Verbal Plurality and Distributivity 9783110293500, 9783110292107

Table of contents :
Authors’ biographies
Introduction
Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect
Pluractionality and accompaniment in Cuzco Quechua
How plural can verbs be?
Event-based additivity in English and Modern Hebrew
On distributivity in Karitiana
Pluractional verbs: An overview
Distributivity is not uniformly over events
Index

Citation preview

Linguistische Arbeiten

546

Herausgegeben von Klaus von Heusinger, Gereon Müller, Ingo Plag, Beatrice Primus, Elisabeth Stark und Richard Wiese

Patricia Cabredo Hofherr, Brenda Laca (Eds.)

Verbal Plurality and Distributivity

De Gruyter

ISBN 978-3-11-029210-7 e-ISBN 978-3-11-029350-0 ISSN 0344-6727 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. © 2012 Walter de Gruyter GmbH & Co. KG, Berlin/Boston Gesamtherstellung: Hubert & Co. GmbH & Co. KG, Göttingen ∞ Gedruckt auf säurefreiem Papier Printed in Germany www.degruyter.com

Table of contents

Authors’ biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi Patricia Cabredo Hofherr & Brenda Laca Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Patrícia Amaral & Chad Howe Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect . . . . . . . . . . . . . . . . . . . . . . . . . . . .

25

Martina Faller Pluractionality and accompaniment in Cuzco Quechua . . . . . . . . . . . . .

55

Abdelkader Fassi Fehri How plural can verbs be? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

87

Yael Greenberg Event-based additivity in English and Modern Hebrew . . . . . . . . . . . . . 127 Ana Müller & Esmeralda Negrão On distributivity in Karitiana . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Paul Newman Pluractional verbs: An overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Asya Pereltsvaig Distributivity is not uniformly over events . . . . . . . . . . . . . . . . . . . . . . 211 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

Authors’ biographies

Patrícia M. Amaral is Assistant Professor in the Department of Romance Languages at the University of North Carolina at Chapel Hill. She received a Ph.D. in Hispanic Linguistics from the Ohio State University in 2007. Prior to her current appointment, she was Lecturer at the University of Liverpool and a post-doctoral fellow in the Department of Linguistics at Stanford University. She has published on the semantics and pragmatics of scalar adverbs in Romance, experimental pragmatics, and language acquisition. Her recent publications include papers in Linguistics, Linguistics and Philosophy and Language and Linguistics Compass. Patricia Cabredo Hofherr is a researcher at the UMR 7023 Structures formelles du langage (CNRS & Paris 8). She has worked on aspect and verbal plurality, the syntax-morphology interface and impersonal pronouns. Her current research focuses on the syntax and semantics of R-impersonal pronouns cross-linguistically. Martina Faller is a Senior Lecturer in the Department of Linguistics and English Language at The University of Manchester. She received her Ph.D. in Linguistics from Stanford University in 2002. Her primary research focus is in semantics and pragmatics with a particular interest in cross-linguistic variation. She has studied a number of issues in Quechua, including evidentiality, modality, and nominal and verbal quantification and has carried out regular fieldwork on this language in and around Cusco, Peru. Abdelkader Fassi Fehri is currently a Senior Researcher at the King Abdallah’s International Center for the Arabic Language (KAICAL), Riyad, and President of the Linguistic Society of Morocco, Rabat. Previously, he was Chair of Arabic & Comparative Linguistics and Director of the Institute for the Study and Research on Arabization at Mohammed V University in Rabat. He is an active Member of Arabic Academies, and has published numerous books and articles in Arabic, English, and French on Arabic Grammar and linguistic theory, including Issues in the Structure of Arabic Clauses and Words (1993, Kluwer), and Lexicon of Linguistic Terms (trilingual, Beyrouth, 2009). His current research focuses on Number generality, the count/ mass classification, silent pronouns, time and space anchoring, pluractionality, variation and parametrization.

Authors’ biographies

vii

Yael Greenberg teaches formal semantics, and semantics-pragmatics interface at Bar Ilan University. Her Ph.D. on two types of nonaccidental generics was published by Routledge Outstanding Dissertations in Linguistics series. Her research interests include the syntax-semantics interface of various constructions in Modern Hebrew, the semantics-pragmatics interface of aspectual particles as well as pluractionality and genericity. She is currently heading an ISF research project on the semantics, pragmatics and focus sensitivity of focus sensitive expressions in Modern Hebrew, which has general and cross linguistic implications to the research of exclusivity, polarity sensitive items and fine grained models of discourse. Chad Howe is an Assistant Professor in the Department of Romance Languages and the Linguistics Program at the University of Georgia. He received his Ph.D. in Spanish Linguistics from The Ohio State University in 2006. His primary areas of research are language change and variation in Romance, with an emphasis on temporal and aspectual forms. He is especially interested in corpus-based and quantitative approaches to explaining processes of semantic change. Brenda Laca is Professor of Linguistics at the University of Paris 8 and a member of the UMR 7023 Structures formelles du langage (CNRS & Paris 8). Her research interests focus on semantics. She has worked on genericity, determinerless nouns, (co-)distributivity, aspectual periphrases and modal verbs. Ana Müller is Assistant Professor of Linguistics at the University of São Paulo, Brazil. She holds a PhD in Linguistics from the Universidade de Campinas. Her primary research areas are formal semantics and the syntaxsemantics interface of Brazilian Portuguese and of Brazilian native languages. Her research interests include the semantics of nominal expressions and of genericity. Her current research focuses on the expression of plurality and distributivity in Karitiana, a Tupi language. She has edited a volume on the semantics of Brazilian Portuguese and has published work in Probus, Journal of Portuguese Linguistics and in Brazilian linguistics journals. Esmeralda Vailati Negrão is Full Professor of Linguistics at the University of São Paulo, Brazil. Her main research area is the interface between the syntax and semantics of natural languages, especially Brazilian Portuguese. She took part in research projects that dealt with quantification in natural languages and the comparison between the grammars of Brazilian Portuguese and European Portuguese. The main results of her research have been published as articles in journals and book chapters in Brazil and abroad. She has coedited a volume on the syntax of Brazilian Portuguese, Brazilian Portuguese and the Null Subject Parameter.

viii

Authors’ biographies

Paul Newman is currently Distinguished Professor Emeritus of Linguistics and Adjunct Professor of Law affiliated with the Center on Law, Society, and Culture at Indiana University. His main research area is Chadic linguistics. He is the author of numerous books, including Nominal and Verbal Plurality in Chadic (1990) and The Hausa Language: An Encyclopedic Reference Grammar (2000). He was the founding editor of the Journal of African Languages and Linguistics, and has been on the editorial board of Language, Current Anthropology, Studies in African Linguistics, and Anthropological Linguistics. He is a life member of the Linguistic Society of America (LSA), serving as a member of the Endangered Languages Committee and as a member and chair of the Social and Political Issues Committee. Asya Pereltsvaig holds a PhD in Linguistics from McGill University. She is currently teaching linguistics at Stanford University, and prior to her current appointment, she has taught at Yale and Cornell. Her current research focuses on the distribution, structure and interpretation of noun phrases in Russian, with special focus on the syntax-semantics interface of numerical quantification and adjectival modification. Recent publications include papers on issues in syntax and semantics of Slavic languages in Natural Language and Linguistic Theory, Lingua, Studia Linguistica and Russian Linguistics.

Patricia Cabredo Hofherr & Brenda Laca

Introduction – event plurality, verbal plurality and distributivity

1. Theoretical issues The articles collected in this volume deal with different aspects of event plurality in a wide variety of languages (Arabic, Cuzco Quechua, European Portuguese, Karitiana, Modern Hebrew, and Russian).* Before we present the individual papers, we would like to lay out the central issues that underlie much recent research on event plurality.1 We consider under the general term EVENT PLURALITY any linguistic means of expressing a multiplicity of events, be they verbal markers (reread), adverbials (twice, often, always, again), or adnominal markers (John lived in different countries, each boy built a canoe, John repaired several bicycles). We use the term VERBAL PLURALITY more narrowly for event plurality marked on the verb. Following the usage in the literature we refer to markers of verbal plurality as PLURACTIONAL MARKERS. A rough typology of the expressions of event plurality should chart the specific restrictions that markers may impose on the event pluralities they describe. Here, we have singled out restrictions bearing on three different aspects of event plurality: (i) variability among singular events within the event plurality, (ii) restrictions on event pluralities made up of (possibly singular) asserted and presupposed events and (iii) the expression of exact cardinality in the event domain. The first type of restriction concerns the degree of variability among the singular events within the event plurality: along which dimensions (partici–––––––—–– *

1

We gratefully acknowledge the support of the Fédération Typologie et Universaux Linguistiques (CNRS FRE 2559) for the program Dépendances distributives: pluralité nominale et verbale. The papers collected here were presented at the Workshop on nominal and verbal plurality organized by the editors at the University Paris 8 in November 2008. In the glosses we have preserved the glossing in the original sources where possible. The abbreviations used are the following: ABS = absolutive, ACC = accusative, ADV = adverb, ASP = aspect, CL = classifier, DEM = demonstrative, DISTR = distributive morpheme, ERG = ergative, IND = indicative, INS = instrumental, LC = limited control, MOD = modal, PF = perfective, PL = plural, PLR = pluractional, PPRT= past participle, PREP = preposition, PRS = present, PST = past, S = subject, SG = singular, SP = perfective past, RED = reduplication, RL = realis, TR = transitive, TRZ = transitivizer, WP = witnessed past (Chechen).

2

Patricia Cabredo Hofherr & Brenda Laca

pants, space, time) are the singular events in the event-plurality allowed or required to differ from each other? As we will see, clear requirements on the spatio-temporal distribution of the event-plurality are found with different markers (section 2.1). The conditions restricting the range of relationships between the arguments of the (sub-)events and the event plurality are more intricate as they may depend on the type of argument-DP and on the argument position (section 2.2). Since the distributive behaviour of many verbal plurality markers is typically scopeless, we will highlight the contrasts found with scope-taking expressions as exemplified by adnominal markers and reduplicated numerals (section 2.3). A second array of restrictions is found with additive markers of event plurality. These are peculiar in that they yield a mixed event plurality resulting from the combination of one (or more) asserted events with one or more presupposed events. As we will see in section 3 the challenge of characterising possible variability between events rearises in a different guise here since the presupposed event is construed on the basis of the asserted event, and different additive markers impose different identity conditions on the description of the presupposed event. The third type of restriction concerns the specification of cardinality. Many pluractional markers are reported to be incompatible with the expression of cardinality of the event plurality they contribute (section 4.1). This incompatibility is unexpected if pluractional markers are seen as the expression of plurality in the verbal domain on a par with inflectional plural number in the nominal domain. For this reason we will place the exact-cardinality restriction in the wider context of similarities and differences observed for plurality in the nominal and in the verbal domain (section 4.2).

2. Event plurality: restrictions on variability among sub-events Markers of event plurality vary with respect to the conditions they impose on the events making up the event plurality.2 Markers of verbal plurality in particular can be differentiated for a number of properties including distribution over participants, places and times as well as causation and multiple displacement (see Mithun 1988b, 217, for North American languages). Here, we –––––––—–– 2

A single language typically has various means of marking event plurality, with a choice of verbal, adverbial or adnominal expressions. Note that languages may also have more than one marker of verbal plurality; for studies comparing different pluractional markers in a single language see e.g. Collins (2001), Faller (this volume), Garrett (2001), Rose (2008).

Introduction

3

will consider examples with single markers that do not have clear additional grammatical meaning such as causation or agency. For markers that distribute the event plurality the distribution requirements can be of different types: the distribution required for the felicitous use of a marker can be over different individuals, places or times, and possibly over more than one dimension simultaneously. For pluractional markers this can be captured by the formal definition proposed by Lasersohn in his seminal analysis (1995:256). (1)

V-PA(X) e, e’
X [P(e) & ¬K(e)
K(e') & card(X)  n] with K = temporal trace or spatial trace or participants of the event

One of the crucial ingredients of Lasersohn’s concerns the individual events of the plurality: the formula requires these events to fulfill a property P that is not necessarily identical to the basic verb V that the pluractional marker PA attaches to. Lasersohn specifically distinguishes cases where P is identical to V from cases where P is otherwise lexically specified. In what follows we will first give examples illustrating markers of event plurality that impose spatial or temporal distribution on the individual subevents of the plurality. We will then turn to examples that include distribution over participants. In terms of Lasersohn's definition, the latter examples show that beyond the basic verb meaning the construction of the property P has to take the nominal arguments into account. 2.1 Temporal and spatial distribution requirements A clear example of a pluractional marker requiring spatial distribution of the event plurality is the complex marker kí-Verb-q||o described for Hoan in Collins (2001). Collins gives the meaning of this construction as marking “that there are several different places at which the event or action is sequentially repeated" (Collins 2001, 467). The fact that kí-Verb-q||o is incompatible with PPs meaning “in one place” shows that the marker requires that different places be involved (2). (2)

Titi ikí-‘am-q||o *(ki ci m
un) (Hoan) Titi PAST kí[pl] eat-around (PREP place one) ‘Titi eats around *(in one place).’ (Collins 2001:467, ex 31a)

An example of a marker including temporal distribution is the West Greenlandic pluractional marker -tar- "repeatedly" analysed in Van Geenhoven (2004, 2005) as imposing distribution in time with temporal gaps between the individual events:

4

Patricia Cabredo Hofherr & Brenda Laca (3)

Nuka ullaap tungaa tamaat sanioqquttarpoq. (West Greenlandic) Nuka ullaa-p tunga-a tama-at N.ABS morning-ERG direction-SG.SG.ABS all-3SG saniuqqut-tar-puq go.by-repeatedly-IND.[–TR].3SG
Nuka went by repeatedly for the whole morning.’ (Van Geenhoven 2004: 146, ex 27)

Another clear case of distribution over different times (i.e. distribution in time with temporal gaps) is provided by the present perfect in a Northeastern variety of Brazilian Portuguese that we have argued should be analysed as a pluractional marker (Cabredo Hofherr, Laca & de Carvalho 2010). The requirement that the event-plurality described by the present perfect be separated by temporal gaps is illustrated by the following minimal pair: (4)

a.

b.

Pedro tem dormido na varanda o inverno inteiro. (NE-BrPort) P. has sleep.PPRT in-the balcony the winter whole ‘Pedro has been sleeping on the veranda all winter.’ # O urso tem dormido na sua caverna o inverno inteiro. The bear has sleep.PPRT in-the his cave the winter whole ‘The bear has been sleeping in his cave all winter.’ (Cabredo Hofherr, Laca & de Carvalho 2010:76)

Markers can also underspecify the dimension of distributivity; in the Hausa example below temporal distribution or spatial distribution are equally possible interpretations, but lack of distribution crucially is not: (5)

ruwaa yaa zuzzuboo (Hausa) water 3SG.PF RED-pour ‘The water was coming/ pouring from different places (or interruptedly); crucially not in one stream.’ (Soucková & Buba 2008: 141)

As suggested by the examples in (3) and (4), markers indicating primarily or exclusively temporal distribution are the main area of overlap between the phenomenon of event plurality and the domain of aspect, in particular lexical aspect in the tradition of Vendler (1957) and much subsequent work, or derived situation aspect as identified by Smith (1991). However, as event plurality, and in particular pluractional markers, can also clearly impose patterns in the distribution over locations (see (2)) and over participants (section 2.2), it does not seem to be empirically adequate to conflate the whole array of verbal plurality with lexical aspect, contrary to the general approach taken in the major cross-linguistic surveys on verbal plurality by Dressler (1968), Cusic (1981) and Xrakovskij (1997).

5

Introduction

2.2 Distribution over participants Event plurality markers not only vary with respect to the requirements they impose on the distribution of the event plurality in space and time; they also allow different distributive dependencies between the plural arguments and the event plurality. Some markers require each individual member of a plural argument to participate in a necessarily plural event. Other pluractional markers do not impose this condition, permitting situations where each individual is only involved in a single event of the type described by the basic verb. In these cases the event plurality can be the distributive key with the plural argument as the distributive share in a dependency between the singular events and their participants.3 The pluractional marking in the Chechen example (6) below illustrates the latter case: The pluractional marking on the verb is possible even though the predicate only applies once to each individual in the subject argument in (6a/c) and the object argument in (6b). This difference appears particularly clearly in the availability of pluractional marking with once-only predicates as in (6b/c) (a bomb can only explode once, a fish is only caught once). (6)

Chechen pluractional marking: once-only predicates possible a.

b.

c.

ceera~duezalsh takhana duqa hxaalkhie ghittira their members.of.family today very early wake.up.PLR.WP ‘Their family members woke up very early today.’ (Yu 2003:297) takhana as duqq'a ch'eerii liicira today 1SG.ERG many=& fish.PL catch.PLR.WP ‘I caught a lot of fish today.’ (each fish is only caught once). (Yu 2003:297) Bombanash lilxira bomb.PL explode.PLR.WP ‘The bombs exploded.’ (Wood 2007:211, ex 17b)

As Yu (2003) points out, example (6a) does not have “the expected repeated event reading [...] [the sentence] means that all the family members woke up more or less around the same time” (Yu 2003:297), suggesting that Chechen does not impose temporal distribution in all cases. –––––––—–– 3

Choe (1987) proposes that distributive dependencies are relations between to a SORTING KEY and DISTRIBUTED SHARE: (i) Each child ate a sweet. sweets per child KEY

(ii)

SHARE

Bombanash bombs

lilxira explosions per bomb explode.PLR.WP (see 6c in the main text)

KEY

SHARE

6

Patricia Cabredo Hofherr & Brenda Laca

Squamish CVC-reduplication, in contrast, is an example of a pluractional marker that requires each member of the plural argument to participate in a plural event (Bar-el 2008:12). (7)

Squamish CVC-reduplication does not allow once-only predicates a.

na

kw'ech-kw'ach-nexw-as 7alhi slhanay' woman (i) ‘He's been watching her [the woman].’ (ii) ‘They have been watching her [the woman].’ (iii) */? ‘They each saw her once.’ (Bar-el 2008:12, ex 24) chet xwet-xwit-im 1S.PL RED-jump-INTR (i) ‘We are jumping.’ (ii) */? ‘We jumped.’ (Context: we each jumped once) (Bar-el 2008:12, ex 25)

RL

b.

RED-look.at-TRZ(LC)-3ERG DEM

While the Squamish CVC-reduplication shows that a temporally distributed event plurality has to apply to each participant, the  Hoan marker kí-VERBq||o provides an example of the same requirement for a spatially distributed event plurality. The interpretation of (8b) shows that spatial distribution of a plurality of events involving different individuals is not enough: the marker imposes the additional restriction that each member of the plural argument has to be involved in a spatially distributed plural event: (8)

a. b.

*Titi ikí-‘am-q||o ki ci m
un (Hoan) Titi PAST kí[pl]- eat-around PREP place one tsi i kí- ‘am-q||o 3PL PAST ki[pl]-eat-around ‘They ate around.’ (Cannot mean: Chris ate in one place, Titi ate in another place and Leha ate in a third place.) They are going around (separately or together) eating in different places. (Collins 2001:467, exs 31a & 32)

The markers of event plurality we have considered here differ in whether they allow distributive dependencies between the plural event and an argument of the main predicate. This plural argument is expressed by different plural DPs in the examples: a possessive plural DP (6a), a plural indefinite (6b), a bare plural (6c) and plural pronouns. To obtain a distributive dependency, however, a plural argument is not necessary; in contexts containing quantificational expressions such as (9) below, singular indefinites can give rise to a reading that distributes over a multiplicity of individuals (multiplication of singular indefinites):

7

Introduction (9)

a. b.

