Decotructing ergativity two type of ergative languges and their features

There are two types of tive languages; in one type, the ergative expression is a PP in the subject position, and in the other type, the ergative is a DP in a structural case, not very di

Deconstructing Ergativity

OXFORD STUDIES IN COMPARATIVE SYNTAX

Richard Kayne, General Editor

Movement and Silence

Richard S. Kayne

Restructuring and Functional Heads:

The Cartography of Syntactic Structures, Volume 4

Guglielmo Cinque

Scrambling, Remnant Movement and

Restructuring in West Germanic

Roland Hinterhölzl

The Syntax of Ellipsis: Evidence from Dutch

Dialects

Jeroen van Craenenbroeck

Mapping the Left Periphery:

The Cartography of Syntactic Structures, Volume 5

Edited by Paola Benincà

and Nicola Munaro

Mapping Spatial PPs: The Cartography of

Syntactic Structures, Volume 6

Edited by Guglielmo Cinque and

Luigi Rizzi

The Grammar of Q: Q- Particles, Wh-

Movement, and Pied- Piping

Functional Heads: The Cartography of

Syntactic Structures, Volume 7

Edited by Laura Brugé, Anna

Cardinaletti, Giuliana Giusti, Nicola Munaro, Cecilia Poletto

Adverbial Clauses, Main Clause Phenomena

and Composition of the Left Periphery:

The Cartography of Syntactic Structures, Volume 8

Edited by Ian Roberts and Enoch Aboh

Aspects of Split Ergativity

Edited by Chris Collins

Japanese Syntax in Comparative Perspective

Edited by Peter Svenonius

Chinese Syntax in a Cross- linguistic Perspective

Edited by Edited by Y.- H Audrey Li, Andrew Simpson, and W.- T Dylan Tsai

The Architecture of Determiners

Thomas Leu

Beyond Functional Sequence:

The Cartography of Syntactic Structures, Volume 10

Edited by Ur Shlonsky

The Cartography of Chinese Syntax: The Cartography of Syntactic Structures, Volume 11

Edited by Wei- Tien Dylan Tsai

Argument Licensing and Agreement

Deconstructing Ergativity

Two Types of Ergative Languages and Their Features

Maria Polinsky

Published in the United States of America by Oxford University Press

198 Madison Avenue, New York, NY 10016, United States of America.

© Oxford University Press 2016

All rights reserved No part of this publication may be reproduced, stored in

a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted

by law, by license, or under terms agreed with the appropriate reproduction rights organization Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above.

You must not circulate this work in any other form

and you must impose this same condition on any acquirer.

Library of Congress Cataloging- in- Publication Data

Names: Polinsky, Maria, author.

Title: Deconstructing ergativity : two types of ergative languages

and their features / Maria Polinsky.

Description: New York, NY : Oxford University Press, [2016] | Series: Oxford Studies in Comparative Syntax | Includes bibliographical references and index Identifiers: LCCN 2015031701| ISBN 9780190256593 (pbk : alk paper) | ISBN 9780190256586 (hardcover : alk paper) | ISBN 9780190256609 (ebook) | ISBN 9780190256616 (online content)

Subjects: LCSH: Grammar, Comparative and general—Ergative constructions | Generative grammar.

In memory of Lana Borodkin (1903– 1980)

