The Distribution of Pronoun Case Forms in English (Linguistik Aktuell Linguistics Today)

The Distribution of Pronoun Case Forms in English

Linguistik Aktuell/Linguistics Today Linguistik Aktuell/Linguistics Today (LA) provides a platform for original monograph studies into synchronic and diachronic linguistics. Studies in LA confront empirical and theoretical problems as these are currently discussed in syntax, semantics, morphology, phonology, and systematic pragmatics with the aim to establish robust empirical generalizations within a universalistic perspective.

Series Editors Werner Abraham

Elly van Gelderen

University of Vienna

Arizona State University

Advisory Editorial Board Cedric Boeckx

Ian Roberts

Harvard University

Cambridge University

Guglielmo Cinque

Ken Safir

University of Venice

Rutgers University, New Brunswick NJ

Günther Grewendorf

Lisa deMena Travis

J.W. Goethe-University, Frankfurt

McGill University

Liliane Haegeman

Sten Vikner

University of Lille, France

University of Aarhus

Hubert Haider

C. Jan-Wouter Zwart

University of Salzburg

University of Groningen

Christer Platzack University of Lund

Volume 82 The Distribution of Pronoun Case Forms in English by Heidi Quinn

The Distribution of Pronoun Case Forms in English

Heidi Quinn University of Canterbury

John Benjamins Publishing Company Amsterdam
/
Philadelphia

8

TM

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

Library of Congress Cataloging-in-Publication Data Heidi Quinn The Distribution of Pronoun Case Forms in English / Heidi Quinn. p. cm. (Linguistik Aktuell/Linguistics Today, issn 0166–0829 ; v. 82) Includes bibliographical references and indexes. 1. English languague--Pronoun. 2. English language--Case. PE1261.Q56 2005 2005050763 isbn 90 272 2806 X (Hb; alk. paper)

© 2005 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

To Mike, and my parents, Hildegard and Helmut Schludermann

TABLE OF CONTENTS

Acknowledgements

xi

Key to abbreviations

xii

Introduction CHAPTER 1 The history of the English case system 1.0 Introduction 8 1.1 Morphological case in Old English 9 1.2 Morphological case in Middle English 14 1.3 Lexical and structural case marking in the history of English 17 1.4 Morphological changes in the pronominal paradigm during the Early Modern English period 20 CHAPTER 2 Formal approaches to case and the three case constraints 2.0 Introduction 26 2.1 Case and argument structure 27 2.1.1 Case assignment by structural linking 27 2.1.2 Instances of case marking not predictable from structural linking 37 2.2 Case and structural position 39 2.2.1 vP and case checking 41 2.2.2 The roles of C and T in case checking 48 2.2.3 Case checking within DP 50 2.2.4 The case properties of prepositions 53 2.3 Argument Case, Positional Case, and (Positional) Default Case 57 2.4 A constraint-based approach to agreement 61

1

8

26

viii

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

CHAPTER 3 Case and the weak/strong distinction in the English pronoun system 65 3.0 Introduction 65 3.1 Strong and deficient pronouns in languages other than English 65 3.2 The syntactic properties of constructions with pronoun case variation 69 3.3 The prosodic properties of pronouns that exhibit case variation 74 3.4 Summary of differences between weak and strong pronouns 76 CHAPTER 4 The empirical survey 4.0 Introduction 78 4.1 Constructions and pronoun combinations tested 79 4.2 Task types 87 4.3 Fillers 96 4.4 The contents of the five questionnaires 97 4.5 The sample 99 4.6 Data collection 100 CHAPTER 5 The survey results 5.0 Introduction 101 5.1 Pronoun case in coordinates 101 5.1.1 Overall trends 102 5.1.2 Individual speaker patterns 113 5.1.3 Implications for formal analyses of coordination 121 5.2 Pronoun-NP constructions 124 5.3 Pronoun case in it-clefts 133 5.4 Pronoun case in than-comparatives 142 5.5 Summary of the most important case trends 145

78

101

CHAPTER 6 Relative Positional Coding and the Invariant Strong Form constraints 148 6.0 Introduction 148 6.1 The limitations of a purely case-based analysis 148 6.2 Relative Positional Coding 151 6.2.1 Phonological complexity and the classification of pronoun forms 156 6.2.2 Vowel quality and the classification of pronoun forms 168 6.2.3 Onsets and the classification of pronoun forms 170

TABLE OF CONTENTS

6.3 6.4

ix

The Invariant Strong Form constraints 171 The interaction of case and non-case constraints 175

CHAPTER 7 Modelling the interaction of the constraints 7.0 Introduction 178 7.1 Optimality Theory (OT) 179 7.1.1 Basic principles of OT 180 7.1.2 OT-based approaches to variation 184 7.2 An alternative constraint-weighting approach 186 7.2.1 Cumulativity and the most common combinations of 1sg and non-1sg patterns 189 7.2.2 Patterns ruled out by the proposed approach 194 7.3 Summary 200 CHAPTER 8 The distribution of personal pronoun forms in other strong pronoun contexts 8.0 Introduction 201 8.1 Topicalised pronouns 205 8.2 Left-dislocated pronouns 214 8.3 Right-dislocated pronouns 219 8.4 Independent pronouns 228 8.5 Pronoun case after be 233 8.5.1 Pronouns in basic identificational sentences 233 8.5.2 Pronouns in it BE sentences 242 8.6 Pronoun case in V-ing constructions 248 8.7 Pronoun case in to-infinitives 258 8.8 Pronoun case in small clauses 262 8.9 Pronoun case in gapping constructions 272 8.10 Pronoun case in bare argument ellipsis 278 8.11 Pronouns following but, save, except 280 8.12 Pronouns following not 283 8.13 Pronoun case after only 286 8.14 The case of modified pronouns 287 8.14.1 Pronouns modified by adjectives 288 8.14.2 Pronouns followed by a numeral 291 8.14.3 Pronouns associated with the quantifiers all and both 294 8.14.4 Pronouns followed by a PP 296

178

201

x

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

8.14.5 Pronouns followed by a self-reflexive 299 8.14.6 Pronouns followed by an appositive 300 8.14.7 Pronouns followed by a relative clause 302 8.15 Summary and conclusions 308 CHAPTER 9 The distribution of wh-pronoun forms in Modern English 9.0 Introduction 310 9.1 wh-pronouns in matrix questions 312 9.2 The case of subject pronouns after fronted auxiliaries 315 9.3 wh-pronouns in echo questions 318 9.4 wh-pronouns in embedded questions 320 9.5 wh-pronouns in sluicing constructions 325 9.6 wh-pronouns in free relatives 331 9.6.1 Free relatives introduced by complex and simplex wh-pronouns 333 9.6.2 Free relatives with VP ellipsis and Null Complement Anaphora 343 9.7 wh-pronouns in headed relatives 352 9.7.1 The extension of who to objective contexts 355 9.7.2 The extension of whom to nominative contexts 357 9.7.3 Implications for formal analyses of headed relatives 359 9.8 Summary and conclusions 367

310

CHAPTER 10 Speculations and conclusions 370 10.0 Introduction 370 10.1 Morphological and positional licensing in the history of English 370 10.2 The rise of Positional Case 373 10.3 The divergence of the weak and strong pronoun series in English 375 10.3.1 The distribution of personal pronouns in Old English 375 10.3.2 The emergence of Relative Positional Coding and the trend towards invariant strong pronoun forms 380 References

384

Name index

398

Subject index

402

ACKNOWLEDGEMENTS

I owe a special thank you to all the students at the University of Canterbury who participated in my empirical survey and helped with the pilot. Thanks also to the University of Canterbury, who supported my research with the University of Canterbury Research Grant U6206. Linguistic thanks go to Kate Kearns for the probing questions and countless suggestions that helped shape the arguments and analysis presented in this book; Liz Pearce, and Diane Massam for their interest, enthusiasm, and invaluable comments on the manuscript; Andrew Carstairs-McCarthy, Jen Hay, Ralf Vogel, Bea and Mike Dukes, K.P. Mohanan, Joan Bresnan, and the audience at the empirical syntax workshop in Berlin for ‘Optimal’ inspiration; Lyle Campbell and Cindy Allen for historical insights; Richard Kayne, Halldor Sigursson, Peter Peterson, Elly van Gelderen, Janne Johannessen, Ed Zoerner, and Kyle Johnson, for stimulating discussions and copies of their work on syntactic structure, case, agreement, coordination and gapping; Alice Harris, Luigi Rizzi, Adriana Belletti, and Ardith Meier for more syntactic stimulation; Shizuka Torii, Philippa Horton, Evelyn Todd, Therese Aitchison, Melanie Owens, Heather McCann, Sarah Macann, and Sharbani Banerji for lengthy conversations and correspondence about pronouns and syntactic structure; and Werner Abraham for helpful advice and encouragement. I would also like to thank Jandy Blundell, Emma Parnell, Wendy Nuthall, Suzanne Nesbitt, Rachel Rowlands, and various University of Canterbury IT technicians, for their help with administrative matters and computer problems, and Kees Vaes for his assistance in getting the manuscript camera-ready. All the tree diagrams in this book were drawn using the ArborWin font created by Cascadilla Press, a truly indispensable tool for any syntactician. It would have been difficult to write this book without the support I received from my friends and family, in particular Soe Tjen Marching, Rebecca Yanxia Huang, Sanya Silver, Thea Cameron-Faulkner, Yoko Mizushima, and my parents, Hildegard and Helmut Schludermann. My final and biggest thank you goes to Mike Quinn, who has always been happy to listen when I needed someone to bounce my ideas off, and who probably knows more about linguistics now than he has ever wanted to know.

KEY TO ABBREVIATIONS

1du 1pl 1pl-NP

1ps 1sg 2du 2pl 2ps 2sg 3pl 3pl-NP

3ps 3sgF 3sgM 3sgN ACC Arg DAT Def EME EModE FEM

1st person dual (pronoun) 1st person plural (pronoun) construction where a 1pl pronoun is followed by a noun phrase (e.g. we New Zealanders) 1st person (pronoun) 1st person singular (pronoun) 2nd person dual (pronoun) 2nd person plural (pronoun) 2nd person (pronoun) 2nd person singular (pronoun) 3rd person plural (pronoun) construction where a 3pl pronoun is followed by a noun phrase (e.g. them politicians) 3rd person (pronoun) 3rd person singular feminine (pronoun) 3rd person singular masculine (pronoun) 3rd person singular neuter (pronoun) accusative Argument (in Arg-Agreement and Arg-Case) dative Default (in Def-Agreement and Def-Case) Early Middle English Early Modern English feminine

GEN INV LF MASC ME NCA NOM NOM/ACC non-1sg

O OBJ OBL OE OT P P comp PF PL Pos RPC S SF SG TS

VPE

genitive Invariant Strong Form Logical Form masculine Middle English Null Complement Anaphora nominative syncretic nominative and accusative form case-alternating pronouns other than 1sg (i.e. he/him, she/ her, we/us, they/them) object of verb objective (syncretic dative and accusative form) oblique (syncretic genitive and dative form) Old English Optimality Theory (complement/object of) preposition complement of preposition Phonological Form plural Positional (in PosAgreement and Pos-Case) Relative Positional Coding subject of a finite clause Semantic Form singular θ-structure (interface-level between Semantic Form and Phrase Structure) VP Ellipsis

INTRODUCTION

The aim of this book is to provide a systematic syntactic analysis of the distribution of pronoun case forms in Modern English, and to relate current pronoun case trends to historical changes in the English case system and in the licensing of structural arguments. The distribution of the pronoun case forms I/me, he/him, she/her, we/us, they/them and who/whom in Modern English has been the subject of much linguistic debate. On the one hand, formal approaches to case tend to assume that the alternation between nominative and objective pronoun forms in English is determined exclusively by structural case mechanisms. On the other hand, the distributional differences between personal pronoun forms and wh-forms, and the variability of pronoun case in coordinates, it-clefts, than comparatives, and similar constructions, are often cited as evidence that pronoun case selection in English is largely unsystematic, and best treated as the product of local rules, grammatical viruses, and hypercorrection (cf. Emonds 1985, 1986; Sobin 1997; Lasnik & Sobin 2000; Peterson 2004). I will demonstrate that the presence versus absence of variation in pronoun case choice largely correlates with the morphosyntactic status of the pronoun (cf. Cardinaletti & Starke 1999). I propose that there is a two-way distinction between weak and strong pronouns in the Modern English pronoun system. Unlike in Romance languages, where some deficient pronouns appear in special (head-adjoined) clitic positions, the deficient personal pronouns found in Modern English only appear in positions that are also available to full noun phrases. The evidence discussed in this book suggests that weak pronouns not only differ from their strong counterparts in their prosodic properties, their internal syntax, and their syntactic distribution, but also in their case morphology. English weak pronouns, which are licensed in structural positions associated with case assignment in formal theory, do indeed exhibit the consistent nominative/objective case distinction predicted by general approaches to case. The distribution of strong pronoun forms, on the other hand, is difficult to capture in a purely case-based account. Despite the wealth of anecdotal evidence and the considerable amount of speculation that surrounds the distribution of strong pronoun forms in English, very little systematic research has been carried out to support the various hy-

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

2

potheses. Early studies of pronoun case variation were largely based on literary corpora and can therefore provide only limited evidence on the pronoun case system of individual speakers (cf. Jespersen & Haislund 1949; Visser 1963; Klima 1964; Erdmann 1978; and Householder 1986, 1987). More recent corpus studies by Wales (1996) and Angermeyer & Singler (2003), and the empirical surveys carried out by Emonds (1985, 1986), Parker et al. (1988), Quattlebaum (1994), Quinn (1995) and Boyland (2001) suggest that there is systematic case variation between pronouns as well as between speakers, but no existing study has so far tested the whole range of alternating personal pronouns in a set of comparable environments on a sizable number of speakers. This book presents the results of a written survey of 90 native speakers of English, which tested the distribution of all alternating personal pronoun forms (i.e. I/me, he/him, she/her, we/us, they/them) in coordinates (1), it-clefts (2), and than comparatives (3), and the distribution of 1pl (we/us) and 3pl forms (they, them) in pronoun-NP constructions (4). (1) (2) (3) (4)

Stuart and I/me It was he/him who/that insisted on going to the rally. Oliver is bound to respond more quickly than she/her. we/us New Zealanders

The pronoun case patterns attested in the survey suggest that case considerations still influence the distribution of pronoun forms in these constructions. However, both the survey results and the case trends reported in existing studies indicate that this case influence is weakened by a trend towards invariant strong pronoun forms. For personal pronouns, the emerging invariant forms are the objectives me, him, her, us, them. For wh-pronouns, the emerging invariant forms are the nominatives who and whoever. The survey results also point to considerable distributional differences between 1sg (I/me) and non-1sg pronoun forms (he/him, she/her, we/us, they/ them), and confirm some of the often noted correlations between conjunct position and pronoun form in coordinates (cf. Schwartz 1985; Zoerner 1995; Sobin 1997; Johannessen 1998). This suggests that strong pronoun forms no longer merely identify the structural case of a pronoun, but also code its position within a syntactic construction. The systematic variability in the responses of individual survey participants indicates that pronoun case is a morphophonological (PF) phenomenon conditioned by a range of syntactic and semantic factors (cf. Sigursson 2003; McFadden 2004). The interaction of these conditioning factors is best captured

INTRODUCTION

3

in a surface-oriented constraint-based approach where constraint violations do not necessarily crash the derivation. I propose that the distribution of pronoun case forms in Modern English is determined largely by the violable constraints given in (5)-(10). Each constraint is followed by examples illustrating its case predictions for conjoined pronouns in subject (a), object (b), and prepositional complement position (c). Since the predictions of the two Relative Positional Coding constraints (RPC 1 and RPC 2) differ only when the final conjunct is occupied by a 3pl pronoun (they/them), these two constraints are followed by an additional set of example sentences with 3pl in final conjunct position (d)-(f). All of the example sentences are taken from the questionnaires used in the empirical survey. The numbers of the relevant questionnaire items are given in square brackets after the sentences illustrating the predictions of RPC 1. Tables detailing the overall survey results for the questionnaire items containing sentences (a)-(d) can be found in Section 5.1. (5)

Argument Case (Arg-Case) The overt case form of any structural argument of a predicate must comply with the structural linking between cases and arguments in the θ-structure (cf. Wunderlich 1997; Kiparsky 1997). (a) [He and I] arrived here three hours ago. (highest argument of V = nominative)

(b) Brenda had promised she would meet [him and me] at the station. (lower argument of V = objective)

(c) The landscapes painted by [him and me] drew huge crowds at the exhibition. (lower argument of P = objective)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

4 (6)

Positional Case (Pos-Case) The overt case form of an argument noun phrase appearing as the specifier of an agreement-related functional head at Spell-Out must match the case/agreement features of this functional head, iff the position of the noun phrase at Spell-Out differs from its θposition. (a) [He and I] arrived here three hours ago. (argument that has raised to [Spec,TP] of a finite clause = nominative)

(b) Brenda had promised she would meet [him and me] at the station. (if the object coordinate has raised to [Spec,vP] = objective; otherwise no predictions)

(c) No predictions for complement of preposition, because no suitable agreement-related functional head is available to check Pos-Case.

(7)

(Positional) Default Case (Def-Case) The overt case form of any noun phrase not influenced by the Pos-Case constraint must match the default case of a language. In Modern English, the default case is the objective case. (a) No predictions for the preverbal subject of a finite clause, because it checks Pos-Case.

(b) Brenda had promised she would meet [him and me] at the station. (if the object coordinate has remained in [Spec,VP] = objective, because it will not qualify for Pos-Case checking; otherwise no predictions)

(c) The landscapes painted by [him and me] drew huge crowds at the exhibition. (objective, because complements of prepositions cannot check Pos-Case)

INTRODUCTION

(8)

5

Relative Positional Coding 1 (RPC 1) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then A must be gracile, and B must be robust. Gracile pronoun forms: me, he, she, we, they, who Robust pronoun forms: I, him, her, us, them, whom (a) [He and I] arrived here three hours ago. [qu057] (b) Brenda had promised she would meet [he and I] at the station. [qu014]

(c) The landscapes painted by [he and I] drew huge crowds at the exhibition. [qu045] (d) [She and them] spent the whole week complaining about the food, the rooms, the service, … [qu134] (e) He’s told [she and them] several times not to disturb him, but do they listen? [qu132] (f) But if you ask me, he’d been lying to [she and them] right from the start. [qu142] (9)

Relative Positional Coding 2 (RPC 2) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then B must be more robust than A. In the set of gracile pronoun forms, they is more robust than me, he, she, we. (a)-(c) Same predictions as RPC 1. (d) [She and they/them] spent the whole week complaining about the food… (e) He’s told [she and they/them] several times not to disturb him, but do they listen? (f) But if you ask me, he’d been lying to [she and they/them] right from the start.

6

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(10) Invariant Strong Form (INV) The morphological form of strong pronoun forms must be invariant in all contexts. There is a separate Invariant constraint for each pronoun. The invariant personal pronoun forms are: me, him, her, us, them The invariant wh-forms are: who, whoever (a) [Him and me] arrived here three hours ago. (b) Brenda had promised she would meet [him and me] at the station. (c) The landscapes painted by [him and me] drew huge crowds at the exhibition. The interaction of the proposed constraints can be modelled in a probabilistic approach where constraints are weighted and the probability of occurrence of a particular variant is determined by the combined weight of the constraint violations it incurs (cf. Guy 1997; Mohanan 1998). For individual speakers, variability in pronoun case occurs in contexts where the combined weight of constraint violations is similar for both forms of a pronoun. Variation between speakers arises from different relative weightings of the various constraints. Discrepancies between the grouping of pronoun forms associated with structural case and the grouping of pronoun forms associated with Relative Positional Coding are largely responsible for the distributional differences between strong 1sg and non-1sg forms. For structural case, I patterns with the non-1sg nominatives he, she, we, they, who, and me patterns with the non-1sg objectives him, her, us, them, whom. For Relative Positional Coding, on the other hand, I groups with him, her, us, them, whom to form the robust series, and me groups with he, she, we, they, who to form the gracile series (8). Both the survey results and data from existing studies of pronoun case highlight the importance of structural position to pronoun case in Modern English. The virtual absence of case variation with weak pronouns, which must be licensed in the specifier of an agreement-related functional head, is most easily captured by assuming that the Positional Case constraint outweighs the Argument Case constraint in Modern English. This relative weighting of the two case constraints is further supported by the pronoun case patterns attested in strong pronoun contexts. In Old English, on the other hand, the case form of a pronoun or noun phrase appears to have been largely determined by Argument Case. The differences between Modern English pronoun case and Old English

INTRODUCTION

7

case marking support Kiparsky’s (1997) hypothesis that the phonology-driven loss of morphological case marking on nouns prompted a shift from morphological to positional licensing of structural arguments during the Middle English period (cf. also Allen 1995). The increasing influence of Positional Case and the reanalysis of pronoun case forms as markers of structural position can be seen as by-products of this shift. The book is structured as follows: Chapter 1 outlines the historical development of the English case system. Chapter 2 argues that the distribution of pronoun case forms in Modern English is influenced by three case constraints (Argument Case, Positional Case, and Default Case), and discusses the theoretical concepts that the proposed constraints are based on. Chapter 3 presents evidence for a weak/strong distinction in the English pronoun system and argues that the presence versus absence of variation in pronoun case choice largely correlates with the morphosyntactic status of the pronoun. Chapters 4 and 5 describe the methodology used in the empirical survey and present the survey results, which provide information about the distribution of pronoun case forms in certain strong pronoun contexts, namely coordinates, pronounNP constructions, it-clefts, and than-comparatives. Chapter 6 introduces the Relative Positional Coding constraints and the Invariant Strong Form constraints, which are needed to account for the pronoun case patterns attested in the empirical survey. Chapter 7 considers ways to model the interaction of constraints and concludes that the combinations of case patterns found in coordinates are most successfully modelled in a probabilistic constraint-weighting approach. Chapter 8 takes a closer look at a wide range of strong pronoun contexts that were not included in the empirical survey, and demonstrates how the personal pronoun case trends reported for these contexts provide further evidence for the interaction of the various constraints proposed in Chapter 2 and Chapter 6. Chapter 9 examines existing evidence of pronoun case variation involving wh-pronouns and shows that the distribution of wh-case forms can be accounted for in the proposed approach if the nominatives who and whoever are treated as emerging invariant forms. And Chapter 10 considers the historical contexts of the pronoun case trends found in Modern English, and argues that both the increasing influence of surface position on pronoun case and the trend towards invariant strong forms can be seen by-products of a diachronic change in the licensing of arguments.

CHAPTER 1 THE HISTORY OF THE ENGLISH CASE SYSTEM

1.0

Introduction This chapter looks at the history of case marking and case assignment in English. Changes in the English case system affected both the way in which case was assigned to noun phrases, and the morphological realisation of case distinctions. As Allen (1995:211f,443f) points out, the loss of lexical case marking and the reduction of overt case morphology appear to have been gradual and systematic. Although the deterioration of the English case system is most evident in the Early Middle English period (EME), we already find a fair amount of syncretism in the case morphology of Old English (OE) nouns and adjectives. Similarly, a tendency to avoid lexical case marking in favour of structural case assignment is already present in late OE. Since the loss of lexical case marking on objects appears to be closely related to the loss of the morphological distinction between accusative and dative case (Allen 1995:445), I present an overview of the changes in case morphology during the OE and ME periods in Sections 1.1 and 1.2, before taking a closer look at the increasing importance of structural case assignment in the history of English in Section 1.3. Section 1.4 concludes the historical overview with a brief discussion of morphological changes in the paradigm of the 2nd person pronoun during the Early Modern English (EModE) period. Much of the discussion in this chapter foreshadows the arguments presented in Chapter 10, where I suggest that the neutralisation of morphological case distinctions at the end of the OE period contributed to a shift in the licensing of arguments. While structural arguments in OE could be licensed either by their case morphology or by the position they occupy in the clause, once the nominative-accusative case distinctions in the nominal paradigm had been lost, morphological licensing was no longer an option. I propose that the shift to exclusive positional licensing and the decline of lexical case marking have led to an increase in the importance of Positional Case, and have indirectly contributed to the reanalysis of strong pronoun forms as markers of relative structural position (Relative Positional Coding) and of strong pronoun status (Invariant Strong Form).

CHAPTER 1

9

The Modern English (ModE) data presented in the following chapters suggest that the trend towards invariant case forms in strong pronoun contexts is more advanced for some pronouns than others. As I will show in Sections 1.1 & 1.2, we also find differences between pronouns when it comes to the neutralisation of the dative-accusative case distinction during the Old and Middle English periods. 1.1

Morphological case in Old English As can be seen from Tables 1-3, OE nouns and adjectives do not distinguish between nominative and accusative in the plural. Masculine and feminine adjectives in the strong declension class have distinct nominative, accusative, and oblique singular forms, but only the masculine and neuter adjectives distinguish between the genitive and the dative in the singular (see Table 1.1). 1 strong stems NOM ACC GEN DAT

masculine singular Ø -ne -es -um

plural -e -e -ra -um

feminine singular Ø / -u -e -re -re

plural -e -e -ra -um

neuter singular Ø Ø -es -um

plural Ø / -u Ø / -u -ra -um

Table 1.1. Case inflections on strong adjectives in OE (Kemenade 1987:103; Allen 1995:164)

The only noun classes to exhibit distinct nominative and accusative inflections in the singular are the strong feminine o-stems and the masculine and feminine nouns in the weak declension class (see Tables 1.2-1.3). These stems nevertheless exhibit case syncretism, but between accusative, genitive, and dative singular forms. The inflections found on weak adjectives are identical to those found on weak nouns, with the exception that the plural genitive suffix may also take the form –ra (Kemenade 1987:102-103; Mitchell 1985:4,§9). Feminine i-nouns and all u-stems have only one form for nominative, accusative, and genitive plural. Singular genitive and dative forms are identical for all ustems and the neuter nouns in the weak declension class (Kemenade 1987:102103).

1

Shading indicates syncretism. Ø indicates that the stem surfaces without a suffix; phonological variants of a suffix are separated by a slash, e.g. Ø / -e. I have followed the transcription conventions adopted in existing theoretical studies of OE syntax, and omitted any indication of vowel length from the historical data presented here (cf. Kemenade 1987; Allen 1995).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

10 weak stems NOM ACC GEN DAT

masculine singular -a -an -an -an

plural -an -an -ena -um

feminine singular -e -an -an -an

plural -an -an -ena -um

neuter singular -e -e -an -an

plural -an -an -ena -um

Table 1.2. Case inflections on weak nouns in OE (Kemenade 1987:103) a-stems NOM ACC GEN DAT i-stems NOM ACC GEN DAT u-stems NOM ACC GEN DAT o-stems NOM ACC GEN DAT

masculine singular Ø Ø -es -e

plural -as -as -a -um

neuter singular Ø / -e Ø / -e -es -e

plural -u / Ø -u / Ø -a -um

masculine singular Ø Ø -es -e

plural -e / -as -e / -as -a -um

neuter singular Ø Ø -es -e

plural -u / Ø -u / Ø -a -um

feminine singular Ø Ø -e -i

plural -a -a -a -um

masculine & feminine singular plural -u / Ø -a -u / Ø -a -a -a -a -um feminine singular -u / Ø -e -e -e

plural -a / -e -a / -e -a -um

Table 1.3. Case inflections on strong nouns in OE (Kemenade 1987:102)

Even the paradigms of the definite determiner, the wh-pronouns, and the personal pronouns exhibit some case syncretism (Kemenade 1987:103-104; Allen 1995:165; Mitchell 1985:4,§9; Jespersen 1949[1927]:116). We find no nominative-accusative distinction in the neuter and plural paradigms of the definite determiner or in the paradigms of the 3rd person singular neuter and the 3rd person plural pronoun (see Tables 1.4-1.5). The nominative-accusative dis-

CHAPTER 1

11

tinction is also absent in the neuter wh-pronoun paradigm (Table 1.6). Syncretism between genitive and dative forms can be found with both the feminine determiner and the 3rd person singular feminine pronoun (Tables 1.4-1.5). OE determiner masculine se Þone Þæs Þæm

NOM ACC GEN DAT

singular feminine seo Þa Þære Þære

plural neuter Þæt Þæt Þæs Þæm

Þa Þa Þara Þæm

Table 1.4. Paradigm of the definite determiner in OE (Kemenade 1987:104; Allen 1995:165) OE 3ps pronoun NOM ACC GEN DAT

masculine he hine his him

singular feminine heo hi(e) hi(e)re hi(e)re

plural neuter hit hit his him

hi(e) hi(e) hi(e)ra him

Table 1.5. 3rd person (3ps) pronominal paradigms in OE (Jember et al. 1975:ix; Kemenade 1987:103f; Gelderen 2000:21) OE wh-pronoun NOM ACC GEN DAT

masculine & feminine hwa hwone hwam hwæs

neuter hwæt hwæt hwam hwæs

Table 1.6. wh-pronoun paradigm in OE (Jespersen 1949[1927]:116; Jember et al. 1975:x)

The first person and second person accusative forms mec, Þec, usic, eowic, uncit,and incit appear to have been used mainly in the oldest texts, and were already supplanted by the dative forms me, Þe, us, eow, unc, and inc during the OE period (Table 1.7). This extension of the dative foreshadows analogous later developments in the paradigms of the 3rd person singular masculine, feminine, and plural pronouns (Jespersen & Haislund 1949:221).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

12 early OE texts

1sg

2sg

1dual

2dual

1pl

2pl

NOM ACC GEN DAT

ic mec min me

Þu Þec Þin Þe

wit uncit uncer unc

git incit incer inc

we usic ure us

ge eowic eower eow

later OE texts

1sg

2sg

1dual

2dual

1pl

2pl

NOM ACC GEN DAT

ic me min me

Þu Þe Þin Þe

wit unc uncer unc

git inc incer inc

we us ure us

ge eow eower eow

Table 1.7. 1ps and 2ps pronominal paradigms in OE (Jember et al. 1975:ix; Kemenade 1987:103f; Jespersen & Haislund 1949:221; Gelderen 2000:21; Gelderen 2004:272)2

The demise of the old 1ps and 2ps accusative forms (mec, þec, etc.) is further highlighted by OE reflexive data. As (1) illustrates, simple objective pronoun forms could also occur reflexively in Old English. (1) No ic me an herewæsmun hnagran talige guþgeweorca, not 1sg.NOM 1sg.DAT on prowess smaller think wardeeds þonne Grendel hine than Grendel 3sgM.ACC “By no means do I consider myself smaller in prowess and wardeeds than Grendel does himself” (Beowulf 677-8) [Gelderen 2000:33]3

2 Gelderen (2000:33f) reports small numbers of mec, þec, usic and eowic in the epic poem Beowulf (MS dated approx. 1000, from an 8th century original); mec, þec, and incit occur in the gospel texts of The Junius MS (approx. 1000, with parts possibly from the 8th century), and mec is also quite prominent in the Riddles part of the Exeter Book (Gelderen 2000:43-46). According to Gelderen (2000:61), mec and þec are more common in The Vespasian Psalter (Mercian dialect, early half of 9th century) than in other texts. For an overview of the occurrence of 1ps and 2ps accusative and dative forms in Old English see Gelderen (2000:200-204). All 1ps and 2ps dual forms appear to have dropped out of the language during the 13th century (Mustanoja 1960:125). 3 Round brackets enclose primary sources, square brackets mark secondary sources. The full references for the primary sources of OE and ME examples are listed at the end of the bibliography.

CHAPTER 1

13

3ps reflexives usually took the accusative form (hine, etc.), whereas 1ps and 2ps reflexives almost always surfaced in the dative form (me, þe, etc.) (Gelderen 2000:43). With 3ps pronouns, the occurrence of dative forms coreferent with the subject was generally restricted to non-argument positions (2), where the use of a reflexive pronoun served to emphasise subject involvement (Gelderen 2000:43; Keenan 1997:3). (2) forðæm hi him ondrædað ða frecenesse ðe hi because 3pl.NOM 3pl.DAT fear the danger that 3pl.NOM ne gesioð not see “because they fear (them) the danger that they do not see” (Pastoral Care 433) [Keenan 1997:3]

The preference for dative 1ps and 2ps forms in all reflexive contexts emphasises the marginal status of the 1ps and 2ps accusative forms in most OE texts. As we might expect, the reflexive use of accusative 1ps and 2ps is largely limited to manuscripts such as the Riddles and The Vespasian Psalter, where 1ps and 2ps accusative forms are generally common (Gelderen 2000:46,61). At the end of the OE period, a number of phonological changes led to a dramatic increase in the syncretism found in the case morphology of nouns and adjectives (Allen 1995:163f):4 (a) The distinction between high, mid, and low vowels disappeared in final unstressed syllables (and thus also in case suffixes); and in the 11th century even the front-back distinction was lost in this environment. (b) Final /m/ was replaced with /n/ in unstressed syllables (c) Final /n/ was lost in unstressed syllables. These changes neutralised the remaining differences between the singular nominative and accusative inflections of weak nouns and adjectives, and ultimately

4

Both Moore (1928:240-243) and Allen (1995:163f) note that the replacement of /m/ with /n/ in unstressed syllables appears to have preceded the loss of final /n/ in unstressed position. The chronological order between these consonant changes and the weakening of unstressed vowels is not entirely clear, although the written evidence seems to indicate that the complete loss of vowel distinctions in unstressed syllables more or less coincided with the loss of final /n/ (cf. Moore 1928:244-247). According to Allen (1995:163), evidence for the levelling of height distinctions between unstressed vowels can already be found in ninth century texts, which suggests that at least in some dialects, the earlier stages of vowel weakening were contemporaneous with the replacement of /m/ with /n/ in unstressed syllables.

14

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

led to the complete loss of the nominative-accusative distinction on nouns in the late 11th century (Allen 1995:165). 1.2

Morphological case in Middle English (ME) During the Middle English period, the remaining nominal case (and gender) inflections gradually disappeared, with only the genitive retained, predominantly as a marker of noun phrases that appear within another noun phrase (Allen 1995:195). The original dative suffix -e of strong masculine and neuter nouns was largely restricted to objects of prepositions (Allen 1995:176,204). We still find different case forms for the definite determiner, especially in the more conservative southern dialects, but even these varieties of ME show some levelling towards Þe in the singular and Þa in the plural. Where the original singular accusative Þone is still used, it has taken over the function of a generic object form. As with nouns and adjectives, there is no longer any evidence for a consistent distinction between dative and accusative forms (Allen 1995:171f, 190f). The dative/accusative distinction also disappeared from the pronominal paradigms during the ME period. Significantly, the distinction between dative and accusative case forms was lost at different times for different pronouns. These φ-related differences in the ME case system (where ‘φ’ stands for person, number and gender features) foreshadow the φ-related variation in the distribution of pronoun case forms in Present-Day English (cf. Chapters 5, 8 and 9), and could be seen to support the links between case and φ-feature agreement proposed in some formal approaches (e.g. Schütze 1997; Chomsky 2000; cf. Chapter 2 for a more detailed discussion). The distinction between dative and accusative 1ps and 2ps forms had already been lost during the OE period, when the dative forms me, Þe, us, eow, unc, and inc were extended to accusative contexts (Table 1.7). For wh-pronouns, the dative/accusative distinction was neutralised on the dative form whom (quam in northern dialects) in the masculine/feminine paradigm, but on the nominative/accusative form what (quat in northern dialects) in the neuter paradigm (Table 1.8). ME wh-pronoun

masc. & fem.

neuter

NOM ACC DAT

who, qua whom, quam whom, quam

what, quat what, quat what, quat

Table 1.8. wh-pronoun paradigms in ME (Mustanoja 1960:180-185,194f,199-201)

CHAPTER 1

15

The changes in the 3ps paradigm happened at different times for the different 3ps pronouns and also varied between dialects (Allen 1995:169-205; Kemenade 1987:229-233): Southern dialects retained the dative/accusative distinction considerably longer than northern dialects, and we still find a difference between the dative and accusative forms of at least some pronouns in late 13th century texts from Kent. Southern English usage during the 12th and 13th centuries is exemplified by the paradigm for 3ps pronouns in the Vices & Virtues (Table 1.9). As Allen (1995:189) notes, the nominative/accusative distinction in the plural actually represents an increase in case marking distinctions when compared to OE (Table 1.5). At the same time, the occasional extension of the 3sgM and 3pl datives him and hem to accusative contexts points to a weakening of the dative/accusative distinction, which is eventually lost in the early 14th century. Vices & Virtues (c.1200) NOM ACC GEN DAT

3ps singular masculine he, hie hine (him) his(e) him

feminine heo, hie, he hie, hes, his hire hire

3ps plural neuter hit hit his him

hie, hi, he hes, his (hem) here, her, heare hem, heom

Table 1.9. 3ps pronominal paradigms in the Vices & Virtues (Southern England) (Allen 1995:189)

In the Northeast of England, the dative/accusative distinction is only in evidence up to the early 12th century (Table 1.10). The scribe who wrote the First Continuation of the Peterborough Chronicle (c.1131), still uses the original 3sgM accusative hine alongside the dative him in accusative contexts. However, in the Second Continuation (written c.1154), hine has been completely replaced by him. According to Allen (1995:176), hine is generally limited to clear accusative contexts in the First Continuation, but she also notes two instances of hine in dative contexts, which could be seen as further evidence that the scribe’s command of the dative/accusative distinction was waning. As in the Southern dialects, the OE syncretism between nominative and accusative plural forms has disappeared in the Peterborough Chronicle, but this time as a result of the extension of the dative 3pl form heom rather than the introduction of a distinct accusative form. The 3sgF and 3pl datives hire and heom have already spread to accusative contexts in the First Continuation,

16

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

whereas for the 3sg neuter, the dative him has been lost and the accusative hit has been extended to all dative contexts.5 Peterborough Chronicle First Continuation (c.1131) masculine NOM he ACC him, hine DAT him (hine)

3ps singular feminine heo hire hire

Second Continuation (c.1154) masculine NOM he ACC him DAT him

3ps singular feminine heo hire hire

3ps plural neuter hit hit hit

hi heom heom 3ps plural

neuter hit hit hit

hi heom heom

Table 1.10. 3ps pronominal paradigms in the Peterborough Chronicle (Northeast) (Allen 1995:176; Kemenade 1987:230f)

It is also in the Northeast that we find the first recorded examples of the Scandinavian 3pl forms they, them, and their (Table 1.11). The author of the Ormulum, a poem probably written just north of Peterborough in the late 12th century, consistently substituted the Scandinavian nominative they for the older hi, but used both English and Scandinavian forms in the genitive and the object case (Allen 1995:179). In keeping with the general extension of dative forms to all objects, the borrowed Scandinavian dative them was used in both accusative and dative contexts (Jespersen & Haislund 1949:221). Ormulum (late 12th c)

3ps plural

NOM ACC DAT GEN

they heom, them heom, them hira, their

Table 1.11. Pronominal paradigm for 3pl in the Ormulum (Northeast) (Allen 1995:179) 5

Allen (1995:189fn.30) suggests that the exceptional direction of the dative/accusative merger in the 3sg neuter paradigm may be due to the replacement of the OE gender distinctions with a human/non-human distinction in ME. The extension of the 3sg neuter accusative hit to dative contexts helped retain the distinction between non-human 3sg neuter objects and human 3sgM objects (which had the form him after the loss of the dative/accusative distinction).

CHAPTER 1

17

The complete loss of the dative/accusative distinction was probably the most influential change in ME case morphology, because it affected all nominal elements and had far-reaching consequences for the way in which case was assigned. 1.3

Lexical and structural case marking in the history of English In Old English (OE), case could be assigned either structurally or lexically. While lexical case, also known as ‘inherent’ or ‘quirky’ case, is specified in the lexical entries of certain verbs or prepositions, structural case is assigned according to the grammatical relation and structural position of the noun phrase in question. As a consequence, syntactic processes like passivisation can affect structural case assignment, but do not have any bearing on lexically assigned case. The difference between structural and lexical case assignment in OE is most clearly seen in verbal passives, such as those given in (3b) and (4). Like all highly transitive verbs (Allen 1995:25), ofslean ‘to slay, kill’ occurs with an object in the structural accusative case (hine) when the sentence is active (3a). When the verb is passivised, the subject appears in the nominative case (se cyning) (3b). (3) a. oþ þæt hie hine ofslægenne hæfdon until 3pl.NOM/ACC 3sgM.ACC slain had “until they had killed him” (Parker Chronicle 48.4 (755)) [Denison 1993:343]

b. þyslic wæs seo syn, þe se cyning such was the.SG.FEM.NOM sin that the.SG.MASC.NOM king fore ofslege wæs for slain was “such was the sin for which the king was slain” (Bede 3 16.228.5) [Denison 1993:130]

The verb deman ‘to judge’, on the other hand, assigns a lexical dative to its direct object (nanum men), and also takes a dative subject in the passive (him) (4).6 Passives with lexically case-marked subjects are generally known as ‘indirect passives’, as opposed to the ‘direct passive’ exemplified in (3b).

6

Note that the lexically case-marked dative subject him does not trigger agreement on the verb be. It appears that lexically case-marked noun phrases generally failed to trigger verbal agreement in OE, even when they exhibited clear subject properties and preceded the verb.

18

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(4) hi ne demað nanum men, ac him bið 3pl.NOM not judge no.PL.DAT men.PL.DAT but 3pl.DAT be.SG gedemed judged “They will not judge any men, but they will be judged.” (Ælfric Homilies XI:369) [Allen 1995:27]

According to Allen (1995:25), many verbs that lexically case-marked their objects in OE could also appear with objects in the structural accusative case (e.g. abelgan ‘to anger’ occurred with both dative and accusative objects).7 Since arguments of prepositions are not normally subject to operations comparable to passivisation, the nature of the case assigned by a given preposition is much harder to determine than the nature of the case assigned by a verb. Nevertheless, prepositions are generally assumed to have assigned lexical case in Old English, because the case of prepositional objects varied with the preposition and/or its interpretation (cf. Gelderen 2000:62; Mitchell 1985:496-499).8 These case differences between objects of different prepositions seem to have been lost during the Early Middle English period. In the Peterborough Chronicle (and also in the later Ormulum), any nominal objects of prepositions could optionally occur with the OE dative suffix -e, regardless of the nature of the preposition (Allen 1995:176,180). As Allen (1995:176) points out, this suggests that prepositions assigned structural dative rather than lexical case in Early Middle English. For verbs, the option of lexical case assignment to objects and passive subjects disappeared once the accusative/dative distinction had been lost around the 13th century (Allen 1995:219f,370,375f). Interestingly, the loss of lexical object case not only affected verbs that selected dative objects, but also verbs that assigned genitive case to their objects, even though nouns and pro7

See Mitchell (1985:529) for a similar observation. Mitchell (1985:455-464) also provides a list of verbs that occur with lexically case-marked objects which illustrates that many verbs in this category may take either lexically case-marked objects or objects in the structural accusative case. As both Mitchell (1985:449-454) and Allen (1995:25f) point out, the case-marking possibilities of Old English verbs are closely linked to the semantic properties of the verb, and the alternation between lexically case-marked objects and structurally case-marked objects often correlates with an alternation in meaning. However, the fact remains that many lexically case-marking verbs were already able to occur with accusative objects in at least some of their uses (or else coexisted with homonymous verbs that required objects in the structural accusative). 8 For example, prepositions like over generally took a dative object when they received a locative interpretation, but an accusative object when they received a path interpretation (Mitchell 1985:496-498).

CHAPTER 1

19

nouns retained distinct genitive case morphology in all dialects (Allen 1995: 195,218). As Allen points out, the loss of genitive objects is indicative of the general shift towards structural case assignment during ME, which is also evident in the earlier development of structural dative case assignment to objects of prepositions (1995:176,180,195,218). However, lexical case assignment was not limited to verbal and prepositional objects and passive subjects in OE. Allen (1995:96-121) presents convincing evidence that the preverbal dative experiencers in sentences like (5) have subject status, even though the verb generally surfaces in the unmarked 3sg form, or agrees with the postverbal nominative target of emotion (Allen 1995:142).9 (5) þæt þe wel licode þæra gewrita andgit that 2sg.DAT well liked the.PL.GEN writings.GEN meaning “that you liked the content of the writing” (COE Ælet 6 (Wulfgeat) 4) [Allen 1995:109]10

According to Allen (1995:104-111), dative experiencers with verbs like lician not only occur more frequently in preverbal position than indisputable objects, but are also attested in contexts where topicalised objects are neither necessary nor expected. What is more, pronominal experiencers of verbs that license preverbal dative experiencers, are able to occur postverbally only in those contexts where we also find nominative pronominal subjects, namely in clauses introduced by certain adverbs and negatives, but not in clauses introduced by a topicalised noun phrase or PP (Allen 1995:107; see Section 10.3.1 for a more detailed discussion of postverbal pronominal subjects in Old English). Further evidence for the subject status of preverbal dative experiencers comes from their ability to control Coordinate Subject Deletion to a greater extent than we would expect of ordinary dative objects (Allen 1995:117), and from the fact that expletive subjects (hit or þær) rarely cooccurred with prever9 Compare the Icelandic case marking patterns discussed in Yip et al. (1987). The availability of the preverbal dative experiencer construction in OE was to some extent linked to the semantic properties of the verb. Preposing of the dative experiencer was most common when the experiencer was a pronoun and the target of emotion was a noun (Allen 1995:109). A preposed dative experiencer was therefore most likely to be found with verbs like lician, which favoured topical experiencers and non-human targets of emotion. Allen notes that verbs of this type typically emphasise the role of the experiencer in the event, whereas verbs that do not exhibit the preposed dative experiencer construction (e.g. (ge)cweman) tend to focus on the role of the target of emotion as the cause of the event (1995:146f). 10 The full details of the primary sources for examples originally taken from the Microfiche Concordance of Old English (COE) can be found in the guide to the concordance.

20

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

bal dative experiencers, even though they were already strongly favoured with weather verbs and verbs that took sentential complements (Allen 1995:117121; Kemenade 1997:351n.18). If preverbal dative experiencer constructions involved a non-overt pro subject (cf. Kemenade 1997:334f), we would expect to find a similar preference for non-overt pro subjects with weather verbs and verbs that take clausal complements in OE. Since this does not appear to be the case, it seems more plausible to assume that the dative experiencer constructions in question already contain an overt subject, namely the dative experiencer. The subjecthood of preverbal dative experiencers may explain why lexical case-marking was retained longer in the preverbal experiencer constructions than on objects and passive subjects (Allen 1995:220,224,231). While there was no longer any morphological evidence for a distinction between lexical dative and structural accusative objects after the loss of the dative/accusative distinction in the pronominal paradigm, there would still have been a clear morphological distinction between lexical dative and structural nominative subjects. Although nouns lost the nominative/dative distinction during the ME period, personal pronouns distinguish between nominative and objective case even today. Since the experiencer in preposed dative experiencer constructions was typically a pronoun, there would have been sufficient morphological evidence to trigger the acquisition of lexical case marking (Allen 1995:100f,109, 111,231). What is more, the dative case on experiencer subjects “was useful in signalling the non-agentivity of the subject” (Allen 1995:446). The eventual replacement of preposed dative experiencers with nominative experiencers in all but a few frozen expressions during the 15th century is best seen as the last step towards the complete predominance of structural case assignment in English (Allen 1995:289). 1.4

Morphological changes in the pronominal paradigm during the Early Modern English period The loss of lexical case marking for experiencer subjects meant that by the start of the Early Modern English (EModE) period, the case system of English had basically reached its present-day form. The only morphological change yet to happen was the drastic simplification of the 2ps paradigm. At the start of the Early Modern English period, we still find distinct 2ps singular and plural forms, as well as a consistent morphological distinction between the nominative (NOM) and the general objective (OBJ) case (Table 1.12). However, Kjellmer (1986:446), drawing on the OED entry for you, argues that “you began to be used for ye between 1300 and 1400 and replaced it in general use by 1600”.

CHAPTER 1

21

Jespersen & Haislund (1949:265-268) suggest that the levelling of the nominative/objective case distinction may in part have been due to phonetic influences. Since the 2ps nominative and objective case forms differed only in their vowels, the phonetic distinction between thou and thee and also the distinction between ye and you would already have been neutralised in spoken English when the 2ps forms appeared in an unstressed position.11 According to Mustanoja (1960:125), the use of e rather than ou as the object of preie in (6) may be due to the lack of stress on the pronoun. (6) y preie e, seie e to me 1sg.NOM pray 2pl.NOM say 2pl.NOM to 1sg.OBJ “I pray you, you say to me” (Pecock Repressor 86) [Mustanoja 1960:125]

2ps singular

2ps plural

thou thee

ye you

intermediate stage suggested by Wallis, Priestley and Jespersen & Haislund

2ps singular

2ps plural

NOM OBJ

thou, you thee, you

ye you, ye

Present-Day English

2ps singular

2ps plural

you you

you you

Early Modern English NOM OBJ

NOM OBJ

Table 1.12. The 2ps pronominal paradigm in Early Modern English and Present-Day English (Allen 1995:210; Jespersen & Haislund 1949:129,270; Priestley 1761:91; Wallis 1765[1653]:98)

11 According to Jespersen & Haislund (1949:270), Shakespeare and Marlowe sometimes use you in emphatic position and ye as the unstressed form. Wales (1996:89) suggests that ye was used as an unemphatic objective 2pl form, and Jespersen & Haislund cite Mason’s English Grammar (p.49) as noting that Spenser uses you as the emphatic nominative and ye as the unemphatic nominative.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

22

Mustanoja (1960:124f) also notes that the 2sg objective form þe appears as an “unstressed form” of the 2sg nominative thou in a number of Middle English texts, “especially in connection with auxiliary verbs” and “in enclisis” (7). (7) wy seiste so why say-2sg.OBJ so “why do you say so?” (Robert of Gloucester 8972, MS B; seistou in other mss) [Mustanoja 1960:125]

Interestingly, we find a similar trend with 1pl pronouns. Jespersen & Haislund (1949:256) observe that unstressed 1pl subjects following a fronted auxiliary verb may surface as (u)s rather than we from the 15th century onwards (8). (8) a. hens must vs flee (The Towneley Plays, ed. England, EETS 1897:31) b. How shal’s get it? (Shakespeare, Timon of Athens:IV.iii.408) [Jespersen & Haislund 1949:256f]

Since (u)s cannot be plausibly analysed as a phonetically reduced form of the corresponding nominative, the occurrence of sentences like (8) indicates that structural case is weaker for subjects that occur after fronted finite verbs than for subjects that precede the finite verb. In Section 9.2, I will argue that the weakening of structural case in this context is due to the lack of an overt specifier-head relationship between the subject and the finite verb, which means that the subject position is no longer necessarily analysed as a nominative case position. While the weakening of structural case in postverbal subject position would be expected to have a bearing on the distribution of the case forms of all pronouns, only the 2ps paradigm could have been affected by the availability of alternative structural analyses for imperatives with a subject-related pronoun in postverbal position. As Jespersen & Haislund (1949:258-261) point out, three types of subject-related 2ps pronouns could follow the verb in EModE imperatives: postverbal subject pronouns, which bore nominative case (9); nonargument pronouns associated with heightened subject involvement (cf. Section 1.1), which bore the objective case (10); and object pronouns coreferent with the subject, which also bore the objective case (11). (9)

sit thou by my bedde [thou = postverbal subject pronoun; nominative case] (Shakespeare, Henry IV, Part 2:IV.v.182) [Jespersen & Haislund 1949:259]

CHAPTER 1

23

(10) dreed thee noght [thee = non-argument pronoun; objective case] (Chaucer, Legend of good women, in Skeat’s six-volume edition:1742) [Jespersen & Haislund 1949:259]

(11) set thee down [thee = object pronoun coreferent with subject; objective case] (Shakespeare, Love’s labour’s lost:IV.iii.4) [Jespersen & Haislund 1949:259]

Although the status of the pronoun is reasonably clear from the context and the nature of the verb in examples (9)-(11), the original difference in meaning between sentences with a non-argument reflexive (12a) and sentences without a non-argument reflexive (12b) had virtually disappeared by the EModE period. (12) a. I fear me. [me = non-argument pronoun] b. I fear. As a result, many intransitive imperative verbs could occur equally plausibly with a nominative postverbal subject pronoun as with an objective non-argument reflexive, no matter what the context. And in the absence of any clear meaning differences, the nominative and objective 2ps forms would have seemed completely interchangeable in many imperatives (13)-(14). (13) a. fare thou well (Shakespeare, The tempest:V.318) b. Far thee well (Shakespeare, Macbeth:IV.iii.34) [Jespersen & Haislund 1949:259]

(14) a. fare ye well (Shakespeare, The merchant of Venice:I.i.58) b. fare you well (Shakespeare, The merchant of Venice:II.vii.773) [Jespersen & Haislund 1949:259]

Jespersen & Haislund (1949:269f) propose that the eventual extension of the 2pl object form you to both singular and plural nominatives may also have been influenced by the phonological similarities between you and the 2sg nominative thou (cf. also Wales 1996:89). They note that the earliest instances of you in a nominative context appear to refer to single individuals, and that 17th century grammarians draw attention to an opposition between singular you and plural ye in colloquial English. Thus both Wallis (1765[1653]:98) and Priestley (1761:9n.m) observe that you is commonly used when addressing a single person, but ye can only have a plural referent (15)-(16).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

24

(15) Wallis’ observations on you and ye (1765[1653]:98) Notandum item apud nos morem obtinuisse […] dum quis alium alloquitur, singularem licet, numerium tamen pluralem adhibendi; verum tunc you dicimus, non ye. “Note also that it is customary with us […] to use the plural number in addressing someone, even when the singular would be appropriate; but then we say you, not ye.”

(16) Example discussed by Priestley (1761:9n.m) a. b.

You (= 2pl) are reading. / You (= 2sg) are reading. Ye (= 2pl) are reading. / * Ye (= 2sg) are reading.

In Chapter 5, I will propose that phonological similarities between certain pronoun forms have contributed in a similar way to the emergence of an alternative classification of 1sg, 3sg, 1pl and 3pl case forms. This alternative grouping of the pronoun forms will be argued to play an important role in the variability of pronoun case in Present-Day English. As we will see in subsequent chapters, the pronoun case variation found in Present-Day English occurs primarily in strong pronoun contexts. Setting aside the changes in the 2ps paradigm and the occurrence of us after fronted auxiliary verbs, we find very little variation or change in the distribution of pronoun case forms in weak pronoun contexts from the Early Modern English period onwards (cf. Denison 1996: 288). Since the demise of dative experiencer subjects, lone pronouns in the (preverbal) subject position of a finite declarative clause consistently surface in the nominative forms I, he, she, we, they, while lone pronominal objects of verbs consistently surface in the objective forms me, him, her, us, them (Allen 1995:210).12 Only the invariant 2ps and 3sg neuter forms you and it occur both in weak subject and in weak object position (Table 1.13).

preverbal subject of finite clause object of verb

1sg

3sgM

3sgF

1pl

3pl

2ps

3sgN

I

he

she

we

they

you

it

me

him

her

us

them

you

it

Table 1.13. The distribution of weak pronoun forms in Present-Day English

12 According to Wales (1996:15), the 3sgF nominative she was introduced in the Late Middle English period. Wales also notes that the source of this form is “much disputed”.

CHAPTER 1

25

This consistency of case choice in weak pronoun contexts indicates that, unlike the changes in the 2ps paradigm, the reanalysis of 1sg, 3sg, 1pl, and 3pl case forms is confined to the strong pronoun series. While the remainder of this book is largely dedicated to the analysis of pronoun case variation in strong pronoun contexts, the distribution of weak pronoun forms in Present-Day English will feature prominently in Chapters 2 and 3. Chapter 2 discusses some of the more influential formal approaches to case, and Chapter 3 takes a closer look at the weak/strong distinction in the English pronoun system. As we will see, theoretical discussions of case assignment tend to concentrate on the case of lone, unmodified noun phrases in canonical subject or object position. Since these positions are associated with the licensing of weak pronouns in English, the distribution of weak pronoun forms is captured rather well in formal approaches to case.