Each boy read a book. (different books for each boy) Mary often fixed a bicycle. (different bicycles on each occasion)

Strikingly, it has been observed that many pluractional markers allowing distributive dependencies with plural arguments do not allow multiplication of indefinite singulars -- and more generally of cardinalized indefinites -- in the same position. As pointed out by Yu (2003) for Chechen, Van Geenhoven (2004) for West Greenlandic, and Laca (2006) for Spanish, pluractional markers that allow distribution over bare plurals (see 10-12a) may bar multiplication of singular indefinites (10-12b). (10)

a.

b.

(11)

a.

b.

(12)

a.

b.

Qaartartut sivisuumik qaaqattaarput (West Greenlandic) qaartartu-t sivisuu-mik qaar-qattaar-put bomb.ABS-PL lengthy.INS explode-QATTAR-IND.[-TR]3PL ‘Bombs exploded again and again for a long time.’ #Qaartartoq sivisuumik qaaqattaarpoq qaartartuq sivisuu-mik qaar-qattaar-puq bomb.ABS lengthy.INS explode-QATTAR-IND.[-TR].3SG #’A/the bomb exploded again and again for a long time.’ (Van Geenhoven 2004: ex. 30-31) Bombanash lilxira bomb.PL explode.PLR.WP ‘The bombs exploded.’ #Bomba lilxira bomb.SG explode.PLR.WP ‘The bomb exploded again and again.’ (one bomb produces several explosions) (Wood 2007:211, ex 17b/c)

(Chechen)

El zorro anduvo matando gallinas. the fox walk.SP killing hens ‘The fox has been killing hens.’ ??El zorro anduvo matando una gallina. the fox walk.SP killing a hen ‘The fox has been killing a hen.’ (same hen) (Laca 2006, ex 20/21)

(Spanish)

As Van Geenhoven (2004) points out, scope-taking elements do not have this restriction: Q-adverbs and quantified arguments can multiply singular indefinites.

8

Patricia Cabredo Hofherr & Brenda Laca (13)

Jim hit a golf ball into the lake for an hour. i. There is a golf ball and Jim hit it into the lake repeatedly for an hour. ii. # For an hour, Jim hit each time another golf ball into the lake.

(14)

Jim hit a golf ball into the lake every five minutes for an hour. i. There is a golf ball and for an hour Jim hit it into the lake every five minutes. ii. For an hour, Jim hit every five minutes another ball into the lake. (Van Geenhoven 2005: exs 40/41)

This suggests that the distributive dependency observed in the examples (1012a) with bare plurals are not due to scopal interactions but rather comparable to distributive dependencies that obtain between two plural arguments. Laca (2006) further shows that verbal plurality markers can differ in the range of DPs that support cumulative distributive dependencies (see Landman 2000) with the event-plurality. While the Spanish verbal periphrases ir/andar+gerund allow distributive dependencies with definites, universally quantified DPs and coordinations, this is not the case for the Czech distributive prefix po- (discussed in Filip 1999): (15)

a.

b.

c.

(16)

a.

b.

c.

Definite plurals El zorro anduvo matando las gallinas. (Spanish) the fox walk.SP killing the hens ‘The fox has been killing the hens.’ (Laca 2006 ex.22a) Quantified NPs Juan anda llamando por teléfono a cada uno de sus amigos. Juan walk.PRS calling by phone to each one of his friends ‘Juan is phoning every one of his friends.’ (does not require more than one phone-call per friend) (Laca 2006 ex.23a) Coordinations Juan se va poniendo la camisa, la chaqueta y la corbata. Juan REFL go.PRS putting.on the shirt, the jacket, and the tie ‘Juan succesively puts on his shirt, his jacket, and his tie.’ (Laca 2006 ex. 23b) ??Po-zamykal zásuvku. (Czech) DISTR-lock.PAST.3.SG drawer.SG.ACC ?? ‘He locked a/the drawer [gradually/successively].’ Po-zamykal zásuvky. DISTR-lock.PAST.3.SG drawer.PL.ACC ‘He locked the drawers [successively, one after the other].’ Quantified NPs ??Po-zamykal kazdou zásuvku DISTR-lock.PAST.3.SG each/every drawer.SG.ACC not: ‘He locked each/every drawer gradually/successively.’ (Filip 1999, ch 5, exs. 63, 9b & 66)

9

Introduction

The behaviour of the Spanish periphrases also contrasts with Chechen pluractional verbs. In Chechen a coordination of singulars is not compatible with a pluractional verb (contrasting with (15c) above). (17)

takhana as c’ii-ch’aara’a jai-ch’aara’a miaq-ch’aara’a today 1SG.ERG sturgeon=& bullhead=& catfish=& leecira / *liicira (Chechen) catch.WP/ *catch.PLR.WP ‘Today I caught a sturgeon, a bullhead, and a catfish.’ (Yu 2003: 317 ex. 58)

Furthermore, Wood (2007:216) observes that demonstrative plurals and plural pronouns in Chechen do not allow distributive dependency readings in contexts where the indefinite plural does. Wood notes that when the absolutive refers to a bounded set of criminals the distinction between a distributed and a non-distributed reading disappears ((18a) vs (18b)), and the pluractional liicira can only mean that the same criminals were caught repeatedly:4 (18)

a.

b.

(19)

a.

b.

Sialxana milcuos tykan chohw duqqa zulamxoi yesterday police.officer.ERG store.GEN inside many criminal.PL leecira (Chechen) catch.WP ‘Yesterday the police officer caught a lot of criminals (together) in the store.’ Sialxana milcuos ghaalaw duqqa zulamxoi yesterday police.officer.ERG city.LOC many criminal.PL liicira catch.PRL.WP ‘Yesterday the police officer caught many criminals in the city (separately).’ Sialxana milcuos hara zulamxoi leecira (Chechen) yesterday police.officer.ERG DEM criminal.PL catch.WP ‘Yesterday the police officer caught these criminals (together or separately).’ As ysh sialxana liicira 1SG.ERG 3PL.ABS yesterday catch.PRL.WP ‘I caught them again and again yesterday.’ (Wood 2007:216, ex.24a/b)

–––––––—–– 4

Wood (2007:216) notes that “the different location expressions in the two sentences were provided spontaneously by a native speaker in order to make the relevant (simulfactive or pluractional) verb form sound natural and appropriate.”

10

Patricia Cabredo Hofherr & Brenda Laca

This again contrasts with the Spanish periphrasis ir+gerund that allows distributive dependencies with plural pronouns: (20)

El bibliotecario recibe los libros, los etiqueta y the librarian receives the books, 3PL.ACC labels and los va poniendo en los anaqueles. (Spanish) 3PL.ACC go.PRS3SG putting on the shelves. ‘The librarian receives the books, labels them and gradually puts them on the shelves.’

The examples discussed here show that the availability of distributive dependencies with pluractional verbs depends on the syntactic type of DP. For all the markers we reviewed, bare plurals allow the dependency, while indefinite singulars block it. For other types of DP, we have seen that there is crosslinguistic variation. Coordinations of singulars and plural pronouns allow distributive dependencies with the pluractional periphrases ir/andar+ gerund in Spanish, but not with the pluractional verb in Chechen. Quantified DPs can enter a distributive dependency with ir/andar+ gerund in Spanish, but not with the po-prefix in Czech.

2.3 Adnominal distributive markers and distributive adverbs Unlike the pluractional markers discussed in sections 2.1 and 2.2, adnominal distributive markers such as each (marking the distributive key) and adnominal distributive expressions such as reduplicated numerals (marking the distributive share) generally allow multiplication of cardinalized indefinite arguments. (21)

a. b.

Each boy read a book. (a different book for each boy) Romanma da Zurabma sam-sami Roman.ERG and Zurab.ERG DISTR-three.ABS anta caio. (Georgian) suitcase. ABS carry.PST.SG (i) ‘Roman and Zurab carried three suitcases each/ a piece.’ (ii) ‘Roman and Zurab carried the suitcases three by three.’ (Gil 1988)

The fact that multiplication of cardinalized indefinites is possible suggests that these types of distributivity markers -- unlike the pluractional markers discussed before -- are scope-taking expressions. The question then arises, how the distributivity of scope-taking expressions should be analysed. An influential proposal (originating in work by

Introduction

11

Schein 1993) holds that distributivity is uniformly over events. Under this view distributivity over individuals arises from distributivity over events when each event involves a single individual. This approach accounts well for some markers, such as the reduplicated numerals in Karitiana studied by Müller & Negrão (this volume); as these authors show, Karitiana reduplicated numerals allow readings distributing over individuals as well as over events while consistently behaving as adverbs syntactically. However, as Pereltsvaig (this volume) points out, there are empirical arguments against generalizing a uniform analysis of distributivity over individuals and distributivity over events. Pereltsvaig shows that many languages have two distinct markers for distributivity over individuals and distributivity over events. Furthermore, the marker for eventdistributivity tends to be morphologically more complex than the marker for distributivity over individuals. Both observations are unexpected if eventdistributivity is taken to be the basic case. A further case of distributivity is found with distributive readings arising from lexical cumulativity. It is generally assumed that the denotation of plural nouns is cumulative (see Link 1983 and much work since); it has further been proposed that basic verb denotations are also cumulative, including singular and plural events (Krifka 1992, Landman 1996, Kratzer 2008). Müller and Negrão (this volume) specifically compare distributive readings arising from lexical cumulativity of noun- and verb-denotations and distributivity contributed by reduplicated numerals in Karitiana. They show that while sentences without a reduplicated numeral allow distributive interpretations -- as well as collective and cumulative interpretations -- the reduplicated numerals force a distributed reading distributing over the object. Among distributive readings at least three cases have to be distinguished: (i) distributivity induced by scope-taking elements such as each, (ii) distributivity requirements that are part of the meaning of markers of event plurality (as discussed above) and (iii) distributive readings that arise as a result of cumulative denotations. 2.4 Summary In the preceding sections we have discussed examples of distributivity requirements imposed by a marker of event plurality. While there are examples that show a clear effect of distributivity over time, space and participants to the event, it is frequently the case that not all examples in a given language display distributivity along the same dimension. For Chechen pluractional marking, e.g., only a lexically specified subset of verbs allows a distributive reading (Yu 2003:315). As pointed out by

12

Patricia Cabredo Hofherr & Brenda Laca

Wood (2007) while some verbs have distributive readings (see (6) above) other verbs such as ai'ira “to lift” that would have an equally plausible distributive reading do not allow it: (22)

a.

b.

As jashchik hwala- ai’ira / ii’ira (Chechen) 1SG.ERG box uplift.WP / lift.PRL.WP ‘I lifted the box once /repeatedly.’ As duqa jashchikash hwala- ai’ira / ii’ira 1SG.ERG many box.PL uplift.WP / lift.PRL.WP ‘I lifted many boxes once / repeatedly (either together or separately).’ (Wood 2007, 213, ex 19a/b)

There have been proposals to link the variation in the readings of pluractional markers with the Aktionsart of the underlying predicate (e.g. HajiAbdolhosseini, Massam & Oda 2002 for Niuean, Yu 2003 for Chechen) but the coverage of these proposals remains partial, suggesting that additional factors have to be taken into account in future studies. Based on the observation that pluractional markers express continuative, frequentative, durative and intensive readings many studies have viewed verbal plurality as a special case of lexical aspect (Dressler 1968, Cusic 1981, Xrakovskij 1997). As we have seen in this section, however, pluractional markers can carry at least two types of additional semantic restrictions that go beyond temporal properties specified by the lexical predicate. First, markers of event plurality can force the distribution of the event-plurality over space and over participants. Secondly, the distributive dependencies available between participants and events can be restricted to plural arguments of particular syntactic types. We therefore conclude that event plurality and more specifically verbal plurality include reference to distributivity in space and over participants and cannot be reduced to the same primitives as lexical aspect if the latter is understood as essentially specifying the temporal contour of the basic eventuality.

3. Mixed event plurality: additive operators While the examples discussed up until now contained markers asserting a plurality of events, there are other markers that impose a mixed event plurality in the sense that the event-plurality is made up of asserted and presupposed events (see Tovena & Donazzan 2008 for discussion). Examples of the latter type of event plurality are additive adverbs such as again and its equivalents, repetitive affixes such as the prefix re- in English

13

Introduction

and Romance (Fr. relire “reread”) or Cuzco Quechua -pa (Faller this volume) and additive nominal expressions such as English some more N and Modern Hebrew od (Greenberg this volume). The study of additive markers opens a series of very relevant questions concerning the identity conditions between asserted and presupposed eventtypes. As shown by Tovena & Donazzan (2008) and Greenberg (this volume) additive markers vary with respect to the events-types that count as possible presupposed events. As Tovena & Donazzan (2008) show, for the Mandarin Chinese additive adverb zai the subject has to be shared by the asserted and the presupposed event parts, while this is not the case for French encore or Italian ancora or English again, with ancora and encore allowing a "choice between keeping the subject, the direct object and/or other parts". (23)

a.

b. c.

#Zhangsan/ wo gangcai qu kan le yi xia, (Mandarin) Zhangsan/ I just go watch ASP a little guo yi huir wo hui zai qu. pass a moment 1SG MOD ZAI go # ‘Zhangsan/ I just went to take a look, I will go again later.’ Marie a mangé trois kiwis et puis encore autre chose. (French) ‘Mary ate three kiwis and then some more stuff.’ Maria ha appena controllato che il bambino dormisse. (Italian) Controllo ancora fra cinque minuti e poi andiamo. ‘Mary just checked that the baby is asleep. I’ll check again in five minutes and then we can go.’ (Tovena & Donazzan 2008, exs 35, 36a/b)

Nominal reference further constrains the computation of the presupposed event: referential NPs and rigid designators, for example, force the identity of arguments between the asserted and the presupposed events (Tovena & Donazzan 2008). The following example shows that verbs of consumption are impossible with a referential NP, since a verb of consumption cannot apply to the same referent twice. (24)

Zhangsan hui zai xizao /#chi zhe ge pingguo. (Mandarin) Zhangsan MOD ZAI wash/ eat this CL apple ‘Zhangsan will rinse/ #eat this apple again.’ (Tovena & Donazzan 2008, ex. 37)

The nominal additive particle od in Modern Hebrew examined by Greenberg (this volume) imposes a much weaker identity condition: subjects, times, locations and even the verbal predicate can differ between the asserted and the presupposed event. Greenberg shows that the variation of the predicate is restricted by a superset condition requiring the two predicates to be instantiati-

14

Patricia Cabredo Hofherr & Brenda Laca

ons of a more abstract predicate. The example in (25a) is possible in the given context since “acquiring sheep” is available as a superset predicate, while (25b) is not felicitous since no plausible superset predicate can be constructed: (25)

a.

b.

kibalti 30 kvasim mi-dod eli. I-got 30 sheep from-uncle mine ‘I got 30 sheep from my uncle.’ be-avu'a ha-ba ekne od 10 in-week the-next I-will-buy OD 10 ‘Next week I will buy 10 more.’ # be-avu'a ha-ba emkor od 10 in-week the-next I-will-sell OD 10 ‘Next week I will sell 10 more.’ (Greenberg this volume, exs 20 and 20')

(Modern Hebrew)

As a mixed plurality is made up of two distinct sets of events, the markers of mixed event plurality can carry additional information concerning the relationship between asserted and presupposed events. Additive markers like again, for example, specify a temporal ordering between the presupposed event(s) and the asserted event(s), as the following example due to Kripke shows (cited by Kamp and Roßdeutscher 1994): (26)

a. b.

We will have pizza on Mary’s birthday. So we should not have pizza on John’s birthday too. We will have pizza on Mary’s birthday. So we should not have pizza again on John’s birthday

As Kripke points out, the example (26b) differs from (26a) since using again as opposed to too supports the inference that John's birthday takes place after Mary's birthday.

4. Exact cardinality and number in the verbal domain Following current usage, we have referred to event plurality expressed by pluractional markers on verbs as verbal plurality. Since the late 1980s structural parallels between the nominal and the verbal domain have been emphasised (Abney 1987, Grimshaw 1991 and a large literature since). This general approach has also been applied in many studies treating verbal plurality as an instance of a more abstract category NUMBER in the verbal domain. One of the fundamental questions is therefore whether verbal plurality should be understood as the verbal counterpart to nominal plurality. To begin

15

Introduction

with, we will review the restrictions observed on exact cardinality specifically for pluractional markings. These restrictions are unexpected if verbal plurality is the counterpart of number in the nominal domain, and we will therefore consider the cardinality restriction in the wider context of differences between nominal and verbal plurality.

4.1 Restrictions on exact cardinality As pointed out by Corbett (2000:250) verbal plural markers differ from nominal plural markers in that they often do not express a plurality corresponding to more than one but rather a plurality corresponding to several. These expressions of event plurality would not be applied to a plurality containing only two or three events; this restriction can be termed the many-ness requirement. There are, however, classes of markers that do not display this limitation. As stressed by Faller (this volume) the repetitive marker -pa in Cuzco Quechua can be applied to an event plurality containing just two events; this applies to additive markers like again more generally. Another class of markers that seems to be exempted from the many-ness requirement are collective plural markers (Mithun 1999:92). Mithun characterises collective and distributive plural markers in North American languages as follows: (27)

a.

b.

Collective plural markers: Sets of events viewed collectively are typically contiguous in space and time, often implying the spatial proximity of their participants. The participants are typically treated as a unit as well, often with the implication that agents cooperate in concerted action, or that patients are affected or manipulated together as a set. Distributive plural markers: with distributives members [of a plural participant, PCH&BL] are presented as distinct individuals, separated in space, type or time. Mithun (1999: 92)

Mithun notes that distributive markers would not be used to describe an action affecting only two elements while collective plural markers can appear in this context (Mithun 1999:93). If we examine the many-ness requirement more closely, it becomes apparent that it conflates three distinct requirements: (i) vague amount as expressed by degree expressions (a lot), (ii) vague cardinality (many times) and (iii) relatively large cardinality (ten times).

16

Patricia Cabredo Hofherr & Brenda Laca

Compatibility with expressions of exact cardinality provides a test bearing more narrowly on the last of these properties, since exact cardinality expressions can refer to relatively large cardinalities. It has been observed that many markers of event plurality do not combine with expressions of exact cardinality: (28)

a.

b.

??mutàanee àshìrin sun firfitoo (Hausa) people twenty 3PL.PF RED-come.out (Souckova & Buba 2008:135) adama takhan yttaza chai melira / *miillira (Chechen) Adam.ERG today ten times tea drink.WP/ drink.PLR.WP ‘Adam drank tea ten times today.’ (Yu 2003: 303, exs 27a/b, pluractional V excluded)

In Chechen, the same effect arises with a coordination of singulars. However, if one of the NPs in the NP-conjunction is plural, the pluractional verb is clearly preferred. (29)

a.

b.

xyyrana johanna’a elita’a so’a niaxar ullie morning.ADV Johanna=& Elita=& 1SG=& door next to dxa-hwettira /*dxa-hittira DX-stand.wp/*DX-stand.PLR.WP ‘Johanna, Elita, and I stood by the door in the morning.’ (Yu 2003: 316, ex. 56) xyyrana beerash’a elita’a so’a niaxar ullie morning. ADV children.pl=&Elita=& 1SG=& door next to *dxa-hwettira / dxa-hittira *DX-stand.WP/DX-stand.PLR.WP ‘The children, Elita, and I stood by the door in the morning.’ (Yu 2003: 316, ex. 56)

The examples discussed here involve different expressions of exact cardinality: an explicit expression of exact cardinality contained in a noun phrase (28a) or an adverbial expression (28b), and a coordination of singulars (29a), where exact cardinality is an accidental byproduct of the construction. It is an open question whether there is a single restriction on cardinality specification covering coordinated singulars, nouns modified by numerals and cardinality adverbs, or whether these three restrictions can be dissociated for particular markers.