1.2.2 The range of the phenomenon 11

1.2.3 The relevance of syntactic ergativity 13

1.3 The importance of starting small 15

1.3.1 Syntactic ergativity broadly defined 15

1.3.2 Not all A- bar movement phenomena are created equal 20

1.3.3 Some methodological odds and ends 21

Appendix: Compensatory strategies under syntactic ergativity 22

2 Proposal 28

2.1 Crucial empirical observations 29

2.1.1 Diachronic pathways to ergativity 29

2.1.2 Oblique subjects 35

2.2 The proposal: Two classes of ergative languages 35

2.3 From a PP specifier to syntactic ergativity 38

2.3.1 The relationship between the verbal functional head and ergative P 38

2.3.2 Ergative P and P- stranding 39

2.3.3 Ergative P and pied- piping 40

2.3.4 From a PP subject to syntactic ergativity 44

2.4 Basic clausal structures in the two types of ergative languages 45

2.4.1 PP- ergative and DP- ergative languages: transitive clauses 45

2.4.2 PP- ergative and DP- ergative languages: unergative clauses 49

2.4.3 PP specifiers everywhere? Preventing overgeneration 53

2.4.4 Compatibility between the ergative and the passive 54

2.5 Summary 55

3 Prepositional phrases: Establishing the diagnostics 56

3.1 PPs have distinct extraction and subextraction properties 57

3.2 Restrictions on PPs as pivots of clefts 59

3.3 PPs have resumptive proforms and may have special modifiers 60

3.4 PPs are less accessible to agreement probes than DPs are 60

3.5 PPs and binding 62

3.6 PPs and A- movement 65

3.7 PPs cannot be at the tail of a control chain 68

3.8 Summary 70

4 Ergative as a PP: Initial evidence 72

4.1 Ergative expressions can be PPs 72

4.2 Subextraction out of the ergative expression 73

4.3 Extraction: Ergative extraction requires resumption 79

4.4 Ergative and agreement 82

4.5 Ergative and depictives 83

4.6 Ergative and quantifier float 84

4.7 Interim summary 87

4.7.1 Silent P head 87

4.7.2 Overt P head 88

4.7.3 The nature of the operator 90

5 Ergative as a PP: Take two 94

5.1 Binding: Reflexives and reciprocals 94

6 Cross- linguistic landscape: Correlates of PP- ergativity 111

6.1 Word order correlates 111

6.2 Expletive subjects 117

6.3 Non- canonical (quirky) subjects 120

6.4 Summary 123

7 The other ergative: A true DP 124

7.1 Extraction of the ergative with a gap 125

7.2 Subextraction from the ergative and the absolutive 127

8 The relationship between PP- ergativity and DP- ergativity:

9 Alternative accounts of variation across ergative languages 160

9.1 Comp- trace versus P- trace 160

9.2 Criterial freezing 163

9.3 Phase boundaries and high-/ low- absolutive languages 166

9.4 Non- syntactic explanations for variation across ergative languages 173

9.5 Summary 180

PART II: Paradigm languages

10 A paradigm PP- ergative language: Tongan 185

10.2.1.1 Basic facts about clitics 195

10.2.1.2 Accounting for Tongan clitics 198

10.2.1.3 Clitic doubling 202

10.2.2 Possessive clitics and possessive markers 203

10.3 Deriving Tongan clause structure 206

10.3.1 Word order: Deriving V1 206

10.3.2 Word order: The right periphery 207

10.3.2.1 The definitive accent 208

10.3.2.2 VOS is not due to scrambling 211

10.3.2.3 VOS as rightward topicalization 212

10.3.3 Basic clause structures 220

10.4.3 Focus: Exceptive constructions 240

10.4.4 Ko- Topicalization 242

10.4.5 Interim summary 243

10.5 Raising and control 244

10.5.1 The status of ke- clauses 244

10.5.2 “Raising” 248

10.5.2.1 Raising- like verbs and their structures 248

10.5.2.2 What moves in ke- clauses, and where does it

move to? 250

10.5.2.3 What is the nature of the operator in ke- clauses? 255

10.5.2.4 The transparency of finite ke- clauses 260

10.5.3 The verb lava 262

10.5.3.1 Monoclausal structure with lava: Restructuring 263

10.5.3.2 Biclausal structures with lava 266

10.6.2 Reciprocals? Just pluractionality 287

10.6.3 Other binding contexts 290

10.7 Summary 291

11 A paradigm DP- ergative language: Tsez 294

11.1 Tsez basics 294

11.1.1 Preliminaries 294

11.1.2 Unergatives and unaccusatives 297

11.1.3 Clauses with two or more arguments 299

11.2 Discontinuous noun phrases 301

11.3 Non- finite forms 306

11.3.1 Infinitival and masdar clauses 306

11.7 Interim summary 331

11.8 Deriving Tsez clauses 333

11.8.1 Two possible analyses 333

11.8.1.1 A single vP 333

11.8.1.2 Layered functional heads in the verb phrase 339

11.8.2 Single heads or layered structure: Which analysis

Linguists often have to deal with the tension between working on language and working with languages The present manuscript is no exception; I  have tried to account for the range of variation in natural language by looking at a number of individual languages, some of them rather “exotic.” The main subject of this book, ergativity, is still a rather alien topic in linguistic theory I do not expect to settle all theoretical debates pertaining to ergativity within these pages; rather, it is my hope that this work will help to make this unusual phenomenon look less mysterious Peering behind the veil of such mysteries is a driving force of linguistic progress, and one that often results in the languages under consideration seeming less exotic when all is said and done Thus, although I may not arrive at a novel theory, as long as fresh data allow me to reduce the bizarre to the familiar, I will consider my job done.The reality of the presence of ergative languages in our midst compels us to under-stand them better In my work on this book, I have made an effort to obtain detailed data on several ergative languages: Tongan, Niuean, Samoan, Chukchi, Q’anjob’al, Tsez, Archi, Avar, and Circassian The encounter with these particular ergative lan-guages has been critical for my own growth as a linguist, and I consider myself for-tunate to have caught a glimpse of their structure; nevertheless, they represent just

a tiny portion of all ergative languages, and many others still remain a puzzle Quite

a few times during the composition of this work, I have been tempted to write, “We need more data to understand this better”; this need for more comprehensive empir-ical coverage will hopefully push us forward as well

The main idea promoted in this book is very simple There are two types of tive languages; in one type, the ergative expression is a PP in the subject position, and in the other type, the ergative is a DP in a structural case, not very different from the familiar nominative or accusative The two ergative types are characterized

erga-by a cluster of correlated properties, and once these properties are considered all together, each type emerges as internally consistent The result is a simpler, more streamlined syntactic representation of ergativity

This book consists of two parts Part I outlines the general principles that lie the division of ergative languages into two types, with a particular emphasis on PP- ergative languages Part II presents two case studies illustrating the two types of ergative languages: the PP- ergative Tongan and the DP- ergative Tsez

under-Part I  is structured as follows Chapter  1 introduces the notion of ergativity, focusing in particular on syntactic ergativity, the puzzle that serves as the starting

point of my investigation Chapter 2 presents the main proposal advanced in this work: that syntactic ergativity follows as a side effect from the combination of two other linguistic properties: (i) that the ergative in the language in question is a prep-ositional phrase, not a DP, and (ii) that the language does not allow extraction of (or subextraction out of) PPs Chapter 3 identifies several diagnostics that can be used to determine the status of an expression as a PP Chapters 4 and 5 build on these diag-nostics to present arguments in favor of the main proposal and introduce the critical properties of PP- ergative languages Chapter 6 explores further cross- linguistic cor-relates of PP- ergativity that follow from the presence of a PP in the subject position Chapter 7 presents an idealized contrast case— that of a DP- ergative language that exhibits morphological, but not syntactic, ergativity In that chapter, the emphasis

is on the properties that separate PP- ergative languages and DP- ergative languages (Chapter 11, which describes an actual example of a “well- behaved” DP- ergative lan-guage, Tsez, can be read as a companion to Chapter 7.)

Chapter 8 considers the possible diachronic relationship between the two types

of ergative languages; I argue that both types can be traced back to the same source and can therefore be viewed as diachronically related

Chapter 9 compares the proposal advanced in this book with other approaches

to syntactic ergativity, of which there are several These alternatives include criterial freezing of a constituent in the subject position (which can lead to the ban on subject extraction or subextraction), the presence of different licensing restrictions on the absolutive (which may incidentally produce concomitant constraints on the extrac-tion of the ergative), and explanations based on processing

In Part II, chapters 10 and 11 offer a more detailed empirical exploration of two languages that characterize the two types:  chapter  10 examines Tongan, which instantiates the PP- ergative type; chapter 11 outlines the grammar of Tsez, which

is unambiguously DP- ergative Chapter 12 presents my conclusions and addresses several unanswered questions that I  hope will form the basis for future work on ergativity

Many people have helped to make this work possible; none bear responsibility for the final product, but many have been gracious in sharing their time, intuitions, insights, and comments with me The course on ergativity that Anoop Mahajan and

I co- taught at UCLA in 2011 provided an important impetus for this work, and I am grateful to Anoop and all the students in the course for their encouragement I have benefitted from discussions with Edith Aldridge, Ivano Caponigro, Sandy Chung, Jessica Coon, Shin Fukuda, Itziar Laka, Eric Potsdam, Omer Preminger, and Yakov Testelets I  am also indebted to the four anonymous reviewers whose comments were extremely helpful in edging this manuscript closer to completion

I am grateful to David Adger, Judith Aissen, Peter Arkadiev, Jeremy Aron- Dine, Winifred Bauer, Abbas Benmamoun, Jason Brown, Seth Cable, Lauren Eby Clemens, Bernard Comrie, Annabel Cormack, Norbert Corver, Marcel den Dikken, David Erschler, Nomi Erteschik- Shir, Anamaria Falaus, Grant Goodall, Boris Harizanov, Robert Henderson, Vincent Homer, Caitlin Keenan, Roni Katzir, Hilda Koopman, Julie Legate, Beth Levin, Nick Longenbaugh, Diane Massam, Pedro Mateo Pedro, Jason Merchant, Adam Milton Morgan, Léa Nash, Yuko Otsuka, Hazel Pearson,

Nina Radkevich, Luigi Rizzi, Jerry Sadock, Bridget Samuels, Peter Sells, Dominique Sportiche, Dieter Wunderlich, and Colin Zwanziger for helpful discussions of this work at its various stages I am sorry that I was unable to follow up on all the excel-lent suggestions I received.