CHAPTER 2 FORMAL APPROACHES TO CASE AND THE THREE CASE CONSTRAINTS

2.0

Introduction Much of current syntactic theory assumes that despite the lack of overt morphological case distinctions on nouns, Modern English still has a healthy case system. The case forms of personal pronouns are often treated as the only remaining overt manifestations of this underlying case system, and the distribution of lone, unmodified pronoun forms (1) tends to be viewed as evidence that subjects of finite clauses receive nominative case, objects of verbs and prepositions receive accusative/objective case, and possessive noun phrases receive genitive case (e.g. McCreight 1988:5f,9f; Chomsky 1993; Pollard & Sag 1994; Burzio 2000).1 (1) a. b. c. d.

I/he/she/we/they welcomed Tom. (subject - nominative case) Tom welcomed me/him/her/us/them. (object of V - objective case) Tom left without me/him/her/us/them. (object of P - objective case) my/his/her/our/their house (possessive - genitive case)

The various formal accounts of case differ primarily in the relative importance they accord to particular syntactic configurations in the context of case assignment. While some approaches assume that case assignment is driven entirely by the semantic form/argument structure of a predicate, and is therefore independent of word order, many others argue that case assignment can proceed only when the noun phrase appears in a particular structural configuration with the case-assigning head. This chapter outlines the most important assumptions and case predictions of argument-based and configurational approaches to case, and presents evidence that we need to distinguish two types of structural case marking in order to account for the case of lone, unmodified pronouns in Modern English: Argument Case, which is determined by structural linking between cases and

1 Since Modern English lacks a morphological difference between dative and accusative case forms, I will refer to the case found in object and prepositional complement position as ‘objective’.

CHAPTER 2

27

arguments of a predicate; and Positional Case, which is checked in a spec-head configuration at Spell-Out. The structural linking approach that forms the basis of Argument Case is introduced in Section 2.1. The configurational approach that forms the basis of Positional Case is discussed in Section 2.2. The most important predictions of Argument Case and Positional Case are summarised in Section 2.3, which also introduces an additional Default Case constraint, and outlines the interaction and relative importance of the three case constraints in Modern English. The chapter concludes with a brief look at agreement, and argues that verb agreement is determined by the interaction of three agreement constraints that have similar properties to the three proposed case constraints. 2.1

Case and argument structure Kiparsky (1997) and Wunderlich (1997) offer the most detailed recent explorations of the links between case and argument structure. Both argue that the case of structural (= noun phrase) arguments is determined at a semantic rather than syntactic level of representation, and assume that structural linking between arguments and cases relies on shared structural features. Their approaches differ mainly in the detailed structure of the semantic levels and associated representations, and in the features assigned to different positions on the argument hierarchy. To avoid confusion, I will adhere to Wunderlich’s (1997: 48) feature system throughout this book, even when illustrating arguments put forward by Kiparsky.2 2.1.1 Case assignment by structural linking Wunderlich (1997:32,46-50) proposes that structural (= noun phrase) arguments of a predicate receive case through structural linking in the θ-structure (TS), an interface level between Semantic Form (SF) and Phrase Structure (PS). How many structural arguments can occur with a particular predicate is determined by the lexical category and the SF representation of a predicate. In English, adjectives can license only one structural argument, at the most; nouns and prepositions are able to license up to two structural arguments; and verbs can licence up to three structural arguments.3 The morphological case of a structural argument depends on its position in the argument hierarchy, which is 2

See Wunderlich (1997:48fn.15) for a discussion of the differences between Kiparsky’s and Wunderlich’s feature systems. 3 If bare noun phrase adverbs are treated as arguments of the verb (cf. Larson 1985:605f,620; Przepiórkowski 1998:239f), the number of structural arguments a verb can license will be even higher.

28

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

determined by the SF representation of a predicate and its arguments. At TS, the argument hierarchy position of every structural argument is encoded with the binary features [± higher] and [± lower]. These features are intended to be interpreted as follows: (2) Wunderlich’s (1997:48) system of structural features [+ higher] = there is a higher argument [- higher] = there is no higher argument [+ lower] = there is a lower argument [- lower] = there is no lower argument In a set of three arguments, the highest structural argument will have the feature-specification [- higher, + lower], the lowest structural argument will have the feature-specification [+ higher, - lower], and the intermediate argument will have the feature-specification [+ higher, + lower]. I would like to propose that three argument cases are available for structural linking in Modern English, namely, nominative, objective, and genitive case. Nominative case (NOM) has the feature specification [-higher] and occurs with all types of predicates. Objective (OBJ) and genitive case (GEN) are not specified for any features, but objective case is available only to [- N] predicates (i.e. verbs and prepositions), and genitive case is available only to [+ N] predicates (i.e. nouns and adjectives).4 Since adjectives are unable to license more than one structural argument (which is either identified with an argument of a higher predicate, or linked to nominative case because it has the structural feature [- higher]), genitive case only ever shows up on arguments of nouns.5 4

Although Wunderlich (1997:48) and Kiparsky (1997:476f) assume that NOM is the structural case not specified for any structural features, I would like to argue that in Modern English (ModE), OBJ is the underspecified structural case, and that NOM is specified for the feature [- higher]. The differences between my proposal (which focuses on ModE) and the proposals put forward by Kiparsky and Wunderlich (which are influenced by Old English and German) could be seen to reflect differences between the ModE case system and the case systems of German and Old English. Kyle Johnson (p.c.) suggests that the nominative is the default case in German, but that the objective is the default case in English. It seems plausible that the default case of a language should correspond to the least specified structural case in the θstructure of verbal predicates. 5 Only arguments realised as noun phrases count as structural arguments in the proposed approach (cf. Wunderlich 1997:38-42,46f). Arguments of a predicate that are realised as PPs are not structural for the purposes of structural linking. This means that in APs such as proud of his achievement, the noun phrase following of is not a structural argument of proud. Similarly, the documents is not a structural argument of destruction in the DP the destruction of the documents.

CHAPTER 2

29

The feature-based structural linking between arguments and cases is demonstrated in (3), which illustrates case-assignment to the structural arguments of double-object give in an active sentence in Modern English (cf. Wunderlich 1997).6 (3) She gave me the keys.

(example sentence)

give [x CAUSE [BECOME [y POSS z]]] λx λy λz [- higher] [+ higher] [+ higher] [+ lower] [+ lower] [- lower]

g

NOM [- higher]

gw

OBJ [ ]

(Semantic Form) (structural arguments at TS) (structural features of arguments) (argument cases associated with [- N] predicates in Modern English, and their structural features)

Argument-case linking obeys the Specificity Principle (4). That is, an argument with the feature [- higher] will be linked to the nominative case, because the nominative is specified for [- higher], whereas the objective is not. Arguments with the feature [+ higher], on the other hand, can only link to the objective case, because their feature specification is incompatible with the [- higher] feature of the nominative. (4)

Specificity Principle (cf. Wunderlich 1997:49; Kiparsky 1997:477) Each argument is linked with the most specific case compatible with its structural features.

The structural nature of the argument cases assigned in (3) becomes evident under passivisation. As mentioned in Section 1.3, lexical case is retained under passivisation, but structural case is lost. In Wunderlich’s approach, passivisation removes the highest structural argument from the TS hierarchy be6

Wunderlich’s (1997) SF representations of verbal predicates also include a referential situational argument (s). I have omitted this referential situational argument from my SF and TS representations, because it does not interact with the individual arguments of a verb, and therefore has no bearing on their feature specifications. In the approach advocated by Wunderlich (1997:38-43), the bracketed SF representation of double-object give corresponds to a binary lexical tree diagram representing the logical types of its constituents and the hierarchical relations between them. For example, POSS is a predicate of type <e,<e,t>> because it takes two entities (y and z in (3)) to form a proposition (t). Whether an individual argument (i.e. an argument of type e) is structural at TS is assumed to depend on the c-command relationships between the individual arguments in the lexical tree.

30

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

fore structural linking can apply.7 As a result, the following argument will bear the feature [- higher], and will be linked to the nominative case (5). (5) I was given the keys.

(example sentence)

give [x CAUSE [BECOME [y POSS z]]] λx λy λz λy λz [- higher] [+ higher] [+ lower] [- lower]

g

NOM [- higher]

g

OBJ [ ]

(Semantic Form) (structural arguments at TS) (outcome of passivisation) (structural features of remaining arguments) (argument cases associated with [- N] predicates in Modern English, and their structural features)

While the features of structurally case-marked arguments are determined solely in relation to other arguments on the argument hierarchy, the structural features of lexically case-marked arguments are (partially) pre-specified in the lexical entry of a predicate. For example, the lexical entry of the Old English experiencer verb lician specifies that the experiencer argument (x) receives the feature [+ higher].8 As can be seen from (6), this lexical specification of the features of one argument can affect the assignment of structural features and case to the remaining arguments of the predicate.9

7

Even though the argument removed from the TS hierarchy is no longer structural (and thus only able to be realised as a PP), it is still present as the highest argument in the SF representation of give. The presence of the agent argument in the SF of passives could explain why the nonstuctural agent argument appears to be able to control a PRO subject in sentences such as (i). I would like to thank Liz Pearce (p.c.) for drawing my attention to this issue. (i) The boat was sunk (by its owneri) [PROi to collect the insurance money].

8

My argument-case linking analysis for lician is based on Wunderlich’s (1997:52) analysis of the German experiencer verb gefallen, which appears to have the same argument structure properties as lician (cf. Allen 1995:141). As noted by Kate Kearns (p.c.), the assumption that x occupies a higher argument structure position than y in [x LIKE y], receives support from Wechsler’s (1995:35ff) Notion Rule. The Notion Rule prohibits any semantic representations where the lower argument of a predicate has a notion of the higher argument, but not vice versa (Wechsler 1995:36,47). 9 Primary sources are given in round brackets, secondary sources in square brackets. The full details of primary sources for examples originally taken from the Microfiche Concordance of Old English (COE) can be found in the guide to the concordance.

CHAPTER 2

31

(example sentence) (6) him gelicade hire þeawas 3sgM.DAT liked her virtues.NOM/ACC “He liked her virtues / Her virtues pleased him” (COE Chron D (Classen-Harm) 1067.1.35) [Allen 1995:142] lician [x LIKE y] λx λy [+ higher] [- higher] [+ lower]

[-lower]

! !

! !

DAT [+ higher] [+ lower]

NOM [ ]

(Semantic Form) (structural arguments at TS) (structural features specified in the lexical entry of lician) (structural features added by default) (structural features encoding the position of the arguments in the argument hierarchy) (argument cases available in Old English, and their structural features)10

Both Wunderlich and Kiparsky focus on verbal predicates in their analyses, but the structural case linking approach can also be extended to nominal, prepositional, and adjectival predicates (cf. Wunderlich 1997:48n.16). Nonverbal predicate types differ from verbal predicates in that they typically function as arguments of a higher predicate.11 Since the highest structural argument of a dependent predicate tends to undergo θ-identification with an argument of its host, structural linking associated with non-verbal predicates often affects only the lower argument(s) of such a predicate (cf. Higginbotham 1985:564; Wunderlich 1997:34). I am assuming that θ-identification of the highest argument does not lead to a change in the features of the remaining arguments on the argument hierarchy. This means that all of the lower structural arguments will have the feature specification [+ higher], and will link to the objective or genitive case (depending on the lexical category of the predicate). 10 As mentioned earlier, I am assuming that Old English and Modern English have different feature-specifications for verb-related argument cases. 11 Verbal predicates can obviously also function as arguments of a higher predicate (i), but unlike non-verbal predicates, they readily stand alone (ii).

(i) We expected [him to give her the keys]. We saw [her leave]. (ii) He gave her the keys. She left. Like Wunderlich (1997:33f), I assume that this independence of verbal predicates is due to the presence of a referential situational argument in the SF representation of a verb, which is able to interact with functional categories such as tense.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

32

The example in (7) illustrates how θ-identification and structural linking apply to the arguments of the preposition to in a sentence where the prepositional predicate functions as an argument of the verb run. The SF representation of the prepositional predicate follows suggestions by Jolly (1993:289). (7)

She ran to him.

(example sentence)

[[x RUN] & [BECOME [x AT y]]] λx λx λy

(combined SF of run and to) (structural arguments at TS) (θ-identification between the external argument of to and the external argument of run) (remaining structural arguments of to) (structural features of arguments)

! ! z--------- m λy [+ higher] [- lower] NOM [- higher]

g

OBJ [ ]

(argument cases associated with [- N] predicates in Modern English, and their structural features)

The availability of nominative case for structural arguments of nonverbal as well as verbal predicates is highlighted by the occurrence of nominative pronouns in absolutive and independent small clauses (8)-(10). (8)

Absolutive small clause with a prepositional predicate A dead man, and [I by] (Richard B. Sheridan, Dramatic works:333) [Jespersen 1949[1927]:374]

(9)

Independent small clause with a nominal predicate [She a beauty]! I should as soon call her mother a wit (Jane Austen 1894[1813], Pride and prejudice:333) [Jespersen & Haislund 1949:239]

(10) Absolutive small clause with an adjectival predicate [they dead], two men only would remain (Anthony Hope 1894, The prisoner of Zenda:227) [Jespersen 1946:57]

CHAPTER 2

33

In (8)-(10), the non-verbal predicates do not function as arguments of a higher predicate. This means, that their highest structural argument is unable to undergo θ-identification, and remains available for argument-case linking (11). (11) She a beauty!

(example sentence)

beauty [x BEAUTY] λx [- higher] [- lower]

g

NOM [- higher]

GEN [ ]

(Semantic Form) (structural arguments at TS) (structural features of arguments)

(argument cases associated with [+ N] predicates in Modern English, and their structural features)

Nouns differ from prepositions and adjectives in that their highest argument is referential. In Wunderlich’s approach, referential arguments interact with functional categories to determine “the anchoring of a linguistic expression in the external world” (1997:33f). This means that the referential individual argument of a noun may be θ-bound by a noun-related functional category such as D, just like the referential situational argument of a verb may be bound by a verb-related functional category such as T (cf. Higginbotham 1985:56; Wunderlich 1997:33f). θ-binding prevents an argument from being structurally realised, but like θ-identification, it has no bearing on the feature specification of the remaining arguments.12 The SF representation relevant for noun-related argument-case linking comes from the qualia structure of a noun (cf. Pustejovsky 1995). Pustejovsky (1995:85-87) argues that qualia structure specifies those aspects of a word’s meaning that constrain the interpretation of other words in the same syntactic environment. He distinguishes four qualia (12), whose values are expressed in terms of predicate relations (Pustejovsky 1995:76-78,85f). A simplified qualia structure for the noun novel is given in (13).

12 Note that my treatment of referential individual arguments of nouns does not entirely correspond to the approach advocated by Wunderlich. While Wunderlich (1997:34) argues that the referential argument of a noun does not participate in structural linking, I am assuming that the individual referential argument of a noun is able to undergo structural linking or θ-identification if it is not θ-bound.

34

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(12) The four qualia distinguished by Pustejovsky (1995:76,85f) (a) CONSTITUTIVE: the relation between an object and its constituent parts (b) FORMAL: properties that distinguish an object in a larger domain (c) TELIC: purpose and function of the object (d) AGENTIVE: factors involved in an object’s origin or ‘bringing about’ (13) novel QUALIA =

CONSTITUTIVE = hold (x,y) FORMAL = book (x) TELIC = read (w,x.y), possess (v,x) AGENTIVE = write (u,x.y)

where the arguments are restricted to the following types: x: physical object y: narrative w, v, u: animate individual and the dotted type x.y combines the properties of x and y (cf. Pustejovsky 1995:78,100f)

In Pustejovsky’s (1995:101-103) approach, the different qualia compete for projection into the syntax, and only a subset can be projected at any given time. I would like to argue that the FORMAL quale provides the referential argument of a noun, and is therefore obligatorily projected. The possible interpretations of the genitive pronoun in a phrase like his novel suggest that an additional structural argument could come from either the TELIC or the AGENTIVE quale. If the AGENTIVE quale is the source of the structural argument, the genitive pronoun will be interpreted as the author of the novel (14). If the TELIC quale is the source of the structural argument, then the genitive pronoun will be interpreted either as the reader, or as the person who possesses the novel (15).13

13

Wunderlich (1997:38-43) assumes that the c-command relationships between individual arguments in the lexical tree of a predicate determine which of the arguments are structural. It appears that for nominal predicates, an individual argument will be structural only if it c-commands all other individual arguments in a projected quale. Thus, x is structural in (14) and (15) because it is the only argument in the FORMAL quale; u is structural in (14) because it c-commands y and x in the AGENTIVE quale; and w is structural in (15) because it ccommands x.y or y in the TELIC quale.

CHAPTER 2

(14) his novel

(example phrase)

[x BOOK] ∧ [[u WRITE y] & [BECOME [x EXIST]]]

λx

λu

(15) his novel

(Semantic Form resulting from the projection of FORMAL and AGENTIVE qualia) (structural args at TS)

(example phrase)

[x BOOK] ∧ [w READ x.y] or [x BOOK] ∧ [w POSS x]

λx

35

λw

(Semantic Form resulting from the projection of FORMAL and TELIC qualia) (structural args at TS)

For event nouns like destruction, the projection of the AGENTIVE quale will yield the ‘active’ version (16), and the projection of the TELIC quale will yield the ‘passive’ version (17). (16) his destruction of the documents (active) (17) their destruction (by enemy agents) (passive) In the ‘active’ version, which is based on the projection of the AGENTIVE quale, the agent of the event is realised as a structural argument and linked with genitive case (18).14 In the ‘passive’ version, which is based on the projection of the TELIC quale, the theme of the event is realised as a structural argument (19). A structural linking approach to case is thus able to deal not only with case assignment to arguments of verbal predicates, but also with case assignment to the arguments of nouns, prepositions, and adjectives. Argument-case linking can even be extended to bare noun phrase adverbials, provided we follow suggestions by Larson (1985:605f,620) and Przepiórkowski (1998) that such adverbials should be treated as optional structural arguments of the verb, 14

Only arguments realised as noun phrases count as structural arguments in Wunderlich’s (1997:38-42,46f) approach. Arguments of a predicate that are realised as PPs are not structural for the purposes of structural linking. Since the documents appears as the complement of the preposition of in (16) and (18), it is not a structural argument of the event noun destruction in these examples.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

36

and occupy the lowest positions on the argument hierarchy. However, some instances of case marking in Modern English would appear to be influenced by factors other than the position of an argument on the argument hierarchy of a given predicate. Two examples of case marking that do not seem to arise from structural linking are discussed in the next section. (18) his destruction of the documents

(‘active’ version)

[x DESTRUCTION] ∧ [y CAUSE [BECOME [z NOT EXIST]]] (SF resulting from the projection of the AGENTIVE quale) λx λy (structural arguments at TS) λy (outcome of θ-binding) [+ higher] (structural features [- lower] of remaining g arguments) NOM GEN (argument cases [- higher] [ ] associated with [+ N] predicates in Modern English and their structural features)

(19) their destruction (by enemy agents) [x DESTRUCTION] ∧ [BECOME [y NOT EXIST]] λx

λy λy [+ higher] [- lower]

NOM [- higher]

g

GEN [ ]

(‘passive’ version) (SF resulting from the projection of the TELIC quale) (structural arguments at TS) (outcome of θ-binding) (structural features of remaining arguments) (argument cases associated with [+ N] predicates in Modern English and their structural features)

CHAPTER 2

37

2.1.2 Instances of case marking not predictable from structural linking The shortcomings of a strict structural linking approach to case assignment are particularly evident in Poss-ing gerunds (20), and in constructions involving dependent predicates where θ-identification with an argument of the host does not seem plausible (21). (20) You must excuse [my telling you] (Charles Dickens 1912 [1865], Our mutual friend:28) [Jespersen 1946:148]

(21) a. We expected [him to give them the keys]. b. We consider [him innocent]. c. [For him not to give them the keys] would have been unexpected. Poss-ing gerunds are characterised by the presence of a genitive subject (my) and the possible occurrence of a structural object (you) (cf. Abney 1987). In the structural linking approach outlined here, genitive case is found only on arguments of [+ N] predicates. This means that we will have to treat telling as a noun if we want to account for the genitive case on the subject. However, nouns can only ever license one structural argument in addition to their referential argument. Any other arguments must be realised as prepositional phrases (22). (22) a. his destruction [of the documents] b. * his destruction [the documents] An analysis of Poss-ing gerunds as verbal predicates will correctly predict the possible appearance of structural objects in the objective case (23), but it cannot account for the genitive case of the subject. (23) You must excuse [my telling him]. As discussed above, the highest structural argument of a verbal predicate must be linked with nominative case, and any lower arguments will link with objective case. This means that the subject of the gerund would be predicted to bear nominative rather than genitive case (24). (24) I telling him

38

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

A similar mismatch between the predictions of structural linking and the actual case form of arguments occurs when a non-finite clause containing an overt subject appears as the argument of a higher predicate (25).15 (25) We expected [him to give them the keys]. The structural linking approach outlined above requires the highest structural argument of any predicate to be linked to nominative case unless it is θ-identified with an argument of a higher predicate.16 This means that the highest argument of give will have to be analysed as θ-identified with an argument of expect in (25).17 However, the semantic properties of expect make such an analysis highly implausible. As (26) illustrates, the verb expect only ever takes two arguments: the person expecting, and either the expected person (26a) or the expected event (26b). (26) a. We expected him. b. We expected [that he would give them the keys]. c. * We expected him [that he would give them the keys]. The interpretation of sentences where expect is followed by a non-finite clause (25) corresponds to the interpretation of sentences where expect is followed by a finite clause (26b). This suggests that him is not an argument of expect in the non-finite clause construction (25). A θ-identification analysis is equally problematic for the subjects of nonfinite clauses introduced by for (27). Nominative case linking to the highest argument of give in (27) can be prevented only by θ-identification of the subject with the lowest argument of for. Such an analysis seems plausible for sentences like (28), where there is a noticeable intonation break after for him, and for has a clearly prepositional interpretation. In (27), however, for does not seem to contribute any prepositional meaning to the sentence, and behaves like a functional head rather than a lexical predicate. (27) [For him not to give them the keys] would have been unexpected. (28) For him, [not to give them the keys] would have been unthinkable. 15

The same problems arise with embedded non-verbal predicates that contain an overt subject, such as They considered [him innocent/a fool]. 16 The referential individual argument of a noun may also be θ-bound by a functional category. 17 See Przpiórkowski (1998:236-238) for an HPSG treatment of raising constructions along these lines.

CHAPTER 2

39

Case-marking in non-finite clauses and gerunds clearly poses a problem for a strictly argument structure-based approach to case assignment. This suggests that not all case-marking in Modern English can be accounted for by argument-structure linking. In Chapters 3, 5 and 8, I will provide further evidence that casts doubt on the assumption that “Case and agreement are entirely independent of word order” (Johnson & Lappin 1999:83). As we will see, many of the pronoun case trends observed in Modern English point to an increasing influence of surface position on case marking. 2.2

Case and structural position The syntactic configurations most often argued to be involved in case assignment are the head-complement relationship (29), which forms the basis of the most restrictive definition of government, and the specifier-head relationship (30), which forms the basis of spec-head agreement. (29)

XP ei ZP X' ei X --> YP

The head X governs its complement YP18

(30)

XP ei ZP X' : ei z> X YP

The head X can undergo spec-head agreement with its specifier ZP

Lexical case assignment is generally assumed to take place in conjunction with theta-marking (θ-marking) when a noun phrase enters the derivation (cf. Chomsky 1995:386n.55). Since lexical case assignment is tied to θ-marking, only lexical heads are able to assign lexical case. Whereas lexical case assignment tends not to be explicitly restricted to a particular structural configuration, many configurational approaches assume that structural case can only be checked in a spec-head configuration. In most current configurational approaches, a noun phrase has to move out of its base-position to check structural case via spec-head agreement with a functional head. Since structural case is often treated as the flip-side of φ-feature agreement, the functional heads associated with structural case checking tend to be agreement-related (cf. Chomsky 1993, 1995, 2000; Schütze 1997).19

18

In most treatments of government, the head will also govern its specifier ZP. As mentioned in Chapter 1, the term ‘φ-features’ is used as shorthand for person, gender, and number features. 19

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

40

Early treatments of case assignment via spec-head agreement assume that all structural case checking takes place in special agreement phrases (AgrPs). The nature of the case checked by a particular AgrP is determined by the properties of a lower head that raises to Agr (cf. Chomsky 1993:7): An Agr containing T(ense) checks nominative case, an Agr containing the V head of a verb phrase checks objective case (31). (31)

AgrsP ei Agrs' DPk [nom] subject ei TP Agrs [nom] 2 3 Ti Agrs T AgroP ei ti Agro' DPl [obj] shifted object ei Agro [obj] VP 2 3 Agro DP V' Vj surface position tk 3 of lexical verb V DP tj tl

As can be seen from (31), the subject DP must move to [Spec,AgrsP] to check nominative case, and the object DP must move to [Spec,AgroP] to check objective case. Since object noun phrases follow the lexical verb in English, the movement of the object DP to the [Spec,Agro] position is generally assumed to take place after Spell-Out. In more recent discussions, case checking no longer involves separate Agr projections, but is associated with functional heads that are already required for other purposes (cf. Chomsky 1995:349-355). Chomsky (1995:282; 2000) proposes that a finite transitive sentence has the basic structure given in (32). As can be seen from (32), the functional heads involved in (verb-related) case and agreement ([φ]-feature) checking are C, T, and v. Finite T checks nominative case (in conjunction with C), and v checks objective case. Such a ‘bare phrase structure’ approach forms the basis of Positional Case (cf. Section 2.3), and will therefore be discussed in more detail in the following sections.

CHAPTER 2

41

(32)

CP 3 wh-phrase C’ 3 C TP [φ] ei subject T’ [nom] wo T vP [φ] wo [nom] shifted object v’ [obj] wo external argument v’ (base position) wo v VP [φ] ei [obj] direct object V’ (base position) 4

2.2.1 vP and case checking The addition of the light verb v to the verb phrase was inspired by Larson’s (1988) VP-shell analysis of double object constructions, and Hale & Keyser’s work on argument structure (cf. Hale & Keyser 1993, 1998). Following suggestions in Hale & Keyser (1993), Chomsky (1995:315) argues that the v-VP configuration expresses “the causative or agentive role” of the external argument.20 Since arguments receive their θ-role when they enter the derivation, the external argument of any (agentive/causative) transitive verb (33) has to be base-generated in [Spec,vP].21 The same goes for the sole (external) argument of unergative verbs (34).22 As can be seen from (33), the internal argument of a simple transitive verb is base-generated in [Spec,VP]. For ditransitive verbs such as double-object give, one of the internal arguments is base-

20 Note that the external argument in [Spec,vP] receives its agent/cause θ-role from the v-VP complex rather than v alone. v is a functional rather than lexical head and is therefore unable to assign a θ-role and/or lexical case (Chomsky 2000:102). The assumption that v is an (agreement-related) functional head is crucial to the proposal that v is able to check the structural case of an object that has raised to a higher [Spec,vP] position. 21 In Chomsky (2001:8) this base-generation of an argument in a θ-position is referred to as ‘external Merge’, in contrast to ‘internal Merge’ which involves movement to a derived position. According to Chomsky, “[a]rgument structure is associated with external Merge (base structure); everything else with internal Merge (derived structure)” (2001:8). 22 For a more detailed discussion of causative, unergative, and unaccusative verbs see Brousseau & Ritter (1991), Levin & Rappaport Hovav (1995), and Hale & Keyser (1998).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

42

generated in [Spec,VP], and the other is base-generated as the complement of V (35). (33)

vP 3 DP v’ They 3 v VP 3 DP V’ him 5 freed

The base positions of internal and external arguments in the transitive sentence They freed him.

(35)

vP 3 DP v’ She 3 v VP 3 DP V’ me 3 V DP gave the keys

(34)

vP 3 DP v’ She 3 v VP 5 laughed

The base position of the (external) argument in the unergative sentence She laughed.

The base positions of the internal and external arguments in the ditransitive sentence She gave me the keys.

Passive verbs (36) and unaccusatives (37) do not license a structural agent or causer. This means that no vP-layer is present, and all argument DPs are basegenerated within VP. (36)

VP2 3 VP1[passive] V2 was 3 DP V’1 I 3 DP V1 given the keys

The base position of the (internal) arguments in the passive sentence I was given the keys. Note that neither the lower verb give nor the passive auxiliary be project a vP-layer.

(37)

VP ei DP V’ They 5 melted

The base position of the (internal) argument in the unaccusative sentence They melted.

CHAPTER 2

43

The vP-layer is similarly absent from copular constructions (38) and sentences involving identificational be (39). Since v is assumed to be required for objective case checking in Chomsky’s (1995) approach, the absence of a vP-layer in (39) means that objective case checking is impossible in constructions involving identificational be. As we will see in Chapters 5 and 8, this has important consequences for pronoun case in it-clefts and it BE sentences. (38)

VP 3 V PP were 3 DP P’ we 3 P DP in the kitchen

The base position of the (internal) arguments in the copular sentence We were in the kitchen.

(39)

VP 3 DP V’ I 3 V DP am me

The base position of the (internal) arguments in the identificational sentence I’m me (advertisement for Cachet perfume) [Wales 1996:95]

Chomsky’s (1995) proposal that v is not only involved in assigning the agent/cause θ-role to the external argument of a verb, but is also responsible for checking objective case, has its origins in Burzio’s Generalization (40). (40) Burzio’s Generalization (adapted from Burzio 1986, 2000) If a verb does not assign an external θ-role, it will not assign structural objective case to its highest internal argument. This generalization falls out naturally from the argument linking approach outlined in Section 2.1.1, where passivisation removes the highest argument of a verbal predicate from the argument hierarchy, and assigns the feature [- higher] to the argument next in line. The Specificity Principle ensures that this argument will be linked to the nominative, which is also specified for the feature [- higher], rather than the objective, which is not specified for any particular features. In configurational approaches, on the other hand, the link between argument structure and case assignment has generally had to be stipulated. In order to make an explicit structural connection between the assignment of an agent/ cause role and the availability of structural objective case, Chomsky (1995: 355-360) posits that v may project more than one specifier (41).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

44 (41)

vP wo Spec2 v’ moved object ei [obj] Spec1 v’ external arg 3 [agent/cause] v [obj] VP 1 4 tk Vk v

The external argument is base-generated in the lower specifier position (Spec1) where it receives the agent/cause role. Since structural case is only checked in derived positions, the external argument has to move out of [Spec1,vP] to check case (i.e. a noun phrase cannot check structural case in the position where it receives its θ-role). The object, which is base generated in [Spec,VP], moves to [Spec2,vP] to check objective case. The lexical verb V always raises out of its base-position and adjoins to v. Since the object generally follows the verb in Modern English surface syntax, the raising of the object DP to [Spec,vP] and the checking of objective case in this position are usually treated as covert.23 However, as I will try to demonstrate in the remainder of this book, there are good reasons to believe that some object DPs undergo overt movement to [Spec,vP], that the lexical verb always raises to the head of a functional projection dominating vP, and that all case checking takes place before Spell-Out. Support for such an analysis comes form the different behaviour of pronouns and full noun phrases in V-particle constructions. As a comparison of (42) and (43) shows, full noun phrase objects are able to appear both after the particle (42a) and between the verb and the particle (42b) in V-particle constructions. Unstressed unmodified lone pronouns, on the other hand, are only able to appear between the verb and the particle (43b). (42) a. b.

Betsy threw out her boyfriend. Betsy threw her boyfriend out.

(43) a. * Betsy threw out him. b. Betsy threw him out. (examples adapted from Johnson 1991:593f)

23 Strictly speaking, the covert movement is assumed to involve only the formal features of DP, which adjoin to the head v rather than v’ (cf. Chomsky 1995:360f,370f). Since I am going to adopt a more surface-oriented approach, where all case checking must happen at Spell-Out, I will not discuss covert movement in any more detail in this book.

CHAPTER 2

45

V-particle constructions need to be distinguished from verb phrases containing directional PPs (cf. Radford 1988:459f).24 Directional PPs differ from particles in that they may be modified by degree adverbs (44), and frequently contain noun phrase complements (45). They also tend to permit a wider range of interpretations than are available for V-particle constructions. (44) a. b. * c. d. *

She threw her boyfriend/him out of the window. She threw out of the window her boyfriend/him. She threw her boyfriend/him out of her flat. She threw out of her flat her boyfriend/him.

(45) a. She threw her boyfriend/him right out. b. * She threw right out her boyfriend/him. As can be seen from (44) and (45), directional PPs are unable to intervene between the verb and its object, even when the object is a full noun phrase and the directional PP has roughly the same interpretation as the particle (44c-d). The differences between (44)-(45) and (42)-(43) suggest that the syntactic status of directional PPs differs from the syntactic status of particles in V-particle constructions. If we assume that objects are base-generated in [Spec,VP], then the easiest way to guarantee the word order in (44a), (44c), and (45a) is to assume that directional PPs are complements of V. Particles, on the other hand, are most plausibly analysed as occupying a position outside VP. Following a suggestion by Kate Kearns (p.c.), I will assume that the telic particle out heads its own phrase (TelP), which immediately dominates VP (46)-(47).25 In the proposed approach, the object noun phrase in a V-particle construction is most plausibly analysed as occupying [Spec,VP] at Spell-Out 24

I would like to thank Kate Kearns (p.c.) for drawing my attention to this issue. Nicol (2002) also assumes that the particle heads its own phrase (wP), but he does not specify the semantic properties of this phrase. The analysis of the particle as the head of a phrase associated with telicity draws on proposals by Sawai (1997) and Ritter & Rosen (2000:202ff), who argue that particles head a delimiting phrase (FP-delim). It also ties in with Bowers’ (2002:191ff) suggestion that telicity markers in Scottish Gaelic appear as the head of a predicate phrase (PrP). However, there are certain important differences between the TelP analysis these alternative proposals. As Kate Kearns (p.c.) notes, “telicity operates at the level of the inner event” and does not require the presence of an external argument. I therefore assume that TelP immediately dominates a VP containing only internal arguments. FP-delim, on the other hand, is assumed to dominate the base positions of both internal and external arguments, while PrP may dominate either VP or a transitivity phrase. The properties of Tel also differ from the properties of F-delim and Pr in that Tel is unable to project a specifier, and is also unavailable as a target for V-movement. TelP is thus neither involved in objective case checking, nor does it establish the predication relation between the subject and the predicate of a clause.

25

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

46

when it follows a particle (46), and [Spec,vP] when it precedes the particle (47). vP (47) TP 3 3 v TelP DPi [nom] T’ 1 2 She 3 VP T [nom] FP Vk v Tel threw out 2 3 DP V’ F vP her boyfriend 4 1 3 vl F DPj [obj] v’ tk 1 him 2 ti v’ Vk v 2 F = functional head that is unrelated to the argument structure v [obj] TelP of the verb. Since the exact identity of the functional head tl 2 the verb moves to does not have any direct bearing on my Tel VP analysis of case checking, I have decided to leave the category out 2 of this functional head unspecified. See Banerji (2003:70n.8) tj V’ and Quinn & Banerji (2003) for evidence that lexical verbs move 4

(46)

to the head of a Focus Phrase between TP and vP in English.

tk

The proposed analysis allows us to argue that the ungrammaticality of (48) arises from the licensing requirements of weak pronouns in English. (48) * She threw out him. While full noun phrases may remain in their θ-position throughout the derivation, unstressed unmodified lone pronouns are weak and can only be licensed if they raise out of their θ-position into the specifier of a functional head associated with Positional Agreement in Present-Day English, that is, T (dominated by C), v, and D (cf. Section 2.4 and Chapter 3 for more discussion). Since weak object pronouns must move into the second [Spec,vP] position before SpellOut, they will automatically precede a particle in Tel (47), and sentences such as (48) will be ruled out. In (47), both the subject pronoun she and the object pronoun him appear in a spec-head relationship with an agreement-related functional head. As a result, both are able to undergo case checking. The object pronoun checks objective case with v, and the subject pronoun checks nominative case with T. To ensure the correct surface order between the verb and the object, we need to assume that the V-v complex undergoes movement to a higher functional head before Spell-Out (cf. Johnson 1991:613; 1996:24,38). Since the object pronoun precedes the particle in sentences like (49), where T

CHAPTER 2

47

is clearly filled by the auxiliary have, I am assuming that the functional head in question occupies a position between T and v. (49) She has thrown him out. Further support for overt verb movement to a position above v comes from locative inversion (50) and deictic there constructions (51).26 As illustrated in (50)-(51), full noun phrase subjects are able to occur after the verb in such constructions, but pronominal subjects are not. (50) a. In came Sue. b. * In came she.

(51) a. There goes John. b. * There goes he.

If we assume that the verb has raised to a functional projection above v and below C in (50)-(51), then we can argue that (50b) and (51b) are ungrammatical because the weak pronoun has failed to raise out of its θ-position.27 (50a) and (51a), on the other hand, will be fine, because full noun phrases do not need to raise to the specifier of an agreement-related functional head to be licensed. Given that only weak pronouns are subject to agreement-related licensing requirements (cf. Chapter 3), we might expect strong pronouns to have the same distribution as full noun phrases in V-particle constructions, locative inversion, and deictic there constructions. As we will see in the following chapters, modified and coordinated strong pronouns, are indeed able to occur in the same position as full noun phrases in these contexts. However, the occurrence of lone unmodified strong pronouns in post-particle position and in post-verbal subject position is extremely rare, and appears to be restricted to certain deictic uses, where the pronoun is strongly stressed, carries a noticeable pitch movement, and is accompanied by a pointing gesture (52). 26 For an in-depth discussion of inversion constructions involving locative or directional PPs see Levin & Rappaport Hovav (1995:215-277), and Bresnan (1994). For more detail on deictic there-constructions see Lakoff (1987:462-585). For additional evidence that lexical verbs undergo overt movement in Modern English, see Gelderen (1997:132-145). 27 This base position will either be [Spec,vP] or [Spec,VP], depending on our analysis of the argument structure of these verbs. Collins (1997:27) assumes that locative inversion is only possible when the verb is treated as an unaccusative, and therefore analyses the postverbal noun phrase as occupying [Spec,VP]. However, if the lexical verb is able to move to a position beyond vP before Spell-Out (as I am arguing here), then the postverbal noun phrase could also be base-generated in [Spec,vP]. This would allow us to capture the fact that many of the verbs that occur in locative inversion constructions are basically unergative, even though the whole construction receives an unaccusative interpretation (cf. Levin & Rappaport Hovav 1995:215277).

48

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(52) a. In came ) SHE. (the traitor!) b. Betsy threw out ) ME. (of all people - how could she have!) c. Betsy threw out ) HIM? (I can’t believe it!) A possible explanation for the general marginality of pronouns in post-verbal subject position and in post-particle position comes from research into the mapping between syntax and Information Structure. Vallduví & Vilkuna (1998), Ambar (1999), and Ouhalla (1999:337f) discuss evidence from a variety of languages which suggests that when movement out of a θ-position is possible but not obligatory, a DP that remains in its base position tends to receive a rhematic (i.e. new information) interpretation at Information Structure. Since all pronouns are fundamentally topics (i.e. given information), and thus not very suited to serving as rhemes, we would expect even strong pronouns to exhibit a distinct preference for raising out of their base position, especially when they are uncoordinated and appear without any modifiers.28 Since the evidence from V-particle constructions, locative inversion, and deictic there-constructions suggests that lexical verbs must be able to move past vP before Spell-Out in Modern English, I will assume from now on, that the lexical verb always undergoes overt movement to a functional head beyond vP. I also assume that all case checking happens at Spell-Out and can therefore only apply to pronouns that have raised to the specifier of an agreement-related functional head before Spell-Out. 2.2.2 The roles of C and T in case checking Chomsky (2001:13) argues that T is not solely responsible for checking the case of a noun phrase (DP) in [Spec,TP], but always interacts with C (cf. also Watanabe 1996:12, 19; Bittner & Hale 1996; Nash 1997:143; Cormack 1999:58). In a finite clause, C endows T with the ability to check nominative case on the DP in its specifier (53). Since the lexical verb raises to the functional head F before Spell-Out, its surface position precedes the trace of the subject DP in the tree diagram.

28 Compare Dehé’s (2002:112,125,201f,245) suggestion that the object-particle order in Vparticle constructions results when objects that fail to introduce new information raise out of their base position within the VP. As Toivonen (2003:168) notes, strong pronouns following particles tend to have an “unusual discourse role” (cf. also Bolinger 1971:39f).

CHAPTER 2

(53)

CP 3 C TP [+ finite] 3 DPi [nom] T’ he 3 T [nom] FP 6 laughedk ti tk

49

Nominative case checking on the preverbal subject of a finite clause

In a non-finite clause, T will only be able to check the case of an overt subject DP if C is filled with the complementizer for (54).29 Since a C containing for is [- finite], T checks objective rather than nominative case. Alternatively, we could argue that the complementizer for has properties similar to T and v (cf. Kayne 2000:314-326; 2001), and is therefore able to attract the subject pronoun to its specifier to check objective case. An analysis along these lines would require us to posit that for moves to a higher functional head (F2) before SpellOut (55). When a to-infinitive with an overt subject DP is not introduced by for, the CP-layer is assumed to be absent, which means that T is unable to check case, and the subject DP has to raise to the specifier of the vP projected by the matrix verb to check objective case (56).

29 As Law (2000:172-177) and Wurmbrand (2001:114) point out, evidence from negative sentences suggests that the inifinitive marker to occupies a position lower than T at Spell-Out. If to appeared in T at Spell-Out, we would expect it to precede not in neutral negative sentences, just like finite auxiliaries do (i).

(i) a. He did/would not laugh. b. He has not laughed. c. He did/would/has not. (VP-ellipsis)

d. * He not did/would laugh. e. * He not has laughed. f. * He not did/would/has.

However, infinitival to can only precede not in sentences involving constituent negation (ii). In order to be interpreted as negating the whole non-finite clause, not has to precede to (iii). This is particularly evident in VP-ellipsis constructions, which are compatible only with sentential negation (iv). (ii) He tried [to not win]. (constituent negation) (iii) a. b.

[For him not to win] would be unexpected. (sentential negation) We expected [him not to win]. (sentential negation)

(iv) a. We expected [him not to]. (VP-ellipsis) b. * We expected [him to not]. For a more detailed discussion of the syntactic and semantic differences between sentential and constituent negation see Potsdam (1998:142-147).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

50

(54)

CP 3 C [- finite] TP for 3 T’ DPi [obj] him 3 FP T [obj] 6 to laughk ti tk

Case checking involving C and T in non-finite clauses introduced by for

(55)

FP2 3 F2 FP1 3 fork DPi [obj] F1’ him 3 F1[obj] TP 2 tk ti T’ 6 to laughj ti tj

Alternative analysis suggested by Kayne’s (2000, 2001) treatment of complementizers

(56)

FP 3 Case checking on the subject of the F vP embedded clause in the sentence 1 3 vl F DPi [obj] v’ We expected him to laugh. 1 him 3 (twe = trace of the matrix subject) Vn v twe v’ expected 3 VP v [obj] tl 3 V TP 3 tn DP T’ 6 ti to laughk ti tk

2.2.3 Case checking within DP In many recent discussions of case checking within DP, D is assumed to interact with a lower functional head in checking genitive case on a DP in the specifier of the lower head (cf. Kayne 1994:26,85f; Siloni 1997:41f; Alexiadou & Wilder 1998). For English, the most readily justifiable functional projection between DP and NP is NumP. NumP is headed by cardinal determiners such as a/an, no, many, and two, and hosts the number features of a noun phase (cf. Lobeck 1995:80-99; Ritter 1991:50-58).30

30 For a discussion of the differences between proportional and cardinal determiners see Kearns (2000:73-76). Like Lobeck (1995:84), I assume that only definite noun phrases project a DPlayer. In an indefinite noun phrase such as two novels, the highest functional projection would be NumP.

CHAPTER 2

51

Since both D and Num are agreement-related functional heads whose presence in the phrase structure is already required for non-case purposes,31 both D and Num qualify as potential case checkers in a bare phrase structure approach. I will nevertheless follow Abney’s (1987:83f,271) suggestion that D alone is responsible for genitive case checking within DP, and that the genitive noun phrase occupies [Spec,DP] at Spell-Out. As discussed earlier, structural case must be checked in a derived position, i.e. in a position distinct from that associated with θ-role assignment. In English, DPs with genitive case marking always precede the adjectives and cardinal determiners within a noun phrase (57). This suggests that the lowest possible base position for genitive DPs in an English noun phrase is [Spec,NumP] (cf. Ritter 1991:47). (57) a. his/whose/Kim’s two new novels b. * two his/whose/Kim’s new novels c. * two new his/whose/Kim’s novels If we assume that genitive noun phrases are base-generated in [Spec,NumP] and receive their θ-role there, then the only derived position available for genitive case checking is [Spec,DP] (58).32 (58)

DP 3

Case checking by D in the noun

DPi [gen] D’ phrase his/whose two novels his 3 whose D [gen] NumP 3 DP Num’ 3 ti Num NP two novels

31 Num specifies the number features of a noun phrase, and D is required for the θ-binding of the referential argument of the noun (cf. Higginbotham 1985:560; Wunderlich 1997:34). 32 While genitive DPs may have either an AGENTIVE or a TELIC interpretation (cf. Section 2.1.1), structural AGENTIVE and TELIC arguments never cooccur in an English noun phrase. I will therefore assume that there is only one θ-position for structural arguments in a noun phrase. As mentioned in Section 2.1.1, only arguments realised as noun phrases count as structural arguments in the approach adopted here (cf. Wunderlich 1997:38-42,46f). Arguments of a predicate that are realised as PPs are not considered to be structural. For example, his is a structural argument of the noun destruction in the DP his destruction of the documents, but the documents is not, because it is contained in a PP headed by of.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

52

Unlike structural linking (cf. Section 2.1.2), case checking by D can account for the occurrence of genitives in Poss-ing gerunds as well as noun phrases, provided we adopt an analysis of Poss-ing gerunds along the lines proposed in Abney (1987). Abney (1987:223f) assumes that gerunds contain an abstract morpheme -ING, which adds the categorical feature [+ N] to the projection it attaches to. For Poss-ing gerunds, -ING attaches to VP and turns it into an NP dominated by a complete DP-layer. The subject of a Poss-ing gerund raises to [Spec,DP] and checks genitive case with D. For Acc-ing gerunds, -ING attaches to TP and turns it into a DP. The subject surfaces in [Spec,TP] and is unable to check structural case. While Abney’s (1987) proposal successfully accounts for the genitive case found on the subjects of Possing gerunds, some aspects of his analysis has to be modified if it is to fit in with current assumptions about phrase structure. As Kate Kearns (p.c.) points out -ING behaves like a functional head, and should therefore be assigned a consistent category and complementation. Since both Poss-ing and Acc-ing gerunds appear to be temporally independent, -ING is best analysed as taking a TP as its complement.33 If we assume that -ING belongs to the category Num and optionally projects a DP-layer, then Poss-ing gerunds could be argued to arise from the projection of this optional DP-layer above NumP, while Acc-ing gerunds result when the DP-layer fails to be projected.34 The presence of the DPlayer in Poss-ing gerunds allows the subject of the gerund to check genitive case in [Spec,DP] (59). As can be seen from the analysis of Poss-ing gerunds (59) and non-finite clauses (54)-(56), a configurational treatment predicts the pronoun case forms found in these constructions more accurately than structural linking. Since case checking captures the distribution of pronoun case forms in finite clauses and noun phrases just as well as case linking, it might seem tempting to completely abandon the structural linking approach in favour of a purely configurational approach to case marking in Modern English.

33 Wurmbrand (2001:100f) suggests that infinitives project a TP-layer if they permit independent temporal reference in the form of a temporal adverbial or the prefective auxiliary have. As the grammaticality of (i) and (ii) demonstrates, perfective have appears readily in both Poss-ing and Acc-ing gerunds, which suggests that both contain a TP-layer.

(i)

There is no record of [his ever having lost his temper] (Stacy Aumonier 1916, Olga Bardel, London:74) [Jespersen 1946:111]

(ii) There is no record of [him ever having lost his temper] 34

As Kate Kearns (p.c.) notes, Poss-ing gerunds bear a structural resemblance to definite noun phrases in this analysis, and Acc-ing resembles indefinites.

CHAPTER 2

(59)

53

DP 3 Case checking by D and v in the DPi [gen] D’ my 3 Poss-ing gerund my telling him D [gen] NumP 3 -ING TP 2 T’ tj 2 T FP 3 F vP 1 3 v’ vl F DPj [obj] 1 him 2 Vk v ti v’ telling 3 VP v [obj] 5 tl tj t k

However, there is one instance of case marking that is straightforwardly predicted in a structural linking approach to case, but has proved notoriously difficult for case checking analyses: the case of prepositional objects. 2.2.4 The case properties of prepositions In the structural linking approach outlined in section 2.1.1, the case marking of a prepositional object is determined in the same way as the case of a verbal object. The case of the argument of any predicate is affected by the nature of the predicate and the position of the argument on the argument hierarchy. Both verbs and prepositions have the category feature [- N], which means that their arguments may be linked either with nominative case or with objective case. The linking of arguments with cases is feature-driven. Any argument available for structural linking must be associated with the case whose structural feature specification is most closely compatible with the structural features of the argument itself. The structural features of the different cases are determined by parameter setting. In Modern English, nominative case is specified as [- higher], whereas objective case is unspecified for structural features.35 The structural features of an argument are determined by its position on 35 That is, objective case is the elsewhere case, which will be linked to any structural argument that does not meet the feature specification for the nominative (cf. Section 2.1.1 for further discussion).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

54

the argument hierarchy of a predicate. Since any object of a verb or preposition is preceded by the external argument of the verb/preposition on the argument hierarchy, it will bear the structural feature [+ higher]. This feature specification is incompatible with the [- higher] feature of the nominative, which means that any verbal or prepositional object will surface with objective case (cf. Section 2.1.1 for a detailed discussion). In the case checking approach discussed in Sections 2.2.1-2.2.3, on the other hand, a noun phrase is only able to check structural case if it appears as the specifier of an independently motivated agreement-related functional head at Spell-Out. This means that we will have to assume either that argumenttaking prepositions are agreement-related functional heads that are able to participate in case checking (60),36 or that a separate agreement-related functional head is present within the prepositional phrase (61). Since the preposition precedes its object in surface syntax, we will also have to posit a further functional head that the preposition must raise to before Spell-Out. (60)

FP 3 F PP 1 3 Pj F DPi [obj] P’ 2 base pos. P’ of higher 2 argument P [obj] ti tj

Preposition-related projections needed for structural case checking if P is treated as an agreementrelated functional head

(61)

FP2 3 FP1 F2 1 3 F1k F2 DPi [obj] F’1 1 3 Pj F 1 F1 [obj] PP tk 3 base pos. P’ of higher 2 argument tj ti Preposition-related projections needed for structural case checking if P is assumed not to check structural case

Neither (60) nor (61) is easy to justify. As mentioned in Section 2.2.2, the complementizer for could be argued to be an agreement-related functional head that is able to participate in objective case checking in the manner outlined in (55). An analysis of argument-taking prepositions as agreement-related functional heads is much less plausible, though. The functional heads v, T, C, and D do not take DP arguments and are unable to assign θ-roles (cf. Chomsky 2000: 102). Even the agent/cause role is assigned by the v-VP complex rather than v

36 See Zubizarreta (1998:25-27) for a proposal that P is involved in the (covert) checking of the formal features of the object DP.

CHAPTER 2

55

(cf. Section 2.1.1). Prepositions such as to, with, about, between, on the other hand, clearly are able to take DP arguments and θ-mark them. If the preposition itself is not able to participate in structural case checking, we will have to assume that the prepositional phrase contains an additional agreement-related functional head (61).37 However, while functional heads suitable for case checking are readily available in clauses and noun phrases, it is difficult to find an independent motivation for the presence of such a head in a prepositional phrase. If we want to maintain a purely configurational approach to case marking, a last remaining solution would be to treat the case assigned by prepositions in Modern English as lexical rather than structural case. As mentioned at the start of Section 2.2, configurational approaches to lexical case assignment generally assume that lexical case is associated with θ-marking rather than spec-head agreement, and only involves lexical heads. Since argument-taking prepositions assign θ-roles to their objects and are clearly lexical rather than functional heads, lexical case assignment would seem a fairly plausible option. However, there are certain factors that speak against an analysis of prepositions as lexical case assigners in Modern English. As noted in Section 1.3, lexical case assignment is generally associated with particular thematic relations, and would be expected to vary with the preposition and/or its interpretation. While different prepositions appear to have assigned different cases in Old English (cf. Gelderen 2000:62; Mitchell 1985: 496-499), there is no evidence for θ-related differences in case assignment between prepositions in Modern English (cf. Runner 1998:31f).38 Further evidence for the structural nature of the case assigned by prepositions comes from preposition stranding (P-stranding) in wh-questions (62) and pseudo-passives (63). (62) Who did you vote for? (based on an example given by Kayne 1984:103) (63) I’ve been shouted at already. Kayne (1984:115) and Arnold (1996:4f) suggest that the stranding of the P (for, at) in sentences like (62) and (63) is possible in Modern English, because 37

See Runner (1998:32) for an analysis along these lines, where the case of the prepositional object is checked in a preposition-related Agr projection. 38 Data discussed in Chapters 5, 8 and 9 suggest that we do find differences in case marking between argument-taking prepositions, complementizers, and focus prepositions. However, these differences would appear to be due to the presence versus absence of a thematic relationship with the following noun phrase rather than differences in the nature of the thematic relation.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

56

the V-P sequence can undergo some kind of reanalysis. According to Kayne (1984:115), the possibility of P-stranding in Modern English suggests that prepositions assign structural case, because “reanalysis between two lexical categories is possible only if they assign Case in the same way”. Since verbal case is clearly structural in Modern English, the case assigned by prepositions will have to be structural as well. In languages where V is involved in structural case assignment but P assigns case inherently, P-stranding of the kind found in (62) and (63) is predicted to be impossible. Interestingly, the history of English provides supporting evidence for Kayne’s reanalysis principle. As discussed in Section 1.3, prepositions appear to have assigned inherent case to their objects in Old English, but many verbal objects already appeared with structural case. Kayne’s reanalysis principle would lead us to expect that P-stranding in wh-questions and pseudo-passives was impossible in Old English, and this is indeed the case. P-stranding in pseudo-passives and wh-questions is first attested in Middle English (cf. Allen 1980:224-230; Arnold 1996), when there was no longer any morphological evidence of lexical case marking by prepositions (cf. Section 1.3).39 The evidence discussed in this section suggests that objects of prepositions receive structural case, but not in quite the same way as verbal arguments 39 Since a detailed investigation of prepositions and case would have been beyond the scope of this book, I am assuming that the observations made here hold for all argument-taking prepositions in Present-Day English. It is however possible that prepositional predicates functioning as the argument of a verb have different case properties from adverbial prepositional predicates such as before, after, outside “which convey circumstatial information about an action, object or process” (Jolly 1993:275; cf. also Kayne 2001). As (62)-(63) and (i) illustrate, preposition stranding is fine when the PP is selected by the verb and has an argumental rather than adverbial function. When the PP is clearly adverbial however, stranding appears much more marginal (ii)-(iii).

(i)

Which officer did you give your keys to?

(ii) *? Which meal did he smoke the cigars after? (cf. After which meal did he smoke the cigars?) (iii) *? Which building did they stage their protest outside? (cf. Outside which building did they stage their protest?) It is likely that the prepositions in (ii)-(iii) are unable to undergo the reanalysis required for preposition stranding because they do not have the close semantic and syntactic relationship with the verb that is exhibited by argument PPs (cf. Arnold 1996:4f). The absence of preposition stranding with adverbial prepositions is thus not necessarily proof that adverbial prepositions assign lexical case to their objects. However, it does mean that we have considerably less evidence for structural case assignment by adverbial prepositions than for structural case assignment by argument prepositions. I would like to thank Liz Pearce and Diane Massam for drawing my attention to this issue.