4.2. Plurality in the nominal and verbal domain We have seen that many markers of event plurality mark a plurality that corresponds to several/ many events. This contrasts with nominal plurality that

17

Introduction

can generally be employed when two or more individuals are involved. This difference between plurality in the nominal and in the verbal domains is also reflected in the distribution of counting expressions in the nominal and in the verbal domain. As Doetjes (2008) points out, exact cardinal modification cannot combine directly with a VP in English, French and Dutch where the insertion of a classifying expression like Engl. times, Fr. fois or Du. keer is needed. (30)

a. b. a’. b’. a’’. b’’.

John went to the movies three times. John slept three times. Jean est allé trois fois au cinéma. (French) Jean a dormi trois fois. Jan is drie keer naar de bioscoop gegaan. (Dutch) Jan heeft drie keer geslapen. (Doetjes 2008:151, ex 27)

Doetjes shows that the need for a classifier cannot be reduced to the lack of number in the verbal domain in English, French and Dutch, since the same effect is found in Hausa which has pluractional verbs, and therefore should qualify as a language with number in the verbal domain. In Hausa, two alternative strategies can be used for exact cardinal event modification: either insertion of a noun sàu “time” or a cognate object structure with cardinal modification of the object: (31)

a.

b.

Sun ci jarràbâawaa sàu ukù (Hausa) 3PL eat exams time three ‘They passed exams three times.’ (Doetjes 2008:153, ex 28b) Taa zàagee shì zaagìi ukù she insult him insulting three ‘She insulted him three times (lit. three insultings).’ (cognate object) (adapted from Newman 2000:89-91)

If plurality in the verbal domain were comparable to nominal plurality, classifying expressions like times/ fois/ keer would be expected behave on a par with nominal classifiers. As Doetjes (2008) argues, however, this is not the case. Doetjes shows that nominal classifiers combine with nouns that have cumulative denotation, excluding singular nouns. This restriction does not apply in the verbal domain with times/ fois/ keer: these expressions can combine with VPs that are the equivalent of singular nouns in that they only have singular events in their denotation such as acheter deux kilos d'olives.5 –––––––—–– 5

Doetjes (2007) shows in detail that singular VPs are not compatible with the degree adverb beaucoup since beaucoup is restricted to plural VPs or mass VPs:

18

Patricia Cabredo Hofherr & Brenda Laca (32)

Elle a acheté trois fois deux kilos d'olives. she has bought three times two kilos of olives ‘She bought two kilos of olives three times.’

Doetjes (2008:155) further notes that the reading of (32) corresponds to the distributive reading of a plural noun phrase in that the cardinality expression multiplies the cardinalized indefinite (yielding three times two kilos of olives in the context where this example is true). This ties in with another difference in scope of the numeral: while the numeral combining with times licenses the internal reading of different, the numeral combining with a nominal classifiers does not: (33)

a.

b.

Three times he gave a different answer. => three different answers (adapted from Doetjes 2008: 155, ex 32a) three pieces of different furniture not three different pieces of furniture

Doetjes concludes that while vague cardinality modification by degree expressions such as Fr. beaucoup “a lot” is indeed similar in the nominal and verbal domains, exact cardinality modification is essentially a nominal property. Fassi Fehri (this volume) argues specifically for Arabic that nominal plurality and verbal plurality should not be assimilated. He shows that nominal plurality can be interpreted in different positions throughout the DP, while verbal plurality is only interpreted in the vP if pluractional marking on the verb is used (i.e. if the lexical head is modified by a derivational process); in all other cases the verbal plurality is interpreted on the nominal dependents.

5. The papers in this volume The papers in the present volume examine morphological markers of event plurality in detail. Apart from the contribution by Paul Newman, which is de–––––––—–– (i) a. Elle she b. Elle she c. #Elle she

a has l'a him-has a has

beaucoup dormi. (mass VP) a.lot slept beaucoup rencontré. a.lot met (count VP, no number) beaucoup acheté trois kilos d'olives. a.lot bought three kilos of-olives (count singular)

Introduction

19

voted to the study of cross-linguistic morphological variation found with pluractional markers, the contributions examine the semantic properties of specific markers of event plurality. These studies focus on different aspects of the semantics of event plurality as set out above. Patrícia Amaral and Chad Howe examine the diachronic development of the European Portuguese Pretérito Perfeito Composto, formed by the present of ter “to have” combined with a non-agreeing past participle. As is well known, this construction does not have the full range of interpretations found with present perfects cross-linguistically: a resultative interpretation is impossible and an iterative interpretation reminiscent of the English Present Perfect Progressive is forced. As Amaral and Howe show, earlier stages of Portuguese such as 15th and 16th century Portuguese did not have this restriction, allowing resultative and recent past interpretations with no implication of event plurality. The authors argue that the interpretation of the perfect including obligatory event plurality arises as a consequence of semantic ambiguity under certain morphosyntactic conditions. The resultative construction has an agreeing past participle, while the Perfect construction has a nonagreeing participle: when the NP-complement is masculine singular, however, the structure is ambiguous between an agreeing resultative and nonagreeing perfect reading. Amaral and Howe propose that a trigger for semantic change is to be found in structurally ambiguous contexts with a semantically plural complement: in these structures the distributivity over the complement is reinterpreted as distributivity over events, transferring nominal plurality to the verbal domain. These cases of ambiguity are then the onset contexts that eventually allow the semantic enrichment of the Present Perfect to be conventionalized (following Traugott & Dasher 2001 and Eckhard 2006). Martina Faller offers a detailed overview of markers of pluractionality in Cuzco Quechua. She shows that an analysis adopting the semantics given in Lasersohn (1995) accounts for the semantic properties of the pluractional markers in their temporal and spatial uses. Beyond these two basic meanings two of the pluractional markers in Cuzco Quechua, namely the frequentative morpheme -paya and the reciprocal -puna can be used to convey meanings of “social accompaniment”. Faller proposes to account for these readings compositionally based on the semantics of pluractionals and additional devices that introduce a beneficiary. The two accompaniment constructions differ from each other with respect to the mechanism introducing the beneficiary. For reciprocal -paya the beneficiary is introduced by the reflexive -ku and identified with the plural subject, in this construction accompaniment is symmetrical between the individuals making up the subject. For frequentative -paya, in contrast the accompaniment is asymmetrical, with the beneficiary generally being expressed as an accusative NP. Strikingly, the accompaniment reading of frequentative -paya introduces a presupposition that the

20

Patricia Cabredo Hofherr & Brenda Laca

beneficiary is engaged in his or her own event; as Faller points out, this presupposition is absent from purely frequentative readings of -paya. In her contribution, Yael Greenberg provides a semantic analysis for the additive expression od in Modern Hebrew and its English counterpart (some) more. Greenberg argues that despite the fact that both markers combine syntactically with a nominal expression, they both express additivity in the domain of eventualities. According to Greenberg's analysis, od and (some) more trigger the presupposition that there is an existing eventuality that the asserted eventuality is added to yielding a larger eventuality. This analysis is formalized using a generalized version of the stage-of-relation proposed by Landman (1992). The analysis of od and (some) more in terms of a more developed summed eventuality accounts for the fact that the presupposed event need not share the same predicate as the asserted event precisely in cases where both events describe sub-types of a bigger eventuality. In a second step Greenberg shows that the proposed analysis can be extended to postpredicate od/ (some) more. She observes that post-predicate and nominal additives differ with respect to the dimension of increase: while nominal additives increase the number of participants, post-predicate additives increase the spatio-temporal properties such as the running time or the spatial path of the eventuality. Comparing od/ some more to the additive particles ancora (It.) encore (Fr.) and zài (Mandarin), examined in Tovena & Donazzan (2008), Greenberg stresses that unlike these expressions od and (some) more do not allow repetitive readings with punctual predicates such as meet John; in these contexts English and Modern Hebrew have to use the modified expressions od pa'am and once more or lexical alternatives such as again. Abdelkader Fassi Fehri's contribution compares the marking of plurality in the verbal and nominal domain in Arabic, arguing that plurality does not project in the same fashion in both domains. In a detailed study of the morphology and syntax of the DP, he shows that nominal number can be interpreted on the root and on several other functional heads in the DP. In particular, Fassi Fehri argues that collective agreement (i.e. the agreement taking the form of the 3rd person feminine singular) and plural agreement are of different types and exhibit different properties; e.g. the former occurs in both VSO and SVO orders, whereas the latter is limited to SVO. He then proposes to analyse the plural in syntactic collectives as a collective classifier, not as an instance of number. Moreover, as there are DP coordinations that can combine with collective agreement this classifier projection has to be located high in the DP. Verbal plurality can only be interpreted on the verbal projection if pluractional or reciprocal morphology directly modifying the verb root is used, in contrast with nominal plurality that is distributed over various projections in the DP. Arabic has various instances of pluractional morphology on the verb root expressed by consonant gemination, partial reduplication or vowel lengthening. Verbal plurality contributed by these markers can be in-

Introduction

21

terpreted on the verb itself, yielding an “intensive” reading, or on a plural argument, yielding a “plural of abundance” interpretation for this argument. Unlike these instances of internal plurality on the verb-root, phrasal plurality of the vP/TP is not interpreted on the vP/TP itself but on its nominal dependents. Fassi Fehri analyses the plural agreement found on the verb with SVO orders as phrasal plurality on the vP/TP and shows that contrary to nominal plurality that can be interpreted at different levels throughout the DP there are no clear cases where phrasal plurality in the verbal domain is interpreted internally to the vP/TP. This phrasal plurality behaves as a purely formal agreement marker, since phrasal plurality on verbs is generally interpreted on the nominal dependents of the verb. Particular attention is given to the expression of reciprocity as a particular case of verbal plurality. Fassi Fehri points out that syntactic reciprocals differ from morphological reciprocals with respect to the scope of frequency adverbs like talaat-a marraat-in “three times”: while the syntactic reciprocal allows three events per participant, morphological reciprocals only allow readings involving exactly three events. Ana Müller and Esmeralda Negrão explore two types of distributivity in Karitiana, contrasting distributivity arising from cumulative denotations of nouns and verbs with distributivity induced by reduplicated numerals. As they show cumulative denotations of nouns and verbs yield distributive readings along with cumulative and collective readings for simple sentences. Once a reduplicated numeral is added, however, only distributive readings over individuals or over occasions are possible; other situations that are possible instantiations of cumulative or collective events of the simple sentences are not in the denotation of the sentence with the reduplicated numeral. Müller and Negrão stress that reduplicated numerals allow two types of distributive readings, namely distribution over individuals and distribution over occasions. However, they show that syntactically, reduplicated numerals behave uniformly as adverbial expressions, providing empirical evidence that reduplicated numerals in Karitiana should not be analysed as ambiguous between a determiner and an adverbial quantifier. Müller and Negrão analyse reduplicated numerals as adverbial operators over a relation between entities and events denoted by VPs or sentences. According to their analysis reduplicated numerals pluralize the events -- removing singular events from the basic denotation of the VP -- and impose a cardinality restriction on the entities. Müller and Negrão attribute the fact that both types of distributive interpretation are available to the vagueness involved in event individuation, which can proceed on the basis of individuals or occasions and possibly other criteria. In his contribution, Paul Newman draws together the generalizations on the morphology of pluractionals emerging from a wide range of studies. It has been observed that pluractional marking often takes the form of stem alternations with derivational rather than inflectional properties, with many languages using full or partial reduplication to mark pluractionals. A clear

22

Patricia Cabredo Hofherr & Brenda Laca

difference between pluractional marking and agreement marking with a plural argument is provided by the characteristically ergative alignment of pluractional marking: even in nominative-accusative languages the pluractional marking targets subjects of intransitives and objects of transitives. Newman further points out that languages vary widely with respect to productivity of pluractional marking: while in some languages a significant proportion of the verbal lexicon has a pluractional counterpart, in others only a handful of forms are attested. Beyond these questions regarding productivity of the pluractional morphology, pluractional forms give rise to three well-known mismatches in morphology cross-linguistically: (i) suppletive forms seem to be disproportionately frequent for pluractional forms, (ii) languages with more than one process of pluractional formation both can be combined in doubly marked pluractionals (Newman's hyperpluractionals), and (iii) languages may have formally pluractional verbs that do not have a simple nonpluractional counterpart (Newman's frozen pluractionals). The volume closes with Asya Pereltsvaig's contribution examining the opposition between distribution over individuals and distribution over events. Pereltsvaig argues against recent proposals in the literature that take distribution to be uniformly over events and analyse distribution over individuals as derived from distribution over events (following Schein 1993). She presents two empirical arguments against this view. The first argument is based on the morphology of distributive expressions cross-linguistically. Pereltsvaig shows that many languages have distinct markers for distribution over events and distribution over individuals. Moreover, marking for distribution over events is often morphologically derived from the distributive marker for distribution over individuals. As Pereltsvaig points out, both observations are unexpected if languages treat distribution over individuals as a special case of distribution over events. The second argument against an analysis of distribution over individuals in terms of distribution over events is based on the interpretation of the distributive marker -nibud and the distributive po construction in Russian. Pereltsvaig shows that with both markers distributivity over events is excluded with a perfective verb since perfective verbs in Russian cannot denote iterated events but only a single event. If distribution over individuals did depend on distribution over events, it would be expected that distributive markers are ungrammatical with perfective verbs. Contrary to this expectation, however, in these contexts with perfective verbs distributivity over participants is still possible given an appropriate DP and ungrammaticality only ensues if the DP argument is itself of a type barring distribution (such as a DP introduced by a collective numeral).

Introduction

23

References Amaral, Patrícia & Chad Howe (this volume). “Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect.” Bar-el, Leora (2008). “Verbal number and aspect in Skwxwú7mesh.” Recherches Linguistiques de Vincennes 37:31–54. Cabredo Hofherr, Patricia, Brenda Laca, & Sandra de Carvalho (2010). “When perfect means plural: The Present Perfect in Northeastern Brazilian Portuguese.” In P. Cabredo Hofherr & B. Laca, (eds.): Layers of Aspect, 67–100. Stanford, CA: CSLI Publications. Choe, Jae-Woong (1987). Anti-quantifiers and a Theory of Distributivity. Ph.D., UMass/Amherst. Collins, Chris (2001). “Aspects of plurality in Hoan.” Language 77:456–476. Corbett, Greville G. (2000). Number. Oxford: Cambridge University Press. Cusic, David Dowell (1981). Verbal Plurality and Aspect. Ph.D. Stanford. Doetjes, Jenny (2007). “Adverbs of quantification: degrees vs frequency.” Lingua 117:685–720. -------. (2008). “Counting and degree modification.” Recherches Linguistiques de Vincennes 37:139–160. Dressler, Wolfgang (1968). Studien zur verbalen Pluralität. Österreichische Akademie der Wissenschaften. Phil.-hist. Klasse. Sitzungsberichte. Bd. 259. Abh. 1. Wien: Bühlau in Kommission. Eckardt, Regine (2006). Meaning Change in Grammaticalization. Oxford: Oxford University Press. Faller, Martina (this volume). “Pluractionality and accompaniment in Cuzco Quechua.” Fassi Fehri, Abdelkader (this volume). “How plural can verbs be?” Garrett, Andrew (2001). “Reduplication and infixation in Yurok: morphology, semantics, and diachrony.” IJAL 67(3):264–312. Gil, David (1988). “Georgian reduplication and the domain of distributivity.” Linguistics 26:1039–1065. Greenberg, Yael (this volume). “Event-based additivity in English and Modern Hebrew.” Haji-Abdolhosseini, Mohammad Haji, Diane Massam, & Kenji Oda (2002). “Number and events: Verbal reduplication in Niuean.” Oceanic Linguistics 41:475–492. Kamp, Hans & Roßdeutscher, Antje (1994). “Remarks on Lexical Structure and DRS Construction.” Theoretical Linguistics 20:97–164. Laca, Brenda (2006). “Indefinites, quantifiers, and pluractionals. What scope effects tell us about event pluralities.” In S. Vogeleer and L. Tasmowski, (eds.): Nondefiniteness and Plurality, 191–217. Amsterdam/ Philadelphia: John Benjamins. Landman, Fred (1992). “The Progressive.” Natural Language Semantics 1:1–32. ------- (2000). Events and Plurality. The Jerusalem Lectures. Dordrecht: Kluwer. Lasersohn, Peter (1995). Plurality, conjunction and events. Dordrecht: Kluwer. Müller, Ana & Esmeralda Negrão (this volume). “On distributivity in Karitiana.” Newman, Paul (1981). The Classification of Chadic within Afroasiatic. Leiden: Universitaire Pers.

24

Patricia Cabredo Hofherr & Brenda Laca

------- (1990). Nominal and verbal plurality in Chadic. Dordrecht: Foris. ------- (2000). The Hausa Language: An Encyclopedic Reference Grammar. New Haven: Yale University Press. ------- (this volume). “Pluractional verbs: An overview”. Pereltsvaig, Asya (this volume). “Distributivity is not uniformly over events”. Rose, Françoise (2008). “Action répétitive et action répétée: aspect et pluralité verbale dans la réduplication en émérillon.” Faits de Langues 28:125–143. Schein, Barry (1993) Plurals and Events. Cambridge: MIT Press. Souková, Kateina & Malami Buba (2008). “Intensive plurality: Hausa pluractional verbs and degree semantics.” In Linguistics in the Netherlands 2008, 133–144. Amsterdam/ Philadelphia: John Benjamins. Tovena, Lucia and Marta Donazzan (2008). “On ways of repeating.” Recherches Linguistiques de Vincennes 37:85–112. Traugott, Elizabeth C. & Richard C. Dasher (2001). Regularity in Semantic Change. Cambridge: Cambridge University Press. Van Geenhoven, Veerle (2004). “For-adverbials, frequentative aspect, and pluractionality.” Natural Language Semantics 12:135–190. ------- (2005). “Atelicity, pluractionality, and adverbial quantification.” In H. Verkuyl, H. De Swart, and A. Van Hout, (eds.): Perspectives on Aspect, 107–125. Berlin: Springer. Wood, Esther (2007). The semantic typology of pluractionality. Ph.D., Berkeley. Xrakovskij, Viktor S. (1997a). “Semantic types of the plurality of situations and their natural classification.” In V. S. Xrakovskij, (ed.): Typology of iterative constructions, 3–68. München: LINCOM Europa. ------- (ed.) (1997b). Typology of iterative constructions. LINCOM Europa. Yu, Alan C. (2003). “Pluractionality in Chechen.” Natural Language Semantics 11:289–321.

Patrícia Amaral & Chad Howe

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect*

1. Introduction: Iteration and the Portuguese Perfect In contemporary Portuguese, the Pretérito Perfeito Composto (henceforth PPC), formed by the Present of ter ‘to have’ + Past Participle, differs from other Romance Perfects in that it does not display the full range of interpretations that are attested for present perfects cross-linguistically (Campos 1986, Giorgi & Pianesi 1997, Schmitt 2001). In main clauses with the indicative, the PPC does not have a resultative interpretation1 and must receive an iterative interpretation, as exemplified in (1):2 (1)

A Ana tem chegado the Ana have. 3SG arrive.PPART ‘Ana has been arriving late (*once).’

atrasada (*uma vez). late ( one time)

The PPC in (1) is interpreted as denoting a repetition of events of arriving late with non-overlapping run times; this requirement of event iteration can be tested since co-occurrence with the adverbial uma vez ‘once’ is ruled out.3 –––––––—–– *

1

2

3

We would like to thank three anonymous reviewers, the audiences at the Workshop on Nominal and Verbal Plurality 2008 in Paris and the American Association for Corpus Linguistics Conference as well as Brenda Laca, Elizabeth Traugott, and Alexander Williams for their helpful comments and feedback on this research. All errors remain our own. In this paper, the expression “contemporary Portuguese” refers to European Portuguese. Throughout this paper, we will use RESULTATIVE INTERPRETATION to refer to the interpretation in which a Present Perfect indicates the existence of a resultant state. We will discuss “durative” readings of the PPC in Portuguese as well: this type of interpretation is also referred to as the PERFECT OF PERSISTENT SITUATION (Comrie 1976), CONTINUATIVE, or UNIVERSAL PERFECT (see Nishiyama and Koenig 2004 for a review of terminology related to readings of the Perfect). Abbreviations used in the glosses are as follows: 3PL = Third Person Plural (Present), 3SG = Third Person Singular (Present), ACC = Accusative, F = Feminine, M = Masculine, FUT = Future, INF = Infinitive, PST = Perfective Past Tense (Pretérito Perfeito Simples), PPART = Past Participle. In subordinate clauses in the subjunctive, the PPC does not receive an iterative interpretation, as shown by (i):

26

Patrícia Amaral & Chad Howe

The unavailability of the resultative interpretation is shown in (2): the PPC cannot be used to refer to a present state resulting from a recent event, hence the unacceptability of (2c). In Portuguese, the resultant state/ recency interpretation required as an appropriate response to the question in (2) would necessitate the use of the aspectual periphrasis acabar de ‘(lit.) to finish’ + Infinitive, as in (2a), or of the simple perfective past with a temporal adverbial, as in (2b). (2)

Onde está a Ana? where be. 3SG the Ana ‘Where is Ana?’ a. Está aqui: acabou de chegar. be. 3SG here finish.PST of arrive.INF ‘She’s here: she has just arrived.’ b. Está aqui: chegou agora. be. 3SG here arrive.PST now ‘She’s here: she has arrived just now.’ c. ??Está aqui: tem chegado. be. 3SG here have. 3SG arrive.PPART ?? ‘She’s here: she has been arriving.’