Empirical data collection for this book was supported in part by NSF grants BCS- 0131993, BCS- 0231946, BCS– 1144223, BCS- 137274, and BCS- 1414318; by the Faculty of Arts and Sciences at Harvard University; by the David Rockefeller Center for Latin American Studies at Harvard; by the Davis Center for Eastern European Studies at Harvard; by the Center for Advanced Study of Language (CASL)

at the University of Maryland; and by the Max- Planck Institute for Evolutionary Anthropology in Leipzig

Unless noted otherwise, the language examples come from my own fieldwork I am indebted to the native speakers who have shared their language data with me: Raxmet Gisheva, Raxmet Esheva, Mira Unarokova, and Miriam Djakoe (Adyghe); Aminat Eldarova, Kerim Kerimov, Madjid Khalilov, Djamilya Magomedova, and Magomed

I. Magomedov (Avar); Peter Inenlikey and Vladimir Raxtilin (Chukchi); Eka Egutia, David Erschler, Maya Iashvili, and Léa Nash (Georgian); Archna Bhatia, Ashwini Deo, and Gyanam and Anoop Mahajan (Hindi); Ana Lopez de Mateo (Kaqchiqel); Tom Etuata, Pefi King, Tammi King, Mele Nemaia, Pat and Granby Siakimotu, Ligi Sisikefu, Lynsey Talagi, and Kara- Ann Tukuitonga (Niuean); Pedro Mateo Pedro (Q’anjob’al); John Frujean and Lotu Sili (Samoan); Kaufo’ou Faletau, Sisilia Lutui, Saia Mataele, Sofia Tolu, and Melenaite Taumoefolau (Tongan); and Arsen Abdulaev, Madjid Khalilov, Paxruddin Magomedinov, and Ramazan Rajabov (Tsez) Without their help, this work would not have been possible

And finally, many thanks to my family for keeping me sane and for having such

a great sense of humor What’s in this book is unlikely to enthrall you, but it is for you— with love

The abbreviations in the glosses follow the Leipzig Glossing Rules

A agent- like argument of canonical transitive verbABS absolutive

Deconstructing Ergativity

Two types of ergatives

intran-is referred to as “accusative” (1a); if the intransitive subject and direct object share encoding (ABS; absolutive case) to the exclusion of the transitive subject (ERG; erga-tive case), then the alignment pattern is referred to as “ergative” (1b) (Comrie 1978; Dixon 1979, 1994; Manning 1996; Aldridge 2008; McGregor 2009; Palancar 2009; among others) These alignments can be expressed not only in case marking but also

in agreement; agreement can group S and A together, in contrast to O, or it can group

S and O together, in contrast to A

(1) a Accusative b Ergative

The realization of ergative alignment in case marking and/ or agreement results

in what is known as “morphological ergativity.” Observing these surface differences

in alignment leads us to the following question: Does alignment have deep syntactic repercussions, or is it only skin- deep, with simple morphological reflexes?

The initial answer to this question is that differences in alignment seem ited to surface morphology Although ergative languages may appear exotic at first glance to those of us who speak nominative- accusative languages, long- standing research has shown them to be less alien than one might expect, especially with respect to familiar structural properties Starting with Anderson’s seminal work

lim-(S Anderson 1973, 1976, 1977), researchers have noted that the ergative DP is structurally superior to the absolutive DP This asymmetry manifests itself in sev-eral ways For instance, it is the ergative DP, not the absolutive DP, that corresponds

to the null pronominal (addressee) of the imperative in (2); this pattern is similar to the pattern observed in accusative languages, where the nominative (not the accu-sative) corresponds to the addressee of the imperative It is also the ergative argu-ment of a transitive verb, rather than the absolutive, that can undergo raising and corefer with the silent subject in control structures (3) Compare illustrative exam-ples from two Nakh- Dagestanian languages spoken in the Northeast Caucasus:

‘Read your letter!’

‘The boy has to chop firewood.’ (Khalilova 2009: 383)

Furthermore, the ergative binds the absolutive, but not the other way around The following example from Tsaxur illustrates this with a monomorphemic reflexive.1 As the examples below show, the order of elements does not affect the interpretation

‘Rasul beat himself up.’

(‘Rasul beat himself up.’)

(‘Rasul beat himself up.’) (Kibrik and Testelets 1999: 346)

In ergative languages that allow multiple wh- fronting and show superiority effects, these superiority effects are just like those found in English This effect is illustrated

by the following examples from Basque (Ortiz de Urbina 1989; Reglero 2004; Jeong 2004; 2007) As in more familiar languages, when a clause contains two wh- words, the one that undergoes wh- movement is the one closest to the interrogative C.2

1 Here and below, it is important to differentiate the binding of monomorphemic ives from the binding of complex reflexives (cf Reinhart and Reuland 1993) The latter often have logophoric properties (cf Reinhart and Reuland 1993; Reuland 2011); the pres-ence of these properties makes complex reflexives a weaker test case, since a logophoric center can take scope over several clauses

reflex-2 The Basque facts are only relevant if we adopt the view that Basque has wh- movement

An alternative, explored by Arregi (2002), is that it has wh- in- situ with pied- piping

(5) a Nork nori zer eman dio? Basque

‘Who gave what to whom?’

(‘Who gave what to whom?’)