CHAPTER 2

57

in canonical subject and object position. In Section 2.3, I will propose that the structural case requirements for prepositional objects are determined solely by structural linking, whereas the structural case requirements for verbal arguments in canonical subject and object position are determined by case checking as well as structural linking. 2.3

Argument Case, Positional Case and (Positional) Default Case Configurational approaches to structural case account rather well for case checking within clauses and noun phrases, but run into problems when applied to prepositions and their objects. Structural linking on the other hand, is able to account for the case marking on prepositional objects, but cannot predict the case properties of Poss-ing gerunds and non-finite clauses. Given that neither case checking nor case linking alone is able to provide a complete account of pronoun case in Modern English, I would like to argue that the distribution of pronoun forms is constrained by both types of structural case, as well as a default case requirement. As I will demonstrate in the following chapters, the case form of a pronoun is determined by the interaction of case checking, case linking, and default case, as well as some factors other than case. The competition between these different influences is the source of pronoun case variation in Modern English. Since violations of structural case requirements do not lead to ungrammaticality, the influence of case checking, case linking and default case is best captured in three violable case constraints: Argument Case, Positional Case, and Default Case. The Argument Case constraint (64) is based on the structural linking approach outlined in Section 2.1.1. (64) Argument Case (Arg-Case) The overt case form of any structural argument of a predicate must comply with the structural linking between cases and arguments in the θ-structure. Arg-Case makes the following case predictions for Modern English: (a)

The highest structural argument of a predicate must bear nominative case, because nominative case has the structural feature [- higher].40

40 The highest structural argument of a noun is referential and structurally realised only in small clauses such as She a beauty!. In non-predicative noun phrases, the highest structural argument is usually θ-bound, and therefore not present in the syntax (cf. Section 2.1.1).

58

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(b)

All other structural arguments of a verb or preposition must surface in their objective case form, because objective case is unspecified for structural features and associated with predicates that have the category feature [- N].

(c)

The remaining structural argument of a noun must bear genitive case, because genitive case is unspecified for structural features and associated with [+ N] predicates.

The Positional Case constraint (65) is based on the spec-head agreement approach to case checking outlined in Section 2.2.41 (65) Positional Case (Pos-Case) The overt case form of an argument noun phrase appearing as the specifier of an agreement-related functional head at Spell-Out must match the case/agreement features of this functional head, iff the position of the noun phrase at Spell-Out differs from its θposition. Although I am assuming that Positional Case only affects argument noun phrases, it is difficult to determine whether non-argument DPs could ever check Positional Case. In English at least, pronouns with alternating case forms are always arguments when they appear in positions covered by Positional Case. As we will see in Chapter 10, Positional Case is likely to have started out as a constraint on argument noun phrases, but it seems conceivable that the argument status of a DP would eventually become irrelevant for Positional Case marking. The tree diagrams in (66)-(69) illustrate how case checking applies to (argument) DPs in the specifier of the various agreement-related functional heads associated with case checking in Modern English.

41

Note however that there is an important difference between minimalist case checking and the Positional Case constraint. In a minimalist approach, structural case features on nouns and agreement features on verbs are uninterpretable and will crash the derivation if they remain unchecked. This means that in any convergent derivation, all argument noun phrases are assumed to have undergone case checking, either overtly or after Spell-Out. In the constraintbased approach adopted here, Positional Case is only checked if an argument noun phrase has raised out of its θ-position and occupies the specifier of an agreement-related functional head at Spell-Out. If the noun phrase remains in its θ-position at Spell-Out, the Positional Case constraint is inapplicable and has no bearing on the convergence of the derivation.

CHAPTER 2

(66)

CP 3 C [+ finite] TP 3 DPi [nom] T’ he 3 T [nom] …ti...

Case checking on an (argument) DP that occupies [Spec,TP] at Spell-Out in a finite clause

(68)

FP wo F vP pos. of lexical V 3 at Spell-Out DPi [obj] v’ him ty v [obj] …ti... Case checking on an (argument) DP that occupies [Spec,vP] at Spell-Out

(67)

59 CP ei C [- finite] TP for 3 DPi [obj] T’ him 3 T [obj] …ti...

Case checking on an (argument) DP that occupies [Spec,TP] at Spell-Out in a non-finite clause introduced by for

(69)

DP 3 DPi [gen] D’ his 3 whose D [gen] NumP 5 … ti …

Case checking on an (argument) DP that occupies [Spec,DP] at Spell-Out

The (Positional) Default Case constraint (70) affects all noun phrases not influenced by the Positional Case constraint. It constrains the case form of any structural argument that does not appear in the specifier of one of the agreement-related functional heads associated with case checking in Modern English, and it also constrains the case form of any pronoun that is not a structural argument of a predicate.42 (70) (Positional) Default Case (Def-Case) The overt case form of any noun phrase not influenced by Pos-Case must match the default case of a language. In Modern English, the default case is the objective case.

42

The Default Case constraint proposed here resembles the notion of default case discussed in Schütze (1997:52-62; 2001) in that it applies to noun phrases that fail to undergo case checking. However, while Schütze’s (1997:44) and McFadden’s (2004:35-37) default case only shows up on noun phrases that are unable to receive any other kind of case, the Default Case constraint defined in (65) affects the form of any pronoun unable to check Positional Case, even when the form of the pronoun is already constrained by Argument Case. So, strictly speaking, the constraint should be called ‘Positional Default Case’. However, to save space and avoid confusion with the Positional Case constraint itself, I have decided to refer to the constraint merely as ‘Default Case’ in the remainder of this book.

60

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Since the Default Case constraint primarily affects DPs that are topicalised, dislocated, or otherwise separated from the functional heads involved in case checking, its influence on pronoun case in Modern English will become more apparent in Chapters 8 and 9, which look at pronoun case variation in precisely these environments. As Pos-Case and Def-Case apply in mutually exclusive contexts, a DP can only ever be subject to either Pos-Case or Def-Case. The case form of a DP may however be simultaneously constrained by Arg-Case and Pos-Case, or by Arg-Case and Def-Case. Table 2.1 provides a brief summary of the constraint combinations applicable in different contexts. status of the DP argument which has raised out of its θ-position and appears in the specifier of an agreement-related functional head at Spell-Out argument that has remained in its θ-position and/or does not occupy the specifier of an agreement-related functional head at Spell-Out non-argument (in all syntactic positions)

applicable case constraints Argument Case (Arg-Case) Positional Case (Pos-Case) Argument Case Default Case (Def-Case) Default Case

Table 2.1. Case constraints that apply to argument and non-argument DPs in different syntactic positions

In most instances of pronoun case marking discussed so far, the applicable case constraints make the same predictions. Thus, both Arg-Case and Positional Case require pronominal subjects of finite clauses to be nominative, and predict that pronominal objects of verbs should be objective (71). Similarly, both Argument Case and Default Case predict that objects of prepositions should be objective (72). (71) She gave me the keys. (case forms predicted by both Arg-Case and Pos-Case) (72) between us (case form predicted by both Arg-Case and Def-Case) For pronouns in the specifier of a noun phrase, both Arg-Case and Pos-Case stipulate genitive case (73). However, in Poss-ing gerunds, Arg-Case would predict nominative subjects, whereas Pos-Case correctly captures that the subject should take the genitive case form (74). (73) his/whose novel (case form predicted by both Arg-Case and Pos-Case)

CHAPTER 2

61

(74) a. You must excuse [my telling him]. (case forms predicted by Pos-Case) b. * You must excuse [I telling him]. (case forms predicted by Arg-Case) The predictions of Positional Case are also more accurate than the predictions of Argument Case when it comes to the case of subjects in embedded nonfinite clauses (75). (75) a. We expected [him to give them the keys]. (Pos-Case predictions) b. * We expected [he to give them the keys]. (Arg-Case predictions) The pronoun case evidence from Poss-ing gerunds (74) and non-finite clauses (75) suggests that any case clash between Arg-Case and Pos-Case is resolved in favour of the pronoun form required by Positional Case. This indicates that Positional Case is more influential than Argument Case in Modern English. Case differences between objects of verbs and objects of prepositions in the empirical data presented in Chapter 5 indicate that Positional Case is also more influential than the Default Case constraint in Modern English. As mentioned above, the form of prepositional objects is constrained by Arg-Case and Def-Case, whereas the form of verbal objects is constrained by Arg-Case and Pos-Case. The results of the empirical survey suggest that pronoun case variation is more likely to occur with objects of prepositions than with objects of verbs. Such a difference between prepositional and verbal objects is expected if the influence of Default Case is weaker than the influence of Positional Case. Further evidence for the relative weighting of Pos-Case and Def-Case in Modern English is discussed in Chapters 8 and 9. As we will see, pronoun case variation tends to arise when the demands of Arg-Case clash with Def-Case requirements, or when Def-Case is the only case constraint affecting the pronoun. Both of these trends indicate that Default Case is more easily overridden by competing case and non-case constraints than Positional Case. In Chapter 10 I will argue that the relative importance of Positional Case (and thus configurational case checking) in Modern English can be seen as a consequence of the shift from morphological to positional licensing at the end of the Middle English period. 2.4

A constraint-based approach to agreement Case morphology is often considered to be closely related to verbal agreement morphology. As mentioned in Section 1.3, Old English verbs usually exhibit agreement with nominative subjects, but surface in the default 3sg form when the subject bears lexical dative case. In Belfast English (cf. Henry

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

62

1995), a verb in the 3sg default form is only compatible with non-pronominal plural noun phrases (76) or plural subject pronouns that bear objective case (77). When a pronominal plural subject is nominative, the verb obligatorily appears in the plural form (78). (76) [The eggs] are / is cracked.

(Henry 1995:5)

(77) a. [Us and them] are / is always arguing. b. [Us students] are / is going. c. Usuns were / was late. (Henry 1995:18,21) (78) a. [We and they] are / *is always arguing. b. [We students] are / *is going. c. We were / *was late. (Henry 1995:18,21) As Henry (1995:25,42) points out, the data in (77)-(78) suggest nominative case and subject agreement are checked in the same syntactic configuration. In the approach advocated here, we can capture this generalization by assuming that the functional heads responsible for Pos-Case checking also check Positional Agreement (Pos-Agreement). Pos-Agreement, like Pos-Case, is determined on the basis of syntactic configurations at Spell-Out. A verb will show subject Pos-Agreement (s-φ) with an argument noun phrase that occupies [Spec,TP] at Spell-Out (79), and object Pos-Agreement (o-φ) with an argument noun phrase that occupies [Spec,vP] at Spell-Out (79).43 As with Pos-Case checking, Pos-Agreement can occur only when the surface position of the argument noun phrase in question differs from its θ-position. When Pos-Agreement is not possible,44 the verb will be influenced by (subject and/or object) Default Agreement (Def-Agreement). In Modern English, the subject Def-Agreement form of a verb is its 3sg form. 43

As we will see in Chapter 3, the syntactic positions associated with Pos-Agreement also serve as licensing positions for weak pronouns in Present-Day English: weak subject pronouns are licensed in [Spec,TP] and weak object pronouns are licensed in [Spec,vP]. Gelderen (1997:106) presents evidence from Early Modern English which suggests that in languages where finite lexical verbs are able to raise to C, the subject Pos-Agreement features of a verb in C are determined by an argument noun phrase in [Spec,CP] rather than [Spec,TP] (i). (i)

44

What cares these roarers for the name of King (Shakespeare, The tempest: I. i. 17)

That is, when the relevant specifier position (i.e. [Spec,TP], [Spec,vP], and possibly [Spec,CP] in languages where finite lexical verbs are able to raise to C) is not filled by an argument noun phrase, or when it is filled by an argument noun phrase whose surface position is identical to its θ-position.

CHAPTER 2

(79)

TP 3 T’ DPi [s-φ] 3 T FP 5 [s-φ] …tj…

(80)

63

vP 3 DPj v’ [o-φ] 3 DP v’ tsubject 3 v VP [o-φ] 5 …tj…

When an argument DP raises to [Spec,TP] before Spell-Out, it will check nominative Pos-Case, and trigger subject Pos-Agreement on the finite verb. If a subject DP fails to raise all the way to [Spec,TP] before Spell-Out, its case form will be constrained by Def-Case rather than Pos-Case, and the verb will be influenced by Def-Agreement rather than Pos-Agreement. We can thus account for occurrence of the default objective case and default 3sg verb form in utterances such as Us students is going by assuming that the subject DP occupies the specifier of a functional head below T at Spell-Out. The nominative case and plural agreement in utterances such as We students are going indicate that the subject has raised to [Spec,TP], giving rise to nominative PosCase checking as well as subject Pos-Agreement. The optional occurrence of subject agreement with subjects in the objective case (Us students are going) could be argued to arise from competition between Default Agreement and an Argument (Arg) Agreement constraint that requires the verb to show subject agreement with the highest structural argument. Further evidence for competition between Default Agreement and ArgAgreement comes from constructions such as locative inversions (81), deictic there constructions (82) and there BE sentences (83)-(84). While the verb in such constructions may agree with the postverbal noun phrase (81)-(83), agreement often ‘breaks down’, and the verb surfaces in the 3sg default form, especially in there BE sentences (84) (cf. Lakoff 1987:547f; Bartlett 1992:13; Sobin 1997: 332-342; Gelderen 1997:105-123; Deevy 1998; Hay & Schreier 2004). (81) In it were cooked the puddings, the pies, the cakes, the waffles, and the pancakes which filled the table meal after meal (L. Bromfield, The Farm:125) [Levin & Rappaport Hovav 1995:251]

(82) There go our profits for the year. (83) There are some letters on your desk.

(84) There’s some letters on the desk. (Sobin 1997:341)

64

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

In all of these constructions, [Spec,TP] is either empty or filled by a nonargument XP at Spell-Out. This means that there is no trigger for subject PosAgreement, and the verb will instead be influenced by 3sg Def-Agreement. Since the postverbal noun phrase is the highest argument of the verbal predicate, the Arg-Agreement constraint will predict that the finite verb agrees with the postverbal noun phrase (I am assuming that copular be combines with the predicate of its small clause complement to form a complex predicate at the level of Semantic Form, which means that the highest argument of the small clause predicate is also the highest argument of be). When the postverbal noun phrase is a plural, the demands of Arg-Agreement will therefore clash with the demands of Def-Agreement, giving rise to the variation between (82) and (83). It is interesting to note that the objective subject pronouns in Henry’s (1995) examples are all either conjoined (77a), modified (77b), or morphologically complex (77c), i.e. they are all strong. Weak subject pronouns like we in (78c) invariably surface in the nominative case and trigger subject agreement. In Chapter 3, I will argue that weak pronouns generally surface in the case form predicted by the Pos-Case constraint, because they must be licensed in the specifier of the functional heads associated with Pos-Agreement. However, Pos-Case checking and Pos-Agreement do not always coincide. As we will see in Section 9.2, T-to-C raising leads to the loss of Pos-Case checking between T and an argument in [Spec,TP]. As a result, a weak subject pronoun following a fronted auxiliary will be influenced by the conflicting case demands of the Arg-Case and Def-Case constraints, and the potential for case variation arises. Unlike weak pronouns, strong pronouns may exhibit case variability even when they are influenced by the Pos-Case constraint. As Henry (1995:25) points out, many varieties of English where a finite verb obligatorily agrees with a preverbal subject still permit utterances such as Us students are going, where a strong subject pronoun surfaces in the objective case. With such varieties, there is no real justification for assuming that preverbal subjects ever appear in a position lower than [Spec,TP], because they always trigger agreement on the verb. This suggests that the objective form of the subject is due to influence from a non-case constraint that is able to override the demands of PosCase and Arg-Case. The exact nature of the non-case influences affecting strong pronouns in Present-Day English will be explored in Chapters 5 & 6.

CHAPTER 3 CASE AND THE WEAK/STRONG DISTINCTION IN THE ENGLISH PRONOUN SYSTEM

3.0

Introduction In this chapter I will demonstrate that the presence versus absence of variation in pronoun case choice largely correlates with the morphosyntactic status of the pronoun. I will argue that English weak pronouns do indeed exhibit the consistent nominative/objective case distinction predicted by general approaches to case (cf. Chapter 2), and I will provide evidence that all instances of case variation that cannot be given a purely case-based account occur in strong pronoun contexts. Section 3.1 looks at the properties and distribution of strong and deficient (i.e. weak or clitic) pronouns across languages, and demonstrates that the contexts associated with pronoun case variation in English tend to trigger the use of strong pronoun forms in other languages. In Sections 3.2 and 3.3, I provide evidence that the syntactic and phonological properties of the pronouns subject to case variation in English do indeed correspond to the syntactic and phonological properties that characterise strong pronouns in other languages. Section 3.4 concludes the chapter with a summary of the morphosyntactic differences between weak and strong pronouns in Modern English. 3.1

Strong and deficient pronouns in languages other than English Cardinaletti & Starke (1999) propose that pronouns in many languages fall into two distinct classes: strong pronouns and deficient pronouns.1 The class of deficient pronouns can be further divided into clitics, which are heads at surface structure, and weak pronouns, which surface as maximal projections.

1

See also Foulet (1935a, 1935b, 1936), Benveniste (1974), Kayne (1975; 2000:163-184), Cardinaletti (1991, 1994, 1999), Cardinaletti & Starke (1995, 1996), Galves (1997), and ZribiHertz & Mbolatianavalona (1999).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

66 (1)

(2)

Deficient pronouns (a) (b)

can undergo phonological reduction must raise out of their θ-position at Spell-Out

(c)

(d)

can only be licensed through a close syntactic relationship with certain functional heads at Spell-Out (for clitics, the required syntactic relation is head-adjunction, whereas for weak pronouns, it is spec-head agreement with the functional head) are unable to take modifiers

(e)

are generally underspecified for semantic features

Strong pronouns (a) cannot be destressed or contracted (b) (c)

may remain in their θ-position throughout the derivation do not need to be licensed by a functional head, and may therefore occur in peripheral positions at Spell-Out

(d) (e)

are able to be coordinated and modified may be specified for semantic feature values such as [+ human]

Many Romance languages have a clear morphological as well as syntactic distinction between a clitic pronoun series and a strong pronoun series (cf. Foulet 1935a, 1935b, 1936; Benveniste 1974; Kayne 1975; Cardinaletti 1999; Cardinaletti & Starke 1999). As can be seen from the pronominal paradigm for French (Table 3.1), clitic pronouns tend to have different subject and object forms, and sometimes even distinguish dative and accusative case. Strong pronouns, on the other hand, generally surface in just one form, regardless of their syntactic position and grammatical relation. Like weak and clitic pronouns in other languages, French clitics have a very limited syntactic distribution, and surface only in positions adjacent to the finite verb (3)-(5). Strong pronouns, on the other hand, appear in a wide variety of syntactic environments and are usually in some way separated from the verb. A closer look at the syntactic distribution of strong pronoun forms in French (6)-(18) reveals that they occur in exactly those syntactic constructions and positions that are associated with pronoun case variation in English (cf. Chapters 5 and 8 for further discussion ).

CHAPTER 3

1sg 2sg 3sgM 3sgF 1pl 2pl 3plM 3plF

NOM je tu il elle nous vous ils elles

clitic series ACC me te le la nous vous les les

DAT me te lui lui nous vous leur leur

67 strong series INVARIANT moi toi lui elle nous vous eux elles

Table 3.1. Clitic and strong pronoun forms in French (cf. Kayne 1975:67f, 84)

(3)

Il partira bientôt. 3sgM.CL.NOM will.leave soon “He will leave soon.”

(4)

Marie lui a donné le livre. Marie 3sgM.CL.DAT has given the book “Marie has given him the book.”

(5)

Marie le voyait. Marie 3sgM.CL.ACC saw “Marie saw him.”

(6)

Left-dislocation (see also Benveniste 1974:199) Moi je mettrai la table. 1sg.STRONG 1sg.CL.NOM will.set the table “Me, I’ll set the table.” (Jakubowicz & Rigaut 1997:59)

(7)

Right-dislocation Ils sont intelligents, eux. 3plM.CL.NOM are intelligent 3plM. STRONG “They are intelligent, them.” (Kayne 1975:67)

(8)

Exception structures Marie ne voyait que lui. Marie not see than/but 3sgM.STRONG “Marie saw only him.”

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

68 (9)

Coordination2 a. [Lui et Paul] partiront bientôt. 3sgM.STRONG and Paul will.leave soon “Him/he and Paul will leave soon.” (based on Kayne 1975:85) b. [Paul et lui] partiront bientôt. Paul and 3sgM.STRONG will.leave soon “Paul and he/him will leave soon.” (based on Kayne 1975:85) c. Marie voyait [lui et Paul]. Marie saw 3sgM.STRONG and Paul “Marie saw him and Paul.”

(10) Comparatives Marie est plus vite que toi. Marie is more fast than 2sg.STRONG “Marie is faster than you.” (11) Identificational / it BE sentences C’est moi. it.is 1sg. STRONG “It’s me.” (Benveniste 1974:199) (12) Clefts with a pronominal focus (see also Kayne 1975:82) C’est moi qui l’ai it.is 1sg.STRONG who 3sg.CL.ACC-has done “It’s me that did it.” (Benveniste 1974:199)

fait.

(13) Stripping / bare argument ellipsis moi aussi 1sg.STRONG too “Me too.” (Benveniste 1974:199)

2

Although all of the examples given here involve pronouns coordinated with a proper noun, Benveniste (1974:199) points out that strong pronouns in French may also be coordinated with other strong pronouns (moi et toi) and with noun phrases containing a common noun modified by a possessive pronoun (moi et mes amis).

CHAPTER 3

69

(14) Independent pronouns a. Qui est là? - Moi. who is there 1sg.STRONG “Who’s there? - Me.” (Benveniste 1974:199) b. Qui as-tu vu? - Lui/elle/eux. who have-2sg.CL.NOM seen 3sgM/3sgF/3pl.STRONG “Who did you see? - Him/Her/Them.” (Kayne 1975:83) (15) Pronouns modified by a numeral [Eux deux] partiront bientôt. 3plM.STRONG two will.leave soon “The two of them will leave soon.” (Kayne 1975:85) (16) Pronouns followed by an apposition moi, Pierre 1sg.STRONG Pierre “I, Pierre” (Benveniste 1974:199) (17) Pronouns modified by a relative clause moi, qui suis … 1sg.STRONG who am … “I, who am …” (Benveniste 1974:199) (18) Pronominal subjects of infinitives and small clauses a. Toi faire ça! 2sg.STRONG do that “You do that?!” (Jespersen 1934[1924]:130) b. Lui avare? 3sgM.strong greedy “He/him greedy?” (Jespersen 1934[1924]:130) 3.2

The syntactic properties of constructions with pronoun case variation As discussed in Section 3.1, deficient pronouns must be licensed by a close syntactic relationship with an agreement-related functional head. In Present-Day English, there is little evidence that any pronouns are able to undergo head-adjunction to a functional head, since even weak pronouns can only occur

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

70

in positions also available to full noun phrases. It therefore seems most plausible to assume that deficient pronouns in English are weak rather than clitic. This means that they will have to be licensed through spec-head agreement with an appropriate functional head. For Present-Day English, the heads most likely to be able to license weak subject and object pronouns are T (if dominated by C), which also checks Positional subject agreement on the verb (cf. Section 2.4), and v, which is involved in checking Positional object agreement in languages that have object agreement morphology on the verb. D could be argued to license weak possessive pronouns. Many of the pronoun constructions where pronoun case variation has been reported in existing studies (cf. Chapters 5, 8 & 9) lack at least one of these agreement-related functional heads involved in the licensing of weak pronouns. V-ing constructions (19), to-infinitives (20), small clauses (21), and gapping constructions (22) lack either the CP-layer required to endow T with the ability to check Pos-Case and subject Pos-Agreement, or lack both a CP and a TP-layer. Independent pronouns (23), than-comparatives, and exception structures (24) are characterised by the absence of any agreement-related functional head, at least in some of the analyses available for these constructions; and constructions involving identificational be (25)-(26) lack the vP-layer associated with object agreement. (19)

TP 3 T’ DPi me/I 3 T FP 3 F vP 1 3 v’ vl F DPj 1 him 5 Vk v t i tl tj tk telling

Tree diagram illustrating the absence of a CP-layer in V-ing constructions (cf. Section 8.6)

(20)

TP 3 DPi T’ he/him 3 T FP 6 to laughk ti tk

Tree diagram illustrating the absence of a CP-layer in to-infinitives that lack an overt complementizer. As discussed in Section 8.7, some to-infinitives can be argued to lack a TP-layer as well.

CHAPTER 3

71

(21) Tree diagrams illustrating the absence of CP and TP in small clauses (cf. Section 8.8) (a)

(b)

PredP 3 DP Pred’ subject 3 Pred XP The basic structure of categorical small clauses

(c)

XP 3 DP X’ subject 2 X …

The basic structure of thetic small clauses, where XP is the (extended) projection of a lexical head

FP 3 F’ DPi Me 3 The structure of thetic small clauses F vP with a verbal predicate. 1 3 v’ vl F DPj 1 him 3 ti v’ Vk v throw 3 v TelP 3 tl Tel VP out 3 V’ tj 5 tk

(22)

vP 2 DP v’ he/him 2 v VP 2 DP V’ her 4

Tree diagram illustrating the absence of a TP-layer in gapping constructions such as She picked him up, not [he__ her] (cf. Section 8.9)

(23)

DP 5 me

(24)

PP 3 P DP than me but

Tree diagrams illustrating the absence of any agreement-related functional heads in a nonsentential constituent analysis of independent pronouns (23), as well as in than-comparatives (than me) and exception structures (but me) (24) (cf. Sections 8.4, 5.4, and 8.11)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

72 (25)

CP 3 Tree diagram illustrating the lack of a vP-layer in basic identificational sentences (I could be her), C TP [+ finite] 3 and in it BE sentences (it could be me) DPi T’ (cf. Section 8.5) I 3 it T VP couldl 3 V FP 3 tl F VP bek 3 DP V’ ti 3 V DP tk her (26) TP me 3 DPj T’ it ei Tree diagram illustrating the lack of a T FP vP-layer in it-clefts (cf. Section 5.3) 1 3 Vk T F VP 3 is/was tk DP V’ tj 3 V VP 3 tk DP V’ 3 him V CP 6 tk that telephoned

In topicalisation (27), left-dislocation (28), and right-dislocation structures (29), the pronoun appears in a position outside [Spec,TP] or [Spec,vP]. The same goes for wh-pronouns in questions and relative clauses (30). (27)

CP 3 DPi C’ pronoun 3 C TP 6 ti

Tree diagram illustrating the surface position of a topicalised pronoun in a main clause (cf. Section 8.1)

(28)

CP 3 DP C’ pronouni 3 C TP 6 pronouni

Tree diagram illustrating the surface position of a left-dislocated pronoun (cf. Section 8.2)

CHAPTER 3

73

(29) Tree diagrams illustrating the syntactic position of a right-dislocated pronoun in a monoclausal analysis of right-dislocation structures (cf. Section 8.3) (a)

TP 2 TP DP

(b)

pronoun

FP 2 CP F’ 2 F DP

if we treat right-dislocated pronouns as TP-adjuncts

pronoun

(c)

CP 2 C’ TPi 2 C FP 2 DP F’ pronoun 5 ti

if we assume that right-dislocated constituents are basegenerated as complements of an abstract functional head (b), or as specifiers of a functional head in the C-system (c)

(30)

CP 3 C’ The surface position of a wh-pronoun in a DPi wh-pronoun 3 wh-question or relative clause (cf. Chapter 9) C TP 6 ti

In at least certain analyses of coordinates (31) and constructions where the pronoun is modified by a preceding adjective (32), the pronoun appears in embedded position and is removed from the direct influence of an agreement-related functional head by intervening projections (highlighted in bold). (31)

TP / vP wo ConjP T’ / v’ 3 2 DP Conj’ T/v … he/him 3 Conj DP and I/me

Tree diagram illustrating the absence of a direct spec-head relationship between a coordinated pronoun and an agreementrelated functional head (cf. Section 5.1.3)

(32)

DP 3 D NumP 3 Num AP 3 AA NP lucky them poor us

The surface position of a pronoun preceded by an adjective, if we assume that prenominal adjectives always take NP complements (cf. Section 8.14.1)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

74

In constructions where a pronoun precedes a modifier (33), the pronoun does appear to head a DP in the specifier of an agreement-related functional head (cf. Sections 5.2 & 8.14). However, because of the presence of the modifier, the DP headed by the pronoun is internally complex, a property usually associated with strong rather than deficient pronouns.3 (33)

DP ei D NumP 1 3 NumP NP (a) / PP (b) / CP (c) Ni D we 2 ei Num NP (a) linguists 4 (b) in this department N (c) who are to blame ti

3.3

The prosodic properties of pronouns that exhibit case variation As discussed in Section 3.1, deficient pronouns are able to undergo prosodic restructuring and phonological reduction, whereas strong pronouns are prosodically independent and always bear some degree of stress. If we want to argue that case variation in English is more or less confined to strong pronoun contexts, we would expect the pronouns affected by case variation to exhibit the same phonological properties associated with strong pronouns in other languages. This does indeed appear to be the case. As was already noted by Selkirk (1980:31f,41), lone pronouns in canonical argument positions tend to be unstressed and often surface without onsets and/or with reduced vowels (34).4 Pronouns in contexts susceptible to case variation, on the other hand, are generally unable to undergo destressing (35).5

3

See Cardinaletti (1994); cf. also Cardinaletti & Starke’s (1999:178f,185f) suggestion that a projection headed by weak pronouns contains fewer layers than a projection headed by strong pronouns, and are therefore unable to license modifiers. It is important to note that Cardinaletti & Starke’s approach differs from the one adopted here, because Cardinaletti & Starke (1999: 214) assume that weak pronouns lack the highest structural layer projected by strong pronouns, whereas I am arguing that a weak pronominal DP lacks the lower structural layers (i.e. NumP, AP, and NP) that are present in DPs headed by strong pronouns (cf. Section 3.4). 4 Reduced vowels other than [] are marked with the diacritic [  ]. Although the reduced subject pronouns in (35a-c) appear in initial position, the phonological reduction of subject pronouns is most likely to occur when the pronoun is preceded by an overt constituent. I would like to thank Liz Pearce (p.c.) for drawing my attention to this point. 5 See Selkirk (1980:31f,41,132f,154f) and Gelderen (2004:24).

CHAPTER 3

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

75

[hi]/[ i] never got a word in edgeways. (he) [ i] got a job last year. (he) (Akmajian 1984:9) [j] always screw up royally. (you) (Akmajian 1984:9) I’ll see [j] before the match. (you) They beamed [m] up. (him) Rob saw them [m] in the library. (them) We’ve sent them [m] to Tim. (them)

(35) a. [Poor [hm]] never got a word in edgeways. b. * [Poor [i]/[m]] never got a word in edgeways. (pronoun premodified by an adjective)

c. Rob saw [[em] and [s]] in the library. d. * Rob saw [[m] and [s]] in the library. (coordinated pronouns) e. I’ll see [ju boys] before the match. f. * I’ll see [j boys] before the match. (pronoun followed by a noun) g. You’re faster [than [s]]. h. * You’re faster [than [s]]. (pronoun in a than comparative) i. It was [em] that voted against the upgrade. j. * It was [m] that voted against the upgrade. (focus pronoun in an it-cleft)

k. [hm] I like. l. * [m] I like. (topicalised pronoun) Selkirk (1980:54,134f,154; 1995) suggests that obligatorily stressed pronouns have the status of prosodic words (36), whereas weak pronouns merely head their own syllable, and lack prosodic word boundaries (37) (φ = prosodic phrase, ω = prosodic word, Σ = foot, σ = syllable). (36)

ω g Σ g σ g It was him they saw.

(37)

σ σ g g I saw him

Cardinaletti & Starke (1999:191) point out, the prosodic deficiency of weak pronouns correlates with their syntactic deficiency. We might thus hypothesise that prosodic strength is linked to syntactic complexity.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

76 3.4

Summary of differences between weak and strong pronouns Weak pronouns are base-generated in D and simultaneously behave like heads and maximal projections (38).6 The obligatory lack of internal structure in a DP headed by a weak pronoun accounts for the absence of modifiers with weak pronouns and correlates with a lack of prosodic structure (39). (38)

DP g D

(39)

σ !

weak pronoun

weak pronoun

Strong pronouns are base-generated in N, and generally raise to D before SpellOut (40). As a result, a strong DP is internally-complex and may contain modifiers. The structural complexity of strong DPs correlates with their prosodic complexity (41).7 (40)

DP 3 D NumP 1 3 NP Ni D Num strong 4 pronoun N ti

(41)

ω g Σ g σ g

strong pronoun

Panagiotidis (2002a:187) argues that functional categories such as D need to dominate a lexical category such as N in order to be licensed. Following this suggestion, I will assume that the syntactic ‘strength’ of strong pronouns arises from their nominal nature. Weak pronouns are syntactically deficient because they are base-generated as intransitive Ds. Since the DP projected by a weak pronoun does not contain a lexical head that could license the functional head D, weak pronominal DPs must raise out of their θ-position to be licensed by an agreement-related functional head. The distribution of weak pronouns in Present-Day English indicates that they can only be licensed in the specifier of functional heads associated with Positional Agreement, that is, T 6 Note that weak pronouns in English cannot appear in any positions that are unavailable to full noun phrases. This suggests that they do project a phrasal (DP) layer and have a spec-head relation rather than a head-head relation with the agreement-related functional heads they are licensed by (cf. Gelderen 2004:18-20 for a different approach). 7 As we will see in Section 8.14.1, strong DPs may also contain an AP-layer between NumP and NP. I have omitted this layer from the tree diagram to avoid unnecessary complexity.

CHAPTER 3

77

(dominated by C), v, and D. Since these heads are also involved in Pos-Case checking, weak pronouns consistently surface in the case required by the PosCase constraint (42). (42) The Pos-Case properties of licensing positions for weak pronouns in Present-Day English (a)

CP ei TP C [+ finite] 3 T’ DPi he [nom] 3 T [nom] …ti…

(c)

FP (d) ei F vP 1 3 DPj v’ vk F him [obj] 3 tsubject v’ 3 v [obj] …tj… tk

(b)

CP ei C [- finite] TP for 3 DPi T’ him [obj] 3 T [obj] …ti… DP 3 D’ DPi [gen] his 3 D [gen] …ti…

Strong pronominal DPs, which do not have to move to be licensed, tend to appear in positions immune from Pos-Case, which makes them susceptible to non-case influences. The exact nature of these non-case influences is the focus of the following two chapters. As we will see, the distribution of nominative and objective forms in strong pronoun contexts is variable, and is difficult to account for in a purely case-based approach. Although the data discussed in existing studies offer some indication of the linguistic factors that may influence the choice of strong pronoun forms, it is difficult to gauge the extent of the variation between different pronouns and speakers when the syntactic context is not controlled for. I therefore decided to carry out my own empirical study on the subject. The survey, which is discussed in more detail in Chapter 4, tested all alternating personal pronouns in a small number of strong pronoun contexts, and focused on differences between subjects, objects and prepositionnal complements, as well as differences between initial and final conjuncts in coordinates. A summary of the survey results is given in Chapter 5.

CHAPTER 4 THE EMPIRICAL SURVEY

4.0

Introduction Sigursson (2003, 2005) and McFadden (2004) present evidence from a number of languages which suggests that morphological case marking is a PF phenomenon conditioned by the interaction of a range of different factors, only some of them syntactic.1 The aim of the empirical study discussed in this chapter was to identify the most important linguistic factors influencing the distribution of strong pronoun case forms in Present-Day English. The survey tested the whole paradigm of alternating personal pronouns (i.e. I/me, he/him, she/ her, we/us, they/them) in coordinates (1), it-clefts (2), and than-comparatives (3), plus the occurrence of alternating plural pronouns (i.e. we/us, they/them) in pronoun-NP constructions (4).2 (1) (2) (3) (4)

[They/them and I/me] were supposed to get everything ready… [qu082] It was he/him who/that insisted on going to the rally. [qu251] Oliver is bound to respond more quickly than she/her. [qu252] We/us New Zealanders must stick together. [qu163]

These constructions feature prominently in existing discussions of English pronoun case, and have been identified as contexts where we may find case variation between pronouns as well as between speakers. However, in the absence of a systematic study testing all the alternating pronouns in comparable syntactic contexts, there has been considerable disagreement over the nature of the case trends observed. Existing research also largely fails to provide information about the pronoun case patterns exhibited by individual speakers which would allow us to gauge the extent of case variation between speakers and within the speech of individuals. I therefore decided to carry out a detailed written survey designed to obtain comprehensive information on individual speakers’ pronoun 1

PF = Phonological/Phonetic Form, the interface level that provides instructions for the articulatory and perceptual systems (cf. Chomsky 2000:90f). 2 Each example is followed by a bracketed reference to the questionnaire item it is taken from. For example, the reference [qu082] in (1) signals that the item had the number 82 in the survey questionnaires.

CHAPTER 4

79

case preferences in the four strong pronoun contexts illustrated in (1)-(4). The distribution of pronoun forms in coordinates and pronoun-NP constructions was tested in four core questionnaires administered to 90 native speakers of New Zealand English between July and November 1996. The use and acceptance of the different pronoun forms in it-clefts and than comparatives was tested in one further questionnaire, which was completed in mid-1997, by 41 of the speakers who had participated in the core part of the survey. 4.1

Constructions and pronoun combinations tested Some formal approaches to coordination assume that coordinates are transparent to outside case influences (cf. Pollard & Sag 1994; Kaplan & Maxwell 1995; Zoerner 1995:35; Johannessen 1993, 1998), while other analyses treat coordinates as fundamentally opaque, and predict that the argument structure status and/or syntactic position of the coordinate as a whole will have no bearing on the case form of its conjuncts (cf. Emonds 1985; Schütze 1997:59; Jones 1988; Peterson 2004:675f). In order to determine whether the form of conjoined strong pronouns is in any way influenced by the three case constraints proposed in Chapter 2, I tested test the use and acceptance of the pronoun case forms I, me, he, him, she, her, we, us, they, them in coordinates that appear as the subject of a finite clause (5), the object of a verb (6), and the complement of a preposition (7). (5) (6)

[He/him and I/me] won all the medals at the last tournament. [qu057] Brenda had promised she would meet [he/him and I/me] at the station. [qu014]

(7)

The landscapes painted by [he/him and I/me] drew huge crowds at the exhibition. [qu045]

If coordinates are transparent to case, the case constraints will predict that pronouns in subject coordinates such as (5) should be nominative, while conjoined pronouns in object and prepositional complement position should surface in their objective forms. Case transparency might also lead us to expect possible case differences between object and prepositional coordinates, because objects of verbs will be influenced by Argument Case (Arg-Case) and Positional Case (Pos-Case) if they raise to [Spec,vP] before Spell-Out, whereas the form of prepositional complements is always constrained by Arg-Case and Default Case (Def-Case). Although Def-Case requires prepositional complements to bear objective case, just like Pos-Case calls for objective case on objects in

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

80

[Spec,vP], different weightings of the two constraints could leave strong pronouns in one of the two contexts more susceptible to non-case influences. While it has often been assumed that all conjuncts of a coordinate receive the same case (cf. Klima 1964; Harris 1981; Emonds 1985, 1986; Lumsden 1987; Pollard & Sag 1994; Kaplan & Maxwell 1995; Dalrymple & Kaplan 2000:779), many recent approaches to coordination allow for, or even predict, case asymmetries between initial and final conjuncts. These asymmetries are assumed to arise either from the non-distributive nature of the case feature (Peterson 2004:676), or from differences in the syntactic status of initial and final conjuncts (cf. Munn 1987, 1993; Johannessen 1993, 1998; Zoerner 1995, Gelderen 1997:176ff). To find out whether conjunct position has any bearing on the case form of strong pronouns in English, I tested each of the case alternating personal pronouns in both initial and final conjunct position in all three of the syntactic contexts given in (5)-(7). Each of the pronouns under investigation was tested with two types of non-pronominal conjunction mates (8), as well as with each of the remaining alternating pronouns as a conjunction mate (9), where possible (1sg and 1pl did not appear conjoined with each other). (8)

a. [She/her and Karen] returned to Wellington last night. [qu059] Conjunction mate = proper noun (abbreviated as Name)

b. [She/her and her fiancé] were conned into selling their unit for less than half the government value. [qu208] Conjunction mate = possessed noun phrase containing the possessive form of the pronoun tested (abbreviated as poss)

(9)

a. [She/her and I/me] arranged to work from home this week. [qu019] b. [She/her and he/him] refused to have anything to do with their scheme right from the start. [qu220] c. [She/her and we/us] came all the way from Dunedin only to cheer you on. [qu104] d. [She/her and they/them] spent the whole week complaining about the food, the rooms, the service, … [qu134]

For each syntactic position and pronoun combination, the survey contained one item where respondents could choose the conjunct order, as well as two items with set conjunct order (cf. Section 4.2). In one of the set conjunct order items, the pronoun was given in the initial conjunct, and in the other set conjunct order item, the pronoun was tested in the final conjunct. Items testing coordinates

CHAPTER 4

81

with pronominal conjunction mates were used to elicit case preferences for both of the pronouns involved in the coordinate, and allow us to see whether or not speakers exhibit a preference for case concord between the conjuncts. All names used in coordinates were at least bisyllabic, and thus represent an intermediate step in terms of phonological complexity between pronouns and possessed noun phrases. One drawback of using polysyllabic rather than monosyllabic names in coordinates is that the phonological differences between the pronouns and names make it difficult to determine the role of syntactic status/complexity in conjunct order choice (cf. Section 6.2). To establish whether there really is a difference in syntactic complexity between strong pronouns and names, we would have to use monosyllabic names in the coordinates. I did not do so, because I wanted a clear prosodic contrast between coordinates testing pronouns conjoined with names and pronouns conjoined with each other, but it might be worth comparing monosyllabic and polysyllabic names in future questionnaires.3 As discussed in Section 2.4, Henry (1995) draws attention to correlations between case and verb agreement in Belfast English. Coordinated subjects of a finite clause may be either nominative or objective when the verb is plural (10), but they are always objective when the verb shows singular agreement morphology (11). (10) [He and I] / [Him and me] are going. (11) [Him and me] is going. I did not include any instances of singular verb agreement with subject coordinates (11) in my survey, because I have not come across such examples in New Zealand English. If I had included sentences such as (11) in the questionnaires, they would probably have been perceived as extremely odd by my informants, and would possibly have encouraged them to look for ‘right’ and ‘wrong’ answers in the remaining questionnaire items. Since 1pl and 3pl are the only pronouns with different nominative and objective case forms that readily co-occur with a (non-appositive) noun phrase in Present-Day English, only the distribution of 1pl and 3pl pronoun forms was tested in pronoun-NP constructions. In some analyses of pronoun-NP construc3

Angermeyer & Singler (2003:188,194) grouped monosyllabic and bisyllabic names together as ‘short names’ when they analysed the influence of the conjunction mate on case and ordering preferences in object and prepositional complement coordinates involving 1sg pronouns. The results they present thus also fail to provide a comparison between coordinates involving pronominal conjunction mates and coordinates where one of the conjuncts is a monosyllabic proper noun.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

82

tions, the pronoun is assumed to fill the specifier position in the noun phrase (cf. Postal 1969[1966]:217-221; Jackendoff 1977:106; Payne & Huddleston 2002:374), where we might expect it to be immune from case constraints on the noun phrase as a whole. Alternatively, the pronoun is treated as the head of the construction, with the following noun phrase in complement position (cf. Abney 1987:282ff; Hudson 1987:122f; Longobardi 1994:636n.31; Cardinaletti 1994:202-205; Spinillo 2004:87-91). Such an analysis would predict that the case form of the pronoun should be quite strongly influenced by the surface position and argument status of the pronoun-NP construction (cf. Hudson 1987:122). In order to test whether Arg-Case, Pos-Case, and Def-Case have any bearing on the form of 1pl and 3pl pronouns in pronoun-NP constructions, I tested both pronouns in items where the pronoun-NP construction appeared as the subject of a finite clause (12), the object of a verb (13), and the complement of a preposition (14). (12) [We/us New Zealanders] must stick together. [qu163] (13) I wouldn’t trust [they/them politicians]. [qu159] (14) It’s a hard life for [we/us students]. [qu141] As mentioned above, the three case constraints would predict that subjects of finite clauses surface in the nominative case. Objective forms should be favoured in both object and prepositional complement position, but non-case influences could be stronger in one of the two non-subject positions if Pos-Case and Def-Case are weighted differently. Formal analyses of it-clefts proposed in the literature predict that the form of the focus pronoun should be at least partly influenced by the syntactic properties of the following clause (cf. Gundel 1977; Delahunty 1982; Heggie 1988; Kayne 1994; Meinunger 1998; Hedberg 2000; Jayaseelan 2001). Factors that have been identified as potentially relevant in studies of pronoun case in itclefts include:4 (a) (b) (c)

the function of the relativised constituent in the clause the presence versus absence of an overt relative pronoun the presence versus absence of the complementizer that

Wales (1996:96) and Sobin (1997:334) also suggest that the case of the focus in it-clefts may depend on the φ-features of the focus pronoun, although they 4

See especially Jespersen & Haislund (1949:254) and Wales (1996:95f).

CHAPTER 4

83

disagree on whether case preferences divide along person or number distinctions. In order to determine the extent to which these different factors affect pronoun case in it-clefts, I tested all case-alternating personal pronouns in three types of it-clefts: It-clefts where the relativised constituent in the clause is the subject, and the clause is introduced either by the complementizer that or by the relative pronoun who. As can be seen from example (15), speakers were asked to choose both the case form of the pronoun and the relative marker in the subject it-cleft items. In order to avoid potential interactions between case and verb agreement in clefts with 1sg focus pronouns, I only used invariant past tense verb forms in the clause.5 All clefts were introduced by it was. (15) Imagine you're in a conversation with friends. [qu256] a. b.

Tick the sentence you would be most likely to come across. Pick the second best sentence and put a 2 in the box beside it.

[] [] [] []

It was we that organised the meeting. It was us that organised the meeting. It was us who organised the meeting. It was we who organised the meeting.

It-clefts where the relativised constituent in the clause is the object of a verb and the clause is not introduced by any overt relative marker (16) (16) It was she/her [they wanted for their commercial]. [qu250] It-clefts where the relativised constituent in the clause is the object of a preposition and the clause is not introduced by any overt relative marker (17) (17) It was he/him [they were looking for]. [qu237] 5

As Akmajian (1970:150-153) points out, a present tense verb in the clause usually shows number agreement with the focus in subject it-clefts. When the focus is a nominative 1sg pronoun, the verb may also exhibit person agreement with the focus (i). This connection between nominative case and verb agreement could lead speakers to favour the 1sg objective form me when a present tense verb fails to exhibit person agreement (ii). (i)

It is I [who am having to do with material things] (Ivy Compton-Burnett 1971[1933], More women than men:131) [Erdmann 1978:78]

(ii)

It is me [who is having to deal with the fallout].

84

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Since relative markers (i.e. relative pronouns or the complementizer that) primarily occur when the relativised constituent in the clause of the cleft is the subject, I did not include any relative markers in items testing pronoun case in it-clefts where the relativised constituent is the object of a verb or preposition. Kayne (1994:153n.6), Meinunger (1998), and Jayaseelan (2001:63) analyse the focus as a constituent of the following clause which has undergone movement to a position on the left periphery. Since focus pronouns are arguments of a following predicate in such an approach, their case form will be constrained by the Arg-Case properties of this predicate. In it-clefts where the relativised constituent is the subject of a clause (15), and thus the highest argument of a verbal predicate, the focus will receive nominative Arg-Case. In itclefts where the relativised constituent is the object of a verb (16) or preposition (17), the focus will be linked to objective Arg-Case. As a fronted constituent, the focus pronoun will be unable to check Pos-Case, in any of the it-clefts included in the survey, because an overt constituent (who, that, or the subject of the clause) intervenes between T and the head of the functional projection occupied by the focus constituent (cf. Section 9.1). This means that the focus pronoun will be influenced by Def-Case rather than Pos-Case. For focus constituents in subject it-clefts, the Def-Case requirement for objective case will clash with the nominative case demanded by the Arg-Case constraint, and we have the potential for variation. However, as we will see in Section 9.2, the Arg-Case constraint seems to outweigh the Def-Case constraint in most varieties of Present-Day English. The focus movement analyses would thus lead us to expect a considerably higher preference for nominative focus pronouns in subject clefts than in non-subject clefts. Kiss (1998:258f) argues that a focus movement analysis is possible only for object and prepositional complement clefts where the clause lacks an overt relative pronoun. In her approach, the target of focus movement is the specifier of a clause-external focus position. Moving a subject focus out of a clauseinternal position would lead to a violation of the Empty Category Principle (ECP), because the intermediate trace in [Spec,CP] would not be properly head-governed (cf. Rizzi 1990; Kiss 1998:259). Kiss (1998:259) therefore assumes that the focus of subject clefts is base-generated in the specifier of the focus projection dominating the clause, and is linked to an (overt or empty) relative pronoun in the clause at Logical Form (LF).6 According to Kiss (1998: 258), the copula be functions as an expletive verb in clefts (cf. also Delahunty 1982:215). This suggests that the focus pronoun in subject clefts has a status similar to that of left-dislocated pronouns (cf. Section 8.2), and is therefore in6

For a detailed discussion of LF composition in it-clefts see Delahunty (1982:213-217).

CHAPTER 4

85

fluenced only by the Def-Case constraint. Since the Def-Case demands for objective case seem rather weak in Present-Day English, Kiss’ (1998) analysis would lead us to expect that focus pronouns in subject clefts might be particularly vulnerable to non-case influences. At the same time, a subject focus could also be affected by case-transmission from the relative pronoun in the clause. Gundel (1977) and Heggie (1988) offer the most explicit discussions of case transmission between the clause and focus in it-clefts. Heggie (1988:134) assumes that the focus constituent forms a small clause with the following clause, and inherits the case and θ-role from the empty operator in [Spec,CP] of the clause. Like Gundel (1977:550), Heggie (1988:138) argues that the case features of the relativised constituent in the clause can only be transmitted to the focus when there is no overt relative pronoun. Such an approach might lead us to expect case differences between subject clefts where the clause is introduced by a wh-pronoun, and subject clefts where the clause is introduced by the complementizer that. It is however difficult to determine the exact case predictions of the analyses proposed by Gundel (1977) and Heggie (1988), because the argument status of the focus pronoun is not entirely clear from their discussion. Hedberg (2000:907,912f) proposes an analysis of it-clefts where the focus pronoun is the lower argument of identificational be, which is base-generated as the complement of be in the matrix clause. The identity between the focus constituent and the relativised constituent in the clause is assumed to be purely semantic, and is therefore captured at LF rather than in the syntax (Hedberg 2000:913). Since the focus is treated as the lower argument of be, the focus pronoun will receive objective Arg-Case, no matter whether the relativised constituent in the clause is the subject, object of a verb, or object of a preposition. As discussed in Section 2.2.1, there are good reasons to assume that the verb be fails to project a vP-layer. This means that the focus pronoun will be unable to check Pos-Case, and will instead be influenced the Def-Case constraint, which calls for objective pronoun forms. Hedberg’s (2000) analysis would thus lead us to expect a general preference for objective case in it-clefts. The final strong pronoun context tested in the survey was the than-comparative construction where than is followed by a noun phrase. The sequence than DP could either be analysed as a preposition followed by a DP, or as a preposition followed by the DP remnant of an elliptical finite clause.7 Pronoun case variation would be most readily expected in than-comparatives where the 7

Compare Jespersen & Haislund’s (1949:227f) suggestion that than and as could be analysed either as ‘conjunctions’ or as prepositions. Jespersen (1946:358f) and Gelderen (2004:125) point out, these two alternative analyses are also available for like in Modern English.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

86

noun phrase following than could be interpreted as the subject of an ellipted finite clause.8 I therefore included only than-comparatives where the pronoun following than would function as the subject of a finite clause if the than comparative was analysed as involving ellipsis (18).9 (18) a. She takes these corners much faster than [I take these corners]. [qu264]

b. I’m sure Andrea will arrive earlier than [he will arrive]. [qu259] c. Oliver is bound to respond more quickly than [she will respond]. [qu252]

d. They obviously thought they would fare better than [we fared]. [qu245]

e. He’s been training much harder than [they have been training]. [qu243]

In an ellipsis approach, the case of the pronoun is constrained by the properties of the ellipted clause. Chao (1987:65-74) argues that LF reconstruction is obligatory for any predicates that are missing at Spell-Out in ellipsis constructions. This means that a remnant subject pronoun will receive nominative Arg-Case from the missing embedded predicate. If the ellipted clause is finite and the pronoun is analysed as occupying [Spec,TP], the pronoun will also be able to check nominative Pos-Case, because T must be present at Spell-

8

Among the pronoun data cited by Jespersen & Haislund (1949:234f), the only nominatives that could not possibly be interpreted as subjects of a clause come from Shakespeare plays. An example is given in (i). (i)

I hope I shall see an end of him; for my soul, yet I know not why; hates nothing more than he. (Shakespeare, As you like it:I.i.171) [Jespersen & Haislund 1949:234]

9

I decided to use only sentences where the verb of the ellipted clause would be identical to the verb in the matrix clause, since the potential subject status of the pronoun following than is clearest in this context. It is however interesting to note that many of Jespersen & Haislund’s (1949:234f) examples of nominatives after than appear in contexts where the pronoun can be interpreted as the subject of a copular construction (i)-(ii). (i)

he did not believe himself better than [they were good] because he had not yielded to their temptations (Hugh Walpole 1920, Captives:85) [Jespersen & Haislund 1949:234]

(ii) To a thinner man than [I am thin], or from a stouter man than [he was stout], the question might have been offensive (James R. Lowell 1871, My study windows:64) [Jespersen & Haislund 1949:234]

CHAPTER 4

87

Out to license the empty FP/VP (cf. Lobeck 1995 and Merchant 2001).10 An ellipsis approach would thus predict that the pronouns in (18) will all surface in their nominative forms. As discussed in Section 2.2.4, complements of prepositions are generally unable to check Pos-Case in English. DP complements of than will also fail to receive Arg-Case, because than does not have a predicate-argument relationship with the following DP. If the pronoun is a DP complement of than, we would thus expect it to surface in the objective form demanded by the DefCase constraint. Given that the Def-Case constraint seems to be rather weak in Present-Day English (cf. Section 9.2), we might also expect the pronoun to be susceptible to non-case influences.11 4.2

Task types Two pairs of related task types occurred in the questionnaire items testing the distribution of pronoun case forms in the constructions discussed in Section 4.1. Task 1 and task 3 are straight multiple choice questions. Task 2 and task 4 are a cross between multiple choice question and cloze test. Speakers are presented with a cloze test sentence involving one of the constructions tested, and are asked to complete the sentence using a choice of different pronoun case forms. Task 1 asks speakers to choose one of two sentences (19), while task 3 asks speakers to choose two of four sentences (20). In task 2 items, speakers are offered the nominative and objective form of one pronoun, and, for coordinates, also a proper noun (21). In task 4 items, speakers are offered the nominative and objective forms of two pronouns, and are expected to pick one case form for each of these pronouns (22).

10

See Section 9.6.2 for a more detailed discussion of VP-ellipsis. Compare Sobin’s (1997:334f,337) suggestion that than lacks the head feature that triggers case checking between argument-taking prepositions and their DP complements. The post-than pronoun in the comparatives can only check case through the default accusative rule (i), and is therefore also susceptible to a nominative virus (ii). 11

(i) The Default Accusative Rule (Sobin 1997:336) If: ...[NP ACC]... 1 then: check ACC on 1. (ii) The ‘...than I’ Rule (Sobin 1997:337) If: ...than [Prn +1, -pl, NOM]... 1 2 then: check NOM on 2.

88

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(19) Imagine you're in a conversation with friends. Tick the sentence you would be most likely to come across. [qu007] (task 1) Maybe you could ask William. [ ] He and Peter helped organise the conference last year. [ ] Him and Peter helped organise the conference last year. (20) Imagine you're in a conversation with friends. [qu014] (task 3) a. Tick the sentence you would be most likely to come across. b. Pick the second best sentence and put a 2 in the box beside it. As you can imagine, Arnold was fuming. [ ] After all, Brenda had promised she would meet him and me… [ ] After all, Brenda had promised she would meet he and I… [ ] After all, Brenda had promised she would meet him and I … [ ] After all, Brenda had promised she would meet he and me… (21) Please complete the sentence using two of the following words: her mother, she, her [qu182] (task 2) Rachel Sinclair is taking over as the new assistant manager. _____________________ and ______________________ used to run that little second hand shop at the corner of Columbo and St Asaph Street. (22) Please complete the sentence using two of the following words: [qu044] (task 4) her, me, she, I Would you please tell Patrick to stop bothering Rebecca. We all know that he blames ______ and ______ for the accident, but there's no call for this sort of behaviour. Task 1 items were used to test pronoun case in (a) coordinates involving one pronominal conjunct (b) 1pl and 3pl forms in pronoun-NP constructions Task 3 items were used to test (a) pronoun case in coordinates with pronouns in both conjuncts (b) the interaction of pronoun case and relative markers (relative pronoun who versus complementizer that) in it-clefts where the relativised constituent in the clause is the subject.