Another test for the iterative interpretation is the incompatibility of the PPC with “once-only” events, as with to be born and to die in (3), adapted from Campos (1986): (3)

*O animal tem nascido (morrido). the animal have. 3SG be_born.PPART (die.PPART) ‘??The animal has been born repeatedly (died repeatedly).’

The same predicates are acceptable with the PPC with a plural subject, since the plurality of being-born or dying events now distributes over each of the plural participants:4 –––––––—–– (i)

4

É possível que a Ana tenha chegado atrasada uma vez. ‘It is possible that Ana has arrived late once.’ When reference time does not coincide with utterance time, the iterative reading does not arise (Mateus et al. 2003), as in (ii) and (iii), from Mateus et al. (2003: 142-143), examples (54) and (55): (ii) Quando a Ana chegar a casa da Maria, já o Rui a tem visitado. ‘When Ana arrives at Maria’s home, Rui will have already visited her.’ (iii) Sempre que a Ana chega a casa da Maria, já o Rui a tem visitado. ‘Everytime Ana arrives at Maria’s home, Rui has already visited her.’ Plural subjects compatible with the PPC are bare plurals and existentially and universally quantified NPs, but not cardinalized nominal expressions, e.g.

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect (4)

27

Muitos animais têm nascido nesta rua. Many animals have.3PL be_born.PPART on-this street ‘There have been many events of animals being born on this street.’

The requirements on event iteration are stricter than mere event plurality: the PPC requires a repetition of events that are regularly distributed over a time interval and the plurality of events cannot be cardinalized, as in (5): (5)

A Ana tem chegado atrasada (*três vezes). the Ana have. 3SG arrive.PPART late (three times) ‘Ana has been arriving late (repeatedly) (*three times).’

Crucially, this requirement on event plurality is true regardless of the number of participants in the eventuality and thus is part of the encoded meaning of the PPC (see Cabredo Hofherr, Laca & de Carvalho 2010 for a treatment of the PPC in a variety of Brazilian Portuguese as a pluractional operator). However, corpus data provide evidence that in earlier stages of Portuguese the PPC denoted the resultant state of a single event, i.e. it received a resultative or recent past interpretation with no implication of plurality of events. This is shown in (6) (in boldface), an example from the 15th century extracted from the Corpus do Português (CdP).5 (6)

[Context: The speaker just found out that the king has died.] ca em el Rey meu Senhor eu tenho perdido um tão bom e verdadeiro amigo (CdP, 15th Century) ‘Since in the person of my Lord the King I have lost such a good and true friend’

In this case, the context rules out a plural interpretation of the form tenho perdido: the speaker is referring to the death of the king, a recent once-only event. In contemporary Portuguese, under the same contextual premises, (6) would be semantically anomalous. Another instance of an unambiguous non-iterative interpretation of the PPC is (7), extracted from the Tycho Brahe Parsed Corpus of Historical Portuguese (TBCHP).6 Here, the complement of the verb requires a single event interpretation. –––––––—–– (i)

5

6

?? Dez animais têm nascido nesta rua. ten animals have.3PL be_born.PPART on-this street We will return to this issue later. Corpus do Português, Mark Davies and Michael Ferreira (2006) (45 million words, 1300s-1900s). Available online at http://www.corpusdoportugues.org. Tycho Brahe Parsed Corpus of Historical Portuguese (approximately 2.3 million words, 1400s-1800s). Available at http://www.tycho.iel.unicamp.br.

28

Patrícia Amaral & Chad Howe (7)

Eu ey por bem que Nicolaao Jusarte, fidalgo de minha casa, a que tenho ffeito merce da capitania de um dos navios que vão pera a India nesta armada d'outubro, vaa no navio do Porto (TBCHP, 16th Century) ‘I order that Nicolau Jusarte, nobleman of my house, whom I have awarded the honor of being the captain of one of the ships that will go to India in the October fleet, shall go on the ship from Oporto.’

In (7), both the number of the NP um dos navios and contextual information force a once-only event interpretation: under normal circumstances, someone is chosen as the captain of a ship going on a specific expedition only once. This paper details the semantic change leading from the interpretation of the PPC as denoting the resultant state of a (possibly single) event to the current non-resultative interpretation of the PPC, with pluractional properties (following the notion of ‘pluractionality’ proposed by Van Geenhoven 2004). In our analysis, the pluractional meaning arises as a consequence of semantic ambiguity given appropriate morphosyntactic conditions. We assume an ambiguity between a resultative reading (i.e. with focus on the resultant state denoted by the participle) and a multiple event reading, the latter type of reading being favored in certain contexts by pragmatic factors (Traugott and Dasher 2001, Eckardt 2006). The multiple event reading serves as the vector for semantic change: the pluractional interpretation of the PPC arises in contexts in which the semantic plurality in the arguments of the verb induces event plurality, which gradually becomes conventionally associated with the PPC form. This paper is organized as follows. In section 2, we present the semantic properties of the PPC in contemporary Portuguese. In section 3, we analyze the periphrasis ter + Past Participle in diachrony. In section 4 we detail our proposal concerning the semantic change undergone by the Portuguese PPC. Section 5 provides concluding remarks.

2.

The Present Perfect in contemporary Portuguese

Before turning to our discussion of the Portuguese PPC in diachrony, we will provide some initial observations concerning those semantic and syntactic properties discernable in present-day Portuguese that bear on our subsequent diachronic explanation. Previous analyses of the PPC in Portuguese have argued that part of its core meaning is the aspectual restriction of required iteration of telic eventualities (see Campos 1986, Giorgi and Pianesi 1997, Schmitt 2001, and Cabredo Hofherr, Laca & de Carvalho 2010). We present the aspectual properties of

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

29

the PPC in section 2.1. In terms of its temporal properties, analyzed in section 2.2, the PPC in Portuguese is similar to perfects in other Romance languages, requiring that an eventuality be evaluated in an interval containing utterance time, where utterance time may or may not be a final subinterval (see Howe 2007).

2.1 Aspectual properties The primary semantic feature distinguishing the PPC in Portuguese from other, typologically related have-perfects is the requirement that a plurality of events be distributed evenly over an interval. In the case of non-stative eventualities, the distribution occurs via event iteration. Recall that in (1), the PPC with the achievement verb chegar ‘arrive’ was incompatible with a one-time event interpretation. For examples (8) and (9) below, both cases refer to an unbounded plurality of blossoming events, where the number of iterations is undetermined. These events are distributed over the relevant time interval, with a left boundary that is located before utterance time and a right boundary that follows utterance time. Thus, utterance time is included in the interval of evaluation. The difference between the two is that (8) denotes a plurality of blossomings with one single participant, a árvore ‘the tree’ (i.e. the same tree blossoming at multiple, non-overlapping times), whereas in (9) there are multiple participants in multiple blossoming events. Thus, (8) is not true if the tree blossomed only once, say in the previous spring, and (9) cannot be true if there are multiple blossomings of different trees that occur once and have the same run time. In sum, it is not enough to simply have a “plurality” of events; these events must also be distributed along the time interval of which the predicate holds. (8)

A árvore tem florido. the tree have. 3SG blossom.PPART ‘The tree has been blooming.’

(9)

As árvores têm florido. the trees have.3PL blossom.PPART ‘The trees have been blooming.’

We also observe iteration of events with complex telic events, as in (10). With this aspectual class, what is iterated is not an eventuality that has reached its completion, which for (10) would be an eventuality of reading the whole book. Rather, the PPC contributes an imperfective interpretation: (10) has an intermittent reading in which subparts of the same book were read on

30

Patrícia Amaral & Chad Howe

different occasions—i.e. a plurality of sub-events of the same type (reading from the book).7 Here we follow Smith (1997) in assuming that, for imperfective aspect, the eventuality denoted by the predicate is evaluated as a superinterval of the interval of evaluation (i.e. the interval of evaluation is contained within the temporal run time of the eventuality). (10) a.

b.

A Maria tem lido As vinhas da ira, the Maria have. 3SG read.PPART the grapes of wrath mas ainda não acabou. but yet not finish.PST3SG ‘Maria has been reading The Grapes of Wrath, but she hasn’t finished it yet.’ A Maria tem lido As vinhas da ira the Maria have. 3SG read.PPART the grapes of wrath (desde o mês passado). since the month past ‘Maria has been reading (parts of) The Grapes of Wrath (since last month).’

The imperfective aspectual value of the Portuguese PPC can be shown by comparing (10) above with the imperfective aspectual periphrasis in (11) and the past perfective form in (12). The imperfective periphrastic form with andar a + Infinitive in (11a) does not entail a completed book-reading event, similar to the PPC in (10a).8 The perfective past form in (12a), however, in an out-of-the-blue context, does carry this entailment. Both the PPC and the imperfective past forms are compatible with adverbials that indicate duration— e.g. desde o mês passado ‘since last month’ as in (10b) and (11b). These adverbials cannot cooccur with perfective forms, as shown in (12b). (11) a.

A Maria anda a ler As vinhas da ira the Maria walk. 3SG to read.INF the grapes of wrath mas ainda não acabou. but yet not finish.PST3SG ‘Maria is currently reading The Grapes of Wrath, but she hasn’t finished it yet.’

–––––––—–– 7

8

This is reminiscent of Van Geenhoven’s observation about the sentence John was eating a fish: “Here, the pluractional operator corresponding to imperfective aspect creates a plurality of eatings which each involve a part of an object. Imperfective aspect thus triggers a partitive interpretation of the accomplishment’s complement.” (Van Geenhoven 2005:118). It can be argued that the intermittent reading for example (10) is the preferred reading due not only to semantic factors but also to pragmatic ones. While it is certainly possible to have repeated instances during which the entire book is read, this reading is marked pragmatically due to the typical length of a book and the recency implications associated with the PPC.

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect b.

(12) a.

b.

31

A Maria anda a ler As vinhas da ira the Maria walk. 3SG to read.INF the grapes of wrath (desde o mês passado). since the month past ‘Maria is currently reading The Grapes of Wrath (since last month).’ ??A Maria leu As vinhas da ira the Maria read.PST the grapes of wrath mas ainda não acabou. but yet not finish.PST3SG ‘??Maria read The Grapes of Wrath, but she hasn’t finished it yet.’ A Maria leu As vinhas da ira the Maria read.PST3SG the grapes of wrath (*desde o mês passado). since the month past ‘Maria read The Grapes of Wrath (*since last month).’

With stative predicates, both iterative and durative interpretations9 are available, as shown in (13). As is typical with perfects with stative predicates (see Dowty 1979, among others), there are multiple readings available for (13): one in which there are iterated states of Pedro being sick which alternate with states of Pedro not being sick during the relevant time interval (analogous to examples 8, 9, and 10), and another that is durative—i.e. refers to only one state of illness that initiated in the past and continues without interruption up to, and possibly after, utterance time. The schemata in (10) and (13) below represent the types of readings described for examples (10) and (13). The schema in (10) depicts the multiple event reading available for both stative and non-stative predicates. The schema in (13) represents the durative interpretation available with stative predicates. In (10) and (13), (i) e is an eventuality of the type denoted by the predicate, (ii) IEval is the interval of evaluation, (iii) LB and RB are the left and right boundaries, respectively, of IEval, (iv)
(e) is the total run time of the eventuality, and (v) UT is utterance time. (13) O Pedro tem estado the Pedro have. 3SG be.PPART ‘Pedro has been sick.’

doente sick

–––––––—–– 9

Here we are departing from the current literature on the Portuguese PPC, which either does not address the availability of durative readings (cf. Schmitt 2001) or argues that they are not available in all varieties of Portuguese (cf. Cabredo Hofherr, Laca & de Carvalho 2010).

32

Patrícia Amaral & Chad Howe (10) Multiple Event/Pluractional Reading:

UT

IEval e1

e2

e3

e4

e5

e6

e7

LB

RB
(e)

(13) Durative Reading:

IEval

UT

e1

LB

RB
(e)

The schema in (10) shows the aspectual requirement that iterated eventualities be distributed as regular and discrete repetitions throughout the interval of evaluation (noted by Cabredo Hofherr, Laca & de Carvalho 2010 as the requirement for discontinuity and regularity). Our diachronic analysis will illustrate that this aspectual property is predictable from the proposed vector of semantic change: the requirement on temporal distribution observed in synchrony arises in diachrony in the interaction between the interpretation of the verbal predicate and nominal quantification in its complements.

2.2. Temporal properties Two main properties characterize the temporal semantics of the PPC in Portuguese. First, the interval of evaluation for the PPC must include utterance time (Campos 1986, Cabredo Hofherr, Laca & de Carvalho 2010). This feature is typical among perfects cross-linguistically and accounts for the incompatibility of the PPC with adverbials that exclude utterance time (see Klein’s 1992 discussion of the “Present Perfect Puzzle”). Thus, in (14), até ontem ‘until yesterday’ cannot co-occur with the PPC because it places the right boundary of the interval of evaluation prior to utterance time.

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

33

(14) Até agora / #Até ontem, a Ana tem viajado sozinha. ‘Until now / #Until yesterday, Ana has been traveling alone.’

The location of the left boundary is also restricted in a PPC, though not in the same manner as the right boundary. In example (15), the incompatibility with desde ontem ‘since yesterday’ arises pragmatically as an indirect result of the iteration of events required by the aspectual profile of the PPC. In order for an interval of evaluation to felicitously contain multiple instances of an iterated event, that interval must be sufficiently extended so as to allow for this iteration, where the required “sufficient” extension is determined on the basis of world knowledge. Multiple (and hence distinct) traveling events cannot, under normal conditions, occur in an interval beginning only one day prior to utterance time. Campos observes that the proximity between the left boundary of the interval of evaluation of the PPC and the utterance time is not part of the semantics of this construction, offering example (16) as evidence (taken from Campos 1986: 41, example 12). (15) Desde sempre/ #Desde ontem, a Ana tem viajado sozinha. ‘Since always/ #Since yesterday, Ana has been traveling alone.’ (16) A língua falada no Brasil tem evoluído muito desde que os portugueses ali chegaram no século XVI. ‘The language spoken in Brazil has evolved a lot since the Portuguese arrived there in the 16th century.’

The temporal characteristics of the PPC discussed here are largely consistent with those of the resultative source construction (discussed below), especially as they relate to utterance time, which must be included in the interval of evaluation of the PPC. The most important distinction between the PPC and the resultative construction is that while the denotation of the latter, a resultant state, is evaluated relative to a point in time (utterance time), the denotation of the PPC (iterated eventualities) can only be evaluated with respect to a time interval. The interval of evaluation of a PPC includes the utterance time and must be a sufficiently extended time interval that allows for the iterated eventualities denoted by the predicate to hold.

3.

The diachronic picture

At least until the 16th century, both the verbs ter (from Latin tenre) and haver (from Latin habre) occurred in the periphrastic construction with the

34

Patrícia Amaral & Chad Howe

Past Participle in Portuguese.10 According to Wigger (2004:178), the overall frequencies of haver + Past Participle and ter + Past Participle were roughly equivalent in the 13th and 14th centuries. The subsequent centuries saw a steady decline of haver + Past Participle, so much so that by the 17th and 18th centuries ter is found almost categorically in collocations with a Past Participle (Wigger 2004:178). For this reason, our analysis focuses on data from the 16th century so that we can target the process of change. Eventually, ter became the auxiliary of the PPC. Note that in contemporary Portuguese the alternation between ter and haver can be marginally found in the Past Perfect, depending on register (tinha saído/ havia saído) but not in the Present Perfect, where the use of haver is ruled out (tem saído/ *há saído). The lexical semantics of ter (originally meaning ‘to obtain, to hold’) has been proposed as a catalyzing factor in explaining the semantic properties of the Portuguese PPC within Romance (see, e.g., Giorgi and Pianesi 1997).11 While we shall not account for the iterative interpretation of the Portuguese PPC as a function of the lexical semantics of ter, we will argue that the choice of ter as the auxiliary in the PPC played a role in the semantic change undergone by this form.12 A study of the constructions ter + Past Participle and haver + Past Participle in texts of the 13th century reveals that ter tended to occur more often in the resultative construction than haver (Cardoso and Pereira 2003). In the following, we introduce this construction within the periphrastic forms with ter + Past Participle. We will focus on the ambiguity triggered by certain collocations of one of these constructions in diachrony.

3.1. Two constructions with ter To begin, it is necessary to distinguish two constructions with ter + Past Participle found in synchrony in Portuguese (and in other Romance languages,

–––––––—–– 10

11

12

We are excluding from our domain of inquiry the verb ser ‘to be’, which could also occur in a periphrasis with the Past Participle in Portuguese. Viotti (1998:44) notes that in Vulgar Latin habre was already significantly desemanticized, a process which is reflected in the distribution of haver as early as in the 10th century. Consequently, haver no longer assigned thematic roles, either to agents or patients, and collocated in generic and existential constructions (see Mattos & Silva 1991, Ribeiro 1993, and Wigger 2004). The verb ter undergoes a later process of desemanticization and subsequently occurs in competition with haver in the periphrastic construction with the Past Participle. Cabredo Hofherr, Laca & de Carvalho (2010) argue that the effect of auxiliary selection in the semantics of the PPC is “indirect” and concerns its functional opposition with the resultative construction.

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

35

like Spanish and Galician), the resultative construction13 and the PPC, exemplified in (17) and (18), respectively. (17) Tenho a porta have. 3SG the door.FSG ‘I have the door closed.’

fechada. closed.FSG

(18) Tenho fechado have. 3SG close.PPART.MSG ‘I have been closing the door.’

a the

porta. door.FSG

In (17), the participial adjective fechada agrees in gender and number with the NP a porta, and the word order is [ter NP PPART]. In (18), an instance of the Portuguese PPC, the word order is [ter PPART NP] and there is no agreement between the direct object of the verb and the past participle, regardless of the position of the direct object with respect to the PPC form.14 In (17), the NP is the direct object of the verb ter and fechada is a secondary predicate modifying the complement of the verb. This periphrasis formed with the verb ter has its roots in the resultative construction which is considered to be the origin of the Present Perfect in the Romance languages, exemplified in (19), from Salvi (1987: 226): (19) habeo epistul-am script-am have.1SG letter-FSG:ACC written-FSG:ACC ‘I have a letter written / I have written a letter.’