Accusative and ergative languages are also alike in having multiple verb- object idioms (Verb+ACCOBJECT in accusative languages, Verb+ABSOBJECT in ergative ones) but few subject idioms (Manning 1996; Ura 2000) For instance, English idiomatic expressions with objects are plentiful, and their number keeps growing, with expres-

sions such as talk smack “insult” or jump the shark “decline (of a show, brand, etc.)”

added on a regular basis At the same time, English has just a handful of subject-

oriented idioms, such as the shit hit the fan, the pot (is) calling the kettle black, cat got

your tongue, and the cat is out of the bag Tsez, similarly, exhibits scores of object

idi-oms Even though the sources available for Tsez are not as extensive as the English Urban Dictionary, a relatively modest collection of 7,500 lexical items presents ten

idioms with the word rok’u ‘heart’ in object position (Xalilov 1999: 220) but only one

with that word in subject position:3

dem.i- gen heart.abs.iv down iv- go- pst.nwit

‘He calmed down.’ (lit.: his heart went down)

In addition, ergative and accusative languages pattern alike in permitting true incorporation or pseudo- incorporation of objects (and sometimes the sub-jects of unaccusatives) Compare the illustrations of pseudo- incorporation in two Polynesian languages— accusative Māori and ergative Niuean— in (7)  and (8)  below; note that incorporation in Niuean is morphologically visible, in the sense that the case of the subject changes from ergative to absolutive (i.e., incor-poration results in detransitivization) Crucially, neither alignment type allows the incorporation of a transitive subject, be it nominative or ergative It is well known that the incorporation of transitive nominative subjects is impossible (Baker 1988); we again observe a parallel between the nominative subject of a transitive and the ergative argument

3 In Tsez glosses, the Roman numeral after a noun indicates the gender (grammatical class) of that noun The verb agrees in gender with the absolutive

(7) a E mahi ana te iwi i te kai Māori

‘The tribe was procuring (the) food.’

‘The tribe was procuring food.’

(‘The tribe was procuring food.’)

‘(The) cats eat the/ a chicken.’

‘(The) cats eat chickens.’

(‘(The) cats eat the/ a chicken.’)

OK in the unintended meaning “The bird eats cats.”

The properties presented above all serve to identify syntactic subjects as ally distinct from other grammatical elements — although researchers do not always agree on the range of the relevant properties or their relative importance For exam-ple, some researchers have argued that imperative addressees and incorporated/ idiomatic NPs are potentially identifiable on a semantic basis.6 To avoid potential controversy, I will not be relying very heavily on these particular properties; there are other unambiguous diagnostics identifying the highest structural argument of a clause In fact, even if we exclude some of the properties discussed above, the erga-tive argument still exhibits plenty of standard properties associated with syntactic subjects This finding suggests that ergativity is only skin- deep: a simple morpholog-ical difference in the encoding of subjects and objects

behavior-Nevertheless, there are also systematic differences between languages with ergative case- marking and those with accusative case- marking The two main areas of dissimilarity have to do with agreement and A- bar movement.7 The

4 See also Bauer (1993: 478– 480)

5 See Seiter (1980: 69– 77); Massam (2001) for a detailed discussion of Niuean pseudo- noun incorporation, and see more discussion of this property in Chapter 8

6 See Dixon 1979, 1994 on the idea that the addressee of the imperative is determined

on the basis of a semantic diagnostic; see Kiparsky 1997 on the semantic conditions for forming idioms

7 In addition, there has been lingering uncertainty concerning the possibility of tactic ergativity in Dyirbal (Dixon 1972, 1979, 1994) If Dyirbal is syntactically ergative

syn-in a profound way, it presents a serious contrast to all other known languages (syn-includ-ing other ergative languages) Some researchers have proposed alternative accounts of

(includ-agreement asymmetry is as follows:  languages with ergative case- marking can have ergative or accusative alignment in agreement, but languages with accu-sative case- marking never show ergative alignment in agreement (S Anderson

1976, 1977, 1984; Comrie 1978; Dixon 1979, 1994; among others) Thus, ergative languages impose fewer restrictions on their agreement alignment than accusa-tive languages do This discrepancy has attracted a substantial amount of atten-tion (cf Bobaljik 2008; Preminger 2014; Režać 2011; see also Corbett 2006; Deal 2010; Markman and Graschenkov 2012), and I  will not discuss it here in great detail (although it is tangentially connected to the main issues I  will address throughout the rest of this work) The second area of significant divergence has to

do with displacement under A- bar movement: this is the major puzzle examined

or object, can undergo A- bar movement

Consider the following Tongan examples, which illustrate syntactic ergativity under relativization Example (9a) is a baseline transitive clause; in (9b), the absolu-tive object of that clause is extracted, leaving a gap at the base position (here and below, unless I am discussing a particular syntactic analysis, I indicate the gap at the extraction site atheoretically as “_ _ ” with a subscript indicating co- indexation) Extraction that produces a gapped structure for an ergative argument is impossible; instead, the extraction position must be occupied by a resumptive pronoun which is cliticized to the tense marker (9c):

Dyirbal that rule out syntactic ergativity and thus make this language look more rable to the other six thousand– odd languages of the world (Jake 1978; Polinskaja 1989; Legate 2008b)

compa-8 Throughout this book, I am using the term relativization only in relation to the

forma-tion of restrictive relative clauses Descriptive (appositive) relative clauses, such as the English (i) below, may be subject to different principles, and even more crucially, not all languages have such relative clauses

(i) The major greenhouse gases are water vapor, which causes about 36– 70% of the

warming”)

(9) a.  baseline transitive sentence

‘Oku fakamolemole‘i ‘e Mele ‘a e kaiha‘a Tongan

prs forgive erg Mary abs det thief

‘Mary forgives the thief.’

b.  ABS object (must relativize with a gap at the extraction site)

e kaiha‘ai [‘oku fakamolemole‘i ‘e Mele _ _ i/ *‘a ia]

det thief prs forgive erg Mary abs 3sg

‘the thief that Mary forgives’

c.  ERG subject (must leave a resumptive pronoun in the relative clause)

e ta‘ahine i [‘oku *(ne i ) fakamolemole‘i _ _ i ‘a e kaiha‘a]

det girl prs rp forgive abs det thief

‘the girl that forgives the thief’

(10) a  baseline intransitive sentence

 ‘Oku tangi ‘a Mele.

 prs cry abs Mary

 ‘Mary is crying.’

b  ABS subject (must relativize with a gap at the extraction site)

 e ta‘ahinei [‘oku (*ne) tangi _ _ _ i/ *‘a ia ]

 det girl prs rp cry abs 3sg

 ‘the girl that is crying’

Likewise, a resumptive pronoun is required in Tongan if the ergative is topicalized or appears as the pivot of a cleft (see also Chung 1978: 226– 229; Otsuka 2000; and see further discussion in chapter 10 of this book)

Contrast this pattern with the pattern of extraction observed in Basque, where both the ergative subject and the absolutive object can leave a gap at the extrac-tion site Thus, the two arguments extract the same way, and there is no contrast between them:

‘The men built this house.’

b ABS object (must relativize with a gap at the extraction site)

‘the house that the men built’

c.ERG subject (must relativize with a gap at the extraction site)

‘the men that built the house’

9 Basque exhibits homophony between the plural absolutive, shown in (11c), and the singular ergative, shown in (11a, b)

(12) a.  baseline intransitive sentence

‘The boy has come to school.’