CHAPTER 4

89

Task 2 items were used to test (a) pronoun case and conjunct order preferences in coordinates involving one pronoun and a non-pronominal conjunction mate, (b) the use of 1pl and 3pl forms in pronoun-NP constructions, (c) the use of pronoun case forms in the focus of it-clefts where the relativised constituent in the clause is the object of a verb or preposition, and (d) pronoun case preferences after than Task 4 items were only used to test pronoun case and conjunct order preferences in coordinates involving two pronouns with alternative case forms. As can be seen from the examples in (19)-(22), all four task types were used to test pronoun case in coordinates. While task 1 and task 2 items targeted case preferences for pronouns coordinated with non-pronominal conjunction mates, every task 3 and task 4 item elicited case preferences for two pronouns (one in initial conjunct position, and one in final conjunct position). A detailed breakdown of the numbers of items testing pronoun case selection in coordinates is given in Tables 4.1-4.4). subject

object

complement of preposition

total

task 1 (non-pronominal conjunction mate)

20

20

20

60

task 2 (non-pronominal conjunction mate)

10

10

10

30

task 3 (pronominal conjunction mate)

18

18

18

54

task 4 (pronominal conjunction mate)

9

9

9

27

total

57

57

57

171

items testing pronoun case in coordinates

Table 4.1. Total number of items testing the distribution of pronoun forms in coordinates: breakdown according to task type and syntactic position of the coordinate

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

90

subject

object

complement of preposition

total

pronoun & Name (t1)

5

5

5

15

Name & pronoun (t1)

5

5

5

15

pronoun & poss (t1)

5

5

5

15

acceptance-oriented task types (task 1 & 3)

poss & pronoun (t1)

5

5

5

15

pronoun & pronoun (t3)

18

18

18

54

total

38

38

38

114

& pronoun, Name12 (t2)

5

5

5

15

& pronoun, poss (t2)

5

5

5

15

& pronoun, pronoun (t4)

9

9

9

27

total

19

19

19

57

production-oriented task types (task 2 & 4)

Table 4.2. Number of items testing pronouns in coordinates: breakdown according to type of conjunction mate, conjunct order, and syntactic position of the coordinate

subject

object

complement of preposition

total

1sg & cm13 (task 1 and 3)

5

5

5

15

cm & 1sg (task 1 and 3)

5

5

5

15

& 1sg, cm (task 2 and 4)

5

5

5

15

total

15

15

15

45

1pl & cm (task 1 and 3)

5

5

5

15

cm & 1pl (task 1 and 3)

5

5

5

15

& 1pl, cm (task 2 and 4)

5

5

5

15

total

15

15

15

45

Table 4.3. Number of items testing 1sg (I, me) and 1pl (us, we) in coordinates: breakdown according to conjunct order (set or free) and syntactic position of the coordinate

12 13

The fronted & indicates that respondents were free to choose the conjunct order. cm stands for conjunction mate.

CHAPTER 4

91

subject

object

complement of preposition

total

3sgM & cm (task 1 and 3)

6

6

6

18

cm & 3sgM (task 1 and 3)

6

6

6

18

& 3sgM, cm (task 2 and 4)

6

6

6

18

total

18

18

18

54

3sgF & cm (task 1 and 3)

6

6

6

18

cm & 3sgF (task 1 and 3)

6

6

6

18

& 3sgF, cm (task 2 and 4)

6

6

6

18

total

18

18

18

54

3pl & cm (task 1 and 3)

6

6

6

18

cm & 3pl (task 1 and 3)

6

6

6

18

& 3pl, cm (task 2 and 4)

6

6

6

18

total

18

18

18

54

Table 4.4. Number of items testing 3sgF (she, her), 3sgM (he, him) and 3pl (they, them) in coordinates: breakdown according to conjunct order and syntactic position of the coordinate

The distribution of pronoun forms in pronoun-NP constructions was tested using task types 1 and 2. Examples of pronoun-NP items that involve task 1 and task 2 are given in (23)-(24). Table 4.5 provides a breakdown of the number of questionnaire items testing pronoun case in pronoun-NP constructions involving 1pl and 3pl pronouns. (23) Imagine you're in a conversation with friends. Tick the sentence you would be most likely to come across. [qu167] (task 1) [ ] We Mainlanders see things a little differently. [ ] Us Mainlanders see things a little differently. (24) Please complete the sentence using one of the following words: us, we [qu141] (task 2) It's a hard life for _________________ students.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

92

total

1

complement of preposition 1

1

1

3

2

2

2

6

3pl-NP task 1

1

1

1

3

task 2

1

1

1

3

2

2

2

6

1pl-NP

subject

object

task 1

1

task 2

1

total

total

3

Table 4.5. Number of items testing 1pl (we, us) and 3pl (they, them) in pronoun-NP constructions: breakdown according to task type and syntactic position of the pronoun-NP

The distribution of pronoun forms in it-clefts was tested using task 2 and task 3 items. Case preferences in subject it-clefts were elicited with questionnaire items involving task 3 (25). Task 2 items were used to test the distribution of pronoun forms in it-clefts where the relativised constituent in the clause is the object of a verb (26) or the object of a preposition (27). As we will see in Section 5.3, the 2nd choice responses to task 3 it-cleft items provide valuable information on the interaction between pronoun case and relative markers in itclefts where the relativised constituent in the clause is the subject. (25) Imagine you're in a conversation with friends. [qu251] (task 3) a. Tick the sentence you would be most likely to come across. b. Pick the second best sentence and put a 2 in the box beside it. Robert feels very strongly about this. [ ] It was him who insisted on going to the rally. [ ] It was him that insisted on going to the rally. [ ] It was he that insisted on going to the rally. [ ] It was he who insisted on going to the rally. (26) Please complete the sentence using one of the following words: me, I [qu239] (task 2) It was ________________ they saw at the races. (27) Please complete the sentence using one of the following words: her, she [qu250] (task 2) It was ________________ they wanted for their commercial.

CHAPTER 4

93

Tables 4.6 and 4.7 give a breakdown of the numbers of questionnaire items testing pronoun case in different types of it-clefts. items testing pronoun case in it-clefts

task 2 (no relative marker) task 3 (clause introduced by who or that) total

role of the relativised constituent in the clause object of object of subject verb preposition 5 5

total 10

5

-

-

5

5

5

5

15

Table 4.6. Number of items testing the distribution of pronoun forms in it-clefts: breakdown according to task type and role of the relativised constituent in the clause items testing pronoun case in it-clefts

1sg (I, me) 3sgM (he, him)

role of the relativised constituent in the clause object of subject object of verb preposition (task 3) (task 2) (task 2) 1 1 1 1 1 1

total

3 3

3sgF (she, her)

1

1

1

3

1pl (we, us)

1

1

1

3

3pl (they, them)

1

1

1

3

total

5

5

5

15

Table 4.7. Number of items testing pronoun case in it-clefts: breakdown according to pronoun and role of the relativised constituent in the clause

The distribution of pronoun case in than-comparatives was tested in five task 2 items (28), one for each of the personal pronouns with alternating case forms in Present-Day English (Table 4.8). (28) Please complete the sentence using one of the following words: me, I [qu264] (task 2) So you think I'm driving dangerously. You should go for a ride with my mum. She takes these corners much faster than _____________.

94

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

items testing pronoun case in than comparatives (task 2) 1sg (I, me)

1

3sgM (he, him)

1

3sgF (she, her)

1

1pl (we, us)

1

3pl (they, them)

1

total

5

Table 4.8. Number of items testing pronoun case in than-comparatives: breakdown according to pronoun

While task 1 and task 3 focus on the acceptance of pronoun case forms in particular constructions, task 2 and task 4 are production-oriented. Task types 1 and 3 were designed to elicit pronoun forms and combinations which are accepted or at least recognised as possible (especially in an informal setting), but not necessarily considered to be correct. These task types thus targeted actual usage rather than perceived prescriptive norm. Unfortunately, speakers not only tend to be reluctant to admit using a form that is branded as incorrect or ungrammatical in prescriptive grammars, but they are often unaware of their own usage. The instructions for task types 1 and 3 were therefore phrased so as to encourage respondents to picture a conversational setting and report usage observed in their friends’ speech.14 If the distribution of pronoun case forms is governed by mechanisms of grammar, speakers would be expected to pick out only those options as plausible utterances by their peers which fit in with their own grammar. Options incompatible with their own grammar should be ruled

14 Virtually all of the speakers participating in the survey were undergraduate students aged in their late teens and early twenties. Many of them knew each other, and they often arrived at the survey sessions in groups. It would obviously have been preferable to supplement the written questionnaires with spoken data collected in an informal conversational setting, but informal group interviews conducted as part of an earlier exploratory study yielded very few instances of the constructions under investigation (cf. Quinn 1995). A brief examination of recordings from the Mobile Unit Archives of the Origins of New Zealand English (ONZE) Project at the University of Canterbury suggested that the pronoun constructions are equally rare in existing corpora of spoken New Zealand English. However, it would be worth investigating the occurrence of coordinated pronouns, pronoun-NP constructions, pronominal it-clefts, and pronominal than comparatives in the Canterbury Corpus, and in the Wellington Corpus of Spoken New Zealand English, neither of which was readily available at the outset of this study.

CHAPTER 4

95

out, because they would not be recognised as possible (New Zealand) English utterances. The production-oriented task types 2 and 4 were designed to elicit both case and ordering preferences. Although all of the task types in the survey contain a multiple choice component, the multiple choice aspect of the task may be overridden by the cloze test component in task 2 and task 4 items. The survey results for 3pl-NP items provide evidence that task 2 is more likely than task 1 to elicit responses that do not necessarily fit in with the options provided, but are closer to speakers’ actual usage. The distribution of the 3pl case forms they and them in pronoun-NP constructions was tested in three task 1 items and three task 2 items. Quite a number of survey participants rejected both they and them in at least some of the 3pl-NP items, and instead offered the distal demonstrative those, which was not included in the options provided in the questionnaire items, but is clearly favoured in this context in many varieties of English.15 As can be seen from Table 4.9, more speakers offered the distal demonstrative those in task 2 pronoun-NP items than in the corresponding task 1 items, which suggests that task 1 more strongly discouraged students from departing from the set options than task 2. subject

they - NP

them – NP

those - NP

task 1 task 2

7.06% (6) 8.64% (7)

82.35% (70) 70.37% (57)

10.59% (9) 20.99% (17)

task 1 task 2

they - NP 0.00% (0) 1.20% (1)

them – NP 83.53% (71) 77.11% (64)

those - NP 16.47% (14) 21.69% (18)

complement of P task 1 task 2

they - NP 0.00% (0) 2.53% (2)

them - NP 84.88% (73) 75.95% (60)

those - NP 15.12% (13) 21.52% (17)

object

Table 4.9. The percentages of speakers who opted for a particular pronoun form in task 1 and task 2 items testing case preferences in 3pl-NP contexts16

Further evidence that the cloze test component of task 2 is more influential than the multiple choice component comes from the task 2 item testing the distribution of 1sg forms in coordinates with a possessed noun phrase (my dad) as a conjunction mate (29). Six speakers offered Dad rather than my dad in re15

See Kjellmer (1986:445) and Wales (1996:100) for some discussion. The numbers in the rows add up to less than 90, because speakers who offered two or more options were excluded from the totals, as were speakers who failed to offer a relevant response. 16

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

96

sponse to this item (30), which suggests that Dad is felt to be more natural than my dad in this context. (29) Please complete the sentence using two of the following words: me, I, my dad [qu180] (task 2) _________ and _________ never cared much for family reunions anyway. (30) List of speakers who used Dad rather than my dad in item qu180 (29) A020 & A113 (me and Dad) A070 & A094 (Dad and I) A059 (Dad and I, me and Dad - offered as equal alternatives) A013 (Dad and me) While the cloze test items provide valuable information on case and ordering preferences, I decided not to rely exclusively on production-oriented task types, because I wanted to examine case trends in both initial and final conjuncts for each of the pronouns involved. If respondents had been free to choose the conjunct order in all coordinate items, they would have been able to avoid disfavoured orders.17 Cloze test items are also more likely to elicit hypercorrect responses than tasks where speakers are simply asked to choose one of a set of given options, because production oriented tasks are more likely to be interpreted as testing a respondent’s knowledge of grammar. 4.3

Fillers Apart from the items testing pronoun case in coordinates, pronoun-NP constructions, it-clefts and than-comparatives, the questionnaires used in the survey also included a comparatively small number of fillers involving syntactic and semantic ambiguities (31) and regional lexical items (32). (31) How would you interpret the following sentence? (please tick) Bill left Harry drunk. [ ] Bill was drunk when he left Harry. [ ] Harry was drunk when Bill left.

17 Angermeyer & Singler (2003:176) suggest that speakers will either favour the ‘Polite order’ [X & 1sg] or the ‘Vernacular order’ [1sg & X], and argue that speakers with the vernacular order will always use objective case in coordinates (cf. Parker et al. 1988 for similar observations).

CHAPTER 4

97

(32) Please tick the sentence you would be most likely to use in a conversation with friends. [ ] I'll just go and get some luncheon from the dairy. [ ] I'll just go and get some Belgium from the dairy. [ ] other (please specify): ____________________________ The main purpose of the fillers included in the survey questionnaire was to combat/relieve judgment fatigue and reinforce the impression that there were no right or wrong answers to the different questionnaire items. Even though the number of fillers was considerably smaller than that of the pronoun items, respondents appeared to spend more time on some of the fillers than on some pronoun items. Comments after the completion of the questionnaires also suggest that respondents did take the fillers seriously, were intrigued by some of the ambiguities, and had strong views on some of the lexical variants. 4.4

The contents of the five questionnaires Questionnaire I (qu001-qu064) tested only coordinated pronouns. The pronoun combinations tested were 1sg & Name, 3sgM & Name, 3sgF & Name, 1sg & 3sgM, 1sg & 3sgF. Each of the combinations was tested in all three syntactic positions (i.e. subject of a finite clause, object of a verb and complement of a preposition), with two items each testing the acceptance of case forms in a set conjunct order and one item where respondents were free to choose the conjunct order. Questionnaire II (qu065-qu124) tested pronouns in coordinates involving the following pronoun combinations: 1pl & Name, 3pl & Name, 1sg & 3pl, 3sgM & 1pl, 3sgF & 1pl. As in questionnaire I, each of the combinations was tested in all three syntactic positions (i.e. subject of a finite clause, object of a verb and complement of a preposition), with two items each testing the acceptance of case forms in a set conjunct order and one item where respondents were free to choose the conjunct order. Questionnaire III (qu125qu180) tested pronoun case in coordinates and pronoun-NP constructions. The pronoun combinations tested in coordinates were 1sg & poss,18 1pl & poss, 3sgM & 3pl, 3sgF and 3pl. Since 1pl and 3pl are the only alternating pronouns to occur readily in pronoun-NP constructions in Present-Day English, only 1pl and 3pl were tested in this context. Like coordinates, pronoun-NP constructions were tested in all three syntactic positions, with one item each asking respondents to choose one of two complete sentences, and one item each requiring 18 poss = possessed noun phrase modified by the possessive form of the pronoun tested in the item (e.g. my dad in the coordinate me and my dad).

98

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

respondents to complete a sentence using one of two pronoun case forms. Questionnaire IV (qu181-qu235) tested pronoun case in coordinates involving the following pronoun combinations: 3sgM & poss, 3sgF & poss, 3pl & poss, 3sgM & 3sgF, 1pl & 3pl. Questionnaire V (qu236-qu265) focused on pronoun case in it-clefts and than-comparatives. I randomised the order of all items in each questionnaire with the help of random number lists. Where necessary, I moved fillers to break up long blocks of test items or to separate successive items testing similar pronoun combinations and/or similar environments. The order of options for each item was randomised too. The order of questionnaire items and options within each item was held constant across all speakers. Differences between speakers can thus not be attributed to differences in the order of items and options. The disadvantage of presenting all survey participants with identical questionnaires is that it is difficult to estimate to what degree the order of items and options may have affected the choice of pronoun case forms. Schütze (1996) points out that the very first items in a questionnaire may trigger different responses from later items, when survey participants have become accustomed to the questionnaire format and tasks. Similarly, existing studies suggest that judgment fatigue may affect the answers to items at the end of a questionnaire. While the comparatively small size of my sample prevented me from testing the effects of item order on survey responses, the individual speaker results provide some supporting evidence for the exceptional status of initial and final questionnaire items. As we will see in Section 5.1, it is highly unusual for speakers to offer instances of I in initial conjuncts of object or prepositional complement coordinates. Interestingly, most of the speakers who did opt for I in these contexts, only (or primarily) selected this option in initial or final questionnaire items. The prepositional complement item given in (33) was the first task 3 coordinate item in Questionnaire I. Three speakers opted for initial I in this item, but did not select I in the initial conjuncts of other prepositional complement coordinates: A002 (I & him), A022 & A045 (I & he).19 Similarly, four respondents opted for initial I in (34), which concluded Questionnaire II, but not in other object or prepositional complement coordinates.

19

The use of the formal ‘are to + passive’ construction in the test sentence may have been another factor prompting these speakers to opt for the nominative I. As Angermeyer & Singler (2003:178) point out, I appears to be perceived as the polite 1sg form, which may lead speakers to favour it over me in formal contexts.

CHAPTER 4

99

[qu009] (33) Gary is off to the national conference on Monday. Any messages for [I/me and he/him] are to be sent on to Auckland.

(34) Well, I know I've borrowed some of Robin's and Kim's stuff on occasions, but the difference between [I/me and they/them] is that I ask first. [qu124] An additional factor contributing to the selection of initial I in (34) could be the close proximity of the corresponding subject item (35), which appeared earlier on the same page. As can be seen from (36), all but one of the speakers who selected initial I in (34) also opted for initial I in (35). (35) All I know is that the article really annoyed Ryan and Alannah. [I/me and they/them] wrote letters to the editor, but I don't think anyone else bothered. [qu122] (36) List of speakers who opted for initial I in item qu124 (34) A006 & A044: I & them in (34), I & they in qu122 (35) A043: I & them in both (34) and (35) A031: I & them in (34) - offered as an equal alternative to me & them I piloted all questionnaires on groups of 15-20 first year linguistics students at the University of Canterbury, and I used the feedback I received from pilot participants to finetune the wording of questionnaire items and instructions. Participants in the pilot study who completed more than two pilot questionnaires received a $20 gift voucher of their choice (music, book, petrol). 4.5

The sample For my core sample I recruited students attending first-year courses in chemistry, management science, geography, economics, and biology at the University of Canterbury in the winter of 1996. I gave a brief presentation (stressing the New Zealand English angle of the survey) at the start of the relevant lectures and passed forms around for students to sign up for the survey. To reach as many students as possible, I selected first-year courses with high enrolment numbers. I limited my sample to native speakers of New Zealand English, and accepted only speakers without any significant background in languages or linguistics. Virtually all of the speakers participating in the survey were undergraduate students aged in their late teens and early twenties. Most had grown up in the South Island of New Zealand, and many knew each other.

100

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Approximately 60% of the survey participants were female and around 40% were male. Of the students who signed up for the survey, 119 completed the first questionnaire. Those who followed the instructions and gave relevant answers (i.e. answers containing the pronoun combinations tested in the item, but not necessarily in the set order) to most of the questions were contacted again and asked to participate in the remainder of the core survey (i.e. Questionnaires II IV). 94 speakers returned to complete the second questionnaire, the third questionnaire was completed by 91 speakers, and 90 speakers completed all four core questionnaires. Only responses from these 90 speakers were included in the analysis of the coordinate and pronoun-NP results. About ten months later, 41 of the 90 core survey participants completed a further, shorter questionnaire on the distribution of pronoun case forms in it-clefts and than comparatives. Speakers who completed all 4 core questionnaires received a $50 gift vouchers of their choice (music, book, petrol) after they had completed the fourth questionnaire. The funding for these rewards was provided by the University of Canterbury Research Grant U6206. 4.6

Data collection Almost all of the survey participants completed their questionnaires in one of the seminar rooms in the department of linguistics at the University of Canterbury. I administered all the questionnaires myself, and always gave a brief verbal summary of the tasks involved in the questionnaire. The core questionnaires took roughly 20-40 minutes to complete. For the first two questionnaires, I arranged for participants to come in blocks, so that I could brief several people at the same time. Although I was not always able to stay in the room with the groups, I have not found any clear evidence in the completed questionnaires that the responses of participants had been influenced by the other participants in the same block. As a rule, speakers tended to be in different blocks for different questionnaires (sometimes even in a room by themselves). Since their responses nevertheless appear to be fairly consistent and systematic across all questionnaires, it seems fair to assume that their answers were not significantly influenced by the other speakers present at the same time. Nevertheless, it is impossible to completely rule out outside influence, and it is not inconceivable that influences from other survey participants lie behind some of the inconsistencies found in individual speakers’ responses to the different questionnaires.

CHAPTER 5 THE SURVEY RESULTS

5.0

Introduction This chapter summarises the most important results of the empirical survey described in Chapter 4 and discusses the problems the attested pronoun case patterns pose for a purely case-based account. Since the main aim of the survey was to identify linguistic factors influencing pronoun case choice for individual speakers, the discussion focuses on individual speaker patterns. However, I also outline the overall case trends for each of the four constructions tested in the survey and compare them to findings reported in existing studies. Section 5.1 discusses the distribution of pronoun case forms in coordinates. Section 5.2 looks at case in pronoun-NP constructions. Case preferences in it-clefts are discussed in Section 5.3, and Section 5.4 presents the survey results for than-comparatives and also looks at pronoun case after as and like. The most important case trends to emerge from the survey are summarised in Section 5.5. 5.1

Pronoun case in coordinates Many early discussions of English pronoun case assume that all conjuncts of a coordinate receive the same case, and that conjoined pronouns will either consistently surface in the same case form as lone pronouns, or always bear objective case (cf. Klima 1964; Harris 1981; Emonds 1985, 1986; and Lumsden 1987).1 However, most of the examples cited in existing studies do not provide any clear evidence for symmetric or asymmetric case assignment within coordinates, because the pronoun is coordinated with a noun phrase (1) & (3) or a pronoun that does not exhibit overt case marking (2). (1) Yeah but [me and Catherine] really don’t talk about you know (Longman Corpus of London Teenager Language, excerpt provided by Gisle Andersen and Anna-Brita Stenström) [Hudson 1995: 380]

Klima (1964), Emonds (1985, 1986), Householder (1986, 1987) and Jones (1988) all report a general trend towards the use of objective forms in 1

Compare also Henry’s (1995) discussion of case in Belfast English (cf. Section 2.4).

102

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

coordinates. At the same time, Householder (1986:152) and Denison (1996: 292) draw attention to Shakespeare’s use of the 1sg nominative I in final conjuncts of coordinates in prepositional complement position (2). (2) All debts are cleared between [you and I]. (Shakespeare 1600, The Merchant of Venice) [Angermeyer & Singler 2003:174]

Recent research by Boyland (2001) and Angermeyer & Singler (2003) indicates that final I is still common in prepositional complement coordinates, and also appears frequently in coordinates that function as the object of a verb (3). (3) Anyone who knows [my wife and I], knows that in our case at least, this stereotype does not hold water (Op-ed piece, Arizona Republic, May 26, 2001) [Angermeyer & Singler 2003:172]

Angermeyer & Singler (2003: 178f) further note that the 3sg nominatives she and he quite readily occur in initial conjuncts of object or prepositional complement coordinates (4), but will only show up in final conjuncts of nonsubject coordinates if the initial conjunct is filled by he, she, or you. (4) a. I can put [she and her daughter] together. (flight attendant) b. We’re very happy for [he and his family]. (football announcer) [Angermeyer & Singler 2003:172]

Similar correlations between pronoun case and conjunct position have been observed by Erdmann (1978), Schwartz (1985), Parker et al. (1988), Quattlebaum (1994), Zoerner (1995), Sobin (1997), Johannessen (1998), and Gelderen (1997). Like Boyland (2001) and Angermeyer & Singler (2003), these studies focus on the distribution of 1sg and 3sg pronoun forms and offer little evidence on the distribution of the 1pl and 3pl case forms we/us and they/them. 5.1.1 Overall trends The survey results provide important insights into many of the issues raised in existing studies. Overall case trends for different pronouns in coordinates are summarised in Tables 5.1-5.5. When considering the percentages in these tables, it is important to bear in mind that they represent averages across items with different conjunction mates. The ‘noun phrase’ averages combine results from coordinates involving proper nouns (e.g. me/I & Patrick) and coordinates involving possessives (e.g. me/I & my parents). Similarly, the ‘pronoun’ averages combine results from items with various different pronominal

CHAPTER 5

103

conjunction mates (e.g. me/I & he/him, me/I & she/her, me/I & they/them). For most of the questionnaire items, virtually all of the 90 core survey participants offered relevant responses (i.e. they selected one of the options presented without making any changes to the order or form of the conjuncts). Items with the set conjunct order 1sg & 3ps (name or pronoun) are the main exception. For some of these items, almost twenty speakers either changed the conjunct order or failed to provide any answer at all. Since only relevant responses were included when calculating the percentages (and thus indirectly, the average percentages presented in the tables), the figures for items with the disfavoured conjunct order are based on lower totals than the figures for items with more readily accepted conjunct orders. As can be seen from the Tables 5.1-5.6, the distribution of 1sg case forms in coordinates differs quite markedly from the distribution of non-1sg forms: With 1sg, the objective form me is clearly favoured in initial conjuncts, regardless of the syntactic position of the coordinate. Most speakers favour I in final conjuncts of subject coordinates, and many also accept and use I in final conjuncts of object and prepositional complement coordinates. With non-1sg pronouns, we find a clear preference for the objective forms him, her, us, them in final conjuncts, regardless of the syntactic position of the coordinate. The nominative forms he, she, we, they are most readily accepted in initial conjuncts of subject coordinates, but many speakers favour the objective non-1sg forms even in this context. The preference for objective forms everywhere is particularly pronounced for 1pl Table 5.5. The distributional differences between 1sg and non-1sg case forms have an important effect on the average percentages for particular pronoun case combinations, which are presented below the heading ‘more detailed breakdown’ in the overview tables. For example, many speakers who opted for the nominative combination she & I in subject coordinates, favoured the nominative-objective combination she & him in the same contexts Table 5.3. These differences are lost when the percentages are averaged out for the overview in Table 5.4. Since more items tested case preferences with non-1sg conjunction mates than with a 1sg conjunction mate, the average percentages for pronoun case combinations are most strongly influenced by non-1sg case trends. Although the average percentages in the overview tables do not reflect case differences between coordinates with 1sg and non-1sg conjunction mates, they show quite clearly that there is no overarching trend for conjuncts in coordinates to have the same case. Nor do we find evidence supporting Angermeyer & Singler’s (2003:179) suggestion that a final conjunct has to be nominative if the initial conjunct appears in the nominative case. Indeed, when the

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

104

final conjunct is filled by a non-1sg pronoun, the combination of initial nominatives with final objective forms is at least as common as NOM-NOM combinations, if not favoured. set order (t1 & t3-1st)

average % of speakers task 1 & 2

subject

I & noun phrase

object

free order (t2 & t4)

P comp subject

object

P comp

6.74

0.00

0.59

0.57

0.00

0.56

me & noun phrase

93.26

100.00

99.41

17.48

35.00

30.00

noun phrase & I

89.87

50.65

46.74

77.27

46.11

52.22

noun phrase & me

10.13

49.35

53.26

4.68

18.89

17.22

task 3-1st & 4 I & pronoun

16.62

3.03

4.66

0.37

0.00

0.00

me & pronoun

83.38

96.97

95.34

10.46

27.04

21.11

pronoun & I

95.13

51.61

65.77

85.80

44.07

43.33

4.87

48.39

34.23

3.37

28.89

35.56

pronoun & me more detailed breakdown I & NOM (same)

7.40

0.00

0.85

0.37

0.00

0.00

I & OBJ

9.23

3.03

3.81

0.00

0.00

0.00

81.99

96.09

95.34

10.46

26.67

21.11

1.39

0.87

0.00

0.00

0.37

0.00

me & OBJ (same) me & NOM NOM & I (same)

58.43

4.20

9.76

51.83

4.07

4.07

OBJ & I

35.70

47.41

56.01

33.97

40.00

39.26

OBJ & me (same)

4.49

48.39

33.49

3.37

28.52

35.56

NOM & me

0.37

0.00

0.74

0.00

0.37

0.00

Table 5.1. Average percentage of speakers who opted for a particular 1sg form in different coordinate contexts2

2

The highest percentage for each set of options is highlighted in bold. Where more than one option was strongly favoured, the relevant percentages are highlighted in bold and italicized. The number of competing options for task 1 (t1) and task 3 (t3) items is lower than for task 2 (t2) and task 4 (t4) items, because in tasks 2 & 4 speakers were free to choose the order of the conjuncts as well as the case of the conjoined pronouns. The figures in the overview tables only take the 1st choice responses to task 3 items (t3-1st) into account, because they are most readily comparable to the responses elicited by the remaining task types. In the more detailed breakdown of the results for coordinates with two pronominal conjuncts, the annotation ‘(same)’ signals that the conjoined pronouns bear the same case.

CHAPTER 5

105

The overall results indicate that pronoun case variation is largely confined to final conjuncts of object and prepositional complement coordinates for 1sg, and to initial conjuncts of subject coordinates for non-1sg pronouns. The nature of the conjunction mate affects both order and case preferences for many speakers, but the syntactic environments where variation occurs are fairly constant across conjunction mates. This suggests that, for most speakers, the nature of the conjunction mate is a minor rather than major factor in determining pronoun case choice.

average % of speakers task 1 & 2

set order (t1 & t3-1st) subject

object

free order (t2 & t4)

P comp

subject

object

P comp

he & noun phrase

74.33

4.44

6.18

45.56

9.04

5.56

him & noun phrase

25.67

95.56

93.82

26.67

63.10

63.33

noun phrase & he

19.19

3.39

2.29

6.11

1.15

2.78

noun phrase & him

80.81

96.61

97.71

21.67

26.70

28.33

he & pronoun

57.19

10.72

11.00

33.76

5.95

6.12

him & pronoun

42.81

89.28

89.00

32.37

64.48

67.19

pronoun & he

19.88

2.91

2.08

5.64

0.29

0.83

pronoun & him

80.12

97.09

97.92

28.23

29.28

25.86

he & NOM (same)

32.16

2.88

6.20

21.11

1.11

2.22

he & OBJ

25.03

7.84

4.80

12.66

4.84

3.90

him & OBJ (same)

30.43

78.75

75.50

20.56

56.15

56.63

him & NOM

12.38

10.53

13.50

11.80

8.33

10.56

NOM & he (same)

17.33

0.00

0.92

4.79

0.00

0.28

task 3-1st & 4

more detailed breakdown

OBJ & he

2.55

2.91

1.16

0.84

0.29

0.56

OBJ & him (same)

42.49

89.57

90.53

10.70

25.65

19.47

NOM & him

37.62

7.52

7.39

17.53

3.62

6.39

Table 5.2. Average percentage of speakers who opted for a particular 3sgM form in different coordinate contexts

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

106

subject coordinates

3sgF & 1sg (qu019)

3sgF & 3sgM (qu220)

3sgF & 1pl (qu104)

3sgF & 3pl (qu134)

she & NOM (same) she & OBJ her & OBJ (same) her & NOM

71.91 0.00 2.25 25.84

16.67 62.22 21.11 0.00

9.20 35.63 50.57 4.60

25.84 37.08 33.71 3.37

Table 5.3. Percentage of speakers who opted for a particular pronoun case combination in subject coordinates with a 3sgF pronoun in initial conjunct position

average % of speakers

set order (t1 & t3-1st) subject

object

she & noun phrase

67.80

3.37

5.00

her & noun phrase

32.20

96.63

noun phrase & she

18.48

1.11

noun phrase & her

81.52

she & pronoun

free order (t2 & t4) object

P comp

52.84

5.00

5.56

95.00

14.20

64.44

65.56

6.24

4.55

0.56

0.56

98.89

93.76

28.41

30.00

28.33

64.64

7.84

7.87

51.08

8.63

9.75

her & pronoun

35.36

92.16

92.13

22.05

60.48

59.56

pronoun & she

19.83

2.33

1.45

6.22

0.00

0.00

pronoun & her

80.17

97.67

98.55

21.33

31.16

30.97

she & NOM (same)

30.90

0.85

1.43

26.35

1.11

1.11

she & OBJ

33.73

6.99

6.45

24.74

7.52

8.64

her & OBJ (same)

26.91

77.67

77.20

12.88

49.65

49.55 10.01

task 1 & 2

P comp subject

st

task 3-1 & 4

more detailed breakdown

her & NOM

8.45

14.49

14.92

9.17

10.84

15.83

1.43

0.00

5.65

0.00

0.00

4.00

0.90

1.45

0.57

0.00

0.00

OBJ & her (same)

52.20

94.98

96.80

11.54

29.78

30.41

NOM & her

27.97

2.69

1.74

9.79

1.39

0.56

NOM & she (same) OBJ & she

Table 5.4. Average percentage of speakers who opted for a particular 3sgF form in different coordinate contexts

CHAPTER 5

107

The results presented in the overview tables suggest that both conjuncts in a coordinate are influenced by Positional Case (Pos-Case) and Argument Case (Arg-Case). Thus, nominative forms of all pronouns are more likely to appear in subject coordinates than in object and prepositional complement coordinates. The results also point to a general trend towards the use of objective pronoun forms in all coordinate contexts. This trend appears to be strongest for 1pl and least pronounced for 1sg.

average % of speakers task 1 & 2

set order (t1 & t3-1st) subject

object

free order (t2 & t4)

P comp subject

object

P comp

we & noun phrase

52.85

1.68

3.35

27.38

1.67

0.56

us & noun phrase

47.15

98.32

96.65

6.69

19.44

35.00

noun phrase & we

8.79

0.56

0.00

2.79

0.00

0.00

noun phrase & us

91.21

99.44

100.00

63.14

78.89

64.44

task 3-1st & 4 we & pronoun

57.94

3.12

0.75

16.72

1.85

0.37

us & pronoun

42.06

96.88

99.25

11.89

15.19

16.88

pronoun & we

10.95

0.75

0.00

8.53

0.00

0.38

pronoun & us

89.05

99.25

100.00

62.85

82.96

82.36

more detailed breakdown we & NOM (same)

16.58

0.41

0.00

4.46

0.00

0.00

we & OBJ

41.36

2.72

0.75

12.26

1.85

0.37

us & OBJ (same)

40.05

95.31

97.65

11.52

15.19

16.51

us & NOM

2.01

1.57

1.60

0.37

0.00

0.37

NOM & we (same)

8.67

0.38

0.00

6.67

0.00

0.00

OBJ & we

2.28

0.37

0.00

1.86

0.00

0.38

OBJ & us (same)

50.50

91.43

93.96

33.85

75.56

79.35

NOM & us

38.54

7.82

6.04

29.01

7.41

3.01

Table 5.5. Average percentage of speakers who opted for a particular 1pl form in different coordinate contexts

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

108

average % of speakers task 1 & 2

set order (t1 & t3-1st) subject

object

free order (t2 & t4)

P comp subject

object

P comp

they & noun phrase

60.84

2.78

1.12

35.00

3.33

3.37

them & noun phrase

39.16

97.22

98.88

18.89

49.44

62.77

noun phrase & they

15.58

1.68

2.24

5.56

1.67

0.00

noun phrase & them

84.42

98.32

97.76

40.56

45.56

33.86

task 3-1st & 4 they & pronoun

49.36

4.53

4.82

29.70

3.07

1.39

them & pronoun

50.64

95.47

95.18

27.13

49.61

52.50

pronoun & they

27.85

1.40

2.00

14.59

0.56

0.83

pronoun & them

72.15

98.60

98.00

28.58

46.77

45.28

they & NOM (same)

26.06

1.68

0.83

16.83

1.11

0.56

they & OBJ

23.31

2.85

3.98

12.87

1.96

0.83

them & OBJ (same)

34.00

82.10

79.05

18.42

38.21

42.50

them & NOM

16.63

13.36

16.13

8.71

11.40

10.00

NOM & they (same)

23.63

0.84

0.86

12.90

0.00

0.28

more detailed breakdown

OBJ & they

4.22

0.56

1.14

1.69

0.56

0.56

OBJ & them (same)

46.93

92.67

93.57

18.21

41.91

40.83

NOM & them

25.22

5.92

4.42

10.37

4.86

4.44

Table 5.6. Average percentage of speakers who opted for a particular 3pl form in different coordinate contexts

When we consider the case forms of the two conjuncts in a coordinate, the results suggest that speakers favour certain pronoun forms in certain conjunct positions, rather favouring identical case on all conjuncts, or treating one position within the coordinate as a privileged case position. This becomes even clearer when we consider the 2nd choice responses elicited in task 3 items. The 2nd choice responses to task 3 items testing pronoun case in coordinates were more varied than the 1st choice responses, and have therefore been omitted from the discussion of individual speaker results. It is however interesting to note that many speakers agreed on the two best options in certain contexts: In subject items with the set word conjunct order [pronoun & 1sg], such as (5), speakers generally selected [he/she/they & I] and [him/her/them & I] as the two best options (Table 5.7).

CHAPTER 5

109

(5) Samantha really counted on her cousins turning up early. [They/them and I/me] were supposed to get everything ready [qu082] for the disco while she set up the barbecue. 1st choice

2nd choice

total number

them & I

46

27

73

they & I

38

28

66

them & me

4

24

28

they & me

1

6

7

me & them

0

1

1

I & them, me & they, I & they

0

0

0 5

number of speakers

changed word order:

not answered

1

4

total number

90

90

Table 5.7. The number of speakers who selected a particular case combination as a 1st choice and 2nd choice in item qu082 (5)

In subject coordinates containing two non-1sg pronouns, such as (6), most speakers opted for objective forms in the final conjunct and varied the case form of the initial conjunct for the two best options (Table 5.8). (6) [We/us and he/him] had been planning to have a holiday together for ages. [qu080] 1st choice

2nd choice

total number

we & him

46

28

74

us & him

33

34

67

we & he

9

15

24

us & he

1

9

10

him & us

1

1

2

he & us, him & we, he & we

0

0

0

not answered

0

3

3

total

90

90

number of speakers

changed word order:

Table 5.8. The number of speakers who selected a particular case combination as a 1st choice and 2nd choice in item qu080 (6)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

110

In object and prepositional complement items with the set word order [pronoun & 1sg], such as (7), most speakers selected [him/her/them & I] and [him/her/them & me] as the two best options Table 5.9. (7) Needless to say, Ruth was disappointed. Even the secretary had refused to greet [she/her and I/me] when we arrived at the office. [qu024] 1st choice

2nd choice

total number

her & I

51

34

85

her & me

37

39

76

she & I

2

16

18

she & me

0

1

1

me & her

1

0

1

I & her, me & she, I & she

0

0

0

not answered

0

1

1

total number

91

91

number of speakers

changed word order:

Table 5.9. The number of speakers who selected a particular case combination as a 1st choice and 2nd choice in item qu024 (7) 3

In object and prepositional complement coordinates combining a 3sg and 1pl pronoun, such as (8), most speakers opted for us in the two responses and varied the case form of the 3sg pronoun, regardless of the conjunct order Table 5.10. (8) Of course you can come. Rick will be there too. If they'll admit [we/us and he/him] into their ranks they can hardly turn you down. [qu090]

3 The numbers in the 1st and 2nd choice columns both add up to 91, because one speaker opted for her & I and her & me as equal 1st choice responses, and another speaker selected both she & I and her & me as 2nd choices.

CHAPTER 5

number of speakers us & him

111

1st choice

2nd choice

total number

80

9

89

us & he

3

51

54

we & him

3

21

24

we & he

0

7

7

him & us

4

0

4

he & us

0

1

1

him & we, he & we

0

0

0

not answered

0

2

2

total

90

91

changed word order:

Table 5.10. The number of speakers who selected a particular case combination as a 1st choice and 2nd choice in item qu090 (8)4

As mentioned above, the overall results point to a clear case distinction between pronouns in subject coordinates and pronouns in object or prepositional complement coordinates (9)-(10). (9)

Nominative pronoun forms (I, he, she, we, they) are more likely to surface in subject coordinates than in object or prepositional complement coordinates.

(10) Objective pronoun forms (me, him, her, us, them) are more likely to surface in object or prepositional complement coordinates than in subject coordinates. While the overall results provide comparatively little evidence for case differences between coordinates in object and prepositional complement position, a comparison of the overall results for individual questionnaire items suggest that, for some speakers at least, nominative pronoun forms are not only more strongly favoured in subject than in non-subject position, but are also more readily accepted in prepositional complement position than in object position. The case differences between the three syntactic contexts tested are particularly evident in the results for the items testing coordinates with the set conjunct order 3sgM & 1sg (Tables 5.11-5.13). 4 The numbers in the 2nd choice column add up to 91, because one speaker selected both us & he and us & him as 2nd choices.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

112

1st choice

2nd choice

total number

he & I

54

16

70

him & I

29

35

64

him & me

6

27

33

he & me

0

9

9

me & him

1

2

3

I & him, me & he, I & he

0

0

0

0

2

2

90

91

number of speakers

changed word order:

not answered 5

total number

Table 5.11. Survey results for the subject item qu057, which tested the use of 1sg and 3sgM forms in the context [He/him and I/me] arrived here three hours ago.

number of speakers

1st choice

2nd choice

total number

him & me

52

27

79

him & I

35

40

75

he & I

6

14

20

he & me

0

3

3

me & him

1

1

2

I & him, me & he, I & he

0

0

0

0

5

5

94

90

changed word order:

not answered 6

total number

Table 5.12. Survey results for the object item qu014, which tested the use of 1sg and 3sgM forms in the context Brenda had promised she would meet [he/him and I/me] at the station.

5 The numbers in the 2nd choice column add up to 91, because one speaker selected both he & I and him & I as 2nd choices. 6 The numbers in the 1st choice column add up to 94, because 3 speakers offered both him & I and him & me as 1st choice responses, and 1 speaker offered both he & I and him & me as 1st choice responses.

CHAPTER 5

number of speakers him & I

113

1st choice

2nd choice

total number

44

36

80

him & me

26

32

58

he & I

19

15

34

he & me

0

4

4

me & him

1

2

3

I & him, me & he, I & he

0

0

0

not answered

0

1

1

total number

90

90

changed word order:

Table 5.13. Survey results for the prepositional complement item qu045 which tested the use of 1sg and 3sgM forms in the context The landscapes painted by [he/him and I/me] drew huge crowds at the exhibition.

As we will see below, the case differences between object and prepositional complement coordinates are quite pronounced in the individual results of a number of speakers, especially where the choice of 1sg forms is concerned. 5.1.2 Individual speaker patterns The pronoun case patterns for individual speakers point to a considerable amount of variation, both between speakers, and within the responses offered by individual speakers. Many of the individual speaker patterns point to significant differences in the distribution of 1sg (I/me) and non-1sg (he/him, she/her, we/us, they/them) pronoun forms, as well as less pronounced distributional differences between the various non-1sg pronoun forms. The most important case patterns for individual pronouns to emerge from the individual speaker results are summarised in Tables 5.14-5.18. A comparison of the patterns in these tables shows that the distribution of 1sg case forms differs markedly from the distribution of non-1sg case forms. The non-1sg patterns are characterised by a tendency towards the use of objective forms in all cells, and suggest that pronoun case is influenced primarily by the position of the coordinate in the sentence: the non-1sg nominatives he, she, we, they are virtually confined to coordinates that appear as the subject of a finite clause (S), and the non-1sg objective forms him, her, us, them are more strongly favoured in object (O) and prepositional complement coordinates (P) than in subject coordinates. The greater preference for non-1sg nominatives in initial than in final conjuncts of coordinates suggests that conjunct position is

114

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

also a factor, but it appears to be less influential than the trend towards objective forms and the subject/non-subject distinction.

1sg = A-range (21 speakers) S

me

I

O

me

I

P

me

I

1sg = D-range (16 speakers) S

me

I

O

me

x

P

me

x

1sg = B-range (12 speakers) S

me

I

O

me

x

P

me

I

1sg = G-range (10 speakers) S

me

I

O

me

me

P

me

me

The pronoun forms in the left-hand column are found in initial conjuncts, the pronoun forms in the right-hand column are found in final conjuncts. x signals the absence of a clear case preference in a particular form (i.e. neither of the two case forms was selected in 75% or more of the relevant questionnaire items). ‘A-range’ is the label identifying the pattern range characterised by the distribution of pronoun forms presented in the table. The A-range comprises not only 1sg = A, which involves a 100% compliance with the pattern given in the table, but also 1sg = A1, 1sg = A2, etc, where the compliance with the pattern in certain cells is between 75% and 100%. The table thus states that 21 speakers showed at least 75% adherence to the pattern in each cell. I decided to set the cut-off point for clear preference at 75%, because of the small token numbers for some cells. Some speakers changed the set conjunct order in certain multiple choice items, and thus produced only 3 or 4 tokens of a particular pronoun form in a given context. While 2 out of 3 tokens in a particular case form do not constitute compelling evidence for a case preference, we can fairly confidently say that a case form is favoured by a speaker if it appears in 3 out of 4 tokens (which equates to 75%).

Table 5.14. The most popular 1sg ConjP pattern ranges to emerge from the survey results (only patterns adhered to by 5 speakers or more are listed)

CHAPTER 5

3sgM = b-range (32 speakers) S

x

O P

115 3sgM = a-range (16 speakers)

him

S

him

him

him

him

O

him

him

him

him

P

him

him

3sgM = j-range (10 speakers)

3sgM = e-range (8 speakers)

S

he

x

S

he

him

O

him

him

O

him

him

P

him

him

P

him

him

Table 5.15. The most popular 3sgM ConjP pattern ranges to emerge from the survey results (only patterns adhered to by 5 speakers or more are listed)

3sgF = b-range (32 speakers)

3sgF = e-range (17 speakers)

S

x

her

S

she

her

O

her

her

O

her

her

P

her

her

P

her

her

3sgF = j-range (13 speakers)

3sgF = a-range (8 speakers)

S

she

x

S

her

her

O

her

her

O

her

her

P

her

her

P

her

her

Table 5.16. The most popular 3sgF ConjP pattern ranges to emerge from the survey results (only patterns adhered to by 5 speakers or more are listed)

1pl = b-range (30 speakers)

1pl = e-range (22 speakers)

S

x

us

S

we

us

O

us

us

O

us

us

P

us

us

P

us

us

1pl = a-range (21 speakers) S

us

O P

1pl = j-range (6 speakers)

us

S

we

x

us

us

O

us

us

us

us

P

us

us

Table 5.17. The most popular 1pl ConjP pattern ranges to emerge from the survey results (only patterns adhered to by 5 speakers or more are listed)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

116

3pl = a-range (22 speakers)

3pl = b-range (22 speakers)

S

them

them

S

x

them

O

them

them

O

them

them

P

them

them

P

them

them

3pl = j-range (12 speakers)

3pl = e-range (9 speakers)

S

they

x

S

they

them

O

them

them

O

them

them

P

them

them

P

them

them

3pl = v-range (9 speakers) x

3pl = P-range (5 speakers)

S

x

S

they

they

O

them

them

O

them

them

P

them

them

P

them

them

Table 5.18. The most popular 3pl ConjP pattern ranges to emerge from the survey results (only patterns adhered to by 5 speakers or more are listed)

The most important factor influencing the distribution of 1sg forms appears to be the position of the pronoun within the coordinate: The 1sg objective form me is clearly favoured in initial conjunct position, while the 1sg nominative I is favoured in final conjunct position. Like the non-1sg nominatives, I is more likely to occur in subject than in non-subject coordinates, but I is much more readily tolerated in object and prepositional complement coordinates than the non-1sg nominatives he, she, we, they. Tables 5.15-5.18 suggests that there is comparatively little difference in the distribution of the various non-1sg forms, although the 3pl nominative they appears to be more readily tolerated in final conjuncts of coordinates than the remaining non-1sg nominatives he, she, we. The distributional differences between 1sg and non-1sg pronoun forms are further highlighted by a comparison of the most popular variation patterns (Tables 5.19-5.23). Variation patterns highlight the contexts where individual speakers consistently offered the same case form, and contrast them with contexts where we find some degree of case variation in the responses offered by individual speakers.

CHAPTER 5

In the variation tables, the occurrence of nominative or objective pronoun forms indicates categorical selection of the form in the relevant cell (i.e. 100% compliance).

1sg = D-variation (23 speakers) S

me

I

O

me

variation

P

me

variation

Any degree of variation is indicated by variation.

1sg = R-variation (15 speakers) S

me

variation

O

me

variation

P

me

variation

The D-variation table thus states that 23 speakers categorically chose me in initial conjuncts and I in final conjuncts of subject coordinates, but exhibited some degree of variation in final conjuncts of object and prepositional complement coordinates.

1sg = Trend 3-variation (7 speakers) S

variation

I

O

me

variation

P

me

variation

117

Table 5.19. The most commonly attested variation patterns in the 1sg ConjP results (only patterns attested for 5 speakers or more are listed)

3sgM = v-variation (15 speakers) variation

3sgM = b-variation (14 speakers)

S

variation

O

him

him

O

him

him

P

him

him

P

him

him

3sgM = z-variation (11 speakers)

S

variation

him

3sgM = Trend 18-variation (7 speakers)

S

variation

variation

S

variation

him

O

variation

him

O

variation

him

P

variation

him

P

him

him

3sgM = c-variation (5 speakers) S

3sgM = variation in all cells (5 speakers)

variation

him

S

variation

variation

O

him

him

O

variation

variation

P

variation

him

P

variation

variation

Table 5.20. The most commonly attested variation patterns in the 3sgM ConjP results (only patterns attested for 5 speakers or more are listed)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

118

3sgF = b-variation (19 speakers)

3sgF = v-variation (11 speakers)

S

variation

her

S

variation

O

her

her

O

her

her

P

her

her

P

her

her

3sgF = d-variation (7 speakers)

variation

3sgF = c-variation (6 speakers)

S

variation

her

S

variation

her

O

variation

her

O

her

her

P

variation

her

P

variation

her

3sgF = x-variation (5 speakers) S

variation

variation

3sgF = z-variation (5 speakers) S

variation

variation

O

her

her

O

variation

her

P

variation

her

P

variation

her

Table 5.21. The most commonly attested variation patterns in the 3sgF ConjP results (only patterns attested for 5 speakers or more are listed)

1pl = b-variation (40 speakers)

1pl = v-variation (16 speakers)

S

variation

us

S

variation

variation

O

us

us

O

us

us

P

us

us

P

us

us

1pl = j-variation (8 speakers) variation

1pl = a-variation (6 speakers)

S

we

S

us

us

O

us

us

O

us

us

P

us

us

P

us

us

Table 5.22. The most commonly attested variation patterns in the 1pl ConjP results (only patterns attested for 5 speakers or more are listed)

CHAPTER 5

3pl = v-variation (27 speakers)

119

3pl = b-variation (17 speakers)

S

variation

variation

S

variation

them

O

them

them

O

them

them

P

them

them

P

them

them

3pl = a-variation (6 speakers)

3pl = w-variation (6 speakers)

S

them

them

S

variation

variation

O

them

them

O

variation

them

P

them

them

P

them

them

Table 5.23. The most commonly attested variation patterns in the 3pl ConjP results (only patterns attested for 5 speakers or more are listed)

The variation patterns reveal that is considerably less variation between speakers in the distribution of 1pl forms than in the distribution of other strong pronoun forms. For example, 40 speakers, almost half of the core sample, only have variation between we and us in initial conjuncts of subject coordinates. Speaker-internal variation found with 1pl is also much less pronounced than for the other pronouns with alternating forms. Thus none of the speakers in the core sample exhibited variation in all cells for coordinated 1pl pronouns, but we do find a few speakers with some degree of variation in all cells for coordinated 3sg pronouns. The comparatively high number of speakers who exhibit a v-variation pattern in the 3pl results suggests that they is rather more readily tolerated in final conjuncts of subject coordinates than any of the other non-1sg nominatives. Like the distributional differences between 1sg and non-1sg pronoun forms, this is a problem for any purely case-based approach. Even though they is more readily tolerated in final conjuncts than other non-1sg nominatives, 3pl forms still pattern much more like other non-1sg forms than 1sg forms. Most importantly, the 3pl objective form them patterns like the other non-1sg objective forms rather than the 1sg objective form me (i.e. them occurs most readily in final conjuncts, whereas me is most strongly favoured in initial conjuncts). For many speakers, the distribution of pronoun forms in object coordinates is very similar to the distribution of pronoun forms in prepositional complement coordinates. However, some of the pattern ranges and variation patterns presented above point to a pronoun case difference between objects and prepositional complements. A small number of patterns suggest that nominatives are more likely in object than in prepositional complement position (see Table 5.24). However, the majority of speakers with a difference between ob-

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

120

jects and prepositional complements exhibit a greater preference for objective pronoun forms in object than in prepositional complement coordinates (Table 5.25). 3sgM = Trend 18-variation (7 speakers) S O P

variation variation him

him him him

3pl = w-variation (6 speakers) S O P

variation variation them

variation them them

non-1sg = Trend 38 variation (7 speakers) S O P

variation variation variation

variation variation OBJ

Table 5.24. Patterns suggesting that nominatives are more likely to occur in object than in prepositional complement position

1sg = B-range (12 speakers) S

me

I

O P

me me

x I

3sgM = c-variation (5 speakers) S O P

variation him variation

him him him

3sgF = c-variation (6 speakers) S O P

3sgF = x-variation (5 speakers) S O P

variation her variation

variation her her

variation her variation

her her her

non-1sg = x-variation (7 speakers) S O P

variation OBJ variation

variation OBJ OBJ

non-1sg = Trend 39-variation (5 speakers) S

variation

variation

O P

variation variation

OBJ variation

Table 5.25. Patterns suggesting that nominatives are more likely to occur in object than in prepositional complement position

CHAPTER 5

121

To sum up, the individual speaker patterns attested in coordinates indicate that the distribution of strong pronoun forms is influenced primarily by the factors given in (11)-(13). (11) The overall position of the coordinate in the sentence: (a) Conjoined pronouns that appear as the subject of a finite clause are more likely to be nominative than conjoined pronouns in object and prepositional complement position. (b) Pronouns in object coordinates are more likely to be objective than pronouns in prepositional complement coordinates. (12) The position of the pronoun within the coordinate: (a) For 1sg, me is favoured in initial conjuncts, and I in final conjuncts. (b) For non-1sg, he, she, we, they are favoured in initial conjuncts, and him, her, us, them are favoured in final conjuncts. Thus, the objective 1sg form me patterns with the nominative non-1sg forms he, she, we, they in being favoured in initial conjunct position, while the nominative 1sg form I patterns with the objective non-1sg forms him, her, us, them in being favoured in final conjunct position. The speaker patterns emerging from the survey suggest that for many speakers the association between conjunct position and pronoun form is stronger for 1sg than for non-1sg. (13) A general tendency towards the use of the objective forms me, him, her, us, them as invariant strong pronoun forms. The speaker patterns emerging from the results suggest that this tendency is stronger for non-1sg than 1sg, and that it may vary between the non-1sg pronouns (the development of an invariant strong form appears to be most advanced for 1pl). 5.1.3 Implications for formal analyses of coordination The clear case distinctions between subject coordinates and non-subject coordinates in the empirical survey suggest that all conjuncts of a coordinate are influenced by the demands of the three case constraints proposed in Chapter 2. At the same time, the correlations between conjunct position and pronoun case preferences indicate that for many speakers, the status of initial

122

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

conjuncts differs from that of final conjuncts in a way that affects pronoun case selection. We can capture both of these properties by treating coordinates as phrases (ConjPs) that are transparent to outside case influences and have a binary internal structure (14), with the initial conjunct in [Spec,ConjP], and final conjunct in complement position (cf. Munn 1987:133,136f; Johannessen 1993, 1998; Zoerner 1995).7 (14)

ConjP 3 DP Conj' 3 Conj DP

Following suggestions by Munn (1987:133,136f) and Zoerner (1995:34), I will assume that Conj is a functional category unspecified for the major category features (±N, ±V). As a result of this underspecification, ConjP inherits the major category features of the constituents conjoined in the coordinate, and any case influences affecting the ConjP as a whole are passed on to the different conjuncts.8 As discussed in Chapter 2, the Arg-Case constraint links the highest argument of a predicate to nominative case, and any lower arguments to objective case (if the predicate is verbal or prepositional). The Arg-Case constraint would thus predict that pronouns in subject coordinates will surface in the nominative case, while pronouns in object and prepositional complement coordinates should surface in their objective case. The predictions of the Pos-Case constraint depend on our assumption about the position of conjoined pronouns at Spell-Out. Preverbal subject coordinates are most plausibly analysed as occupying [Spec,TP] at Spell-Out, and will therefore be able to enter into nominative Pos-Case checking, provided the clause is finite (15).

