Secondary Predication and Adverbial Modification: The Typology of Depictives

Secondary Predication and Adverbial Modification This page intentionally left blank Secondary Predication and Adver...

Author: Nikolaus P. Himmelmann | Eva F. Schultze-Berndt

97 downloads 1467 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Secondary Predication and Adverbial Modification

This page intentionally left blank

Secondary Predication and Adverbial Modification The Typology of Depictives

Edited by NIKOLAUS P. HIMMELMANN AND EVA SCHULTZE-BERNDT

AC

AC

Great Clarendon Street, Oxford ox2 6dp Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York #

editorial matter and organization Nikolaus P. Himmelmann and Eva Schultze-Berndt 2005 # the chapters their authors 2005 The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published 2005 First published in paperback 2006 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose the same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Secondary predication and adverbial modification : the typology of depictives / edited by Nikolaus P. Himmelmann and Eva Schultze-Berndt. p. cm. Summary: ‘‘This book approaches depictive secondary prediction from a crosslinguistic perspective. It maps out all the phenomena and brings together critical surveys and new contributions on their morphosyntactic and semantic properties’’---Provided by publisher. Includes bibliographical references and index. 1. Grammar, Comparative and general---Syntax. 2. Grammar, Comparative and general--Verb phrase. 3. Grammar, Comparative and general---Adverbials. 4. Semantics. 5. Typology (Linguistics) I. Himmelmann, Nikolaus. II. Schultze-Berndt, Eva. P295.S413 2005 415---dc22 2004024152 Typeset by Newgen Imaging Systems (P) Ltd., Chennai, India Printed in Great Britain on acid-free paper by Biddles Ltd, King’s Lynn ISBN 0--19--927226--3 978--0--19--927226--6 ISBN 0--19--920434--9 (Pbk.) 978--0--19--920434--2 (Pbk.) 10 9 8 7 6 5 4 3 2 1

Contents Preface and chapter summaries Acknowledgements The contributors Abbreviations and glossing conventions 1 Issues in the syntax and semantics of participant-oriented adjuncts: an introduction Nikolaus P. Himmelmann and Eva Schultze-Berndt 2 Depictives in English and Warlpiri Jane Simpson 3 Adverbials and depictives as restrictors Thomas Mu¨ller-Bardey 4 Depictive agreement and the development of a depictive marker in Swiss German dialects Claudia Bucheli Berger 5 Quantifying depictive secondary predicates in Australian languages William B. McGregor 6 Depictives in Kartvelian Winfried Boeder 7 On depictive secondary predicates in Laz Silvia Kutscher and N. Sevim Genç 8 Participant agreement in Panoan Pilar M. Valenzuela 9 Secondary predicates and adverbials in Nilotic and Omotic: a typological comparison Azeb Amha and Gerrit J. Dimmendaal 10 Asyndetic subordination and deverbal depictive expressions in Shona Tom Gu¨ldemann 11 Forms of secondary predication in serializing languages: on depictives in Ewe Felix K. Ameka

vii xv xvi xx

1 69 107

141

173 201 237 259

299

323

355

vi

Contents

12 Depictive and other secondary predication in Lao Nicholas J. Enfield

379

13 A semantic map for depictive adjectivals Johan van der Auwera and Andrej Malchukov

393

References Index of languages Index of terms

423 441 443

Preface and Chapter Summaries This book is concerned with depictive secondary predicates such as fresh in George bought the carrots fresh and related constructions in crosslinguistic perspective. These constructions have special semantic and syntactic properties and therefore have played a major role in modern theories of the syntax and semantics of natural languages. This book provides for new perspectives on these constructions in two ways. First, it considerably expands the database for them by showing that, in addition to the well-studied adjectival depictives such as fresh, a wide range of other expression types warrant an analysis as depictives, including, for example, some types of subordinate clause. This becomes apparent when considering data from a much wider range of languages than those which usually feature in the literature on this topic, especially since many of these languages have specific morphosyntactic means of marking depictives. Second, the book brings together different research strands which have tended largely to ignore each other because they diverge considerably both in terms of topic area (syntax vs. semantics) and in terms of their basic theoretical persuasions (cognitive linguistics, typology, lexical-functional grammar, neo-Davidsonian semantics, etc.). The contributions thus point to the need for a new, considerably richer and more complex systematics for the grammatical domain of participant-oriented adjunct constructions. Participant-oriented adjunct constructions differ from other types of adjunct construction (most importantly event-oriented adjunct constructions) in that they primarily convey additional information regarding one of the participants in the eventuality expressed by the main predicate (fresh in the example above tells us primarily something about the carrots and not about George or the manner of buying carrots). Although secondary predicates are perhaps prototypical examples of participant-oriented adjuncts, it has been noted that other types of adjuncts may also show participant orientation, and therefore exhibit a strong semantic affinity to depictives. These include, for example, certain subject-oriented adverbials such as enthusiastically in John opened the door enthusiastically. Moreover, across languages we frequently find that the same construction may render both participant- and event-oriented content, even beyond those adverbials that are typically regarded as subject-oriented. For this reason, the present volume also explores commonalities and differences between adverbials and depictives, based on the hypothesis that depictive and adverbial constructions cover a common semantic space. With regard to the more ambitious goal of presenting a comprehensive typology of the participant-oriented adjunct constructions attested in the world’s

viii

Preface and chapter summaries

languages, this book is clearly premature, in that very little is in fact known about depictive secondary predicates and related constructions in most of the world’s languages. That is, this is not a topic which is regularly studied in work on lesserknown languages, and there are only few grammars of such languages where the topic is mentioned at all (and none where it is extensively dealt with). This deplorable state of affairs is not going to change as long as adjuncts in general, and participant-oriented adjuncts in particular, do not make their way into the regular work package of fieldworkers and descriptive linguists (along with standard topics such as grammatical relations and relative clauses). Thus, this book also has the purpose of contributing to this expansion of the standard set of descriptive topics by offering ideas and research strategies for tackling the language-specific intricacies of depictive secondary predicates and other participant-oriented adjuncts. Regarding the genesis of the volume, it may be useful to take note of the following. With the exception of the chapter by Valenzuela, the chapters appearing in this collection were first prepared for the Workshop on Depictives in Crosslinguistic Perspective at the Ruhr-Universita¨t Bochum (7–9 June 2001), which was organized by the editors. All of these contributions were based on a position paper by the editors. The contributors considerably revised their papers after the workshop, taking into account feedback during the workshop as well as further comments by the editors. The introductory chapter was drafted after the workshop, when editing the individual chapters. Although it includes some aspects of the original position paper, it offers a different approach to the whole subject matter, reflecting a number of changes in our thinking which evolved during and in the aftermath of the workshop. It also contextualizes the other contributions with regard to a set of general themes. (Major parts of the original position paper have since—with considerable delay not attributable to the authors—appeared in Linguistic Typology, as Schultze-Berndt and Himmelmann 2004.) The current ordering of the chapters is motivated by the breadth of their approach, their theoretical orientation and the languages they deal with. Each chapter exhibits particularly close ties with the one preceding and the one following it. The chapters by Simpson and Mu¨ller-Bardey complement and expand major themes of the introductory chapter, which immediately precedes them. Simpson provides further details regarding syntactic issues in the analysis of depictives, and contrasts a language with clearly identifiable depictives (English) with a language where depictives appear to convey a broader range of meanings and functions (Warlpiri). Mu¨ller-Bardey offers a more detailed discussion of some of the core semantic issues, focusing on the distinction between depictives and manner adverbs, based on English and German data. The following three chapters share a focus on depictive constructions marked by agreement and other unambiguous markers of depictive status.


ix

Bucheli Berger complements the discussion of English and German in Simpson and Mu¨ller-Bardey by providing data from non-standard varieties of German which exhibit interesting similarities to the Australian data discussed in the editors’ introduction, as well as in Simpson’s and McGregor’s contributions. McGregor elaborates on some of the intricacies involved in the interpretation of the Australian data, focusing on a subtype of depictives which to date has been largely ignored in the literature (quantifying depictives). While still focusing on agreement marking of depictives, Boeder’s chapter on Georgian is the first in a series of chapters which take into view a much broader range of expression types than those usually considered in the literature on depictives, and which also discuss problems of delimitation between depictives and adverbials. All but the last of the remaining chapters share this common focus, which in turn is based on the semantic space for participant-oriented adjuncts established in the editors’ introduction. Apart from this common focus these chapters are interconnected by various areal, typological and topical links (see the short chapter summaries below). The final chapter, by van der Auwera and Malchukov, puts the semantic space for participant-oriented adjuncts into a larger perspective by showing various links to neighbouring semantic spaces such as the ones for predicative complements, appositive attributes or pure manner adverbials. This chapter, which presents data from a worldwide sample of languages, heavily draws on the data presented in the preceding chapters.

Chapter summaries Nikolaus P. Himmelmann and Eva Schultze-Berndt, ‘Issues in the syntax and semantics of participant-oriented adjuncts: an introduction’ The editors’ introduction begins with a critical survey of previous classifications of adverbials and secondary predicates. It is argued that there exist significant overlaps between participant-oriented manner adverbials (John angrily read the review), depictive secondary predicates, and so-called weak free adjuncts (Standing on a chair, John can touch the ceiling). Thus, from a crosslinguistic point of view, it is necessary to make use of an overarching category, participantoriented adjunct, which subsumes all of these construction types. Having established this category, the introduction then presents the range of semantic expressions which may show formal correlates of participant-orientation, and the morphosyntactic features typically associated with participant-oriented adjuncts crosslinguistically. Jane Simpson, ‘Depictives in English and Warlpiri’ Simpson compares depictive secondary predicates in English and Warlpiri. In English, there are heavy constraints on depictives in terms of their

x


phrase structure position, their co-occurrence with other adjunct types and resultatives, their controllers, and their semantics, as well as the semantics of the main predicate. Simpson shows that most of these constraints are absent in Warlpiri, and proposes that these differences stem from basic syntactic differences between the two languages. These include the use of word order (English) vs. case marking (Warlpiri) to mark grammatical relations, and the possibility, in Warlpiri, of making use of temporal-aspectual clitics on depictives, which allow for an independent specification of the relation of the state denoted by the secondary predicate to the event denoted by the main predicate. Simpson concludes that in Warlpiri, depictives are true adjuncts, while in English they are comparable to complements. Thomas Mu¨ller-Bardey, ‘Adverbials and depictives as restrictors’ Mu¨ller-Bardey applies analytical concepts originally developed for the analysis of indefinite NPs to the analysis of adverbials and depictives, and as a result proposes a new subclassification of circumstantial adverbials and depictives. Of fundamental importance is the partition of the clause, i.e. the idea that various kinds of operators including frequency adverbs such as often and always divide clauses into two parts, one functioning as the restrictor for the operator, the other as its nuclear scope. Mu¨ller-Bardey claims that various kinds of circumstantial adverbials (including manner, time, and location) do not behave uniformly with regard to restrictor formation, since only some of these adverbials allow readings in which they are part of the restrictor. ‘Pure’ manner adverbials, in particular, are always part of the nuclear scope, and in this regard they differ from depictives, which behave like those adverbials which can be part of the restrictor. Claudia Bucheli Berger, ‘Depictive agreement and the development of a depictive marker in Swiss German dialects’ Bucheli Berger discusses the different morphological markers found on (adjectival) predicative complements and depictive secondary predicates in Swiss German dialects. Three main dialect areas can be distinguished: one where secondary predicates and predicative complements are unmarked and are therefore not formally distinguished from adverbials (thus comparable to Standard German), one where both depictives and predicative complements show obligatory agreement with their controller (thus comparable to Romance and many Australian languages), and a third area where depictives (but not predicative complements) take a marker which is restricted to this construction, a crosslinguistically rare phenomenon. The emergence of this depictive marker is explained in terms of reanalysis of a former agreement marker, and the syntactic and semantic properties of depictive constructions showing the marker are discussed in some detail.


xi

William McGregor, ‘Quantifying depictive secondary predicates in Australian languages’ McGregor investigates participant-oriented quantifiers in Australian languages, which comprise two main expression types: those that indicate the cardinality of a set of participants (e.g. ‘alone’, ‘as a pair’, ‘as a group of three’), and those that indicate iterated involvement (e.g. ‘one by one’, ‘few by few’). The formal manifestation of these expression types is subject to variation: constructions found in Australian languages may involve agreement of a numeral with its controller (‘identical marking’), use of a numeral ‘one’ in the sense of ‘together’, locative marking of the quantifying expression, and other, more specific, lexical and grammatical means. For each construction, the question is addressed of whether or not it is an instance of a dedicated depictive construction, which can be distinguished from NP-internal attributes on the one hand, and from adverbials on the other. Winfried Boeder, ‘Depictives in Kartvelian’ In his survey of depictives in Georgian and Svan, Boeder gives special attention to the overlapping semantic range of depictive and adverbial expressions. Depictives in Georgian show case agreement with their controller, and may convey a condition or state, an intention, a phase or life stage, and quantity. There exist, however, a number of non-agreeing, adverbial constructions involving the socalled adverbialis case or the instrumental, which may also receive a participantoriented interpretation, e.g. in the domains of posture, manner, comparison, function/role, and concomitance. The close semantic relationship between depictive and adverbial adjuncts in Georgian manifests itself in considerable variation in the use of the two construction types, both within and between dialects. It is also shown by the fact that adjuncts of the two types can be coordinated, and that the same interrogative marker can be used with both. Boeder provides an explanation of this overlap between adverbial and depictive constructions in terms of a regular metonymic relationship between the property predicated of a participant and characteristics of the event itself. Silvia Kutscher and N. Sevim Genç, ‘On depictive secondary predicates in Laz’ Kutscher and Genç discuss participant-oriented adjuncts in Ardes¸en Laz, a Kartvelian language spoken in Turkey. Laz, unlike the related language Georgian discussed in the contribution by Boeder, does not exhibit case agreement of depictives with their controllers. Consequently, it is argued that Laz only has general adjunct constructions which may have participant-oriented or event-oriented interpretations, including unmarked adjectives and adverbials, and instrumental or locational adjuncts. One important difference between participant-oriented adjuncts and manner expressions is possibly found in the domain of prosodic realisation, though: the former may constitute an intonation unit of their own, while manner expressions are always integrated into the verb phrase.

xii


Pilar M. Valenzuela, ‘Participant agreement in Panoan’ The Panoan languages of the Amazon region exhibit a very extensive system of participant agreement in that participant orientation is overtly marked on both clausal and phrasal adjuncts, and generally S, A, and O controllers are all formally distinguished. In her contribution, Valenzuela explores the participant orientation system in the Panoan language Shipibo-Konibo. Here, marking of participant orientation occurs on expressions of manner, quantity, location, and life stage, among others, as well as on emphatic pronouns and same-subject clauses. Valenzuela discusses the semantic distinctions conveyed by different types of agreement marking, and accounts for differences in marking patterns on adjuncts of various types by establishing a semantic hierarchy. Azeb Amha and Gerrit J. Dimmendaal, ‘Depictive secondary predicates in Nilotic and Omotic: a typological comparison’ Amha and Dimmendaal discuss various types of participant-oriented adjuncts in languages belonging to two distinct language groups spoken in adjacent areas of (north-)eastern Africa, the Nilotic branch within Nilo-Saharan and the Omotic branch of Afroasiatic, and address the correlation between the formal properties of these adjuncts and other typological properties of the languages in question. In Turkana as well as in other Nilotic languages, participant-oriented adjuncts are predominantly verbal, and formally distinct from manner adverbials, but the authors show that the distribution of expression types across the two constructions may vary, thus providing further evidence for the view that, in a crosslinguistic perspective, adverbial and depictive constructions ‘compete’. In contrast to the Nilotic languages, the Omotic languages Wolaitta and Maale systematically distinguish between same-subject converbs showing agreement with the main clause subject, and free-subject converbs which allow the independent expression of a subject. The authors argue that the former should be analysed as depictives and the latter as adverbial clauses, and that the switchreference function of these converb clauses is parasitic upon the depictive/ adverbial distinction. Tom Gu¨ldemann, ‘Asyndetic subordination and deverbal depictive expressions in Shona’ In Shona as well as other Bantu languages, participant-oriented expressions mainly take the form of converb clauses (or free-subject dependent predicates), which are prosodically marked and which may also function as part of a complex predicate, or as a relative clause. Gu¨ldemann reviews depictive-like uses of clauses of this type and discusses the arguments for and against assigning the status of a (morphosyntactically distinct) depictive construction to them. A major concern in this exploration is the investigation of the role of focus structure


xiii

for the analysis of depictives. In the concluding sections of his chapter, Gu¨ldemann goes well beyond the Shona data and assembles crosslinguistic support for the view that depictives in general represent the assertive focus of a clause. Felix K. Ameka, ‘Forms of secondary predication in serializing languages: on depictives in Ewe’ Ameka shows in his contribution that meanings conveyed by depictives e.g. in European languages are often expressed by means of multiverb constructions (serial verb constructions) in languages like Ewe, a Kwa language spoken in Ghana and Togo. However, different types of participant-oriented adjuncts, including ideophones, prepositional phrases, and noun phrases, also occur in this function, contra Larson’s (1991) hypothesis that serializing languages do not have nominal secondary predicates. Ameka, moreover, presents a new analysis of a ubiquitous enclitic ¼i in Ewe which serves as a marker of secondary predicates and predicative complements, but also as an adverbializing suffix. It also appears on some verb phrases in a series, which raises the intriguing question as to how serial verbs and verbal adjuncts can be distinguished, in particular in an isolating language such as Ewe. Nicholas J. Enfield, ‘Depictive and other secondary predication in Lao’ In Lao, depictive expressions are predominantly verbal. Since Lao, like Ewe, is a typical isolating language, lacking inflectional morphosyntactic categories such as case-marking and agreement, verbs are defined distributionally, and there is no distinction between finite and non-finite verb forms. Enfield suggests, therefore, that the distinction between a primary and a secondary predicate can only be made in the following way: the secondary predicate is an adjunct in the sense that it can be omitted without changing the basic semantics of the primary predication. While Lao clearly has secondary predicates in this sense, it does not have a specific depictive construction involving verbs, since not only depictive but also resultative and adverbial meanings can be conveyed by unmarked verbs functioning as secondary predicates. Enfield also addresses pragmatic restrictions on specific combinations of primary and secondary predicate, related to the relevance and/or conventionality of the combination. Johan van der Auwera and Andrej Malchukov, ‘A semantic map for depictive adjectivals’ Van der Auwera and Malchukov explore the formal relationships that are found crosslinguistically between adjectives in depictive function, and adjectives in other, arguably related functions (where ‘adjective’ is taken in a broad sense to include stative verbs, property nominals, participles, and relative clauses). The results are presented in the form of semantic maps, which are based on the

xiv


assumption that crosslinguistic similarities in terms of formal properties can be taken to reveal proximity in semantic space. The findings show that depictives occupy an intermediate position between predicatives (i.e. adjectives used as a main predicate), predicative complements, appositive structures, and adverbials, and therefore can be said to have semantic affinities with all four of these categories. London April 2004

NPH ESB

Acknowledgements The editors are very grateful to the contributors, who have been not only exceptionally cooperative in sticking to deadlines and tailoring their contributions to the design of the book but also extremely patient, staying on board despite the fact that the review and editing process took much longer than originally expected. The editors also gratefully acknowledge the generous support of the German Research Foundation (Deutsche Forschungsgemeinschaft), which included a research fellowship for Schultze-Berndt from 1999 to 2001 as well as the complete funding of the workshop in June 2001. Many thanks to Judith Ko¨hne and Jan Strunk for help with the proofs and indices. Finally, we thank John Davey at OUP for his encouragement and for seeing this work through.

The Contributors Felix K. Ameka teaches in the Department of African Languages and Cultures, Leiden University, and is a Research Associate in the Language and Cognition Group, Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands. His main interests are in language description, typology, semantics, pragmatics, anthropological linguistics, and ethnography of communication with special reference to West African languages. Currently he is working on the documentation of the language and culture of Likpe (Ghana) and (with James Essegbey) preparing a modern comprehensive grammar of Ewe. He is Editor-in-Chief of the Journal of African Languages and Linguistics. Azeb Amha is a post-doctoral researcher at the Leiden University Centre for Linguistics and teaches in the Department of African Languages and Cultures at Leiden University, The Netherlands. She works on Omotic languages, and is interested in typological/areal comparative studies. She has published The Maale Language (2001) as well as articles in scholarly journals and collections. She is currently writing a grammar of Wolaitta (with Gerrit Dimmendaal) and is documenting Zargulla, an endangered Omotic language. Winfried Boeder is Professor of Linguistics at the University of Oldenburg, Germany. His main interests are language typology and historical linguistics. He works on Kartvelian (South Caucasian) comparative and historical morphosyntax, and on the historical sociolinguistics of Georgian. Claudia Bucheli Berger is a research associate in an empirical project investigating the syntax of Swiss German dialects at the University of Zurich, Switzerland. She just finished her doctoral thesis on serial verbs in French Creoles, focussing on their implications for typology and theories of creolization. Gerrit J. Dimmendaal is Professor of African Linguistics at the University of Cologne, Germany. His main interests are language typology and Nilo-Saharan languages. His more recent publications include Surmic Languages and Cultures (1998, with Marco Last), ‘Morphology’ (2000, in African Linguistics: An Introduction, edited by Heine and Nurse), ‘Number marking and noun

The contributors

xvii

categorization in Nilo-Saharan languages’ (2001, Anthropological Linguistics 42), and ‘Places and people: field sites and informants’ (2001, in Linguistic Fieldwork, edited by Newman and Ratliff). He is currently editing a book on the Central Sudanic family. Nicholas J. Enfield is a scientific staff member in the Language and Cognition Group, Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands. Recent publications include Linguistic Epidemiology: Semantics and Grammar of Language Contact in Mainland Southeast Asia (2003), Ethnosyntax: Explorations in Grammar and Culture (edited volume, 2002), and numerous journal articles on issues in grammar, semantics, semiotics, and interaction. Continuing research is based on regular fieldwork in mainland Southeast Asia, especially Laos. N. Sevim Genç graduated in linguistics at the University of Cologne, Germany, with a thesis on Laz word formation. She is currently committed to helping preserve Laz as a native language. Tom Gu¨ldemann is a Research Associate at the Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany, and teaches African linguistics at Leipzig University. He specializes in Bantu and Khoisan languages and has a major interest in typology, areal and historical linguistics. He is currently the director of two major projects: ‘Focus in South African Languages’ at the ZAS Berlin, and ‘Documentation of Western !Xoõ’, funded by the Volkswagen Foundation. His more recent works include ‘Khoisan’ (2000, in African Linguistics: An Introduction, edited by Heine and Nurse), Quotative Constructions in African Languages: A Synchronic and Diachronic Survey (2001), and ‘Present progressive vis-a`-vis predication focus in Bantu: a verbal category between semantics and pragmatics’ (2003, Studies in Language 27). Nikolaus P. Himmelmann is Professor of Linguistics at the University of Bochum, Germany. His major research interests include typology, grammaticization theory, prosody and grammar, as well as language documentation and description. He is the author of an in-depth study of the typology and grammaticization of demonstratives and articles (Deiktikon, Artikel, Nominalphrase, 1997) and co-editor of The Austronesian Languages of Asia and Madagascar (2005) and Essentials of Language Documentation (2006). He has done fieldwork in the Philippines, Sulawesi and East Timor, and published widely on a number of core issues in Austronesian grammar, including the nature of lexical and syntactic categories and voice.

xviii

The contributors

Silvia Kutscher has written a doctoral thesis on the noun phrase structure of Ardes˛en-Laz, and currently holds a Lise Meitner scholarship at the University of Cologne. Her research interests include lexical semantics, argument structure, and linguistic typology with a focus on Germanic, South-Caucasian, and Finno-Ugric languages. Together with N. Sevim Genç she is involved in documenting the endangered language Laz. William B. McGregor is Professor of Linguistics at Aarhus University in Denmark. His main interests are Aboriginal languages of the Kimberley region of the far north-west of Australia, in particular Gooniyandi, Nyulnyul, Warrwa, and Unggumi. Among other things, he is also interested in language typology, numerals, the origins and evolution of language, and linguistic theory (specifically, the development of an adequate formal-functional theory). He has published grammars of four languages, and is working on descriptions of other. His most recent publications include Verb Classification in Australian Languages (2002, Berlin: Mouton de Gruyter) and The Languages of the Kimberley, Western Australia (2004, London: Routledge Curzon). He has just finished revising Frs. Nekes and Worms’ 1953 monograph Australian languages, which appeared in February 2006, published by Mouton de Gruyter, and has edited a collection of papers on the history of Australian Aboriginal linguistics (submitted to Pacific Linguistics). Andrej Malchukov is a member of the St Petersburg Typology Group (Institute of Linguistic Research, Russian Academy of Sciences). He specializes in Tungusic languages, and his habilitation thesis deals with aspects of the morphosyntax of the North Tungusic language Even. His main interests are in morphosyntactic and semantic typology. Recently finished projects include the typology of attributive relations (Dependency Reversal in Noun-attribute Constructions: Towards a Typology, 2000), the typology of transcategorial operations (Nominalization/verbalization: Constraining a typology of transcategorial operations. LINCOM, 2004), and the typology of adversative coordination. Currently he is working on the typology of case-marking systems within the PIONIER project ‘Case cross-linguistically’ at the University of Nijmegen, The Netherlands. Thomas Mu¨ller-Bardey was a researcher in the department of General and Comparative Linguistics at the University of Mainz, Germany, from 1992 to 2001. His main research interests include switch-reference, converbs and related phenomena both from a typological and a formal semantics point of view. He now lives in Cologne and works in public relations. Eva Schultze-Berndt is Professor of Linguistics at the University of Graz, Austria. Her research interests include typology, grammar of spoken

The contributors

xix

language, lexical semantics, language contact, and language documentation and description. She has published on complex and secondary predication, verb semantics, word classes, and construction-based approaches to grammar, both from a typological perspective and with a focus on the Northern Australian language Jaminjung, based on her own fieldwork. Jane Simpson teaches linguistics at the University of Sydney, Australia. She works on central Australian languages (Warumungu and Warlpiri) and southern Australian languages (Kaurna and Thura Yura languages). She is interested in syntax, lexical semantics, comparative linguistics, child language acquisition, language maintenance, and lexicography. She has co-edited books on Indigenous Australian place names, Australian languages, and Aboriginal history. Her most recent book is A Learners’ Guide to Warumungu (2002). She is presently working with a team studying children’s acquisition of mixed languages and code-switching in Central Australian languages. Pilar M. Valenzuela teaches Spanish linguistics at Chapman University in Southern California, USA. She is interested in functional-typological syntax, language contact and change, indigenous South American languages, and anthropological linguistics. She has published a book and various articles on Panoan languages. Topics of in-depth study include synchronic and historical syntax, ethnobiological taxonomy, and borrowing of verbs. Her doctoral dissertation, Transitivity in Shipibo-Konibo Grammar, was awarded the Mary R. Haas Book Award by SSILA. Her professional experience includes work as researcher, instructor, and consultant in language development and intercultural-bilingual education among the indigenous peoples of Peru and Bolivia, as well as teaching linguistics to speakers of native languages. Together with a group of collaborators, she has compiled and edited the bilingual autobiography of a Shipibo woman, which will be published in Peru. Johan van der Auwera teaches English and general linguistics at the University of Antwerp, Belgium. His research, conducted within Antwerp’s Centre for Grammar, Cognition and Typology, concerns the typology and areal typology of grammatical semantics. Currently the focus is on modality, mood and sentence type, and negation. Earlier work concerned connectives, relative clauses, Functional Grammar, and phasal adverbials.

