents four lexical combinations of objects in ditransitive constructions (see section 4) of which only the first three (4.1, 4.2 and 4.3) are documented with the typical marking scheme of objects, that is, with the DO without a marking and the 10 with the prepositional marking a\ a.

4.1 with a concrete inanimate (thing) DO combined with an person 10, that is, the typical ditransitive b. 4.2 with abstract entity DO combined with person IO, and c. 4.3 with both DO and IO inanimate. These three combinations documented with the typical marking scheme, that is, direct object without a marking and indirect object with the prepositional marking a, have a common characteristic: they maintain a certain but decreasing degree of asymmetry among themselves.

276

Rosa Maria Ortiz Ciscomani

Table 8. % Definiteness in objects (common nouns) by century +definition DO

IO

13th

54(199/369)

96(220/230)

14th

47 (70/150)

70(105/150)

15th

37(130/349)

88(265/302)

16th

29(207/724)

84(369/440)

17th

42(145/346)

86(213/249)

18th

47(82/173)

93(131/141)

19th

38(213/567)

90(375/416)

20th

45(171/381)

93(243/261)

Taking into account that concurrent objects of combination 4.1 are those that possess the asymmetrical characteristics of ditransitivity, that is, an inanimate direct object, flexible as far as individuation is concerned, combined with an animate indirect object, predominantly a person, which is highly individuated, it can be asserted that combination 4.1 is profiled as the 'most ditransitive' - the most prototypical one (Company 1997; Givon 1984a, among others). I will analyse this combination of objects below.

6.1. Combination scheme thing DO + person IO It is worth pointing out that 22% of the finite constructions covered by the study have combination 4.1 with a concrete entity, a thing, as DO and a person as IO. Table 9 registers the total constructions with this combination, by century:

Non-diachrony of ditransitivity

111

Table 9. % Construction with thing DO/ person 10 13th

16% (50/316)

14th

26% (30/115)

15th

13% (32/239)

16th

24% (137/567)

17th

24% (59/249)

18th

23% (28/120)

19th

21% (81/381)

20th

32% (89/282)

Total

22% (506/2269)

In this lexical combination, objects show their most typical features: the direct object is lexically inanimate, a thing (but relatively flexible in the sense that it admits concrete, as well as abstract and animate nouns, even persons, all of them with a different degree of individuation as Table 3 shows). Formally, direct objects always appear unmarked. In contrast, the indirect object is predominantly a person, therefore, animate, and in this sense, it is lexically less flexible than the direct object (see Table 5). Furthermore, it is formally rigid, since it never loses the prepositional marking a, and as I show in the data I present below, it is thematically flexible.

6.2. Verbs in the typical marking scheme The prototypical ditransitive construction, as I already stated, involves a successful act of transfer that has already been achieved and completed (from the aspect point of view), in which a concrete entity is transferred to an animate entity, in an event in which the causing subject and the indirect object enter into contact. This ditransitive construction is instanced in Spanish by the verb dar 'to give', whose lexical value coincides with the construction's meaning (7):

278

Rosa Maria Ortiz Ciscomani

(7) a. e dio el vaso al rev que gelo and gave the glass to the king that him-IO-it-DO diese e/(GEII 87) [13th] had given he 'and gave the king the glass so that he could give it to him' b. Ε por le facer merced, dio a su musser and for her-IO to make pleasure, gave to his wife la ropa quella vistiese todos los aiios de the clothes that she-would wear all the years of th su vida el dia de reyes (CRCII 94.20) [15 ] her life the day of kings 'And to please her, he gave his wife the clothes she would wear every year of her life on the Day of the Holy Three Kings' c. El senorles habia dicho en efecto que Luis the man them-IO had told in fact that PN Mesa y yo lo ibamos a estropear todo Mesa and I it-ACC we-went to to-spoil everything si lei ddbamos el cheque a la if her-DAT gave the check to the Oortera, (Suerte, 122) [20th] janitor 'The man had told them, in fact, that Luis Mesa and I were going to spoil everything if we gave the check to the janitor' In this construction, the subject is a volitional agent, an entity that acts consciously to cause or provoke a change in the indirect object. The direct object is a theme in the sense that it does not undergo an observable physical change, but rather experiences a change of place; it is moved from the subject agent's context into a new one, that of the 10 entity that receives it. This indirect object is a recipient (not a possessor). It is a conscious entity that receives the indirect object in its environment. In this sense, a prototypical ditransitive construction involves a completed act of transfer. There are many lexical verbal entries with this combination during the eight centuries under study: 97 in total. Among them, those with the greatest historical recurrence (documented for six or more centuries), besides dar 'to give', are entregar 'to deliver' (8) and enviar 'to send' (9): (8) a. Cecilia abriö Cecilia opened

su ropero y entrego a la criada her closet and delivered to the maid

Non-diachrony of ditransitivity

279

un canasta Ueno de raices, de yerbas seca y de a basket full of roots, of herbs dry and of pedacitos de palo de diverses tamanos little pieces of wood of different sizes y colores (Bandidos, 224) [ 19th]. and colours 'Cecilia opened her closet and gave to the maid a basket full of roots, dry herbs and little pieces of wood of different sizes and colours' b. Herr professor sirvio dos copas. Camino hasta Herr Professor poured two glasses, he-walked to la mesa y le, entrego una de ellas the table and him-DAT delivered one of them al Archiduauei (Noticias, 95) [20th] to the Archduke 'Herr Professor poured two glasses. He walked to the table and gave the Archduke one of them.' (9) a. luego enviaron sus cartas e mensageros a todos then they-sent their letters and messengers to all los cavalleros de las cibdades ν villas de Andaluzia the gentlemen of the cities and villages of Andalusia 'then they sent their letters and messengers to all the gentlemen in the cities and villages of Andalusia' (CRCII 13.22) [15th] b. e dende envio sus cartas al Conde Don Alfonso, and from-there he-sent his letters to the PN, su hermano, que estava en Bresanza trayendo sus his brother, who was in Breganza bringing his pleytesias con el rev don Ferrando de Portoeal homage with the king don Fernando of Portugal (Juan th I, 76) [14 ] 'and from there, he sent his letters to Count Don Alfonso, his brother, who was in Breganza paying homage to the king Don Fernando of Portugal' Other recurrent verbs documented for five centuries were: pedir 'to ask', escribir 'to write', hacer 'to make'; while besar 'to kiss', cortar 'to cut', dejar 'to leave', demandar 'to demand', ensenar 'to teach, to show', poner 'to put', pagar 'to pay', repartir 'to distribute', were documented for four centuries; and llevar 'to take', mostrar 'to show', ofrecer 'to offer', presentar 'to present, to introduce', quebrar 'to break', quitar 'to take away', tirar

280

Rosa Maria Ortiz Ciscomani

'to throw, to pull', tornar 'to turn', traer 'to bring', ver 'to see' were registered only in three centuries (table 10): Table 10. Recurrent verbs in typical marking scheme (3 or more centuries) 13 th

Pedir 'to ask' Escribir 'to write' Hacer 'to make' Besar 'to kiss' Cortar 'to cut' Dejar 'to leave' Demandar 'to demand' Ensenar 'to teach, to show' Poner 'to put' Pagar 'to pay' Repartir 'to distribute' Llevar 'to take' Mostrar 'to show' Ofrecer 'to offer' Presentar 'to present, to introduce' Quebrar 'to break' Quitar 'to take away' Tirar 'to throw' Tornar 'to turn' Traer 'to bring' Ver 'to see'

14 th

15 til

16 th

•

17 th V»

•

18 th