b  ABS subject (must relativize with a gap at the extraction site)

[_ _ _ i eskola- ra etorri zuten] mutil- a- ∅i

school- to come aux boy- det- abs

‘the boy that came to school’

For the purposes of this book, I will be referring to this inaccessibility to extraction

on the part of the ergative argument as syntactic ergativity. Thus,

(13) Syntactic ergativity: the inaccessibility of ergative arguments to A- bar movement

with a gap in the base position, as contrasted with the accessibility of absolutive arguments to such movement

There are other conceptions of syntactic ergativity, some more inclusive than the one adopted here I will review them later on in this chapter, but for now suffice it to state that I am adopting a more restrictive view of syntactic ergativity than is found

in some other studies

The examples I have just offered all compare the absolutive and ergative under relativization, which is just one type of A- bar movement In fact, although rela-tivization, focusing, wh- question formation, topicalization, and tough move-ment are all types of A- bar movement, the ergative argument does not always behave identically in each of those constructions, even within a single ergative language There are several reasons for this variation For instance, wh- questions and focus structures are not always formed by movement; they can be formed with the questioned or focused elements in situ, in which case their syntax is differ-ent from the syntax of relative clauses and scrambled topics Yet another strategy

is to form wh- questions or focus constructions by clefting or pseudo- clefting, as illustrated below:

(Pseudo- )clefts are biclausal structures that include a relative clause (cf the CP in (14)); therefore, we should in principle expect to observe the same constraints under (pseudo- )clefting as we do under relativization However, the relative clauses used

in clefts are not necessarily equivalent to those found in noun- modifying relative clauses; it is possible that a language could use movement for headed relatives and a non- movement strategy for headless relatives In that case, wh- questions and focus expressions could retain their freedom from syntactic ergativity

Topicalization and other processes of left dislocation, too, need not necessarily involve movement Extensive research on left dislocation shows that it is not a uni-form phenomenon In particular, some left- dislocation processes rely on movement

structures, in which an element literally dislocated from a clause- internal tion, while others rely on base- generation structures, in which the “left- dislocated” element is base- generated in its peripheral position and linked to its clause- internal position via interpretive mechanisms; no movement is involved Both left- dislocation strategies are observed in English (cf Prince 1981; Gregory and Michaelis 2001 for examples) In the examples below, the initial constituent in (15)

posi-is viewed as dposi-islocated from its base position as a result of A- bar movement In (16), however, this initial constituent is base- generated in its left- periphery position and

is linked to its clause- internal position via interpretive mechanisms that are much freer than the mechanisms operating in syntax In principle, any element can be base- generated as an external topic and co- indexed with clause- internal material lower in the structure No difference in alignment is expected in such constructions Based on this discussion, if a language has only left- dislocation and no topicaliza-tion, there is no expectation of syntactic ergativity (or accusativity) in that domain

Additionally, even in those languages where all the relevant A- bar processes tivization; wh- question formation; topicalization; tough movement) are subject

(rela-to movement, there may be a difference in the nature of the movement opera(rela-tor For instance, it is often the case that wh- questions and focus constructions have

an overt movement operator, while the operator for relative clauses is covert; this system is observed particularly often in head- final languages (a common headedness type among ergative languages) I will return to the implications of differences in the phonetic content of movement operators in subsequent chapters As a descriptive observation, we can conclude that relativization is the most reliable test of syntactic ergativity, since it allows for cross- linguistic comparison even when the behaviors

of ergative DPs in other A- bar processes are not uniform Furthermore, since some constructions that are in principle representative of A- bar movement may be based

on relativization, it appears that relativization allows us to start small and expand if necessary In the discussion below, I will be using the terms “extraction” and “A- bar movement” to describe instances where it is possible to generalize across different A- bar movement processes; I will address discrepancies between relativization and other A- bar movement phenomena where they are known to occur.10

10 Before moving on, I will to raise briefly another issue relevant to syntactically tive languages Asking questions about an event participant and forming a relative clause that restricts a set of entities are indispensable tasks of communication However, a syn-tactically ergative language cannot use the readily available question templates to ask

erga-“Who did that?” or readily available relative clause templates to express “This is the rat that ate the cheese,” as these structures require extraction of the ergative argument Therefore,

in order to perform these necessary linguistic operations, syntactically ergative languages must find ways of circumventing the ban on the extraction of the ergative In the appendix

to this chapter, I present a short summary of different options available to syntactically

1.2.2 The range of the phenomenon

How common is syntactic ergativity? The answer to this question is inevitably tative, since we know less about ergative languages than we do about nominative- accusative languages and our sample of ergative languages is relatively small Still,

ten-it is possible to offer some inten-itial considerations A cursory look at morphologically ergative languages shows that, within this domain, the phenomenon of syntactic ergativity is widespread: in the WALS sample of thirty- two morphologically erga-tive languages, Comrie (2008) and Comrie and Kuteva (2008) list only twelve lan-guages that allow subject relativization with a gap:  Bawm, Burushaski, Chukchi, Lower Grand Valley Dani, West Greenlandic, Hunzib, Ingush, Lezgian, Tukang Besi, Ngiyambaa, Pitjantjatjara, and Wardaman The notion of subject extraction sub-sumes extraction of the absolutive (intransitive subjects) and ergative (transitive subjects) However, extraction of the ergative with a gap happens in only a subset

of these languages.11

Upon closer examination of these thirty- two languages, the list of candidates for genuine ergative extraction (with a gap left at the extraction site) shrinks further

As I will show later on, Chukchi does not allow direct extraction of the ergative with

a gap; instead, it requires that the clause be detransitivized (via antipassivization) before extraction of the subject can take place After detransitivization, the DP to

be extracted is no longer in the ergative case but in the absolutive, and can therefore leave a gap Lower Grand Valley Dani and West Greenlandic also require antipassiv-ization as a way station for subject extraction with a gap.12

The Austronesian language Tukang Besi follows the subject- only restriction

commonly found in the Austronesian family, regardless of alignment Under this restriction, the only DP argument that can be extracted by A- bar movement is the structurally highest one (sometimes referred to as the “external argument,”13

“pivot,” or “trigger”); different light verbs (or voice projections) serve to promote a particular argument to the highest structural position The subject- only restriction

is widespread in Austronesian: it occurs, for instance, in Malagasy, Philippine guages, Formosan languages, languages of Indonesia (where Tukang Besi is spoken), and some Oceanic languages (see Wechsler and Arka 1998; Aldridge 2004, 2008; and, for overviews, Gärtner et al 2006; Chung and Polinsky 2009) Depending on which DP serves as the trigger, the form of the verb has to change, assuming what

lan-ergative languages to work around the impossibility of A- bar movement of the lan-ergative The problem cannot be solved, but it can be avoided