7

When a coordinate contains more than two conjuncts, the ConjP shells out (Zoerner 1995). Since I did not test case patterns in coordinates conjoining three pronouns or more, the discussion here focuses on coordinates with two conjuncts. It is however interesting to note that the 1sg case forms found in the initial and final conjuncts of the formulaic triple coordinate me, myself, and I match the 1sg case preferences attested for initial and final conjuncts of twoconjunct coordinates (cf. me & him vs. him & I). 8 Zoerner (1995:35) assumes that ConjP is immediately dominated by a category node that bears the category of the conjuncts. Positing such an additional node seems rather ad hoc, and becomes unnecessary if we assume that ConjP itself can inherit the category features of its conjuncts.

CHAPTER 5

(15)

CP wo C TP [+ finite] qp ConjP T’ ru ty DP1 Conj’ T [nom] ... he [nom] ru Conj DP2 and I [nom]

123

The influence of Pos-Case on conjoined pronouns that occupy [Spec,TP] of a finite clause

The relevance of Pos-Case and Def-Case to pronouns in object coordinates depends on whether the object coordinate is analysed as having remained in its VP-internal base position (16),9 or whether it is analysed as having raised to the [Spec,vP] position (17). (16) Object coordinates in the VP-internal base position are affected by Arg-Case and Def-Case requirements. For speakers who adopt this analysis, object coordinates will have the same case status as prepositional complement coordinates (cf. the majority of distribution patterns attested in the survey). (17) Object coordinates in [Spec,vP] are affected by Pos-Case (18) and Arg-Case requirements. For speakers who adopt this analysis, object coordinates will have a case status different from that of prepositional complement coordinates. (18)

FP The influence of Pos-Case on qp conjoined pronouns that occupy F vP pos. of lexical V qp [Spec,vP] at Spell-Out at Spell-Out ConjP v’ ru ty Conj’ v [obj] … DP1 him [obj] ru Conj DP2 and me [obj]

9

Support for the assumption that object coordinates may remain in [Spec,VP] comes from V-particle constructions. As can be seen from (i)-(ii), an object coordinate may either precede or follow the particle, which indicates that it may either raise to [Spec,vP] or remain in [Spec,VP] (cf. Section 2.2.1 for further discussion of object movement, verb movement, and V-particle constructions). (i) Why would anyone want to lock up [me and him]? (ii) Why would anyone want to lock [me and him] up?

124

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

As discussed in Section 5.1.2, most speakers with case differences between object and prepositional complement coordinates exhibit a stronger preference for objective forms in object coordinates than in prepositional complement coordinates. These case differences between objects of verbs and complements of prepositions provide support for the distinction between Arg-Case and Pos-Case, and for the assumption that prepositions are unable enter into Pos-Case checking with their complements. The greater popularity of objective forms in object coordinates suggests that the combination of Pos-Case and Arg-Case is stronger than the combination of Arg-Case and Def-Case. Although the occurrence of nominative pronoun forms in prepositional complement coordinates cannot be predicted by any of the case constraints proposed here, a low weighting of Arg-Case and Def-Case will render pronouns in prepositional complement coordinates more susceptible to non-case influences. From a case point of view, the ready acceptance and use of I in final conjuncts of object and prepositional complement coordinates and the distributional differences between 1sg and non-1sg forms are the most problematic aspects of the survey results. Any purely case-based approach would predict that nominative and objective pronoun forms will occur in complementary contexts, with all nominative forms confined to subjects of finite clauses. In Chapter 6, I will argue that pronoun form selection in strong pronoun contexts is not only influenced by case considerations, but also by the relative structural complexity of the position the pronoun occupies in the construction concerned. As we will see in the remainder of this chapter as well as in Chapters 8 & 9, the pronoun forms favoured in initial conjuncts of coordinates are also favoured in pronounNP constructions and other contexts where the pronoun asymmetrically ccommands the remainder of the construction. Similarly, the pronoun forms favoured in final conjunct position are also favoured when the pronoun appears as the complement of a head (e.g. V or P), and is asymmetrically c-commanded by the head and other constituents of the phrase it appears in (cf. Section 6.2 for further discussion of asymmetric c-command). The ConjP analysis outlined in (14) offers an elegant way to capture these pronoun case similarities between coordinates and other strong pronoun contexts. 5.2

Pronoun-NP constructions Wales (1996:100) observes that us is favoured in informal spoken and written English when the 1pl pronoun is followed by an NP, even when the whole pronoun-NP construction appears as the subject of a finite matrix clause (19) or embedded clause (20). As Kjellmer (1986:445) points out, even the OED recognises the use of us NP in finite subject position as acceptable in col-

CHAPTER 5

125

loquial English, whereas the use of them in the same context tends to be stigmatised in Standard English (cf. also Wales 1996:100; Denison 1996:295). (19) [Us idlers] find it harder to admire Prince Charles (Daily Mail, 12 December 1993) [Wales 1996:100]

(20) You think [us old fellows] are fools (Jonathan Swift 1892[1738], Polite conversation:116) [Jespersen & Haislund 1949:272]

At the same time, existing studies also report instances of the nominative form we + NP, when the pronoun construction appears as the complement of a preposition (21) or as the subject of a small clause following let (22). (21) a. no mere mortal man is a match for [[we women], let alone Wonder Woman Maggie] (The Daily Express, 20 March 1990) [Wales 1996:101] b. I’m the only single one out of [we three musketeers] (Arthur W. Pinero 1895, The benefit of the doubt:12) [Jespersen & Haislund 1949:272]

c. But to [we nostalgic post-Imperial powers] it was what the Scandinavian sexologists call an anticlimax (The Guardian, 27 March 1980) [Kjellmer 1986:448]

(22) LET [[WE WORKERS] DECIDE] (sign displayed during an unofficial strike in London, 2 January 1979) [Kjellmer 1986:448]

The survey results provide further evidence for the 1pl trends and also offer more detailed information on the distributional differences between 1pl and 3pl case forms in pronoun-NP constructions (Table 5.26). As can be seen from the overview in Table 5.26, a fair number of speakers opted for us when the 1plNP construction appeared in subject position, although the majority favoured we. Us is clearly favoured in object and prepositional complement 1pl-NPs, but there are some speakers who will use the nominative we in this context. In 3plNP constructions, the objective form them is favoured over the nominative they in all syntactic positions. Although demonstratives were not given as an option in questionnaire items testing 3pl-NPs, quite a few speakers offered the distal demonstrative those in place of a personal 3pl form. While this reaction to 3plNP items ties in with Wales’ (1996) and Kjellmer’s (1986) observations about the stigmatisation of them-NP, the survey results also show that the use of the

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

126

nominative they is even more disfavoured in this context. This suggests that, for many speakers, the syntactic status of the pronoun in 3pl-NP constructions is quite different to the syntactic status of the pronoun in 1pl-NPs. % of speakers (numbers in brackets)

task 1 subject

object

task 2 comp of P

subject

object

comp of P

we – NP

59.55 (53)

3.33 (3)

5.62 (5)

68.54 (61)

11.24 (10)

11.11 (10)

us – NP

40.45 (35)

96.67 (87)

94.38 (84)

31.46 (28)

88.76 (79)

88.89 (80)

they – NP

7.06 (6)

0.00 (0)

0.00 (0)

them - NP

82.35 (70)

83.53 (71)

84.88 (73)

70.37 (57)

8.64 (7)

77.11 (64)

1.20 (1)

75.95 (60)

2.53 (2)

those - NP

10.59 (9)

16.47 (14)

15.12 (13)

20.99 (17)

21.69 (18)

21.52 (17)

Table 5.26. Percentage of survey participants that opted for particular 1pl and 3pl case forms in pronoun-NP constructions appearing in different syntactic positions

A comparison of the most common individual case patterns in pronounNP items (Tables 5.27-5.28) further confirms the sharp difference between 1pl and 3pl that is evident from the overall results. 1pl = pn7 (28 speakers)

1pl-NP = pn4 (25 speakers)

S

we

S

x

O

us

O

us

P

us

P

us

1pl = pn1 (19 speakers) S

us

O

us

P

us

Table 5.27. The most popular 1pl-NP patterns to emerge from the survey results (only patterns attested for 5 speakers or more are listed)

3pl = pn1 (64 speakers)

3pl = pn7 (5 speakers)

S

them

S

they

O

them

O

them

P

them

P

them

Table 5.28. The most popular 3pl-NP patterns to emerge from the survey results (only patterns attested for 5 speakers or more are listed here)

CHAPTER 5

127

For many speakers, the results for 1pl-NP resemble the results for 1pl in initial conjuncts of coordinates (ConjPs). That is, the nominative we tends to be favoured in subject position, and the objective us in object and prepositional complement position. However, most speakers exhibit a greater preference for we in 1pl-NP constructions than in initial conjuncts of both subject and nonsubject ConjPs. Only six of the 90 core survey participants exhibit a greater tolerance of we in initial conjuncts of ConjPs than in pronoun-NPs (Table 5.29). For these speakers, the 1pl pattern would seem to correspond to the most popular 3pl pattern in pronoun-NP constructions. A042 A043 A068

distribution of 1pl in pronoun-NP

S

us

distribution of 1pl in initial conjuncts of coordinates

we (us)

O

us

us

P

us

us

we

A050 S

us

O

us

us

P

us

us

A100 S

x

we

O

us

us

P

us

us

x

we

A106 S O

us

x

P

us

us

Table 5.29. List of speakers who were more ready to accept/use we in initial conjunct of coordinates than in pronoun-NP constructions

Almost all of the respondents who are ready to accept we in initial conjuncts of object and prepositional complement ConjPs also offered we in object and prepositional complement pronoun-NPs (Table 5.30).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

128

A027 S

distribution of 1pl in pronoun-NP we

distribution of 1pl in initial conjuncts of coordinates we

O

x

we (us)

P

x

x

A066 S

we

we

O

x

x

P

x

x

A081 S

we

O

x

x

P

x

us (we)

we

Table 5.30. The distribution of 1pl case forms in pronoun-NP constructions and initial conjuncts of coordinates for speakers with we in initial conjuncts of object and/or prepositional complement coordinates (x = clear variation between we and us)

It is possible that the greater popularity of us in ConjPs than in 1pl-NP constructions is due to the absence of a direct relation between the individual conjuncts of the ConjP and the agreement-related functional head that triggers Pos-Case on the whole ConjP. While the lack of a direct spec-head relation would not appear to prevent Pos-Case from influencing the form of conjoined pronouns altogether, its influence could well be weakened by the presence of the ConjP structure. In 1pl-NP constructions, on the other hand, the pronoun will have a direct relation with the agreement-related functional head that triggers Pos-Case, as long as the pronoun is analysed as the head of the pronounNP construction. Unlike 1pl pronouns, 3pl pronouns almost always surfaces in the objective form them in pronoun-NP constructions, no matter which 3pl forms a speaker favours in initial conjuncts of coordinates, and regardless of their 1pl-NP pattern (Table 5.31). They-NP was offered almost exclusively by respondents who also showed a strong preference for they in initial conjuncts of coordinates. The only exception is A025, who has categorical them in initial conjuncts of subject coordinates, but variation between they and them in subject 3pl-NP constructions. Nine of the twelve speakers who offered one or more instances of they-NP also favoured we-NP in the same position. Of the three speakers who bucked this trend, A070 at least offered an instance of we in a subject 1pl-NP, but A042 and A043 have us in all 1pl-NP items (see Table

CHAPTER 5

129

5.32). It is interesting to note that A042 and A043 also exhibit exceptional distribution patterns in other constructions, and could have influenced each other’s responses because they completed the questionnaires at the same time. A020 S

pronoun-NP us (t1, t2)

A010

them (t1, t2)

S

pronoun-NP we (t2) us (t1)

them (t1, t2)

O

us (t1, t2)

them (t1, t2)

O

us (t1, t2)

them (t1, t2)

P

us (t1, t2)

them (t1, t2)

P

us (t1, t2)

them (t1, t2)

A002

pronoun-NP

A006

pronoun-NP

S

we (t1, t2)

them (t1, t2)

S

we (t1, t2)

them (t1, t2)

O

us (t1, t2)

them (t1, t2)

O

us (t1, t2)

them (t1, t2)

P

us (t1, t2)

them (t1, t2)

P

we (t1, t2)

them (t1, t2)

A066

pronoun-NP

A004

pronoun-NP

S

we (t1, t2)

them (t1, t2)

S

we (t1, t2)

them (t1, t2)

O

we (t2) us (t1)

them (t1, t2)

O

we (t2 disfav) us (t1)

them (t1, t2)

P

we (t2) us (t1)

them (t1, t2)

P

we (t1, t2 disfav)

them (t1, t2 disfav)

Table 5.31. Speaker patterns which illustrate that the preference for them in 3pl-NPs does not necessarily correlate with a preference for us in 1pl-NPs A070

pronoun-NP

A043

pronoun-NP

S

we (t2) us (t1)

them (t1) n/a (t2)

S

us (t1, t2)

they (t1, t2)

O

us (t1, t2)

they (t2) them (t1)

O

us (t1, t2)

them (t1, t2)

P

us (t1, t2)

them (t1, t2)

P

us (t1, t2)

them (t1, t2)

A042 S

pronoun-NP us (t1, t2)

them (t1, t2)

O

us (t1, t2)

them (t1, t2)

P

us (t1, t2)

they (t2) them (t1)

Table 5.32. The pronoun-NP results for speakers with they-NP in syntactic positions where they did not select we-NP

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

130

Another factor that sets the 3pl-NP results apart from the 1pl-NP results is the tendency towards replacing the personal 3pl pronoun with the distal demonstrative those (Table 5.33). The interchangeability between them and those in 3pl-NP constructions, and the sharp differences in the 1pl-NP and 3pl-NP case patterns raises the question whether 1pl-NP and 3pl-NP have the same syntactic structure. 3pl = dn5 (7 speakers)10

3pl = dn1 (7 speakers) S

those

S

them/those

O

those

O

them/those

P

those

P

them/those

Table 5.33. The most popular demonstrative patterns to emerge from the 3pl-NP results (only patterns attested for 5 speakers or more are listed here)

The distributional differences between 1pl and 3pl pronouns in pronounNP constructions suggest that the two pronouns occupy different syntactic positions. The susceptibility of 1pl-NPs to Pos-Case, Arg-Case, and Def-Case requirements indicates that, for most speakers, the 1pl pronoun heads the 1pl-NP construction at Spell-Out (cf. Abney 1987:282ff; Hudson 1987:122f; Longobardi 1994:636n.31; Cardinaletti 1994:202-205; and Spinillo 2004:87-91). Following Cardinaletti (1994:202-205), I will assume that pronouns able to take modifiers are base-generated in N but generally raise to D before Spell-Out (19). (19)

DP ei D NumP 1 3 NumP NP Ni D we 2 New Zealanders Num NP 4 N ti

As I will demonstrate in Sections 8.14.1 and 8.14.2, the assumption that strong pronouns start out in N allows us to account for the possible occurrence of adjectives in premodifying position (e.g. poor us), and offers an elegant way to 10 Note that all of the speakers with Pattern dn5 were also counted as exhibiting Pattern pn1 in Table 5.27.

CHAPTER 5

131

analyse reflexive constructions such as he himself. Cardinaletti’s (1994) analysis of pronoun-NP constructions also has the advantage of capturing the similarity between pronouns modified by an NP/NumP and pronouns modified by a PP or a restrictive relative clause (cf. Sections 8.14.4 and 8.14.7). The consistent selection of them rather than they in 3pl-NP constructions suggests that a 3pl pronoun generally occupies a different syntactic position from 1pl in pronoun-NP constructions. The position in question would appear to be unaffected by nominative Pos-Case when the whole DP appears in [Spec,TP]. I will therefore assume that a 3pl pronoun (and an unstressed demonstrative such as those) generally occupies [Spec,DP] at Spell-Out when it appears in a pronoun-NP construction (20).11 (20)

DP ei DP D’ those 3 them D NumP Australians

Since the pronoun in [Spec,DP] does not head the DP dominating the whole noun phrase, it will be unable to receive an Arg-Case or Pos-Case assigned to the whole DP (the case only percolates to the head of the phrase).12 It will also be unable to receive Arg-Case or Pos-Case from within the noun phrase, because it is not an argument of the noun Australians. This means that the surface form of a 3pl pronoun in pronoun-NP constructions will be influenced only by the Def-Case constraint, which calls for objective pronoun forms.

11 As Liz Pearce (p.c.) points out, this proposal strongly resembles the analysis of Romance pronouns and demonstratives proposed by Giusti (2001:159,166f). Giusti (2001:166) argues that the pronominal/demonstrative DP in utterances like (20) is base-generated in [Spec,NP] and raises to [Spec,DP] before Spell-Out. Since those/them is not an argument of Australians in (57), the pronoun in [Spec,DP] will be unable to check genitive Pos-Case with D, no matter whether it is base-generated in [Spec,DP] or has undergone overt movement to this position. 12 Rigter (1980), Hudson (1987:122), and Spinillo (2004:90,146) all note that an important argument against a specifier analysis of pronouns in pronoun-NP constructions is their influence on the person marking of reflexives. Thus, the person of the reflexive generally matches the person of the pronoun rather than the person of the noun phrase (i).

(i) We linguists are proud of ourselves/*themselves. (Spinillo 2004:90) However, this argument does not apply to 3pl-NP constructions, because the pronoun and the following noun phrase both have 3pl features, and thus both predict the occurrence of a 3pl reflexive.

132

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Possible independent support for a syntactic distinction between 3pl-NP and 1pl-NP constructions comes from the most natural interpretation associated with the two types of constructions.13 In 1pl-NP constructions, the noun phrase is generally interpreted as a restriction on the set of referents picked out by the 1pl pronoun (21). This interpretation of we/us-NP was already noted by Jespersen (1949[1927]:102), who compares the construction to pronoun + relative clause, and comments that the NP and relative clause are restrictive in the same way, because they indicate “what person or persons should be included besides the speaker”. (21) [We/us nurses] are overworked and underpaid. = We who are nurses are overworked and underpaid. 3pl-NPs and noun phrases introduced by unstressed those, on the other hand, basically have a generic interpretation, with them and those used primarily to introduce the phrase as old information and convey the emotional distance between the speaker and the set of referents denoted by the noun phrase (22).14 (22) [Them/those politicians] would … = Politicians would … & I don’t like politicians For speakers who favour us in all 1pl-NP constructions, us could be seen to convey the speaker’s identification with the group of people denoted by the noun (cf. Panagiotidis 2002b:191). In that case, the whole noun phrase will basically have a generic interpretation, and the pronoun will be interpreted primarily as expressing the speaker’s allegiance with the set of referents picked out by the noun (23). (23) [Us nurses] are overworked and underpaid. = Nurses are overworked and underpaid & I am a nurse myself When it receives the interpretation in (23), the 1pl pronoun will appear in

13

Compare suggestions by Ritter (1995:405) and Ghomeshi & Ritter (1996:96-100) that the referential and morphosyntactic properties of 3ps pronouns differ from those of 1ps & 2ps pronouns, and that these differences are reflected in the syntactic structures the pronouns project. 14 The semantic differences between 1pl and 3pl pronouns in pronoun-NP constructions are reminiscent of observations by Benveniste (1966:228) and Harley & Ritter (2002:26), that ‘3rd person’ is really “the absence of person”. I would like to thank Liz Pearce (p.c.) for drawing my attention to Harley & Ritter’s work on this topic.

CHAPTER 5

133

[Spec,DP] at Spell-Out (24), rather than heading the DP itself.15 (24)

DP ei DP D’ us 3 D NumP nurses

The survey results suggest that a small number of speakers favour they in subject 3pl-NPs as well as 1pl-NPs. These speakers would appear to have extended the favoured analysis for 1pl-NPs to 3pl-NPs, and analyse the 3pl pronoun as occupying D rather than [Spec,DP] at Spell-Out (25).16 (25)

DP ei D NumP they Australians

5.3

Pronoun case in it-clefts The survey results for it-clefts (Tables 5.34–5.35) resemble the results for pronouns in initial conjuncts of object and prepositional complement coordinates. In keeping with observations by Jespersen & Haislund (1949:254) and Wales (1996:95f), use and acceptance of nominative pronoun forms is almost completely confined to clefts where the relativised constituent in the clause is the subject, and nominatives are considerably more likely to co-occur with the relative pronoun who than with the complementizer that.

15 A specifier analysis of the pronoun in 1pl-NP constructions is less plausible than that of the pronoun in 3pl-NP constructions, because a reflexive will generally agree in person with the 1pl pronoun rather than the following noun phrase (i). Interestingly, the use of a 3pl reflexive seems much less degraded when the 1pl pronoun in the pronoun-NP construction appears in the objective case (ii). The analysis proposed here would lead us to expect that speakers who exhibit a stronger preference for us in subject 1pl-NP constructions than in initial conjuncts of subject coordinates (Table 5.29) will also accept/use a 3pl reflexive in sentences like (ii).

(i) We linguists are proud of ourselves/*themselves. (Spinillo 2004:90) (ii) Us linguists are proud of themselves. 16 Compare Giusti’s (2001:166f) suggestion that a demonstrative or pronoun in [Spec,DP] may be reanalysed as occupying D rather than the specifier position. As Giusti (2001:167) points out, the structures in (20) and (25) may well coexist in a language for several generations, because English (and Romance) do not appear to permit the co-occurrence of an overt element in [Spec,DP] with an overt D head in pronoun-NP constructions.

134

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

% of speakers (numbers in brackets)

it was I it was me it was I who it was I that it was me who it was me that it was he it was him it was he who it was he that it was him who it was him that it was she it was her it was she who it was she that it was her who it was her that it was we it was us it was we who it was we that it was us who it was us that it was they it was them it was they who it was they that it was them who it was them that

the relativised constituent in the clause is the subject (task 3) 10.00 90.00 7.50 2.50 65.00 25.00 37.50 62.50 30.00 7.50 37.50 25.00 22.50 77.50 20.00 2.50 27.50 50.00 14.63 85.37 12.20 2.44 29.27 56.10 21.95 78.05 21.95 0.00 56.10 21.95

(4) (36) (3) (1) (26) (10) (15) (25) (12) (3) (15) (10) (9) (31) (8) (1) (11) (20) (6) (35) (5) (1) (12) (23) (9) (32) (9) (0) (23) (9)

the relativised constituent in the clause is the object of a V (task 2)

the relativised constituent in the clause is the object of a P (task 2)

2.44 97.56 n/a n/a n/a n/a 4.88 95.12 n/a n/a n/a n/a 9.76 90.24 n/a n/a n/a n/a 2.44 97.56 n/a n/a n/a n/a 7.32 92.68 n/a n/a n/a n/a

2.44 97.56 n/a n/a n/a n/a 9.76 92.50 n/a n/a n/a n/a 7.50 92.50 n/a n/a n/a n/a 2.44 97.56 n/a n/a n/a n/a 4.88 95.12 n/a n/a n/a n/a

(1) (40)

(2) (39)

(4) (37)

(1) (40)

(3) (38)

(1) (40)

(4) (37)

(3) (37)

(1) (40)

(2) (39)

Table 5.34. Percentage of speakers who opted for a particular pronoun case form and relative marker in the various it-cleft items included in the survey17

17 The highest percentage for a particular set of options is highlighted in bold. In the more detailed breakdowns for subject clefts, the highest percentage of nominatives is given in bold print and italicised.

CHAPTER 5

135

It-cleft summary pattern 1 (14 speakers) role of the relativised case of non-1sg constituent in the clause case of 1sg pronouns subject OBJ OBJ 1st choice OBJ OBJ 2nd choice object of V or P OBJ OBJ It-cleft summary pattern 3 (8 speakers)18 role of the relativised case of non-1sg constituent in the clause case of 1sg pronouns subject OBJ NOM and OBJ 1st choice OBJ (NOM and) OBJ 2nd choice object of V or P OBJ OBJ It-cleft summary pattern 2 (3 speakers) role of the relativised case of non-1sg constituent in the clause case of 1sg pronouns subject st OBJ OBJ 1 choice OBJ NOM and OBJ 2nd choice object of V or P OBJ OBJ It-cleft summary pattern 10 (3 speakers) role of the relativised case of non-1sg constituent in the clause case of 1sg pronouns subject OBJ OBJ 1st choice OBJ NOM and OBJ 2nd choice object of V or P OBJ NOM and OBJ Table 5.35. The most popular it-cleft patterns attested in the empirical survey (only patterns exhibited by at least 3 speakers are listed)

As in initial conjuncts of object and prepositional complement coordinates, we find hardly any instances of the 1sg nominative I, and the 1pl nominative we is also fairly rare. Third person nominatives are more common, with the 3sgM form he particularly well-accepted. These results resemble the trends Wales (1996:96) found in her Survey of English Usage corpus, but they do not match Sobin’s (1997:334) claim that singular pronouns are generally more likely to surface in the nominative than plural pronouns.

18 All but one of the eight speakers with this summary pattern offered at least one non-1sg nominative as a 2nd choice in subject coordinates.

136

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The individual speaker results (Table 5.35) confirm both the overall preference for objective pronoun forms in the focus of it-clefts, and the difference between subject and non-subject it-clefts that is suggested by the overall results. 14 of the 41 speakers who completed the relevant questionnaire items opted for objective pronoun forms in all it-cleft items, both as a first and second choice. A further 3 speakers only selected nominative pronoun forms as a 2nd choice. Virtually all speakers who offered nominative pronoun forms in nonsubject it-clefts also offered nominative pronoun forms in at least some subject it-clefts. The only exception is speaker A054, who offered both she and her in an it-cleft where the relativised constituent in the clause is the object of a preposition, and offered no instances of nominative pronoun forms in subject it-clefts (Table 5.36). A054 role of the relativised constituent in the clause

pronoun forms and relative markers selected as a 1st choice (2nd choices are given in brackets)

object of V

me who (me that) me

him who (him that) him

object of P

me

him

subject

her that (her who) her she her

us who (us that) us

them who (them that) them

us

them

Table 5.36. The it-cleft results for speaker A054

The individual speaker result also reveal that speakers are much more likely to accept a non-1sg nominative in the focus of it-clefts than the 1sg nominative I. Only speaker A018 opted for I in a subject it-cleft but failed to offer any non-1sg nominatives (Table 5.37). A018 role of the relativised constituent in the clause

pronoun forms and relative markers selected as a 1st choice (2nd choices are given in brackets)

object of V

I who (me who) me

him who her who us who (him that) (her that) (us that) him her us

them who (them that) them

object of P

me

him

them

subject

her

us

Table 5.37. The it-cleft results for speaker A018

The case patterns exhibited by individual speakers provide further evidence that nominative focus pronouns more likely to occur when a following subject clause is introduced by the relative pronoun who than by the comple-

CHAPTER 5

137

mentizer that. 12 speakers offered nominatives with who, but not with that (Table 5.38), and 12 speakers offered nominatives with both who and that (Table 5.39). Only one speaker (A070) offered a nominative with that (the 2nd choice he that) but failed to offer any nominatives with who. speaker A082 A076 A089 A014 A041 A073 A079 A104 A018 A040 A111 A117 A096

pronoun forms and relative markers selected as a 1st choice (2nd choices are given in brackets) me that (me who) me who (me that) me that (me who) me who (me that) me who (me that) n/a (me who) me who (I who) me that (I who) I who (me who) me who (me that) me who (me that) me who (me that) me who (me that)

him that (him who) him that (he who) him who (him that) he who (him who) he who (him that) he who (him who) he who (him who) he who (him who) him who (him that) him who (he who) him who (him that) him that (him who) him that (he who)

her that (her who) her that (she who) her that (she who) her who (her that) she who (her that) n/a (her who) she who (her that) she who (her that) her who (her that) her that (her who) her who (she who) her who (she who) she who (her that)

us that (us who) us that (us who) us who (us that) us who (us that) us who (us that) us who (we who) us who (us that) we who (us that) us who (us that) us that (us who) us that (us who) us that (we who) us that (us who)

them that (they who) them who (them that) them who (they who) them who (them that) them who (they who) they who (them who) them that (them who) they who (them that) them who (them that) them who (they who) them that (them who) them who (them that) they who (them that)

Table 5.38. Pronoun forms and relative markers selected in subject it-clefts by speakers who offered at least one nominative with who, but none with that

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

138

speaker A004 A045 A047 A068 A113 A061 A002 A008 A011

A051 A010 A066

pronoun forms and relative markers selected as a 1st choice (2nd choices are given in brackets) me that him that her that we that them that (me who) (he that) (she who) (us that) (them who) me that he that her that us that them who (me who) (him that) (her who) (us who) (they who) me who he who she who us that they who (me that) (he that) (her that) (we who) (them who) me that he who she who we who they who (me who) (he that) (she that) (us who) (them that) me who he who her who us that them that (me that) (him that) (she that) (us who) (they who) I who he who her who us that them who (me who) (he that) (she who) (us who) (they who) him who, me who her that us who they who him that (me that) (her who) (us that) (they that) (n/a) me who he that her that we who them who (I that) (him who) (she that) (us who) (they who) me that (I who)

he who (him that)

she that (her that, she who)

us that (us who)

they who (them that)

me who (I who) I that (I who) I who (I that)

he who (he that) him who (him that) he who (he that)

she who (she that) her who (her that) she who (her who)

us who (we who) we who (us who) we who (we that)

them that (them who) they who (them who) they who (they that)

Table 5.39. Pronoun forms and relative markers selected in subject it-clefts by speakers who offered nominatives with both who and that

The survey results suggest that the focus position of an it-cleft is basically an objective case position. The occurrence of nominative pronoun forms, especially in subject clefts, indicates that constraints other than the case constraints have some influence on pronoun case selection in this position. I propose that the pronoun case patterns attested in the survey are best captured in a biclausal analysis of it-clefts, where the focus appears as the lower argument of

CHAPTER 5

139

identificational be (cf. Hedberg 2000:907),19 and the clause is base-generated as an adverbial-like complement of V (26).20 (26)

TP ei T’ The structure of it-clefts in Present-Day English DPj it ei T FP 1 3 F VP Vk T is/was tk 3 DP V’ 3 tj V VP tk 3 DP V’ focus 3 V CP 5 tk clause

In the proposed analysis, the pronoun in the focus of the it-cleft is the lower argument of identificational be and will therefore be linked to objective ArgCase. Since identificational be does not project a vP-layer, the pronoun will not be able to undergo objective Pos-Case checking, but will instead be subject to the Def-Case constraint, which also calls for objective case. The absence of Pos-Case checking means that the focus pronoun will be more susceptible to additional case and non-case influences than weak object pronouns in [Spec,vP].

19 Akmajian (1970:150), Harris (1981:19f), Emonds (1986:96), Quirk et al. (1985:338), and Wales (1996:95) all draw attention to the similarities between the preference for objective forms in it-clefts and it BE sentences. However, empirical evidence discussed by Erdmann (1978:75-78) and Wales (1996:95f) suggests that nominatives are more readily tolerated in the focus of it-clefts than in it BE sentences. The greater acceptance of nominative pronoun forms in it-clefts would seem to support the analysis proposed here. Since the focus pronoun asymmetrically c-commands the following clause, the Relative Positional Coding constraints will predict that non-1sg pronouns will surface in their gracile nominative forms he, she, we, they in this position. In it BE constructions, on the other hand, the lower argument of BE appears in an asymmetrically c-commanded position, which means that Relative Positional Coding will call for the robust non-1sg objective forms him, her, us, them (cf. Sections 6.2 and 8.5). 20 See Delahunty (1982) for arguments in favour of assuming that the clause in an it-cleft is base-generated in its surface position rather than extraposed or dislocated from a noun phrase headed by it.

140

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The comparatively ready occurrence of nominative pronouns in subject it-clefts could be accounted for by assuming that the focus is able to inherit the Arg-Case of the relativised constituent in the clause through its interpretive relationship with the clause (cf. below for more discussion). However, the (admittedly marginal) occurrence of nominative pronoun forms in nonsubject clefts and the distributional differences between 1sg and non-1sg case forms in it-clefts both suggest that the surface form of the focus pronoun is determined at least in part by factors other than case. As can be seen from the diagram in (26), the focus of a cleft asymmetrically c-commands the clause in the proposed syntactic analyses for it-clefts. The greater preference for I in it BE sentences than in it-clefts could thus be argued to arise from the preference for me in asymmetrically c-commanding positions, and I in c-commanded positions. Similarly, the ready use of he, she, we, they in it-clefts could be argued to be at least partly due to the association between these pronoun forms and asymmetrically c-commanding positions. If the occurrence of non-1sg nominative forms in it-clefts was entirely due to Arg-Case agreement with the open position in the clause, we would expect the 1sg nominative I to be just as readily used in subject clefts as the non-1sg nominatives he, she, we, they. This expectation is not borne out by the individual speaker results. Many of the speakers in the survey opted more readily for non-1sg nominatives than for I. This is exactly what we would predict if the surface form of pronouns was influenced by asymmetric c-command relationships as well as case considerations. The proposed biclausal analysis of it-clefts has the added advantage of allowing us to capture the similarities in the diachronic development of it-clefts and it BE sentences (cf. Section 8.5.2), and the parallels between cleft clauses and restrictive relatives.21 The preference for objective focus pronouns and the lack of agreement between the verb BE and the following focus pronoun in Modern English both support an analysis of the focus pronoun as the lower argument of identificational be. However, in Chaucer’s work the verb still agrees with the focus constituent, which consistently surfaces in the nominative case (27). This indicates that in Late Middle English the focus was still analysed as the higher argument of be (28), and therefore received nominative Arg-Case and triggered subject Arg-Agreement on the verb.22 21

See Visser (1963:41), Gelderen (1997:149-151,157) for more details on the history of itclefts and it BE sentences, and Ball (1994) for empirical evidence that use of relative pronouns and complementizers in cleft clauses has undergone the same diachronic changes as the use of relative pronouns and complementizers in restrictive relatives. 22 See Section 2.4 for a more detailed discussion of Arg-Agreement

CHAPTER 5

141

(27) and it am I That loveth so hote Emelye the brighte. (Chaucer, Knight’s Tale:1736-7) [Gelderen 1997]

(28)

TP ei DPj T’ The structure of it-clefts in Late Middle English23 it ei T FP 1 3 F VP Vk T 3 am tk DP V’ I 3 focus V VP 3 tk DP V’ tj 3 V CP 6 tk That loveth so hote Emelye the brighte clause

The similarities between cleft clauses and restrictive relatives can be argued to arise from the fact that both are non-argument CPs containing a variable (i.e. an empty operator or a wh-pronoun) that must be bound by an antecedent at LF (cf. Delahunty 1982:213-224). In view of the importance of Relativised Minimality (29) in antecedent-government (cf. Rizzi 1990:1-27), we might expect that the open position in the clause will tend to be bound by the closest available overt antecedent.24 (29) Relativized Minimality as defined for antecedent-goverment (cf. Rizzi 1990:7) X antecedent-governs Y only if there is no Z such that (i) Z is a typical potential antecedent-governor for Y (ii) Z c-commands Y and does not c-command X 23 In the absence of independent evidence for verb movement to C in identificational sentences, I am assuming that a postverbal focus in it-clefts always occupies its VP-internal base position. 24 See Delahunty (1982:93) for the suggestion that “the semantics operate compositionally upon the structure provided by the syntax”, and Rizzi (1990:15ff) for evidence that the construal of a focus constituent with the clause in an it-cleft is indeed subject to Relativised Minimality.

142

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

In it-clefts with a pronominal focus, the closest suitable antecedent for the variable in the clause will always be the focus pronoun, because it raises to TP before Spell-Out. As a result, it will be the focus pronoun that binds the open position in the clause. Since the construal of the focus with the clause effectively links the focus to a position on the argument hierarchy of a predicate in the clause, we might expect a focus pronoun to be able to inherit the Arg-Case linked to that position.25 If we assume that the Arg-Case inherited from the clause may override the Arg-Case assigned to the focus pronoun by be, the proposed analysis will correctly predict that focus pronouns in subject clefts should be more likely to surface in the nominative case than focus pronouns in non-subject clefts. What this analysis cannot readily account for is the correlation between the case of the focus pronoun and the nature of the relative marker in the following clause. It is possible that phonological factors play a role here. Who resembles the nominative forms I, he, she, we, they (as well as me and also her in non-rhotic varieties of English) in that it is an open syllable, lacking a coda consonant. That, on the other hand, is a closed syllable with a consonant following the nucleus, and in this resembles the objective forms him, us, and them. Speakers may be more likely to accept and use the (generally disfavoured) nominatives in clefts when the following relative marker has the same the same syllabic structure. Interestingly, the most popular sequence of nominative + relative marker is he who, where the focus pronoun and relative marker not only have the same open syllable structure, but also the same onset. 5.4

Pronoun case in than-comparatives Data cited by Visser (1963:340) suggest that in Old English, a pronoun following comparative þonne ‘than’ generally surfaced in the nominative case (30). (30) a. ic eom betre þonne heo 1sg.NOM am better than 3sgF.NOM ‘I am better than she/her.’ (Old English Riddles (Ex. Bk., Krapp):40.28) [Visser 1963:340]

b. He is strengra ðonne ic 3sgM.NOM is stronger than 1sg.NOM ‘He is stronger than I/me.’ (Old English Gospel: Matthew III.ii) [Visser 1963:340] 25 Compare the potential effects of Discourse Inference on the Arg-Case of an independent pronoun (cf. Section 8.4).

CHAPTER 5

143

In Present-Day English, on the other hand, both objective and nominative pronoun forms are possible, as can be seen from the translations of the Old English examples. According to Jespersen & Haislund (1949:236), “the construction with the obl. [i.e. objective] case is now so universal as to be considered the normal one” with than, and the use of nominatives in than comparatives constitutes “an artificial reaction against the natural tendencies of the language”. Wales (1996:97f) draws attention to a suggestion in the literature that 3ps pronouns are more likely to take the objective case in than-comparatives than 1ps pronouns. Although Wales (1996:98) herself did not find any evidence to support this hypothesis in the Survey of English Usage corpus, she comments that the 1sg nominative I seems considerably better than the 1pl nominative we after than. The survey results provide support both for the general trend towards than + objective forms noted by Jespersen & Haislund (1949), and for the case differences between 1sg and non-1sg pronouns hinted at by Wales (1996). Pronouns in than-comparatives almost always surface in their objective forms me, him, her, us, them, although a handful of speakers opted for the nominatives I or she (Table 5.40). Interestingly, a similar (marginal) use and acceptance of nominative forms (especially I and the 3sg nominatives he and she) has been reported after other unstressed and predominantly functional prepositions such as as, like, but, except, and save (cf. Section 8.11; Jespersen & Haislund 1949: 227-236; and Denison 1996). task 2 than I than me than he than him than she than her than we than us than they than them

percentage of speakers (numbers in brackets) 12.20 (5) 87.80 (36) 0.00 (0) 100.00 (41) 7.32 (3) 92.68 (38) 0.00 (0) 100.00 (41) 0.00 (0) 100.00 (41)

Table 5.40. Percentage of survey participants that opted for particular pronoun case forms in the various than-comparatives

144

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The clear preference for objective pronoun forms after than suggests that than comparatives are generally analysed as simple prepositional phrases (31) rather than ellipsis constructions. (31)

PP 3 P DP than g D I

Since than is unable to check Pos-Case, and does not take the following noun phrase as its argument, a pronoun in a than comparative will be influenced only by Def-Case, and should be particularly susceptible to non-case influences. Only three different pronoun case patterns emerged from the survey results for than comparatives (Table 5.41). The majority of respondents opted for objective forms of all pronouns. Five speakers opted for I and non-1sg objective forms, and three speakers opted for she and otherwise objective forms. None of the survey participants offered more than one nominative pronoun form in the than comparative items. than 1 (33 speakers) me

him

her

us

them

than 2 (5 speakers) I A008 A011 A033 A068 A096

him

her

us

them

non-1sg nominatives popular in it-clefts; I offered as 2nd choice for subject cleft non-1sg nominatives popular in it-clefts; I offered as 2nd choice in subject cleft only objective forms it-clefts non-1sg nominatives favoured in subject it-clefts; 1sg always me in clefts non-1sg nominatives popular in it-clefts; 1sg always me in clefts than 3 (3 speakers)

me

him

she

us

them

A066 non-1sg nominatives generally favoured in clefts and common in all constructions A079 he and she offered as 1st choice in subject cleft, and she in all subject coordinates A117 she also offered in object it-cleft and as 2nd choice for subject cleft Table 5.41. Individual speaker patterns attested in than-comparatives

CHAPTER 5

145

As can be seen from Table 5.41, the three speakers who used the non-1sg nominative she after than also opted for she in clefts, and in various coordinate positions. This suggests that their selection of she in than-comparatives reflects a general preference for the 3sgF nominative form in strong pronoun contexts. Speakers who chose the 1sg nominative I in than-comparatives generally favoured the 1sg objective form me in it-clefts and initial conjuncts of coordinates. This suggests that the use of I after than is due to factors other than a general preference for I in strong pronoun contexts. The 1sg case differences between than comparatives and it-clefts for these speakers provide further evidence for a link between pronoun form and structural position. In it-clefts and initial conjuncts of coordinates, where the objective form me is favoured even by speakers who readily accept/use the non-1sg nominatives he, she, we, they, the pronoun asymmetrically c-commands the remainder of the construction (the clause in it-clefts, and the final conjunct in coordinates). In than comparatives and final conjuncts of coordinates, on the other hand, where these speakers favour the forms I, him, her, us, them, the pronoun is asymmetrically ccommanded by the rest of the construction.26 5.5

Summary of the most important case trends

(a)

For almost all speakers, nominative pronoun forms are most readily tolerated when (the construction containing) the pronoun appears as the subject of a finite clause.

(b)

For many speakers, the objective non-1sg forms him, her, us, them are categorical in final conjuncts of object and prepositional complement coordinates, although they frequently favour the nominative 1sg form I in this position.

26 The case trends for pronoun-NP constructions, it-clefts, than-comparatives, and pronouns premodified by an adjective (cf. Section 8.14.1) suggest that the internal structure of the DP headed by a strong pronoun only influences pronoun case selection when the DP contains an overt constituent that asymmetrically c-commands the pronoun (e.g. premodifying adjectives) or is asymmetrically c-commanded by it (e.g. the noun phrase in pronoun-NP constructions). So although strong pronouns are assumed to be base-generated in N before raising to D (cf. Chapter 3), the NumP and NP dominated by the DP the pronoun heads at Spell-Out will only have a bearing on the surface form of the pronoun if they are filled by overt constituents. I have therefore omitted the detailed internal structure of any strong pronoun DP from the tree diagrams when the lower projections within the DP are empty.

146

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(c)

For most speakers, the objective 1sg form me is categorical in initial conjuncts of object and prepositional complement coordinates, even when they tolerate some non-1sg nominatives (he, she, we, they) in this position.

(d)

Quite a few speakers more readily tolerate the 3pl nominative they in final conjuncts of subject coordinates than any other non-1sg nominative form.

(e)

For the majority of speakers, non-1sg nominatives (he, she, we, they) are most readily accepted in subject pronoun-NPs (for we) and in initial conjuncts of subject coordinates. The next most likely position to attract non1sg nominatives is the final conjunct of subject coordinates, followed closely by initial conjuncts of object and prepositional complement coordinates. Final conjuncts of object and prepositional complement coordinates are least likely to attract non-1sg nominatives.

(f)

For the majority of speakers, the 1sg nominative I is most strongly favoured in final conjuncts of subject coordinates. The next most likely position to attract the 1sg nominative I is the final conjunct of prepositional complement coordinates, followed closely by the final conjunct of object coordinates. I is generally disfavoured in initial conjunct position, but is generally more readily tolerated in initial conjuncts of subject coordinates than in initial conjuncts of object and prepositional complement coordinates.

(g)

For some speakers, there appears to be a pronoun case difference between objects and prepositional complements. (i) A small number of patterns suggest that nominatives are more likely in object than in prepositional complement position. (ii) The majority of speakers with a difference between objects and prepositional complements exhibit a greater preference for objective pronoun forms in objective than in prepositional complement coordinates For speakers in category (i), we either find a greater tolerance for the 1sg nominative I in final conjuncts of prepositional complement coordinates than in final conjuncts of object coordinates [cf. B-range pattern], or a trend towards the non-1sg nominatives he, she, we, they in initial conjuncts of prepositional complements but not in initial conjuncts of objects, or both.

CHAPTER 5

147

(h)

Many speakers favour the non-1sg objective forms him, her, us, them in all positions, but hardly any speakers exhibit a general preference for the non-1sg nominative forms he, she, we, they.

(i)

Many speakers more readily tolerate 3sg nominatives he and she than the 1pl and 3pl nominatives we and they.

(j)

Many speakers exhibit a preference for the 1pl objective form us in all positions, even when they will quite readily accept and use other non-1sg nominatives.

(k)

The 1sg nominative I is generally more readily tolerated than the non-1sg nominatives he, she, we, they. Thus, quite a number of speakers exhibit a preference for non-1sg objective forms in all positions, but at the same time favour I in final conjunct position.

(l)

Those speakers who do tend towards me in all positions invariably also exhibit a strong preference for non-1sg objective forms.

The most significant issue to emerge from the survey data is the discrepancy between the distributions of strong 1sg and non-1sg case forms. No purely case-based approach will ever be able to account for the case differences between strong 1sg and non-1sg pronouns, unless me is reanalysed as a nominative in strong pronoun contexts, and I as an objective. This is quite counterintuitive, since weak me still clearly marks the objective case, and weak I is unambiguously nominative. It seems fairly unlikely that speakers should reanalyse the strong 1sg forms as marking the inverse of the grammatical relations represented by their weak counterparts while the phonological similarity between the two is still so strong. At the same time, strong pronouns do differ from their weak counterparts in exhibiting case variability even when they are clearly influenced by Argument Case, Positional Case and/or (Positional) Default Case. In the next two chapters, I will argue that the case variation found in strong pronoun contexts is due to a clash between the demands of the three case constraints on the one hand, and the requirements imposed by two additional sets of constraints that affect only strong pronouns.

CHAPTER 6 RELATIVE POSITIONAL CODING AND THE INVARIANT STRONG FORM CONSTRAINTS

6.0

Introduction The survey results reported in Chapter 5 indicate that the distribution of strong pronoun forms in Present-Day English is at least partly determined by factors other than case. In this chapter, I will demonstrate that we need at least two additional sets of constraints to account for the most important pronoun case patterns attested in the empirical survey: the Relative Positional Coding constraints, which relate pronoun form to syntactic position, and the Invariant Strong Form constraints, which capture the tendency towards the use of invariant pronoun forms in strong pronoun contexts. Section 6.1 provides a brief summary of the pronoun case trends that emerged from the previous chapters, and argues that the surface form of a strong pronoun not only reflects the structural case of the pronoun, but also codes its position within a syntactic construction, and identifies its morphosyntactic status as a strong pronoun. Section 6.2 introduces the two Relative Positional Coding constraints, and proposes that distributional differences between strong 1sg and non-1sg forms are largely due to discrepancies between the grouping of pronoun forms for structural case, and the grouping of pronoun forms for Relative Positional Coding. Section 6.3 discusses the Invariant Strong Form constraints, and presents evidence that we need to posit a separate Invariant constraint for each strong pronoun. Section 6.4 summarises the predictions of all of the constraints proposed in this book. Ways of modelling the interaction of these constraints will be considered in Chapter 7. 6.1

The limitations of a purely case-based analysis The case trends discussed in Chapters 2 and 5 indicate that pronoun case choice is influenced by three different case constraints: Positional Case, Argument Case, and Default Case. Argument Case restricts the case form of all pronouns that are structural arguments of a predicate; Positional Case affects the case form of argument noun phrases that appear in the specifier of certain agreement-related functional heads at Spell-Out; and Default Case restricts the case form of pronouns unable to enter into Positional Case checking.

CHAPTER 6

149

None of these case constraints is able to predict the occurrence of nominative pronoun forms in prepositional complement position and in other non-subject contexts. Even more importantly, a case-based approach cannot readily accommodate the considerable distributional differences between strong 1sg and non-1sg pronoun forms, which are particularly evident in coordinates. The following pronoun case trends identified in Chapter 5 are problematic for a purely case-based approach: (a)

For many speakers, the objective non-1sg forms him, her, us, them are categorical in final conjuncts of object and prepositional complement coordinates, although they frequently favour the nominative 1sg form I in this position.

(b)

For most speakers, the objective 1sg form me is categorical in initial conjuncts of object and prepositional complement coordinates, even when they tolerate some non-1sg nominatives (he, she, we, they) in this position.

(c)

Quite a few speakers tolerate the 3pl nominative they in final conjuncts of subject coordinates more readily than any other non-1sg nominative.

(d)

For the majority of speakers, non-1sg nominatives (he, she, we, they) are most readily tolerated in initial conjuncts of subject coordinates, and we is particularly popular in subject pronoun-NPs. The next most likely position to attract non-1sg nominatives is the final conjunct of subject coordinates, followed closely by initial conjuncts of object and prepositional complement coordinates. Final conjuncts of object and prepositional complement coordinates are least likely to attract non-1sg nominatives.

(e)

For a number of speakers, the 1sg nominative I is more readily tolerated in final conjuncts of prepositional complement coordinates than in final conjuncts of object coordinates. Similarly, some speakers will tolerate the non-1sg nominatives he, she, we, they in initial conjuncts of prepositional complement coordinates but not in initial conjuncts of object coordinates. These trends point to the interaction of the case constraints with the noncase constraint responsible for the patterns in (a) and (b). As discussed in the preceding chapters, the absence of a Pos-Case position within PPs could explain why prepositional complements are less strongly affected by case requirements than objects of verbs, but none of the case constraints can account for the occurrence of nominative pronoun forms in this context.

150

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(f)

For the majority of speakers, the 1sg nominative I is most strongly favoured in final conjuncts of subject coordinates. The next most likely position to attract the 1sg nominative I is the final conjunct of prepositional complement coordinates, followed closely by the final conjunct of object coordinates. I is generally disfavoured in initial conjunct position, but is generally more readily tolerated in initial conjuncts of subject coordinates than in initial conjuncts of object and prepositional complement coordinates.

(g)

Many speakers favour the non-1sg objective forms him, her, us, them in all positions, but hardly any speakers exhibit a general preference for the non-1sg nominative forms he, she, we, they.

(h)

Many speakers exhibit a preference for the 1pl objective form us in all positions, even when they will quite readily accept and use other non-1sg nominatives.

(i)

Many speakers more readily tolerate 3sg nominatives he and she than the 1pl and 3pl nominatives we and they.

(j)

The 1sg nominative I is generally more popular than the non-1sg nominatives he, she, we, they. Thus, quite a number of speakers exhibit a preference for non-1sg objective forms in all positions, but at the same time favour I in final conjunct position. Speakers with a general preference for me, on the other hand, invariably favour non-1sg objective forms in all strong pronoun contexts.

These trends suggest that English strong pronoun forms not only serve to identify the position of the pronoun on the argument hierarchy (Argument Case), and the overt syntactic position of a pronoun relative to an agreement-related functional head (Positional Case vs. Default Case), but also code the position of a pronoun within a syntactic construction (Relative Positional Coding). The distributional differences between strong 1sg and non-1sg forms arise from discrepancies between the grouping of pronoun forms associated with structural case and the grouping of pronoun forms associated with relative positional coding. For the purposes of relative positional coding, the 1sg nominative I groups with the non-1sg accusatives him, her, us, them, while the 1sg accusative me patterns with the non-1sg nominatives he, she, we, they. As we will see in Section 6.2, these pronoun groupings can be argued to have a phonological and cognitive basis.

CHAPTER 6

151

6.2

Relative Positional Coding When we compare the most popular distribution and variation patterns for 1sg and non-1sg pronouns in coordinates, it soon becomes evident that 1sg case forms pattern quite differently from non-1sg case forms: (1)

For 1sg, the objective form me is favoured in initial conjuncts, and the nominative form I in final conjuncts.

(2)

For non-1sg, the nominative forms he, she, we, they are favoured in initial conjuncts, and the objective forms him, her, us, them are favoured in final conjuncts.

The correlation between pronoun form and conjunct position is particularly evident in the coordinate results for the following speakers: A027 S me 75 O me P me

A090 S me O me P me A033 S me 67 I 33 O me P me 86 A029 S me O P

he 89 him 82 him 56 he 44

initial conjunct she 92 we her 75 we 89 her 83 we 63

initial conjunct he 60 she 70 us 63 him 92 her 78 us 83 him 82 her 91 us

he 86

initial conjunct she 80 we 86

him 60 him 67

her 92 her 67

us us

initial conjunct she 50 we 50 her 50 us 50 me him 85 her 91 us me 83 he 50 her 91 us him 50 he 64

they 83 them 91 them

I I I

final conjunct him 90 her us him her us him 90 her us

them them them

them 63 them them

I I I

final conjunct him 75 her us 86 them him her us them him her us them

they 80

I

final conjunct him her us them 63

them them

I 90 I 75

him him

her her

us us

them them

they 83 I

final conjunct him 86 her us

them

them 82 I them I 89

him him

them them

her her

us us

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

152 A081 S me

he 92

O

me 83 he 64

P

me

A106 S me O P

me me

he 69

initial conjunct she we she 54 her 46 she 67

they

us 67 they

I 90

final conjunct him 83 her 71 us

I

him

her 83

us

they 50 them 50 them

him

her

us

them 88

us 83 they 90 I 83

initial conjunct he 50 she 67 we him 50 him 75 her 70 us 67 him her 70 us

they 55 I them 45 them 89 I 80 them 70 I 70

him

final conjunct her us

him him

her her

us us

them them them

The distribution of pronoun forms in coordinates suggests that the strong pronoun forms investigated in this study not only divide according to their case status (i.e. the nominatives I, he, she, we, they versus the accusatives me, him, her, us, them), but also along slightly different lines, to yield what I will term the gracile (3) and robust series (4).1 As I will discuss in more detail below, the series of gracile pronoun forms also includes the nominative wh-form who, and the robust series also comprises the objective wh-form whom. (3)

gracile series me he she we they who

(4)

robust series I him her us them whom

Gracile pronoun forms differ from robust pronoun forms in the way their syntactic distribution may be restricted.

1

(5)

If the distribution of gracile forms is restricted, the forms will be limited to initial conjunct position.

(6)

If the distribution of robust forms is restricted, the forms will be limited to final conjunct position.

I would like to thank Kate Kearns (p.c.) for suggesting these terms.

CHAPTER 6

153

While these restrictions on gracile and robust forms are particularly obvious in the distribution patterns of the speakers listed above, almost all speakers in the survey appear to be susceptible to the gracile-robust distinction. A speaker’s acceptance and/or use of me, he, she, we, they in final conjunct position implies that the speaker in question also uses and/or accepts these forms in initial conjunct position. A speaker’s acceptance and/or use of I, him, her, us, them in initial conjunct position implies that the speaker in question also accepts and/or uses these forms in final conjunct position. No speaker accepted the gracile forms me, he, she, we, they only in final conjuncts and the robust forms I, him, her, us, them only in initial conjuncts. The pronoun case trends found in other strong pronoun contexts provide further evidence for the distributional differences between gracile and robust pronoun forms: (a)

for some speakers the gracile 1pl form we is particularly favoured in pronoun-NP constructions (cf. Section 5.2)

(b)

3sg pronouns modified by a relative clause tend to be gracile rather than robust: e.g. she who must be obeyed (cf. Section 8.14.7)

(c)

gracile pronoun forms occur in the focus of it-clefts (cf. Section 5.3)

(d)

gracile forms occur in left-dislocated and topicalised position, and as the subject of absolutive V-ing constructions and small clauses (cf. Sections 8.1, 8.2, 8.6, and 8.8)

(e)

the robust wh-form whom is largely confined to prepositional complement position, while the gracile form who is strongly favoured in initial position; this is particularly evident in sluiced questions involving prepositions (cf. Section 9.5)

(f)

pronouns after than, as, save, but tend to be robust, as do pronouns in it BE sentences (cf. Sections 5.4, 8.5.2, and 8.11)

The distribution of pronoun forms in the constructions mentioned in (a)-(f) suggests that the gracile/robust distinction encodes asymmetric c-command (7)-(9). (7)

C-command (cf. Reinhart 1976; 1981:612) In a syntactic tree, a node α c-commands a node β iff the branching node most immediately dominating α also dominates β, and β is not a constituent of α.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

154 (8)

Asymmetric c-command (cf. Kayne 1994:4) α asymmetrically c-commands β iff (a) α c-commands β and (b) β does not c-command α

(9)

YP 3 The head (Y) and the specifier (XP) asymmetrically XP Y’ c-command the head of the complement (Z). ! 3 X Y ZP g Z

All things being equal, a pronoun will tend to be gracile if it asymmetrically c-commands other overt constituents in the same construction at Spell-Out, but robust if it appears in a position asymmetrically c-commanded by all other overt elements in the same construction (10)-(11). (10)

XP 3 X’ DP1 [gracile] 3 X DP2

(11)

DP 3 D XP [gracile] 5

[robust]

The basic distributional differences between robust and gracile forms can be captured in a constraint I will call Relative Positional Coding 1 (12).2 (12) Relative Positional Coding 1 (abbreviated as RPC 1) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then A must be gracile, and B must be robust.3 2

As I will discuss in more detail below, the survey results suggest that we need to posit an additional related constraint (Relative Positional Coding 2) to capture the distribution of the 3pl nominative they. 3 The distribution of pronoun forms in Present-Day English suggests that Relative Positional Coding primarily applies within local domains (i.e. within a coordinate, within a modified DP headed by a pronoun, within the VP containing the pronominal focus in an it-cleft). The requirement that an asymmetrically c-commanding constituent should be gracile while an asymmetrically c-commanded constituent should be robust, theoretically applies to any heads and phrases in a relationship of asymmetric c-command within a local domain. However, since alternative robust and gracile forms are available only for strong pronouns in Present-Day English, the RPC constraints effectively influence only the surface form of strong pronouns.

CHAPTER 6

155

Assuming that coordinates have the structure in (13), the initial conjunct (DP1) will asymmetrically c-command the final conjunct (DP2). (13)

ConjP ei Conj’ DP1 he ei [gracile] Conj DP2 and I [robust]

RPC 1 therefore predicts that the initial conjunct of a coordinate should be gracile and that the final conjunct should be robust. It rules out coordinates such as him and me or I and he, where the initial conjunct is robust and the final conjunct is gracile, coordinates like he and me or me and he, where both conjuncts are gracile, and coordinates such as him and I or I and him, where both conjuncts are robust. RPC 1 also predicts that pronouns followed by a restrictive modifier should be gracile (= me, he, she, we, they), because the head of a phrase a-symmetrically c-commands any constituents contained within its complement (14).4 Pronouns following prepositions, on the other hand, are predicted to surface in their robust forms (I, him, her, us, them), because the preposition asymmetrically c-commands the head of its complement (15).5 (14)

DP ei D NumP 1 3 NumP NP (a) / PP (b) / CP (c) Ni D we 2 ei [gracile] Num NP (a) linguists 4 (b) in this department N (c) who are to blame ti

(15)

PP 3 P DP than g for D I whom [robust]

As I will demonstrate in the next sections, phonology and cognitive psychology offer a number of possible explanations for the association of gracile forms 4

See Sections 5.2 and 8.14 for a detailed discussion of the syntactic relationship between a pronoun and different modifiers. 5 As mentioned in Chapter 3, I am assuming that strong pronouns start out in N but generally raise to D before Spell-Out. However, I have omitted the lower projections within the DP from all tree diagrams involving unmodified strong pronouns, because the internal structure of the DP is only relevant to Relative Positional Coding if it contains overt constituents that are in an asymmetric c-command relationship with the pronoun in D (cf. Section 5.4).

156

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

with initial/c-commanding positions and the association of robust forms with final/c-commanded positions. 6.2.1 Phonological complexity and the classification of pronoun forms In this section I will argue that the phonological complexity of pronoun forms correlates with the relative ‘structural integration’ of the syntactic position they are associated with in Relative Positional Coding. Hawkins (1994: 26) defines the ‘structural integration’ of a node within a constituent in terms of the complexity of its structural domain (16). The more structurally complex the structural domain of a node, the more complex the relationships that determine the structural integration of the node. (16) Structural domain (SD) of a node X in a constituent C (Hawkins 1994:28) The SD of a node X in C consists of the following nodes that structurally integrate X in C: (a) all nodes dominating X within C (including C itself) (b) all or some sisters of X (c) all sisters of the nodes dominating X within C Hawkins (1994:29) proposes that “complexity involves the number of structural relations within different portions of a tree, measured in terms of sets of structurally related nodes, and relative complexity involves relative numbers of these nodes”. That is, the more tree nodes a constituent contains, the greater its complexity and grammatical weight (cf. Wasow 1997:85). In Hawkins’ (1994: 42f) approach, the complexity of the structural domain of a node is directly related to asymmetric c-command (17). (17) Structural complexity and asymmetric c-command (Hawkins 1994:29,43) The structural complexity of a Structural Domain (SD) is measured by counting the set of nodes within it: the complexity of SDi exceeds that of SDj iff SDi > SDj. If A asymmetrically c-commands B, then the SD of B contains A but not vice versa. Since the structural complexity of a Structural Domain (SD) is determined by the number of nodes contained in it, the SD of a node asymmetrically c-commanded by other nodes will be more complex than the SD of the nodes that asymmetrically c-command it. As discussed in the previous section, the initial

CHAPTER 6

157

conjunct of a coordinate asymmetrically c-commands the final conjunct in a ConjP. This means that the SD of the final conjunct will be more complex than the SD of the initial conjunct. Hawkins (1994:42f) discusses work by Primus (1987, 1991), which suggests that the relative structural complexity of the SD of a noun phrase tends to correlate with the morphological complexity of the case assigned to that noun phrase. According to Primus (1987, 1991), noun phrases with the least complex SD will tend to surface in the nominative case, while noun phrases with more complex SDs will tend to bear dative or other oblique cases. Interestingly, the complexity of morphemes used to mark the relevant cases tends to correspond to the complexity of the SD: the nominative case is often marked with a zeroaffix, while the dative and other oblique cases tend to involve overt case morphemes. I would like to propose that structural complexity may not only correlate with morphological complexity, but also with phonological complexity (in the absence of morphological differences). Since the addition of overt morphemes also results in an increase in the phonological complexity of a stem, a correlation between structural and phonological complexity does not seem all that implausible. As Gil (1982) notes, “all other things being equal, an order in which small constituents precede large ones is preferable to an opposite order, in which large constituents precede small ones”. According to Gil (1982:130-2, 137f), one of the strategies languages employ to increase the phonological size of final constituents is the addition of case morphemes, agreement morphemes, or conjunctions. This generalisation is captured in his Iambic Marking Principle, which stipulates that constituents which follow other constituents in a sentence or grammatical construction will be more likely to attract morphological markers than constituents which precede others (1982:135f). Although Gil focuses on the use of morphemes to increase the phonological size of constituents, he also points out that the respective phonological size of two constituents may also affect constituent order: “given that one constituent is larger than the other, the two constituents will then be ordered in such a way that the smaller one precedes the larger one” (1982:138). Similar observations can already be found in the works of Jespersen (1894) and Behaghel (1909/10:39).6 Jespersen’s (1894:57) Principle of Relative Weight (18) captures the tendency for heavy constituents to appear in more peripheral positions than lighter constituents.

6

See Harris & Campbell (1995:21,24) and Wasow (1997:82) for more detailed discussions of Jespersen’s and Behaghel’s proposals.

158

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(18) Jespersen’s Principle of Relative Weight (Jespersen 1894:57) Lighter elements can be placed near the center while heavier ones are relegated to more peripheral places. (quoted in Harris & Campbell 1995:21) Behaghel (1909/10:34) comments on a rhythmic tendency towards an increase in length from the first to the second in a pair of constituents, which he terms ‘das Gesetz der wachsenden Glieder’ (translated as ‘the law of growing elements’ by Wasow 1998:103n.1). As Hawkins (2004:49-51) notes, ordering short constituents before longer ones tends to facilitate on-line processing, because it allows for an early identification of the syntactic structure of the utterance containing the constituents. For example, we only need to process two words (the pronoun and conjunction) to be able to identify the string they and their friends as a coordinate structure. With their friends and them, on the other hand, we have to process three words (their, friends, and) before the structure can be identified as a ConjP. While length undoubtedly influences the relative grammatical weight and ordering of constituents, their internal complexity seems to be even more important. Chomsky (1975) points out that the complexity rather than length of the object noun phrase determines the naturalness of particle movement: It is interesting to note that it is apparently not the length in words of the object that determines the naturalness of the transformation, but, rather, in some sense, its complexity. Thus “they brought all the leaders of the riot in” seems more natural than “they brought the man I saw in”. The latter, though shorter, is more complex. (Chomsky 1975:477)

Allan (1987:72-73) and the studies discussed by Bock (1982) provide further evidence for the role of phonological size and complexity in cognitive processing. As Bock (1982:18) points out, one of the factors likely to affect retrieval of a given lexical item “is the complexity or the sheer amount of information in the phonological representation of the word”. Differences in retrievability may affect the order of elements in sentences: words with a less complex phonological representation should precede words containing a greater amount of phonological information (Bock 1982:18). Robust pronoun forms do not contain more syllables or exhibit any obvious additional morphology when compared to gracile forms7, but most of them 7

The final -m in him and them might be argued to constitute a separate morpheme marking accusative case, but I can see no obvious way in which the differences between me-I, she-her, and we-us could be reduced to simple affixation. Since most varieties of New Zealand English today are non-rhotic, I am assuming that the phonological representation of her lacks a final /r/ in the speech of the survey participants.

CHAPTER 6

159

can nevertheless be argued to have greater phonological complexity than their gracile counterparts (Table 6.1). gracile mi hi i wi ei hu

1sg 3sgM 3sgF 1pl 3pl wh

robust ai hm h s em hum

Table 6.1. Phonological representations of gracile and robust pronoun forms

As distinctions between syllable types (e.g. heavy versus light) are generally made on the basis of the material in the rhyme rather than the onset, we might expect the rhyme to have the greatest bearing on the robustness of a monosyllabic item. It is important to note that the difference between gracile and robust forms is not one of syllable weight. As discussed in Chapter 3, all strong pronouns bear a certain degree of stress, and must therefore be bimoraic. What distinguishes most robust forms from their gracile counterparts is the number and featural complexity of the rhyme segments in the rhyme. In feature geometric terms, the rhymes of the gracile forms me, he, she, we, and who consist of just one vowel segment (19). This is also true for the robust 3sgF form her in non-rhotic varieties of English. Assuming that a diphthong is a sequence of two tautosyllabic vowels (cf. e.g. Kenstowicz 1994:46), the rhymes of they and I consist of two vowel segments (20). The rhymes of him, us, them, and whom consist of a vowel segment followed by a consonant segment (21). (19) Feature geometric representation of the vowel [i] (cf. Kenstowicz 1994:154) root cavity articulator terminal

 - consonantal   + sonorant  qp Oral Pharyngeal ! ! Dorsal Radical wgo ! [+ high] [-low] [-back] [+ATR]

Glottal features ([±voiced]) are not specified for [i], because voicing is not distinctive for unmarked [-consonantal, +sonorant] segments specified for Oral features (cf. Dogil & Luschützky 1990:9).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

160

(20) Feature geometric representation of the diphthong [ei] σ qp root  - consonantal   - consonantal   + sonorant   + sonorant  eo eo cavity Oral Pharyngeal Oral Pharyngeal ! ! ! ! articulator Dorsal Radical Dorsal Radical egi ! egi ! terminal [- high] [-low] [-back] [+ATR] [+ high] [-low] [-back] [+ATR]

syllable

It is possible that the Pharyngeal specification of the vowel [e] is not necessary in English, because a given variety of English will generally have only one distinctive [-high, -low, -back] vowel (realised as [+ATR] in some varieties and [-ATR] in others).

(21) Feature geometric representation of the rhyme sequence [em], which occurs in the robust form them σ qp root  - consonantal   + consonantal   + sonorant   + sonorant  eo ru cavity Oral Pharyngeal Nasal Oral ! ! ! !o stricture ! ! ! ![+continuant] ! ! ! ! articulator Dorsal Radical Soft Palate Labial egi ! ! terminal [- high] [-low] [-back] [+ATR] [+nasal]

syllable

I am following Dogil & Luschützky (1990:9) in assuming that all and only [+consonantal] segments are specified for stricture. Like Dogil & Luschützky (1990), I am treating stricture as a dependent of the Oral cavity node, but nothing rides on this assumption. My arguments would be equally consistent with approaches where stricture is associated with the root node rather than the Oral cavity node.8 Glottal features ([±voiced]) are not specified for [m], because voicing is not distinctive for unmarked [+consonantal, +sonorant] segments specified for Nasal features (cf. Dogil & Luschützky 1990:11).

8

See Kenstowicz (1994:480-90) for a detailed discussion of different proposals concerning the apparent interdependence between stricture and Oral place/articulator features.

CHAPTER 6

161

The feature geometric representations of diphthongs (20) and vowelconsonant sequences (21) are more complex than the feature geometric representations of monophthongs (19), because both diphthongs and VC sequences involve two sets of features, while monophthongs only involve one set of features. The difference in complexity between gracile and robust forms is most obvious for 1sg (me vs. I), 3sgM (he vs. him), and 1pl (we vs. us), where the rhyme of the gracile form is clearly monophthongal and the rhyme of the robust form is a diphthong or VC sequence both underlyingly and on the surface. The 3sgF forms she and her both contain a monophthongal rhyme, which suggests that the classification of she as gracile and her as robust is due to factors other than the complexity of the rhyme. Interestingly, there is some support for the potentially gracile status of her in the survey results. A number of speakers seem to be more ready to accept gracile final conjuncts when her occupies the initial conjunct position (e.g. her and we) than when any of the other robust forms appear in the initial conjunct.9 The featural representations of the 3pl forms they and them both involve two sets of features in the rhyme. They is thus closer to the robust forms in complexity than the other gracile pronoun forms. As this greater robustness of they would lead us to expect, the distribution of they differs from that of the other gracile pronoun forms, at least for some speakers (cf. Section 5.1 for more discussion). A080 S me O P

me me

A002 S me O P

9

he 91

initial conjunct she we

him 89 her 77 him her 91

us us

they

I

them I 82 them 90 I 73

initial conjunct her 56 us 67 they 80 I she 44 me 80 him her us them I 60 me 83 him 90 her 82 us them 89 I 89 he 60

him him him

him him him

final conjunct her 83 us her her

us us

final conjunct her us her her

us us

they 50 them 50 them 86 them

they 50 them 50 them them

As Liz Pearce (p.c.) points out, the sequence her and would generally be pronounced with an [r] between the two vowels. Research by Hay (2001) and Sudbury & Hay (2001) suggests that in Present-Day New Zealand English, the occurrence of linking [r] across word boundaries cannot be seen as evidence for the presence of a final /r/ in the underlying representation of the word preceding the [r], because the distribution of linking [r] across word boundaries strongly resembles the distribution of intrusive [r], whose occurrence is phonologically conditioned.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

162

A004 S me 80 he O P

me me

A007 S me O me P me

initial conjunct she we

him 91 her him her

us us

initial conjunct he 75 she 92 we him 75 her us him 88 her 82 us

they

I

them them

me 88 him me him

they 88 I 67 them I 75 them I 90

he 63

final conjunct she 50 we 57 they her 50 us 43 her us them her us them

final conjunct him 80 her us 89 them 60 him her us them him her us them

A016 initial conjunct S me 75 him 67 her 60 us 57 them 89 I we 43 O me him her us them I 89 P me him her us them I 88

him him

A074 S me O me P me

him him him

final conjunct her us her us her us

him

final conjunct her us 88 them 67

A100 S me O P

me me

him him him

initial conjunct her 78 us 75 them 75 I her us them I 78 her us them I 89

initial conjunct he 50 she 62 we him 50 him her us him her us

they

I

them I 90 them 86 I

him

him him

final conjunct her 88 us 63 them 67 her her 86

her her

us us

us us

them them

them 70 them them

them them

A101 initial conjunct final conjunct S me 80 he 91 she 70 we 75 them 60 I 90 him 71 her us 86 them 63 O me him her us them I 78 him her us them P me him her us them I 56 him her us them me 44

A closer look at the survey results for the speakers in question reveals that final they is particularly favoured when the initial conjunct is one of the other gracile pronoun forms (me, he, she, we). This use of they in final conjuncts of subject coordinates with the more gracile forms he, she, we in initial conjuncts can be captured by positing an additional constraint relating to the distribution of gracile and robust pronoun forms. I will call this constraint Relative Positional Coding 2.

CHAPTER 6

163

(22) Relative Positional Coding 2 (abbreviated as RPC 2) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then B must be more robust than A. While RPC 1 rules out they in final conjunct position, RPC 2 permits they in final conjuncts iff the initial conjunct of the coordinate is less robust than they. Since all the other gracile forms lack a segmentally complex rhyme, any coordinate involving one of the unambiguously gracile forms me, he, she, we followed by they will meet RPC 2, even though it violates RPC 1 (23). (23)

ConjP ei DP1 Conj’ he ei [gracile] Conj DP2 and they [gracile]

(but more robust than he) The relative weighting of RPC 1 and RPC 2 will determine how readily a speaker will use and accept they in final conjunct position. If RPC 1 and RPC 2 have the same weighting, or RPC 2 outweighs RPC 1, they will be readily tolerated in final conjuncts (at least in subject coordinates, where the nominative form also meets Pos-Case and Arg-Case demands). If RPC 1 outweighs RPC 2, the distribution of the 3pl nominative they and and the 3pl objective them will be fairly similar to the distribution of the remaining non-1sg nominative and objective forms, respectively. Despite its relative robustness compared to the remaining gracile forms, they still tends to be less likely to occur in asymmetrically c-commanded positions than them. Dogil & Luschützky (1990:9,18) argue that (unmarked) vowel segments need to be specified for place of articulation in the oral cavity, but not for stricture, voicing, or articulation in the nasal cavity. Consonants, on the other hand, have to be specified for at least place of articulation and stricture. The feature geometric representation of a diphthong can thus be argued to involve less branching at a non-terminal level than the feature geometric representation of a VC sequence: in (20), neither segment of the diphthong [ei] needed to be specified for stricture or Nasal cavity features, whereas in (21), the segment [m] in the VC sequence [em] needed to be specified for [-continuant] stricture and Nasal cavity features.

164

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

One advantage of defining robustness in terms of complexity/branching is that it allows us to account not only for the ordering preferences between monomorphemic monosyllables (determined by phonological complexity), but also for the ordering preferences between constituents of different morphological or syntactic complexity.10 Thus the RPC constraints predict that morphological markers such as the case and agreement morphemes discussed by Gil (1982) should be more likely to occur on final than initial constituents in a construction. What is more, the RPC constraints also predict that syntactically complex phrases will tend to be ordered after syntactically simple constituents. Possible evidence for the relevance of syntactic complexity to the order of conjuncts in coordinates comes from the ordering differences between items such as (24), which involve coordination between a pronoun and a proper noun, and items like (25), which involve coordination between a pronoun and a possessed noun phrase. (24) Please complete the sentence using two of the following words: them, Raymond, they [qu071] Just remember that none of this would have been possible without Leigh and Amanda. I think you owe a big thank you to _________ and _________ for all the work they put in. (25) Please complete the sentence using two of the following words: their skinhead friends, them, they [qu204] I suppose it was a bit stupid to leave Mark and Ted to look after the house. John had warned us about ________________________ and ________________________ from the start, but they all seemed so helpful and friendly. In both (24) and (25), the coordinate appears as the complement of a preposition. The task type requires respondents to choose not only the case form of the 3pl pronoun but also the order of the conjuncts (either 3pl & conjunction mate or conjunction mate & 3pl). Tables 6.2 and 6.3 present the survey results for (24) and (25), respectively. The results for these items are representative of a more general trend. 10 My assumptions tie in with Allan’s (1987:70) observation that the most general formal (structural) principle governing constituent order is the ‘complexity hierarchy’: structurally simpler constituents will precede structurally more complex constituents, all things being equal.

CHAPTER 6

165

they and Raymond them and Raymond

number of speakers 3 43

Raymond and they Raymond and them

0 42

total number of speakers who offered a relevant response

88

pronoun case form and conjunct order

Table 6.2. Survey results for item qu071(24)

they and their skinhead friends them and their skinhead friends

number of speakers 3 69

their skinhead friends and they their skinhead friends and them

0 18

total number of speakers who offered a relevant response

90

pronoun case form and conjunct order

Table 6.3. Survey results for item qu204 (25)

In keeping with the predictions of the case constraints, most speakers used the accusative case form them in both (24) and (25). But speakers differ in their ordering preferences for the two items: the majority of speakers opted for them and their skinhead friends in response to (25), whereas only about half of the speakers offered them and Raymond in (24), with the other half opting for Raymond and them. A more detailed look at individual speaker patterns (Table 6.4) reveals that speakers fall mainly into two groups where ordering preferences in these items are concerned. group (a): speakers who ordered the 3pl pronoun before both the proper noun Raymond and the possessed noun phrase their skinhead friends group (b): speakers who ordered the 3pl pronoun after the proper noun Raymond, but before the possessed noun phrase their skinhead friends A small group of speakers (group (c)) ordered the 3pl pronoun after both the proper noun and the possessed noun phrase, and even fewer ordered the pronoun before the proper noun, but after the possessed noun phrase (group (d)).

166

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

label used in text group (a)

response pattern

number of speakers

they/them and Raymond they/them and their skinhead friends

39

group (b)

Raymond and they/them they/them and their skinhead friends

31

group (c)

Raymond and they/them their skinhead friends and they/them

11

group (d)

they/them and Raymond their skinhead friends and they/them

7

total number of speakers who offered a relevant response for both items

88

Table 6.4. Speaker response patterns for items qu071 (24) and qu204 (25)

The results presented in Table 6.4 would appear to confirm my hypothesis about the influence of syntactic complexity on conjunct order in coordinates.11 Whatever our approach to syntactic structure and possessives, possessed noun phrases such as their skinhead friends will have a more complex syntactic representation than (unmodified) pronouns or proper nouns. What is more, the syntactic representation of strong pronouns (i.e. the type of pronoun we find in coordinates) will be similar in complexity to the syntactic representation of proper nouns (cf. Chapter 3 and Section 8.14). We would thus expect speakers to order pronouns before possessed noun phrases, but not necessarily before proper nouns. Since there is no clear difference between the syntactic complexity of strong pronouns and the syntactic complexity of proper nouns, the relative ordering of pronouns and proper nouns in a coordinate will depend on other factors. Speakers in group (a), who ordered the 3pl pronoun before the proper noun Raymond, are likely to be especially susceptible to the relative prosodic 11

This is not to say that syntactic complexity is the only factor influencing the relative order of pronouns and possessed noun phrases in a coordinate. As Liz Pearce (p.c.) points out, we would expect binding effects to influence ordering preferences in sentences like (25), where the pronoun (they/them) is coreferential with the possessor (their) in the possessed noun phrase. Since all the possessors in my survey items are coreferential with the personal pronoun tested in the item, I do not have any data on ordering preferences in coordinates involving a pronoun and a possessed noun phrase with a non-coreferential possessor (i). Nor do I have any data on ordering preferences in coordinates involving a pronoun and a simple noun phrase headed by a common noun (ii). It would be worth testing these in a future study. (i) they/them and our children vs. our children and they/them (ii) they/them and the park rangers vs. the park rangers and they/them

CHAPTER 6

167

complexity of syntactically simple constituents. The 3pl pronoun forms they and them are both monosyllabic, and might therefore be expected to be ordered before the polysyllabic proper noun Raymond.12 Speakers in group (b), who ordered the proper noun Raymond before the 3pl pronoun, would appear to be more sensitive to a phi-feature-related ordering constraint which stipulates that 3sg is ordered before 3pl (cf. Angermeyer & Singler 2003: 195). Such a constraint would predict that the 3sg noun Raymond should be ordered before the 3pl pronoun they/them. Since both they/ them and the possessed noun phrase their skinhead friends are 3pl, a phi-related ordering constraint will have no bearing on their relative order in a coordinate.13 At this stage, I do not have any explanation for the patterns exhibited by speakers in groups (c) and (d). Most of the speakers in group (c) show a strong general preference for ordering pronouns after non-pronominal conjunction mates, as do some of the speakers in group (d). The group (d) response pattern appears to be exceptional even for the speakers who exhibit the group (d) pattern in the two items discussed here. For the remaining items involving pronouns and non-pronominal conjunction mates, these speakers tend to pattern with either group (a) or group (c). While an approach based on phonological complexity provides some explanation for the position-related pronoun case differences in coordinates, other factors are likely to have contributed to the development of the gracile-robust distinction. The gracile pronoun forms differ from their robust counterparts not only in the complexity of the segments in the rhyme, but also in the quality of the vowels, the presence versus absence of an onset, and the type of consonant in the onset. It therefore seems quite likely that the emergence of the pronoun groupings for positional coding is the product of the interaction of a number of phonological factors. The influence of vowel quality and onset consonants is likely to have been particularly important for the respective classification of the 3sgF case alternants she and her as gracile and robust.

12

The conjunct order preferences exhibited by group (a) could also be influenced by the topic status of the 3ps pronouns (all of which have a discourse antecedent in the questionnaires). As discussed in Section 2.2.1, topics tend to occupy higher positions within a clause than constituents expressing new information. It seems possible that such information-structure based ordering preferences may also apply within coordinates. 13 All proper nouns used in the questionnaire are 3sg, and the possessed noun phrases in items testing the 3pl pronouns are all 3pl.

168

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

6.2.2 Vowel quality and the classification of pronoun forms As was already noted by Jespersen & Haislund (1949:262-4), the distributional similarities between me, he, she, and we (as well as ye and thee) may have their origin the fact that all of the forms end in the high front vowel [i]: After the old case-rules had been shaken off in different ways, instinctive feeling seized upon this similarity; and likeness in form has in part led to likeness in function. (Jespersen & Haislund 1949:262)

The preference for monosyllables with high front vowels to appear in initial rather than final conjuncts fits in with two of Allan’s (1987:72) formal ordering hierarchies (26)-(27). Since the nuclei of the robust forms I, him, her, us, them are arguably lower and/or more back than those of the corresponding gracile forms me, he, she, we, they (Table 6.5), Allan’s hierarchies would predict that gracile forms should be ordered before robust forms in a coordinate. (26) monosyllables with high vowels > monosyllables with low vowels (27) monosyllables with front vowels > monosyllables with back vowels 1sg 3sgM 3sgF 1pl 3pl wh

gracile mi hi i wi ei hu

robust ai hm h s em hum

Table 6.5. Phonological representations of gracile and robust pro noun forms

The difference in vowel quality is greatest for 1sg, 1pl, and 3sgF, where the nucleus of the robust form is clearly lower and more back than its gracile counterpart. The similarity between the nuclei of the 3pl forms they and them might have further reinforced the patterning of they with both gracile and robust forms. While the hierarchies in (26) and (27) correctly predict that the gracile forms should be ordered before robust forms in a coordinate, they do not in themselves offer any explanation for the correlation between vowel quality and conjunct position. We might speculate that there is some link between sonority or second formant (F2) frequency and relative order,14 but it is not entirely clear 14 See Selkirk (1984:112) for a sonority scale where high vowels have a lower sonority index than low vowels.

CHAPTER 6

169

how the frequency or sonority of segments would relate to lexical retrieval or linear/hierarchical order. Dogil and Luschützky (1990) propose that the sonority of a segment is determined by its articulatory/featural phonological complexity. In their feature geometrical approach, the sonority of a segment is inversely proportional to the number of nodes that have to be consulted on the way to the place node (28). Laryngeals are lowest in sonority because they are not specified for articulator nodes at all. (28) Universal sonority scale (Dogil and Luschützky 1990:18) I h

S h OC

I h

h PL

I h S S h th OC SP OC th th ST PL ST PL

Vowels > Approximants (highest sonority)

> Nasals

I th L S th SP OC th ST PL >

Obstruents

I f L

> Laryngeals (lowest sonority)

Key to abbreviations: I = Initiator, S = Supralaryngeal, L = Laryngeal, OC = Oral Cavity, SP = Soft Palate, PL = Place, ST = Stricture

Although Dogil & Luschützky’s (1990) hierarchy treats all vowels as equal, it predicts that highly sonorous segments should be easier to process than segments with lower sonority, because the feature geometric representation of a highly sonorous segment should be less complex (fewer branching nodes) than the representation of a less sonorous segment. In this approach, low vowels will only count as more sonorous than high vowels if low vowels can be shown to have less articulatory complexity than high vowels. However, if low vowels are less complex, then Bock’s (1982) observations on lexical retrieval would predict that pronoun forms with a low nucleus should be easier to retrieve than pronoun forms with a high nucleus. This is exactly the opposite of the pattern indicated by the survey results. A possible relation between vowel pitch and (physical) distance is suggested by Woodworth’s (1991) study of sound symbolism in the deictics of 26 languages. Woodworth (1991:280-5,291) observes that where there is a difference in vowel quality between proximal and distal deictics, the pitch of the vowel in the proximal deictic (e.g. English this, here) will tend to be higher than the pitch of the vowel in the distal deictic (e.g. English that, there). It would obviously be premature to draw any firm conclusions about the correlation between pitch/F2 frequency and distance from one study based on gram-

170

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

matical descriptions of phonetic data, but if Woodworth’s hypothesis turns out to have some general validity, then it might provide a partial explanation for the ordering hierarchies presented in (26) and (27). If low back vowels are associated with distance and high front vowels with proximity, then we might expect monosyllables with high front vowels to precede monosyllables with low back vowels in a coordinate, all things being equal. 6.2.3 Onsets and the classification of pronoun forms As Downing (1998:1-5) points out, onsetless syllables are often excluded from prosodic processes such as reduplication, tone association, and stress assignment. She argues that “the most plausible explanation for the exceptional prosody of onsetless syllables lies in their relative ill-formedness: optimal syllables have onsets” (1998:5). In order to capture the need for prosodic constituents to start with an optimal syllable, Downing proposes that the traditional ONSET constraint (syllables must have onsets) can conjoin with any alignment constraint that requires a prosodic constituent to be left-aligned with a syllable (1998:24f). All of the subject coordinates in my questionnaire items appear at the start of a clause. This means that the initial conjuncts of these subject coordinates will invariably form the start of a prosodic constituent and be susceptible to the onset requirement. Downing’s discussion focuses on instances where her conjoined ONSET-Alignment constraint can be satisfied through extraprosodicity of the onsetless syllable, but extraprosodicity is not the only strategy languages could adopt to ensure the satisfaction of the onset requirement for initial syllables in a prosodic constituent. In English, all conjuncts of a coordinate must bear a certain degree of stress (cf. Selkirk 1980:31). Extraprosodicity of an initial onsetless syllable is thus not an option if the initial conjunct is monosyllabic. Where the monosyllabic initial conjunct is a 1sg pronoun, we do however have another means of avoiding a violation of the onset requirement: All we need to do is opt for the 1sg objective form me rather than the onsetless 1sg nominative I. For 1pl pronouns, on the other hand, the selection of the nominative we will ensure compliance with the onset requirement.15 The preference for onsets in initial conjuncts is thus likely to have further reinforced the group-

15

As Liz Pearce (p.c.) notes, both I and us would also meet the onset requirement for initial syllables if they were realised with an initial glottal stop (i.e. as [ai] and [s]). In the absence of detailed phonetic data, I am unable to comment on whether speakers of New Zealand English readily adopt this strategy to avoid onsetless syllables in initial position. It would certainly be interesting to investigate this question in more detail.

CHAPTER 6

171

ing of me and we with gracile forms, and the grouping of I and us with robust forms. Sonority sequencing generalisations and observations by Everett & Everett (1984) suggest that an optimal syllable should not only have an onset, but that this onset should be maximally salient/obstruent, because this will ensure a steep sonority cline between onset and nucleus. For 3sgM and 3pl, and the wh-pronoun, the onsets of both case alternants are identical ([h] in he and him, and also who and whom; [] in they and them), so the nature of the onset could have had no bearing on the assignment of a particular case form to the gracile or robust series. For 3sgF, on the other hand, the nominative form she has a considerably more salient/obstruent onset than the objective form her (cf. Keating 1983). These onset differences may thus have contributed to the classification of she as gracile, and her as robust. 6.3

The Invariant Strong Form constraints Quite a number of speakers exhibit a strong preference for the robust objective forms him, her, us, and them. This suggests that the objective is the emerging invariant strong form for 3sgM, 3sgF, 1pl, and 3pl. As the tables below illustrate, most of the speakers concerned also exhibit a tendency towards invariant me. However, the trend towards an invariant strong 1sg form is generally less advanced than the development of invariant strong non-1sg forms. A062 S me O P

me me

A015 S me O P

me me

A056 S me O P

me me

initial conjunct final conjunct him 57 her 70 us 60 them 89 me 63 him 91 her us he 43 we 40 him her us them 90 me him her us him her us them me him her us

them them 90 them

initial conjunct him 89 her us 71 them 88 me 57 him I 43 him her us them me 89 him him her us them me 71 him

final conjunct her us us us

them them

initial conjunct him 89 her 56 us she 44 him her us him her us

final conjunct her us

them

them 90 I 63 them them

him

me 89 him me 88 him

her her

her her

us us

them

them them

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

172 A091 S me O me P me

initial conjunct him 75 her 60 us 80 them 89 I 58 him him her us them me 78 him him 91 her us them me 78 him

final conjunct her us her us her us

them them them

A118 S me O me P me

him him him

initial conjunct her 91 us 83 them 86 I him her us them me 80 him her us them I 57 him me 43

final conjunct her us her us her us

them them them

initial conjunct him 60 she 67 we 50 them 71 I 83 him us 50 him her us them me him him her us them me 86 him

final conjunct her us

them

A098 S me 60 I 40 O me P me

initial conjunct he 55 she 56 us 60 them 80 I 90 him him 45 her 44 we 40 him 90 her us them me 89 him him her us them me 56 him I 44

final conjunct her us

A025 S me

initial conjunct him 60 her 63 us 80 them

final conjunct her us

A116 S me O P

me me

them them

I 57 him me 43 me 60 him I 63 him

A020 S me O me P me

initial conjunct him 91 her 64 us 75 them him 92 her us them him her us them

A076 S me O me P me

O P

me me

him him

her 92 her 91

us us

her her

her her

her her

us us

us us

them them

them 88 them them

them 86

us us

them them

I him me 71 him me 71 him

final conjunct her us her us her us

them them them

initial conjunct him 63 her 73 us 88 them 63 I 89 him him her us them me him him her us them me 89 him

final conjunct her us her us her us

them them them

CHAPTER 6

173

A077 S me O me P me

initial conjunct final conjunct him 60 she 90 us 86 them 89 I him her us him her us them me him her us him her us them me 88 him 88 her 86 us

A087 S me O me P me

him him him

initial conjunct her 92 us 83 them her us them her us them

I him me 86 him me 88 him

final conjunct her us her us her us

them them them

them them them

For the following speakers, the trend towards invariant non-1sg forms is quite advanced, but there is little evidence for a similar trend in the distribution of 1sg forms. A067 S me

initial conjunct she 54 us 67 them her 46 we 33 him her us them him her us them

I 57 him me 43 I 80 him I 86 him

A095 S me O me P me

initial conjunct him 80 her 90 us him her us him her us

I 88 I I

A104 S me

he 83

O P

O P

me me

me me

him

initial conjunct she we 75 they 50 I them 50 him her us them I 83 him her 91 us them I 90

A060 initial conjunct S me 75 him 80 her 64 us O me 80 him her us P me him her us A072 S I 60 me 40 O me P me

them them them

them 80 I them I 90 them I 80

initial conjunct he 50 she 67 we 89 they 73 I him 50 him 90 her us them I him her us them I

him him him

final conjunct her us her her

us us

final conjunct her us her us her us

them 82 them them 88

them them them

final conjunct him 83 her us

them 80

him him

us us

them them

him him him

final conjunct her us her us her us

them them them

him

final conjunct her us

them

him him

her her

her her

us us

them them

174

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

A074 S me O me P me

him him him

initial conjunct her 78 us 75 them 75 I her us them I 78 her us them I 89

A083 S me O me P me

initial conjunct him 63 her us him her us him her us

them 92 I 90 them I 67 them I 63

him him him

final conjunct her us her us her us

them 70 them them

him him him

final conjunct her 89 us her us her us

them them them

The tendency towards objective pronoun forms in all strong pronoun contexts can be captured in a set of invariant strong form constraints, which stipulate that the morphological form of a strong pronoun must be invariant. The invariant strong form of the various pronouns is determined by parameter setting. For English pronouns it is the objective form. The survey results suggest that we need to postulate a separate invariant constraint for each strong personal pronoun (29)-(33), since the trend towards an invariant strong form may be more pronounced for some pronouns than for others (cf. also Denison 1996). For many speakers, 1sg INV appears to be weaker than any of the non1sg INV constraints, while 1pl INV is stronger than the other INV constraints. (29) Invariant strong 1sg form (abbreviated as 1sg INV) A strong 1sg pronoun will surface as me. (30) Invariant strong 3sgM form (abbreviated as 3sgM INV) A strong 3sgM pronoun will surface as him. (31) Invariant strong 3sgF form (abbreviated as 3sgF INV) A strong 3sgF pronoun will surface as her. (32) Invariant strong 1pl form (abbreviated as 1pl INV) A strong 1pl pronoun will surface as us. (33) Invariant strong 3pl form(abbreviated as 3pl INV) A strong 3pl pronoun will surface as them. As we will see in Chapter 9, the distribution of wh-forms indicates that the emerging invariant form for [+ human] wh-pronouns is the nominative who, whoever rather than the objective whom, whomever (34). (34) Invariant wh-form A personal wh-pronoun will surface as who(ever)

CHAPTER 6

175

The development of invariant strong pronoun forms can be seen as part of the divergence of the weak and strong pronoun series in English (cf. Chapter 10 for further discussion). At the moment, the weak pronouns are still phonetically similar to their strong counterparts, and are generally given the same orthographic representation.16 This similarity may contribute to the continuing influence of the three case constraints on strong pronoun forms. As the case constraints appear to be the most important factor favouring the occurrence of gracile nominatives, we might expect that the gracile nominative forms he, she, we, and they will drop out of the strong pronoun series once the phonetic similarity to the weak nominative forms is lost. The strong 3sgM, 3sgF, 1pl and 3pl pronouns will then surface only in the robust accusative forms him, her, us, and them. It seems likely that the strong 1sg form will eventually be invariant me, but the results of the present survey suggest that speakers may retain distinct variants in the strong 1sg paradigm for longer than in the paradigms of strong non-1sg forms. 6.4

The interaction of case and non-case constraints Tables 6.7-6.9 illustrate the predictions of both case and non-case constraints for the surface form of strong pronouns in different syntactic contexts. Since the interaction of case and non-case constraints is most evident in the distribution of pronoun forms in coordinates, the tableaux focus on the predicted distribution of pronoun case forms in coordinates. As can be seen from the tableaux, all applicable constraints call for the same 1sg form in initial conjuncts of object and prepositional complement coordinates, and for the same non-1sg forms in final conjuncts of object and prepositional complement coordinates. Pronoun case is thus predicted to be categorical in these positions. These predictions correspond to trends attested in the survey results (Table 6.6). 1sg - categorical pronoun choice and variation across speakers S variation variation O me variation P me variation

3sgM, 3sgF, 1pl, 3pl - categorical pronoun choice and variation across speakers S variation variation O variation him, her, us, them P variation him, her, us, them

Table 6.6. Contexts where we find categorical 1sg and non-1sg forms (across speakers), and contexts where we find variation (both between speakers and within individual speaker results) 16 But note the New Zealand slogan for water safety promotion (i), where the weak 2ps pronoun is orthographically incorporated into the preceding word.

(i) Have fun in the water and do wotcha oughta. (= do what you ought to)

176

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

In Chapter 7, I will explore ways in which the interaction of the different constraints can be modelled to yield the distribution patterns attested in the empirical survey, and compare the predictions of OT-based approaches to variation with an alternative constraint-weighting approach, where the cumulative weight of constraint violations determines the probability of occurrence of a particular variant.

initial conjunct of subject (S) I me final conjunct of subject I me initial conjunct of object (O) I me final conjunct of object I me initial conjunct of prepositional complement (P) I me final conjunct of prepositional complement I me

ArgCase

PosCase

DefCase

*

*

n/a n/a

*

*

n/a n/a

*

*

n/a n/a

*

*

RPC 1

RPC 2

*

*

*

*

*

*

n/a n/a

*

*

*

*

*

*

*

n/a n/a

*

*

n/a n/a

*

total number 1sg INV of constraint violations * 3 2

*

1 4

*

5 0

*

3 2

*

5 0

*

3 2

Table 6.7. Tableau illustrating 1sg pronoun case predictions for different syntactic and conjunct positions (if the status of object ConjPs differs from that of prepositional complements)

CHAPTER 6

initial S he, she, we him, her, us final S he, she, we him, her, us initial O he, she, we him, her, us final O he, she, we him, her, us initial P he, she, we him, her, us final P he, she, we him, her, us

ArgCase

PosCase

*

*

DefCase n/a n/a

*

*

n/a n/a

*

*

*

n/a n/a

*

n/a n/a

*

n/a n/a

*

*

RPC 1

RPC 2

*

*

*

*

*

*

*

*

*

*

*

*

INV *

n/a n/a

*

177 total number of violations 1 4

*

3 2

*

3 2

*

5 0

*

3 2

*

5 0

Table 6.8. Tableau illustrating 3sgM, 3sgF, and 1pl pronouns case predictions for different syntactic and conjunct positions (difference between objects and prepositional complements)

initial S they them final S they them initial O they them final O they them initial P they them final P they them

ArgCase

PosCase

*

*

DefCase n/a n/a

*

*

n/a n/a

*

*

n/a n/a

*

*

n/a n/a

*

n/a n/a

*

n/a n/a

*

*

RPC 1

RPC 2

*

*

*

*

*

total number of violations 1 4

*

2 2

*

3 2

*

4 0

*

3 2

*

4 0

*

*

*

3pl INV *

*

Table 6.9. Tableau illustrating which constraints predict which 3pl pronoun form in particular syntactic and conjunct positions (difference between objects and prepositional complements)

CHAPTER 7 MODELLING THE INTERACTION OF THE CONSTRAINTS

7.0

Introduction In the preceding chapters, I argued that the distribution of pronoun case forms in Present-Day English is largely determined by the constraints outlined in (1)-(6). (1)

Argument Case (abbreviated as Arg-Case in the tableaux) The overt case form of any structural argument of a predicate must comply with the structural linking between cases and arguments in the θ-structure. In Modern English, nominative case has the structural feature [- higher], which means it must be linked to the highest structural argument of any predicate. Objective case is restricted to arguments of [- N] predicates, but unspecified for any structural features, which means it must be linked to all other structural arguments of a verb or preposition.

(2)

Positional Case (abbreviated as Pos-Case in the tableaux) The overt case form of an argument noun phrase appearing as the specifier of an agreement-related functional head at Spell-Out must match the case/agreement features of this functional head, iff the position of the noun phrase at Spell-Out differs from its θposition. In Modern English, finite C combines with T to check nominative case on a noun phrase in [Spec,TP]; v requires objective case on its specifier.

(3)

(Positional) Default Case (abbreviated as Def-Case in the tableaux) The overt case form of any noun phrase not influenced by Pos-Case must match the default case of a language. In Modern English the default case is the objective.

CHAPTER 7

(4)

179

Relative Positional Coding 1 (abbreviated as RPC 1 in the tableaux) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then A must be gracile, and B must be robust. The set of gracile pronoun forms comprises: me, he, she, we, they The set of robust pronoun forms comprises: I, him, her, us, them

(5)

Relative Positional Coding 2 (abbreviated as RPC 2 in the tableaux) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then B must be more robust than A. In the set of gracile pronoun forms, they is more robust than me, he, she, we.

(6)

Invariant Strong Form constraints (abbreviated as INV) The morphological form of strong pronoun forms must be invariant in all contexts. There is a separate Invariant constraint for each pronoun (abbreviated as 1sg INV, 3sgM INV, 3sgF INV, 1pl INV, and 3pl INV in the tableaux). The invariant strong forms for the five pronouns are: me, him, her, us, them.

This chapter explores ways in which we can model the interaction of these constraints, and compares the predictions of the different models to the pronoun case patterns attested in the empirical survey. Since the survey yielded the most detailed information on the distribution of pronoun case forms in coordinates, I will focus on pronoun case in coordinates when illustrating and evaluating the different approaches. Section 7.1 discusses optimality-theoretic approaches to variation, and demonstrates that they cannot adequately account for the survey results. Section 7.2 presents an alternative, constraint-weighting approach, which is not only able to capture all of the most popular pronoun case patterns attested in the survey, but also correctly predicts the scarcity or non-occurrence of other patterns. 7.1

Optimality Theory (OT) Optimality Theory (OT) has been the most popular theory of constraint interaction in recent years. As I will show in Section 7.1.1, most of the funda-

180

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

mental principles of OT make it rather attractive for modelling variation, but the assumption of strict dominance of constraints severely limits the way in which variation can be captured in an OT-based approach. Section 7.1.2 evaluates OT-based approaches to variation that comply with the principle of strict dominance, and concludes that none of these can adequately model the pronoun case variation attested in coordinates. 7.1.1 Basic principles of OT Some of the core principles identified in Prince & Smolensky’s seminal 1993 paper on OT are listed in (7). (7)

Some basic principles of OT (cf. Prince & Smolensky 1993) (a) constraints are violable (b) constraints apply simultaneously (c) constraints are ranked

The three principles in (7) encapsulate the basic requirements for any successful constraint-based approach to variation. As we will see in the following sections, pronoun case variation occurs when constraints clash, that is, when some constraints call for one pronoun form, but others call for the alternative form. This competition between constraints and the resulting variation is most easily modelled in a framework where constraints are violable and apply simultaneously. Constraint rankings allow us to capture the relative importance of individual constraints in the grammars of different speakers, and thus provide a means of modelling variation between speakers. However, there is one central tenet of OT that poses problems for modelling variation, especially the kind of speaker-internal variation attested in the survey: (8) Strict domination of constraints as the constraint hierarchy is descended, each constraint acts to disqualify remaining competitors with absolute independence from all other constraints. A parse found wanting on one constraint has absolutely no hope of redeeming itself by faring well on any or even all lower ranking constraints (Prince & Smolensky 1993:78)

If one constraint is enough to rule out a candidate, then any traditional OT approach will predict that, for individual speakers, pronoun case should be categorical in all contexts, and should be determined by the highest-ranking constraint(s). This prediction is clearly not borne out by the survey results. The

CHAPTER 7

181

only attested patterns that are entirely predictable from the restrictions imposed by the highest applicable constraint(s) are (a) Pattern P, which is attested for both 1sg and non-1sg pronouns (b) Pattern a and Pattern e, which are attested only for non-1sg pronouns (c) Pattern A and Pattern G, which are attested only for 1sg Pattern P results if Arg-Case outranks all other constraints (Table 7.1). For both 1sg and non-1sg pronouns, Pattern P will also result if (a)

Pos-Case outranks all other constraints, and the next-highest constraint is either Arg-Case, Def-Case, or INV

(b)

Def-Case outranks all other constraints, and the next-highest constraint is either Arg-Case or Pos-Case

Pattern a results if INV outranks all other constraints (Table 7.2). For both 1sg and non-1sg pronouns, Pattern a will also result if Def-Case outranks all other constraints, and INV is the second-highest constraint. Pattern A results for 1sg if RPC 1 or RPC 2 outrank all other constraints (Table 7.3). Both the non-1sg Pattern e and the 1sg Pattern G result if Def-Case is the highest-ranking constraint, followed by RPC 1 (Table 7.4), provided we assume that object coordinates remain in [Spec,VP], where they will be influenced by Def-Case rather than Pos-Case. The following abbreviations are used in the pattern tables and tableaux throughout this chapter: NOM = nominative pronoun form; OBJ = objective pronoun form; S = coordinate that appears as the subject of a finite clause; O = object coordinate; P = coordinate that appears as the complement of a preposition; initial = initial conjunct; final = final conjunct. For a more detailed discussion of the various case patterns attested in the survey, see Chapter 5. In the tableaux, winning candidate(s) are marked with an arrow (=>), constraint violations are indicated with an asterisk (*), and fatal violations are marked with an exclamation mark preceding the asterisk (!*). Cells not involved in determining the winning candidate are shaded, and n/a indicates that a particular constraint is not applicable in the given environment.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

182

Pattern P NOM NOM OBJ OBJ OBJ OBJ

S O P

initial S =>

ArgCase

PosCase1

I, he, she, we, they me, him, her, us, them

!*

*

DefCase

RPC 12

RPC 2

INV

n/a

(* 1sg)

(* 1sg)

*

n/a

* (1sg ok)

* (1sg ok)

n/a

* (1sg ok)

* (1sg, 3pl ok)

final S =>

I, he, she, we, they me, him, her, us, them

!*

*

n/a

(* 1sg)

(* 1sg)

!*

* (n/a)

* (n/a)

(* 1sg)

(* 1sg)

(n/a)

(n/a)

* (1sg ok)

* (1sg ok)

* (n/a)

* (n/a)

* (1sg ok)

* (1sg, 3pl ok)

(n/a)

(n/a)

(* 1sg)

(* 1sg)

*

initial O & P I, he, she, we, they => me, him, her, us, them

*

final O & P I, he, she, we, they => me, him, her, us, them

!*

*

Table 7.1. Tableau illustrating the constraint ranking required to yield Pattern P in a standard OT approach

1

Pos-Case influences pronoun case in subject coordinates, and it does not influence pronoun case in prepositional complement coordinates (shown by n/a in the relevant rows). For object coordinates, Pos-Case applies if the coordinate is assumed to occupy [Spec,vP] at Spell-Out, but not if it is assumed to remain in [Spec,VP] (shown by (n/a) in the relevant rows). Def-Case influences pronoun case in prepositional complement coordinates, and in object coordinates that occupy their VP-internal base position. 2 1sg case forms violate RPC 1 and RPC 2 in different conjunct positions than their non-1sg counterparts, because the 1sg nominative I patterns with the non-1sg objective forms him, her, us, them for the purposes of Relative Positional Coding, and the 1sg objective form me patterns with the non-1sg nominatives he, she, we, they. RPC 2 permits the 3pl form they in final conjunct position, provided that the initial conjunct of the coordinate is gracile (i.e. me, he, she, or we).

CHAPTER 7

S O P

Pattern a OBJ OBJ OBJ OBJ OBJ OBJ

initial S he, she, we, they => him, her, us, them final S he, she, we, they => him, her, us, them initial O & P he, she, we, they => him, her, us, them final O & P he, she, we, they =>

183 1sg = Pattern A me I me I me I

S O P

non-1sg INV !*

PosCase

ArgCase

*

*

!*

!*

!*

him, her, us, them

DefCase n/a n/a n/a

*

*

n/a

* (n/a) (n/a)

*

* (n/a) (n/a)

* (n/a) (n/a)

*

* (n/a) (n/a)

RPC 1

RPC 2

*

*

*

* (3pl ok)

*

*

*

* (3pl ok)

Table 7.2. Tableau illustrating the constraint ranking required to yield Pattern a in a standard OT approach

initial S I => me final S => I me initial O & P I

RPC 1/ RPC 2 !*

PosCase

ArgCase

*

!* !*

=> me final O & P => I me

!*

1sg INV *

*

DefCase n/a n/a

*

n/a n/a

*

* * (n/a) (n/a)

*

* (n/a) (n/a)

*

* (n/a) (n/a)

*

* (n/a) (n/a)

*

Table 7.3. Tableau illustrating the constraint ranking required to yield Pattern A for 1sg 3 in a standard OT approach 3

The tableau contains only one column for RPC constraints, because RPC 1 and RPC 2 make the same predictions for the distribution of 1sg forms. That is, Pattern A results if either RPC 1 or RPC 2 outranks the remaining constraints.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

184

S O P

non-1sg Pattern e NOM OBJ OBJ OBJ OBJ OBJ

Definitial S Case => me, he, she, we, they n/a I, him, her, us, them n/a final S me, he, she, we, they n/a => I, him, her, us, them n/a initial O & P I, he, she, we, they !* => me, him, her, us, them final O & P I, he, she, we, they => me, him, her, us, them

S O P

RPC 1

RPC 2

!*

*

!*

* (3pl ok)

(* 1sg)

(* 1sg)

1sg Pattern G me I me me me me Arg-Case

* * (1sg ok) (1sg, 3pl ok) (* 1sg) (* 1sg)

INV

(* 1sg) (* 1sg) * (1sg ok) * (1sg ok) * (1sg ok) (* 1sg) *

* (1sg ok) * (1sg ok) !*

Pos-Case

(* 1sg) (* 1sg) * (1sg ok) * (1sg ok) * (1sg ok) (* 1sg)

n/a

*

n/a *

n/a

*

n/a

Table 7.4. Tableau illustrating the constraint ranking required to yield non-1sg Pattern e and 1sg Pattern G in a standard OT approach

7.1.2 OT-based approaches to variation In an approach that assumes strict domination of constraints, the only way to account for speaker-internal variation is to argue that speakers have simultaneous access to different constraint rankings, and thus different grammars. In existing OT approaches to variation, this simultaneous availability of different constraint rankings is represented either by constraint ties (cf. Anttila 1995; Nagy & Reynolds 1997; Büring 2001; Schmid 2001), or by overlaps between the probability distributions of adjacent constraints (Zubritskaya 1997; Boersma & Hayes 2001). The probability of occurrence of a particular variant is determined by the number of possible constraint rankings it wins out on and/or by the degree of overlap between the probability distributions of the constraints. Probabilistic OT models have the advantage of being able to capture not only the patterns shown above Tables 7.1-7.4, but also the very popular patterns given in Table 7.5. The most basic combinations of constraint rankings and ties that will yield the patterns in Table 7.5 are given below the table.

CHAPTER 7

185

S O P

non-1sg b-range x OBJ OBJ OBJ OBJ OBJ

S O P

non-1sg j-range NOM x OBJ OBJ OBJ OBJ

S O P

1sg D-range me I me x me x

S O P

1sg B-range me I me x me I

Table 7.5. Popular patterns that cannot be modelled in classical OT, but are predictable in a probabilistic approach using constraint ties or overlapping probability distributions

Non-1sg b-range patterns arise when Def-Case outranks all other constraints, followed by tied INV and RPC 1, or when there is a tie between Def-Case, INV, and RPC 1.4 Non-1sg j-range patterns arise when Def-Case outranks all other constraints, followed by tied/overlapping RPC 1 and Arg-Case or PosCase. 1sg D-range patterns arise when tied Def-Case and RPC 1 (or RPC 2) outrank all other constraints, or when there is a tie between RPC 1, RPC 2, Def-Case, and 1sg INV.5 1sg B-range patterns arise when tied RPC 1, RPC 2, Pos-Case, and 1sg INV outrank all other constraints, and object coordinates appear in [Spec,vP] at Spell-Out.6 Even OT models using constraint ties or overlapping probability distributions seem to be unable to account for commonly attested combinations of 1sg and non-1sg patterns, however, especially the ready co-occurrence of both non1sg b-range and non-1sg e-range patterns with 1sg A-range, B-range and Drange patterns. Any approach yielding the constraint ties and rankings discussed above would predict that only 1sg G-range patterns should be able to co-occur with non-1sg e-range patterns, and only 1sg G-range and D-range patterns should be able to co-occur with non-1sg b-range patterns. 1sg A-range and B-range patterns require a higher ranking of RPC 1 (or RPC 2) than is compatible with non-1sg b-range and e-range patterns. And the overlap between RPC 1 (or RPC 2) and Def-Case that yields a 1sg D-range pattern is incompatible with the constraint rankings required for a non-1sg e-range pattern.

