Deconstructing Constructions [1 ed.] 9789027289605, 9789027205742

This article focuses on the so-called get-passive, frequently regarded as a problematic construction in the linguistic l

194 51 2MB

English Pages 328 Year 2008

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Deconstructing Constructions [1 ed.]
 9789027289605, 9789027205742

Citation preview

Deconstructing Constructions

Studies in Language Companion Series (SLCS) This series has been established as a companion series to the periodical Studies in Language.

Editors Werner Abraham University of Vienna

Michael Noonan

University of Wisconsin, Milwaukee

Editorial Board Joan Bybee

Christian Lehmann

Ulrike Claudi

Robert E. Longacre

Bernard Comrie

Brian MacWhinney

University of New Mexico University of Cologne Max Planck Institute, Leipzig University of California, Santa Barbara

William Croft

University of New Mexico

Östen Dahl

University of Stockholm

Gerrit J. Dimmendaal University of Cologne

Ekkehard König

Free University of Berlin

University of Erfurt

University of Texas, Arlington Carnegie-Mellon University

Marianne Mithun

University of California, Santa Barbara

Edith Moravcsik

University of Wisconsin, Milwaukee

Masayoshi Shibatani

Rice University and Kobe University

Russell S. Tomlin

University of Oregon

Volume 107 Deconstructing Constructions Edited by Christopher S. Butler and Javier Martín Arista

Deconstructing Constructions Edited by

Christopher S. Butler Swansea University

Javier Martín Arista University of La Rioja

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Deconstructing constructions / edited by Christopher S. Butler and Javier Martín Arista. p. cm. (Studies in Language Companion Series, issn 0165-7763 ; v. 107) Includes bibliographical references and index. 1.  Linguistics models. 2.  Grammar, Comparative and general. 3.  Typology (Linguistics)  I. Butler, Christopher, 1945- II. Martín Arista, Javier. P128.M6D43

2009

415--dc22 isbn 978 90 272 0574 2 (Hb; alk. paper)

2008038844

© 2008 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents

Contributors

vii

Introduction   Christopher S. Butler & Javier Martín Arista

xv

Part i. Theoretical issues Innovative coinage: Its place in the grammar   Daniel García Velasco

3

The construction of macro-events: A typological perspective   Johan Pedersen

25

Constructions, co-composition and merge   Beatriz Martínez Fernández

63

A typology of morphological constructions   Javier Martín Arista

85

Part ii. The Lexical Constructional Model: An overview The Lexical Constructional Model: Genesis, strengths and challenges   Christopher S. Butler

117

Levels of description and explanation in meaning construction   Ricardo Mairal Usón & Francisco Ruiz de Mendoza Ibáñez

153

Part iii. Studies of specific constructions Measuring out reflexivity in secondary predication in English and Spanish: Evidence from cognition verbs   Francisco Gonzálvez-García

201

The inchoative construction: Semantic representation and unification constraints   Francisco Cortés Rodríguez

247

Semantic and pragmatic constraints on the English get-passive   Pilar Guerrero Medina

271



Table of contents

Name index

295

Language index

299

Subject index

301

Contributors Christopher s. Butler took a first degree in biochemistry at the University of Oxford and taught biochemistry and microbial physiology for some years at what is now Nottingham Trent University, UK, before making a career move into linguistics. He worked at the University of Nottingham from 1971 to 1992; first in the Department of English Studies and later in the Department of Linguistics, where he was Head of Department from 1986 to 1992. He also directed the university’s Language Centre from 1986 to 1990. In 1992 he took up the post of Head of English Language and Linguistics at what is now York St John University. From 1994 to 1998 he acted as Director of Research at that institution. He was awarded a Professorship of Linguistics in 1994. In 1998 he took early retirement in order to devote more time to research and writing. In 2000 he was made Honorary Professor in what was then the Centre for Applied Language Studies at Swansea University. His main research interests are concerned with theoretical and descriptive issues in functional grammars, with particular reference to English and Spanish. Much of his work is corpus-based and frequently makes use of statistical techniques. A number of publications, particularly in recent years, are concerned with both of the above aspects, attempting to show how insights from corpus linguistics can help us to formulate more adequate functional grammars. He has published around seventy articles, as well as four books, the most recent of these being Structure and function: A guide to three major structural-functional theories (2 vols., John Benjamins, 2003), comparing Functional Grammar, Systemic Functional Grammar and Role and Reference Grammar. He has also co-edited a number of other volumes, including Nuevas perspectivas en Gramática Funcional (Ariel, 1999), The dynamics of language use: Functional and contrastive perspectives (John Benjamins, 2005) and Functional perspectives on grammar and discourse. In honour of Angela Downing (John Benjamins, 2007). He has given presentations, often invited plenary lectures, at many international conferences in Europe, Australia and the USA. He serves on the editorial committees of a number of scholarly journals, and is currently Reviews Editor for Functions of Language (John Benjamins). He is a member of SCIMITAR, a research group based in Santiago de Compostela, Spain. E-mail: [email protected] Francisco Cortés Rodríguez is Professor of English Language and Linguistics in the Department of English and German Philology, University of La Laguna, Spain. His areas of research interests are word-formation, lexical structure, the interaction between lexis and grammar within functional and cognitive models, and diachronic lexicology. He has been the head of various research projects funded by the Spanish

  Deconstructing constructions

Ministry of Education and the Regional Government of the Canary Islands (HUM2005 07651-C02–01, BFF2002 0659; PB1999/136) that dealt with the structure of the lexicon and linking mechanisms in Old English. The research carried out within these projects stems from the application of the theoretical and methodological guidelines for the analysis of historical lexica that were propounded in Cortés Rodríguez, F.J. and R. Mairal Usón “A preliminary design for a syntactic dictionary of Old English on semantic principles”, in Díaz Vera, J. (Ed.) A changing world of words. Studies in English historical lexicography, lexicology and semantics, 3–46 (Rodopi, 2002). An extensive application of this proposal is offered in the work entitled “Anglo-Saxon verbs of sound: semantic architecture, lexical representation and constructions”, Studia Anglica Poznaniensia 42 (2006), 249–284 (co-authored with M. González Orta). An important contribution to the understanding of the interface between Anglo-Saxon motion lexical predicates and grammatical structures is to be found in Cortés Rodríguez, F.J. and D. Torres Medina “Old English verbs of running: linking semantic representation and morphosyntactic structure”, Folia Linguistica Historica XXIV/I (2003), 153–174. Other relevant contributions are devoted to the field of derivational morphology in present-day English and Spanish, as is the case of Cortés Rodríguez, F.J. and M.J. Pérez Quintero “On the syntax-semantics interface in word-formation: The case of –er nominalizations”, in Mairal, R. and M.J. Pérez Quintero (Eds) New perspectives on predicate structure in Functional Grammar, 213–246. (Mouton de Gruyter 2002). The paper “Derivational morphology in Role and Reference Grammar: a new proposal” RESLA (Revista Española de Lingüística Aplicada) 19 (2006), 41–66, is a study that offers a wider perspective on how this topic can be approached within functionally-oriented theories. E-mail: [email protected] Daniel García Velasco is Lecturer in English Linguistics at the University of Oviedo, Spain, where he teaches courses on syntax and semantics from both a functional and a generativist perspective. He is an active researcher in Simon Dik’s Functional Grammar and its successor Functional Discourse Grammar. In 2003, he published Funcionalismo y lingüística: La Gramática Funcional de S.C. Dik (University of Oviedo Press) an exhaustive critical presentation of Dik’s model in Spanish. He has contributed to the development of the theory in different areas. In 2002, together with Kees Hengeveld (University of Amsterdam, The Netherlands), he published “Do we need predicate frames?” (Mairal Usón, R. and M.J. Pérez Quintero (Eds), New perspectives on argument structure in Functional Grammar, Mouton de Gruyter) an article in which the authors propose to separate lexemes from the frames in which they appear in the lexicon and dispense with the notion of “predicate frame”. More recently, he has proposed a new approach to the semantics of lexical items with significant consequences for the organization of the lexicon in Functional Discourse Grammar (“Lexical competence and Functional Discourse Grammar”.



Contributors  

Alfa – Revista de Lingüística, 51). The adequacy of this proposal has been tested in a forthcoming article on the process of conversion in the English language (“Conversion in English and its implications for Functional Discourse Grammar”, Lingua, 2008). He has also co-edited the volume The Noun Phrase in Functional Discourse Grammar (Mouton de Gruyter), together with Jan Rijkhoff (University of Aarhus, Denmark). This book contains a contribution by Daniel García Velasco (“Functional Discourse Grammar and extraction from (complex) noun phrases”), in which the author offers a pragmatics-based approach to Ross’s Complex Noun Phrase Constraint. E-mail: [email protected] Francisco Gonzálvez-García is Lecturer in English Language and Linguistics in the Department of English and German Philology, University of Almería, Spain. His research interests center on the areas of syntax, semantics, pragmatics and discourse analysis, as well as on functionalist and constructivist approaches. His early work, deriving from his Ph.D. received at the University of Bologna (Royal Spanish College) “The syntax-semantics interface in complex-transitive complementation in contemporary English”, was geared towards an eclectic approach to the syntactic analysis and semantico-pragmatic import of verbless complement clauses in English. This synthesising analysis is outlined in “A modality view of predicate selection in small clauses” (1997; Texas Linguistic Forum 38:  101–119). A much more ‘delicate’ version of this account, drawing on the Goldbergian version of Construction Grammar and applied to both English and Spanish subcategorized verbless complement clauses, has been recently published as “Re-constructing object complements in English and Spanish” (in Martínez Vázquez, M. (Ed.) (2003), Gramática de Construcciones: Contrastes entre el inglés y el español. Huelva: Grupo de Gramática Contrastiva, pp. 17–58). Over the last few years, he has published a number of articles focusing on argument structure in English and Spanish from a constructionist standpoint. His most recent publications offer a fine-grained comparison of 11 functionalist, cognitivist and/or constructionist models in terms of a list of 36 features (with Christopher Butler, “Mapping functionalcognitive space”, Annual Review of Cognitive Linguistics 4 (2006), 39–96), a Goldbergian analysis of passive verbless configurations in Spanish as constructions in their own right (“Passives without actives: evidence from verbless complement clauses in Spanish”, Constructions SV1–5/2006), an analysis of coercion via reflexive pronouns in secondary predication in English and Spanish (“‘Saved by the reflexive’: Evidence from coercion via reflexives in verbless complement clauses in English and Spanish”. Annual Review of Cognitive Linguistics 5 (2007), 193–238) and a Goldbergian analysis of the family of object-related depictives in English in Spanish (“The family of object-related depictives in English and Spanish: Towards a usage-based, constructionist analysis”, Language Sciences (2008; in press). [Available online from www.science-direct.com]. E-mail: [email protected]

  Deconstructing constructions

Pilar Guerrero Medina is Lecturer in English Grammar at the University of Córdoba, Spain. She has mainly conducted her research within a functionalist framework, focusing on the relationship between lexis and grammar on the one hand, and grammar and discourse on the other. In her early article “A prototype approach to transitivity: its implications for the FG typology of SoAs” (in Olbertz, H. et al. (Eds) (1998) The structure of the lexicon in Functional Grammar. Studies in Language Companion Series. John Benjamins, 215–232) she integrates the notion of prototypical transitivity within S.C. Dik’s Functional Grammar theory. Other FG articles are: “Reconsidering aspectuality: interrelations between grammatical and lexical aspect” (Working Papers in Functional Grammar 75 (2001), 1–11) and “Object assignment in S.C. Dik’s Functional Grammar: marginal accessibility in English” (Odisea. Revista de Estudios Ingleses 4 (2004), 9–57). A constructionist approach to the study of argument structure has been adopted in two of her later publications: “A constructional account of ‘lexical ergativity’ in English” (in Carretero González, M. et al. (Eds) (2002) A life in words. University of Granada, 199–215) and “Lexically ergative constructions in English and Spanish” (in Iglesias Rábade, L. and S. Doval (Eds) (2002) Studies in Contrastive Linguistics. University of Santiago de Compostela, 465–473). Her most recent work explores the grammar-discourse interface: “Cardinal transitivity in foregrounded discourse. A contrastive study in English and Spanish” (in Butler, C.S. et al. (Eds) (2005) The dynamics of language use: Functional and contrastive perspectives. John Benjamins, 349–369), which aims to link the semanticogrammatical properties of transitivity to its discourse function in English and Spanish, and “A discourse-based approach to English de-transitivization: middle vs passive” (in Guerrero Medina, P. and E. Martínez Jurado (Eds) (2006) Where grammar meets discourse. Functional and cognitive perspectives. University of Córdoba, 133–150). E-mail: [email protected] Ricardo Mairal Usón has been a Full Professor of English Language and Linguistics in the Department of Modern Languages at the Spanish National DistanceLearning University (UNED) since 2002. His main areas of research interest are the architecture of the English lexicon, the representation of lexical knowledge, linguistic universals and the interactions between lexical semantics, syntax and morphology with particular reference to theoretical models, both formal and functional. He has been the head of various research projects funded by the Spanish Ministry of Education and the Regional Government of Madrid and has additionally participated in other projects that deal with various aspects of language research such as terminology, the compilation of lexical representations and linking mechanisms in Old English, natural language processing and the development of lexical databases for lexicography. He has co-authored or co-edited a number of books including: Nuevas perspectivas en Gramática Funcional (Ariel, 1999), Constructing a lexicon of English verbs (Mouton de Gruyter, 1999), New perspectives on argument structure in Functional Grammar



Contributors  

(Mouton de Gruyter, 2002), En torno a los universales lingüísticos (Cambridge University Press, 2003), Linguistic Universals (Cambridge University Press, 2006). He has also published over fifty scholarly articles which have appeared in specialised national and international journals. He has served as a scientific committee member for several specialised journals, including Cuadernos de Investigación Filología, Atlantis, RESLA, Estudios Ingleses de la Universidad Complutense, Onomázein and Functions of Language. He has done occasional review work for Cognitive Linguistics and Language Sciences and has also been a member of the advisory committee of various international conferences on Role and Reference Grammar. He has also lectured extensively as keynote speaker at national and international conferences on Applied and Theoretical Linguistics. E-mail: [email protected] Javier Martín Arista was appointed senior lecturer in English at the University of La Rioja in 1996; two years after defending his Ph.D. dissertation, entitled SVO and passive order in English: Synchronic, diachronic and typological perspectives (Servicio de Publicaciones de la Universidad de Zaragoza). He has been a visiting scholar at the Universities of Sheffield (1996), Pennsylvania (1999), SUNY Buffalo (2000) and Amsterdam (2001) and has delivered lectures by invitation at several European and American universities, including Newcastle, Strathclyde, Sheffield Hallam, Queen Mary London, Copenhagen and Toronto. He delivered a plenary lecture at the 2004 International Conference of the Spanish Association for Applied Linguistics. He has published more than fifty book chapters and articles in journals specialising in theoretical linguistics, English studies and diachronic studies. He has made around seventy presentations at conferences, of which around thirty were international. He has supervised four doctoral dissertations in theoretical linguistics and is supervising another six dealing with Old English derivational morphology. He has taken part in about twenty research projects in synchronic and diachronic linguistics, and at the moment is the leading researcher of the Research Group in Functional Grammars of the University of La Rioja. Dr Martín Arista was a member of the editorial board of Shakespeare Quarterly-WSB until 2004; and is currently a reviewer for ATLANTIS, STVDIVM, Journal of English Studies, Revista de Lingüística y Lenguas Aplicadas, Revista Española de Lingüística Aplicada and The Open Applied Lingusitcs Journal. He is also editor of RæL-Revista Electrónica de Lingüística Aplicada. He has been a reviewer of the National Agency for Evaluation and Planning since 1999 and was Pro-Vice-Chancellor (Research and Staff) of the University of La Rioja from 2001 until 2004. E-mail: [email protected] Beatriz Martínez Fernández teaches English Language and Linguistics in the Department of Modern Languages, University of La Rioja, Spain. She obtained a Ph.D. in Linguistics from the University of La Rioja in 2007 for her thesis entitled Verbs of

  Deconstructing constructions

pure change of state in English: projection, constructions and event structure. Her research focuses on the syntax-semantics interface, lexicology, and pragmatics, mainly from a functionalist and cognitivist approach. She has published articles on ergativity and possessed-raising constructions, markedness and subject prototypicality, nominal aspect, syntactic and semantic valence, and, from a more pragmatic perspective, on the generation of humour. She has been a visiting researcher at SUNY at Buffalo, University of Essex and Queen Mary, University of London, and has collaborated in various research projects on diachrony and functional theories of language. She is a member of the Research Group of Functional Grammars of the University of La Rioja, and she is co-editor of RæL (Revista Electrónica de Lingüística Aplicada [Electronic Journal of Applied Linguistics]). E-mail: [email protected] Johan Pedersen is Associate Professor in Spanish Language and Linguistics in the Department of English, Germanic and Romance Studies, University of Copenhagen, Denmark. His research interests are in syntax and semantics, including constructionist approaches to grammar. Of particular interest is Spanish grammar in a contrastive/typological perspective. He received a Ph.D. in Spanish linguistics at the University of Copenhagen (2000) for a dissertation on aspects of the relation between grammar and pragmatics in Spanish, “Conectores y relaciones gramaticales. El caso de aunque en español” (Connectors and grammatical relations. The case of aunque in Spanish). He has published a number of articles on different aspects of Spanish grammar and linguistics. Selected publications: “Lexical and constructional organization of argument structure. Typological differences” (in Zlatev, J. et al, Studies in language and cognition, Cambridge Scholars Publishing, forthcoming). “Organización léxica y esquemática – una perspectiva tipológica (Lexical and schematic organization  – a typological perspective)” (Papers from the VIII congress on general linguistics 2008; Madrid, forthcoming). “The impersonal se-construction in Spanish. Constructional variation and change” (Constructions, 2005, 1, 1–49). “Reflexive intensification in Spanish. Toward a complex reflexive?” (in Fortescue, M. et al (Eds), Selected papers from the XVIth International Conference on Historical Linguistics, John Benjamins 2005). “Construction Grammar and Second Language Acquisition. A cognitive understanding of language in a contrastive perspective” (in Hansen, H.L. (Ed.), Disciplines and interdisciplinarity in foreign language studies, Copenhagen: Museum Musculanums Forlag, 2004). “The Spanish connective and verbal mood. Changes in the mood selection of de ahí/aquí que” (in Eksell, K. & T. Vinther (Eds), Explanatory models on language change within the verbal system, Peter Lang Publishers, 2006). “La base discursiva de la oración compuesta (The discourse basis of the compound sentence)” (in Delbecque, N. (Ed.), Aproximaciones cognoscitivo-funcionales al Español, Foro hispánico (Cognitive-functional approaches



Contributors  

to Spanish, Hispanic Forum, Rodopi, 2003). He is editor (Section of Spanish Linguistics) of the journal Revue Romane (John Benjamins). E-mail: [email protected] Francisco Ruiz de Mendoza Ibáñez has been a full professor of English Linguistics at the University of La Rioja since 1999. He has been the head of a number of research projects financed by the Spanish Ministry of Education and Science that deal with language acquisition, cognitive modeling, and the compilation of lexical databases for lexicography. He has published close to 100 articles and book chapters. Many have appeared in journals such as Journal of Pragmatics, Language and Communication, LAUD-Essen, and in series such as Cognitive Linguistics Research (Mouton), Topics in English Linguistics (Mouton), Pragmatics and Beyond (John Benjamins), Lodz Studies in Language (Peter Lang). He has been invited to present his research in universities and other research institutions in Madrid, Pisa, Galati, Hamburg, Copenhagen, Lodz, Las Palmas, Provo, Beijing, among other places. He has also been an invited speaker in national conferences such as the International Conference on Cognitive Linguistics (Pavia 2003), EURESCO (Granada 2004), International Contrastive Linguistics Conference (Santiago 2005), Converging and Diverging Tendencies in Cognitive Linguistics (Dubrovnik 2005, 2008), Perspectives on Metonymy (Lodz 2005), the 10th International Cognitive Linguistics Conference (Krakow 2007), and the Seventh International Conference on Researching and Applying Metaphor (Cáceres 2008). He serves on the editorial and scientific boards of a number of journals such as Miscelánea, Atlantis, Estudios de Filología Moderna, Jezikoslovlje, Estudios Ingleses de la Universidad Complutense, Revista Alicantina de Estudios Ingleses, Revista Española de Lingüística Aplicada, ITL-International Journal of Applied Linguistics, Revue Romane, and Cognitive Linguistics. He was the editor of the Revista Española de Lingüística Aplicada from 1998 to 2005 and is currently the editor of the Annual Review of Cognitive Linguistics (John Benjamins) and co-editor of the series Applications of Cognitive Linguistics (Mouton de Gruyter). He has done occasional review work for the Journal of Pragmatics, TEXT, Cognitive Linguistics, Journal of Applied Linguistics, and has been a member of the advisory committee of a number of international conferences including EURESCO in 2003; and the International Cognitive Linguistics Conference in 2005 and 2007. He is the current president of the Spanish Association of Applied Linguistics. E-mail: [email protected]

Introduction Christopher S. Butler & Javier Martín Arista These first years of the new century are turning out to be an exciting time in linguistics. In the last decade or so, we have finally begun to witness a breaking down of the parochialism which has characterised much linguistic theorising for far too long. After years of controversy characterised by partisan and often militant positions, we have perhaps reached a point where eclecticism (in the positive sense of the feasibility of combining concepts and/or methodological insights from different but compatible approaches in a coherent, consistent way) is no longer and embarrassing side-effect of particular fundamental assumptions, but is seen, at least in some quarters, as adding value to linguistic theories. Dialogue between different ways of conceptualising the linguistic enterprise is gaining momentum, and studies of the relationships between approaches are appearing in the literature. One obvious area for such comparative studies centres on the debate between ‘formalist’ mainstream generative linguistics and those approaches which label themselves as ‘functionalist’ and/or ‘cognitivist’. However, as Culicover and Jackendoff remind us, it is also worth comparing the alternatives among themselves: We would like to see more investigation that compares frameworks dispassionately, for it is only by doing such comparisons that we pit them against each other scientifically rather than merely sociologically. In particular, most of the alternative frameworks conceive of themselves primarily in opposition to mainstream generative grammar and not in comparison to one another – despite the evident conceptual affinities among them that we have reviewed above. The goal of such comparison would be to distil out of each framework the essence of what it thinks language is like, free of the peculiarities of its formalism. Again, this is hard; it’s tough enough to master one framework, much less know two or more well enough to do a comparison. But it can be done, especially in the context of collaboration. Given the fragmentation of frameworks that prevails today, such integration is crucial for the future of the field. (Culicover & Jackendoff 2005: 546–547)

Culicover and Jackendoff ’s own work offers a very promising synthesis of ideas from formalist linguistics and proposals which are closely related to what functionalists and constructionists have been saying for some time. Discussion of the relationships between formalism and functionalism can be found in Newmeyer (1998, 2003, 2005a, 2005b, 2005c), Darnell et al. (1999) and Butler (2003: Chapter 1, 2006), while functional

  Christopher S. Butler & Javier Martín Arista

and cognitive and/or constructivist approaches are compared in Nuyts (2005, 2007), Butler and Gonzálvez-García (2005) and Gonzálvez-García & Butler (2006). Particularly relevant to the present book are relationships between functionalist and cognitivist and/or constructionist approaches. Gonzálvez-García & Butler (2006) show, through close analysis of eleven approaches to language covering a wide spectrum from Jackendoff ’s parallel architecture model to radically functionalist emergentist proposals, that there is a core of features which are relatively homogeneous across the models investigated, while other sets of features mark differences between more clearly functionalist and essentially cognitivist approaches within what the authors dub ‘functional-cognitive space’. Although some of the contributions to the present volume are firmly placed within models which are situated in either the more clearly functionalist or the cognitivist areas of functional-cognitive space, the majority demonstrate how the commonalities between the two areas may be insightfully exploited, and the tensions (whether real or alleged) resolved, within a single overarching framework. One particularly interesting and important aspect of recent work within functionalist models is the attention being given to work in constructionist approaches, particularly those with a clearly cognitive basis, such as Goldberg’s model of Construction Grammar, which she now refers to as Cognitive Construction Grammar (Goldberg 2006: 214). A key distinction here is between two ways of looking at the relationship between semantics and syntax, particularly in the area of argument structure. Functionalist approaches have traditionally taken a projectionist view, according to which the syntax associated with a verbal predicate can be determined on the basis of the semantic argument structure. Proponents of most constructionist approaches, on the other hand, show that verb semantics is insufficient to account for all the structural constituents which may appear with a particular type of verb, and conclude that the construction itself contributes a meaning, which may even override the contribution made by the verbal predicate.1 The concept of construction is thus a central point around which much debate currently revolves. Functionalist models have begun to incorporate insights from Construction Grammars: for instance, Role and Reference Grammar, though essentially projectionist in its orientation, also makes use of constructional schemas (see Van Valin 2005: 131–135), and Belgian linguists whose background is in Systemic Functional Grammar have for some years now been combining ideas from this model with insights from cognitive and/or constructionist approaches

.  Hans Boas, however, argues that altough constructional semantics is important, it is insufficient in itself to predict the whole range of constructions from the point of view of decoding and, more crucially, from encoding. He therefore invokes mini-constructions, which capture delicate lexico-semantic and encyclopaedic information.



Introduction  

in order to give a more powerful account of constructions (see e.g. Davidse 2000; Laffut & Davidse 2000; Taverniers 2002). With the above relationships as a backdrop, a workshop was organised by the Functional Grammars Research Group at the University of La Rioja in Logroño, Spain, in May 2006; on the topic ‘Deconstructing Constructions’. According to the philosopher David Allison, deconstruction “signifies a project of critical thought whose task is to locate and ‘take apart’ those concepts which serve as the axioms or rules for a period of thought” (Allison 1973: xxxii, note 1). The idea behind the workshop was thus to bring together a group of linguists, some from functionalist backgrounds, others working within Construction Grammars, to explore the concept of construction from different angles and try to arrive at a better understanding of what a construction is, and what roles constructions play in the frameworks which constitute functionalcognitive space. Six of the chapters of the present volume are derived from papers given at the Logroño workshop. This book is heterodox not only in bringing together different theoretical approaches, but also in the range of methodologies it illustrates. Both qualitative and quantitative approaches are taken, in an effort to shed maximum light on the complex issues under discussion. At the same time, this set of articles preserves a perspective on tradition, for example tracing the historical background of recent proposals. The volume is organised in three sections: the first consists of papers dealing with a range of particular theoretical issues; the second is devoted to a model, the Lexical Constructional Model, which is eclectic in the positive sense defined earlier, combining functionalist and constructionist ideas into a single framework; the third section consists of studies of particular constructions. In the first section, the paper by Daniel García Velasco, ‘Innovative coinage: its place in the grammar’, examines coinages such as denominal verb conversion (e.g. faxN → to faxV , bottleN → to bottleV), and argues that these should be seen as an integral part of the grammar. The paper then explores the consequences of this position for both functional and constructional approaches to language. Johan Pedersen’s paper, ‘The construction of macro-events: a typological perspective’, presents a critical constructionist analysis which takes as its point of departure the work of Talmy on satellite-framed Germanic languages vs. verb-framed Romance languages, and argues that Talmy’s generalisation of the pattern to macro-events (i.e. combinations of a main and a subsidiary event) relies too much on lexicalisation patterns, and should include a more schematic constructional level of analysis. The analysis is illustrated from three Germanic and three Romance languages. Beatriz Martínez Fernández’s chapter on ‘Constructions, co-composition and merge’ discusses examples of the use of verbs of breaking in English which, she claims, to not fit the definition of construction as formulated in Goldbergian Construction Grammar because they express the semantics of motion while retaining that of a change of state. Martínez Fernández draws on

  Christopher S. Butler & Javier Martín Arista

work in Role and Reference Grammar and Pustejovsky's Generative Lexicon in order to give a more explanatory account of these phenomena in terms of the operations of co-composition and merge. The final paper in this section, Javier Martín Arista’s ‘A typology of morphological constructions’, also exploits the mechanisms of Role and Reference Grammar, generalising previous work in this model to cover the internal structure of morphological words. In the course of this paper, proposals are made for the replacement of descriptive notions such as ‘compounding’ and ‘derivation’ by principled types of morphological construction. The analysis is illustrated from the Australian language Pitjantjatjara/Yankunytjatjara. The second section, devoted to the Lexical Constructional Model, begins with a paper by Christopher Butler, which documents the stages through which this model has arisen from its antecedents in Dik’s Functional Grammar, Coseriu’s Lexematics, Van Valin’s Role and Reference Grammar, Wierzbicka’s Natural Semantic Metalanguage, Mel’čuk’s Meaning-Text Theory and, most recently, work in Goldbergian Construction Grammar and other areas of Cognitive Linguistics. Butler then summarises what he perceives to be the strengths of the LCM and discusses a number of areas which he considers to constitute challenges for the further development of the model. The paper by Ricardo Mairal Usón and Francisco Ruiz de Mendoza Ibáñez, ‘Levels of semantic representation in meaning construction’, then gives a more detailed description of the LCM, and discusses the central concepts of lexical template and constructional template which are at the heart of the model, and which are key elements in enabling it to synthesise functional and constructional approaches into a powerful overall theory of meaning construction which goes beyond ‘core grammar’ to incorporate areas such as illocution, implicature and discourse coherence. The third and final section begins with Francisco Gonzálvez-García’s article ‘Measuring out reflexivity in secondary predication in English and Spanish: evidence from cognition verbs’. By extensive reference to corpus materials, Gonzálvez-García documents the properties of instances of secondary predication, in both English and Spanish, involving find/encontrar (‘find’), used as a cognition verb, together with an object reflexive pronoun, as in I find myself incredulous. In his usage-based, bottom-up analysis, within the framework of Goldbergian Construction Grammar, the author argues for two distinct though related constructions, in the sense of learned formfunction pairings, which can be represented by this one structure, namely the reflexive subjective-transitive construction and the self-descriptive subjective-transitive construction. This final section of the book continues with Francisco Cortés Rodríguez’s paper ‘The inchoative construction: semantic representation and unification constraints’. The analysis is framed within the Lexical Constructional Model, and makes use of one of the central features of that model, namely the subsumption of lexical templates by constructional templates. Cortés Rodríguez shows that, in keeping with the proposals of the LCM, this lexical subsumption process is subject to two types of



Introduction  

constraint, external and internal. The external constraint on unification is based on the high-level process for action metonymy, whereas the internal constraints are of two types, one relating to the telicity and causativity of causative/inchoative verbs, the other to the arguments in the lexical templates. The final paper in the volume, Pilar Guerrero Medina’s ‘Semantic and pragmatic constraints on the English get-passive’, aims to show that the properties of the get-passive cannot be satisfactorily accounted for in purely lexical terms, but require reference to discourse-pragmatic factors. Guerrero Medina shows, on the basis of corpus data, how the semantics of the get-passive interacts with five semantic classes of verb which can appear together with it. Her conclusion is that, consistently with a Goldbergian type of analysis, there is a family of constructions associated with the structure, grouped into two basic subtypes, ‘causative’ and ‘spontaneous’. As this brief introduction has shown, the nine papers collected in this volume provide a snapshot of current thinking about constructions in relation to the rapprochement between functional and cognitive linguistics which is gaining ground in the 21st century. We believe that it will help to clarify some of the basic issues involved, and we hope that it will stimulate further debate in what promises to be a rapidly expanding field of research.

References Allison, David B. 1973. Introduction. In Jacques Derrida, Speech and phenomena and other essays on Husserl’s theory of signs. Trans. David B. Allison. Evanston IL: Northwestern University Press. Butler, Christopher S. 2003. Structure and function: A guide to three major structural-functional theories. Part 1: Approaches to the simplex clause. Amsterdam: John Benjamins. Butler, Christopher S. 2006. On functionalism and formalism: A reply to Newmeyer. Functions of Language 13(2): 197–227. Butler, Christopher S. & Francisco Gonzálvez-García 2005. Situating FDG in functional-cognitive space: An initial study. In Studies in Functional Discourse Grammar [Linguistic Insights 26], J. Lachlan Mackenzie & María L.A. Gómez-González (Eds), 109–158. Bern: Peter Lang. Culicover, Peter W. & Ray Jackendoff. 2005. Simpler Syntax. Oxford: OUP. Darnell, Michael, Edith Moravcsik, Frederick Newmeyer, Michael Noonan & Kathleen Wheatley (Eds), 1999. Functionalism and formalism in linguistics. 2 Vols. Amsterdam: John Benjamins. Davidse, Kristin. 2000. A constructional approach to clefts. Linguistics 38: 1101–1131. Goldberg, Adele. 2006. Constructions at work: The nature of generalization in language. Oxford: OUP. Gonzálvez-Garciá, Francisco & Christopher S.  Butler. 2006. Mapping functional-cognitive space. Annual Review of Cognitive Linguistics 4: 39–96. Laffut, An & Kristin Davidse. 2000. Verb meaning and construction sets: The case of caused NP-PrepP relations. LACUS Forum 1999: 293–304. Newmeyer, Frederick J. 1998. Language form and language function. Cambridge MA: The MIT Press.

  Christopher S. Butler & Javier Martín Arista Newmeyer, Frederick J. 2003. Grammar is grammar and usage is usage. Language 79(4): 682–707. Newmeyer, Frederick J. 2005a. Review of Christopher S. Butler Structure and function: A guide to three major structural-functional theories. Part 1: Approaches to the simplex clause. Part 2: From clause to discourse and beyond. Functions of Language 12(2): 275–283. Newmeyer, Frederick J. 2005b. Possible and probable languages: A generative perspective on linguistic typology. Oxford: OUP. Newmeyer, Frederick J. 2005c. A reply to critiques of ‘Grammar is grammar and usage is usage’. Language 81: 229–236. Nuyts, Jan. 2005. Brothers in arms? On the relations between cognitive and functional linguistics. In Cognitive Linguistics: Internal dynamics and interdisciplinary interaction [Cognitive Linguistics Research 32], Francisco J. Ruiz de Mendoza Ibáñez & Sandra Peña Cervel (Eds), 69–100. Berlin: Mouton de Gruyter. Nuyts, Jan. 2007. Cognitive linguistics and functional linguistics. In Handbook of cognitive linguistics, Dirk Geeraerts & Hubert Cuyckens (Eds), 543–565. Oxford: OUP. Taverniers, Miriam. 2002. Systemic-functional linguistics and the notion of grammatical metaphor. A theoretical study and a proposal for a semiotic-functional integrative model. Ph.D. dissertation, Ghent University. Van Valin, Robert D., Jr. 2005. Exploring the syntax-semantics interface. Cambridge: CUP.

part i

Theoretical issues

Innovative coinage Its place in the grammar* Daniel García Velasco The innovative creation of lexical items is not a topic of chief relevance among those which tend to mark the development of grammatical models. Yet nonce-formations and neologisms show the speaker's remarkable ability to create new lexical items in speech interaction and should thus be accounted for by linguistic theories of functional orientation. This article deals with a particular case of innovative lexical creation: the use of proper nouns in verbal function. Following the analysis in Clark & Clark (1979), I will defend the view that verbal eponyms are contextuals, that is, expressions whose interpretation is strongly tied to the context in which they are used. Unlike authors who have argued that the meaning of these units is predictable, I will show that verbal eponyms may receive multiple interpretations out of context, and that their meaning may shift from one setting to another. Next I will examine the extent to which the observed facts can be incorporated in two functional theories of language: Construction Grammar and Functional Discourse Grammar. After discarding the possibility of treating verbal eponyms as an example of coercion, a notion frequently adduced to justify a constructional approach to grammar, I will propose that the organization of Functional Discourse Grammar, which systematically links up with a conceptual and a contextual component, offers a more adequate architecture to implement the analysis proposed.

1.  Introduction The architecture of most grammatical theories includes a lexicon which lists the lexical items of a given language and their conventional senses. The stock of lexemes of a language, however, is not a fixed set, as speakers may creatively increase their basic vocabulary in two different ways: first, they can extend the meaning of an already existing lexical item in novel ways, and secondly, they can create new items which can be interpreted by the rest of speakers on the basis of linguistic, contextual and general knowledge.

*I am grateful to Gerry Wanders and two anonymous referees for their comments on a previous version of this article.



Daniel García Velasco

Thus, grammatical models which try to account not only for linguistic competence, but also for linguistic usage, cannot simply assume that word meanings are statically listed in the lexicon and retrieved every time a lexeme is used; they should also provide the means to account for lexical creativity from a production perspective and to explain how novel lexical creations are interpreted on line by other speakers. This article explores this question, which is illustrated with the case of verbal conversion from eponyms in the English language, and evaluates the extent to which the analysis presented can be incorporated in Construction Grammar and Functional Discourse Grammar.

2.  The problem: Lexical creativity New words may enter the lexicon of a language through borrowing or calques from other languages or as innovations coined by speakers themselves. The field of science and technology is obviously a source of lexical creations, as well as slang and the speech of the youth (Quirk et al. 1985: 1534ff.). Necessarily, new words enter the language at some point in time and it is sometimes possible to establish the moment at which a particular word was coined (e.g. Hohenhaus 2006 on the form bouncebackability). It is only at this moment that the word has the status of a nonce-formation, defined by Bauer (1983: 45) as “a new complex word created by a speaker/writer on the spur of the moment to cover some immediate need”. Many nonce-formations will die in a few minutes, but others will make their way through the language to become neologisms first and fully conventional lexemes later. Following Lyons (1977), Bauer (1983:  63) establishes a distinction between productivity and creativity along the following lines: Productivity is one of the defining features of human language, and is that property of human language which allows a native speaker to produce an infinitely large number of sentences, many (or most) of which have never been produced before. It is assumed that productivity is to be accounted for by the rules of a generative grammar. Creativity, on the other hand, is the native speaker’s ability to extend the language system in a motivated, but unpredictable (non-rule-governed) way.

Bauer illustrates the distinction with the case of the word headhunter, whose meaning ‘member of a tribe which keeps and preserves the heads of human victims’ would be a case of predictable productivity based on the rules of the grammar, whereas the metaphorical extension ‘one who recruits executives’ would be an example of unpredictable creativity. Thus, morphological innovations are predictable in meaning since they follow the syntactic rules of word creation. Speakers often make use of productive derivational rules to create words which may be interpretable by other speakers even out of context. Despite general mockery, when President Bush stated that his critical opponents had misunderestimated him



Innovative coinage: Its place in the grammar

he simply made use of existing word formation processes in the English language to create a novel expression whose meaning is predictable and can be easily understood by other speakers. Novel creations based on standard morphological rules are a common feature in the use of language (Katamba 1993: 296, Quirk et al. 1985: 1534). However, strict lexical creativity, in the form of meaning extensions and meaning creations, is unpredictable: even if the process underlying a given meaning extension can be characterized as a metonymy or a metaphor, still it will not be rule-governed in Bauer’s sense. A similar observation is provided by Nunberg (1995: 109) in his study of reference transfers (see below). He notes that these processes may be classified as metonymies or metaphors which may give rise to recurrent patterns of lexical alternation, but still this observation should not preclude the analyst from explaining why the process has taken place at all. A reference transfer, as illustrated in the following examples from English and Spanish respectively, could be considered an extreme case of meaning creation which is crucially dependent on the context:

(1) The ham sandwich is at table 7. (Nunberg 1995: 115)

(2) La semana que viene haré Bilbaos. The week that come.pres.3sg do.fut.1sg Bilbaos ‘Next week I shall be flying to Bilbao’ Lit.: ‘Next week I shall be doing Bilbaos’

Both (1) and (2) involve reference transfers. (1) is assumed to be employed by a waiter to identify a particular customer, whereas (2), which I attested myself, was used by a stewardess on a plane commenting on her work schedule. In these restricted contexts, ham sandwich is interpreted as ‘person who has ordered a ham sandwich’ and Bilbao can be paraphrased as ‘flights to Bilbao’. However, it would seem inadequate to assume that those meanings are listed in the lexical entries of ham sandwich and Bilbao, respectively. What is crucial about these examples is their context dependence. As noted by Nunberg (1995: 116), it would be odd for the same waiter to say I saw the ham sandwich at table 7 the other day driving a Mercedes, when the context is not directly concerned with the relations among customers, waiters and orders. Crucially, the author argues that this can be called a type of occurrent metonymy. Occurrent metonymies may become less context-specific through usage, Nunberg argues, and become systematic lexical alternations such as the following examples (Nunberg 1995: 117): (3) transmissions for cars: 4 speed, automatic, etc. painters for works: a Picasso, a Derain, etc. containers for volumes of stuff: She drank two glasses, etc. writer for oeuvre: fifty pages of Wordsworth place for inhabitants: Indianapolis voted for the referendum tree for wood: The table is made of oak





Daniel García Velasco

What is relevant about these processes is that they seem to be originally based on the same mechanisms which underlie occurrent metonymies, in particular, the fact that the property which warrants the transfer is ‘noteworthy’, that is, appropriate for identifying or classificatory purposes in the relevant context. It seems adequate, therefore, to distinguish two basic parameters of lexical creativity on the basis of the conditions which govern the interpretation of the new form: rule-governed lexical creations (e.g. headhunter in the first sense discussed) and context-governed lexical creation (e.g. ham sandwich in 1a). The distinction, however, is not so neat, as the two dimensions frequently overlap: context-governed creations are necessarily constrained by rules of the grammar and rule-governed processes may require contextual information to be properly interpreted. Consider the case of the form: Chomskyan. According to the meaning provided by the affix, the adjective should mean something like ‘associated to Chomsky’. Thus, given the fact that Noam Chomsky is known both as a political activist and as a linguist, the expression a Chomskyan argument could either mean a political or a linguistic argument, a matter which should be settled on the basis of the available context. Another obvious case is that of N+N compounds, whose interpretation, unlike that of verbal compounds, is not constrained by thematic relations and can thus receive multiple readings which can only be resolved in context. Thus, as is well-known in the literature, a compound such as marble museum (Plag 2003: 150) may have several readings, including: ‘museum built with marble’ or ‘museum in which marble statues are displayed’, which may be distinguished in speech through different stress patterns. Certainly, more interpretations are possible, but all of them will refer to a museum of some kind, since it is a basic fact of English derivational morphology that the rightmost element functions as the head of the entire construction. Thus, the interpretation of marble museum is determined by both linguistic and contextual knowledge. This means that the possibility of creating new meanings and interpreting them is constrained by both grammar rules and the context. It seems adequate, therefore, to propose that the meaning predictability of a given innovation is dependent on the relative weight of grammar and context. Intuitively, if a new creation results from the application of a productive morphological rule, the meaning of the new derivative would be more predictable than that of those creations whose interpretation is strongly context-dependent. Between the two poles on this predictability scale, there may be a good number of intermediate forms whose interpretation should rely on both grammatical and contextual knowledge to varying degrees. In this article I will study a specific type of lexical creation which may be seen as standing at the contextual pole on the scale just proposed. These innovations, which are called contextuals in the work of Clark & Clark (1979), have, as a defining feature, the fact that their meaning needs to be constructed on line by the addressee. They have the following properties which distinguish them from conventional expressions (Clark & Clark 1979: 782):



–– –– ––

Innovative coinage: Its place in the grammar

Number of senses: unlike conventional units, contextuals have an indefinite number of potential senses. Dependence on context: as contextuals have a shifting sense and reference, their interpretation depends on the context in which they appear. Cooperation between speaker and addressee: as contextuals have an indefinite number of potential senses, arriving at the intended interpretation requires cooperation between speaker and addressee.

Clark & Clark (1979) argue that verbs converted from nouns in English were first introduced in the language as innovative contextuals and, to the extent they were accepted by the linguistic community, became conventional units of the language. This is clearly seen in the case of denominal verbs which originate from proper nouns. The authors mention examples such as boycott, diddle, dun, finagle, fudge, lynch, pander, philander, charleston, xerox and more. These forms may now be non-transparent to many speakers, who may not recognize their origin as proper nouns, but the important thing is that, at the time they were created, they should have been felt as innovations or nonce-formations which then became neologisms and later conventional units of the language. Moreover, the authors argue that denominal verbs obtained from common nouns are also contextuals. Although a common noun, unlike proper nouns, has sense and not only reference, the interpretation of its verbal use offers several potential senses which have to be selected on the basis of context and general knowledge until the expression becomes conventional. The meaning of deverbal nouns is constructed, they claim, upon speakers’ ‘theories’ of categorized objects. Within each theory some properties will carry more weight as they may be considered more central to the characterization of an object. These will be more likely employed in the construction of the meaning of a deverbal converted noun.1 Given the coincidence of this analysis with his ‘noteworthy’ condition, Nunberg (1995: 117) admits that meaning transfers or occurrent metonymies may be understood as a type of contextual. Contextuals are thus particularly interesting for theories of grammar, since they require special mechanisms of interpretation on the listener's part and, especially, because their meaning cannot be assumed to be included in ready-made lexical entries. Of course, these problems, to the extent they have been given attention in formal theories of grammar, have usually been seen as belonging to the theory of language use and have, therefore, been excluded from the study of linguistic competence. Functional linguistics, however, cannot just use the same arguments to obviate an explanation of the use of contextuals in verbal interaction. The functional methodology, with its emphasis on the way we use language to achieve effective communication, should

.  See also Štekauer (2006: 492), who in a similar fashion speaks of prototypical features.





Daniel García Velasco

thus pay attention to the cognitive processes which underlie the creation of meaning in context. In the following section, I will present additional properties of eponyms used in verbal function as a type of expression which requires both general and contextual information to be felicitously interpreted. In section 4, I will compare the way they can be analysed in Construction Grammar and Functional Discourse Grammar.

3.  “He’s Beckhamed it”: Eponyms as contextuals Concerning the linguistic status of proper names, Lyons (1977: 219) notes that “… the most widely accepted philosophical view nowadays is that they may have reference, but no sense, and that they cannot be used predicatively purely as names”.2 If that is true, then, the meaning of, say, the name David Beckham is merely the reference to that particular individual. The rest of the information we associate with the football player and celebrity David Beckham belongs to general or encyclopaedic knowledge and is not part of linguistic competence. This received view is called into question whenever one encounters expressions such as (4):

(4) Oh my, he’s Beckhamed it!

Expression (4) is to be understood in the context of a football player missing a penalty kick. The expression seems to have originated after David Beckham’s miss of a crucial penalty kick in the 2004 European Football Cup. According to several Internet sources, a TV sports commentator coined the phrase during a subsequent match in the tournament, in which another player missed a penalty kick.3 Hence, within this restricted setting, the nonce-formation to Beckham is to be interpreted as ‘to miss a penalty kick in the game of soccer’. In principle, one might think that the relevance of this creation is not too significant. As mentioned earlier, the scarce literature on the theoretical aspects of nonce-formations and neologisms agrees that most nonce-formations die quickly (Bauer 1983; Crystal 2000; Hohenhaus 2005; Štekauer 2002) and very few get the status of neologisms in the language or of fully conventional lexical units. In the case of (4), it might be argued that the use of that expression was in vogue immediately after the soccer event, when the memory of the missed penalty was more alive in the supporters’ minds, than it is today.

.  The standard view has been challenged by Russell’s Theory of Descriptions which was in turn criticized by Kripke. For a nice short summary see Lycan (2000: chapter 3). .  See e.g. http://www.urbandictionary.com/define.php?term=beckham%27d+it.



Innovative coinage: Its place in the grammar

Presumably, the expression will fall into oblivion once David Beckham retires from football or his role as a modern icon is replaced by another superstar. Granted that, what is undeniable is that the process of converting proper nouns into contextually dependent expressions will not disappear. Quite on the contrary, the process seems productive and subject to norms of verbal interaction. For example, from Suzanne Kemmer’s neologisms database, I have obtained the novel creations based on eponyms which are shown in Table 1.4 It should be noted that I have excluded those obtained by affixation or blending in which just one of the members was a proper noun (e.g. Clintonomics). Although they necessarily involve access to general knowledge on the figure named (cf. Chomskyan above) their interpretation may be guided by the derivational process and they are thus not ‘pure’ contextuals. Of course, some of these novel creations may be unknown to many native speakers. Sometimes they involve very specific knowledge available to just the speakers and his/her closest friends (e.g. Claire, Gregariah). Still, they illustrate that the process of eponym transfer is very much alive in the English language and merits theoretical discussion even if most of these words have a rather limited future. The properties of meaning construction in contextual eponyms have been studied by Clark & Gerrig (1983), who discuss the case of the constructed example do a Napoleon. In the context of picture taking, they argue, the expression please, do a Napoleon for the camera is naturally interpreted as posing in the way Napoleon was famous for (i.e. with one hand tucked inside his jacket). As the authors observe, the interpretability of eponyms in verbal function depends on the interaction of the familiarity with the character and his acts and the restricting function of the context which guides the addressee to the right interpretation. Crucially, then, it is both the listener’s general knowledge and the contextual information that interact in the process. Of course, since we are dealing with nonce creations, it is only the speaker that knows the right interpretation and the addressee that has to decode the meaning. This means that s/he may have several potential interpretations available which somehow compete for recognition. Since, by definition, the meaning of a contextual can only be arrived at in context and through the cooperation of the speaker and the addressee, these expressions are likely to be multiply ambiguous out of context. Significantly, Balteiro Fernández (2001) notes that, when speakers are asked to provide different

.  The Rice University Neologisms Database: http://esa4.rice.edu/~ling215/is a collection of neologisms compiled by students at Rice University under the supervision of Prof. Suzanne Kemmer. The definitions and explanations of the neologisms have also been provided by students themselves.



Steven

McGyver Sancho

Illiana John

Gregariah

Federer Fredo

Bennifer Brangelina Britney Claire

An Apple computer A handsome male A newly married man who has long been a bachelor A one-hundred dollar bill/money in general

Apple Baldwin Benedict Benjamin

Eponym/origin

Apple Macintosh (computer brand) Alec Baldwin (actor) Benedick (a Shakespearian character) Benjamin Franklin (whose face is portrayed on one-hundred dollar bills) The relationship between Ben Affleck and Jennifer López Ben Affleck (actor)/Jennifer López (actress) The relationship between Brad Pitt and Angelina Jolie Brad Pitt (actor)/Angelina Jolie (actress) A young woman of dubious moral integrity Britney Spears (pop-singer) An instance when a student stays up all night working on Claire (the speaker’s roommate with that characteristic behaviour) homework because he or she procrastinated and did not begin working on it earlier Paradigm or example of perfection Roger Federer (tennis player) A person who is considered the blacksheep of a family or Fredo (A character in The Godfather) organization Person whose conduct resembles that of Rice student Greg X (the speaker’s roommate) Greg X The University of Indiana and the University of Illinois University of Indiana/University of Illinois An excuse for not winning, esp. at a video game John (a player named John who constantly had to make up excuses for losing) One who can use materials in novel ways McGyver (A TV show) Half man and half coward with a speck of metrosexual; Sancho (a character from Don Quixote) an individual who commits adultery, while he pimps the person who he is with A possum, particularly one who resides in a residential Steven (boy’s name) area and is not especially afraid of humans

Meaning

Noun

Table 1.  Eponyms in nominal and verbal function in The Rice Neologisms Database

 Daniel García Velasco

Oprah Photoshop Ralph Schumacher

Michael Jackson

Martha Steward

Louie

Internet Jude Law Leopold

Google

To participate in a performance of the song “Louie, Louie” with the Marching Owl Band To perform domestic chores such as cleaning or organizing a house or room in order to improve the aesthetic appearance (of room) and the quality of life (of owner) To touch or fondle somebody inappropriately in a joking, non-threatening manner To exploit or use To edit a picture To vomit To create a sequel to or continuation of a work of media that does not stand up to the original creator’s intent and/ or talent

To change sine and cosine to complex numbers, or the reverse operation To display a company’s ad when a person visits a rival company’s Web site To perform an online search for a certain subject or interest, either on google.com or another search engine To use or connect to the Internet To cheat (on a romantic partner) To perform in the style of Leopold Stokowski

Euler

Gator

Meaning

Verb

Oprah Winfrey (American TV celebrity) Adobe Photoshop (software program) Ralph (probably after a person named Ralph who vomited often) Joel Schumacher (American film director)

Michael Jackson (pop-singer)

Martha Steward (American TV celebrity)

Internet Jude Law (actor) Leopold Stokowski (a famous director of the Philadelphia orchestra known for his eccentric style) “Louie, Louie” (song title)

Google (Internet search engine)

Gator (software program)

Leonard Euler (Swiss mathematician)

Eponym

Innovative coinage: Its place in the grammar 

 Daniel García Velasco

meanings for the verbal use of proper nouns such as Shakespeare or Spielberg, they give definitions as disparate as the following: (5) Spielberg to make a film or direct it in the style of Spielberg to be directed by Spielberg to obtain huge profits with a film to become rich and successful in the film industry to triumph at the Oscars (6) Shakespeare to write in the manner or style of Shakespeare to perform the plays or roles of Shakespeare to make to resemble a play by Shakespeare to get every word right in the writing of a scene to write good poetry to put into archaic language to turn into iambic pentameter to take an existing story and transform it into a work of genius to embellish a script to take on some characteristics of Shakespeare’s speech to be insulted in the style of Shakespeare

The existence of so many potential interpretations for the use of these proper nouns as verbs clearly corroborates their contextual nature. Štekauer (2006) also reports on a similar survey (originally included in Štekauer 1997), in which 75 Slovak students of English were asked to provide meanings for verbal uses of famous people’s names. The author believes that there are no significant differences between the meaning of common nouns and proper names, since both can be structured around prototypical properties or behaviour/function.5 He claims (Štekauer 2006: 500): We cannot predict the meaning of, for example, the converted naming unit displayV if we do not understand the meaning, function, purpose, etc., of the

.  Coherently, Štekauer (1996: chapter 7) adheres to Russell’s Description Theory which defends the view that proper names have sense and function as predicates. However, an obvious criticism to this position is related to the fact that the sense of a proper name cannot possibly be the same for all speakers. As Lycan (2000: 41) puts it: “Undeniably, different people know different things about other people. In some cases, X’s knowledge about Z and Y’s knowledge about Z may not even overlap. It follows from the Name Claim (names as definite descriptions [DGV]) that the same name will have (many) different senses for different people; every name is multiply and unfathomably ambiguous”. Therefore, the assumption that names have sense does not solve the problem of explaining how the right sense is selected in a given speech act.



Innovative coinage: Its place in the grammar 

converting displayN (of a computer). Similarly, we cannot predict the meaning of the recategorized naming unit HavelV (a famous personality of the Czechoslovak velvet revolution) if we do not know who he is (was) and what the characteristic features of this personality are (were).

Štekauer thus assumes that the meaning of converted proper nouns can be activated even out of context as long as speakers have the necessary world knowledge. Reasonably, he notes that the meaning of these units is not crucially context-dependent as the context is of little help if one does not know who the named person is. While this is undoubtedly true, the problem with famous personalities is that the number of beliefs and pieces of knowledge we associate with them may vary significantly across speakers and, therefore, unless the interpretation is severely restricted by the (extra)linguistic context, listeners are likely to fail in the process. Thus, Štekauer also tests the verbal use of the noun Spielberg, which receives the interpretations ‘to succeed’ or ‘to direct’ among his informants. The author assumes that these interpretations simply “miss the point” as they are too general given the fact that there are many successful directors and, therefore, those definitions do not refer to Spielberg’s prototypical behaviour or actions. Consider another of his examples (Štekauer 2006: 501):

(7) He Ben Johnsoned his promising swimming career.

The author claims that the right interpretation for this unit is ‘to spoil one’s career as a result of doping’ because that is what Ben Johnson did. He reports that students were successful at arriving at this interpretation. But note that the sentence provided facilitates that reading with the sequence his promising swimming career. I presume that, if students had just been given the sequence He Ben Johnsoned, with no visible object, arriving at that particular interpretation would have been more difficult. Take the example to bogart which I comment upon in García Velasco (in press). If I hear the expression to bogart a cigarette, given my knowledge of the individual, I would come up with interpretations such as ‘to hold a cigarette in the way Bogart does in his films’ or ‘to smoke a cigarette in the way Bogart does’, because smoking/holding is what one usually does to a cigarette. Yet, the verb bogart is defined in The Urban Dictionary as:6 (8) bogart (slang verb) To keep something all for oneself, thus depriving anyone else of having any. A slang term derived from the last name of famous actor Humphrey Bogart because he often kept a cigarette in the corner of his mouth, seemingly never actually drawing on it or smoking it. Often used with weed or joints but can be applied to anything.

.  http://www.urbandictionary.com/define.php?term=bogart.

 Daniel García Velasco

Given my ignorance on the particulars of joint-sharing and the fact that I never realized that Humphrey Bogart never smoked the cigarettes he kept in his mouth, I could not arrive at the right interpretation. This point is acknowledged by Štekauer himself, who admits that some features may be prominent in the context of small communities only. But this is the crucial question: it is the context that determines which feature should be considered more central and that is what makes these units contextuals in Clark and Clark’s sense. True, with some historical figures, one property may be central for most speakers, and reaching the right interpretation may be easy. With others, however, the situation may be different: speakers may have very limited knowledge about one individual (and fail to provide interpretations or provide too general ones) or they may have a great deal of information associated to one person and, for exactly the opposite reason, will not know which piece of knowledge to choose to construct the verbal meaning unless the context provides the necessary help.7 Crucially, this view can also be extended to the analysis of verbal conversion from common nouns. In her treatment of conversion in English, Lieber (2004) assumes that the range of meanings exhibited by converted verbs is much broader than that of verbs created by affixation, and therefore concludes that the analysis of conversion as zeroderivation is untenable. Given this unpredictability, she argues that denominal verbal conversion is a type of innovative coinage by which any noun can be used as a verb “as long as it is interpretable in context” (Lieber 2004: 94). If the novel verb gets reused and becomes conventional, it is ‘relisted’ in the mental lexicon as a unit of different syntactic category. As mentioned earlier, the context dependence of the verbal use of proper nouns poses an interesting challenge to linguistic theories, and one that cannot be solved by simply assuming that innovative meanings are retrieved from the mental lexicon. In the following section I will explore how two linguistic theories take up this challenge.

4.  L  exical innovation in Construction Grammar and Functional Discourse Grammar If on line meaning creation, as reflected in innovative coinage, is a common feature of verbal interaction, it is obvious that usage-based or functional theories of language should provide the mechanisms to account for this ability. In what follows I will

.  It is interesting to see that some of the examples in the Rice Neologisms Database receive different interpretations in other sources. The Urban Dictionary defines the verb to federer as ‘to completed [sic] dominate someone in any sport or event, particularly tennis’.



Innovative coinage: Its place in the grammar 

compare how this problem may be treated in Construction Grammar and Functional Discourse Grammar. Construction Grammar (CG) may be referred to as a family of theories, as described by Goldberg (2006: chapter 10). The basic tenet which all constructionist approaches seem to share is the view of grammar as a declarative stock of conventional constructions, defined as learned pairings of form and meaning. According to Goldberg (2006: 5), the whole grammatical system may be characterized as a set of constructions, from morpheme to sentence level. Although the question addressed in this article does not seem to be a matter of primary concern in CG, we do find some interesting observations which are relevant in the present context. In particular, Fillmore (1997: 5) seems to assume Bauer’s distinction above between productive and unpredictable processes of lexical creation when he claims: In one case there is the ability of speakers, using existing resources in the language, to produce and understand novel expressions. In the other case, the one for which we use the term coining, a speaker uses existing patterns in the language for creating new resources.

It is interesting to note that Fillmore argues that patterns of coining, although unproductive, should be considered part of the grammar, whereas Kay (2002: 15) explicitly argues that they should not. Kay believes that the grammar should contain only those constructions which produce predictable results. He argues: Rejecting the construction/coining distinction and accepting both types of patterns as constructions of equal status has the undesirable result of robbing grammatical research of one of its main methods: prediction of grammaticality. I have proposed that patterns of coining be excluded from a grammar, although they do deserve study as meta-grammatical patterns likely to influence grammatical change.

Kay thus seems to exclude verbal conversions from eponyms from the study of grammar proper. Yet, it seems to me that understanding the dichotomy construction/ coining in terms of prediction of grammaticality is not correct. Even though the use of eponyms in verbal function creates new lexemes in the language system, the process cannot be assumed to be a pattern of coining which cannot be evaluated in terms of prediction of grammaticality. The process is productive and yields predictable results (i.e. the unit will function as a verb which can be inflected, presents an argument structure, etc.) and, in principle, the grammar of English permits its application to just any proper noun. The only difference from other grammatical processes is that its meaning (or rather, its interpretation) is context-bound. Rather than a pattern of coining, it seems to me that eponym conversion might be understood in CG as an example of coercion or coercion effects. Although the

 Daniel García Velasco

interpretation of the phenomenon varies slightly (see Ziegeler 2007: 992), coercion can be seen as a process in which, given a semantic incongruity between a lexical unit and a syntactic context, the need arises to apply special interpretative mechanisms to process the expression. Significantly, coercion effects have provided an important argument for the role of constructions in grammatical description. Crucially, the existence of novel uses of verbal expressions such as the following (discussed in Michaelis 2006), is seen as clear evidence (see also Goldberg 2006: 6–7): (9) a. Most likely they were fellow visitors, just panting up to the sky-high altar out of curiosity. (Lindsey Davis, Last Act in Palmyra, p. 28) b. When a visitor passes through the village, young lamas stop picking up trash to mug for the camera. A gruff ‘police monk’ barks them back to work. (Newsweek 10/13/97) c. Although he professed to like the sweater she knit him for his birthday, he wouldn’t wear it in public […]. (www.knitty.com/ISSUEwinter02/FEATsweatercurse.html)

As Michaelis notes, following the analysis in Goldberg (1995), the verbs pant, bark and knit participate in unexpected syntactic environments which impose additional meaning components onto the verbal expression. Thus, pant and bark in (9a) and (9b) get a sense of caused movement which is not part of the verb’s original meaning, whereas knit in (9c) would get a sense of transfer. Under a compositional view of sentence meaning, one would have to assume the rather counterintuitive claim that these verbs are polysemous, in much the same way as ham sandwich and Bilbao in (1) and (2) above would have to be assumed to be polysemous. This undesirable conclusion can be avoided if one admits that constructions are also symbolic units which impose meaning on verbal items whenever there is a conflict between the verb and the syntactic construction itself.8 This analysis has been adopted by Michaelis (2003, 2004, 2005, 2006) who has studied aspectual coercion effects, and mass count alternations, as in the following examples: (10) a. I had a tea. b. Give me some blanket.

.  Significantly, Kay (2002) suggests that Goldberg’s caused-motion construction should be considered a pattern of coining given the fact that it does not produce predictable results. The following examples seem to be ungrammatical (2002 : 13):

a. *He bragged her out of the room. b. *She screamed him under the bed.

It thus seems reasonable to consider the use of eponyms in verbal function as a coercion effect.



Innovative coinage: Its place in the grammar 

In (10a), the mass noun tea receives a countable interpretation when it is combined with the indefinite article, whereas in (10b) a count noun receives a mass construal when combined with some. In her unification-based version of CG, the author assumes that the nouns and the nominal construction present conflicting feature values which compete for priority in the interpretative process. Which of the two wins out is determined by the override principle (Michaelis 2004: 25): The override principle. If a lexical item is semantically incompatible with its morphosyntactic context, the meaning of the lexical item conforms to the meaning of the structure in which it is embedded.

Crucially, it is also possible to use proper nouns in contexts in which they have to be interpreted as mass nouns. Nunberg (1995: 115) offers the following example: (11) They played lots of Mozart.

in which the proper noun Mozart is interpreted as a mass noun (i.e. music composed by Mozart). I would assume, given the analysis of the mass/count alternation, that this would also be the analysis for this example. The case of verbal eponyms could thus be treated along similar lines. An expression such as he Beckhamed it could be analysed as resulting from the application of the override principle: given a mismatch between the semantic properties of the proper noun Beckham and those of the transitive verbal construction in which it is inserted, the verbal construction wins out and the noun gets a ‘verbal’ meaning. While this analysis makes sense, it obviates the fact that Beckham has no sense, only reference. The grammatical meaning could be imposed through the verbal valency and even the denotation type may be altered, but this by itself is not enough to explain the interpretation the lexeme gets. There is nothing in a transitive construction which could lead one listener to interpret the expression he Beckhamed it as ‘he missed a penalty kick’ among the set of potential senses that could be attributed to the expression. Similarly, some authors see no difference between coercion effects and metonymy (see the discussion in Ziegeler 2007). Ziegeler argues that coercion effects may not be necessarily connected to the syntax of the construction in which the lexical item appears. It is true, she admits, that in the case of argument structure constructions such as the examples in (9), lexical meaning interacts with the syntactic construction, but still there is a lexical shift which cannot be explained in all cases by the semantic nature of the construction. Basically, coercion effects and meaning transfers involve shifts of meaning based on polysemy extension and metonymy and metaphorical processes which may or may not be related to the use of a lexical item in an unconventional syntactic frame. If that is the case, then, there is no reason to assume that the process is constructionally guided. Note, furthermore, that this would mean that the meaning created is somehow predictable, which is not the case.

 Daniel García Velasco

The view that I shall defend here, however, differs from the constructional analysis in that it attributes a crucial role to both the context and general knowledge in the understanding of innovative verbal eponyms. Indeed, the verbal construction does play a role in the interpretation of the expression. However, rather than imposing its meaning onto the proper noun, the construction serves as a trigger for the addressee to arrive at the right interpretation. The use of a proper noun in verbal function forces an event reading which, as mentioned in the previous section, is open to a great number of interpretations. Clark & Gerrig (1983) have noted that the selection of the right interpretation is hierarchically constrained according to the following sequence: 1. Identity of the eponym 2. Acts by the eponym 3. Relevant acts by the eponym 4. The type of act referred to In order to arrive at the right interpretation, listeners should first be familiar with the identity of the eponym. Otherwise, they cannot move to stage 2, where they would highlight relevant acts by the eponym. Stages 3 and 4 are clearly contextually guided: 3 refers to those acts which are relevant within speech context, and 4 to the identification of one act as the one intended for that particular context. This process of interpretation, then, crucially relies on the interaction of general knowledge (in this case about a famous person) and the particular situational context in which the speech act is performed. This model should also be seen as part of a cooperative model of verbal interaction in which speakers evaluate their addressees’ pragmatic knowledge and conclude they can arrive at a unique interpretation for the expression in question (see Dik 1997 for a sketch of such a model). It should be noted, moreover, that this analysis is also adequate from both a speaker and a listener’s point of view. The listener decodes the meaning intended, but speakers simply assign a novel meaning to the nonce-formation which they then make compatible with a verbal construction. This supports the view that the argument structure is not responsible for the meaning itself, but merely hints at a meaning which is then decoded by the listener on the basis of syntactic, contextual and general information. As argued by Štekauer (2002), neologisms are clearly definable from the speaker’s point of view, but they imply a special interpretation strategy on the listener’s part, at least until the meaning of the new coinage becomes conventional. This analysis fits well within Functional Discourse Grammar (FDG), the most recent upshot of Dik’s Functional Grammar, presented in Hengeveld & Mackenzie (2006, 2008). The authors propose the general organization for the model which is shown in Figure 1. As the figure shows, the grammatical component is surrounded by both a conceptual and a contextual component. Crucially, although these components are not part



Innovative coinage: Its place in the grammar 

Primitives

Formulation

Primitives

Encoding

Contextual Component

Grammatical Component

Conceptual Component

Output Component Figure 1.  General layout of FDG (adapted from Hengeveld & Mackenzie 2006: 669).

of the grammatical system proper, they are assumed to be necessary in the characterization of an adequate theory of verbal interaction. The phenomenon studied here illustrates the benefits of this proposal as it clearly shows that these components are necessary in accounting for the process of meaning creation. The interpretation of eponyms in verbal function requires the consultation of the conceptual component in order to look for relevant facts about the eponym. Note that this information cannot be present within the lexicon as it is assumed that proper nouns do not have sense, but only reference. Hence, speakers, in order to construct the meaning of the verbal use to Beckham, must look up the conceptual component for the relevant piece of information about this individual which they wish to convey in the linguistic expression. Moreover, they will have to assume, in a Gricean cooperative spirit, that those beliefs are also part or their addressees’ conceptual knowledge. Listeners, in turn, will make use of syntactic information, but crucially, also of contextual information as described in the contextual component, to arrive at the right interpretation among the set of beliefs they associate to David Beckham in their conceptual component. In order to illustrate this proposal, let us assume the following situation. Speaker A and speaker B are watching a football match on TV. A football player, say Luis Figo, has just missed a penalty kick, a fact which is described by speaker A as: he’s Beckhamed it. Figure 2 could be a simplified characterization of the contextual component. If speaker B is to arrive at the intended interpretation (under the assumption he has never heard the expression before) Figure 3 could be a rough characterization of his conceptual component. Following Clark & Gerrig’s (1983) classification, I have accordingly classified the speaker’s beliefs in terms of their role in the interpretative process.

 Daniel García Velasco Contextual component (at time t1) General Situation: Speaker and addressee are watching a football game on TV Active Referent: Luis Figo Status: visible for speech participants Active event: Luis Figo is about to shoot a penalty kick

Figure 2.  Speech participants’ contextual component.

Speaker B’s conceptual component (at time t1) There’s a individual named David Beckham (Identity) David Beckham is a famous football player (Identity) Etc. David Beckham married a famous pop singer (Acts by the eponym) David Beckham has acted as a fashion model (Acts by the eponym) Etc. David Beckham has scored many goals with penalty and free kicks (Relevant Act) David Beckham missed a crucial penalty kick in the 2004 European Cup (Relevant Act) Etc.

Figure 3.  Speaker B’s conceptual component.

Needless to say, the set of beliefs included in Figure 3 will not necessarily exhaust the amount of information a given individual may have on David Beckham. They are just included there to show that, following the hierarchy proposed by Clark and Gerrig, the interpretation process should proceed downwards, disregarding beliefs about the individual until the relevant piece of information is reached. As shown in García Velasco (2007, in press) the analysis proposed here fits nicely within Functional Discourse Grammar as long as one assumes a theory of lexical meaning in which lexical definitions are not constructed on the basis of fixed sets of necessary and sufficient features. Rather, lexical meaning varies across individual competences (although certainly some features in the meaning of lexical items will be perceived as constitutive of normal competence) and it is necessary therefore for speakers to converge dynamically on the set of relevant beliefs associated to lexical items for communication to take place. The notion of convergence is borrowed from Marconi (1997) and it underlines the idea that meaning is created cooperatively by speakers on the basis of common knowledge, assessment of each others’ beliefs and contextual clues (for similar views see Allwood 2003 and Evans 2006).



Innovative coinage: Its place in the grammar 

It should also be noted that, unlike previous Functional Grammar, FDG is not a projection-based theory any more. That is, the construction of a linguistic expression is not constrained by the syntactic requirements of lexical items. After García Velasco & Hengeveld’s (2002) proposal, lexemes do not appear in the lexicon in the form of semantico-syntactic structures (predicate frames) defining the syntactic and semantic valency of predicates. Rather, lexemes are separated from the frames in which they can be inserted, which are now seen as belonging to the set of primitives permitted for each language. Crucially, and unlike constructional approaches, frames in FDG are not assumed to be pairs of form and meaning. Therefore, this view coincides with the idea that constructions or frames operate as indicators that a given lexeme has a particular function in the construction of a linguistic expression. As mentioned earlier, this is exactly what happens whenever a proper noun is inserted in a verbal frame. The frame does not impose its meaning on the lexeme (as it is not associated with any descriptive content), but it merely has a triggering effect for the listener to look for the appropriate interpretation of the expression in the terms described above.

5.  Conclusion In this article, I have studied a particular case of innovative lexical creation: the use of eponyms in verbal function. Following the analysis presented in Clark & Clark (1979), I have defended the view that verbal eponyms are contextuals, that is, expressions whose felicitous interpretation depends on the context and general knowledge. The function of the context, I have shown, is precisely to guide the addressee to the right interpretation among the set of potential senses that the verbal eponym might receive. Unlike authors who have argued that the meaning of these units is predictable, I have shown that (i) verbal eponyms may receive multiple interpretations out of context, and (ii) the meaning of a verbal eponym may shift from one context to another. In the second part of this article I have examined the extent to which the observed facts can be incorporated in two functional theories of language: Construction Grammar and Functional Discourse Grammar. First, I explored the possibility of treating verbal eponyms as examples of coercion, a notion which has spurred much research and provided strong arguments for a constructional approach to language. I concluded, however, that the meaning of verbal eponyms cannot be assumed to result from the interaction of a verbal construction and the meaning of the proper noun. By contrast, I proposed that Functional Discourse Grammar, which is characterized as the grammatical component of a wider theory of verbal interaction which also includes a conceptual and a contextual component, offers an adequate architecture to implement the analysis proposed.