11 Note, however, that for some languages the presence or absence of the relevant ject extraction feature may simply not be noted in WALS (Dryer and Haspelmath 2008),

sub-or the infsub-ormation therein may be based on incomplete data

12 See the appendix to this chapter and Polinsky (forthcoming- b) for a more detailed discussion of antipassivization

13 “External argument” is perhaps not the best term for this item, because it implies that the argument is somehow “externalized,” often with some restructuring of the argu-ment structure of the verb This externalization does not necessarily take place In discuss-ing the relevant data below, I will be using the term “trigger.”

some researchers call “voice” and others call “argument- topic” (agent- topic, theme- topic, etc.) form Compare the following minimal pair (with some glosses changed from the original presentation) In (17a), the DP “child” appears as the trigger,

marked with the nominative na, and is the only argument accessible to A- bar

move-ment operations The verb appears in what seems to be the agent- topic (AT) form,

without special marking, and the non- nominative marker te cliticizes to that form

In (17b), where the verb is in a different form, presumably theme- topic (TT), the trigger is the DP “friend,” which is now the sole argument that can undergo A- bar movement

‘The child saw his friend.’

‘The child saw his friend.’ (~ “The friend was seen by the child.”)

In quite a few languages, including Tukang Besi, the trigger (the structurally est argument) is subject to semantic constraints that more properly define topics than subjects For example, this argument must be specific and/ or definite Donohue (1999: 467– 468) provides examples indicating that Tukang Besi follows the subject- only restriction Independent of this, the status of morphological ergativity in Tukang Besi is controversial The morphological ergativity of Bawm, another lan-guage on the WALS list, is similarly in dispute

high-When all the accounting is done, we are left with only seven languages that are clearly free of syntactic ergativity: Hunzib, Ingush, Lezgian, Burushaski, Ngiyambaa, Pitjantjatjara, and Wardaman These languages are not distributed randomly across the globe Five of them belong to one of two language families: Nakh- Dagestanian (Hunzib, Ingush, Lezgian) and Pama- Nyungan (Ngiyambaa, Pitjantjatjara); Wardaman is a non- Pama- Nyungan Australian language.14 Burushaski is an isolate surrounded by the Indic languages of Pakistan and the Tibetan languages of China and northern Kashmir At least some varieties (e.g., Nagar Burushaski, Srinagar Burushaski) reflect Urdu influence, including the introduction of relativizing ele-ments (Gregory Anderson, pers comm.)

Up to this point, I  have focused my attention on identifying those languages, reported in WALS and beyond, that have syntactic ergativity One wonders, however,

14 In building his sample for WALS, Comrie does not count Basque among tive languages He characterizes it as an active- inactive language, where some intransi-tive subjects (S) can appear in the ergative, and others in the absolutive Comrie also excludes Hindi from the list of ergative languages because he characterizes its align-ment as tripartite (separate marking for S, A, and O) Additionally, one could include some other ergative languages excluded by the design of WALS, but the overall pattern remains the same

erga-if we should make an effort to correct errors in the reporting of patterns that go the other way; it may be that there are morphologically ergative languages that are not reported to have gaps in subject position, but on closer examination actually do At least one candidate comes immediately to mind: Shipibo is morphologically erga-tive but clearly not syntactically ergative (Valenzuela 2002; Baker 2014) It is a valu-able addition to the geographical distribution of languages that are not syntactically ergative, as it is spoken in Peru and does not represent either of the two geographi-cal areas discussed above It also offers a cautionary note: we do not have sufficient information on the extraction patterns in many languages to make definition state-ments, so any judgment calls about the relative frequency of certain patterns over others are necessarily preliminary.

Nevertheless, let us pause here and draw a comparison between the ergative and the accusative case In the next section, I will discuss the rationale for this compar-ison; for now, let us just take it for granted The salient comparative fact is as fol-lows: if we were to select a random sample of thirty- plus accusative languages from across the world, the number of those languages that showed restrictions on the extraction of the accusative would be very small, and such languages would in gen-eral be hard to find This discrepancy alone makes the pervasive nature of syntactic ergativity puzzling The final numbers may not be identical to what I report here, but there is clearly a sizeable cohort of morphologically ergative languages that behave like Tongan rather than Basque: they allow the extraction of the absolutive with a gap, but their ergative argument is inaccessible to A- bar movement in general, or at least to some subtypes of A- bar movement

1.2.3 The relevance of syntactic ergativity

The contrast between the absolutive and the ergative with respect to A- bar ment constitutes the puzzle of syntactic ergativity:  What prevents the ergative from being extracted with a gap, despite the fact that it shows subject properties with respect to other diagnostics? This is a particularly vexing question given the well- known accessibility hierarchy of relativization, which predicts that subjects should undergo relativization with a gap (Keenan and Comrie 1977, slightly modi-fied below):

arti-languages as Warlpiri or Hindi as allowing relativization of ergatives Second, they discuss Dyirbal’s syntactic ergativity and propose a passive analysis of that language

in which the ergative is essentially a by- phrase, hence a low- ranking oblique object

on (18) Their approach to Tongan is similar to their approach to Dyirbal; they pose that the Tongan ergative is also a passive agent (i.e., a by- phrase) Curiously, they justify this claim on the basis of the Tongan ergative’s historical status rather than its synchrony (Keenan and Comrie 1977: 87– 88), yet they somehow project this historical status into the synchronic properties of the ergative with respect to rela-tivization With the present availability of more extensive data, including data from more ergative languages, the fact that ergative arguments are broadly unable to undergo relativization in the same way that absolutive subjects do remains a prob-lem for the Accessibility Hierarchy

pro-To recapitulate, syntactic ergativity is puzzling in light of the ergative case’s subject- like properties; these subject- like properties manifest under most diagnos-tics, but not under A- bar movement, contrary to the predictions of the Accessibility Hierarchy In addition, syntactic ergativity is important because it underscores the asymmetry between the ergative and accusative cases, something I  have already alluded to in the preceding section These two cases appear to be very much alike, but they are also very different

The reasons for the (superficial) similarity between the ergative and accusative are at least twofold First, ergative and accusative cases tend to be morphologically more marked than the other cases within their alignment (i.e., ergative is typically more marked than absolutive, while accusative is typically more marked than nomi-native) One of the ways in which this relative markedness manifests itself is in the fact that the ergative and the accusative tend to be marked overtly, while the absolu-tive and the nominative often appear with zero marking in the case paradigm.16