Abbreviations and Glossing Conventions In glossing example sentences, the following conventions obtain: Prefixes and suffixes are separated by hyphens, clitics by an equal sign (¼), and infixes are enclosed by . If root and affixes are not separated in the vernacular, a colon is used between grammatical category labels and root meaning in the gloss (e.g. rls.ug:getting). If one formative represents two or more morphosyntactic categories, a period is used to separate the categories in the gloss (e.g. rls.ug). In the case of pronouns/pronominal clitics and affixes, an exception is made with regard to the period convention in order to save space, hence: 1sg, 2pl, etc. (but: 1sg.acc). The following abbreviations are used in this volume: 1 2 3 a abl abs acc act adess adj adjz adv aff agr all anaph andat ant antip aor

first person second person third person agent-like argument of canonical transitive verb; A-orientation ablative absolutive accusative actor adessive adjective adjectivizer adverb(ial(izer)) affirmative agreement allative anaphoric (article or deictic) andative (case) anterior (event) antipassive aorist

Abbreviations and glossing conventions appl art asp assoc aux ben caus clf coi com comp compl cond cont cop cvb dat decl def dem der des det detr dim dir dist distr do.along dpct ds du dur eff emph erg ess evid excl exp

applicative article aspectual marker associative auxiliary benefactive causative classifier coincidence comitative complementizer completive (aspect) conditional continuative (aspect) copula converb dative declarative definite demonstrative derivational marker desiderative determiner detransitivizer diminutive directional distal distributive do continuously while in motion depictive (marker) different subject dual durative effector (agent marker) emphatic ergative essive evidential exclusive experiential

xxi

xxii

Abbreviations and glossing conventions

f fam foc fqy fs fut g gen hab hon hsy hts ideo ill imp imprs inan inch incl incomp ind indf iness inf ingr ins intens intent intr intrss io ipfv iprf irr lig lk loc m mal man

feminine familiar (referent) focus frequency (‘times’) free-subject dependent predicate future general (gender) genitive habitual honorific hearsay evidential high tone suffix ideophone illative imperative impersonal (third person) inanimate inchoative inclusive incompletive (aspect) indicative indefinite inessive infinitive ingressive instrumental intensifier intentive (personal purposive) intransitive interessive (complement of interest) indirect object imperfective imperfect irrealis ligature linker locative; location masculine malefactive manner

Abbreviations and glossing conventions med mid min mod mot n nneg nmlz nom o obj obl onom opt ord orig p part pass pau perl pfv pl pn poss pot pred pref prev prf priv prog proh prol pron prop prox

xxiii

medial (distance) middle minimal modal element motative case (‘motion to or from’) neuter non- (e.g. nsg nonsingular, nf non-feminine, npst nonpast) negation, negative nominalizer/nominalization nominative second argument of transitive verb; O-orientation object oblique onomatopoeia optative (Georgian examples: optative of aorist series) ordinal (form of numeral) origin(ative) (case) patient-like argument of canonical transitive verb partitive passive paucal perlative perfective plural personal name (marker) possessive potential predicative (marker) prefix preverb perfect privative (‘without’) progressive prohibitive prolative pronoun proprietive (‘having’) proximal/proximate

xxiv

Abbreviations and glossing conventions

prs pst ptcl ptcp purp pv q quot rdp rec recp ref refl rel rem rep res restr rls rs s sbj sbjv seq sg sim simil spec ss stat sub subs subsec suff sup temp top tr transl

present past particle participle purposive patient voice question particle/marker; interrogative quotative reduplication recent past reciprocal referential reflexive relative remote past repetitive aktionsart resultative restrictive marker realis restricted-subject dependent predicate single argument of canonical intransitive verb; S-orientation subject subjunctive sequentive/sequential (relator) singular simultaneous converb/participle similative specific (article) same subject stative subordinator subsequent (event) subsecutive suffix superlative temporal (marker) topic transitive/transitivizer translative

Abbreviations and glossing conventions ua ug uq util vblz vis vn voc vsf

unit augmented undergoer unified quantifier utilitive verbalizer visible verbal noun vocative verb stem formative

xxv


1 Issues in the syntax and semantics of participant-oriented adjuncts: an introduction NIKOLAUS P. HIMMELMANN AND EVA SCHULTZE-BERNDT

1.1 Depictive secondary predicates in crosslinguistic perspective Constructions such as (1), where fresh functions as a depictive secondary predicate, provide a number of challenges for syntactic and semantic theory, and are frequently invoked as crucial evidence in syntactic analyses. (1) George bought the carrots fresh. Recent accounts have focused in particular on the following issues: Since depictive constructions contain two different predicates—the main predicate and the secondary predicate—they are a crucial test for theories of predication, in particular predication theory within generative grammar (Williams 1980; 1983; see Rothstein 2001 for a summary of the relevant literature). Depictive secondary predicates, being simultaneously predicates and adjuncts, take part in two relations within the clause, and thus pose problems for an analysis either in terms of constituency or in terms of dependency. Nevertheless, they are also invoked as evidence for controversial proposals in phrase structure theory such as the subjectinternal hypothesis (e.g. Roberts 1988). The role of the subject (or controller) of a depictive secondary predicate (the carrots in (1)) is often restricted to core argument functions (subject, object). For this reason they are also used as evidence in discussions of grammatical relations (e.g. Hale 1983; Kroeger 1993: 30).

N. P. Himmelmann and E. Schultze-Berndt

2

Depictive secondary predicates provide a major diagnostic for the stage-level vs. individual-level distinction (Carlson 1977; Kratzer 1995; Ogawa 2001) and currently play an important role in discussions of verbal aspect/aktionsart (e.g. Rapoport 1999). Depictive secondary predicates also provide evidence for focus theories (e.g. Winkler 1997). Discussions of these issues tend to make use of a very limited number of example types drawn from a few European languages (most importantly English, of course) and Japanese. This book, in contrast, attempts to present the full range of depictive constructions in crosslinguistic perspective. It is thus concerned with questions such as: Do all languages have (the same kinds of) depictive constructions? How are depictive constructions distinguished from related constructions in a given language? What kind of semantic content is typically rendered by a depictive construction? The contributions in this volume address these questions drawing on data from languages belonging to more than a dozen different families. This book thus builds on the short pioneering explorations of the typology of depictives by Nichols (1978a; 1978b) and Mu¨ller-Bardey (1990), and complements and expands the recent survey by Schultze-Berndt and Himmelmann (2004). The crosslinguistic perspective immediately brings to the fore an issue which is almost totally ignored in the literature on secondary predicates, but regularly surfaces in the literature on the classification of adverbs: the problem of distinguishing between depictives and (certain kinds of) adverbials. These are widely considered two distinct types of adjunct constructions, but the boundary between them is far from clear. As a very preliminary illustration, compare the two English examples in (2) with the German example in (3): (2) a. Claire left the room angry. b. Claire left the room angrily. (3)

Claire hat wu¨tend das Zimmer verlassen.

Both (2a) and (2b) have to be translated into German as (3). That is, in English there (often) is a fairly straightforward formal contrast between a depictive construction and an adverbial construction, but no such clearcut contrast exists in German. This difference between English and German is not restricted to morphosyntactic form. German native speakers would not normally consider (3) as semantically ambiguous, but rather as vague with regard to the meaning difference between the two English examples.

Issues in the syntax and semantics of participant-oriented adjuncts

3

German is not unique in this respect: time and again when looking for depictive constructions in the languages of the world we find that the same morphosyntactic construction may render depictive and adverbial content. For this reason, this chapter begins with a brief survey of some of the proposed classifications of adverbials and secondary predicates and then explores commonalities and differences between adverbials and depictives (section 1.2). It is argued that a number of adjunct types, including those referred to as depictives, circumstantials, oriented manner adverbs, and weak free adjuncts in studies of English syntax, share so many characteristics that at times they become virtually indistinguishable and form a single domain for crosslinguistic comparison. The term participant-oriented adjuncts is used here to refer to this domain. Participant orientation is found not only with standard examples of depictive adjuncts, i.e. adjuncts expressing a physical or mental state or condition (e.g. ‘angry’, ‘alive’, ‘hungry’, ‘drunk’, ‘raw’, ‘hot’), but also with a much larger semantic range of adjuncts, including quantificational, comitative, and locative adjuncts. Section 1.3 provides more details on the semantic range of participant-oriented adjuncts from a crosslinguistic perspective. Section 1.4 is a programmatic sketch of a morphosyntactic typology of participant-oriented adjuncts, taking into account the syntactic properties, the word class and internal structure, and the morphological marking displayed by the adjuncts in question.

1.2 The classification of adverbials and secondary predicates In this section, after introducing some basic distinctions and terminology regarding adverbials and secondary predicates (section 1.2.1), we first focus on participant-oriented adverbials and argue that in crosslinguistic perspective these are not necessarily distinguished as syntactic and semantic categories from depictive secondary predicates (section 1.2.2). In Section 1.2.3 we compare depictives with another type of secondary predicate, i.e. circumstantials, and again argue that these two types of secondary predicate may be clearly distinguishable in some but not necessarily in all languages. In section 1.2.4, we turn to so-called free adjuncts which have been a major concern in the semantics literature. We argue that weak free adjuncts are in fact identical to what is called circumstantial in the literature on secondary predicates. strong free adjuncts are considered independent clausal constructions only loosely connected to the matrix clause they modify. Finally, we briefly consider predicative complements, another construction type which is closely related to, and sometimes indistinguishable from,

4


depictives (section 1.2.5). Section 1.2.6 summarizes the major distinctions established in this section. 1.2.1 Some basic distinctions At first sight, the classification of secondary predicates, which are also known as predicative attributes (e.g. Paul 1919; Halliday 1967), predicative adjuncts (Hengeveld 1992a), and copredicates or copredicatives (e.g. Nichols 1978b; Plank 1985), appears to be straightforward. They are generally subdivided into depictives and resultatives. Depictives express a state that holds during the reference time of the event encoded by the main predicate. Example (1) above states that the carrots were fresh at the time that George bought them. Resultatives, on the other hand, express a state which is interpreted as a result of the state of affairs encoded by the main predicate. In (4) the carrots are soft as a result of having been boiled. (4) George boiled the carrots soft. Depictives and resultatives have in common the predicative nature of the adjunct (hence the term secondary predicate). That is, the state encoded by the secondary predicate is interpreted as holding for one of the participants of the main predicate (henceforth termed the controller), which in (1) and in (4) happens to be the direct object (for other possibilities see section 1.4.2). This participant orientation is often seen as the main feature distinguishing secondary predicates from adverbials, which are considered to be exclusively event-oriented. As we will see shortly, however, participant orientation and event orientation are not clearly aligned with depictive and adverbial constructions, respectively. Before turning to the classification of adverbials, we may note two problems with the simple classification of secondary predicates into depictives and resultatives. First, other more complex types of classification have been proposed (Halliday 1967; Nichols 1978a; 1978b; 1981; Plank 1985; Mu¨llerBardey 1990), and we will have a closer look at a putative third type of secondary predicates in section 1.2.3. Second, it is not clear whether resultatives are in fact secondary predicates. Crosslinguistically, it is quite clear that resultative notions are often expressed by complex predicates, not by adjunct constructions. This is true for Warlpiri (Simpson, Ch. 2, this volume), but also for English and German—at least this has been argued by Dowty (1979), and is the central claim in Neeleman (1994) and Winkler (1997). Consequently, resultatives only play a minor role in this volume. The classification and analysis of adverbials remains a highly controversial issue. Numerous subdivisions have been proposed (for a recent overview see


5

Tenny 2000), and the classes resulting from the application of different criteria hardly overlap. Furthermore—and this is perhaps the most confusing phenomenon in this grammatical domain—it is the rule rather than the exception that a given adverb or adverbial allows for a number of different readings and thus often also belongs to a number of different classes. Compare the following three examples: (5) a. Elaine spoke naturally. b. Naturally Elaine spoke. c. Elaine spoke, naturally. The difference between (5a) and (5b) is obvious: In (5a) naturally says something about the way Elaine spoke, i.e. it in some sense modifies the verb and is thus called a predicate-level adverb. In (5b) naturally conveys an evaluative stance of the speaker towards the proposition as a whole (for the speaker, Elaine speaking was a natural thing to happen). Here, naturally functions as a sentence-level adverb. In these two examples, the two quite different functions correlate with a difference in position (clause-initial vs. clause-final). However, it is not the case that differences in meaning and syntactic scope are always clearly correlated with positional differences, as shown by (5c). Here naturally is prosodically detached from the rest of the clause and forms an intonation unit of its own, and both interpretations are possible.1 Variable position, sometimes correlating with a meaning difference and sometimes not, is a highly conspicuous feature common to both adverbials and depictive secondary predicates, as will become apparent throughout this chapter as well as the whole book. There are four basic kinds of parameter employed in the classification of adverbials. Adverbials may be classified by their internal constituency (e.g. simple (lexical) adverbs, adjective-derived adverbs, prepositional phrases, adverbial clauses) or by their morphological marking (e.g. English adverbs in -ly, adverbials marked with an instrumental adposition or case). While there is little controversy about these two ways of classifying adverbials, these morphosyntactic features rarely correlate to a significant degree with the two remaining parameters, semantics and syntactic distribution and scope (such that, for example, -ly adverbs in English would always be manner adverbs semantically). Hence, internal constituency and morphological marking are of rather limited interest for the classification of adverbials. 1 In spoken language, the two interpretations may often be distinguishable by prosodic cues in rendering naturally, which is of no direct import to the current argument.

6


The classification of adverbials according to semantics or syntactic distribution and scope is much more problematic. This is because there are different subparameters in each domain, and many taxonomies in fact mix semantic with syntactic criteria. Syntactic sub-parameters include positional variability (e.g. for English: adverbials with fixed sentence-final position vs. adverbials with variable position, see Filipenko 2000) and syntactic scope, which in part at least correlates with position. The best-known scope distinction is the one between sentence-level and predicate-level adverbials, which is as much a syntactic as a semantic distinction. The most common and widespread semantic classification of adverbials is the one by semantic function (or logical role), distinguishing adverbials of manner, location, time, degree, etc. (see also section 1.3). Here some confusion easily arises because terms such as manner, location, time, reason/ cause, condition, which are widely used in descriptive and pedagogical grammars, are used in a different, much more specific sense in the more recent theoretical literature, where a much larger number of semantic (or syntactic-semantic) adverbial classes are distinguished. For example, ‘manner adverb(ial)’ may be used either in its broad descriptive sense of ‘an item which usually/potentially conveys something about the manner in which an action is performed’, or in the narrower sense of ‘an item which actually conveys the manner in which an action is performed, and nothing else’. Geuder (2000), for example, distinguishes the following four subtypes of manner adverb functions: pure manner, transparent, agentive, and resultative (not to be confused with resultative secondary predicates). These are illustrated in (6)–(9) (taken from Geuder 2000: 29–35). Note in particular that the same lexical item, e.g. angrily or stupidly, can appear in more than one function. (6) a. John shouted at them angrily. (pure manner) b. John answered the question stupidly. (pure manner) (7)

He angrily broke the door open. (transparent)

(8)

John stupidly answered the question. (agentive)

(9)

They loaded the cart heavily. (resultative)

In (7)–(9) the ‘manner’ adverbs do not exclusively convey the manner in which an action is performed in the same sense as in (6). Instead, at least (7) and (8) also convey statements about one of the participants of the main event. In (7) John was angry while breaking the door open; in (8) it was stupid of John to answer the question (for (9) see Geuder 2000: Ch. 3). In this


7

sense they are participant-oriented adverbs rather than pure manner adverbs, which are exclusively event-oriented. Before taking a closer look at participant-oriented adverbials, a note on terminology is in order. We will follow Geuder in using the term pure manner when referring to adverbials (or more precisely: adverbial uses) which convey manner and nothing else. Adverbials in the uses illustrated in (7) to (9) will be termed oriented (manner) adverb(ial)s. We will use the unmodified term manner adverb(ial) in the broad sense, i.e. with reference to an adverb(ial) which conveys manner (and thus is event-oriented) but may also have an additional orientation. In fact, as will be shown in sections 1.2.2 and 1.3.3 below, and in several of the contributions to this volume, for manner adverbials in this broad sense it can often be said that participant orientation and event orientation are simultaneously present, or can be derived from one another by metonymic shift. Note also that participant orientation is not the only kind of additional orientation. There is, for example, also speakerorientation, which characterizes evaluative uses such as (5b) above. 1.2.2 Oriented adverbials vs. depictives The fact that adverbials are not necessarily and exclusively event-oriented but instead may exhibit semantic orientation towards a participant has been widely noted in the (semantics) literature; see Platt and Platt (1972), Jackendoff (1972: 47–107), Bartsch (1976: 144), Dik et al. (1990: 31), McConnell-Ginet (1982), Frey and Pittner (1998), and Ernst (2000b), among others. Most of these authors only discuss manner adverbials (in the broad sense), although occasionally other types of adverbial are also taken into consideration. Jackendoff (1972: 57–8), for example, also regards in order to clauses (as in Bill has ruined his chances of an inheritance in order to kill his mother) as subject-oriented expressions. This issue is the main concern of section 1.3 where we make an attempt to determine the full semantic range of participant-oriented adverbials. In the current section, the main concern is as follows. Orientation towards a participant is a key feature shared by oriented adverbials and secondary predicates. This immediately raises the question of how the two construction types differ. One of the few authors who devotes extensive discussion to this question is Geuder (2000), on which the first part of this section therefore heavily draws. We limit the discussion to two of the three types of oriented manner adverbs distinguished by Geuder, i.e. agentive and transparent manner adverbs. Agentive adverbs such as stupidly in (8) above have also been called subject-oriented adverbs and syntactically belong to the sentence-level adverbials. They ascribe a certain characteristic to the agent on the basis of the

8


event which it performs. The agentive orientation of the adverb in (8) is clear from paraphrases such as (10). (10)

It was stupid of John to answer the question.

In the ‘pure manner’ use of the same adverb in (6b), in contrast, stupidity is ascribed to the way John answers the question. Ernst (2000a; 2000b) makes the intuitions behind these paraphrases more explicit by stating that agent-oriented adverbials evaluate the agent with respect to the event that is performed, where the comparison class consists of other possible events. In the case of (8), these alternatives would include silence, or an explicit refusal to answer the question. In the case of pure manner adverbials, the comparison is with other events of the same specific type. In the case of (6b) these would be other events of answering the same question, which could take place in more or less stupid ways.2 These differences in interpretation here clearly correlate with a different position of the adverbial (preverbal vs. clause-final). Although agentive adverbs are participant-oriented, they differ from depictives in a number of ways. Semantically, depictives do not convey any evaluation of the agent’s action and thus, among other things, may be oriented towards participants other than the agent (as in John drank his coffee hot). Depictives are also never to be interpreted as predicating over the whole state of affairs: John answered the questions drunk cannot be paraphrased as ??It was drunk of John to answer the question (compare (10) above). These differences in scope often correlate with differences in syntactic position between agentives and depictives. In English, agentives are only found sentence-initially and in pre-auxiliary position, while depictives generally occur in post-predicate position. In other languages, other formal characteristics may set agentives apart from depictives. In German, for example, agentives have a special suffix (-weise). Without having investigated this in detail, we think it very likely that agentive adverbials are relatively easily distinguishable from depictives on formal as well as semantic grounds in all languages. Therefore, they will not be considered any further in the remainder of this chapter. The second type of participant-oriented adverbs—transparent adverbs3— are very similar to depictives, however. Compare the following examples: (11) a. John read the review slowly. (pure manner) b. John angrily read the review. (transparent) c. John left the party angry. (depictive) 2 Geuder (2000: 137) proposes to re-analyse Ernst’s ‘comparison classes’ as alternatives in the sense of focus semantics. 3 Regarding the rationale for this less than transparent term, see Geuder (2000: ch. 2.1.1).


9

As shown in detail by Geuder (2000: ch. 5), angrily in (11b) is not a pure manner adverb like slowly in (11a). It does not exclusively indicate the manner of reading, which is why (11b) cannot be paraphrased as John read the review in an angry manner. Rather, it is participant-oriented in that it describes the emotional condition of the subject at the time of reading. In this respect it bears a close semantic relationship to the depictive in (11c), which also encodes that John was angry at the time of leaving the party. There is a rather subtle semantic difference between the transparent and the depictive construction. The depictive construction in (11c) merely entails that the two states of affairs of leaving the party and being angry overlap in time. Note in particular that the departure here is not necessarily related to the state of being angry. The adverb in the construction in (11b), on the other hand, conveys a closer factual link between the activity denoted by the main predicate and the bodily or psychological condition of the subject. This link pertains to the fact that the way in which the subject is engaged in the activity denoted by the main predicate suggests something about its bodily or psychological condition which in turn suggests that activity and bodily or psychological condition are inherently linked in some way. The precise nature of the factual link varies. Two possibilities mentioned by Geuder are causal consequence and motive. In (11b), the link appears to consist in a causal consequence: John becomes angry because of reading the review. An example showing a motivational link between the state conveyed by the transparent adverb and the activity denoted by the main predicate is (12), where the state of being hungry is understood as the motivation for the boy’s return to his parents. (12) The boy hungrily returned to his parents. (Geuder 2000: 204) The major semantic difference between depictives and transparent adverbs thus pertains to the relation established between the state of affairs denoted by the main predicate and the condition denoted by the participant-oriented adjunct. In the case of depictives, the relation is one of mere temporal overlap, while in the case of transparent adverbs it goes beyond mere temporal coincidence and reflects a ‘deeper’, factual link (often causal or motivational). An important correlate of this semantic difference would appear to be the fact that transparent adverbs, unlike depictives, always seem to exhibit agent orientation (on the range of controllers for depictives, see section 1.4.2 below). Geuder discusses the intricacies involved in providing a detailed formal representation of the meaning differences between depictives and transparent adverbials. There is one point in this discussion that we wish to

10


emphasize here. Within the framework of Davidsonian semantics, where verbs carry an event variable and adverbials are analysed as predicates of events, there is an apparently straightforward way of representing the difference in meaning between (11a) (pure manner) and (11c) (depictive), as shown in (13). (13) a. read(e) (x, review) & slow(e) (pure manner) b. leave(e) (x, party) & angry(x) (depictive) The representations in (13a) and (13b) make a clear-cut distinction between event orientation and participant orientation: in (11a) John read the review and his reading was slow while in (11b) John left the party and was angry at the same time. As Geuder notes, there are two major problems with this approach. First, it does not offer an equally simple representation for transparent adverbs. Second, and more importantly, the representation of the depictive construction in (13b) only captures the participant orientation of angry. It does not represent the relation of temporal overlap between the two states of affairs of leaving and being angry, and thus, for example, does not explain why individual-level predicates such as clever cannot function as depictives (consider *John left the party clever). Geuder consequently proposes more complex representations for depictives and transparent adverbs, and states in conclusion: In fact, depictives have been found to carry an almost ‘adverbial’ reading: Their interpretation is linked to the event variable of the verb. The difference between depictives and transparent adverbs lies merely in the fact that depictives assert the independence of a concurrent state while the adverbial forms assert the existence of a closer factual connection to the event. (Geuder 2000: 213)

That is, at least in some of their uses, manner adverbs and depictives are very similar indeed. It will thus not come as a surprise that the different meanings conveyed by depictives and (in particular participant-oriented) manner adverbials are not distinguished on the formal, i.e. morphosyntactic, level in all languages, as seen in the following discussion. The semantic differences between depictives and participant-oriented uses of manner adverbs in English may be rather subtle, and native speakers not specializing in semantics usually find it very hard to make them explicit. But there is little doubt that such differences exist. This may be due to the fact that in English most manner adverbs involve the suffix -ly and thus differ formally from depictives. In German, on the other hand, for many bases expressing bodily or psychological conditions such as traurig ‘sad’, wu¨tend


11

‘angry’, etc. there is no morphological difference between the use as a manner adverbial and as a (primary or secondary) adjectival predicate, making it questionable whether these uses can be distinguished at all.4 Geuder (2000: 192) mentions the possibility that they may be distinguished by position (and stress), such that (14a) only allows the depictive reading and (14b) only allows a manner reading (pure manner or transparent). German (14) a. daß Hans den Raum traurig verließ that John the room sad left [with the major accent on traurig] ‘ . . . that John left the room sad’ b. daß Hans traurig den Raum verließ that John sad the room left ‘ . . . that John left the room sadly’ (Geuder 2000: 192) However, to us—and, we believe, most native speakers—this difference is far from clear-cut, and we would argue that both (14a) and (14b) are vague rather than ambiguous with respect to a depictive or an adverbial (pure manner or transparent) interpretation. While the emotional condition traurig in (14b), but probably not in (14a), can be construed as a motivation for leaving the room (which may simply be due to the iconicity of ordering; cf. Geuder 2000: 212), in both examples an interpretation of pure temporal overlap as well as of pure manner can be construed. (See also our discussion of example (3) above.) In languages such as German where depictive and adverbial constructions are often not distinguished by morphological marking, it is common to diagnose the differences between the two by the paraphrases they allow (cf. e.g. Bartsch 1972: 140f.; Pittner 1999: 97). It is assumed that depictive constructions, but not adverbials, can be paraphrased by a biclausal construction where the element corresponding to the depictive functions as a primary predicate. Thus, (15) is a possible paraphrase of (1). (15)

The carrots were fresh (when George bought them).