 Daniel García Velasco

References Allwood, Jens. 2003. Meaning potentials and context: Some consequences for the analysis of variation in meaning. In Cognitive approaches to lexical semantics, Hubert Cuyckens, Rene Dirven & John Taylor (Eds), 29–65. Berlin: Mouton de Gruyter. Balteiro Fernández, Isabel. 2001. Native speakers’ use, interpretation, and awareness of the conversion of proper names into verbs. In Proceedings of the 25th International Conference of AEDEAN, Marta Falces Sierra, Mercedes Díaz Dueñas & José María Pérez Fernández (Eds), Granada: Servicio de Publicaciones de la Universidad de Granada. CD-ROM edition. Bauer, Laurie. 1983. English word-formation. Cambridge: CUP. Clark, Eve & Herbert H. Clark. 1979. When nouns surface as verbs. Language 55: 767–811. Clark, Herbert H. & Richard J.  Gerrig. 1983. Understanding old words with new meanings. Journal of Verbal Learning and Verbal Behavior 22: 591–608. Crystal, David. 2000. Investigating nonceness: Lexical innovation and lexicographic coverage. In Manuscript, narrative and lexicon: Essays on literary and cultural transmission in honor of Whitney F Bolton, Robert Boenig & Kathleen Davis (Eds), 218–231. Lewisburg: Bucknell University Press. Dik, Simon C. 1997. The theory of Functional Grammar. Part 1: The structure of the clause. 2nd revised Edn. Kees Hengeveld (Ed.), Berlin: Mouton de Gruyter. Evans, Vyvyan. 2006. Lexical concepts, cognitive models and meaning-construction. Cognitive Linguistics 17(4): 491–534. Fillmore, Charles. J. 1997. Lecture on idiomaticity. Available at http://www.icsi.berkeley. edu/~kay/bcg/lec02.html García Velasco, Daniel. 2007. Lexical competence and Functional Discourse Grammar. Alfa – Revista de Lingüística 51(2): 165–187. García Velasco, Daniel. In press. Conversion in English and its implications for Functional Discourse Grammar. Special issue of Lingua on Semantic Representation in Functional Discourse Grammar. García Velasco, Daniel & Kees Hengeveld. 2002. Do we need predicate frames? In New perspectives on argument structure, Ricardo Mairal Usón & María J. Pérez Quintero (Eds), 95–123. Berlin: Mouton de Gruyter. Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago IL: Chicago University Press. Goldberg, Adele E. 2006. Constructions at work. Oxford: OUP. Hengeveld, Kees & J. Lachlan Mackenzie. 2006. Functional Discourse Grammar. In Encyclopedia of language and linguistics, 2nd Edn. Vol. 4, Keith Brown (Ed.), 668–676. Oxford: Elsevier. Hengeveld, Kees & J. Lachlan Mackenzie. 2008. Functional Discourse Grammar: A typologicallybased theory of language structure. Oxford: OUP. Hohenhaus, Peter. 2005. Lexicalization and institutionalization. In Handbook of word-formation, Pavol Štekauer & Rochelle Lieber (Eds), 353–373. Berlin: Springer. Hohenhaus, Peter. 2006. Bouncebackability: A web-as-corpus-based case study of a new formation, its interpretation, generalization/spread and subsequent decline. SKASE, Journal of Theoretical Linguistics 3(2): 17–27. Katamba, Francis. 1993. Morphology. London: MacMillan. Kay, Paul. 2002. Patterns of coining. Paper presented at ICCG2, Helsinki, 9/08/02. (http://www. icsi.berkeley.edu/~kay/coining.pdf)



Innovative coinage: Its place in the grammar 

Lieber, Rochelle. 2004. Morphology and lexical semantics. Cambridge: CUP. Lycan, William G. 2000. Philosophy of language. London: Routledge. Lyons, John.1977. Semantics. Cambridge: CUP. Marconi, Diego. 1997. Lexical competence. Cambridge MA: The MIT Press. Michaelis, Laura A. 2003. Word meaning, sentence meaning and constructional meaning. In Cognitive approaches to lexical semantics, Hubert Cuyckens, Rene Dirven & John Taylor (Eds), 163–210. Berlin: Mouton de Gruyter. Michaelis, Laura A. 2004. Type shifting in Construction Grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15(1): 1–67. Michaelis, Laura A. 2005. Entity and event coercion in a symbolic theory of syntax. In Construction Grammar(s): Cognitive grounding and theoretical extensions, Jan-Ola Östman & Mirjam Fried (Eds), 45–87. Amsterdam: John Benjamins. Michaelis, Laura A. 2006. Construction Grammar. In Encyclopedia of language and linguistics, 2nd Edn. Vol. 3, Keith Brown (Ed.), 3–84. Oxford: Elsevier. Nunberg, Geoffrey. 1995. Transfers of meaning. Journal of Semantics 12: 109–132. Plag, Ingo. 2003. Word-formation in English. Cambridge: CUP. Quirk, Randolph, Sydney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. London: Longman. Štekauer, Pavol. 1996. A theory of conversion in English. Berlin: Peter Lang. Štekauer, Pavol. 1997. On the semiotics of proper names and their conversion. Arbeiten aus Anglistik und Amerikanistik 22(1): 27–36. Štekauer, Pavol. 2002. On the theory of neologisms and nonce-formations. Australian Journal of Linguistics 22: 97–112. Štekauer, Pavol. 2006. On the meaning predictability of novel context-free converted naming units. Linguistics 44(3): 489–539. Ziegeler, Debra. 2007. A word of caution on coercion. Journal of Pragmatics 39: 990–1028.

The construction of macro-events A typological perspective* Johan Pedersen In this article Talmy’s influential typology of macro-events (Talmy 1985, 1987, 1991, 2000) is discussed from the point of view of construction grammar (Goldberg 1995, 2006). Talmy has described typological differences of lexicalization between what he calls satellite-framed languages and verb-framed languages. The discussion originates in a contrastive analysis of a short story by H.C. Andersen available in six parallel versions: the original Danish version, an English, a German, a Spanish, an Italian and a French version. The article argues that the generalized version of the typology (Talmy 1991, 2000) suffers from being formulated exclusively in terms of lexicalization patterns, and that the typology should include both the lexical level and a schematic constructional level of analysis. A framework is proposed in which the typological patterns are interpreted as an information structure phenomenon. Constructions of the main information (MIC) and the supportive information (SIC), of varying degree of specificity, are the basic constituents of the typology. From this point of view, Germanic languages tend to map the main information (MI) onto a complex schematic construction and the supportive information (SI) onto a lexical (verbal) construction. Romance languages tend to map the MI onto the verb, while the SI may be mapped onto a complex schematic construction. The article hypothesizes that MIC and SIC stem from generalizations from usage, that they have their own, procedural role in grammar, as a device for organizing the information, and that the typology is anchored in this task. The interpretation of Talmy’s descriptive typology is in this perspective that some pairs of MIC/SIC are more entrenched in the grammar of some languages than in others. The proposed framework is well suited for analyzing usage data that does not fit the basic patterns. It is also adequate for identifying patterns in data that are similar to those recognized in Talmy’s work, yet not recognized as part of his typology.

*I would like to thank the editors of this volume and two anonymous referees for their valuable criticism and suggestions.

 Johan Pedersen

1.  Introduction The aim of this article is to discuss the distinction in cognitive semantics between verbframed and satellite-framed languages (Talmy 1985, 1987, 1991, 2000) in the theoretical context of a construction grammar approach to argument structure (e.g. Goldberg 1995, 2006; Croft 2001), as opposed to a lexical approach (e.g. Grimshaw 1990; Levin & Rappaport Hovav 1995; Pinker 1989). In several publications, Talmy has thoroughly described characteristic typological differences of lexicalization, e.g. differences between Germanic and Romance languages. The classic example is the expression of motion. In Danish and English, which are said to be satellite-framed languages, the path of movement is mapped onto a satellite,1 while the manner of movement is mapped onto the verb. In Spanish, a verb-framed language, the path is mapped onto the verb, and the manner may be expressed by an adverbial expression. In Talmy’s work, this typology is formulated in terms of lexicalization patterns. In Talmy (1991, 2000), the typology is modified. It now has a broader scope and no longer concerns only the motion event. The attempt to generalize the typology makes it clear that a generalizing framework requires a slightly different conception of grammar, particularly with respect to the principles of Talmy’s componential lexical approach. More specifically, I will claim that the typology would profit from being fitted into a construction grammar framework (e.g. Goldberg 1995, 2006; Croft 2001).2 I will argue that even though the typology of motion events, originally presented in Talmy 1985, very convincingly shows important patterns of lexicalization, the idea of extending its applicability to other semantic domains requires a framework that goes beyond having the lexeme as the basic unit of the typology. The essential question that this study raises is the following: do we have a general typology of lexicalization, as claimed by Talmy, or should we instead develop a general typology of constructions, in which patterns of lexicalization are special cases? I will argue that the latter is the most rational and fruitful strategy. The argument follows, thus, a more general trend in typological research away from typologizing languages as a whole, to typologizing particular situation types expressed in a language (see e.g. Croft et al. 2008). As pointed out by Croft and his collaborators, the former strategy usually leads to declaring that all languages are a “mixed” type (Croft et al. 2008: 25). The descriptive typology developed by Talmy is based on data from a broad, though still far from complete, sample of the world’s languages. In recent years, the framework has been applied to an extremely wide range of languages. It should be

.  Talmy defines a satellite as a grammatical constituent, other than a nominal argument, that has a sister relation to the verb. This includes a wide variety of grammatical entities, including, e.g. English verb particles and verb prefixes in German (Talmy 2000). .  Ruiz de Mendoza Ibáñez and Mairal Usón (in press) offers a similar framework.



The construction of macro-events: A typological perspective 

emphasized that the present study is not itself a typological study. It is rather to be considered as a criticism of certain established assumptions in current typological work on macro-events, a criticism that leads to a new interpretation of Talmy’s findings. The analysis makes use of data from six European languages, which represent two major families: the Romance languages and the Germanic languages. It is my hope that my proposals will turn out to have a broader application, and that this cross-linguistic study, which is more contrastive than typological, may be a contribution to a revised, and improved, typology of macro-events. First, I will present the topic of this article: the classic distinction in cognitive semantics between verb-framed and satellite-framed languages (Section 2). Secondly, I will discuss some problems that are related to Talmy’s idea of designing a typology with a broader scope (Section 3). Thereafter (Section 4), I will tentatively suggest a construction grammar based theoretical framework. This framework will be exemplified with data from some major European languages (Section 5). Finally, some methodological principles for an empirical study on a larger scale will be outlined, and some preliminary results from a pilot study will be presented (Sections 6 and 7). In the first part (Sections 2–5), whenever examples are not provided with references, they are made-up, or translated, examples, checked by native speakers. In the last part (Sections 6–7), data are taken from the parallel corpus Andersen (2005), created specifically for this study.

2.  Verb-framed and satellite-framed languages Expressions of motion, as in (1), have been studied extensively and they are often used as the prototypical example of typological differences between e.g. Germanic and Romance languages (e.g. Aske 1989; Berman & Slobin 1994; Gennari et al. 2002; Ibarretxe-Antuñano 2004a, 2004b; Slobin 1996, 1997, 2000, 2004; Talmy 1985, 1987, 1991, 2000). Talmy has identified similar typological patterns in other semantic domains, such as state change and temporal contour (aspect), and he generalizes the typology to be valid for what he defines as macro-events (Talmy 1991, 2000). Macro-events are complex semantic structures comprising a main event (ME), the framing event, and a co-event (CE). He claims, thus, that in (1) the motion event is a macro-event, in which the path of motion is the ME, and the manner of motion is the CE: (1) The bottle floated Flasken flød CE

into the cave. ind i hulen. ME

La botella entr-ó en la cueva flot-ando. the bottle enter-pst.3sg in the cave float-ger ME CE

Talmy (1985) (Danish) (Spanish)

 Johan Pedersen

According to Talmy, some languages, e.g. English and Danish, map the ME onto the satellite and the CE onto the verb. Other languages map the ME onto the verb and the CE outside the verb, for instance, as in Spanish, onto an adverbial expression. The term event, as used by Talmy, is indeed highly abstract, and it has to be understood in the context of certain general cognitive processes that in Talmy (2000) are termed conceptual partitioning and ascription of entityhood. According to Talmy, the human mind in perception or conception can extend a boundary around a portion of what would otherwise be a continuum (space, time and other domains) and ascribe to this content the property of being a single unit entity. One category of such an entity is perceived or conceptualized as an event (Talmy 2000: 215). The basic idea is that the macro-event is organized as a gestalt, as a figure-ground relation between the main-event and the co-event, which is also characterized as a support event for the main event (e.g. the meaning of manner, or cause). This theorizing is thus analogical with Talmy’s previous analysis of the expression of complex events in compound sentences (Talmy 1978, 2000): (2) Since his wife was tired, Ground

they went home early. Figure

In (2), the principal event is that they went home early, and the backgrounded causal event is that his wife was tired. It is not clear what exactly are the constraints on what may count as a macro-event in Talmy’s framework. Building on Talmy’s work, Bohnemeyer et al. introduce the term macro-event property (MEP) in a recent, large scale, cross-linguistic study of event segmentation. MEP is a property of clausal expressions that assesses the tightness of packaging of subevents in the expression. An expression has the MEP if it packages event representations such that temporal operators (e.g. tense and time adverbial) necessarily have scope over all subevents (Bohnemeyer et al. 2007). Talmy’s typology includes five types of macro-events in different semantic domains. In (3) the macro-event consists of two subevents: the state change, which is the principal meaning component, the main event (ME), and the causing event (CE). The overall interpretation of the macro-event is that some activity, or action, directed toward some entity causes a state of change to come about: (3) I blew the candle out. Jeg pustede stearinlyset ud. CE ME Apag-ué la vela de un soplido. put out-pst.1sg the candle by a blow ME CE

(Talmy 1991) (Danish) (Spanish)

In Danish and English, the ME, the state change, is expressed by a satellite (ud, out), whereas the CE (pustede, blew) is expressed by the verb. In Spanish, the ME and the CE is expressed by the verb (apagué) and an adverbial element (de un soplido) respectively.



The construction of macro-events: A typological perspective 

Viewed as a macro-event, a temporal contour, or aspect (Comrie 1976), may be expressed distinctly in different language types as well, according to Talmy (2000). The typological difference looks like the one observed for the construction of motion events, or state change. The ME is the temporal contour (the aspectual structure) and the CE is the backgrounded supporting event (the process in question). A number of Germanic languages typically express the ME in a satellite, whereas the same meaning component in a Romance language like Spanish is constructed by the verb. In (4), English and Danish express the meaning of recency as an adverbial satellite, while Spanish expresses it within the verb: (4) I had just eaten. Jeg havde lige spist. ME CE Acab-aba de Finish-pst.ipfv.1sg to ME

(Talmy 2000) (Danish) com-er. eat-inf CE

(Spanish)

The implicit qualification of the relation between ME and CE as a figure-ground relation is in this case not very clear. The relative prominence of the ME and the CE seems to depend on the specific construal in usage. Another semantic domain that Talmy mentions is action correlation, i.e. co-activity. Co-activity may be expressed as a macroevent, in which the co-activity is the ME and the activity in question is the CE. Germanic languages tend to construe the co-activity (ME) according to the general pattern, i.e. in a satellite, whereas the activity in question (CE) is expressed verbally (Talmy 2000): (5) I walked Jeg gik CE Yo le I acc.3sg

along with him. sammen med ham. ME acompañ-é accompany-pst.1sg ME

(Adapted from Talmy 2000) (Danish) and-ando. walk-ger CE

(Spanish)

An event of realization/completion, e.g. the police hunted the fugitive down (Talmy 2000), is the fifth type of macro-event (including the motion event) that Talmy analyzes as a conflation of two subevents, the framing event and the supporting event. This type is, however, not illustrated with Romance examples. In Section 5.2, I will show that the category resultative is highly relevant for a revised, construction-based, typology, and particularly that this is easily exemplified with Spanish (or Romance) examples.

3.  Some problems in Talmy’s typology I have already mentioned that Talmy’s typology relies on principles of figure/groundorganization. In my view, as I will argue in more detail below, this idea lacks motivation

 Johan Pedersen

since figure-ground conceptualization in linguistic theorizing does not operate across languages. Another strategy for a further development and elaboration of Talmy’s typology, would be (and this is what I will suggest, tentatively, in this article) to analyze the typological differences in terms of information structure constructions (Lambrecht 1994). The function of information structure constructions is to organize the essential contribution to meaning – the main information – in relation to a supportive element of meaning – a secondary item of information. Talmy’s typology is a theory about lexicalization patterns. In this article I will argue that, from the point of view of the construction grammar framework (Goldberg 1995, 2006), the generalized version of the typology (as described in detail in Talmy 2000) suffers from being formulated exclusively in terms of lexicalization patterns. This is a theoretical matter that has important implications for the interpretation of the huge amount of data that does not fit the typology suggested by Talmy. “Unfitting” data are mostly an expected, and to some extent, acceptable outcome that is not necessarily a threat to a proposed theory. In this case, though, it is evident that substantial amounts of “unfitting” data weaken the accuracy, and particularly the scope, of a generalizing typology such as the one suggested by Talmy. Therefore, “unfitting data”, as widely documented in the literature (e.g. Aske 1989; Berman & Slobin 1994; Gennari et al. 2002; Ibarretxe-Antuñano 2004a, 2004b; Slobin & Hoiting 1994; Slobin 1996, 1997, 2000, 2004; Zlatev & Yangklang 2004) call for a revision of the theory.

3.1  Macro-events – Complex information units In Talmy’s terminology, an event is understood in a strictly cognitive sense: a cognitive unit of perception or conception. He assumes, as we have seen in the previous section, that the macro-event is organized as a figure-ground relation between the main event (ME) and the co-event (CE), which is a supporting event. I will argue against this latter idea. Gestalt-organization in linguistic structure is due to a fundamental human cognitive ability, which is widely, though not exclusively, reflected in the principles of construal in cognitive linguistics (e.g. Langacker 1987). When it is claimed in cognitive linguistics that some complex linguistic structures in a given language are organized as a figure-ground relation, an important motivation for such an analysis is provided when the structure is characterized by a potential alternation, i.e. when there is a possibility of choosing an alternative linguistic construal that corresponds to a figure-ground alternation. The construction of macro-events in fact involves a potential alternation since the main event and the co-event may be construed differently. Talmy’s typological claim about the clausal construction of macro-events, however, is concerned with the existence of fixed patterns in one language type (i.e. not alternating construal in one language) opposed to distinct fixed patterns in other language types. It is therefore an objection to Talmy’s typology that even though his macro-events,



The construction of macro-events: A typological perspective 

(the semantic constituents of the typology) are assumed to be organized on the basis of figure-ground organization, the alternative construal of the macro-event, which is the essence of the typology, is still supposed to take place, exclusively, across languages. In other words, it may be objected that figure-ground conceptualization in linguistic theorizing does not operate across different languages. Talmy’s use of the figure-ground distinction is supposedly analogical with his previous analysis of complex events in compound sentences (Talmy 1978, 2000), as mentioned above. In this early work, the figure-ground distinction accounts for the relative prominence (importance or essentiality) of the events involved in the compound sentence, and it has undoubtedly motivated the choice of terminology in Talmy’s later work on conflated events (macro-events) in simple clauses. This has in fact been pointed out by Talmy himself (e.g. Talmy 2000, vol.II: 215). The main event is defined as the framing event, the basic constituent of the macro-event. In that sense, the main event represents the core information, and the co-event specifies the main event. In terms of prominence, the co-event thus represents a secondary information. Given that this is a correct characterization of the figureground distinction in Talmy’s typology, the role of the gestalt theory in his framework is reduced to be a distinction between main information and secondary information in the construction of macro-events. A different issue is whether the main event and the co-event are always related as figure and ground respectively. If we apply Talmy’s rationale in his analysis of aspectual expressions, explicit expressions of aspect should be treated as the main event and expressions of the verbal process as the co-event, as in he kept [ME] eating [CE]. Firstly, the distribution of ME and CE, and correspondingly of figure and ground, in this example, which represents a very common usage in English, goes counter to Talmy’s typology, which predicts that the aspectual structure (ME) may be explicitly expressed by a satellite, not by the verb as in Spanish. Secondly, in examples like he kept eating and eating, which gives prominence to the activity of eating, eating and eating would not only represent the figure, but also the ground (CE).3 The theoretical point of departure in the present framework that will be developed in the subsequent sections is different. I assume, following basic principles of construction grammar (e.g. Goldberg 2006), that grammatical structure is formed on the basis of generalizations from usage. The implication for the typology is that the generalized asymmetric bipolarity in event conflation is motivated by simple patterns of information structure constructions, abstracted from usage and entrenched in the grammar. Expressions of complex events have a principal schematic meaning that determines the type of event, e.g. ‘X moves Y’ (motion event) or ‘X causes Y to move Z’ (caused motion event), and a specifying meaning element (e.g. the specific manner

.  This example has been suggested to me by an anonymous reviewer.

 Johan Pedersen

or cause of motion). I hypothesize that users tend to make constant generalizations from usage, across different types of complex event expressions: in general complex events involve a delimited principal information and a supportive, specifying chunk of information. In the present framework, the term macro-event is thus used in a slightly different way as compared to the use in Talmy’s work. A macro-event is a complex cognitive event that in the linguistic encoding (conceptualization) process may be subdivided into a main information unit (MI) and a supportive information unit (SI). The MI is the schematic information that defines the semantic domain. The SI specifies the MI. Notice that while figure-ground organization is asymmetric in nature, the present framework, on the contrary, does not exclude the possibility that subevents combine in a symmetric relation maintaining an equal status of prominence as information units. This is important since it may, in some cases, be difficult clearly to distinguish a specific asymmetric relation, to differentiate the main information from the secondary information, or the main event from the co-event in Talmy’s terminology. See e.g. the analysis of aspectual structure as a macro-event (Talmy 2000), cf. the example provided above and example (4) in Section 2. Typological alternations in expressions of macro-events seem to have direct consequences for information structure.4 Talmy’s findings should therefore be interpreted as an information structural phenomenon rather than a matter of lexicalization. This is the reason why I believe that it is reasonable to substitute the terms main event and co-event in Talmy’s framework with main information (MI) and supportive information (SI) in the present account. We may ask: how is the main information (MI) and the supportive information (SI) expressed? In (3), reproduced here as (6), the basic meaning is a state change (‘X causes Y to change’), which is the main information (MI). This information is expressed by the construction as a whole in English and Danish since the meaning of the lexemes out/ud and blew/pustede, per se are not sufficient input for the listener to decipher this basic information.5 In Spanish, however, the same information is encoded by the verb, in the sense that the meaning of the verb is a specific state change, including a complementary valence structure (‘X causes Y to be put out’). The SI, the specifying information about what motivated the state change, is expressed by the verb in English and Danish (‘I blew the candle’), and possibly by an adverbial construction in Spanish: (6) [I blew SI [Jeg pustede SI

the candle

out] MI

(Talmy 1991)

stearinlyset

ud] MI

(Danish)

.  On information structure, see Lambrecht 1994: 5f. and chapter 5. .  The meaning of out/ud is directed motion: he went out, and the meaning of blew/pustede is a specific activity/action: he blew the whistle.



The construction of macro-events: A typological perspective 

Apag-ué la vela [ de un soplido] SI put out-pst.1sg the candle by a blow MI

(Spanish)

To summarize, in the present framework the generalized typology does not assume a general cognitive ability of figure/ground organization of the complex event, and it is not directly concerned with specific patterns of conceptualization. It is rather concerned with regularities in the distribution of information. Hence, the constituents of the typology are, as we shall see, different types of information structure constructions.

3.2  “Unfitting” data Many studies have shown that there is a substantial amount of data that does not fit the typology proposed by Talmy (see e.g. Ibarretxe-Antuñano 2004a, 2004b; Slobin 2004; Slobin & Hoiting 1994; Zlatev & Yangklang 2004). Firstly, as I have already pointed out, the whole idea of typologizing languages has often led to declaring that all languages are a “mixed” type with respect to Talmy’s typology (Croft et al. 2008). In (6), for instance, the macro-event is state change. The English and Danish versions may show the “Germanic” pattern, see section 1.1. However, the user may also choose a “Romance” type. In Danish, it is equally correct and common to construe the main information (MI = ‘X causes Y to change’) by the verb: jeg slukkede (‘I put out’) stearinlyset, as in (6a). In the Spanish version in (6), the MI is expressed by the verb apagué, whereas the SI, the causal factor, may be expressed by an adverbial construction. An alternative would be, as in (6a), to express the SI verbally: soplé (‘I blew’) la vela, in which only the supportive information is expressed, leaving the MI for inferential interpretation: (6) a. Jeg slukke-de stearinlyset. I put out-pst the candle MI Sopl-é la vela. Blow-pst.1sg the candle SI

(Danish)

(Spanish)

When it comes to aspect, temporal contour construal does not always follow the typological pattern suggested by Talmy, see (4) in Section 1.1. The Germanic type as well as the Romance type, e.g. may explicitly encode aspectual structure (= MI) by means of the verbal lexeme: (7) He continued eating. Han fortsatte med at spis-e. he continue-pst with to eat-inf El segu-ía he continue-pst.ipfv.3sg MI

com-iendo. eat-ger

(Danish) (Spanish)

 Johan Pedersen

Another important issue is that some languages do not seem to fit in Talmy’s binary typology (see e.g. Slobin 2004; Slobin & Hoiting 1994; and Zlatev & Yangklang 2004). This is, for instance, the case in serial verb languages like Thai. Slobin suggests the addition of a third type of lexicalization pattern in order to account for such “unfitting” languages. In his typology, he includes, thus, what he calls equipollently-framed languages, in which both manner and path are encoded as main verbs.6 “Unfitting” data is obviously not per se an argument for rejecting a theoretical framework. Nevertheless, massive amounts of negative data have to be taken seriously, and may be an indication that the theory should be revised. The essential question is how to interpret the mismatch. What should its precise impact on the typological theory be? In this article, I will suggest a strategy that instead of typologizing languages, is centred in typologizing particular constructions expressed in a language.

3.3  The need for a constructional approach Sometimes the information cannot be localized in one single clausal element. Instead, it is mapped onto a combination of constituents (Sinha & Kuteva 1995). For instance, very often the path is expressed by a satellite in combination with a preposition: he ran out of the room. Distributedness is not per se a problem for the theory, in fact, as pointed out by Talmy himself, it has been an integrated part of the framework from the outset (Talmy 2005). What I want to point out here is that the typology (Talmy 1991, 2000) suffers from being formulated exclusively in terms of lexicalization, understood as information lexicalized in one constituent, or distributed over a combination of constituents. What the lexical approach, regardless of the admittance of complex units, does not capture is that the schematic construction seems to play a crucial role in clause structure, as an essential part of clausal meaning. That was exactly what Goldberg found in her study of argument structure constructions (Goldberg 1995). In general terms, and according to Goldberg, a schematic construction has features that are not derivable from other form-meaning pairings (such as lexemes) (Goldberg 1995, 2006). The schematic construction type is needed in a generalized typology to be able to account for the basic meaning of the macro-event, i.e. the main event in Talmy’s work and the main information (MI) in the present framework. This becomes clear if we consider one of Goldberg’s classic examples:

(8) He sneezed the napkin off the table.

.  See also the discussion in Talmy (2005).

(Goldberg 1995)



The construction of macro-events: A typological perspective 

Goldberg argues that it is not plausible to claim that the lexical meaning of sneezing, or that of off for that matter, may account for the basic meaning of (8). Therefore, we have to accept the existence and crucial role of schematic constructions in the user’s grammar. Schematic constructions are instantiated by, but not derived from, lexical items, and they have their own schematic meaning. The constructional information is thus independent from the information contributed by the verb, which is an intransitive verb meaning in (8), and it cannot be derived from the lexically encoded information. Since English speakers do understand perfectly well the basic transitive meaning of (8), we are forced to accept that a transitive schematic construction, the so called Caused motion construction, plays the principal role in the encoding of that basic meaning. The caused motion construction has the form: [subj, v, obj, obl] and the meaning: ‘X causes Y to move Z’. The point here is: If a generalized typology has to account for motion events in general, not only intransitive motion events as in (1), but also transitive motion events (caused motion events), the schematic meaning and the schematic construction should play a crucial role. In his analysis of (9), Talmy states that the main event is the transitive motion event (‘I moved the ball into the box’); whereas the supportive co-event is the causal event (‘I kicked the ball’):

(9) I kicked the ball into the box. (Talmy 2000; II: 228)

Correspondingly, the main information communicated in (9) is the caused motion event: ‘X causes Y to move Z’, or as paraphrased by Talmy: ‘I moved the ball into the box’, while the secondary information is the specification of the causal factor (‘I kicked the ball’). However, Talmy’s typological model does not work in this case. In accordance with his typological hypothesis, the main event ‘I moved the ball into the box’, is mapped onto the satellite into. The meaning of into may be defined as ‘entering a container’ (Rudzka-Ostyn 2003), and it is not plausible to assume that the transitive causal element: ‘I caused the ball to move’ should be part of the meaning of into.7 So my point is complementary to the one made by Goldberg in her analysis of the caused motion construction, cf. (8), in which she claims that the (lexical) verbal meaning cannot account for the basic meaning of the construction. I claim that the (lexical) meaning of the satellite into in (9) cannot account for the basic meaning of (9) either.

.  This kind of example shows why it is not convincing when Mendívil Giró, in his contrastive analysis (English-Spanish) of the resultative construction, argues that English may be characterized as a satellite structuring language, and Spanish as a verb structuring language (Mendívil Giró 2003).

 Johan Pedersen

If we analyze expressions of temporal contour (aspect), which is one of the semantic domains that plays a dominant role in Talmy’s generalized typology: (10) I [had just eaten] MI SI Jeg [havde lige spist] MI SI [Acab-aba de com-er] MI finish-pst.ipfv.1sg to eat-inf SI

(Talmy 2000) (Danish) (Spanish)

it becomes clear that the MI, the aspectual structure, in the English/Danish version, is not simply expressed by a lexical item (just). The aspectual meaning in (4), reproduced here as (10), that something happened a short time ago, is expressed by a perfective verb phrase construction plus a lexical item (the satellite just/lige). Likewise, in the Spanish version the aspectual meaning is expressed by the construction: [imperfective verb phrase + de + inf.] plus a lexical item (acabar ‘finish’). In fact, when we in a Danish version choose another verbal form, e.g. the present tense, or the simple past tense: jeg spiser/spiste lige, the meaning of the particle lige (‘just’) will no longer be aspectual, it will be modal. We may say that in the three languages the aspectual meaning is encoded constructionally (in a schematic construction) as well as lexically (in a lexical construction). In all three languages, the SI, the activity in question, is lexically expressed by the second verb (eat, spise, comer). Examples like (9) and (10) point out that a generalized version of the typology, that includes e.g. temporal contour and caused motion, should include the lexical level as well as a schematic constructional level. If schematic, not only lexical, formmeaning pairs should be a central and integrated part of the typology as argued above, and if it is supposed to have a more general scope and not only be concerned with the motion event (Talmy 2000), the form-meaning units of the typology should have a very general nature and should be applicable in various semantic domains. In fact, the argument sketched above applies for a number of basic clause structures, including the intransitive expression of the motion event: (11) The fly buzzed into the room. MI = schematic meaning: ‘X moves Y’ SI = lexical meaning: ‘X buzzed’

(Goldberg 1995)

If we accept that an essential part of the meaning communicated in complex expressions like (8), (9), (10) and (11), and English expressions of macro-events in general, is schematic in nature, how do we then capture the typological differences in expressions of macro-events? The answer is that we need to reformulate the typology in terms of constructions of different specificity, including both lexical and schematic constructions. Such a typology thus needs to be construction-based.



The construction of macro-events: A typological perspective 

This point is further indicated by the way the supportive information (SI) typically is expressed in e.g. Romance languages, particularly in Spanish. In Romance languages the supporting information (e.g. manner or cause) is typically expressed by an adverbial construction: (12) Sal-ió a la calle corr-iendo. go out-pst.3sg onto the street run-ger MI SI (adv)

In terms of lexicalization, the expression of the SI in (12) (the lexical category verb) does not represent a typological regularity. Very frequently not only the MI (e.g. ‘X moves Y’) but also the SI (e.g. manner of motion) are formally expressed by a verb, as exemplified in (12). From a lexical point of view, both the MI and the SI are expressed by a verbal predicate. Consequently, a lexical approach will not capture the typological regularity, ascribable to Romance languages, which concerning the SI is not lexical in nature, but a matter of a schematic (adverbial) construction. In other words, if the expression of the SI (or co-event in Talmy’s terminology) in different languages has to be part of the typology, as it is in Talmy (2000), the typology has to be formulated and interpreted in terms of constructional features: lexical constructions versus schematic constructions.

4.  Macro-event constructions – The constituents of the typology In this section I will reformulate the typology in terms of constructions of different specificity. As we shall see, a constructional unit in this revised typological framework may be lexical or schematic in nature. The basic constituents of the typology will be termed macro-event constructions. Recall that by the term event we do not refer directly to semantic/conceptual content. It has to be understood in a strictly cognitive sense, as a delimited unit of information related to linguistic processing. Macro-event constructions are thus information structure constructions that may either be constructions of the main information (MI-constructions, or MIC), or constructions of the supportive information (SI-constructions, or SIC). [SUBJ, V, OBJ, OBL]/‘MI’ and [V]/‘SI’ formalizes the MIC and SIC, respectively, related to expressions like: Peter kicked the ball into the room, in which the main information (MI) is ‘X caused Y to move Z’, and the supportive information is ‘X kicked Y’. These theoretical points will be explained, and exemplified, in detail below.

4.1  Constructions It should be emphasized that I am not using the term construction in the traditional sense, as a complex syntactic structure. The present use of the term is linked to a

 Johan Pedersen

specific theoretical context. This study has been carried out within the framework of construction grammar (e.g. Goldberg 1995, 2006; Croft 2000, 2001). Constructions are form-meaning pairings of different specificity (Goldberg 2006). Construction grammar is a usage-based approach (Bybee 1985; Langacker 1987, 1988). In usagebased approaches to grammar, grammatical structure emerges from language use in the sense that linguistic units are seen as being abstracted from usage events. Grammatical development and change is thus grounded on abstractions made from actual usage events. The fundamental implication of the usage-based model is that the existence of constructions (pairings of form and meaning) in grammatical representation is a function of frequency and similarity in form and meaning. Frequency is thus an important parameter in construction grammar. When a construction has a high frequency in usage, it is considered to have a high degree of entrenchment, i.e. cognitive automation, in grammar (Langacker 1987). A high degree of entrenchment means that the linguistic structure in question has a stable status in grammar. Constructions are basic rather than epiphenomenal, and rules in grammar are abstracted schematic constructions (Goldberg 1995). Grammar is represented as a network of constructions of different specificity. Schematic constructions and more substantial constructions are thus the basic grammatical constituents of what is sometimes called the constructicon (e.g. Jurafsky 1996). Hence, the grammar contains information about very specific (substantial) elements of language as well as more general patterns (schematic constructions). Constructions form a network and are linked by inheritance relations, which motivate many of the properties of particular constructions. Inheritance allows us to capture generalizations across constructions and particularly the fact that two constructions may be in some ways the same and in other ways distinct (Goldberg 1995: 72). It is further hypothesized that since constructions are the primitive units of representation, the categorical status of their elements is dependent on the construction(s) in which they occur, not the other way around (Croft 2001). Grammatical categories and relations are thus construction specific and undergo constant abstraction and (re)analysis by the users. As language users and language learners we face the task of categorizing utterances into discrete (construction) types. The question of how to identify constructions is thus essentially a categorization problem (Croft 2001). This is an empirical question that, in spite of its subtlety, in principle is testable. The schema is a central term in the usage based model, and hence in construction grammar. A schema is defined as a cognitive representation of user’s generalizations from structural similarities in usage (Goldberg 2006). Schemas are representations of patterns, used in the process of production and comprehension of linguistic expressions. In syntax, where focus is on linguistic form and combinations of form, schemas are often referred to simply as (schematic) constructions. In the following examples, different construction types, relevant for Spanish, are listed:





The construction of macro-events: A typological perspective 

Construction type Lexeme: Morpheme: Impersonal se, e.g. Idiom:

Form [casa] [-s] [se, V3sg, a, OBJ] [más vale tarde que nunca]

Meaning ‘house’ ‘plural’ ‘impersonal meaning’ ‘better late than never’

Another important observation is that most of what we think of as constructions, for instance clausal constructions, contain a range of different constructions in their internal structure. For example (cf. Goldberg 2006): (13) ¿Qué le hizo Pedro a la hija de Fernando? what dat.3sg do-pst.3sg Pedro to the daughter of Fernando ‘What did Pedro do to Fernando’s daughter?’

In the internal structure of the construction in (13) there are at least: [Interrogative] – construction [dative] – construction [VP] – construction [NP] – construction [Lexeme] – constructions [Flexive] – constructions (e.g. mode/aspect/tense)

In Goldberg’s framework, it is not entirely clear which principles govern the interaction of the various constructional levels of the grammar. Apart from being a construction grammar interpretation of Talmy’s descriptive typology, this article is also intended to be a contribution to a better understanding of this issue.

4.2  Macro-event constructions – Procedural function Macro-event constructions are information structure constructions, i.e. pairings of form and information-units; either units of the main-information (MI-constructions = MIC), or units of supportive information (SI-constructions = SIC).8 For instance, [SUBJi V POSSi way OBL]/‘MI’ and [V]/‘SI’ are MIC and SIC, respectively, related to expressions like Peter fought his way out of the restaurant, in which ‘X moving Y by creating intentionally a path’ is the main information (MI) and ‘(X) fought’ represents the supportive information (SI). Macro-event constructions are information structure constructions in the sense that they represent entrenched generalizations about how the information is organized in the clause. Consequently, their basic function is procedural. Macro-event constructions are hypothesized to exist as pairs of MIC and SIC.

.  For a general treatment of relations between form and information-units, see Lambrecht (1994).

 Johan Pedersen

Knowledge of MIC/SIC in grammar helps the user to organize grammatical information, and in processes of production and reception to generate and interpret complex expressions. I hypothesize that the existence of MIC/SIC types in grammar is due to users’ constant generalizations from usage, cf. the usage based approach, and that their function is related to grammatical procedure rather than conceptual representation. An important question is: under which conditions are macro-event constructions, i.e. pairs of MIC/SIC, activated in usage? My proposal is tentatively that MIC/SIC-pairs are activated in production and comprehension when the clausal expression has the macro-event property (MEP) (Bohnemeyer et al. 2007). An expression has the MEP if it packages event representations such that temporal operators necessarily have scope over all subevents. Expressions that have the MEP present an event in terms of a unique initial and/or terminal boundary, a unique duration, and a unique position on the time line (Bohnemeyer et al. 2007:  524). I will further hypothesize that the MEP itself is stored in the grammar as an abstract MEP-construction (form-meaning pair), distilled out of usage via generalization, and linked to (the internal structure of) expressions of macro-events. Further research, however, will have to reveal the exact nature of the MEP. See also the discussion in Bohnemeyer et al. (2007).

5.  A generalized typology of macro-event constructions There seems to be a certain regularity in the way different languages organize the information in a number of complex expressions. Different patterns of information structure may be observed in different language types. In this study it is hypothesized that these patterns are distilled out of usage, in the sense that they stem from users’ constant generalizations from usage. I will suggest that main information constructions (MIC) and supportive information constructions (SIC) of varying degrees of specificity should be the basic constituents of a generalized typology of macro-events. [SUBJ, V, OBL …]/‘MI’, a schematic construction of the main information, and [V]/‘SI’, a lexical (verbal) construction of the supportive information, are typical MIC/SIC in Germanic languages like Danish, English and German. In Romance languages, [V]/‘MI’, a lexical (verbal) construction of the main information, and [advform]/‘SI’, a schematic (adverbial) construction of the supportive information, are typical patterns. MIC/SIC are entrenched in the grammar to different degrees in different languages. Some MIC/SIC play a dominant role, while others are less prominent. This framework provides us with a typological characterization that is not reduced to one determined set of lexicalization patterns. I believe this is an advantageous, and necessary, theoretical adjustment of Talmy’s typology. In the first place, it offers a more adequate description of typological regularities. It is, moreover, also important from



The construction of macro-events: A typological perspective 

an empirical point of view, since we have to face the existence of a vast amount of data, in all the languages studied, which do not follow the typological patterns suggested by Talmy. The constructional approach permits us (as we shall see in Section 5.1 to 5.4) on the one hand to analyze an even broader range of expressions as instances of the same typological patterns, and on the other hand, to handle the substantial amount of data that do not fit the basic patterns. When qualifying the MIC and the SIC with respect to schematicity, only the possession of the schematic feature will be focused on. For instance, when a MIC is qualified as being schematic, it is implied that the MI is not organized and encoded, and cannot be identified solely by a lexical item; it is organized and encoded, and can only be identified by a more complex schematic construction. A schematic MIC is precisely in this sense different from a lexical MIC, though it is obvious that lexical information in general is relevant for the specification of the content. How do we know that information structure constructions (MIC/SIC) actually exist, and that they are relevant for the typology? it may be objected. Are they not just the linguist’s generalizations upon clausal content and form, and cannot they simply be derived from specific complex expressions of conceptual content, such as expressions of motion, that have the macro-event property (Bohnemeyer et al. 2007)? MIC/ SIC are, in fact, derived from expressions of conceptual content, but importantly, only if we look at it in a diachronic perspective. This is just as true as the fact that Goldberg’s schematic constructions over time are derived, via constant generalizations over usage, from verbal (lexical) content. Synchronically, MIC/SIC, and Goldberg’s schematic constructions, are not derived elements of grammar. That is exactly what defines them as constructions in the construction grammar framework (Goldberg 1995, 2006). Goldberg’s schematic constructions are not synchronically derived from lexical content since they may be used independently from the verbal meaning (valence structure). MIC/SIC are not synchronically derived from complex expressions of conceptual content since they have their own, independent, procedural role in grammar: they organize the information in chunks of main information and chunks of supportive information. The hypothesis of this article is that the typological patterns are anchored in this task, and that MIC/SIC, therefore, are the basic constituents of the typology. Thereby, the typology is better suited for analyzing actual usage data since information structure is highly sensitive to users’ (including translators’) individual choice and strategies in performance.

5.1  Motion events The expressions of transitive motion (caused motion events) in (14) and (15) contain two information units whose formal expressions are packed in a simple clause structure: The main information (MI) is about causing an entity to move: ‘X causes Y to

 Johan Pedersen

move Z’, and the supportive information (SI) is a specification of the causal factor, i.e. more specifically, how did the caused motion event come about?9 (14) Fred stuffed the papers in the envelope. Fred presse-de papirerne ned i konvolutten. Fred press-pst the papers down into the envelope Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

(Goldberg 1995) (Danish)

Fred met-ió con dificultad los papeles en el sobre. Fred place-pst.3sg with difficulty the papers in the envelope Lexical MIC: [V]/‘MI’ (Spanish) Schematic SIC: [adv-form]/‘SI’

In (14) the MI (‘X causes Y to move Z’) is in English and Danish schematically expressed by the formal pattern: [SUBJ, V, OBJ, OBL], while the SI (‘the specification of the causal factor’ = ‘X stuffed/pressede Y’) is lexically expressed. Notice, in continuation of the discussion of (9), that in terms of lexicalization it is implausible to claim that the MI is lexically encoded in the English and Danish version (in …/ned i …). In Spanish the MI (‘X causes Y to move Z’) is lexically expressed (metió), whereas the SI, the specification of the main information, is expressed by an adverbial construction, [X … con dificultad]. Here it is implausible to claim that the SI is lexically encoded. The reason is that dificultad simply represents information about ‘a difficulty’, while the adverbial construction [X … con dificultad] provides specifying information, with respect to ‘X causes Y to move Z’ (MI) about the causal factor. We find the same patterns in (15): (15) Sam washed the soap out of her eyes. Sam vaskede sæben ud af hendes øjne. Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

(Goldberg 1995) (Danish)

Sam le quit-ó el jabón de los ojos con agua. Sam dat.3sg take away-pst.3sg the soap from the eyes with water Lexical MIC: [V]/‘MI’ (Spanish) Schematic SIC: [adv-form]/‘SI’

In English and Danish the MI (‘X causes Y to move Z’) and the SI (the specification) is expressed by a schematic and a lexical (verbal) construction respectively. In Spanish the MI is constructed lexically (quitó) while the expression of the SI is an adverbial construction (con agua).

.  The English version in some of the examples in the following sections is taken from Goldberg (1995), who argues for the existence of constructional meaning in general. The Spanish version is provided by native speakers at the University of Copenhagen.



The construction of macro-events: A typological perspective 

If this analysis is applied to expressions of intransitive motion events, see (1), reproduced here as (16), we will find that the MI, ‘X moves Y’, is encoded by a schematic construction, [SUBJ, V, OBL], in English and Danish. The SI, ‘the manner of motion’, is lexically expressed: (16) The bottle floated into the cave. Flasken flød ind i hulen. Schematic MIC: [SUBJ, V, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

La botella entr-ó en la cueva flot-ando. the bottle enter-pst.3sg in the cave float-ger Lexical MIC: [V]/‘MI’ Schematic SIC: [adv-form]/‘SI’

(Talmy 1985) (Danish)

(Spanish)

In Spanish the MI is lexically expressed by the main verb (entró), whereas the SI is expressed by an adverbial construction. Notice that the specifying information cannot be identified lexically by the verb float, per se. The schematic adverbial form [(main clause) V-gerund] provides the supportive specification.

5.2  Resultatives Talmy analyzes (resultative) expressions of what he terms (macro) events of realization/ completion, though he does not provide any Spanish examples. See (Talmy 2000: 262f): (17) I kicked the hubcap flat. (18) I washed the shirt clean.

He argues that the satellites flat and clean encode the main event, and that the verbs kicked and washed encode the co-event (Talmy 2000: 262ff + 278ff). However, it is not very convincing that the basic meaning component ‘X causing Y to become flat/clean’ should be attributed to the meaning of the satellite flat/clean per se, cf. the analysis of (9): he kicked the ball into the box. Goldberg (1995) does not discuss Talmy’s work, but in her analysis of the English resultative she provides basically the same counter-argument, though she focuses on the verb and its contribution to the clausal meaning. In short, she argues that the basic meaning of the clause, i.e. ‘X causes Y to become Z’, cannot convincingly be assigned to the lexical meaning of the verb, kicked in (17) and washed in (18). Instead, she claims that the basic meaning of the clause is provided by a schematic construction. The independent role of the schematic construction, which is perhaps the most important principle in Goldberg’s 1995 book, is criticized by Hans Boas in his work on the resultative construction (e.g. Boas 2003). In particular, he criticizes what he sees as a top-down approach that is not detailed enough to account for the licensing

 Johan Pedersen

of resultative phrases and non-subcategorized NP’s. He argues that a more precise lexical analysis of the different senses of a verb provides a more fine-grained system that may account for the distribution of resultative expressions that occurs with a given verb. Boas’ proposals implicate a shift of the explanatory burden from the level of abstract constructional semantics to the level of concrete verbal semantics (Boas 2003:  313ff.). It should be noted, though, that in her 2006 book, Goldberg is much more explicit about the usage based status of her framework. She makes it very clear that the existence of schematic constructions in language is due to users’ generalizations over usage, and that the variety of constructions within a given language exists to enable speakers to package information in useful ways (Goldberg 2006: 228). Boas’ approach is interesting for the analysis of Spanish, and other Romance languages, in which expressions of resultative meaning (and other meaning structures, as exemplified in this article) tend to be lexically organized and centred in verb semantics. In fact, contrastive data seem to indicate that there are systematic differences between English and Spanish with respect to the way information is organized in the clause (Mendívil Giró 2003, Pedersen in press, Snyder 2001). Constructions of resultative meaning, for instance, appear to be relatively schematic in English as compared to parallel Spanish versions, which tend to be centred in the verb. Elaborating on Goldberg’s analysis (Goldberg 1995, 2006), I will suggest that in resultative expressions, the MI is the schematic meaning ‘X causes Y to become Z’; and the SI is a specification of that meaning: (19) She kissed him unconscious. Schematic MIC: [SUBJ, V, OBJ, PRED]/‘MI’ Lexical SIC: [V]/‘SI’

Le desmay-ó con un beso. acc.3sg.m faint-pst.3sg with a kiss Lexical MIC: [V]/‘MI’ Schematic SIC: [adv-form]/‘SI’

(Goldberg 1995)

(Spanish)

It seems implausible to claim that the MI is lexically constructed in the English version in (19). The lexeme unconscious does not encode, either directly or indirectly, the basic meaning ‘X causes Y to become Z’ since unconscious, per se could also refer to she is unconscious, whose meaning is by no means ‘resultative’. Alternatively, I suggest that the encoding of the MI (‘X causes Y to become Z’) is centred in a schematic construction in English and in a lexical (verbal) construction in Spanish. The supportive information is expressed lexically in English and schematically in Spanish. (20) may be analyzed as (19): (20) She licked the plate clean. Hun slikkede tallerkenen ren. Schematic MIC: [SUBJ, V, OBJ, PRED]/‘MI’ Lexical SIC: [V]/‘SI’

(Danish)





The construction of macro-events: A typological perspective 

Ella limp-ió el plato con la lengua. she clean-pst.3sg the plate with the tongue Lexical MIC: [V]/‘MI’ Schematic SIC: [adv-form]/‘SI’

(Spanish)

The MI, the resultative meaning, is encoded by a schematic construction. It is not encoded lexically by clean for the same reason that the MI is not expressed lexically by unconscious in (19). In the Spanish version, we see the characteristic pattern, cf. (19). Notice again that while the MI is encoded lexically by the verb, the expression of the supportive information (SI) shows typological regularity as a schematic adverbial construction. The lexical specification of the adverbial construction is not relevant as a typological regularity, as I have already pointed out. It would make little sense to claim, as a statement about typological patterns, that the SI in the Spanish version is expressed by a noun (lengua ‘tongue’). The supportive information may be provided in a number of ways (different construction types), most frequently perhaps the SI is expressed by means of the gerund (verb phrase construction), and specified by a verbal lexeme. Some expressions of ‘state change’ (Talmy 2000), see above and (3), reproduced here as (21), may also be analyzed as a specific kind of resultatives in terms of macroevent constructions. The MI is ‘X causes Y to become Z’ and the SI is a specification of the causal factor. See also Section 6.3. (21) I blew the candle out. Jeg pustede stearinlyset ud. Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

Apag-ué la vela de un put out-pst.1sg the candle by a Lexical MIC: [V]/‘MI’ Schematic SIC: [adv-form]/‘SI’

soplido. blow

(Talmy 1991) (Danish)

(Spanish)

5.3  Temporal contour (aspect) As demonstrated in Section 1.1, aspectual structures may also be expressed as macroevents. The MI is the aspectual structure and the supportive information (SI) is the activity/process in question. However, aspectuality is a very special semantic domain for two particular reasons. In the first place, no action, activity or state is involved as main information (MI), only as secondary information (SI). Secondly, aspectuality is an integrated part of the lexical meaning of every verb phrase. Therefore, the claim maintained by Talmy (2000), that the MI (the main event in Talmy’s terminology) is expressed specifically by a satellite in some languages, e.g. Germanic languages, and

 Johan Pedersen

by the verb, in other languages, like e.g. Spanish, is not completely convincing. See (4), reproduced here as (22):10 (22) I had just eaten. Jeg havde lige spist. Schematic MIC: [AUX, SAT, V]/‘MI’ Lexical SIC: [V]/‘SI’

Acab-aba de comer. finish-pst.ipfv.1sg to eat Schematic MIC: [acab-pst.IPFV, de, V-inf]/‘MI’ Lexical SIC: [V]/‘SI’

(Talmy 2000) (Danish)

(Spanish)

In English and Danish, the MI (the aspectual structure) is expressed by a perfective verb phrase-construction in combination with a satellite. The SI is lexically expressed by the verb. In Spanish, the MI is expressed by an imperfective verb phrase-construction in combination with the verb acabar, the preposition de and a verb in the infinitive flexional form. The SI is expressed by a specific verbal lexeme. In short, in the English and Danish version, as well as in the Spanish version, the MI, the aspectual structure, cannot be delimited to stem from a specific lexeme. In the three languages, the MI is encoded by a complex schematic construction.

5.4  Perception Some frequent expressions of perception may be analyzed in terms of macro-event constructions. The main information (MI) is ‘X perceives Y’ and the supportive information (SI) is ‘the manner of perception’: (23) She looked happy. Schematic MIC: [SUBJ, V, PRED]/‘MI’ Lexical SIC: [V]/‘SI’

Hun så glad ud. Sie sah froh aus. she looked happy out Schematic MIC: [SUBJ, V, PRED, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

(Danish) (German)



Parec-ía contenta (observ-ándo-la). seem-pst.ipfv.3sg happy observe-ger-acc.3sg.f Lexical MIC: [V]/‘MI’ Schematic SIC: [adv construction]/‘SI’

(Spanish)

.  According to Talmy (2000: 233), English seems to represent a mixed typological picture, both in the domains of ‘motion’ and ‘temporal contouring’ (aspect), though it leans toward qualifying as being satellite-framed.



The construction of macro-events: A typological perspective 

In the English version, the perception path, someone else (e.g. the speaker) per­ ceiving the subject (her), is not lexicalized solely by the perception verb look since this verb could also appear in frequent expressions like: she looked out of the window in which, even though the basic meaning of visual perception would be the same, the perception path would be the opposite: perception by the subject (her) of something else.11 Neither is it lexicalized, obviously, in the adjective happy. This demonstrates, again, that the main information (MI), the act of perception by X directed toward Y, is expressed by a schematic construction. The supportive information (SI), the visual manner of perception, on the other hand, is expressed lexically by the verb (look). The Danish and the German version show a similar pattern: neither the verb så/sah nor the satellite ud/aus lexicalize the act of perception (MI) since both lexemes appear in the frequent expression type: hun så ud af vinduet / sie sah aus dem Fenster (‘she looked out of the window’), in which, as in the English version, the perception path is: perception by the subject (her) of something else. And it is not convincing at all to claim that the MI is lexicalized by the adjective glad/fröhlich. Also in this case the only plausible solution is to say that a schematic construction is an essential part of the encoded act of perception (MI). As regards the supporting information (SI), the visual manner of perception, the SI is lexically expressed by the verb (så/sah). In the Spanish version, the act of perception (MI) is expressed lexically by the verb parecer, while the visual manner of perception, the supportive information (SI), may be expressed by an adverbial construction. We use the same construction when the act of perception is auditory: (24) (Escuch-ándo-la) parec-ía irrit-ada. listen-ger-acc.3sg.f seem-pst.ipfv.3sg irritate-ptcp.f ‘When I listened to her, she seemed irritated’ Lexical MIC: [V]/‘MI’ Schematic SIC: [adv construction]/‘SI’

(Spanish)

Notice that English and Danish have similar constructions, in which the act of perception (MI) is expressed verbally, while the manner of perception (SI) is unspecified: (25) She seemed (to be) irritated. Hun virkede irriteret. Lexical MIC: [V]/‘MI’

(Danish)

.  Expressions of this kind of (basic) visual perception by the clausal subject have been studied in a crosslinguistic perspective by Slobin, among others. See Slobin (forthcoming) and references cited there.

 Johan Pedersen

Moreover, in Spanish we also find a pattern that is similar to the typical English one (she looked tired), cf. (23): (26) Pedro la ve-ía cans-ada. Pedro acc.3sg.f see-pst.ipfv.3sg tire-ptcp.f Schematic MIC: [SUBJ, V, PRED]/‘MI’ Lexical SIC: [V]/‘SI’

(Spanish)

It is interesting to notice that in all four languages discussed in this section, the act of perception (MI) and the specification of the manner of perception (SI) are in some cases expressed verbally, in the same lexical construction:12 (27) She sounded irritated. Sie klang gereizt. Hun lød irriteret. Ella sonaba irritada. Lexical MIC: [V]/‘MI, SI’

(German) (Danish) (Spanish)

We have seen in this section that contrastive analysis of expressions of macro-events indicate that a generalized version of the typology has to include not only a lexical level of analysis, but also a more schematic constructional level of analysis. In this perspective, the typology is a matter of mapping out macro-event constructions in different languages in terms of lexical constructions versus schematic constructions. Hence, the typology is not simply a question of identifying patterns of lexicalization. MIC/SIC, the constituents of the typology, are entrenched generalizations about how the principal and the secondary information are organized. The function of MIC/SIC is thus procedural rather than related to symbolic representation of specific conceptual structures. This revised version of the typology permits us to analyze a broader range of expressions as instances of the same typological pattern. Moreover, the constructional approach permits us to include, in a predicted and systematic way, more variation in the typological description. As will be demonstrated in the following sections, in order to work out a typological description that has a broader and more general scope, on the one hand, and includes more variation on the other, the first step has been to carry out a pilot study.

6.  A contrastive analysis of macro-event constructions As the empirical part of this study, and as a pilot study, I have carried out a contrastive analysis of a short story (The Snowman) by H.C. Andersen. The corpus material is

.  In the data only one particular expression type represents this pattern: She sounded irritated. See also Rojo and Valenzuela’s contrastive study of English and Spanish verbs of perception (Rojo & Valenzuela 2005).



The construction of macro-events: A typological perspective 

available, digitally, in six parallel versions: the original Danish version, an English, a German, a Spanish, an Italian and a French version (Andersen 2005). In a joint project (The Mulinco Project), the Centre for Language Technology and the Department of English, Germanic and Romance Studies at the University of Copenhagen have collaborated in the development of a corpus platform that contains parallel corpora. Parallel corpora are a particularly suitable tool for contrastive and typological linguistic research. A set of parallel corpora contains translated texts in various languages. The texts are aligned by sequence or period, so that by searching for an expression in one language you will get direct access to the corresponding sequences of text in other languages. The creation of large, digital parallel corpora, sequentially aligned, makes it possible to carry out multilingual typological research on macro-event constructions. So far macro-event constructions (MI-constructions and SI-constructions) have been identified on the basis of semantics, cf. the headings of the subsections in Section 5. It is a very complicated task, however, to search for semantically defined entities because the available software can search only on surface form. This is a serious matter that needs to be solved in the planning of larger scale research projects, for instance, by identifying recurrent formal patterns that determine a basic inventory of macro-event constructions. In the following subsections, I will exemplify the general analysis by analyzing a number of text sequences from The Snowman (Andersen 2005) in six parallel versions: The original Danish version, an English, a German, a Spanish, an Italian and a French version.13 I am aware that some of the examples, e.g. (23), represent metaphorical, or figurative, extensions from the conceptual categories that have defined the inventory of macro-events in Talmy’s work. One of them, (29), is even completely unrelated to the Talmian framework. In fact, it is easy to find prototypical examples, which are directly comparable with the examples in Talmy’s work, to exemplify the present framework, see this section and Section 5. Some of the examples chosen for this section are therefore intended to demonstrate that there is in principle no reason why the typology should be limited to include only the five specific semantic domains identified by Talmy.14 Of course, we have to recognize, though, the significance of the huge amount of descriptive typological work, mostly on expressions of the motion event, that represent evidence for his proposals from an extremely broad range of languages.

.  In the analysis, I have, for reasons of simplicity, not taken into account constructional variation due to word order. .  I.e. motion event, event of state change, event of temporal contouring, event of action correlating and event of realization (Talmy 2000).

 Johan Pedersen

6.1  Motion events (28) a. Det lys-te rødt lige op af hans Bryst. it shine-pst red right up of his breast Schematic MIC: [SUBJ, V, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ b. It gleamed red upon his bosom. Schematic MIC: [SUBJ, V, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ c. Es leucht-ete rot seine ganze Brust herauf. It shine-pst.3sg red his whole breast up Schematic MIC: [SUBJ, V, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ d. Su his

pecho desped-ía también un brillo rojizo. chest emit-pst.ipfv.3sg also a light reddish

Lexical MIC: [V]/‘MI’ Lexical SIC: [NP]/‘SI’

e. Illumin-ava di rosso il suo petto. Illuminate-pst. ipfv.3sg with red his breast Lexical SIC: [V]/‘SI’ f. La blanche poitrine du Bonhomme de neige the white chest of the man of snow

en recev-ait des reflets rouges. receive-pst. ipfv.3sg reflections red



Lexical MIC: [V]/‘MI’ Lexical SIC: [NP]/‘SI’

In (28), the Danish, the English and the German version, as expected, express the MI (‘X moves Y’) in a schematic construction [SUBJ, V, OBL]/‘MI’). Specifying information (SI) about the motion event, which has to be understood in a figurative sense as light that moves, is verbally expressed. In the Spanish and the French versions, the MI is expressed verbally (= emit and receive respectively). The supportive SI is expressed by a NP (= a redish light/reflections). Notably, the Italian version does not express the motion event. Instead, it focuses exclusively on the event of illumination, which is expressed by means of a lexical construction centred in the verb (illuminava). In (29), in which the motion event is transitive, we find the same pattern, though in this case the Italian version follows the “Germanic” pattern. In the Danish, English, German and Italian version, the MI (‘X causes Y to move Z’) is expressed by a schematic construction, whereas



The construction of macro-events: A typological perspective 

the SI, the secondary specification, is verbally expressed. In Spanish and French, the MI is verbally constructed, and not further specified, i.e. there is no SI encoded. (29) a. Hvor det klæde-r hende at rækk-e Tungen how it suit-prs her to stretch-inf the tongue Schematic MIC: [(SUBJ), V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

ud! out

b. How beautiful it looks when it stretches out its tongue. Schematic MIC: [SUBJ, VP, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ c. Wie schön es ihr steh-t, die Zunge so how beautiful it her stand-prs.3sg the tongue so heraus-zu-streck-en. out-to-stretch-inf Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ d. Qué bien le sient-a eso de sac-ar how well dat.3sg become-prs.3sg that of take-inf out la lengua. the tongue Lexical MIC: [V]/‘MI’ e. Come le don-a quando tir-a how dat.3sg give-prs.3sg when stretch-prs.3sg

fuori la lingua. out the tongue



Schematic MIC: [SUBJ, VP, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

f. C’est si bon lorsque la langue lui sort that’s so fine when the tongue dat.3sg hang-prs.3sg out

de la bouche. of the mouth



Lexical MIC: [V]/‘MI’

Example (30) shows the same patterns. In the Danish, English, German and Italian version a lexical (verbal) construction is used to specify the causal factor of the motion event (SI), while the lexical (verbal) construction expresses the MI (‘X causes Y to move Z’) in the Spanish and the French version (quitó, prendre). Notice that the Italian version of the motion event is not constructed in the same way as a macro-event with main information and supportive information. Only a specifying information

 Johan Pedersen

unit is expressed. The main information (MI), which in the other languages is explicitly encoded, has to be inferred in the Italian version. Correspondingly, only the MI (‘X causes Y to move Z’) is explicitly expressed (verbally) in the Spanish and the French version, leaving the specification (SI) for inferential contextualization. (30) a. Han stød-te fra mig det Been, jeg gnave-de paa. he push-pst away from me the bone I gnaw-pst on Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ b. He kicked away the bone I was gnawing. Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ c. Er mir den Knochen weg-stieß, an dem he dat.1sg the bone away-push-pst.3sg on which

ich nag-te. I gnaw-pst.1sg



Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’

d. Me quit-ó un hueso que dat.1sg take-pst.3sg away a bone that

est-aba ro-yendo. be-prog.pst.ipfv.1sg gnaw-ger Lexical MIC: [V]/‘MI’

e.

Aveva da-to un calcio all’ osso che aux-pst. ipfv.3sg give-ptcp a kick to the bone that



stavo rosicchi-ando. be-prog.pst. ipfv.1sg gnaw-ger Schematic SIC: [SUBJ, V, OBJ]/‘SI’

f. qui ven-ait de me prend-re un os. which come-pst.ipfv.3sg of dat.1sg take-inf away a bone ‘Which just had taken a bone away from me’ Lexical MIC: [V]/‘MI’

6.2  Aspect (31) exemplifies the macro-event ‘temporal contour’. It contains the basic information unit (MI) about the aspectual structure: something happens again, and a specification (SI) of the involved verbal process (something appears). In the Danish, English, German and French versions, the MI is lexicalized in a satellite: igjen, again, wieder, de nouveau,



The construction of macro-events: A typological perspective 

while the SI is expressed verbally: viste sig, showing himself, zeigt sich, se montrait. In the Spanish and the Italian versions, the MI is expressed lexically (verbally) in a specific schematic construction: [volvía/tornava + a + infinitive], whereas the SI is expressed lexically: aparecer, mostrarsi: (31) a. Han tro-ede, at det var Solen, der vis-te he think-pst that it be-pst the sun that show-pst

sig igjen. itself again



Lexical MIC: [SAT]/‘MI’ Lexical SIC: [V]/‘SI’

b. He intended to say the sun is showing himself again. Lexical MIC: [SAT]/‘MI’ Lexical SIC: [V]/‘SI’ c. Damit woll-te er sag-en: die Sonne zeig-t Thus want-pst.3sg he say-inf the sun show-prs.3sg

sich wieder. refl.3sg again



Lexical MIC: [SAT]/‘MI’ Lexical SIC: [V]/‘SI’

d. Cre-ía que era el sol que think-pst.ipfv.3sg that be-pst.ipfv.3sg the sun that

volv-ía a aparec-er. return-pst.ipfv.3sg to appear-inf



Schematic MIC: [Volvía, a, V-inf]/‘MI’ Lexical SIC: [V]/‘SI’

e. Cred-eva che fosse il sole che think-pst.ipfv.3sg that be-pst.ipfv.sbjv.3sg the sun that

torn-ava a mostr-ar-si. return-pst.ipfv.3sg to appear-inf-refl.3sg



Schematic MIC: [tornava; a; V-inf]/‘MI’ Lexical SIC: [V]/‘SI’

f. Il pens-ait que c’ était le soleil qui he think-pst.ipfv.3sg that be-pst.ipfv.3sg the sun that

se montr-ait de nouveau. refl.3sg show-pst.ipfv.3sg again



Lexical MIC: [SAT]/‘MI’ Lexical SIC: [V]/‘SI’

 Johan Pedersen

6.3  State change Some expressions pertain to a semantic domain that I will term change of condition, which is a subcategory of the more general category state change.15 Expressions of change of condition may profitably be analyzed typologically in terms of macro-event constructions. The main information (MI) is about the change of condition and can be formalized as: ‘X causes Y to be in condition Z’. The supportive information (SI) specifies how this change comes about (‘the manner’). In (32), the MI is about a change of condition: ‘X causes Y to be in condition Z’ (from being in a state of freedom to being chained up). The SI specifies how this change of condition has come about. In the Danish, German, Italian and the French versions, the MI is expressed in a complex schematic construction: [SUBJ, V, OBJ, OBL]. In the English and the Spanish versions, the MI is expressed verbally: chained, encadenaron. The SI is expressed verbally in the Danish and the German versions: De satte mig, Man legte mich, while the SI is not expressed in the English, Spanish, Italian and the French versions: (32) a. De satte mig her i Lænke. they seat-pst me here in chain Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIS: [V]/‘SI’ b. Man leg-te mich hier an die Kette. they lay-pst.3sg me here in the chain Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIS: [V]/‘SI’ c. They chained me up here. Lexical MIC: [V]/‘MI’ d. Me encaden-aron. acc.1sg chain-pst.3pl Lexical MIC: [V]/‘MI’ e. Mi hanno messo qui al-la catena. acc.1sg aux-prs.3pl put-ptcp here in-the chain Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ f. On me mit à l’-attache. one acc.1sg put-pst.3sg in the-chain Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’

.  Talmy presents examples of what he calls change in condition, which is a slightly different category.



The construction of macro-events: A typological perspective 

(32) divides the languages into three groups: 1) Danish/German; 2) English/Spanish; and 3) Italian/French, demonstrating that Italian and French in this case share features with Danish and German as well as with English and Spanish. In (33), the MI is figuratively a state change, ‘X causes Y to become Z’ (from not being alive to being alive). The supportive information is a specification of how the change of state has come about. In the Danish, English and German versions, the MI is constructed by a schematic construction: [SUBJ, V, OBJ, OBL]. The SI is encoded by verbal organization (predicate + arguments). In the Spanish and Italian versions, the MI is encoded lexically by verbal organization: infundir (vida), ridare (vita), and the SI, the way in which the change of state has come about, is expressed nominally: el viento cortante/el vento: (33) a. Vinden kan rigtignok bid-e Liv i Een! the wind may-prs.3sg indeed bite-inf life into one Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ b. This is a kind of wind that can blow life into one. Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ c. Der Wind kann einem wirklich Leben ein-beiß-en. the wind may-prs.3sg to one indeed life into-blow-inf Schematic MIC: [SUBJ, V, OBJ, OBL]/‘MI’ Lexical SIC: [V]/‘SI’ d. El viento cortante pued-e infund-ir vida en uno. the wind cutting may-prs.3sg instill-inf life into one Lexical MIC: [V]/‘MI’ Lexical SIC: [NP]/‘SI’ e. Il vento sa proprio rida-re vita. the wind know-prs.3sg really give-inf back life Lexical MIC: [V]/‘MI’ Lexical SIC: [NP]/‘SI’ f. Et ce vent cinglant, comme il vous fouett-e and this wind biting how it you whip-prs.3sg

agréablement! pleasantly



Lexical MIC: [V]/‘MI’ Lexical SIC: [NP]/‘SI’

 Johan Pedersen

6.4  Complex circumstances The MIC/SIC-model helps us to recognize general typological patterns in usage that we otherwise would not be aware of. A description of complex circumstances may be performed by organizing the information as macro-event constructions. In (34), the complex situation is the following: the dog is fastened to a chain, and the dog lies out in the cold. The main issue (MI), ‘X is in circumstance Y’, is (or rather: may be chosen to be) that the dog is fastened to a chain. The specifying information (SI), ‘X is in circumstance Z’, is that this scenario takes place out in the cold. In the Danish, the German, the Italian and the French versions, the MI is expressed by a schematic construction, e.g. jeg stod ikke i lænke (Danish), [SUBJ, V, OBL]/‘MI’. The expression of the SI is likewise schematic, e.g. jeg stod ikke her I kulden (Danish), [SUBJ, V, OBL]/‘SI’. The Spanish version follows a typical pattern, i.e. the MI is verbally organized: (estar) encadenado, [V]/‘MI’, while the SI is added as an adverbial construction: a la intemperie, [adv-form]/‘SI’. The English version has mixed properties. The MI is verbally organized, as it is in the Spanish vesion: fastened to a chain, while the specifying information (SI) is expressed by a schematic construction: I (did not) lie out here in the cold, [SUBJ, V, OBL]/‘SI’, as it is in the Danish, German, Italian and the French versions. (34) a. en Tid, hvor jeg ikke stod her a time when I not stand-pst.1sg here

i lænke. in chain



Schematic MIC [SUBJ, V, OBL]/‘MI’ Schematic SIC [SUBJ, V, OBL]/‘SI’

i kulden in the cold

b. eine Zeit, da lag ich nicht hier in der Kälte a time when lie-pst.1sg I not here in the cold

an der Kette. in the chain



Schematic MIC [SUBJ, V, OBL]/‘MI’ Schematic SIC [SUBJ, V, OBL]/‘SI’

c. a time when I did not lie out here in the cold, fastened to a chain. Schematic MIC: [Free predicative construction]/‘MI’ Schematic SIC: [SUBJ, V, OBL]/‘SI’ d. un tiempo en que no ten-ía que est-ar a time when not have-pst.ipfv.1sg to be-inf

encaden-ado a la intemperie. chain-ptcp in the bad weather



Lexical MIC: [V]/‘MI’ Schematic SIC: [adv-form]/‘SI’



The construction of macro-events: A typological perspective 

e. tempi in cui non st-avo qui al freddo time when not be-pst.ipfv.1sg here in the cold

al-la catena. in-the chain



Schematic MIC [SUBJ, V, OBL]/‘MI’ Schematic SIC [SUBJ, V, OBL]/‘SI’

f. un temps où je n’-ét-ais pas dans la cour, a time when I neg-be-pst.ipfv.1sg not in the yard

au froid à l’-attache. in the cold in the-chain



Schematic MIC [SUBJ, V, OBL]/‘MI’ Schematic SIC [SUBJ, V, OBL]/‘SI’

The analysis of (34) shows, to some extent, similarities with the analysis of macroevent constructions in other semantic domains, particularly with respect to the Spanish type. This indicates that a typology formulated in terms of MIC/SIC has an extended, more general application. But it also indicates that by applying the MIC/ SIC-model for typological analysis, we may account for important typological peculiarities in each language.

7.  Some results Table 1 summarizes the analysis by listing the most commonly occurring pairs of macroevent constructions that have been identified in the languages examined in this pilot study. Table 1.  Macro-event Constructions in six languages Schematically encoded MI [SUBJ, V, OBL]/‘MI’ [V]-‘SI’ [SUBJ, V, OBJ, OBL]/‘MI’ [V]/‘SI’ [SUBJ, V, OBJ, PRED]/‘MI’ [V]/‘SI’ [SUBJ, V, PRED, OBL]/‘MI’ [V]/‘SI’ [SUBJ, V, PRED]/‘MI’ [V]/‘SI’ [..] /‘MI’ [V]/‘SI’ Lex./verbally encoded MI   7. [V]/‘MI’ [adv-form]/‘SI’   8. [V]/‘MI’ [NP]/‘SI’   9. [SAT]/‘MI’   1.   2.   3.   4.   5.   6.

[V]/‘SI’

10. [V]/‘MI’ [..]/‘SI’

Danish English German Spanish Italian French X X X X X X

X X X X X

X X X X X X

(X) (X)

(X) (X)

(X) (X)

(X) X

X

X

X X X

X X X

X

X

X

X

X

X X X

(X)

(X)

(X)

X

 Johan Pedersen

In the first part of Table 1, construction type 1–6, we have MIC/SIC in which the main information (MI) is not, while the supportive information (SI) is, organized and encoded by the verb. The typical pattern in this group is that the MI is schematically encoded. In the second part of the table, construction types 7–10, the MI is lexically organized and encoded, typically by the verb. Table 1 seems to confirm the existence of a well-known, though very complex, typological picture, since some pairs of MIC/SIC may, to some degree, be observed systematically across the traditionally defined categorization of the languages in question. According to the (very sparse amount of) data in this pilot study, Germanic languages tend to express the MI in a schematic construction and the SI by the verb. These languages may in some cases express the MI by a lexical item (verb, satellite) in which case the SI tends to be lexically expressed as well, or unexpressed. Romance languages, and particularly Spanish, tend to construct the MI by a lexical item, mostly the verb, and the SI by an adverbial construction, though a schematic construction of the MI and a lexical construction of the SI may be found as well. The Italian versions have, surprisingly, been shown to follow almost systematically the “Germanic” type.16 This may be due to translators’ strategy, and individual choices, but it may also reflect the fact that typological categorizations are not so clear-cut as sometimes assumed in the literature. English is typologically ambivalent, as pointed out by Talmy (1991, 2000) and other scholars. Noticeably, all the languages show a considerable typological variation in the construction of macro-events. We have seen various examples that show that in a specific language, the users, when choosing a construction, have more than one option among the identified typological patterns.

8.  Conclusion and some perspectives In this article I have discussed Talmy’s typological distinction in cognitive semantics between verb-framed and satellite-framed languages. The focus has been on the theoretical underpinnings of the typology. On the one hand, the typology seems to qualify as a general theory of expressions of macro-events that goes beyond the mere study of motion events. On the other hand, many studies have shown that there are substantial deviations from the basic typological patterns, which were originally suggested in Talmy’s work. This study suggests that a generalized version of the typology, originally proposed by Talmy, should include both the lexical level and more schematic construc-

.  Similar observations have been made in Masini (2005), which provides a synchronic and diachronic analysis of verb-particle constructions in Italian, and in Bernini et al. (2006) (from an L2-acquisition perspective).



The construction of macro-events: A typological perspective 

tional levels of analysis. Constructions of different degree of specificity (schematic and lexical constructions) should be the basic constituents of the typology. From this point of view, it is argued that Germanic languages tend to map the main information of expressions of macro-events onto a complex schematic construction and the secondary information onto a lexical (verbal) construction. Romance languages, particularly Spanish, tend to map the main information onto the verb, i.e. a lexical construction, while the secondary information may be mapped onto a complex schematic construction. In this revised version, the typology is thus not merely a matter of lexicalization patterns, as it is in Talmy’s work. It is about constructional patterns, the internal structure of constructions, and patterns of combined constructions of varying specificity in different language types. It is further suggested that the generality of the typological patterns is due to the ontology of the typology: it is only indirectly a typology of conceptualization patterns. It is basically an information structure phenomenon. Macro-event constructions, i.e. constructions of the main information (MIC) and the supportive information (SIC), are the basic constituents of the typology. This framework has proved adequate to identify patterns in data that are very similar to those recognized in Talmy’s work, yet not recognized as part of his typology. It is hypothesized that pairs of MIC/SIC are distilled out of usage due to the user’s constant generalizations from usage. They are procedural devices for organizing the clausal information. Their existence in the grammar is reflected in usage as an information structure phenomenon, which has obvious stylistic consequences when translating a text from one language to another. MIC/SIC cannot be synchronically derived from complex expressions of conceptual content. They have their own, independent, procedural role in grammar, as a device for organizing the information. The hypothesis of this article is that the typology is anchored in this task. Thereby, the typology is better suited for analyzing actual usage data since information structure is highly sensitive to the user’s, including translator’s, individual choice and strategies in performance. For instance, the framework has proved to be well designed to account for the substantial deviation from basic typological patterns in the Italian data that has been observed in this study, and in other studies. In general, the present study confirms what has been pointed out by several scholars (on the basis of much larger amounts of data), that even though there are important typological differences between e.g. Germanic and Romance languages in expressions of macro-events, there is no simple clear cut distinction. In this article, the interpretation of Talmy’s descriptive typology is that some MIC/ SIC are more entrenched in the grammar of some languages than in others. This is the essence of the typology in the present framework. A survey of the inventory, and frequency, of different types of MIC/SIC is thus an interesting research question for larger scale research projects on clausal typology. Such studies require a large, advanced multilingual parallel corpus.