(Latin)

This construction, found as early as in Pre-Classical Latin texts, displayed the following properties: (i) the verb habeo is a main verb with a full meaning of possession, (ii) the past participle scriptam has a predicative function and is a secondary predicate of the direct object, displaying the behavior of an adjective, (iii) there is no obligatory coindexation between the subject of habeo (in (19), the speaker) and the logical subject of the participle (i.e. in (19) the letter that is in possession of the speaker may or may not have been written by her/him), and (iv) this construction was restricted to past participles of telic verbs. The structure assumed for this construction is that of a small clause, given in (19): –––––––—–– 13

14

Note that the small clause “resultative” construction in Romance discussed in section 3.1 is not the same as the construction in English discussed by Dowty (1979) and Williams (2008)—e.g. Al pounded the cutlet flat. (Williams 2008:5). In contemporary Portuguese, word order is a reliable criterion to distinguish the two constructions. In the case of the PPC, there may be interpolation of adverbs like lá or até, but not of a complement NP.

36

Patrícia Amaral & Chad Howe (19) [VP habeo [A NP A]]

(Salvi 1987: 228)

Thus, (19) can be glossed as ‘I own a written letter.’ Accordingly, the resultative construction exemplified in (17) and the PPC in (18) have different entailments. The resultative construction denotes a state that holds at utterance time, whereas the PPC denotes an eventuality that is distributed over a time interval right-bounded by utterance time but which may or may not be true at utterance time,15 as shown by the contrast between (20a) and (21a). Hence, the resultative construction accepts modification by an adverbial that refers to utterance time (or an extended present), whereas the PPC may only co-occur with an adverbial that introduces a time interval (cf. (20b) vs (21b)). (20) a. b.

(21) a. b.

??Tenho a porta fechada, mas a porta não está fechada. ?? ‘I have the door closed, but the door is not closed.’ Agora tenho a porta fechada. ‘Now I have the door closed.’ Tenho fechado a porta, mas a porta não está fechada. ‘I have been closing the door, but the door is not closed (now).’ Até agora tenho fechado a porta. ‘Up till now I have been closing the door.’

Compare the schematic representation of the interpretation of the resultative construction, given below as (20), with the one provided in (10), which can also depict the PPC in (21).

–––––––—–– 15

Peres (1996:36) considers the location of the event with respect to utterance time to be context-dependent, allowing for the possibility that an eventuality denoted by a PPC may not hold at utterance time. Consider (i) and (ii): (i)

(ii)

O Paulo tem estado muito doente. Não sei se agora já estará recuperado, porque não falo com ele há dois dias. [undetermined] ‘Paulo has been very sick. I don’t know if he has already recovered because I haven’t spoken with him in two days.’ O Paulo tem estado muito doente. Olha como está pálido. [includes the utterance time] ‘Paulo has been very sick. Look how pale he is.’

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

37

(20) Resultative

UT

e1 IEval The resultative construction is aspectually stative and hence contributes no entailment of a prior event (although one may infer the existence of an event yielding the resultant state). For this reason, it is neither compatible with instrumental phrases (e.g. com o pé ‘with my foot’) nor with adverbials that modify an eventive predication (e.g. rapidamente ‘quickly’), as in (22a). The PPC, on the other hand, may occur with such modifiers, as in (22b): (22) a.

b.

??Tenho a porta fechada com cuidado / ‘I have the door closed carefully / rapidamente / frequentemente. quickly / frequently.’ Tenho fechado a porta com cuidado / ‘I have closed the door carefully / rapidamente / frequentemente. quickly / frequently.’

com o pé / with my foot /

com o pé / with my foot /

In contemporary Portuguese, there is also a difference in the participial forms that may occur in each of these constructions. For verbs that allow a weak and a strong participle like acender 'to light' (cf. Nunes 1989, Maia 1986), e.g. acendido and aceso, respectively, the former occurs in the PPC and the latter in the resultative construction: (23) a.

b.

Tenho acendido/ *aceso velas. I have lit1.msg / lit2.msg candles ‘I have been lighting candles.’ Tenho a vela acesa/* acendida. I have the candle.f lit2.fsg / lit1.fsg ‘I have the candle lit.’

We will return to these properties as we analyze the emergence of the Portuguese PPC.

38

Patrícia Amaral & Chad Howe

3.2. Collocational tendencies of ter + Past Participle An analysis of the periphrases of ter + Past Participle in the corpus data from the 16th century reveals three distinguishable patterns: 1. Ter + Participial Adjective (referred to as STRUCTURALLY RESULTATIVE in what follows): there is overt agreement between the complement NP and the past participle (the participle is a secondary predicate of the direct object), as in (17); 2. STRUCTURALLY PERFECT: there is overt non-agreement between the complement NP and the past participle (for transitive verbs, as in 18); and 3. STRUCTURALLY AMBIGUOUS: both the NP complement and the past participle are [MSG], so it is not possible to determine on morphosyntactic grounds whether the periphrasis is an instance of the resultative construction or an instance of the Present Perfect, as in (27) below.

Despite the fact that in synchrony the resultative construction displays the order [ter NP PPART], word order was not used as a discriminating criterion in the analysis of the corpus data because it has proven not to be a reliable criterion in previous diachronic stages of Portuguese (cf. Wigger 2004, Cardoso & Pereira 2003), as shown by some of the examples below. Besides the agreement criterion, it is possible to identify other distributional facts that relate to the syntactic properties of the resultative construction. In (24), the two participial adjectives cerrada and selada are coordinated with a prepositional phrase which predicates a property of the argument a carta ‘the letter’. Crucially, a coordination structure of this type would be anomalous with the PPC. (24) Vejo…que temos a carta cerrada, selada e com sobre escrito (TBCHP, 16th century) ‘I see that we have the letter closed, sealed and inside an envelope.’

Evidence for treating these examples as instances of the resultative construction comes from the predicative function of the participle, as attested by its syntactic behavior. In (25), the past participle behaves as an adjective, since it may undergo degree modification (tão dilatada ‘so much expanded’). (25) os Padres da Companhia de Jesus, que nelas tem tão dilatada a fé de Cristo (TBCHP, 16th century) ‘the Priests of the Company of Jesus, who have the faith in Christ so much expanded [in the provinces of the Empire].’

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

39

Note that in (25) the NP complement a fé occurs after the participle and yet we find the properties of the resultative construction, thus supporting the assumption that at this diachronic stage word order is not a reliable criterion for discriminating the construction type (see Cardoso & Pereira 2003). The second type of pattern identified above is exemplified in (26). Here, there is no agreement between the past participle amostrado ‘demonstrated’, which is [MSG], and the NP muita amizade ‘much friendship’, headed by a noun that is [FSG]. (26) e pela muita amizade que tenho amostrado a el Rey de França (TBCHP, 16th century) ‘and for all the friendship that I have repeatedly demonstrated to the King of France’

The presence of a recipient argument (a el Rey de França ‘to the King of France’) is also a formal test to identify the eventive interpretation of the ter + Past Participle form. The resultative construction, which is aspectually stative, would not be compatible with such an argument. Finally, (27) shows a structurally ambiguous case: the head of the direct object NP nome ‘name’ is [MSG], and the participle displays the -o ending. On morphosyntactic grounds, (27) could either be an instance of the resultative construction or of the PPC.16 (27) e o nome deste soldado também o tempo tem gastado, como o tem a outras muitas cousas bem dignas de memória (TBCHP, 16th century) ‘and the time has also eroded the name of this soldier, as has been the case with many other things well worth remembering’

The structural ambiguity exemplified in (27) will be an important component of the process of semantic change, which, in our proposal, leads to the iterative meaning of the Portuguese PPC. However, we will claim that the contexts in which semantic change was triggered must display yet another type of ambiguity, semantic ambiguity, described in the next section. –––––––—–– 16

As one reviewer points out, the verb gastar ‘to spend/to erode’ in contemporary Portuguese has both a weak and a strong, or “truncated” (Nunes 1989: 318, Lindsay 1894: 543) form of the past participle, gastado (as in example 27) and gasto, respectively. In synchrony, the former occurs in compound tenses and cannot have an adjectival use, while the latter is limited to adjectival contexts, including the resultative construction (e.g., A placa tem o nome gasto ‘The sign has the name worn out’) and the passive voice. Before the 18th century, however, only the form gastado was available, further corroborating our observation about the potential ambiguity of examples like (27). See Bosque (1990) for a discussion of “perfective adjectives” of the type represented by gasto.

40

4.

Patrícia Amaral & Chad Howe

The proposal: Mechanisms for change

4.1. Contexts of Semantic Change We propose that the contexts inducing the semantic change undergone by the Portuguese PPC constitute a subset of the examples that instantiate pattern 3 above (STRUCTURALLY AMBIGUOUS). This subset is formed by examples with transitive verbs with a direct object that is [MSG] but semantically contributes a plural interpretation. We argue that this plural interpretation arises when the direct object contains either a universal quantifier or a mass noun. What is crucial about these examples is that they present two types of ambiguity. First, they are ambiguous on structural grounds, since the direct object is syntactically [MSG] and the past participle displays a [MSG] ending. That is, from their morphosyntactic properties it is not possible to determine whether they are instances of the resultative construction or if they are instances of the PPC. In the former case, the past participle is analyzed as a predicate of the nominal complement of the verb and agrees with it. In the latter case, there is no agreement, and auxiliary verb and participle constitute a morphosyntactic unit. The examples conforming to this pattern are ambiguous due to the reanalysis permitted by their structural properties. Second, they are ambiguous on semantic grounds, as the semantic plurality of the internal argument of the verb triggers a plural interpretation of the eventuality described by the verb. If an example is interpreted as the resultative construction, it denotes a state that holds at utterance time and is true of multiple participants. On the other hand, if an example is interpreted as the PPC, it denotes a series of (multiple, distinct) events leading to the resultant state. Each of the events corresponds to one of the participants affected by the eventuality. The driving assumption of our proposal is that this ambiguity arises VP-internally, i.e. plurality in the verbal domain is triggered by semantic plurality of the internal argument of the verb. The two types of ambiguity are exemplified in (28). The morphosyntactic ambiguity between the resultative construction and the PPC arises via [MSG] agreement between the participle (escrito) and the antecedent of the relative pronoun, tudo ‘everything’. The semantic ambiguity is triggered by the meaning of this antecedent, which is a universal quantifier. The wh-phrase here is the nominal complement of the verb escrever ‘to write’. (28) em tudo o que escrito tenho, o tenho mizclado (TBCHP, 16th century) ‘in everything that I have written, I have mentioned it (the decorum)’

Under a resultative interpretation, (28) denotes a set of written works in which the issue of ‘decorum’ is mentioned. It can be inferred that this set of works is a result of either one or multiple writing events by the author. Alter-

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

41

natively, under the PPC interpretation, (28) denotes multiple events of writing within a time interval, such that in all the events the author mentioned the issue of ‘decorum’. The interpretation inducing temporal distribution of the writing events (i.e. plurality in the event domain) is favored over a singleevent interpretation, since presumably the same author could not have written all his works on a single occasion. The ambiguity here consists of the possibility of universal quantification over written works versus quantification over writing events. We claim that the resultative construction is the historical precursor of the PPC in Portuguese. Note that the emergence of the iterative interpretation requires a stage at which the construction ter + Past Participle entails the prior occurrence of the event denoted by the verb, as the PPC in contemporary Portuguese denotes regular event iteration over a time interval. A semantically plural object triggers a plural interpretation of the eventuality denoted by the verb also when it is not possible to differentiate the participants affected by the eventuality. In fact, this is true of many corpora examples from the 16th century, like (28). This semantic feature of the direct object favors the ambiguity between: (i) an entailment of existence of a single event and (ii) an entailment of existence of plural events whose cardinality is undetermined. As noted by Cusic (1981), certain types of nominal complements may lead to diffuseness and multiplicity at the event level, i.e. semantic plurality in the nominal domain has a bearing on event individuation. We argue that in the types of contexts described, the implications resulting from event individuation eventually led to semantic change, resulting in the present-day aspectual properties of the PPC. From a semantic point of view, two features in the nominal domain contributed to induce plurality in the verbal domain: (i) the nominal complement of the verb often contains an indefinite pronoun (e.g. tudo, pouco) that may have either a singular or a plural referent, or (ii) the nominal complement of the verb contains a universal quantifier or a mass noun, inducing a distributive interpretation. Following Schein (1993, 2003), we assume that distributivity over individuals is mediated by distributivity over subevents; in other words, distributive quantification requires concurrent quantification over events. This interaction between nominal and event quantification plays a role in the emergence of the pluractional meaning in the diachrony of the Portuguese PPC. In sum, on our analysis, nominal quantification involving the direct object of the verb has a bearing on temporal distribution, inducing a change in the aspectual properties of the PPC. We assume that the nominal properties that play a role in the semantic change undergone by the PPC are found in the direct object, which is in line with the “ergative-like” pattern found in the literature on pluractionality. Cross-linguistically, the arguments of the verb that can be shown to systematically interact with pluractional operators are

42

Patrícia Amaral & Chad Howe

subjects of intransitive verbs and objects of transitive verbs (see Cusic 1981, Newman 1990).

4.2. Onset contexts Before turning to the analysis of specific examples from the corpora, we want to discuss briefly the role that context and ambiguity play in semantic change. The cases of ambiguity discussed in the previous sections represent “onset contexts”—contexts in which an item can be understood as having “older” and “newer” uses (Eckardt 2006:42). For our purposes, the older meaning of the ter + Participle construction is the resultative meaning (with the concomitant resultant state entailment) and the newer meaning is the pluractional interpretation. The transition between different stages of meaning has been argued to occur as a result of inference on the part of the speaker who seeks to enhance the expressive content of an item beyond its conventional meaning, i.e. through a process of pragmatic enrichment (Traugott and Dasher 2001, Heine 2002, Diewald 2002). However, Eckardt argues that “[m]ore than mere pragmatic inference is required in order to force a construction’s meaning to shift” (2006:53). On her view, inferential (i.e. pragmatic) tendencies alone do not explain how and when an item will come to take on a new meaning (or meanings) in a given context; there must also be accompanying structural and semantic ambiguities that both allow and constrain the processes of inference. The onset contexts for the emergence of pluractional meaning with the Portuguese ter + Past Participle construction, we argue, are only those that satisfy the syntactic and semantic criteria detailed in the previous section. An analysis based only on the conventionalization of possible pragmatic inferences associated with the ter construction would not distinguish the pattern of semantic change exhibited in Portuguese from those of other Romance languages with similar structures.17

4.3. Nominal quantification and temporal distribution We turn now to the instances of onset contexts for the shift from resultative to pluractional meaning. There are two necessary components of ambiguity in these cases. One is associated with the morphologically [MSG] participle –––––––—–– 17

Indeed, Spanish displays a resultative construction with tener (< tenre), e.g. Tengo todos los libros leídos ‘I have all of the books read’ and a Present Perfect with haber that does not require an iterative interpretation. See Harre (1991) for a detailed discussion of the tener construction in Spanish.

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

43

and the other is associated with the semantics of the nominal complement. This is exemplified in (29). In addition to the structural ambiguity (i.e. cada hospital MSG ‘each hospital’ and provido MSG ‘granted’), we find a universal quantifier (cada ‘each’) that distributes a set of physicians over a set of hospitals (where the terms “hospital” and “physician” should not be taken literally, since the text builds on the metaphor of “spiritual health”): (29) tenho provido cada hospital de seu físico, que são os abades, retores, vigários e curas (TBCHP, 16th century) ‘I have granted each hospital with a physician, who are the abbots, rectors, vicars, and priests’

The schema in (29) depicts the ambiguity between the two interpretations of (29). Here, hn is a hospital belonging to a set of hospitals, dn is a doctor, and en is the event of assigning a doctor to a hospital. The square brackets represent the eventualities whose truth is evaluated for each case. Under the resultative reading, it must be the case that each hospital have a doctor assigned to it where the set of hospitals is exhausted. From this stative interpretation, one may infer a mapping between each hospital-doctor pair and an assignment (sub)event, yielding a set of assignments of the same type. Under the pluractional reading, it must be the case that there are multiple sub-events resulting in hospital-doctor mappings. The change in the bracketing between the resultative and the pluractional interpretations is meant to capture the semantic change. For the resultative interpretation, the relevant ingredients are the resultant hospital-doctor assignments, whereas the pluractional use requires that multiple distinct events of hospital-doctor assignments have occurred.

{

Resultative h1 h2 h3 h4 h5 }

{

{

d 1 d2 d3 d4 d5 }

{

(29)

h1 | e1 | d1

Pluractional h2 h3 h4 h5 } | | | | e2 e3 e4 e5 | | | | d2 d3 d4 d5 }

The hospital-doctor assignments could have been achieved either through a single event (e.g. through an event of signing a letter) or through multiple temporally-sequenced events, each pertaining to a different hospital-doctor pair. Whereas in this case both explanations seem equally plausible, in other examples the plural event interpretation seems more appropriate on pragmatic grounds (if, for instance, the same individual could not have performed mul-

44

Patrícia Amaral & Chad Howe

tiple actions of the same type simultaneously). We believe that this type of pragmatic factor may have played a role in supporting the contextual adequacy of the plural interpretation. However, our claim is that the ambiguity between the plural interpretation in the nominal domain and in the verbal domain is semantic, not pragmatic. From a syntactic perspective, there are several factors favoring reanalysis where the complement of the verb is instantiated by a pronoun displaying [MSG] agreement, hence creating a morphosyntactic ambiguity between the resultative construction and the PPC. There are a number of cases in which the PPC occurs in a relative clause with a pronominal antecedent, with the relative pronoun preceding the PPC (as in (28) and (30) below). The word order in these tokens favors the analysis of the ter + Past Participle construction as a syntactic-semantic unit, suggesting the increased syntactic cohesion typical of verbal periphrases. In example (30) below, the NP (n)este pouco ‘this little bit’ is the antecedent of a relative clause headed by que. This surface structure allows for two distinct parses, one associated with the resultative construction, shown in (30), and the other with the PPC, as in (30). (30) D'aqui se póde conhecer que cousa he pintura e pintor neste pouco que tenho dito. (TBCHP, 16th century) ‘From here it can be known what painting and painter are, from this little bit that I have said.’ (30) Resultative: [NP este pouco [COMP quei [S [NP (eu)] [VP tenho [NPi [AdjP dito]]]]]] (30) PPC: [NP este pouco [COMP quei [S [NP (eu)] [VP tenho dito [NPi ]]]]]

This type of syntactic indeterminacy contributes to reanalysis in the presence of semantic factors that induce ambiguity. A crucial component of the semantic change we are analyzing is the entailment of a prior event yielding a resultant state. When the occurrence of the event denoted by the verb becomes part of the encoded meaning of the form, we expect to see an expansion in the aspectual classes of the verb phrases as well as other distributional facts ruled out in the stative resultative construction. This is exemplified in (31), where the factor inducing event plurality is the presence of the mass noun contentamento.18 Mass noun –––––––—–– 18

An anonymous reviewer rightly pointed out that (31) could not have been interpreted as an instance of the resultative construction given that in synchrony such an analysis is ruled out. We believe that this is due to the lexical semantics of the verb receber and not, as the reviewer suggested, to the nominal complement of the verb. Our point is precisely that (31) exemplifies the expansion of the ter + Past

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

45

complements adhering to the criteria for structural ambiguity (i.e. [MSG]) also give rise to a multiple event interpretation: (31) Eu tenho recebido tanto contentamento com vossas cartas, pelas quais tenho visto, e pelas obras sabido, como me tendes bem servido (TBCHP, 16th century) ‘I have received so much satisfaction with your letters, by which I have seen, and by the works known, how you have served me well.’