Conversely, adverbial constructions, but not depictives, can be paraphrased with a clause where the element corresponding to the adverbial functions as the main predicate, or by an attributive construction where the element corresponding to the adverbial is a modifier of a verbal noun. Thus, both (17a) and (17b) are possible paraphrases of (16), but (18a) and (18b) are 4

See also van der Auwera and Malchukov (Ch. 13, this volume) for Dutch.

12


not possible paraphrases of (1). (16)

George bought the carrots fast.

(17) a. The buying of the carrots (by George) was fast. b. The fast purchase of the carrots (by George). (18) a. ??The buying of the carrots (by George) was fresh. b. ??The fresh buying of the carrots (by George). However, even these examples may be used to show that the supposedly clear semantic difference between participant-oriented (depictives) and event-oriented adjuncts (adverbials) is not quite so straightforward. That is, the paraphrase tests which are supposed to be diagnostic for this difference are flawed in that it is possible to paraphrase the ‘adverbial’ construction in (16) as (19), in analogy to the paraphrase of the depictive construction in (15). (19) George was fast (when he bought the carrots). Thus, as has also been pointed out by Plank (1985), semantic intuition and paraphrase tests are often at a loss to distinguish between a participantoriented and an event-oriented adjunct in the absence of overt formal marking, since it appears that one and the same adjunct (such as fast in (16)) can be construed as simultaneously participant-oriented and event-oriented. From a wider crosslinguistic perspective, there is further evidence for a much more pervasive overlap between participant-oriented and eventoriented adjuncts than the traditional dichotomy between secondary predicates and manner adverbials suggests. One similarity between manner adverbials and depictives in many languages, including English, is that both can be substituted by the manner interrogative (‘how’) (see Boeder, Ch. 6, this volume, for Georgian). Moreover, constructions that can be identified as depictive constructions on formal grounds may also encompass expressions that we might regard as primarily event-oriented on semantic grounds—in other words, the morphosyntactic marking contradicts our semantic intuitions in these cases. For example, adverbs like slowly and fast are classified by almost all authors as pure manner adverbs, i.e. as exclusively event-oriented adverbs, on semantic grounds (although, as we have already seen in (19), speed can also be ascribed to a participant directly). However, there are languages in which the translation equivalent of slowly behaves exactly like a depictive. As (20) shows, in the Australian language Martuthunira, the manner adjunct jarruru ‘slow(ly), careful(ly)’ agrees in case with a controller in the same way as the typical depictive wanka ‘alive’. (In (20a), the controller is not overtly expressed.)


13

Martuthunira (20) a. ngunhaa malumalu-npa-waa¼rru jarruru-u waruul dem.nom dark-inch-purp¼now slow-acc still ‘(It’s slowly getting cool now as the sun goes down). That will make everything go dark slowly.’ (Dench 1988: 121) b. nhulaa miyu mungka-rnuru wajupi-i wanka-a¼l dem.nom cat eat-prs grasshopper-acc alive-acc¼temp ‘this cat eats grasshoppers alive.’ (Dench 1995: 182) A number of authors have proposed that a process of metonymic shift may provide the explanation for the fact that event-oriented adjuncts may be interpreted as participant-oriented (and vice versa), a possibility already hinted at in our discussion of the transparent use of angrily illustrated in (11b) above. Consider (21). (21)

Fred ate the sausages ravenously.

In the discussion of this example, Platt and Platt (1972: 237) note: We do perhaps talk about ravenous eating but in saying this we imply that the eater is ravenous. It may be objected that this ravenous quality displays itself in the manner of the eating. This is not denied. The manner of eating is an outward and visible sign of an inner ravenous quality of the eater. To say that someone eats ravenously is a judgement about the eater.

The authors then conclude that ravenously is primarily a participant-oriented adverbial (where the participant in this case is the agent), but that the nature of the agent may be inferred from the manner in which the event is performed, and this is how the adverbial derives a secondary, event-oriented sense. A similar conclusion is drawn by Geuder (2000: 214–16) for the adverbial sadly in the translation of (14): the manner of leaving the room is not sad as such, but it allows the conclusion that the subject is sad at the time of leaving the room. In an analogous fashion, Vogel (1997: 412) interprets German leise ‘quiet’ as a primarily event-oriented adverb. When it is predicated of an animate being (as in John is quiet), this is interpreted as ‘John generally behaves in a quiet manner’ (see also Eisenberg 2002: 63). In the analyses just referred to, the adverbials in question are interpreted as polysemous, being basically event-oriented but also allowing a derived participant-oriented reading (or vice versa). However, this would not appear to be the only possible analysis. Instead, the adverbials in question could be considered to be semantically general, ‘vague’, or ‘underspecified’ (Do¨lling 2003). That is, they exhibit both event-oriented and participant-oriented

14


components simultaneously, without one being primary and the other secondary. The metonymic transfer is inherent in their lexical semantics, because it is always possible to ascribe the manner of performing an event to the agent, and to conclude the agent’s nature from the way in which the event is performed. A more detailed argument along these lines is put forward by Boeder (Ch. 6, this volume). It is this dual nature of manner adverbials (in a broad sense) that allows us to paraphrase (16) as both (17) and (19). The dual nature of many types of adjunct with regard to event- and participant-orientation also points to an explanation for the crosslinguistic differences in coding hinted at above. In Schultze-Berndt and Himmelmann (2004) we have suggested that there exists a continuum between expression types (such as expressions of physical condition) where the participantoriented component prevails and those (such as expressions of time) where the event-oriented component prevails. Manner expressions are situated in the middle of this continuum and are therefore encoded as adverbials in some languages (as in the case of agent-oriented adverbials in -ly in English), and as depictives in others, as shown for Martuthunira in (20). Languages may also exhibit an overlap of both adverbial and depictive coding strategies for certain semantic types of adjunct, in particular manner adjuncts, with no or only very subtle differences in meaning (as shown by Geuder (2000) for English, and Boeder (Ch. 6, this volume) for Georgian), or as a stage in a diachronic development from one preferred strategy to another (see Pinkster 1988: 227 for Latin). As a third possibility, a language may choose a nonspecific coding strategy, as has just been argued for German manner adjuncts. Given that participant orientation and event orientation may thus overlap and cooccur in a broad variety of linguistic structures, the following terminological conventions will be useful for keeping track of crosslinguistically recurrent (non-)distinctions. The terms construction and adjunct are used in this chapter only in reference to morphosyntactically definable units while expression is used primarily in reference to semantically definable units. Participant orientation and event orientation are understood here to be primarily semantic relations. A depictive construction (or secondary predicate) is a construction primarily dedicated to conveying participant orientation, although, as we will see, certain expressions involving depictive constructions probably have to be regarded as event-oriented in some languages. Frequently, participant orientation is iconically reflected in the formal marking of a depictive construction, e.g. in the agreement of the depictive with its controller. The important point, though, is that the depictive construction differs in at least one morphosyntactic feature from other adjunct constructions. This feature may pertain to syntactic position or


15

distribution or to the fact that the construction is morphologically unmarked while all other adjunct constructions are marked. Similarly, an adverbial construction is a distinctive construction (e.g. one marked by adverbial affixation, or by a distinct position) which is primarily dedicated to conveying event orientation, although in some cases it may be interpreted as participant-oriented. The term general adjunct construction, on the other hand, is used for adjuncts which are neutral with regard to the event-oriented/participant-oriented distinction and thus allow for both a participant- and an event-oriented interpretation as an unmarked reading. With regard to (morphosyntactically defined) adjuncts, we thus distinguish between depictive, adverbial, and general adjunct constructions (see also Schultze-Berndt and Himmelmann 2004). As these definitions make clear, semantic interpretation as participant- or event-oriented and morphosyntactic properties of a construction do not necessarily align. Consequently, in the current framework it is not a contradiction to speak of participant-oriented uses of an adverbial construction (e.g. the transparent uses of English manner adverbs discussed above) or event-oriented instances of a depictive constructions (e.g. the agreeing temporal expressions discussed in section 1.3.11 below.). 1.2.3 Depictives vs. circumstantials In this section, we will introduce a further subdivision in the domain of depictive constructions as defined in the previous section, into depictives proper and participant-oriented adjuncts of another type, so-called circumstantial (or conditional) secondary predicates. The latter are usually morphologically identical to depictives; in English, for example, neither depictives nor circumstantials carry a special morphological marking. The main crosslinguistic finding is similar to the one in the preceding sections. In some languages, including English, depictives proper and circumstantials are two closely related but nevertheless distinct constructions, the main difference resulting from their interaction with focus. In other languages there may be just a single construction for both functions. As briefly mentioned in section 1.2.1 above, it is common in the literature on secondary predicates to distinguish between depictives and resultatives, and especially in the more recent literature this already completes the top level of the taxonomy. Some authors, however, discuss a third, related construction. Halliday (1967: 78–81) terms this construction condition(al) (short for ‘conditional attribute’). Halliday himself notes that ‘the distinction between attribute [i.e. depictive—NPH and ESB] and condition may seem somewhat arbitrary; and certainly it may not be very clear to which type a

16


given token should be assigned’ (1967: 80). Despite these problems of delimitation and the fact that they share a number of important features with depictives and resultatives (1967: 62), Halliday clearly sets conditionals apart from depictives/resultatives in his overall systematics. It is thus unclear whether or not conditionals should also be considered secondary predicates in his taxonomy. In some of the examples that Halliday provides for conditionals, e.g. those in (22a) and (22b), the conditional meaning seems to arise from the presence of a modal element in the main predicate. For example, the relation between the adjunct raw and the main predicate (and the undergoer) in (22b) seems to be identical to that in you eat them raw, which Halliday considers a depictive construction (cf. also Aarts 1995: 78f.). (22) a. I can carry it empty. b. You can’t eat them raw. c. She died young. (Halliday 1967: 78) In other examples, e.g. (22c) (which according to Halliday is ‘surely conditional’ as opposed to depictive she died happy), it appears to be the secondary predicate itself which contributes the ‘conditional meaning’. The term condition is used here in two quite different senses. In (22a) and (22b) it refers to a precondition which could also be rendered by a conditional clause (I can carry it if it is empty). In (22c), on the other hand, condition refers to a life stage (she died when she was young). We argue here that there may be some support for distinguishing the former usage (i.e. examples (22a) and (22b)) from depictives while the latter is in fact depictive. To date, conditionals have generally not been acknowledged as a distinct type of secondary predicate in the literature, with a few exceptions. The major exception is Nichols (1978a: 117; 1981; see also Plank (1985: 169–70), and Simpson (Ch. 2, this volume)), who regards conditionals as a subtype of what she calls circumstantial secondary predicates, thus making it clear that they are secondary predicates. Examples of conditional circumstantial secondary predicates in Nichols’s sense are found in (23). (23) a. This food is not supposed to be nice cold. [overheard utterance] b. I can’t work hungry. The other two subtypes of circumstantial secondary predicate recognized by Nichols are temporal (As a child he lived in Paris or I knew him young) and concessive (Even dead I won’t forget). Quite clearly, then, Nichols’s term


17

circumstantial, which we will adopt here, covers the same ground as Halliday’s and Simpson’s conditional.5 At first sight, the distinction between circumstantial and depictive secondary predicates appears to be primarily a semantic one, pertaining to the logical relation between main and secondary predicate. In the case of depictives (He bought the carrots fresh, They left outraged), the logical relation between main and secondary predicate is exclusively one of temporal overlap: the state referred to by the depictive holds true at the same time as the event expressed by the main predicate (and it may have held true before that point in time and keep on holding true after it). In the case of circumstantials, other semantic links are evoked in addition to the basic pure temporal overlap. Following Nichols, these ‘other semantic links’ could be classified as condition, concession, and temporal (in this regard circumstantials are similar to the transparent uses of manner adverbs discussed above).6 Since this semantic distinction at times ‘may seem somewhat arbitrary’, as Halliday rightly stresses, one may wonder whether there are any formal correlates which would allow us to make the distinction more operational. Two major candidates for such formal correlates which are occasionally invoked in the literature are negation and position. We take up negation first. Circumstantial secondary predicates, such as cold and hungry in (23) and as a young girl in (24), appear to be outside the scope of negation, i.e. in these examples it is possible to negate the main predicate without negating the secondary predicate (see also Simpson, Ch. 2, this volume). (24) As a young girl Sarah did not travel to Paris alone. This clause can certainly not be understood as denying that Sarah was a girl in the same way that (25) can be read as denying that they were outraged. (25) They didn’t leave outraged (they left happy). That is, the negator in depictive constructions such as (25) can be interpreted as having either the depictive alone in its scope, or both the depictive and the main predicate (No! They didn’t leave outraged, they didn’t leave at all), but not just the main predicate as is usually the case with circumstantials.

5 Note that Halliday also uses the term circumstantial, but in a very different sense. His circumstantial is a cover term for the following four constructions: beneficiaries (roughly: indirect objects), range (e.g. wall in he jumped the wall), attribut(iv)es (depictives, resultatives), and his condition(al)s (1967: 52–3). 6 In the case of temporal circumstantials (As a child he lived in Paris) the additional semantic link consists in the fact that the secondary predicate (as a child) contributes to the determination of the reference time of the clause in a way similar to temporal adverbials (Twenty years ago he lived in Paris).

18


With regard to the second formal parameter, position, it has been noted that in English depictives usually have to occur in post-predicate position, while circumstantials often occur in clause-initial position. This is completely unproblematic for As a child he lived in Paris and Even dead I won’t forget. In the other examples of circumstantials mentioned above, initial position is at least marginally possible, although for some of the presumed temporal circumstantials, e.g. (26e, f) the result is not too good (as indicated by the question marks): (26) a. Empty I can carry it. b. Raw you can’t eat them. c. Cold this tea isn’t good. d. Hungry I can’t work. e. ??Young I knew him. f. ??Young she died. The preceding observations point to the fact that what is actually at stake here is focus. Typical depictive secondary predicates always provide focal information, as has been argued at length by Winkler (1997) (see also Geuder 2000: 199–200; Gu¨ldemann, Ch. 10, this volume). They are part of the focus domain of the clause they appear in and, at least in standard examples such as They left angry and She returned drunk, they in fact carry the sentence accent and thus function as the focus exponent. Being part of the focus domain explains why depictives are usually not acceptable in initial position in English (except in contrastive contexts), and why they are within the scope of clause-level negators. Circumstantials, on the other hand, contribute presupposed information to an utterance. They are generally unstressed (unless they attract contrastive stress) and outside the scope of clause-level negators. Thus, they appear either in pre-predicate position (as in (24)), or they are clause-final and unstressed (see also Pittner 1999: 115 for German). Thus (26c) is equivalent to (27a), while (27b) is only acceptable in contrastive contexts (such as this tea isn’t good cold, it’s good hot). (27) a. This tea isn’t good cold. b. ?? This tea isn’t good

COLD. COLD.

In this view, the fact that young in I knew him young and she died young is usually stressed and strongly disfavours initial position implies that it is a depictive and not a circumstantial in these examples (pace Halliday and Nichols). This is also supported by the negation test, since She didn’t die young allows a reading in which only the secondary predicate is in the scope of negation (She didn’t die young, she died in her seventies).


19

To sum up this section: in English it is possible to distinguish two kinds of otherwise very similar secondary predicates, depictives proper and circumstantials. Both are participant-oriented adjuncts which convey a state of affairs which temporally overlaps with the state of affairs conveyed by the main predicate. They differ, however, in that depictives are part of the focus domain and convey focal information while circumstantials do not. This explains their positional and prosodic differences (or rather preferences) as well as the fact that depictives, but not circumstantials, can be in the exclusive scope of a negator. On the semantic level, circumstantials often evoke a specific semantic link (condition, concession) to the main predicate in addition to the basic pure temporal overlap characterizing depictives. This can be regarded as a pragmatic inference deriving from the non-focal quality of these adjuncts, not as a defining characteristic of the construction: if the adjunct is not in focus, it is part of the presupposition and in restrictor position (cf. Mu¨ller-Bardey, Ch. 3, this volume), and its function is that of restricting the interpretation of the constituent(s) in focus. While it is very likely that participant-oriented adjuncts may appear in focus or in the presupposition in all languages where participant-oriented adjunct constructions occur, it is not necessarily the case that this difference with regard to information structure is also manifest on the morphosyntactic level. That is, it would appear to be possible that a language does not distinguish between depictive and circumstantial constructions, but has a single construction conveying participant orientation which is open to both a focal and a presuppositional interpretation, based primarily on pragmatic inferences. The fact that depictives and circumstantials thus are very closely related constructions gives rise to different readings of the term depictive: In one sense, it refers to participant-oriented adjuncts which not only are formally distinct from event-oriented adjuncts, but also belong to the (morphosyntactically delimited) focus domain. These may be called depictives proper. On the other hand, it may refer to adjunct constructions conveying participant orientation regardless of their information-structural status, hence encompassing depictives proper and circumstantials. In line with widespread usage in the current literature (including most contributions to this volume), these may be simply called depictives. We will be using depictives in the broad sense whenever it is important to emphasize that a given argument or statement pertains to both depictives proper and circumstantials. 1.2.4 Depictives vs. free adjuncts So-called free adjuncts (e.g. Having unusually long arms, John can touch the ceiling) are also participant-oriented and share essential similarities with

20


depictives such as denoting a condition or state which temporally overlaps with the state of affairs denoted by the main predicate. Here we argue that so-called weak free adjuncts are actually identical to what has been called circumstantial secondary predicates in the preceding section, thus unifying insights from two different strands in the literature which appear to have been unaware of the fact that they deal with essentially the same phenomenon. Strong free adjuncts are only loosely attached to, and thus in fact outside, the clause headed by the primary predicate, and hence strictly speaking not secondary predicates. The distinction between strong and weak free adjuncts has been proposed by Stump (1985), who provides the most detailed semantic analysis of free adjuncts to date. The following two of his examples (1985: 41–2) may serve as an initial illustration: (28) a. Standing on a chair, John can touch the ceiling. (weak) b. Having unusually long arms, John can touch the ceiling. (strong) In (28)a, the initial adjunct is similar to a conditional clause in that it restricts the interpretation of the modal, as in If he stands on a chair, John can touch the ceiling. In (28b), on the other hand, the adjunct does not have such a restricting function and rather resembles a causal clause (Because he has unusually long arms, John can touch the ceiling). This difference correlates with a difference in entailments. The actual truth of the strong adjunct in (28b) (i.e. that John has unusually long arms) is always entailed by the truth of the whole sentence, while the actual truth of the weak adjunct in (28a) (i.e. that John is standing on a chair) is not necessarily entailed by the truth of the whole sentence (Stump 1985: 41–64). The difference between weak and strong free adjuncts is linked in part to the meaning of the adjunct. If the adjunct involves an individual-level predicate (roughly: the subset of stative predicates which refers to inherent, essential properties) then it is always strong. That is, individual-level predicates like having unusually long arms in (28b) may not function as weak free adjuncts. Only stage-level predicates such as standing on a chair or drunk allow weak uses. However, a weak use is always just a possibility, and very much depends on the larger context in which the adjunct occurs, as repeatedly emphasized by Stump (1985: 55 and passim).7 Thus given the right context, standing on a 7 Note that we deviate here in a minor way from the terminological conventions used by Stump. For Stump, there are two kinds of free adjunct, strong and weak, the latter allowing two kinds of use, which are referred to in various ways including strong vs. weak. In the present chapter, weak free adjunct refers to a stage-level predicate used in weak free adjunct function, while strong free adjunct refers to an individual-level or a stage-level predicate used in strong free adjunct function.


21

chair in (28a) can be interpreted as a description of an actual occurrence of John standing on a chair and can thus function as a strong free adjunct. In this case the actual truth of the adjunct would be entailed, and a causal rather than a conditional link between adjunct and main predicate would be evoked. Furthermore, the difference between strong and weak adjuncts only becomes apparent in a limited number of morphosyntactic contexts. One such context are modally marked main predicates, as in the above examples. Other contexts discussed by Stump are frequency adverbs (e.g. Lying on the beach, John sometimes smokes a pipe; see also Mu¨ller-Bardey, Ch. 3, this volume) and generic or habitual sentences (e.g. Drunk, he drives very dangerously). In these contexts, weak free adjuncts function as arguments of a clause-level operator (such as a modal or a frequency adverb), while strong free adjuncts remain outside the scope of this operator. Importantly, then, weak free adjuncts are part of the same clause as the (finite) main predicate because both function as arguments of the same operator. Strong free adjuncts, on the other hand, express ‘a proposition logically distinct from that expressed by the main clause’ (Stump 1985: 276–7). The preceding short summary of the major differences between strong and weak free adjuncts will have made it clear that weak free adjuncts have much in common with the circumstantial secondary predicates discussed in the preceding section. Both are non-focal participant-oriented adjuncts which preferably occur in modal or habitual expressions and often involve a conditional link between adjunct and main proposition. Stump (1985: 87f., 272–7) in fact claims that some of those prepositions which mark circumstantial secondary predicates such as as in as a child serve to turn strong free adjuncts into weak ones (or individual-level predicates into stage-level predicates). Stump’s example is: A blonde, Mary might look something like Jane (only strong) vs. As a blonde, Mary might look something like Jane (weak or strong). A possible objection to collapsing weak free adjuncts and circumstantials in a single category may be differences with regard to lexical category, position, and prosody. Standard free adjunct examples usually consist of participial expressions in initial position, which is the most commonly attested position. Circumstantials are often illustrated with single adjectives, usually also in initial position but occasionally also in other positions. But this difference with regard to standard examples is purely superficial. Weak free adjuncts may also simply consist of a single adjective; they may also occur after the superordinate clause and sometimes even immediately after the subject of the superordinate clause (Stump 1985: 6). As far as prosody is concerned, note that the fact that both strong and weak free adjuncts are generally separated with a comma from the superordinate


22

clause is an orthographic convention in English and does not necessarily indicate a prosodic break. It is not unlikely that strong and weak adjuncts are actually distinguished by the presence vs. the absence of a prosodic break. But this is an empirical issue impossible to decide at this point, as we lack crucial evidence. However, we may note that the occurrence of a clear prosodic break appears to turn typical depictives into independent clause-like units or, in Stump’s terminology, strong free adjuncts, even when they occur in utterancefinal position. Compare the following examples (the comma here represents a clear prosodic break).8 (29) They returned early in the morning, (totally) drunk. (30) They didn’t leave, outraged. In example (30), the scope of negation makes it clear that we no longer deal with a ‘true’ depictive (in contrast to example (25) above). The only possible interpretation of this sentence is that the subject didn’t leave at all, being outraged. Because of its prosodic detachment outraged here is no longer part of the same clause (and proposition) as leave and hence it is not a secondary predicate but rather constitutes a reduced clause of its own. To conclude—strong free adjuncts are another type of participantoriented expression which differs quite clearly from depictives and circumstantials in that it is not part of the same clause as the primary predicate and hence not, strictly speaking, an adjunct (nor a secondary predicate). Weak free adjunct, on the other hand, appears to be simply another term for circumstantial secondary predicates. In the preceding section we suggested that circumstantials/weak free adjuncts are non-focal participant-oriented adjuncts, differing from depictives proper in that they provide presupposed information rather than conveying focal information. In Stump’s discussion of circumstantials/weak free adjuncts, which is concerned with distinguishing these from strong free adjuncts rather than from depictives proper, the major characteristic of this type of adjunct is its interaction with an operator in the main clause (e.g. a modal element, a frequency adverb). Mu¨ller-Bardey (Ch. 3, this volume) adopts basic insights from Stump’s analysis to the analysis of depictives (in the broad sense) and recasts it in terms of the so-called partition of the clause. In this analysis, the operator divides the clause into two parts, with one part functioning as a restrictor for the operator and the other part functioning as its nuclear scope. Depictives (in the broad sense) can occur in either part. When they occur as part of the nuclear scope (as in George usually 8

For a spontaneous example, see ex. (18) in Schultze-Berndt and Himmelmann (2004: 68).