 Johan Pedersen

Finally, we may speculate whether the observed typological patterns reflect a general clausal typology. We may ask whether some languages, e.g. English, tend to organize the principal clausal information by means of complex, schematic constructions, complementing this information lexically, whereas other languages, e.g. Spanish, tend to organize the principal clausal information lexically around the verb and its valence structure, complementing the principal information by means of more schematic constructions.

References Andersen, Hans C. 2005. The Snowman. Six different versions. The Mulinco project, University of Copenhagen. Aske, Jon. 1989. Path predicates in English and Spanish: A closer look. In Proceedings of the fifteenth annual meeting of the Berkeley Linguistics Society, Kira Hall, Michael Meacham & Richard Shapiro (Eds), 1–14. Berkeley CA: Berkeley Linguistics Society. Berman, Ruth A. & Dan, I. Slobin. 1994. Relating events in narrative: A crosslinguistic developmental study. Hillsdale NJ: Lawrence Erlbaum. Bernini, Giuliano, Lorenzo Spreafico & Ada Valentini. 2006. Acquiring motion events in a second language: The case of Italian L2. Linguistica e Filologia 23: 7–26. Boas, Hans. 2003. A constructional approach to resultatives. Stanford CA: CSLI. Bohnemeyer, Jürgen et al. 2007. Principles of event segmentation in language: The case of motion events. Language 83(3): 495–532. Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam: John Benjamins. Comrie, Bernard. 1976. Aspect. Cambridge: CUP. Croft, William. 2000. Explaining language change. London: Longman. Croft, William. 2001. Radical Construction Grammar. Oxford: OUP. Croft, William et al. 2008. Revising Talmy’s typological classification of complex events. Ms. (http:// www.unm.edu/~wcroft/WACpubs.html) Gennari, Silvia P. et al. 2002. Motion events in language and cognition. Cognition 83: 49–79. Goldberg, Adele E. 1995. Constructions. A construction grammar approach to argument structure. Chicago, IL: University of Chicago Press. Goldberg, Adele E. 2006. Constructions at work. The nature of generalization in language. Oxford: OUP. Grimshaw, Jane. 1990. Argument structure. Cambridge MA: The MIT Press. Ibarretxe-Antuñano, Iraide. 2004a. Motion events in Basque narratives. In Relating events in narrative: Typological and contextual perspectives, Sven Strömqvist & Ludo Verhoeven (Eds), 89–111. Mahwah NJ: Lawrence Erlbaum Associates. Ibarretxe-Antuñano, Iraide. 2004b. Dicotomías frente a continuos en la lexicalización de los eventos de movimiento. Revista Española de Lingüística 34(2): 481–510. Jurafsky, Dan. 1996. A probabilistic model of lexical and syntactic access and disambiguation. Cognitive Science 20: 137–94. Lambrecht, Knud. 1994. Information structure and sentence form. Cambridge: CUP. Langacker, Ronald W. 1987/91. Foundations of Cognitive Grammar, Vol. I + II.  Stanford CA: Stanford University Press.



The construction of macro-events: A typological perspective 

Langacker, Ronald W. 1988. A usage-based model. In Topics in cognitive linguistics, Brygida Rudzka-Ostyn (Ed.), 127–161. Amsterdam: John Benjamins. Levin, Beth & Malka Rappaport Hovav. 1995. Unaccusativity: At the syntax-lexical semantics interface. Cambridge MA: The MIT Press. Masini, Francesca. 2005. Multi-word expressions between syntax and the lexicon: The case of Italian verb-particle constructions. SKY Journal of Linguistics 18: 145–173. Mendívil Giró, José L. 2003. Construcciones resultativas y gramática universal. Revista Española de Lingüística 33(1): 1–28. Pedersen, Johan. In press. Lexical and constructional organization of argument structure: A contrastive analysis. In Studies in language and cognition, Jordan Zlatev et al. (Eds), Cambridge: Cambridge Scholars Publishing. Pinker, Steven. 1989. Learnability and cognition: The acquisition of argument structure. Cambridge MA: The MIT Press. Rojo, Ana & Javier Valenzuela. 2005. Verbs of sensory perception in English and Spanish. Languages in Contrast 5(2): 219–243. Rudzka-Ostyn, Brygida. 2003. Word power: Phrasal verbs and compounds. A cognitive approach. Berlin: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. In press. Levels of description and constraining factors in meaning construction: An introduction to the Lexical Constructional Model. Folia Linguistica 42(2). Sinha, Chris & Tania Kuteva. 1995. Distributed spatial semantics. Nordic Journal of Linguistics 18: 167–199. Slobin, Dan I. 1996. Two ways to travel: Verbs of motion in English and Spanish. In Grammatical constructions: Their form and meaning, Masayoshi Shibatani & Sandra A. Thompson (Eds), 195–219. Oxford: OUP. Slobin, Dan I. 1997. Mind, code and text. In Essays on language function and language type, Joan Bybee, John Haiman & Sandra Thompson (Eds), 437–68. Amsterdam: John Benjamins. Slobin, Dan I. 2000. Verbalized events: A dynamic approach to linguistic relativity and determinism. In Evidence for linguistic relativity, Suzanne Niemeier & René Dirven (Eds), 107– 138. Amsterdam: John Benjamins. Slobin, Dan I. 2004. The many ways to search for a frog: Linguistic typology and the expression of motion events. In Relating events in narrative: Typological perspectives, Sven Strömquist & Ludo Verhoeven (Eds), 219–257. Mahwah NJ: Lawrence Erlbaum Associates. Slobin, Dan I. Forthcoming. Relations between paths of motion and paths of vision: A crosslinguistic and developmental exploration. Title to be announced, Virginia Gathercole (Ed.), (http://ihd.berkeley.edu/slobinpapers.htm) Slobin, Dan I. & Nini Hoiting. 1994. Reference to movement in spoken and signed languages: Typological considerations. In Proceedings of the twentieth annual meeting of the Berkeley Linguistic Society, Susanne Gahl, Andy Dolbey & Christopher Johnson (Eds), 487–505. Berkeley CA: Berkeley Linguistics Society. Snyder, William. 2001. On the nature of syntactic variation: Evidence from complex predicates and complex word-formation. Language 77(2), 324–342. Talmy, Leonard. 1978. Figure and ground in complex sentences. In Universals of human language, Vol.  4, Joseph Greenberg et al. (Eds), 627–49. Stanford CA: Stanford University Press. Talmy, Leonard. 1985. Lexicalization patterns: Semantic structure in lexical forms. In Language typology and syntactic description, Vol. 3: Grammatical categories and the lexicon, Timothy Shopen (Ed.), 57–149. Cambridge: CUP.

 Johan Pedersen Talmy, Leonard. 1987. Lexicalization patterns: typologies and universals [Berkeley Cognitive Science Report 47]. Berkeley CA: Cognitive Science Program, University of California. Talmy, Leonard. 1991. Path to realization: A typology of event conflation. In Proceedings of the Seventeenth Annual Berkeley Linguistics Society, Laurel A Sutton, Christopher Johnson & Ruth Shields (Eds), 480–519. Berkeley CA: Berkeley Linguistics Society. Talmy, Leonard. 2000. Toward a cognitive semantics. Vol.1 & 2. Cambridge MA: The MIT Press. Talmy, Leonard. 2005. Written interview on my work conducted by Iraide Ibarretxe: Part 1. Annual Review of Cognitive Linguistics 3: 325–347. Amsterdam: John Benjamins. Zlatev, Jordan & Peerapat Yangklang. 2004. A third way to travel: The place of Thai (and other serial verb languages) in motion event typology. In Relating events in narrative: Typological and contextual perspectives, Sven Stromqvist & Ludo Verhoeven (Eds), 159–190. Mahwah NJ: Lawrence Erlbaum Associates.

Constructions, co-composition and merge* Beatriz Martínez Fernández This article follows the line of research proposed in this volume by Mairal and Ruiz de Mendoza, Cortés Rodríguez, and Martín Arista, combining functional and constructional models of language. I discuss some interesting examples of break verbs with argument-adjuncts of motion which, being syntactically similar to Goldberg’s caused-motion construction, do not meet the definition for constructions devised by Goldberg. These structures are characterised by acquiring the semantics of motion without losing the semantics of change of state. Therefore, I have labelled them merge structures. Since Construction Grammar focuses mainly on constructions, I resort to Pustejovsky’s Generative Lexicon (henceforth GL) in search of a mechanism that allows me to explain this type of structure, because it is purposely designed to deal with creative uses of language and, more precisely, with polysemy. In a sense, it is not that different from Goldberg’s constructions, but it has the advantage that using the same system of lexical representation that is used for cocomposition, it can also explain merge. The comparison of the two representations, co-composition and merge, indicates that the main differences can be explained by putting some of the semantic weight on the event and qualia structures. Although these representations capture the main differences existing between merge and co-composition, this argument only works at the level of interpretation or comprehension, a problem that was already pointed out by Goldberg (1995, 2006). Therefore, this article has also addressed the question of how to account for the production of those structures. Van Valin (in press) foresees the possibility of reconciling projectionist accounts like Role and Reference Grammar (RRG) and constructionist approaches like Pustejovsky’s GL and argues that the interpretive role of co-composition is helpful in the linking from syntax to semantics in RRG. Thus, when presented with a certain syntactic structure, the choice between a merge or a co-compositional reading will be determined by their lexical representations.

1.  Introduction In this article, I discuss some interesting examples of break verbs with argument-adjuncts of motion. Despite their syntactic resemblance to the caused-motion construction, these

*This research has been funded through the project HUM2005–07651-C02–02/FILO. I should like to thank Javier Martín Arista and three anonymous reviewers for their comments on an earlier draft of this article. The usual disclaimers apply.

 Beatriz Martínez Fernández

examples do not comply fully with the semantic behaviour assigned to these constructions by traditional construction grammars. For this reason, they have not received enough attention. After detecting and explaining the differences existing between them and the traditional constructions, I turn to Pustejovsky’s Generative Lexicon (GL), based on a solid system of qualia and some generative operations like co-composition, to account for the new constructs, which I label merge structures. This analysis is compatible with the Role and Reference Grammar (RRG) linking algorithms from syntax to semantics and vice versa, which I then use to integrate the more descriptive mechanism of co-composition into a theory that also accounts for the productivity of new structures. The article proceeds as follows: first I describe break verbs lexically; then I bring forward some examples which illustrate the caused-motion construction as described in Goldberg (1995, 2006), and some other examples which are syntactically alike but which, I argue, are not constructions. Second, I explain the difference between constructions (or co-composition) and merge. Third, I discuss how two constructional approaches, Goldberg’s Construction Grammar (CG) and Pustejovsky’s Generative Lexicon, account for structures of this type. Fourth, I opt for an approach to merge and co-composition that combines a projectionist approach like that of Role and Reference Grammar and a constructional approach like GL, along the lines of Van Valin (in press). Finally, I present the conclusions in section 5. This investigation follows the line of research delineated by Mairal and Ruiz de Mendoza (this volume), Cortés Rodríguez (this volume), and Martín Arista (this volume; forthcoming), which combines functional and constructional models of language.

2.  The structural pattern of break verbs The examples I bring up for discussion here are characterised in the first place by denoting a change of state or, to be precise, a pure change of state. I take this notion from Levin (1993: 165), who distinguishes between verbs of pure change of state and those that only denote change of state. Both are found in the middle and resultative alternations and both refer to actions that bring about a change in material integrity, but the latter do not turn up in the causative/inchoative alternation. I illustrate the causative/inchoative alternation in (1).

(1) a. I broke the twig off (of) the branch/The twig broke off (of) the branch.  (Levin 1993: 166)

b. Carla slid the books across the table/The books slid across the table.  (Levin 1993: 133)



Constructions, co-composition and merge 

The contrast between pure and non-pure verbs of change of state is exemplified by verbs like break and cut, respectively. I leave aside cut verbs and focus exclusively on break verbs. This group comprises verbs like break, crack, crash, crush, fracture, rip, shatter, smash, snap, splinter, split and tear.1 Break verbs are characterised, like many other verbs, by an argument structure that expands beyond that of their basic predicates, as illustrated in (2).

(2) 〈FA5 1938〉2 He saw his hand come up and smash itself into Guy’s amiable face.

Despite being built around break verbs, some of these extended expressions do not denote a change of state. Instead, their meaning derives into a new one like, for instance, caused-motion:

(3) Sam tore a piece off the block. (Goldberg 1995: 154)

Goldberg (1995, 2006) refers to sentences such as (3) as caused-motion constructions. These constructions have been widely discussed and accurately defined. However, little attention has been paid to other examples which, being syntactically alike, still preserve the sense of change of state and are therefore not constructions, because the meaning of the whole is derivable from that of the parts. I illustrate the two cases in (4).

(4) a. 〈C8E 73〉 Saturday afternoons, when the streets were solid with white faces, was a carnival of consumerism as goods were ripped from shelves.

b. 〈FA5 1938〉 He saw his hand come up and smash itself into Guy’s amiable face. c. 〈ANF 594〉 When Marevna finally went home at dawn: “Max Jacob stood with a missal in his hand and Modigliani was still on his feet methodically tearing the rest of the wallpaper off the walls and singing an Italian song.” d. 〈APC 1457〉 On his back like a beetle, three sat on him while two ripped his daysack from his chest. e. 〈CA1 106〉 The kite might start to climb, then perform a curving sweep up to about 45 degrees, execute a beautiful turn and point head downwards until it crashes into the ground. f. 〈CK4 1970〉 LETHAL INCENDIARY grade performance on and off stage (Keith Moon brings chat show host Russell to the edge of a Harty attack, terrorist Townshend smashes guitar through a club roof).

.  Following Lemmens (1998: 37), I leave chip out of this group because it should be considered a cut verb. .  All examples introduced by a series number in angled brackets have been taken from the British National Corpus.

 Beatriz Martínez Fernández

I explain the notion of construction in more depth in section 3, but for the time being, it is useful to compare the examples to see where the main differences between constructions and non-constructions lie. In example (4a), the goods are not necessarily broken when ripped from the shelves. Rip just indicates the violent way in which the change of position has been carried out. Hence, (4a) illustrates a construction because its semantics is not compositionally derivable from the lexical items that instantiate it; more precisely, it is an example of the caused-motion construction. In example (4b), the hand is not broken; that is, it does not experience a change of state. All it does is land on Guy’s amiable face; in short, it undergoes a change of position, and, thus, it is also a construction. Examples (4c), (4d), and (4f) may at first sight look like constructions. However, they preserve the meaning of change of state. Thus, in (4c) the wallpaper undergoes a change of state (it is being ripped; that is, broken) and then a change of position (it is being removed from the wall). In (4d), the most logical interpretation seems to be that the daysack (or at least its strap) will be broken when ripped. Otherwise, the verb rip, which indicates some violence in the performance of the action, would not be appropriate in this context. (4e) is more open to interpretation, because the kite may or may not have broken when crashing into the ground. In short, these examples add new semantics to the original semantics of break verbs. They are a merge of change of state + change of position. My claim is that these examples must also be accounted for by a theory of syntax. Construction grammar focuses on constructions, and hence it does not explain examples like these. Pustejovsky’s notion of co-composition covers more or less the same ground. However, here I argue that a combination of RRG’s syntax-semantics interface and Pustejovsky’s theory of co-composition through qualia can explain examples like these while accounting for structures like the caused-motion construction. In the next section, I explain the different approaches to constructions and cocomposition and how to approach the new structures.

3.  Constructions, co-compositon, and merge In this section, I introduce Goldberg’s constructions and discuss why the examples mentioned above are not constructions. Having done that, I suggest a name for these new structures and turn to Pustejovsky for an alternative approach to the new structures. As mentioned above, break verbs participate in what Goldberg (1995: 152–179) calls the English Caused-Motion Construction. This construction is semantically characterised by a causer argument which directly causes the theme argument to move to a new location which is designated by the directional phrase, and it covers the following types of expressions:





Constructions, co-composition and merge 

(5) Expressions covered by the English Caused-Motion Construction (Goldberg 1995: 152):



a. b. c. d. e. f.

They laughed the poor guy out of the room.  rank sneezed the tissue off the table. F Mary urged Bill into the house. Sue let the water out of the bathtub. Sam helped him into the car. They sprayed the paint onto the wall.

Goldberg tries to justify the existence of this construction by showing that its semantics is not compositionally derived from other constructions existing in the grammar, such as the lexical items that instantiate it.3 In other words, the combination of the lexical items that instantiate the construction must not lead to the interpretation of the construction (Goldberg 1995: 153). Focusing on the English Caused-Motion Construction in particular, Goldberg (1995: 153) explains that some of these verbs do not encode the caused-motion semantics when used in isolation: in some cases they do not encode causation, as illustrated by kick in (6a); in others they do not encode motion, as shown in (6b): (6) Examples taken from Goldberg (1995: 153) a. Joe kicked the wall. b. Frank squeezed the ball.

Nonetheless, when used in the construction, these verbs encode both causation and motion: (7) Examples taken from Goldberg (1995: 153) a. Joe kicked the dog into the bathroom.(He caused the dog to move into the bathroom.) b. Frank squeezed the ball through the crack. (The ball necessarily moves.)

What is more, many transitive verbs which participate in this construction do not bear the same semantic relation to their direct object as they do in simple transitive sentences. Goldberg (1995: 153–154) illustrates this idea by claiming that (8) does not entail (9):

(8)  Sam tore a piece off the block.



(9)  Sam tore a piece.

.  Constructions also include fully predictable patterns as long as they occur with sufficient frequency (Goldberg 2006:  5). Nonetheless, this statement mainly refers to highly common patterns such as interrogatives, relatives, passives, and so on.

 Beatriz Martínez Fernández

The construction in (8) does not encode a change of state, but only a change of position – plus the manner specification in which the change of position took place. Besides the basic sense posited above, the caused-motion construction is associated with a number of related senses such as X enables Y to move Z, illustrated in (10a), X prevents Y from moving Comp(Z), illustrated in (10b), X helps Y to move Z, illustrated in (10c), and X causes Y to move Z, illustrated in (10d). In this sense, the construction shows a certain degree of polysemy. (10) Examples taken from Goldberg (1995: 161–162) a. sam allowed Bob out of the room. b. Sam barricaded him out of the room. c. Sam helped him into the car. d. Sam invited him out of her cabin.

The latter sense of the caused-motion construction, illustrated in (10d), also includes examples like Frank sneezed the tissue off the nightstand, the difference being that (10d) involves a force-dynamic verb (Talmy 1985) that encodes a communicative act and, consequently, the motion is not strictly entailed (that is, the fact that Sam invites someone out of the house does not necessarily entail that the person actually leaves the house). Frank sneezed the tissue off the nightstand, by contrast, entails motion. Although the construction is claimed to exist independently of the verbs which instantiate it, there seem to be some idiosyncrasies which can only be accounted for through specific semantic constraints such as direct causation or one event per clause. Constraints on what kinds of situations can be encoded by the caused-motion construction are further specified and, with regard to change of state verbs in particular, Goldberg claims that there is some incidental motion implied in the action denoted by the verb and only if the motion is intended are the examples acceptable. Summing up these constraints, Goldberg (1995: 174) writes: if the verb is a change-of-state verb (or a verb of effect), such that the activity causing the change of state (or effect), when performed in a conventional way, effects some incidental motion and, moreover, is performed with the intention of causing the motion, the path of motion may be specified.

Despite the depth of the treatment of constructions by Goldberg, none of the above helps explain the examples I have brought up for discussion here. Goldberg argues against the idea that the semantics of the caused-motion construction can be derived from the inherent semantics of the verb, the semantics of the preposition, or the semantics of the combination of verb plus preposition (Goldberg 1995: 154). While this approach works for examples such as Sam tore a piece off the block which, as explained above, does not necessarily entail Sam tore a piece, there are other examples where the verb does not change its inherent semantics. Consider the following examples:



Constructions, co-composition and merge 

(11) a. 〈A0R 2161〉 Rodney cracked two eggs into the frying pan. b. 〈AR9 823〉 All the floorboards had been removed, and the ceilings and plasterwork had crashed down into the basement with the weight of water pouring through the roof. c. 〈ANF 594〉 When Marevna finally went home at dawn: “Max Jacob stood with a missal in his hand and Modigliani was still on his feet methodically tearing the rest of the wallpaper off the walls and singing an Italian song.”

Although examples (11a) and (11b) are open to more than one interpretation, the most straightforward one entails that Rodney cracked two eggs and that the plasterwork crashed. Likewise, (11c) may be interpreted as entailing that Modigliani methodically tore the rest of the wallpaper, because wallpaper is stuck to the wall and, consequently, it tends to break while being removed from it. Therefore, part of the semantics of the sentence is derived from the combination of the verb plus the semantics of the preposition. Independently of which interpretation is most appropriate, what these examples point out is that the same syntactic form may have more than one interpretation, and Goldberg’s model only explains those which conform to the idea of construction. In other words, although Goldberg’s CG satisfactorily accounts for expressions with an idiosyncratic semantics (Croft 2001), her model does not account for this kind of polysemy where one syntactic form gives rise to different meanings. CG explains constructional polysemy (Kay 2005:  73), but not polysemy within one verbal class as shown here.4 Furthermore, Goldberg “does not place any limits on the range of possible individual links that explain metaphorical senses through polysemy, perhaps considering the inventory of individual links unbounded (…)” (Kay 2005: 75). The problem of the proliferation of constraints is recurrent in the latest versions of the model; Goldberg (2006) revises some principles put forward in previous works, such as the Argument Realization Principle which, in her own words, “cannot be correct as it stands” (Goldberg 2006: 24), but this version still contains many constraints. These problems are partly overcome in Pustejovsky’s (1991, 1992, 1995, 1998, 2000) theory of the Generative Lexicon. Pustejovsky (1991) claims that the meaning of the whole sentence can be compositionally derived from the combination of the meanings of the constituent parts, thus avoiding positing additional verb senses. By co-composition must be understood “a ‘bilateral semantic selection’ between the verb and its complement, giving rise to a novel sense of the verb in each context of use” (Pustejovsky 2000: 8).

.  Kay (2005: 73) defines constructional polysemy as follows: “Each sense of the construction corresponds to a distinct set of entailments along the lines discussed above. Each sense of the construction also combines with verbs of a distinct semantic class or set of classes”.

 Beatriz Martínez Fernández

When a verb and a prepositional phrase (PP) coocur in the syntax, an interpretive process, co-composition, combines their representations to yield the active accomplishment interpretation, as in float into the cave. The resulting expression is not derived from the lexicon, as in projectionist approaches; rather, the new interpretation is the result of the interaction of the semantics of the verb with the semantics of the PP itself, both of which are available in the lexical representations of the generative lexicon. In the case of break verbs and the causative/inchoative alternation that characterises them, Pustejovsky adopts a projectionist approach and assumes that, instead of having two lexical entries, they have a single, underspecified lexical representation. The event structure of crack, for example, has two subevents, a process and a result state, and either may be selected as the ‘head’ of the structure. If the process is taken as the head, the sentence is causative, as in Mary cracked the vase; if the result state is the head, it is non-causative, as in the vase cracked. The idea behind the label event headedness is that subevents are ordered not only temporally, but also according to their prominence. This is represented in the event structure by means of the marker e*: (12) [eσe1* transitive verb) kawantjingaNi ‘cause to lose one’s way’ (intransitive verb > transitive verb) kuTintjingaNi ‘turn over’ (intransitive verb > transitive verb) mayaltjingaNi ‘make give off a rebound noise sound’ (intransitive verb > transitive verb) munturtjingaNi ‘cause to give out a rhythmic running noise’ (noun > transitive verb) puultjingaNi ‘raise dust, make dust come up’ (adjective > transitive verb) puyuntjingaNi ‘make smoke’ (noun > transitive verb) raartjingaNi ‘make rattle’ (noun > transitive verb) takaltjingaNi ‘make give out a knocking or chopping sound’ (noun > transitive verb) tuulytjingaNi ‘cause to give out a thumping sound’ (noun > transitive verb) ulatjingani ‘cause to cry’ (intransitive verb > transitive verb) wangkatjingaNi ‘bring someone to the point where they must speak up’ (noun > transitive verb) waRaratjingaNi ‘bounce or toss (baby) in the air’ (noun > transitive verb) punkaltjingaNi ‘drop, make fall over, fell’ (intransitive verb > transitive verb) taatjingaNi ‘make something burst open, tear open’ (intransitive verb > transitive verb) tjirkaltjingaNi ‘make dodge, move out of the way’ (intransitive verb > transitive verb) c.

Zero-derivation of transitive verbs amiRinanyi ‘lead by the arm (a child)’ (noun > transitive verb) ampuNi ‘hold someone in your arms’ (noun > transitive verb)



A typology of morphological constructions 



aNangunanyi ‘bring up, raise’ (noun > transitive verb) iNini ‘loosen’ (adjective > transitive verb) itjilinanyi ‘cut meat into pieces’ (noun > transitive verb) iwarananyi ‘make a track or road’ (noun > transitive verb) katuNi ‘lift, raise’ (adverb > transitive verb) kuTi-kuTini ‘turn over and over’ (active adjective > transitive verb) uRini ‘brush clear, sweep, clear’ (noun > transitive verb) waRuly-waRulyinanyi ‘make steaming hot’ (adjective > transitive verb)

d.

Reduplication arkai-arkai ‘difficult to see, faint’ (adjective > adjective) ilkaRi-ilkaRi ‘rolling of eyes’ (noun > adjective) iTi-iTini ‘treat as if a baby, spoil, child’ (noun > transitive verb) takal-takal(pa) ‘repeated knocking, chopping’ (noun > noun) waRara-waRara ‘by hopping’ (noun > adverb)

Word-formation is often recursive in P/Y, as can be seen in (8): (8) waRu ‘fire’ (basic) waRu-waRu ‘pushily, roughly, too directly’ (reduplication) waRuly-waRulyi ‘steaming hot’ (active adjective > adjective) waRuly-waRulyinanyi ‘make steaming hot’ (adjective > transitive verb)

Relators, as in the formation of adjectives from nouns by means of -tja, take part in very regular morphological processes, which strongly resemble inflection in the sense of semantic transparency and generalization in a given category: (9) ilkaRitja ‘of, from the sky, heaven’ (noun > adjective) kungkatja ‘of or relating to women’ (noun > adjective) lirutja ‘associated with snakes’ (noun > adjective) mamutja ‘of or from spirit-monsters’ (noun > adjective) nguratja ‘in or belonging to a camp or place’ (noun > adjective) piriyitja ‘associated with the spring windy season’ (noun > adjective) puTitja ‘associated with the bush’ (noun > adjective) tjuwatja ‘associated with the shop’ (noun > adjective) unngutja ‘from or associated with the inside’ (noun > adjective) walypalitja ‘from or brought by white people’ (noun > adjective)

Given the general characterization of P/Y morphology just provided, the next step in this discussion is the application of the constructional schemas identified in the previous section to this Australian language. To begin with, the endocentric and the exocentric morphological constructions proposed on language-general grounds appear in P/Y, in instances like those represented by figures 16 and 17 respectively:

 Javier Martín Arista COMPLEX WORDN COREN NUCN

ARGN COREN NUCN

PREDN wangka

PREDN La

Figure 16.  Endocentric morphological construction: wangkaLa ‘chatterbox, neverending talker’. COMPLEX WORDN COREN NUCN

ARGN COREN NUCN

PREDN paul

PREDN pi

Figure 17.  Exocentric morphological construction: paulpi ‘a cook’.

The instantiations of the endocentric and the exocentric morphological constructions in P/Y are arranged in a language-specific morphological template that can be tentatively formulated as in Figure 18: [PREFIELDα] NUCLEUS (P-N SLOT) [POSTFIELDα]..[POSTFIELDΩ] Figure 18.  A morphological template for P/Y.

The morphological template proposed for P/Y accounts for the fact that this language is mainly suffixal by establishing an open number of Postfield semantic positions and a single Prefield position. This is also consistent with the fact that directionals are mutually incompatible and do not occur in combination in the



A typology of morphological constructions 

verbal Prefield. A formal position has been defined to the right of the Nucleus, the P-N (Post-Nuclear) Slot, in order to allow for the insertion of the duplicated element in reduplication, as in aLa-aLa ‘feeling expectant’ (base aLa ‘view’) or ilytji-ilytji ‘not fully mashed up’ (base ilytji ‘bush’). This element takes up a special position defined in structural terms with respect to the Nucleus. The fact that this position is post-nuclear rather than pre-nuclear follows from the general orientation of the language towards the Postfield. Further derivation or inflection of reduplicated predicates is carried out by free or bound predicates that occupy Prefield or, mainly, Postfield semantic positions, as in the intracategorial derivations in (10a) and the intercategorial derivations in (10b): (10) a. Intracategorial derivation of reduplicated predicates arkai-arkai ‘difficult to see, faint’ (adjective > adjective) iTi-iTi ‘girls’ game pretending to have babies’ (noun > noun) laka-lakaNi ‘clatter’ (transitive verb > transitive verb) nguLu-nguLu ‘anxiously’ (active adjective > active adjective) taa-taaNi ‘pop, crack or burst open’ (intransitive verb > intransitive verb) b. Intercategorial derivation of reduplicated predicates akuRi-akuRirinyi ‘sweat’ (noun > intransitive verb) aRuTu-aRuTu ‘moving around aimlessly’ (transitive verb > adjective) ilkaRi-ilkaRi ‘rolling of eyes’ (noun > adjective) iluRu-iluRurinyi ‘feel miserable’ (adjective > intransitive verb) itaRitaRi ‘dragging on the ground’ (adjective > adverb)

Considering the constructional schemas proposed in the previous section as language-specific instantiations of the language-general endocentric and exocentric morphological constructions, P/Y shows continuous constructional schemas such as the augmented stem in nominalizations (Goddard 1993:  28) given in (11a) and discontinuous constructional schemas, as in the derivation of previously reduplicated bases like the one provided by (11b): (11) a. Augmented stem tju-nanyi ‘put’~tju-nku-nytja ‘putting’ pu-nganyi ‘hit’~pu-ngku-nytja ‘hitting’ b. Reduplication waRu > waRu-waRu > waRuly-waRulyi ‘steaming hot’

The augmentation of stem accommodates to the template devised in Figure 10, while the suffixation of a reduplicated predicate that takes place in waRu-waRu〉waRulywaRulyi requires the template given in Figure 11 with the addition of the Post-Nuclear slot as rendered in Figure 18. This is shown, respectively, by figures 19 and 20. It must be borne in mind that the suffix –ly performs a single function in spite of the fact that

 Javier Martín Arista

it occupies two positions. This is represented by the semantic feature α that turns up to the left and to the right of the Post-Nuclear Slot: NUCLEUS

Tju

-nku

[POSTFIELDα]

-nytja

‘putting’

Figure 19.  Augmented stem as a continuous constructional schema: Tjunkunytja ‘putting’. NUC [POSTFIELDα]

waRu

(P-N SLOT) [POSTFIELDα] [POSTFIELDβ]

ly

waRu

ly

i

Figure 20.  Suffixation of reduplicated predicate as a discontinuous constructional schema: waRuly-waRulyi ‘steaming hot’.

The continuous constructional schema in Figure 19 represents an instantiation of the exocentric construction because the nominal feature is provided by the nonnuclear element whereas the discontinuous constructional schema in Figure 20 constitutes an instantiation of the endocentric construction because there is no percolation of morphological features in the derivation of waRuly-waRulyi ‘steaming hot’ (adjective) from waRu-waRu (adjective). Turning to the analytic vs. synthetic constructional schema, compound case markers as defined by Schweiger (2000: 279) represent the analytic constructional schema. Indeed, in P/Y the allative case ending kutu, as in (12a), and the ablative case ending nguRu, as in (12b), require the locative case la (Goddard 1993: 19): (12) a. Allative on locative case Wataru -la -kutu Wataru -loc-all ‘to Wataru’ b. Ablative on locative case Wataru -la -nguRu Wataru -loc-abl ‘from Wataru’

Compound case markers resulting in a morphological constructional schema of the analytic type constitute instantiations of the endocentric morphological construction. Actually, inflection is always endocentric since it does not allow for



A typology of morphological constructions 

category change and therefore no percolation of morphological features can take place. The analytic constructional schema has been represented in Figure 13 as a template of the type NUCLEUS [POSTFIELDβ][POSTFIELDβ], into which the semantic features expressed by the allative (or ablative) and locative case of the previous examples can be inserted, as is displayed by Figure 21: NUCLEUS [POSTFIELDα][POSTFIELDβ]

Wataru

-la

-kutu

Figure 21.  Double case as an analytic constructional schema: Wataru-la-kutu ‘from Wataru’.

Regarding the synthetic morphological construction, the P/Y suffix kira/Rara/ ra, which derives nouns from nouns thus belonging in the endocentric construction, expresses the inflectional feature of number as well as the derivational feature of kinship: (13) katjaRara ‘a person together with their son’ (noun > noun) kuriRara ‘a person together with their spouse’ (noun > noun) kamuRura ‘a person together with their mother’s brother’ (noun > noun) kangkuRura ‘a person together with their senior sister’ (noun > noun) kuNTilira ‘a person together with their father’s sister’ (noun > noun) kuTara ‘a person together with their senior brother’ (noun > noun) mamaRara ‘a person together with their father’ (noun > noun) mayatjara ‘a person together with their boss or supervisor’ (noun > noun) maLanykira ‘a person together with their junior brother’ (noun > noun) uNTalkira ‘a person together with their daughter or niece’ (noun > noun)

The suffixation by means of -Rara represents an instance of the synthetic constructional schema, for which the template NUCLEUS [POSTFIELDαβ] has been devised in Figure 12. -Rara-derivatives give rise to synthetic constructional schemas corresponding to the endocentric morphological construction. This is illustrated by Figure 22: NUCLEUS [POSTFIELDαβ]

Kuri

Rara

‘a person together with their spouse’

Figure 22.  Rara-suffixation as a synthetic constructional schema, KuriRara ‘a person together with their spouse’.

 Javier Martín Arista

The nouns in example (13) result from non-recursive morphological processes. On the other hand, the compound case markers in (12) could count as a recursive morphological constructional schema. Figure 23 also involves recursivity in the morphology of P/Y, although this is in the line of derivation rather than inflection, as is the case with (12). The template and the tree of this instantiation of the endocentric construction follow: [[NUCLEUS (P-N SLOT)] PREFIELDα] NUCLEUS [POSTFIELDΩ] COMPLEX WORDV COREV ARGN

NUCV

WORDN COREN NUCN PREDN

PREDv

COMPLEX WORDN takal-takal

mananyi

Figure 23.  Takal-takalmananyi ‘give out a repeated knocking’.

5.  Conclusion: The role of morphological constructions Lexicalism, following the first proposal made by Chomsky (1970), has opted for defining word-formation rules which account for morphological processes that produce complex words in a pre-syntactic lexicon. In the LSW the structural and functional similarites between morphology and syntax are stressed by generalizing to morphology the descriptive and explanatory principles of syntax. Additionally, a concept of construction shared by units larger and smaller than a phrase and motivated functionally finds its way into the theory, which is in tune with the more central role that



A typology of morphological constructions 

constructions play in the latest version of RRG. Morphological constructions, as far as they are based on the distribution of markedness with resulting asymmetry, account for the maximization of the meaningful devices of the language that morphological processes require. Morphological constructions also contribute to distinguishing morphological processes in a principled way: derivation (including compounding and affixation) can be endocentric or exocentric, whereas inflection is endocentric. Even though the language-general distinction between endocentric and exocentric morphological constructions has turned out more distinctive for morphological processes than the typology of constructional schemas, the latter also yield significant generalizations. Derivation and inflection can be synthetic or analytic, as well as continuous or discontinuous. Recursivity appears as a derivational phenomenon, but the instances of double case in P/Y probably count as recursive inflection. Conversely, synthesis seems to be more widespread in inflection than in derivation, but the synthetic constructional schema of P/Y derivational morphology demonstrates, at least, that synthesis in derivation or even at the boundary between derivation and inflection, deserves more attention. On the theoretical side, the discussion of P/Y has shown that the interaction of Word and Phrase morphology entailed by the existence of phrases in which the last word regularly carries the inflectional ending for case, as happens in P/Y, calls for a model that allows for interaction between Word and Phrase morphology. The Complex Word layer of the LSW represents the boundary between relational and nonrelational Word morphology, in the sense that it delimits the inheritance of relational morphological features and the percolation of non-relational morphological features. Also of theoretical interest is the question of lexical integrity. Lexical integrity, whereby derivation is prior to inflection, is hard to support on a language-general basis, as is demonstrated by Toman (1998), Rood (2005) and Martín Arista (fc.), among others. The question is avoided here by identifying semantic features (both inflectional and derivational) that take up structural positions to the left (Prefield) or the right (Postfield) of the Word Nucleus. On the typological side, while P/Y is mainly dependent marking-suffixal, this Australian language also shows traces of head-marking and prefixal morphology. It remains to determine if dependent-marking only languages make use of exocentric morphological constructions exclusively, following Nichols’s (1986:  109) argument that head-marked syntax implies endocentricity. In P/Y inflection is consistent in appearing in the Postfield, whereas derivation can be either prefixal or suffixal. From the perspective of the morphological constructions advanced in this article, P/Y shows endocentric and exocentric morphological constructions in its morphology, which are realized through constructional schemas that belong to all the categories defined in the theoretical part of this work: continuous, discontinuous, synthetic, analytic, nonrecursive and recursive.

 Javier Martín Arista

As a final assessment, the morphological nature of the constructions under scrutiny prevents them from being as dynamic as syntactic constructions are. The interaction with context is far more limited and the semantic-syntactic motivation less direct. Moreover, inheritance and combination of constructions of the kind found in syntax are far less likely to appear. Constructions and constructional schemas, however, are not a by-product. They account for universal as well as language-specific properties of word-formation and inflection.

References Baker, Mark. 1988. Incorporation. Chicago IL: University of Chicago Press. Baker, Mark. 2003. Lexical categories. Cambridge: CUP. Bakker, Dik. 2001. The FG Expression Rules: A dynamic model. Revista Canaria de Estudios Ingleses 42: 15–54. Bauer, Laurie. 2005. The borderline between derivation and compounding. In Morphology and its demarcations, Wolfgang Dressler, Dieter Kastovsky, Oskar Pfeiffer & Franz Reiner (Eds), 97–108. Amsterdam: John Benjamins. Beard, Robert & Mark Volpe. 2005. Lexeme-Morpheme Base Morphology. In Handbook of word-formation, Pavol Štekauer & Rochelle Lieber (Eds), 189–205. Dordrecht: Springer. Bloomfield, Leonard. 1984 [1933]. Language. Chicago IL: The University of Chicago Press. Booij, Geert. 2005. Compounding and derivation. In Morphology and its demarcations, Wolfgang Dressler, Dieter Kastovsky, Oskar Pfeiffer & Franz Reiner (Eds), 109–132. Amsterdam: John Benjamins. Butler, Christopher. 2003a. Structure and function: A guide to three major structural-functional theories, Part 1: Approaches to the simplex clause. Amsterdam: John Benjamins. Butler, Christopher. 2003b. Structure and function: A guide to three major structural-functional theories, Part 2: From clause to discourse and beyond. Amsterdam: John Benjamins. Bybee, Joan. 1985. Morphology: A study of the relation between meaning and form. Amsterdam: John Benjamins. Chomsky, Noam. 1970. Remarks on nominalization. In Readings in transformational grammar, Roderick Jacobs & Peter Rosenbaum, 184–221. Waltham MA: Ginn and Co. Cortés Rodríguez, Francisco. 2006a. Negative affixation within the Lexical Grammar Model. RæL-Revista Electrónica de Lingüística Aplicada 5: 27–56. Cortés Rodríguez, Francisco. 2006b. Derivational morphology in Role and Reference Grammar: A new proposal. Revista Española de Lingüística Aplicada 19: 41–66. Croft, William. 2001. Radical Construction Grammar. Syntactic theory in typological perspective. Oxford: OUP. Di Sciullo, Anna Maria. 2003. Morphological relations in asymmetry theory. In Asymmetry in grammar, Vol. 2: Morphology, phonology, acquisition, Anna Maria Di Sciullo (Ed.), 9–36. Amsterdam: John Benjamins. Di Sciullo, Ana Maria. 2005. Asymmetry in morphology. Cambridge, MA: The MIT Press. Dik, Simon. 1986. On the notion ‘functional explanation’. Belgian Journal of Linguistics 1: 11–52.



A typology of morphological constructions 

Dik, Simon. 1997a [1989]. The theory of Functional Grammar, Part 1: The structure of the clause. Edited by Kees Hengeveld. Berlin: Mouton de Gruyter. Dik, Simon. 1997b. The theory of Functional Grammar, Vol. 2: Complex and derived constructions. Edited by Kees Hengeveld. Berlin: Mouton de Gruyter. Dixon, R.M.W. 2003. Australian languages. Cambridge: CUP. Dryer, Mathew. 1991. SVO languages and the OV: VO typology. Journal of Linguistics 27: 443–482. Dryer, Mathew. 1992. The Greenbergian word order correlations. Language 68(1): 81–138. Eckert, Paul & Joyce Hudson. 1988. Wangka Wiru. A handbook for the Pitjantjatjara language learner. Adelaide: University of South Australia. Everett, Daniel. 2002. Towards an RRG theory of morphology. Lecture delivered at The 2002 International Conference on Role and Reference Grammar, held at the University of La Rioja. Everett, Daniel. 2005. Cultural constraints on grammar and cognition in Pirahã. Current Anthropology 46(4): 621–46. Foley, William & Robert D. Van Valin, Jr. 1984. Functional syntax and universal grammar. Cambridge: CUP. Givón, Talmy. 1979. On understanding grammar. New York NY: Academic Press. Glass, Amee & Dorothy Hackett. 1970. Pitjantjatjara grammar. Camberra: Australian Institute for Aboriginal Studies. Goddard, Cliff. 1993. A learner’s guide to Pitjantjatjara/Yankunytjatjara. Alice Springs: Institute for Aboriginal Development Press. Goddard, Cliff. 2001. Pitjantjatjara/Yankunytjatjara to English Dictionary. Alice Springs: Institute for Aboriginal Development Press. Goldberg, Adele E. 1995. Constructions. A Construction Grammar approach to argument structure. Chicago: Chicago University Press. Goldberg, Adele E. 2006. Constructions at work: The nature of generalization in language. Oxford: OUP. Greenberg, Joseph. 1991. The last stages of grammatical elements: contractive and expansive desemanticization. In Approaches to grammaticalization, Elizabeth Closs Traugott & Bernd Heine (Eds), 301–314. Amsterdam: John Benjamins. Hay, Jennifer. 2002. From speech perception to morphology: Affix ordering revisited. Language 78(1): 527–555. Hay, Jennifer. 2003. Causes and consequences of word structure. London: Routledge. Hengeveld, Kees. 1989. Layers and operators in Functional Grammar. Journal of Linguistics 25: 127–157. Kay, Paul. 1997. Words and the grammar of context. Stanford CA: CSLI. Langacker. Ronald. 1987. Foundations of Cognitive Grammar I: Theoretical prerequisites. Stanford CA: Stanford University Press. Langacker, Ronald. 1991. Foundations of Cognitive Grammar II: Descriptive application. Stanford CA: Stanford University Press. Langacker, Ronald. 2005. Integration, grammaticization, and constructional meaning. In Construction Grammars. Cognitive grounding and theoretical extensions, Jan-Ola Östman & Mirjam Fried (Eds), 157–189. Amsterdam: John Benjamins. Leino, Jaakko. 2004. Frames, profiles and constructions. Two collaborating CGs meet the Finnish Permissive Construction. In Construction Grammars. Cognitive grounding and theoretical extensions, Jan-Ola Östman & Mirjam Fried (Eds), 89–120. Amsterdam: John Benjamins.

 Javier Martín Arista Lieber, Rochelle. 1992. Deconstructing morphology. Chicago IL: University of Chicago Press. Lieber, Rochelle. 2004. Morphology and lexical semantics. Cambridge: CUP. Mairal Usón, Ricardo & Francisco Cortés Rodríguez. 2000–2001. Semantic packaging and syntactic projections in word formation processes: the case of agent nominalizations. Revista Española de Lingüística Aplicada 14: 271–294. Marchand, Hans. 1969. The categories and types of present-day English word-formation. Munich: Beck’sche. Martín Arista, Javier. Forthcoming. Unification and separation in a functional theory of morphology. In Investigations of the syntax-semantics-pragmatics interface, Robert D. Van Valin, Jr. (Ed.), Amsterdam: John Benjamins. Michaelis, Laura & Knud Lambrecht. 1996. Toward a construction-based theory of language function: The case of nominal extraposition. Language 72(2): 215–247. Nichols, Johanna. 1986. Head-marking and dependent-marking grammar. Language 62(1): 56–119. Nichols, Johanna. 1992. Linguistic diversity in space and time. Chicago IL: University of Chicago Press. Nichols, Johanna. 1993. Heads in discourse: Structural versus functional centricity. In Heads in grammatical theory, Greville Corbett, Norman Fraser & Scott McGlashan (Eds), 164–185. Cambridge: CUP. Rijkhoff, Jan. 2002. The noun phrase. Oxford: OUP. Rood, David. 2005. Wichita word formation: Syntactic Morphology. In Morphology and its demarcations, Wolfgang Dressler, Dieter Kastovsky, Oskar Pfeiffer & Franz Reiner (Eds), 1–15. Amsterdam: John Benjamins. Schweiger, Fritz. 2000. Compound case markers in Australian languages. Oceanic Linguistics 39(2): 256–284. Sosa Acevedo, Eulalia. 2007. The semantic representation of Anglo-Saxon (Ge-)Séon and (Gé-)Lócian: Syntactic evidence for meaning decomposition. RæL-Revista Electrónica de Lingüística Aplicada 6: 92–107. Štekauer, Pavol. 2005. Compounding and affixation: Any difference? In Morphology and its demarcations, Wolfgang Dressler, Dieter Kastovsky, Oskar Pfeiffer & Franz Reiner (Eds), 151–159. Amsterdam: John Benjamins. Toman, Jindrich. 1998. Word Syntax. In The handbook of morphology, Andrew Spencer & Arnold Zwicky (Eds), 306–321. Oxford: Blackwell. Van Valin, Robert D., Jr. 2005. Exploring the syntax-demantics interface. Cambridge: CUP. Van Valin, Robert D., Jr. & Randy LaPolla. 1997. Syntax: Structure, meaning and function. Cambridge: CUP. Williams, Edwin. 1981. On the notions ‘lexically related’ and ‘head of a word’. Linguistic Inquiry 12: 51–97. Zwicky, Arnold. 1993. Heads, bases and functors. In Heads in grammatical theory, Greville Corbett, Norman Fraser & Scott McGlashan (Eds), 293–315. Cambridge: CUP.

part ii

The Lexical Constructional Model An overview

The Lexical Constructional Model Genesis, strengths and challenges* Christopher S. Butler This article reviews briefly some recent work on relationships across a spectrum of functionalist, cognitivist and constructionist approaches to language. It then goes on to chart the history of the model currently known as the Lexical Constructional Model, showing how it has developed from work in a range of earlier approaches: Dikkian Functional Grammar, Coseriu’s Lexematics, Role and Reference Grammar, Wierzbicka’s Natural Semantic Metalanguage model, Mel’čuk’s Meaning-Text Theory, and most recently Goldberg’s version of Construction Grammar and the theory of metaphor and metonymy. The strengths of the model are then summarised, and a number of challenges for the future are discussed.

1.  Introduction The early years of the 21st century are proving to be an interesting time for linguistic theory. The last few years have seen a welcome increase in the discussion of similarities and differences both within groupings of theories (e.g. those which would claim to be functionalist in orientation) and across such groupings (e.g. formalist, functionalist, cognitivist, constructionist). This in turn is leading to increasing awareness of the possibilities for rapprochement between models, and to the realisation that a combination of ideas from different approaches may turn out to be much more powerful than any of the models taken by itself. It is against this background that the present article is written. In it, I shall first review briefly some recent work on relationships across a spectrum of functionalist, cognitivist and constructionist theories, as a background to the rest of the discussion. I shall then look at one recent model, the Lexical Constructional Model (henceforth LCM), which richly embodies the principle of combining good ideas from a variety of

*I am grateful to Francisco Gonzálvez García and to two reviewers for their comments on an

earlier version of this paper. The work reported here was done within the framework of the project ‘Estudio comparativo de la interfaz gramática-discurso en lengua inglesa con especial referencia a la coherencia y a la subjetividad’ (A comparative perspective on the grammardiscourse interface in English, with special reference to coherence and subjectivity), financed by the Spanish Ministry of Education and Science (HUM2007-62220).

 Christopher S. Butler

compatible sources, and which is described in greater detail in other papers in this collection. The LCM is a complex model, with antecedents in a whole range of functional, cognitivist and constructionist approaches. For this reason it is useful, in understanding and evaluating the current proposals, to look at how the model came into being. The first part of this article therefore demonstrates how and why the model has arisen, taking material from various approaches. I then summarise the strengths of the current model which emerge from the foregoing discussion, and conclude the paper with a discussion of some important challenges for the future.

2.  Relationships across theories Culicover & Jackendoff (2005:  546–547), at the end of a book in which they make a number of crucial modifications to Chomskyan linguistic theory with which functionalists and proponents of constructionist approaches would feel comfortable, say that they “would like to see more investigation that compares frameworks dispassionately, for it is only by doing such comparisons that we pit them against each other scientifically rather than merely sociologically”. Of some relevance to this enterprise is Butler (2003a, 2003b), which first discusses how we might characterise the set of functional approaches to language, and within that a set of functional grammars, which itself contains a sub-set of structural-functional grammars, focused on the relationship between structure and function, and capable, in principle, of describing languages in terms of a set of explicit and interlocking rules and principles. It is argued that three grammars, Dik’s Functional Grammar, Role and Reference Grammar and Systemic Functional Grammar, could be seen as central to the set of structural-functional grammars. The rest of the work then presents a detailed comparative and critical discussion of these three grammars. More comprehensive, though necessarily less detailed, is the comparison of eleven approaches, spread across a range of functional, cognitive and constructionist models, presented in Gonzálvez-García & Butler (2006), which in turn builds on previous work limited to just six models, and reported in Butler & Gonzálvez-García (2005). The larger project covered the following approaches: Dikkian Functional Grammar (FG: e.g. Dik 1997a, 1997b); its latest manifestation in Functional Discourse Grammar (FDG: e.g. Hengeveld & Mackenzie 2006, 2008); Role and Reference Grammar (RRG: e.g. Van Valin & LaPolla 1997; Van Valin 2005); Systemic Functional Grammar (SFG: e.g. Halliday & Matthiessen 1999, 2004); the work of Givón (e.g. Givón 1995, 2001a, 2001b); the Emergent Grammar (EG) of Hopper, Thompson and other ‘West Coast’ functionalists (e.g. Hopper 1998; Hopper & Thomspon 1985, 1993); Langacker’s Cognitive Grammar (CG: e.g. Langacker 1987, 1991); and three Construction Grammar models (Fillmore et al. e.g. Fillmore, Kay & O’Connor 1988[2003]; Fillmore & Kay 1995; Goldberg (e.g. 1995, 2006); Croft (e.g. 2001, 2005)); as well as the work of Jackendoff, most recently expounded in the Culicover & Jackendoff book mentioned



The Lexical Constructional Model: Genesis, strengths and challenges 

above, which it was thought would make an interesting point of comparison. A close reading of the literature led to the establishment of a set of thirty-six features on which the approaches could profitably be compared. It was found that approximately half these features were shared fairly homogeneously across the set of models investigated, and formed a common core for functionalism, cognitivism and constructionism. Other, smaller sets of features clearly set off two major groupings: on the one hand a cognitive/constructionist group consisting of EG, CG, Goldberg and Croft; and on the other a clearly functionalist group comprising FG, FDG and RRG. SFG, Givón’s work, the approach of Fillmore et al. and Jackendoff ’s ‘Simpler Syntax’ model share some features with the cognitivist/constructionist group, but differ in having other characteristics which are not common in the main groups, and which give these four approaches a somewhat anomalous profile with respect to the other models. One of the most noteworthy features of the model to be explored in this paper is that it capitalises on the similarities between functionalist and cognitivist approaches and achieves a synthesis of the two strands of what we might call functional-cognitive linguistics.

3.  The genesis of the Lexical Constructional Model In order to help the reader follow more easily the various stages of development which have resulted in the LCM, Figure 1 presents an overview of the model’s genesis. Dik’s Functional Grammar

Coseriu’s Lexematics

Functional Lexematic Model

Lexical Grammar Model

[further semantic enrichment]

Role and Reference Grammar

Natural Semantic Metalanguage

Construction Grammar

Lexical Constructional Model Figure 1.  The genesis of the Lexical Constructional Model.

Meaning-Text Theory

metaphor/metonymy theory

 Christopher S. Butler

3.1  From Dik’s FG and Coseriu’s Lexematics to the Functional Lexematic Model The Functional Grammar proposed by Dik (1978, 1989, 1997a, 1997b) allocated a central role to the lexicon, which contained all the basic predicates and terms of a language.1 The selection of a predicate from the lexicon was the starting point for the building up of the underlying structure of the clause. Predicates were listed in the lexicon in predicate frames, specifying the syntactic category of the predicate, the number of arguments (quantitative valency), the semantic functions of these arguments (qualitative valency), any semantic selection restrictions imposed on the arguments in non-metaphorical usage, and a meaning definition, expressed in the form of meaning postulates which link the predicate, by means of entailment relations, to other predicates of the same language. For example, Dik (1997a: 100) gives the following meaning definition for the predicate bachelor, in which both sides of the bilateral entailment relation are predicate frames:

(1) (= Dik’s (52)) bachelor [N] (xi: 〈man [N]〉)Ø ↔ unmarried [A] (xi: man [N])

This tells us that the predicate bachelor, which is nominal, has a single argument with Zero semantic function (the function proposed for states in FG), and must be applied to an entity with the feature ‘man’, and that the predicate can be defined as equivalent to an expression in which the adjectival predicate unmarried restricts the nominal predicate man. The predicates used in meaning definitions may themselves be complex, as in the case of unmarried and man above, and may therefore need to be defined in terms of simpler predicates of the language, in the process of stepwise lexical decomposition. For instance, for the series of predicates assassinate, murder, kill, die, Dik (1997a: 100) provides meaning definitions which can be stated informally as follows (Dik also provides more formal definitions): (2).

(= Dik’s (54)) a. assassinate murder in a treacherous way b. murder kill a human being intentionally c. kill cause an animate being to die d. die become dead

Dik thus provided a mechanism which allows us to relate sets of predicates within a language, in an onomasiological (thesaurus-like, rather than alphabetical, as in a dictionary) manner. However, as pointed out by Martín Mingorance (1990: 232–233), “it

.  Terms are what fill the argument slots of predicates. The lexicon and derived predicates and terms (created by predicate formation rules and term formation rules respectively) together form the fund in Dik’s model.



The Lexical Constructional Model: Genesis, strengths and challenges 

is somewhat paradoxical that within FG no coherent methodology has been devised for the onomasiological structuring of the lexicon which would make possible its organization in lexical fields and, consequently, the stepwise decomposition of the groups of lexemes of each field”.2 It was this gap in FG which Martín Mingorance set out to fill in his Functional Lexematic Model (henceforth FLM).3 In essence, the FLM is a combination of Dik’s proposals for the lexicon in FG and Lexematics, a model developed by Coseriu in the 60s and 70s and refined by Geckeler (see Coseriu 1981). The latter was an elaboration of the structural semantic model, allowing the mapping of the lexical structure of a language in terms of lexical fields or domains. In the FLM, as in Coseriu’s model, the primary task is to investigate the paradigmatic structure of the lexicon. In Lexematics this task is carried out in terms of the investigation of fields and classes, the former being characterised by semes (lower level features), classemes (higher level features) and dimensions (intermediate between sets of lexemes and the lexical field itself). Once the paradigmatic structure has been worked out, the syntagmatic potential of lexemes can be investigated, in terms of selection restrictions of a semantic and syntactic nature. Martín Mingorance’s work, and that of his colleagues and students in the 80s and early 90s, provides many examples of such analyses, largely though by no means exclusively in English and Spanish. One feature of the early FLM which is of particular relevance to its later development is the fact that it develops considerably the meaning definitions of predicates in FG. No longer do we simply have a set of meaning postulates which are meant to give a (partial) characterisation of the meaning of the lexeme. Rather, the definition takes the form of a more finely nuanced predicate frame which indicates various components of the meaning. Consider, for example, the definitions for be drowsy and drop off (in the sense of going to sleep) offered by Martín Mingorance (1990: 246): (3) BE DROWSYV (x1)Proc def = [beginV (x1)Proc (x2: [fall asleepV (x1)Proc] (x2))Goal]Process (y1: [appearV (x1: calmAdj & relaxedAdj (y1))0] (y1))Circumstance (4) DROP OFFV (x1)Proc def = [fall asleepV (x1)Proc]Process i (y1: [Neg intendV (x1)Ag (x2: Processi)Goal] (y1))Circ

.  In a similar vein, Schack Rasmussen (1994: 41) observes that meaning definitions play no part in grammatical analysis in FG. She goes on to elaborate her own model of lexical semantic patterning in terms of action schemes and semantic fields. .  For a somewhat fuller description than can be provided here, see Butler (2003a: 99–105). Several of the key papers on the early development of the FLM are collected in Martín Mingorance (1998).

 Christopher S. Butler

In the mid to late 90s, the FLM was developed further, largely by two of Martín Mingorance’s former students, Pamela Faber and Ricardo Mairal Usón (see especially Faber & Mairal Usón 1994, 1998a, 1998b, 1999; Mairal Usón 1994). Here, these developments will be illustrated from Faber & Mairal Usón’s 1999 book, Constructing a lexicon of English verbs, which presents the most detailed account of the later stages of the model. Faber and Mairal Usón see the lexicon not as a static store-house, but rather as a “dynamic, text-oriented network of information about words and their contexts” (1999: 57), and as serving not only as the basis for the language user’s mental lexicon, but also as the lexical part of a model of language, an important element in Natural Language Processing by computer, and a dictionary. The dictionary itself is conceived as a grammar, in which words are allocated their semantic, pragmatic, syntactic and morphological properties, an idea which is fully consonant with those of Martín Mingorance. The methodology adopted for the elaboration of lexical networks is a bottom-up process in which similarities and differences of meaning in the definitions of individual lexemes in a range of monolingual dictionaries are used to establish lexical dimensions, and so to create a hierarchy of domains and subdomains within an overall field. All lexemes within a given field have the same nuclear defining word (definiens), and the various lexical dimensions then permit differentiation among the set of lexemes. Factorisation of the meaning elements of a group of lexemes leads to the definition of a subdomain, and further factorisation of the elements of the subdomains leads to the establishment of higher level domains. It will be clear that this methodology is in complete harmony with the stepwise lexical decomposition proposals of Dik. An example of a lexical hierarchy is shown in (5), taken from Faber & Mairal Usón (1999: 188). This set of verbs is taken from the lexical field of CONSUMPTION. (5) (= Faber & Mairal Usón’s (232)) drink to consume liquid, taking it into one’s mouth and swallowing it.   imbibe to drink alcohol [formal].   gulp (down) to drink something very quickly.   quaff to drink something quickly [old-fashioned].   swig to drink something quickly in large amounts in a series of big swallows [informal].   swill to drink something quickly and greedily in large amounts [informal].   guzzle to drink something (especially alcohol) very quickly, greedily and noisily in an unattractive way.   tipple to drink something (especially alcohol) secretly and in small amounts [informal].   sip to drink something slowly in very small amounts.

As can be seen, the eight verbs indented to show their subordinate status in the hierarchy are all defined in terms of the superordinate predicate drink, and are



The Lexical Constructional Model: Genesis, strengths and challenges 

differentiated by various lexical parameters such as manner and speed of drinking and the substance drunk. An important feature of Faber and Mairal Usón’s account is that, again in accordance with the principles advanced by Martín Mingorance, they investigate the syntagmatic potential of lexemes within hierarchies, in terms of their syntactic complementation patterns. They propose a Principle of Lexical Iconicity, according to which “The greater the semantic coverage of a lexeme, the greater its syntactic variation” (Faber & Mairal Usón 1994: 211, 1998a: 8, 1999: 187). For instance, in the hierarchy shown in (5) the verb drink can take zero complementation, a NP, a PP with to (e.g. drink to someone’s health) and a reflexive plus a PP with to/into (e.g. drink oneself to death/into a stupor), whereas imbibe has just the first two possibilities, and each of the other verbs only one (the NP in most cases, zero for tipple) (Faber & Mairal Usón 1999: 189). Thus complementation patterns are shown not to be arbitrary, but rather to be semantically motivated. Faber and Mairal Uson’s proposals thus involve considerable enrichment of the FG concept of the predicate frame. In order to reflect this enrichment these authors set up the category of predicate schema, defined as … a modular, dynamic characterization that subsumes linguistic symbolic units obtained through the activation of lower-level schemas. These schemas are linguistically motivated and reflect our perceptions of reality. (Faber & Mairal Usón 1999: 213)

It is important to note here the strongly cognitive orientation of the definition: schemas “reflect our perceptions of reality”. Indeed, Faber and Mairal Usón’s contention is that “lexical structure on both the paradigmatic and syntagmatic axes can be said to reflect cognition through the codification of linguistic knowledge” (1999: 203), and they explicitly link their concept of the schema with Langacker’s (1987: 371) definition (Faber & Mairal Usón 1999: 212). Finally, Faber and Mairal Usón show that there are connections among schemata, giving rise to what the authors (1999: 228) call semantic macronets. For instance there are clear links between visual perception and cognition, in the use of predicates such as see and show. We must now leave the FLM as an extension of FG, and turn to the next development on the way to the LCM, namely the synthesis of the FLM with ideas from Role and Reference Grammar.

3.2  The Functional Lexematic Model and Role and Reference Grammar: A synthesis Faber & Mairal Usón’s 1999 book, as we have seen, paved the way for an integration of the semantic aspects of lexical structure with the syntactic aspects, in terms of the

 Christopher S. Butler

linkage between semantically-based hierarchies and syntactic complementation patterns. However, the model had no explicit component which provided a fully systematic account of the mapping of semantics on to syntax. Unfortunately, the parent model FG was being criticised for its own lack of a syntactic level (see Van Valin 1990; also the discussion in Butler 2003a: 205–209). There was, however, another functional theory, Role and Reference Grammar (RRG) which did indeed have a clearly defined syntax, in addition to a cross-linguistically validated subtheory of the semantic structures of predicates, and sets of mapping algorithms for the linkage of semantic representations to syntactic representations and vice versa. At the heart of the semantic level in RRG are the logical structures (LS) of predicates, which form the core of the entry for a given predicate in the lexicon. In the standard late-90s version of the theory, a typical LS would appear as in the examples in (6) and (7) below, taken from Van Valin & LaPolla (1997: 155).

(6) drink do´(x, [drink´(x, y)])



(7) melt

BECOME melted´(x)

Here, the elements in bold with primes are abstract predicates, while elements such as BECOME are semantic modifiers. Thus the meaning of drink is decomposed into the general activity predicate do´ and an abstract predicate labelled drink´, and melt into the inchoative modifier BECOME and the abstract predicate labelled melted´. A more complex decomposition is shown in the LS for kill given in (8).

(8) kill [do´(x, Ø)] causE [BECOME dead´(y)]

Lexical decomposition is seen as essential in order to generalise across semantically related verbs and their arguments (Van Valin & LaPolla 1997: 90–91). The use of abstract predicates and modifiers reflects the concerns of RRG practitioners that the theory should have a high degree of typological adequacy, so that predicates in all languages should be analysed, ultimately, in terms of a set of universal semantic elements. However, it is patently obvious (and was fully accepted by RRG linguists) that while elements such as do´ and BECOME are perhaps plausible candidates for universal status, those such as drink´ or melted´ are clearly not. Some progress was being made in formulating more delicate decompositions of lexemes (see Van Valin & LaPolla 1997: 116–118 on verbs of saying, and Van Valin & Wilkins 1993 on remember and its equivalents in the Australian language Mparntwe Arrernte), but nothing like a detailed and systematic account of decomposition was available. Mairal Usón and his colleagues saw this as an opportunity to combine the logical structures of RRG, and the explicit mapping rules available within that model, with the greater lexical sophistication of the FLM. It was clear, however, that any such move would necessarily involve the abandonment of the Dikkian principle of stepwise lexical decomposition, in favour of the use



The Lexical Constructional Model: Genesis, strengths and challenges 

of abstract predicates. Mairal Usón & Van Valin (2001: 157–159) and Mairal Usón & Faber (2002: 41ff.) give three pieces of evidence to show that the FG predicate frame, with its meaning definitions based on the principle of stepwise lexical decomposition, is not an appropriate mechanism to respond to the challenge of constructing a lexically-based grammar. Firstly, FG gives no account of how predicate frames arise or how the argument structure shown in them can be linked with the meaning definition of the predicate. Secondly, each of the alternations shown by certain classes of verb (Levin 1993) would demand the postulation of a separate predicate frame, with no way of showing what information is shared across predicates, or how lexical classes are linked systematically with syntactic configurations. Thirdly, there exist constructions which cannot be dealt with adequately using the machinery available in FG. In (9), for example, it is not clear whether the resulting entity encoded in into pieces should be seen as an argument of the predicate or a satellite (aka adjunct).

(9) You sprinkle it on and then cut the pie into pieces. (BNC A3C 319)4

Mairal Usón and his colleagues demonstrate that all these problems can be solved if the RRG system of lexical semantic representation is adopted. Firstly, the LS in RRG provides a mechanism for integrating the argument structure of a predicate with a definition of its meaning. Secondly, the LS can be expanded in such a way that additional arguments, which may not be syntactically obligatory, can be accommodated, so providing a way of dealing with alternations. Thirdly, the result component in examples such as (9) is handled in the RRG LS in the form of a structure of the type BECOME pred´(y) (in this instance, BECOME pieces´ (pie)). The crucial concept in the new model is that of the lexical template which characterises each lexical class and “encodes regularities and maximises information in the lexicon with a minimum cost of representation” (Mairal Usón & Faber 2002: 54). Such templates include both syntactically-relevant aspects of the meaning of a predicate and those semantic features which are relevant for distinguishing a particular lexical class from others. These correspond to the external and internal variables, respectively, of RRG: see Van Valin & LaPolla (1997: 117–118), where internal variables are proposed in order to account for the properties of verbs of saying. Effectively, lexical templates refine the meaning definitions proposed in the FLM and re-express them in terms of the abstract semantic predicates and modifiers of RRG, distinguishing notationally between those arguments which are linked to the syntax and those which are needed for the internal semantic characterisation of the class of verbs concerned. Mairal Usón & Faber (2002: 45, fn. 2) see lexical templates as similar to the predicate schemas of their earlier work: however, as we have seen, the two differ crucially in the change from

.  Examples marked as BNC are taken from the British National Corpus, World Edition.

 Christopher S. Butler

language-specific stepwise lexical decomposition to an RRG-based abstract metalanguage. In (10) is shown the template proposed by Mairal Usón & Faber (2002: 55) for verbs of cutting.5 (10) [[do´(w, [use.sharp-edged.tool(α)in(β)manner´(w, x)]) & [BECOME be-at´ (y, x)]] causE [[do´(x, [make.cut.on´(x, y)])] causE [BECOME pred´ (y, (z))]]], α = x.

This formula can be unpacked in ordinary language as follows: … an effector (w) uses a sharp-edged tool (x) in such a way that the tool becomes in contact with a patient (y), causing an event such that x makes a cut on y, and this, in turn, causes that y becomes cut. Furthermore, a new variable (z) is introduced to account for those cases where the final result is further specified (into pieces, in strips, open etc.). (Mairal Usón & Van Valin 2001: 159)

The external, syntactically-relevant variables are w (with the semantic function of Effector), x (Instrument), y (Patient) and z (Result State), and the internal variables α and β. This lexical template is seen as valid for the whole of the class of cutting verbs, and can be adapted in various ways in order to account for the properties of specific verbs within that class. These adaptations involve the specification of values for internal variables, which include not only Instrument and Manner (α and β above), Manner being subdivided into an Effector type and a Movement type, but also Affected Object and Result. For instance, the verb hew indicates that a large rock, stone or other hard material (Affected Object) is cut in a rough way, with difficulty (Manner); shave encodes the cutting of hair from the face or other parts of the body, very close to the skin (Affected Object), with a razor or shaver (Instrument); chop indicates that something is cut into pieces (Result state), by repeatedly hitting it (Manner) with a sharpedged tool such as an axe or knife (Instrument); and so on (Mairal Usón & Faber 2002: 60, 62). Mairal Usón and Faber propose to link the specification of object types, such as cutting instruments and type of affected object, to a well-developed ontological semantics network such as that used in the Mikrokosmos project (see e.g. Mahesh & Nirenburg 1995). Mairal Usón & Faber (2002: 75–85) also show how templates can be formulated for each of the syntactic alternations proposed by Levin (1993) for verbs of cutting: transitive, conative, middle, unspecified object, instrument subject, characteristic property of instrument, unintentional interpretation, path phrase, resultative phrase and creation/transformation. The minimal expression of the lexical template corresponds to the transitive alternation (see (11)), for which the template shown in (12) is proposed:

.  A minimally different template is presented in Mairal Usón & Van Valin (2001: 159).



The Lexical Constructional Model: Genesis, strengths and challenges 

(11) Lady Braithwaite cut the celebration cake … (BNC KAF 124) (12) (= Mairal Usón & Faber’s (46), 2002: 75) [do´(x, Ø)] causE [BECOME cut´(y)]

For the conative alternation, as in (13), the template is as in (14). .

(13) A figure popped out of a doorway to Alexei’s left, and he cut at it with his sword and ran on. (BNC G17 1991) (14) [do´(w, [use.sharp-edged.tool(α)in(β)manner´(w, x)]) & [BECOME be-at´ (y, x)]]

The instrument subject alternation, exemplified in (15), is analysed according to the template in (16). (15) The knife cut its throat. (BNC HTM 2742) (16) [[do´(Ø, [use.sharp-edged.tool(α)in(β)manner´(Ø, x)])] & [BECOME be-at´ (y, x)]] causE [[do´(x, [make.cut.on´(x, y)])] causE [BECOME pred´(y)]]

As a final example, the resultative phrase alternation, as in (9), receives the interpretation in (17). (17) [[do´(w, [use.sharp-edged.tool(α)in(β)manner´(w, x)])] & [BECOME be-at´(y, x)]] causE [[do´(x, [make.cut.on´(x, y)])] causE [BECOME pred´(y, (z))]]

where pred´ represents the result predicate, which functions together with cut as a complex predicate in what RRG terms a nuclear juncture (see Van Valin & LaPolla 1997: 442–444). The question which now arises is how the relationships between the maximal template in (10) and the more specific templates in, for example, (12), (14), (16) and (17), can best be described and indeed constrained. The answer given by Mairal Usón and Faber is to postulate a Lexical Template Modelling Process which allows operations of particular kinds: Lexical Template Modeling Process Lexical templates can be modeled by accommodating external variables, instantiating internal variables and operators (e.g. causE), or else, by introducing elements resulting from the fusion with other templates iff there is a compatibility between the features in the lexical template and the syntactic construction under scrutiny. (Mairal Usón & Faber 2002: 87)

The details of this modelling process, and the inventory of lexical rules involved, are worked out in much more detail in Mairal Usón (2002), which unfortunately remains unpublished. Here, the modelling process is reconceptualised in terms of fusion between two types of template: the lexical template for a particular lexical class, and templates for various types of the construction into which verbs can enter (e.g.

 Christopher S. Butler

transitive, causative/inchoative, instrument subject, etc.). We shall see later that this is an important step in the evolution of the model into its current form. Seven types of rule are postulated (Mairal Usón 2002: 56–66): i.  Full matching. This occurs when variables, subevents and operators are identical in the two templates. An example of this is that the lexical template for verbs of cutting and the template for the transitive construction have elements which can be matched exactly in this way, so accounting for the ability of this class of verbs to occur in the transitive construction (see example (11)). ii.  Suppression of variables. In this case, a variable in the lexical template is suppressed in order to fit with the number of variables in the constructional template, provided that the basic interpretation of the constructional template is not thereby violated. For instance, the agent of the lexical template for verbs of cutting can be suppressed in the instrument subject alternation, where the instrument functions as the effector of the action (see example (15)). iii.  Fusion of internal variables. For this type of modelling to occur, the internal variables of the canonical lexical template must be compatible with the semantic content of the construction with which the lexical template fuses. An example is the inability of the verb jab to occur in the resultative construction, explained by the fact that the lexical template for this verb includes ‘repeatedly’ as an internal variable, and this is incompatible with the semantics of the resultative construction. iv.  Event identification condition. In this case, the semantics of the construction must permit it to occur as a proper subevent of the lexical template. For example, the semantics of the conative construction (see example (13)) can be identified with the subevent of the lexical template for, for example, cutting and hitting verbs which is expressed as BECOME be-at´(y, x). v.  Predicate integration condition. This occurs where a new predicate introduced by the constructional template itself is compatible with the semantic content of the lexical template. An example is the caused motion construction illustrated in (18): (18) They would have laughed Philip out of such a hopeless misalliance; … (BNC CDY 1796)

vi.  Partial matching. This can occur when the semantics of the constructional template is compatible with at least one part of the lexical template. For instance, the template for verbs of breaking includes a component which specifies the final resulting state of being broken (i.e. involving a one-place predicate), and this licenses the use of this class of verbs in the causative/inchoative alternation, while verbs such as hit, which are two-place predicates with no final result state, cannot occur in this alternation:



The Lexical Constructional Model: Genesis, strengths and challenges 

(19) I broke the glass angrily … (BNC GWH 176) (20) The glass broke with a loud noise. (BNC GVM 162) (21) One bullet had hit the windscreen. (BNC K5M 4396) (22) *The windscreen hit.

vii.  Lexical blocking. If one of the components of a lexical template for a given verb corresponds to a suppletive form, it can block fusion with a construction which would normally allow the verb to occur in that construction. For example, one would expect the causative/inchoative alternation to occur with kill, except for the fact that the BECOME dead´(x) component is lexicalised in English as die, hence the possibility of (23) and (24) but the impossibility of (25) with the same meaning as (24): (23) But the police have killed 46 people in the past five years, … (BNC ABD 667) (24) 46 people have died in the past five years. (25) *46 people have killed in the past five years.

The 2002 paper is important in other ways besides the spelling out of the lexical rules, in that for the first time it situates lexical templates and their modelling within an overall model, dubbed the Lexical Grammar Model, whose architecture is shown in Figure 2, adapted slightly from Mairal Usón (2002: 25). Several features of the model deserve comment here.