Second, there is a general intuition that, in many situations, the ergative and the accusative can only appear in a clause that already contains another case; for exam-ple, for the ergative to appear, there has to be an absolutive in the same clause (Yip

et al 1987; Marantz 1991; Bobaljik 1993; Bittner and Hale 1996a, b; Woolford 2000,

15 Hindi, however, uses correlatives, so its relativization facts are irrelevant for ing syntactic ergativity

assess-16 It is worth noting, however, that this is only a tendency There are languages where both absolutive and ergative are marked (e.g., Chukchi, Adyghe, Kabardian), and languages where the opposite pattern holds:  the ergative is unmarked and the absolu-tive has an overt morphological exponent For instance, in several Australian languages (Warluwara, Pitjantjatjara, Yulbaridja, Gugada), animate nouns used as intransitive sub-

jects have to occur with the suffix - nya (historically, the accusative marker) The same case

is used in the object position of a transitive clause, and the unmarked case continues to

be used as the transitive subject (Blake 1977: 13) The resulting case alignment is ergative,

but with unmarked ergative case: the ergative is null, and the absolutive ends in - nya Still

another instance of an unmarked ergative and marked absolutive may be found in Nias Selatan (Brown 2001, 2005), although Crysmann (2009) argues against Brown’s analysis that this unmarked form is ergative Such a reversal of markedness is not unique to erga-tive languages; for example, in Oromo, the nominative is unmarked and the accusative marked (Owens 1985)

2006; Legate 2008a; Merchant 2009; Deal 2010; among others) This intuition has

been explicitly formalized in the idea of dependent case, developed by Marantz (1991,

2000) Informally, the notion behind dependent case is that the presence of case

X is generally dependent on the availability of another case within the same case- assigning domain Thus, the presence of the ergative depends on the availability of the absolutive, but not vice versa; likewise, the presence of the accusative depends

on the availability of the nominative The main idea behind this proposal is that case realization is subject to a number of different constraints and follows a markedness hierarchy of the following nature:

(19) lexically specified (“quirky”) case > dependent case > unmarked (obligatory) case

For now, let us just note that, for both the ergative and accusative cases, their ment is contingent on the presence of another case With this in mind, let us return

assign-to our putative sample of thirty- something morphologically accusative languages, the parallel to the WALS sample of ergative languages discussed above Within the accu-sative sample, it will be hard to find languages that do not allow relativization of the accusative with a gap in the base position The only type of accusative language that restricts object relativization is the subject- only type— but as I already mentioned, subject- only languages are in a class by themselves If we exclude them, we find that the accusative, although it is a dependent case just as the ergative, freely extracts with

a gap We are thus dealt a puzzle: Why is A- bar extraction of the ergative severely restricted, while extraction of the accusative is relatively free? This conundrum takes

us back to the puzzle of syntactic ergativity

In sum, syntactic ergativity matters because it highlights an unexpected tion on subjects, which are otherwise favored by language design in terms of acces-sibility to the syntactic process of extraction It matters because it seems widespread among morphologically ergative languages And it matters because it offers an intriguing contrast between two dependent cases, the ergative and the accusative

restric-1.3 THE IMPORTANCE OF STARTING SMALL

1.3.1 Syntactic ergativity broadly def ined

As I already mentioned, the interpretation of syntactic ergativity adopted for the purposes is this book is restricted, and refers specifically to the inability of the erga-tive argument to be extracted with a gap under A- bar movement Other researchers view syntactic ergativity as a broad phenomenon, so an explanation for my adoption

of a notably narrower understanding of this phenomenon is in order

Several researchers, most notably Konstantin Kazenin (1994) and Christopher Manning (1996), include a wide range of contrasts under the rubric of syntactic erga-tivity: the contrast between absolutive arguments (S and O) and the ergative argu-ment with respect to A- bar movement, coreference across clauses, scope, binding, quantifier float, raising, control, and possibly other dependencies Based on these criteria, a number of languages could be characterized either as fully syntactically

ergative or as “mixed-pivot” languages (Manning 1996) It seems that this more sive approach to syntactic ergativity relies on two particular assumptions First, sub-jects are uniformly defined by a cluster of properties that are responsible for the more

inclu-“active” status of subject in syntactic processes (as compared to more “inert” non- subjects) The corresponding properties do not necessarily have to be uniform, as long

as the tally of what subjects can do is greater than the tally of what non- subjects can

do Second, the absence of a constituent from a particular position (say, a gap in a tive clause or the missing subject of a clause conjoined with another clause) follows from a coherent set of underlying principles, with the result that all such absences are similar enough to warrant comparison In other words, the following structures are viewed as qualitatively similar with respect to their silent elements:

deletion under coreference

(purpose clause)

Manning (1996: 34) suggests that the three structures in (20) through (22) low either the ergative pattern or the neutral pattern (where the absolutive and the ergative arguments are treated alike) Furthermore, they seem to be arranged hier-archically; for instance, the presence of syntactic ergativity under relativization is more likely than the presence of syntactic ergativity under conjunction reduction and purpose- clause formation Thus, a language that exhibits syntactic ergativity under purpose- clause formation should also exhibit it under conjunction reduction and relativization, etc

fol-(23) Cline of syntactic ergativity (Kazenin 1994; Manning 1996)

relativization > conjunction reduction > purpose- clause formation

The problem with such a hierarchy is that it does not consider the possibility that the reasons for deletion of a constituent may vary across the three structures One can reject the machinery of a particular theory and its concomitant specific theoretical constructs— such as traces or null pronominals— while accepting that the silent elements in control clauses, relative clauses, and conjoined clauses have systematically different properties Examples abound, and I limit the discussion here to a single illustration pertaining to silent subjects in Spanish embedded clauses

Consider the paradigm in (24), below (see also Freidin 2007: 204) In all three

sentences in (24), Juan is the subject of the matrix clause However, the rules of

core-ference in each clause are quite different, and clearly depend on the matrix verb as well as on the distance between the possible antecedent and the gap Furthermore, only in (24b) and in the lower clause of (24c) can the empty category alternate with

an overt pronoun without changing the interpretation

(24) a Juani quiere [que _ _ *i/ k se vaya]. Spanish

Juan wants comp refl go

‘Juan wants for you/ for someone else to go.’

NOT: ‘Juan wants to go.’

b Juan i no cree [que _ _ i/ k se vaya].

Juan neg believes comp refl go

‘Juan does not believe that he is going/ you are going/ someone is going.’

c Juani no quiere [que _ _ *i/ k crea [que i/ k se vaya]].

Juan neg wants comp believe comp refl go

‘Juan does not want you/ someone else to believe that he/ you/ someone else will go.’