4

The latter option will result in a b-range pattern provided we assume that the predicted probabilities apply to large data samples, where a 60%+ occurrence of a particular form will constitute a clear case preference (cf. Section 7.2). 5 The second combination of tied constraints will only result in a D-range pattern if we assume that a predicted probability of 0.6 or more constitutes a clear case preference. 6 Provided we assume that a predicted probability of 0.6 or more constitutes a clear case preference.

186

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Boersma & Hayes’ (2001) Gradual Learning Algorithm (GLA), which uses overlapping probability distributions to account for variation, is more successful at modelling the co-occurrence of individual 1sg and non-1sg patterns than an approach based on constraint ties. The unidirectional GLA option on Jäger’s (2002) program evolOT (http://www.ling.uni-potsdam.de/~jaeger/evolOT) produced suitable constraint rankings for two-pattern combinations involving any of the most common 1sg patterns and a non-1sg e-range pattern or a non-1sg b-range pattern (provided a predicted probability of 0.6 or more is assumed to constitute a clear case preference). However, neither the unidirectional GLA nor the additional bidirectional learning and evaluation options available in evolOT (cf. Jäger 2002) appear to be able to generate constraint rankings that could account for the co-occurrence of 1sg A-range, B-range, or D-range patterns with both non-1sg e-range and non-1sg b-range patterns in the speech of an individual. The inability of probabilistic OT models to account for the pattern combinations attested in the survey suggests that we might have to sacrifice the principle of strict dominance of constraints in order to account for the variation of pronoun case in Present-Day English. In the following section, I will present an alternative approach which focuses on the weighting rather than ranking of constraints, and assumes that the success of a particular candidate is determined by the cumulative weight of the constraint violations it incurs.7 Since constraint violations are additive, weaker constraints can ‘gang up’ on stronger constraints, and a candidate violating several weaker constraints may incur the same number of violation points as a candidate violating one or two stronger constraints (cf. Mohanan 1998; Guy 1997; Goldwater & Johnson 2003; Jäger n.d.; Jäger & Rosenbach 2003; and also Prince & Smolensky’s 1993:200-202 discussion of Harmonic grammar). 7.2

An alternative constraint-weighting approach The constraint-weighting approach I will outline in this section is strongly influenced by Mohanan’s (1998) strength specifications for competing constraints, and Guy’s (1997) adaptation of variable rule analysis. Both Guy (1997) and Mohanan (1998) reject the principle of strict dominance. They propose that each constraint has a certain weight in the grammar of a speaker, and 7

The Gradual Learning Algorithm (GLA) also assigns numerical values to constraints. However, constraint violations are not assumed to be additive, and the constraint values merely serve to highlight the relative distances between the different constraints, and thus provide an indication of the likelihood of ranking reversals resulting from overlapping probability distributions.

CHAPTER 7

187

candidates that violate a particular constraint incur violation points that correspond to the weighting of this constraint. Since violation points are cumulative, the success of a particular candidate depends on the combined weight of the constraints it violates rather than on its compatibility with the highest-ranked constraint(s). As a result, a form that violates a strong constraint may win out over a form that violates several weaker constraints. The difference between Guy’s (1997) and Mohanan’s (1998) proposals and standard OT is illustrated in Table 7.6. constraints weighting winner in probabilistic approach OT winner

x y z

A 5 *

B 4

C 3

D 2

E 1

*

*

*

*

Table 7.6. Predictions on the interaction of weighted constraints in a cumulative probabilistic approach and standard OT

In standard OT, z will be the winning candidate, because x and y both violate more highly ranked constraints than z. In a probabilistic approach, on the other hand, y will be the winning candidate, because it has only 4 violation points (from its violation of constraint B), whereas x has 5 violation points (from its violation of constraint A) and z has 6 violation points (from its violation of constraints C, D, and E). While Mohanan (1998) is primarily concerned with capturing the cumulativity of constraint violations, Guy’s (1997) main reason for proposing a variable rule based approach is his desire to provide a satisfactory account of variation. In Guy’s (1997) approach, variation arises from competition between the different constraints. The proportion of violation points incurred by individual forms determines the likelihood of their occurrence. The higher the number of violation points for a particular form in a particular context, the lower its likelihood of occurring in the position concerned. The degree of preference for one form over another thus depends on the relative weighting of competing constraints in a speaker’s grammar. As can be seen from (9), the approach adopted here shares the basic properties of the models proposed by Guy (1997) and Mohanan (1998).

188 (9)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Basic properties of the proposed constraint-weighting approach (a) all constraints have some weight for all speakers (b) each candidate incurs violation points corresponding to the weight of each constraint it violates (c) violation points are cumulative (i.e. the total number of violation points incurred by a candidate is determined by the combined weight of all the constraints violated by the candidate) (d) the success of a candidate is determined by the proportion of violation points it incurs (e) the candidate with the lowest number of violation points is predicted to occur more frequently than any other candidates (f) when all constraints call for the same candidate (i.e. if one candidate does not incur any violation points at all), the form in question is predicted to be categorical (or as close to categorical as it can get - there may be performance factors playing into it)8

The probability of occurrence for a particular candidate can be calculated using the formula given in (10). (10) 1 —

violation points incurred ----------------total number of violation points

For example, if the total number of possible violation points that can be incurred in a particular context is 28, and candidate A incurs 10 violation points, then the predicted probability of occurrence for candidate A is 0.643 (= 1(10/28)), or 64.3%. The probabilities predicted on the basis of this calculation are most likely to be accurate for large data samples, and would not be expected to directly match the percentages calculated from the small token numbers in this study. Since a 60% selection of a particular case form in a given context constitutes a clear preference when token numbers are large, any candidate with a predicted probability of 0.6 is treated as a clearly favoured form 8

I have decided to focus only on major constraints in my analysis of English pronoun case, but some marginal occurrences of I in initial conjuncts of object and prepositional complement coordinates may be due to influences from other, less important factors, such as conjunction mate and task type. The same goes for instances of non-1sg nominatives in final conjuncts of object and prepositional complement coordinates.

CHAPTER 7

189

in the tableaux. An absence of a clear case preference is assumed to occur only when both of the candidates have a predicted probability of less than 0.6. 7.2.1 Cumulativity and the most common combinations of 1sg and non-1sg patterns A central characteristic of the proposed approach is the cumulativity of constraint violations (9c). Jäger & Rosenbach (2003) argue that we need to distinguish the two types of cumulativity defined in (11) and (12).9 As I will demonstrate below, we need to permit both types of cumulativity if we want to capture the distribution of English pronoun case forms in coordinates. (11) Ganging-up cumulativity A candidate which violates a number of lower-ranked constraints may lose out to a candidate that violates only one higher-ranked constraint, if the cumulative weights of the lower-ranked constraints outweigh the weight of the higher-ranked constraint. (12) Counting cumulativity Each constraint violation needs to be counted separately. If a candidate violates a particular constraint twice, it will incur double the violation points for this constraint. Ganging-up cumulativity allows us to model combinations of 1sg and non-1sg patterns like those exhibited by speakers A003, A023, and A040. Speaker A003 combines a 1sg A-range pattern with e-range patterns for 3sgF and 3sgM, and b-range patterns for 1pl and 3pl.10 Speaker A023 combines a 1sg B-range pattern with a 3sgF a-range pattern, a 3pl e-range pattern, and brange patterns for 3sgM and 1pl. And Speaker A040 combines a 1sg D-range pattern with a 3sgF b-range pattern, a 1pl e-range pattern, and a-range patterns for 3sgM and 3pl. A003 S me O P 9

me me

he 92 him him 90

initial conjunct she 82 us 71 her her

us us

they 50 them 50 them them

I 90

final conjunct him 83 her us 88

them 90

I 75 I 82

him him

them them

her her

us 89 us

I would like to thank Joan Bresnan for alerting me to Jäger and Rosenbach’s work on cumulativity effects in constraint evaluation. 10 With 71% us in initial conjuncts of subject coordinates, the distribution of 1pl forms almost corresponds to an a-range pattern.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

190

A040 initial conjunct S me him 80 she 64 we 86 them 91 I 89 O me 88 him her us them I 57 me 43 P me him her 91 us them I 57 me 43 A023 S me O P

me me

he 70 him him

initial conjunct her 83 us 60 we 40 her 92 us her us

final conjunct him 88 her 88 us 88 them him her us them him

they 92 I 93

him

them them

him him

I 67 I 75

her 86 us

final conjunct her us her her

us us

them

them them them

As discussed in Section 7.1.2, these pattern combinations are problematic for all of the OT-based models to variation that assume a strict dominance of constraints, because neither constraint ties nor overlaps can yield constraint rankings that will account for the co-occurrence of both e-range and b-range pattern with 1sg A-range, B-range, or D-range patterns. In the cumulative constraint-weighting approach proposed here, on the other hand, a-range, e-range, and b-range patterns are expected to cooccur both with each other and with all of the popular 1sg patterns. As can be seen from Table 7.7, non-1sg a-range, b-range, e-range patterns all result when the relevant Invariant Strong Form (INV) constraint outweighs all other constraints. The weight of the INV constraint relative to the combined weight of the remaining constraints determines which of the three patterns will be instantiated for a particular non-1sg pronoun. A non-1sg a-range pattern results when the weight of the relevant INV constraint (3sgM INV and 3pl INV in this example) exceeds the combined weight of the remaining constraints. A non-1sg b-range pattern results when the weight of the relevant INV constraint (3sgF INV in this example) is roughly equivalent to the combined weight of the remaining constraints that influence these pronouns. A non-1sg e-range pattern results when the relevant INV constraint (1pl INV in this example) outweighs any other relevant constraint, but the combined weight of the remaining constraints is greater than the weight of the 3sg INV constraint by itself. The three patterns are readily compatible with each other, because each pronoun has its own INV constraint, whose weight is independent of the weightings of other INV constraints.

CHAPTER 7

3sgM/3pl 3sgF

initial S he, they

INV (40)

INV (20)

1pl PosINV RPC 1 Case (13) (7) (6)

*

n/a

n/a

=> him, them

n/a

n/a

*

n/a

=> she

n/a

=> her

n/a

=> we

n/a

n/a

us

n/a

n/a

*

n/a

n/a

n/a

n/a

*

n/a

n/a

191

RPC 2 (4)

Arg- DefCase Case (3) (2) n/a

* *

* *

* *

* *

* *

*

*

*

combined weight of violations 40

n/a

20

n/a

20

n/a

20

n/a

13

n/a

20

n/a

51 (3pl: 47)

n/a

9

n/a

31

final S he, they => him, them she

n/a

=> her

n/a

we

n/a

n/a

=> us

n/a

n/a

*

n/a

n/a

n/a

n/a

*

n/a

* *

n/a *

* (3pl ok) *

* *

* *

*

n/a

9

n/a

24

*

n/a

9

*

*

* *

initial O & P he, they => him, them she

n/a

=> her

n/a

we

n/a

n/a

=> us

n/a

n/a

*

n/a

n/a

n/a

n/a

*

n/a

n/a

n/a *

n/a

*

n/a *

*

n/a

*

*

*

*

*

*

n/a n/a

*

n/a

* (3pl ok)

45 11 25 11 18 11

final O & P he, they => him, them she

n/a

=> her

n/a

we

n/a

n/a

=> us

n/a

n/a

* *

n/a *

*

*

n/a n/a

0 *

*

*

36

*

*

*

29

n/a *

n/a n/a

56 (3pl: 52)

0 0

Table 7.7. Tableau illustrating the constraint weightings for A040 and their implications for the distribution of non-1sg pronoun forms (a-range patterns for 3sgM and 3pl; b-range pattern for 3sgF; e-range pattern for 1pl)

192

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Since it is the cumulative weight of the remaining constraints rather than their relative ranking that determines whether the distribution of non-1sg case forms follows an a-range, b-range, or e-range pattern, the approach proposed here correctly predicts that all three patterns should co-occur with any of the common 1sg patterns (see Quinn 2005 for detailed tableaux). A 1sg A-range pattern will occur when the combined weight of the RPC constraints is higher than the combined weight of Arg-Case, Def-Case, and 1sg INV, and preferably also the combined weight of Arg-Case and Pos-Case. A 1sg B-range pattern will result when object coordinates appear in [Spec,vP] at Spell-Out, and the combined weight of RPC 1 and RPC 2 is equal to the combined weight of Pos-Case, Arg-Case, and 1sg INV, but outweighs the combination of Arg-Case, Def-Case, and 1sg INV. A 1sg D-range pattern arises when at least one of the RPC constraints outweighs the individual case constraints, and the combined weight of the two RPC constraints is slightly less than the combined weight of 1sg INV, Arg-Case, and Def-Case. 1sg G-range patterns resemble non-1sg e-range patterns in that they result when the relevant INV constraint (in this case 1sg INV) outweighs all other individual constraints, but is weaker than the combined weight of the remaining constraints. For this reason, 1sg G-range patterns, are also compatible with non-1sg j-range patterns, which require that the combination of Arg-Case and Pos-Case has the same weight as the combination of the RPC constraints and the relevant non-1sg INV constraint. Since final they violates RPC 1 but not RPC 2, the weight of RPC 2 may be reasonably high and still yield a 3pl j-range pattern, provided RPC 1 and 3pl INV are comparatively weak. For 3sg and 1pl, on the other hand, a j-range pattern will result only if both RPC 1 and RPC 2 are fairly weak.3pl j-range patterns might therefore be expected to cooccur with 1sg and non-1sg patterns that require a strong RPC constraint (e.g. 1sg A-range, B-range, and D-range), whereas 3sg and 1pl j-range patterns should be unable to co-occur with such 1sg and non-1sg patterns. The absence of any clear empirical evidence for the expected trend would seem to suggest that RPC 1 generally outweighs RPC 2, and thus ensures that 3pl j-range patterns are most likely to co-occur with 1sg and non-1sg patterns that are compatible with weak RPC constraints. While ganging-up cumulativity is sufficient to model the co-occurrence of the popular 1sg and non-1sg patterns, we also need to assume counting cumulativity to capture the fact that many speakers quite readily accept both she/he/they & I and her/him/them & I in subject coordinates (cf. Section 5.1.1).

CHAPTER 7

193

A030 has a b-range pattern for all non-1sg pronouns, and a D-range pattern for 1sg (Table 7.8). A closer look at individual questionnaire items further reveals that A030 exhibits variation between she & I and her & I in subject coordinates. The constraint weightings required to yield this combination of 1sg D-range and non-1sg b-range patterns are illustrated in Table 7.9. A030 S me O P

me me

initial conjunct him 56 she 56 we 60 he 44 her 44 us 40 him her us him her us

they 55 I them 45 them me 67 them me 71

final conjunct him 89 her us

them 86

him 89 her him her 88

them them

us us

Table 7.8. Summary of the coordinate results for A030

subject coordinates => she & I => her & I she & me her & me

3sgF INV (20) * *

1sg INV (8) * *

RPC 1 (7)

PosCase (6)

RPC 2 (4)

ArgCase (3)

* * **

* * **

* * *

* * **

combined weight of constraint violations 28 28 40 36

Table 7.9. Tableau illustrating the constraint weightings for A030 and the resulting pronoun case predictions for subject coordinates with 3sgF in initial conjunct position and 1sg in final conjunct position, if we assume counting cumulativity

As can be seen from Table 7.9, coordinates containing more than one pronoun may incur multiple violations of the Case constraints as well as RPC 1, because these constraints have to be satisfied by each of the pronouns in the coordinate. She & I and her & I are correctly predicted to be joint winners if we assume that constraint violations are cumulative, and each violation of the different constraints needs to be counted separately. If we did not count each violation of the different constraints separately, her & me would only incur 20 violation points. An approach that excludes counting cumulativity would thus incorrectly predict that her & me should be the winning candidate. The two types of cumulativity that characterise the proposed approach thus allow us to account for various aspects of the survey results, many of which are difficult to capture in OT-based approaches to variation. The question is whether this greater descriptive adequacy comes at the cost of a loss of predictive power. Any approach that allows cumulative constraint violations, is necessarily less constrained and predictive than an approach like OT, which

194

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

assumes a strict dominance of constraints. However, the weight-based model proposed here does still rule out certain distribution patterns, namely (a)

distribution patterns where pronoun forms favoured in particular cells violate all of the proposed constraints

(b)

distribution patterns where the constraint weighting suggested by the pronoun case choice in one cell (e.g. initial conjuncts of subject coordinates) differs from the constraint weighting required to yield the results attested for another cell (e.g. final conjuncts of object coordinates)

A comparison of predictions and survey data reveals that the types of patterns described in (a) and (b) are indeed unattested or very rare. 7.2.2 Patterns ruled out by the proposed approach The non-1sg patterns given in Table 7.10 are ruled out because we find clear variation between nominatives and objective forms in final conjuncts of object and/or prepositional complement coordinates, where all constraints call for the objective non-1sg forms him, her, us, them. n-range S NOM x O x x P OBJ OBJ attested for: 3sgM (A008)

o-range S NOM x O x x P x OBJ attested for: 3pl (A042)

p-range S NOM x O x OBJ P OBJ x attested for: 3sgM (A089)

q-range S NOM x O x OBJ P x x attested for: 3sgF (A042)

r-range S NOM x O x x P x x attested for: 3sgM (A066 & A119)

O-range S NOM NOM O OBJ x P OBJ OBJ attested for: 3pl (A066)

u-range S NOM NOM O x x P OBJ OBJ attested for: 3sgF (A066) Table 7.10. Non-1sg patterns where case variation occurs in final conjuncts of non-subject ConjPs

CHAPTER 7

195

Similarly, the 1sg patterns presented in Table 7.11 are ruled out because we find variation between I and me in initial conjuncts of object and/or prepositional complement coordinates (ConjPs), where all of the constraints call for me.

S O P

1sg L-range (attested for A037) me I me I x I S O P

S O P

1sg M-range (attested for A006) x I me I x I

1sg N-range (attested for A012) x I x I me I

Table 7.11. 1sg patterns where case variation occurs in initial conjuncts of non-subject ConjPs

The constraint-weighting model further predicts that the robust 1sg nominative I should be more likely to occur in final conjuncts of subject coordinates than in initial conjuncts of subject coordinates or in final conjuncts of object and prepositional complement coordinates. I violates only the 1sg INV constraint in final conjuncts of subject coordinates, but it violates both the RPC constraints and the 1sg INV constraint in initial conjuncts of subject coordinates, and the relevant Case constraints as well as the 1sg INV constraint in final conjuncts of object and prepositional complement coordinates. If we assume that all (applicable) constraints have some weight for every speaker, then only speakers with strong RPC constraints would be predicted to exhibit a clear preference for I in final conjuncts object and prepositional complement coordinates. As illustrated in Table 7.12, such a weighting of the RPC constraints will invariably yield I in final conjuncts of subject coordinates.

final S => I me final O & P => I me

RPC 1 (10)

RPC 2 (8)

PosCase (5)

ArgCase (4)

1sg INV (2) *

*

DefCase (3) n/a n/a

combined weight of violations 2 27

*

*

*

*

*

n/a n/a

*

*

*

9 18

Table 7.12. Constraint weighting that will yield a clear preference for I in final conjuncts of non-subject coordinates, and its predictions for final conjuncts of subject coordinates

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

196

As a consequence, the constraint-weighting approach proposed here will rule out 1sg patterns such as the Q-range pattern (cf. Table 7.13), where a clear preference for I in final conjuncts of object and prepositional complement coordinates contrasts with variation between me and I in final conjuncts of subject coordinates. 1sg Q-range (attested for A007 & A067) S me x O me I P me I S O P

1sg I-range (attested for A075) x I me I me x

S O P

1sg O-range (attested for A044) I I me x me me

Table 7.13. 1sg patterns with incompatibility between case trends in subject coordinates and final conjuncts of objects or prepositional complements

The 1sg I-range and O-range patterns in Table 7.13 are ruled out because the constraint weightings required to yield 1sg nominative forms in initial conjuncts of subject coordinates are incompatible with the distribution of 1sg forms found in final conjuncts of object and prepositional complement coordinates. In both the I-range and the O-range pattern, there is a difference between final conjuncts of object and prepositional complement coordinates. This indicates that object coordinates are analysed as having raised to [Spec,vP], and are therefore influenced by Pos-Case rather than Def-Case. With the I-range pattern, the clear preference for I in final conjuncts of object coordinates suggests that the combination of RPC 1 and RPC 2 outweighs the combined weight of Pos-Case, Arg-Case and 1sg INV (cf. Table 7.12). The variation in initial conjuncts of subject coordinates, on the other hand, indicates that the combined weight of Pos-Case and Arg-Case is roughly equal to the combined weight of RPC 1, RPC 2, and 1sg INV (Table 7.14).

initial S => I => me

PosCase (6) *

RPC 1 (5) *

ArgCase (4) *

RPC 2 (3) *

1sg INV (2) *

combined weight of violations 10 10

Table 7.14. Constraint weighting that will yield variation between I and me in initial conjuncts of subject coordinates

CHAPTER 7

197

With the O-range pattern, the variation in final conjuncts of object coordinates indicates that the combined weight of RPC 1 and RPC 2 is roughly equal to the combined weight of Pos-Case, Arg-Case, and 1sg INV. The clear preference for I in initial conjuncts of subject coordinates, on the other hand, suggests that the combination of Pos-Case and Arg-Case outweighs the combined weight of RPC 1, RPC 2, and 1sg INV. The 1sg patterns in Table 7.15 are ruled out because the heavy weighting of 1sg INV that will yield clear variation between me and I in final conjuncts of subject coordinates, would predict that me should be clearly favoured in final conjuncts of object and prepositional complement coordinates. Similarly, the non-1sg patterns given in Table 7.16 are ruled out, because the nominative non-1sg forms found in final conjuncts of subject coordinates are incompatible with the heavy weighting of the INV constraint that is required to yield objective non-1sg forms in initial conjuncts of subject coordinates. 1sg R-range (attested for A113 & A025) S me x O me x P me x

S O P

1sg S-range (attested for A015) me x me me me x

Table 7.15. 1sg patterns with case variation in final conjuncts of both subject and non-1subject ConjPs non-1sg K-range S x NOM O OBJ OBJ P OBJ OBJ attested for: 3pl (A034 & A078)

non-1sg T-range S OBJ x O OBJ OBJ P OBJ OBJ attested for: 3sgF (A008, A092) 3pl (A016, A070, A074)

Table 7.16. Non-1sg patterns with incompatible case trends in initial and final conjuncts of coordinates in subject position

The non-1sg distribution patterns presented in Table 7.17 are ruled out because the heavy weighting of the INV constraint required to yield objective non-1sg forms in initial conjuncts of subject coordinates is incompatible with clear variation in initial conjuncts of object and/or prepositional complement coordinates. Most of the patterns in Table 7.17 also violate other predictions of the proposed approach. The c-range and y-range patterns contain variation between nominative and objective non-1sg forms in final conjuncts of object coordinates, where all constraints call for the objective forms him, her, us, them.

198

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The c-range, w-range, x-range, and y-range patterns are also ruled out because they exhibit clear variation between nominative and objective non-1sg forms in both initial and final conjuncts of subject coordinates. c-range S x x O OBJ x P x OBJ attested for: 3sgM (A022 & A029) 3sgF (A070) 3pl (A106) d-range S x OBJ O x OBJ P x OBJ attested for: 3sgF (A106)

w-range S x x O x OBJ P OBJ OBJ attested for: 3sgM (A013)

x-range S x x O OBJ OBJ P x OBJ attested for: 3sgF (A034)

y-range S x x O x x P OBJ OBJ attested for: 3sgM (A042)

Table 7.17. Non-1sg patterns with incompatible case trends in initial conjuncts of subject and non-subject coordinates

The occurrence of clear variation between nominative and objective forms in both initial and final conjuncts of subject coordinates is also a characteristic of the v-range pattern (Table 7.18), which is the only comparatively well-attested non-1sg pattern that fails to match the predictions of the proposed approach. As can be seen from Table 7.19, the constraint weighting required for clear variation between nominative and objective non-1sg forms in initial conjuncts of subject coordinates will always predict a preference for objective non-1sg forms in final conjuncts of subject coordinates. In a model based on constraint-ties or overlaps, such as Boersma & Hayes’ (2001) Stochastic OT, v-range patterns will result when overlapping Arg-Case and INV outrank all other constraints. We could achieve a similar effect by assuming that both RPC 1 and RPC 2 have zero weight for speakers with these patterns, while the combined weight of Arg-Case and Pos-Case is equal to the weight of the relevant INV constraint. However, such a step would clash with our basic assumption that every constraint counts and has some weight for each speaker. What is more, it would lead us to predict that non-1sg

CHAPTER 7

199

v-range patterns can only co-occur with 1sg patterns not influenced by the RPC constraints. This prediction is not borne out by the survey results: most of the speakers listed in Table 7.18 exhibit a 1sg A-range, B-range, or D-range pattern, and all of these 1sg patterns require a reasonable weighting of RPC 1 and/or RPC 2. v-range S x x O OBJ OBJ P OBJ OBJ attested for: 3sgM (A034, A070, A105) 3sgF (A022) 1pl (A016, A096, A111, A039) 3pl (A008, A009, A045, A073, A075, A086, A101, A105, A111) Table 7.18. The pronouns and speakers exhibiting a v-range pattern

initial S => he, she, we, they => him, her, us, them final S he, she, we, they => him, her, us, them

non-1sg INV (15) *

PosCase (6)

ArgCase (4)

RPC 1 (3)

RPC 2 (2)

*

*

*

*

*

* (3pl ok)

* *

*

combined weight of violations 15 15 20 (3pl: 18) 10

Table 7.19. Constraint weighting that will yield variation between nominative and objective non-1sg forms in initial conjuncts of subject coordinates, and its predictions for final conjuncts of subject coordinates

Interestingly, virtually all of the speakers that have a v-range pattern for one or more non-1sg pronouns, do exhibit some difference in case preferences between initial and final conjuncts of subject coordinates for the pronoun(s) concerned. What is more, the v-range pattern is more commonly attested for 3pl than for other non-1sg pronouns. This would be predicted by my approach, because only RPC 1 requires the 3pl objective form them in final conjuncts of subject coordinates. RPC 2 predicts that either they or them should be fine in this position. For the remaining non-1sg pronouns, on the other hand, both

200

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

RPC 1 and RPC 2 demand the objective rather than nominative form. The attested v-range patterns can thus be seen to provide additional evidence supporting the approach proposed here, even though they would at first glance appear to violate its predictions. 7.3

Summary As I have tried to demonstrate in this chapter, the interaction of the constraints on the distribution of pronoun forms in coordinates is most readily modelled in a constraint-weighting approach which rejects the strict dominance of constraints that forms the cornerstone of OT-based models of variation. A constraint-based approach is particularly suitable for dealing with interface phenomena such as morphological case, because it allows us to capture the interaction between the various syntactic, semantic, morphological, and phonological conditioning factors (cf. Sigursson 2003 and McFadden 2004). While an OT-based approach could successfully account for nearly all of the most common 1sg and non-1sg pattern ranges attested in the empirical survey, many of the readily occurring combinations of 1sg and non-1sg patterns appear to defy an optimality theoretic analysis. The proposed approach may be less constrained and predictive than OT models that assume strict constraint domination, but it is able to accommodate all the most popular combinations of 1sg and non-1sg patterns attested in the survey and correctly rules out a range of pronoun case patterns that are unattested or rare in the empirical data.

CHAPTER 8 THE DISTRIBUTION OF PERSONAL PRONOUN FORMS IN OTHER STRONG PRONOUN CONTEXTS

8.0

Introduction Chapters 8 and 9 look at pronoun case in strong pronoun contexts other than those tested in the survey. As mentioned at the start of Chapter 2, formal approaches to case tend to treat pronoun case forms as overt manifestations of an underlying case system. Such a view of pronoun case would lead us to predict that pronoun case variation will arise only from the interaction of different types of case assignment and/or from parametric differences in the syntactic properties of the construction concerned. Chapters 8 and 9 examine some of the existing evidence for pronoun case variation in contexts other than coordinates, pronoun-NP constructions, it-clefts, and than-comparatives. As I will demonstrate, the variation reported in the constructions featured in Chapters 8 & 9 provides further support for the interaction of Argument Case (1), Positional Case (2), and Default Case (3) with Relative Positional Coding (4)-(5) and the Invariant Strong Form constraints (6). (1) Argument Case (abbreviated as Arg-Case) The overt case form of any structural argument of a predicate must comply with the structural linking between cases and arguments in the θ-structure. In Modern English, nominative case has the structural feature [- higher], which means that it must be linked to the highest structural argument of a predicate. Both objective and genitive case are unspecified for structural features, but objective case is restricted to arguments of [- N] predicates, while genitive case is limited to arguments of [+ N] predicates. This means that objective case must be linked to all remaining structural arguments of a verb or preposition, while genitive case must be linked to the remaining structural argument of a noun.

202

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(2) Positional Case (abbreviated as Pos-Case) The overt case form of an argument noun phrase appearing as the specifier of an agreement-related functional head at Spell-Out must match the case/agreement features of this functional head, iff the position of the noun phrase at Spell-Out differs from its θposition. In Modern English the following functional heads are involved in the checking of Positional Case: (a) finite C combines with T to check nominative case on a noun phrase in [Spec,TP] (b) non-finite C filled by the complementizer for combines with T to check objective case on a noun phrase in [Spec,TP] (c) v checks objective case on its specifier (d) D checks genitive case on its specifier (3) (Positional) Default Case (abbreviated as Def-Case) The overt case form of any noun phrase not influenced by Pos-Case must match the default case of a language. In Modern English, the default case is the objective case. (4) Relative Positional Coding 1 (abbreviated as RPC 1) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then A must be gracile, and B must be robust. The set of gracile pronoun forms comprises: me, he, she, we, they, who The set of robust pronoun forms comprises: I, him, her, us, them, whom (5) Relative Positional Coding 2 (abbreviated as RPC 2) If a constituent A asymmetrically c-commands a constituent B in a given syntactic construction, then B must be more robust than A. In the set of gracile pronoun forms, they is more robust than me, he, she, we.

CHAPTER 8

203

(6) Invariant Strong Form (abbreviated as INV) The morphological form of strong pronoun forms must be invariant in all contexts. There is a separate Invariant constraint for each pronoun. The invariant personal pronoun forms are: me, him, her, us, them The invariant wh-forms are: who, whoever Denison (1993:22) suggests that the main change in the English case system during the Modern English period “has been a continued shift towards objective as unmarked form, most noticeable in such patterns as It’s me and taller than me […], where subjective I would have been normal at earlier times” (cf. also Denison 1996). This view appears to be shared by many, and similar statements can be found in Harris (1981), Emonds (1985:237f; 1986:93-96,121), Kjellmer (1986), and Wales (1996:88f,93,107). If these observations are correct, we would expect the Default Case constraint to be very strong in PresentDay English. However, a closer look at the available evidence suggests that the distribution of pronoun case forms in Present-Day English is characterised by more than just a uniform trend towards objective case. Although we do find objective case forms in positions previously reserved for nominatives, we also find nominative pronoun forms (especially who and I) in environments traditionally considered to require objective case (cf. Jespersen & Haislund 1949: 274). As we will see in the remainder of this chapter and in Chapter 9, the case form of both personal pronouns and wh-pronouns tends to correlate with the structural position of the pronoun at Spell-Out, its status in the argument hierarchy of a predicate, and the presence versus absence of agreement. Case variation typically occurs when a pronoun is in some way separated from an appropriate case-agreement head, and/or where the position of the pronoun is structurally ambiguous. Competition between the three case constraints, and the availability of alternative structural analyses can account for some of the case variation reported in existing studies. However, many of the trends discussed in Chapters 8 and 9 are difficult to account for in a purely case-based approach. The distribution of case forms in the various strong pronoun contexts considered here provides further evidence that strong pronoun forms not only identify the structural case of a pronoun in Modern English, but also code its position within a syntactic construction, and identify its morphosyntactic status as a strong pronoun. Thus, strong pronouns in asymmetrically c-commanding positions tend to surface in

204

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

the forms who, me, he, she, we, they, while strong pronouns in asymmetrically c-commanded positions tend to surface in the forms whom, I, him, her, us, them, as predicted by the Relative Positional Coding constraints (4)-(5). In keeping with the predictions of RPC 2 (5), they is more likely to surface in asymmetrically c-commanded positions than the less complex gracile forms who, me, he, she, and we. And we find a general trend towards towards invariant me, him, her, us, them, who in all strong pronoun contexts, as predicted by the Invariant Strong Form constraint. Although the main aim of Chapters 8 and 9 is to provide further evidence for the relevance of the proposed constraints to pronoun case choice in Modern English, the data discussed in these chapters also serve to illustrate the shortcomings of any study based solely on literary corpora and anecdotal evidence. The lack of exhaustive pronoun case data from individual authors means that it is often difficult to determine whether observed differences in pronoun case represent different diachronic stages, synchronic variation between speakers, or case variability within the speech of individuals.1 Chapter 8 is organised as follows: Section 8.1 looks at pronoun case in topicalisation structures. Topicalised pronouns resemble the wh-pronouns to be discussed in Chapter 9, in that they appear in clause-initial position at SpellOut and are linked to an empty argument position within the clause. Sections 8.2 to 8.4 are dedicated to left-dislocated, right-dislocated, and independent pronouns, which can be argued to lack the argument status required for Argument Case assignment and Positional Case checking. Section 8.5 considers the case of pronouns after be in basic identificational sentences and it BE constructions. As already mentioned in the discussion of it-clefts (Section 5.3), I propose that the occurrence of case variation after be arises from the lack of an agentive/causative vP-layer in the extended projection of the verb be. The absence of the vP-layer means that the pronoun following be is unable to check objective Positional Case. V-ing constructions, non-finite to-clauses, and small clauses are the focus of Sections 8.6 to 8.8. What all of these constructions have in common is the ability to occur as arguments of a higher predicate, and the lack of a construction-internal functional projection associated with nominative Pos-Case assignment. Sections 8.9 and 8.10 examine the case variation found with pronouns occurring in gapping and bare argument ellipsis. Both of these constructions will be argued to lack certain agreement-related functional projections at Spell-Out. Focus prepositions such as but feature in Sections 8.11 and 8.12. Since focus prepositions are neither predicates nor agreementrelated functional heads, pronouns following a focus preposition are only sub1

I would like to thank Diane Massam (p.c.) for alerting me to the importance of this issue.

CHAPTER 8

205

ject to the Default Case constraint. Section 8.13 discusses the case of pronouns following focus markers such as only, and Section 8.14 takes a look at case variation in constructions where the pronoun is in some way modified. The case variation found with modified pronouns in canonical argument positions supports the analysis of modified pronouns as strong (cf. Chapter 3). 8.1

Topicalised pronouns When a noun phrase is topicalised, it raises to a clause-initial position. Like relativisation, this fronting of the noun phrase is not accompanied by an associated raising of the finite verb (cf. Sections 9.6-9.7). From a semantic perspective, a topicalised constituent is a secondary focus, which is embedded in the presupposition of another focus (Williams 1997:614f). In metalinguistic responses to sentences that already contain a topicalisation structure (7), a topicalised element may be elevated to primary focus (Williams 1997:611f).2 (7) A: John would eat beans and corn, but broccoli he WOULDN’T EAT. (broccoli = topicalised & secondary focus) B: No, CABBAGE he wouldn’t eat. (cabbage = topicalised & primary focus) While both subject and non-subject noun phrases can be semantically topicalised, the syntactic consequences of the operation are most evident when the topicalised phrase is the object of a verb or preposition. Although Jespersen & Haislund (1949:223f) and Householder (1987: 164) argue that topicalisation encourages the use of nominative pronoun forms, the case of lone topicalised pronouns generally corresponds to the case of lone pronouns in canonical argument positions (cf. Lasnik & Sobin 2000:353). When the subject of a finite clause is topicalised, it obligatorily surfaces in the nominative case, no matter whether it functions as a secondary (8)-(9) or primary focus (10)-(11).3 (8) So what about you? What did you eat? a. I/We ate BEANS. b. * Me/Us ate BEANS.

2

The examples in (7) are taken from Williams (1997:612). Topicalised constituents are given in bold print, and the primary focus of the sentence is capitalised. 3 Examples (8) and (9) are based on an example discussed in Williams (1997:615).

206 (9)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

So what about Kim? What did she/he eat? a. She/he ate SPAGHETTI. b. * Her/him ate SPAGHETTI.

(10) So Kim ate the beans. a. No, I/WE ate the beans. b. * No, ME/US ate the beans. (11) So you ate the spaghetti? a. No, ) HE/SHE/THEY ate the spaghetti.4 b. * No, ) HIM/HER/THEM ate the spaghetti. Topicalised objects of verbs or stranded prepositions tend to appear in their objective case forms (12)-(13), as do topicalised subjects of embedded nonfinite clauses and small clauses (14).5 (12) a. Her I like. b. Them I would never (ask anyone else to) taste. (Lasnik & Sobin 2000:353)

(13) a. Me they never listen to. b. Him they never talk about. (14) a. Him I consider [to be a genius]. b. Me they would never consider [suitable]. A closer look at the data collected by Householder (1987:181) and Jespersen & Haislund (1949:224f,264), reveals only one clear instance of a lone topicalised nominative pronoun that functions as the object of a verb (15). (15) But shee, I can hooke to me (Shakespeare, The winter’s tale:II.iii.6) [Jespersen & Haislund 1949:264]

Note that 3rd person pronouns can only occur as a primary focus if they are used deictically and are accompanied by a pointing gesture. 5 Since the case preferences illustrated in (12)-(14) appear to hold both when the pronoun bears secondary focus (i) and when it bears primary focus (ii), I will not mark the primary focus in any of the remaining examples in this section. 4

(i) Her I LIKE. (but he really gets on my nerves) (ii) ) HER I like. (not him)

CHAPTER 8

207

Virtually all of the remaining examples of unexpected nominatives in topicalised position involve 3sg pronouns modified by a relative clause. In some of these sentences, the nominative form of the topicalised pronoun could be argued to have been influenced by the function of the relativised constituent in the clause. Thus, the topicalised object of the verb treat in (16) and the topicalised subject of the small clause complement of see in (17) are both modified by a subject relative. However, nominative 3sg forms also occur when the relativised constituent in the clause is the object of a verb or preposition (18) (the examples given below are all cited in Jespersen & Haislund 1949:224f). (16) [She, who had been the bane of his life] … he treated with the respect a good son might offer a kind mother (Charlotte Brontë 1867[1852], Villette:378)

(17) [He, who had always inspired in herself a respect which almost overcame her affection], she now saw the object of open pleasantry (Jane Austen 1894[1813], Pride and prejudice:475)

(18) a. [She whom thine eie shall like], thy heart shall haue (Christopher Marlowe, Doctor Faustus:594)

b. [She in whom I might have inspired a dearer love], I had taught to be my sister (Charles Dickens 1897[1849-50], David Copperfield:775) The use of she in (18) is reminiscent of the preference for the nominative whform whoever in topicalised free relatives where both the wh-pronoun and the free relative itself function as the object of a verb (19) (cf. Section 8.5.1). (19) And generally, [who euer the King fauours], The Cardinall instantly will finde imployment (Shakespeare, Henry VIII:II.i.47) [Jespersen 1949[1927]:58]

The surface word order created by the topicalisation of verbal and prepositional objects in English matrix clauses suggests that topicalised pronouns occupy a position higher than [Spec,TP]. As mentioned above, topicalised elements normally bear secondary focus (20), but can be elevated to primary focus in metalinguistic responses to sentences that already contain a topicalisation structure (21).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

208

(20) So what about Kimberley? How do you get on with her? Her I LIKE. (her = topicalised & secondary focus) (21) So your problem with Kimberley and Kevin is really Kimberley; him you like. - No, HER I like. (HER = topicalised & primary focus) Both Williams (1997:611f) and Rizzi (1997:285ff) argue that the syntactic status of the topicalised element in (20) differs from that in (21). In the approach proposed by Rizzi (1997:287, 297), a topicalised pronoun with primary focus (21) occupies the specifier of a Focus Phrase (FocP) in the C-system and directly binds a phrasal trace in argument position (22). A topicalised pronoun with secondary focus (20), on the other hand, occupies the specifier of a Topic Phrase (TopP) in the C-system, and is assumed to be unable to bind the empty argument position directly (23). Instead, the argument trace is bound by an empty anaphoric operator in the specifier of a lower functional projection associated with finiteness (FinP). The interpretive relation between the empty anaphoric operator and the phrase in [Spec,TopP] is assumed to be similar to the relation between a relative pronoun and its antecedent (Rizzi 1997:292f; cf. also Section 9.7).6 (22)

FocP 3 DP Foc’ PRONOUNi 2 Foc TopP 2 Top FinP 2 Fin TP 2 tj

The syntactic status and surface position of a pronoun with primary focus, in the approach advocated by Rizzi (1997:287, 297)

(23)

TopP 3 DP Top’ pronouni 2 Top FinP 2 Fin’ DPj Oi 2 Fin TP 2 tj The syntactic status and surface position of a pronoun with secondary focus, in Rizzi’s approach (1997:286, 314)

While the analysis in (23) may seem fairly plausible for topicalised objects of verbs and prepositions, it appears problematic for sentences where the subject is topicalised. As we will see in Section 9.7, restrictive subject relatives are generally introduced by a wh-pronoun or the complementizer that (24), 6

The interpretive relation between the null operator and its antecedent in (23) is represented by coindexation.

CHAPTER 8

209

even though both the relative pronoun and the overt complementizer are preferentially omitted when the relativised constituent is the object of a verb or stranded preposition (25). (24) a. You’ll have to talk to the people [who had beans]. b. You’ll have to talk to the people [that had beans]. c. * You’ll have to talk to the people [Ø had beans]. (25) a. You are talking about the man [Ø I love]. b. They remind him of the horrible men [Ø he used to go about with]. If we assume that the relative clauses lacking an overt wh-pronoun are introduced by an empty operator binding an argument trace, the data in (24) and (25) suggest that Present-Day English readily permits a null operator in the Csystem to bind object traces, but that the binding of a subject trace by a null operator is possible only if the operator in some way interacts with that (24b). Since the complementizer that is clearly not involved in the licensing of topicalised subjects with secondary focus, we will have to argue that the syntactic and interpretive relationship between the topicalised pronoun and the null operator in (23) is sufficiently strong to license the binding of a subject trace. This raises the question whether we really have enough syntactic evidence to assume that a separate null operator is present in sentences where the topicalised element bears secondary focus in Present-Day English. Rizzi (1997: 329n.11,331n.26) notes that there appear to be subtle differences in subjacency and that-trace effects between primary and secondary focus topicalisations, but the constructions involved seem rather complex and the differences difficult to judge. As mentioned above, the case preferences for topicalised pronouns with secondary focus would appear to be identical to the case preferences for topicalised pronouns with primary focus: topicalised subjects obligatorily surface in the nominative case, and topicalised objects of verbs and stranded prepositions tend to surface in the objective case, unless the topicalised pronoun is modified by a relative clause. This suggests that topicalised pronouns have a case status very similar to that of wh-pronouns in embedded questions and relative clauses. I will therefore assume that all topicalised pronouns appear in [Spec,CP] at Spell-Out and directly bind a phrase in argument position (26).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

210 (26)

CP 3

DPi

pronoun

C’ 3 C TP 6 ti

The syntactic status and surface position of a topicalised pronoun in a matrix clause, in the approach adopted here (cf. Riemsdijk 1997:9n.3; Demirdache 1997:224)

As Riemsdijk (1997:9n.3) points out, topicalisation is marginal in embedded clauses, but when topicalisation does occur in an embedded clause, the topicalised constituent always follows the complementizer that (cf. also Potsdam 1998:325,350n.7-8). If we assume that topicalisation involves movement to the specifier of a C-related head, then the word order in embedded clauses with topicalisation could be seen as evidence for the existence of more than one functional head in the C-system (cf. Rizzi 1997). Topicalised phrases could then be argued to occupy the lowest specifier position in the C-system, with the higher specifier position reserved for wh-phrases, which typically precede the complementizer that in varieties of English that allow the cooccurrence of an overt wh-pronoun and complementizer (cf. Browning 1996: 252 and Rizzi 1997:308 for analyses along these lines). However, since the presence of additional empty positions in the C-system has no bearing on the predictions of the three case constraints proposed here, I will continue to show only a single C-related functional layer in my tree diagrams unless there is clear overt evidence for an additional projection. The strong preference for the nominative wh-form who in matrix questions (cf. Section 9.1) suggests that finite C is able to check nominative PosCase on a pronoun in [Spec,CP] if T has raised to C, or if there is surface adjacency between C and T. While topicalisation never triggers T-to-C raising in English, C and T are adjacent at Spell-Out when the topicalised constituent is the subject of the matrix clause (27). (27)

CP 3

C’ DPi [nom] I 3 C [nom] TP [+ finite] 3 ! DP T’ ! ti 3 z------ T FP 6 ate beans

CHAPTER 8

211

The use of the nominative in this context is further reinforced by the Arg-Case constraint, which requires a topicalised subject to surface in the nominative form because it is the highest argument of a predicate. The case constraints thus correctly predict that the topicalised subject of a matrix clause should always appear in the nominative case (8)-(11). When the topicalised pronoun is the object of a verb or stranded preposition, the subject of the clause will intervene between C and T at Spell-Out and thus prevent C from acquiring the ability to check nominative Pos-Case (28). The case status of topicalised objects of verbs or prepositions resembles the case status of object wh-pronouns in embedded questions (cf. Section 9.4). They are arguments of a predicate but do not occupy a surface position covered by Pos-Case, which means that their surface form will be affected by Arg-Case and Def-Case requirements, but not by Pos-Case. The same applies to topicalised pronouns that function as the subject of an embedded nonfinite clause or small clause (29). (28)

CP (29) CP 2 2 C’ DPj C’ DPj her 2 him 2 C TP C TP [+ finite] 2 [+ finite] 2 ! DPi T’ ! DPi T’ 2 ! I 2 ! I z--=- T FP z--=- T FP 6 ei ti consider[tj (to be) a genius] ti like tj never ti listen to tj

While Def-Case calls for objective case forms in all positions not covered by Pos-Case, the Arg-Case constraint predicts that topicalised pronouns and whpronouns should surface in their objective form only if they function as the object of a verb or preposition. Subjects of all types of embedded clauses are linked to nominative Arg-Case, because they function as the highest argument of a predicate. The case constraints proposed here would thus lead us to expect case differences between topicalised objects and topicalised subjects of embedded clauses: topicalised objects of verbs and prepositions should always appear in the objective case, while topicalised subjects of embedded clauses should be able to surface either in the objective or the nominative case. This prediction does not seem to be confirmed by the case trends reported in topicalisation structures. Lone topicalised pronouns preferentially surface in the objective

212

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

form even when they function as the subject of an embedded non-finite clause or small clause (30), while topicalised pronouns modified by a relative clause may appear in the nominative even when they function as the object of a verb or preposition (31). (30) a. Him I consider [to be a genius]. b. Me they would never consider [suitable]. (31) [She, who had been the bane of his life] … he treated with the respect a good son might offer a kind mother (Charlotte Brontë 1867[1852], Villette:378) [Jespersen & Haislund 1949:225]

The preference for objective personal pronoun forms sentences like (30) could be seen as evidence for the trend towards invariant me, him, her, us, them in strong pronoun contexts. We would expect the influence of such a trend to be particularly noticeable in positions not covered by Pos-Case, because ArgCase and Def-Case appear to be less influential than Pos-Case in Present-Day English (cf. Chapter 2), and Def-Case also calls for objective forms. While the trend towards invariant me, him, her, us, them, can help us account for the selection of objective personal pronoun forms in sentences like (30), we will have to find some other explanation for the occurrence of nominative personal pronoun forms in topicalised constituents where the pronoun is modified by a relative clause (31). As discussed in Section 5.3, the interpretive relationship between an overt or empty wh-pronoun and its antecedent (represented by coindexation on the pronouns in the tree diagrams) could be argued to lead to case agreement. Under such an analysis, we might expect the antecedent to be influenced by the Arg-Case assigned to the wh-pronoun in the relative clause. Thus, the use of the nominative she in sentences like (31) could be argued to arise from case agreement with the wh-pronoun, which receives nominative Arg-Case, because it functions as the subject of the relative clause (32). (32)

CP qp DP C’ ei 6 he treated with respect DP CP shei 3 [nom] DPj C’ 6 whoi [nom] tj had been the bane of his life

CHAPTER 8

213

Neither relative-internal Arg-Case requirements nor the trend to invariant strong forms offer a possible explanation for the selection of the nominative she in sentences like (33), where both the personal pronoun and the relative pronoun function as the object of a verb, or in (34), where the topicalised pronoun appears without a modifying relative clause. (33) [She whom thine eie shall like], thy heart shall haue (Christopher Marlowe, Doctor Faustus:594) [Jespersen & Haislund 1949:224]

(34) But shee, I can hooke to me (Shakespeare, The winter’s tale:II.iii.6) [Jespersen & Haislund 1949:264]

As we will see in Section 8.14.7, 3sg pronouns modified by a restrictive relative clause readily surface in the nominative forms he and she, even when they are not topicalised. It thus seems that the mere presence of a relative clause has a bearing on the case form of a pronoun. In Section 8.14.7, I will propose that the use of he and she with a restrictive relative is related to the position of the pronoun within the relative construction: a pronoun followed by a restrictive relative will always asymmetrically c-command the relative CP (35). Similarly, a lone topicalised pronoun will always asymmetrically c-command the remainder of the clause (36). (35)

DP 3 Di NumP 3 she NumP CP 4 6 ti

whom thine eie shall like

(36)

CP 2 DPj C’ shee 2 C TP 2 DPi T’

I 6

can ti hooke tj to me

The ready use of the 3sg nominatives he and she before a restrictive relative, and the marginal occurrence of lone 3sg nominatives in topicalisation (34), could thus be seen as evidence for influence from the Relative Positional Coding constraint, which calls for the gracile forms he and she in asymmetrically c-commanding positions.

214

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

8.2

Left-dislocated pronouns As Erdmann (1978:69) points out, left-dislocation of lone pronouns most commonly involves a 1sg pronoun coreferent with the subject of the clause (37).7 (37) Me, I usually end up giving them their stupid incompletes. (X., e-mail message, 15-12-1986) [Prince 1998:288]

Non-1sg pronouns rarely occur in left-dislocated position by themselves, but when they do, they generally surface in their objective forms, just like 1sg pronouns, no matter whether they are coreferent with the subject of the clause (38a-c), or the object of a verb (38d). (38) a. Him, he’s crazy. (Rodman 1997 [1974]:53n.8) b. Us, we’ll go together. (Ross 1986[1967]:259) c. Them, they can’t stand each other. (Ross 1986[1967]:259) d. Him, I can’t stand him. (Potsdam 1998:302) Both Householder (1987:164) and Jespersen & Haislund (1949:223f) suggest that left-dislocated pronouns coreferent with the object of a verb or preposition may surface either in the nominative or objective case. However, only one of their examples involves a lone left-dislocated pronoun (39). (39) She, I’ve never spoken to her. (George Higgins 1985[1975], A city on a hill:251) [Householder 1987:181]

While lone nominatives are rather exceptional in left-dislocated position, left-dislocated non-1sg pronouns quite readily surface in their nominative form if they are modified by a relative clause or form part of a coordinate. Leftdislocated non-1sg pronouns modified by a relative clause frequently appear in their nominative form, especially when the relative clause modifying the pronoun is a subject relative (40)-(42). This trend applies to pronouns modified by restrictive relatives (41)-(42) as well as pronouns modified by non-restrictive relatives (40). As (40)-(42) illustrate, the case and grammatical function of the intrusive pronoun has little bearing on the surface form of a left-dislocated pronoun modified by a relative clause (for a detailed discussion of the status of the

7

See Erdmann (1978:69) and Householder (1987:181) for further examples.

CHAPTER 8

215

pronoun associated with the left-dislocated constituent see Demirdache 1997: 198). (40) Left-dislocated pronouns modified by a relative clause and coreferent with the subject of the main clause [Thou, who didst subdue Thy country’s foes ere thou wouldst pause to feel The wrath of thy wrongs, or reap the due Of hoarded vengeance] … [thou who with thy frown Annihilated senates] … thou didst lie down (Byron, Sulla:Ch.H.IV.83) [Jespersen 1934[1924]:27] (41) Left-dislocated pronoun modified by a relative clause and coreferent with the subject of a small clause [He that is able to receiue it], let him receiue it (The authorized version of the Bible 1611 (1833 Facsimile ed.):Matthew 19.12) [Jespersen & Haislund 1949:224]

(42) Left-dislocated pronouns modified by a relative clause and coreferent with the object of a verb or preposition a. [Hee that rewards me], heauen reward him (Shakespeare, Henry IV, Part 1:V.iv.167) [Jespersen & Haislund 1949:224]

b. But [we indeed who call things good and fair], The evil is upon us while we speak (Elizabeth Barret Browning 1856, Aurora Leigh:42) [Jespersen & Haislund 1949:224]

As Riemsdijk (1997:4) points out, the kind of left-dislocation we find in English is commonly referred to as Hanging Topic Left Dislocation (HTLD). Kayne (1994:78), Vat (1997:82), and Demirdache (1997:198) all propose that left-dislocated constituents in English HTLD are base-generated in the specifier of a functional head. Anagnostopoulou (1997:167f), on the other hand, argues that the left-dislocated constituent in English HTLD is adjoined to CP, and suggests that this is why embedded HTLD is possible only in environments that permit CP recursion, namely when the CP is selected and governed by a bridge verb. However, like the topicalised phrase in an embedded topicalisation, the left-dislocated constituent in embedded HTLD always follows the complementizer that, which suggests that left-dislocated constituents do not adjoin to CP. I will therefore assume that all left-dislocated pronouns are basegenerated in the specifier of a functional head in the C-system (43).

216 (43)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

CP 3 The syntactic position of a left-dislocated pronoun (bold print)8 XP C’ pronouni 3 C TP 3 pronouni

As discussed in Section 8.1, Pos-Case checking between C and a pronoun in [Spec,CP] is possible only if no overt constituent intervenes between C and T at Spell-Out. Since C and T are separated by an overt subject in all English sentences involving left-dislocation, a left-dislocated pronoun will be unable to check Pos-Case, even if we assume that Pos-Case applies to both argument and non-argument noun phrases.9 Unlike topicalised DPs, left-dislocated constituents do not function as arguments of a predicate (cf. Riemsdijk 1997:2-4), which means that a left-dislocated pronoun will be unaffected by any Arg-Case requirements. We might speculate that left-dislocated pronouns could receive an Arg-Case through agreement with a coreferent intrusive pronoun, but the case data discussed above suggest that the semantic and syntactic relation between the intrusive pronoun and its left-dislocated antecedent are not strong enough to license case agreement.10 If we assume that Arg-Case agreement with the intrusive pronoun is not an option, left-dislocated lone pronouns will be subject only to Def-Case. The Def-Case constraint calls for objective pronoun forms in all contexts not covered by Pos-Case, and thus correctly predicts that left-dislocated lone pronouns should always surface in their objective form.11 8 The possible occurrence of topicalised or left-dislocated constituents after that (cf. Section 8.1) suggests that such constituents actually appear as specifiers of a lower functional head within the C-system (cf. Rizzi 1997). However, since the presence of additional empty positions in the C-system has no bearing on the predictions of the three case constraints, I will show only a single C-related functional layer in my tree diagrams unless there is clear overt evidence for an additional projection. 9 I am assuming that Pos-Case only affects argument DPs in Modern English, although the argument status of a DP could conceivably be irrelevant for Pos-Case checking (cf. Section 2.3). 10 As Riemsdijk (1997:5) notes, “case agreement is generally optional and often blocked in HTLD constructions”. Thus, left-dislocated noun phrases in German (i) tend to appear in the nominative case, even when the associated pronoun in the following clause is in the accusative.