Lexical Entries

ONTOLOGY

Lexical Templates

Word Formation Lexical Templates

Affixal Entries

Lexical Template Modelling Process Constructional Lexical Templates

Constructional Lexical Templates

Output: Fully Specified Semantic Representation

Output: Fully Specified Semantic Representation

Linking rules

Linking Rules Macroroles Determinant PSA Determinatum Syntactic Templates

Clauses

Figure 2.  The Lexical Grammar Model.

Expression rules

Derived words

 Christopher S. Butler

Firstly, note that the input to the mechanism of lexical templates and their modelling is an ontology. This is consistent with the proposals made in Mairal Usón & Faber (2002), where, as we have seen, the specification of the properties of instantiations for internal variables, such as the Affected Object of verbs of cutting, relies on linkage to a conceptual ontology. It will be remembered that Mairal Usón & Faber (2002) suggest linking elements of lexical templates to an ontology developed within a computational approach such as the Mikrokosmos project of Nirenburg and his colleagues. While this possibility is still recognised in the Lexical Grammar Model, Mairal Usón (2002: 18) also makes an alternative proposal which will be of importance in the later development of the model, namely that a more linguistically-oriented and culturally less biased approach might be to adopt, or adapt, the proposals made by Wierzbicka and her colleagues within the framework of the Natural Semantic Metalanguage (NSM) model (see e.g. Goddard & Wierzbicka 2002), a possibility which, as Mairal Usón recognises, was adumbrated in Butler (2002: 273–274). This second possibility is not, however, further developed in the 2002 paper. Secondly, the model envisages a parallel set of templates for word formation, which was studied extensively within the framework of the Functional Lexematic Model (see e.g. Cortés Rodríguez 1994, 1997a, 1997b, 1997c, 1997d, Mairal Usón 1999, Mairal Usón & Cortés Rodríguez 2000–2001, also the brief summary of this work in Butler 2003a: 104–105). Thirdly, the lexical template modelling process is seen as the initial phase in the linking of lexical templates to the final form of the clause. The second phase of linking consists of a modification of the RRG semantics-to-syntax linking algorithm, which takes as its input the result of fusion between the lexical template and the constructional template, and produces as its output the final syntactic structure of the clause. This final, and crucial, linking phase merits more detailed discussion. RRG posits two sets of linking rules, one mapping from semantics on to syntax and corresponding to language production, the other going from syntax on to semantics, and so relevant to language comprehension (see Van Valin & LaPolla 1997: Chapters 7 and 9; Van Valin 2005: Chapters 5 and 7). Both algorithms make crucial use of the concepts of macrorole (MR), privileged syntactic argument (PSA) and syntactic template. Macroroles are the semantic roles of actor and undergoer, which generalise across more specific thematic roles such as Agent, Patient, Theme, Experiencer, etc. They are needed because many rules in the grammar refer to these more generalised roles rather than to the specific ones, which in fact have no theoretical status within RRG, since they are predictable from the forms of logical structures (e.g. the first argument of an activity predicate, with do´ in its LS, is an Effector, while the argument of a one-place stative predicate is a Patient). The prototypical actor is an Agent (an Effector performing a deliberate action), while the prototypical undergoer is a Patient, but other specific roles can also have macrorole status (e.g. the Experiencer in a state of feeling is an undergoer).



The Lexical Constructional Model: Genesis, strengths and challenges 

The privileged syntactic argument is the category proposed in RRG to deal with syntactic relations, and is anchored to a particular construction in a language, so that in Icelandic, for example, we have a PSA for finite verb agreement, one for passive participle/predicate adjective agreement, and so on. For most constructions in English and some other languages, the PSA is equivalent to the traditional Subject, though Van Valin & LaPolla (1997: 263–274) argue persuasively that the categories of Subject and Object are not generalisable across a wide range of language types and so should find no place within a typologically adequate theory. PSAs are of two broad types, controllers (concerned with phenomena internal to the syntactic ‘core’ of the clause in RRG) and pivots (concerned with complex constructions such as clause linkage), and there are subtypes of pivots, but these details need not concern us here. Syntactic templates are the building blocks for the construction of syntactic structures in RRG. They are language-specific, and make reference to the syntactic units recognised in the cross-linguistically validated layered structure of the clause: the clause consists of a core and a periphery; the core contains the nucleus (housing the semantic predicate) and the core arguments of the predicate, while the periphery contains non-arguments (adjuncts).6 So, for example, Van Valin (2005: 15) provides a set of templates for various possible configurations of the core for English, together with further templates for the pre-core slot (into which certain items such as wh-constituents and fronted elements which are integral to the clause obligatorily go) and the leftdetached position (a position outside the clause proper which can be used to announce topical elements about which the clause will say something, as in As for the Normandy campaign, I wouldn’t have missed it for the world. (BNC A61 2468)). The important point about the linking algorithms of RRG is that they provide a set of universal rules, tested against a wide range of language types, and also language-specific rules, which together allow us to link the logical structures which form the semantic basis of clauses and sentences to the syntactic structures which realise the meanings conveyed. They are thus an essential part of the generative and interpretive mechanisms proposed in RRG, and constitute a very strong set of hypotheses with a great deal of cross-linguistic support. The job of the semantics-to-syntax linking algorithm is to determine the assignment of actor and undergoer macroroles, and then to decide on the morphosyntactic coding of arguments, including which macrorole (if there is more than one) will be the PSA. The algorithm also makes reference to the selection of appropriate syntactic templates which will be assembled to form the final structure, and finally assigns arguments to positions in the syntactic representation of the sentence. The syntax-to-semantics algorithm, on the other hand, first determines the macrorole(s) and any other core

.  Recently, Van Valin (2005: 21) has proposed that there is a periphery for each layer of the clause. This proposal need not concern us here.

 Christopher S. Butler

arguments in the clause, then retrieves the LS of the predicate in the nucleus of the clause and maps the macroroles on to it. Detailed exemplification of the linking algorithms is given in the chapters of Van Valin & LaPolla (1997) and Van Valin (2005) cited earlier, and a briefer, simplified account in Butler (2003a: 143–148). As can be seen in Figure 2, the Lexical Grammar Model makes use of an adaptation of the RRG semantics-to-syntax algorithm, including the use of intermediate semantic roles, or macroroles, the concept of the privileged syntactic argument, and syntactic templates. Mairal Usón (2002: 74) proposes a set of linking rules which simply makes some terminological changes and a few minor additions to the RRG algorithm, and he goes on (in Chapter 4) to give an informal account of how the rules work for a number of different verb classes and constructions.

3.3  Further enrichment of the semantics: The role of Natural Semantic Metalanguage and Meaning-Text Theory We have seen that considerable enrichment of the semantics of RRG was achieved by the incorporation of the decompositional techniques pioneered in the Functional Lexematic Model, but that there was a pressing need for a semantic metalanguage in terms of which meanings in any language could be expressed. We have also seen that Mairal Usón (2002) mentions the work of Wierzbicka and her colleagues within the Natural Semantic Metalanguage model as a possible way of achieving this goal. As noted by Mairal Usón & Faber (2007: 147),7 the primitives used in the semantic decompositions of Faber & Mairal Usón (1999) actually corresponded quite closely to a subset of those proposed in the NSM.8 However, a number of disadvantages have been noted in relation to the meaning explanations which are given for lexical items in the NSM: Mairal Usón & Faber (2007: 143) point out that they are unwieldy and so not conducive to a concise representation; Nichols (1982: 698) observes that Wierzbicka’s approach is highly contentoriented and does not make links with the properly syntactic properties of the items being characterised. For this reason, Mairal Usón and Faber, while still striving to use NSM-like primitives in lexical decomposition, have sought to enrich their semantic descriptions of lexemes by making use of the Meaning-Text Theory (MTT) of Mel’čuk (1981, 1988, 1989). Mel’čuk’s ‘Explanatory and Combinatorial Lexicology’ framework makes use of a set of lexical functions of the form f(x) = y, where f is the lexical function, x is its argument, and y is the result of applying the function to the argument. An example

.  Mairal Usón & Faber (2007) is a revised version of a presentation given at the 2005 RRG conference. In terms of chronology, therefore, it presents the next stage in the development of the LCM. .  For a list of the NSM primitives see the NSM website at http://www.une.edu.au/lcl/nsm/ nsm.php#model, consulted 14.1.2008.



The Lexical Constructional Model: Genesis, strengths and challenges 

is the function Magn, which expresses intensity, and can be applied to a range of lexemes: Magn (smoker) = heavy, Magn (bachelor) = confirmed, and so on. In Mel’čuk’s own work, these functions are used largely to account for syntagmatic relations between lexemes; however Mairal Usón and Faber, following a suggestion by L’Homme (2005), use them paradigmatically, to distinguish between lexemes within a given domain. In Mairal Usón & Faber (2007:  148), a lexical template consists of two parts: the semantic properties which differentiate one lexeme from another within a given domain, specified in terms of MTT lexical functions, and the event structure of the predicate, showing its grammatically-relevant properties, in the form of a RRG logical structure. An example, taken from Mairal Usón & Faber (2007: 149), is shown in (26), which represents the meaning of regret: (26) [sympt (sadness) involv1,2 (want) degrad (do)2 locin temp ←/(become)2 locin temp←] feel´(x, y)

The LS for this predicate, on the right of the definition, uses the predicate feel´, and what distinguishes regret from other verbs of feeling is encapsulated in the first part of the definition, composed of lexical functions and arguments. The feeling of regretting generates a symptom (sympt), in that ‘x’ feels sad about ‘y’. There is also a sub-activity, represented by involv1,2, specifying that ‘x’ wants (or, more exactly, would prefer) the event (which is the second argument) not (degrad) to have been carried out (do) or to have happened (become), in the past (temp). Mairal Usón and Faber offer many more examples showing how different combinations of lexical functions and primitives can be used to define a range of lexemes.9 At this point, then, the model has reached a stage where the syntactically relevant properties of a predicate are formalised, so potentially allowing linkage to a syntactic structure, and where the meaning of a given lexeme is related to, and differentiated from, that of other lexemes by means of appropriate putatively universal semantic primitives taken from the NSM and lexical functions derived from MTT.

3.4  Synthesis with Construction Grammar and metaphor/metonymy theory: The birth of the Lexical Constructional Model In Mairal Usón (2002), as we have seen, the idea is put forward that lexical templates fuse with templates representing constructions. However, these templates are discussed informally, and no clear connection is made with proposals made in Construction Grammars. In recent work Mairal Usón has combined forces with the cognitive

.  In recent work it has been proposed that lexical templates should be remodelled in terms of the qualia proposed by Pustejovsky (1995), leading to a closer integration of the two halves of the template: see Cortés Rodríguez, this volume; Mairal Usón & Ruiz de Mendoza Ibáñez, this volume, fn 11.

 Christopher S. Butler

linguist Ruiz de Mendoza, one of the consequences of this alliance being that the implicit link with cognitively-oriented versions of Construction Grammar such as that of Goldberg is made explicit and becomes a central plank of the model, a development which is documented in detail in Ruiz de Mendoza & Mairal Usón (2006a, 2006b, in press) and Mairal Usón and Ruiz de Mendoza (in press). Indeed, Gonzálvez-García (2007) comes to the conclusion that that the LCM, as currently conceived, leans more towards the cognitivist pole than the functionalist one. Since these ideas are spelled out fully in other contributions to this volume, I shall present only a very brief outline here. Ruiz de Mendoza & Mairal Usón (2006a: 26–28) contrast the views of functionalist and constructionist approaches on the relationship between lexicon and grammar. Functionalist theories such as RRG regard lexicon and grammar (in the sense of morphosyntax) as separate,10 and postulate that information from lexical semantic representations can be projected on to the morphosyntax via linking rules. On the other hand, cognitive theories, including those constructionist theories which subscribe to the main tenets of Cognitive Linguistics11 (e.g. that of Goldberg 1995, 2006), posit that lexicon and grammar form a continuum, and that linking rules are not required. Rather, there is a large set of form-meaning pairings, or constructions, which together constitute the ‘constructicon’. Ruiz de Mendoza and Mairal Usón’s view is that both approaches have their weaknesses: functionalist approaches do not pay sufficient attention to the importance of constructions in determining morphosyntactic structure,12 while on the other hand constructionist approaches do not offer detailed accounts of the constraints on combining particular lexical entries with particular constructions. As an example, Ruiz de Mendoza and Mairal Usón consider the caused motion construction discussed by, among others, Goldberg (1995: 152–179). Example (18), discussed earlier in relation to lexical template modelling, is repeated for convenience as (27) below: (27) They would have laughed Philip out of such a hopeless misalliance;… (BNC CDY 1796)

This is an example of what in Cognitive Linguistics is called coercion, regulated by the Override Principle (Michaelis 2003: 268) according to which if there is a clash between a lexical entry and a construction, the former adapts itself to the latter. Ruiz de Mendoza and Mairal Usón see this as one application of a more general principle whereby .  Note that there is one significant exception, in that Hallidayan Systemic Functional Grammar, like many cognitive approaches, regards grammar and lexical items as forming a continuum. .  I shall use initial capitals (Cognitive Linguistics) to distinguish the movement associated with scholars such as Langacker and Lakoff from the wider area which could be termed ‘cognitive linguistics’. .  It should be noted that in this section I am dealing only with Mairal Usón and Ruiz de Mendoza’s own position. In §5.1 I shall suggest that they have underestimated the extent to which RRG is already a constructionist model.



The Lexical Constructional Model: Genesis, strengths and challenges 

more general, higher level cognitive patterns take in, or subsume, less general, lower level patterns (Ruiz de Mendoza & Díez Velasco 2002). In the LCM, the more general pattern corresponding to the construction is described in the same metalanguage as that used for lexical templates, so facilitating fusion of the two. For instance, the caused motion construction can be represented as in (28): (28) [LS] causE [BECOME NOT be-in´(x, y)]

The fusion of this structure with the lexical template for laugh formalises the way in which ‘their laughing’ causes Philip to move (metaphorically) out of the misalliance he is in. This fusion, or ‘lexical-constructional subsumption’, can be represented as in Figure 3 (for analysis of a similar example see Mairal Usón and Ruiz de Mendoza in press: the asterisk before NOT represents optionality).13 Lexial template external to the construction: laugh-at´ (x, y)

Abstract semantic representation of the Caused Motion construction: [Lexical template] CAUSE [BECOME *NOT be-LOC´ (y, z)]

Constructionally coerced modification of the lexical template laugh´ (x, y)

Unification of the modified template with the construction: [laugh´ (x, y)] CAUSE [BECOME NOT be-LOC´ (y, z)]

Fully specified semantic representation: [laugh´ (they, Philip]) CAUSE [BECOME NOT be-LOC´ (Philip, hopeless misalliance)] Figure 3.  Lexical-constructional subsumption: the caused motion construction.

.  An analysis of the caused motion construction is also given in Ruiz de Mendoza Ibáñez and Mairal Usón 2006b: 125), but this differs in small but important ways from the more recent version. In particular, the initial lexical template is shown as laugh´(x, y) rather than laughat´(x, y). See further discussion below.

 Christopher S. Butler

Showing how the fusion process works is not, however, enough. As Ruiz de Mendoza and Mairal Usón point out, we also need to specify the constraints on the fusion process. They specify two types of constraint, internal and external. Internal constraints are those specified in the lexical template modelling process of Mairal (2002), minus what was then called partial matching: full matching, variable suppression, internal variable fusion, event identification condition, predicate integration condition, lexical blocking (Mairal Usón and Ruiz de Mendoza in press). Since these were dealt with earlier, I shall not elaborate further here. External constraints are concerned with the licensing of fusion types by the higher level cognitive processes of metaphor and metonymy. As an example, the fusion between the lexical template for laugh at and the caused motion construction requires the conversion of the basic activity predicate laugh-at´(x, y) into a causative accomplishment predicate laugh´(x, y). Mairal Usón and Ruiz de Mendoza (in press) point out that this is a metaphorical process, in which information about a source domain allows us to reason about another domain, the target. In this particular case, the metaphor proposed is EXPERIENTIAL ACTION IS EFFECTUAL ACTION, which works through the correspondence between two set of participants: in the source domain, the effector of the causative accomplishment has an effect on the effectee, while in the target domain, we have an actor who is just a ‘doer’ of the action which is then experienced by a goal/ experiencer (see also Ruiz de Mendoza & Mairal Usón 2007). However, it is clearly necessary to constrain such mappings so that inappropriate metaphors are ruled out, and to this end previous work in Cognitive Linguistics by Ruiz de Mendoza and his colleagues is brought in. More concretely, three principles are invoked to constrain mappings and so explain the appropriateness or inappropriateness of metaphors: Extended Invariance Principle (Ruiz de Mendoza 1998): based on the original Invariance Principle of Lakoff (1993), this states that “the generic-level structure of a target domain has to be preserved in such [sic] a way that is consistent with the topological structure of the source domain” (Ruiz de Mendoza & Mairal Usón 2006a: 39). Correlation Principle (Ruiz de Mendoza & Santibáñez 2003): “for a metaphoric source element to qualify as the counterpart of a target domain element, the source element needs to share the relevant implicational structure of the target element” (Ruiz de Mendoza & Mairal Usón 2006a: 40). Mapping Enforcement Principle (Ruiz de Mendoza & Mairal Usón 2006a: 40): “no item in the source is to be discarded from a mapping system if there is a way to find a corresponding source element in the target domain”. Ruiz de Mendoza & Mairal Usón (2006a: 40) claim that these three principles “stipulate all possible correspondences between a source and a target domain”, and go on to show how they can account for felicitous and infelicitous metaphors. They also demonstrate that metonymies, characterised as “domain-internal mappings where one of



The Lexical Constructional Model: Genesis, strengths and challenges 

the domains involved provides a point of access to the other” (2006a: 41) form the conceptual, cognitive basis for certain alternations, such as the causative/inchoative alternation illustrated earlier with break in (19) and (20), where the metonymy involved is PROCESS FOR ACTION. For much more detail of this and other metonymies and metaphors see Ruiz de Mendoza & Mairal Usón (2007). In the latest version of the LCM (Ruiz de Mendoza & Mairal Usón in press; Mairal Usón & Ruiz de Mendoza, this volume) cognitive models play an even more extensive role. The model is expanded to include no fewer than four levels, of which we have so far discussed only the first. All four levels involve constructional templates and subsumption processes, regulated by internal and external constraints. The first level, dealing with lexical templates and their subsumption into level 1 constructional templates, is now regarded as the ‘core grammar’.14 This level also allows for some inferential activity (‘conceptual cueing’), such as that involved in She’s ready (for the party) or I will (marry you). At level 2 further low-level inferential processes are handled, such as those involved in the interpretation of the Who do you think you’re X construction. Level 3 deals with the high-level inferences relevant to the determination of illocutionary force, which refers to social conventions for acceptable behaviour. Finally, level 4 is concerned with high-level non-situational frames to do with the often implicit logical, temporal and conceptual relations between propositions. Examples of phenomena at each level can be found in Ruiz de Mendoza and Mairal Usón (in press) and Mairal Usón and Ruiz de Mendoza (this volume).

4.  Strengths The strengths of the LCM will, I hope, have emerged from the historical sketch I have already presented, so I shall merely summarise them briefly here, in the form of a list. – That part of the LCM which is inherited from the Functional Lexematic Model provides a mechanism for describing lexical domains paradigmatically in terms of hierarchical structures based on similarity and difference of meaning. –– The combination of the FLM-derived proposals with the account of predicateargument structure in terms of the logical structures of RRG provides, in principle, a powerful means of relating semantics to syntax in this area.

.  It is important to note that this use of the term ‘core grammar’ does not coincide with that in generativist models, where a distinction is made between the core, which is the object of study, and a periphery which is left aside.

 Christopher S. Butler

–– The use of primitives taken from the list proposed by Wierzbicka and her colleagues in the Natural Semantic Metalanguage framework, together with the use of lexical functions taken, or adapted, from Mel’čuk’s Meaning-Text Theory approach, allows very rich semantic descriptions of lexemes of individual languages with a basis in components which have been validated through the study of a typologically diverse range of languages. –– Thanks to the concept of lexical-constructional subsumption, the model is able to account for the ways in which particular (sets of) predicates can occur in the various types of construction whose properties have been studied in Construction Grammars. Particularly important here are the internal constraints which regulate the fusion of a lexical template with a constructional template. –– Despite the close links with Goldbergian Constructional Grammar, the LCM differs from that model in that it is concerned with providing a principled account of the division of labour between lexical semantics (aka lexical templates) and constructional semantics (aka constructional templates), instead of positing constructional templates as better overall predictors of sentence meaning. –– The current model provides cognitive underpinning for the process of lexicalconstructional subsumption and its regulation, by pointing to the role of metaphor and metonymy in external constraints on subsumption processes, thereby acknowledging the systematicity of these cognitive operations as explanatory, motivating factors. –– The model also provides, albeit so far only programmatically, further levels of structure, each based on the common mechanisms of subsumption into a higher-level constructional template and conceptual cueing, which account, in principle, for various types of inferential processing, illocutionary meaning and implicit logical, temporal and conceptual connections between propositions. Again, these processes have a strong underpinning in terms of Cognitive Linguistics. Nevertheless, the LCM, for all its history in a diverse range of functional, cognitive and/or constructionist approaches, is still very much in its infancy, and there remain a number of questions to be answered, and challenges to be faced. These are the topic of discussion in the rest of this article.

5.  Challenges 5.1  The relationship between semantics and morphosyntax The most obvious lacuna in the current version of the LCM is that it does not make clear how semantic representations are mapped on to morphosyntactic representations. As we have seen, it is envisaged in Mairal (2002) that the fused template structure



The Lexical Constructional Model: Genesis, strengths and challenges 

will still feed into the semantics-to-syntax linking algorithm of RRG for production or the syntax-to-semantics mapping algorithm for comprehension. The current version of the LCM, however, contains no such mechanism. Furthermore, in Ruiz de Mendoza & Mairal Usón (2006a: 28) we find the following: Within the broader context of a functional and cognitive paradigm, the LCM provides an alternative to the relationship between lexicon and grammar and offers a framework which bridges the theoretical gap between projectionist and construction-based approaches by developing an inventory of constraints that simulate the role of interface (or linking rules) on the one hand, and by vindicating the role of constructions as a crucial part in the semantic representation of the theory. (emphasis added to the original)

Presumably the linking rules referred to here are those of RRG, perhaps modified as suggested in Mairal (2002), and the constraints are the internal and external constraints which govern the fusion of lexical and constructional templates. However, these do quite different jobs: the linking rules tell us how to get from semantic representations to syntactic ones (and vice versa), while the constraints tell us either about possible modifications to the structures of lexical and constructional templates or about which types of lexical template can fuse with which types of constructional template, both specified in semantic terms. Certainly, the templates specify syntactically relevant information, but this information is still semantic and still needs to be linked with the morphosyntax: templates in themselves contain the information necessary for linking to the relevant syntactic arguments, but do not specify explicitly how that linking should occur, although syntactic consequences of constraints are discussed informally. Presumably there are two possibilities: a projectionist account as in RRG (but see also below), or a non-projectionist account in terms of mapping between parallel structures as in Goldbergian Construction Grammar. The form of the templates used in the LCM, and especially their predictive nature with respect to the syntax, suggest a projectionist account, and this is also adumbrated by a remark in Mairal Usón and Ruiz de Mendoza (this volume), to the effect that lexical templates “combine (encyclopedic) semantic and logical variables that are linked to one another in readiness for syntactic projection” (emphasis added). I would therefore suggest that the original proposals of Mairal (2002) be revived, if necessary revised, and incorporated into the LCM (see also Cortés Rodríguez, this volume footnote 1). Once this link to the syntax is established, the proponents of the model will be in a better position to address other aspects of syntax which are not so closely related to predicate-argument structures. A further problem related to the interaction between semantics and syntax is that the proponents of the LCM appear to have underestimated the extent to which RRG (which, as we have seen, forms the basis for the logical structures in their model) is already a constructionist model. Constructional schemas play a very important part in present-day RRG (Van Valin 2005: 131–135). Whereas generalisations which apply across constructions and across languages are captured by means of general principles of the

 Christopher S. Butler

grammar, the idiosyncratic, language-specific properties of constructions are accounted for in terms of constructional schemas which can themselves make reference to the more general principles. For instance, the general, cross-linguistic properties of the passive construction are stated in terms of general principles for two types of voice constructions: ‘PSA modulation voice’, which allows a non-default argument to act as PSA, and ‘argument modulation voice’, which realises a macrorole argument in a non-canonical form (Van Valin 2005: 116). Those syntactic, morphological, semantic and pragmatic properties which specifically characterise the English passive, on the other hand, are stated in the constructional template for that construction (Van Valin 2005: 132), which makes reference to general principles of the grammar. A considerable number of such constructional schemas are detailed in Van Valin (2005). Constructional schemas are particularly important because of the part they play in linking semantics to syntax and vice versa. In the semantics-to-syntax linking algorithm they provide the language- and construction-specific details needed for the appropriate encoding of the semantics in the morphosyntax, while in the syntax-to-semantics algorithm they specify what the PSA is, in languages where there can be different PSAs for different constructions. Crucially for the LCM, constructional schemas in RRG can also be used to account for coercion phenomena, including those which can result from the caused motion construction, as in example (27) of §3.4. Ruiz de Mendoza & Mairal Usón (2006a: 27) use this construction to illustrate their view that “functional projectionist theories ignore the unquestionable theoretical weight of constructions in predicting morphosyntactic structure, an issue that undermines a theory of linking”. However, Van Valin (2005: 239) demonstrates that the constructional schema for the resultative construction in English accounts perfectly well for linking in the famous case of sneeze (see Goldberg (1995: 9), shown in (29), with the logical structure in (30): (29) Chris sneezed the napkin off the table. (30) [SEML do´(Chris, [sneeze´(Chris)])] causE [BECOME NOT be-on´ (table, napkin)]

The constructional schema for the resultative specifies that the construction is of the serial verb type, the syntax of clause linkage involving nuclear cosubordination. The syntactic template is selected by means of the appropriate general rule, there is no PSA, and the construction obeys the default linkage rules. The semantic properties of the construction are specified in terms of the predicate corresponding to the first nucleus in the syntactic structure causing the predicate corresponding to the second nucleus, which must be static rather than dynamic. Importantly, there is no need to convert the predicate sneeze from intransitive to transitive (compare the analysis in Figure 3 of §3.4), since the arguments of the logical structures involved come together in a nuclear juncture, so that the napkin, although it is an argument of the whole logical structure, is not an argument of sneeze itself. The LCM analysis goes beyond that of



The Lexical Constructional Model: Genesis, strengths and challenges 

RRG in discussing the internal and external constraints which govern the possibility of combining a particular predicate with a given construction, but proponents of the LCM clearly need to demonstrate the superiority of their own valency-changing analysis over the simpler and more elegant solution proposed in RRG. Recently, Van Valin (in press) has proposed an even more strongly constructionist component for RRG. He suggests that while speakers adopt a projectionist system, in which morphosyntactic structures are projected from meanings, hearers necessarily operate in a constructionist manner, in that comprehension requires ‘co-composition’ processes in order to arrive at the meaning of an utterance from the morphosyntactic components which are progressively assembled. Projectionist and constructionist views are thus seen as compatible, and equally necessary in a theory which takes seriously the requirements of cognitive adequacy. This is a viewpoint which proponents of the LCM might want to take into consideration when they examine in more detail the roles of projectionist and constructionist viewpoints in their work.

5.2  The relationship between grammar and the lexicon In the following, Ruiz de Mendoza and Mairal Usón appear to reject the claim, made by many cognitive and/or constructionist linguists, that grammar and the lexicon form a continuum: One of the weaknesses of the excessive emphasis put by cognitive linguists on the non-discreteness of categories is the “overapplication” of this idea to all areas of linguistic enquiry, including the relationship between lexicon and grammar. The concept of non-discreteness of categories initially came from observations on the internal semantic makeup of concepts associated with concepts (prototype theory versus traditional feature theory). It is fairly uncontroversial there, but there are no iron-clad arguments in Cognitive Linguistics why it should apply to the relationship between lexicon and grammar. (Ruiz de Mendoza & Mairal Usón 2006b: 122)

This view is very much in line with that of Van Valin (2007:  236), who points out problems with the proposal, central to the constructionist postulation of the integration of grammar and lexicon, that “all levels of grammatical analysis involve constructions: learned pairings of form with semantic or discourse function, including morphemes or words, idioms, partially lexically filled and fully general phrasal patterns” (Goldberg 2006: 5, emphasis in original). If this is so, then what, Van Valin asks, is the content, both theoretical and empirical, of the claim that everything is a construction? Furthermore, he observes, since constructions are learned form-meaning pairings, they must be language-specific, so raising the question of how cross-linguistic generalisations are to be captured. According to Van Valin, Goldberg finally takes a position quite close to that of Croft (2001), who claims that there are no cross-linguistic generalisations, only cognitive ones.

 Christopher S. Butler

However, in another paper, Ruiz de Mendoza & Mairal (2006a: 29) claim that their proposal “captures relevant features that lexical template representations share with constructional representations, which makes our description fully at home with the idea of a lexical-constructional continuum”. Furthermore, the subtitle of Mairal Usón and Ruiz de Mendoza (in press) is ‘The lexicon-grammar continuum’, again suggesting that this claim is accepted. However, although the fact that the lexical and constructional templates share the same format makes it easier to show relationships between them and to allow their fusion, this does not entail that they form a continuum rather than constituting separate levels. Obviously there is a need for clarification here. Langacker (1987: 26) puts the case for the lexicon-syntax continuum very clearly. Contrasting his view with that of formal linguistics, where “[s]yntax was thought of as the domain of generality and regularity, of productive rules with fully predictable outputs; anything falling short of these standards was relegated to the purgatory of lexicon”, Langacker states that he is “aware of no a priori or factual grounds for believing that grammatical constructions divide nearly into two groups on the basis of generality, or that the regular aspects of language structure can be segregated in any meaningful way from the irregular ones”. Crucial to the debate are what Langacker (1987: 35) calls ‘conventional expressions’, comprising “stock phrases, familiar collocations, formulaic expressions, and standard usages that can be found in any language and thoroughly permeate its use”. These, as Langacker (p.36) points out, do not fit into the lexicon, as conceived in formal grammars, since they are larger than prototypical lexical items, and many of them have meanings which are derivable from those of their components. But neither do they sit easily in the syntax, since this is conceived as dealing with general rules rather than with specific combinations. Unlike formal accounts, cognitive and functional approaches cannot simply turn a blind eye to a set of constructions which is so important in communication. More recently, even Culicover & Jackendoff (2005), erstwhile stalwarts of Chomskyan linguistics, have totally rejected the lexicon-syntax divide in favour of a continuum, for largely the same reasons as advanced many years earlier by Langacker. As we have seen, some constructions which fall into Langacker’s class of ‘conventional expressions’ have been examined within the LCM: for instance, level 2 of the model deals with expressions of the form What do you think you’re X. The implications of such expressions for the question of the relationship between lexicon and syntax have not, however, been explored. Also unexplored are the implications of the degree of generality of the construction: expressions such as What do you think you’re X fall between fully abstract constructions (e.g. X causes Y to receive Z) and fully local, specific ones such as holophrases (e.g. How do you do?)

5.3  Extending lexical coverage The LCM has inherited the emphasis placed, within the Functional Lexematic Model, on the verbal lexicon and its relationship with argument structure. Clearly, future



The Lexical Constructional Model: Genesis, strengths and challenges 

developments in the model need to give attention to nominal, adjectival and adverbial aspects of the lexicon.

5.4  The upper levels of the model The non-core levels of the model (i.e. levels 2, 3 and 4) are still somewhat programmatic, though the account given by Mairal Usón and Ruiz de Mendoza in the present volume adds some interesting detail. Further development of these levels can benefit from previous work by Ruiz de Mendoza and his colleagues (e.g. on illocution see Ruiz de Mendoza 1999; Pérez Hernández 1997, 1998/1999, 2001; Pérez Hernández & Ruiz de Mendoza 2001; Ruiz de Mendoza & Otal Campo 1997, 2002: 145–158; Ruiz de Mendoza & Baicchi 2007).

5.5  Criteria of adequacy for functional theories A set of criteria of adequacy to which a functional theory should aspire was proposed by Dik (1989) and is discussed critically by Butler (1991, 1999, 2003a, 2003b). In Butler (2003b: 485–489) the following modified set of criteria of adequacy is proposed:











Descriptive adequacy – Attested linguistic productions, such as those found in corpora, should be used as an important source of data, though other sources, including intuition, informant testing and psycholinguistic experimentation are also important. – Typological adequacy, the requirement that a theory should be capable of accounting for the full range of phenomena found across the whole range of the world’s languages, can be seen as a type of descriptive adequacy (see Hengeveld & Pérez Quintero 2001). Explanatory adequacy – Psychological/cognitive adequacy: A functionalist theory should take into account what we know of the cognitive structures and mechanisms involved in the storage and processing of language. – Sociocultural adequacy: We must account for the ways in which texts/discourses are shaped by, and in turn help to shape, sociocultural relationships, and sociocultural features must be cognitively represented. – Discoursal adequacy: A functionalist theory must give an account of the structure and functioning of discourse, seen as a dynamic, rule-governed, contextually-related activity, leading to structures composed of units with functional relations between them, and subject to coherence constraints. – Acquisitional adequacy: A truly functionalist theory must give an account, inevitably largely constructivist in orientation, of how the properties of languages proposed in that theory can be learned.

 Christopher S. Butler

More recently (Butler fc.) a further criterion, that of diachronic adequacy, has been added to the descriptive set, and discoursal adequacy is seen as concerned with both descriptive and explanatory adequacy. In what follows I shall look briefly at the LCM in the light of these criteria. Firstly, we may observe that the analysis of authentic data from corpora has so far played a rather minor role in the development of the LCM. The hierarchical meaning descriptions of Faber & Mairal Usón (1999) were developed on the basis of factorisation of meanings taken from a range of dictionaries, and it has always been maintained that corpus analysis, though important, should serve the role of providing examples and checking on the basic correctness of the proposals put forward. However, it is now widely accepted that when corpus analysis is taken as the starting point for an investigation, rather than as merely corroborative, in-depth description of the data leads, more often than not, to a very different description. I would therefore like to see corpora used as a test-bed for rigorous testing of the meanings proposed for lexemes in the LCM. Corpus studies would also be useful at a more general level, for instance in assessing the importance of frequency on issues ranging from the definition of constructional templates to a psycholinguistically adequate account of metaphor and metonymy.15 In terms of typological adequacy, it is clear that the data base for the LCM needs expanding beyond the study of English. There are indications that this is already under way, in the form of a contrastive English-Spanish dictionary project (see Mairal Usón 2007). The LCM has not so far given consideration to matters of diachronic development. A full assessment of the psychological/cognitive adequacy of the LCM is beyond the scope of the present article, since it would involve discussion of the extent to which Cognitive Linguistics itself is cognitively adequate, and this in turn requires careful definition of the term ‘cognitive’. I can do no more here than offer my own opinion on this matter, which is that Peeters (1998:  226) is right to criticise much of Cognitive Linguistics for not being cognitive in the sense of “the sort of linguistics that uses findings from cognitive psychology and neurobiology and the like to explore how the human brain produces and interprets language”. Langacker has based his whole theory of Cognitive Grammar on the fact that the mechanisms which underpin it, such as perception, categorisation and the like, are ones which are central to cognitive psychology, but the crucial point is that the theory is then elaborated without any concern for whether the detailed proposals made in the grammar themselves have any psychological validity. In other words, “the connection between Cognitive Linguistics and cognitive science remains weak” (Peeters 2001: 103). One important question

.  For discussion of the possible roles of corpus analysis in functional linguistics, see Butler (2004).



The Lexical Constructional Model: Genesis, strengths and challenges 

which remains for the LCM, then, is whether there is any evidence that the ‘cognitive’ principles, processes and structures proposed in that part of the LCM which derives from Cognitive Linguistics (e.g. lexical constructional subsumption, and all the different principles of metaphor and metonymy) are cognitively adequate in the sense of corresponding to the mechanisms that are actually used in processing. In attempting to answer this question, it would be important to look at the quite extensive psycholinguistic literature on, for example, the processing of metaphors.16 Level 4 of the latest version of the LCM gives us reason to hope that the model will pay attention to the study of discourse phenomena, since it gives a (so far programmatic) account of cohesion and coherence constraints. However, no model of discourse structure as such is presented. One fruitful area for future research might be the study of the interaction between lexical and constructional meanings in language acquisition. Such work could build on the information already available about the learning of constructions (see e.g. Kelly & Clark 2005). Finally, it would be interesting to study phenomena such as coercion and the effects of high level metaphor and metonymy from a sociocultural perspective, for instance examining the use of these devices in particular genres and registers of languages, and their function in the establishment, maintenance and breakdown of social relationships.

6.  Conclusion I hope to have shown in this article that the Lexical Constructional Model offers a very attractive account of many aspects of the semantics and pragmatics of predicateargument relations, deriving its strengths from the wide range of approaches which have been incorporated, often in a modified form, into the model. It has also begun to deal with other (e.g. illocutionary, discoursal) aspects of language structure, and has elaborated on the proposal that a single set of cognitive principles underlies all levels of the model. Although it has a fairly long history, the LCM itself is very new, and so has a long way to go before it reaches maturity. I have pointed out some of the areas in which I feel development is required. Of these, the most urgent is surely the need to

.  There are some honourable exceptions to the generalisation made in this paragraph. For instance, Gibbs and his colleagues have investigated experimentally the processing of idioms (for an overview see Gibbs 2007), and Bencini and Goldberg (2000) have made use of empirical techniques in their study of sorting sentences either by constructions or by the morphological form of the verb.

 Christopher S. Butler

provide the model with a syntax and with clear mechanisms for linking this to the rich semantic descriptions which the LCM provides.

References Bencini, Giulia M.L. & Adele E. Goldberg. 2000. The contribution of argument structure constructions to sentence meaning. Journal of Memory and Language 43: 640–651. Butler, Christopher S. 1991. Standards of adequacy in Functional Grammar. Review article on Simon C. Dik (1989) The theory of Functional Grammar. Part I: The structure of the clause. Journal of Linguistics 27: 499–515. Butler, Christopher S. 1999. Nuevas perspectivas de la Gramática Funcional: Los estándares de adecuación de la teoría. In Nuevas perspectivas en Gramática Funcional, Christopher Butler, Ricardo Mairal, Javier Martín Arista & Francisco J. Ruiz de Mendoza Ibáñez (Eds), 219–256. Barcelona: Ariel. Butler, Christopher S. 2002. Catching a glimpse of linguistic reality: Modelling the complexity of catch in the Functional Lexematic Model. In New perspectives on argument structure in Functional Grammar, Ricardo Mairal Usón & María J. Pérez Quintero (Eds), 247–279. Berlin: Mouton de Gruyter. Butler, Christopher S. 2003a. Structure and function: A guide to three major structural-functional theories, Part 1: Approaches to the simplex clause. Amsterdam: John Benjamins. Butler, Christopher S. 2003b. Structure and function: A guide to three major structural-functional theories, Part 2: From clause to discourse and beyond. Amsterdam: John Benjamins. Butler, Christopher S. 2004. Corpus studies and functional linguistic theories. Functions of Language 11(2): 147–186. Butler, Christopher S.  Forthcoming. Criteria of adequacy in functional linguistics. Folia Linguistica. Butler, Christopher S. & Francisco Gonzálvez-García. 2005. Situating FDG in functional-cognitive space: An initial study. In Studies in Functional Discourse Grammar [Linguistic Insights 26], J. Lachlan Mackenzie & María L.A. Gómez-González (Eds), 109–158. Bern: Peter Lang. Cortés Rodríguez, Francisco J. 1994. Lexicón onomasiológico de afijos nominalizadores españoles e ingleses. Ph.D. dissertation, Universidad de La Laguna, Tenerife. Cortés Rodríguez, Francisco J. 1997a. La creación léxica. Una aproximación funcional. La Laguna, Tenerife: Servicio de Publicaciones, Universidad de La Laguna. Cortés Rodríguez, Francisco J. 1997b. The Functional Lexematic Model of word formation. Atlantis XIX(1): 79–98. Cortés Rodríguez, Francisco J. 1997c. La morfología derivativa en la Gramática Funcional de Dik: ¿Formación de palabras o formación de predicados? Alfinge 9: 117–34. Cortés Rodríguez, Francisco J. 1997d. Sobre la relación entre léxico-genesis, diacronía y categorización. In Functional lexicology, Gerd Wotjak (Ed.), 227–243. Frankfurt: Peter Lang. Coseriu, Eugenio. 1981. Lecciones de lingüística general. Madrid: Gredos. Croft, William. 2001. Radical Construction Grammar: Syntactic theory in typological perspective. Oxford: OUP. Croft, William. 2005. Logical and typological arguments for Radical Construction Grammar. In Construction Grammars: Cognitive grounding and theoretical extensions [Constructional



The Lexical Constructional Model: Genesis, strengths and challenges 

Approaches to Language 3], Jan-Ola. Östman & Miriam Fried (Eds), 273–314. Amsterdam: John Benjamins. Culicover, Peter & Ray Jackendoff. 2005. Simpler Syntax. Oxford: OUP. Dik, Simon C. 1978. Functional Grammar. Amsterdam: North Holland. Dik, Simon C. 1989. The theory of Functional Grammar, Part I: The structure of the clause [Functional Grammar Series 9]. Dordrecht: Foris. Dik, Simon C. 1997a. The theory of Functional Grammar, Part 1: The structure of the clause [Functional Grammar Series 20], 2nd Edn. Kees Hengeveld (Ed.), Berlin: Mouton de Gruyter. Dik, Simon C. 1997b. The theory of Functional Grammar, Part 2: Complex and derived constructions [Functional Grammar Series 21], Kees Hengeveld (Ed.), Berlin: Mouton de Gruyter. Faber, Pamela & Ricardo Mairal Usón. 1994. Methodological underpinnings for the construction of a functional lexicological model. Miscelánea 15: 193–217. Faber, Pamela & Ricardo Mairal Usón. 1998a. Methodological criteria for the elaboration of a functional lexicon-based grammar of the semantic domain of cognitive verbs. In The structure of the lexicon in Functional Grammar, Hella Olbertz, Kees Hengeveld & Jesús Sánchez García (Eds), 3–24. Amsterdam: John Benjamins. Faber, Pamela & Ricardo Mairal Usón. 1998b. Towards a semantic syntax. In Lexical studies towards the year 2000 [Monographic section of Revista Canaria de Estudios Ingleses 36], Francisco J. Cortés Rodríguez (Ed.), 37–64. La Laguna, Tenerife: Servicio de Publicaciones, Universidad de La Laguna. Faber, Pamela & Ricardo Mairal Usón. 1999. Constructing a lexicon of English verbs [Functional Grammar Series 23]. Berlin: Mouton de Gruyter. Fillmore, Charles J. & Paul Kay. 1995. Construction Grammar coursebook, Chapters 1 thru 11 (Reading materials for ling. X20). Berkeley CA: University of California, Berkeley. Fillmore, Charles J., Paul Kay & Mary C.  O’Connor. 1988[2003]. Regularity and idiomaticity in grammatical constructions: The case of let alone. In The new psychology of language [Cognitive and Functional Approaches to Language Structure], Michael Tomasello (Ed.), 243–270. Vol. 2. Mahwah NJ: Lawrence Erlbaum. Gibbs, Raymond W., Jr. 2007. Idioms and formulaic language. In The Oxford handbook of Cognitive Linguistics, Dirk Geeraerts & Herbert Cuyckens (Eds), 697–725. Oxford: OUP. Givón, T. 1995. Functionalism and grammar. Amsterdam: John Benjamins. Givón, T. 2001a. Syntax: An introduction, Vol. I. Amsterdam: John Benjamins. Givón, T. 2001b. Syntax: An introduction, Vol. II. Amsterdam: John Benjamins. Goddard, Cliff & Anna Wierzbicka (Eds), 2002. Meaning and universal grammar: Theory and empirical findings (2 vols.). Amsterdam: John Benjamins. Goldberg, Adele E. 1995. Constructions: A Construction Grammar approach to argument structure. Chicago IL: Chicago University Press. Goldberg, Adele E. 2006. Constructions at work: The nature of generalization in language. Oxford: OUP. Gonzálvez-García, Francisco. 2007. Situating the Lexical Constructional Model in functionalcognitive space: Some preliminary considerations. Paper given at workshop on ‘Bridging the gap between functionalism and cognitivism: the Lexical Constructional Model’, as part of the 40th annual meeting of the Societas Linguistica Europaea on ‘Functionalism in Linguistics’, Joensuu, Finland, 29 August – 1 September 2007.

 Christopher S. Butler Gonzálvez-García, Francisco & Christopher S.  Butler. 2006. Mapping functional-cognitive space. Annual Review of Cognitive Linguistics 4: 39–96. Halliday, M.A.K. & Christian M.I.M. Matthiessen. 1999. Construing experience through meaning: A language-based approach to cognition. London: Cassell. Halliday, M.A.K. & Christian M.I.M. Matthiessen. 2004. An introduction to functional grammar. 3rd Edn. revised by Christian M.I.M. Matthiessen. London: Arnold. Hengeveld, Kees & J. Lachlan Mackenzie. 2006. Functional Discourse Grammar. In Encyclopedia of language and linguistics. 2nd Edn. Vol. 4, Keith Brown (Ed.), 668–676. Amsterdam: Elsevier. Hengeveld, Kees & J. Lachlan Mackenzie. 2008. Functional Discourse Grammar: A typologicallybased theory of language structure. Oxford: OUP. Hengeveld, Kees & María J. Pérez Quintero. 2001. Descriptive adequacy in Functional Grammar. In Challenges and developments in Functional Grammar [Monographic section of Revista Canaria de Estudios Ingleses 42], María J. Pérez Quintero (Ed.), 103–17. La Laguna, Tenerife: Servicio de Publicaciones, Universidad de La Laguna. Hopper, Paul J. 1998. Emergent Grammar. In The new psychology of language [Cognitive and Functional Approaches to Language Structure], Vol. 1, Michael Tomasello (Ed.), 155–175. Mahwah NJ: Lawrence Erlbaum. Hopper, Paul J. & Sandra A. Thompson. 1985. The iconicity of the universal categories ‘noun’ and ‘verb’. In Iconicity in syntax [Typological Studies in Language 6], John Haiman (Ed.), 151–183. Amsterdam: John Benjamins. Hopper, Paul J. & Sandra A. Thompson. 1993. Language universals, discourse pragmatics, and semantics. Language Sciences 15(4): 357–376. Kelly, Barbara & Eve V. Clark (Eds), 2005. Constructions in acquisition. Stanford CA: CSLI. Lakoff, George. 1993. The contemporary theory of metaphor. In Metaphor and thought, 2nd Edn. Andrew Ortony (Ed.), 202–251. Cambridge: CUP. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. I: Theoretical prerequisites. Stanford CA: Stanford University Press. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, Vol. 2: Descriptive application. Stanford CA: Stanford University Press. Levin, Beth. 1993. English verb classes and alternations: A preliminary investigation. Chicago IL: University of Chicago Press. L’Homme, Marie-Claude. 2005. Using explanatory and combinatorial lexicology to describe terms. In Selected lexical and grammatical topics in the Meaning-Text Theory. In honour of Igor Mel’čuk, Leo Wanner (Ed.), 163–198. Amsterdam: John Benjamins. Mahesh, Kavi & Sergei Nirenburg. 1995. A situated ontology for practical NLP. In Proceedings of the Workshop on Basic Ontological Issues in Knowledge Sharing. International Joint Conference on Artificial Intelligence (IJCAI-95), August 1995. Montreal. Mairal Usón, Ricardo. 1994. Parámetros para la organización de una sintaxis léxica funcional. In Estudios de Gramática Funcional, Javier Martín Arista (Ed.), 23–73. Zaragoza: Mira. Mairal Usón, Ricardo. 1999. El componente lexicón en la Gramática Funcional. In Nuevas perspectivas en Gramática Funcional, Christopher Butler, Ricardo Mairal, Javier Martín Arista & Francisco J. Ruiz de Mendoza Ibáñez (Eds), 41–98. Barcelona: Ariel. Mairal Usón, Ricardo. 2002. Lexical templates and syntactic structures: the design of a Lexical Grammar Model. Unpublished manuscript, Universidad Nacional de Educación a Distancia, Madrid.



The Lexical Constructional Model: Genesis, strengths and challenges 

Mairal Usón, Ricardo. 2007. Building a dictionary of lexical and constructional templates in English and Spanish. Paper given at International Informal Colloquium ‘Possible Dictionaries’. Università Roma Tre, July 6–7, 2007. Mairal Usón, Ricardo & Francisco Cortés Rodríguez. 2000–2001. Semantic packaging and syntactic projections in word formation processes: The case of agent nominalizations. Revista Española de Lingüística Aplicada 14: 271–294. Mairal Usón, Ricardo & Pamela Faber. 2002. Functional Grammar and lexical templates. In New perspectives on argument structure in Functional Grammar, Ricardo Mairal Usón & María J. Pérez Quintero (Eds), 39–94. Berlin: Mouton de Gruyter. Mairal Usón, Ricardo & Pamela Faber. 2007. Lexical templates within a functional cognitive theory of meaning. Annual Review of Cognitive Linguistics 5: 137–172. Mairal Usón, Ricardo & Francisco Ruiz de Mendoza Ibáñez. In press. Internal and external constraints in meaning construction: The lexicon-grammar continuum. In Estudios de filología inglesa: Homenaje a la Dra. Asunción Alba Pelayo [Colección Varia], Teresa Gibert Maceda and Laura Alba-Juez (Eds). Madrid: Universidad Nacional de Educación a Distancia. Mairal Usón, Ricardo & Robert D.  Van Valin, Jr. 2001. What Role and Reference Grammar can do for Functional Grammar. In Challenges and developments in Functional Grammar [Monographic section of Revista Canaria de Estudios Ingleses 42], María J. Pérez Quintero (Ed.), 137–66. La Laguna, Tenerife: Servicio de Publicaciones, Universidad de La Laguna. Martín Mingorance, Leocadio. 1990. Functional Grammar and lexematics in lexicography. In Meaning and lexicography, Jerzy Tomaszczyk & Barbara Lewandowska-Tomaszczyk (Eds), 227–253. Amsterdam: John Benjamins. Martín Mingorance, Leocadio. 1998. El modelo lexemático-funcional: El legado lingüístico de Leocadio Martín Mingorance. Amalia Marín Rubiales (Ed.), Granada: Universidad de Granada. Mel’čuk, Igor. 1981. Meaning-text models: A recent trend in Soviet linguistics. Annual Review of Anthropology 10: 27–62. Mel’čuk, Igor. 1988. Semantic description of lexical units in an explanatory combinatorial dictionary: Basic principles and heuristic criteria. International Journal of Lexicography 1(3): 165–188. Mel’čuk, Igor. 1989. Semantic primitives from the viewpoint of the Meaning-Text Linguistic Theory. Quaderni di Semantica 10(1): 65–102. Michaelis, Laura A. 2003. Headless constructions and coercion by construction. In Mismatch (Form-function incongruity and the architecture of grammar), Elaine J.  Francis & Laura A. Michaelis, 259–310. Stanford CA: CSLI. Nichols, Johanna. 1982. Review of Anna Wierzbicka (1980) Lingua mentalis. Language 59: 654–659. Peeters, Bert. 1998. Cognitive musings. Review article on Richard A.  Geiger & Brygida Rudzka-Ostyn (Eds) Conceptualizations and mental processing in language. WORD 49(2): 225–237. Peeters, Bert. 2001. Does Cognitive Linguistics live up to its name? In Language and ideology, Vol.  1: Cognitive theoretical approaches. René Dirven (Ed.), 83–106. Amsterdam: John Benjamins. Pérez Hernández, Lorena. 1997. FG, illocution, and cognition. Revista Española de Lingüística Aplicada 12: 7–25.

 Christopher S. Butler Pérez Hernández, Lorena. 1998/1999. A cognitive revision of the FG treatment of illocution. Revista Española de Lingüística Aplicada 13: 207–23. Pérez Hernández, Lorena. 2001. Illocution and cognition: A constructional approach. Logroño: Servicio de Publicaciones, Universidad de La Rioja. Pérez Hernández, Lorena & Francisco J. Ruiz de Mendoza Ibáñez. 2001. Towards a pragmaticallyoriented cognitive Functional Grammar. In Challenges and developments in Functional Grammar [Monographic section of Revista Canaria de Estudios Ingleses 42], María J. Pérez Quintero (Ed.), 187–214. La Laguna, Tenerife: Servicio de Publicaciones, Universidad de La Laguna. Pustejovsky, James. 1995. The Generative Lexicon. Cambridge MA: The MIT Press. Ruiz de Mendoza Ibáñez, Francisco J. 1998. On the nature of blending as a cognitive phenomenon. Journal of Pragmatics 30(3): 259–274. Ruiz de Mendoza Ibañez, Francisco J. 1999. La ilocución y la gramática. In Nuevas perspectivas en Gramática Funcional, Christopher Butler, Ricardo Mairal, Javier Martín Arista & Francisco J. Ruiz de Mendoza Ibáñez (Eds), 99–171. Barcelona: Ariel. Ruiz de Mendoza Ibáñez, Francisco J. & Annalisa Baicchi. 2007. Illocutionary constructions: cognitive motivation and linguistic realization. In Explorations in pragmatics: Linguistic, cognitive, and intercultural aspects, Istvan Kecskes & Laurence Horn (Eds), 95–128. Berlin: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco J. & Olga I.  Díez Velasco. 2002. Patterns of conceptual interaction. In Metaphor and metonymy in comparison and contrast, René Dirven & Ralph Pörings (Eds), 489–532. Berlin: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. 2006a. Levels of semantic representation: Where lexicon and grammar meet. Interlingüística 17: 26–47. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. 2006b. Lexical representation and constructions: Bridging the gap between the constructional and process models of grammar. In A pleasure of life in words: A Festschrift for Angela Downing, Marta Carretero et al. (Eds), 107–134. Madrid: Departamentos de Filología Inglesa I/II, Universidad Complutense de Madrid. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Aspects of meaning construction in lexicon and grammar, Günter Radden, Klaus-Michael Köpcke, Thomas Berg & Peter Siemund (Eds), 33–49. Amsterdam: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. In press. Levels of description and constraining factors in meaning construction: An introduction to the Lexical Constructional Model. Folia Linguistica 42(2). Ruiz de Mendoza Ibáñez, Francisco J. & José Luis Otal Campo. 1997. Communication strategies and realization procedures. Atlantis 19(1): 297–314. Ruiz de Mendoza Ibáñez, Francisco J. & José Luis Otal Campo. 2002. Metonymy, grammar, and communication [Colección Estudios de Lengua Inglesa 7]. Albolote: Editorial Comares. Ruiz de Mendoza Ibáñez, Francisco J. & Francisco Santibáñez Sáenz. 2003. Content and formal cognitive operations in construing meaning. Italian Journal of Linguistics 15(2): 293–320. Schack Rasmussen, Lone. 1994. Semantic functions in perspective  – reconsidering meaning definitions. In Function and expression in Functional Grammar [Functional Grammar Series 16], Elisabeth Engberg-Pedersen, Lisbeth Falster Jakobsen & Lone Schack Rasmussen (Eds), 41–63. Berlin: Mouton de Gruyter, 41–63.



The Lexical Constructional Model: Genesis, strengths and challenges 

Van Valin, Robert D., Jr. 1990. Layered syntax in Role and Reference Grammar. In Layers and levels of representation in language theory [Pragmatics & Beyond, New Series 13], Jan Nuyts, A. Machtelt Bolkestein & Co Vet (Eds), 193–231. Amsterdam: John Benjamins. Van Valin, Robert D. Jr. 2005. Exploring the syntax-semantics interface. Cambridge: CUP. Van Valin, Robert D., Jr. 2007. Review of Goldberg (2006). Journal of Linguistics 43: 234–240. Van Valin, Robert D., Jr. In press. Lexical representation, co-composition, and linking syntax and semantics. In New Developments in the Generative Lexicon, James Pustejovsky et al. (Eds), Dordrecht: Kluwer. Van Valin, Robert D., Jr. & Randy J.  LaPolla. 1997. Syntax: Structure, meaning and function. Cambridge: CUP. Van Valin, Robert D., Jr. & David P. Wilkins. 1993. Predicting syntactic structure from semantic representations: Remember in English and its equivalents in Mparntwe Arrernte. In Advances in Role and Reference Grammar, Robert D. Van Valin, Jr. (Ed.), 499–534. Amsterdam: John Benjamins.

Levels of description and explanation in meaning construction* Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez This article proposes the Lexical Constructional Model (LCM) as an explanatorily adequate model for the investigation of meaning construction at all levels of linguistic description, including pragmatics and discourse. The LCM has an argument structure module or level 1 module consisting of elements of syntactically relevant semantic interpretation. Then it has three basically idiomatic modules dealing with cognitively entrenched meaning implications deriving from the application of low-level and high-level inferential schemas (levels 2 and 3 respectively) and with discourse aspects of meaning, especially cohesion and coherence phenomena (level 4). Each level is either subsumed into a higher-level constructional configuration or acts as a cue for the activation of a relevant conceptual structure that yields an implicit meaning derivation. Interaction between lexical and constructional configurations (at whatever description level) is regulated by a number of constraints that are either internal or external to the process.

1.  Introduction Over the past few years the proponents of some linguistic approaches have relaxed the harsh tone that had presided over the linguistic debate for well over three decades and have gradually come closer to recognizing that a comprehensive theory of language demands the joint efforts and serious commitment of supposedly competing theories (cf. Jackendoff 2002: xv). A clear example of this gradual approximation of positions is the one that is taking place between the functional and the cognitively-oriented

*Financial support for this research has been provided by the DGI, Spanish Ministry of Edu-

cation and Science, grants HUM2004–05947-C02–01/FILO, HUM2005–02870/FILO, and HUM2007–65755/FILO. The research has been co-financed through FEDER funds. We are grateful to two anonymous referees, to Professor Christopher S. Butler (Swansea), and to the members of the Lexicom research group (www.lexicom.es) for comments and advice on a preliminary version of this contribution. Any remaining error is our own responsibility.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

constructionist approaches to language (e.g. Culicover & Jackendoff 2005; Jackendoff 2002, 2007; Levin & Rappaport 2005; Nuyts 2005). This approximation is not surprising since the two paradigms share their emphasis on the communicative dimension of language. But the explanations offered on each side are too frequently based on opposing views on crucial theoretical issues, such as the role of verbal semantics, the nature of the syntax-semantics interface, and the role of constructions, to name just a few.1 This fundamental problem, coupled with the vast amount of work carried out within each perspective, supplying explanations at all levels of linguistic enquiry, makes the task ahead a gigantic one. But it is a necessary enterprise if we want to understand, in full detail and in an integrated manner, the many dimensions of language. It is in this context of reconciliation between paradigms that the Lexical Constructional Model (hereafter LCM) has to be placed (Ruiz de Mendoza and Mairal 2006, 2007, in press a, in press b).2 The primary concern of the LCM is to develop a usage-based, comprehensive theory of meaning construction that aims to give explanations of how all aspects of meaning, including those that go beyond so-called coregrammar (e.g. traditional implicature, illocutionary force, and discourse coherence) interact with one another. For reasons that will become apparent later on, this concern leads to the rejection of unsubstantiated dichotomies that have become part of the most recent linguistic debate, especially verb-centered projectionism (e.g. Role and Reference Grammar; RRG; S.C. Dik’s Functional Grammar; FG) versus the constructional approach, as propounded by Kay & Fillmore (1999), Goldberg (1995, 2006), Croft (2001), and Bergen & Chang (2005), among others. The LCM argues that both perspectives are really necessary if we want to account for the vast range of phenomena involved in meaning construction. Another relevant aspect of the LCM is its special focus on finding unifying features across the various levels of linguistic description and explanation. This is achieved by working under the assumption that, unless there are well-evidenced reasons to the contrary, all levels of linguistic description and explanation make use of the same or at least comparable cognitive processes (cf. Ruiz de Mendoza 2007). This assumption, which we call the equipollence hypothesis, is a methodological one and has allowed us to introduce a large degree of regularity and parsimony in our exploration. Thus, we postulate that metaphor and metonymy go beyond the lexical level of explanation and have a place in constraining lexical-constructional interaction and also in supplying

.  For a detailed discussion of the typology of functional models and of their points of convergence and divergence with cognitive accounts of Construction Grammar, we refer the reader to Butler (2003), Butler & Gonzálvez-García (2005) and Gonzálvez-García & Butler (2006). .  For further information on the Lexical Constructional Model we refer the reader to the LEXICOM research webpage: 〈www.lexicom.es〉.



Levels of description and explanation in meaning construction 

inference schemas at the pragmatic level. Other cognitive processes, such as generalization and parametrization, also have a role as constraining factors at all levels of meaning construction. We further argue that inferential activity, which we call cued inferencing or cueing, is not only a matter of the pragmatic and discourse levels, but also has a place on the predicate-argument level of grammatical description, which constitutes the central or “core” level of the LCM. Finally, we see idiomaticity as an active process not only with reference to the lexicon but also constructionally and at all levels of description. In order to give an account of the way the LCM differs from other cognitive and functional models in dealing with the interrelation between the various levels of linguistic description and explanation in meaning construction, we have designed the present article as follows. First, we place our model within the context of the projectionist versus constructionist dichotomy. Second, we discuss the problem of lexical representation by contrasting the decompositional and frame-based perspectives and argue for the notion of lexical template as an alternative form of lexical representation that integrates relevant elements from the other two perspectives. Third, we discuss syntactically-relevant non-lexical representations. These are divided into idiomatic and non-idiomatic (or argument) constructions. The latter have been the main object of research in Construction Grammar accounts, so we focus our attention on the former and on the role these constructions play in meaning derivation. We distinguish implicative, illocutionary, and discourse constructions and study their main features from a descriptive perspective. Finally, in order to give our account full explanatory adequacy, we explore the way in which the meaning creation process is constrained by cognitive principles. We distinguish between two forms of working out meaning, cued inferencing and subsumption, and address the way each step of the meaning generation process is internally and externally constrained on the basis of cognitive mechanisms. In general, our discussion addresses the various aspects of the overall architecture of the LCM, as diagrammed in Figure 1, which is borrowed from Ruiz de Mendoza & Mairal (in press a).

2.  Projections vs. constructions The proliferation of frameworks (both formal and functional) that have been propounded in linguistics over the last years, or even the different offshoots that compete within the same linguistic school, such as the various versions of Construction Grammar or the many different functional theories,3 are an eloquent proof that linguistic

.  According to Nichols (1984), these range from ‘extreme’ and ‘moderate’ to ‘conservative’ (see also Butler 2003). A much more detailed and updated description of functional approaches in comparison with cognitive theories can be found in Gonzálvez-García & Butler (2006).

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

LTs subsumption/ conceptual cueing Level 1 CTs

Level 1 internaland external constraints ARGUMENTAL OR “CORE” LEVEL

subsumption/ conceptual cueing

Level 2 internal and external constraints

Level 2 CTs/CSs

subsumption/ conceptual cueing Level 3 CTs/CSs

Level 3 internal and external constraints

subsumption/ conceptual cueing

Level 4 constraints

Level 4 CTs/CSs Discourse representations conceptual cueing Final meaning interpretation

Figure 1.  The overall architecture of the Lexical Constructional Model LT = lexical template; CT = constructional template; CS = Conceptual Structure.



Levels of description and explanation in meaning construction 

inquiry has been intense and active, thus offering a wide range of different methodological options for the researcher. As things stand, it might sound somewhat bizarre to offer a new linguistic product in a really saturated linguistic market. What are the added advantages that a new model can offer over the already existing proposals? In connection with this, the LCM emerges as an attempt to reconcile a spurious debate between the projectionist and constructivist accounts of language and reorient this cardinal theoretical issue within the larger context of a theory of language that aspires to provide a fully-fledged and fine-grained description of all aspects of meaning construction. If this is our goal, we certainly need to capture the most relevant elements of the projectionist and the constructionist accounts. But before we discuss what these elements are, let us briefly discuss some of the competing theoretical issues that have fragmented functional and constructionist approaches. This will serve as a backdrop against which we will briefly outline the architecture of the LCM. One of the central assumptions of many functional accounts of grammar is that the syntactic configuration of predicates can largely be determined on the basis of their argument structure, which is described in terms of logical configurations. For this reason, these functional accounts can be labeled “projectionist”, i.e. they claim that the logical structure of predicates constrains syntax by being mapped onto it. However, as has been extensively shown in Construction Grammar circles, there are instances where the argument structure of a predicate proves insufficient to explain the occurrence of one constituent. Consider the examples in (1) which exemplify the fact that constructions may contribute arguments to yield the final semantic interpretation of a particular expression:4 (1) a. Te quier-o fuera de mi vida. acc.2sg want-prs.1sg outside of poss.1sg life ‘I want you out my life’ b. Me gust-aría ver tu tesis dat.1sg like-conditional.1sg see.inf poss.2sg thesis

termin-ad-a cuanto antes. finish-ptcp-f.sg as much before



‘I would like to see your thesis completed as soon as possible’

c. Ver-emos su tesis termin-ad-a pronto. see-fut.1pl poss.2sg thesis finish-ptcp-f.sg soon ‘We will see your thesis completed soon’

.  Interlinear morpheme-by-morpheme glosses are supplied for the Spanish examples following the Leipzig Glossing Rules (see http://www.eva.mpg.de/lingua/resources/glossing-rules. ph). The following additional abbreviations will be used in this paper: conditional (conditional or potential verb tense), and reflex pass (reflex passive).

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

d. Te ve-o en la Moncloa acc.2sg see-prs.1sg in def.f.sg moncloa[name] en cinco año-s. in five year-pl ‘I see you in Moncloa in five years’

In (1a) the PP fuera de mi vida cannot be derived from the argument structure of the predicate ‘quiero’. In our view, (1a) exemplifies in Spanish the caused-motion construction, which has been studied for English (e.g. He sneezed the napkin off the table) in quite a lot of detail by Goldberg (1995, 2006). The caused-motion construction is used with causative and volition predicates and expresses a categorical and strong manipulation of the state of affairs denoted in the NP and the XPCOMP. In (1b) the resultative predicate terminada (completed) is not part of the argument structure of the predicate ver (see). In much the same way, in (1c) the complex resultative expression [en la Moncloa en cinco años], where ‘Moncloa’ is metonymic for ‘the Spanish government’, originates in a non-explicit action with an observable or imaginable endpoint (i.e. someone’s political career). In essence, the examples in (1) show that constructions should not be regarded as pure epiphenomena since they play an active role in determining the type of syntactic configuration in which a predicate occurs. As a matter of fact, the LCM agrees with this aspect of Godlberg’s Construction Grammar, thereby incorporating into its core-grammar level of description an inventory of argument constructions. But the LCM differs from Goldberg’s account in significant respects, especially the descriptive metalanguage, which resembles that used for lexical characterizations of verbal predicates, and the specification of the ways in which core-grammar argument constructions interact with lexical configurations and with other constructions, whether constructional or idiomatic, at other levels of description.

3.  The role of verbal semantics Lexical representation has become a crucial issue in linguistic theory, especially after the discovery that many of the syntactic properties of a predicate are in large part determined by its argument structure, a methodological stance that has become a hallmark for projectionist theories. This obviously contrasts with the ancillary status to which verbal semantics has been relegated in constructionist circles, where verbal semantics is not an area of emphasis.5 Consider the following projectionist representational

.  While not opposed to the study of verbal semantics, most constructionist approaches agree that constructions are better predictors of sentence meaning than verbal semantics (cf. Goldberg,



Levels of description and explanation in meaning construction 

format for three semantically close verbal predicates, viz. assassinate, murder, and kill (Dik 1997a: 101): assassinate [V] (x1: 〈human〉)Ag (x2: 〈human〉)Go ↔ murder [V] (x1)Ag (x2)Go (x3: treacherous [A]))Manner murder [V] (x1: 〈human〉)Ag (x2: 〈human〉)Go ↔ kill [V] (x1)Ag (x2)Go (x3: intentional [A]))Manner kill [V] (x1)Ag/Fo (x2: 〈human〉)Go ↔ cause [V] (x1)Ag/Fo (e1: [die [V] (x2))Proc ])Go

Each characterization is a predicate frame. For each argument of the predicate frame, the characterization specifies its selection restrictions (between angled brackets) and its semantic role (or function). Other distinguishing features are part of the definition part of the formalism. The explicit connection between the argument structure of the predicate and two syntactically relevant dimensions of semantic interpretation (selection restrictions and semantic functions) endows the representational format with a clear semantically-motivated syntactic projection potential. The rest of the features allow us to assign each item its due place in a lexical class. Consider now the different projectionist system of lexical representation provided by Van Valin & LaPolla (1997):

melt: shatter: destroy: break:

BECOME melted´ (x) INGR shattered´ (x) [do´ (x,φ)] causE [[do´ (y,φ)] causE [BECOME destroyed´ (z)] [do´ (x,φ)] causE causE [BECOME broken´ (y)]

These formalisms do not include selection restrictions or semantic functions. Instead, they make use of an abstract semantic metalanguage which consists of a number of primes (or constants) together with a list of operators and variables. Another variant of this line of research is represented by Levin & Rappaport’s (2005) event structure templates. As argued in Ruiz de Mendoza & Mairal (in press b), one of the problems of this approach is that these representations only capture those aspects of the meaning of a word that have syntactic projection, at the cost of ignoring relevant knowledge parameters that are also part of a speaker’s lexical competence. One further problem is the inability of these representations to account for many every-day uses of concepts that would require a broader definitional approach. Thus, there is no way in which representations like these can allow us to predict why break may be used intransitively

Casenhiser & Sethuraman 2005). Croft (2001) and Boas (2003) are exceptions. These two scholars vindicate the relevance of verbal semantics in the context of constructional accounts. The LCM takes sides with this position but also seeks to maximize it by endowing the LCM with a robust lexical component, as described in sections 4 and 5 below.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

(e.g. The window broke) but destroy may not (e.g. *The building destroyed). The predicates broken´ and destroyed´ have to be decomposed further, an issue that we shall deal with in section 7.2 below. Alternatively, cognitive linguists make use of encyclopedic information in the form of frames as developed in Frame Semantics and more recently in the Frame Net project.6 Semantic frames, which have been described as “specific unified frameworks of knowledge, or coherent schematizations of experience” (Fillmore 1985: 223), are schematic representations of situation types (e.g. ‘buying’, ‘drinking’, ‘reading’, etc.) describable in terms of participants and their roles (cf. Fillmore & Atkins 1992, 1994). Frames are very different from event structure templates or logical structures. Thus, frames do not include formal representations of the parameters that are determinant for argument realization. Instead, frames provide a comprehensive account of the conceptual framework underlying the meaning of a predicate, which can account for many aspects of its use. Thus, the commercial frame specifies a buyer, a seller, some merchandise, money, and a marketplace, among other elements. The ingestion frame specifies an ingestor, a thing that is ingested (an “ingestible”), an instrument, a place, a time, the manner of ingesting, and so on. The reading frame has, among others, slots for a reader, something that is read (e.g. a book), a place (e.g. a library, a study room). The advantage for grammatical explanation of this kind of account is that it deals with the different ways in which verbal predicates profile relations and how this affects grammatical realization. For example, pay relates to the part of the transaction frame that deals with the transfer of money from buyer to seller in order for the latter to receive goods. Not all these elements of semantic structure need to be realized in syntax: (2) a. He didn’t want to pay the car dealer (the agreed price) (for the car). b. He didn’t want to pay the agreed price (to the car dealer) (for the car).

It is possible to contrast the use of pay with other related verbal predicates instantiating the same frame in terms of how each profiles the same set of relations: (2’) a. He didn’t want to buy the car (from the car dealer) (for the agreed price). b. *He didn’t want to buy the car dealer. c. *He didn’t want to buy the agreed price.

While pay may take two different frame elements as its grammatical object, this is not the case for buy. The motivation for this difference can be attributed to the fact that the verb buy profiles a relation between the buyer and the goods bought (i.e. the focus is on the buyer obtaining the goods), but pay profiles a relation between the buyer and the seller in terms of the price paid for the merchandise (i.e. the focus is on the buyer and the seller being involved in an exchange of money for goods).

.  For an updated account of the FrameNet Project, see http://framenet.icsi.berkeley.edu.



Levels of description and explanation in meaning construction 

One insufficiency of this approach, however, lies in its lack of representational systematicity, since labels for semantic structure elements are stipulated in a rather ad hoc fashion without any standardized procedure. Thus, for the ingestion frame, we wonder why there is no specification of the way the actor and the affected entity interact: there is a causal connection that results in consumption of the affected entity, which will vary with the form of ingesting (eating, gulping, chewing, munching, drinking, sipping, devouring, consuming, gobbling, slurping, etc.). For example, one can eat a sandwich by taking small bites at it or by taking one or two big bites, but not by licking it. However, one can eat egg yolk this last way. The Frame Semantics system would need to postulate a virtually limitless number of subframes (eating a sandwich, a cake, ice cream, egg yolk, lamb, fish, rice, corn on the cob, etc.) for each way of ingesting food and make explicit the possible and impossible combinations between ways of ingesting and food types. Another problem for the Frame Semantics approach is the lack of clear mechanisms to predict different forms of syntactic realization for each element of semantic structure. In order to do so, it is necessary to do much more than simply spell out connections between frames and the verbal predicates realizing them. For example, for a verb like charge we can postulate its potential to realize that part of the commercial transaction frame where someone (a customer) pays an amount of money for some work (e.g. The mechanic charged me 60$ for just half an hour’s work). However, the person that receives the money is not a seller of goods but a service provider. Are we to understand the concept of ‘seller’ in the commercial transaction frame in a broad manner? If so, how do we make explicit the restrictions that charge places on the type of seller on the basis of Frame Semantics? We could postulate a ‘charging’ subframe, but then we would need to postulate subframes for just about any verbal predicate connected with trading goods or services (e.g. bargain, haggle, deal, traffic, auction, transact, swap, market, retail, vend, supply, etc.) and then formulate the various syntactic realization restrictions. Evidently, this step would draw Frame Semantics very close to a projectionist account of lexical semantics. This is a theoretical development that frame semanticists have never wanted to make,7 but in our view it is a necessary one unless we want encyclopedic semantics to remain largely dissociated from syntactic theory, which is the actual situation in Cognitive Linguistics, where the different versions of Construction Grammar and even Langacker’s Cognitive Grammar do not really incorporate it explicitly.8 There is thus an important mismatch within Cognitive

.  Advocates of Frame Semantics have explicitly written against the projectionist account of verb classes proposed by Levin (1993) (cf. Baker & Ruppenhofer 2003; Boas 2006). .  The focus of Langacker’s Cognitive Grammar is on cognitive phenomena such as construal, perspective, subjectivity and mental scanning (cf. Langacker 1987, 1991, 1999, 2005, 2008), and Construction Grammar mainly deals with the distinct grammatical properties of constructional

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

Linguistics between its theoretical emphasis on the encyclopedic conception of meaning and its actual incorporation into an account of grammar. The LCM is an attempt to solve this problem. In order to combine the two systems, i.e. constructionalism and projectionism, the LCM has developed a system of lexical representation in terms of lexical templates, which combine (encyclopedic) semantic and logical variables that are linked to one another in readiness for syntactic projection.