Given the significant observed differences between different types of silent ments, it is unclear how much can be gained by collapsing all of these elements into one group Certainly, we stand to lose important structural distinctions when we dis-regard the smaller details of individual structures and of different types of silence In admitting that not all types of silence are the same, we recognize that some types of silence are more constrained in terms of their distribution and interpretation— for example, the silent element in the Spanish example in (24a) is more constrained than the silent element in (24b) If not all types of silence are created equal, compar-ing them may not always be straightforward

ele-In generative grammar, and in minimalism in particular, extraction is accounted for under Move (or, strictly speaking, under copying or internal re- Merge of an already merged syntactic object); together with Agree and external Merge, Move is part of the narrow syntax (for arguments, see Safir 2008, among others) The output

of narrow syntax is then enriched by further vocabulary to fully represent logical structure It’s within this logical structure that operations such as scope- reading and operator- variable binding occur (Reuland 2011) Coreference across clauses is part of

an even broader system— that of discourse The discourse component of the mar situates the logical syntax in a larger context that includes world knowledge, speaker intent, and the full linguistic context Discourse is where reference relations are established; it includes coreference relations across clauses and sentences, and possibly reference of the imperative addressee On this approach, generalizations related to A- bar movement pertain to narrow syntax, as do generalizations related

gram-to obligagram-tory control and raising (although these two phenomena do not involve A- bar movement) What happens in conjunction reduction pertains in large measure

to discourse; together, these facts argue for treating conjunction reduction effects separately from movement effects.17

The hierarchical relationship between structure- building mechanisms ing extraction), on the one hand, and logical- syntactic and discourse relation-ships, on the other, gives us a natural way of dividing the dependencies discussed

(includ-17 In functional approaches to syntax, all types of silence (if they are allowed at all) are represented along a cline; this amounts to treating all the variants of silence as stemming from the same source, with their differences being determined by specific lexical verbs

here; the boundary between extraction (A- bar movement) and A- movement, binding, and various anaphoric dependencies related to binding is not ad hoc Some of these dependencies— most notably coreference across clauses— belong

to a part of language that lies outside the purview of syntax and concerns the course integration of language material These facts alone call for a more restric-tive view of syntactic ergativity, which excludes these phenomena as diagnostics for syntactic ergativity In addition, as I will show below, the division of erga-tive languages into two types based on their A- bar movement properties actually allows us to account for a larger cluster of syntactic dependencies in a principled way In other words, what has been presented as a series of implicational proper-ties by other researchers, as in (23), can be shown to follow from a single contrast between two types of ergative languages, differing in the nature of their ergative expression

dis-The result is a more predictable, uniform syntax, which is always desirable If the cost of that predictability involves giving up some of what is traditionally assumed to

be part of syntax, that does not seem to be an unwelcome trade- off

Accordingly, narrowing down the domain in which syntactic ergativity is ated allows us to treat (with certainty) cross- linguistic instances of syntactic erga-tivity as realizations of the same phenomenon, not different phenomena whose connection to each other may not be entirely clear or apparent By starting small, we can maintain the explanatory power of our observations

evalu-Limiting the range of phenomena under consideration allows us to arrive at generalizations drawn from the same grammatical domain But in addition to that, there is also a more practical consideration for starting small, which has to do with the accuracy of cross- linguistic comparisons In a broad cross- linguistic compari-son, it is not uncommon to compare phenomena from several languages even if we are not certain how similar these phenomena are In these cases, we run the risk of

or constructions That, in turn, entails the possibility that a particular instantiation of silence (“deletion” of an element) may display properties associated with different con-texts A cline of properties, such as the one shown in (23), is therefore useful in capturing the facts as they are observed, but, crucially, it also allows for a mixture of properties— for instance, if a silent element is observed in a purpose clause, its properties might show some of the signatures of a silent element in a relative clause, plus additional purpose- clause- specific properties The result is an absence of discrete categories While this approach has benefits, failure to recognize discrete categories (including discrete types

of silent elements) prevents us from establishing a priori a set of features that determine how a given category enters into a relationship with other categories This, in turn, could lead to missed generalizations

In unification- based frameworks, it is common to distinguish between grammatical tions and thematic (argument) structure, with relativization, wh- question formation, and topicalization being representative of the former and control (purpose) clauses and bind-ing representative of the latter (it is not entire clear where conjunction reduction belongs)

rela-A cline such as in (23) is useful in that it treats phenomena that are linked to grammatical relations as more constrained than phenomena associated with argument structure Put differently, there is an expectation that cross- linguistic variation is more likely to occur at the level of grammatical relations than at the level of argument structure

comparing structures whose similarities are only superficial Purpose clauses are

a particularly telling case; in some languages, they are always non- finite, in ers, they are always finite, and for quite a few languages, they can appear in both structures Without taking finiteness into consideration, it is difficult to compare the types And if finiteness is taken into consideration, the cline in (23) needs to

oth-be revised

To offer another illustration, consider coordination: how clear can it be, from a cursory survey of cross- linguistic data, that a particular instance of coordination represents coordination of clauses as opposed to coordination of verb phrases? Clausal coordination and VP- coordination are often difficult to tell apart without

an extensive language- particular analysis The difference is not always clear even for such well- studied languages as English; consider examples such as (25a, b), for which one may need to posit IPs at a lower level of representation or posit various types

of null categories in the second conjunct (Van Valin 1986; Goodall 1987; Burton and Grimshaw 1992; McNally 1992; among others):

(25) a The criminal will be arrested and will confess to the crime

(Burton and Grimshaw 1992: 310)

b The employees complained and were given more vacation time

(McNally 1992: 336)

Along the same lines, consider the ambiguity in the following sentence; the rial deleted in each of the readings is indicated by strikethrough

(i) Must she make John go to the dentist and will she make John go

to the dentist?

(ii) Must John go to the dentist and will she make John go to the dentist?

Under reading (i), John is construed as the object in both instances; under ing (ii), John is the subject of the event of going and the object of her causing him

read-to go (Wilder 1997: 92) In (i), the two constituents joined by and are verb phrases,

which share the same subject Under reading (ii), the two constituents are clauses, not verb phrases

If we cannot definitively determine what happens in English, how can we be sure whether a paratactic combination of two clauses in a lesser- known language, avail-able only from a grammatical description, stands for conjunction reduction?

Next, conjunction reduction and reduction in purposive clauses are not always easy to tell apart without an in- depth linguistic analysis Imagine a language— let’s

call it English- 1— which has ergativity, pro- drop, and the functional element lest

Furthermore, imagine that we only have a couple of examples of sentences with that functional element, such as those shown below: