CAMBRIDGE STUDIES IN LINGUISTICS General editors: §. R. ANDERSON, J. BRESNAN, B. COMRIE, W. DRESSLER, C. EWEN, R. HUDDLESTON, R. LASS, D. LIGHTFOOT, J. LYONS, P. H. MATTHEWS, R. POSNER, S. ROMAINE, N. V. SMITH, N. VINCENT PossessionInihis series 52. MICHAEL 5. ROCHEMOWT and PETER W. CULICOVER: English focus constructions and the theory of grammar theory forthe generative Sf EvE Sweersen: Prom etymology to pragmatics: metaphorical and cultural aspects of semantic stracture 55. REGINA BLASS: Relevance relations in discourse: a study with special reference to Sisala 56 ANDREW CHESTERMAN: On definiteness: a study with specal reference to English ‘and Finnish 57 ALLESSANDRA GIORGI and GuisePre LONGODARDI: The syntax of noun phrases configuration, parameters and empty categories SON: A-Morphous morphology 63. Lesuey stiRLino: Switch reference and discourse 1: The sytan ad pragmatics of anaphora /AMMRECHY. Information structure and sentence form: Topi, focus, andthe reretations of deco n {URZIO: Principles of English stress B HAWKINS: A performance theory of order and constituency " - HARRIS and LYLE CAMPBELL: Historical syntax in crosfinguistic WAEOEMAN: The syntax of negation Syntax and parsing 82. JOHN. mM, ANDERSON: A notional theory of symtactic categories 83. BERND WetwE: Possession cognitive sources, forces and grammaticalzation Supplementary volumes LILIANE HAEGEMAN: Theory and desription in generative syntax: A cae study in West Flemish faves wacknouse: Thelen ld of ast en NiKOLAUS RIT: Quantity adjustment: Vowel le Middle English Earlier issues not listed are also available POSSESSION Cognitive sources, forces, and grammaticalization BERND HEINE Profesor of African Stules, University of Cologne CAMBRIDGE UNIVERSITY PRESS‘CAMBRIDGE UNIVERSITY PRESS ‘Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sto Paulo For Tom Givin Published inthe United States of America by Cambridge University Press, New York le: www.cambridge.orp/9780521550376 ‘© Cambridge University Press 1997 published 1997 A catalogue record for this publication is availabe rom the Briish Library Library of Congress Cataloguing in Publication data Heine, Bernd, 1939 ive sources, ores, and gram p._ em. (Cambridge Stadis in Ling Includes bibliographical references and inde. ISHN0521 550378 $83) 1. Series. P299P67HAS 1997 415-de20 96-26940 CIP 5037-6 hardback 17-8 hardback 978-045: 0: ISBN-13 978-0-521-02413-6 paperback ISBN: 10,0-521-02613-7 paperbackContents List of tables and figures Preface Abbreviations The state Introduction ‘Some possessive notions Problems ‘The process Sources Grammati Targets How to reconstruct schemas Language-internal variation Schemas and possessive notions Further issues tion On attributive possession From source to target, Specification On ‘possessor ascension’ On inalienability Attributive and predicative possession Conclusions From possession to aspect Parallels Specifying possessionExistence, possession, location, and other domains Conclusions Evaluation Alternative approaches Event schemas On categories and universals ‘On explanation Conclusions Appendix: A world-wide survey of ‘have'-constructions References Index of Authors Index of Languages, Index of Subjects 195 202 207 209 209 22 224 28 233 240 245 264 267 270 Tables Definiteness vs. indefiniteness in ‘have'- and ‘belong’- constructions: Typical associations. Some prototypical properties of possessive notions. A formulaic description of schemas used for the expression of predicative possession. Source schemas for ‘have'-constructions in 100 languages according to continents. Contrastive properties of source schemas for predicative possession, ‘Typical correlations between source schemas and ‘have’ and ‘belong’-constructions. ‘The main possessive constructions in Manding and the notions expressed by them. ‘The main stages in the development from location to possession in Ewe, ‘The main stages in the development from existence to possession in Ewe. ‘The main constructions of predicative possession in Ewe and the notions expressed by them, ‘Typical participant encoding in ‘have'-constructions according to source schema. A formulaic description of source schemas used for the expression of attributive possession, ‘The distribution of inalienable and alienable nouns in Pima-Papago, Correspondences between event schemas and Locker’s Propositional types. Correspondences between event schemas and the classifications proposed by Hengeveld (1992) and Stassen (1995). iionships according to Casagrande and Hale 30 39 47 15 a1 92 123 126 128 133 135 181 210 213 223LL 2. 3. Figures A prototype characterization of possessive notions. The Overlap Model. A radial network of genitival meanings (based on Nikiforidou 1991), 156 Preface ‘The claim made in this work is that the structure of grammatical categories is predictable toa large extent once we know the range of possible cogni- tive structures from which they are derived, This claim is tested here reference to one example, viz. the structure of predicative possession. It ‘would seem that most of the possessive constructions to be found in the lan- ‘guages of the world can be traced back to a small set of basic conceptual patterns. These patterns are ident ied with reference to the principles of those works. The present work has profited greatly from the assistance and co-opera- tion of a number of colleagues. Most of all, my gratitude is due to Ulrike Claudi who, in addition to accompanying my research with constructive criticism, also gave me access to her unpublished paper on “To have or not to have: on the conceptual base of predicative possession in some African languages’ (Claudi 1986), which already contains a number of the basic notions to be discussed below. Furthermore, my gratitude is due to the members of the research team working on grammaticalization at the University of Cologne, in particular to Tom Giildemann, Ingo Heine, Christa Kilian-Hatz, Christa Kénig, Tania Kouteva, Heinz Roberg, and ‘Mathias Schladt, and to my colleagues Leila Behrens, Jirgen Broschart, Brugman, Joan Bybee, Norbert Cyffer, Karen Ebert, Michel in, Suzanne Fleischman, Zygmunt Frajzyngier, Orin Gensler, Paul Hopper, George Lakoff, Frank Lichtenberk, Doris ld Reh, Hans-Jirgen Sasse, Fritz Serzisko, Leon Stassen, lz, Thomas Stolz, Eve Sweetser, Leonard Talmy, Elizabeth Christelxiv Preface ‘Traugott, and Peter Trudgill for critical comments and advice. I also wish to thank Hassan Adam (Swahi and Mohamed Touré (Bambara) for their patience in providing me with information on their mother tongue, Haspelmath, Tania Kouteva, Dirk Otten, Thomas Stolz, and Nigel Vincent for substantive comments on earlier versions of the manuscript on which this book is based, Finally, 1 wish to thank the Volkswagen-Stiftung (Volkswagen Foundation) for having sponsored my visit to the University of California at Berkeley and Stanford University during the Winter Semester of 1993/4, and to the Deutsche Forschungsgemeinschaft for generously sponsoring ‘my research on grammaticalization and the development of grammatical functions. Abbreviations subject of a transitive or ditransitive verb ablative case absolute absolutive abstract possession ry body-part term carrier noun class classifier classifier i connector copula dative case marker of definiteness demonstrative EXIST F Foc GEN MP INAL IN/A INA existential marker feminine gender inalienable possession inanimate alienable possession inanimate inalienable possession instrument imperfective kinship term locative ‘masculine gender noun neuter gender non-feminine gender non-specific object of a transitive verb oblique case possessivexvi Abbreviations PRES PRs PRT @ REC s sc spcr SPEC present tense TAM presentative marker particle TEMP question marker TNS recipient Tor subject of an intransitive UNSP verb 1 singular 2 specifier 3 specific 4, ete. subject of intransitive verb tense, aspect, and modality temporary possession tense topic unspecified third-person first person, class 1 second person, class 2 third person, class 3 class 4, etc. 1 The state 1.1 Introduction Possession is @ universal domain, that is, any human language can be expected to have conventionalized expressions for it. Nevertheless, when working on the linguistic expression of possession one is likely to be con- fronted with a number of problems. One of these problems relates to the cognitive nature of possession. Possession belongs to the kind of concepts that tend to be described as being inherently vague or fuzzy. The English verb have has been called ‘col- ‘orless’ (Buck 1949:740) and the possessive concepts expressed by it are said to be indeterminate; have has even been described as an ‘unsuitable lexical item’: like @ tree has leaves. In passing we may observe the ‘have, which can appear in a wide variety of contexts, and express a variety of very diferent relations: You have a cold, Mary has {sick grandmother, Bill has a good job, Who has the exact rime? and so on. All of these at least share the feature that if you were to take away your , Mary's grandmother, or Bill's ob, you, Bill, and Mary would still be there intact. But in addition to a tree has leaves, we can say a tree has brane has lar fashion, possession has occasionally been described as a concept that is neither conceptually nor linguistically basic, or that is not of universal significance. Miller and Johnson-Laird say: ‘The ordinary business of lfe can be conducted without explicit recourse to the conceptual core of possession; people need only grasp the interre- lations between having, buying, giving, and so forth, Indeed, many people hhave only the vaguest notion of the conceptual core. There may be whole that do not grasp the core exp!2 The stare Irrespective of how this statement is to be interpr am not aware of any language that would not i it means for expressing, for example, ‘This is my wife’ or ‘I have no food.’ But pos are used for a wide range of experiences and concept some authors therefore claim that linguistic expressions for possession are meaningless, that is, that English items like have or of are semantically vacuous (ef. Bach 1967). Furthermore, the wide range of meanings expressed by possessive constructions has induced some authors to propose fairly abstract descriptions of possession. For Langacker (1993:8), for ‘example, the various uses of the English genitive have in common that one entity ‘is invoked as a reference point for purposes of establishing mental ‘contact with another’, and some authors would go so far as to claim that possession simply involves any abstract relation between two entities. We noted above that the range of meanings expressed by possessive con- structions is so wide that referring to all of these meanings as ‘possessive’ ‘ones would be misleading, and that possessive expressions are likely have ‘other, non-possessive, meanings