4.  Lexical templates The notion of lexical template is originally a development of the logical structures in Role and Reference Grammar (RRG) (cf. Van Valin & LaPolla 1997; Van Valin 2005, in press). RRG uses a decompositional system for representing the semantic and argument structure of verbs and other predicates (their Logical Structure, LS). The verb class ascription system is based on the Aktionsart distinctions proposed in Vendler (1967), and the decompositional system is a variant of the one proposed in Dowty (1979). Verb classes are divided into states, activities, achievements, semelfactives, and accomplishments, together with their corresponding causatives. In Table 1 we give a representation of each verb class with its corresponding formalism (cf. Van Valin 2005: 45). Table 1.  Verb classes in RRG Verb class

Logical structure

Example

Instantiation of LS

State Activity Achievement

predicate´ (x) or (x,y) do´ (x, [predicate´ (x) or (x,y)] INGR predicate´ (x) or (x,y), or

see´ (x,y) do´ (x,[run´ (x)]) INGR popped´ (x)

Semelfactive

INGR do´ (x, [predicate´ (x) or (x,y)] SEML predicate´ (x) or (x,y)

see run pop (burst into tears) glimpse, cough

SEML see´ (x,y)

Accomplishment

SEML do´ (x, [predicate´ (x) or (x,y)] BECOME predicate´ (x) or (x,y), or

receive

BECOME have´ (x,y)

Active  accomplishment

BECOME do´ (x, [predicate´ (x) or  (x,y)] do´ (x, [predicate1´ (x, (y))] & BECOME predicate2´ (z,x) or (y)

drink

do´ (x,[drink´ (x,y)]) & BECOME consumed’ (y)

Causative  accomplishment

α causES ß where α, ß are LS of any  type

kill

[do´ (x, Ø)] causE [BECOME [dead´ (y)]

patterns based on thematic roles. The need to incorporate a proper syntactic component has also been explicitly put forward in Boas (2008).



Levels of description and explanation in meaning construction 

RRG maintains that state and activities are primitives and thus form part of the logical representation of the other types of predicates; by way of example, an accomplishment is either a state or activity predicate modified by the telic operator BECOME. However, Van Valin & Wilkins (1993) and Van Valin & LaPolla (1997) all claim quite explicitly that state and activity atomic predicates need further semantic decomposition and thus provide a first approach for the predicate remember and speech act verbs respectively.9 In an attempt to provide logical structure with a more robust semantic decomposition, we decided to develop the notion of lexical template. A lexical template consists of two modules: (i) the semantic module, and (ii) the logical representation or Aktionsart module, each of which is encoded differently. Here is the basic representational format for a lexical template: predicate: [semantic module〈lexical functions〉 [aktionsart module 〈semantic primes〉]

The rightmost part of the representation includes the inventory of logical structures as developed in RRG with the proviso that the predicates used as part of the meaning definition are putatively candidates for semantic primes, or else, these cannot be further decomposed. This allows us to avoid the problem of having to regard as undefinable predicates which can be further semantically decomposed, e.g. defining the predicate redden in terms of BECOME red´, or popped in terms of INGR popped´, or activity predicates like sing or drink in terms of do´ (x,[drink´ (x)]) or do´ (x,[sing´ (x)]). The innovation here with respect to the original RRG proposal resides in finding a systematic procedure to identify the correct prime together with a uniform framework for decomposing every predicate semantically until we arrive at the undefinable elements. The semantic and pragmatic properties of the semantic module, as shown in the leftmost part of the representation, are formalized by making use of lexical functions such as those used in Mel’čuk’s Explanatory and Combinatorial Lexicology (ELC) (cf. Mel’čuk 1989; Mel’čuk et al. 1995; Mel’čuk & Wanner 1996; Alonso Ramos 2002).10 These lexical functions have also been shown to have a universal status

.  For a discussion of the exact details of the formalism of the first lexical templates, we refer the reader to Van Valin & Wilkins (1993), Van Valin and LaPolla (1997), Mairal & Faber (2002; 2007). .  According to Mel’čuk et al. (1995: 126–127), a lexical function (LF) is written as: f(x) = y, where f represents the function, x, the argument, and y, the value expressed by the function when applied to a given argument. The meaning associated with an LF is abstract and general and can produce a relatively high number of values; e.g. Magn expresses intensification and can be applied to different lexical units thus yielding a high set of values:

Magn Magn

(Engl. smoker) = (Engl. bachelor) =

heavy confirmed

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

(cf. Mel’čuk 1989), something which is in keeping with our aim of providing typologically valid representations. In contrast to the use of lexical functions in Mel’čuk’s work and the complete literature on the Explanatory Combinatorial Dictionary, in our approach such functions are essentially paradigmatic and capture those pragmatic and semantic parameters that are idiosyncratic to the meaning of a word, which allows us to distinguish one word from others within the same lexical hierarchy. For example, if we want to account for the semantic differences between mandar (‘command’), ordenar (‘order’), decretar (‘decree’), preceptuar (‘set up a precept’), preinscribir (‘preregister’) from the lexical domain of speech acts or cautivar (‘captivate’), arrebatar (‘seize’), arrobar (‘entrance’), embelesar (‘enrapture’), extasiar (‘send into an ecstasy’), hechizar (‘bewitch’) from the domain of feeling in Spanish, we would certainly need some mechanism that allows us to discriminate and encode those meaning elements that differentiate one predicate from others. Then, we have devised a semantic module that consists of a number of internal variables, i.e. world knowledge elements of semantic structure, which relate in very specific ways to the external variables that account for those arguments that have a grammatical impact. Now, let us consider the following examples: captar: [MagnObstr & Culm12[ALL]] know´ (x, y) x = 1; y = 2 consider: [locin temp↔1,2 Cont] think´ (x, y) x = 1; y = 2

In Spanish captar (‘grasp’) the logical structure represents a state predicate with know´ as primitive and modified by two variables x and y. On the right hand side, the semantic module includes two parameters that encode the culmination of knowing the propositional meaning of something [Culm12[ALL]] and the fact that this process has been done with great difficulty [MagnObstr]. Then, in consider, we have a two place state predicate with the primitive think´ as definiens. Within the lexical domain of cognition, Faber & Mairal (1999) note two types of indefinables, know and think, that serve to define the rest of the predicates in this class. The case that concerns us here belongs to the think type and is modified by two lexical functions, [locin temp↔1,2] and [Cont], that express duration and temporal setting (in the present).11



Magn Magn

(Sp. error) = (Sp. llorar) =

craso llorar como una Magdalena

.  A further issue that arises is if lexical functions provide a complete catalogue of the different semantic and pragmatic parameters that are involved in the meaning definition of a word. Moreover, it would be desirable to regroup lexical functions into broader categories such that the inventory becomes more transparent. Within this context, a very recent development proposes to readapt the semantic modules of the lexical templates in terms of Pustejovsky’s (1995) qualia (cf. Mairal & Cortés, in prep.; Cortés, this volume).



Levels of description and explanation in meaning construction 

Lexical templates are built into constructional templates, which have a more abstract nature and are largely based on Goldberg’s (1995) notion of argument constructions. Argument constructions are associated with Level 1 constructional templates in the LCM, which form the core grammar module and are made up of elements of semantic interpretation that can be realized syntactically. Hence, constructional templates thus consist of sets of arguments that relate to one another on the basis of abstract predicates such as causE, BECOME, MOVE, and HAVE. In the LCM we have adapted these representations to the same formalism as used in a lexical template, which means that constructional templates at this level, which are largely grounded in an Aktionsart characterization, are described by using a semantic metalanguage and an inventory of operators. As a result, both lexical templates and constructional templates are based on the same metalanguage, which makes the unification of the two formalisms a straightforward task. For example, let us see the format of the causedmotion construction: a. [do´ (x, y)] causE [BECOME *NOT be-LOC´ (y, z)] b. [pred´ (x, y)] causE [BECOME *NOT be-LOC´ (y, z)]

Although this construction is not very productive in Spanish, there are just a few examples like (3), where the causing subevent is a state predicate:12 (3) Juan lanzó la pelota a través del campo. John throw-indefpret.3sg def.f.sg ball across.def.m.sg field ‘John threw the ball across the field.’ [do´ (John, ø)] causE [BECOME be-LOC´ (pelota, campo)]

Recall that, as discussed in Goldberg (1995), and in Van Valin & LaPolla (1997), the caused-motion construction, which might be well equally interpreted as a case of the resultative construction, introduces another logical structure that focalizes a further degree of the result obtained by the state part of the lexical template. This subevent is saturated by either an adjectival or a prepositional phrase.

5.  The possible relations between verbs and constructions Construction Grammar approaches have postulated very general principles to account for the possible relations between verbs and constructions: for example, the Semantic Coherence Principle, the Correspondence Principle, the Causal Relation Hypothesis

.  For a representation of the inchoative construction and the subjective transitive construction, we refer the reader to Cortés (this volume) and Gonzálvez-García (2008).

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

(Goldberg 1995:  50; 61, 62), or ‘coercion’ and its associated principle the Override Principle (Michaelis 2003). However, these principles are too general to account for the specificities and exact conditions that actually regulate the integration of a verb’s lexical entry with the construction in which it is embedded. The Semantic Coherence Principle simply states that the participant role of the verb and the argument role of the construction must be semantically compatible. The Correspondence Principle specifies that semantically salient roles must be encoded in such a way that they receive sufficient discourse prominence (that is, the principle ensures that lexical specifications and discourse structure are generally aligned). The Causal Relation Hypothesis, based on previous work by Croft (1991), states that verbal and constructional meaning must be integrated via a temporally contiguous causal relationship (e.g. sound emission verbs, such as screech in The train screeched into the station can be used to designate motion if the sound results from and occurs simultaneously with motion; cf. *The dog barked into the room). Finally, the Override Principle stipulates that the meaning of a lexical item must conform to the meaning of the structure in which it is embedded. For example, in Spanish (as in English) we find that the predicates romper (‘break’), cortar (‘cut’) and destruir (‘destroy’) exhibit distinct syntactic behavior in the context of the inchoative construction and the impersonal reflexive construction (see Cortés, this volume): (4) a. Romp-ieron la ventana (con facilidad). break-indefpret.3pl def.f.sg window with ease ‘They broke the window easily’ b. La ventana se romp-ió (con facilidad) def.f.sg window reflex pass break-indefpret.3sg with ease por sí mism-a]. by itself-f.sg



[como as.if



‘The window broke easily (as if by itself)’

c. Destruy-eron el edificio. destroy-indefpret.3pl def.m.sg building ‘They destroyed the building’ d. *El edificio se destruy-ó [como def.m.sg building reflex pass destroy-indefpret.3sg as.if

por sí mism-o] by itself-m.sg



‘The building destroyed itself (as if by itself)’

e. Est-e pan se prox-m.sg bread reflex pass ‘This bread cuts easily’

cort-a (con facilidad). cut-prs.3sg with ease



Levels of description and explanation in meaning construction 

f. #Est-e pan se prox-m.sg bread reflex pass *‘This bread cuts itself ’

cort-a. cut-prs.3sg

We claim that it is necessary to identify the type of elements that are part of the structure of verbs like romper or cortar, which allow the impersonal reflexive in (a) and the evaluative reflexive construction in (b). In much the same way, it is necessary to determine why a predicate like destruir blocks out a reflexive construction as shown in (d). The LCM posits two major mechanisms that account for the possible relations that are established between a verb and a construction, on the one hand, and between a constructional leveln and a higher constructional leveln+1, on the other hand. One of the mechanisms, which works purely at the grammatical level, is the one we have referred to above by the label subsumption. The other mechanism, cueing, deals with inferences developed on the basis of the blueprint provided by the output of lexical and constructional integration at whatever level of representation. The two mechanisms are in turn regulated by what we have termed internal and external constraints (cf. Section 7).

6.  The pragmatic and discourse dimensions of constructional meaning Functional approaches have been for some time concerned with the nature of the pragmatic and discourse dimensions of meaning. The goal of giving pragmatic adequacy to grammatical explanation influenced much of Dik’s Functional Grammar, which also addressed some discourse phenomena such as topicality and focalization (cf. Dik 1997b). The more recent Functional Discourse Grammar (Hengeveld 2004; Hengeveld & Mackenzie 2006, 2008) has gone so far as to take the discourse act as the basic unit of analysis. In Cognitive Linguistics, the concern with pragmatics and discourse has guided an increasing amount of research (e.g. Langacker 2001; Otal & Ruiz de Mendoza 2007; Panther & Thornburg 2003; Ruiz de Mendoza & Pérez 2003; Steen 2005) but, in contrast to Functional Grammar, neither dimension has been made a central part of any of the existing accounts. The tendency in CL is to show that pragmatic and discourse phenomena can be accounted for by making use of the same mechanisms used for other domains of linguistic enquiry. Thus, Langacker (2001) applies the notions of profile and base, which are pervasive in grammar, to an understanding of the relationship between finite clauses and the context in which they are produced. He also contends that the basic units of discourse structure are relationships (propositions) rather than things (nominal expressions), another basic division of his Cognitive Grammar. But this treatment of discourse makes it epiphenomenal, the result of opening up attentional frames that often correspond to clauses that one after another serve to create and update a discourse space. There is no discussion of how

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

some linguistic mechanisms serve as cues to creating conceptual connectedness, nor is there any indication of what principles regulate the creation of discourse structure. In other words, discourse meaning is not made part of a unified framework. The LCM does aim to offer a unified approach to the pragmatic and discourse dimensions of meaning construction. In order to do so, it must accomplish three crucial tasks: one is to distinguish between different levels of non-argumental meaning; a second task consists in finding to what extent a non-argumental level of description is to be accounted for in terms of grammatical mechanisms, especially since much of the meaning at the pragmatic and discourse levels is obtained inferentially; the third task is to determine the principles that regulate interaction between the different layers. First, since the LCM has a constructional orientation, it focuses its attention on postulating constructions that capture in a highly conventionalized way different layers of non-argumental meaning. In the LCM argument constructions are considered part of the core-grammar level of description, or level 1, where content-bearing lexical items and predicate-argument constructions interact to produce what is normally called propositional meaning. Then, the LCM distinguishes between three kinds of non-argumental constructions: (i) level 2 or implicational constructions, which capture meaning that arises from the way the speaker interacts with the lexical and grammatical properties of utterances but does not affect the basic relationship between predicates and their arguments; (ii) level 3 or illocutionary constructions, which deal with meaning that arises from the way speakers interact on the basis of argumentpredicate configurations; (iii) level 4 or discourse constructions, which deal with how the speaker creates connectedness in his speech production on the basis of all other aspects of the semantic configuration of utterances. One interesting feature of levels 2, 3, and 4 constructions is their higher degree of idiomaticity when compared to level 1 constructions. In the case of level 1 constructions, we sometimes come across expressions that are either fully idiomatic, i.e. that admit a very low degree of elaboration or none at all, such as kick the bucket, spill the beans, beat about the bush, bear the brunt of, let the cat out of the bag, walk the second mile, and others that are midway between being idiomatic and being argumental: look someone in the eyes, make oneself at home. But even the more flexible level 1 idiomatic constructions do not match the degree of elaboration that constructions at other levels can reach. This is so because these higher-level constructions combine idiomatic elements with highly parametrizable variables. We deal with levels 2, 3, and 4 constructions in the following subsections.

6.1  Implicational constructions Let us consider the following utterances: (5) a. Who’s been messing with my computer? b. It was your son that broke the window.



Levels of description and explanation in meaning construction 

Example (5a) is a case of a presuppositional construction and (5b) of a focus construction. Let us deal with each construction in turn. In (5a), the speaker is bothered by the situation described at the argument structure level through a presupposition (i.e. someone has been messing with the speaker’s computer). The idea that the speaker is bothered is not directly derivable from the presupposition, but rather from an implication based on the fact that no one is expected to use another person’s computer without permission. But note that a similar implication may also be obtained from other related sentences that make use of the same structural configuration: (6) a. Who’s been crying the whole night? b. Who’s been reading my journal? c. Who’s been reciting Homer? d. Who’s been sitting here?

This happens because the pragmatic implication that the speaker is bothered by the situation described in the propositional structure of wh-sentences has become “entrenched” to a fairly large extent in the linguistic system. Entrenchment, as defined by Langacker (1999:  105), is a function of the frequency of association between a meaning element, whatever its origin, and any given structural configuration. There are other related configurations that tend to carry the same meaning implication that we have observed for example (5a) and all the examples in (6): (7) a. Where’s our son been the whole night? b. Why’s he behaving like that? c. When’s that order been issued? d. What’s your sister been doing today?

All these configurations presuppose that whatever is described after the wh- interrogative pronoun has actually been the case. But we can have the same meaning effect even if we make use of a non-presuppositional construction. The clearest case is provided by the What’s X Doing Y? construction, discussed by Kay & Fillmore (1999): (8) a. What’s the child doing in the garden? b. What’s your sister doing in the theatre? c. What’s your mother doing in the kitchen? d. What’s the Prime Minister doing in China?

In all the examples in (8) there is a situation simultaneous with the time of utterance. In contrast with the presuppositional examples (5a), (6a–d), and (7a–d) the actuality of the situation is not taken for granted on the basis of a level 1 property of the construction, but rather on the basis of a default assumption whose origin is to be found in a potential pragmatic implication. If any of the sentences of the form ‘What’s X Doing Y?’ is produced in a context in which it is evident to the speaker that the hearer is already aware of the situation, the only way to make them relevant is to shift the

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

meaning interpretation from one where the hearer is being required to describe the situation to one in which the hearer needs to determine how the speaker feels about the situation. In this connection, note that sentences exploiting this construction are often interpreted as forms of calling the hearer’s attention to a negative situation about which the hearer (or even someone else) would have been expected by the speaker to do something or as ways of complaining about the situation if fixing it is beyond other people’s ability. Compare: (9) a. The child is up to something in the garden. b. Your sister is again in the theater with that boyfriend of hers. c. The doctor told you your mother should have full rest. d. The Prime Minister was not supposed to go to China.

Following the interpretive rationale specified above, sentences (9a), (9b), and (9c) would have a default interpretation as requests for remedial action. Sentence (9d) would be a complaint. However, we may find contexts where the first three sentences are either pure complaints where the speaker does not expect the hearer to act in any way or a conceptual combination of complaining and requesting remedial action. For example, for (9b), it is possible to think of a context where the speaker feels that the hearer has no possibility of doing anything to prevent his sister from dating her boyfriend. Then, (9b) would be understood as a complaint and nothing else. But if the speaker feels the hearer could stop the situation, then (9b) would also be a request for action. These observations will be taken up again later on, when we discuss level 3 meaning constructions. For the time being, it is sufficient to note that the call for action and its associated complaint interpretation are only possible because of a highly conventionalized (or entrenched) meaning implication that is produced at level 2 through a constructional mechanism. We also observe that sentence (5a) and the sentences in (6) and (7) above give rise to similar level 3 interpretations. A different case of level 2 construction is provided by focus constructions, like (5b). Focalization of a level 1 constructional element is a widespread phenomenon in many languages. Dik (1997b) has discussed what he calls focus constructions in a number of unrelated languages. Focalization can be achieved through a variety of linguistic mechanisms involving cleft structures, intonation, accentual prominence, and reduplication. Each mechanism has the function of presenting part of the information of the sentence as more relevant than the rest either because it is new or because the speaker thinks it is worthy of note for the hearer. For example, in (5b) we can have a situation where both speaker and hearer are aware that the hearer’s son broke the window. In this context the piece of information that receives prominence is intended to emphasize the hearer’s son’s responsibility and probably the hearer’s own responsibility to deal with the situation as needed on the basis of the established socio-cultural patterns. But the same information could be totally new to the hearer, in other contexts, and be therefore



Levels of description and explanation in meaning construction 

taken as an initial call of the hearer’s attention to deal with the situation. Since focus constructions are forms of restructuring level-1 information in such a way that it shows how the speaker interacts with it, the LCM deals with them at level 2.

6.2  Illocutionary constructions Level 3 constructions give rise to conventionalized illocutionary meaning. Here the LCM departs rather drastically from functional models such as those propounded by Halliday (1994) (see also Halliday & Matthiessen 2004) and Dik (1997b) and comes somewhat closer to some proposals within Cognitive Linguistics, especially Stefanowitsch (2003), Thornburg & Panther (1997), Panther & Thornburg (1998, 2003), and Panther (2005), although there are still some crucial differences. Halliday & Matthiessen (2004: 107) do not think of speech act meaning or illocutionary meaning as a meaning dimension apart from grammar. In their view, the clause can be seen as a grammatical unit that combines meanings of three different kinds: ideational, interpersonal, and textual, each of them corresponding to one of the general functions of language. Each meaning dimension presents the clause from a different perspective: as representation (ideational meaning), as exchange (interpersonal meaning), and as message (textual meaning). Speech act meaning is thus treated – with mood, polarity, and modality systems – as a part of the clause-as-exchange dimension of grammar. More specifically, it is seen as a matter of giving or demanding either information or so-called “goods-&-services”. This activity results in four basic speech functions and their associated responses (see Table 2). Table 2.  Primary speech functions and their associated responses

Giving information Giving goods-&-services Demanding information Demanding goods-&services

Primary speech functions

Expected response Discretionary response

Stating John will come Offering Shall I carry your bag? Questioning What did John buy? Commanding Wash the dishes!

Acknowledging Will he? Accepting Yes, please, do! Answering A new car Undertaking Sure!

Contradicting No, he won’t Rejecting No, thanks Disclaiming I couldn’t tell Refusing I won’t

Halliday & Matthiessen’s analysis is faced with two important challenges. First, while it is easy to accept that we can make use of language to give or demand information and to demand “goods-&-services”, it is not at all clear in what way we can give “goods-&-services” by means of language. Rather, what we do is indicate that we have the desire or the willingness to give “goods-&-services”, as is the case with offers and

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

promises. The system is therefore not as symmetrical as proposed by Halliday and Matthiessen.13 Second, there is no indication in this account of how we can deal with non-primary speech functions (requesting, begging, promising, warning, threatening, condoling, boasting, among many others that have been discussed in the traditional speech act literature) and their associated responses, which are not necessarily the same as the ones proposed for the primary speech functions. For example, while it might be argued, following Halliday and Matthiessen’s rationale, that promising is, like offering, a way of “giving goods-&-services”, it would not be reasonable to argue that we can “accept” a promise. Promising is more naturally followed by an expression of gratitude as a response: (10) A: OK, you shall have the new bicycle you wanted, I promise B: Great! Thank you so much! (cf. # Yes, please, do!).

Note that when we find an expression of acceptance following a promise, the scope of the response is not the actual promise but the future action involved in it: (11) A: OK, I’ll buy you the bicycle you wanted, I promise. B: Yes, please, do!

B’s response in (11) is not an expression of acceptance, but a form of encouraging his interlocutor to act as specified in his speech turn, i.e. to live up to his promise. Or consider expressive speech acts such as congratulating, thanking, apologizing, condoling, and boasting. They are forms of giving information while expressing the speaker’s attitude to it. However, the expected responses are not generally those of acknowledgement or contradiction: (12) A: You gave a great talk. B: Thank you (#Yes, I did) (#No, I didn’t). (13) A: We are grateful that you are serving your country. B: It’s an honor (#Yes, you are/#Yes, I am) (#No you’re not/#No, I’m not). (14) A: I’m sorry I forgot to call you. B: That’s fine. Don’t worry (#Yes, you are/#Yes, you did) (#No, you’re not/#No, you didn’t). (15) A: I’m sorry your mother died. B: Thank you (#Yes, you are/#Yes, she did) (#No, you’re not/#No, she didn’t). (16) A: I’m the best in town. B: Yes, you are/No, you’re not.

.  A similar point is found in Fawcett (1980), who remarks that offers are normally realized nonlinguistically but are often accompanied by an information-giving utterance.



Levels of description and explanation in meaning construction 

Except for the act of boasting in (16), which can be acknowledged or contradicted, the rest expect responses to the first speaker’s attitude in the form of thankfulness (for congratulating, condoling) or acceptance (thanking, apologizing). As is evident from the odd responses between brackets, expressive acts are not very sensitive to responses based on the non-attitudinal content of the initiating turn. In fact, such responses, when they take place, break politeness expectations. Thus, in (12), the acknowledging answer (Yes, I did) is a form of inappropriate boasting, and the contradicting response (No, I didn’t) of humility. In (13), we can acknowledge or contradict the content of the matrix or of the subordinate clause. If the response refers to the matrix clause, which carries the explicit illocutionary predicate, the meaning effect is one of pragmatic inadequacy: Yes, you are is likely to be interpreted as ironical (the second speaker believes the opposite of what he says) as a way of resolving the absurdity of reasserting the truthfulness of the first speaker’s expressed attitude; No, you’re not is offensive since it directly contradicts the sincerity condition of the explicit illocution. Similar considerations hold for (14) and (15), where reaffirming the matrix clause may readily be taken as ironical and negating the matrix clause suggests insincerity. If, on the other hand, the response refers to the subordinate clause, the expression of acknowledgement could be acceptable as a form of self-assurance, while the negation could be taken as a rather inappropriate way of showing humility or as a (probably impolite) contradiction of the reason for the expression of thankfulness. Here, examples (14) and (15) work differently. Thus, in (14) reaffirming the subordinate clause of an explicit apology (Yes, you did) is a way of manifesting that the apology may not be accepted. This is so because of the conceptually iterative character of the reaffirmation: if an apology is accepted, there is forgiveness, but reminding the offender again about his offence suggests lack of forgiveness. In condolences, however, reaffirming the subordinate clause is tantamount to a reiteration of the conditions that cause sorrow, which may suggest that the expression of sympathy will not be useful to soothe the emotional pain or that there is no such pain, so the condolences are not really necessary. In turn, negating the subordinate clause of an apology directly renders the apology unnecessary, since it is the same as saying that the conditions that motivated the apology are not valid. The same applies to the negation of the subordinate clause in (15), where the speaker rejects the validity of the conditions that would call for condoling. Since the range of possible responses to variants of the primary speech functions is very complex, Halliday and Matthiessen’s account needs to be improved. In our view, a possible improvement is found in the account by Dik (1997b), who observes that, according to typological data (Sadock & Zwicky 1985), most languages code four basic speech acts, i.e. statements, questions, commands, and exclamations. Except for exclamations, each basic speech act corresponds to one of the widely recognized sentence types: declarative, interrogative, and imperative. Exclamations are obtained by applying special suprasegmental features to sentences of any of the three basic types.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

Human languages then have special grammatical mechanisms to derive other illocutions from the basic ones. In English, it is possible to convert declarative, interrogative, and imperative sentences into requests by adding please: Please, I can’t stand that noise!; Pass me the salt, please!; Can you please swim?. Declaratives can be converted into questions by means of a tag: She’s the best in class, isn’t she? An imperative can become an exclamation through the adequate combination of stress and intonational features: See who’s COMING! Dik also admits the existence of non-grammatical, pragmatic mechanisms to derive illocutionary meaning, which explain, for example, why I’m thirsty is likely to be interpreted as a request for water. However, there are two problems with the idea of derived illocutions. One has to do with the fact that the degree of necessity of a conversion device like please in order to disambiguate a given structural configuration can vary. Thus, please is more necessary in (17a) than in (17b) below for a request interpretation. And sometimes, for the same structural configuration (e.g. a Can You sentence), the conversion device cannot be used, as in (17c). (17) a. Can you swim? > Can you swim, please? b. Can you listen to what I’m saying (please)? c. Can you see the island from here? > *Can you please see the island from here?

The other problem is related to the fact that many non-basic conventional illocutions seem to be obtained directly, without any derivational (whether grammatical or pragmatic) activity: (18) a. Can’t you be quiet for a minute? b. Won’t you help me at all? c. Shall I put the light on? d. You shall leave tomorrow. e. Why don’t you buy that book? f. You shall have all the benefits listed below.

These observations strongly suggest that grammar is capable of coding speech act meaning directly by developing entrenched meaning-form associations to that effect. For this reason, the LCM postulates the existence of level-3 constructions capturing conventionalized illocutionary meaning. In the LCM a Can You structural configuration is potentially ambiguous between two interpretations, one as a question about ability and the other as a request for action. Each interpretation crucially hinges upon the activation of a different construction: the polar interrogative construction, whose Aux-NP constituents can be realized (and thus parametrized) by can you; the Can You request construction, where can-you is idiomatic. In Dik’s approach adverbs like please or kindly are illocutionary converters. For example, please can convert any basic illocution into a request (Dik 1997b: 246): (19) a. Can you pass me the salt, please? (Interrogative > request) b. Please, John, stop tickling me! (Imperative > request) c. Please, John, it’s broad daylight! (Declarative > request)



Levels of description and explanation in meaning construction 

In our approach, devices like please and kindly are optional elements of the request construction and have the function of mitigating the directive force of the construction or of urging the addressee to act in the way described by its non-idiomatic elements. Mitigation effects can also be captured through other mechanisms, like the substitution of could or would for can in (19a), which has no effect on the overall request value of the resulting utterances (cf. Would/could you pass me the salt?), but simply affects their degree of politeness and/or formality. Since idiomatic constructions admit a degree of variation in their non-parametrizable elements, we consider will you/would you/could you/do you think you could forms as varieties of the same construction. In general, we will not postulate the existence of a completely different illocutionary construction provided that the variation of the idiomatic elements in it is not open-ended and that the variation does not bring about changes in the illocutionary status of the configuration.

6.3  Discourse constructions Discourse constructions capture the meaning implications of discourse relations, i.e. those relations that underlie discourse coherence. Discourse coherence is a very complex phenomenon all of whose intricacies can hardly be addressed in a short space. The phenomenon involves a rich network of principled connections between the various levels of linguistic analysis, thus accounting for the potential of some linguistic mechanisms to organize discourse while specifying in what ways discourse activity may constrain certain semantic and pragmatic choices. Although the LCM takes into account these connections (cf. Otal & Ruiz de Mendoza 2007), in this section we will restrict our discussion to the question of how we interpret the meaning of a sentence in the light of the meaning implications (at whatever level) of other sentences within the same communicative event. This understanding of discourse activity is just a development of the basic insights provided by early discourse analysts such as Winter (1982), Hoey (1983), and Longacre (1972) on so-called clausal or interclausal relations. The difference is that, in the LCM, discourse connections go beyond what is signaled through linguistic mechanisms. Thus, the model admits, as we have noted in previous sections, the possibility of interpreting meaning at any level through linguistically-cued inferences. At the level of core grammar, linguistic cueing results in explicature derivation. At the other levels it results in inferred implicatures, illocutionary values, or discourse relations. As was mentioned in the introduction section, the LCM focuses much of its attention on finding out what the different levels of linguistic description and explanation have in common. Many of the meaning relations at the discourse level can be found at other levels. This observation is not entirely new. Beaugrande (1980:  37) made a similar point with respect to cause-consequence relations, which can be inferred as in (20a) below or made explicit as in (20b): (20) a. Peter burned the book. He didn’t like it. b. Peter burned the book because he didn’t like it.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

In this view, the semantic classification of complex sentences would be a reasonable guide to a classification of discourse relations of coherence. Here, we find Halliday & Matthiessen’s (2004: 378) distinction between elaborating, extending, and enhancing – the three logico-semantic relations in the clause complex – an adequate starting point. In elaboration, one clause expands another by addressing part of it in greater detail (e.g. restating, commenting, specifying, or exemplifying the relevant part). In extension, a clause is expanded by another clause that adds a new element, gives an exception, or offers and alternative. In enhancement, a clause is expanded by qualifying it with some circumstantial feature of time, place, cause, or condition. These three relations are general labels for more specific clause-complex relations that have been discussed by Halliday and Matthiessen in some detail. In our view, many, if not all, of these relations parallel discourse relations. For the purpose of illustration we supply in Table 3 a tentative list of discourse relations that can be related to some of the categories discussed by Halliday and Matthiessen.14 The list of relations in Table 3 is by no means exhaustive. More relations can be added and some of the relations that have been identified can be further refined or subdivided. For example, we may have contrastive relations, usually signaled by however, but, nevertheless, which are a form of extension: John was born in New York; his wife is from Texas (cf. John was born in New York but his wife is from Texas). We can also treat some forms of implicit comparison as cases of extension: He is careless; his partner is careless too (cf. He is careless, and so is his partner). And we may distinguish between at least two kinds of alternation relation: one where we have contrast (contrastive alternation), as in Either you win or you lose; another one, where the two alternates are not necessarily antithetical but can be complementary (complementary alternation): No one insulted him or did physical harm to him. Finally, as recognized by Halliday & Matthiessen (2004: 411) themselves, time relations can be of at least two kinds: temporal sequence (First he killed the cat, then he buried it), or temporal overlap (Tony proposed to her while she was on her knees cleaning a toilet). As is evident from Table 3, discourse relations can be signaled linguistically or left for the addressee to infer. In the first situation, we have discourse constructions, while in the second we have cued inferencing, just as is the case with the rest of descriptive/ explanatory levels of our model and as should be expected on the basis of the equipollence hypothesis. It must be borne in mind that discourse constructions vary in their degree of specificity. A case in point is the well-known X Let Alone Y construction, discussed by Fillmore, Kay & O’Connor (1988). In our view, this construction is a

.  For a more fine-grained analysis of discourse semantic relations from a systemic perspective see Martin (1992); see also Butler (2003, part 2, chapters 3 and 4) for an overview of this area of enquiry.



Levels of description and explanation in meaning construction 

Table 3.  Some discourse relations General Specific semantic semantic relation relation Elaboration

Extension

Enhancement

Examples Implicit

Explicit

She has the right work She has the right work experience and skills. She can experience and skills; in do the job. other words, she can do the job. Comment He spoke calmly; she was He spoke calmly, which surprised. surprised her. Specification We all knew something: the We all knew that the car had car had been stolen. been stolen. Exemplification Big companies are growing Big companies are growing bigger; witness the number of bigger, as is evidenced by the mergers. number of mergers. Addition The bride wore a handThe bride wore a handknitted gown; she also carried knitted gown and carried a a woolen bouquet. woolen bouquet. Exception Chuck Norris does not Chuck Norris does not believe there should be any believe there should be any action heroes; he is the only action heroes except for exception. himself. Alternation Perhaps she didn’t know how Either she didn’t know how to reach out; perhaps she to reach out or she didn’t didn’t really love me. really love me. Time He had surgery; then he After he had surgery, he checked himself out of the checked himself out of the hospital. hospital. Location We found the mummy; it was We found the mummy just inside a huge chamber. where the old manuscript said it was located. Cause They were pleased; the engine They were pleased because worked well. the engine worked well. Condition You can have the day off You can have the day off tomorrow. But you have to tomorrow on condition agree to work on Saturday. that you agree to work on Saturday. Restatement

specific case of the complementary alternation construction involving two nonantithetical alternates (which in practice may be argued to shade off into cases of addition). Compare: (21) a. I won’t eat that garbage; and I won’t pay for it. b. I won’t eat that garbage, nor pay for it. c. I won’t eat that garbage, let alone pay for it.

For example (21a) there is a strong default interpretation: the speaker dislikes the food that someone is trying to sell him and is not willing either to eat it or pay for it. Example

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

(21b) has the same interpretation, but makes explicit the fact that the two clauses are complementary alternates through the use of the complementary alternation construction (not X nor Y). Now, in many contexts both (21a) and (21b) may carry an additional implication: the idea that the speaker finds it outrageous to pay for such as bad quality food as he is being offered. In connection with this implication, there is still another one: that he is much less willing to pay for the food than to eat it. These two related implications are made explicit in (21c) through the use of the ‘X Let Alone Y’ construction. Since this construction conveys a broader range of meaning implications than the complementary alternation construction, we consider it a specific case of the latter. According to Fillmore, Kay & O’Connor (1988), in the ‘X Let Alone Y’ construction, there is an entailment relationship between the elements X and Y such that X either expresses or implicates a negative situation and Y is considered less likely to happen than X. Since the constraints on the X and Y elements do not restrict their realization to a closed class of items (i.e. the number of possible instantiations is limitless), X and Y are constructional variables. In contrast, the coordinating conjunction joining X and Y has to be chosen from among a closed set of options, such as much less, not to mention, and never mind. The coordinating conjunction is thus a non-variable element of the basic ‘X Let Alone Y’ construction. Note that the use of one or another conjunction does not affect the meaning entailments described above: (22) a. I won’t eat that garbage, much less pay for it. b. I won’t eat that garbage, not to mention pay for it. c. I won’t eat that garbage, never mind pay for it.

Second, observe that there is a construction that is apparently the converse of ‘X Let Alone Y’. Consider: (23) a. He isn’t a skilled worker; he’s not even an apprentice. b. He isn’t an apprentice, let alone a skilled worker.

The contrast between (23a) and (23b) points to a preservation of the general constructional constraints on the X/Y variable elements of the ‘X Let Alone Y’ configuration in Y not even X: in both constructions X is negative and Y is less likely to be the case than X. However, there is a clear difference between the constructional presuppositions: in (23a) the emphasis is on the claim that the protagonist’s skills are worse than those of an apprentice; in (23b) the emphasis is on negating the presupposed addressee’s assumption that the protagonist could be a skilled worker. This difference explains why it is odd to convert any of the examples in (22) into a ‘Y not even X’ format: (24) #I won’t pay for that garbage, not even eat it.

The reason for this oddity is to be found in the fact that if the speaker-protagonist is not willing to pay for the bad-quality food, it is only natural that he will not want to



Levels of description and explanation in meaning construction 

eat it. So, putting the emphasis on the natural consequence, which is already implied by the first clause, is absurd. Since the meaning implications of (23b) differ to a significant extent from those of the ‘X Let Alone Y’ configuration and its variants, we will assign Y not even X an independent constructional status. However, since there is a degree of overlap between the two constructions and both can be considered specific cases of the more generic complementary alternation construction, we will regard ‘X Let Alone Y’ and ‘Y not even X’ as sister constructions. Another discourse construction that deserves some attention because of its large degree of specificity is Just Because X Doesn’t Mean Y or its variant Just Because X Is No Reason to Think Y (Holmes & Hudson 2000; Bender & Kathol 2001; Weilbacher & Boas 2006). This construction is a specific case of the general causative construction that parametrizes evidence-conclusion relations. Let us compare the following sets of sentences: (25) a. Just because it’s digital, doesn’t mean it sounds good. b. It is digital, but this doesn’t mean it sounds good. c. It doesn’t have to sound good simply because it is digital.

The context for the examples in (25) is a warning about the loss of audio quality when songs are converted from CD to mp3 format. The speaker presupposes that the hearer believes that digital formats for sound are better than more traditional audio formats and casts doubt on that assumption. The ‘Just Because X Doesn’t Mean Y’ construction, used in (25a), indicates that the second constituent (captured by the Y variable) does not necessarily follow from the first. This meaning effect can also be obtained by a discourse implication, as in (25b), or by expressing the central constructional meaning effect explicitly through a more general causative construction, as in (25c). There are other ways of parametrizing evidence-conclusion relations that make use of the discourse connectors so and after all: (26) a. It is digital; so it must sound good. b. It sounds good; after all, it’s digital.

The X So Y and Y After All X patterns are converses of each other. The former expresses the evidence first and then the conclusion that follows. The latter expresses the conclusion and then the evidence on which it is based. The meaning effects associated with each pattern are slightly different. Thus, (26a) gives prominence to the conclusion, while (26b) emphasizes the evidence. The same effects hold in the case of the negative counterparts of (26): (27) a. It can’t sound good; after all, it’s not digital. b. It is not digital; so it can’t sound good.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

Like the examples in (26), (27a) and (27b) explicitly deny the presupposition that we find in (25a) that the speaker believes that digital formats are better than traditional analogical audio. As in other cases, these discourse connections may be obtained inferentially, as in (28a) and (28b) below: (28) a. It can’t sound good; it’s not digital. b. It’s not digital; it can’t sound good.

These are examples of cued inferencing at the discourse level. Note that the general default use of can’t as a modal auxiliary indicating a logical deduction exhibits a strong potential to act as a cue for the preferred interpretation of the whole utterance in terms of the evidence-conclusion pattern.15 Note that without this level-1 grammatical pointer, the inferential process would be less constrained from a discourse perspective: (29) a. It doesn’t sound good; it’s not digital. b. It’s not digital; it doesn’t sound good.

The examples in (29) work better as ways of realizing a general cause-effect relation than an evidence-conclusion pattern: (30) a. It doesn’t sound good because it’s not digital. b. It’s not digital, that’s why it doesn’t sound good.

7.  C  ognitive constraints on meaning construction: Cueing and subsumption Meaning construction in the LCM revolves around two key processes, cueing and subsumption. The former is linguistically very economical since it leaves it up to the addressee to determine in what way a message should be interpreted at any level of meaning construction. It may also be a communicative strategy to avoid the responsibility of being explicit. But it can sometimes be risky and lead to misinterpretation that has to be solved on the basis of repair and meaning negotiation strategies (see Ruiz de Mendoza & Otal 1997; Otal & Ruiz de Mendoza 2007). We consider each process in turn.

.  Of course, it is also possible to have a non-default interpretation of can’t in terms of capacity (i.e. ‘it doesn’t have the capacity to sound good’). In this case, the sentence It can’t sound good would act as the effect element of a cause-effect pattern (‘it can’t sound good because it’s not digital’). In either situation, a relevant part of the inferential process at the discourse level is guided by lexico-grammatical cues that act in one way or another depending on complementary contextual cues that help to determine the value of the modal auxiliary.



Levels of description and explanation in meaning construction 

7.1  Cued inferencing As has been noted throughout our discussion, cueing or cued inferencing is a way of constraining interpretation on the basis of linguistic clues. We have seen how cued inferential activity takes place at the level of discourse relations in our brief discussion of (28a) and (28b) in the previous section. We have parallel situations at the other three levels. Let us first deal with cued inferencing at the core grammar level or level 1. Take the following utterances: (31) a. We are all ready now [for the party]. b. I certainly will [attend the meeting]. c. The president has finished [speaking]. d. They have just started [singing].

The examples in (31) are all cases of so-called ‘underdetermined’ linguistic expressions, i.e. expressions that can only make full sense if completed with contextual material. The study of underdetermined expressions has a long tradition. In the philosophy of language, Frege discussed the notion of sense completers (see Perry 1977) in relation to his distinction between sense and thought, where the former, which was guided by the structure of language, could be incomplete, while the latter had a truth value and was therefore complete. The items between square brackets in each of the examples in (31) would be clear cases of sense completers, since without them it is impossible to assign a truth value to any of the sentences to which they apply. For example, if for (31a) to be true it is necessary that the protagonists are ready for a party, then simply saying that they are ready in general will not make the sentence true. In pragmatics, Frege’s sense completers are not treated as mechanisms to obtain the truth-value of an incomplete sentence, but rather as contextually-derived information that is necessary to obtain the meaningful interpretation of an utterance. Different authors have used different names for this process of filling in implicit contextual information: saturation (Recanati 1989), completion (Bach 1994), and enrichment (Sperber & Wilson 1995). But what is important is the fact that this process of filling in missing information is regarded by these scholars as an inferential one and, we may add, a constrained one too: the language user needs to find adequate contextual information that matches the requirements imposed by the expressed utterance. Note that such requirements are largely constructional rather than based on lexical projection. Thus, in (31a), the Y element in the X Be Ready Y construction has a syntactically optional complement expressing the event or activity for which arrangements have been made either on the basis of a ‘for+NP’ structure or a to-infinitival clause. Other options are excluded (cf. *He was ready going to the party/*He was ready for going to the party). This constructional character of ‘X Be Ready Y’ is further reinforced by other formal features of this configuration, such as the reluctance of the verbal predicate to appear in the progressive form (*We are being ready for the party). In turn, modal constructions allow for

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

the elision of easily recoverable previous textual material, as in (31b). Finally, (31c) and (31d), which omit the non-finite verbal complement, are based on two converse phasal aspect constructions: (31c), which signals the end point of an action, is egressive; (31d), which signals the initial point of an action, is ingressive. There are other ways of cueing inferences at level 1. Consider the examples in (32), (33), and (34) below: (32) a. Something [bad] has happened. b. Your father [habitually] drinks [alcohol]. c. She has a [higher-than-normal body] temperature. (33) a. My sister has [exactly] three children. b. There were [approximately] three hundred thousand demonstrators. c. John and Alex went to the city museum [together]. (34) a. Nice day! b. You there! c. Morning! d. Look out!

Carston (2002, 2004) has discussed examples like these as different forms of free enrichment, a term which suggests that the completion or specification process is not guided by the linguistic structure of the message. The first form is illustrated by the examples in (32), where a literal (i.e. descriptive) reading of each utterance yields an obvious truth: events happen, humans drink liquids, and people who are alive have some body temperature. But, interpretively, each of the utterances in (32) is fully meaningful. The second form of free enrichment, which is illustrated by the examples in (33), holds for utterances whose basic conceptual layout contains elements that need to be spelled out in greater detail. Note that the bracketed specifications are default interpretations: (33a) would be true too if the speaker’s sister had more than three children; (33b) would be just as true if there were exactly three hundred thousand demonstrators; (33c) could also apply to a situation in which John arrived at the museum first and Alex some time later. Finally, the third form of free enrichment is illustrated by so-called subsentential utterances, which are the equivalent of what Halliday and Matthiessen call minor clauses. Clauses of this kind are different from elliptical clauses in which the elided elements are presupposed either on the grounds of the preceding linguistic context, as in question-answer pairs, or of the grammatical structure of the explicit elements (Seen Fred? [Have you seen Fred?]). Minor clauses do not have a verb (or another form of predicator) and they typically realize such functions as exclaiming, calling, greeting, and alarming, as illustrated respectively by each of the examples in (34). They may also be used presentatively (e.g. Mr. Jones, uttered by way of introduction of Mr. Jones) or even indexically (e.g. A stork!, uttered with surprise by a small child as he points at the bird). The overall function of minor clauses is to call the addressee’s attention to some entity, situation, or event, while elliptical clauses are simply ways of avoiding repetition of recoverable material.



Levels of description and explanation in meaning construction 

Cued inferencing at levels 2 and 3 take the form of traditional implicatures and illocutionary force respectively. In the LCM implicatures are seen as the result of affording access to low-level situational models by mentioning a relevant part of them (cf. Ruiz de Mendoza 2007). This is in fact a metonymic operation: (35) Mary: Is your tooth still hurting? Fred: Finally, I called the dentist.

The brief conversational exchange in (35) is based on every-day knowledge that we have about ‘going to the dentist’. By mentioning a relevant part of the frame, we give the addressee access to the whole frame: in (35) we are explicitly told that the second speaker called the dentist, and we interpret that the second speaker made an appointment, that he went to the dentist’s office, that the dentist took care of his dental problem, and that the tooth is probably not bothering the speaker any more. In Frame Semantics it would be argued that ‘going to the dentist’ is a frame, in much the same way as going to school, visiting a museum, dining at a restaurant, buying and selling, and the like. The notion is equivalent to constructs like scripts and scenarios postulated in the 1970s by a number of Artificial Intelligence workers that elaborated computer programs capable of generating inferences within restricted world-knowledge domains (e.g. Schank & Abelson 1977). In the LCM, situational frames are regarded as low-level situational models, to be differentiated from high-level situational models and from non-situational models. Both low-level and high-level situational models can be exploited metonymically in order to produce low-level and high-level implicated meaning. The former is the basis of traditional situation-based implicatures; the latter is traditional illocutionary meaning. Thus, we can compare the interpretation process for (35) with the one for (36) below: (36) Mary: I’m terribly thirsty. Fred: I’ll get you something to drink.

When people are thirsty, they usually look for something to drink. In every-day life we encounter hundreds of situations in which people face problematic situations and try to find ways to solve them. In many of these situations, people simply expect other people to be obliging and help them to solve their problem. On the basis of previous work by Pérez & Ruiz de Mendoza (2002), Ruiz de Mendoza & Baicchi (2007) have captured the essentials of these cultural expectations in part of a complex high-level cognitive model, which they call the Cost-Benefit Idealized Cognitive Model. The subdomain that applies to the interpretation of (35) as a request reads as follows (Ruiz de Mendoza & Baicchi 2007: 111): If it is manifest to A that a particular state of affairs is not beneficial to B, and if A has the capacity to change that state of affairs, then A should do so.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

Mary’s strategy in (35) is to make Fred aware that she has a problem, in the hope that Fred will act as is to be expected on the basis of cultural convention and solve her problem. In acting this way, the first speaker affords access to the whole subdomain by making explicit the part where a non-beneficial situation is affecting her. This is a metonymic exploitation based on a high-level situational model, which yields an interpretive situation that parallels the one we identified for implicatures, the only difference being the greater degree of genericity of the situational model in the case of illocution.

7.2  Subsumption Subsumption is a stepwise meaning production mechanism that consists in the principled incorporation of lower levels of semantic structure into higher levels of syntactically-oriented structure (Ruiz de Mendoza & Mairal 2006, in press a; Mairal & Ruiz de Mendoza in press). Subsumption is a constrained process that takes place at all levels of meaning derivation. At the level of core grammar (Level 1), internal constraints specify the conditions under which a lexical template may modify its internal configuration or the conditions for (part of) a constructional template to be realized by a given lexical item. They take the form of licensing or blocking factors that depend on lexical class ascription, on lexical-constructional compatibility, and on either predicate or internal variable conditioning of external variables. External constraints, in contrast, do not relate to the internal adaptation of lexical items to constructions, but rather to potential changes in the global category ascription of a lexical item when subsumed into a construction. In order to better appreciate the difference between the two kinds of constraint, let us consider the following sentences: (37) a. The audience laughed the actor off the stage. b. She drove me into a depression.

Sentence (37a) illustrates the caused-motion construction, which has been extensively discussed by Goldberg (1995, 2006). The construction typically requires the use of a caused-motion verb, as in She blew the napkin off the table, He pushed me into the room, They shoved me out of the car, but there many examples of the construction where motion is expressed but no motion verb is used, as in (37a). Other examples are possible: (38) a. She could smile him into abject submission. b. Finally, I could feel I was listened into existence. c. Sandra stared him into silence. d. His colleagues shouted him out of the lecture hall. e. She winked him into her bedroom.



Levels of description and explanation in meaning construction 

In other cases, we can make use of a caused-motion verb but there is no real motion. One example of this is (37b). Other examples are easy to come by: (39) a. How will he get us out of the quagmire of war? b. His policy is pushing our cause into oblivion. c. The discovery threw her into a state of great excitement. d. Hard thumping sounds pulled him out of his thoughts. e. When sin brought me into trouble, I found my refuge in you.

So, we have two different situations: one in which we need to account for the incorporation of a non-motion predicate into a constructional configuration that requires a motion predicate; another in which caused motion is figuratively used to express a change of state. In the LCM, the first situation, which involves the subcategorial conversion of verbal predicates, is considered a case of external constraints, and the second situation, which has to do with modifications in the internal configuration of lexical and constructional templates, is a matter of internal constraints on lexicalconstructional subsumption. Let us discuss each kind of constraint in greater detail. Consider again example (37a). The verb laugh is basically intransitive (e.g. Today we laughed a lot during the performance), but it can be used transitively by taking an oblique complement (e.g. Why are you laughing at the actor?). But this transitive use of laugh is qualitatively different from other forms of transitivity where the verbal action has a direct effect on the object, as in break, hit, kill, kiss, pull, push, and put, which we will refer to as effectual action verbs. In English this difference between effectual and non-effectual transitivity is marked by the use of a non-prepositional versus a prepositional object. Now, effectual action verbal predicates mark some form of cause-effect connection between agent and object. In some cases, the effect is an inherent part of the meaning of the predicate (e.g. killing results in death); in others, the effectual predicate only codes a very generic form of effect that is open to further specification by means of explicit resultative predicates, some of which express motion. Consider: (40) a. The child broke the vase [into pieces]. b. He hit the ball [off the field]. c. She kissed him [unconscious]. d. He pulled me out of the car. e. She pushed me into the pool. f. She put the book back on the shelf.

Note that the greater the genericity of the effect coded by the verbal predicate, the greater the need for a resultative complement. Thus, while the resultative specification is fully optional in (40a–c), it is slightly less so in (40d–e) and it is obligatory in (40f). Observe additionally that in (40b) and (40d–f) the result of the action is motion. In (40a) and (40c) the result of the action is a state, which in (40c) is expressed through a resultative adjective, whereas in (40a) it is expressed by means of figurative motion.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

It is evident from all these observations that the caused-motion construction requires effectual action verbal predicates whose effect component can be made explicit in the form of a prepositional complement expressing motion. This would mean that non-effectual action predicates (e.g. die, slide, flow) and effectual action predicates that code a specific effect (e.g. kill, poison, hurt) are not adequate candidates to take part in the construction. If this is the case, then what allows a verb like laugh to be used in the caused-motion construction? In our view, the answer lies in the possibility of treating this verb, which is not an effectual verb, as if it were an effectual verb (cf. Ruiz de Mendoza & Mairal 2007). Obviously, this special treatment of laugh is metaphorical in the sense given to this term in Cognitive Linguistics (Lakoff 1987; Lakoff & Johnson 1980, 1999); that is, there is a mapping or set of correspondences between two conceptual domains whereby one of the domains, the source, is used to think and reason about some aspects of another domain, the target. In the metaphor from an effectual action to non-effectual actions we treat actors of various kinds as if they were effectors, and goals as if they were effectees (i.e. direct objects of the effectual action). The metaphor can thus be applied, as a licensing factor, when verbs like laugh and others like those in the examples in (38) above are subsumed into the caused-motion construction, which requires an effectual predicate without an in-built result of the action. Note, in this connection, that in this subsumption process the object loses the grammatical mark of obliqueness. This mark is necessary outside the construction. Consider, in contrast to the correctness of the examples in (38) above, the impossibility of those in (41) below, where the constructional indicators of caused-motion have been removed: (41) a. *She could smile him. b. *They listened me. c. *Sandra stared him. d. *His colleagues shouted him. e. *She winked him.

There are other possible metaphors constraining grammatical conversion phenomena motivated by lexical-constructional subsumption: (42) a. We all drank (a lot of alcohol). b. We all drank (a toast) to John Rambo and the United States of America. (43) a. The cook separated the yolk from the white. b. The cook separated the yolk and the white. (44) a. Mary cut the bread. b. Mary cut at the bread.

First, let us consider example (42b) in relation to (42a). The use of drink in (42a) is its regular use as either an intransitive or a transitive predicate specifying the object of the action. In (42b), however, drink is adapted to make it part of the benefactive construction.



Levels of description and explanation in meaning construction 

This construction borrows part of its makeup from the ditransitive construction (e.g. We gave a machine gun to John Rambo), which in turn makes use of caused-motion constructional elements (cf. We sent/threw a machine gun to John Rambo ‘We caused a machine gun to move to John Rambo’s location by sending/throwing it’). In so doing, the benefactive construction treats institutionalized activities like drinking a toast, erecting a building, painting a picture, reciting a poem, among others, as transfers of possession that benefit the receiver. This is again metaphorical thinking. Interestingly enough, the idea of benefit is the result of a default implication – grounded in experience – from what we know about objects being transferred: when the object is at its destination, the receiver gains control of it (usually in the form of possession). So the roles of destination, receiver, and controller (or possessor) are conflated in our experience (cf. Lakoff & Johnson 1999, for other examples of experiential conflation and their role in metaphor). Example (43b) is a case of the reciprocal construction: the source domain is the domain of companionship (cf. We didn’t separate John and Mary), and the target is the domain of whole-part relations. Either putting together parts of an object or separating an object into parts can be seen as creating or destroying companionship relationships. Finally, the verb cut, which does not generally take an oblique object, does so in example (44b), where it is subsumed into the conative construction, thus suggesting an attempted action. This use is licensed by a metaphor with a non-effectual action in the source and an effectual action in the target, i.e. we understand the action of cutting the bread as if it was attempted but not necessarily carried into effect. This case is therefore the converse of the one for the adaptation of laugh to the caused-motion construction. What the metaphor does is ask us to understand one type of actor-object relationship in terms of another type. This cognitive operation is possible because the two kinds of action have a number of relevant features in common: they are goal-oriented (there is an intended effect on the object), there is motion and potential contact by impact, and the intended effect on the object is generic (i.e. not specific such as a change of state). Thus, change-of-state verbs such as bend, break, and smash cannot be mapped onto generic effectual actions (i.e. those that only code a very generic effect). It could be argued that this meaning effect does not originate in a metaphorical mapping but rather results from constructional coercion as defined by Michaelis (2003). Constructional coercion is the meaning adaptation process whereby a construction imposes part of its meaning structure on a lexical configuration. In this view, the conative construction would require the adaptation of the meaning of cut from an effectual action with a visible impact on the object to an attempted action directed to an object. The LCM admits the existence of constructional coercion, but it further argues that coercion is not arbitrary but motivated. We need reliable criteria that allow us to determine and therefore predict when coercion is possible and when it is not. The possibility or impossibility of setting up metaphorical correlations between

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

two verbal predicate types is one such criterion. There are other criteria – the internal constraints on lexical-constructional subsumption – that look at coercion as a matter of the internal configuration of lexical and constructional templates. We now turn to such constraints. There are two broad groups of internal constraints on lexical-constructional subsumption. One group relates to the conditions for a lexical template to be regarded as internally compatible with or in need of being adapted to a constructional template. The other group deals with the instantiation conditions of constructional variables. Let us first deal with the first kind of internal constraint. The simplest case is full matching. Here, a lexical item is required to respond to all constructional requirements without doing violence to its internal configuration. For example, the verb drink has a transitive use (e.g. John drank the beer). Following Van Valin’s (2005) Aktionsart distinctions, drink is an activity predicate that can be converted into an active accomplishment by the addition of a resultative component. This is captured in a lexical template through a BECOME operator that ranges over the second argument or complement variable. The transitive construction similarly requires two argument variables, the second of which is affected by a BECOME operator: (45) do´ (x, [drink´ (x, y)]) & BECOME consumed´ (y)

do´(x, y)

& BECOME pred´ (y)

But drink can also appear in the intransitive form (e.g. We just want to drink and be merry). The intransitive form requires the elimination of the second argument variable. In a constructional framework there are two ways of explaining this phenomenon. One is to postulate that, since drink is not a naturally intransitive predicate like breathe, run, or walk, making it part of the intransitive construction requires some internal adaptation of drink, which thus loses its second argument variable in its formal expression. But note that the intransitive use of a transitive verb is different in meaning from the intransitive use of a naturally intransitive verb. Even though the second argument has been dropped from formal expression, it is still latent, which does not suggest a valence reduction phenomenon of the kind postulated by Dik (1997a).16 Another way of addressing this issue is to postulate a special form of transitive construction where the object has been omitted (the objectless transitive construction; cf. Lemmens 2006) or, as Goldberg (2001) puts it, de-emphasized or de-profiled (the

.  In a similar way, Van Valin & LaPolla (1997: 122-125) propose the existence of an inherent argument, specifying the nature of the process but not obligatorily present in the syntax.



Levels of description and explanation in meaning construction 

deprofiled object construction) for discourse reasons while the action receives prominence. This formulation is in agreement with our previous discussion on level-1 cued inferencing: the hearer is cued by the deprofiled object construction to recover the missing argument inferentially from the context or from world knowledge. The intransitive construction, on the other hand, cannot cue such a process since it is by definition a true objectless construction. Furthermore, our account dispenses with the need to postulate verbal valency reduction processes in cases of intransitivization like the one for drink discussed above, or for other verbal predicates such as read (cf. He read the whole book; He read for hours before going to sleep), write (cf. She wrote a novel; She didn’t write last night), or kill (cf. The sheriff killed the gunman; Stop me before I kill again) in contexts in which there is some discourse reason for placing prominence on the verbal process (e.g. iteration, contrast, topicality). Since the object is recoverable, there is no real reduction process of the verbal arguments. This also means that we cannot postulate a subsumption process of these predicates into the intransitive construction, which would require full deletion of the object. Another case of internal constraint is the event identification condition, which requires the various subevents (i.e. bundles of operators and variables) into which a lexical and a constructional configuration can be segmented to match. It can be illustrated by the adscription of the verb break to the inchoative construction (cf. Cortés 2007, this volume). This construction, which can be listed as an objectless construction where the object has been promoted to subject status, occurs with predicates expressing either a telic accomplishment or an achievement, but not with states, activities – or their corresponding causatives – and with active accomplishments. Thus, it can be applied to pure change of state verbs like break (cf. The child broke the window; The window broke) but not to verbs like see (cf. We all saw the picture; *The picture saw), which denotes a state, or to predicates like eat or drink (cf. The child drank her milk; *Her milk drank), which can be active accomplishments. These situations are represented in (46)–(49) below, where the asterisks represent lexical-constructional mismatches: (46)

(47)

[Caus1Fact1]

[BECOME/INGR pred´ (x)] 1 = x

do´ (x, ∅) CAUSE

[BECOME broken (y)]

[Caus1Fact1]

***

[BECOME/INGR

***

pred´ (x)] 1 = x

see´ (x, y)

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

(48)

[Caus1Fact1]

[BECOME/INGR

do´ (x, ∅) CAUSE

(49)

[Caus1Fact1]

***

pred´ (x)] 1 = x

[feel´ (y, [pred´])]

[BECOME/ INGR pred´ (x)] 1 = x

***

*** do´ (x, [pred´ (x, y)])

[BECOME consumed´ (y)]

A third constraint is the lexical class constraint, which accounts for restrictions that class ascription places on lexical-constructional subsumption. Compare: (50) a. Your son broke my videotape. b. The CIA destroyed the videotape. (51) a. My videotape broke. b. *The videotape destroyed. (52) a. do´ (x, Ø) causE [BECOME broken´ (y)] b. do´ (x, Ø) causE [BECOME destroyed´ (y)]

We may wonder why if break and destroy have the same logical representation, as can be seen in (52), and the same transitive use, as evidenced in (51), only break can take part in the inchoative construction. The reason is that break and destroy belong to different lexical classes: the former is a change-of-state predicate, while the latter is an existential predicate indicating cessation of existence. The properties inherited from the lexical class can be built into the lexical template thus going beyond the basic logical form. The lexical template for the verb destroy should then include a specification of cessation of existence in the Aktionsart module based on semantic primes, further decomposing [BECOME destroyed´ (y)] into [BECOME NOT exist´ (y)], while the idea of severe damage involved in destroyed should be transferred to the semantic module, which is based on amalgams of lexical functions (or operators) ranging over internal variables. We thus have the following refined representation of destroy, where Caus, Magn, and Dam are lexical functions that indicate causation, intensity, and physical damage respectively: (53) [CausMagnDam12] do´ (x, Ø) causE [BECOME NOT exist´ (y)] x = 1, y = 2



Levels of description and explanation in meaning construction 

The representation in (53) reads as follows: there is an action performed by an actor such that the action has caused great damage to an object, thus resulting in the object ceasing to exist as such. The mismatch between this description and the inchoative construction is evident from a consideration of (54) below, where the resultative part of the logical structure of the lexical template is different from its expected counterpart in the constructional template. In the lexical template pred´ is non-existence, while in the constructional template pred´ is any state different from the initial state. (54)

[Caus1Fact1]

[BECOME/INGR pred´ (x)] 1 = x

do´ (x, ∅) CAUSE

*** [BECOME NOT exist´ (y)]

A fourth internal constraint is lexical blocking. Here, one of the components of the lexical template can block the fusion with a certain construction given that this component is a suppletive form. Accordingly, kill does not occur in the inchoative construction since its corresponding objectless form is suppletive, i.e. die, as in He killed the goose/The goose died. Since the linguistic system has already coded the form die in the lexicon with the same meaning, then the expected inchoative form of kill is blocked. In Spanish we have a comparable situation with only one difference. Spanish codes three values: matar (‘kill’), morir (‘die’), matarse (‘kill oneself ’). Note that an expression like El niño se mató (‘The child killed himself [accidentally]’) usually suggests that the child accidentally did something and as a result he died but there is no external agent as a possible cause of his death as in El asesino mató al niño (‘The murderer killed the child’) or La enfermedad mató al niño (‘The disease killed the child’). Now we turn our attention briefly to the second broad group of constraints, i.e. those pertaining to the limitations on the way constructional variables are realized as subsumption takes place. One such constraint is what we call predicate-argument conditioning. Sometimes the co-instantiation of the verbal predicate and one of its arguments can place restrictions on the kind of instantiating element that we can have for other constructional arguments. Thus, in the caused-motion construction we have a basic constructional structure of the type X-pred-Y(=NP)-Z(=PP). In principle, the constructional template can take any verb participant role to instantiate the Y element, which can be either human or non-human, as illustrated by the following Spanish variant of the caused-motion construction: Quiero a este niño lejos de aquí (‘I want this child far from here’); Quiero esta silla en otra parte (‘I want this chair somewhere else’). However, once the predicate and PP slots have been filled in, this choice constrains the kind of Y element that we can have. Me metió en una profunda depresión (‘He drove me

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez

into a deep depression’), the Y element (me) has been realized by a human verb role; we cannot have a non-human element (cf. *Metió a la silla en una profunda depresión ‘He drove the chair into a deep depression’). A second constraint within this group, internal variable conditioning, occurs when the internal predicate variables place constraints on the nature of both the predicate and constructional arguments. A clear example is supplied by the use of the Spanish verb volver in caused-resultative configurations: Me vas a volver loco (‘You’re driving me crazy’). There is a tendency for the resultative predicate to be axiologically negative: #Me vas a volver amable (‘You’re driving me kind’). To end our discussion of subsumption, we must also consider the way it works outside core grammar. In general, levels 2, 3, and 4 subsumption processes are a matter of semantic compatibility between the idiomatic parts of the construction and the conceptual material that is intended to realize its variable parts. We refer to this phenomenon as morphosyntactic parametrization. This constraint spells out the actual morphosyntactic type of the constructional unit in question. For example, as argued above the Can You Y? construction blocks out certain lexical classes (e.g. states and non-active accomplishments) as possible candidates for the Y element. In a similar way, the X and the Y elements in the ‘What’s X Doing Y?’ construction are constrained such that the X is usually a NP, while the Y element is a PP. The same can be said of the Double Be construction (McConvell 1988; Tuggy 1996; Massam 1999), which is used to draw attention to a given situation while asserting its truthfulness, as in The thing is, is that he didn’t tell the truth. In this construction, which takes the form ‘X is, is Y’, the X and Y elements are clearly defined by a number of features: X, which is the topic, is marked by a high tone, whereas Y, which is the focus, takes a low tone; Y can be freely realized by any that-clause but X is limited to a few options: the thing, the problem, the question, what I mean, and what happens.

8.  Conclusion The present paper has proposed the Lexical Constructional Model as an explanatorily adequate model for the investigation of the way in which lexical and constructional representations interact. The LCM is intended to be operational at all levels of linguistic description, including pragmatics and discourse. Thus, it has a level 1 or core module consisting of elements of syntactically relevant semantic interpretation, a pragmatic or level 2 module that accounts for low-level inferential aspects of linguistic communication, a level 3 module dealing with high-level inferences (i.e. illocutionary force), and finally a level 4 module that accounts for the discourse aspects of the LCM, especially cohesion and coherence phenomena. Each level is either subsumed into a higher-level constructional configuration or acts as a cue for the activation of a



Levels of description and explanation in meaning construction 

relevant conceptual structure that yields an implicit meaning derivation. At the heart of the LCM we find the notions of lexical and constructional template, which are the building blocks of the model. The principled interaction between lexical and constructional templates supplies the central or core meaning layer for other more peripheral operations – involving implicated meaning – to take place. Meaning construction in the LCM revolves around two key processes, cueing and subsumption. Subsumption is a key meaning production mechanism that consists in the principled incorporation of lexical and/or constructional templates from one representational level into higherlevel constructional representations. At the level of core grammar constructional templates “coerce” lexical templates. We distinguish two kinds of constraints on coercion: internal and external. The former arise from the semantic properties of the lexical and constructional templates and do not affect the Aktionsart ascription of the predicates involved. The latter do involve Aktionsart changes and result from the possibility or impossibility of performing high-level metaphoric and metonymic operations on the lexical items involved in the subsumption process. Internal constrains specify the conditions under which a lexical template may modify its internal configuration. Finally, cueing or cued inferencing is a form of constraining non-explicit meaning on the basis of lexical and constructional clues. It takes place at all levels of meaning derivation as an alternative to subsumption. Thus, at the level of core grammar, it accounts for inferences obtained by making contextual adjustments to the meaning of some predicates. At other levels it accounts for meaning implications based on potential conceptual connections between propositions (the case of discourse), or on metonymic activations or high-level (for illocution), and low-level (for implicature) situational models or scenarios.

References Alonso Ramos, Margarita. 2002. Colocaciones y contorno en la definición lexicográfica. Lingüística Española Actual 24(1): 63–96. Bach, Kent. 1994. Conversational impliciture. Mind and Language 9(2): 124–162. Baker, Collin F. & Josef Ruppenhofer. 2003. FrameNet’s Frames vs. Levin’s Verb Classes. In Proceedings of the 28th annual meeting of the Berkeley Linguistics Society, Julie Larson & Mary Paster (Eds), 27–38. Berkeley CA: University of California Linguistics Department. Beaugrande, Robert de. 1980. Text, discourse and process. Norwood NJ: Ablex. Bender, Emily & Andreas Kathol. 2001. Constructional effects of Just Because… Doesn’t Mean… In Proceedings of the 27th annual meeting of the Berkeley Linguistics Society. Berkeley CA: CLS. Bergen, Benjamin K. & Nancy Chang. 2005. Embodied construction grammar in simulationbased language understanding. In Construction grammars: Cognitive grounding and theoretical extensions, Jan-Ola Östman & Mirjam Fried (Eds), 147–190. Amsterdam: John Benjamins.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez Boas, Hans C. 2003. A constructional approach to resultatives. Stanford, CA: CSLI Publications. Boas, Hans C. 2006. A frame-semantic approach to identifying syntactically relevant elements of meaning. In Contrastive studies and valency, Petra Steiner, Hans C. Boas & Stefan Schierholz (Eds), 119–149. Frankfurt: Peter Lang. Boas, Hans C. 2008. Resolving form-meaning discrepancies in Construction Grammar. In Constructional reorganization, Jaakko Leino (Ed.), 11–36. Amsterdam: John Benjamins. Butler, Christopher S. 2003. Structure and function. A guide to three major structural-functional theories. 2 vols. Amsterdam: John Benjamins. Butler, Christopher S. & Francisco Gonzálvez-García 2005. Situating FDG in functionalcognitive space: An initial study. In Studies in Functional Discourse Grammar, J. Lachlan Mackenzie & María L.A. Gómez-González (Eds), 109–158. Berlin: Peter Lang. Carston, Robyn. 2002. Thoughts and utterances: The pragmatics of explicit communication. Oxford: Basil Blackwell. Carston, Robyn. 2004. Explicature and semantics. In Semantics: A reader, Stephen Davis & Brendan S. Gillon (Eds), 817–845. Oxford: OUP. Cortés, Francisco. 2007. The English constructicon. Ms, University of La Laguna. Cortés, Francisco & Mairal Usón, Ricardo. In preparation. Constructing meaning in Role and Reference Grammar. Croft, William. 1991. Syntactic categories and grammatical relations. Chicago IL: The University of Chicago Press. Croft, William. 2001. Radical Construction Grammar. Oxford: OUP. Culicover, Peter W. & Ray Jackendoff. 2005. Simpler Syntax. Oxford: OUP. Dik, Simon C. 1997a. The Theory of Functional Grammar, Part 1: The structure of the clause. Kees Hengeveld (Ed.), Berlin: Mouton de Gruyter. Dik, Simon C. 1997b. The Theory of Functional Grammar, Part 2: Complex and derived constructions. Kees Hengeveld (Ed.), Berlin: Mouton de Gruyter. Dowty, David. 1979. Word meaning and Montague Grammar. Dordrecht: Foris. Faber, Pamela & Ricardo Mairal. 1999. Constructing a lexicon of English verbs. Berlin: Mouton de Gruyter. Fawcett, Robin P. 1980. Cognitive linguistics and social interaction: Towards an integrated model of a Systemic Functional Grammar and the other components of an interacting mind. Heidelberg: Julius Groos. Fillmore, Charles. 1985. Frames and the semantics of understanding. Quaderni di Semantica 6(2): 222–254. Fillmore, Charles J. & Beryl T.S. Atkins. 1992. Towards a frame-based organization of the lexicon: The semantics of RISK and its neighbors. In Frames, fields, and contrast: New essays in semantics and lexical organization, Adrienne Lehrer & Eva F. Kittay (Eds), 75–102. Hillsdale NJ: Lawrence Erlbaum Associates. Fillmore, Charles J. & Beryl T.S. Atkins. 1994. Starting where the dictionaries stop: the challenge for computational lexicography. In Computational approaches to the lexicon, Beryl T.S. Atkins & Antonio Zampolli (Eds), 349–393. Oxford: OUP. Fillmore, Charles J., Paul Kay & Mary Catherine O’Connor. 1988. Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64: 501–538. Goldberg, Adele. 1995. A Construction Grammar approach to argument structure. Chicago IL: University of Chicago Press.