23

drives home drunk), they are depictives proper in our terminology. When they occur as part of the restrictor (as in Drunk, George drives very dangerously where drunk restricts the occurrences of George’s driving dangerously), they are circumstantials in our terminology. Thus, there appear to be two different, not necessarily overlapping properties distinguishing between depictives proper and circumstantials/weak free adjuncts. On the one hand, the distinction between depictives proper and circumstantials appears to correlate with being in focus vs. being in the presupposition (as proposed in the previous section). On the other hand, it also appears to correlate with being in the restrictor vs. being in the nuclear scope, as suggested by Mu¨ller-Bardey’s analysis. A further exploration of these relationships is not a straightforward matter, as it involves a fair number of complex issues. To begin with, it is not clear whether the focus/ presupposition and the restrictor/nuclear-scope distinctions themselves differ substantially (regardless of how they relate to depictives) or whether they are just different ways of making essentially the same distinction. Mu¨ller-Bardey (Ch. 3, this volume, sections 3.2.3 and 3.2.4) argues that they are not, although examples where items are distributed across non-matching categories (i.e. an item in the restrictor is in focus or an item in the nuclear scope is presupposed) are hard to come by (but see Mu¨ller-Bardey’s example (47) and discussion). Assuming that focus/presupposition and restrictor/nuclear scope actually refer to substantially different distinctions, one relating to information structure, the other to logical form, it then becomes a rather intricate empirical problem to determine whether they make different predictions with regard to the distinction between depictives proper and circumstantials. One major difference here is that the restrictor/nuclear-scope distinction requires the presence of an operator inducing the partition of the clause, which in turn predicts that circumstantials only occur in clauses which include such operators (recall from section 1.2.3 above that circumstantials generally involve some additional semantic link between secondary and main predicate which we broadly characterized as condition, concession, or temporal).9 We are currently not in a position to pursue this issue any further. In fact, it is not clear to us whether there is much to be gained by further pursuing it, since for practically all uncontroversial examples of either depictives proper 9 So far, we have not come across any reasonably natural examples which would allow us to resolve this issue. A relevant example would have to look something like ??Fresh we drank orange juice neat that night, where fresh would have to be presupposed information (see also Simpson, Ch. 2, this volume, section 2.2.5).

24


or circumstantials known to us, the two properties match perfectly: depictives proper are both in focus and in nuclear scope, circumstantials are presupposed and in the restrictor. To simplify the ensuing exposition, we simply assume that being non-focal is the major and defining feature of circumstantials. 1.2.5 Depictives vs. predicative complements For some authors (including Halliday 1967; Nichols 1978a; Napoli 1989) the notion of (depictive) secondary predicate also includes constructions where the second predicative element is not an adjunct but a complement, i.e. where it is in some sense selected by the main predicate, and is therefore obligatory, as in (31) (from Nichols 1978a: 114). (31) a. They elected him president. b. Rocks serve them as support. It cannot be denied that there are certain functional and formal relationships between predicative complements and depictives. Functionally, both are obviously participant-oriented and convey partially independent predications about one participant of the main predicate. Moreover, a secondary predicate can rarely be selected completely independently of the main predicate (see also section 1.4.1 below). Simpson (Ch. 2, this volume) in fact argues that in a language like English, these restrictions are so strong that the depictive elements in standard depictive examples such as They returned drunk may have to be analysed as complements rather than adjuncts. Even if one does not want to follow Simpson’s argument for English, it is clear that once one acknowledges that there are at least weak selectional restrictions between main and secondary predicates, the question of distinguishing depictives and predicative complements becomes a complex issue. In fact, given the close surface similarities between predicative complements and depictive adjuncts, distinguishing them would appear to be just another instance of the well-known problem of distinguishing complements and adjuncts in general (see Dowty 2003 for a recent survey and discussion of complement/adjunct similarities and historical transitions between them). Formally, in many (though by no means all) languages, predicative complements and depictives may have the same internal constituency and the same morphological marking (if any). Examples of the latter include the predicative marker as in English, illustrated in (31b), the Russian instrumental (van der Auwera and Malchukov, Ch. 13, this volume) and the predicative marker ¼ i in Ewe (Ameka, Ch. 11, this volume). Plank (1985) in fact advances the hypothesis that depictive constructions are always derivative of


25

predicative complement constructions, i.e. that they always exhibit the same structure and marking as predicative complement constructions. However, Bucheli Berger (Ch. 4, this volume) provides counter-evidence from Swiss German dialects with a generalized depictive marker which is not found on predicative complements. Nevertheless, in those instances where predicative complements and depictives have the same internal constituency and the same morphological marking, one may wonder whether we are dealing with distinct construction types. The answer depends very much on whether selectional restrictions between the main predicate and its complements are formally manifest in a given language. Of major importance in this regard is clear-cut evidence for the obligatoriness of some argument expressions but not others. It would appear that once again this is a grammatical domain where there is ample room for crosslinguistic variation. Simpson’s comparison of Warlpiri and English adjuncts and complements in Chapter 2 illustrates one fairly clear case of such variation. See also Ogawa (2001: 9), who claims that the multiple object construction in Korean is a depictive secondary predicate construction despite the fact that the second object is obligatory. On the basis of the pattern of formal distinction vs. indistinction, van der Auwera and Malchukov (Ch. 13, this volume) conclude that depictives and predicative complements are in principle distinct, but adjacent on their semantic map of construction types. 1.2.6 Summary of basic distinctions regarding participant-oriented adjuncts In this section, we have discussed a number of adjunct types in English which share the characteristics of allowing participant-oriented readings and of denoting a state or condition which temporally overlaps with the state of affairs designated by the main predicate, i.e. depictive secondary predicates, participant-oriented manner adverb(ial)s, circumstantial secondary predicates, weak free adjuncts, and strong free adjuncts. The discussion suggests a somewhat revised set of terminological distinctions, with their crosslinguistic applicability in view. In particular, we propose that the term depictive proper (or depictive in the narrow sense) be reserved for participantoriented adjuncts which are part of the focus domain of a sentence, i.e. which function as focus exponents. Such adjuncts encode a state which contributes a significant characteristic to the main event—for example, leaving drunk or outraged is different from simply leaving (cf. also Geuder 2000: 197). In English, they typically occur in post-predicate position and carry the main sentence stress.

26


Participant-oriented adjuncts outside the focus domain will be termed circumstantials, following Nichols (1978b). As we have argued in sections 1.2.3 and 1.2.4, this category also comprises Halliday’s conditionals and the adjunct type that has been referred to as weak free adjunct in the semantics literature. Whether the distinction between depictives proper and circumstantials is universally applicable is an empirical question; it is questioned for Warlpiri by Simpson (Ch. 2, this volume). For the time being, we propose the term depictives in the broad sense to comprise both depictives proper and circumstantials. In line with much of the recent literature, most contributors to this volume use the term depictive in this broad sense.10 The term depictive will, moreover, be restricted to those participantoriented adjuncts which can be distinguished from event-oriented adjuncts— adverbials—on morphosyntactic grounds. This terminology allows us to describe cases of mismatch between morphosyntactic and semantic properties: a construction type which generally serves to convey event orientation may still receive a participant-oriented interpretation in some instances. For English, a case in point are the transparent adverbs discussed by Geuder (2000), which are formally distinguished from depictives by the presence of the suffix -ly. As briefly mentioned in section 1.2.2 and further discussed in section 1.3, the reverse type of mismatch can also be found: for some languages we can speak of event-oriented depictives, i.e. constructions which usually convey participant orientation, but which may receive a primarily event-oriented interpretation under certain circumstances. The term general adjunct construction was introduced in section 1.2.2 for constructions which are not morphosyntactically marked for either event- or participantorientation. So far, we have only considered participant-oriented expressions which clearly form a constituent of their own within the matrix clause headed by the main predicate. However, there are further morphosyntactic options for participant-oriented expressions, which are not always easily distinguishable from depictives in the broad sense as just defined. The main types here are loosely adjoined reduced clauses, and incorporated expressions, discussed in turn below. 10 In passing, it may be noted that the currently most widespread usage of depictive is actually somewhat narrower than just defined, since the term is generally used only in reference to adjectival or participial adjuncts exhibiting participant orientation. However, this appears to be based on the fact that the most common and straightforward examples of depictive secondary predicates in English and other European languages involve adjectives and participles. As further discussed in section 1.4, there is no systematic basis for limiting depictive status to these two lexical categories. Some of the contributions to this volume restrict their discussion to depictives of this type, though.


27

Loosely adjoined reduced clauses which are participant-oriented in that they share an argument with the matrix predicate have also been termed strong free adjuncts. Unlike depictives, they render a logically completely independent proposition. One indication of this independence is the fact that they are usually prosodically detached from the main clause. Furthermore, reduced clauses are always outside the scope of a negator of the main predicate (as in They didn’t leave, outraged), and impose fewer restrictions on the semantic nature of the participant-oriented expression. Thus, in English, participant-oriented reduced clauses, in contrast to depictives, may contain individual-level predicates like that in (28b). There is, however, some evidence that the restriction of depictives to stage-level predicates may not be universal (which reduces the number of criteria distinguishing depictives from reduced clauses/strong free adjuncts). For example, Simpson (Ch. 2, this volume) points out that individual-level predicates could be analysed as depictives in Warlpiri, e.g. when they function as predicates on the objects of perception verbs. In addition to the cases of participant orientation discussed so far, we may note in passing that in the literature, the term (depictive) secondary predicate has occasionally also been applied to participant-oriented expressions which do not have adjunct status, but are part of the main predicate (i.e., the two jointly form a complex predicate). This issue is taken up again briefly in section 1.4.1; see also Schultze-Berndt and Himmelmann (2004: 69–72) for discussion and illustration. Having introduced the most important terminological distinctions, we are now in a position to review the range of semantic domains for which depictive constructions can be used (section 1.3), as well as the morphosyntactic features of depictive constructions found in different languages (section 1.4).

1.3 The semantic range of participant-oriented adjuncts While, as we have seen in section 1.2, the oriented nature of manner adverbials has received some attention in the literature, there is only occasional mention of the participant-oriented nature of e.g. adjuncts of quantity, concomitance, comparison, or location.11 In this section, we will briefly summarize the evidence, from a crosslinguistic perspective, for the claim that 11 Discussions of the Latin praedicativum are a major exception, in that they tend to include a detailed listing of the different semantic domains for which secondary predicates can be used. In fact, in works such as Menge (2000: 354–9) roughly the same kind of semantic domains are mentioned as in the following sections (thanks to Utz Maas for reminding us of this very useful source).

28


expressions that have traditionally been assigned to the realms of ‘adverbials’ and ‘depictives’, respectively, in fact form part of the same semantic domain, in that they all share the feature of participant orientation (as well as event orientation), albeit to different degrees. Such evidence, we will argue, comes from the fact that the encoding of these adjuncts, at least in some languages, differs in no way from that of prototypical depictive secondary predicates (such as fresh in example (1) above), which express a physical state or condition that holds of one of the participants of the main predicate at the time of the main event. Such ‘prototypical depictives’, in many languages, are rendered by general adjunct constructions which allow for both participantand event-oriented readings (see section 1.2.6). For languages where participant orientation and event orientation are distinguished by morphosyntactic means, the resulting two construction types—depictives in the broad sense and adverbials—are in competition within the semantic domain outlined here. Building on the hierarchy ranging from prototypical participantoriented expressions to prototypical event-oriented expressions in SchultzeBerndt and Himmelmann (2004: 119–23), but incorporating the more complex findings presented in the contributions to this volume, we propose in Figure 1.1 a (very tentative) semantic map for this semantic domain.12 The expression types included in the map will be briefly discussed in sections 1.3.1–1.3.14 below. In section 1.3.14, the semantic map will be tested against four of the languages represented in this volume, the Appenzell dialects of Swiss German (Figure 1.2), Georgian (Figure 1.3), Warlpiri (Figure 1.4), and Shipibo-Konibo (Figure 1.5). The semantic map, as presented here, centres around the most prototypical participant-oriented adjuncts, i.e. those encoded as depictives in all languages that have depictive constructions at all. As mentioned above, these are adjuncts conveying a physical condition or state. Since this adjunct type has already been discussed extensively in the preceding sections, it is not included in the following review. For expository reasons, the order of presentation of the semantic categories in the following subsections also does not follow the semantic map in all respects. The arrangement of the other expression types on the semantic map is designed to reflect, first and foremost, their likelihood of depictive encoding— a semantic domain that is encoded as a depictive adjunct in more of the languages surveyed by us is placed closer to the centre. The dotted lines connecting various expression types should be read as hypothesized paths of 12 For the methodology of semantic maps, see van der Auwera and Malchukov (Ch. 13, this volume), and references therein.

Concomitance (with a dog)

Intention (wanting to be beautiful)

Benefactive/ Malefactive (for/on you)

Resultant state / anterior event (drunk ; having burnt the field)

Association (with a hat) Distributivity (three by three)

Collective (alone ; as a group of three)

Emphatic pronoun (herself )

Mental/emotional condition (sad ; purposefully)

Physical condition (fresh ; green)

Posture/ configuration (flat ; in a circle)

Purpose (in order to train the dog) Function (purpose) (in order to be skinned) Simultaneous event (grinding grain)

Manner (fast ; carefully)

Comparison (like a horse) Location (of participant) (in the canoe ; from the balcony) Location (of event) (in the camp)

Function (Role) (as a teacher)

Life stage (as a child) Atmospheric condition (in the dark)

Order ([as the] first)

Time (in the morning ; yesterday)

Figure 1.1 A tentative semantic map for participant-oriented expressions

Frequency (twice)

30


context expansion of depictive marking (alternatively, they can of course also be read as paths of context expansion of adverbial marking encroaching on the core domain of participant orientation, but this possibility will not be further discussed below). The expression types enclosed in a box connected to the centre by straight lines are envisaged as located on a separate plane, i.e. in a third dimension of the semantic map; these are the expression types which are most likely to be encoded by clausal adjuncts, with somewhat different properties from the other adjuncts (see further sections 1.3.12 and 1.3.13). Semantic domains found adjacent to each other in the semantic map are likely to receive the same type of morphosyntactic coding, whether in a depictive construction or in an adverbial or general adjunct construction. An example of the latter would be the construction involving the marker als ‘as’ in German, which covers expressions of life stage (‘as a child’), function and role (‘as a teacher’), and order (‘as the first’). However, due to its twodimensional nature, the map cannot always adequately represent semantic closeness of this type, and no attempt has been made to visually represent potential identity of marking other than by spatial contiguity. The major evidence that we draw on in this section is the semantic range of genuine depictive constructions (in the broad sense), which, as the contributions in this volume clearly show, can vary considerably. Depictive constructions usually display some form of agreement between the depictive adjunct and its controller (see section 1.4.5.1). Alternatively, there is some other clear formal marker of the construction such as the dedicated depictive marker in a few Swiss German dialects. Constructions showing agreement, in our view, constitute the paradigm case of participant-oriented adjunct constructions, because the predicative relationship between controller and adjunct is overtly reflected in their morphosyntactic structure. Although evidence from agreement is not always unequivocal,13 in all of the examples of relevance here participant orientation provides the most straightforward explanation of the agreement facts. 1.3.1 Mental or emotional condition Expressions of a mental or emotional condition such as sad(ly), angry/angrily, or ravenous(ly), when appearing with adverbial marking, belong to those expressions typically subsumed under ‘oriented adverbials’. The use of this term already suggests that, at least when predicated of the subject, expressions of this type not only convey information about the internal state or condition 13

See Schultze-Berndt and Himmelmann (2004: 82–4) for further discussion.


31

of a participant, but are also event-oriented in that they almost inevitably make a statement about the manner in which an event is performed, which after all is often the only clue to the internal state of a participant (see section 1.2.2 for a detailed discussion). This distinguishes expressions of a mental or emotional condition from expressions of a physical condition, which are found at the centre of the semantic map in Figure 1.1. The latter are more exclusively participant-oriented because the time-stable nature of the physical state or condition means that it is less likely to be perceived as influencing the manner of the event. Needless to say, the boundary between physical and mental condition is not always clear-cut: consider e.g. concepts like ‘hungry’. This also holds for the other semantic categories discussed. The prediction made here is that there will be languages which distinguish expressions of physical condition from expressions of mental or emotional condition by encoding the former as depictives and the latter as adverbials. While we have not yet found an example of a language where this distinction in encoding is consistently made, there are a number of languages which behave like English in that expressions of mental or emotional condition, unlike expressions of physical condition, allow for a variation between depictive and adverbial coding, focusing on either the participant-oriented aspect or the event-oriented aspect of the expression; other relevant examples are Georgian (Boeder, Ch. 6, this volume), and Latin (see e.g. Menge 2000: 354f.). Finally, there should be languages where mental or emotional condition is consistently encoded by a depictive construction. Shipibo Konibo (see Valenzuela, Ch. 8, this volume, ex. (64)) and many Australian languages including Warlpiri appear to be languages of this type (cf. Hale 1982; Simpson 1991; for an example, see Simpson, Ch. 2, this volume, ex. (29b)). 1.3.2 Posture and configuration Like expressions of a mental or emotional condition, expressions of posture such as backwards and expressions of configuration such as in a circle describe a property of a participant. At the same time, when the posture is that of the agentive participant, it may alter significantly the way in which an event is conceived. Semantically, therefore, these expressions are situated between expressions of a physical condition (the prototypical depictives) and manner expressions, and have been represented in this way in the semantic map in Figure 1.1. The formal evidence available to us so far seems to support this intermediate position, in that expressions of posture and configuration can oscillate between adverbial and depictive coding (in languages where the distinction

32


between the two construction types exists).14 Examples are provided by Bucheli Berger (Ch. 4, this volume) and Boeder (Ch. 6, this volume, section 6.4.1); in both of the languages discussed by these authors—the north-eastern dialects of Swiss German, and Georgian, respectively—a subgroup of posture expressions are coded as depictives, while the remaining posture expressions and prototypical manner expressions (e.g. of speed) are coded as adverbials. 1.3.3 Manner Even the most prototypical manner adjuncts usually associated with event orientation, e.g. expressions of speed such as quickly and slowly, are not distinguished from depictives in their morphosyntactic coding in all languages. For example, manner expressions may be unambiguously marked as depictive by agreement with their controller. Relevant examples can be found in Martuthunira (see ex. (20)), Warlpiri (Simpson, Ch. 2, this volume, ex. (20)), Diyari (see ex. (48)), and Shipibo-Konibo (Valenzuela, Ch. 8, this volume). In some Omotic languages, obligatorily controlled converb constructions (see section 1.4.4.5), identified as genuine depictive constructions by Amha and Dimmendaal (Ch. 9, this volume), are used to encode both physical condition and manner (see in particular their ex. (36b)). In Ewe (Ameka, Ch. 11, this volume), the predicative marker ¼i found on nominal depictives and some expressions of physical condition occurs—albeit in a more strongly grammaticalized form—on expressions of manner as well. Van der Auwera and Malchukov (Ch. 13, this volume) discuss examples from several other languages where manner adverbials and depictives share major formal characteristics, e.g. Basque and the Tungusic language Even. In some languages, e.g. Italian (Napoli 1975), Turkana (Amha and Dimmendaal, Ch. 9, this volume), and Shona (Gu¨ldemann, Ch. 10, this volume), manner is a semantic domain where adverbial and depictive coding alternate with the same expression, or positive and negative versions of the same expression, with often only very subtle semantic differences. A similar variation can be observed between corresponding expressions in closely related languages (for examples see Amha and Dimmendaal, Ch. 9, this volume, and Boeder, Ch. 6, this volume). This further supports the hypothesis that expressions of manner are simultaneously event-oriented and participantoriented, and that one or the other component can be ‘highlighted’ by the morphosyntactic characteristics of the construction involved. 14 In serializing languages, posture is often encoded by a verb in a verb series (see Ameka, Ch. 11, this volume, and Enfield, Ch. 12, this volume, for examples), and thus falls outside the domain of adjuncts as defined here.


33

1.3.4 Comparison Closely related semantically to expressions of manner are expressions of comparison (similatives) such as like a horse in He eats his food like a horse. From a semantic point of view, adjuncts of this type clearly have a participantoriented component: the similarity to another entity is ascribed to one of the participants (generally the agent; in the example above it is the subject and not the food which is compared to a horse). At the same time, the comparative expression also says something about the manner in which the event (in this case, the eating) is performed, and in this respect it is event-oriented. From a crosslinguistic perspective, we therefore expect a similar variation in coding to other manner adverbials. This expectation is indeed borne out. As Schultze-Berndt and Himmelmann (2004: 112–13) show, in languages where depictive secondary predicates denoting condition or state agree with their controller in case, similative expressions also tend to show agreement. This is illustrated in (32) for the Australian language Gooniyandi. A paraphrase that captures the relationship between the similative adjunct and the agent is ‘the woman fights being like a man (at the time of fighting)’. Gooniyandi (32) goornboo15 thirri gardboowoona yoowooloo-jangi-ngga woman fight she:belts:them man-simil-erg ‘the woman fights like a man’ (McGregor 1990: 346) The participant-oriented nature of similative expressions is not always reflected formally, though. In Georgian, for example, secondary predicates of condition or state, but not similative expressions, show case agreement with their controller, although the latter still show semantic agreement in number (Boeder, Ch. 6, this volume). As Haspelmath and Buchholz (1998: 322) point out, in many languages expressions of comparison are formally related to expressions of role or function (see section 1.3.5). This is true, for example, for Russian, where both expression types are marked with instrumental case, and for Ewe, where both may take the connector abe´ ‘like’ and/or the predicative marker ¼i (Ameka, Ch. 11, this volume). For this reason, expressions of comparison are placed between manner expressions and expressions of function/role in the semantic map.

15 The NP representing the controller, goornboo ‘woman’, here does not carry the ergative marker, since ergative marking of agents is optional in Gooniyandi (cf. McGregor 1992b; 1998b).

34


1.3.5 Function and role A typical example of an adjunct of function or role is as a present in They gave him the book as a present, which semantically clearly relates only to the object (the theme of the transfer), and neither to the subject (the agent) nor to the indirect object (the recipient). Expressions of function/role have therefore been adduced as typical examples of participant-oriented adjuncts (see Nichols 1978a). This analysis fits in with the view expressed by Stump (1985) that in English, the marker as (and presumably its translation equivalents) serves to convert individual-level predicates—which cannot function as depictives—into stage-level predicates, i.e. expressions of a temporary state (see also section 1.2.4 above). Arguably, however, expressions of function or role also have an event-oriented component (e.g. as a present above can also be interpreted as making a statement about the manner of presenting). Thus, in Georgian, where depictives and adverbials are distinguished by the presence vs. absence of agreement, expressions of function and role receive adverbial coding (Boeder, Ch. 6, this volume), and good examples of depictive coding of these expressions are surprisingly hard to come by. In some languages, expressions of function and role carry special predicative markers such as the so-called ‘essive’ case (see section 1.4.5.2). In other languages, they take a copula or auxiliary verb, just as they would as main predicates (cf. Enfield, Ch. 12, this volume, for Lao; Gu¨ldemann, Ch. 10, this volume, for Shona; and Amha and Dimmendaal, Ch. 9, this volume, for Turkana). They may also be marked with an instrumental or locative case or adposition, and/or share their marking with expressions of comparison (see section 1.3.4) or expressions of life stage (see section 1.3.6); this is the case e.g. in English, German, Russian, Finnish, and also in Ewe (Ameka, Ch. 11, this volume). Most of these constructions probably have to be regarded as general adjunct constructions. Although participant-oriented adjuncts of role or function are fairly frequent in English and many other languages, there appear to be languages which cannot express these functions by an adjunct, but which have to resort to biclausal constructions. One of these is Laz (Kutscher and Genç, Ch. 7, this volume). 1.3.6 Life stage Adjuncts of life stage such as as a young girl in (24) belong to the standard examples of depictive or circumstantial secondary predicates in the literature (see section 1.2.3). Their participant-oriented nature may be reflected in their formal marking, as in Georgian (Boeder, Ch. 6, this volume).


35

In other languages, however, either the predicative nature (see the remarks in section 1.3.5 above) or the temporal nature of life-stage expressions is coded in preference over their participant-oriented nature; expressions of life stage are therefore placed between prototypical participant-oriented expressions and temporal expressions in the semantic map. Schultze-Berndt and Himmelmann (2004: 121) have found that in many Australian languages with a strict requirement of case agreement of participant-oriented adjuncts with their controller, expressions of life stage receive temporal or locative marking and thus fall outside the domain of (formally marked) participant-oriented adjuncts. The same is true for a subset of life stage expressions in ShipiboKonibo (Valenzuela, Ch. 8, this volume), while others carry the formal markers of participant orientation. In this language, therefore, life stage expressions exhibit variation between depictive and adverbial coding. 1.3.7 Quantification and order Numerals and other quantifiers in adjunct position also often exhibit a semantic relationship to one participant of the main predicate, i.e. participant orientation. The clearest case can be made for quantificational adjuncts in a collective reading, which indicate the size of the set of entities involved as a participant in an event. An example is alone in Mark ate the cakes alone again; the adverb here only indicates the quantity of the subject, not of the object. Formally, collective quantificational adjuncts, in particular numerals expressing the cardinality of a set of participants, are often rendered by a distinctive construction which only applies to these adjuncts and which can arguably be regarded as a depictive construction (see Schultze-Berndt and Himmelmann 2004: 107–9). McGregor (Ch. 5, this volume) provides a detailed survey of constructions involving participant-oriented quantificational adjuncts in several Australian languages. The characteristic formal property of some of the constructions he discusses is agreement, making these adjuncts clearly depictive. Further examples of agreeing quantificational adjuncts are provided by Boeder (Ch. 6, this volume) for Georgian. Distributive quantificational adjuncts, or, in McGregor’s terms, adjuncts expressing iterated co-participation (as in The schoolchildren walked two by two/in pairs) could also be regarded as participant-oriented. From a crosslinguistic perspective, however, there seems to be less evidence for a formal manifestation of participant orientation in this type of expressions, which very frequently consist in reduplicated forms of a numeral (for more discussion, see McGregor, Ch. 5, this volume, and Kutscher and Genç, Ch. 7, this volume).

36


Ordinal numerals in adjunct function, as in She wanted to arrive home first, are also arguably participant-oriented. In some languages, such as German, these expressions share their marking with expressions of function/role and life stage (see sections 1.3.5 and 1.3.6). For Latin, ordinal adjuncts, displaying agreement with their controllers, are presented as a clear case of praedicativa, i.e. depictives (e.g. Menge 2000: 356). The final expression type involving quantifiers to be considered here consists of expressions of frequency such as twice. While modifying the event as a whole rather than being participant-oriented, and therefore found at the margin of the semantic map in Figure 1.1, in Shipibo-Konibo they receive the type of depictive encoding also found in expressions of manner (Valenzuela, Ch. 8, this volume). 1.3.8 Emphatic pronouns Emphatic pronouns in adjunct position can be illustrated with the example She drove the truck herself. Trivially, there exists a relationship of coreference between the emphatic pronoun herself and the subject of the sentence. Expressions of this type could thus also be regarded as participant-oriented adjuncts. One reason for adopting such an analysis is their close semantic relationship to participant-oriented quantifiers (since e.g. herself in the example above can be interpreted as alone); see McGregor (Ch. 5, this volume) for a discussion of this issue. The morphosyntactic facts of Panoan languages provide an additional argument. In Shipibo-Konibo, agreement between the emphatic pronominal and the participant of the main clause does not just manifest itself in shared values for person, number, or case; rather, emphatic pronouns display the type of agreement specifically restricted to participant-oriented adjuncts (i.e. depictives), and this is therefore the analysis that Valenzuela (Ch. 8, this volume) adopts. If we accept this evidence, emphatic pronouns may well be considered as very close to prototypical participant-oriented adjuncts, which is where they have been placed in the semantic map in Figure 1.1. 1.3.9 Concomitance and association Expressions of concomitance—marked with a comitative or a comparable case or adposition—are generally regarded as adverbials. However, some authors (see in particular Frey and Pittner 1998 and Pittner 1999: 101) have noted that they exhibit a special relationship to one of the participants of the main predicate, in that the latter is said to be accompanied by the referent of the comitative phrase. This can be illustrated with the example They brought


37

in the prisoner with his accomplice, where the comitative phrase is interpreted with respect to only the object, not the subject. Still, the participant-oriented nature of the adpositional phrase in English is not obvious from its encoding. For this reason, the comitative phrase here is probably more appropriately characterized as a general adjunct construction. Evidence for an analysis of comitative expressions as participant-oriented adjuncts comes from languages where they show agreement with the noun phrase whose referent is the participant that is accompanied. In (33), from Warlpiri, the comitative expression is in addition ergative-marked in agreement with the agent (see Schultze-Berndt and Himmelmann 2004: 110f. for further examples). Warlpiri (33) kurdu ka ngurlu kipi-rni karnta-ngkajinta-rlu child prs mulga.seed winnow-npst woman-com-erg ‘The child is winnowing mulga seed with the woman.’ (Hale 1982: 272) Similarly, in Shona (Gu¨ldemann, Ch. 10, this volume), accompaniment is expressed by a converbal construction in the same way as other participantoriented adjuncts. However, this formal manifestation of participant orientation is not found in all of the languages with dedicated depictive constructions. An example is Georgian (Boeder, Ch. 6, this volume), which has participant-oriented expressions of concomitance which are marked with instrumental case rather than showing agreement. Interestingly, many languages appear to make a formal distinction between expressions of concomitance or accompaniment where the concomitant has a certain degree of independence (i.e. it is typically animate) and expressions where the concomitant is more closely associated with, and less independent of, the controller. A prototypical example of such an associative relation is adornment (e.g. with a hat), although any inanimate entity carried in close physical contact, and sometimes also a person’s children, may fall under this category. Compare the Warlpiri example in (33) with the one in (34); the associative relationship in the latter is indicated by a different marker glossed traditionally as ‘proprietive’. (34) kurdu ka-rla ngarrka-ku rdanpa-rni kuyu-kurlu-ku child prs-3sg.io man-dat accompany-npst meat-prop-dat ‘The child is accompanying [the man]i [with meat]i.’ (Hale 1982: 277) While in Warlpiri, as these examples show, both comitative- and proprietive-marked adjuncts exhibit case agreement with a controller, in

38


Shipibo-Konibo participant agreement is restricted to proprietive expressions (which have a similar semantic range to the Warlpiri ones), but in some other Panoan languages it also extends to comitatives (Valenzuela 2003 and Ch. 8, this volume). From a semantic point of view, too, it makes sense to regard associative expressions, where the association between the accompanying and the accompanied entity is more direct, as more strongly participant-oriented than expressions of concomitance. Further evidence comes from the fact that if a distinction between the two construction types is made, it is the marker signalling an associative relationship which is also often involved in the marking of expressions of mental or emotional condition, manner or posture. ‘Association’ is therefore singled out as a separate category in the semantic map in Figure 1.1, and placed in between expressions of physical condition and expressions of concomitance. This constitutes a refinement and, to some extent, revision, of the hierarchy proposed in Schultze-Berndt and Himmelmann (2004: 119–23), where expressions of concomitance are regarded as a single category which is ascribed a relatively high degree of participant orientation. 1.3.10 Location and direction It has occasionally been noted in the literature that locative expressions are not necessarily exclusively event-oriented, i.e. indicate the location of an event as a whole. Rather, they can also have a participant-oriented reading in that they indicate the location of only one of the participants in an event, as in (35) (see e.g. Mu¨ller-Bardey 1990; Maienborn 2000; 2001; Takezawa 1993: 55). (35) The cook prepared the chicken in a Marihuana sauce. (Maienborn 2000: 155) As demonstrated by Maienborn (2000; 2001), these two cases are distinguished by syntactic position in languages like German (see also the discussion in Mu¨ller-Bardey, Ch. 3, this volume), as well as by the fact that participant-oriented locative phrases are substituted by the manner interrogative, not the locative interrogative. It is not surprising, therefore, that participant-oriented locative phrases occur in depictive constructions. In Warlpiri, for example, participantoriented locative phrases show agreement with their controller (see examples (18) and (19b) in Simpson, Ch. 2, this volume). With verbs of ‘emission’, e.g. speaking or throwing, a source (or ‘ablative’) expression can also have a participant-oriented interpretation and indicate the location of the ‘emittor’ or ‘source’, as in The boy threw stones from the


39

roof. Examples of this type are discussed for Laz by Kutscher and Genç (Ch. 7, this volume) and for Panoan languages by Valenzuela (Ch. 8, this volume). Finally, directional phrases can also be participant-oriented. An example from Walmatjarri, with agreement of the directional phrase, is given in (36). Being directed towards the sandhill is conceptualized here as a temporary condition of the participant who is moving. Walmatjarri (36) jilji-karti-rlu ma-nya wanyjani yapa-warnti nganpayi-rlu sandhill-all-erg aux-3pl.o leave:pst child-pl man-erg ‘The man left the children on his way to the sandhill.’ (Hudson 1978: 35-6) While in the preceding examples the agreement facts appear to have a semantic basis, reflecting the participant-oriented nature of the adjunct, there are a number of languages where agreement of locative phrases is generalized in that it occurs in all instances, even if the location is that of the event as a whole (including all participants) rather than just that of a single participant. Typically, the participant triggering agreement in this case is the agentive, and presumably most prominent, participant, as in the Warlpiri example in (37); see also Valenzuela (Ch. 8, this volume) for relevant examples from ShipiboKonibo. Warlpiri (37) ngarrka-ngku man-erg ‘The man is and Bresnan

ka jarnti-rni karli ngurra-ngka-rlu prs trim-npst boomerang camp-loc-erg trimming the boomerang in the camp.’ (Simpson 1983: 57)

Similarly, in the East Georgian mountain dialect Tush, agreement seems to have been generalized to all expressions of a spatial source (see Boeder, Ch. 6, this volume). We consider these cases as instances of overgeneralization of an originally more limited agreement pattern. Consequently, location of a participant and location of an event are distinguished in the semantic map in Figure 1.1, with the former closer to the core than the latter. 1.3.11 Time and atmospheric condition Temporal expressions constitute the perhaps most puzzling case of expressions whose formal make-up, in at least some languages, suggests an analysis as participant-oriented adjuncts—even though temporal expressions are generally considered a paradigm case of event-oriented adverbials. Still, it is

40


well known that they may agree with a participant of the main predicate in Old Greek and in Latin (Menge 2000: 356f.; Pinkster 1988). In several Australian languages, too, including Warlpiri and Yankunytjatjara, a subset of temporal nominals show optional case agreement with the subject (A or S argument) of the clause (Hale 1982: 281; Goddard 1985: 256–9; SchultzeBerndt 2006: 203; see also ex. (21) in Simpson, Ch. 2, this volume). As suggested in Schultze-Berndt and Himmelmann (2004: 119), we consider the depictive coding of time expressions as an overgeneralization of a depictive construction beyond its core semantic domain. Depictive encoding is always ‘exceptional’ in the sense that to our knowledge there is no language where all temporal adjuncts regularly agree with another constituent in the matrix clause. Rather, this type of agreement is usually limited to a subset of temporal adjuncts and often optional. Time expressions are therefore included in the semantic map shown in Figure 1.1, albeit at the very margin. Interestingly, Bucheli Berger (Ch. 4, this volume, ex. (47)) reports that the depictive marker in the north-eastern Swiss German dialects is marginally acceptable with expressions of atmospheric circumstance such as ‘dark’ in a sentence such as They went home (in) dark(ness), which can be interpreted as a metonymic reference to a time of the day, and hence as event-oriented. On the other hand, these expressions seem to be more strongly participantoriented than other temporal expressions, since the participant involved may be affected directly by an atmospheric circumstance such as darkness (consider also the discussion of a similar Latin expression in Boeder, Ch. 6, this volume, ex. (126)). We very tentatively propose here that such expressions may provide a link between more prototypical participant-oriented expressions and temporal expressions; they are therefore placed between the two in the semantic map in Figure 1.1. 1.3.12 Resultant state (anterior event) and simultaneous event In Schultze-Berndt and Himmelmann (2004: 98–106), we argue at length that event nominals, converbs or participles, and consequently also clauses headed by these, can have depictive status if they are obligatorily controlled by a participant of the main clause. This means that the subject of the subordinate clause is unexpressed and understood as coreferential with a participant of the main clause. This and the following subsection are devoted to semantic expression types most likely to be encoded as clausal constructions. A particularly clear case for depictive status can be made if these constructions either show agreement with a controller or involve what is generally termed ‘switch-reference marking’, which usually involves at least a distinct marker (‘same-subject marking’) for subject orientation of the clause (see further


41

sections 1.4.4.5 and 1.4.5.1). Because the latter constitutes a specific coding property, expressions of this type are represented on a different plane in the semantic map; this issue is taken up again in section 1.3.14. In many languages, including several discussed in this volume, converbs with overt same-subject marking display a contrast in taxis or ‘relative tense’. Wolaitta and Maale (Amha and Dimmendaal, Ch. 9, this volume) distinguish anterior forms (expressing an event that has been completed prior to that of the main predicate) and simultaneous forms (expressing an event simultaneous to that of the main predicate). Shipibo-Konibo (Valenzuela, Ch. 8, this volume), in addition, also has subordinate clauses displaying subsequent marking. The taxis system in Shona is even more complex (Gu¨ldemann, Ch. 10, this volume). We will discuss anterior and simultaneous expressions before turning to subsequent expressions. Anterior converbs, ‘past participles’, or ‘resultative participles’, and clauses headed by these, encode a state resulting from a completed event, and are semantically very close to ‘prototypical’ depictives (compare the participial drunk with the adjective sober). In all the languages with dedicated depictive constructions surveyed by us, expressions of this type in fact receive depictive encoding. Examples can be found in several contributions to this volume, including Amha and Dimmendaal (Ch. 9), where the relevant examples are those for anterior same-subject converbs in Wolaitta and Maale. For further examples and discussion, see the semantic maps and references for the north-eastern Swiss German dialects, Georgian, Warlpiri, and Shipibo-Konibo in section 1.3.14, as well as Schultze-Berndt and Himmelmann (2004: 103–6). The following Warlpiri example illustrates agreement of the simultaneous converb with the controller in ergative case. Simultaneous event expressions straightforwardly fulfil the criterion of temporal overlap between a depictive and a main clause: the participant of the main clause (here, the man) is said to be (in the state of) running at the same time as being involved in the event encoded by the main clause (seeing a kangaroo). Warlpiri (38) wati-ngki marlu nya-ngu parnka-nja-karra-rlu man-erg kangaroo see-pst run-cvb-ss-erg ‘The man saw the kangaroo while [the man was] running.’ (Hale et al. 1995: 1442) Other examples of simultaneous same-subject clauses with depictive coding can be found in the contributions by Valenzuela (Ch. 8), Amha and Dimmendaal (Ch. 9), and Gu¨ldemann (Ch. 10).

42


1.3.13 Subsequent event It appears that a subsequent event is less readily attributed to a participant of the main clause (in terms of formal correlates of participant orientation) than an anterior or simultaneous event, judging from the finding that dedicated marking of participant orientation is more frequently found on anterior and simultaneous expressions than on subsequent expressions. In a subsection entitled ‘Prospective deverbal depictives’, Schultze-Berndt and Himmelmann (2004: 104–6) demonstrate that subsequent expressions, at least in the Australian languages surveyed for that paper, are more likely to receive depictive coding if they specifically convey an intention, rather than subsequent taxis in general. In other words, it is the intention as a pre-state of the event, not the event itself, that is predicated of the controller. A specialized converbal form of this type, the preparatory purposive form in Warlpiri, is illustrated in example (10d) in Simpson (Ch. 2, this volume). In the semantic map in Figure 1.1, it is proposed (in the form of a link in the semantic map) that depictive coding may spread from expressions of intention to purposive or subsequent expressions in general. An example of a language with generalized agreement marking on subsequent expressions appears to be Shipibo-Konibo (Valenzuela, Ch. 8, this volume; see also Figure 1.5 below). A third type of subsequent event, conceptualized as a pre-state and attributed to a participant, has to be identified in order to account for the agreement facts of Latin and Georgian (see Boeder, Ch. 6, this volume, ex. (33)). These languages exhibit ‘gerundive’ forms conveying a purpose in the sense of an intended function, as in The bear lay there to be skinned. A comparable example is (39). Martuthunira (39) murna-ngka-nguru warra, ngayu thathu-lalha mulurru close-loc-abl emph 1sg.nom send-pst straight [thanturri-waa yanti-ngka waruul] go.down-purp.o side-loc still ‘From close up, I sent it [the spear] straight [for it] to go down into [the emu’s] side.’ (Dench 1995: 264, l. 7) It is hardly possible to ascribe an intention to the bear in the example above, or the spear in the Martuthunira example in (39). Rather, it is the intended function that is ascribed to the participant, and that explains the formal marking of participant orientation on subsequent clauses of this type.


43

1.3.14 Summarizing the evidence for the semantic map Throughout this section (and in fact throughout this volume), it has become evident that many adjunct types that are widely analysed as adverbials and hence, at least implicitly, as event-oriented, have a participant-oriented semantic component which may or may not be formally reflected in a given language. It has been argued that at a semantic level, participant orientation can be regarded as a feature of adjuncts conveying e.g. a mental or emotional condition, an associative relationship, a quantity, or a life stage, but also of expressions of manner, function or role, and concomitance, and of a subset of locative expressions. As argued in section 1.3.8 above, emphatic pronouns are also good candidates for participant-oriented adjunct status. In the semantic map introduced in Figure 1.1, the expression types just listed are arranged according to their postulated degree of participant orientation. At the margin of the map are found, furthermore, semantic categories such as temporal and frequency expressions which are only rarely encoded as depictives (but rather as adverbials or general adjunct constructions), but which nevertheless should be considered as part of the same semantic domain because of the morphosyntactic facts in some languages.16 Finally, adjuncts representing an event as anterior, simultaneous or subsequent with respect to the event encoded by the main predicate are represented on a separate plane from the other expressions in Figure 1.1. This is because the type of depictive marking involved may have a different origin from that employed on other adjuncts, i.e. in switch-reference (same-subject) morphology (see further sections 1.4.4.5 and 1.4.5.1). As both Valenzuela (Ch. 8, this volume) and Gu¨ldemann (Ch. 10, this volume) argue, same-subject morphology may constitute the main source for depictive coding in a language and may in fact spread to other, intraclausal adjuncts. In other words, formal marking of participant orientation may arise on these constructions independently, without any implicational relationship to depictive marking even on the core semantic category (physical condition) in the lower plane of the map—the constructions in question are represented as a ‘second core type’ of depictive marking, as it were, on a different level. The links between these and other expression types that are nevertheless present in the map can, in this case, be read in at least three ways (and note that no exhaustive representation has been aimed at). First, depictive marking of the switch-reference type as typically found on the second plane may spread to constructions on the first 16 Note also that only the positive version of the various expression types is represented in the map, even though the negative counterparts (e.g. ‘without’ vs. ‘with’ in the domain of concomitance and association) may exhibit interesting differences in behaviour (see section 1.4.1), and would therefore have to be included in a more fine-grained version of the map.

44


plane—at least this is what Valenzuela argues for Panoan languages. Second, clausal constructions involving switch-reference marking may in fact be the major strategy in a language to encode expression types on the lower plane such as location or concomitance (the Shona case, according to Gu¨ldemann). Finally, if in a language a depictive construction involves e.g. case agreement, this marker may be found both on expressions on the lower plane and on the more clausal expressions represented on the upper plane, regardless of whether the latter carry additional markers of restricted reference (e.g. same-subject marking), as in Warlpiri (cf. Hale 1994), or not, as in Georgian (see Boeder, Ch. 6, this volume). Figures 1.2–1.5 illustrate a few test cases for the semantic map outlined in this and the preceding subsections. They represent the findings for those four languages discussed in some detail in this volume which exhibit genuine depictive constructions. These are the Appenzell dialects of Swiss German (Figure 1.2), Georgian (Figure 1.3), Warlpiri (Figure 1.4), and Shipibo-Konibo (Figure 1.5). The prediction tested for these languages is that the semantic range corresponding to the depictive construction is a subset of the domain of participant-oriented adjuncts included in the semantic map, and that this semantic range is represented by a contiguous segment of the map, including the centre (expressions of physical condition). In the maps below, semantic expression types which, when occurring in adjunct position, are encoded as depictives in the language in question are shaded in grey. Partial shading indicates that only a subset of the relevant expression types receives depictive encoding. Hatching represents variation between depictive and adverbial or general adjunct coding. Absence of information on a particular expression type is indicated by a question mark next to that segment in the map. Figure 1.2 represents the expression types on which the dedicated depictive marker is found in the north-eastern (Appenzell) dialects of Swiss German (Bucheli Berger, Ch. 4, this volume). Depictive marking is basically restricted to the core area of the map—adjectives encoding a physical condition, and anterior-resultative participles (with adjectival morphology). The only extensions beyond that core domain are a subset of posture expressions, and, for some speakers, expressions of atmospheric condition such as ‘dark’ (see also section 1.3.11). In Georgian, depictive constructions can be clearly distinguished from adverbial or general adjunct constructions by the presence of case agreement (and sometimes number agreement). Agreement is found on depictive adjectives, participles, numerals, and nouns in the genitive and instrumental case. The range of expression types displaying agreement is shown in Figure 1.3,







?







Frequency (twice)



Function (Role) (as a teacher) Order ([as the] first)


Figure 1.2 Semantic range of depictive marking in the north-eastern (Appenzell) dialects of Swiss German (Bucheli Berger, Ch. 4, this volume)

46


following Boeder (Ch. 6, this volume). Not only expressions of physical condition (often encoded by means of anterior participles), but also expressions of life stage and collective quantification, and a subset of posture expressions and expressions of mental or emotional condition, regularly involve depictive agreement. Variation between depictive and adverbial or general adjunct encoding is found in the domains of function/purpose (expressed by means of ‘future’ or ‘gerundive’ participles) and association. The hatched section in the ‘physical condition’ segment represents the encroachment of a general adjunct marker, the adverbialis case, into the core domain of participant orientation. Finally, agreement on locative constructions is only found in the Tush dialect of Georgian, and is restricted to source expressions. Figure 1.4 shows the semantic range of case agreement on adjuncts in Warlpiri, based on a number of sources including Hale (1982; 1994), Hale et al. (1995), Simpson (1991; Ch. 2, this volume; p.c.). In Walpiri, the semantic range covered by case agreement—interpreted by us as a depictive marker— goes far beyond that seen for Georgian, in that it also includes expressions of manner, association, and participant-oriented location. Furthermore, case agreement is also found in clausal expressions which take specialized ‘complementizer’ or converb markers, and encode simultaneous events, anterior events, and intention or preparation with respect to a subsequent event. Variation between the use of agreeing and non-agreeing expressions has been described for event-oriented location, time, and concomitance. The shaded section of the ‘frequency’ area tentatively represents agreement with the subject on the adverb tarnnga ‘always’ (see example (10e) in Simpson, Ch. 2, this volume). Collective meanings (‘alone’, ‘as a group’) are expressed by preverbs, i.e. as part of a complex predicate, rather than by adjuncts, so the absence of agreement on these expressions cannot be regarded as counterevidence to the semantic map proposed here. As Figure 1.5 shows, the Panoan language Shipibo-Konibo, as described by Valenzuela (2003; Ch. 8, this volume), is another case of a language where formal marking of participant orientation is found in a very large, and sometimes surprising, semantic range, extending to expression types at the margin of the map such as location of a participant, distributive quantification, subsequent and simultaneous event, manner, comparison, and even frequency. Expressions of life stage exhibit variation between depictive and adverbial (temporal) coding (see section 1.3.6). In Shipibo-Konibo, participant agreement is also (marginally) attested on an expression type not discussed so far, termed ‘interessive expressions’ by Valenzuela, which comprises both benefactive and malefactive adjuncts. The motivation for depictive coding of expressions of this type is not






?








Comparison (like a horse) [Source expressions in Tush]

Location (of event) (in the camp)

Location (of participant) (in the canoe ; from the balcony)




Figure 1.3 Semantic range of depictive agreement in Georgian (Boeder, Ch. 6, this volume)

Frequency (twice)













Frequency (twice)




Time (in the morning; yesterday)

Figure 1.4 Semantic range of case agreement on adjuncts in Warlpiri (Hale 1982; Simpson 1991 and Ch. 2, this volume)






? Function (purpose) (in order to be skinned)



Purpose (in order to train the dog)

Simultaneous event (grinding grain)








? ?


Figure 1.5 Semantic range of participant agreement in Shipibo-Konibo (Valenzuela, Ch. 8, this volume)

Frequency (twice)

50


completely clear, but a possible link, indicated in the semantic map, connects ‘interessives’ to expressions of mental or emotional condition, since a benefactive expression can be interpreted as the intention of the agentive participant to benefit the referent of the adjunct expression. Still, this potential motivation cannot be extended to the malefactive use of these expressions. The semantic map proposed here is compatible with the hierarchy that Valenzuela (Ch. 8, this volume) proposes, based on purely language-internal criteria, and according to which emphatic pronouns and life-stage expressions exhibit the greatest degree of participant orientation, and expressions of time and concomitance (which do not receive participant agreement marking) exhibit the greatest degree of event orientation, with expressions of location, manner, quantification, and clausal expressions located in between the two poles. The semantic map is also supported by findings for other languages, such as Latin, or the Australian languages Martuthunira and Yankunytjatjara, which were included in our survey but are not discussed in detail in this volume. Needless to say, the map and the predictions associated with it are still tentative and preliminary in that they remain to be tested on many more genetically and areally diverse languages. In particular, the proposed paths of context expansion of depictive marking would have to be corroborated by historical and comparative research (of the type undertaken by Valenzuela 2003 for Panoan languages).