in addition. For example, there is no doubt ‘that (Ia) is an instance of posse: what about ( ()— (@ Lizhasacar. (b) Lizhasa problem. To take care of cases like (1b), alternative terms such as ‘relational’, ‘asso- ", ete. have been proposed to refer to concepts that include posses- Creissels 1979). Ina similar vein, Hawkins in English, is merely one particular case of what, following Chomsky (1972), he calls ‘intrinsic connection’; whether or not possession obtains, he argues, is determined by context and depends on world knowledge and possibly bel ‘Nevertheless, even if possessive constructions tend to cover a wide range of relations or associations be ample, reversing the two participants in (1) would res sentence: *A car has Liz is not considered Similarly, while the cat's tail is acceptable, the i's eat is usually not. LLL How to deal with possession In earlier accounts, interest in possession has focussed on how to delimit this domain and to define its status vis-a-vis other ontological entities. A ‘number of problems have been highlighted, the main ones being discussed 1.1 Introduction 3 in section 1.4. Among the descriptive concepts that were proposed to deal with possession, ‘control’ has perhaps most frequently been named, for obvious reasons: prototypical instances of possession imply some kind of control of the possessor over the possessee (see especially Hagége 1993:93f1.). The relevance of the notion becomes debatable in the case of inalienable possession, especially in many cases of kinship relations (eg John's father, John has three uncles), and in the case of inanimate possessors (eg. the windows of this house, This house has ten windows). Other problems surrounding the use of the term ‘control’ relate to the fact th associated ability of the possessor to manipulate the possessee; the qu manipulation includes the possi ‘The phrase my head implies cont Tcan manipulate my head ‘ways but I cannot normally terminate while some authors argue that body-part pos- session does not involve control (Chappell and McGregor 1996b), others say it does (ef. Lynch 1973:6; Brugman 1988:229), jother frequently named property of possession is contiguity of loca- ty between possessor and possessee (Brugman As we will see in section 1.3, thi tances of possession, but, like control, does in fact relevant to canonical ‘not qualify as a definitional criterion, This does not exhaust the list of concepts that have been proposed to define or characterize possession. One might mention, for example, Langacker's (1987) definition of the relation between the possessor and Possessee in terms of a ‘sphere of influence’, or Brugman’s (1988:231ff) “schema of interest or involvement’. Students of law and other people draw a di ‘and ownership. This distinction is in fact releva between possession the present topic and I not deal wi reat detail, first, because it appears to be highly culture-speci interest is primarily with cross-linguistic regul hhumber of quite divergent ways in which this. evant literature. For example, while some authors argue ownership are clearly diferent things (eg. Bickerton I two as being essential4 Thestate ‘Another problem concerns the cross-linguistic and cross-cultural signif- icance of the possessive domain. Is it in fact a universal domain, as some argue, or is it culture-specific, that is, does it occur in certain parts of the world but not in others, or did it evolve during certain periods in the history of mankind but not during others? ‘There is a related problem that has to do with the fact that possession has been widely studied in western societies but much less so in other parts of the world. The implications this fact may have for a theory of possession are considerable, For example, Bach (1967:479) notes that the situation in English, where we have a special verb-like form for ‘have’, is almost patho- logical’ considering the fact that constructions in other languages corre- sponding to the English ‘have’-sentences are notoriously varied. Bach's observation was in fact not new; rather, he was echoing what other authors hhad claimed earlier (e.g. Locker 1954; Lafstedt 1963). From such observations it follows that defining possession is perhaps the ‘most crucial problem. For example, should one aim at a definition in terms istic properties? While most linguists will probably answ ition. Two proposals inthis directo Seiler (1983:4-7), possession is essen pattern: he defines possession as ‘the rel ‘conceptual relationship hip between a human being, longings, his cultural and Tb) proposes a cros ‘kinds of semanto-syntat yas the ‘Agent of an A¢ ‘Compared to the former, the ‘Possessor of an Act’- cor have the following properties (Seiler 1977b:174-9): between the two configurations is portrayed in the following formulas, where (2a) represents the active and (2b) the passive form of the ‘Agent of an Act’-configuration, while (2c) isthe ‘Possessor of an Act’-configuration (parentheses indicate that the 1.1 Introduction § (2) Three kinds of semanto-syntactic configurations accordis ier ae igurations according to Seil AGENT ACT (oBsECT) in fact argued for independently by Taylor (1989b:202-3, 1989a:679fT). He views possession as an experiential gestalt, and defines it 48 a prototypical notion involving a constellation of properties st ones listed in (3), oa (3) (@) The possessor is a specific human being, ‘The posseseeis a specific concrete thing (usually inanimate), not an abstract. The por it ssi other le can make use ofthe poses only wh the permsion of the possessor. ps * m= / (©) Therelaonship of possesion ia longterm one, measured in year rather than in minutes or hours ic discourse, the possessor is presented as a referential This does not conclude the list of properties that are associated with pro- totypical instances of possession. Taylor (1989a, 1989b:202) proposes the following properties in addition: the possessor’s rights over the possessce are invested in him/her in virtue of a transaction, i.e. through purchase, the possessor is responsible for the possessee, and stances of possession discussed here ie entire range of these properties. Nevertheless, the more of the instances of possession of the properties listed lisagreement arises as to whether a given linguistic tance of possession one may return to this characteri- Most treatments of the subject do in fact take some highly prototypical instance as a point of departure for understanding and/or defining posses- the reader is referred to Snare (1972) and Miller and Johnson-Laird for what may be said to be classical examples. But things are more6 The state complex: which particular kind of possessive relationship is most ‘proto- typical’ or ‘basic’ depends on the perspéctive one adopts and on the con- ceptual distinctions one decides to consider and ignore, respectively. 1.1.2 The present volume The objective of this work is to study why possession is expressed the way it is, Thus, our concern is with explanation. The observations to be made are based on findings on grammaticalization, in particular on the following, assumptions: (The structure of grammatical categories is predictable to a large extent once we know the range of possible cos from which they are derived. Underlying this assumptions of the following kind: (i) Grammatical categories can be traced back to semantically con- crete source concepts For each grammatical category there is only a small pool of pos- sible source concepts (iv) While the choice of sources is determined primarily by universal ways of conceptualization, itis also influenced by other factors, especially by areal forces. ‘These assumptions are based on research in the course of the last decade within the paradigm of gramm: is referred to the relevant works for more details (especi Heine 1991b; Heine, Claudi, and Hinnemeyer 1991; Hopper and. ‘Traugott 1993; Bybee, Perkins, and Pagliuca 1994; Stolz 1991, 1994). We will argue here that most expr in the languages of the world for what corresponds to ‘have’-constructions in English can be described as con- forming to the assumptions just made. ‘The label ‘have’-construction to be used throughout this book is sugges- tive of a Euro-centric perspective according to which verbs meaning ‘have’ form the or a primary means of expressing possession (Hilary Chappk pc). As we shall see below, a verb corresponding to the notion of a ‘have’- verb in many European languages is much less common than one might be inclined to believe. The book also falls within the scope of what is sometimes referred to as typological universal grammar as developed by Greenberg (1963b; see also 1978a, 19786), Givon (1979, 1995), Comrie (1981), Mallinson and Blake 1.1 Introduction 7 (1981), Bybee (1985), Bybee, Perkins, and Pagliuca (1994), Croft (1991), and others, these works have in common in particular is that they aim at establishing cross-linguistic regularities based on world-wide samples of languages. At the same time, however, the approach used here differs from that tradition in arguing that language structure is derivative of the cogni- tive forces that gave rise to it and, hence, our concern is primarily with extra-linguistic forces. From what has just been said it follows that, more than in previous works ‘on typological universal grammar, our concern will be with explanation, ‘more precisely, with external explanation. The main explanatory parame- ters used are cognition and diachrony. Cognition involves the acquisition, storage, retrieval, and use of knowl- edge. We will confine ourselves to one aspect of cognition, namely to the interrelationship between different concepts and the way linguistic expres- ns used for one of them are extended to also refer to other concepts. We I call the process concerned conceptual transfer, Our task will be to iden- tify and describe salient processes of conceptual transfer relating to the domain of possession and, by doing so, to understand why possessive con- structions are formed the way they are, Conceptual transfer takes place in time and, hence, will be treated as a iachronie notion, Our findings thus are based on hypotheses on diachronic development. This means that the processes that we shall be concerned with can be accounted for with reference to diachronic prin- ciples, and that our findings are falsifiable by means of diachronic evidence. that cor to synchronic language use, must not be at variance with the interpretation these terms would receive if used in a strictly diachronic sense. ‘Thus, the evidence on which our hypotheses rest are lt hand, and diachronic ‘Suppose we discover prior. The evidence for such a conclusion is twofold, attested cases of diachronic change; we know, for ex: time. Second, itis based on generalizations on grammaticalization, accord- ing to whi istic item, like English have, that combines the functions