Levels of description and explanation in meaning construction 

Goldberg, Adele. 2001. Patient arguments of causative verbs can be omitted: The role of information structure in argument distribution. Language Sciences 34: 503–524. Goldberg, Adele. 2006. Constructions at work: The nature of generalization in language. Oxford: OUP. Goldberg, Adele, Devin M.  Casenhiser & Nitya Sethuraman, 2005. The role of prediction in construction-learning. Journal of Child Language 32(2): 407–426. Gonzálvez-García, Francisco. 2008. The family of object-related depictives in English and Spanish: Towards a usage-based, constructionist analysis. Language Sciences. doi:10.1016/j. langsci.2008.01.003. Gonzálvez-García, Francisco & Christopher, S.  Butler. 2006. Mapping functional-cognitive space. Annual Review of Cognitive Linguistics 4: 39–96. Halliday, Michael A.K. 1994. An introduction to functional grammar, 2nd Edn. London: Edward Arnold. Halliday, Michael A.K. & Christian M.I.M.  Matthiessen. 2004. An introduction to functional grammar, 3rd Edn. London: Hodder Arnold. Hengeveld, Kees. 2004. The architecture of a Functional Discourse Grammar. In A new architecture for Functional Grammar, J. Lachlan Mackenzie & María L.A. Gómez-González (Eds), 1–21. Berlin: Mouton de Gruyter. Hengeveld, Kees & J.  Lachlan Mackenzie. 2006. Functional Discourse Grammar. In Encyclopaedia of language and linguistics, Vol. 4. 2nd Edn. Keith Brown (Ed.), 668–676. Oxford: Elsevier. Hengeveld, Kees & J. Lachlan Mackenzie. 2008. Functional Discourse Grammar. A typologicallybased theory of language structure. Oxford: OUP. Hoey, Michael. 1983. On the surface of discourse. London: George Allen and Unwin. Holmes, Jasper & Richard Hudson. 2000. Just Because X Doesn’t Mean Y. Paper delivered at the Linguistics Association of Great Britain, Spring Meeting, University College London. Jackendoff, Ray. 2002. Foundations of language. Oxford: OUP. Jackendoff, Ray. 2007. A whole lot of challenges for linguistics. Journal of English Linguistics 35: 253–262. Kay, Paul & Charles J. Fillmore. 1999. Grammatical constructions and linguistic generalizations: The ‘What’s X doing Y’ construction. Language 75: 1–33. Lakoff, George. 1987. Women, fire and dangerous things: What categories reveal about the mind. Chicago IL: University of Chicago Press. Lakoff, George & Mark Johnson. 1980. Metaphors we live by. Chicago IL: The University of Chicago Press. Lakoff, George & Mark Johnson. 1999. Philosophy in the flesh. New York NY: Basic Books. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1: Theoretical prerequisites. Stanford CA: Stanford University Press. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Vol. 2: Descriptive application. Stanford CA: Stanford University Press. Langacker, Ronald W. 1999. Grammar and conceptualization. Berlin: Mouton de Gruyter. Langacker, Ronald W. 2001. Discourse in Cognitive Grammar. Cognitive Linguistics 12(2): 143–188. Langacker, Ronald W. 2005. Construction Grammars: cognitive, radical, and less so. In Cognitive Linguistics. Internal dynamics and interdisciplinary interaction, Francisco J. Ruiz de Mendoza Ibáñez & Sandra Peña Cervel (Eds), 101–159. Berlin: Mouton de Gruyter.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez Langacker, Ronald W. 2008. Cognitive Grammar: A basic introduction. Oxford: OUP. Lemmens, Maarten. 2006. More on objectless transitives and ergativization patterns in English. Constructions SV1–6/2006. (www.constructions-online.de) Levin, Beth. 1993. English verb classes and alternations. A preliminary investigation. Chicago IL: The University of Chicago Press. Levin, Beth & M. Rappaport Hovav. 2005. Argument realization. Cambridge: CUP. Longacre, Robert E. 1972. Hierarchy and universality of discourse constituents in New Guinea languages: Discussion. Washington DC: Georgetown University Press. Mairal Usón, Ricardo & Pamela Faber. 2002. Functional Grammar and lexical templates. In New perspectives on argument structure in Functional Grammar, Ricardo Mairal Usón & María J. Pérez Quintero (Eds), 41–98. Berlin: Mouton de Gruyter. Mairal, Ricardo & Pamela Faber. 2007. Lexical templates within a functional cognitive theory of meaning. Annual Review of Cognitive Linguistics 5: 137–172. Mairal Usón, Ricardo & Francisco José Ruiz de Mendoza Ibáñez. In press. Internal and external constraints in meaning construction: the lexicon-grammar continuum. In Estudios de Filología Inglesa: Homenaje a la Dra. Asunción Alba Pelayo [Colección Varia], Teresa Gibert Maceda and Laura Alba-Juez (Eds). Madrid: UNED. Martin, James R. 1992. English text: System and structure. Amsterdam: John Benjamins. Massam, Diane. 1999. Thing is constructions: The thing is, is what’s the right analysis? English Language and Linguistics 3: 335–352. McConvell, Patrick. 1988. To be or double be: Current change in the English copula. Australian Journal of Linguistics 8: 287–305. Mel’čuk, Igor. 1989. Semantic primitives from the viewpoint of the Meaning-Text Linguistic Theory. Quaderni di Semantica 10(1): 65–102. Mel’čuk, Igor, André Clas & Alain Polguère. 1995. Introduction à la lexicologie explicative et combinatoire. Louvain-la-Neuve: Duculot. Mel’čuk, Igor & Leo Wanner. 1996. Lexical functions and lexical inheritance for emotion lexemes in German. In Recent trends in Meaning-Text Theory, Leo Wanner (Ed.), 209–227. Amsterdam: John Benjamins. Michaelis, Laura. 2003. Word meaning, sentence meaning, and syntactic meaning. In Cognitive approaches to lexical semantics, Hubert Cuyckens, René Dirven & John R.  Taylor (Eds), 93–122. Berlin: Mouton de Gruyter. Nichols, Johanna. 1984. Functional theories of grammar. Annual Review of Anthropology 13: 97–117 Nuyts, Jan. 2005. Brothers in arms? On the relations between Cognitive and Functional Linguistics. In Cognitive Linguistics. Internal dynamics and interdisciplinary interaction, Francisco José Ruiz de Mendoza Ibáñez & Sandra Peña Cervel (Eds), 69–100. Berlin: Mouton de Gruyter. Otal Campo, José Luis & Francisco José Ruiz de Mendoza Ibáñez. 2007. Modeling thought in language use: at the crossroads between discourse, pragmatics, and cognition. Jezikoslovlje 8(2): 115–167. Panther, Klaus-Uwe. 2005. The role of conceptual metonymy in meaning construction. In Cognitive Linguistics. Internal dynamics and interdisciplinary interaction, Francisco José Ruiz de Mendoza Ibañez & M. Sandra Peña Cervel (Eds), 353–386. Berlin: Mouton de Gruyter. Panther, Klaus-Uwe & Linda Thornburg. 1998. A cognitive approach to inferencing in conversation. Journal of Pragmatics 30: 755–769.



Levels of description and explanation in meaning construction 

Panther, Klaus-Uwe & Linda Thornburg. 2003. Metonymies as natural inference and activation schemas: the case of dependent clauses as independent speech acts. In Metonymy and pragmatic inferencing, Klaus-Uwe Panther & Linda Thornburg (Eds), 127–147. Amsterdam: John Benjamins. Pérez, Lorena & Francisco Ruiz de Mendoza. 2002. Grounding, semantic motivation, and conceptual interaction in Indirect Directive Speech Acts. Journal of Pragmatics 34(3): 259–284. Perry, John. 1977. Frege on demonstratives. Philosophical Review 86: 474–497. Pustejovsky, James. 1995. The Generative Lexicon. Cambridge MA: The MIT Press. Recanati, François. 1989. The pragmatics of what is said. Mind and Language 4: 294–328. Ruiz de Mendoza Ibáñez, Francisco José. 2007. High level cognitive models: In search of a unified framework for inferential and grammatical behavior. In Perspectives on metonymy, Krzysztof Kosecki (Ed.), 11–30. Frankfurt: Peter Lang. Ruiz de Mendoza Ibáñez, Francisco José & Annalisa Baicchi. 2007. Illocutionary constructions: Cognitive motivation and linguistic realization. In Explorations in pragmatics: Linguistic, cognitive, and intercultural aspects, Istvan Kecskes & Laurence Horn (Eds), 95–128. Berlin: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José & Ricardo Mairal Usón. 2006. Levels of semantic representation: Where lexicon and grammar meet. Interlingüística 17: 26–47. Ruiz de Mendoza Ibáñez, Francisco José & Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Aspects of meaning construction, Günter Radden, Klaus-Michael Köpcke, Thomas Berg & Peter Siemund (Eds), 33–49. Amsterdam: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco José & Ricardo Mairal Usón. In press a. Levels of description and constraining factors in meaning construction: An introduction to the Lexical Constructional Model. Folia Linguistica 42. Ruiz de Mendoza, Francisco José & Ricardo Mairal Usón. In press b. Challenging systems of lexical representation. Journal of English Studies 4. Ruiz de Mendoza Ibáñez, Francisco José & José Luis Otal Campo. 1997. Communication strategies and realization procedures. ATLANTIS. Revista de la Asociación Española de Estudios Anglonorteamericanos 19(1): 297–314. Ruiz de Mendoza Ibáñez, Francisco José & Lorena Pérez Hernández. 2003. Cognitive operations and pragmatic implication. In Metonymy and pragmatics, Klaus-Uwe Panther & Linda Thornburg (Eds), 23–49. Berlin: Mouton de Gruyter. Sadock, Jerrold M. & Arnold M. Zwicky. 1985. Speech act distinctions in syntax. In Language typology and syntactic description, Timothy Shopen (Ed.), 155–196. Cambridge: CUP. Schank, Roger C. & Robert P. Abelson. 1977. Scripts, plans, goals and understanding: An inquiry into human knowledge structures. Hillsdale NJ: Lawrence Erlbaum. Sperber, Dan & Deirdre Wilson. 1995. Relevance. Communication and cognition, 2nd Edn. Oxford: Basil Blackwell. Steen, Gerard. 2005. Basic Discourse Acts: Towards a psychological theory of discourse segmentation. In Cognitive Linguistics. Internal dynamics and interdisciplinary interaction, Francisco José Ruiz de Mendoza Ibáñez & Sandra Peña Cervel (Eds), 283–312. Berlin: Mouton de Gruyter. Stefanowitsch, Anatol. 2003. A construction-based approach to indirect speech acts. In Metonymy and pragmatic inferencing, Klaus-Uwe Panther & Linda L. Thornburg (Eds), 105–126. Amsterdam: John Benjamins.

 Ricardo Mairal Usón & Francisco José Ruiz de Mendoza Ibáñez Thornburg, Linda & Klaus-Uwe Panther. 1997. Speech act metonymies. In Discourse and perspective in Cognitive Linguistics, Wolf-Andreas Liebert, Gisela Redeker & Linda R. Waugh (Eds), 205–219. Amsterdam: John Benjamins. Tuggy, David. 1996. The thing is is that people talk that way. The question is is why? In Cognitive Linguistics in the redwoods: The expansion of a new paradigm in linguistics, Eugene H. Casad (Ed.), 713–752. Berlin: Mouton de Gruyter. Van Valin, Robert D. Jr. 2005. Exploring the syntax-semantics interface. Cambridge: CUP. Van Valin, Robert D. Jr. In press. Lexical representation, co-composition, and linking syntax and semantics. In New developments in the Generative Lexicon, James Pustejovsky et al. (Eds), Dordrecht: Kluwer. Van Valin, Robert D. Jr. & Randy LaPolla. 1997. Syntax: Structure, meaning and function. Cambridge: CUP. Van Valin, Robert D. Jr. & David P. Wilkins 1993. Predicting syntactic structure from semantic representations: remember in English and its equivalents in Mparntwe Arrernte. In Advances in Role and Reference Grammar, Robert D. Van Valin Jr. (Ed.), 499–534. Amsterdam: John Benjamins. Vendler, Zeno. 1967. Linguistics in philosophy. Ithaca NY: Cornell University Press. Weilbacher, Hunter & Hans C. Boas. 2006. ‘Just Because’ two constructions look alike in two languages doesn’t mean that they share the same properties: Towards contrastive Construction Grammars. Presented at the 4th International Construction Grammar Conference, Tokyo, Japan. Winter, Eugene O. 1982. Towards a contextual grammar of English. London: George Allen and Unwin.

part iii

Studies of specific constructions

Measuring out reflexivity in secondary predication in English and Spanish Evidence from cognition verbs* Francisco Gonzálvez-García This article provides a usage-based, bottom-up constructionist analysis à la Goldberg (2006) of formally identical instances of secondary predication featuring find/encontrar (‘find’) and a reflexive pronoun in the object slot in English and Spanish. Specifically, it shows that these two configurations can be aptly regarded as two different, though closely connected, constructions (i.e. learned form-function pairings), namely, the reflexive subjective-transitive construction and the self-descriptive subjective-transitive construction. At a higher level of resolution, it is argued that the reflexive subjective-transitive construction imposes an agentive, intentional construal on the event/state of affairs in question, while the self-descriptive subjective-transitive construction requires a non-volitional, non-agentive construal. Crucially, configurations of the reflexive subjective-transitive construction are shown to be closer to a two-participant event elaboration in which the entity encoded in the main clause and the reflexive is construed as a ‘divided self ’ (Haiman 1998) between an experiencer subject and an affected object. By contrast, instances of the self-descriptive subjective-transitive construction can be aptly considered to be functionally equivalent to oneparticipant events, and can thus often be paraphrased by means of (in-)transitive or intensive clauses. At an ever higher level of granularity, the continuum between reflexives (encoding two-participant events) and middles (encoding one-participant events) can also be observed vertically within the self-descriptive

*Financial support for this paper has been provided by the DGI, Spanish Ministry of Education

and Science and the European Regional Development Fund, grants HUM 2004–05947-C02–01/ FILO, HUM 2005–02870/FILO, HUM 2007–65755/FILO and HUM 2007–62220. This research is a revised and extended version of Gonzálvez-García (2001) and is also part of more wideranging work in progress undertaken within the research group PAI HUM 0269. I am most grateful to two anonymous reviewers as well as Chris Butler for most helpful comments that led to substantial improvements in the final version. I am also indebted to Mira Ariel, who generously granted me access to unpublished material. Last but not least, over the last few years I have benefited from enlightening discussions of this topic with Adele Goldberg and Hans Boas, who have also been kind enough to supply unpublished material. All usual disclaimers apply.

 Francisco Gonzálvez-García

subjective-transitive construction. The crucial determinant in this respect is taken to be the inherent meaning and form properties of the object-related predicative phrase (i.e. the XPCOMP) and its transitivity properties (Hopper & Thompson 1980). Finally, corpus-based evidence is provided that although English and Spanish share a considerable number of morphosyntactic realizations of the XPCOMP, the inventory of such realizations is not fully symmetrical, thus lending further credence to Croft’s (2003) contention that argument structure is not only construction-specific but also language-specific.

1.  Introduction This article constitutes a first step towards a better understanding of the complex intricacies regarding the continuum between transitive reflexive constructions, on the one hand, and intransitive/middles, on the other, drawing on naturally-occurring data from the original edition of the British National Corpus (BNC henceforth) and the Corpus de Referencia del Español Actual (CREA henceforth) in English and Spanish, as in (1)–(2) below:1

(1) a. Peter looked at himself in the mirror.

b. Pedro se mir-ó en el espejo. Pedro[name] 3sg.refl look-indefpret.3sg in def.m.sg mirror. ‘Peter looked at himself in the mirror’.

(2) a. Peter died.

b. Pedro se muri-ó. Pedro[name] 3sg.pronomclitic die-indefpret.3sg ‘peter died’.

Specifically, the focus of this article will be primarily on instances of secondary predication in the sense of e.g. Aarts (1995) and Demonte and Masullo (1999) (or, alternatively, ‘small clauses’/‘complex transitive’ complementation) featuring cognition verbs such as find and encontrar (‘find’) as well as a reflexive pronoun in the object slot and an object-related predicative phrase (XPCOMP). Thus, consider (3)–(4) below:

.  From now on, interlinear morpheme-by-morpheme glosses will be supplied for the Spanish examples following the Leipzig Glossing Rules (see http://www.eva.mpg.de/lingua/files/morpheme.html). The following three additional abbreviations will be used in this paper: impers (impersonal), pronomclitic (pronominal clitic), and conditional (conditional or potential verb tense).





Measuring out reflexivity in secondary predication in English and Spanish 

(3) a. I find myself incredulous. There is nothing whatever upon which Mariology could be built! (BNC EF0 1199).

b. Mi bis-abuelo 1sg.poss great-grandfather

vest-ía así wear-imppret.3sg like.this

y yo me encuentr-o muy atractiv-o con ellos. and 1sg 1sg.refl find-prs.1sg very attractive-m.sg with 3pl

‘My great-grandfather would be dressed like this and I find myself very attractive in these’. (CREA, 2003, Prensa, Tp Teleprograma, nº 1963, 17–23/11/2003: TVE 1 estrena “Arroz y tartana” con Maura y Sancho, TeleRevistas S.A. Hachette Filipacchi (Madrid), 2003).

(4) a. The young James found himself a virtual prisoner of the Red Douglasses in Edinburgh castle. (BNC AS7 0994).

b. Al final de to.def.m.sg end of

es-a temporada dist-f.sg season

pens-é think-indefpret.1sg



que acab-aría mi trayectoria en comp finish-conditional.1sg 1sg.poss career in



el Barcelona, pero sin def.m.sg Barcelona.football.team but without



dar-me cuenta, me encontr-é give.inf-acc1sg account 1sg.refl find-indefpret.1sg



en in

el paro. def.m.sg unemployment

‘At the end of the season I thought that I would end my professional career in the Barcelona football team, but all of the sudden I found myself out of work’. (CREA, ABC Electrónico, 15/11/1997: Valencia. Manuel Conejos).

In this article, I will show that, despite their formal similarity, the configurations in (3) and (4) above are, in these two languages, instances of different, though nonetheless related, constructions, viz. the reflexive evaluative subjective-transitive construction and the self-descriptive subjective-transitive construction, respectively. At a higher level of resolution, it will be argued that the latter construction can be taken to illustrate different points in a continuum between two-participant events and one participant events (cf. Kemmer 1993, 1994, 2003: 109, Maldonado 1999, 2007, 2008). The structure of this article is as follows: section two provides some methodological preliminaries on the data on which our constructionist, usage-based analysis is grounded. Section three provides an overall picture of the continuum between reflexives and middles/intransitives in English and Spanish. Section four examines in some detail the most distinctive semantico-pragmatic and discourse-functional properties

 Francisco Gonzálvez-García

of the self-descriptive subjective-transitive construction in English and Spanish, with special focus on its impact on the continuum between reflexives and middles/intransitives in general and the elaboration of the event in particular. In line with the bottom-up, corpus-driven usage-based approach invoked here, attention is also paid to the most frequent collocates attested in the data in the two languages. Finally, section five summarizes the main findings and addresses some important future challenges for a usage-based, constructionist analysis of the issues under investigation here.

2.  Some methodological preliminaries In line with the usage-based stance taken in the Goldbergian strand of Construction Grammar (CxG henceforth), the methodological focus here is on the use of authentic data extracted from corpora routinely supplemented with data gained from introspection by native speakers (see e.g. Goldberg 1996: 69, 2006; Thompson 2002; Boas 2003; Bybee 2006; Bybee & Eddington 2006; inter alios). Specifically, within the bottom-up, corpus-driven approach invoked in this article, searches were conducted in the original edition of the BNC (see Burnard 1995 for further details on the source of the examples) and the CREA (see the Real Academia Española website in the bibliographical section) for instances of find and encontrar (‘find’) taking a reflexive pronoun and an obligatory object-related complement (XPCOMP). The searches were conducted in all text categories and modes in both corpora. In the case of Spanish, only the variety of Castilian Spanish was taken into account for practical reasons, viz. to guarantee a quantitative parity in the raw amount of data extracted. Thus, in the case of English, our searches yielded a raw total of 3,131 tokens, while in Spanish a number of 2,815 instances were found. In order to ensure maximum precision and recall (Gries, Hampe & Schönefeld 2005: 13), the raw tokens were manually coded, and only those featuring instances of secondary predication of the type illustrated in (3)–(4) were computed for analysis (see Tables 1–2). Moreover, all the examples as well as the material reproduced in this article, whether taken from the BNC or CREA or from other sources, were previously rated as (a) acceptable, (b) marginally acceptable or (c) unacceptable by a group of 30 educated British and American native speakers aged between 20 and 50 and by a group of Spanish university students aged between 21 and 22 at the University of Almería, Spain, respectively.2 The native speakers were given the following instructions as

.  Example (49) has been taken from the Great Britain component of the International Corpus of English (ICE-GB) for ease of exposition.



Measuring out reflexivity in secondary predication in English and Spanish 

to how to interpret the acceptability labels above. ‘Acceptable’ was taken to mean “the sentence is possible in English/Spanish”, while ‘unacceptable’ was intended to reflect that “the sentence is impossible in English/Spanish”. Moreover, the ‘marginally acceptable’ label was meant to capture the following acceptability judgements: (a) “the sentence is not altogether impossible but does not sound completely OK either” and/or (b) “I’m not quite sure about whether this sentence is acceptable or unacceptable”. The ratings given by the native informants to the examples and the verbs listed in 1–10 are reflected as follows: the unmarked sentences were unanimously deemed to be acceptable by 100% of the informants; sentences considered marginally acceptable by at least 90% of the informants are marked as #; while the examples marked as * were unambiguously taken to be ungrammatical by 100% of the informants. Before proceeding further, two important considerations are in order regarding the use of the term ‘usage-based’ from both a theoretical and methodological standpoint. With regard to the former, this term will be taken to imply the acceptance of the following premises: (a) usage, synchronic variation and diachronic change are intimately connected, (b) the focus is placed on communicative competence, and the strict division between competence and performance is rejected, (c) redundant generalizations concerning (highly) frequent item-specific patterns/expressions are allowed, even if these are fully compositional, and, (d) extensive use should be made of data from naturally occurring data in the investigation of language use (see Gonzálvez-García and Butler 2006:  82–83). From a methodological point of view, although the acceptability judgements presented in this article have been validated against elicitation tests with native speakers, I concur with Boas (2005, 2008a, 2008b) that taking a usage-based approach seriously implies taking into account encoding in conjunction with decoding. Thus, the generalizations presented in this article emerging from the examination of corpus data should be tested with a number of psychological experiments involving e.g. sentence completion tasks or reading of specific material to gather relevant information from the point of view of encoding. As advanced at an earlier point in this introduction, this article is exclusively concerned with the analysis of a fairly productive polysemous verb between a cognitive and sensory perception interpretation (see Sweetser 1990; Ibarretxe Antuñano 2006, inter alios), viz. find and encontrar (‘find’). In line with the importance accorded to frequency of occurrence within the usage-based approach, the primary reason why these matrix verbs have been selected for analysis here is the striking productivity of the former (Barlow 1996; Fukaya 2002) and the considerable frequency of the latter with a reflexive pronoun in the secondary predication environment. However, many of the claims made about these verbs can be duplicated for other verbs likely to be construed as verbs of cognition (e.g. see, imagine, ver (‘see’), imaginar (‘imagine’) in these two languages.

 Francisco Gonzálvez-García

3.  A  n overview of the continuum between reflexives and middles/ intransitives Before going into a fine-grained characterization of reflexives in English and Spanish, it may be convenient to provide an overview of the continuum between reflexives and intransitive constructions. To this end, and for reasons which will become clearer at a later stage in this article, I have selected the classification proposed in Martín Zorraquino (1979: 87–120) originally conceived for Spanish (see Ruiz de Mendoza 2008a: 135 for an alternative taxonomy) as the point of departure for our discussion of the reflexivity continuum in English and Spanish in the secondary predication environment. Following Martín Zorraquino (1979), a ten-point continuum can be posited, illustrated in (I)–(X) below: I.  ‘Proper reflexive constructions’, that is, configurations which unambiguously allow the substitution of the reflexive in the object slot with an object clitic, as in (5a)–(5b) below: (5) a. Mi padre se afeit-a con un-a poss.1sg father refl.3sg shave-prs.3sg with indf-f.sg

phillips. phillips[name]



‘My father shaves himself with a Philips’.

b. A Pedro lo afeit-a su barber-o. obj Pedro acc.m.sg shave-prs.3sg poss.3sg barber-m.sg ‘Peter, his barber shaves him’. (Examples taken from Martín Zorraquino 1979: 87, translations and glosses added).

An additional hallmark exclusive to Spanish reflexive configurations is that the reflexive form can be expanded by a sí mismo/a (‘to himself/herself ’). Semantically, the subject of this type of configuration can be agentive or experiencer in English and Spanish, as in (6)–(7) below, respectively, and they more likely than not imply varying degrees of intentionality. (6) 

Él se vi-o (a sí mism-o) 3sg.m refl.3sg see-indefpret.3sg obj dat.3sg same-m.sg



en in



‘He looked at himself in a mirror’.

(7) 

Ella se encuentr-a muy fe-a. 3sg.f refl.3sg find-prs.3sg very ugly-f.sg



‘She finds herself very ugly’.



un espejo. indf.m.sg mirror

(Examples taken from Martín Zorraquino, 1979, p. 90, translations and glosses added).



Measuring out reflexivity in secondary predication in English and Spanish 

II.  Pronominal constructions which accept the replacement of the putative reflexive form with an object clitic but nonetheless disallow the expansion of a sí mismo (‘to oneself ’), as illustrated in the acceptability differences reproduced in (8) below:  (8) a. Él se levant-ó a 3sg.m pronomclitic.3sg get.up-indefpret.3sg at

las diez. def.f.pl ten



‘He got up at ten o’ clock’.

b. Él lo 3sg.m acc.m.sg

levant-ó a get.up-indefpret.3sg at



las diez. def.f.pl ten



‘He woke him up at ten’.

c. *Él se levant-ó a 3sg.m pronomclitic.3sg get.up-indefpret.3sg obj

sí mism-o a las diez. dat.3sg same-m.sg at def.f.pl ten

‘He got himself up at ten’. (Examples taken from Martín Zorraquino 1979: 91, translations and glosses added).

III.  Pronominal constructions which accept an agentive subject or an experiencer subject, the expansion with a sí mismo (‘to oneself ’) yielding an acceptable result only with the former type of subject. Thus consider (9)–(10) below: (9) a. María se asust-ó María[name] pronomclitic.3sg frighten-indefpret.3sg

cuando oy-ó la noticia. when hear-indefpret.3sg def.f.sg item.of.news



‘Mary got frightened when she heard the item of news’.

b. A María la obj María[name] acc.3sg.

asust-aron frighten-indefpret.3sg

con aquell-a noticia. with dist-f.sg item.of.news

‘They frightened María with that item of news’. (Examples taken from Martín Zorraquino 1979: 92, translations and glosses added).

 Francisco Gonzálvez-García

(10) María se asust-ó a María[name] refl.3sg frighten-indefpret.3sg obj



sí mism-a con pensamiento-s trágic-o-s. dat.3sg same-f.sg with thought-pl tragic-m-pl

‘maría frightened herself with tragic thoughts’. (Example taken from Martín Zorraquino 1979: 93, translations and glosses added).

IV.  Configurations in which the reflexive pronoun functions as indirect object rather than as direct object. The reflexive unambiguously allows the substitution for an object clitic and some of these configurations admit the expansion by a sí mismo, (‘to oneself ’) as in (11) below: (11) a. Él se concedi-ó a sí 3sg.m refl.3sg allow-indefpret.3sg dat dat.3sg

mism-o unas vacaciones. same-m.sg indf.f.pl vacations



‘He allowed himself a vacation’.

b. El jefe le concedi-ó unas vacaciones. def.m.sg boss dat.3sg.m allow-indefpret.3sg indf.f.pl vacations ‘The boss gave him a vacation’. (Examples taken from Martín Zorraquino 1979: 95, translations and glosses added).

V.  Configurations in which the reflexive does not function syntactically as indirect object of the main verb despite the fact that it can be replaced with dative clitic forms such as le and les. In some cases, the expansion with a sí mismo (‘to oneself ’) is possible if the subject is agent rather than experiencer. Thus consider (12): (12) a. La señorita – que es muy def.f.sg young.lady rel be.prs.3sg very

masoquista – masochistic



se ha cortad-o el dedo refl.3sg pfvaux.3sg cut-ptcp def.m.sg finger

a dat



sí mism-a, voluntariamente. dat.3sg same-f.sg willingly



‘The young lady, who is very masochistic, has cut her finger, on purpose’

b. *La señorita se ha cortad-o def.f.sg young.lady refl.3sg pfvaux.3sg cut-ptcp

el def.m.sg

dedo a sí mism-a trabaj-ando en la fábrica. finger dat dat.3sg same-f.sg work-ger in def.f.sg factory

‘The young lady has cut her finger working at the factory’. (Examples taken from Martín Zorraquino 1979: 99, translations and glosses added).



Measuring out reflexivity in secondary predication in English and Spanish 

VI.  Reciprocal constructions: These configurations do not allow the expansion by a sí mismos (‘to themselves’) but rather those of the type uno-a a otro-a (‘each other’), mutuamente (‘mutually’). (13) 

Pedro y Juan se cre-en Pedro[name] and Juan[name] recipr.3pl think-prs.3pl



tont-o-s mutuamente. fool-m-pl mutually

‘Pedro and Juan consider each other fools’. (Example taken from Martín Zorraquino 1979: 101, translations and glosses added).

VII.  Constructions with a ‘superfluous reflexive’ pronoun. These configurations do not allow the expansion with a sí mismo (‘to oneself ’), although some of them allow the substitution of the se element with an object clitic, as shown in (14) below: (14) a. María se com-ió el pastel. María[name] pronomclitic.3sg eat-indefpret.3sg def.m.sg cake ‘María ate (up) the cake’. b. María le com-ió el pastel María[name] dat.3sg eat-indefpret.3sg def.m.sg cake

(a Juan). dat Juan[name]

‘María ate Juan’s cake’. (Examples taken from Martín Zorraquino 1979: 104, translations and glosses added).

VIII.  Constructions in which the se element cannot be felicitously replaced with an object clitic pronoun (/lo, la, los, las/ and/or a dative one (/le, les/), do not allow the expansion by a sí mismo (‘to oneself ’). In addition, the se element appears to be highly redundant, insofar as its omission does not yield an ungrammatical result. Thus, consider the acceptability results reflected in (15) below: (15) a. Juan se fue/ Juan[name] pronom.clitic.3sg go.indefpret.3sg

march-ó/ muri-ó/ ri-ó. leave-indefpret.3sg die-indefpret.3sg laugh-indefpret.3sg

‘juan left/went away/died/laughed’. (Examples taken from Martín Zorraquino 1979: 109, translations and glosses added). b. *Juan lo/ le muri-ó. Juan[name] acc.3sg dat.3sg die-indefpret.3sg *‘Juan died him/to him’.

 Francisco Gonzálvez-García

c. *Juan se fue Juan[name] pronomclitic.3sg go.indefpret.3sg

a obj

sí mism-o. dat.3sg same-m.sg *‘Juan went to himself ’.

d. Juan muri-ó/ fue a Barcelona. Juan[name] die-indefpret.3sg go.indefpret.3sg to Barcelona[name] ‘Juan died/went to Barcelona’. (Examples taken from Martín Zorraquino 1979: 110, translations and glosses added).

IX.  Middle configurations which do not accept the opposition with object pronouns, as shown in (16d) and do not allow the expansion by a sí mismo (‘to oneself ’) (cf. (16c)). (16) a. La nieve se derrit-ió. def.f.sg snow mid melt-indefpret.3sg ‘The snow melted’. b. La puerta se abr-ió de repente. def.f.sg door mid open-indefpret.3sg of sudden ‘The door opened all of a sudden’. c. #La nieve se derrit-ió a sí mism-a. def.f.sg snow mid melt-indefpret.3sg obj dat.3sg same-f.sg ‘The snow melted itself ’ d. #La puerta lo abr-ió de repente. def.f.sg door acc.3sg open-indefpret.3sg of sudden

‘The door opened it all of the sudden’.

X.  Pseudo-reflexive constructions, which, according to Martín Zorraquino (1979: 115) are closer in meaning to impersonal and reflex passive constructions (i.e. configurations featuring the clitic se and a verb in the active voice but likely to be construed as passives from a semantico-pragmatic viewpoint) (see further Mendikoetxea 1999; Sánchez López 2002; Gonzálvez-García 2006): (17) a. Se vend-e el libro. impers sell-prs.3sg def.f.sg book ‘The book is for sale’. b. Se viv-e bien en España. impers live-prs.3sg well in Spain ‘One lives well in Spain’. (Examples taken from Martín Zorraquino 1979: 116, translations and glosses added).



Measuring out reflexivity in secondary predication in English and Spanish 

From the general picture provided in this section, two important parameters can be concluded to be diagnostics of a reflexive construction, namely, (i) the possibility of replacing the reflexive pronoun with a full lexical NP or a clitic in English and Spanish, and (ii) the feasibility of expanding the reflexive by means of the phrase a sí mismo (‘to oneself ’) in Spanish. It should be immediately remarked that this expansion, far from being meaningless, may often serve to convey a contrastive or emphatic interpretation in contrast to their counterparts without the expansion in question (see example (20) below). In the next section, I will address how these parameters have been further elaborated within the Cognitive Grammar literature.

4.  The reflexive subjective-transitive construction Reflexive constructions have been extensively discussed in the literature, especially in connection with the so-called middle voice (Kemmer 1993, 1994, 2003; Barlow 1996; Martínez Vázquez 1998; Maldonado 1999, 2008; Sánchez López 2002:  74–79, Fried 2007; Ariel 2008, inter alios). Although coreferentiality (i.e. the fact that the subject and the object of a transitive verb have the same referent) is often pointed out in the literature as the distinctive property of reflexives (Quirk et al. 1985:  357–361, Biber et al. 1999:  342, Huddleston & Pullum 2002: 1487), Maldonado (2008: 154) notes that this should not be considered alone as the most distinctive property of reflexives. Purely semantic analyses of reflexives have also been posited in the literature in terms of selfaffectedness (Janda 1993) or the degree of distinctiveness of event participants (Croft 2001). More recently, Fried (2007: 750) has demonstrated that in the case of Czech a relatively wide number of semantic and pragmatic properties must be taken into account, such as absence vs. anonymity vs. coreference with patient, interaction with discourse roles, the pattern’s overall communicative function (event-centered or participant-centered), the meaning of the verb, aspectual differences and so forth. In the case of Spanish, Maldonado (2008), drawing on Kemmer (1993), suggests that reflexives “profile the way an agent acts on himself ” and can thus be expanded by the prepositional phrase a mí mismo ‘to myself ’, a ti mismo ‘to yourself ’, a sí mismo ‘to himself/herself ’, which must agree in person and number with the subject. By contrast, middles focus on the change of state undergone by the experiencer. This is reflected by the fact that middles do not have a split representation and cannot be expanded by a + clitic + mismo phrase (Maldonado 2008: 155). Thus, for instance, example (18a) below is a reflexive construction (or a ‘pure direct reflexive’ construction according to Martín Zorraquino 1979), while (18b) is not, given that it does not allow the expansion by a sí mismo (‘to oneself ’), thus qualifying as a case of ‘intransitive reflexive’ for Martín Zorraquino 1979.

 Francisco Gonzálvez-García

(18) a. Valeria se vio en el espejo. Valeria[name] refl.3sg see.indefpret.3sg in def.m.sg mirror ‘Valeria saw herself in the mirror’. b. Don Nico se murió. Don Nico[name] mid.3sg die.indefpret.3sg ‘Don Nico died (unexpectedly)’. (Examples and translations from Maldonado 2008: 151, glosses adapted).

In the case of secondary predication, as in (19) below, the clitic se unambiguously functions as a reflexive marker according to Maldonado’s definition, given the fact that it signals coreferentiality and allows a split representation of the event, since the reflexive can be expanded by a sí mismo (‘to oneself ’). (19)  (…) Me consider-o [a 1sg.refl consider-prs.1sg obj

mí mism-o]. dat.1sg same-m.sg

bastante extrovertid-o. quite extroverted-m.sg

(material within brackets added by the author) ‘I consider myself quite extroverted’. (CREA, La Vanguardia, 24/10/1994: “He logrado mi sueño de ser útil”. Patarroyo espera culminar las vacunas… ).

Following Maldonado (2008: 155), I concur that the expansion of the reflexive by means of the a sí mismo (‘to oneself ’) phrase may serve emphatic or contrastive purposes given an adequate supporting context in Spanish. Thus, for instance, it can serve to emphasize the fact that the subject himself/herself is the authority on which the judgemental stance expressed in the XPCOMP is grounded. However, it can also serve e.g. as a corrective statement in which the subject/speaker makes it abundantly clear that it is him (and not any other person) who considers himself quite extroverted, as in (20) below: (20) No me consider-an extrovertid-o, me neg acc.1sg consider-prs.3pl extroverted-m.sg 1sg.refl

consider-o a mí mism-o extrovertid-o. consider-prs.1sg obj dat.1sg same-m.sg extroverted-m.sg



‘they do not consider me extroverted, I consider myself extroverted’.



Reflexive configurations have been extensively discussed in the functionalist and cognitivist literature (see e.g. García 1975; Faltz 1977; Martín Zorraquino 1979; Kemmer 1993, 1994, 2003; Barlow 1996; Vera Luján 1996; Martínez Vázquez 1998; Maldonado 1999, 2008; Fried 2007; Ariel 2008, to name but a few), especially in relation to the so-called middle voice. While admitting that reflexive constructions are best handled in terms of a four-fold continuum between (i) two-participant events, (ii) reflexive events, (iii) middle events, and (iv) one-participant events (see Kemmer 1993, 1994, 2003, p. 109; also Maldonado 1999, 2007, 2008 for a similar position), at least the



Measuring out reflexivity in secondary predication in English and Spanish 

following distinguishing properties can be pinpointed regarding a finer-grained characterization of reflexive constructions in English and Spanish. Thus consider (i)–(iii) below (see further Sánchez López 2002: 74–79): i.  Reflexive configurations, unlike middle voice configurations (see e.g. (21d) below), allow, under normal circumstances, for the possibility of replacing the reflexive pronoun in the object slot with a non-reflexive clitic functioning as e.g. a direct or indirect object, as summarized in (20a)–(20d) below: (21) a. Pedro se lav-a. Pedro[name] refl.3sg wash-prs.3sg ‘Pedro washes himself ’. b. Pedro lo lav-a (a él/ Pedro[name] acc.3sg.m wash-prs.3sg obj 3sg.m

es-o/ a su herman-o). dist-m.sg obj poss.3sg brother-m.sg



‘Pedro washes him/it/his brother’.

c. Pedro se desmay-ó Pedro[name] pronomclitic.3sg vanish-indefpret.3sg ‘pedro vanished’. d. *Pedro lo desmay-ó. Pedro[name] acc.3sg.m vanish-indefpret.3sg *‘pedro vanished it’.

ii.  In Spanish, reflexive constructions allow for a duplication of the reflexive element through the form mismo/a (‘same’) preceded by the preposition a (‘to’), such a redundant reduplication yielding an unacceptable result in the case of middle voice constructions, as in (22a)–(22b) below: (22) a. Pedro se lav-a a sí mism-o. Pedro[name] refl.3sg wash-prs.3sg obj dat.3sg same-m.sg ‘Pedro washes himself ’. b. Me consider-o a mí mism-o refl.1sg consider-prs.1sg obj dat.1sg same-m.sg

Cervantes, el poeta. Cervantes[name] def.m.sg poet



‘I consider myself Cervantes, the poet’.

c. *La

puerta se

cerr-ó



def.f.sg door



mism-a de repente. same-f.sg of sudden ‘the door closed itself all of the sudden’.

a



mid close-indefpret.3sg obj dat.3sg

 Francisco Gonzálvez-García

These two facts taken together can be taken to point to a crucial property of reflexives in English and Spanish, namely, that they encode a split representation of the event in question, unlike their non-reflexive counterparts (e.g. intransitives/middles). iii.  Reflexive constructions involve at least some degree of intentionality on the part of the subject/speaker regarding the action encoded in the clause (as in (23a)–(23b) below), unlike their corresponding middle voice configurations (see further Luján 1977; Lázaro Mora 1983: 305, inter alios). However, interestingly enough, even within the secondary predication frame, some configurations involve an intentional action, while others do not, as illustrated for English and Spanish in (23)–(24) below: (23) a. (Willingly)/(On purpose) I find myself incredulous. (BNC EF0 1199; material within brackets added by the author). b. (…) (Por voluntad propia)/(A propósito) me For willingness own On purpose refl.1sg

consider-o Cervantes, el poeta (…). consider-prs.1sg Cervantes[name] def.m.sg poet

‘(Willingly)/(On purpose), I consider myself Cervantes, the poet’. (CREA, Tiempo, 19/02/1990: Fernando Arrabal/Lanza una novela y estrena una obra de teatro en Madrid). (24) a. (Suddenly/Unexpectedly) I find myself alone with the steward who tells me has been working on this line for the past twenty-five years. (BNC ABS 761, material within brackets added by the author). b. Después (sin quer-er/ sin dar-me Later without want-inf without give.inf-pronomclitic

cuenta) me encontr-é aisl-ad-o. account refl.1sg find-indefpret.1sg isolate-ptcp-m.sg



‘Later on, (unwillingly/suddenly) I found myself isolated’. (CREA, El País 11/12/1979: Balbín se Niega a Salir en Pantalla).

c. Después, (#por voluntad propia), (# a propósito) Later for willingness own on purpose

me encontr-é aislad-o. refl.1sg find-indefpret.1sg isolate-ptcp-m.sg



‘Later on (#willingly) (#on purpose) I found myself isolated’.

Before proceeding further, an important clarification is in order at this stage in relation to the sense in which the term ‘construction’ is used in this article. According to Goldberg (2006: 3, in press), constructions are taken to be “conventionalized pairings of form and function” (see also Croft 2001:  17–21, Langacker 2003; Fried & Östman 2005: 18–23 for alternative definitions of constructions). In this respect,



Measuring out reflexivity in secondary predication in English and Spanish 

an important difference from earlier definitions of the construction in the Goldbergian strand of CxG (Goldberg 1995: 4), is that idiosyncrasy is no longer considered to be a sine qua non condition for the status of construction (see further Goldberg 1998: 205). Thus, in agreement with the usage-based model invoked in this article, (highly) frequent configurations will be considered here to be constructions, even if these are fully compositional and can thus be predicted from a corresponding higherlevel construction at a given level of specificity (Goldberg 2006: 64, 214–215, Bybee & Eddington 2006: 328, inter alios). However, it should be immediately emphasized that when the labels ‘secondary predication’, or, alternatively, ‘subjective-transitive construction’, and so forth, are invoked in this article, these should not be taken to detract from Croft’s observation (2003: 59–65) that much of argument structure is construction-specific and languagespecific. Specifically, the labels ‘reflexive subjective-transitive’ and ‘self-descriptive subjective-transitive’ construction in Figures 1 and 2 are not intended to mean that the constructions in question are the same in English and Spanish. The labels and their corresponding anatomies, as represented in Figures 1 and 2, are rather shorthand for expository convenience as well as graphic representations of the considerable degree of convergence between these two configurations in English and Spanish. Our position in this respect is therefore in agreement with Goldberg (2006: 226), who opts for retaining “the more traditional emphasis on trying to capture and motivate generalizations, imperfect though we recognize them to be”. (Goldberg 2006: 225–226). To round off this section, I shall outline only briefly the main semantico-pragmatic properties of the reflexive subjective-transitive construction in English and Spanish (see further Gonzálvez-García 2007, 2008 for a more exhaustive characterization of the construction and the constraints on its constituents than can be afforded here). The reflexive subjective-transitive construction: The subject/speaker (NP1) expresses a (non-cancellable) direct, personal, judgemental stance involving a high degree of commitment towards the state of affairs/ action envisioned about himself/herself (NP2). Let us briefly illustrate the relevance of this constructional meaning with respect to examples (23a)–(23b) above. They both convey the result of a self-perception by the subject/speaker, which boils down to an evaluation fully endorsed by the subject/ speaker. Thus, by way of illustration, examples (23a)–(23b) above could not be felicitously combined with e.g. (25)–(26) below: (25) a. # but I really do not think that I am incredulous at all. b # but I personally have doubts as to whether I am really incredulous. (26) a. #pero yo ni siquiera he leído a Cervantes. but 1sg neg even pfvaux.1sg read.ptcp obj Cervantes[name] ‘but I haven’t even read anything by Cervantes’.

 Francisco Gonzálvez-García

b. # pero but



yo no cre-o en absoluto 1sg neg think-prs.1sg in absolute

que sea como Cervantes. comp be.subjv.prs.1sg like Cervantes[name] ‘but I do not think that I’m like Cervantes at all’.

Within a Goldbergian constructionist analysis, the encoding of a direct, personal judgement encoding a high degree of commitment is most parsimoniously attributed to the semantics of the construction rather than to that of the verb. Moreover, these semantico-pragmatic features are inherited from the higher-level subjective-transitive construction. However, the lower-level reflexive-subjective transitive contributes a further meaning nuance, viz. that of self-judgement/perception (with or without an introspection nuance). Compelling evidence for the need to posit this construction in English and Spanish stems from coercion effects of the type exemplified in (27)–(29) below (see Michaelis 2003, 2004a, 2004b for further discussion of the relevance of coercion for a constructionist analysis of the type invoked in Goldberg’s Cognitive Construction Grammar): (27) a. By that reckoning I can certainly count myself a ‘real climber’. (BNC ECH 324). b. (…) *I can certainly count all my friends real climbers. (28) a. I knew myself in his power; as never before I felt myself powerless, a thin benavelled creature without understanding. (BNC HA0 645). b. *I knew all those people around him in his power. (29) a. Ramón quizá se piens-a líder Ramón[name] perhaps 3sg.refl think-prs.3sg leader.m.sg

de es-e otr-o fascismo musolinian-o of dist.m-sg other-m.sg fascism.m.sg Mussolinean-m.sg



y español and Spanish.m.sg



‘Ramón perhaps thinks himself the leader of that other Mussolinean and Spanish fascism’. (CREA, 1995, Francisco Umbral, Leyenda del César Visionario, Novela). b. *La opinión públic-a piens-a a def.f.sg opinion.f.sg public-f.sg think-prs.3sg obj

Ramón líder de es-e otr-o Ramón[name] leader.m.sg of dist.m.sg other-m.sg



fascismo musolinian-o y español fascism.m.sg Mussolinean-m.sg and Spanish.m.sg

‘Public opinion thinks Ramón the leader of that other Mussolinean and Spanish fascism’.



Measuring out reflexivity in secondary predication in English and Spanish 

in the light of the coercion effects illustrated in (27)–(29), two points should be highlighted as compelling arguments for a constructionist account in general and for the existence of the reflexive subjective-transitive construction in particular. First, a number of verbs such as count, or know in English and pensar (‘think’) in Spanish are acceptable in the secondary predication environment with the proviso that the object slot is realized by a reflexive pronoun rather than a lexical NP. Second, and more crucially for our purposes here, a number of cognition verbs which lexically encode little or no judgemental meaning at all (e.g. know, pensar ‘think’) end up being construed as ‘consider’-type verbs in this syntactic environment, thus conveying the expression by the subject/speaker of a judgemental stance/perception about himself/herself. A necessarily brief account of the main restrictions impinging on the construction’s elements, most notably, the subject and the XPCOMP is provided below (see Gonzálvez-García 2007 for further reference). The lower-level reflexive subjectivetransitive configuration can be seeing as inheriting two important properties from the higher-level subjective-transitive construction in English and Spanish: First, the main clause subject must be human or likely to be at least metonymically construed as such, as illustrated in (30)–(31) below: (30) Encima, est-e hotel, que se las da besides prox.m.sg hotel m.sg rel refl.3sg acc.3pl give-prs.3sg de estrella-s, es mucho peor que of star.f-pl be.prs.3sg much worse than

el otr-o. def.m.sg other-m.sg

‘Besides, this hotel, which fancies itself a luxurious one, is much worse than the other’. (CREA, 1981, Carlos Pérez Merinero, Días de Guardar). (31) *Est-a manzana se las da de dulce. prox.f.sg apple.f.sg refl.3sg acc.3pl give.prs.3sg of sweet.sg *‘This apple fancies itself sweet’.

Thus, este hotel (‘this hotel’) can be understood as an instance of people-involving entity for the people involved type of metonymy, and can be taken to refer to the people working in the hotel or the experts who have accorded this hotel the degree of excellence in question (see further Panther & Thornburg 1999; Ruiz de Mendoza & Otal 2002; Barcelona 2003; Dirven 2005, and references therein, inter alios). By contrast, no such metonymic explanation is feasible in the case of esta manzana (‘this apple’), which violates the restriction that the subject must be construable as a human entity, and thus yields an unacceptable result in the reflexive subjectivetransitive construction. Second, the XPCOMP must be construed as entertaining an evaluative, characterizing (rather than an identifying/equative) relation with the preceding reflexive pronoun. As in the case of the subjective-transitive construction, proper nouns are only

 Francisco Gonzálvez-García

felicitous in the xpcomp slot with the proviso that these can be construed metonymically as involving a characterization, as illustrated in (32)–(33) below: (32) Era un tipo atildad-o, un be.imppret.3sg indf.m.sg fellow.m.sg affected-m.sg indf.m.sg niñat-o de San Gervasio que brat.m.sg from San Gervasio[name] rel

pese a despite to

dár-se-las de Mozart, a mí, give.inf-refl.3sg-acc.3f.pl of Mozart[name] dat 1sg.dat rezum-ando brillantina, me record-aba más ooze-ger brilliantine.f.sg acc.1sg remind-imppret.3sg more

a Carlos obj Carlos

Gardel. Gardel [name]

‘He was an affected fellow, an immature guy from San Gervasio, who, despite fancying himself Mozart, to me, oozing brilliantine, seemed more like Carlos Gardel’. (CREA, 2001, Carlos Ruiz Zafón, La Sombra del Viento). (33) *Se las dab-a de Juan Fernández Pérez. refl.3sg 3f.pl.acc give.imppret-3sg of Juan Fernández Pérez [name] *‘He fancied himself Juan Fernández Pérez’.

Crucially, this generalization can be taken to be semantically motivated insofar as the semantic classes of elements occupying the XPCOMP slot in the reflexive subjectivetransitive construction are clearly subjective in the sense of e.g. Scheibman (2002) (see further Gonzálvez-García 2007). In particular, most of the adjectives occurring in this slot in both languages belong to the category of what Dixon (1991: 79) calls ‘human propensity’, such as lucky, satisfied, pleased, interested, etc, in English and satisfecho con algo (‘satisfied with something’), listo (‘clever’), gracioso (‘funny’), macho (‘macho’) in Spanish, which Scheibman (2002: 130) identifies as one of the most frequent classes of adjectives associated with the expression of subjectivity in American English conversation. the anatomy of the reflexive subjective-transitive construction can be formally represented as in Figure 1. This figure follows by and large the notation system invoked in Goldberg (1995, 2006), which is grounded on the assumption that linking patterns are directly correlated with one or more semantic structures. One of these linking patterns is the reflexive-subjective transitive construction, whose core semantics is summarized in the top left corner. Below the core semantic representation of the construction we can find a brief list of verbs which can instantiate this construction in English and Spanish. Recall that the lumping of English and Spanish verbs in this figure has been done for expository convenience. In order to do full justice to the complexity of the data, different anatomies would have to be posited for the Spanish and English reflexive-subjective transitive constructions, respectively. The pairing of the semantic roles of the construction’s constituents (as well as their information structure functions)



Measuring out reflexivity in secondary predication in English and Spanish  Sem SELF-EVALUATION

〈conceptualizer + human

R: instance, PRED means 〈CONFESS, PROFESS, DISCOVER, KNOW, REPUTE, SHOW/ RECONOCER, CONFESAR, PENSAR, SABER〉 Information〈 TOPIC Structure Syn

V

theme

attribute〉

TOPIC

FOCUS 〉

SUBJ

OBJ

NP

NP

XPCOMP NP (characterizing) AP PP (non-literal, nonlocative) AdvP (non-literal, nonlocative) (−ED/−ING) participle (characterizing)

Figure 1.  The anatomy of the reflexive subjective-transitive construction.

with the grammatical functions (i.e. subject, direct object and XPCOMP, respectively) is regulated by the Semantic Coherence Principle and the Correspondence Principle (Goldberg 1995: 50). However, Goldberg (2006: 38–41) places added emphasis on the importance of pragmatic context and frame semantic knowledge (see Fillmore 1985) for determining when a verb’s participant roles can fuse with a construction’s argument roles. Finally, it must be emphasized that Figure 1 (and Figure 2) also differ from the standard Goldbergian formalism in making room for the mappings between syntactic functions, on the one hand, and their morphological realizations in relation to their inherent meaning-function properties.3 To illustrate the point, consider the case of PPs as XPCOMPs in the secondary predication syntactic environment. Both the higher-level subjective-transitive construction and the lower-level reflexive subjectivetransitive require that that the XPCOMP should be construed as evaluative, in keeping with the semantics of the construction. This would explain why, for instance, literal,

.  The reader is referred to Gonzálvez-García (2006) for a discussion of the changes introduced in the mapping of the morphosyntactic realizations of the XPCOMP onto their inherent meaning properties in the Goldbergian-type of construction.

 Francisco Gonzálvez-García

locative PPs are disallowed in these two constructions. Thus, consider (34) below in the case of English (although the same claim can be made for Spanish): (34) a. I consider him off his rocker/*in room 8 b. He considers himself out of place /*in room 8

Thus, whether one is off his rocker or not is a subjective state of affairs which can be subjected to evaluation on the part of the subject/speaker. By contrast, whether one is in room 8 or not is, under normal circumstances, a fairly objective state of affairs which clashes with the overwhelming subjective (i.e. evaluative) semantics of these two constructions. In the current formalism employed by Goldberg, the construction only maps from meaning to syntax, and thus no regulations regarding word order are in principle captured in the anatomy of the construction. To round off this necessarily brief explanation of how Figure 1 (and Figure 2) should be read off, an important observation needs to concern us in the context of the present volume. However, while Goldberg-type constructions of the type examined here are recognized as important for the comprehension of conventionalized utterances, it has been pointed out that they are insufficient “to predict the full range of a verb’s argument in novel utterances based on non-conventionalized verb senses” (Boas 2003: 100–103, 2008a, 2008b for further discussion of the problems impinging on the semantic constraints proposed by Goldberg). Much in line with Boas’ proposal to assign a more crucial role to verbal semantics in the production and interpretation of constructions is the driving force behind the Lexical Constructional Model (LCM henceforth), as outlined in Ruiz de Mendoza Ibáñez & Mairal Usón 2006, 2007, in press; Mairal Usón 2008; Ruiz de Mendoza 2008b, Mairal & Ruiz de Mendoza this volume. This model is explicitly advertised as bridging the gap between a ‘moderate functional model’ such as Role and Reference Grammar (Van Valin 2005) (RRG henceforth) and the non-monotonic, cognitively-influenced Goldbergian strand of CxG (1995, 2006). In this connection, it is worth emphasizing that the LCM builds on the achievements of the Functional Lexematic Model (Faber & Mairal Usón 1999) and, unlike CxG and RRG, places added emphasis on the lexicon, hierarchically organized into semantic classes, to provide robust generalizations regarding the fusion of verbs with constructions (see Butler this volume for a thorough examination of the phylogenesis of the LCM).

5.  Th  e self-descriptive subjective-transitive construction in English and Spanish Let us start off by considering again examples (4a)–(4b), repeated below for ease of exposition as (35a)–(35b) below: (35) a. The young James found himself a virtual prisoner of the Red Douglasses in Edinburgh castle. (BNC AS7 0994).



Measuring out reflexivity in secondary predication in English and Spanish 

b. Al final de es-a temporada pens-é to.def.m.sg end of dist-f.sg season think-indefpret.1sg

que acab-aría mi trayectoria en comp finish-conditional.1sg 1sg.poss career in



el Barcelona, pero sin dar-me def.m.sg Barcelona.football.team but without give.inf-acc1sg



cuenta, me encontr-é en el paro. account 1sg.refl find-indefpret.1sg in def.m.sg unemployment

‘At the end of the season I thought that I would end my professional career in the Barcelona football team, but all of the sudden I found myself out of work’. (CREA, ABC Electrónico, 15/11/1997: Valencia. Manuel Conejos).

Gonzálvez-García (2001) treats configurations like the ones in (35a)–(35b) as instances of the self-descriptive subjective-transitive construction, whose general skeletal meaning can be glossed as follows. It should be emphasized from the start that the information represented in Figure 2 is not necessarily of a coarser grain than that presented in Figure 1:4 The self-descriptive subjective-transitive construction:  Xj UNEXPECTEDLY REALIZES Yj IS UNINTENTIONALLY Z Sem.

REALIZE-BE Z,

〈Experiencer

Undergoer

Attribute〉

TOPIC

FOCUS

INVOLVEMENT R:instance PRED [FIND/ENCONTRAR SEE/VER] Pragm. Syn.

TOPIC V

SUBJECT

OBJ./SUBJ.2 XPCOMP NP# (characterizing) AP PP AdvP (−ED/−ING) participle

Figure 2.  The anatomy of the self-descriptive subjective-transitive construction.

Xj=NP1, Yj= NP2, Z=XPCOMP .  The hash sign after NP in the anatomy of this construction is taken to imply that this morphosyntactic realization is marginally acceptable in English, but not in Spanish.

 Francisco Gonzálvez-García

In addition to the comments made on the interpretation of Figure 1, it should be noted that the construction in question is meaningful, and its core semantics is represented in capitals, following the Goldbergian system of notation. The elements X, Y and Z have the status of arguments of the construction, that is, semantically (and syntactically) obligatory elements. In other words, their omission yields a dramatic meaning change or an unacceptable result. Thus, consider (36) below: (36) a. I found myself a virtual prisoner of the Red Douglasses in Edinburgh castle : : ↛I found myself : : ↛ I found a virtual prisoner of the Red Douglasses in Edinburgh castle. b. Me encontr-é en el paro : : ↛ Me 1sg.refl find-indefpret.1sg in def.m.sg unemployment 1sg.refl

encontr-é ↛ Encontr-é en el paro. find-indefpret.1sg find-indefpret.1sg in def.m.sg unemployment

‘I found myself out of work’ ↛ ‘I found myself ’↛ * ‘I found out of work’.

Moreover, the sub-index ‘j’ indicates that X and Y (that is, the subject and the object) must be coreferential. In order to demonstrate that the self-descriptive subjective-transitive construction qualifies as a construction in its own right, I will provide a brief summary of the most distinctive semantico-pragmatic and discourse-functional inherent properties of this construction which cannot be observed in other complementation strategies attested with cognition verbs in English and Spanish. These are presented in (i)–(vi): i.  Non-agentive, non-volitive perception/cognition: Those verbs implying processes of perception requiring an agentive rather than an experiencer subject appear to be incompatible with this configuration, as shown in the acceptability contrasts reproduced in (37)–(38) below: (37) a #I didn’t look at/encounter/contemplate/watch myself whispering “Let him go and you stay, Izzie”. (Examples adapted from BNC FAJ 0670). b. I found myself whispering “Let him go and you stay, Izzie”. (38) a. *De Of

repente sudden

me mir-é/ refl.1sg look-indefpret.1sg



vigil-é/ observ-é escrib-iendo watch-indefpret.1sg observe-indefpret.1sg write-ger



sobre ti el día de about dat.2sg def.m.sg day of

tu muerte. poss.2sg death

*‘All of a sudden I looked at/watched/observed myself writing about you the day of your death’. (Examples adapted from CREA, La Vanguardia, 10/03/1994).



Measuring out reflexivity in secondary predication in English and Spanish 

b. De repente me encontr-é escrib-iendo sobre Of sudden refl.1sg find-indefpret.1sg write-ger about

ti el día de tu muerte. dat.2sg def.m.sg day of poss.2sg death



‘All of a sudden I found myself writing about you the day of your death’.

Further compelling evidence for the fact that these configurations are construed as non-intentional actions is that they are felicitous with e.g. the “What happened was (that) …” test (Jackendoff 2007: 246). Thus consider the acceptability contrasts reproduced in (39) for English below (the same claim can be duplicated for Spanish): (39) a. What he did was find a couple of tickets for the concert : : #What he did was find himself with a couple of tickets for the concert at the last minute. b. What happened was that he found himself with a couple of tickets for the concert at the last minute.

ii.  Unexpected state of affairs/action/process: Those matrix verbs conveying an anticipated prediction or expectation by the subject/speaker clash with the default unpredictability implication associated with this syntactic environment. Thus consider (40)–(41) below: (40) #As expected/To my surprise, I found myself appointed chairman of the research committee. (Example adapted from Van Ek and Robat 1984: 312 and approved by native informants, material in italics inserted by the author). (41) Teng-o yo un-a pesadilla en la que have-prs.1sg 1sg indef-f.sg nightmare in def.f.sg rel #como cabr-ía esper-ar/ as there.be.room-conditional.3sg expect-inf sorpresa, surprise

para for

mi poss.1sg

me ve-o decrépit-o dentro de refl.1sg see-prs.1sg decrepit-m.sg within of

un-o-s año-s. indf-m-pl year-pl ‘I have a nightmare in which #as is only to be expected/to my surprise, I see myself decrepit within a few years’. (Example adapted from CREA, El País, 02/08/1987; material in italics inserted by the author).

iii.  Stage-level construal of the XPCOMP: The XPCOMP must encode a stage-level predicate (i.e. one denoting temporary, accidental properties of the entity encoded in the NP2 occupying the object slot) (Bolinger 1973; Milsark 1974; Carlson 1977; Kratzer 1988; Escandell-Vidal & Leonetti 2002, inter alios). This sheds some light on

 Francisco Gonzálvez-García

why PPs and adverbial phrases with a literal locative meaning and, to some extent also, ‘-ed’ and ‘-ing’ clauses with a dynamic flavour are favoured in this construction. By contrast, individual-level predicates (whether characterizing or identifying), which encode permanent or inherent properties, invariably yield an unacceptable result in this syntactic environment. (42) *I found myself blue-eyed/intelligent/a man/George Bush. (Example created by the author and approved by native informants). (43) *La verdad es que, def.f.sg truth be.prs.3sg comp

por la mañana, for def.f.sg morning

me encontr-é con ojo-s azul-es/ refl.1sg find-indefpret.1sg with eye-pl blue-pl inteligente/ un hombre/ George intelligent indf.m.sg man George

Bush. Bush[name]

*‘In actual fact, in the morning, I found myself blue-eyed/intelligent/a man/ George Bush’. (Example adapted from CREA, ABC electrónico, 06/8/1992).

This semantic requirement may also shed some light on why NPs are not very frequent in this construction and practically non-felicitous in Spanish. NPs are not prototypical categories to encode stage-level properties. The example reproduced in (45) below is somewhat exceptional insofar as it allows the omission of the non-finite form, while also yielding an acceptable result with the nominalized ordinal NP. (44) The BBC found itself the unwilling accomplice of the advertisers. (BNC A6Y 0738). (45) En la vida política nunca he quer-ido in def.f.sg life political never pfvaux.1sg want-ptcp ser el primer-o, siempre he prefer-ido be.inf def.m.sg first-m.sg always pfvaux.1sg prefer-ptcp est-ar en la trastienda. Pero los últim-o-s be-inf in def.f.sg backstage but def.m.pl last-m-pl año-s ha-n sido de hiperactividad, porque year-pl pfvaux-3pl be.ptcp of hyperactivity because

me encontr-é (siendo) el primer-o de refl.1sg find-indefpret.1sg be.ger def.m.sg first-m.sg of



la lista sin hab-er-lo previsto. def.f.sg list without pefv.aux-inf-acc.3sg.n foresee.ptcp

‘In political life, I have never wanted to be the first, I have always preferred to remain in the background. But the last few years have been hyperactive because I found myself being the first in the list without having foreseen it’. (CREA, Diario de Navarra, 05/05/1999: Fermín Ciaurriz, brackets added by the author).



Measuring out reflexivity in secondary predication in English and Spanish 

(46) *Me encontr-é médico/ profesor de física. refl.1sg find-indefpret.1sg doctor teacher of physics *‘I found myself a doctor/professor of physics’.

iv.  Preference for a contingent, negative state of affairs/process/action: The XPCOMP in this construction prototypically encodes a negative, risky, difficult state of affairs or situation which the main clause subject is unexpectedly faced with. Thus, consider (47)–(48) below (see Tables 3–4): (47) Contemporary scholars find themselves in difficulty to accept this. (BNC CB6 0418). (48) Entre otr-a-s catástrofe-s, me ve-o sin Among other-f-pl disaster-pl refl.1sg see-prs.1sg without tema de conversación con es-o-s matrimonio-s. topic of conversation with dist-m-pl couple-pl ‘Among other disasters, I see myself without a topic of conversation with those couples’. (CREA, El País, 01/06/1988).

However, it must be stressed that the prosody of the XPCOMP does not necessarily have to be negative. Some positive states of affairs may also occur, albeit marginally: (49) I find myself pleased when patients fail to attend. (ICE-GB W1B-013–12). (50) Y si ya ven-ía alucin-ando And if already come-imppret.1sg hallucinate-ger lo mío con la mism-a desde mediados def.n.sg poss.1sg with def.f.sg same-f.sg from middle de su segund-a temporada (…) me of poss.3sg second-f.sg season refl.1sg encontr-é con un final de temporada find-indefpret.1sg with indf.m.sg end of season diez veces mejor de lo que esper-aba. De es-o-s ten times better of def.n.sg rel expect-imppret.1sg of dist-m-pl que te rel acc.2sg

dej-an clav-ad-o en el sillón. leave-prs.3pl stick-ptcp-m.sg in def.m.sg armchair

‘And if I was already having a superb time with the same [TV show] from half way through its second time, I found myself with a season finale ten times better than I had expected. It was one of those that leaves you flabbergasted in your armchair’. (CREA, 2003, Efímero, Páginas web).

v.  Informational salience of the XPCOMP: The more informationally salient a given XPCOMP is, the more felicitous it will be in this frame. Following Lambrecht (1994: 207) (see also Ackerman & Goldberg 1996), the term ‘focus’ is here taken to

 Francisco Gonzálvez-García

refer to the unpredictable or pragmatically non-recoverable element in an utterance. The focus element is represented in the examples below in small capitals. (51) A. I found myself the object of a takeover. B. That’s a lie! You found yourself the victim of a terrorist attack. B’. That’s a lie! # You did not find yourself in any specific state/situation. (52) A. Y entonces rápidamente me encontr-é And then quickly refl.1sg find-indefpret.1sg

escrib-iendo por dinero. write-ger for money



‘And then quickly I found myself writing for money’. (CREA, Lo más plus, 13/12/96, Canal Plus).

B. Es-o es mentira. Tú te dist-n.sg be.pres.3sg lie. 2sg refl.2sg

encontr-aste escrib-iendo por pur-a vocación. find-indefpret.2sg write-ger for pure-f.sg vocation



‘That’s a lie! You found yourself writing out of true vocation’.

B.’ #Es-o es mentira. Tú no te dist-n.sg be.pres.3sg lie. 2sg neg refl.2sg

encontr-aste hac-iendo nada. find-indefpret.2sg do-ger nothing



# ‘that’s a lie. You didn’t find yourself doing anything’.

It should be remarked that sometimes informational salience may save an otherwise unacceptable result. Thus, consider by way of illustration (53) below: (53) #He finds himself the only person : : He finds himself the only person in the building.

vi.  Preference for a human main clause subject: The main clause subject prototypically designates a human being in this frame. However, non-human or even nonanimate subjects may occur if construed metonymically as standing for human beings (see also example (49) above): (54) Britain finds itself with no recourse. (BNC CAF 1342) (=the people(s) of Britain). (55) El país se encontr-ó def.m.sg country refl.3sg find-indefpret.3sg

sin without

profesorado: (…). teachers ‘The country found itself without teachers’. (CREA, El País. Educación, 10/03/2003: Donde la Iglesia se moja).



Measuring out reflexivity in secondary predication in English and Spanish 

The examples reproduced in (54)–(55) can be characterized in terms of the peopleinvolving entity for the people involved type of metonymy. In (54)–(55), Britain and El país (‘the country’) can be taken to refer either to the leaders or the inhabitants of the countries involved. Note that the occurrence of a non-human subject without a metonymic interpretation is highly indicative of the fact that the configuration is not reflexive at all. Thus, consider (56)–(57) below: (56) # The house finds itself abandoned. (57) La casa se encuentr-a def.f.sg house pronomclitic.3sg find-prs.3sg abandon-ad-a. abandon-ptcp-f.sg

# ‘The house founds itself abandoned’.

Thus, despite their reflexive formal appearance, the sentences in (56)–(57) are not reflexive at all. They do not fulfil the two criterial properties of reflexives outlined in sections three and four. The reflexive cannot be felicitously replaced with a clitic without yielding a dramatic meaning difference. In addition, the subject cannot be expanded by a sí misma (‘itself ’). (58) a. La casa la encuentr-a abandon-ad-a. def.f.sg house acc.3sg.f.sg find-prs.3sg abandon-ptcp-f.sg ‘S/he finds the house abandoned’. b. #La casa se encuentr-a a sí def.f.sg house pronomclitic.3sg find-prs.3sg obj dat.3sg

mism-a abandon-ad-a. same-f.sg abandon-ptcp-f.sg



#‘the house founds itself abandoned’.

i will now go on to demonstrate that the features outlined in (i)–(vi) above qualify as inherent semantico-pragmatic hallmarks of the self-descriptive subjective-transitive construction which cannot be derived from any other complementation strategies attested with cognition verbs in English and Spanish. (59) a

Me encontr-é escrib-iendo por dinero. refl.1sg find-indefpret.1sg write-ger for money ‘I found myself writing for money’.

b. *Me consider-é escrib-iendo por dinero. refl.1sg consider-indefpret.1sg write-ger for money *‘I considered myself writing for money’.

 Francisco Gonzálvez-García

c. Encontr-é a mi-s amig-o-s find-indefpret.1sg obj poss.1sg-pl friend-m-pl

escrib-iendo por dinero. write-ger for money



‘I found my friends writing for money’.

d. Encontr-é que est-aba escrib-iendo por dinero. find-indefpret.1sg comp be-imppret.1sg write-ger for money ‘i found that I was writing for money’.

Example (59a), which qualifies as an instance of the self-descriptive subjectivetransitive construction, conveys an unexpected situation in which the subject/speaker finds himself/herself probably without his/her intending to do so. By contrast, such an event is incompatible with the reflexive subjective-transitive construction, among other things because the inherently objective, factual-like meaning of the event in question clashes with the semantics of the construction, which requires that the state of affairs/ situation encoded in the complement clause is a matter of judgement/evaluation on the part of the subject/speaker. The non-reflexive counterpart of (59a), as in (59c), is normally associated with the expression of a direct perception by the subject speaker, which can be paraphrased as follows: “When I found my friends, they were writing for money”. Finally, (59d) is, unlike the previous examples, not an instance of secondary predication, but rather of monotransitive complementation with a finite complement clause. Configurations of this kind express the discovery of a fact based on indirect rather than direct evidence (see further Rodríguez Espiñeira 1989; Gonzálvez-García 2003, 2006; Langacker 2004; García-Miguel & Comesaña 2004, inter alios). Having provided a general overview of the main semantico-pragmatic motivations for treating the configurations as constructions in their own right, namely, instances of the self-descriptive subjective-transitive construction, I shall now go on to provide a finergrained bottom-up corpus-driven analysis of this construction in English and Spanish. The general distribution of find and encontrar (‘find’) in relation to the morphosyntactic realization of the XPCOMP in the BNC and CREA corpora are reproduced in Tables 1–2: From the examination of the data reproduced above, a number of considerations emerge which need to concern us here: First, English find displays a considerably higher degree of association (98.55%) with the self-descriptive subjective-transitive construction than its Spanish counterpart (62.69%). Second, an interesting feature is that there are important differences regarding the choice of the subject (and thus the person of the reflexive form) in English and Spanish. Thus, the English construction has an overriding preference for first person singular subjects, a feature inherited from the subjective-transitive construction (see Borkin 1973: 45–46, Gonzálvez-García 2003, 2006, 2008). By contrast, the most productive

24/3086

(0.77%)

187/3086

(6.05%)

Total



(0.72%)

– 4/552

(1.06%)

1/94

(0.52%)

1/189

(0.11%)

1/889

(5.55%)

16/880



(4.89%)

– 27/552

(2.12%)

2/94

(7.93%)

15/189

(7.42%)

66/889

(7.38%)

65/880

(0%)

oneself

yourselves themselves

ourselves

itself

herself

himself

(3.44%)

0/29

(0.44%)

(2.43%)

1/29

2/452

11/452

myself

yourself

np

ap

Find

(25.11%)

775/3086



(29.89%)

– 165/552

(54.23%)

51/94

(33.33%)

63/189

(15.86%)

141/889

(24.54%)

216/880

(31.03%)

9/29

(28.76%)

130/452

pp

(2.72%)

84/3086



(4.89%)

– 27/552

(7.44%)

7/94

(2.11%)

4/189

(2.47%)

22/889

(1.93%)

17/880

(3.44%)

1/29

(1.32%)

6/452

adv. p

(43.22%)

1334/3086

(100%)

1/1

(29.16%)

– 161/552

(27.65%)

26/94

(22.75%)

43/189

(50.61%)

450/889

(39.77%)

355/880

(58.62%)

17/29

(62.16%)

281/452

gerund

(20.09%)

620/3086



(30.43%)

– 168/552

(7.44%)

7/94

(33.33%)

63/189

(73.01%)

211/889

(23.97%)

211/880

(3.44%)

1/29

(4.86%)

22/452

‘-ed’ part.

(98.56%)

3086/3131

(100%)

1/1

(98.92%)

– 552/558

(98.94%)

94/95

(96.92%)

189/195

(98.34%)

889/904

(90.09%)

880/888

(82.85%)

29/35

(99.34%)

452/455

Total

Table 1.  Distribution of the morphosyntactic realizations of the XPCOMP with find + reflexive pronoun in the secondary predication frame in the BNC

Measuring out reflexivity in secondary predication in English and Spanish 

TOTAL

uno (‘one’)

2pl.refl 3pl.refl

1pl.refl

3sg.refl

2sg.refl

1sg.refl

ENCONTRAR

(55.55%)

(11.11%)

(24.48%)

(14.59%)

(28.57%)

(40.21%)

452/1765

(25.60%)

(36.94%)

652/1765

2/64

(3.12%)

21/64

– 41/281

(32.81%)

– 113/281

12/49

(24.32%)

(40.11%)

14/49

245/1007

404/1007

5/9

(41.40%)

(27.88%)

1/9

147/355

con +   alguien

99/355

con + algo

(8.25%)

145/1756

(7.81%)

5/64

(10.32%)

– 29/281

(4.08%)

2/49

(9.33%)

94/1007



(4.22%)

15/355

con + queclause

(9.04%)

157/1735

(18.75%)

12/64

(21.35%)

– 60/281

(8.16%)

4/49

(5.85%)

59/1007

(11.11%)

1/9

(5.91%)

21/355

pp

(5.34%)

91/1701



(1.77%)

– 5/281

(10.20%)

5/49

(5.46%)

55/1007

(11.11%)

1/9

(7.04%)

25/355

gerund

86/1756

(4.89%)

(7.81%)

5/64

(4.27%)

– 12/281

(18.36%)

9/49

(5.26%)

53/1007



(1.97%)

7/355

‘ed’participle

88/1716

(5.12%)

(9.37%)

6/64

(2.49%)

– 7/281



(5.75%)

58/1007

(11.11%)

1/9

(4.50%)

16/355

ap

94/1756

(5.35%)

(20.31%)

13/64

(4.98%)

– 14/281

(6.12%)

3/49

(3.87%)

39/1007



(7.04%)

25/355

phrase

adverbial





– –









NP

9/12

(62.69%)

1765/2815

(94.11%)

64/88

(44.25%)

--281/635

49/60

(81.66%)

(63.57%)

1007/1584

(75%)

(81.42%)

355/436

Total

Table 2.  Distribution of the morphosyntactic realizations of the XPCOMP with encontrar + reflexive pronoun in the secondary predication frame in the CREA

 Francisco Gonzálvez-García



Measuring out reflexivity in secondary predication in English and Spanish 

combination in absolute terms is the impersonal form uno (‘one’), which is scarcely found in its English counterpart. What is more, the first person plural is slightly more frequent than the first person singular in Spanish. Third, the distribution of the English construction is more homogeneous than that of its Spanish counterpart. In particular, the English construction is frequently attested with the third person (singular and plural), while the frequency of these forms is considerably lower in Spanish. Fourth, an interesting point of convergence is that the second person plural is not attested in this construction in English or Spanish. These two languages differ dramatically in one interesting respect which is crucial for our purposes here, namely, the morphosyntactic realization of the XPCOMP. Thus, while the English construction frequently selects as XPCOMP gerunds, prepositional phrases and ‘-ed’ participles (see (60) below), a different picture emerges in the case of Spanish. The three most frequent combinations are con algo (‘with something’), con alguien (‘with somebody’) and prepositional phrases, which are only slightly more frequent than con que-finite clause (‘with that-clause’). Interestingly enough, the con alguien and con que-finite clause realizations of the XPCOMP exemplified in (61) below do not have a felicitous English counterpart. (60) a. (…) I found myself wanting to alert readers to an increasing amount of detailed literature, (…). (BNC B15 553). b. (…) I found myself in the dim bathroom of my hut staring at a haunted face. (BNC APC 986). c. I found myself frustrated by unanswered questions at every turn: (…). (BNC AKB 190). (61) a. (…) Me refl.1sg

encontr-é con un e-mail find-indefpret.1sg with indf.sg e-mail



suy-o en mi correo. 3sg.poss-m.sg in poss.1sg mail



‘I found myself with an e-mail written by him/her in my mailbox’. (CREA, 2003, Efímero, 03206004. Weblog 2003; España, Páginas Web).

b. Cuando volv-ía a casa me when return-imppret.1sg to home refl.1sg

encontr-é con un grupo de find-indefpret.1sg with indf.sg group of



gente muy people very

numeroso (…). numerous

‘As I was going back home, I ran into a very large group of people’. (CREA, 1992, José Antonio Piqueras, El movimiento obrero, España, Historia, Anaya, Madrid).

 Francisco Gonzálvez-García

c. En plen-a noche, nev-ando, me in full-f.sg night snow-ger refl.1sg

encontr-é find-indefpret.1sg

en in

la def.f.sg

calle. street

‘In the middle of the night, snowing, I found myself in the street’. (CREA, 1998, Carmen del Val, Nacho Duato. Por vos muero. España. Danza. Martínez Roca. 1998). d. (…)

Me encontr-é con que refl.1sg find-indefpret.1sg with comp

te hab-ías cas-ado (…). pronomclitic.2sg pfvaux-imppret.2sg marry-ptcp

‘I found out that you had got married’. (CREA, 1985, Adelaida García Morales, El sur seguido de Bene, España, Novela, Anagrama, Barcelona, 1994).

In addition, English allows, albeit marginally, NPs as XPCOMPs, while no instance of this realization has been attested in our Spanish data. Thus, consider (62) below: (62) I found myself second curate in a large mission called Kibuye on the outskirts of Kisumu town. (BNC A7K 1363).

At this stage a digression is necessarily in order regarding the relevance of the morphosyntactic realization of the XPCOMP to the continuum between reflexives and middles in English and Spanish. Other things being equal, an important difference between instances of the reflexive subjective-transitive construction and the self-descriptive subjective-transitive construction in English and Spanish concerns the degree of elaboration of the state of affairs/event encoded. Thus, the former conveys what Haiman (1998) labels a ‘divided self ’, where the very same referent is profiled by two distinct discourse roles (often, a more conscious one – an experiencer – and a more emotional self – an affected self – see further Ariel 2008 and references therein). By contrast, the latter is closer to a (one-participant) internal event/state of affairs, and can very often be paraphrased as an intransitive or intensive configuration. (63) a. I find myself incredulous. (BNC EF0 1199) → According to my personal views, I am incredulous. b. (…) me encuentr-o muy atractiv-o con ellos. → 1sg.refl find-prs.1sg very attractive-m.sg with 3pl ‘I find myself very attractive in these’. En mi opinión, estoy muy atractiv-o con ellos. In poss.1sg opinion be.prs.1sg very attractive-m.sg with 3pl ‘In my opinion, I am very attractive in these’. (CREA, 2003, Prensa, Tp Teleprograma, nº 1963, 17–23/11/2003: TVE 1 estrena “Arroz y tartana” con Maura y Sancho, TeleRevistas S.A. Hachette Filipacchi, Madrid 2003).



Measuring out reflexivity in secondary predication in English and Spanish 

(64) a. The young James found himself a virtual prisoner of the Red Douglasses in Edinburgh castle. (BNC AS7 0994) → The young James was a virtual prisoner of the Red Douglasses in Edinburgh castle. b. (…) me encontr-é en el 1sg.refl find-indefpret.1sg in def.m.sg

paro → (…) est-aba en el paro. unemployment imppret.1sg in def.m.sg unemployment



‘I was unemployed’. (CREA, ABC Electrónico, 15/11/1997: Valencia. Manuel Conejos).

However, the continuum between the elaboration of events in terms of one or two participants cannot only be discerned between the reflexive subjective-transitive construction, on the one hand, and the self-descriptive subjective-transitive construction, on the other. Crucially, within the latter construction, a continuum can also be posited vertically depending on the inherent meaning and form properties of the XPCOMP. Thus, for instance, examples (61a), (61b), and (61d) above are closer to two-participant transitive events than (61c), which can be paraphrased as a one-participant intensive clause. Thus consider (65): (65) a.

Encontr-é un e-mail suyo. find-indefpret.1sg indf.m.sg e-mail 3sg.poss-m.sg ‘i found one of his/her emails’ (two-participant event).

b. Encontr-é un grupo de gente muy numeroso. find-indefpret.1sg indf.m.sg group of people very numerous ‘I ran into a very large number of people’ (two-participant event). c.

Est-aba en la calle. be-imppret.1sg in def.f.sg street ‘I was in the street’ (one-participant event).

Those realizations of the XPCOMP through a gerund may rank higher or lower on the scale of event elaboration depending on the degree of transitivity of the verb in question (in the sense of Hopper & Thompson 1980), as illustrated in (66) below: (66) a. (…) De est-e modo, me encontr-é of prox-m.sg way refl.1sg find-indefpret.1sg

presid-iendo un mitin socialista. chair-ger indf.m.sg political.meeting socialist

‘This way I found myself chairing a socialist political meeting’ → ’I was chairing a socialist political meeting’ (two-participant event). (CREA, 1982, PRENSA, ABC, 11/10/1982; Prensa española, S.A., Madrid, 1982).

 Francisco Gonzálvez-García

b (…) me encontr-é pregunt-ándo-me a refl.1sg find-indefpret.1sg wonder-ger-refl.1sg obj

mí mism-o por qué dat.1sg same-m.sg for what

Casablanca Casablanca[name]



hab-ía sido en su momento pfvaux-imppret.3sg be.ptcp in poss.3sg moment



Casablanca y no Tánger, (…). Casablanca[name] and neg Tangier[name]

‘I found myself wondering myself why Casablanca had been in its day Casablanca and not Tangier’ → ‘I was wondering why Casablanca had been in its moment Casablanca and not Tangier’ (one-participant event). (CREA, 1994, Juan José Armas Marcelo, Madrid, distrito federal, España, Novela, Seix Barral, Barcelona, 1994).

Lower on the scale are prepositional phrases, adverbial phrases, past participles, which are near-equivalent to one-participant events, as shown, for instance, in (61c) and (62b), to cite two telling examples.5 Before going into a more detailed analysis of the specific combinations attested in our data, a brief comment is in order regarding polarity. All the English combinations were attested in the declarative form. The Spanish combinations were also found in the declarative form, with only two exceptions in the interrogative form. A representative example is reproduced in (67) below: (67) ¿ A que no sab-e usted con lo que To comp neg know-prs.2sg 2sg with def.n.sg rel me encontr-é en México capital?. refl.1sg find-indefpret.1sg in Mexico[name] capital ‘Surely you don’t know the situation in which I found myself in Mexico City’. (CREA, 2001, Santi Ortiz Trixac, Lances que cambiaron la Fiesta, España, Tauromaquia, Madrid, Espasa Calpe).

This observation can be taken to point to the fact that the primary function of the construction is to make an assertion regarding an unexpected event/state of affairs envisioned in the complement clause.

.  An anonymous reviewer points out that my claim that reflexives encode two-participant events and middles only one participant events may well be right from a syntactic point of view, but not from a semantic viewpoint, given that both configurations describe two-participant events. The reviewer then suggests that perhaps it is a question of how the profiling of each construction influences the mapping of the semantics to syntax. I find myself in complete agreement with this suggestion. However, given that the syntax of argument structure still remains a challenge for CxG (Boas 2008a, 2008b), I leave it for future research to further explore the viability of this working hypothesis.



Measuring out reflexivity in secondary predication in English and Spanish 

In line with the bottom-up, corpus-driven usage-based approach invoked here, the remainder of this section will outline the most frequent combinations attested in our data. These are shown in Table 3 and Figure 3 for English and Table 4 and Figure 4 for Spanish. Table 3.  The most frequent collocates with find in the self-descriptive subjective construction in the BNC find oneself

Tokens

wondering something thinking something unable alone saying something back surrounded in difficulties/in trouble able involved in something doing something Total

52 50 43 34 31 24 15 13 13 10 7 292

60 50 40 30 20 10

eth

ing

ed

som

olv

do ing

inv

abl e

rou bl e

in

di cul t

ies /

in t

k

nd ed

rou

bac

sur

ing say

ne alo

abl e un

ing nk

thi

wo nd

eri

ng

0

Figure 3.  The most frequent collocates with find in the self-descriptive subjective-transitive construction in the BNC.

 Francisco Gonzálvez-García

Table 4.  The most frequent collocates with encontrar (‘find’) in the self-descriptive subjective construction in the CREA Encontrarse (‘find oneself ’) en una (difícil) situación/posición ‘in a (difficult) situation/position’ en un lugar ‘at a given place’ solo/a ‘alone’ con un problema ‘with a problem’ con una (agradable/desagradable) sorpresa ‘with a(n) (pleasant/unpleasant) surprise’ a gusto ‘comfortable/at home’ mal ‘unwell’ bien/mejor well/better con una (buena/mala) noticia ‘with a (good/bad) item of news’ sentado/a ‘seated’ Total

Tokens 45 25 24 23 20 12 12 10 10 6 187

In the light of the data reproduced in Tables 3–4, a number of observations suggest themselves which need to concern us here: First, in agreement with the usage-based approach invoked in this article, the frequent collocates outlined in Tables 3–4, together with the sequence “find/encontrar (‘find’) + reflexive pronoun” can be considered constructions in their own right, even if their properties can be fully predicted from the self-descriptive subjective-transitive construction in English and Spanish (Langacker 1988: 23, 2005: 139–143; Croft 2001: 28 Goldberg & Jackendoff 2004: 533, fnt. 1; Goldberg 2006: 224, in press; Bybee 2006: 714 Bybee & Eddington 2006: 328, inter alios). Second, the semantico-pragmatic profiles of the XPCOMPs in English and Spanish by no means coincide perfectly, but there are striking similarities regarding the states of affairs/events/situations likely to occur (e.g. alone/solo/a, in a difficult situation/position, in trouble/en una (difícil) situación/posición (‘in a difficult situation/position’), con un problema (‘with a problem’). Third, while the actions/situations/states of affairs encoded in the self-descriptive subjective-transitive construction are not exclusively negative, it is nonetheless evident that hostile, negative or dangerous situations prevail over positive ones. By way of illustration, consider the fact that unable is almost three times more frequent than able.



Measuring out reflexivity in secondary predication in English and Spanish  45 40 35 30 25 20 15 10 5

o/a tad

cia

sen

oti

ejo a/m

ala

)n

n/m bie

uen a (b un con

le/d dab con

un

a (a

gra

r

l ma

ust o

ag

sa pre

ma

sor

le)

ble

esa

gra

un con

situ il) ifíc a (d un en

dab

pro

sol

o/a

ar lug

un

en

aci

ón

/po

sic

ión

0

Figure 4.  The most frequent collocates with encontrar (‘find’) in the self-descriptive subjectivetransitive construction in the CREA.

A similar claim can be made about the Spanish collocates, where, for instance, con una mala/desagradable noticia (‘with a bad/unpleasant item of news’) is twice more frequent than its positive counterpart.

6.  Closing remarks and outlook This article has been concerned with showing that formally identical instances of secondary predication featuring find/encontrar (‘find’) and a reflexive pronoun in the object slot can be aptly considered to be two different, though closely connected, constructions

 Francisco Gonzálvez-García

(i.e. form-function pairings), namely, the reflexive subjective-transitive construction and the self-descriptive subjective-transitive construction. While both constructions qualify as reflexive, since they allow the substitution of the reflexive pronoun for a clitic and, in the case of Spanish, also for the expansion through a a sí mismo (‘to oneself ’) phrase, a number of interesting differences have nonetheless been pinpointed: Instances of the reflexive subjective-transitive construction impose an agentive, intentional construal on the event/state of affairs in question, while the self-descriptive subjective-transitive construction requires a non-volitional, non-agentive construal. Configurations of the reflexive subjective-transitive construction are closer to a two-participant event elaboration in which the entity encoded in the main clause and the reflexive is construed as a ‘divided self ’ between an experiencer subject and an affected object. By contrast, instances of the self-descriptive subjective-transitive construction are functionally equivalent to one-participant events, and can often be paraphrased by means of intransitive or intensive clauses. The continuum between reflexives (encoding two-participant events) and middles (encoding one-participant events) can also be observed vertically within the selfdescriptive subjective-transitive construction. The crucial determinant in this respect has been argued to be the inherent meaning and form properties of the XPCOMP and its transitivity properties (Hopper & Thompson 1980). Empirical evidence has been provided that English and Spanish share a considerable number of morphosyntactic realizations of the XPCOMP. However, the inventory of such realizations is not fully symmetrical. Thus, for instance, certain NPs yield an acceptable result in the selfdescriptive subjective-transitive construction in English, while these are unacceptable in Spanish. By contrast, the Spanish counterpart favours con + algo (‘with something’) and con + que-clause (‘with + finite that-clause), which are not felicitous in English in this environment. The contrastive analysis presented in the preceding pages thus lends further credence to Croft’s (2003) contention that argument structure is not only construction-specific but also language-specific. Figure 5 graphically represents the macroscopic view of the continuum between reflexives and middles in the secondary predication frame in English and Spanish. Following a default inheritance system of the type invoked in Cognitive Construction Grammar (Goldberg 2006), the reflexive-subjective transitive construction is taken to be an instance of the higher-level construction, viz. the subjective-transitive construction. An added advantage of a default or partial inheritance system of the type invoked in Goldberg’s Construction Grammar (Goldberg 2006) is that it allows us to capture the commonalities as well as the idiosyncratic particulars among higher-level and lower-level constructions. This point is of paramount importance for constructionist approaches which acknowledge a non-trivial interaction of constructions in the construct-i-con (Jurafsky 1992). Specifically, this representation system enables us to capture the commonalities between the lower-level construction (i.e. the reflexive



Measuring out reflexivity in secondary predication in English and Spanish 

configuration) in relation to the higher-level subjective-transitive construction, especially those concerning the type of subject and the semantico-pragmatic profile of the XPCOMP (see further Gonzálvez-García 2003, 2006, 2008). Closely connected with the subjective-transitive construction is the self-descriptive subjective-transitive construction which, unlike the reflexive-subjective transitive construction, may encode twoparticipant or one-participant events. Further down in the cline are pseudo-reflexive configurations which, despite their formal reflexive appearance, cannot be considered to be reflexive on semantico-pragmatic grounds. Rather, these unambiguously encode one-participant events/situations and are closer to middles than to reflexives.

− re exive

+re exive SUBJECTIVETRANSITIVE CONSTRUCTION I find her a bore La encuentro aburrida IE

(PSEUDO)-REFLEXIVE CONSTRUCTION

SELF-DESCRIPTIVE SUBJECTIVETRANSITIVE CONSTRUCTION

*The house finds itself abandoned La casa se encuentra abandonada

I found myself alone in a room Me encontré solo en una habitación

− re exive

REFLEXIVE SUBJECTIVETRANSITIVE CONSTRUCTION

He finds himself very attractive Él se encuentra muy atractivo +re exive

Figure 5.  The reflexivity continuum in secondary predication in English and Spanish.

A few comments are in order regarding the interpretation of the picture presented in Figure 5. On the one hand, the figure captures the fact that the reflexive subjective-transitive construction, which encodes the highest degree of reflexivity in the secondary predication frame, is a lower-level configuration of the higher-level subjective-transitive construction. The arrow connecting the two constructions

 Francisco Gonzálvez-García

indicates that the relation between them is one of elaboration. In other words, the verbs which occur in the reflexive subjective-transitive construction modulate the general meaning of the subjective-transitive construction in terms of e.g. selfevaluation, self-perception, etc. The lower-level inherits a number of properties from the higher-level construction regarding the semantico-profile of the subject, the direct object, and more crucially of the XPCOMP, which must encode an state of affairs likely to be construed by the subject/speaker as conveying a judgemental rather than factual-like stance. Horizontally, from right to left, the reflexive subjective-transitive construction, the self-descriptive subjective-transitive construction and the (pseudo)-reflexive constructions represent three points in the reflexivity continuum. The arrows with discontinuous dots are meant to reflect the fact that there is a smooth transition and even overlap between the different constructions in the reflexivity continuum (e.g. “I found myself pleasant to Munro” : : ‘It happened that Munro found me pleasant’ : : ‘After having a long conversation, I thought I was pleasant to Munro’). Note that the reflexive subjective-transitive construction would correspond to the subtypes I and IV in Martín Zorraquino’s classification, while what has been provisionally labelled here as (pseudo-) reflexive constructions would correspond to Martín Zorraquino’s subtype IX. Interestingly enough, the self-descriptive subjective-transitive construction would fall under the I subtype in Martín Zorraquino’s classification, thus obscuring the fact that this construction has a different semantics from that of the reflexive subjective-transitive construction and, what is more important for our purposes here, the fact that they instantiate two different points in the reflexivity continuum. However, the proposal presented here has been quite modest: the generalizations emerging from the preceding pages have been drawn on data from decoding in conjunction with native speaker’s judgements through questionnaires. As Boas (2005, 2008a, 2008b) has rightly argued, encoding is as important as decoding, especially if one is to take a usage-based approach seriously. In the case of the constructions under examination here, experimental evidence of all sorts (e.g. sentence completion tasks, reading experiments, etc) is necessary to further refine the sketchy picture of the construction and the frequent collocates which has been provided. In addition, it might be interesting to explore, for instance, the implications of the distinction between a past tense and present tense in the continuum. Instances of the selfdescriptive subjective construction are considerably more frequent with the matrix verb in the past than in the present. A second important avenue for future research lies in the investigation of the discourse-functional properties of this construction in larger pieces of discourse than the ones employed in this article. In any case, we hope to have succeeded in arguing the case for a constructionist, usage-based account of the continuum between reflexives and middles in the secondary predication frame in English and Spanish.



Measuring out reflexivity in secondary predication in English and Spanish 

References Aarts, Bas. 1995. Secondary predicates in English. In The verb in contemporary English. Theory and description, Bas Aarts & Charles F. Meyer (Eds), 75–101. Cambridge: CUP. Ackerman, Farrell & Adele E. Goldberg. 1996. Constraints on adjectival past participles. In Discourse and language, Adele E. Goldberg (Ed.), 17–30. Stanford CA: CSLI. Ariel, Mira. 2008. Pragmatics and grammar. Cambridge: CUP. Barcelona, Antonio. 2003. Metonymy in Cognitive Linguistics. In Motivation in language. Studies in honour of Günter Radden, Hubert Cuyckens, Thomas Berg, René Dirven & KlausUwe Panther (Eds), 223–255. Amsterdam: John Benjamins. Barlow, Michael. 1996. Corpora for theory and practice. International Journal of Corpus Linguistics 1(1): 1–37. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman grammar of spoken and written English. Harlow: Pearson Education. Boas, Hans C. 2003. A constructional approach to resultatives. Stanford CA: CSLI. Boas, Hans C. 2005. Determining the productivity of resultative constructions: A reply to Goldberg and Jackendoff. Language 81: 448–464. Boas, Hans C. 2008a. Resolving form-meaning discrepancies in Construction Grammar. In Constructional reorganization, Jaako Leino (Ed.), 11–36. Amsterdam: John Benjamins. Boas, Hans C. 2008b. Identifying syntactically relevant units of meaning. Plenary delivered at the 26th AESLA Conference. University of Almería, Spain. April, 2008. Bolinger, Dwight. 1973. Essence and accidence: English analogs of Hispanic ser-estar. In Issues in Linguistics: Papers in Honor of Henry and Renée Kahane, Braj B. Kachru, Robert B. Lees, Yakou Malkiel, Angelina Pietrangeli & Sol Saporta (Eds), 57–69. Urbana IL: University of Illinois Press. Borkin, Ann. 1973. To be or not to be. In Papers from the ninth regional meeting of the Chicago Linguistic Society, Claudia Corum, Cedric Smith-Stark & Ann Weiser (Eds), 44–56. Chicago IL: The Chicago Linguistics Society. British National Corpus. http://sara.natcorp.ox.ac.uk/. Burnard, Lou. 1995. Users reference guide for the British National Corpus. Oxford: Oxford University Computing Services. Bybee, Joan L. 2006. From usage to grammar: The mind’s response to repetition. Language 82(4): 711–733. Bybee, Joan L. & David Eddington. 2006. A usage-based approach to Spanish verbs of ‘Becoming’. Language 82(2): 323–355. Carlson, Greg N. 1977. Reference to kinds in English. Ph.D. dissertation, University of Massachusetts at Amherst. Croft, William. 2001. Radical Construction Grammar: Syntactic theory in typological perspective. Oxford: OUP. Croft, William. 2003. Lexical rules vs. constructions: a false dichotomy. In Motivation in language. Studies in honour of Günter Radden, Hubert Cuyckens, Thomas Berg, René Dirven & Klaus-Uwe Panther (Eds), 49–68. Amsterdam: John Benjamins. Demonte, Violeta & Pascual Masullo. 1999. La predicación: Los complementos predicativos. In Gramática descriptiva de la lengua española, Vol. 2: Las construcciones sintácticas fundamentales. Relaciones temporales, aspectuales y modales, Ignacio Bosque & Violeta Demonte (Eds), 2461–2523. Madrid: Espasa Calpe.

 Francisco Gonzálvez-García Dirven, René. 2005. Major strands in Cognitive Linguistics. In Cognitive Linguistics: Internal dynamics and interdisciplinary interaction, Francisco José Ruiz de Mendoza Ibáñez & Sandra Peña Cervel (Eds), 69–100. Berlin: Mouton de Gruyter. Dixon, Robert M.W. 1991. A new approach to English grammar, on semantic principles. Oxford: Clarendon Press. Escandell-Vidal, Victoria & Manuel Leonetti. 2002. Coercion and the stage/individual distinction. In From words to discourse: Trends in Spanish semantics and pragmatics, Javier Gutiérrez-Rexach (Ed.), 159–179. Amsterdam: Elsevier. Faber, Pamela B. & Ricardo Mairal Usón. 1999. Constructing a lexicon of English verbs [Functional Grammar Series 23]. Berlin: Mouton de Gruyter. Faltz, Leonard M. 1977. Reflexivization: A study in universal syntax. Ph.D. dissertation, University of California at Berkeley. (Reprinted 1985; New York: Garland) Fillmore, Charles J. 1985. Frames & the semantics of understanding. Quaderni di Semantica 6: 222–254. Fried, Mirjam. 2007. Constructing grammatical meaning. Isomorphism and polysemy in Czech reflexivization. Studies in Language 31(4): 721–764. Fried, Mirjam & Jan-Ola Östman. 2005. Construction Grammar: A thumbnail sketch. In Construction Grammar in a cross-language perspective [Constructional Approaches to Language 2], Mirjam Fried & Jan-Ola Östman (Eds), 11–86. Amsterdam: John Benjamins. Fukaya, Teruhiko. 2002. On viewing reflexives in the Bank of English. Their distribution and function. In English corpus linguistics in Japan, Toshio Saito, Junsaku Nakamura & Shunji Yamazaki (Eds), 77–91. Amsterdam: Rodopi. García, Erica. 1975. The role of theory in linguistic analysis: The Spanish pronoun system [NorthHolland Linguistic Series 19]. Amsterdam: North-Holland. García-Miguel, José M. & Susana Comesaña. 2004. Verbs of cognition in Spanish: constructional schemas and reference points. In Linguagem, cultura e cognição: Estudos de Linguística Cognitiva, Amadeu Torres, Augusto Soares de Silva & Miguel Gonçalves (Eds), 367–384. Coimbra: Almedina. Goldberg, Adele E. 1995. Constructions (A Construction Grammar approach to argument Structure). Chicago & London: The University of Chicago Press. Goldberg, Adele E. 1996. Construction Grammar. In Concise encyclopedia of syntactic theories, Keith Brown & Jim Miller (Eds), 68–71. Oxford: Elsevier. Goldberg, Adele E. 1998. Patterns of experience in patterns of language. In The new psychology of language [Cognitive and Functional Approaches to Language Structure], Vol. 1, Michael Tomasello (Ed.), 203–219. Mahwah NJ: Lawrence Erlbaum. Goldberg, Adele E. 2006. Constructions at work: The nature of generalization in language. Oxford: OUP. Goldberg, Adele E. In press. The nature of generalization in language. Cognitive Linguistics. Goldberg, Adele E. & Ray Jackendoff. 2004. The resultative as a family of constructions. Language 80: 532–568. Gonzálvez-García, Francisco. 2001. ‘I found myself led into the path of constructions’: A brief overview of how construction-based approaches can furnish new ways of understanding English grammar. In Recent approaches to English grammar, Luisa González Romero, Montserrat Martínez Vázquez, Beatriz Rodríguez Arrizabalaga & Pilar Ron Vaz (Eds), 151–182. Huelva: Grupo de Gramática Contrastiva.



Measuring out reflexivity in secondary predication in English and Spanish 

Gonzálvez-García, Francisco. 2003. Reconstructing object complements in English and Spanish. In Gramática de Construcciones (Contrastes entre el inglés y el español), Montserrat Martínez Vázquez (Ed.), 17–58. Huelva: Grupo de Gramática Contrastiva. Gonzálvez-García, Francisco. 2006. Passives without actives: Evidence from verbless complement constructions in Spanish. Constructions SV1–5/2006. Gonzálvez-García, Francisco. 2007. ‘Saved by the reflexive’: Evidence from coercion via reflexives in verbless complement clauses in English and Spanish. Annual Review of Cognitive Linguistics 5: 193–238. Gonzálvez-García, Francisco. 2008. The family of object-related depictives in English and Spanish: Towards a usage-based, constructionist analysis. Language Sciences doi: 10.1016/j. langsci.2008.01.003. Gonzálvez-García, Francisco & Christopher S.  Butler. 2006. Mapping functional-cognitive space. Annual Review of Cognitive Linguistics 4: 39–96. Gries, Stefan T., Beate Hampe & Doris Schönefeld. 2005. Converging evidence: Bringing together experimental and corpus data on the association of verbs and constructions. Cognitive Linguistics 16(4): 635–676. Haiman, John. 1998. Talk is cheap: Sarcasm, alienation, and the evolution of language. Oxford: OUP. Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56(2): 251–299. Huddleston, Rodney & Geoffrey Pullum. 2002. The Cambridge grammar of the English language. Cambridge: CUP. Ibarretxe-Antuñano, Iraide. 2006. Cross-linguistic polysemy in tactile verbs. In Cognitive Linguistics investigations: Across languages, fields and philosophical boundaries [Human Cognitive Processing 15], June Luchjenbroers (Ed.), 235–253. Amsterdam: John Benjamins. International Corpus of English (ICE). http://www.ucl.ac.uk/english-usage/ice. Jackendoff, Ray. 2007. Language, consciousness, culture. Essays on mental structure. Cambridge MA: The MIT Press. Janda, Laura A. 1993. The semantics of Russian and Czech reflexives. In American contributions to the eleventh international congress of Slavists, Robert A.  Maguire & Alan Timberlake (Eds), 310–319. Columbus OH: Bratislava Slavica. Jurafsky, Dan. 1992. An on-line computational model of human sentence interpretation: A theory of the representation and use of linguistic knowledge. Ph.D. dissertation, University of California, Berkeley, and Report No. UCB/CSD 92/676, Computer Division, University of California, Berkeley. Kemmer, Suzanne E. 1993. The middle voice [Typological Studies in Language 23]. Amsterdam: John Benjamins. Kemmer, Suzanne E. 1994. Middle voice, transitivity and events. In Voice: Form and function, B. Fox & Paul J. Hopper (Eds), 179–230. Amsterdam: John Benjamins. Kemmer, Suzanne E. 2003. Human cognition and the elaboration of events. In The new psychology of language [Cognitive & Functional Approaches to Language], Vol. 2, Michael Tomasello (Ed.), 89–118. Mahwah, NJ: Lawrence Erlbaum. Kratzer, Angelika. 1988. Stage-level and individual-level predicates. In Genericity in natural language, Manfred Krifka (Ed.), 247–284. University of Tübingen: SNS-Bericht 88–42, Seminar für natürlich-sprachliche Systeme.

 Francisco Gonzálvez-García Lambrecht, Knud. 1994. Information structure and sentence form. A theory of topic, focus, and the mental representation of discourse referents [Cambridge Studies in Linguistics 71]. Cambridge: CUP. Langacker, Ronald W. 1988. An overview of Cognitive Grammar. In Topics in Cognitive Linguistics [Current Issues in Linguistic Theory 50]. Brygida Rudzka-Ostyn (Ed.), 3–48. Amsterdam: John Benjamins. Langacker, Ronald W. 2003. Constructions in Cognitive Grammar. English Linguistics 20: 41–83. Langacker, Ronald W. 2004. Aspects of the grammar of finite clauses. In Language, culture and mind, Michel Achard & Suzanne Kemmer (Eds), 535–577. Stanford CA: CSLI. Langacker, Ronald W. 2005. Construction Grammars: Cognitive, radical and less so. In Cognitive Linguistics: Internal dynamics and interdisciplinary interaction [Cognitive Linguistics Research 32] , Francisco J. Ruiz de Mendoza Ibáñez & Sandra Peña Cervel (Eds), 101–159. Berlin: Mouton de Gruyter. Lázaro Mora, Fernando A. 1983. Observaciones sobre se medio. In Serta Philologica F. Lázaro Carreter: Natalem diem sexagesimun celebranti dicata [Estudios de lingüística y lengua literaria 1], Emilio Alarcos Llorach & Fernando Lázaro Carreter, 301–307. Madrid: Cátedra. Luján, Marta. 1977. El análisis de los verbos reflexivos incoativos. Revista de la Sociedad Española de Lingüística 7: 97–120. Mairal Usón, Ricardo. 2008. An overview of the Lexical Constructional Model (LCM). Part I: Lexical templates. Presentation delivered at the 26th AESLA Conference. University of Almería, Spain. April, 2008. Maldonado, Ricardo. 1999. A media voz: Problemas conceptuales del clítico se. [Publicaciones del Centro de Lingüística Hispánica 46]. Universidad Nacional Autónoma de Méjico: Instituto de Investigaciones Filológicas, Ciudad de Méjico. Maldonado, Ricardo. 2007. Grammatical voice in Cognitive Grammar. In The Oxford handbook of Cognitive Linguistics, Dirk Geeraerts & Hubert Cuyckens (Eds), 829–868. Oxford: OUP. Maldonado, Ricardo. 2008. Spanish middle syntax: A usage-based proposal for grammar teaching. In Cognitive approaches to pedagogical grammar – Volume in honour of René Dirven, Sabine De Knop & Teun De Rycker (Eds), 150–190. Berlin: Mouton de Gruyter. Martín Zorraquino, María Antonia. 1979. Las construcciones pronominales en español: Paradigmas y desviaciones. Madrid: Gredos. Martínez Vázquez, Montserrat. 1998. Diátesis. Alternancias oracionales en la lengua inglesa. Huelva: Grupo de Gramática Contrastiva. Mendikoetxea, Amaya. 1999. Construcciones inacusativas y pasivas. In Gramática descriptiva de la lengua español, Vol. 2: Las construcciones sintácticas fundamentales. Relaciones temporales, aspectuales y modales, Ignacio Bosque & Violeta Demonte (Eds), 1575–1722. Madrid: Espasa Calpe. Michaelis, Laura A. 2003. Headless constructions and coercion by construction. In Mismatch (Form-function incongruity and the architecture of grammar), Elaine J.  Francis & Laura A. Michaelis (Eds), 259–310. Stanford CA: CSLI. Michaelis, Laura A. 2004a. Why we believe that syntax is construction-based. Plenary delivered at the 3rd International Conference on Construction Grammar(s), Université de Provence, Marseille, July 9, 2004. Michaelis, Laura A. 2004b. Type shifting in Construction Grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15(1): 1–67.



Measuring out reflexivity in secondary predication in English and Spanish 

Milsark, Gary L. 1974. Existential sentences in English. Ph.D. dissertation, Cambridge MA. Panther, Klaus-Uwe & Linda L. Thornburg. 1999. Coercion and metonymy: The interaction of constructional and lexical meaning. In Cognitive perspectives on language [Polish Studies in English Language and Literature 1], Barbara Lewandowska-Tomaszczyk (Ed.), 37–52. Frankfurt: Peter Lang. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. London: Longman. Real Academia Española: Banco de datos (CREA) [on line]. Corpus de referencia del español actual. http://www.rae.es. Accessed February-July 2006. Rodríguez Espiñeira, María José. 1989. El complemento predicativo del complemento directo en Español. Ph.D. dissertation, University of Santiago de Compostela, Spain. Ruiz de Mendoza Ibáñez, Francisco J. 2008a. Cross-linguistic analysis, second language teaching and cognitive semantics: The case of Spanish diminutives and reflexive constructions. In Cognitive approaches to pedagogical grammar – Volume in honour of René Dirven, Sabine De Knop & Teun De Rycker (Eds), 121–152. Berlin: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco J. 2008b. An overview of the Lexical Constructional Model (LCM), Part II: Subsumption processes. Presentation delivered at the 26th AESLA Conference. University of Almería, Spain. April, 2008. Ruiz de Mendoza Ibáñez, Francisco J. & José L. Otal Campo. 2002. Metonymy, grammar, and communication. Albolote, Granada: Comares. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. 2006. Levels of semantic representation: where lexicon and grammar meet. Interlingüística, 17: 26–47. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Aspects of meaning construction, Günter Radden, Klaus-Michael Köpcke, Thomas Berg & Peter Siemund (Eds), 33–51. Amsterdam: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco J. & Ricardo Mairal Usón. in press. Levels of description and constraining factors in meaning construction: an introduction to the Lexical Constructional Model. Folia Linguistica 42(2). Sánchez López, Cristina. 2002. Las construcciones con se. Estado de la cuestión. In Las Construcciones con Se [Colección Gramática del Español 8], Cristina Sánchez López (Ed.), 13–163. Madrid: Visor Libros. Scheibman, Joanne. 2002. Point of view and grammar. Structural patterns of subjectivity in American English conversation. Amsterdam: John Benjamins. Sweetser, Eve. 1990. From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure [Cambridge Studies in Linguistics 54]. Cambridge: CUP. Thompson, Sandra A. 2002. Object complements and conversation: Towards a realistic account. Studies in Language 26: 125–163. Van Ek, Jan A. & Nico J. Robat. 1984. The student’s grammar of English. Oxford: Basil Blackwell. Van Valin, Robert D., Jr. 2005. Exploring the syntax-semantics interface. Cambridge: CUP. Vera Luján, Agustín. 1996. Esquemas oracionales ergativos reflexivos. Estudios Lingüísticos de la Universidad de Alicante 11: 385–409.

The inchoative construction Semantic representation and unification constraints* Francisco J. Cortés Rodríguez This article provides an analysis of English inchoative structures within the framework of a functionally-based conception of language and, specifically, of the lexicon. This theoretical framework – the Lexical Constructional Model (LCM henceforth) – proposes a lexical component composed of two central elements: a repository of lexical units grouped into lexical classes, which are established on the basis of the commonality of meaning of predicates, and a catalogue of constructions, which is also devised as having internal organization. The LCM also postulates that lexical-constructional subsumption is subject to the conditions imposed on the semantic compatibility between predicates and constructions. Conditions invoke higher level cognitive mechanisms like metonymy and metaphor and lower-level semantic restrictions affecting event or argument structure in semantic representations. The analysis of lexical subsumption within the inchoative construction will be subject to two types of restrictions: firstly, there is an external constraint affecting the unification of causative predicates and inchoative structures. This external constraint is based on a high-level metonymic process which has been labelled process for action: an action is treated as if it were a process that in turn stands for the action. Secondly, unification is conditioned by some internal constraints imposed upon the semantic structure of predicates. Among these there are also two subtypes: (1) constraints on the event structure of predicates, which make reference to the codification of telicity and causativity in the case of causative/inchoative verbs; (2) constraints on the arguments of lexical templates, among which the ‘agentcauser blocking’ and the ‘cause expletivization’ constraints play a crucial role. The analysis of these constraints will in fact reveal the feasibility and explanatory potential of the LCM for meaning construction.

*Financial support for this research has been provided by the DGI, Spanish Ministry of Educa-

tion and Science, grants FF12008–05035-C02–02 and HUM2005–07651-C02–01. The research has been co-financed through European Regional Development Funds. I am most grateful to Prof. C. Butler and the two anonymous reviewers for their comments on this article since they have led to improvements on both the format and content of the original.

 Francisco J. Cortés Rodríguez

1.  Introduction The study presented in this article is part of a more ambitious research project with two major goals: first, the establishment of the structure of the so-called ‘constructicon’ or catalogue of constructions that forms part of the lexical component of the Lexical Constructional Model (Ruiz de Mendoza & Mairal 2006, 2007a, 2007b; Mairal and Ruiz de Mendoza in press). Second, the development of the process of Unification between the constructicon and the repository of lexical units, the Thesaurus. Because of its functional and cognitive-constructionist spirit, the LCM proposes that the derivation of syntactic structures from the lexical component involves the unification between predicates and constructions. The LCM postulates that unification is subject to the conditions imposed on the semantic compatibility between predicates and constructions. Conditions invoke higher level cognitive mechanisms like metonymy and metaphor and lower-level semantic restrictions affecting event or argument structure in semantic representations. The second section of this article provides a general description of the components of the LCM and a more detailed description of the system of semantic representation utilized for both lexical units and constructions. In this regard, section 2.2 proposes a partial reformulation of LCM semantic structures by means of the integration of qualia theory as developed in Pustejovsky’s (1995) Generative Lexicon. After establishing the framework for the analysis, section 3 offers a detailed description of the inchoative construction, in order to illustrate in detail in what way constructions should be encoded in the constructicon. I will seek to unravel the semantic complexity of the inchoative construction and the constraints that play a role in the fusion – and in the blocking as well – of several lexical classes. I will also show that the adoption of a unitary system for representation both in the thesaurus and in the constructicon is in fact a great advantage to spell out the internal constraints that mediate unification processes, as they can be described with the same language used for lexical and constructional templates. The article is rounded off in section 4.

2.  An overview of the LCM As mentioned before, this study adopts the LCM as a model for description. Figure 1 shows a schematic diagram of this model in its most recent version (Ruiz de Mendoza & Mairal 2007b, see also this volume). Thus, the LCM has a level 1 or core grammar module consisting of elements with syntactically relevant semantic interpretation. It also has a pragmatic or level 2 module that accounts for low-level inferential aspects of linguistic communication. There is a level 3 module dealing with high-level inferences (i.e. illocutionary force). Finally, a level 4 module accounts for the discourse aspects of the LCM, especially cohesion and coherence phenomena. The lexical component, as



The inchoative construction: Semantic representation and unification constraints 

envisaged in the LCM, is located at Level 1 and is composed of a thesaurus, or repository of lexical units with their corresponding lexical templates and a catalogue of Level 1 constructions of a language – the constructicon – also endowed with a semantic description. The interaction between these two modules is determined by a set of subsumption operations and constraints – the Unification Process.

LTs

Level 1 internal and external constraints

Level 1 CTs

CORE GRAMMAR

subsumption/ conceptual cueing

subsumption/ conceptual cueing Level 2 CTs/CSs

subsumption/ conceptual cueing Level 3 CTs/CSs

subsumption/ conceptual cueing

Level 2 internal and external constraints

Level 3 internal and external constraints

Level 4 constraints Level 4 CTs/CSs Discourse representations

conceptual cueing Final meaning interpretation Figure 1.  The overall architecture of the Lexical Constructional Model (from Ruiz de Mendoza & Mairal 2007b) LT = lexical template; CT = constructional template; CS = Conceptual Structure.

 Francisco J. Cortés Rodríguez

Within the context of this model, semantic interpretation is the result of the unification of a lexical template (i.e. a low-level representation of the semantic and syntactic properties of a predicate) and a constructional template (i.e. a high-level – conceptual – representation of the semantic properties of a construction).1 This process integrates predicate argument structure (functional approaches) with linking constructions (constructionist approaches). There is an underlying attitude towards the lexicon and its relation to grammar in the LCM that is obviously also assumed in this article: the LCM places itself in an intermediate position between projectionist and constructionist models, and in doing so aims to reconcile both approaches.

2.1  Microstructure: Semantic representation in the LCM At the heart of the model lies the notion of the lexical template, intended to capture both the semantic and syntactic information encoded within a lexical class.2 This means that a lexicon hierarchically organized into coherent semantic classes – the socalled thesaurus – each of which is represented in terms of a lexical template, figures prominently in this approach. The type of lexical representation in the lexical template is different from some proposals for lexical representation in that the latter use syntax as a means of isolating and identifying those aspects of the meaning of a predicate which form part of a lexical representation. The representation proposed here, however, is one that contains a much more fine-grained semantic decompositional system without losing sight of the syntactically-grounded basis of previous proposals. This means that a lexical template attempts to reach a compromise between the demands of semantics and those of syntax. With regard to the syntactically-driven semantic aspects, it must be stated that lexical templates are enriched semantic representations of the logical structures proposed in RRG; hence they are part of an aspectual event structure theory: logical

.  Some of the semantic components of templates are relevant to account for certain aspects of the syntactic behaviour of lexical units. A case in point is the behaviour of different classes of predicates with regard to the inchoative construction, which is the topic of our analysis in section 3. It is still an issue pending resolution to spell out in detail the linking process from semantic representations to actual syntactic structures. By now, and since LCM templates incorporate Role and Reference Grammar’s Logical representations (Van Valin & LaPolla 1997; Van Valin 2005), it seems plausible to adopt the semantics-to-syntax interface mechanisms propounded by this theory. .  The following description follows in part the works of Mairal & Faber (2005) and Ruiz de Mendoza & Mairal (2006, 2007a, 2007b).



The inchoative construction: Semantic representation and unification constraints 

structures are a means to represent event structures in terms of the temporal internal properties of the predicate. Logical structures are based on the Aktionsart distinctions proposed in Vendler (1957 [1967]), and the decompositional system is a variant of the one proposed in Dowty (1979). Verb classes are divided into states, activities, achievements, semelfactives and accomplishments together with their corresponding causatives. States and activities are primitives, whereas accomplishments and achievements consist of either a state or activity predicate plus a BECOME and an INGR operator; semelfactives – the non-telic variants of achievements – are marked by the operator SEML. The inventory of RRG logical structures is shown in Figure 2. Verb class

Logical structure

State

predicate´ (x) or (x, y)

Activity

do´ (x, [predicate´ (x) or (x, y)]

Achievement

INGR predicate´ (x) or (x, y), or INGR do´(x, [predicate´ (x) or (x, y)]

Accomplishment

BECOME predicate´ (x) or (x, y), or BECOME do´(x, [predicate´ (x) or (x, y)]

Semelfactive

SEML predicate´ (x) or (x, y), or SEML do´(x, [predicate´ (x) or (x, y)]

Active accomplishment

do´ (x, [predicate1´ (x, (y))] & INGR predicate2´ (z, x) or (y)

Causative

α CAUSES ß where α, ß are LS of any type

Figure 2.  Logical structures in RRG.

As has been pointed out on several occasions both outside (Mairal & Faber 2002, 2005; Mairal 2004) and within (Van Valin & LaPolla 1997:  116–118, and

 Francisco J. Cortés Rodríguez

Van Valin & Wilkins 1993) the RRG model, not everything about the semantic complexities of predicates is captured by these formulas. Logical structures do not even explain everything about syntactic variation phenomena, as will be made obvious in the analysis of the inchoative construction that is developed in the following sections of this article. It is necessary to enrich logical structures with other semantic non-aspectual components. This led Mairal & Faber (2005) to propose semantic representations devised as lexical templates consisting of two subparts: a logical structure plus a semantic module. Thus, the skeletal structure of semantic definitions is as follows:

(1) [semantic module] + logical structure = predicate’s meaning

Mairal & Faber (2005) propose the use of dictionaries for the extraction of conceptual information regarding meaning parameters, arguments, semantic roles, etc. The extensive analysis of the English verbal lexicon in Faber & Mairal (1999) led these authors to build up the structure of a verbal thesaurus which takes as a starting point, both for its organization and for the definition of lexical units, a restricted set of archi-units (happen, become, move, go, say, feel, etc.). For example, they use know, which is one of the two superordinate terms for the domain of COGNITION, to define a number of hyponyms like learn, study, understand, fathom, grasp, among others. In fact, they propose to go one step further and consider the superordinate terms for each basic conceptual category as possible candidates for the inventory of more basic terms or primitives. Faber and Mairal observe that their proposal coincides to a great extent with that of Wierzbicka’s Natural Semantic Metalanguage (NSM) research program, which has been shown to be valid for over a hundred languages (cf. Wierzbicka 1996, 1999; Goddard & Wierzbicka 2002). For this reason the proposal in the LCM is to use these primes for the description of the verbal lexicon as well. A further set of elements that can be part of the semantic specification of a predicate in terms of lexical templates is the group of ‘Lexical functions’ as propounded in Mel’čuk’s Explanatory and Combinatorial Lexicology (ECL) (cf. Mel’čuk 1989; Mel’čuk et al. 1995; Mel’čuk & Wanner 1996; Alonso Ramos 2002). It is important to note that these lexical functions have also been shown to have a universal status (cf. Mel’čuk, 1989), something which is in accord with our aim of providing typologically valid representations. One difference with regard to the original conception of Lexical Functions is that they are not used horizontally, but vertically, i.e. to produce paradigmatic distinctions. That is, they will be treated as a mechanism that allows us to combine the set of primes to arrive at more specific lexical items within a lexical class (see Mairal and Ruiz de Mendoza’s article in this volume for a more detailed explanation of these issues). The following example is the lexical entry for Spanish captar (Mairal and Ruiz de Mendoza, this volume):





The inchoative construction: Semantic representation and unification constraints 

(2) captar: [MagnObstr & Culm12[ALL]] know´ (x, y)

The entry for captar has two parts: (i) the semantic component in brackets; (ii) the representation of the logical structure. In this case, this predicate is represented by a state logical structure which takes know´ as a primitive and has two arguments (x, y). This logical structure is in turn modified by the lexical functions (or operators) MagnObstr, which specify the large degree of difficulty involved in carrying out the action. The other lexical function, Culm, captures the end-point of knowing something (i.e. understanding). ALL is another lexical function – of the kind postulated by Mel’čuk – that falls within the scope of the internal variables. This lexical function refers to the propositional content of the object of apprehension.

2.2  Enriching lexical templates: Qualia structure As can be seen, lexical templates are not only enriched semantic representations but also have the advantage of being heavily constrained, since they are made up of elements drawn from a finite set of primitives, lexical functions and logical structures. However, it is possible to further improve their format and theoretical significance by adapting some notions from Pustejovsky’s (1995) lexical theory. The aim of this will be to establish the interrelations that hold between the two types of information encoded in the two modules of lexical templates, the semantic component and the logical structure description. A close scrutiny of the semantic import of lexical functions and primes reveals that they are related to the constituents of the event structure or the argument structure characterization that the logical structures provide in the second module of our templates. Thus, a lexical function like ALL that is part of the representation for captar in Mairal and Ruiz de Mendoza (this volume) as it refers to the object of apprehension can be integrated as a restriction over the second argument (y) in the logical structure. On other occasions the elements of the semantic module complement the semantics of (one of) the event(s) in the logical structure, in a vein similar to what is described as ‘qualia structure’ by Pustejovsky (1995: Chapter 6). Qualia structure would be one level of semantic interpretation, representing the modes of predication possible with a lexical item, the other levels of representation being event, argument and lexical inheritance structures. Qualia structure specifies four essential aspects of a word’s meaning (or qualia) (Pustejovsky 1995: 76, 85–86): –– –– –– ––

constitutive (QC): the relation between an object and its constituent parts formal (QF ): that which distinguishes it within a larger domain telic (QT): its purpose and function agentive (QA): factors involved in its origin or ‘bringing it about’

 Francisco J. Cortés Rodríguez

The following are examples of lexical representations based on this system (adapted from Pustejovsky 1995: 101–102): (3) book argstr = [arg1 = x: information] [arg2 = y: phys_obj] qualia = information ·phys_obj_lcp formal = hold (x,y) telic = read (e,w,x.y) agent = write (e’, v, x.y) (4) kill eventstr = [e1= e1: process/e2 = e2: state] restr = )]

Figure 4.  The cause expletivization constraint.

One interesting effect of this constraint is that it can also work in reverse: many inchoative-only verbs are not constructed causatively if the kind of result state cannot

 Francisco J. Cortés Rodríguez

be caused by an external performer. This is the case of the verbs bloom, blossom and glide in the following examples. (46) The Germans in particular will take an indefinite wordage about land where the lemon blooms. (128.068 \lob_g.txt68) ←→ *They bloomed the lemons. (47) Japan, despite all those paper flowers that blossom in a jam-jar is becoming […] (1.492\lob_r.txt7) ←→*The heat blossomed the flowers. (48) In the old days sprit-sailed barges glided, red-sailed, above the land to village hithes with cattle and […] (61.793\lob_f.txt58) ←→ *The oarsman glided the barge.

Note the unnaturalness of asking about the causing performer participant in the above examples: (49) *Who bloomed the lemons?/*Who blossomed the flowers?/*Who glided the barge?

But it is not so odd to ask about the involvement of the theme participant: (50) What did the lemons do? They bloomed. (51) What did the barge do? It glided above the land.

All this argumentation supports the view that the theme participant of inchoative structures is ‘involved’ in the realization of the change of state or position (which is the essential feature of the notion ‘internal-causation’). It also proves the impossibility of tracing an external causer back to these inchoative-only verbs.

4.  Conclusions This article has offered a study of the inchoative construction in which the most relevant factors for unification with a wide array of lexical classes have been accounted for. From the analysis of inchoatives some interesting conclusions can be drawn: a.  The semantic interpretation of sentences is based on the interrelation of two representational structures, the lexical and the constructional template. Both are central to the LCM. b.  Such an interrelation is based on a process of unification which is subject to a set of internal and external constraints. External constraints are based on cognitive mechanisms that make use of conceptual metaphor and metonymy. c.  External constraints find lower-level correlates in internal constraints. The conceptual mechanisms underlying external constraints have a parallel in the conditions imposed on the semantic components of lexical templates.



The inchoative construction: Semantic representation and unification constraints 

d.  Finally, the fact that constructional and lexical templates are represented semantically by a unitary system makes it more feasible to design internal constraints, as they are also devised in the same format, thus rendering a more elegant and coherent explanation. I believe that the ascertainment of the unification processes associated with constructions along the lines of the LCM proposal is a promising research frame for the exploration of the internal make-up of the constructicon.

References Alexiadou, Artemis, Elena Agnasnostopoulou & Florian Schäfer. 2006. The properties of anticausatives crosslinguistically. In Phases of interpretation, Mara Frascarelli (Ed.), 187–211. Berlin: Mouton de Gruyter. Alonso Ramos, Margarita. 2002. Colocaciones y contorno en la definición lexicográfica. Lingüística Española Actual 24(1): 63–96. Chierchia, Gennaro. 2004. A semantics for unaccusatives and its syntactic consequences. In The unaccusative puzzle. Explorations in the syntax-lexicon interface, Artemis Alexiadou, Elena Agnasnostopoulou & Martin Everaert (Eds), 22–59. Oxford: OUP. Dowty, David R. 1979. Word meaning and Montague Grammar. Dordrecht: Reidel Faber, Pamela & Ricardo Mairal. 1999. Constructing a lexicon of English verbs. Amsterdam: Mouton de Gruyter. Goddard, Cliff & Anna Wierzbicka (Eds), 2002. Meaning and universal grammar: Theory and empirical findings. Amsterdam: John Benjamins. Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago IL: University of Chicago Press. Goldberg, Adele E. 1996. Making one’s way through the data. In Grammatical constructions. Their form and meaning, Masayoshi Shibatani & Sandra A.  Thompson (Eds), 29–53. Oxford: OUP. Goldberg, Adele E. 2006. Constructions at work. The nature of generalization in language. Oxford: OUP. Koontz-Garboden, Andrew. 2006. Monotonicity and the semantics of word formation. In Workshop on syntax, lexicon and event structure. The Hebrew University of Jerusalem. July 4–7, 2006. Levin, Beth. 1993. English verb classes and alternations: A preliminary investigation. Chicago IL: University of Chicago Press. Levin, Beth & Malka Rappaport, H. 1995. Unaccusativity: At the syntax-lexical semantics interface. Cambridge MA: The MIT Press. Mairal, Ricardo. 2004. Reconsidering lexical representations in Role and Reference Grammar. Proceedings of the 27th international AEDEAN conference. University of Salamanca. Mairal, Ricardo & Pamela Faber 2002. Functional Grammar and lexical templates. In New perspectives on argument structure in Functional Grammar, Ricardo Mairal & María J. Pérez (Eds), 41–98. Berlin: Mouton de Gruyter. Mairal, Ricardo & Pamela Faber. 2005. Decomposing semantic decomposition. Towards a semantic metalanguage in RRG. In Proceedings of the 2005 international conference on Role and Reference Grammar: 279–308. Taiwan: Academia Sinica.

 Francisco J. Cortés Rodríguez Mairal, Ricardo & Francisco J. Ruiz de Mendoza. In press. Internal and external constraints in meaning construction: The lexicon-grammar continuum. Estudios de filología inglesa: Homenaje a la Dra. Asunción Alba Pelayo [Colección Varia], Teresa Gibert Maceda and Laura AlbaJuez (Eds). Madrid: UNED. (Ms. Available at http://www.lexicom.es/drupal/publications) Mel’čuk, Igor. 1989. Semantic primitives from the viewpoint of the Meaning-Text linguistic theory. Quaderni di Semantica 10(1): 65–102. Mel’čuk, Igor. 1996. Lexical functions: a tool for the description of lexical relations in the lexicon. In Recent trends in Meaning-Text Theory, Leo Wanner (Eds), 37–102. Amsterdam: John Benjamins. Mel’čuk Igor, André Clas & Alain Polguère. 1995. Introduction à la lexicologie explicative et combinatoire. Louvain-la-Neuve: Duculot – De Boeck. Mel’čuk, Igor & Leo Wanner. 1996. Lexical functions and lexical inheritance for emotion lexemes in German. In Recent trends in Meaning-Text Theory, Leo Wanner (Ed.), 209–227. Amsterdam: John Benjamins. Michaelis, Laura A. 2003. Word meaning, sentence meaning, and syntactic meaning. In Cognitive approaches to lexical semantics, Hubert Cuyckens, René Dirven & John Taylor (Eds), 163–210. Berlin: Mouton de Gruyter. Piñón, Christopher. 2001. A finer look at the causative-inchoative alternation. In Proceedings of semantics and linguistic theory 11, Rachel Hastings, Brendan Jackson & Zsofia Zvolenszk (Eds), 346–364. Ithaca NY: CLC Publications. (Ms. available at http://pinon.sdf-eu.org/ papers/pinon_flcia.pdf) Pustejovsky, James. 1995. The generative lexicon. Cambridge MA: The MIT Press. Reinhart, Tanya. 2002. The theta system – An overview. Theoretical Linguistics 28(3): 229–290. Ruiz de Mendoza, Francisco J. & Olga Díez. 2002. Patterns of conceptual interaction. In Metaphor and metonymy in comparison and contrast, René Dirven & Ralph Pörings (Eds), 489–532. Berlin: Mouton de Gruyter. Ruiz de Mendoza, Francisco J. & Ricardo Mairal. 2006. Levels of semantic representation: where lexicon and grammar meet. Interlingüística 17: 26–47. Ruiz de Mendoza, Francisco J. & Ricardo Mairal. 2007a. High-level metaphor and metonymy in meaning construction. In Aspects of meaning construction, Günter Radden, Klaus-Michael Köpcke, Thomas Berg & Peter Siemund (Eds), 33–49. Amsterdam: John Benjamins. Ruiz de Mendoza, Francisco J. & Ricardo Mairal. 2007b. The LCM. The general architecture of the model. Ms. available at http://www.lexicom.es/drupal/modelarchitecture. Ruiz de Mendoza, Francisco J. & Lorena Pérez. 2001. Metonymy and the grammar: Motivation, constraints, and interaction. Language and Communication 21: 321–357. Van Valin, Robert D. Jr. 2005. Exploring the syntax-semantics interface. Cambridge: CUP. Van Valin, Robert D.  Jr. & Randy J.  LaPolla. 1997. Syntax: Structure, meaning and function. Cambridge: CUP. Van Valin, Robert D. Jr. & David P. Wilkins. 1993. Predicting syntactic structure from semantic representations: Remember in English and its equivalents in Mparntwe Arrernte. In Advances in Role and Reference Grammar, Robert D. Van Valin Jr. (Ed.), 499–534. Amsterdam: John Benjamins. Vendler, Zeno. 1957[1967]. Linguistics in philosophy. Ithaca NY: Cornell University Press. Wierzbicka, Anna. 1996. Semantics: Primes and universals. Oxford: OUP. Wierzbicka, Anna. 1999. Emotions across languages and cultures. Diversity and universals. Cambridge: CUP.

Semantic and pragmatic constraints on the English get-passive* Pilar Guerrero Medina This article focuses on the so-called get-passive, frequently regarded as a problematic construction in the linguistic literature. It is my contention that a lexically-based approach is insufficient to account for the appropriateness of the get-passive, since pragmatic and contextual factors are also crucial to ascertain its acceptability. Basing my discussion on corpus data, I will analyze how the semantics of the get-passive interacts with the semantic properties of verbs from five semantic types that can be integrated within the construction: affect, giving, motion (take-subtype), corporeal and annoying. Along the lines of Goldberg & Jackendoff (2004: 563), I suggest that the get-passive should be treated as a family of constructions in order to account for its semantic and pragmatic properties. Two main subconstructions will be posited: the “causative” get-passive and the “spontaneous” get-passive.

1.  Introduction The English get-passive is a construction with strong aspectual values which is commonly associated with the meanings of causation and result and which often lacks an active counterpart, as illustrated by the examples below (Arce-Arenales, Axelrod & Fox 1994: 15): (1) a. Kelly got lost. (*Someone lost Kelly.) b. He got fogged in. (*The weather fogged him in.)

The get-passive has been regarded as a “linguistic puzzle” (Carter & McCarthy 1999: 54), as “a contentious point of discussion” (Chappell 1980: 411) or even as “the subject of widespread disagreement” (Collins 1996: 43). Following Goldberg (1995: 1),

*The research for this article has been conducted in the framework of the research project HUM2005–07651-C02–02/FILO, funded by the Spanish Ministry of Science and Education and European Regional Development Fund. I would like to thank the editors and two anonymous reviewers for their helpful comments on an earlier version of this article.

 Pilar Guerrero Medina

I start from the assumption that a verb-centered account is clearly unable to account for the appropriateness of the construction: the get-passive should rather be regarded as a contentful unit, contributing its own semantic and pragmatic properties. This article is organized as follows. In section 2 I will focus on the main semantic and pragmatic features commonly associated with the get-passive in the linguistic literature. In section 3 I present my constructional approach. Basing my discussion on examples taken from the British National Corpus (BNC),1 I will analyze the interaction between the semantics of the construction and the semantics of five verb classes associated with it, concluding that two main subconstructions of the get-passive should be postulated. Section 4 contains some final remarks.

2.  Semantico-pragmatic features of the get-passive According to Siewierska (1984: 134), the use of be and get in English is “largely governed by stylistic factors, get-passives being more characteristic of colloquial speech”. Weiner & Labov (1983: 31), who contend that be- and get-passives are semantically equivalent “in general contexts”, conclude that the choice of the get- vs. the be- passive is strongly dependent on “social variation” (1983: 43).2 However, as Downing (1996: 181) rightly observes, “the semantic [and I would add “pragmatic”: PGM] features inherent to the get-passive are fundamental in conditioning appropriateness”. In (2) I reproduce these semantico-pragmatic features, as summarized by the author (1996: 183): (2) a. “actional-causative with resultative meaning” b. “tendency to occur without an overtly expressed agent” c. “adverse or beneficial consequences of the event described” d. “involvement in and responsibility for the event described” e. “attitude of the speaker towards the event”

In what follows I will briefly discuss each of these features, commonly attributed to the get-passive as opposed to the be-passive. In section 3 I will present my own (constructional) account of the get-passive in the light of the evidence offered by the corpus analysis.

.  The British National Corpus (BNC) comprises over 100 million words and represents both spoken and written English. (The spoken component of the BNC constitutes, however, only 10% of the total.) The source of the BNC examples will be identified by means of a three-letter code and the sentence number within the text where the hit was found. .  Collins (1996:  53–55) suggests that get-passives are subject to regional and diachronic variation, as well.

Semantic and pragmatic constraints on the English get-passive 



2.1  Actional-causative with resultative meaning The “actional”, “causative” and “resultative” elements are frequently regarded as fundamental in the characterization of the get-passive (see Downing 1996: 184–185 and Vanrespaille 1991: 95). By “actional” Downing (1996: 184) means that “the process in which the getpassive is involved is basically of the material type, in which something is done to or happens to the Medium”.3 The author compares (3a) and (3b) in this regard: (3) a. He got hurt. b. He was hurt.

According to Downing, the construction in (3a) is unambiguously “actionalcausative” (“X caused him to be hurt”), while (3b) is ambiguous between a “material process” (“X hurt him”) and an “attributive relational” process (“he was in a hurt condition”).4 Similarly, Ward, Birner & Huddleston (2002:  1441) observe that the example in (4a) may have a verbal reading (“Someone or something broke the window”) and an adjectival reading (“The window was in the state resulting from prior breaking”), while (4b) has only the verbal reading: (4) a. The window was broken. b. The window got broken.

But the verbal-adjectival distinction with get-passives is also subtle (see e.g. Huddleston 1984: 445 and Gronemeyer 1996: 6), particularly when the passive participle and the adjectival past participle coincide in form, as illustrated in (5) below, where the two meanings —verbal and adjectival— are fused:5

(5) The mould got broken and we never replaced it. (BNC AT7 406)

As already stated, the get-passive is commonly associated with the meaning of result.6 According to Vanrespaille (1991:  97–98), resultativeness is “one of the most central characteristics” of the get-passive: while the be-passive “simply refers to the action

.  Halliday (1994:  163) defines “Medium” as “the entity through the medium of which the process comes into existence”. .  Downing (1996: 184) admits, however, that many verbs are “unambiguously actional” with be, since, as pointed out by Hatcher (1949: 435), a static interpretation is impossible with constructions like “the man was run over, introduced, arrested, fired”, etc. .  See Levin & Rappaport (1986: 625–627) and Ward, Birner & Huddleston (2002: 1436–1440) for a revision of the diagnostic morphosyntactic environments of adjectival and verbal passive participles. .  Alexiadou (2006: 19) contends that “get is actually a light verb that receives a resultative phrase (…), i.e. the resultative participle as its complement”.

 Pilar Guerrero Medina

predicated in the verb”, the get-passive “conveys the sense of an action leading up to a significant result which is sometimes felt as irreversible”. The BNC examples in (6) are illustrative in this regard: (6) a. Timings can be quite accurate so you could have a ‘splodge’ sound as a character gets hit by a custard pie. (BNC HAC 3515) b. Shortly after appearing on an independent television programme to discuss his views (no BBC programme invited him on), Altrincham was hit in the face by a furious official of the League of Empire Loyalists. (BNC ADB 74)

Although both (6a) and (6b) are agentful passives focusing on the action, only (6a) can be regarded as “actional” and “resultative” at the same time. Note that in (6b) there is contextual reference to the action leading up to the result denoted by the verb, whereas in (6a) the event is presented “free from the context of a given situation in time” (Hatcher 1949: 440). However, as illustrated by Downing’s (1996: 185) examples in (7), “resultative” is not necessarily equated with “resulting state”:

(7) a. A lot of education got poured into you.

b. She’d merely got splashed with the lethal water and it hadn’t entered her body.

Downing observes that “resulting state” is expressed only with participles that function as predicative adjectives, such as broken in (5) above. Actually, the examples in (8) show that this “resulting state” can also be coded by a resultative adjective phrase applying to the subject of a get-passive:

(8) a. I charged with them, and got knocked silly for my pains. (Rider Haggard, “King Solomon’s Mines” (1889), in Goldberg (1995: 181))

b. Yes, the good old US fifties, where trade unionists got shot dead for striking in Detroit (…). (BNC ACP 537) c. We had the same bills after he got made redundant. (BNC EGO 1671)

2.2  Tendency to occur without an overtly expressed agent According to Hatcher (1949: 435–436), the most clear-cut restriction on the use of get as a passive auxiliary “is seen in the difficulty of adding the preposition by + name of [human] agent”. Similarly, in his 1966 study of the passive voice in English, Svartvik (1966:  149) observes that “there are not half a dozen agentive get-passives, none of which is agentful”. Downing’s (1996: 192) findings also confirm that get-passives typically occur without an overt agent. However, the tendency to occur without an overt agent is not a distinguishing feature between get- and be-passives, since the English be-passive is also predominantly agentless (see e.g. Givón & Yang 1994: 141 and Svartvik 1966: 141).



Semantic and pragmatic constraints on the English get-passive 

When the agent is expressed, reference to a human agent in get-passives seems to be dependent on the “degree to which this agent is individualized” (Hatcher 1949: 436; fn 4). the author regards (9a) as hardly acceptable, whereas (9b), where the role of the agent in the action is “subordinated”, is considered to be fully grammatical: (9) a. ?He got fired by the superintendent. b. He got run over by a drunken driver.

Note also that, as illustrated by Hatcher’s examples above, expressed agents in getpassives do not usually rank higher than the patient subjects on the animacy hierarchy. According to Carter & McCarthy (1999: 52), “the usual lack of focus on agency reinforces focus on the event and on its effect on the patient”. The authors, who present a description of the construction in spoken British English, very aptly offer this absence of explicit agents as evidence of the pragmatic function of get-passives, whose central characteristic would be the “expression of stance towards the patient and the events in which the patient is involved, and the backgrounding of agents except where they are indispensable to the coherence and information content of the message”.

2.3  Adverse or beneficial consequences of the event described According to Hatcher (1949:  441), the get-passive is used for two types of events: “those felt as having either fortunate or unfortunate consequences for the subject”. This attribution of adversity or benefit is, as Collins (1996:  50) rightly observes, partly determined by the meaning of the main verb, and partly by features of the context, but it is also clearly connected with the speaker’s attitude towards the event (see section 2.5).7 The tendency to associate get-passives with an adversative implicature is widely attested in the literature (see e.g. Givón 1993: 70; Downing 1996: 194; Collins 1996:  52 and Gronemeyer 1999: 6–7). However, Carter & McCarthy (1999: 50) report an interesting number of instances depicting neutral events. I reproduce one of their examples, taken from the Cambridge and Nottingham Corpus of Discourse in English, in (10) below: (10) S1 S2 S3 S4

Do you know how much lawyers get paid for an hour the best ones? I don’t I don’t care. Six hundred pound an hour. I don’t care.

.  As Carter & McCarthy (1999: 42; their emphasis) note, Hatcher “seemed also to be considering the role played in expressing the speaker’s/writer’s attitude by the get-construction when she referred to events ‘felt as having either fortunate or unfortunate consequences.’”

 Pilar Guerrero Medina

The circumstances in (10) are not inherently negative. Similarly, in the BNC example presented in (11), it is the surrounding circumstances that seem to be perceived as problematic by the speaker: (11) I hope I ain’t lost it. Well I did a quick trip to our Arthur’s cos I lent him some money today. Do you know, I wish he’d get paid same week we do! He gets Yeah paid monthly, and we get paid monthly. Yeah. And we get paid the week before he does. (…) (BNC KCX 1444)

As Chappell (1980:  444) writes: “Verbs occurring in get-passives cannot be divided into an ‘adversative’ set and a ‘beneficial’ set. The speaker’s intentions, aided by the context, determine which of the two interpretations is appropriate”. She also notes that the dichotomization into adversity and beneficial applies also to get-passives with inanimate subjects: (12) a. Jane’s bike got stolen yesterday. b. Jane’s bike got fixed yesterday.

In (12a) and (12b), illustrating the “adversative” and “beneficial” get-passive, respectively, the topic is not the inanimate subject, but “rather the person who owns it, or else stands in a relationship to this object equivalent to that of ownership” (Chappell 1980: 440). In the same vein, Langacker (2000: 314) states that the subject of get-passives “must in some way be implicated in the experiential relation, not necessarily as the locus of experience, but maybe just providing a link to the implicit experiencer”. However, there are cases where such a relation does not exist and it is not possible to identify an “adversity/benefit” attribution (Collins 1996: 53). The BNC examples in (13), both depicting natural phenomena, are illustrative in this regard: (13) a. Where the climate has frosty winters, containers enable the tender herbs to be grown through the summer (…). Some will survive winter outdoors with plenty of lagging round the container so that the roots do not get frozen (…). (BNC FEB 201) b. Open-sea fish, like cod, lay several million eggs each breeding season, but they are left to drift in the plankton (…). This is all part of nature’s balance, of course, as these small animals get eaten by fish. (BNC C95 1035)

2.4  Involvement in and responsibility for the event described Chappell (1980: 418–419), regards the terms “responsibility” and “involvement”, used by other writers like Hatcher (1949) and Lakoff (1971) in their semantic analyses of the get-passive, as “vague” and “confusing” (see Downing 1996: 197). In what follows, I will focus on the patient’s involvement. The notion of speaker’s involvement will be discussed in section 2.5.



Semantic and pragmatic constraints on the English get-passive 

Lakoff (1971:  155) writes that “the get passive refers to the involvement of the superficial subject, while the be passive (…) is concerned with the logical subject”. The different assignment of responsibility to the logical and superficial subjects in be-and get-passives is illustrated by the author’s (1971: 156) examples below: (14) a. Radicals must get arrested to prove their machismo. b. Radicals must be arrested if we are to keep the Commies from overrunning the U.S. (15) a. Mary was shot on purpose, the bastards! b. *Mary got shot on purpose, the bastards!

Whereas in (14a) the choice of get indicates that the modal should be taken to refer to the superficial subject (“the radicals”), in (14b) the obligation expressed in the modal rests on the logical subject (“we”, “Americans”, etc.). Likewise, only in (15a) can we ascribe responsibility to the logical subject (i.e. the non-overt agent), which explains the unacceptability of (15b). The ascription of responsibility to the patient (or “superficial subject”) is commonly associated with get-passives in the literature. Barber (1975:  22) writes: “The choice of a form like The window got broken over The window was broken seems to imply that the window somehow brought the catastrophe onto itself (…)”. In Downing’s (1996: 200) words: “when shielding the identity of the Agent (…), the speaker shifts some, though not all, of the responsibility for the action off the implicit Agent onto the Medium”. Similarly, Vanrespaille (1991: 97) speaks of “partial agentivity of the subject”. She writes: What makes the difference is that the action referred to in the be-passive originates entirely in the subject of the active sentence, whereas with the get-passive, the subject of the passive sentence can get part of the agentivity attributed to it.

According to Marín Arrese (2003: 238), “this attribution of responsibility to the subject favours the choice of referents with high inherent topicality as subjects”. It is therefore to be expected that human subjects should be more frequent in get-passives than in be-passives (see e.g. Arce-Arenales et al. 1994: 14). However, inanimate subjects also occur in get-passives, where the ascription of responsibility to the subject is not always possible. As Dixon (2005: 359) observes, (16b) is not a felicitous expansion of (16a): (16) a. That meeting got postponed. b. That meeting got itself postponed.

Givón (1993:  69) points out that the conditions under which inanimate subjects appear in get-passives suggest “a natural extension of the notion of responsibility towards other manners of human involvement”. This “involved” or “responsible”

 Pilar Guerrero Medina

entity may be the possessor (see (12a/b) above) or the hearer, as illustrated by Lakoff ’s (1971: 155) question-answer sequence in (17), where the assumption is that the speaker holds the addressee responsible: (17) “How did this window get opened?” “Sir, I cannot tell a lie: I did it.”

2.5  Speaker’s attitude towards the event As shown above, Lakoff (1971:  154) explains the difference between get- and bepassives in terms of the speaker’s involvement in the action: The get passive in English, unlike the be passive, is frequently used to reflect the attitude of the speaker toward the events described in the sentence: whether he feels they are good or bad, or reflect well or poorly on him or the superficial subject of the sentence.

The examples used by the author (1971: 154–155) to corroborate her statement are reproduced in (18): (18) a. My cache of marijuana got found by Fido, the police dog. b. A cache of marijuana was found by Fido, the police dog.

(18a) conveys “the sense of the speaker’s involvement in the action, and his unhappiness with it”, while be in (18b) suggests that the speaker “has no feelings about what took place, or certainly no feeling that it was detrimental”. Chappell (1980: 429) claims that Lakoff ’s use of the term “involvement” is misleading. On the one hand, there is “lack of clarification concerning whether or not it is the speaker or the superficial subject who is ‘involved’ in the event”; on the other hand, it is not clear whether this involvement is “active” or “emotional”. Downing (1996:  197) observes that the speaker’s involvement has to do “with attitude and evaluation of the Medium’s responsibility” and it is frequently “emotional”, either sympathetic or disapproving. However, Visser (1973:  2031) suggests a more active involvement when he speaks of “an additional connotation of the reaching of the specified state in consequence of some kind of exertion on the part of the speaker”. The examples in (19), where the speaker is clearly involved in the event, are illustrative in this regard: (19) a. Getting elected president of the study body took a lot of time and politicking. (BROWN-P27–280, in Collins (1996: 51)) b. “We used to have to promise to do things to get elected, and now we have to promise not to.” – David Blunkett, in the Red Review. (BNC A3G 431)

As the examples in (19) show, there are pragmatic factors, determined by the situational context, which make it difficult to define the get-passive exclusively in semantic terms. In Carter & McCarthy’s (1999: 51) words:



Semantic and pragmatic constraints on the English get-passive 

(…) explaining the speaker’s choice of a get-passive in preference to a be-passive involves us in questions of stance, defined as the overlay, on to the propositional core of the clause, of meanings connected with speaker attitude, judgement and affective posture.

be- and get-passives “carry different conversational implicatures” (Siewierska 1984: 134) and cannot thus be regarded as pragmatically equivalent: while the get-passive usually reflects the speaker’s (approving or dispproving) attitude towards the subject and the event, the be-passive is more neutral, as illustrated by Lakoff ’s examples in (18) above. Finally, it should be noted that not all the features in (2) are equally relevant in the characterization of the get-passive. A careful inspection of the corpus data leads me to conclude that just two of these features can be said to be contributed by the construction as a whole: i.e. the attribution of “partial agentivity” to the subject and the “resultative” feature (see section 3.3). In line with Goldberg (1995, 1998, 2002), I claim that a construction-based approach, where constructions are defined as pairings of form with meaning and contextual use, can attempt to address some of the “tantalising unresolved questions” (Chappell 1980: 417) concerning the nature of the get-passive.