1.4 Parameters for a morphosyntactic typology of participant-oriented adjuncts In this section, we will discuss five parameters that play a role in a morphosyntactic typology of participant-oriented adjuncts. These are restrictions on the combination of oriented adjuncts and main predicates (section 1.4.1), restrictions on the syntactic function or semantic role of the participant that functions as the ‘controller’ (section 1.4.2), restrictions on the syntactic position of the adjunct (section 1.4.3), the word class and internal structure of the adjunct (section 1.4.4), and its morphological marking (section 1.4.5). As we have seen in section 1.3, languages may have many subtypes of participant-oriented adjuncts, and the parameters do not necessarily align for each subtype. Therefore, establishing a comprehensive typology of participantoriented adjuncts is a huge task which in a volume like this can hardly be begun. Furthermore, in a functionally oriented typology, participant-oriented elements other than secondary predicate and adverbial constructions should also be considered. For example, in many languages the participant-oriented element and the main predicate together form a single complex predicate (see


51

further section 1.4.1). A further alternative is biclausal constructions (see Schultze-Berndt and Himmelmann 2004: 67–9). Thus, in Laz (Kutscher and Genç, Ch. 7, this volume) expressions of function/role are usually not adjuncts, but separate main clauses. In this programmatic survey, we will restrict ourselves to participant-oriented adjuncts, which, however, as already noted, include some types of subordinate clause. 1.4.1 Restrictions on the combination of oriented adjuncts and main predicates Hardly any crosslinguistic research has been undertaken so far with regard to restrictions on the classes of main predicates that occur in constructions with a participant-oriented adjunct. For English, Rapoport (1993: 178) suggests that depictive secondary predicate constructions ‘are restricted to verbs that cause a change in the state or location of their objects’. Rapoport (1999) claims that depictives with object controllers are not possible with activity verbs in English, but Warlpiri appears to be less restrictive (see Simpson, Ch. 2, this volume, for further discussion and examples). Bucheli Berger (Ch. 4, this volume) provides a good overview of possible predicate classes of depictive and main predicate in another rather ‘restrictive’ language, a north-eastern dialect of Swiss German. A number of authors, without stating absolute restrictions on the classes of predicates involved, note certain collocational restrictions between depictives and main predicates occurring with depictives (see e.g. Enfield, Ch. 12, this volume). Nichols (1978a) treats such lexically fixed combinations as a distinct subtype of depictive constructions (‘bound co-predicates’; cf. also Simpson, Ch. 2, this volume). Grammatical correlates of a conventionalized combination of depictive and main predicate have been noted by Kuno and Takami (1993: 130–4). They show that restrictions on Heavy NP Shift with depictives in English, which had previously been attributed to syntactic differences between subject and object depictives, depend on the predictability of the combination of verb and depictive, as illustrated in (40). Heavy NP Shift of the object NP leaves the verb and the depictive in immediate syntactic contiguity, thus making them more like complex predicates, which is only warranted if the combination encodes a conventional event (leaving a party sober as opposed to leaving a party angry). (40) a. ?John left angry [the reception for the ambassador from Ulan Bator]. (Kuno and Takami 1993: 131) b. Mike never leaves sober [parties that he goes to]. (Kuno and Takami 1993: 132)

52


The tendency for the combination of depictive and main predicate to form fixed collocations is also evident in the formal and diachronic relationship between depictive constructions and periphrastic predicates or copular constructions, where the former main predicate takes on an auxiliary or copular function and the former depictive becomes the main predicate (see e.g. Paul 1919: 52; Haspelmath 1995: 43; Nedjalkov 1995: 99–100; Johanson 1995; Hengeveld 1992a: 237–49). Obviously, when the first verb is semantically general but not completely devoid of semantic content, as e.g. the positional verb in The man lay dead in his home for three days, it becomes difficult to distinguish between copular and depictive construction. Some further examples of this type are discussed by Boeder (Ch. 6, this volume). The conventionalized nature of certain combinations may also account for the fact that in many languages, translation equivalents of English depictive constructions are complex predicates. For example, the participant-oriented element may be an incorporated noun or adjective, a construction type not attested in English or in the languages discussed in this book (see SchultzeBerndt and Himmelmann 2004: 69–72 for some examples and discussion). A second possibility is the occurrence of a participant-oriented expression as a preverbal element; an example from Warlpiri can be found in the contribution by Simpson (Ch. 2, this volume, ex. (13)). Alternatively, the translation equivalent of a depictive in English may appear as one of the verb phrases in a serial verb construction, as shown for Ewe by Ameka (Ch. 11, this volume), for Lao by Enfield (Ch. 12, this volume), and for Mandarin Chinese by van der Auwera and Malchukov (Ch. 13, this volume, exx. (25) and (26)). Here, as with copular constructions, the distinction between complex predicates and adjunct constructions may become blurred in some instances, in particular if the language is of the isolating type with no distinction between finite and nonfinite verb forms. Arguments that can be adduced for a more adjunct-like status of one of the verb phrases in a series are its optionality (Enfield, Ch. 12, this volume) and its formal marking (Ameka, Ch. 11, this volume). Notwithstanding the difficulties of distinguishing complex predicates and participant-oriented adjunct constructions in some cases, crosslinguistically speaking the prediction is that the more expected and conventionalized an event is, the more likely it is to be expressed by a complex predicate or an incorporating structure, where the formal expression iconically reflects conceptual closeness. In the extreme case, a highly conventionalized complex event may be expressed by a single predicate. Amha and Dimmendaal (Ch. 9, this volume) briefly discuss the case of highly specific verbs derived from ideophones in both Nilotic and


53

Omotic languages, which, they argue, ‘preempt the need for other strategies rendering a similar content, e.g. the need for a separate manner adverb or secondary predicate’. In Ilokano, the need for a separate participantoriented adjunct may also be circumvented, due to the highly flexible and productive voice morphology which is characteristic of Philippine languages. In the following example, the main predicate (kinilaw) semantically corresponds to a depictive in other languages. Ilokano (41) kinilaw¼da ti sida rls.pv:raw¼3pl.poss art fish ‘They ate the fish raw.’ (lit.: ‘They ‘‘rawed’’ the fish.’) (C. Rubino, p.c.) Note that this strategy is available only for a rather small number of predicates, i.e. those where a conventionalized, culturally well-established practice is being referred to in an abbreviated way and recovered by pragmatics, as it were: the most likely interpretation of ‘raw’ as a main predicate in (41) is as ‘eat raw’. Constructions involving participant-oriented adjuncts, on the other hand, are expected to express more remarkable, unexpected events. This is explicitly commented on by Schroeder (2003) with reference to Turkish and by Gu¨ldemann (Ch. 10, this volume) in his discussion of participant-oriented adjuncts in Shona. Often, the unexpected state of affairs has a negative form. In German, for example, adjectives with the negative prefix un- are frequently found as depictives, and sometimes are actually restricted to this function, while their positive counterparts are not found in adjunct position. Bucheli Berger (Ch. 4, this volume) discusses the form ooggesse ‘uneaten’ in a northeastern dialect of Swiss German. This form, for pragmatic reasons, can take on an agent-controlled interpretation (i.e. ‘not having eaten’). A counterpart from standard German is ungefru¨hstu¨ckt ‘not having had breakfast’, which likewise only appears in depictive function, whereas its positive counterpart cannot be used as an adjunct at all. On the other hand, some examples also point to the tendency for negative states of affairs to be encoded as adverbials, even when their positive counterparts are encoded as depictives. This seems to hold for some manner expressions in Nunggubuyu (see van der Auwera and Malchukov, Ch. 13, this volume, ex. (46)). Boeder (Ch. 6, this volume, section 6.5) also discusses some minimal pairs of this type in Georgian; he in fact concludes that negative properties are less likely to be conceived of as qualities attributable to a participant.

54


1.4.2 Restrictions on the controller A second parameter in the typology of participant-oriented adjuncts concerns restrictions on the syntactic role of the controller (i.e. the participant to which the adjunct relates). For depictive secondary predicates, it is generally assumed that they can be controlled by subjects and objects or, more precisely, by the single core argument of an intransitive predicate (S), the actorlike core argument (A) of a transitive predicate, and the undergoer-like argument (O) of a transitive predicate. It is widely agreed that controllers in English appear to be more or less restricted to subject and object functions (but see below). This finding has sometimes been generalized, and it has been claimed that there are general restrictions either on the syntactic function or on the thematic role of potential controllers, e.g. to theme, agent, and patient (see Williams 1980; McNulty 1988; for counter-arguments to these positions, see Bayer 1997: 210–24 and Mu¨ller 2002: 180–9). However, languages differ considerably with regard to which participants are easily accessible as controllers. Simpson (Ch. 2, this volume) contrasts English and Warlpiri in this respect: in Warlpiri, but not in English, datives and other oblique objects may function as controllers. Examples of depictives controlled by dative-marked indirect objects can also be found in Georgian (Boeder, Ch. 6, this volume). Oblique agents in passive clauses, too, are frequently found as controllers, e.g. in Australian languages like Martuthunira (Dench 1995; for examples see Schultze-Berndt and Himmelmann 2004: 73, and McGregor, Ch. 5, this volume). Finnish also regularly allows oblique controllers (Nichols 1978a: 120–1). Furthermore, one has to be careful in making generalizing statements even for a single language because there appear to be considerable differences with regard to acceptability and usage depending on speaker, context, genre, and medium (spoken vs. written usage). For example, Nichols (1982) reports that the acceptability of depictives with object controllers in Russian varies dramatically depending on factors including the case and topicality of the controller, word order, the presence of other potential controllers, and the involvement of the controller in more than one control relationship, but also on individual preferences. German, just like English, is generally rather restrictive in allowing mainly subject and object controllers (as is also demonstrated for Swiss German dialects by Bucheli Berger, Ch. 4, this volume). Still, in our corpus of spontaneous spoken German we also found a number of examples with oblique controllers which would not be judged acceptable in written German. An example is (42), where the depictive clearly predicates on Tasse ‘cup’, the


55

complement of the preposition aus. Similar examples are cited by Paul (1919: 49–57), Plank (1985: 175), and Schultze-Berndt and Himmelmann (2004: 74). German (42) und sie haben dann am Nachmittag and they. nom aux then in:the:dat afternoon [aus [derselben Tassei]]PP unausgespu¨lti den from the.same:dat cup unrinsed the:acc Kaffee getrunken coffee drunk ‘and in the afternoon they then drank their coffee from the same cup unrinsed.’ [overheard utterance] Similarly, the following English utterance, where the controller of the depictive unlocked is embedded in a PP, is attested, although out of context it is judged unacceptable by most speakers. (43) she came in [through [the back doori]]PP unlockedi [overheard utterance] Thus, it may well be the case that medium and text types (communicative genres) within a given language vary with regard to restrictiveness in the choice of the controller, with less planned and more contextualized types such as everyday conversations being less restrictive in this regard. 1.4.3 Syntactic position/phrase structure properties It would appear obvious that participant-oriented adjuncts differ in their positional characteristics across languages, depending on the general linguistic type of the language in question (configurational vs. non-configurational, VO vs. OV, etc.). Amha and Dimmendaal (Ch. 9, this volume) provide an illustrative case study with data from Nilotic and Omotic languages. In terms of surface structure, crosslinguistic positional differences pertain not only to broad topological parameters (e.g. preverbal vs. postverbal position) but also to positional variability. As adjuncts, it is to be expected that participant-oriented adjuncts show at least some positional variability (which in some languages would set them apart from arguments which are positionally more restricted), but languages clearly differ as to how much positional variability they allow for. Of major import in this regard is the question of whether or not different positions correlate with different meanings and constructions, and in particular with focus assignment.

56


While the preceding observations should be largely uncontroversial, the syntactic position of participant-oriented adjuncts is in fact one of the most problematic issues in a typology of depictives and adverbials simply because the precise position in the hierarchical syntactic structure is still a matter of debate even for better-analysed languages (see e.g Winkler 1997: 17–91; Mu¨ller 2002: 173–207 for recent surveys of some of the positions proposed for depictives proper in English and German). While it has been argued by some authors that depictives—in particular, subject-oriented depictives—are clause-level adjuncts (see e.g. Williams 1980; Rothstein 1985; Nakajima 1990), most authors follow the arguments of Andrews (1982) in assuming that depictives are adjoined somewhere within the VP. Attachment to different levels of the VP is used to account for the distinction between manner adverbs and depictives (see section 1.2.2), and also for the different subtypes of depictives such as depictives proper and circumstantials (see section 1.2.3), and subject- vs. object-oriented depictives (see section 1.4.2). For further discussion and references, see Simpson (Ch. 2, this volume), and Mu¨ller-Bardey (Ch. 3, this volume). Note that, in line with the argument in section 1.2.3, depictives proper (as distinct from circumstantials) by definition are restricted to focus position, which more often than not will also be a verb-phrase internal position (see also Kutscher and Genç, Ch. 7, this volume; Gu¨ldemann, Ch. 10, this volume). Thus, the possibility for making a principled distinction between depictives proper and circumstantials rests on the availability of a reasonably circumscribed focus position. In a language like Warlpiri, with a considerably greater freedom in word order and very little surface evidence for a VP, the distinctions relying on constituency become blurred (Simpson, Ch. 2, this volume). Crosslinguistically, therefore, the syntactic position of participantoriented adjuncts of different types appears to be much more variable than is usually assumed, and an investigation of this topic has only barely begun. The problems regarding the precise phrase structural position of participantoriented adjuncts are but one aspect in unravelling their complex syntax. Another major challenge is to model the ‘dual’ role of these constituents as participant-oriented ‘predicates’ and as verbal adjuncts. The predicative nature cannot be directly captured in models relying exclusively on constituency, except by positing a small clause analysis (see Winkler 1997: 18–50 for an overview of the debate on this controversial analysis). Other suggestions for capturing the particular nature of depictives, depending on the model assumed, include positing an irreducible syntactic relationship of predication between the depictive and its controller (see Rothstein 2001 for an overview and references), or a double relation of dependency (e.g. Nichols 1978a: 120; McGregor 1997c: 171–3, and Ch. 5, this volume). A further


57

alternative is to limit the syntactic analysis of depictives to their syntactic role as adjuncts, as is standard practice for other participant-oriented adjuncts, and to relegate all aspects of participant orientation to semantics. In fact a number of authors, including Plank (1985: 183), Steube (1994), and Du¨rscheid (2002: 70–1), have argued for just this approach to depictives in German. But while German depictives are morphologically unmarked and thus favour such an account, in those instances where participant orientation has morphological repercussions such as agreement with the controller, a ‘semanticsonly’ approach obviously faces the challenge of providing a convincing account of the morphological facts. For further discussion and references, see also Schultze-Berndt and Himmelmann (2004: 74–7). 1.4.4 Word class and internal structure The fourth parameter in a typology of participant-oriented adjuncts concerns the word class and internal structure of the adjunct. Of course, the constructions to be included here depend on the semantic range of adjuncts that one considers. Work on depictive secondary predicates has largely concentrated on adjectival secondary predicates, but, as shown in section 1.3, many types of adverb, adpositional phrase, or case-marked noun phrase, and other complex ‘adverbials’, including subordinate clauses, may also be considered participant-oriented. 1.4.4.1 Adjectives Simple adjectives and adjective phrases are the most widely recognized instances of participant-oriented adjuncts; they are amply illustrated throughout this chapter and in the contributions to this volume. A number of contributors, including Bucheli Berger, Simpson, Mu¨llerBardey, McGregor, and van der Auwera and Malchukov, mainly restrict their discussion to adjuncts of this type. However, in not all languages do adjectival depictives constitute the most ‘basic’ type of depictives. Obviously, languages lacking adjectives, or possessing only a small number of adjectives, have to rely on alternative means of forming participant-oriented adjuncts, which may include adverbs, ideophones, deverbal forms, and subordinate clauses. In fact, the use of adjectives as participant-oriented adjuncts seems to presuppose the possibility of employing adjectives as main predicates—Ewe, for example, lacks both (Ameka, Ch. 11, this volume). 1.4.4.2. Nouns In languages where nouns generally are individual-level predicates, i.e. express time-stable characteristics,17 participant-oriented 17 But see Ogawa (2001), who proposes to apply the stage-level vs. individual-level distinction also to nominal predicates.

58


adjuncts tend to be adjectival or verbal. If a noun appears in adjunct position in these languages, it is in a construction with a dedicated predicative marker or a copula (see section 1.4.5.2), or in a construction with a special case marker or adposition. In these cases, the construction as a whole serves to convert the nominal into a stage-level predicate (cf. also section 1.2.3). In languages like Warlpiri, on the other hand, with no clear distinction between nouns and adjectives, or stage-level and individual-level predicates, nominals are much more freely employed as participant-oriented adjuncts (see Simpson, Ch. 2, this volume). 1.4.4.3 Adverbs and related parts of speech A part of speech restricted to adjunct function is generally classified as ‘adverb’. Adverbs may be participantoriented, event-oriented, or both, as discussed in detail in section 1.2. Examples of participant-oriented adverbs in English are the oriented manner adverbs in -ly discussed in section 1.2.2, and unmarked quantifying adverbs such as alone (see section 1.3.7). If a class of ‘adverbs’ always conveys participant orientation, it may be justified to speak of a word class restricted to functioning as a depictive constituent; this special case will be discussed in 1.4.4.4. below). A class of words which may be used as adjuncts, but also as main predicates (with or without a copula), but not as attributes, are often subsumed under adverbs, although traditionally, the term ‘predicative adjectives’ has also been used. Typically, these exhibit participant orientation. An example from German is barfuß ‘barefoot’ and its dialectal variants, which can be used as a main predicate (e.g. sie war barfuß ‘she was barefoot’) or as an adjunct, as in (44), but not as a modifier (*die barfuße Frau ‘the barefoot woman’). German (Cologne dialect) (44) un alle Kinder durfte fro¨her bla¨ckfo¨ß loufe, and all children may:pst.3pl formerly barefoot walk nur ich nit only 1sg not ‘And in those days all children were allowed to walk barefoot, except for me.’ (Bhatt and Lindlar 1998: 173) In a number of languages, a special word class of (often sound-symbolic) uninflecting words, variously termed ‘ideophones’, ‘expressives’, or ‘soundsymbolic adverbs’, may appear in adjunct position and allow for participantor event- orientation (cf. the contributions in Hinton et al. 1994 and Voeltz and Kilian-Hatz 2001). Participant-oriented ideophones are illustrated with


59

the following two examples from Hausa18 (for similar examples from Ewe see Ameka, Ch. 11, this volume). Hausa (45) kana`r yaa tsayaà k`ıika`m colonel 3sg.m.pst stay ideo:tall&motionless ‘The colonel stood there tall and motionless.’ (Newman 2000: 254) (46) naa gan shı` tik, haihu`war 1sg.pst see 3sg.m.o ideo:stark.naked birthgiving:poss uwarsa` mother.3sg.m.poss ‘I saw him naked as the day he was born.’ (Jaggar 1992: 92) Amha and Dimmendaal (Ch. 9, this volume) mention the case of several Nilotic and Omotic languages where ideophones may have participantoriented manner interpretations. Unlike in Hausa and Ewe, these have to be introduced by a defective verb ‘say’. This is also true for ideophones in a similar function in Shona (Gu¨ldemann, Ch. 10, this volume). Yet another type of adverbial-like part of speech allowing participantoriented uses is found in many northern Australian languages which have a closed class of verbs. Most ‘verbal’ notions, but also notions encoded by adverbs and verbal particles in Germanic languages, are expressed not by verbs, but by uninflecting, inherently predicative elements which form an open class. These are variously known as ‘preverbs’, ‘coverbs’, and ‘verbal particles’ in the literature (see also Simpson, Ch. 2, this volume, for preverbs in Warlpiri). They mainly function as constituents of complex predicates together with a verb (like mung ‘look at’ in (47)), but may also occur independently as participant-oriented adjuncts. In (47), the allative marker on the coverb gurdij ‘stand’ indicates that this is oriented towards the undergoer (O) of the main predicate—in this case, that the pig and not the horse is standing up (see further Schultze-Berndt 2000: 112–13, 2001). Jaminjung gurdij-bina (47) mung gani-ngayi-m¼biyang pigibigi look.at 3sg:3sg-see-prs¼now pig(abs) stand-all(o) ‘It [the horse] is looking at the pig [that is] standing up.’ (fieldnotes, E. Schultze-Berndt)

18

We would like to thank Carmen Dawuda for drawing our attention to these Hausa examples.

60


1.4.4.4 Word classes restricted to occurrence as a depictive adjunct Little attention has been paid in the literature to the fact that some languages appear to have word classes restricted to depictive function. Members of these word classes only occur in adjunct position and in this respect resemble adverbs. However, these lexemes show obligatory agreement with a participant of the clause. For example, Pinkster (1988: 224–6) discusses Latin forms such as invitus ‘reluctant’ which are traditionally classified as adjectives because of agreement in case, number, and gender, but which can function neither as attributes nor as main predicates, but only as depictive secondary predicates (praedicativa). Similar nonverbal word classes which show obligatory agreement and which are restricted to secondary predicate function have been described for some Australian languages, e.g. Kayardild (Evans 1995: 227–31), Martuthunira (Dench 1995: 53), Yankunytjatjara (Goddard 1985: 57), Diyari (Austin 1981a: 107–8), and Warlpiri (Hale 1982: 279–80; 1983; Simpson 1991: 123ff., and Ch. 2, this volume). Semantically, they comprise notions of posture, quantification, and manner. An example is (48) from Diyari. Diyari (48) a. wata yini parraparra pithi-ya not 2sg.sbj energetic(abs) fart-imp ‘Don’t fart loudly!’ (Austin 1981a: 107) b. nhulu karna-li kirra parraparra-li warra-yi 3sg.nf person-erg boomerang(abs) energetic-erg throw-prs ‘The man throws the boomerang energetically.’ (Austin 1981a: 107) From a typological perspective, these forms are probably best regarded as a distinct part of speech, the only function of which is occurrence as depictive secondary predicates. 1.4.4.5 Verbal and deverbal depictives The use of deverbal predicates as depictives is well attested for European languages, where participles can be used just like adjectives, although, as shown by Bucheli Berger (Ch. 4, this volume) for some Swiss German dialects, there may in fact be restrictions on the verb classes that these participles may come from. Following Haspelmath (1995), we use the term converb for ‘participles’ which are used primarily as adjuncts. In Schultze-Berndt and Himmelmann (2004: 98–106) we argue that converbs with restricted reference and null subject—Haspelmath’s (1995: 9) ‘implicit subject’ converbs—are always participant-oriented because they are obligatorily controlled (see also Mu¨ller-Bardey 1990: 2–3; Haspelmath 1995: 17–20).


61

Often, ‘implicit subject’ converbs with an S/A controller are in functional opposition with converbs which are not obligatorily controlled (or which even obligatorily express their own subject) and which typically have a differentsubject interpretation (Haspelmath’s ‘free subject’ and ‘explicit subject’ converbs). Functionally, in this case, the contrasting forms are part of a switchreference system (see e.g. Austin 1981b; Haiman 1983; Nichols 1983; SchultzeBerndt 2006: 186–9), and there has been a tendency to analyse both as adjuncts of the same type (i.e. calling them simply ‘adverbials’). Formally, however, in a given language, same-subject converbs may constitute genuine participant oriented adjuncts, while other converb types (which are neutral with regard to participant orientation) instantiate a general adjunct construction. Two languages which exhibit exactly this type of contrast, the Omotic languages Wolaitta and Maale, are discussed by Amha and Dimmendaal (Ch. 9, this volume). Another language where same-subject clauses, unlike different-subject clauses, formally mark participant orientation is Shipibo-Konibo, discussed by Valenzuela (Ch. 8, this volume). See sections 1.3.12–14 for further discussion. In many languages, the boundary between biclausal constructions, i.e. constructions involving two main predicates, and deverbal adjuncts showing participant orientation is difficult to draw. This is either because the language lacks a finite/nonfinite distinction, as does Lao (Enfield, Ch. 12, this volume) or because translation equivalents of ‘standard’ depictive examples involve a verb with the properties of finite verbs, often an auxiliary or copular verb ‘be’ linking a predicative nominal to the rest of the clause. Borderline cases of this type are presented by both Turkana (Amha and Dimmendaal, Ch. 9, this volume) and Shona (Gu¨ldemann, Ch. 10, this volume). In these languages, predicates of subordinate clauses in ‘depictive’ function take obligatory bound subject pronominals. They thus seem to express their own subject and not to meet the condition of obligatory control proposed for depictives by Schultze-Berndt and Himmelmann (2004). As the above-mentioned authors argue, though, clauses of this type do not allow NPs (including free pronouns) as subjects, and in this sense they can be said to be obligatorily controlled. 1.4.5 Morphological marking Morphological marking constitutes the fifth and probably most straightforward parameter in a typology of participant-oriented adjuncts. In the simplest case, of course, participant-oriented adjuncts do not have any morphological marking. This is the case for adjectival depictives in English, which are distinguished from e.g. manner adverbials (including oriented adverbials) by the absence of formal marking (see section 1.2).