The complement tanto contentamento ‘so much satisfaction’ in (31), though morphologically singular, allows for an interpretation of multiple receiving events, strengthened by the prepositional adjunct com vossas cartas ‘with the letters from you’ which distributes the satisfaction-receiving events over individual letters (cf. Schein 2003). Note that the presence of this prepositional phrase with an instrument role (com vossas cartas) confirms the emerging eventive interpretation of the construction. Such an adjunct could not occur with the aspectually stative resultative construction, as shown in section 3.1. In (31) there is a potential ambiguity between a single event reading—i.e the writer received several letters on a single occasion—and a multiple event reading. In the latter, there is an incremental effect; the writer’s degree of satisfaction increases with each non-overlapping event of letter receiving. Here, the mass noun allows for a similar distribution of participants over eventualities as observed in example (29). In this case, the mapping of each letter to a degree of contentamento induces a plurality of letter-receiving events.19 On the pluractional reading, the focus is on the repetition of letterreceivings with the complement being interpreted as incrementally increased with each iteration. Note that due to the homogeneity of mass nouns, each part of contentamento is not distinct from another part; hence, on the plural interpretation of the events, event individuation yields events that are all of the same type (i.e. “satisfaction-receiving” events). This condition conforms to the requirement on event iteration of the PPC in synchrony. It is not surprising that a multiple event reading would be available given the quantificational nature of the nominal modifiers in examples (28), (29), (30), and (31) (i.e. tudo, cada, pouco, and tanto). Crucially, the semantics of these quantifiers induces a distributive interpretation of the event denoted by the verb with respect to the participants in the event but does not ‘count’ the –––––––—––

19

Participle construction to predicates that were not licensed by the resultative construction. One reviewer pointed out that nouns like contentamento are not simply mass nouns but rather represent what Tovena (2001) refers to as an "intensive quantity" with an inherent degree structure. We believe that this analysis is compatible with the incremental interpretation of the example, which would favor, on our proposal, the emergence of the multiple event meaning.

46

Patrícia Amaral & Chad Howe

iterated events introduced by the pluractional reading, unlike cardinal expressions. The lack of adverbial modifiers expressing cardinality, as well as cardinalized NPs in subject position, is a property of the distribution of the PPC in contemporary Portuguese (recall (5), repeated here) . The corpus data suggest that this type of modification was never available. (5)

A Ana tem chegado atrasada (*três vezes). ‘Ana has been arriving late (repeatedly) (*three times).’

With respect to modification with cardinal adverbials, we want to point out one crucial distinction between the diachronic collocational tendencies of the PPC in Portuguese and those of the Spanish haber Perfect (from Latin habre). Thibault (2000:97) draws a comparison between the plural event meaning of the Portuguese PPC and the possibility of iterated events with the compound past in Spanish, which is claimed to be the required meaning in some Spanish dialects (e.g. Mexican Spanish, see Moreno de Alba 1978). In synchrony, the Portuguese PPC is not compatible with cardinal adverbials. For Mexican Spanish, however, the haber Perfect can co-occur with cardinal modification, as shown in (32). (32) Sí; he ido dos ocasiones [a su tierra]. (from Lope Blanch 1976) ‘Yes. I have gone on two occasions [to his hometown].’

Thibault presents historical data that illustrate the emergence of the so-called plural meaning of the Spanish haber Perfect as a function of frequent collocation with “indicadores iterativos” ‘iterative indicators’ like dos vezes ‘two times’ and muchas vezes ‘many times’ (Thibault 2000:98). Our analysis of the diachronic data from Portuguese revealed no instances of the ter + Past Participle construction co-occurring with these types of adverbials. In light of these distinct collocational patterns in diachrony, it is not surprising that the contemporary data from Portuguese, as in example (5), and Spanish, as in (32), display divergent semantic properties with respect to modification with cardinal adverbials. This observation provides some corroboration for our claim regarding the interaction between nominal and verbal plurality as the locus for semantic change and the subsequent semanticization of event plurality that is unique to the semantics of the PPC in contemporary Portuguese. To summarize, the contexts described in the above examples represent a necessary condition for the shift from resultative to pluractional meaning. We have presented several factors pertaining to plurality in the nominal domain that play a role both in the mapping between the eventuality denoted by the participle and the time interval as well as in the process of event individuation. For example (29), the DO complement cada hospital ‘each hospital’ indicates a plurality of participants in an event and induces a distributive interpretation, resulting in the subsequent interpretation of multiple events. This

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

47

effect is also obtained in example (28) as as result of the meaning of the universal quantifer tudo ‘everything’. Finally, the plural adjunct com vossas cartas ‘with your letters’ combines with the NP tanto contentamento ‘so much satisfaction’ in (31) to produce an incremental interpretation with the predicate recebido ‘received’. These contexts are the relevant vectors of change that precipitate the transition from resultative to pluractional in the presence of morphosyntactic ambiguity. In the 16th century TBCHP corpus, we observed several instances of ter + Past Participle that, despite adhering to our primary criterion for structural ambiguity (i.e. both the NP complement and the past participle are [MSG]), were ruled out as cases resulting in a possible multiple event reading. Such cases do not display the semantic properties that might favor the pluractional interpretation. Note that (33) would, on structural grounds, be a suitable candidate for an onset context for the semantic change undergone by the PPC. (33) e tem um templo alevantado a êste ídolo, que se chama o Paraiso de Amida (TBCHP, 16th century) ‘and (they) have a temple raised up for this idol who is called Amida Paradise’

Nevertheless, the possibility of a plural event reading in (33) can be ruled out because of the number of the direct object and of the nature of the predicate alevantar ‘to raise up/to erect’, which, in an unmarked context, would not allow multiple instances of the same event. Thus, it is not simply structural ambiguity that allows for this transition but rather the possible interpretations licensed by the interaction between the verb ter and one of the pluralizing factors described above. Finally, we have claimed that the source of plurality that gives rise to the pluractional meaning of the PPC in Portuguese is found in the interaction between the verbal complex ter + Past Participle and a direct object complement. Another possible catalyst for this trend might be the influence of a plural subject, which, like plural objects, can induce a multiple event reading. Thus, compare example (34a), which refers unambiguously to a single washing event, to (34b), which allows for either a single event reading (i.e. the collective reading) or a multiple event reading (i.e. the distributive reading). Moreover, if the indefinite complement in (34b) takes narrow scope, the presence of a plural subject would allow for a multiple event reading in which Peter, Paul, and Mary all wash different cars at the same time. In section 2.1, we demonstrated that a multiple, simultaneous event reading with a plural subject is not possible with the PPC in contemporary Portuguese; the multiple events must be distributed over the time interval and have non-overlapping run times (see example 9).

48

Patrícia Amaral & Chad Howe (34) a. b. c.

Peter washed a car. [PL Peter, Paul, and Mary] washed a car. Peter washed [PL some cars].

In comparison, example (34c) entails multiple car-washing events, under the assumption that any one person can only wash one car at time. Here we can see evidence for how object plurality might play a role in the process of event individuation that is not systematically required with plural subjects. Taking the resultative construction as the diachronic source of the PPC, the type of event plurality required by the PPC in synchrony (i.e. multiple events distributed without overlap over the relevant interval of evaluation) is only possible with a singular subject and plural objects, as shown in (35). (35) O Pedro tem [PL os carros lavados]. ‘Peter has the cars washed.’

4.4. Distribution of PPC predicates in diachrony The predicates that occur with the Portuguese resultative construction in synchrony must be (i) telic and (ii) transitive. Analyses of the development of periphrastic past forms from resultative constructions in Romance frequently underscore the gradual expansion of these predicates to include both atelic and intransitive predicates (see Detges 2000, among others). What is particular to the Portuguese PPC is that the intransitive cases are accompanied by further semantic information that allows for the emergent pluractional meaning (see 39 below). As a further test case, we will also observe changes in the collocational patterns with the adverbial até agora ‘until now’, which can be used as a heuristic for the increased semanticization of the pluractional meaning of the PPC in diachrony. In the 16th century data, we observed only eight cases (3%) of stative predicates with ter. In the 18th century data, this number increases to 105 (21.3%), a result consistent with the general trends in expansion of semantic classes across periphrastic forms in Romance. Of the eight cases from the 16th century, almost all stative predicates are with verbs of cognition, as shown in example (36). Several studies have argued that the collocation of resultative constructions with stative verbs was an initial stage in the transition to becoming a periphrastic past (see e.g. Detges 2000). Analysis of the 18th century data, however, revealed a number of tokens in which noncognition statives were attested, as in example (37).

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

49

(36) quero nesta declarar o que tenho disto entendido e que queria que lá se fizesse (TBCHP, 16th century) ‘I want in this (letter) to declare what I have understood of this and what I would like to be done there’ (37) tenho sido vossa inimiga até o presente (TBCHP, 18th century) ‘I have been your enemy up till the present’

What is notable about the example in (37) is the availability of a durative interpretation. In this case, the eventuality denoted by the stative predicate is initiated in the past and continues, without interruption, into the present. This reading is reinforced in example (37) by the boundary modifier até ao presente ‘until the present’ which brings the right boundary of the interval at which the eventuality holds up to utterance time. This reading of the PPC is available in synchrony despite claims that stative predicates with the PPC are also coerced into an iterative reading (cf. Schmitt 2001). Also evident in the data is the variable compatibility with achievements. In synchrony, achievement verbs can only be used in the PPC if an iterative interpretation is coerced. Thus, without a context that allows for multiple instantiations of the eventuality, as provided in example (38) by the overt adverbial modification with muitas vezes ‘a lot’, the PPC would be infelicitous. (38) O João tem chegado ao cimo do monte muitas vezes. ‘João has been arriving at the top of the hill a lot (recently).’

Moreover, we would not expect in contemporary Portuguese to have intransitive verbs occur with the resultative construction—e.g. *Tenho a carta chegada ‘*I have the letter arrived’. It seems then that we can use the cooccurrence of these predicates with the PPC as a test for the expansion of the semantic domain in diachrony. In addition, if these predicates do co-occur with the PPC, we would only expect to find them in contexts for which an interpretation of multiple events is clearly possible if not required. Observe the following example: (39) que tal nome merecem os extremos a que o mundo tem chegado nesta materia (TBCHP, 16th century) ‘that such a name deserve the extremes that the world has reached with this issue’

In example (39), we have selected a protypical achievement predicate— chegar ‘reach/arrive’—which must occur in a context that allows for the eventuality to be repeated when used with the PPC in synchrony. This token co-occurs with a pluralizing element—i.e. os extremos ‘the extremes’— allowing for compatibility with the emergent pluractional meaning of the ter + Past Participle periphrasis. In a separate survey of diachronic data from

50

Patrícia Amaral & Chad Howe

Corpus do Português, we observed no examples of ter + {chegado/ alcançado} attested before the 16th century, further corroborating the claim that the 15th and 16th centuries represent a crucial point in the development of the ter + Past Participle constructions (see Wigger 2004). Finally, the resultative construction is incompatible with interval adverbials. The PPC, however, is acceptable with these types of modifiers. Compare the following with example (14) above. (40) #Tenho a carta escrita até agora. ‘#I have the letter written up till now.’

Therefore, we would expect that as the pluractional emerges we should see increased compatibility with this modifier. Of the uses of até agora in the 16th century data, all occur either with stative predicates or under the scope of negation—an atelicizing context (see Smith 1997). For the 18th century, we expect to find até agora with a wider range of predicates as the meaning of the PPC becomes semanticized, generalizing across verb classes. As expected, até agora is attested with atelic or atelicizing elements, such as a stative predicate, as in (41) or a plural object, as in (42). Crucially, these contexts are found across a variety of verb types—e.g. non-stative visto ‘seen’. (41) Nem o Gabinete de Espanha tem tido até agora vigor (TBCHP, 18th century) ‘And the Spanish Office has not had the strength up till now’ (42) as que tendes visto até agora. (TBCHP, 18th century) ‘those that you have seen up till now’

Taken together, the observations made in this section offer further evidence of the diachronic trajectory of the ter + Past Participle construction as it develops from a structure that indicates the resultant state of a past action to a marker of event plurality. We have argued that a transition of this type would be evidenced by specific changes in the compatibility of ter with different predicates, namely atelic and intransitive predicates. Our preliminary comparison of the 16th and 18th century TBCHP data supports this claim and provides evidence of the increased grammaticalization of pluractional meaning with the PPC. These results are consistent with the observation that, cross-linguistically, the diachronic change undergone by the resultative construction involves semantic widening, expanding the class of verbs with which this structure can occur (see Dahl and Hedin 2000:393).

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

5.

51

Conclusions

In this paper we have proposed an account of the interpretation of the Portuguese PPC by observing its diachronic development from the original resultative source construction (following Wigger 2004 and others). In contemporary Portuguese, the PPC, which takes the verb ter as an auxiliary, requires event plurality; its resultative ter counterpart does not. We conclude that at some point in the diachrony of the resultative construction, there was a shift in the meaning of the ter + Past Participle periphrasis that resulted in speakers using this form to make reference to a plurality of events. We have proposed a mechanism to account for this change: the pluractional meaning of the PPC emerges from transfer of semantic plurality in the nominal domain to the verbal domain. This transfer results in the interpretation of event plurality, which is subsequently semanticized, becoming part of the conventional meaning of the PPC. We have analyzed various cases in which nominal plurality due to universal quantification (examples 28 and 29) or to the meaning of a mass noun (example 31), is argued to derive from a complement of the verb. From a structural perspective, event plurality arises only in the cases in which reanalysis is possible, i.e. those cases in which morphosyntactic ambiguity licenses either a resultative construction or a PPC. Given this confluence of factors, the interpretation and semanticization of event plurality can be viewed as emerging under a specific set of semantic and structural conditions. We have described these conditions as the onset contexts for this change. We follow Eckardt (2006) in assuming that analyses of language change can benefit from the rigors of formal semantic description, especially in the identification of the mechanisms that precipitate and transmit structural and semantic change. In the present analysis, we have defended this approach to modeling language change, using the interaction between the nominal and verbal domains as a test case. The success of this enterprise in the study of language change will be measured by the degree of explanatory precision that semantic analysis can offer.

References Bosque, Ignacio (1990). “Sobre el aspecto en los adjectivos y en los participios.” In: Ignacio Bosque (ed.): Tiempo y aspecto en español, 177–211. Madrid: Cátedra. Cabredo Hofherr, Patricia, Brenda Laca, & Sandra de Carvalho (2010). “When Perfect means Plural: the Present Perfect in North-Eastern Brazilian Portuguese.” In: Patricia Cabredo Hofherr & Brenda Laca, (eds) Layers of Aspect, 67–102. Stanford, CA: CSLI. Campos, Maria Henriqueta C. (1986). “L’opposition du portugais Pretérito Perfeito Simples – Pretérito Perfeito Composto : un cas singulier dans l’ensemble des

52

Patrícia Amaral & Chad Howe

langues romanes.” Morphosyntaxe des Langues Romanes. Actes du XVIIe Congrès International de Linguistique et Philologie Romanes (Aix-en-Provence), 411–422. Cardoso, Adriana & Susana Pereira (2003). “Contributos para o estudo da emergência do tempo composto em Português.” Revista da ABRALIN, vol. II, nº2, 159–181. Comrie, Bernard (1976). Aspect. Cambridge: Cambridge University Press. Cusic, David (1981). Plurality and Aspect. PhD Dissertation. Stanford. Dahl, Östen & Eva Hedin (2000). “Current relevance and event reference.” In Ö. Dahl (ed.) Tense and aspect in the languages of Europe, 386–401. Berlin and New York: Mouton de Gruyter. Detges, Ulrich (2000). “Time and truth: The grammaticalization of resultatives and perfects within a theory of subjectification.” Studies in Language 24:345–377. Diewald, Gabriele (2002). “A model for relevant types of contexts in grammaticalization.” In: I. Wischer & G. Diewald (eds) New Reflections on Grammaticalization, 103–120. Amsterdam: John Benjamins. Dowty, David (1979). Word Meaning and Montague Grammar. Dordrecht: Reidel. Eckardt, Regine (2006). Meaning Change in Grammaticalization. Oxford: Oxford University Press. Giorgi, Alessandra & Fabio Pianesi (1997). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Harre, Catherine (1991). Tener + Past Participle: A Case Study in Linguistic Description. London: Routledge. Heine, Bernd (2002). “On the role of context in grammaticalization.” In: I. Wischer & G. Diewald (eds) New Reflections on Grammaticalization, 83–102. Amsterdam: John Benjamins. Hopper, Paul & Elizabeth Traugott (1993). Grammaticalization. Cambridge: Cambridge University Press. Howe, Chad (2007). “A Semantic/Pragmatic Analysis of Present Perfects in Peninsular Spanish” In: J. Cihlar, A. Franklin, D. Kaiser, & I. Kim-bara (eds) Papers from the 39th Regional Meeting of the Chicago Linguistic Society, 144–161. Klein, Wolfgang (1992). “The Present Perfect ‘Puzzle’.” Language 68:525–552. Lindsay, Wallace M. (1894). The Latin Language. Oxford: Clarendon Press. Lope Blanch, Juan M. (1976). El habla popular de México: materiales para su estudio. México: UNAM. Maia, Clarinda de Azevedo (1986). História do Galego-Português: Estudo Linguístico da Galiza e do Noroeste de Portugal desde o século XIII ao XVI. Lisboa: Fundação Calouste Gulbenkian & Instituto Nacional de Investigação Científica. Mateus, Maria H. M. et al. (2003). Gramática da Língua Portuguesa. Lisboa: Caminho. Mattos e Silva, Rosa V. (1991). “Caminhos de mudanças sintático-semânticas no português antigo.” Cadernos de Estudos Linguísticos 20: 59–74. Moreno de Alba, José (1978). Valores de las formas verbales en el español de México. México: UNAM. Newman, Paul (1990). Nominal and verbal plurality in Chadic. Dordrecht: Foris.

Nominal and verbal plurality in the diachrony of the Portuguese Present Perfect

53

Nishiyama, Atsuko & Jean-Pierre Koenig (2004). “What is a perfect state?” In: B. Schmeiser, V. Chand, A. Kelleher & A. Rodriguez (eds) WCCFL 23 Proceedings, 101–113. Somerville, MA: Cascadilla Press. Nunes, José Joaquim (1989). Compêndio de Gramática Histórica Portuguesa (Fonética e Morfologia). (9th ed.). Lisboa: Livraria Clássica Editora. Peres, João (1996). “Sobre a semântica das construções perfectivas do Português.” Congresso Internacional sobre o Português V. 2:33–58. Ribeiro, Ilza (1993). “A formação dos tempos compostos: A evolução histórica das formas ter, haver, e ser.” In: I. Roberts & M.A. Kato (eds) Português brasileiro: Uma viagem diacrônica, 343–86. Campinas, São Paulo: Editora da Unicamp. Salvi, Giampaolo (1987). “Syntactic Restructuring in the Evolution of Romance Auxiliaries.” In: M. Harris & P. Ramat (eds) Historical Development of Auxiliaries, 225–36. Amsterdam: Benjamins. Schein, Barry (1993). Plurals and Events. Cambridge: MIT Press. Schein, Barry (2003). “Adverbial, Descriptive Reciprocals.” In: J. Hawthorne & D. Zimmerman (eds) Language and Philosophical Linguistics, 333–68. Malden, MA: Blackwell. Schmitt, Cristina (2001). “Cross-linguistic variation and the Present Perfect: The case of Portuguese.” Natural Language and Linguistic Theory 19:403–453. Smith, Carlota (1997). The Parameter of Aspect. 2nd ed. Dordrecht: Kluwer. Thibault, André (2000). Perfecto simple y perfecto compuesto en español preclásico. Tübingen: M.N. Verlag. Tovena, Lucia M. (2001). “Distributional restrictions on negative determiners.” In: K. Jaszczold & K. Turner (eds) Meaning Through Language Contrast, 3–28. Amersterdam/Philadelphia: John Benjamins. Traugott, Elizabeth C. & Richard C. Dasher (2001). Regularity in Semantic Change. Cambridge: Cambridge University Press. Van Geenhoven, Veerle (2004). “For-adverbials, frequentative aspect, and pluractionality.” Natural Language Semantics 12:135–190. Viotti, Evani (1998). “Uma história sobre ter e haver.” Cadernos de Estudos Linguísticos 34:41–50. Wigger, Lars-Georg (2004). Die Entwicklungsgeschichte der romanischen Vergangenheitstempora am Beispiel des Pretérito Perfeito Composto im Portugiesischen. PhD Dissertation. Universität Tübingen. Williams, Alexander (2008). “Patients in Igbo and Mandarin.” In: J. Dölling, T. Heyde-Zybatow, & M. Schäfer (eds) Event Structures in Linguistic Form and Interpretation, 3–30. Berlin/New York: Mouton de Gruyter.