3.  A constructional account of the get-passive Following Goldberg (1995, 1998, 2002), I start from the assumption that an entirely lexically-based approach is clearly unable to account for the appropriateness of the construction: the get-passive should rather be regarded as a contentful unit, contributing its own semantic and pragmatic properties. Goldberg (1995: 1) defines constructions as “form-meaning correspondences that exist independently of particular verbs”. The concept of meaning is “construed broadly enough to include contexts of use, as well as traditional notions of semantics” (Goldberg 1995: 229; fn 6). In fact, the notion of construction is later defined by the author as “a form-function pair” (Goldberg 1998: 205):8 C is a construction iff defn C is a form-function pair, such that some aspect of the form or some aspect of the function is not strictly predictable from C’s component parts.

In section 3.1 I will present the semantic types on which I base my analysis, discussing how the semantics of the construction interacts with the semantic properties of the verbs in each of these types.

.  See also Goldberg’s (2002:  813) definition of a construction as a “a pairing of form with meaning/use such that some aspect of the meaning/use is not strictly predictable from the component parts or from other constructions already established as existing in the language”.

 Pilar Guerrero Medina

3.1  The data: Semantic types This study will be based on get-passives containing verbs from five different semantic types, i.e. affect, giving, motion (take-subtype), corporeal and annoying. Dixon (2005: 96–97) classifies these semantic types, which directly refer “to some activity or state”, as primary, as opposed to secondary types, which provide “semantic modification of some other verb”.9 In Dixon’s (2005: 112) taxonomy, affect verbs are further subdivided into eight subtypes, and I will analyze verbs from six of these subtypes: the touch subtype (i.e. touch and stroke), the hit subtype (i.e. hit and shoot), the stab subtype (i.e. stab and cut), the stretch subtype (i.e. burn and freeze), the build subtype (i.e. build and make) and the break subtype (i.e. break and smash). Most of these verbs are prototypically transitive verbs with Effector subjects (typically agentive)10 and highly individuated (usually animate) affected objects. Verbs from four of these subtypes, i.e. stab, stretch, build and break, are “change-of-state” verbs, with affected or effected Patient objects.11 Verbs of giving are less prototypically transitive verbs, involving three semantic roles: an Effector, a Theme and a Recipient. I will analyze examples containing two of these verbs: give and pay. In transitive structures with give and pay both Th(eme) and Rec(ipient) can become direct objects. According to Dixon (2005: 301), the basic syntactic frame for giving verbs is with the Theme as Object and the Recipient introduced by the preposition to, as in “I’ve lent all my phonetics books (Th/O) to different people (Rec)”. However, when the Recipient participant is the most salient non-agentive role, it can then be assigned the Object function, dropping its preposition and being placed immediately after the verb, as in “I’ve lent my favourite student (Rec/O) a bunch of different sorts of books”. Two verbs of motion from the take subtype have been chosen: i.e. send and steal, both transitive verbs with a human subject (typically agentive) and a Theme in Object

.  Primary verbs are further subdivided into primary-a verbs, which “take concrete nouns as heads of their subject and object NPs when used in a literal sense” and primary-b verbs, which also allow complement clauses in subject and object positions. The affect, giving, motion and corporeal verbs are then primary-a types, while the annoying type is a primary-b verb. .  Van Valin & LaPolla (1997: 85) define “Effector” as “the doer of an action, which may or may not be wilful or purposeful”. Compare with “Agent”, defined as the “wilful, purposeful instigator of an action or event”. “Theme” labels the entity undergoing a change of location (Van Valin & LaPolla ibid), avoiding more specific terms such as Dixon’s (2005: 106, 119) “Gift” or “Moving” roles. .  Following Dowty (1991: 574), I assume that “change of state” includes “coming into existence, going out of existence and both definite and indefinite change of state”.

Semantic and pragmatic constraints on the English get-passive 



function. A Source or Goal participant (indicating destination) is often included. Send is an interesting verb, as it “straddles the giving and motion types” (Dixon 2005: 95). When used as a giving verb, the Recipient participant can be coded as the Object as in “Mary sent the doctor (Rec/O) a present”. However, when used as a motion verb, it is possible to say “Mary sent John to the doctor (Goal)”, but not “*Mary sent the doctor John”, since the Goal is not suitable for Object assignment. Steal is also a specific verb that focalizes the Theme, and not the victim. In Dixon’s (2005: 106) example “He stole ten dollars (Th/O) (from Mary’s purse (Source))”, the focus is on the thing stolen (“ten dollars”) and not on the Source. (The BNC examples in (24) below are also illustrative in this regard.) corporeal verbs like eat and dine, categorized as “verbs of ingesting” in Levin’s (1993: 213) taxonomy of verb classes, involve a human participant in Subject position and an omissible second argument referring to the ingested substance (usually food or drink) or to a particular meal. Finally, verbs from the annoying type like annoy and please co-occur with Experiencer objects and Stimulus subjects. According to Dixon (2001:  360), annoying verbs which are rarely used with get include delight, please, satisfy or amuse, verbs referring to “feelings that tend to be experienced naturally, with the Experiencer having little or no role in bringing them about”. Table 1 presents the frequencies of occurrence of these semantic types in getpassives in the BNC corpus.12 Table 1.  Frequencies of occurrence of affect, giving, motion, corporeal and annoying verbs in get-passives affect touch hit Ø

stab

stretch build

11.75% 2.47% 2.32%

3.91%

break giving motion corporeal annoying 4.93%

58.92% 8.56%

2.32%

4.78%

Only 175 of the 689 occurrences (25.39%) correspond to affect verbs. Four of the analyzed subtypes (i.e. stab, stretch, build and break) can be defined as “changeof-state” verbs. touch verbs are “pure verbs of contact” (Levin 1993: 10),13 while verbs in the hit subtype are “verbs of contact by impact” (Levin 1993: 148), which do not

.  Percentages are based on 689 occurrences of get/gets/got followed by the past participle of the above-mentioned verbs. The total of instances corresponding to each verb is as follows: affect verbs: touch and stroke (ø), hit (50), shoot (31), stab (5), cut (12), burn (14), freeze (2), build (9), make (18), break (27) smash (7); giving verbs: give (14), pay (392); motion verbs: send (49), steal (10); corporeal verbs: eat (16), dine (ø); annoying verbs: annoy (33), please (ø). .  No instances were found of the sequence NP + get/gets/got + “touched”/“stroked”.

 Pilar Guerrero Medina

necessarily involve the change of state of some entity.14 stab verbs specify both the change of state and how the change of state was brought about. The stretch and break subtypes are also interesting in that they include verbs with dual transitivity on the S = O pattern (Dixon 2005: 309). Whereas break verbs can be regarded as “basically transitive”, verbs in the stretch subtype are considered as “basically intransitive” by most native speakers. According to Dixon (2005: 310), the principle in operation is the following: If the S/O role often gets into the state described by the activity on its own, without outside assistance, then the verb is thought to be basically intransitive. But if one would normally expect there to be a Causer (even if one might not know who or what it is) then the verb is thought to be basically transitive.

As stated by Levin & Rappaport (1993:  105), verbs can be classified “according to whether or not they describe an externally caused eventuality or according to whether or not they describe an eventuality that can occur spontaneously”. We could therefore distinguish between verbs describing “internally caused” eventualities (e.g. freeze and burn) and verbs like break and smash, which describe “externally caused” eventualities.15 Get-passives with verbs of the build subtype were not frequent (3.91%, 27 occurrences). Both Lakoff (1971: 154) and Chappell (1980: 443) claim the incompatibility of the get-passive with verbs of creation. In Chappell’s (1980: 421) words, “the subject has to be thought of as ‘pre-existing’ before it can be considered as causally involved in the event”. However, the BNC examples in (20) clearly contradict this claim:16 (20) a. One example of the monumental style which did get built was the Wesleyans’ new London home, the central Hall in Westminster. (BNC AE6 810) b. Corman was following me around with the camera directing my speaking parts. It was a fiasco. But the film got made. (BNC APO 556)

.  As Levin (1993:  6) observes, there are actually two senses of hit: “contact through the motion of an instrument” (as in “Carla hit the door”), where the object is an immovable entity, and “contact using an instrument and set in motion” (as in “The batter hit the ball over the fence”). All the instances analyzed in section 3.1 correspond to the first use of the verb: e.g. “But how on earth can you get hit by a train?” (BNC A7A 1022). .  The verb burn, one of Levin’s (1993:  246) “verbs of entity-specific change of state”, can be predicated of two kinds of entities: entities that can be consumed by fire (such as paper or wood), and entities which have been designed to “burn” (such as candles). It is only this second sense that involves external causation (see Levin & Rappaport 1993: 235). .  Likewise, Downing (1996:  187) shows that verbs of creation are compatible with a getpassive, as in “The ordinary folk are going to see how foreign policy gets made”.



Semantic and pragmatic constraints on the English get-passive 

corporeal verbs constitute the least frequent semantic type (2.32%, 16 instances). Only three of the analysed examples were agentful constructions, as in (13b) above, here reproduced as (21). No occurrences where found of get/gets/got + “dined”. (21) This is all part of nature’s balance, of course, as these small animals get eaten by fish. (BNC C95 1035)

A verb like eat may have two interpretations: an activity interpretation (e.g. “Mario eats pizza”), involving just one participant, usually agentive, and a nonreferential “inherent object” (Van Valin & LaPolla 1997: 123), and an active accomplishment interpretation (e.g. “Mario ate a slice of pizza”), where two participants are involved. In the analysed examples of get-passives the verb had the second interpretation, as in (21) and also (22) below: (22) (…) and if we do not cater for these fish they may find all the food gets eaten before any sinks down to them. (BNC C96 311)

The largest proportion corresponded to verbs of giving (58.92%, 406 instances). Most of the analysed examples (83.33%) were agentless constructions, where the Recipient was lexicalized as the subject: (23) a. You have to have a licence to have a C.B. and when you get your licence you also get given a rule book which tells you all the things you are and are not allowed to do (…) (BNC HAD 448) b. “Coal miners” wives get paid danger money,” he explains. “I get paid grief money.” (BNC C9K 2275)

Verbs of motion were not so frequent (8.56%, 59 instances). Most of the analysed examples were also agentless: (24) a. A lot of that was also due to the fact that we just didn’t get into it, because all of the stuff got stolen after we’d started (…) (BNC HNW 785) b. How many times have you read an interview with a musician where he or she says, “Yeah, that was my favourite guitar, but it got stolen, and I never found another like it”? (BNC C9K 1175)

Finally, the number of annoying verbs was also low (4.78%, 33 instances). No occurrences were found of “got/get pleased”, and just one example including the sequence “gets pleased”, reproduced in (25a) below: (25) a. But there’s a twist. After she gets pleased, she kicks him out. (BNC ED7 23) b. Mother always got annoyed when that story came up in conversation, and I made sure that it frequently did. (BNC AS7 1504) c. I was talking to that presenter of that alternative politics programme (…) and he said that he’d got annoyed with a colleague because she criticized his pink bow-tie. (BNC CA3 1009)

 Pilar Guerrero Medina

Note that a certain degree of initiative may be ascribed to the subject referent in these examples: in (25a) the Experiencer becomes the volitional agent in the main clause: “[when she (Exp) gets (sexually) pleased], she (Ag) kicks him out”. Likewise in (25b) and (25c) the notion of partial agentivity of the subject can be said to apply if we assume that the Experiencer subjects (“Mother” and “that presenter”) made themselves reach a state of being annoyed (see Dixon 2005: 360). The case of annoy is also interesting in that some measure of control may be assigned to the Stimulus: in (25b), where the agent is recoverable from the context, “the person bringing up the story” clearly had the intention of annoying Mother; in (25c) it is the “quasi-agent” (Svartvik 1966: 102), “a colleague”, which seems to be in control.

3.2  The interaction between verb and construction It is obvious that not all verb classes can be integrated within the get-passive. As stated by Goldberg (1995: 24), “the meanings of constructions and verbs interact in nontrivial ways”, and therefore there is bound to be “some cross-reference between verbs and argument structures”. The author (1995: 50) postulates the following two principles to determine which roles of the verb are “fused” with which roles of the construction:17 1.  The Semantic Coherence Principle: “Only roles which are semantically compatible can be fused.”18 2.  The Correspondence Principle: “Each participant role that is lexically profiled and expressed must be fused with a profiled argument role of the construction.” There are specific conditions under which a profiled participant role may fail to be overtly expressed. In be- and get-passives, frequently agentless, the “highest ranked participant” is usually “shaded”, i.e. no longer profiled (Goldberg 1995: 57). The getpassive, however, often serves to “cut” the agent participant, as in the examples in (1) above, here repeated as (26), where the agent cannot be expressed: (26) a. *Kelly got lost by him. b. *He got fogged in by the weather.

Goldberg (1995: 116–117) adapts Dowty’s (1991: 572) observations on argument selection to her constructional framework. Dowty postulates two thematic macro-roles

.  Thematic roles are not defined as theoretical primitives but as “semantically constrained relational slots in the dynamic scene associated with the construction or the verb” (Goldberg 1995: 49) .  Two roles r1 and r2 are semantically compatible if either r1 can be construed as an instance of r2, or r2 can be construed as an instance of r1.



Semantic and pragmatic constraints on the English get-passive 

(Proto-Agent and Proto-Patient), each characterized by a set of verbal entailments, as detailed below:19 (27)

Contributing properties for the Agent Proto-Role: a. “volitional involvement in the event or state” b. “sentience (and/or perception)” c. “causing an event or change of state in another participant” d. “movement (relative to the position of another participant)” e. “exists independently of the event named by the verb”

(28) Contributing properties for the Patient Proto-Role:20 a. “undergoes change of state” b. “incremental theme” c. “causally affected by another participant” d. “stationary relative to movement of another participant” e. “does not exist independently of the event, or not at all”

The following Argument Selection Principle is then proposed to determine the way these proto-roles are associated with grammatical relations (Dowty 1991: 576): In predicates with grammatical subject and object, the argument for which the predicate entails the greatest number of Proto-Agent properties will be lexicalized as the subject; the argument for which the predicate entails the greatest number of Proto-Patient properties will be lexicalized as the direct object.

The argument selection principle only applies when the transitive construction is involved, as illustrated in Figure 1:21 The Proto-Agent and Proto-Patient roles are prototypical concepts, and “arguments can differ in the ‘degree’ to which they bear each role” (Dowty 1991: 613). In Figures 2 and 3 I have made an attempt to distribute the first and second arguments of the different semantic types presented in section 3.1 along their corresponding semantic continua, depending on how many of these verbal P-Agent and P-Patient properties each

.  Compare with Van Valin & LaPolla’s (1997: 143) Actor and Undergoer, defined as “generalized semantic roles whose prototypes are the thematic relations Agent and Patient, respectively”. .  The Incremental Theme is the role of an NP argument that undergoes a “definite” change of state “in distinguishable separate stages” (Dowty 1991: 568). Examples of Incremental Themes are “effected” and “destroyed” objects (e.g. build a house, eat a sandwich). .  Dowty (1991: 582) distinguishes between “event-dependent role notions which are associated with lexical verbs” and “discourse-dependent semantic associations of subjects”. This distinction accounts for the fact that passive subjects are not Agents.

 Pilar Guerrero Medina

Sem

Proto-Agent

Proto-Patient

Syn

↓ SUBJ

↓ OBJ

Figure 1.  Transitive construction (Goldberg 1995: 117).

semantic type entails.22 I have focused on the following properties for each proto-role: “volition”, “sentience” and “causation” (as P-Agent entailments) and “change of state” and “causally affected” (as P-Patient entailments).  

/ /

-P-Agent

 /  +P-Agent

Figure 2.  Semantic continuum in terms of Proto-A entailments.

/ 





-P-Patient

 / / +P-Patient

Figure 3.  Sematic continuum in terms of Proto-P entailments.

It is interesting to observe that the affectedness of the object does not always determine its liability to occur in get-passives. The verbs which most frequently occur in the construction (i.e. hit, pay and send), corresponding to three different semantic types, are inherently atelic verbs with objects which do not necessarily undergo a change of state.

3.3  The semantics of the construction Along the lines of Goldberg & Jackendoff (2004:  563), it is my contention that the get-passive should be treated as a family of constructions, “united by related but not identical syntax and by related but not identical semantics”. Two main subconstructions .  Each of the P-Agent and P-Patient properties is presented as semantically independent, although a verb like build, for example, has all the entailments in (27) for subject and all the entailments in (28) for object.



Semantic and pragmatic constraints on the English get-passive 

should be posited, i.e. the “causative” get-passive and the “spontaneous” get-passive, as represented in (29) and (30).23 (29) “Causative” get-passive a. Syntax: Subj get V-en (OblAg) Semantics: cause-become 〈Eff/Ag Pat/Exp〉 instance [verbal subevent] b. Syntax: Semantics:

Subj get V-en (OblAg) cause-receive 〈Ag Rec Th 〉 instance/means*[verbal subevent] *“transfer of possession to a recipient”

c. Syntax: Semantics:

Subj get V-en (OblSource) (OblGoal)(OblAg) cause-have/not-have 〈Ag Th Source/Goal〉 instance/means*[verbal subevent] *“transfer of possession from a source/to a direction”

(30) “Spontaneous” get-passive Syntax: Subj get V-en Semantics: become 〈Pat〉 instance [verbal subevent]

The “causative” get-passive in (29a) is available to affect verbs (of the hit, stab, stretch, build and break subtypes), corporeal and annoying verbs. As shown above, it is associated with the semantics “y is caused to become z by x”, represented as cause-become 〈Eff/Ag Pat/Exp〉, where the responsible Patient (or Experiencer) is the profiled argument role. The “causative” get-passive is frequently agentless, although the “shaded” Agent may be expressed as an oblique argument (OblAg), as in (31c):24 (31) a. They don’t want to get stabbed or sent away for six months. (BNC B17 560). b. The gunshot is disputed by locals. Police are adamant they heard it. (…) No one got shot. (BNC H7F) c. Jack er put my red tractor away in the dump and it got smashed, smashed up by a digger. (BNC KB3 1442)

.  Profiled argument roles (i.e. arguments expressed as direct grammatical relations) are printed in bold type (see Goldberg 1995: 49). .  Only 17 (9.71%) out of the 175 get-passives with affect verbs were agentful constructions. 13 of these 17 examples had inanimate agents, and just one of the human agents was singular and definite: “Say you got me to Ireland and dropped me off, then got shot down by a British night-fighter off the French coast on your way back.” (BNC HTW 3613)

 Pilar Guerrero Medina

I have postulated two additional subtypes of the “causative” get-passive. In the subtype in (29b), associated with the semantics “y is caused to receive z from x”, the Recipient and Theme are the profiled participants and the Agent may also be expressed as an oblique argument. This type is available to verbs of giving, as illustrated by the examples in (23) above, here reproduced as (32), where the Theme is lexically profiled and expressed: (32) a. You have to have a licence to have a C.B. and when you get your licence you also get given a rule book which tells you all the things you are and are not allowed to do (…) (BNC HAD 448) b. “Coal miners’ wives get paid danger money,” he explains. “I get paid grief money.” (BNC C9K 2275)

As shown in section 3.1, the Recipient was lexicalized as the subject in most of the analyzed get-passives with pay and give. However, in be-passives with these verbs, the Theme is frequently expressed as the subject. Compare the examples in (32) and (33) in this regard: (33) a. They provided more circumstantial evidence that (…) big bribes were paid to companies controlled by Indian agents. (BNC A5M 100) b. This measure was repeated in 1297 and fines were paid during the next four years. (BNC F9L 375) c. During the early part of that day, orders were given to the army to arrest strike organizers over much of Ulster. (BNC CCC 1126) d. Prophylactic intravenous antibiotics (…) were given for 24 hours.

In the third subtype, allowing verbs of motion from the take subtype, whose semantics is represented in (29c) as “y is caused to have/not to have z”, the Theme is the only profiled argument role. The Agent, Source and/or Goal participants can be expressed as oblique arguments: (34) a. We could keep you. “It makes me sound like a parcel that got sent to the wrong address.” (BNC JY0 1348) b. Men get sent back from specialist posts to uniform duties as a punishment (…). (BNC A0K 570)

It is interesting to observe that in be-passives with send the human Recipient is frequently expressed as the subject: (35) a. Older women are sent daily signals that they must struggle to preserve their attractiveness. (BNC CFE 222) b. You will then be sent your Policy Document to study at home without obligation. (BNC ABC 1101)



Semantic and pragmatic constraints on the English get-passive 

In the retrieved examples with steal neither the Agent nor the Source were expressed, as illustrated in (24) above. Examples where the Source is lexically profiled and expressed, are, however, frequent on Google: (36) Responsible gun owners don’t allow their guns to be stolen. Therefore, if they were irresponsible enough to have the gun stolen, they don’t deserve to get it back. You get it? Guns don’t get stolen from responsible owners. They get stolen from irresponsible owners. (〈www.city-data.com/forum/politicsother-controversies/37403–2nd-amendment-8html〉)

The second main subconstruction, i.e. the “spontaneous get-passive” in (30), associated with the semantics “y becomes z”, is represented as become , where the (nonvolitional) patient is profiled and the Agent (or Effector) participant is often “cut”, as in (37): (37) a. And Carrie Fisher as Leia (…) Oh And the bit with Harrison Ford when he gets frozen (…). (BNC KEO 4964) b. The tops of their thighs, that little bit just below the cheeks that always get burnt first, would be as red as her Tequila Sunrise. (BNC EFG 1156)

This subconstruction, where agent extension is unlikely or impossible, is available for the description of spontaneous events with alternating “change-of-state” verbs of the stretch and break subtypes, describing “internally caused eventualities” (as in (37) above) or “externally caused eventualities” (see (5) in section 2.1). On Goldberg’s constructional approach, constructions must be able to specify which verbs can be integrated with them and “the way in which the event type designated by the verb is integrated into the event type designated by the construction” (Goldberg 1995: 49). Notice that in (29a) and (30) the event type designated by the verb is an instance of the event type designated by the construction. However, in (29b) and (29c) the relation between the verbal subevent and the constructional subevent include both instance and means, i.e. “transfer of possession to a recipient” (with verbs of giving) or “transfer of possession from a source/to a direction” (with verbs of motion). These two subconstructions subsume what Lakoff (1971: 157) regards as the two main senses of the get-passive. According to the author (ibid), these two uses “mirror” the two main uses of transitive get: “one in which the subject is the agent and also the initiator of the action; the second in which the subject is the patient, to whom the action happens”, illustrated in (38):25

.  In the same vein, Bonnefille (2006: 32-33) states that the semantics of the get-passive (“Y is caused to be in state z by X”) is constrained by the blending of “X acts on Y” and “X experiences Y”.

 Pilar Guerrero Medina

(38) a. John got a present for his mother. b. John got a present from his mother.

While (38a) can be said to mirror the use of the “causative” get-passive, where some measure of initiative is attributed to the subject, (38b) reflects the use of the so-called “spontaneous” get-passive, closer to the inchoative construction. The semantic relation between the get-passive and the ergative (and middle) construction has been widely explored in the linguistic literature (see e.g. Hundt 2001: 53).26 According to Downing (1996: 194), “the get-passive makes available a typically one-participant causative construction for use with verbs that do not allow the ergative”. In much the same vein, Marín Arrese (2003: 240) states that “the get-passive makes available an inchoative reading in the case of events involving agent-oriented components”, as illustrated in (39): (39) a. The clothes got washed. (*The clothes washed) b. The book got read. (*The book read)

With “change-of-state” verbs of the stretch and break subtypes, the get-passive may alternate with the ergative construction to codify some eventualities that are presented as occurring spontaneously (see Marín Arrese 2003: 240). Compare (5) above, here reproduced as (40a), and (40b), both depicting events that are presented as selforiginated: (40) a. The mould got broken and we never replaced it. (BNC AT7 406) b. The accelerator cable broke in the middle of nowhere so we had to tie some string to it (…). (BNC CRP 267)

It is, therefore, the second subconstruction, the “spontaneous” get-passive in (30) above, where agent extension is unlikely or impossible, which shows stronger semantic connections with the ergative construction, where the focus is also on the change of state undergone by the Patient subject. Get-passives of the “spontaneous” type can be thus regarded as “inchoative” constructions, where the event presented seems, in some way, “the result of chance” (Hatcher 1949: 440).

.  If we use the term “middle” to refer to a “semantically unitary category” associated with different typological marking patterns where the Initiator is the same as the Endpoint entity (Kemmer 1994: 209- 210), get can actually be regarded as a “middle marker”. See e.g. Arce-Arenales et al. (1994: 12), who hold that get is a marker of middle diathesis and conclude that get-passives “are not derived from active constructions but rather are themselves active constructions”.



Semantic and pragmatic constraints on the English get-passive 

As already stated, the features presented in (2) in section (2) above cannot be presented as defining characteristics of the get-passive as a whole. In (41) and (42) I summarize the properties of each subconstruction: (41) “Causative” get-passive a. “Partial agentivity” of the subject (sometimes shared by Patient/Experiencer subject and Agent/Stimulus) b. “causative” with resultative meaning c. Speaker’s involvement in the event (approving or disapproving attitude) d. “Shading” of the agent (42) “Spontaneous” get-passive

a. resultative meaning b. “Partial agentivity” of the subject c. Speaker’s interest in the referent of the subject d. “Cutting” of the agent

In the “spontaneous” get-passive, the feature of “partial agentivity of the subject” is clearly less salient than the “resultativeness” characteristic. Both features are, however, equally relevant in the characterization of the “causative” get-passive, with the exception of get-passives with corporeal verbs, where the attribution of responsibility to the subject is not always possible. It is precisely the emphasis on the result of the action which accounts for the incompatibility of many durative and stative verbs with get-passives (“*he got watched”, “*he got liked”) (see Vanrespaille 1991: 97 and Hatcher 1949: 434; fn 2). However, as we have seen in section 3, some durative verbs (e.g. annoy) are fully compatible with the get-passive. The “adversity/benefit” feature, usually attributed to get-passives in the literature, has not been regarded as central in the definition of the construction. As Downing (1996: 201) observes, it is not easy to draw a clear-cut distinction between adverse or beneficial consequences. As reflected in (41) and (42), I have rather focused on the notion of speaker’s involvement, closely connected with the attribution of responsibility to the subject. In the “causative” get-passive the speaker’s attention is directed towards the involvement of the patient subject (frequently animate) in the event, indicating a sympathetic or disapproving attitude towards the consequences that this event may have on the subject-referent. (As shown in section 2.5, there are cases that suggest a more active involvement of the speaker.) However, in the “spontaneous” getpassive, where the resultative meaning acquires prominence, the notion of speaker’s emotional involvement is less appropriate, and we should rather speak of “speaker’s interest” in the subject and “what happens to it as a result of the event described by the verb” (see Downing 1996: 204).

 Pilar Guerrero Medina

Finally, there is an important profiling difference between two subconstructions: while the “causative” get-passive “shades” or “deprofiles” the agent participant, which can be expressed by an adjunct, the agent is “cut” in the “spontaneous” get-passive, where the Patient is the only profiled argument role.

4.  Final remarks In section 3 I have treated the get-passive as forming a family of constructions in order to account for its semantic and pragmatic properties. Two main subconstructions, i.e. the “causative” and the “spontaneous” get-passive, have been posited. Their properties have been summarized in section 3.3. The constructional approach presented in this article can easily accommodate the conception of the get-passive as a prototypically structured category (see Shibatani 1985 & Collins 1996),27 with meanings distributed along a “passive scale” (Svartvik 1966: 138). The “causative” get-passive mainly corresponds to Collins’s subclass of “central” get-passives (with an explicit or accessible agent-phrase and a dynamic verb) also including instances of what he labels “psychological” get-passives, “whose members exhibit a mixture of verbal an adjectival properties” (Collins 1999: 46). The “spontaneous” get-passive, where get can be regarded as a “change-of-state marker” (Gronemeyer 1999: 6), is semantically and syntactically closer to the class of “adjectival” get-passives, where there is just a tenuous relationship with an active construction. In Goldberg’s (1995: 67) framework “constructions form a network and are linked by inheritance relations which motivate many of the properties of particular constructions”. The two subconstructions in (29) and (30) above can thus be said to “inherit” the linking specifications of the transitive construction in Figure 1, reversing the syntactic association (the Proto-Patient is now lexicalized as subject) and deprofiling the Agent, which may be “shaded” (in the three “causative” subtypes) or “cut” (in the “spontaneous” get-passive).28

.  Collins (1996: 44 ff) holds that get-passives form a “fuzzy set”: “central” get-passives are “core” or “genuine” get-passives which may alternate with an active clause, while “adjectival” getpassives (e.g. “get drunk”, “get tired”) are close to the periphery of the class. Collins’s subclasses of “psychological” get-passives (e.g. “get frustrated”, “get bored”, “get worried”) and “reciprocal/ reflexive” get-passives (e.g. “get dressed”, “get married”) are intermediate between “central” and “adjectival” get-passives. .  The subtypes in (29b) and (29c) can be said to be motivated by the ditransitive construction (“X causes Y to receive Z”) and by the so-called “transfer-caused-motion” construction, encoding the transfer of possession to or from a possessor (see Goldberg 1995: 90).



Semantic and pragmatic constraints on the English get-passive 

I have also tried to show that the get-passive is not semantically or pragmatically equivalent to the be-passive: on the one hand, the get-passive is associated with its own semantic and pragmatic constraints, contributing additional meanings of responsibility and (active or emotional) involvement; on the other hand, as stated in section 3.3, the choice of the get-passive entails some important profiling differences, particularly evident in the case of giving and motion verbs. The “causative” and “spontaneous” get-passive thus impose their own “profiled status” (Goldberg 1995: 53) on the semantic roles lexically profiled by the verbs associated with them.

References Alexiadou, Artemis. 2006. A note on non-canonical passives: the case of the get-passive. In Organizing grammar. Linguistic studies in honor of Henk van Riemsdijk, Hans Broekhuis, Norbert Corver, Riny Huybregts, Ursula Kleinhenz & Jan Koster (Eds), 13–21. Berlin: Mouton. Arce-Arenales, Manuel, Melissa Axelrod & Barbara A. Fox. 1994. Active voice and middle diathesis. A cross-linguistic perspective. In Voice. Form and function, Barbara Fox & Paul J. Hopper (Eds), 1–21. Amsterdam: John Benjamins. Barber, E.J.W. 1975. Voice – beyond the passive. In Proceedings of the first annual meeting of the Berkeley Linguistics Society, Cathy Cogen, Henry Thompson, Graham Thurgood, Kenneth Whistler & James Wright (Eds), 16–24. Berkeley CA: Berkeley Linguistics Society. Bonnefille, Stephanie. 2006. Constructions with get. How to get the picture without getting confused. Annual Review of Cognitive Linguistics 4(1): 21–37. Carter, Ronald & Michael McCarthy. 1999. The English get-passive in spoken discourse: descriptions and implications for an interpersonal grammar. English Language and Linguistics 3(1): 41–58. Chappell, Hilary. 1980. Is the get-passive adversative? Papers in Linguistics 13: 411–452. Collins, Peter C. 1996. Get-passives in English. World Englishes 15(1): 43–56. Dixon, R.M.W. 2005/1991. A semantic approach to English grammar, 2nd Edn. Oxford: OUP. Downing, Angela. 1996. The semantics of get-passives. In Functional descriptions. Theory in practice, Ruqaya Hasan, Carmel Cloran & David G. Butt (Eds), 179–205. Amsterdam: John Benjamins. Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67(3): 549–619. Fox, Barbara & Paul J. Hopper (Eds), 1994. Voice. Form and function [Typological Studies in Language, 27]. Amsterdam: John Benjamins. Givón, Talmy 1993. English Grammar. A function-based introduction. Vol II. Amsterdam: John Benjamins. Givón, Talmy & Lynne Yang. 1994. The rise of the English get-passive. In Voice. Form and function, Barbara Fox & Paul J. Hopper (Eds), 119–149. Amsterdam: John Benjamins. Goldberg, Adele E. 1995. Constructions. A Construction Grammar approach to argument structure. Chicago IL: The University of Chicago Press. Goldberg, Adele E. 1998. Patterns of experience in patterns of language. In The new psychology of language structure, Vol.1, Michael Tomasello (Ed.), 203–219. Mahwah NJ: Lawrence Erlbaum. Goldberg, Adele E. 2002. Construction Grammar. In Encyclopedia of Cognitive Science, Lynn Nadel (Ed.), 813–816. London: MacMillan.

 Pilar Guerrero Medina Goldberg, Adele E. & Ray Jackendoff. 2004. The English resultative as a family of constructions. Language 80(3): 532–567. Gronemeyer, Claire. 1999 On deriving complex polysemy: the grammaticalization of get. English Language and Linguistics 3(1): 1–39. Halliday, M.A.K. 1985[1994]. An introduction to Functional Grammar, 2nd Edn. London: Edward Arnold. Hatcher, Anna G. 1949. To get/be invited. Modern Language Notes 64: 433–446. Huddleston, Rodney. 1984. Introduction to the grammar of English. Cambridge: CUP. Hundt, Marianne. 2001. What corpora tell us about the grammaticalisation of voice in getconstructions. Studies in Language 25(1): 49–87. Kemmer, Suzanne. 1994. Middle voice, transitivity, and the elaboration of events. In Voice. Form and function, Barbara Fox & Paul J. Hopper (Eds), 179–230. Amsterdam: John Benjamins. Lakoff, Robin. 1971. Passive Resistance. In Papers from the seventh regional meeting of the Chicago Linguistic Society, Douglas Adams, Mary Ann Campbell, Victor Cohen, Julie Lovins, Edward Maxwell, Carolyn Nygren & John Reighard (Eds), 149–162. Chicago IL: Chicago Linguistic Society. Langacker, Ronald W. 2000. Grammar and conceptualization. Berlin: Mouton de Gruyter. Levin, Beth. 1993. English verb classes and alternations. Chicago IL: Chicago University Press. Levin, Beth & Malka Rappaport. 1986. The formation of adjectival passives. Linguistic Inquiry 17: 623–661. Levin, Beth & Malka Rappaport. 1993. Unaccusativity. At the syntax-lexical-semantics interface. Cambridge MA: The MIT Press. Marín Arrese, Juana I. 2003. The middle domain in English and Spanish. Middle and related situation types. In Cognitive linguistics in Spain at the turn of the century, Vol. I: Grammar and Semantics, Clara Molina, Marisa Blanco, Juana I. Marín Arrese, Ana. L. Rodríguez & Manuela Romano (Eds), 229–252. Madrid: AELCO and Universidad Autónoma de Madrid. Shibatani, Masayoshi. 1985. Passives and related constructions: a prototype analysis. Language 61(4): 821–848. Siewierska, Anna.1984. The passive. A comparative linguistic analysis. London: Croom Helm. Svartvik, Jan. 1966. On voice in the English verb. The Hague: Mouton. Van Valin, Robert D. & Randy J. LaPolla. 1997. Syntax. Structure, meaning and function. Cambridge: CUP. Vanrespaille, Mia. 1991. A semantic analysis of the English get-passive. INTERFACE. Journal of Applied Linguistics 5(2): 95–112. Visser, F.T. 1973. An historical syntax of the English language, Part III. Syntactical units with two and more verbs. Leiden: E.J. Brill. Ward, Geoffrey, Betty Birner & Rodney Huddleston. 2002. Information packaging. In The Cambridge grammar of the English language, Rodney Huddleston & Geoffrey K. Pullum (Eds), 1363–1447. Cambridge: CUP. Weiner, E.  Judith & William Labov. 1983. Constraints on the agentless passive. Journal of Linguistics 19: 29–58.

Name index

A Aarts, Bas  202 Abelson, Robert P.  183 Ackerman, Farrell  225 Alexiadou, Artemis  266, 273 Allison, David  xvii Allwood, Jens  20 Alonso Ramos, Margarita  163, 252 Andersen, Hans C.  25, 27, 48, 49 Arce-Arenales, Manuel  271, 277, 290 Ariel, Mira x, 202, 211, 212, 232 Aske, Jon  9, 12, 27, 30 Axelrod, Melissa  271 B Bach, Kent  181 Baicchi, Annalisa  143, 183 Baker, Collin F.  161 Baker, Mark  89, 90 Bakker, Dik  90 Balteiro Fernández, Isabel  9 Barber, E.J.W.  277 Barcelona, Antonio  217, 232, 234 Barlow, Michael  205, 211, 212 Bauer, Laurie  4, 5, 8, 15, 92 Beard, Robert  93 Beaugrande, Robert de  175 Bencini, Giulia M.L.  145 Bender, Emily  179 Bergen, Benjamin K.  154, 256 Berman, Ruth A.  27, 30 Bernini, Giuliano  58 Biber, Douglas  211 Birner, Betty  273 Bloomfield, Leonard  85, 86 Boas, Hans  xvi, 43, 44, 159, 161, 162, 179, 201, 204, 205, 220, 234, 240 Bohnemeyer, Jürgen  28, 40, 41 Bolinger, Dwight  223 Bonnefille, Stephanie  289

Booij, Geert  95, 96 Borkin, Ann  228 Burnard, Lou  204 Butler, Christopher S.  xv, xvi, xviii, 85, 87, 97, 118, 121, 124, 130, 132, 143, 144, 153–155, 176, 201, 205, 220, 247 Bybee, Joan  38, 93, 204, 215, 236 C Carlson, Greg N.  223 Carston, Robyn  182 Carter, Ronald  271, 275, 278 Casenhiser, Devin M.  159 Chang, Nancy  154 Chappell, Hilary  271, 276, 278, 279, 282 Chierchia, Gennaro  266 Chomsky, Noam  6, 97, 110 Clark, Eve V. 4, 6, 7, 14, 21, 145 Clark, Herbert H.  4, 6, 7, 9, 14, 18, 19–21 Collins, Peter C.  271, 272, 275, 276, 278, 292 Comesaña, Susana  228 Comrie, Bernard  29 Cortés Rodríguez, Francisco  xviii, 63, 64, 89, 90, 92, 130, 133, 139, 164–166, 189 Coseriu, Eugenio  xviii, 117, 119–121 Croft, William  26, 33, 38, 69, 92, 95, 96, 118, 119, 141, 154, 159, 166, 202, 211, 214, 215, 236, 238 Crystal, David  8 Culicover, Peter W.  xvi, 118, 142, 154 D Darnell, Michael  xv Davidse, Kristin  xvii

Demonte, Violeta  202 Di Sciullo, Anna Maria  97 Díez Velasco, Olga I.  135, 255 Dik, Simon C.  xviii, 18, 86, 87, 89, 90, 94, 97, 100, 118–122, 143, 154, 159, 167, 170, 171, 173, 174, 188 Dirven, René  217 Dixon, R.M.W  102, 218, 277, 280–282, 284 Downing, Angela  272–278, 282, 290, 291 Dowty, David R.  71, 162, 251, 265, 280, 284, 285 Dryer, Mathew  100 E Eckert, Paul  102 Eddington, David  204, 215, 236 Escandell Vidal, Victoria  223 Evans, Vyvyan  20 Everett, Daniel  89, 90, 97 F Faber, Pamela  122, 123, 125–127, 130, 132, 133, 144, 163, 164, 220, 250–252 Faltz, Leonard M.  212 Fawcett, Robin P.  172 Fillmore, Charles J.  15, 118, 119, 154, 160, 169, 176, 178, 219, 255 Foley, William  89, 100 Fox, Barbara A.  271 Fukaya, Teruhiko  205 G García, Erica  212 García Miguel, José M.  228 García Velasco, Daniel  xvii, 13, 20 Gennari, Silvia P.  27, 30 Gerrig, Richard J.  9, 18, 19, 20 Gibbs, Raymond W.  145

  Name index Givón, Talmy  93, 118, 119, 274, 275, 277 Goddard, Cliff  102, 107, 108, 130, 252 Goldberg, Adele  xvi, 15, 16, 25, 26, 30, 31, 34–36, 38, 39, 41–44, 63–69, 72, 73, 77, 82, 83, 86, 96, 117–119, 134, 140, 141, 145, 154, 158, 165, 166, 184, 188, 202, 204, 214–216, 219, 220, 225, 236, 238, 255, 271, 274, 279, 284, 286, 287, 289, 292, 293 Gonzálvez García, Francisco  118, 119, 134, 154, 155, 165, 201, 205, 210, 212, 215, 217–219, 221, 228, 239 Greenberg, Joseph  93 Gries, Stefan T.  204 Grimshaw, Jane  26 Gronemeyer, Claire  273, 275, 292 H Haiman, John  201, 232 Halliday, M.A.K.  118, 134, 171–173, 176, 182, 273 Hampe, Beate  204 Hatcher, Anna G.  273–276, 290, 291 Hay, Jennifer  94 Hengeveld, Kees  18, 19, 21, 89, 118, 143, 167 Hoey, Michael  175 Hohenhaus, Peter  4, 8 Holmes, Jasper  179 Hopper, Paul J.  118, 202, 233, 238 Huddleston, Rodney  211, 273 Hudson, Joyce  101, 102 Hudson, Richard  179 Hundt, Marianne  290 I Ibarretxe-Antuñano, Iraide  27, 30, 33, 205 J Jackendoff, Ray  xv, xvi, xvii, 118, 119, 142, 153, 154, 223, 236, 270, 286

Janda, Laura A.  211 Jurafsky, Dan  38, 238 K Katamba, Francis  5 Kathol, Andreas  179 Kay, Paul  15, 16, 69, 86, 96, 118, 154, 169, 176, 178 Kelly, Barbara  145, 271, 284 Kemmer, Suzanne E.  9, 203, 211, 212, 290 Koontz-Garboden, Andrew  266 Kratzer, Angelika  223 Kuteva, Tania  34 L L’Homme, Marie-Claude  133 Labov, William  272 Laffut, An  xvii Lakoff, George  134, 136, 186, 187 Lakoff, Robin  276–279, 282, 289 Lambrecht, Knud  32, 39, 87, 225 Langacker, Ronald W.  30, 38, 86, 87, 118, 123, 134, 142, 144, 161, 167, 169, 214, 228, 236, 276 LaPolla, Randy J.  77, 82, 86–90, 94, 100, 118, 124, 125, 127, 130–132, 159, 162, 163, 165, 188, 250, 251, 280, 283, 285 Lázaro Mora, Fernando A.  214 Leino, Jaakko  87 Lemmens, Maarten  65, 188 Leonetti, Manuel  223 Levin, Beth  26, 64, 125, 126, 154, 159, 161, 259, 263–266, 273, 281, 282 Lieber, Rochelle  14, 89, 90 Longacre, Robert E.  175 Lycan, William G.  8, 12 Lyons, John  4, 8 M Mackenzie, J. Lachlan  18, 19, 118, 167 Mahesh, Kavi  126 Mairal Usón, Ricardo  viii, xviii, 26, 62, 64, 89, 90,

92, 122–130, 132–144, 154, 155, 159, 163, 164, 184, 186, 220, 248–253, 258–261 Maldonado, Ricardo  203, 211, 212 Marconi, Diego  20 Marín Arrese, Juana I.  277, 290 Martin, James  176 Martín Arista, Javier  xviii, 62, 64, 89, 93, 95, 111 Martín Mingorance, Leocadio  120–123 Martín Zorraquino, María Antonia  206–212, 240 Martínez Vázquez, Montserrat  211, 212 Masini, Francesca  58 Massam, Diane  192 Masullo, Pascual  202 Matthiessen, Christian M.I.M  118, 171–173, 176, 182 McCarthy, Michael  270, 275, 278 Mel’čuk, Igor  xviii, 117, 132, 133, 138, 163, 164, 252, 253, 258 Mendikoetxea, Amaya  210 Mendívil Giró, José L.  35, 44 Michaelis, Laura A.  16, 17, 87, 134, 166, 187, 216, 258 Milsark, Gary L.  223 N Newmeyer, Frederick  xv Nichols, Johanna  97, 98, 100, 102, 132 Nirenburg, Sergei  126, 130 Nunberg, Geoffrey  5, 7, 17 Nuyts, Jan  xvi, 154 O O’Connor, Mary C.  118, 176, 178 Östman, Jan-Ola  214 Otal Campo, José Luis  143, 167, 175, 180, 217 P Panther, Klaus-Uwe  167, 171, 217 Pedersen, Johan  xii, 44 Peeters, Bert  144

Name index    Pérez Hernández, Lorena  143, 167, 183, 261 Pérez Quintero, María J.  viii, 143 Perry, John  181 Pinker, Steven  26 Piñón, Christopher  264, 265 Plag, Ingo  6 Pullum, Geoffrey  211 Pustejovsky, James  66, 69–74, 77, 253–255 Q Quirk, Randolph  4, 5, 211 R Rappaport Hovav, Malka  26, 154, 264, 265, 282 Recanati, François  181 Reinhart, Tanya  267 Robat, Nico J.  223 Rodríguez Espiñeira, María José  228 Rojo, Ana  48 Rudzka-Ostyn, Brigida  35 Ruiz de Mendoza Ibáñez, Francisco J.  xiii, xviii, 63, 64, 134–137, 139, 140–143, 153–155, 159, 167, 175, 180, 183, 184, 186, 206, 217, 220, 248, 249, 252, 253, 255, 258–261 Ruppenhofer, Josef  161

S Sadock, Jerrold M.  173 Sánchez López, Cristina  210, 211, 213 Santibáñez Sáenz, Francisco  136 Schack Rasmussen, Lone  121 Schank, Roger C.  183 Scheibman, Joanne  218 Schönefeld, Doris  204 Sethuraman, Nitya  159 Siewierska, Anna  272, 279 Sinha, Chris  34 Slobin, Dan, I.  27, 30, 33, 34 Snyder, William  44 Steen, Gerard  167 Stefanowitsch, Anatol  171 Štekauer, Pavol  8, 12–14, 18, 92 Svartvik, Jan  274, 284, 292 Sweetser, Eve  205 T Talmy, Leonard  xvii, 25–37, 39–41, 43, 45, 46, 49, 58, 59, 68 Taverniers, Miriam  xvii Thompson, Sandra  118, 202, 204, 233, 238 Thornburg, Linda  167, 171, 217 Tuggy, David  192 V Valenzuela, Javier  48 Van Ek, Jan A.  223

Van Valin, Robert D., Jr.  xvi, xviii, 63, 64, 77–79, 83, 86–90, 94, 100, 118, 124–127, 130–132, 139–141, 159, 162, 163, 165, 188, 220, 251, 252, 256, 263, 283 Vanrespaille, Mia  273, 277, 291 Vendler, Zeno  71, 162, 251 Vera Luján, Agustín  212 Visser, F.T.  278 Volpe, Mark  93 W Wanner, Leo  163, 252 Ward, Geoffrey  273 Weilbacher, Hunter  179 Weiner, E. Judith  272 Wierzbicka, Anna  130, 132, 138, 252 Wilkins, David P.  124, 163, 252 Winter, Eugene O.  175 Y Yangklang, Peerapat  30, 33, 34 Z Ziegeler, Debra  16, 17 Zlatev, Jordan  xii, 30, 33, 34 Zwicky, Arnold M.  100

Language index

D Danish  25–29, 32, 33, 36, 40, 42–52, 54–57 E English  xvii-xix, 4–7, 9, 12, 14, 15, 25, 26, 28, 29, 31, 33, 35, 36, 40–44, 46–52, 54–58, 60, 78, 88, 117, 121, 129, 131, 140, 144, 158, 166, 174, 185, 201–206, 211, 213–218, 220–223, 227, 228, 231, 232, 234–236, 238–240,

247, 252, 254, 256, 257, 271, 272, 274, 275, 278 F French  25, 49–52, 54–57 G German  25, 26, 40, 46–52, 54–57, 95 I Italian  25, 49–59, 266

O Old English  92, 95 P Pitjantjatjara/Yankunytjatjara  xviii, 85, 88, 102–110 S Spanish  xviii, 5, 25–29, 31–33, 35–38, 42–60, 95, 121, 144, 157, 158, 164–166, 191, 192, 201–218, 220–228, 231–240

Subject index

A acquisitional adequacy (see criteria of adequacy) action correlating event (see event) action realization event (see event) activity  28, 29, 31, 32, 36, 45, 68, 77–79, 124, 130, 133, 136, 162, 163, 181, 188, 251, 257, 260–262, 265, 267, 280, 282, 283 ACTOR (see Role and Reference Grammar) adjunct  90, 96, 125, 131, 257, 292 adverbial construction (see construction) AFFECT verb (see verb) affixation (see Word structure) Agent (see semantic role) Agent Proto-Role/ Proto-Agent  285, 286 ‘Agent-causer Blocking’ constraint (see Lexical Constructional Model) agentivity  277, 279, 284, 291 Aktionsart (see Role and Reference Grammar) alternative construal  31 analytic constructional schema (see constructional schema) ANNOYING verb (see verb) argument  xvi, xix,15, 17, 18, 26, 34, 55, 65, 66, 69, 71–73, 75, 78, 79, 81–83, 88–90, 92, 94, 101, 120, 124, 125, 130–133, 137, 139, 140, 142, 145, 153, 155–160, 163–166, 168, 169, 188, 189, 191, 192, 202, 215, 220, 222, 234, 238, 247, 248, 250, 252–254, 258, 262–266, 281, 284, 285, 287, 288, 292

argument structure  xvi, 15, 17, 18, 26, 34, 65, 71, 72, 83, 125, 137, 139, 142, 153, 157–159, 162, 169, 202, 215, 234, 238, 247, 248, 250, 253, 254, 284 argument construction (see construction) Argument Selection Principle  285 argument structure construction (see construction) Argument-Adjunct (see Role and Reference Grammar) asymmetric relation  32 asymmetry  93, 97, 111 B base (see Word structure) be-passive (see passive) BREAK verb (see verb) C causative get-passive (see passive) causative/inchoative alternation (see syntactic alternation) ‘Cause Expletivization’ constraint (see Lexical Constructional Model) caused motion construction (see construction) change-of-state verb (see verb) characteristic property of instrument alternation (see syntactic alternation) co-composition  xvii, xviii, 63–83, 141 coercion  3, 15–17, 21, 71, 134, 140, 145, 166, 187, 188, 193, 216, 217, 258, 259, 261, 266, 267

co-event  27, 28, 30–32, 35, 37, 43 coinage  3–21 common noun (see noun) complex cognitive event  32 complex event  28, 31–33 Complex Word (see Word structure) compounding (see Word structure) conative alternation (see syntactic alternation) Conceptual component (see Functional Discourse Grammar) conceptual cueing  137, 138, 156, 249 conflated events  31 conflation  29, 31, 187 constraint external constraint (in Lexical Constructional Model)  xix, 136–139, 141, 156, 167, 184, 185, 247, 249, 255, 259, 261, 268 internal constraint (in Lexical Constructional Model)  xix, 136, 138, 184, 185, 188, 189, 191, 247, 248, 255, 268, 269 construal  17, 29, 30, 31, 33, 161, 201, 223, 238 construction adverbial construction  32, 33, 37, 40, 42, 43, 45, 47, 56, 58 argument construction  155, 158, 165, 168 argument structure construction  17, 34 caused motion construction  16, 35, 63–68, 82, 83, 128, 134–136, 140, 158, 165, 184, 186, 187, 191, 292

  Subject index construction (Cont.) constructional information  35 imperfective verb phrase construction  46 illocutionary construction  168, 171, 175 inchoative construction  xviii, 165, 166, 189–191, 247–269, 290 lexical construction  36, 37, 48, 50, 58, 59 macro-event construction  25–60 main information construction (MIC)  25, 37, 39, 40–48, 50–59 morphological construction  xviii, 85–112 perfective verb phrase construction  36, 46 resultative construction  35, 43, 128, 140, 165 schematic construction  25, 34–38, 40, 41, 43–48, 50, 53–56, 58–60 supportive information construction (SIC)  25, 37, 39–48, 50–53, 55–59 verb phrase construction  36, 45, 46 constructional information (see construction) constructional schema (in Layered Structure of the Word, see Word structure) analytic constructional schema  101, 108, 109 continuous constructional schema  100, 107, 108 discontinuous constructional schema  100, 101, 107, 108 non-recursive constructional schema  101 recursive constructional schema  101 synthetic constructional schema  101, 108, 109, 111

constructional template (in Lexical Constructional Model, see template) constructivism/constructivist  xvi, 86, 143, 157 consumption verb (see verb) context-governed lexical creation (see lexical creation) Contextual component (see Functional Discourse Grammar) continuous constructional schema (see constructional schema) contrastive analysis  25, 35, 48, 238 Core (see Role and Reference Grammar) CORPOREAL verb (see verb) corpus  xviii, xix, 27, 48, 49, 59, 65, 125, 144, 202, 204, 205, 228, 235, 256, 271, 272, 275, 279, 281 Correspondence Principle  165, 166, 219, 284 count noun (see noun) creation/transformation alternation (see syntactic alternation) criteria of adequacy acquisitional adequacy  143 diachronic adequacy  144 discoursal adequacy  143, 144 descriptive adequacy  143 explanatory adequacy  143, 144, 155 psychological/cognitive adequacy  143, 144 sociocultural adequacy  143 cueing (see Lexical Constructional Model) D dependent-marking  98, 102, 111 descriptive adequacy (see criteria of adequacy) Destination (see semantic role) ‘destroy’ verb (see verb) Determinant (see Word structure)

Determinatum (see Word structure) diachronic adequacy (see criteria of adequacy) discontinuous constructional schema (see constructional schema) discoursal adequacy (see criteria of adequacy) distribution of information  33 E Effector (see semantic role) elaboration (see Lexical Constructional Model) Endocentric morphological construction (see morphological construction) entrenchment  38, 169 eponym  3, 4, 8–11, 15–21 equipollently-framed language  34 event event of action correlating  49 event of action realization  49 event of state change  49 event of temporal contouring  49 event segmentation  28 event structure  70–72, 75–77, 81, 83, 84, 133, 159, 160, 247, 250, 251, 253, 254, 256, 259 macro-event  25–60 motion event  26, 27, 29, 31, 35, 36, 41–43, 49–51, 58 event segmentation (see event) event structure (see event) Exocentric morphological construction (see morphological construction) Experiencer (see semantic role) EXPERIENTIAL ACTION IS EFFECTUAL ACTION metaphor  136 explanatory adequacy (see criteria of adequacy)

Subject index    Explanatory and Combinatorial Lexicology  132, 163, 252 Extended Override Principle  258, 259, 261 external constraint (see Lexical Constructional Model) F figure-ground alternation  30 figure-ground relation  28–30 First Argument (see Role and Reference Grammar) framing event  27, 29, 31 full-matching (see Lexical Constructional Model) Functional Discourse Grammar (FDG) Conceptual component  19, 20 Contextual component  3, 18–21 Grammatical component  18, 19, 21 functional/functionalism/ functionalist  xv–xix, 3, 7, 14, 21, 63, 64, 86–90, 92, 94–97, 99, 101, 110, 117–119, 124, 134, 138–140, 142–144, 153–155, 157, 171, 203, 212, 220, 222, 238, 240, 248, 250 G Generative Lexicon  xviii, 63, 64, 69, 70, 83, 84, 248 genesis of Lexical Constructional Model (see Lexical Constructional Model) Germanic language  xvii, 25, 27, 29, 40, 45, 58, 59 get-passive (see passive) GIVING verb (see verb) Goal (see semantic role) Grammatical component (see Functional Discourse Grammar) grammaticalization  93 H head-marking  98, 111 headedness  70, 98, 100

‘hit verb’ (see verb) HIT verb (see verb) I illocutionary construction (see construction) imperfective verb phrase construction (see construction) inchoative construction (see construction) inflection (see Word structure) information structure  25, 30–33, 37, 39–41, 59, 218, 219 inheritance relation  38, 292 Instrument (see semantic role) instrument subject alternation (see syntactic alternation) internal constraint (see Lexical Constructional Model) intransitive motion event  35, 43 L Layered structure of the clause (see Role and Reference Grammar) Layered Structure of the Word (LSW, see Word structure) levels of description  153–193 lexical blocking (see Lexical Constructional Model) lexical construction (see construction) Lexical Constructional Model (LCM) ‘Agent-causer Blocking’ constraint  247, 265 ‘Cause Expletivization’ constraint  247, 263, 267 cueing  137, 138, 155, 156, 167, 175, 180–182, 193, 249, 258 elaboration  168, 176, 177, 201, 204, 232, 233, 238, 240, 258, 259, 261 external constraint  xix, 247, 261 full-matching  258 genesis of Lexical Constructional Model  119–146 internal constraint  188, 189, 191

lexical blocking  129, 136, 191, 259, 264 lexical-constructional subsumption  xviii, 135, 137, 138, 145, 156, 168, 182, 184–186, 188, 189, 191–194, 248, 249, 255, 256, 258 subevent selection  259, 261 suppression of variables  128, 259 unification process  249, 258 lexical creation context-governed lexical creation  6 rule-governed lexical creation  6 lexical function  132, 133, 138, 153, 163, 164, 190, 252, 253, 256, 258, 262, 267 lexical level  25, 36, 48, 58, 154 lexical representation  63, 70–75, 77, 79, 83, 84, 155, 158, 159, 162, 250, 254 lexical template (in Lexical Constructional Model, see template) lexical template modelling process (in Lexical Constructional Model, see template) lexical-constructional subsumption (see Lexical Constructional Model) lexicalization  25, 26, 30, 32, 34, 37, 40, 42, 48, 59 lexicon-grammar continuum  142 linearization  86, 90, 93, 94, 100 linking algorithm (see Role and Reference Grammar) linking rule (see Role and Reference Grammar) logical structure  72, 76, 78, 79, 82, 124, 130, 131, 133, 136, 137, 139, 140, 157, 160, 162–165, 191, 250–254, 258, 262–266 M macro-event (see event) macro-event construction (see construction)

  Subject index macro-event property  28, 40, 41 macrorole (see Role and Reference Grammar) main event  27, 28, 30–32, 34, 35, 43, 45 main information construction (see construction) manner of movement  26 manner of perception  46–48 markedness  85, 96–99, 111 mass noun (see noun) merge  xvii, xviii, 63–83 metonymy occurrent metonymy  5 PROCESS FOR ACTION metonymy  xix, 137, 247, 261, 263 PART OF A PROCESS FOR ACTION metonymy  261 middle alternation (see syntactic alternation) morphological construction (see also construction) Endocentric morphological construction  85, 98, 99, 106, 108, 109 Exocentric morphological construction  85, 98, 99, 101, 102, 105–107, 111 morphological template (of Role and Reference Grammar, see template) motion event (see event) motion inchoative structure  257 MOTION verb (see verb) N neologism  3, 4, 7–10, 14, 18 nonce-formation  3, 4, 7, 8, 18 non-recursive constructional schema (see constructional schema) noun common noun  7, 12, 14, 17 count noun  17 mass noun  17 proper noun  3, 7–9, 12–15, 17–19, 21, 217 Nucleus (see Role and Reference Grammar)

O occurrent metonymy (see metonymy) operator  28, 40, 89–95, 101, 127, 128, 159, 163, 165, 188–190, 251, 253, 256, 257, 259 Override Principle  17, 134, 166, 258, 259, 261 P parallel corpus  27, 59 PART OF A PROCESS FOR ACTION metonymy (see metonymy) passive be-passive  272–274, 277, 279, 288, 293 causative get-passive  271, 287, 290–293 get-passive  xix, 271–293 spontaneous getpassive  271, 287, 289–293 path of movement  26 path phrase alternation (see syntactic alternation) Patient Proto-Role/ Proto-Patient  285, 286, 292 pattern of coining  15, 16 perception path  47 perception verb  47 percolation (see Word structure) perfective verb phrase construction (see construction) Performer (see semantic role) Periphery (see Role and Reference Grammar) Postfield  93, 94, 98, 100–102, 106–111 Prefield  93, 94, 98, 100, 101, 106, 107, 110, 111 prefix (see Word structure) PROCESS FOR ACTION metonymy (see metonymy) projection  21, 85–95, 98, 139, 155, 159, 162, 181, 255 projectionism/projectionist  xii, 63, 64, 70, 77, 83,

139–141, 154, 155, 157–159, 161, 162, 250 proper noun (see noun) psychological/cognitive adequacy (see criteria of adequacy) Q qualia structure  63, 71, 72, 75, 77, 83, 253, 254 R realization/completion  29, 43 Recipient (see semantic role) recursive constructional schema (see constructional schema) recursivity  101, 110, 111 reduplication (see Word structure) reference transfer  5 reflexivity  201–240 relative prominence  29, 31 result state inchoative structure  257 resultative construction (see construction) resultative phrase alternation (see syntactic alternation) Role and Reference Grammar (RRG) ACTOR  76, 78–81, 88, 89, 130, 131, 136 Aktionsart  162, 163, 165, 188, 190, 193, 251 Argument-Adjunct  63, 90, 94, 101 Core  78, 94, 124, 131 First Argument  79, 92, 94, 130, 258, 265 Layered structure of the clause  88, 90, 93, 131 linking algorithm  64, 78–80, 83, 88, 130–132, 139, 140 linking rule  129, 130, 132, 134, 139 macrorole  78, 88, 89, 129–132, 140 Nucleus  85, 89–91, 93–95, 98, 100, 101, 106–111, 131, 132, 140

Subject index    Periphery  90, 131, 137, 292 Second Argument  79, 92, 94, 133, 188, 253, 263, 281, 285 semantics-syntax linking  94 UNDERGOER  76, 78, 80–82, 88, 89, 130, 135, 285 Romance language  xvii, 25–27, 29, 37, 40, 44, 58, 59, 95 rule-governed lexical creation (see lexical creation) S satellite-framed language  25–27, 58 schematic construction (see construction) Second Argument (see Role and Reference Grammar) secondary information  31, 32, 35, 45, 48, 59 secondary predication  201–240 Semantic Coherence Principle  165, 166, 219, 284 semantic role Agent  128, 130, 185, 191, 208, 211, 254, 265, 266, 272, 274, 275, 277, 280, 284–292 Destination  76, 82, 187, 257, 281 Effector  126, 128, 130, 136, 186, 264–266, 280, 289 Experiencer  130, 136, 201, 206–208, 211, 221, 222, 232, 238, 276, 281, 284, 287, 291 Goal  82, 121, 136, 186, 187, 257, 265, 281, 287, 288 Instrument  126–128, 160, 265, 266, 282 Performer  265, 266, 268 Recipient  280, 281, 283, 287–289 Source  281, 287–289 Stimulus  281, 284, 291 Theme  66, 130, 219, 268, 280, 281, 285, 288 semantics-syntax linking (see Role and Reference Grammar)

serial-verb language  34 Simplex Word (see Word structure) sociocultural adequacy (see criteria of adequacy) Source (see semantic role) spontaneous get-passive (see passive) STAB verb (see verb) state change event (see event) Stimulus (see semantic role) subevent selection (see Lexical Constructional Model) suffix (see Word structure) support event  28 supportive information construction (see construction) suppression of variables (see Lexical Constructional Model) symmetric relation  32 syntactic alternation causative/inchoative alternation  64, 70, 128, 129, 137, 260 characteristic property of instrument alternation  126 conative alternation  126 creation/transformation alternation  126 instrument subject alternation  126–128 middle alternation  126 path phrase alternation  126 resultative phrase alternation  126, 127 transitive syntactic alternation  126 unintentional interpretation alternation  126 unspecified object alternation  126 syntactic template (of Role and Reference Grammar, see template) synthetic constructional schema (see constructional schema)

T telicity  xix, 247, 260, 263 template constructional template (in Lexical Constructional Model)  xviii, 128, 130, 137–140, 142, 144, 156, 165, 184, 185, 188, 191, 193, 249, 250, 255, 258, 259, 261, 268 lexical template (in Lexical Constructional Model)  xviii, xix, 125–130, 133–139, 142, 155, 156, 162–165, 184, 188, 190, 191, 193, 247, 249, 250, 252–255, 258, 259, 261, 264, 268, 269 lexical template modelling process (in Lexical Constructional Model)  127, 129, 130, 136 morphological template (of Role and Reference Grammar)  93, 94, 98, 106 syntactic template (of Role and Reference Grammar)  130, 140 template slot (of Role and Reference Grammar)  94 template slot (of Role and Reference Grammar, see template) temporal contouring event (see event) Theme (see semantic roles) TOUCH verb (see verb) transitive motion event  35 transitive syntactic alternation (see syntactic alternation) U UNDERGOER (see Role and Reference Grammar) unification process (see Lexical Constructional Model) unintentional interpretation alternation (see syntactic alternation) unspecified object alternation (see syntactic alternation) usage event  38

  Subject index usage-based  xviii, 14, 38, 154, 201, 203–205, 215, 235, 236, 240 V valence structure  32, 41, 60 verb AFFECT verb  280, 281, 287 ANNOYING verb  281, 283, 287 BREAK verb  63–66, 70, 75, 79, 81, 266, 282 change-of-state verb  68, 187, 189, 259, 260, 280, 289, 290 consumption verb  263 CORPOREAL verb  271, 280, 281, 283, 287, 291 ‘destroy’ verb  264 GIVING verb  280, 281 ‘hit’ verb  262, 263 HIT verb  280, 281, 287 MOTION verb  184, 185, 281, 293 STAB verb  282

TOUCH verb  281 verb of contact-byimpact  262, 281 verb of cutting  126, 128, 130, 260 verb of existence  264 verb phrase construction (see construction) verbal conversion  xvii, 4, 14, 15 verbal semantics  44, 154, 158, 159, 220 verb-framed language  25, 26 verb of contact-by-impact (see verb) verb of cutting (see verb) verb of existence (see verb) W Word structure affixation  9, 14, 85, 90, 111 base  90, 93, 95, 96, 107 Complex Word  4, 89, 90, 92, 94, 95, 99, 100, 106, 110 compounding  xviii, 85, 89, 90, 92, 95, 96, 104, 111

constructional schemas  xvi, 85, 88, 95–97, 100–102, 105, 107–112, 139, 140 Determinant  90, 129 Determinatum  90, 129 inflection  85, 89, 90, 93, 95, 100, 105, 107, 108, 110–112 Layered Structure of the Word (LSW)  89–91, 93, 95, 98, 101, 110, 111 percolation  84, 90, 95, 98, 108, 109, 111 prefix  26, 102 reduplication  103–105, 107, 170, 213 Simplex Word  90, 91 suffix  98, 102, 107, 109, 257 Word Syntax  89, 90 zero-derivation  104 Word Syntax (see Word structure) Z zero-derivation (see Word structure)

Studies in Language Companion Series A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 110 Dimmendaal, Gerrit J. (ed.): Coding Participant Marking. Construction types in twelve African languages. xiii, 386 pp. + index. Expected March 2009 109 Narrog, Heiko: Modality in Japanese. The layered structure of the clause and hierarchies of functional categories. xxii, 276 pp. + index. Expected March 2009 108 Barðdal, Jóhanna and Shobhana L. Chelliah (eds.): The Role of Semantic, Pragmatic, and Discourse Factors in the Development of Case. xx, 422 pp. + index. Expected February 2009 107 Butler, Christopher S. and Javier Martín Arista (eds.): Deconstructing Constructions. 2009. xx, 306 pp. 106 Vanhove, Martine (ed.): From Polysemy to Semantic Change. Towards a typology of lexical semantic associations. 2008. xiii, 404 pp. 105 Van Valin, Jr., Robert D. (ed.): Investigations of the Syntax–Semantics–Pragmatics Interface. 2008. xxiv, 484 pp. 104 Mushin, Ilana and Brett Baker (eds.): Discourse and Grammar in Australian Languages. 2008. x, 239 pp. 103 Josephson, Folke and Ingmar Söhrman (eds.): Interdependence of Diachronic and Synchronic Analyses. 2008. viii, 350 pp. 102 Goddard, Cliff (ed.): Cross-Linguistic Semantics. 2008. xvi, 356 pp. 101 Stolz, Thomas, Sonja Kettler, Cornelia Stroh and Aina Urdze: Split Possession. An areallinguistic study of the alienability correlation and related phenomena in the languages of Europe. 2008. x, 546 pp. 100 Ameka, Felix K. and M.E. Kropp Dakubu (eds.): Aspect and Modality in Kwa Languages. 2008. ix, 335 pp. 99 Høeg Müller, Henrik and Alex Klinge (eds.): Essays on Nominal Determination. From morphology to discourse management. 2008. xviii, 369 pp. 98 Fabricius-Hansen, Cathrine and Wiebke Ramm (eds.): 'Subordination' versus 'Coordination' in Sentence and Text. A cross-linguistic perspective. 2008. vi, 359 pp. 97 Dollinger, Stefan: New-Dialect Formation in Canada. Evidence from the English modal auxiliaries. 2008. xxii, 355 pp. 96 Romeo, Nicoletta: Aspect in Burmese. Meaning and function. 2008. xv, 289 pp. 95 O’Connor, Loretta: Motion, Transfer and Transformation. The grammar of change in Lowland Chontal. 2007. xiv, 251 pp. 94 Miestamo, Matti, Kaius Sinnemäki and Fred Karlsson (eds.): Language Complexity. Typology, contact, change. 2008. xiv, 356 pp. 93 Schalley, Andrea C. and Drew Khlentzos (eds.): Mental States. Volume 2: Language and cognitive structure. 2007. x, 362 pp. 92 Schalley, Andrea C. and Drew Khlentzos (eds.): Mental States. Volume 1: Evolution, function, nature. 2007. xii, 304 pp. 91 Filipović, Luna: Talking about Motion. A crosslinguistic investigation of lexicalization patterns. 2007. x, 182 pp. 90 Muysken, Pieter (ed.): From Linguistic Areas to Areal Linguistics. 2008. vii, 293 pp. 89 Stark, Elisabeth, Elisabeth Leiss and Werner Abraham (eds.): Nominal Determination. Typology, context constraints, and historical emergence. 2007. viii, 370 pp. 88 Ramat, Paolo and Elisa Roma (eds.): Europe and the Mediterranean as Linguistic Areas. Convergencies from a historical and typological perspective. 2007. xxvi, 364 pp. 87 Verhoeven, Elisabeth: Experiential Constructions in Yucatec Maya. A typologically based analysis of a functional domain in a Mayan language. 2007. xiv, 380 pp. 86 Schwarz-Friesel, Monika, Manfred Consten and Mareile Knees (eds.): Anaphors in Text. Cognitive, formal and applied approaches to anaphoric reference. 2007. xvi, 282 pp. 85 Butler, Christopher S., Raquel Hidalgo Downing and Julia Lavid (eds.): Functional Perspectives on Grammar and Discourse. In honour of Angela Downing. 2007. xxx, 481 pp. 84 Wanner, Leo (ed.): Selected Lexical and Grammatical Issues in the Meaning–Text Theory. In honour of Igor Mel'čuk. 2007. xviii, 380 pp.

83 Hannay, Mike and Gerard J. Steen (eds.): Structural-Functional Studies in English Grammar. In honour of Lachlan Mackenzie. 2007. vi, 393 pp. 82 Ziegeler, Debra: Interfaces with English Aspect. Diachronic and empirical studies. 2006. xvi, 325 pp. 81 Peeters, Bert (ed.): Semantic Primes and Universal Grammar. Empirical evidence from the Romance languages. 2006. xvi, 374 pp. 80 Birner, Betty J. and Gregory Ward (eds.): Drawing the Boundaries of Meaning. Neo-Gricean studies in pragmatics and semantics in honor of Laurence R. Horn. 2006. xii, 350 pp. 79 Laffut, An: Three-Participant Constructions in English. A functional-cognitive approach to caused relations. 2006. ix, 268 pp. 78 Yamamoto, Mutsumi: Agency and Impersonality. Their Linguistic and Cultural Manifestations. 2006. x, 152 pp. 77 Kulikov, Leonid, Andrej Malchukov and Peter de Swart (eds.): Case, Valency and Transitivity. 2006. xx, 503 pp. 76 Nevalainen, Terttu, Juhani Klemola and Mikko Laitinen (eds.): Types of Variation. Diachronic, dialectal and typological interfaces. 2006. viii, 378 pp. 75 Hole, Daniel, André Meinunger and Werner Abraham (eds.): Datives and Other Cases. Between argument structure and event structure. 2006. viii, 385 pp. 74 Pietrandrea, Paola: Epistemic Modality. Functional properties and the Italian system. 2005. xii, 232 pp. 73 Xiao, Richard and Tony McEnery: Aspect in Mandarin Chinese. A corpus-based study. 2004. x, 305 pp. 72 Frajzyngier, Zygmunt, Adam Hodges and David S. Rood (eds.): Linguistic Diversity and Language Theories. 2005. xii, 432 pp. 71 Dahl, Östen: The Growth and Maintenance of Linguistic Complexity. 2004. x, 336 pp. 70 Lefebvre, Claire: Issues in the Study of Pidgin and Creole Languages. 2004. xvi, 358 pp. 69 Tanaka, Lidia: Gender, Language and Culture. A study of Japanese television interview discourse. 2004. xvii, 233 pp. 68 Moder, Carol Lynn and Aida Martinovic-Zic (eds.): Discourse Across Languages and Cultures. 2004. vi, 366 pp. 67 Luraghi, Silvia: On the Meaning of Prepositions and Cases. The expression of semantic roles in Ancient Greek. 2003. xii, 366 pp. 66 Nariyama, Shigeko: Ellipsis and Reference Tracking in Japanese. 2003. xvi, 400 pp. 65 Matsumoto, Kazuko: Intonation Units in Japanese Conversation. Syntactic, informational and functional structures. 2003. xviii, 215 pp. 64 Butler, Christopher S.: Structure and Function – A Guide to Three Major Structural-Functional Theories. Part 2: From clause to discourse and beyond. 2003. xiv, 579 pp. 63 Butler, Christopher S.: Structure and Function – A Guide to Three Major Structural-Functional Theories. Part 1: Approaches to the simplex clause. 2003. xx, 573 pp. 62 Field, Fredric: Linguistic Borrowing in Bilingual Contexts. With a foreword by Bernard Comrie. 2002. xviii, 255 pp. 61 Goddard, Cliff and Anna Wierzbicka (eds.): Meaning and Universal Grammar. Theory and empirical findings. Volume 2. 2002. xvi, 337 pp. 60 Goddard, Cliff and Anna Wierzbicka (eds.): Meaning and Universal Grammar. Theory and empirical findings. Volume 1. 2002. xvi, 337 pp. 59 Shi, Yuzhi: The Establishment of Modern Chinese Grammar. The formation of the resultative construction and its effects. 2002. xiv, 262 pp. 58 Maylor, B. Roger: Lexical Template Morphology. Change of state and the verbal prefixes in German. 2002. x, 273 pp. 57 Mel’čuk, Igor A.: Communicative Organization in Natural Language. The semantic-communicative structure of sentences. 2001. xii, 393 pp. 56 Faarlund, Jan Terje (ed.): Grammatical Relations in Change. 2001. viii, 326 pp. 55 Dahl, Östen and Maria Koptjevskaja-Tamm (eds.): Circum-Baltic Languages. Volume 2: Grammar and Typology. 2001. xx, 423 pp. 54 Dahl, Östen and Maria Koptjevskaja-Tamm (eds.): Circum-Baltic Languages. Volume 1: Past and Present. 2001. xx, 382 pp. 53 Fischer, Olga, Anette Rosenbach and Dieter Stein (eds.): Pathways of Change. Grammaticalization in English. 2000. x, 391 pp.