62


In the following subsections, only the main types of overt morphological marking will be considered (Schultze-Berndt and Himmelmann (2004) discuss a few other types which are more marginal crosslinguistically, such as nonspecific linkers and genitive marking of adjuncts). The types of morphological marking—if any—of participant-oriented adjuncts that are found crosslinguistically formally reflect, on the one hand, the semantic subclasses of participant-oriented adjuncts (e.g. special markers for comitative and similative adjuncts: see section 1.3). On the other hand, they reflect functional relationships that the participant-oriented adjuncts bear to various other construction types, such as adverbials and predicative complements. These functional relationships are represented in the form of a semantic map in Mu¨ller-Bardey (1990) and, in a much more elaborate form, in van der Auwera and Malchukov (Ch. 13, this volume). Disregarding the specialized markers just mentioned, it seems to hold as a crosslinguistic generalization that participant-oriented adjuncts generally capitalize on marking strategies that are also found in other constructions. A specific marker of participantoriented adjuncts (more specifically in this case, of depictives) was only found in one language so far, the north-eastern (Appenzell) variety of Swiss German (Bucheli Berger, Ch. 4, this volume), where it can be shown to originate from agreement marking. 1.4.5.1 Agreement In many languages, agreement is used to establish a link between a participant-oriented adjunct and its controller. In these languages, agreement can be regarded as a major criterion for the status of an adjunct as a depictive, as reflected throughout section 1.3. Agreement, as usually understood, involves the categories of gender, number, and/or case. Languages exhibiting agreement of this type usually have NP-internal agreement as well, although, as the contribution by Boeder (Ch. 6, this volume) on Georgian shows, the latter does not necessarily display the same characteristics as the marking on depictives. Further examples, discussed in this volume, come from Swiss German dialects (Bucheli Berger, Ch. 4) and from Australian languages (Simpson, Ch. 2, and McGregor, Ch. 5). It is also well known that many Romance and Slavic languages exhibit depictive agreement. Agreement can also be used in a wider sense, for a marking strategy that does not involve any independently existing category (such as gender, number, case), but involves forms which have the sole purpose of unambiguously identifying the controller of an adjunct (such as S or A)—which implies, by definition, that there is no corresponding ‘agreement’ within NPs. This is discussed under ‘other strategies for indicating restricted reference’ in


63

Schultze-Berndt and Himmelmann (2004: 84f.). If this marking strategy is employed on subordinate clauses and indicates co-reference of the (often null) subject of this clause and the controller, this is commonly referred to as switch-reference marking rather than agreement. The most common type is same-subject marking (i.e. orientation towards an S/A pivot), as illustrated for Wolaitta and Maale by Amha and Dimmendaal (Ch. 9, this volume). However, specific forms signalling O orientation are also attested, e.g. in the Australian languages Warlpiri (Simpson 1988; Hale 1994), Martuthunira (Dench 1988; see example (39) above), and Jaminjung (Schultze-Berndt 2001; see example (47) above), and in Shipibo-Konibo (Valenzuela, Ch. 8, this volume). The actual markers of participant orientation can be of different types, e.g. specialized converb forms (as in Wolaitta and Maale), or case markers in subordinating function which restrict the reference of a controller (as in Warlpiri and Jaminjung). For the Panoan language family, of which ShipiboKonibo is a member, Valenzuela (Ch. 8, this volume) argues that markers of participant-orientation, although synchronically unanalysable forms, arose diachronically through a combination of a specialized subordinating case marker and an additional case marker added in agreement (in the traditional sense). In Shipibo-Konibo, the same markers are found not only on subordinate clauses, but also on intraclausal adjuncts of lower complexity, which justifies the use of the term ‘participant agreement’ rather than ‘switchreference marking’. In a language like Warlpiri, too, case markers in agreement with the controller appear both on clause-internal depictives, and in addition to the switch-reference marking in participant-oriented subordinate clauses.19 Yet another type of agreement between a subordinate clause and a participant of the main clause can be illustrated with Shona (Gu¨ldemann, Ch. 10, this volume). As already mentioned in section 1.4.4, subordinate predicates in 19 Corbett (1998: 195) argues that case-marking on a modifier within a noun phrase is not triggered by the head noun but ‘imposed’ on both NP constituents by the governing verb, and that it should therefore not be subsumed under agreement. The same would presumably hold for identical casemarking of a controller and a depictive. Two arguments can be made in defence of the use of the term ‘agreement’ to include case-marking. The first—also acknowledged by Corbett—is that in languages of the fusional type, the categories of number, gender, and case are often expressed by portmanteau morphemes and that it therefore makes little sense to separate them. The second is that, as indicated above, the involvement of case in depictive marking cannot be clearly separated from switch-reference marking. Clearly, one would want to analyse e.g. same-subject marking as involving the marking of a relationship between a target and its controller (i.e. agreement proper), and not maintain that it is assigned by the verb of the first clause. This analysis of identical case marking as directly signalling a relationship between the two case-marked constituents could then be extended to non-clausal depictives.

64


depictive function in Shona, although being obligatorily controlled in the sense that they do not allow an NP subject, take an obligatory bound pronominal. The orientation of the subordinate predicate can thus be indicated by person/noun class agreement of this bound pronominal with the controller. 1.4.5.2 Copulas and other predicative markers In many languages participantoriented adjuncts take predicative markers, i.e. markers which are also found on main predicates and on predicative complements, thus reflecting the function of the adjuncts in question as (secondary) predicates. A paradigm case of a predicative marker is the so-called essive case in Finnic languages. In Finnish, it is found not only on predicative complements and expressions of life stage and function or role (in these functions, its English translation equivalent is as), but also on participant-oriented manner expressions, as example (49) shows. Finnish (49) ha¨n la¨hti hiljaise-na huoneesta he left quiet-ess from:room ‘He went quiet out of the room.’ (Nichols 1978a: 123) The clitic ¼i in Ewe, which, as Ameka (Ch. 11, this volume) argues, is a predicative marker, has remarkably similar functions to the essive in Finnish. A predicative marker found with nominals both in main predicate and depictive function in the Eastern Nilotic language Bari is briefly discussed by Amha and Dimmendaal (Ch. 9, this volume). There are also languages where not only main predicates but also secondary predicates appear with an obligatory copula, another type of predicative marker. The copula may be an invariable particle, as in Berber (Maas, p.c.), or belong to the verb class, as in Lao (Enfield, Ch. 12, this volume). In the latter case, it may appear in a converbal form (see section 1.4.4.5) when marking secondary predicates. This is a crosslinguistically frequent strategy, found in languages as diverse as Turkish (Boeder and Schroeder 1998; Schroeder 2003; Johanson 1995), Shona (Gu¨ldemann, Ch. 10, this volume), and Wolaitta (Amha and Dimmendaal, Ch. 9, this volume). 1.4.5.3 Adpositional/case marking In many languages, certain semantic cases or adpositions such as locative or comitative are regularly widely employed on participant-oriented adjuncts. A well-known case of instrumental marking on participant-oriented adjuncts is Russian (see Jakobson 1936; Nichols 1978a; 1981; Janda 1993). Further examples of participant-oriented adjuncts taking an instrumental case or adposition can be found in Japanese


65

(Takezawa 1993), Georgian (Boeder, Ch. 6, this volume) and the related language Laz (Kutscher and Genç, Ch. 7, this volume), and in Ewe (Ameka, Ch. 11, this volume). In section 1.3.10 we pointed out that certain locative adjuncts have to be regarded as primarily participant-oriented rather than event-oriented, in that they do not locate the event as a whole but only one of the participants. Locative cases and adpositions, moreover, may take on functions going beyond the purely spatial ones. In this case, too, the resulting constituents may have participant-oriented interpretations; examples can be found in Ewe (Ameka, Ch. 11, this volume, ex. (5b)), and Wolaitta (Amha and Dimmendaal, Ch. 9, this volume). Usually, locative-marked adjuncts of this type will also have to be regarded as instances of general adjunct constructions. There are other types of case marking or adpositional marking on participant-oriented adjuncts which cannot be assigned a prototypical function such as instrumental or locative. An example is the adverbialis case in Georgian, the functions of which are discussed in some detail by Boeder (Ch. 6, this volume). 1.4.5.4 Restrictive markers In section 1.2.3, depictive secondary predicates in the narrow sense were defined as those participant-oriented adjuncts which are part of the focus domain in a given clause. This is consistent with the observation that depictives are often semantically more specific than the main predicate, and thus make the main contribution to the information conveyed by the clause. Depictives also often seem to be accompanied by certain particles or clitics which are associated with the focus of an utterance (see Ko¨nig and van der Auwera 1990: 344; Gu¨ldemann, Ch. 10, this volume). Schultze-Berndt (2002) discusses particles and clitics with primarily restrictive function (e.g. ‘just’, ‘only’) in several Australian languages, which are frequently found on secondary predicates. She argues that the restrictive marker indicates that only the specific event expressed by the depictive occurs, to the exclusion of possible alternatives. It seems that restrictive markers are obligatory or near-obligatory with depictives in some languages. In the Australian language Jaminjung, a restrictive clitic is obligatory with quantifiers in depictive function (fieldwork Schultze-Berndt; see exx. (27) and (28) in McGregor, Ch. 5, this volume). Nichols (1982: 339) explicitly regards what she calls the ‘delimiter’ esˇˇce (lit. ‘still’) in Russian as a near-obligatory grammatical marker in the subtype of depictive constructions which convey life stage. Bucheli Berger (Ch. 4, this volume) reports that in the Swiss German dialect of Diepoldsau, the restrictive

66


particle asa ‘so, so much’ appears to be obligatory on depictives, while in the surrounding dialects it is merely frequent in this construction. Finally, in Shipibo-Konibo, the restrictive or ‘emphatic’ marker -bi appears to be obligatory on expressions of life stage and emphatic pronouns, both considered as highly participant-oriented by Valenzuela (Ch. 8, this volume). Since the same marker also frequently occurs on adjunct types other than depictives in Shipibo-Konibo and the other languages mentioned here, it cannot be regarded as a depictive marker as such, but its high frequency and strong grammaticalization in this construction type constitutes further evidence for the close link between restrictive marking and depictive constructions.

1.5 Summary In this chapter, we have outlined a number of issues arising in a crosslinguistic approach to the syntax and semantics of depictive secondary predicates. Most importantly, we have seen that such an approach forces us to treat as a single domain a number of semantic adjunct types which to date have generally been treated as two distinct domains, that of depictives and adverbials. Here we have argued that the domain as a whole can be described in terms of the semantic features of participant orientation and event orientation. Thus, this chapter can also be read as a contribution to the further systematization of adverbials (see in particular section 1.2.1).20 The crosslinguistic approach shows that the morphosyntactic manifestation of participant- or event-orientation is an area of considerable variation (section 1.2). In some languages there exist specific depictive constructions, i.e. constructions which mainly or exclusively convey participant orientation (e.g. by means of agreement), which differ from adverbial constructions, i.e. constructions which mainly or exclusively convey event orientation. There are also adjunct constructions which remain unspecified for either event- or participant-orientation (these constructions are termed general adjunct constructions here), and all three construction types can be found in the same language. From the crosslinguistic variation in the semantic range of depictive and adverbial constructions, we have concluded that all of the semantic adjunct types discussed in section 1.3 are semantically both participant- and eventoriented (to very different degrees), and that there is competition as to which orientation is highlighted in morphosyntactic form through the use of a depictive or an adverbial construction. Perhaps the clearest example of this double nature is provided by manner expressions, which are discussed at 20

See also Mu¨ller-Bardey (Ch. 3, this volume) for a somewhat different approach to this topic.


67

length in sections 1.2.2 and 1.3.3. Instances where morphosyntactic form appears to contradict semantic preference (as when temporal adjuncts agree in case with one of the arguments of the main predicate) are seen as cases of overgeneralization. The different degrees of participant- and event-orientation at the semantic level, as well as the variation in encoding, can be represented by way of a semantic map. In the map proposed in section 1.3, primarily or exclusively participant-oriented adjuncts, i.e. adjuncts conveying a condition or state of one of the arguments of the main predicate, are found at the centre, and primarily or exclusively event-oriented adjuncts, e.g. temporal adjuncts, at the periphery. The semantic map predicts, on the basis of the findings reported in this volume, that if participant orientation has a morphosyntactic manifestation in a given language, then the construction in question will cover contiguous segments of the map. Not only the semantic range but also the formal properties of adjunct constructions exhibiting participant orientation are subject to crosslinguistic variation (section 1.4). For example, participant-oriented adjuncts may pose more or fewer restrictions on the syntactic function (e.g. subject, object, indirect object) of possible controllers. And while the standard examples of depictives usually discussed in the literature are adjectives, participant-oriented adjuncts may have a more complex constituency and be built around members of various word classes. Noun phrases and adpositional phrases, ideophones, and subordinate clauses headed by ‘participial’ or converbal forms of verbs can all serve as participant-oriented adjuncts, and languages may exhibit a preference for adjuncts of one or the other type. In section 1.4.4 the possibility is raised that some languages even have a word class restricted to depictive function. In preparation of the crosslinguistic comparison of depictive secondary predicates, a number of concepts and distinctions discussed in the literature had to be critically reviewed and modified (section 1.2). Perhaps the most important point here is the hypothesis advanced in section 1.2.3 that depictives (in the broad sense) can be either in focus or part of the presupposition, and that at least in some languages this difference is manifest on the constructional level, such that a distinction has to be made between depictives proper and circumstantials. Much remains to be done to test this hypothesis, which among other things needs to be properly specified for each of the semantic adjunct types discussed in section 1.3. Further cross-linguistic research is thus likely to lead to further refinements in the typology of participant-oriented adjuncts in terms of their semantics, their morphosyntax, and their status in terms of information structure.


2 Depictives in English and Warlpiri JANE SIMPSON

2.1 Introduction Typical and often-cited examples of ‘depictive attributes’ (Halliday 1967) in English include: (1) a. I ate the meat raw. b. She left drunk. c. I ate the meat drunk. The adjectives (in boldface) in (1) act as secondary predicates, attributing a quality to a participant (underlined) at the time of action denoted by the verb. The term ‘attribute’ was introduced by Halliday to label an English construction involving a ‘characteristic ascribed to one of the participants . . . that relates specifically to the process in question’ (Halliday 1967: 62). ‘Depictives’ are an ‘attribute’ subtype ‘which characterises the attribuant in relation to the process, but as a concomitant, not a result of the process’ (p. 63). Thus, attributes in Halliday’s sense are both participant-oriented and event-oriented, like manner adverbs; but unlike English manner adverbs, they focus more on the participants than on the process (Schultze-Berndt and Himmelmann 2004, and introduction to this volume). The sentences in (1) have translation equivalents in Warlpiri. (2) contains a nominal, wanka ‘raw’, predicated of the understood object, and a nominal yarnunjuku-rlu¼ju, ‘hungry’, predicated of the subject, expressed as a clitic -lu. The verb ngarni ‘eat’ always takes an Ergative subject and so yarnunjuku has Ergative case agreeing with that subject. Thus in both English and Warlpiri I thank the participants in the Workshop on Depictives, Ruhr-Universita¨t Bochum, 7 June 2001, and participants at the 2002 Blackwood Workshop on Australian Languages for stimulating papers and discussion. And I am very grateful to Nikolaus Himmelmann and Eva Schultze-Berndt for excellent editorial remarks which have made me rethink much of the paper.

70

J. Simpson

a state of affairs can be predicated of a participant by means of a nominal, without mediation by an adposition or semantic case-marking.1 (2) Nga-rni¼lki ka¼lu wanakiji¼ji. eat-npst¼then prs¼333.sbj bush.tomato¼top yarnunjuku-rlu¼ju Nga-rni ka¼lu eat-npst prs¼333.sbj hungry-erg¼top ngula¼ju ka¼lu wanka nga-rni. that¼top prs¼333.sbj raw eat-npst ‘Bush tomatoes, they eat them then; they eat them when hungry, and they eat them raw.’ [Hale 1966: 82]

1 Glossing conventions used only in this paper: 111: first person plural exclusive, 222: second person plural, 33: third person dual, 333: third person plural, 122: first person plural inclusive. The boundary between clitics and suffixes is not clear-cut in Warlpiri; endings that are uncontroversially clitics are marked with an equals sign, ¼, while other endings are marked with a hyphen, -. In some Warlpiri examples prosodic breaks are indicated by one or two dashes (- or --), in line with conventions used in their sources. Warlpiri sentences are followed by reference to their source. Published sources are given in parentheses. Unpublished sources are given in square brackets. Much of this material is also contained in digital form in ASEDA (Australian Studies Electronic Data Archive). The following abbreviations are used:

Darby Jampijinpa 1984¼Darby Jampijinpa Warumungku Watikirli, Warlpiri Literature Production Centre Inc. Yuendumu, 1994. Dinny 1979¼Dinny Japaljarri France, Tape #402 (copy of #63). Recorded at Yuendumu 1979. Transcribed by Tess Ross Napaljarri. Hale 1959¼Hale, Kenneth (1959). Walbiri Field Notes. ms 3171, also digitally in the collection ASEDA 0050. Canberra: Australian Institute of Aboriginal and Torres Strait Islander Studies. Followed by page number or section reference. Hale 1966¼Hale, Kenneth (1966–7). Warlpiri Field Notes: Yuendumu, Hooker Creek, Lander River, Warrabri, N.T. ms 865, also digitally in the collection ASEDA 0050. Canberra: Australian Institute of Aboriginal and Torres Strait Islander Studies. Followed by page number or section reference. Ngarlinjiya¼Robertson, Ngarlinjiya Mary Nungarrayi, Transcript of tape no. 629a recorded at Yuendumu, 23 May 1990 by Peggy Rockman Napaljarri and Lee Cataldi, 5 June 1990. Later published (1994) as ‘Witikirli’, in Peggy Rockman Napaljarri and Lee Cataldi (ed.), Warlpiri Dreamings and Histories: Yimikirli. (The Sacred Literature Series.) San Francisco, London, Pymble: HarperCollins. Swartz¼Swartz 7 June 1996: Luke 17:26 Toby Japangardi¼Transcription of 60-minute tape recording made at Willowra by Toby Japangardi around 1984/5, transcribed by Mary Laughren, Dec. 1988. Warlpiri Dic.¼The Warlpiri Dictionary, digital version May 2000, by permission of Mary Laughren. Sentences are cited by reference to the lemma within which they occur (e.g. (Warlpiri Dic: ngakaparla)), and if there is a source given for the example sentence, that also is included (e.g. (Warlpiri Dic. wankaru: C:ML warrarna)). Where I have been able to find the original source, I have checked that, because the example sentences in dictionaries sometimes have to be edited to make the illustration clear. The translations are those given in the Dictionary if possible, modified occasionally if they are misleading.

Depictives in English and Warlpiri

71

In these examples wanka and yarnunjuku-rlu superficially fit the criteria of Schultze-Berndt and Himmelmann (2004: section 2.8) for depictives. Both are nominals without tense or mood marking. Neither is obligatory. Wanka, yarnunjuku-rlu and the verb ngarni are separate predicates, and the state of affairs expressed by wanka/yarnunjuku-rlu holds at a time within the timeframe expressed by ngarni. The subject of wanka is the same as the object of ngarni, while the subject of yarnunjuku-rlu is the same as the subject of ngarni. Neither wanka nor yarnunjuku-rlu forms a complex predicate or a low-level constituent with ngarni, and equivalent secondary predicates appear in a variety of sentence positions. (Warlpiri does have many complex predicates consisting of preverbs and verbs (section 2.2.3 below).) However, while in (1) and (2) the Warlpiri and English secondary predicates appear to be doing much the same thing semantically, there are more syntactic and semantic constraints on the appearance of the English depictives in (1) than on secondary predicates in Warlpiri. I suggest that many of these differences stem from basic syntactic differences between Warlpiri and English. In each language, these secondary predicates act syntactically as part of different systems of grammar. In Warlpiri, as I shall show, these secondary predicates are part of a general system of secondary predication using nominals, which act syntactically as adjuncts, while in English, depictives have closer ties with the ways of expressing complements. In the first part of the chapter, I describe and contrast the systems within which depictives operate in the two languages. In the second part I compare and contrast the constraints on depictives in English with the relative freedom of secondary predicates in Warlpiri.

2.2 Differences between Warlpiri and English Obvious differences in behaviour between depictives in Warlpiri and English stem from: the way grammatical functions are expressed using word order in English and using case agreement in Warlpiri; the wide range of denotation of Warlpiri nominals; the complementarity of depictives and resultatives in English, and related to this, the constraints on the appearance of more than one secondary predicate; the use of clitics in Warlpiri to express temporal information, information structure and connections between predicates. I discuss each in turn.

J. Simpson

72

2.2.1 Word order vs. case agreement In English, depictive predicates normally occur following an intransitive verb (He died young) or the object of a transitive verb (He ate the meat drunk), and, for some verbs, following certain prepositional complements (He served the meat to them raw). They differ from manner adverbs in that they cannot precede the verb (*He drunk ate the meat, he drunkenly ate the meat). In terms of phrase structure position, Andrews (1982) showed that English depictives, regardless of whether they are predicated of the subject or object, occur inside the Verb Phrase (VP) constituent, as (3) illustrates. His findings are repeated by Legendre (1997) for French. Koizumi (1994) argues that in Japanese depictives predicated of objects and some subjects are in the VP, while some depictives predicated of the subject are generated as sisters to VP. Koizumi extends the findings to English to account for the marginal acceptability of sentences like (3h). Eating fish raw upsets Mary. Eating fish naked upsets Mary. What Mary did was eat the fish naked. What Mary did was eat the fish raw. Eat the fish raw though Mary did, she was nonetheless blamed for her attitude. f. Eat the fish naked though Mary did, she was nonetheless blamed for her attitude. (Koizumi’s 108a) g. What John did was drive his car drunk. h. ?What John did drunk was drive his car. (Koizumi’s 108b)

(3) a. b. c. d. e.

Non-depictive secondary predicates are expressed in English by other phrase structure positions. Sentence margins (clause-initial and clause-final) are places where adjuncts modifying the whole clause, or the subject, may appear (4a–d), often resembling depictives in meaning. This is especially true of longer adjuncts which may simultaneously say something about the event (4b–d). Clause-final position can also be used for heavy adjuncts modifying arguments (and even adjuncts) other than the subject (Quirk et al. 1985: 425) (4e–h). They may be ‘strong’ detached adjuncts (Stump 1985; Himmelmann and Schultze-Berndt, Ch. 1, this volume) which may have causal or concessive readings, or they may be ‘weak’ and integrated, as English depictives are. (4) a. b. c. d.

Drunk, John went home. John locked himself out of his house, drunk and bewildered. Crazy as it seems/Crazy as she is, Mary loves John. Mary loves John, crazy as it seems/crazy as she is.


73

e. Mary loves John, crazy as he is/even drunk. f. Mary gave John her car, drunk as he was/brand-new as it was. g. Mary gave her car to John, drunk though he was/brand new though it was. h. ?For Mary he’d do anything, pregnant or not. Depictives and sentence adjuncts in English behave differently under negation, regardless of whether the depictive is predicated of the subject or the object. Depictives, being weak and integrated, are in the scope of negation. Negation of the verb can negate both the verb and the depictive, or just the depictive. (5) a. He didn’t arrive drunk. (either He arrived sober, or He didn’t come at all.) b. He didn’t eat the meat raw. (either He ate it cooked, or He didn’t eat it at all.) c. He didn’t eat the meat naked. (either He ate it fully clothed, or He didn’t eat it at all.) The verb can be negated on its own only if there is a heavy pause following the verb, in which case the most plausible interpretation is not as a depictive, but as a strong detached sentence-margin adjunct, and, as Quirk et al. (1985: 425) argue, the heavier the adjunct is the more acceptable the sentence (6b and c, compared with 6a; 6g compared with 6f). These adjuncts say something about the state of the participant they are predicated of, and also about the reason the main event did not take place (see Nichols 1981: 42 for similar findings in Russian), and acceptability often depends on this (6d compared with 6e). Getting a similar reason reading when a bare adjective or adjective phrase is predicated of an object is harder (6f, g), and is helped by the presence of an overt subject (6h). ??He didn’t arrive, drunk. ??He didn’t eat the meat, naked. He didn’t arrive, too drunk to remember the appointment. He didn’t arrive, drunk as usual. ?He didn’t die, naked as usual. ??He didn’t eat the meat, raw. ?He didn’t eat the meat, disgustingly raw and crawling with maggots. h. He didn’t eat the meat, raw as it was/it being raw.

(6) a. b. c. d. e. f. g.

Hence English depictives are best tested inside a VP or inside a sentential argument or adjunct, as in (3), since detached adjunct readings are much less plausible in these environments.

74

J. Simpson

Inside the VP a depictive predicated of the subject or object must follow NP arguments, but may follow or precede PP arguments with intransitive verbs, as in (7a–h), unless the preposition is acting more like a casemarker and less like a predicate in its own right (7i, j). (7a–d) also show that there need be no overt noun phrase of which the secondary predicate is predicated. The subject of the secondary predicate can have an unspecified referent (Laughren 1992; Himmelmann and Schultze-Berndt, Ch. 1, this volume). (7) a. b. c. d. e. f. g. h. i. j.

Jumping fully clothed into the water was a bad idea. Jumping into the water fully clothed was a bad idea. Prancing naked around the room for an hour was a bad idea. Prancing around the room naked for an hour was a bad idea. Running naked into the street was Jones’s first solecism. Running into the street naked was Jones’s first solecism. Arriving drunk at the party was Jones’s second solecism. Arriving at the party drunk was Jones’s second solecism. Staring at him drunk was Jones’s third solecism. *Staring drunk at him was Jones’s third solecism.

Depictives may also follow or precede PP arguments with transitive verbs, when predicated of the object (8a–d). However, when predicated of the subject (8e, f), sentence final position seems preferable. (8) a. After handing the baby to its mother crying, Jones sighed with relief. b. After handing it crying to its mother, Jones sighed with relief. c. After having waiters serve food raw to the visitors, they brought in the elephants. d. After having waiters serve food to visitors raw, they brought in the elephants. e. ??After having waiters serve food naked to visitors, they brought in the elephants. f. After having waiters serve food to visitors naked, they brought in the elephants. Depictives predicated of objects must precede nonfinite VP adjuncts (9a, b). Depictives predicated of subjects of transitive verbs are difficult to get with nonfinite VP adjuncts (9c, d). However, if the verb is intransitive, then depictives are more acceptable, as in (9e) (thanks to the editors for this example).


75

(9) a. Burning her alive to justify their cause must have seemed like a good idea. b. *Burning her to justify their cause alive must have seemed like a good idea. c. ??After eating the meat drunk to satisfy their hunger they went home. d. *After eating the meat to satisfy their hunger drunk they went home. e. After driving home drunk to avoid being stuck at the host’s place, they went to sleep Differences between depictives predicated of subjects and depictives predicated of objects have received much attention (Legendre 1997 for French; Koizumi 1994 and Yatsushiro 1999 for Japanese; Winkler 1997 for English). In the generative literature the usual solution has been to assume that depictives predicated of objects are attached close to the V (e.g. as sisters of V0 ), and depictives predicated of subjects are attached further away from the V (e.g. as sisters of a VP whether inside a VP or not). Thus phrase structure position in English determines in part whether or not a secondary predicate is interpreted as a depictive, what the scope of negation is, and whether a depictive can be interpreted as predicated of a subject or object. In Warlpiri, phrase structure position is used for information structure purposes (for example, to indicate what is questioned), but is not used for grammatical function or to indicate that a phrase is predicated of another word. For example, the secondary predicate wankaru in (10) appears in a variety of positions, sentence-final (10a), sentence-initial (10b), before the verb (10c), after the object and before an adjunct (10d), before the object (10e)—and other places are possible too. It can be predicated of the subject of an intransitive verb (10a, c), of the object of a transitive verb (10b, d), of the subject of a transitive verb (10e, f). parnka-ja wankaru. (10) a. Jinta-kari kala one-other used.to run-pst alive ‘One would run off alive.’ [Hale 1966: 1138–9] b. Wankaru ka¼jana ngaja-rni kurdu warrarna-rlu. live prs¼333.sbj bear-npst child skink-erg ‘The warrarna skink gives birth to its young live.’ [Warlpiri Dic. wankaru: C:ML warrarna]

76

J. Simpson c. Marlu -- panti-rninja-warnu, ngula; kangaroo spear-nmlz-assoc that wankaru wuruly-parnka-ja. alive hide-run-pst ‘That kangaroo, after being speared, it would run off and hide alive.’ [Hale 1966: 349–50] d. Puuly-marda-rni ka wardapi wankaru¼wiyi hold-npst prs goanna live¼before paka-rninja-kungarnti-rli. hit-nmlz-purp2-erg ‘He holds onto the goanna still alive before killing it.’ wankaru: C]

[Warlpiri Dic.

e. Kirdanyanu-kirdanyanu-rlu kala¼lu¼nganpa father-poss-father-poss-erg used.to¼333.sbj¼111.obj wankaru-rlu¼wiyi yujuku ngurrju-ma-nu- tarnnga-ngku. alive-erg¼before hut good-cause-pst always-erg ‘Our fathers when they were still alive used always to build us huts.’ (Warlpiri Dic. wiyi: BWJ1979:20) f. Yungu¼rlupa¼nyanu warrawarra.ka-nyi wankaru-rlu¼(yi)jala. so.that¼122.sbj¼refl look.out.for-npst live-erg¼also Kula¼rlipa jarda-nguna-mi jalangu¼ju. not¼122.sbj sleep-lie-npst today¼top Yungu¼rlupa¼nyanu wankaru-rlu warrawarra.ka-nyi. so.that¼122.sbj¼refl live-erg look.out.for-npst ‘So we can look out for ourselves (when) lively also. We won’t sleep now, so we can look out for ourselves (while feeling) lively.’ [Hale 1959: 131ns; also Warlpiri Dic. wankaru] We have not found evidence to link phrase structure position with whether the secondary predicate is predicated of a subject or object. Instead, what the secondary predicate is predicated of is shown by case agreement. In (10a–d) above, absence of case is interpreted as Absolutive, the case of the subject of intransitive verbs. In (10e) the Ergative case on wankaru-rlu shows that it is predicated of the Ergative case-marked subject, kirdanyanu-kirdanyanu-rlu. (10f) has no overt Ergative case-marked subject, but the verb is transitive and a predicate agreeing with the understood subject must have Ergative case. As in English, the secondary predicate can be within the scope of negation, by being in the same prosodic unit as a clause starting with the sentential negator kula. The secondary predicate may precede or follow the verb—the examples in (11) are predicated of the object.


77

(11) a. Kula¼lpa¼lu wankaru ka-nja-nu, na:. not¼pst¼333.sbj alive carry-nmlz.go-pst no ‘They did not take them along alive, no.’ [context: they did carry them along, and they made them alive afterwards] [Hale 1966: 892] yinjalykinyi¼ji. b. [..] yirnmi ka¼rnalu nga-rni cooked prs¼111.sbj eat-npst tuber¼top Kula¼lpa¼rnalu nga-njarla rdarri, not¼pst¼111.sbj eat-irr raw purra-nja-warnu-mipa ka¼rnalu nga-rni yinjalykinyi¼ji. cook-nmlz-after-only prs¼111.sbj eat-npst tuber¼top ‘Cooked is how we eat the yinjalykinyi tuber. We cannot eat it raw. Only after it has been cooked do we eat the yinjalykinyi tuber.’ [Warlpiri Dic: yinjalykinyi] There are examples in which the verb is denied while what appears to be a secondary predicate remains taken as true (12). Yarnunjuku-ku in (12) is most probably a secondary predicate, predicated of the Dative first person inclusive object. Its interpretation is vague between unrestrictive (as the translation suggests), concessive (‘even though we are hungry’), or restrictive (‘us hungry ones’). Such vagueness of interpretation is discussed in Section 2.2.2. (12)

yarnunjuku-ku. Kula¼ka¼ngalpa yi-nyi not¼prs¼122.obj give-npst hungry-dat ‘He doesn’t give any to us who are hungry.’ (context: speaker is starving, and complains about not being given anything) [Warlpiri Dic: ngurru: rno]

Warlpiri has little surface evidence for a VP. It is thus hard to find parallels in Warlpiri for the constraints on ordering depictives depending on what they are predicated of. There are phrases consisting of nominalized verbs preceded by their arguments and to a limited extent other constituents, but the preference is for very little material to appear in such phrases. (13) is a possible example; however it is usually possible to interpret such apparent depictives as ‘preverbs’ (predicates that modify the verb) (Section 2.2.3). (13)

Lani parnka-nja-wangu¼(l)pa¼npa karri-yarla¼rni. scared run-nmlz-without¼pst¼2.sbj stand-irr¼hither ‘You would stay without running away scared.’ [Hale 1959: 7.238]

Example (14) shows other relevant syntactic features of Warlpiri. Nominals can act as the main predicator of a clause, since there is no obligatory

78

J. Simpson

copula (14i, iii, iv), but stance verbs can be used as copulas (ngunami in 14vii). There is no requirement for an overt subject (14ii). As will be discussed in section 2.2.2, there is no morphosyntactic distinction between nouns and adjectives and manner adverbs. Consequently (14iii) is syntactically ambiguous between a coordination of two primary predicates, or a primary predicate and a secondary predicate: it is still unripe, it is green first or it is still unripe, first green. These factors combine to make it difficult in isolation to parse nominal predicates following a pause as primary or secondary, let alone as depictives or as detached adjuncts or even manner adverbs. (14)

i. Pirarrpirarrpa nyanungu¼ju yipintiri wita. orange that¼top berry small ii. Pirarrpirarrpa-jarri ka. [ . . . .] orange-become.npst prs iii. Ngula¼ju ka2 wanka¼juku; yarringkarringki¼wiyi. [..] that¼top prs raw¼still green¼before iv. Wanka¼ju yarringkarringki¼wiyi. raw¼top green¼before v. Ngula-jangka¼ju pirarrpirarrpa-jarri¼lki that-from¼top orange-become.npst¼then vi. nyanungu¼ju - yipintiri¼ji. [ . . . ] that¼top berry¼top vii. Pirarrpirarrpa¼lku¼lu nguna-mi yurnmi¼ji. orange¼then¼333.sbj lie-npst ripe¼top i. ‘It is orange, that little yipintiri berry. ii. It becomes orange [..] iii. It is still unripe, first green. iv. The unripe one is first green. v. After that it becomes orange then, vi. that berry.[..]. vii. The ripe ones are orange then.’ [Hale 1966: 834–5]

Example (14) above also shows another feature of Warlpiri secondary predicates—their appearance with temporal-aspectual clitics (Green 1987) (italicized in 14). These provide a precise indication of how to connect the time-reference of the clause with the time-reference of the state associated with the element to which they are attached.

2

It is unusual to have the auxiliary ka without a verb as the main predicate.


79

(15) a. ¼ lku, ¼lki:3 X is in state S at time T, and X wasn’t in state S before time T (14v, vii) b. ¼ wiyi: X was in state S before time T, and X is not in state S after time T (14iii, iv) c. ¼ juku, ¼jiki: X is in state S at time T, and X was in state S before time T (14iii) The use of these clitics allows great flexibility for Warlpiri secondary predicates to appear with verbs of different aspectual classes (sections 2.2.2, 2.2.6, 2.3.3). 2.2.2 Types of secondary predicate As the previous section has suggested, there is no clear-cut distinction between secondary predicates translating as depictives and other nominals. In both instances, case is used to determine what participant the predicate is predicated of, that is, orientation towards a participant (Himmelmann and Schultze-Berndt, Ch. 1, this volume). The first and most obvious instance are the so-called discontinuous noun phrases (Hale 1983; Nash 1986; Simpson 1991; Hale et al. 1996). Nominals with the same case can be interpreted as denoting the same entity, regardless of whether they are contiguous in the sentence. (16)

Maliki-rli¼ji yarlku-rnu wiri-ngki. dog-ERG¼1.obj bite-pst big-erg ‘The/a big dog bit me.’; ‘The/a dog bit me, a big one.’; etc.

In (16) both ‘dog’ and ‘big’ have Ergative case, and the interpretation of the nominals and the relation between them is vague as to whether the nominals are attributes or heads and, if attributes, restrictive or non-restrictive. Obviously the normal interpretation would be for malikirli to be the head and wiringki the attribute. But it seems that only context determines whether the attribute is restrictive, ‘The big dog bit me’ (the ‘merged’ reading, terms due to Nash 1986 and Hale 1983), or non-restrictive (‘unmerged’ reading), ‘The dog bit me, a big one’. I have found no morphosyntactic means for distinguishing these uses of nominals from the depictive use in (2). The question then arises about those nominals which, unlike wiringki in (16), can’t really be interpreted as restrictive attributes. Warlpiri nominals range from definite and fully referential expressions to ones which are almost 3 Warlpiri clitics and suffixes with high vowels usually agree in backness with the last vowel of the stem to which they attach.

80

J. Simpson

exclusively predicative in use. One such class are ‘action nominals’ (Simpson 1991), secondary predicates which, unlike depictives, describe activities or emotions, rather than states, of some participant. These agree in case with the case assigned to that participant, manyu ‘play, fun’ in (17a, b). (17) a. Maliki-jarra-rlu ka¼pala¼nyanu manyu-ngku yarlki-rni. dog-two-erg prs¼33.sbj¼refl play-erg bite-npst ‘The two dogs are biting each other in play.’ (Warlpiri Dic.: yarlkirni) manyu ngari yangka b. Ngula-piya¼jala ka¼lu that-like¼contrast prs¼333.sbj play just anaph junpurr-ma-ni flap.lips-do-npst ‘Like that they playfully blow (flap their lips)’ [Hale 1966: 0102/Warlpiri Dic.: junpurr(pa)] [junpurr-mani’s subject has Absolutive case.] Another such class are secondary predicates indicating place. If these attribute location to an object only, they agree in case with the object, whether this is Absolutive (unmarked) (kuyu in 18a), or Dative (the understood Dative argument cross-referenced by the auxiliary clitic ¼rla in 18b). If the sentence is intransitive, then the location nominal is unmarked. If the sentence is transitive, then the location nominal agrees in case with the subject, but only if the place is more closely associated with the subject (18d, e) than with the object or the event (Hale 1982; Laughren in press). (18f) shows an example where the location nominal is unmarked, despite the subject having Ergative case. (18e) also shows that the nominal may receive overt case-marking in agreement with the subject whether the nominal is inherently locative (yantarli) or marked with Locative case (ngurrangka). kuyu. (18) a. Parlku-ngka ka¼rnalu purra cooking.trench-loc prs¼111.sbj cook.npst meat. ‘We cook meat in a pit.’ [Warlpiri Dic.: jarnngi] walya karrkurr-pu-ngu b. Kala¼lu¼jana¼rla used.to¼333.sbj¼333.obj¼dat earth crunch-pst rdaku-ngka-ku¼ju. hole-loc-dat¼top ‘They (children) would crunch the earth over them (tadpoles) in the hole.’ [Warlpiri Dic.: karrkurrpa: Nora Nungarrayi] [The understood object, the tadpoles, has Dative case.]


81

c. Ngurra-ngka¼lpa nguna-ja purlka¼pardu. camp-loc¼pst lie-pst old.man¼dear ‘The dear old man was lying in camp.’ [Dinny 1979] d. Nga-rnu¼lpa yangka ngurra-ngka-rlu eat-pst¼pst anaph camp-loc-erg ‘He was eating in camp.’ [Dinny 1979] (Overt nominal subjects of ‘eat’ always have Ergative case.) wapal.nya-nyi yantarli-rli e. [ . . . ] yapa-kari-rli ka¼rla person-other-erg prs¼dat look.for-npst at.home-erg ngurra-ngka-rlu. camp-loc-erg ‘[ . . . ] another person who is at home in camp looks out for it.’ [Warlpiri Dic.: riwi: jne] ngurra-ngka¼jala f. Kaji¼ka¼ngku¼lu if¼prs¼2.obj¼333.sbj camp-loc¼contrast Kulu-parnta-mipa-rlu¼ju. panti-ni. spear-npst anger-full-only-erg¼top ‘They might spear you in camp indeed. All the fighters.’ 1959: 7.221]

[Hale

Another kind of secondary predication is found in the part–whole syntax of Warlpiri (Hale 1981), which permits nominals denoting body parts to be predicated of some participant in the clause. Furthermore, these body parts can themselves have modifiers, forming noun phrases which agree in case with the case assignable to any nominal denoting that participant, unmarked Absolutive in (19a), Ergative in (19b). There seem to be no intonational or morphosyntactic grounds for distinguishing these noun phrases from other nominals used as secondary predicates in Warlpiri; thus compare (19a) with (19c, d).

(19) a. Parrulka ka

kakarda kirrirdi wapa. turkey prs neck long walk.npst ‘The turkey walks with long neck.’ [Hale 1959: 799]

b. Jirri¼kila ka nga-rni, kalwa-ngku, tadpole¼foc prs eat-npst egret-erg ngapa-ngka-rlu, kakarda kirrirdi-rli water-loc-erg neck long-erg ‘It’s tadpoles that it eats in the water, the egret, that long necked (one).’ [Hale 1966: 590]

82

J. Simpson c. Yangka yi¼ka kurdu-wangu wapa. anaph cause¼prs child-without move.npst ‘One who walks around without children’ [in definition of ‘childless’] [Hale 1966 pt. 3: 0378] d. Nyurrurla ka¼nku¼lu pingka wapa. you.pl prs¼222.sbj slow move.npst ‘You walk slowly.’ [Hale 1966 pt. 3: 0239]

Thus case agreement on action nominals, locations and body parts indicates orientation towards a participant, as Himmelmann and SchultzeBerndt suggest (Ch. 1, this volume). But case agreement cannot always be easily interpreted in this way. For example, manner nominals such as those denoting speed of an action denote a property of the event, rather than of the participant expressed by the subject. They still receive case-marking in agreement with the subject, as in (20a, b); see Himmelmann and SchultzeBerndt (Ch. 1, this volume) for discussion and similar examples from other languages. (20) a. Turlka-ni kapi¼rna¼ngku wakurturdu-rlu. pinch-npst fut¼1.sbj¼2.obj hard-erg ‘I’ll pinch you hard.’ [Hale 1959: 933] (turlkani’s subject has Ergative case.) b. Yankirri¼ji ka wakurturdu-nyayirni emu¼top prs fast-very jukajuka-parnka. head.back.forth-run.npst ‘The emu runs fast with head going back and forth.’ 1959: 8.55]

[Hale

Finally, nominals denoting time also may agree in case with the subject. (21a) shows that if the subject is Absolutive, then the time nominal has no overt case-marking. In (21b, c) the transitive verbs require Ergative subjects and the time nominals also have Ergative case (see also (10e)).

(21) a. Pirrarni

nyurnu-jarri-ja kurdu. yesterday sick-become-pst child ‘The baby got sick yesterday.’ [Hale 1959: 7]

b. Nya-ngu¼rna¼ngku pirrarni-rli. see-pst¼1.sbj¼2.obj yesterday-erg ‘I saw you yesterday.’ [Hale 1959: 19]


83

c. Jalangu-rlu ka¼rna¼nyarra yirri-pura today-erg prs¼1.sbj¼222.obj tell.npst luwa-rnu kuyu yinya¼ji yangkurra-rlu¼nya¼lu yesterday-erg¼emph¼333.sbj shoot-pst meat that¼top yapa-patu-rlu¼ju puluku manu marlu. person-many-erg¼top cow and kangaroo ‘Today I tell you that yesterday the people shot a cow and a kangaroo over there.’ [Warlpiri Dic.: yangkurra: ppj 10/87] Not all time nominals agree in case all the time with the subject, but what determines the variation has not yet been clarified. In sum, Warlpiri nominals have a wide range of uses. Many of these can be seen as participant-oriented adjuncts (Himmelmann and Schultze-Berndt, Ch. 1, this volume), where case agreement indicates the participant to which they are oriented. However, case agreement with the grammatical function Subject sometimes has no obvious semantic connection with the participant expressed by the subject, as with the time nominals, and perhaps manner nominals such as ‘fast’. For time nominals, in fact, their case-marking seems to have clausal scope (Laughren, in press; Simpson 1991). Such case-marking could result from the extension of a pattern that was originally semantically motivated by participant orientation (Himmelmann and Schultze-Berndt, Ch. 1, this volume). 2.2.3 Resultative/depictive contrast An important difference between English and Warlpiri is that in English syntactically similar structures are used for both ‘predicatives of being’, or depictive secondary predicates (22a), and ‘predicatives of becoming’, or resultative secondary predicates (22b) (Jespersen 1965: ch. 27.17; Halliday 1967). In Warlpiri, ‘predicatives of being’ and ‘predicatives of becoming’ are handled in different parts of the grammar. (22) a. They burned Joan of Arc alive. b. They burned the steak black. Burning Joan alive was an error. depictive Burning the steak black was an error. resultative (22) shows how similar the structures for expressing depictives and resultatives can be. In fact, out of context some English sentences can be ambiguous between a resultative reading and a depictive reading (23a). It is also hard to find acceptable sentences containing both a resultative and a depictive (23b) (but see also section 2.2.4 below).

84

J. Simpson

(23) a. She cooked it dry. Her hair fell loose. The water ran cold. (Halliday 1967: 65) b. ??They burned Joan of Arc black alive. *They burned Joan of Arc alive black. Moreover, both depictives and resultatives are also in approximate complementary distribution with nonfinite VP complements of verbs (24a–d). ((24a, c) are acceptable in the reading ‘He was to become teetotal drunk’.) (24) a. b. c. d.

*Ordering him to become teetotal drunk was a bad idea. *Ordering him drunk to become teetotal was a bad idea. *Convincing him to become teetotal drunk wasn’t hard. *Convincing him drunk to become teetotal wasn’t hard.

The major difference between English resultatives and depictives has to do with what they can be predicated of. Resultatives must be predicated of the object of a transitive verb, can only be predicated of the subjects of certain intransitive verbs, and can appear in a fake reflexive construction, as in She cried/ate herself sick. Resultatives furthermore are unique—not more than one resultative can occur in a single clause (*He boiled the pot dry black) unless they are overtly conjoined. Depictives can be predicated of the subject or object of a transitive verb, and the subjects of many intransitive verbs, but do not appear in fake reflexive constructions. In Warlpiri, free nominals which agree in case with what they are predicated of predicate a state or location or an action or emotion of their subject. Change of state is handled by other means. These include: the use of compound verbs (-jarri(mi) ‘become’ in (14ii, iv) and (21a) indicates a change of state), and the use of special ‘translative’ resultative case suffixes (-karda in 25aii) which normally attach to nominals, or complementizer suffixes (-kijaku in 25aiii) which can attach to nominals or nominalized verbs. These resultative case suffixes create secondary predicates that are not as restricted as their English counterparts; they can be predicated of the subject of a transitive verb (25c). Moreover, even the special resultative case suffix is not restricted: two secondary predicates with the resultative suffix can occur in the same clause (25b), and a secondary predicate with a resultative suffix can co-occur with a secondary predicate indicating the state of a participant (25c). (25) a.

i Ngardaly-kiji-rni ngula¼ji over-turn-npst that¼top ii warlu-ngka kuyu wardapi fire-loc meat goanna

yangka kuja¼ka yapa-ngku anaph when¼prs person-erg yurnmi-karda cooked-transl


85

iii ngardaly-ngardaly-kiji-rni kuyu wardapi linji-kijaku. over.over-turn-npst meat goanna dry-lest ‘Ngardaly-kijirni is when a person keeps turning a goanna over on the fire from one side to the other to cook it and to prevent it from becoming dried out [lest it become dry].’ [Warlpiri Dic.: ngardaly-kijirni: PPJ

Secondary Predication and Adverbial Modification: The Typology of Depictives

Prepositions and Adverbial Particles

Identity and Predication

Plural Predication

Adverbs and Adverbial Adjuncts at the Interfaces

The typology and dialectology of Romani

The Typology of Asian Englishes

The Typology of Semantic Alignment

Philosophical Method and The Theory of Predication and Identity

Truth & Predication

Modification of Proteins

Surface Modification of Textiles

Truth & Predication

Aristotle's Theory of Predication (Philosophia Antiqua)

Behaviour Modification: Principles and Procedures

RNA Interference Editing and Modification

Behavior Modification: Principles and Procedures

Plasma-surface modification of biomaterials

Predication and Simulation Methods for Geohazard Mitigation

Copular Clauses: Specification, Predication And Equation

Polymer Modification

Polymer modification

FYI. Theory and typology of information packaging

The Secondary Spectrum of Hydrogen and the Occurrence of H3+

Synthesis of secondary amines

Approaches to the Typology of Word Classes

Behavior Analysis And Behavior Modification

Characterization and Chemical Modification of the Silica Surface

Chemical Modification of Lignocellulosic Materials

Chemical Modification of Lignocellulosic Materials

Lifestyle Modification for the Prevention and Treatment of Hypertension

Secondary Predication and Adverbial Modification: The Typology of Depictives

Prepositions and Adverbial Particles

Identity and Predication

Plural Predication

Adverbs and Adverbial Adjuncts at the Interfaces

The typology and dialectology of Romani

The Typology of Asian Englishes

The Typology of Semantic Alignment

Philosophical Method and The Theory of Predication and Identity

Truth & Predication

Modification of Proteins

Surface Modification of Textiles

Truth & Predication

Aristotle's Theory of Predication (Philosophia Antiqua)

Behaviour Modification: Principles and Procedures

RNA Interference Editing and Modification

Behavior Modification: Principles and Procedures

Plasma-surface modification of biomaterials

Predication and Simulation Methods for Geohazard Mitigation

Copular Clauses: Specification, Predication And Equation

Polymer Modification

Polymer modification

FYI. Theory and typology of information packaging

The Secondary Spectrum of Hydrogen and the Occurrence of H3+

Synthesis of secondary amines

Approaches to the Typology of Word Classes

Behavior Analysis And Behavior Modification

Characterization and Chemical Modification of the Silica Surface

Chemical Modification of Lignocellulosic Materials

Chemical Modification of Lignocellulosic Materials

Lifestyle Modification for the Prevention and Treatment of Hypertension

Recommend Documents