Martina Faller

Pluractionality and accompaniment in Cuzco Quechua*

1. Introduction This paper gives an overview of the pluractional suffixes of Cuzco Quechua (CQ) and develops a first formal analysis of a subset of them using the semantic schema proposed by Lasersohn (1995). Pluractionality, a term originally coined by Newman (1990), is the grammatical marking of verbal or event plurality and is usually marked on the verb or within the VP.1 Pluractional verbs by definition involve multiple events of the same kind, typically this is the kind described by the verb root.2 Plural events may involve “multiple participants, times or locations” (Lasersohn 1995:240), and all three of these are found with the CQ pluractionals. The basic temporal and spatial pluractional uses of the CQ suffixes are exemplified in (1) and are straightforwardly accounted for with the semantics proposed by Lasersohn (1995).3 –––––––—–– *

1

2

3

I’m indebted to my main bilingual consultants Inés Callalli Villafuerte, Natalia Pumayalli Pumayalli, Edith Zevallos Apaza, and Gloria Canal for their time and willingness to share their language expertise with me. For this paper, they wereconsulted to confirm the data taken from other sources and they participated in the elicitation of original data. The paper has been much improved by comments I received from two anonymous reviewers and from Brenda Laca. All remaining errors are of course mine. I’d also like to thank the audience at the Workshop on Nominal and Verbal Plurality, CNRS, Paris, November 2008, and the participants at the first SPINFest in Manchester in May 2009 for their feedback. Though see Matthewson (2000) for an analysis of a quantifier-like element as a pluractional. An exception to this is so-called event-internal pluractionality, where the verb root describes a collection of multiple events but where the individual events may not fit this description. An example from English is nibble which involves multiple small biting events, each of which on its own would not constitute nibbling (Cusic 1981, cited in Lasersohn 1995). This paper will not be concerned with eventinternal pluractionality. Unless otherwise indicated, examples were elicited by the author. I adopt the convention of capitalizing names of morphemes. Key to abbreviations used in examples: 1, 2, 3: first, second, third person, ACC: accusative, ADD: additive, AST: assistive, AUG: augmentative, BEN: benefactive, CAUS: causative, CISL: cislocative, COM: comitative, COND: conditional, DIR: direct evidence, DIM: diminutive, EMO: emotive, FREQ: frequentative, GEN: genitive, HORT: hortative, ILLA: illative, IMP:

56

Martina Faller (1)

a.

b.

Tashi p’unku-ta wisq’a-paya-n. Tashi door-ACC close-FREQ-3 ‘Tashi closes the door again and again.’ (elicited) Ama ichaqa phawa-kacha-nki=chu. NEG.IMP but run-SIM-2=NEG ‘But don’t run around all over the place.’ (“Pero no vayas a corretear por uno y otro lado”) (Itier 1999:20)

However, two of the CQ pluractionals can, in addition, be used to convey meanings of “social accompaniment” (term from Lasersohn 1995:201) as illustrated in (2). (2) is an example involving the Reciprocal -puna, which requires the Reflexive -ku to derive this meaning. The paper aims to derive these meanings compositionally from the semantics of the pluractionals and the devices that introduce a beneficiary such as the reflexive. (2)

Chay=pis waqa-puna-ku-n-ku. this=ADD cry-RECIP-REFL-3-PL ‘They cried with each other.’ (“Lloraron en los brazos el uno del otro.”) (Itier 1999:196)

The remainder of this introduction gives basic grammatical background information on CQ. The basic meaning of CQ pluractional suffixes will be described in section 2. Section 3 provides an account of the main types of pluractional meaning using Lasersohn’s (1995) semantic schema for pluractionality, which is then extended to account for the accompaniment readings in section 4. Section 5 provides a conclusion and highlights some issues for further investigation. Background on Cuzco Quechua The term Quechua refers to a family of languages spoken in the Andes region. The variety studied here is Cuzco Quechua (CQ), spoken in the Department of Cuzco, Peru. The basic word order of CQ is SOV, but it is quite flexible. There is widespread pro-drop of both subjects and objects. It is a synthetic, agglutinative language and possesses a large number of derivational and inflectional suffixes as well as enclitics. Verbal inflectional suffixes mark tense, aspect (progressive), mood, number, and person. The verbal derivational suffixes, of which Cusihuaman (2001) lists 26, express a variety of meanings, including pluractionality. The suffixes and enclitics are arranged in a template after the root as shown in (3a), with a number of slots within each category. (3b), for example, contains –––––––—–– imperative, INCH: inchoative, LIM: limitative, LOC: locative, NEG: negative, NMLZ: nominalizer, NX.PST: non-experienced past, O: object, PL: plural, PROG: progressive, PST: past, QUEST: question, RECIP: reciprocal, REG: regressive, REFL: reflexive, REP: repetitive, SS: same subject, SIM: simulative, TOP: topic.

Pluractionality and accompaniment in Cuzco Quechua

57

the derivational Hortative, Assistive and Cislocative suffixes, inflectional person and mood suffixes and the enclitic used for forming yes/no-questions. (3)

a. b.

ROOT–derivational–inflectional=enclitics Paqarin llank’a-rqa-ysi-mu-wa-nki-man=chu. tomorrow work-HORT-AST-CISL-1O-2-COND=QUEST ‘Would you help me to work (in the field) tomorrow?’ (Cusihuaman 2001:202)

2. Overview of CQ pluractional suffixes This section gives a brief overview of the suffixes with pluractional meanings found in Cusihuaman (2001) to provide the wider context for the subsequent analysis of a subset of them. There are at least six verbal suffixes which have pluractional uses, namely Durative -raya (allomorph -nraya), Continuative -nya (allomorphs -miya and -ma), Frequentative -paya, Simulative -kacha (allomorph -ykacha),4 Reciprocal -na (allomorph -puna), and Repetitive -pa. The labels and meaning descriptions of these suffixes are largely based on Cusihuaman (2001), with the exception of -na, which is analyzed in detail in Faller (2007). Note that some of these suffixes have non-pluractional uses (not discussed here) in addition to the pluractional ones. They are classed as derivational suffixes (Cusihuaman 2001) and occur in different slots in the suffix sequence, as shown in (4).5 As also shown, all of the pluractional suffixes precede the reflexive suffix -ku, which plays an important role in deriving the accompaniment readings. (4) ROOT

-

 raya 

 kacha 

 paya 

 nya 

 pa 

-ku  (pu)na 

(based on Cusihuaman 2001:205) –––––––—–– 4

5

The name of this suffix is based on its non-pluractional use, in which it indicates that an object pretends to have/simulates a certain property, e.g., qhari ‘man’ and qhari-kacha- ‘pretend to be brave/a man’. Suffixes that occur in the same slot are mutually exclusive, but others can in principle co-occur with each other. Their relative order appears to be somewhat flexible, though there is some controversy as to how flexible exactly (Muysken 1988, van de Kerke 1996). Since this paper looks at single occurrences only, this issue can be set aside.

58

Martina Faller

Apart from descriptions of their basic meanings in grammars, there exist very few studies of pluractional markers in Quechua. Detailed studies only exist of the reciprocal marker -na (Faller 2007, van de Kerke 1992, van de Kerke 2007). Since the focus of this paper is on developing a compositional account of the accompaniment readings found with some of the markers, it will also only provide a very brief overview of the entire system, which should, however, enable the reader to appreciate the richness of pluractionality in Quechua. As mentioned above, pluractionals indicate that there are multiple events of the kind described by the verb. In addition, pluractional markers may specify a dimension along which the individual events are separated from each other, e.g., by requiring “multiple participants, times or locations” (Lasersohn 1995:240). At the same time, the individual events are conceptualized as constituting a single, overarching event (Wood 2007:92ff). The CQ verbal pluractional suffixes all have temporal uses, encoding repetition in time, and with the exception of Reciprocal -na, this appears to be their basic meaning. Two suffixes, namely Reciprocal -na and the Simulative -kacha, also allow a spatial interpretation, and three, Reciprocal -na, Frequentative -paya and Repetitive -pa can be used for participant-based pluractionality, including reciprocity and accompaniment. Table 1 gives an overview of the attested uses of the 6 pluractional suffixes.6 The following subsections describe the meanings of the six pluractional suffixes as well as reduplication as a means of conveying plurality of events. Repetition

(Duration)

+

+

-nya continuative

+

+

-paya frequentative

+

+

-kacha simulative

+

+

-pa repetitive

+

-na reciprocal

+

-raya durative

Spatial

Participant

(Conative)

+ + + +

+

+

Table 1: Attested uses of the six pluractional suffixes –––––––—–– 6

No minus signs were used in this table to indicate that a pluractional does not have a certain interpretation, as further fieldwork is required to obtain such negative evidence. Duration and Conative are given in parentheses as it is not clear that these uses should be considered a type of pluractionality.

Pluractionality and accompaniment in Cuzco Quechua

59

2.1. Pluralizing times In temporal pluractionality, the individual events are separated from each other along the temporal dimension, that is, a temporal pluractional requires there to exist multiple events which take place at different times, but which may have the same participants and take place at the same location. Temporal pluractionality is closely related to aspect. For example, iterativity is often considered a type of aspect (Corbett 2000), but since it involves the repetition of an event in time and therefore indicates event plurality, it can also be considered a type of pluractionality. The meaning of a pluractional verb is also dependent on the interaction of the pluractional morpheme with the base verb’s Aktionsart or telicity. Thus, as we will see below, some of the temporal pluractionals give rise to meanings of durativity with certain verbs but to iterativity with others. This interaction and the relationship between aspect and pluractionality more generally is of great interest and requires further study, but the study of this relationship in CQ will have to await another occasion. Readers interested in this issue are referred to Cusic (1981), HajiAbdolhosseini, Massam & Oda (2002), Yu (2003), van Geenhoven (2005), Wood (2007) and references cited therein.7 As mentioned above, all pluractionals in CQ have a temporal use, indicating iteration of events. With the exception of Repetitive -pa (see below), all specify that an event is repeated an indefinite number of times, usually taken as ‘many’, and it is not immediately obvious how they differ from each other in this use. Three of them, namely, Durative -raya, Continuative -nya and Frequentative -paya have a purely aspectual meaning with certain verbs (and it remains to be investigated what unifies these verbs), expressing that the event or state described persists for a duration or “a long time”, as shown in (5)-(7):

–––––––—–– 7

Another type of meaning that is sometimes subsumed under the term pluractionality is degree of intensity of the action described. Wood (2007:15) argues, however, that intensity by itself cannot be considered a type of pluractionality as it does not necessarily involve multiple events. However, some pluractionals may sometimes give rise to a secondary meaning of intensity, see for example Repetitive -pa in (17) below. Two suffixes in CQ that have only intensive meanings, and which should therefore not be considered pluractionals, are the Exaggerative -tiya and the Augmentative -yu, exemplified in (i) and (ii). (i) t’aqta- ‘level/flatten’, t’aqta-tiya- ‘stamp one’s feet very hard’ (ii) Nishu-ta-n para-yu-sha-n. much-ACC=DIR rain-AUG-PROG-3 ‘It is raining very hard.’

60

Martina Faller (5)

a. b. c.

qhawa- ‘look’, qhawa-raya- ‘look for a long time’ suya- ‘wait’, suya-raya- ‘wait for a long time’ chura- ‘put’, chura-raya- ‘for something to stay in a place for a long time (where it was put)’ (Cusihuaman 2001:187)

(6)

qhawa- ‘look’, qhawa-nya- ‘observe, guard, watch’ (Cusihuaman 2001:190)

(7)

qhawa- ‘look’, qhawa-paya- ‘look for a long time’ (Cusihuaman 2001:189)

Since there is no (necessary) plurality of events, these uses are purely aspectual. Judging by the translations, which are based on Cusihuaman’s (2001) Spanish translations, there appears to be no difference between -raya and -paya in this use, though, given that Cusihuaman (2001) used a different set of verbs to translate qhawa-nya, -nya appears to have a more idiomatic use. It is also interesting to note that chura-raya, (5c), expresses not the duration of the event denoted by the base verb chura- ‘put’ but of the result state of having been put in some place. With other verbs, these suffixes indicate iteration of events, a clear instance of pluractionality, as shown in (8)-(10). (8)

kuti- ‘return’, kuti-nraya- ‘return repeatedly’

(9)

a. b.

astakuti-

(10)

a. b. c.

much’a- ‘kiss’, much’a-paya- ‘kiss frequently’ wisq’a- ‘close’, wisq’a-paya- ‘close again and again’ muyu- ‘round, turn’, muyu-paya‘circle, stay close’ (Cusihuaman 2001:189)

‘transfer’, ‘return’,

asta-nyakuti-nya-

(Cusihuaman 2001:187)

‘transfer again and again’ ‘return continuously’ (Cusihuaman 2001:190)

The number of repetitions implied by these suffixes is ‘many’,8 and this is also the case with the temporal uses of Simulative -kacha and Reciprocal -na exemplified in (11), (12) (a full-sentence example found in a folktale), and (13). It should be noted that examples like (13) are rather rare.9 –––––––—–– 8

9

It remains to be tested whether this is a strict entailment or an implicature, that is, it is not clear whether the examples are felicitous if the event is repeated just once. Cusihuaman (2001) does not mention repetition of the event as a possible interpretation for -na, but examples which may be analyzable in terms of repetition have been reported for the closely related variety of Bolivian Quechua by van de Kerke (1996), e.g., (i)

Pluractionality and accompaniment in Cuzco Quechua (11) a. b.

kumparit’i-

‘fall’, ‘snow’,

61

kumpa-kacha- ‘fall again and again’ rit’i-kacha‘snow with interruptions’ (Cusihuaman 2001:188)

(12) Waqya-kacha-n=si Pascual=qa. call-SIM-3=REP Pascual=TOP ‘Pascual started to insult (lit. call again and again) it.’ (Itier 1999:23) (13) . . . mana=n saru-na-wa-na-nchis=chu ka-sqa-nchis-wan. . . . not=DIR step.on-RECIP-1O-NMLZ-1INCL=NEG be-NMLZ-1INCL-COM ‘. . . (that) they must not discriminate against us (lit.: ‘repeatedly step on/ trample on us’) for what we are.’ (“que no nos discriminen por lo que somos)” (FARTAC:46)

Repetitive -pa differs from the pluractionals discussed so far in that it expresses that an event is repeated once, (14), rather than many times. For example, if someone swept the floor badly, and has to redo it, the verb picha-pa ‘sweep again’ could be employed. Note, though, that -pa can also be used for multiple repetitions as in (14c). (14) a. b. c.

picha- ‘sweep’, picha-payapa- ‘add’, yapa-pathuqa- ‘spit’, thuqa-pa-

‘sweep again’ ‘add more’ ‘spit repeatedly’ (Cusihuaman 2001:192)

Repetitive -pa furthermore differs from the other pluractionals in that it usually presupposes at least one previous occurrence of the described event.10 To illustrate, (15a) asserts that Pedro sweeps the floor repeatedly, without presupposing anything about previous sweepings, whereas (15b) asserts that Pedro sweeps the floor and presupposes that someone had swept it before (either Pedro or someone else). Thus, -paya corresponds roughly to English repeatedly and -pa to again.11 –––––––—–– (i)

10

11

pay turril-is-man taqha-na-ku-sqa he vessel-PL-ILLA bump-RECIP-REFL-NX.PST ‘He bumped into the vessels one after the other.’ That there is not always a presupposition present is suggested by example (14c), which seems to be no different from the non-presuppositional repetitive uses of the other pluractionals. While I have to confirm this with other speakers of Quechua, one speaker also accepts -pa with a restitutive interpretation like again. E.g., in a situation in which I am sent a package, open it, then close it again, because I noticed that it is not for me, (i) is felicitous. (i) Wisq’a-pa-sha-ni. close-REP-PROG-1 ‘I am closing it again.’

62

Martina Faller (15) a.

b.

Pedru pampa-ta picha-paya-n. Pedro ground-ACC sweep-FREQ-3 ‘Pedro sweeps the floor again and again.’ Pedru pampa-ta picha-pa-n. Pedro ground-ACC sweep-REP-3 ‘Pedro sweeps again.’

In conjunction with the Reflexive -ku, Repetitive -pa has two uses which may be considered extensions of its basic temporal meaning. Firstly, -pa-ku can be used with a conative interpretation, that is, it expresses that the subject is making an effort to realize an action, or rehearses an action various times without entailing the completion of that action.12 This is shown in (16). (16) a. b.

siqa- ‘climb’ kicha- ‘open’

siqa-pa-ku- ‘try to climb (a tree, a wall)’ kicha-pa-ku- ‘make an effort to open s.th.’ (Cusihuaman 2001:192)

Secondly, with some verbs, -pa-ku expresses a high degree of intensity, which at least in some cases is a consequence of the repetition of the described event, (17). (17) a. b.

wikch’u- ‘throw away, lose’ saksa‘satisfy’

wikch’u-pa-ku- ‘vomit’ saksa-pa-ku- ‘eat too much’ (Cusihuaman 2001:192)

It will be left for further research to establish how these extensions can be analyzed and what the contribution of the reflexive -ku is in deriving these meanings. In addition to verbal affixes, reduplication of (part of) the verb can also express temporal pluractionality, as shown in (18).13 –––––––—–– 12

13

This appears to be a common extension of pluractionals (Wood 2007, Cusic 1981). Wood (2007:130) puts forward an interesting proposal to account for conative readings arising when pluractionals attach to telic verbs, which relies on Cusic’s (1981) distinction between event-internal and event-external pluractionality. In this paper, I am not concerned with this distinction, and will therefore leave an analysis of these conative meanings to future work. Reduplication can also be used to express notions of collectivity and frequency in the nominal domain, as shown in (i)–(iii). (i) mallki ‘tree’ mallki mallki ‘wood’ (ii) wasi ‘house’ wasi wasi ‘complex of houses’ (iii) p’unchay ‘day’ p’unchay p’unchay ‘day after day’ Further research is required to ascertain the different uses of reduplication in both the nominal and verbal domains and how it differs from plural marking by means of affixes.

63

Pluractionality and accompaniment in Cuzco Quechua (18) a.

b.

Hamu hamu-lla-n=mi pasaqpas pay=qa. come come-LIM-3=DIR always (s)he=TOP ‘(S)he always comes without fail.’ Ch’aki-ra-pu-n ch’aki-ra-pu-lla-n=mi chay pukyu=qa. dry-HORT-REG-3 dry-HORT-REG-LIM-3=DIR this spring=TOP ‘This spring is drying up each time.’ (Cusihuaman 2001:177)

2.2. Pluralizing locations Spatial uses are only attested for two of the pluractional verbal suffixes, namely Simulative -ykacha and Reciprocal -na. Both express that there are multiple events fitting the verb’s description which take place at multiple locations, while the participants and times may be the same. Cusihuaman (2001) offers the examples in (19) for -kacha and (20) is taken from a published folktale. (19) a. b. c.

qhawaapapuri-

‘look’, ‘take’, ‘walk’,

qhawa-kachaapa-kachapuri-ykacha-

‘look everywhere’ ‘take everywhere’ ‘walk aimlessly’

(20) Ama ichaqa phawa-kacha-nki=chu. NEG.IMP but run-SIM-2=NEG ‘But don’t run around all over the place.’ (“Pero no vayas a corretear por uno y otro lado.”)

(Itier 1999:20)

Like its temporal use, the spatial use of -na is also very rare and I have only one attested example, namely (21). (21) Maskha-na-ri-ku-spa puri-sha-n alqu. search-RECIP-INCH-REFL-NMLZ.SS walk-PROG-3 dog ‘The dog walks, searching (for food) all over the place.’ (spontaneous utterance in casual conversation)

2.3. Pluralizing participants: Reciprocity As was shown in the previous two sections, the suffix -na has (rare) uses in which it serves to pluralize times or locations. Its main use, however, is as a marker of reciprocity, hence its label. In this use, -na is almost always accompanied by the reflexive -ku, as shown in (22). 14,15 –––––––—–– 14

Due to these non-reciprocal uses, Faller (2007) argues that -na should not be analyzed as a reciprocal but as a general pluractional marker. For simplicity’s sake, I

64

Martina Faller (22) a. b.

hayt’a- ‘kick’, qhawa- ‘look’,

hayt’a-na-ku‘kick each other’ qhawa-puna-ku- ‘look at each other face to face’

The allomorph -puna (as opposed to just -na) is, according to Cusihuaman (2001:193–194), only used with some verbs, and indicates that two things are coming into contact with each other or that two people look at each other face to face as in (22b).16 For the time being, I will follow Cusihuaman (2001) in assuming that -puna is a single suffix. However, the alternative of considering it to be composed of -pu and -na will be discussed in section 4.2. In its reciprocal use, -na requires multiple participants (Faller 2007, van de Kerke 1996). For example, (23) involves a plurality of events because more than one person gives something to another person. The giving events may or may not happen at the same time and/or location, but what is crucial is that there are several participants involved. (23) Pay-kuna pura qu-na-ku-sha-n-ku. (s)he-PL amongst give-PA-REFL-PROG-3-PL ‘They are giving each other (things).’ (Description of video clip, Faller 2007)

2.4. Pluralizing participants: accompaniment Two of the pluractional suffixes can be used to express accompaniment. Like reciprocity, this use requires the involvement of multiple participants. For example, if Marya accompanies Jose walking, there are two individual walking events, one with Marya as an agent and one with Jose. Accompaniment readings require the individual events to occur in temporal and spatial proximity. The Frequentative -paya can express accompaniment on its own, as shown in (24). (24) a. b. c.

tiya- ‘sit/live’, tiya-paya- ‘accompany s.o. sitting/living’ saya- ‘stand’, saya-paya- ‘stand next to’ puñu- ‘sleep’, puñu-paya- ‘sleep next to’ (Cusihuaman 2001:189)

–––––––—––

15

16

will continue to use Cusihuaman’s (2001) label Reciprocal for this morpheme and gloss it as RECIP. -na can only denote reciprocity on its own when certain morphological conditions hold, e.g., when it occurs with the causative morpheme -chi, hayt’a-na-chi ‘make some people kick each other’, or with the participial form -sqa, hayt’a-na-sqa ‘being in a state of having kicked each other’ (Faller 2007). Cusihuaman (2001) does not indicate what defines the set of verbs that -puna is restricted to and I have so far not been able to identify a criterion either.

Pluractionality and accompaniment in Cuzco Quechua

65

The Reciprocal -puna, however, requires the Reflexive -ku to express the meaning of accompaniment, as in (25), and it is usually the allomorph -puna that is involved, rather than -na.17 (25) a.

b.

Chay-pis waqa-puna-ku-n-ku. this-ADD cry-RECIP-REFL-3-PL ‘They cried with each other.’ (“Lloraron en los brazos el uno del otro.”) (Itier 1999:196) Asi-puna-ku-n-ku pay-kuna pura laugh-RECIP-REFL-3-PL (s)he-PL amongst ‘They are laughing together/with each other.’ (description of a video clip in which two people are laughing)

Having provided an overview of all pluractional suffixes in CQ, the next section will show how the basic temporal and spatial pluractional interpretations can be captured within Lasersohn’s (1995) account before addressing the question of how the accompaniment reading of the Frequentative -paya and the Reciprocal -puna can be derived from pluractionality in section 4. How reciprocity can be composed from the meanings of pluractional -na and the Reflexive -ku is discussed in detail in Faller (2007) and will not be repeated here.

3.

Analyzing temporal and spatial pluractionality

Most formal accounts of pluractionality adopt a version of the semantics first proposed by Lasersohn (1995), and I will follow suit. Lasersohn’s is a neoDavidsonian event semantics,18 that is, verbs are analyzed as predicates of events and their nominal arguments are linked to the events via thematic roles (Lasersohn 1995:253). Thematic roles are partial functions from events to individuals (Landman 2000:38).19 For example, the verb hayt’a ‘kick’ denotes –––––––—–– 17

18

19

Though van de Kerke (2007:1385) presents an example of what he calls the sociative use of Bolivian Quechua -na-ku: (i) Pedru tata-n-wan Cliza-man puri-na-ku-n-ku. Pedro father-3-COM Cliza-ILLA walk-RECIP-REFL-3-PL ‘Pedro and his father walk together to Cliza.’ (van de Kerke 2007:1385, his ex. (36b)) See van Geenhoven (2005) for an alternative account of pluractionality which does not rely on events. Parsons (1990:74) calls thematic roles ‘relations’, but at the same time adopts the Unique Role Requirement, making them effectively functions: “Each of these roles relates an event (or a state) to a thing. No event stands in one of these relations to

66

Martina Faller

the set of kicking events and its subject and object arguments map onto the agent and theme of these events. In lambda notation, the meaning of hayt’a ‘kick’ is represented as in (26), and the sentence in (27a) is translated as in (27b) (disregarding tense).20 (These representations will be revised slightly in light of the cumulativity assumption discussed below.) (26) hayt’a  x. y. e[kick(e)
AG(e) = y
TH(e) = x]

In words, hayt’a is a function that combines with two arguments, x, the direct object argument, and y the subject argument, and returns a set of kicking events with x as the theme and y the agent. (27) a.

b. c.

Asnu=qa Tashi-ta=n hayt’a-rqa-n. donkey=TOP Tashi-ACC=DIR kick-PST-3 ‘The donkey kicked Tashi.’ e[kick(e)
AG(e) = the-donkey
TH(e) = tashi] There is a kicking event, the agent of which is the donkey and the theme of which is Tashi.

A temporal pluractional verb such as hayt’a-paya- ‘kick repeatedly’ denotes a set of plural events the subevents of which are required to be separated in time. That is, a sentence such as (28a), The donkey kicked Tashi repeatedly asserts that there is a plural or non-atomic (¬AT(e)) kicking event, that the agent of this event is the donkey and its theme Tashi, and that the distinct individual kicking events making up this plural event take place at different times.21 In Lasersohn’s (1995) semantics, temporal separation of events is captured by requiring non-overlap of the events’ run times  (overlap of two times t1, t2 is represented as t1  t2). If two events do not overlap in time, then there will be two different times at which an event of the described type occurs. That is, the plurality of subevents leads to plurality of times. These dif–––––––—––

20

21

more than one thing; thus, each event possesses at most one Agent, at most one Experiencer, and so on.” Here, e is a variable over events, x, y are variables over individuals, and AG and TH are the thematic role functions for agent and theme. Existential closure is applied by default to bind the event variable e (Landman 2000:50). In event semantics, events are taken to be ontological primitives. Following Link (1998), I assume that both the domain of individuals De and the domain of events Ds contain both singular individals/events as well as plural individuals/events and that these domains form a part-whole structure, closed under the sum operation : if a and b are individuals/events, then ab is an individual/event. Sums and their constituent parts form a join semi-lattice, structured by the ordering relation . Thus, we write ab to capture the statement that an event a is part of a plural event b. That is, in (27b) and subsequent formulae, the events e and e are subevents of the plural event e.

Pluractionality and accompaniment in Cuzco Quechua

67

ferent aspects of a pluractional sentence are formally represented for example (28a) as in (28b).22 (28) a.

b. c.

Asnu=qa Tashi-ta=n hayt’a-paya-rqa-n. donkey=TOP Tashi-ACC=DIR kick-FREQ-PST-3 ‘The donkey kicked Tashi repeatedly.’ e[*kick(e)
AG(e) = the-donkey
TH(e) = tashi
¬AT(e)
e, e  e[(e
e
kick(e)
kick(e))  ¬[(e)
(e)]] There is a plural kicking event, the agent of which is the donkey and the theme of which is Tashi, and any two kicking subevents do not overlap in time.

The semantic contribution of a temporal pluractional is then (i) to require that the event denoted by the verb is plural and (ii) to require temporal nonoverlap of its subevents. Spatial pluractionality requires instead non-overlap of locations, and participant-based pluractionality non-overlap of thematic roles. Again, requiring non-overlap of two events in these dimensions results in a plurality of locations and a plurality of participants. A pluractional in general, then, can be analyzed as a function that takes the predicate P or relation R denoted by the verb as its argument and adds to it these requirements as in (29) (Faller 2007).23 The output of this function is again a predicate or relation which can combine with its nominal arguments. (29) PAvi  P. x. e[P(x)(e)
¬AT(e)
(|e|  n) (
e, e  e[(e
e
P
(e)
P(e))  ¬[ƒ(e)
ƒ(e)]])] PAvt  R. x. y. e[R(x)(y)(e)
¬AT(e) (
|e|  n) (
e, e  e[(e
e
R(e)
R(e))  ¬[ƒ(e)
ƒ(e)]])] (based on Lasersohn 1995:256)

–––––––—–– 22

23

I use e  e[. . .] as short-hand for e[e < e  . . .], and e, e as shorthand for ee. That is, the full version of the second line in (28) reads: ee[e  e
e  e
e
e  ¬[(e)
(e)]]. *kick' is a predicate of plural and atomic events, see section 3.1, thus allowing e to be plural. We restrict the non-overlap condition for e' and e'' to atomic events that can be described as kicking. Applying this condition to all subevents of e would make this sentence always false. This is because every kicking event has many subevents which do not constitute kicking events themselves, such as moving one's leg. These "internal" subevents necessarily overlap with the kicking events they are part of. Similarly, we cannot require non-overlap for two events e' and e'' if one is part of the other, even if both are kicking events. I assume that a pluractional marker attaches to the lexical verb, and therefore provide two versions, one that combines with intransitive (PAvi) and one that combines with transitive verbs (PAvt). P and R in (29) correspond to the lambda expression denoted by the verb such as (26), which captures the syntactic combinatorial properties of the verb. P and R are variables for the predicates over events in this lambda expression, e.g., kick.

68

Martina Faller

In (29), event plurality is captured requiring that e is not atomic (¬AT(e)). Non-overlap of the subevents is represented as ¬[ƒ(e)
ƒ(e)], where the variable f can be instantiated either to the temporal trace function, the spatial trace function or a thematic role. However, some pluractionals do not specify a dimension along which the subevents are separated, e.g., Reciprocal -na is, as shown above, compatible with temporal, spatial and participant-based interpretations. For such pluractionals the second line in the definitions is irrelevant and therefore given in parentheses. Moreover, a pluractional marker may specify that the number of subevents is greater than a pragmatically determined number n (|e|  n), and this is therefore also presented as an optional requirement in (29).

3.1. Cumulativity Before applying this semantics to the CQ pluractionals, one additional theoretical assumption needs to be discussed, which will lead to a slight revision of the above semantics for verbs and thematic roles. This assumption is that all predicates in CQ are cumulative, which means that their denotations contain both singular and plural individuals.24 Making this assumption accounts for the fact that, e.g., the noun uwiha ‘sheep’ can be used to refer to singular sheep and plural sheep without requiring an overt number morpheme. Thus, in (30a), uwiha refers to more than one sheep (as is clear from the Spanish translation of the published Quechua text from which this example is taken), though theoretically, this example could also mean that there was just one sheep. When the plural suffix -kuna is added, as in (30b), the singular interpretation is eliminated. (30) a.

b.

Uwiha-q qhepa-n-ta urqo-ta ri-spa=n, ... sheep-GEN behind-3-ACC mountain-ACC go-NMLZ.SS=DIR ‘Walking behind the sheep (pl.) to the mountains . . . ’ (“Yendo a los cerros tras las ovejas”) (Valderrama Fernandez and Escalante Gutierrez 1982:26) Urqo-pi hina uwiha-cha-kuna michi-mu-sqa-y-man hina ... mountain-LOC like sheep-DIM-PL herd-CIS-NMLZ-1-ILLA like ‘As I was herding the sheep in the mountains . . . ’ (“En lo que pasteaba a las ovejitas en los cerros” . . .) (Valderrama Fernandez and Escalante Gutierrez 1982:26)

–––––––—–– 24

Some researchers have argued that this is an assumption that should be made for all languages, e.g. Krifka (1998:200) and Kratzer (2008).

Pluractionality and accompaniment in Cuzco Quechua

69

Verbs, too, can be interpreted as denoting plural events even when they do not contain any plural marker. Thus, (31a) describes a plurality of catching events, not just a single one in which all the thieves were caught collectively. (31b) shows that the same verb form can also denote a single event. Adding a pluractional marker, e.g., the Frequentative -paya as in (31c) forces there to be more than one catching event. (31) a.

b.

c.

Chay-pi-ya suwa-kuna tarpa-chi-ku-n. this-LOC-EMO thief-PL catch-CAUS-REFL-3 ‘In this, the thieves let themselves be caught.’ (“Es en eso pues que los ladrones se hacen atrapar.”) (Espinoza 1997:82) Chay-pi-ya Tashi tarpa-chi-ku-n. this-LOC-EMO Tashi catch-CAUS-REFL-3 ‘In this, Tashi let himself be caught.’ (elicited) Chay-pi-ya Tashi tarpa-paya-chi-ku-n. this-LOC-EMO Tashi catch-FREQ-CAUS-REFL-3 ‘In this, Tashi let himself be caught again and again.’ (elicited)

In order to capture these empirical facts formally, I assume, as mentioned above, that all predicates in CQ have a cumulative denotation. Cumulativity can be defined using the *-operator in (32) (Sternefeld 1998), where a and b may be individual objects or events. (32) Cumulativity: For any set P, P is the smallest set such that a. P  P, and b. if a
P and b
P , then a  b
P

The extension of a starred predicate contains the atomic individuals of the unstarred predicate and all plural individuals that can be formed from them. For example, if the extension of dog is (33a), the extension of dog is (33b). (33) a. b.

dog dog

= {tashi, miskhi, buddy} = {tashi, miskhi, buddy, tashi  miskhi, tashi  buddy, miskhi  buddy, tashi  miskhi  buddy}

Similarly, if the extension of kick is (34a), then the extension of kick is (34b). (34) a. b.

kick kick

= {e1, e2, e3} = {e1, e2, e3, e1  e2, e2  e3, e1  e3, e1  e2  e3}

As a consequence of assuming that nouns and verbs denote cumulative predicates in CQ, a plural marker has to “remove” atoms from the extension of

70

Martina Faller

starred predicates, instead of constructing sums from atoms as is the case for plural markers applying to unstarred predicates. In the semantics proposed for pluractional markers in (29), this is achieved by the condition ¬AT(e). Following Landman (2000) and Kratzer (2008), thematic roles are also assumed to be cumulative.25 The cumulative agent of a cumulative event predicate is the set of the singular and plural agents of the events in the event predicate, where a plural agent of an event “is the sum of the agents of the atomic parts” (Landman 2000:185), and similarly for other roles. For example, in a model with two kicking events in one of which the donkey kicked Tashi and in the other of which the llama kicked Miskhi, the plural agent of the sum of the two kicking events is the sum of the donkey and the llama. The cumulative agent and theme roles of kicking in this model can be represented (“[w]ith some notational fantasy” (Landman 2000:184) as in (35a) and (35b) respectively. (35) a. b.

AG = {, , } TH = {, , }

In a model with two kicking events, in each one of which the donkey kicked Tashi, the extensions of the roles are as in (36a) and (36b). (36) a. b.

AG = {, , } TH = {, , }

In light of the cumulativity assumption, the translation of (27a), repeated here as (37a), needs to be revised to (37b). (37) a.

b. c.

Asnu=qa Tashi-ta=n hayt’a-rqa-n. donkey=TOP Tashi-ACC=DIR kick-PST-3 ‘The donkey kicked Tashi.’ e[kick(e)
AG(e) = the-donkey
TH(e) = tashi] There is a (plural) kicking event, the (plural) agent of which is the donkey and the (plural) theme of which is Tashi.

This semantics leaves open the possibility that the event is a plural event, that is, that there might be more than one kicking event of Tashi by the donkey, which is a possible interpretation of (37). If a speaker wanted to make explicit that this is the case, they can add an overt pluractional marker. In fact, –––––––—–– 25

To be precise, Landman (2000:167ff) argues that only basic predicates have thematic roles, starred predicates have instead plural roles which are derived from thematic roles.

Pluractionality and accompaniment in Cuzco Quechua

71

there might be more than one donkey involved, which could be made explicit by adding the nominal plural marker -kuna to asnu ‘donkey’.

3.2. Basic spatial and temporal pluractionality With these theoretical background assumptions and the semantic schema for pluractionality in (29), we can now look in more detail at the semantics for spatial and temporal pluractionality. Consider (38a) involving the Simulative -kacha as an example of plurality of events involving multiple locations. Setting ƒ in (29) to the spatial trace function , and assuming that -kacha involves ‘many’ events,26 we get the semantic representation in (38b). (38) a.

b. c.

Qhawa-kacha-sha-n. look-SIM-PROG-3 ‘(S)he is looking everywhere.’ e[look
(e)
AG(e) = (s)he
¬AT(e)
|e| = many
e, e  e[(e
e
look
(e)
look
(e))  ¬[(e)
(e)]]] There is a plural looking event with many looking subevents, the agent of which is (s)he, and any two atomic subevents of looking (by her/ him) do not overlap in space.

The plural agent of the plural event is a third person singular individual. Since plural agents are the sum of the agents of the subevents, it follows that this individual is also the agent in each of the subevents, and this does therefore not have to be stated explicitly.27 The looking subevents are required to be “many” and to take place at non-overlapping locations, which captures the spatial interpretation of -kacha.28 For temporal pluractionality, we set ƒ to the temporal trace function , as was already shown for example (28). It has been argued that repetition of –––––––—–– 26

27

28

This assumption is based on the translations offered by Cusihuaman (2001) for the examples with -kacha, which, recall, involve adverbial phrases such as aimlessly, everywhere. It will still have to be confirmed in further fieldwork that this suffix cannot be used when only a few locations are involved. In general, for subsequent formulae, whenever the plural or thematic roles are not explicitly stated for a predication of the verb to an event, they can be calculated from the roles of an event in the same lattice. Note that the cardinality condition will technically have to be more involved. Almost any event will have numerous subevents, which however would not count in and of themselves as events of the relevant type. Thus, a kicking event has many subevents of a leg moving just millimeters, and a single kicking event therefore already has cardinality many. We only want to count complete kicking events. In order not to make the formulae too complicated, though, I will appeal to the most intuitive understanding of this condition.

72

Martina Faller

events necessarily involves strict temporal separation of the events. For example, if John sleeps at two contiguous temporal intervals, then it seems we are dealing with just one continuous sleeping event rather than two (Lasersohn 1995:253). Mere non-overlap of run times, which is all that is required by the semantics in (29), however, allows for two events to be contiguous. We therefore have to add a clause to the semantic representation that requires that there is a temporal interval (a “hole” (Tovena & Donazzan 2008) or “hiatus” (van Geenhoven 2005)) between the run times of any two subevents at which no event of the relevant type occurs (Lasersohn 1995:254). Example (39a) with the frequentative suffix -paya then translates as (39b) (the predicate between(t1, t2, t3) is true iff t2