(i) Der Hans, ich kenne ihn schon seit zwölf Jahren. know 3sgM.ACC already since twelve years the.MASC.NOM Hans I “Hans, I’ve known him for twelve years.” (Riemsdijk 1997:5) 11 Compare Ross’ (1986[1967]:259fn.19) suggestion that nominative case forms are blocked by the feature [+ objective] which is assigned to a left-dislocated NP by the Left Dislocation rule.

CHAPTER 8

217

The ready occurrence of non-1sg nominatives when the left-dislocated pronoun is modified by a subject relative, points to possible influence from relative-internal Arg-Case. As discussed in Sections 5.3 and 8.1, the interpretive relationship between the wh-pronoun and its antecedent may lead to ArgCase agreement (44). As a result of this case agreement, the wh-pronoun may surface in the Arg-Case assigned to its antecedent, or the antecedent may surface in the Arg-Case assigned to the wh-pronoun (cf. Section 8.14.7 and 9.7). (44)

CP qp DP C’ ei 6 thou didst lie down DP CP 3 thoui [nom] DPj C’ 6 whoi

[nom] with thy frown tj Annihilated senates

The data presented in (41)-(42) suggest that Arg-Case agreement is not restricted to overt wh-pronouns and their antecedents, but is also possible between empty wh-operators and their antecedents (45). (45)

CP qp DP C’ ei 6 there is no lady in the world DP CP yei 3 too good for yow [nom] DPj C’ 6 Oi [nom] that tj be soo wel borne a man

(Thomas Malory 1889[1485], Morte d’Arthur:150) [Jespersen & Haislund 1949:224]

While case agreement between the wh-pronoun (or operator) and its antecedent could account for the use of the nominative when a left-dislocated pronoun is modified by a subject relative, the occurrence of nominative pronoun forms in left-dislocated coordinates such as (46) defies any purely casebased analysis. (46) But yesternight my lord, [she and that fryer] I saw them at the prison (Shakespeare, Measure for measure:V.134) [Jespersen & Haislund 1949:224]

218

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Given that coordinates appear to be transparent to outside case influences (cf. Section 5.1.3), a purely case-based approach would lead us to predict that the conjuncts of a left-dislocated coordinate should surface in the objective forms required by the Def-Case constraint, as in (47). (47) [me and her], we just never got on together (Harris 1981:19) This suggests that the selection of she in (46) is due to factors other than case. If we assume that coordinates are ConjPs headed by the conjunction, a pronoun occupying the initial conjunct of a coordinate will asymmetrically c-command the remainder of the ConjP (cf. Section 5.1.3). The use of she and other gracile nominatives in initial conjuncts of a left-dislocated coordinate could thus be seen as further evidence for the influence of Relative Positional Coding on strong pronoun forms in English. The assumption that 3sgF pronouns preferentially surface as she in asymmetrically c-commanding positions, would also allow us to account for the occurrence of she in example (39), whose structure is given in (48). (48)

CP 3 DP C’ she 3 C TP 6

I’ve never spoken to her

Since left-dislocated pronouns do not function as arguments of a predicate, a purely case-based approach would predict that the left-dislocated 3sgF pronoun in (48) will be influenced only by the Def-Case constraint, and should therefore surface in the objective form her. Even Arg-Case agreement with the coreferent pronoun in the following clause could not account for the use of the nominative she, because in (48) the left-dislocated pronoun is coreferent with the object of a preposition rather than the subject. However, if we assume that pronoun forms not only encode case, but also asymmetric c-command relationships, then the selection of she in (48) could be argued to arise from the fact that the left-dislocated pronoun asymmetrically c-commands the remainder of the clause.

CHAPTER 8

219

8.3

Right-dislocated pronouns According to Visser (1963: 56), the use of right-dislocated lone pronouns was particularly common in the 16th and 17th century, but can also be found in more recent texts (49). (49) a. I have stood up and defended you, I. (Ben Jonson [1602], Poetaster:III.i.306) [Visser 1963:56]

b. He was not going to be a snuffy schoolmaster, he. (George Eliot 1860, Mill on the Floss:Vol.II,Ch.1,191) [Visser 1963:56]

c. Nay, we have no art to please our friends, we! (Ben Jonson [1599], Every man out of his humour:IV.v.16) [Visser 1963:56]

In all of the examples cited by Visser (1963:56), the right-dislocated pronoun surfaces in the nominative case and is coreferent with the subject of the preceding clause. Indeed, the ungrammaticality of sentences like (50) and (51) suggests that right-dislocated lone pronouns must generally be coreferent with the subject of the matrix clause.12 (50) Right-dislocated lone pronouns may not be coreferent with a possessive pronoun (a), or with the object of a verb (b) or preposition (c) a. Sarah’s always been on our side, *we / *us. b. Alicia took me off the list yesterday, *I / *me. c. We talked about him yesterday, *he / *him

12 As Diane Massam (p.c.) points out, we need to be circumspect when we compare the historical data in (49) with the present-day grammaticality judgments given in (50)-(51). In PresentDay English, the sentences in (49) seem rather archaic, but they are still noticeably less degraded than the examples in (50)-(51). Since it is impossible for us to obtain grammaticality judgments form the 16th-19th century authors cited by Visser, we can never be certain whether they would have considered the sentences in (50)-(51) to be ungrammatical or not. However, the apparent absence of examples such as (50)-(51) from earlier texts and grammars, would seem to suggest that right-dislocated pronouns were preferentially coreferent with the subject of the matrix clause in earlier varieties of Modern English.

220

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(51) Right-dislocated lone pronouns may not be coreferent with the subject of an embedded non-finite clause (a-b) or small clause (c) a. They expected me to win, *I / *me.13 b. Kevin arranged for her to win, *she / *her. c. The police let him go yesterday, *he / *him. (based on Ross 1986[1967]:258)

Ross (1986[1967]:258) argues that in Present-Day English, the rightdislocation of lone pronouns is impossible even when the pronoun is coreferent with the subject of the preceding matrix clause. However, while I would agree that Present-Day English speakers are unlikely to use nominative pronouns in right-dislocation (49), the right-dislocated objective forms in (52) strike me as quite acceptable. (52) a. I like beer, me. (Ross 1986[1967]:258)14 b. He doesn’t drink beer, him. c. She wouldn’t touch a broom, her. As Ross (1986[1967]:258f) points out, right-dislocated pronouns readily surface in either their nominative or objective forms when they form part of a coordinate (53). (53) a. We’ll do it together, [you and I/me]. b. They can’t stand each other, [he and she] / [him and her] (Ross 1986[1967]:259)

13

Note the difference between (51a), where the pronoun appears in right-dislocated position, and (i), where the pronoun forms a nonsentential constituent. (i) They expected me to win. Me! Unlike the right-dislocated pronouns in (49)-(52), the independent pronoun in (i) is strongly stressed, carries a noticeable complex pitch-movement and is preceded by a clear pause. The contrast in grammaticality between (51a) and (i) suggests that right-dislocated lone pronouns must be coreferent with the subject of a finite (matrix) clause, whereas independent pronouns need not be. 14 Although Ross (1986[1967]:258) generally considers the right-dislocation of lone pronouns to be ungrammatical, his annotations suggest that he viewed the use of the objective form me in right-dislocation as slightly more acceptable than the use of the nominative I in the same context (i)-(ii). (i) * I like beer, I. (ii) *? I like beer, me.

CHAPTER 8

221

Unlike right-dislocated lone pronouns, right-dislocated coordinates containing pronominal conjuncts may be coreferent with constituents other than the subject of the matrix clause. Although pronouns are most likely to surface in their objective forms when the coordinate is coreferent with a pronoun in the objective case (54), the example in (55) suggests that nominative pronoun forms are possible in this context (the antecedent of the pronoun is underlined in the examples). (54) a. The police are bound to let them go, [him and Andrea]. b. Alicia took us off the list yesterday, [you and me]. (55) We have’em to the house for dinner … [they and their wives] (Ring Lardner 1975[1915-28], The best short stories of Ring Lardner:337) [Householder 1987:181]

Right-dislocated pronouns modified by a noun phrase (56) or a relative clause (57) are also able to appear with coreferents other than the subject of the matrix clause, and again need not agree with their coreferent in case. (56) I speke of us, [we mendiants], [we freres] (Chaucer, Skeat’s six-volume edition, Canterbury tales:Group D, 1912) [Jespersen & Haislund 1949:225]

(57) the auxiliaries which then stood Upon our side, [us who were strong in love]! (William Wordsworth, The prelude:11.107) [Jespersen & Haislund 1949:225]

There is no clear agreement in the literature on the exact syntactic relation between a right-dislocated constituent and the associated clause. In a syntactic approach that permits right-adjunction, right-dislocated constituents are best treated as adjuncts to a high-level verbal projection, presumably TP (58). (58)

TP ei TP XP

Right-dislocation as adjunction to an extended projection of the verb

right-dislocated constituent

In a strictly antisymmetric approach to syntactic structure, on the other hand, a right-dislocated constituent will have to occupy the specifier or complement position of a functional head (cf. Kayne 1994).

222

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Kayne (1994:78) suggests that right-dislocation involves an abstract functional head (F in the diagram), which takes the right-dislocated constituent as its complement and the preceding clause as its specifier (59). Alternatively, right-dislocated constituents could be argued to be base-generated in the specifier of a functional projection in the C-system, just like left-dislocated constituents. In such an approach, the difference in surface order between right-dislocation and left-dislocation constructions would be due to the presence versus absence of TP-raising. While the TP remains in-situ in left-dislocation, it raises past the dislocated constituent in right-dislocation structures (60).15 (59)

FP 3 CP F’ 3 F XP right-dislocated constituent

Right-dislocated constituents as complements of an abstract functional head (Kayne 1994:78)

(60)

CP 3 C’ TPi 3 C FP 3 XP F’ dislocated 2 constituent F TP ti

Right-dislocated constituents as specifiers of a functional head in the C-system.

Although the relationship between the right-dislocated constituent and the clause differs quite markedly in (58), (59), and (60), the structural differences between the analyses have little bearing on the predictions of the three case constraints. Since the right-dislocated constituent does not function as an argument of a predicate in the clause, we would predict that the only external case-influence on the right-dislocated constituent should come from the DefCase constraint. However, a right-dislocated pronoun could also be subject to case influences internal to the right-dislocated constituent, if we assume that right-dislocated constituents have a more complex internal structure than is apparent from their phonological form.

15 Compare Kayne’s (1994:151n.18) suggestion that sentences like (i) could be derived from (ii) through VP-preposing.

(i) He is real smart, John is. (ii) John is [he is real smart]. As discussed in Sections 8.1 & 8.2, the word order found in embedded clauses provides independent evidence that topicalised and left-dislocated constituents do not occupy the highest specifier position in the C-system (cf. also Rizzi 1997).

CHAPTER 8

223

Kayne (1994:78) suggests that right-dislocated constituents are best analysed as clauses reduced by the ellipsis of a verbal projection. As discussed in more detail in Sections 9.5 and 9.6.2, there is general agreement in current theory that the ellipsis of a constituent must be licensed by a higher functional head, either through government (cf. Lobeck 1995:50ff), or through head-head agreement (cf. Merchant 2001:60). If we follow Kayne (1994:78) in assuming that right-dislocated pronouns occupy the subject position in a reduced clause, the ellipted constituent is most likely to be the verbal projection immediately dominated by TP, with the ellipsis licensed by T. Since T must be present at Spell-Out to license the ellipsis of the following FP, the right-dislocated pronoun in [Spec,TP] will occupy a nominative Pos-Case position, no matter whether we assume that ellipsis involves PFdeletion (cf. Merchant 2001) or the base-generation of empty categories (cf. Chao 1987; Lobeck 1995). As discussed in Section 2.3, an argument DP is able to check Pos-Case if it appears in the specifier of an agreement-related functional head at Spell-Out and its surface position differs from its θ-position. If we treat ellipsis constructions as the result of PF-deletion (cf. Merchant 2001), the pronoun will have been base-generated within a lower verbal projection (either vP or VP), where it would have received its θ-role. Its surface position in [Spec,TP] will thus be different from its θ-position. If we follow Chao (1987) and Lobeck (1995) in assuming that ellipsis involves the base-generation of empty categories, a right-dislocated pronoun will have to be inserted directly into [Spec,TP], because the FP complement of T will lack internal structure in the overt syntax (cf. Sections 9.5 & 9.6.3). However, it could nevertheless be argued to receive its θ-role in a position other than [Spec,TP]. Given that θ-role assignment is generally assumed to take place at a semantic level of representation, we could argue that the θ-role of the subject DP is assigned to a reconstructed trace in [Spec,vP] or [Spec,VP] (cf. Chao 1987:73f). Both a PFdeletion analysis and a base-generation approach to ellipsis will thus predict that the pronoun in [Spec,TP] should be able to check nominative Pos-Case with T when the reduced clause is finite (61). (61)

CP 3 The brackets around the trace of the subject DP C TP indicate that the trace must be present at some [+ finite] 3 level of representation (in order to receive the DPi [nom] T’ θ-role), but not necessarily at Spell-Out. I 3 FP T [nom] 3 ( ti [θ-role] )

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

224

Since the missing predicate in an ellipsis construction must be present at SF (Semantic Form) to ensure the correct interpretation of the overt constituents (cf. Chao 1987:65-74; Lobeck 1995:32-35; Chung et al. 1995), the selection of nominative pronoun forms will be further encouraged by the Arg-Case constraint, which stipulates that the highest argument of a predicate should surface in the nominative case. A reduction approach to right-dislocation would thus predict that rightdislocated pronouns should always surface in their nominative forms. While this appears to be exactly what we find in the 16th-19th century data cited by Visser (1963:56), the Present-Day preference for objective pronoun forms in right dislocation (52) would seem to suggest that Present-Day speakers tend to assign a monoclausal analysis to right-dislocation constructions. In a monoclausal analysis, any right-dislocated constituent overtly realised as a pronoun will be base-generated as a DP rather than a CP. As a result, the surface form of a right-dislocated pronoun will be determined entirely by the position and function of the right-dislocated constituent in the clause (62). (62) Tree diagrams illustrating the syntactic position of a right-dislocated pronoun in a monoclausal analysis of right-dislocation structures (a)

TP 2 TP DP

(b)

pronoun if we treat right-dislocated pronouns as TP-adjuncts

FP 2 CP F’ 2 F DP pronoun

(c)

CP 2 TPi C’ 2 C FP 2 DP F’ pronoun 5 ti

if we assume that right-dislocated constituents are basegenerated as complements of an abstract functional head (b), or as specifiers of a functional head in the C-system (c)

Since the right-dislocated pronoun in (62a-c) does not function as an argument of a predicate in the associated clause, it will fail to receive an Arg-Case, and will also be unable to enter into Pos-Case checking. This means that the surface form of the right-dislocated constituent in (62a-c) will be influenced only by the Def-Case constraint, which calls for objective pronoun forms. A monoclausal analysis of right-dislocation structures thus correctly predicts the present-day preference for objective pronoun forms in right-dislocated position.

CHAPTER 8

225

A monoclausal analysis also provides a straightforward account for the possible occurrence of coordinated and modified right-dislocated pronouns with a non-subject coreferent (63). (63) a. pray eek for us, [we sinful folk unstable] (Chaucer, Skeat’s six-volume edition, Canterbury tales:Group B,1877) [Jespersen & Haislund 1949:225]

b. the auxiliaries which then stood Upon our side, [us who were strong in love]! (William Wordsworth, The prelude:11.107) [Jespersen & Haislund 1949:225]

If we assume that ellipsis/deletion of material is possible only under identity, then a reduced clause analysis will require right-dislocated pronouns to appear in the same syntactic configuration as their antecedents: A right-dislocated pronoun coreferent with the object of a verb or preposition would have to occupy a position within a verb phrase or prepositional phrase. A right-dislocated pronoun coreferent with a possessive would have to occupy a specifier position within a noun phrase. In order to account for the ellipsis of constituents before (and after) an object or possessive pronoun in a right-dislocated clause, we would have to extend our reduction analysis to include a wider range of ellipsis types than originally suggested by Kayne (1994:78) (see Section 8.4 for further discussion). What is more, even an extended reduction analysis would be unable to account for the case differences between the right-dislocated pronouns and their coreferents in (63). In a monoclausal analysis, on the other hand, the surface position and case status of a dislocated constituent is not directly related to the position and case of the coreferent intrusive pronoun, which means that the case differences between the right-dislocated pronoun and its coreferent in (63) are much more easily accommodated. While the majority of the data presented at the start of this section would seem to support a monoclausal analysis of right-dislocation, the occurrence of nominative pronoun forms in both conjuncts of right-dislocated coordinates coreferent with a subject (64), is more readily captured in a reduced clause approach. (64) They can’t stand each other, [he and she]. (Ross 1986[1967]:259) In a reduced clause approach, the right-dislocated coordinate will function as the subject of the ellipted predicate and occupy [Spec,TP] at Spell-Out. Provided we assume that coordinates are transparent to outside case influences, the

226

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

conjuncts of a coordinate in this position will receive nominative Arg-Case and check nominative Pos-Case (65). (65)

CP ei C TP [+ finite] wo ConjPi [nom] T’ 3 3 Conj’ T [nom] FP DP[nom] he 3 3 Conj DP [nom] ( ti [θ-role] ) and she

In order to account for the use of objective pronoun forms in the same context (66), we could assume that the speakers concerned assign a monoclausal analysis to right-dislocation structures. (66) They can’t stand each other, [him and her]. (Ross 1986[1967]:259) Since a right-dislocated constituent does not have argument status in a monoclausal analysis, it will fail to receive either Arg-Case or Pos-Case, and will instead be predicted to surface in the objective form required by the Def-Case constraint. Alternatively, speakers who produce utterances such as (66) might be particularly susceptible to the trend towards invariant me, him, her, us, them in all strong pronoun contexts, and thus favour objective pronoun forms in coordinates, no matter whether they treat right-dislocation structures as biclausal or monoclausal. The availability of a reduced clause analysis for right-dislocated constituents would allow us to provide a purely case-based account for the use of they in (67). (67) We have’em to the house for dinner … [they and their wives] (Ring Lardner 1975[1915-28], The best short stories of Ring Lardner:337) [Householder 1987:181]

In a reduced clause analysis, the right-dislocated coordinate in (67) will function as the highest argument of the preposition to in the small clause they and their wives to the house, and will therefore receive nominative Arg-Case. As discussed in Section 2.2.2, the pronominal subject of an embedded non-finite clause usually raises to [Spec,vP] of the matrix verb, where it checks objective Pos-Case. Since the Pos-Case constraint is more influential than the Arg-Case

CHAPTER 8

227

constraint in Present-Day English, the nominative Arg-Case assigned within the non-finite clause will generally be overridden by the objective Pos-Case checked in [Spec,vP] of the matrix clause. The obligatory use of the objective 3pl form (th)em in (68), suggests that a lone pronoun functioning as the subject of an embedded small clause also raises to [Spec,vP] before Spell-Out (69). (68) We have (th)em / * they to the house for dinner. (69)

FP 3 F vP 1 3 v’ vl F DPi [obj] 1 them 3 twe v’ Vn v have 3 VP v [obj] 3 tl V PP 3 tn DP P’ ti 3 P DP to 3 the house

If movement out of the small clause was equally obligatory for coordinated pronouns as for lone pronouns, we would predict that objective Pos-Case will also override nominative Arg-Case in right-dislocated coordinates coreferent with the subject of an embedded small clause. However, if we assume that coordinated pronouns are able to remain within the small clause throughout the derivation,16 the surface form of the coordinated pronoun in (69) will be influenced by Def-Case rather than Pos-Case. Although the Def-Case constraint also calls for objective pronoun forms, it is weaker than the Pos-Case constraint in Present-Day English, and could therefore be overridden by nomina16

As discussed in Section 2.2.1 and Chapter 3, only weak pronouns must raise to [Spec,vP] to be licensed. Since coordinated pronouns are strong, they may remain in [Spec,VP] throughout the derivation. When movement is optional, an argument remaining in its base position tends to receive a rhematic interpretation at Information Structure. Since pronouns are essentially topics, they will tend to raise out of their base-position before Spell-Out, even if they are strong. However, the (novel) coordination of two noun phrases could be argued to constitute new information, because it requires the creation of a new file card (cf. Heim 1982 for a detailed discussion of file change semantics). This could explain why coordinated pronouns appear to be more likely to remain in their base position than lone strong pronouns (cf. Section 5.1.3).

228

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

tive Arg-Case. The selection of they in the initial conjunct of the right-dislocated coordinate, could of course also be due to non-case influences. As mentioned in Section 8.2, the initial conjunct of a coordinate asymmetrically ccommands the final conjunct, provided we analyse coordinates as ConjPs headed by the conjunction (cf. Section 5.1.3). Thus, the use of they in initial conjuncts could be seen to arise from Relative Positional Coding demands for gracile forms in asymmetrically c-commanding positions. The use of nominative pronoun forms in examples like (70) is difficult to account for in a purely case-based approach, no matter what syntactic analysis we assign to right-dislocation structures. (70) pray eek for us, [we sinful folk unstable] (Chaucer, Skeat’s six-volume edition, Canterbury tales:Group B,1877) [Jespersen & Haislund 1949:225]

A monoclausal analysis would predict that the right-dislocated pronoun should surface in the objective form required by the Def-Case constraint, because it does not function as the argument of a predicate. A reduction analysis would predict that the right-dislocated pronoun should surface in its objective form, because it functions as the object of an ellipted preposition. As such, it will be influenced by the Arg-Case and Def-Case constraint, both of which call for objective forms in this context. The selection of the nominative form we in (70) would thus appear to be due to factors other than case. As discussed in Section 5.2, 1pl pronouns followed by a noun phrase often surface in their nominative form, even when the whole construction appears as the object of a verb or preposition. This tendency towards the gracile form we in 1pl-NP constructions is predicted by the Relative Positional Coding constraints, as long as we assume that the pronoun asymmetrically c-commands the following noun phrase (71). (71)

8.4

DP 3 D NumP we 6 sinful folk

Sinful folk could either be treated as an AP complement of Num (cf. Abney 1987:284; Longobardi 1994:635ff), or it could be analysed as an adjunct to NP or NumP (cf. Cardinaletti 1994:202-205)

Independent pronouns The data cited in existing studies suggest that in Present-Day English, independent pronouns consistently surface in their objective forms when they are interpreted as the object of a verb or preposition (72).

CHAPTER 8

229

(72) a. A: Who did the police question after the accident? B: Me. / Him. / Her. / Us. / Them. b. A: John gave a book to someone. B: Me. / Him. / Her. / Us. / Them. (Barton 1990:89) Objective pronoun forms would appear to be favoured even when the independent pronoun is interpreted as the subject of a finite clause (73), but nominatives may also occur in this context (74).17 (73) a. what would you have done? Me? Let her stew in her own juice (John Galsworthy 1925, Caravan:452) [Jespersen & Haislund 1949:275]

b. I believe he wants us to make you a minister. - Him! He’d lose the war first (Arnold Bennett 1926, Lord Raingo:20) [Jespersen & Haislund 1949:275]

c. ‘Isn’t that a pretty girl!’ Henry exclaimed… ‘Who?’ Willy looked round. ‘The little Carter.’ ‘Oh her. Yes, of course.’ (P.H. Johnson 1971[1968], Catherine Carter:143) [Erdmann 1978:68]

d. Who’s to stop it - us? (Doris Lessing. The golden notebook:172) [Erdmann 1978:68]

(74) a. ‘who was in your room?’ Withouth waiting for Lovell to answer, the other boys, each in turn, said, ‘I, sir,’ or ‘Me, sir.’ (Horace A. Vachell 1905, The hill:64) [Jespersen & Haislund 1949:275]

b. A: Someone gave a book to John. B: I. / He. / She. / We. / They. Me. / Him. / Her. / Us. / Them. (Barton 1990:90) While the choice between nominative and objective pronoun forms is sometimes seen as a matter of style (cf. Kjellmer 1986:445), one example cited in Erdmann (1978: 68) suggests the case form of an independent pronoun may in part depend on its φ-features. Thus the independent 3sgF pronouns in (75) surface in the nominative form she, whereas the 1sg pronouns take the objective form me.

17 Cf. Erdmann (1978:68), Quirk et al. (1985:337), Householder (1987), and Wales (1996:99f). Denison (1996:288) observes that the use of the nominative is also characteristic of earlier stages of English.

230

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(75) She: Oh! Me: That’s just what you are. She: You’re just hard and mean. Me: Who’s mean? (W. Sansom 1969[1958], The cautious heart:137) [Erdmann 1978:68]

In the existing literature, independent pronouns are generally analysed either as nonsentential constituents or as remnants of extensive ellipsis (see Barton (1990:23-42) for an overview). In an ellipsis approach, any independent pronoun interpreted as an argument of a predicate, will be influenced by the Arg-Case constraint, because the missing predicate is assumed to be present at least at a semantic level of representation (cf. Chao 1987:65-74; Lobeck 1995: 32-35; Merchant 2001). When an independent pronoun is interpreted as a subject, it will be linked to nominative Arg-Case, while independent pronouns interpreted as the object of a verb or preposition will be linked to objective ArgCase. An independent pronoun interpreted as the subject of a finite clause will occupy [Spec,TP] at Spell-Out in an ellipsis approach, and will therefore have the same syntactic status as a right-dislocated pronoun in a reduction analysis (cf. Section 8.3). Since T will have to be present at Spell-Out in order to license the ellipsis of the missing verbal projection (cf. Lobeck 1995; Merchant 2001), the independent pronoun will be able to check nominative Pos-Case (76), no matter whether we assume that ellipsis involves PF-deletion (cf. Merchant 2001) or the base-generation of empty categories (cf. Chao 1987, Lobeck 1995). (76)

CP The brackets around the trace of the subject DP 3 C TP indicate that the trace must be present at some [+ finite] 3 level of representation (in order to receive the θ-role), but not necessarily at Spell-Out. 18 DPi [nom] T’ I 3 T [nom] FP 3 ( ti [θ-role] )

An independent pronoun functioning as the object of a verb, would most plausibly occupy [Spec,VP] in an ellipsis analysis. In Section 2.2.1, I argued that lone object pronouns generally need to raise to [Spec,vP] before Spell-Out even when they are strong, because their topic status is incompatible with the new information focus associated with VP-internal object positions. However, when 18

As discussed in Sections 2.2 and 8.3, a DP that occupies the specifier of an agreementrelated functional head at Spell-Out, will only be able to undergo Pos-Case checking if its surface position differs from its θ-position.

CHAPTER 8

231

a strong object pronoun contributes some new information to the discourse, we might expect it to remain in [Spec,VP]. Since independent object pronouns do just that, I will assume that they would appear in [Spec,VP] rather than [Spec,vP] at Spell-Out (77), and will therefore be unable to check objective Pos-Case.19 (77)

VP 3 DP V’ me 4

The surface position of an independent object pronoun in an ellipsis approach

This means that an independent pronoun interpreted as the object of a verb will be influenced by Def-Case rather than Pos-Case. The same goes for any independent pronoun interpreted as the object of a (covert) preposition. In an ellipsis approach, an independent object pronoun will thus have the same status as an object pronoun in bare argument ellipsis (cf. Section 8.10) or in the second conjunct of a gapped sentence that lacks an overt lexical verb (cf. Section 8.9). If we treat independent pronouns as nonsentential constituents (cf. Barton 1990), they will have the status of independent DPs throughout the derivation (78). As a consequence, even independent pronouns interpreted as the subject of a finite clause will be influenced by the Def-Case constraint rather than PosCase requirements. (78)

DP 5 me

The surface position of an independent pronoun in a nonsentential constituent approach (cf. Barton 1990)

Barton (1990:112) argues that the interpretation of all nonsentential constituents is determined by Discourse Inference. According to Barton (1990: 112), Discourse Inference “is triggered by a match between an independent constituent utterance [i.e. nonsentential constituent] and an expansion possibility of a previous element within a structure of linguistic context”, where ‘expansion possibility’ refers to the ability to take arguments or modifiers. Since Discourse Inference effectively links a nonsentential pronoun to an open or filled position in the argument structure of a predicate in the linguistic context (cf. Barton 1990:112-128), we might expect an independent pronoun to be influenced by Arg-Case requirements even when it is analysed as a nonsentential 19 Since the presence or absence of higher functional projections at Spell-Out has no bearing on the case status of an object in [Spec,VP], I have decided to show only the immediate environment of the pronoun in the tree-diagram.

232

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

constituent. The variation between nominative and objective pronoun forms in contexts where the independent pronoun is interpreted as the subject of a finite clause, could then be seen as the result of competition between Arg-Case, which requires the highest argument of a predicate to be nominative, and DefCase, which calls for the use of objective pronoun forms.20 If we want to maintain a purely case-based approach, we will only be able to account for the general preference for objective independent pronoun forms if we assume that the influence of the Arg-Case constraint is weaker than the influence of the Def-Case constraint when the argument status of a pronoun is determined through Discourse Inference. Alternatively, the preferential selection of objective pronoun forms could be seen as evidence that the Arg-Case and Def-Case constraint interact with the trend towards invariant me, him, her, us, them. While the interaction of the case constraints will predict variation between nominative and object pronoun forms when an independent pronoun is interpreted as a subject, no purely case-based analysis can account for the case differences between 1sg and 3sgF pronouns in examples like (79). (79) She: Oh! Me: That’s just what you are. She: You’re just hard and mean. Me: Who’s mean? (W. Sansom 1969 [1958], The cautious heart:137) [Erdmann 1978:68]

The occurrence of the 1sg objective form me in the same environment as the 3sgF nominative she, is reminiscent of the case trends found in coordinates. In coordinates, me and she tend to be found in initial rather than final conjuncts, and in the passage quoted in (79), both she and me appear in initial position, before the quote. As discussed in Sections 5.1.3 and 6.2, there are good reasons to assume that initial conjuncts asymmetrically c-command the remainder of the coordinate, and that the preference for gracile pronoun forms such as she and me reflects the syntactic status of the initial conjunct within the construction. If the pattern in (79) is representative of pronoun case preferences before direct quotes, then the use of the gracile forms me and she could signal that, at

20

Barton herself proposes the following case-rule for non-sentential constituents (1990:89-91): (i)

Case Rule for NP constituent structures If N’’ is the initial node, then assign any Case.

The drawback of Barton’s case rule is that it is unable to explain why independent pronouns are more likely to surface in the nominative if they are interpreted as the subject of a clause than as the object of a verb or preposition.

CHAPTER 8

233

some level of representation, the pronouns in (79) are analysed as asymmetrically c-commanding the following stretch of direct speech. 8.5

Pronoun case after be In Present-Day English, a lone pronoun generally surfaces in its objective form when it appears immediately after the verb in a basic declarative sentence (80). (80) a. I saw him / * he. b. They visited me / * I. c. We gave them / * they the keys.

When a lone pronoun occurs after main verb be, on the other hand, it may take either the nominative or the objective form. The following constructions involving be are particularly prone to case variation: (a) basic identificational sentences (e.g. I am he/him) (b) it BE sentences (e.g. It is I/me) (c) it-clefts with a pronominal focus (e.g. It was they/them who’d taken it) Since I have already discussed it-clefts in Section 5.3, I will focus on basic identificational sentences and it BE sentences in the following sections. As mentioned in Sections 2.2.1 and 5.3, I assume that the potential for case variation in (a)-(c) arises from the fact that all three constructions involve identificational be, which fails to project a vP-layer, and is thus unable to check objective Pos-Case on an argument in postverbal position. 8.5.1 Pronouns in basic identificational sentences Many of Jespersen & Haislund’s (1949:251) and Visser’s (1963:236ff) examples of nominatives after be in basic identificational sentences, come from Early Modern English texts (81)-(82). (81) a. [if that I am I], then well I know your weeping sister is no wife of mine (Shakespeare, Comedy of errors:III.ii.41) [Visser 1963:236] b. that shall not be I ([Udall] [1553 ?], Ralph Roister Doister:21) [Jespersen & Haislund 1949:251]

c. this is not she ([Udall] [1553 ?], Ralph Roister Doister:26) [Jespersen & Haislund 1949:251]

234

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(82) a. they sayd alle, O my lord sir launcelot, [be that ye], and he sayd [Truly I am he] (Thomas Malory 1889[1485], Morte d’Arthur:713) [Jespersen & Haislund 1949:251]

b. [If thou art she], tell me where is that son that floated with thee on the fatal raft (Shakespeare, Comedy of errors:V.i.349) [Visser 1963:238] As can be seen from these examples, the occurrence of a nominative form in postverbal position is independent of verb agreement. In the second example in (82a) and in (82b) the verb clearly agrees with the preceding 1sg and 2sg pronoun, respectively, yet the 3sg pronoun following the verb still surfaces in the nominative case. We still find instances of postverbal nominatives in more recent texts (83),21 but objective forms are clearly more common than nominatives in Present-Day English, no matter whether the other argument shares the φ-features of the postverbal pronoun (84), or has different φ-features (85).22 21

It is important to note, that the definite noun phrase the butler in (83) is referential rather than predicative. As Heggie (1988) points out, definite noun phrases in apparently identificational sentences often have a predicative character. Thus the noun phrase the teacher in (i) and (ii), is most readily analysed as a property predicated of John, rather than a referential noun phrase that picks out an individual identified with John. (i) John is the teacher. (ii) The teacher is John. In (82), on the other hand, the butler is used to refer to a particular individual (the person dressed as the butler at the festivities), which suggests that the sentence is indeed identificational rather than predicative. The subject status of the butler in (83) is highlighted by its ability to be modified by only (iii) and to appear as the focus of a subject cleft (iv). (iii) Only the butler is him [i.e. the Vicomte de Blissac]. (iv) It’s the butler that is him [i.e. the Vicomte de Blissac]. As can be seen from the ungrammaticality of (v)-(vi) under a predicative interpretation, fronted predicate phrases are unable to be modified by only and cannot appear as the focus of a subject cleft (cf. Heggie 1988:133f). The only way to render (v) and (vi) grammatical is to interpret the teacher as picking out a particular individual, which is subsequently identified with John. (v) * It’s the teacher who/that is John. [if the teacher is interpreted as a predicate] (vi) * Only the teacher is John. [if the teacher is interpreted as a predicate] 22 For further examples and discussion see Jespersen & Haislund (1949:251-253), Visser (1963:236-243), Erdmann (1978:74), Emonds (1986:95f, 104f, 115-120), Householder (1987: 179f), Wales (1996:94f), Denison (1996), and Sobin (1997:334). Jespersen & Haislund (1949: 255) argue that the nominative is obligatory when the two pronouns in an identificational sentence have the same φ-features (i.e. the same person, number, and gender), but they themselves quote examples like (84b), which violate this generalisation.

CHAPTER 8

235

(83) as the only celebrant at the festivities so costumed had been the Vicomte de Blissac, [the butler consequently must have been he] (P.G. Wodehouse [1932], Hot water:122) [Visser 1963:237]

(84) a. I’m not a fish in the sea. I’m me. (advertisement for Cachet perfume) [Wales 1996:95]

b. since we are us (Edward F. Benson 1894[1893], Dodo:2.330) [Jespersen & Haislund 1949:253]

(85) a. [You’re him] (Dick Francis 1981[1979], Whip hand:180) [Householder 1987:180]

b. [If I was her], I would not have to put up with it (Jane Austen 1989[1813], Pride and prejudice:284) [Jespersen & Haislund 1949:253]

Wh-pronouns in identificational wh-questions generally surface in the nominative form who, regardless of whether they function as the subject or not. Thus, we not only find a preference for who in metalinguistic wh-questions, where the wh-pronoun functions as the subject (86), but also in identificational questions where the wh-pronoun functions as the lower argument of be (87).23 (86) a. ‘It’s me,’ I said. ‘Who’s me?’ he growled. (Alan Sillitoe 1972[1970], A start in life:48) [Erdmann 1978:68]

b. We had a breakdown … Who’s we? (Arnold Bennett 1922, Mr. Prohack:311) [Jespersen & Haislund 1949:141]

c. We were under the impression that you had a case this morning. Who is we? (Somerset Maugham, Plays:3.154) [Jespersen & Haislund 1949:141]

(87) a. Who am I? b. They don’t know [who I am]. c. ‘But who is she?’(Kingsley Amis 1970[1960], Take a girl like you:276) [Erdmann 1978:68]

d. We don’t know [who she is]. 23 In metalinguistic wh-questions, the verb shows 3sg agreement with the wh-pronoun and the postverbal pronoun resembles an echo, which just copies a part of a preceding utterance (cf. Barton 1990:223n.28).

236

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Similarly, who appears both before and after be in multiple wh-questions where the first wh-pronoun functions as the subject, and the second wh-pronoun functions as the lower argument of be (88). (88) a. Who’s who (title of a biographical lexicon) [Jespersen 1946:495] b. I showed the Bishop of Clogher, at Court, [who was who] (Jonathan Swift 1901, Journal to Stella:487) [Jespersen 1946:495]

In embedded questions, an initial wh-pronoun may sometimes take the objective form whom (89), but only when it does not function as the subject (cf. Section 9.4). In free relatives involving VP-ellipsis (cf. Section 9.6.2), the whpronoun may surface as whom when the whole relative appears after identificational be (90). (89) She did not know [whom this strange young man might be] (Hugh Walpole [1913], Fortitude:138) [Jespersen 1946:484]

(90) General Baird announced … that any thief detected in the fact, be he [whom he might], should be hung (Wilkie Collins [1868], The moonstone:5) [Jespersen 1946:483]

I would like to propose that identificational be takes two arguments, both of which are base-generated within VP in basic identificational sentences (91). Arg-Case linking assigns nominative Arg-Case to the higher argument (DP1), and objective Arg-Case to the lower argument (DP2). Since identificational be does not project a vP-layer, only one of its two arguments will be able to raise out of VP and enter into Pos-Case checking with a higher agreement-related functional head. The argument remaining within the VP will be subject to ArgCase and Def-Case requirements. If we assume that it is the higher argument that raises to [Spec,TP] in a basic identificational sentence (91), we would predict that the argument following be should always surface in the objective form demanded by the Arg-Case and Def-Case constraints. The analysis in (91) correctly predicts the present-day preference for objective personal pronoun forms after be in basic identificational sentences (84)(85), and can also account for the occurrence of whom in embedded questions (89) and free relatives involving VP-ellipsis (90) where the wh-pronoun functions as the lower argument of identificational be. As can be seen from (92)(94), the wh-pronoun appears in a position not covered by Pos-Case in both of

CHAPTER 8

237

these constructions, and is therefore subject to Arg-Case and Def-Case requirements, just like the VP-internal argument in (91). (91)

CP 3 I am assuming that lexical verbs generally raise out of their base-position before Spell-Out (cf. Section 2.2.1). As the C TP [+ finite] 3 functional head (F) targeted by this movement is not DPi [nom] T’ associated with the argument structure of the I 3 verb, it is present even in identificational T [nom] VP sentences, which lack a vP-layer. couldl 3 V FP tl 3 F VP Identificational be must raise bek 3 at least as far as FP before SpellDP1 V’ Out, but may undergo further raising ti 3 to T and C, because of its auxiliary qualities V DP2 her tk

(92)

CP 3 The surface position and case status of the arguments in an C’ embedded wh-question where the wh-pronoun functions DPj as the lower argument of identificational be whom 3 C TP [+ finite] 3 DPi [nom] T’ 3 you T [nom] FP 5 arek tk ti tk tj

(93)

FP 3 The surface position of a wh-pronoun in a free relative functioning as the lower argument of identificational F VP 1 3 be, if we adopt a pro-head approach to free Vk F DP V’ relatives (cf. Section 9.6) be he 3 V DP tk ei D CP ei proi [φ] DP C’ whomi 3 [φ] C TP [φ] 6 he might e

238 (94)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

FP 3 The surface position of a wh-pronoun in a free relative functioning as the lower argument of identificational F VP 1 3 be, if we adopt a wh-head approach to free Vk F DP V’ relatives (cf. Section 9.6) be he 3 V DP ei tk D CP 1 ei DP C’ Di D 6 whom g D he might e ti

In a matrix question corresponding to the embedded question in (92), the wh-pronoun can be argued to occupy a nominative Pos-Case position, because T raises to C before Spell-Out, and thus endows C with the ability to check nominative Pos-Case on a DP in its specifier (95). We would therefore predict that an initial wh-pronoun in matrix questions involving identificational be will always surface in the nominative form who (87a & c). (95)

CP 3 DPj [nom] C’ who 3 TP C [nom] 1 3 Tk C DPi T’ 3 are you T FP tk 5 tk ti tk tj

The ready occurrence of who in embedded questions and free relatives like (96), on the other hand, is more difficult to account for in a purely casebased approach. (96) a. They don’t know [who I am]. b. …let her be [who she might]. The wh-pronoun could be argued to acquire nominative Arg-Case through its interpretive relationship with the higher argument of identificational be (cf. Section 9.4.). However, if the selection of who in (96) was due to Arg-Case agreement in Present-Day English, we would expect to find an equally high

CHAPTER 8

239

preference for nominative pronoun forms after be in all declarative identificational sentences. This does not appear to be the case. In Present-Day English, personal pronouns following identificational be are much more likely to surface in their objective form than in their nominative form. It thus seems more plausible that the use of who in embedded indentificational questions and free relatives is triggered by non-case factors, most notably a trend towards invariant who in all wh-contexts (cf. Chapter 9 for further discussion). Although there is little evidence for general Arg-Case agreement in identificational sentences, we might speculate that Arg-Case agreement is a factor in the clear preference for who after be in multiple wh-questions involving identificational be (97), and in the occurrence of postverbal nominatives in identificational sentences where both arguments have identical φ-features (98). (97) she nyste [who was who] (Chaucer, Skeat’s six-volume ed., Canterbury tales:Group A,4300) [Jespersen 1946:495]

(98) a. [if that I am I], then well I know your weeping sister is no wife of mine (Shakespeare, Comedy of errors:III.ii.41) [Visser 1963:236] b. the axiom [that he was he] (Arnold Bennett 1909[1908], Old wives’ tale:2.50) [Jespersen & Haislund 1949:255]

As the word order in embedded contexts shows (97), the initial wh-pronoun has subject status in multiple wh-questions, which means that the wh-pronoun following be occupies a VP-internal position at Spell-Out (99). The postverbal wh-pronoun in (97) thus has the same syntactic status as the postverbal personal pronouns in (98).24 Since C is able to acquire the ability to check nominative Pos-Case from T when no overt element intervenes between the two heads at Spell-Out (cf. Sections 8.1 and 9.1), the Pos-Case constraint would predict that the initial whpronoun in a multiple wh-question will surface in the nominative form who. The use of of who in initial position will be further reinforced by the Arg-Case constraint, which requires the highest argument of a predicate to surface in the nominative case. In the absence of Arg-Case agreement, a wh-pronoun or personal pronoun following be would be predicted to surface in its objective form, because it is the lower argument of the verb, and occupies a position not cov24 Compare the tree diagram given in (91), which shows the position of a personal pronoun following be in a basic identificational sentence.

240

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

ered by Pos-Case. However, if we assume that the lower argument of identificational be may inherit the Arg-Case of the higher argument, provided the two arguments have identical φ-features, we will be able to account for the use of who and he after be in (97) and (98). (99)

CP The case status of the two arguments in multiple 3 DPi [nom] C’ wh-questions involving identificational be, if we who 3 assume that the higher argument raises to TP [Spec,TP] before Spell-Out C [nom] [+ finite] 3 ! DP T’ 3 ! ti z-----_ T FP 3 isk wask F VP 3 tk DP V’ 3 ti V DP who(m) tk

While Arg-Case agreement can account for the possible occurrence of nominative pronoun forms after be in (97)-(98), it cannot explain why who is virtually obligatory after be in multiple wh-questions like (97), even though objective personal pronoun forms may occur after be in identificational declaratives like (100). (100) I’m not a fish in the sea. I’m me. (advertisement for Cachet perfume) [Wales 1996:95]

This case difference between wh-pronouns and personal pronouns in identificational sentences suggests that the selection of who after be in multiple whquestions may be further reinforced by the trend towards invariant who (cf. Chapter 9). Since the emerging invariant forms for personal pronouns are objective rather than nominative (cf. Section 6.3), no such reinforcement of the nominative form is available for personal pronouns. If we assume that Arg-Case agreement is possible only in identificational sentences where the two arguments of be have identical φ-features, we will have to find a different explanation for the occurrence of postverbal nominatives in sentences like (101).

CHAPTER 8

241

(101) a. that shall not be I ([Udall] [1553 ?], Ralph Roister Doister:21) [Jespersen & Haislund 1949:251]

b. they sayd alle, O my lord sir launcelot, [be that ye], and he sayd [Truly I am he] (Thomas Malory 1889[1485], Morte d’Arthur:713) [Jespersen & Haislund 1949:251]

I would like to propose that the pronouns following be in (101) surface in their nominative form because they are analysed as the highest argument of the verb. As illustrated in (102), the higher argument of identificational be in a basic identificational sentence is base-generated in [Spec,VP], while the lower argument is base-generated as the complement of V. (102)

VP ei V’ DP1 higher 3 argument V DP2 lower argument

Collins (1997:27f) argues that two phrases are equidistant from a specifier position targeted by movement if they are base-generated within the minimal domain of the same head. This means that an argument base-generated as the complement of V should be equally able to raise to [Spec,TP] as an argument base-generated in [Spec,VP]. If we assume that the base position of arguments is associated with new information focus (cf. Sections 2.2.1, 8.3 and 8.4), then we might expect the lower rather than the higher argument to raise out of VP when the higher argument receives some kind of rhematic interpretation. This appears to be exactly what happens in (101). In the proposed analysis, the examples in (101) would have the syntactic structure given in (103). In (103), the lower argument occupies [Spec,TP] at Spell-Out, while the higher argument has remained within VP. The argument in [Spec,TP] will thus receive objective Arg-Case and nominative Pos-Case, while the postverbal argument will be linked to nominative Arg-Case and objective Def-Case. Since Pos-Case overrides ArgCase in Present-Day English, the preverbal argument will always surface in its nominative form. The postverbal argument, on the other hand, may surface either in the nominative form demanded by the Arg-Case constraint, or in the objective form required by the Def-Case constraint.

242 (103)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

CP 3 C TP [+ finite] 3 DPi [nom] T’ I 3 FP T [nom] amk 3 F VP 3 tk DP V’ he/him 3 V DP ti tk

As can be seen from the examples in (101), the verb always agrees with the preverbal argument in a basic identificational sentence. The proposed analysis therefore requires us to assume that verb-agreement is primarily determined by surface configurations rather than argument status. If we assume that Pos-Agreement overrides Arg-Agreement in Present-Day English, just like Pos-Case overrides Arg-Case, we would predict that the verb will agree with the preverbal argument, even when the postverbal argument is the higher argument of be (cf. Section 2.4 for further discussion of a constraint-based approach to agreement). 8.5.2 Pronouns in it BE sentences As can be seen from Table 8.1, the word order, case, and agreement properties of it BE sentences have undergone a number of changes throughout the history of English. In Old English the focus pronoun appeared in preverbal position, triggered agreement on the verb, and always surfaced in the nominative case (104). (104) ic hit eom 1sg.NOM it be.1SG “It is I/me” This suggests that in Old English, the focus pronoun was analysed as the highest argument of be in it BE sentences. At the end of the Middle English period, the focus pronoun started to appear in postverbal position, but it still surfaced in the nominative form and, at least initially, also triggered agreement on the verb (105).

CHAPTER 8

243

(105) ‘Quy la’ quod he. ‘Peter, it am I’, Quod she. (Chaucer, Shipman’s Tale:VII.214-5) [Gelderen 1997:157]

In Section 10.3.1, I will present evidence indicating that even in Old English, the tensed verb failed to raise beyond T except in direct questions, V-initial declaratives and imperatives, narrative-advancing clauses with an adverb in initial position, and some clauses with a negated V (cf. Pintzuk 1995, 1996). Since there does not appear to be any independent evidence for verb raising to C in identificational sentences at any period in the history of English, I will assume that any postverbal focus pronoun in an identificational sentence occupies its VP-internal base position at Spell-Out (cf. also Section 5.3). Old English & Early Middle English The focus pronoun is in the nominative and precedes both it and the verb. The verb agrees with the nominative pronoun.

Late Middle English (Chaucer) The focus pronoun is in the nominative and follows it and the verb. The verb agrees with the postverbal nominative pronoun.

Early Modern English onwards The focus pronoun is in the nominative and follows it and the verb. The verb shows invariant 3sg inflection, and could thus be argued to agree with it. Modern English The focus pronoun appears in its objective form and follows it and the verb. The verb shows invariant 3sg inflection. The sequence it is is often contracted to it’s.

ic hit eom Þu hit eart he hit is heo hit is we hit sind ge hit sind hie hit sind it am I it art thou it is he it is she it are we it are ye it are they it is I it is thou it is he it is she it is we it is ye it is they it is me it is thee/you it is him it is her it is us it is you it is them

Table 8.1. The historical evolution of it BE sentences (cf. Foulet 1936:54; Jespersen & Haislund 1949:251f; Visser 1963:41f,236-241; Traugott 1972:126; Gelderen 1997:150,157f)

244

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The obligatory use of nominative pronoun forms in the focus position during the Late Middle English and Early Modern English period, indicates that the focus pronoun was still analysed as the highest argument of be. By the start of the Early Modern English period, the verb no longer agrees with the focus pronoun, even if it occurs in the nominative case, but instead shows invariant 3sg inflection (106), which could be seen as agreement with preverbal it. (106) it is I (The Townley plays, ed. England, EETS 1897:129) [Jespersen & Haislund 1949:251]

As discussed in Section 8.5.1, a pronoun following be in an identificational sentence would only be expected to surface in the nominative case if it received nominative Arg-Case through structural linking or through Arg-Case agreement with a nominative argument. Since Arg-Case agreement appears to be limited to identificational sentences where the two arguments have identical φ-features, the only possible source for the nominative case on the postverbal pronoun in (105)-(106) would be its own position on the argument hierarchy. The changes in verb agreement between (105) and (106) could be seen as indicative of the increasing importance of Positional Agreement during the Middle English period. In the approach to case and agreement proposed here, the verb will only show consistent subject agreement with a postverbal noun phrase if (a)

the postverbal noun phrase is the highest argument on the argument hierarchy and

(b)

Argument Agreement (Arg-Agreement) is more influential than Positional Agreement (Pos-Agreement).

The agreement between the verb and the focus pronoun in (105) thus suggests that Arg-Agreement was able to override Pos-Agreement during Chaucer’s time. By the start of the Early Modern English period however, Pos-Agreement is clearly dominant, and the verb obligatorily surfaces in the 3sg form required by subject Pos-Agreement with the preverbal 3sg pronoun it (107).

CHAPTER 8

(107)

245

CP 3 The structure and Pos-Agreeement properties C TP [+ finite] 3 of it BE sentences in Early Modern English DPi [s-φ] T’ it 3 FP T [s-φ] isk 3 F VP 3 tk DP V’ I 3 V DP ti tk

The consistent use of the 3sg verb form in it BE sentences from the Early Modern English period onwards would seem to provide evidence that it is an argument of the verb and as such is able to trigger Pos-Agreement rather than Def-Agreement.25 Further evidence for the argument status of it in it BE constructions comes from the favoured Modern English version of the construction. In Modern English, the finite verb always appears in the 3sg form in it BE sentences, and the postverbal focus pronoun preferentially surfaces in its objective form (108).26 (108) it is me - I know it is me (Hall Caine 1897, The Christian:40) [Jespersen & Haislund 1949:252]

This preference for objective focus pronouns indicates that the postverbal pronoun is no longer analysed as the highest argument of be, and thus receives objective rather than nominative Arg-Case. Such an analysis is possible only if the verb be in it BE sentences takes two arguments, one of which is the preverbal pronoun it (cf. Gelderen 1997:148-151; Vikner 1995:233; Everett 1996: 42). I will therefore follow Erdmann (1978:75) in treating it BE sentences as a subtype of identificational be constructions, and will assume that in Modern English, it is generally analysed as the higher argument of the verb (109). 25 As discussed in Section 2.3, a clash between Pos-Agreement and Arg-Agreement is always resolved in favour of Pos-Agreement in Present-Day English. A clash between Def-Agreement and Arg-Agreement on the other hand tends to give rise to variation. If the 3sg form of the verb in it BE sentences was the result of Def-Agreement rather than Pos-Agreement, we would expect to find variation between 3sg agreement and agreement with the postverbal noun phrase. 26 For further data and discussion see Jespersen (1949[1927]:403), Jespersen & Haislund (1949:253ff), Visser (1963:240-243), Klima (1964:3f), Erdmann (1978:75), Householder (1987), and Wales (1996:94f).

246

(109)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

CP 3 The structure and Pos-Agreement properties C TP [+ finite] 3 of it BE sentences in Modern English DPi [s-φ] T’ it 3 FP T [s-φ] 3 isk F VP 3 tk DP V’ 3 ti V DP

tk

me

The similarity between basic identificational sentences and it BE sentences is highlighted by passages like (110), where the basic identificational sentence they aren’t us is rephrased as it isn’t us (cf. also Gundel 1977:555). (110) [they aren’t us] … [it isn’t us], Stephen, really. It can’t be us (H.G. Wells [1913], The passionate friends:229) [Jespersen & Haislund 1949:253]

As mentioned above, the focus pronoun of an it BE sentence usually surfaces in the objective case in Present-Day English. However, we do still find instances of nominative focus pronouns in it BE sentences (cf. Klima 1964:3f; Wales 1996:94f; Sobin 1997:334; Lasnik & Sobin 2000:344). Apart from who, which is as strongly favoured in it BE questions as in identificational whquestions (111), the nominative pronoun form most readily used in it BE sentences today appears to be the 1sg nominative I (112).27 Non-1sg nominatives would seem to be extremely rare, and restricted to certain discourse contexts (113). (111) a. Who was it?

b. I don’t know [who it was].

(112) It is I. / It was I. (Sobin 1997:334) (113) ‘Oh, look Dad, there’s John Cleese’, he said excitedly. It was, of course, he. (The London Evening Standard, 11 February 1993) [Wales 1996:95]

27 See Sobin (1997:334) and Lasnik & Sobin (2000:349f); note also the general focus on the variation between it (i)s me and it (i)s I in the literature (cf. Jespersen & Haislund 1949:251; Visser 1963:238-243; Klima 1964:4; Erdmann 1978; Harris 1981; Quirk et al. 1985:337).

CHAPTER 8

247

The Present-Day English case differences between wh-pronouns, 1sg pronouns, and non-1sg pronouns in it BE sentences suggest that the surface form of the focus pronoun may be influenced by factors other than case. In examples like (113), the discourse context could be argued to favour an analysis of the focus pronoun as the higher argument of the verb, and thus give rise to the selection of a nominative rather than objective form. However, the general preference for who in it BE questions (111), and the comparatively ready occurrence of I (112) need to be accounted for in a different way.28 As discussed in Section 8.5.1, the use of who in matrix questions involving identificational be can be seen as the result of nominative Pos-Case checking between the verb in C and the wh-pronoun in [Spec,CP] (114). In embedded it BE questions, nominative Pos-Case checking between the wh-pronoun and C is impossible, because it intervenes between C and T at Spell-Out (115). (114)

CP (115) CP 3 3 DPj C’ DPj [nom] C’ who 3 who 3 TP C TP C [nom] [+ finite] 3 1 2 DPi T’ ! DPi T’ Tk C was it 2 ! it 3 T FP z--=-- T FP 5 wask 5 tk tk ti tk tj tk ti t k tj

Given that both the Arg-Case constraint and the Def-Case constraint require the wh-pronoun to surface as whom if it functions as the lower argument of be and appears in a position not covered by Pos-Case, the use of who in embedded questions like (111b) is most plausibly analysed as the result of influence from the trend towards invariant who. Since the emerging invariant forms for personal pronouns correspond to their objective rather than nominative forms, the ready occurrence of I in it BE sentences will have to be due to influence from a different non-case factor. As discussed in Section 8.5.1, the lower argument of identificational be is basegenerated as the complement of V. This means that the focus pronoun is asymmetrically c-commanded by the remainder of the clause when it is analysed as the lower argument of be (116). 28 If we assumed that the use of who and I in it BE sentences arises from an analysis of the focus pronoun as the higher argument, then we would expect to find just as many instances of he, she, we, they in basic it BE sentences.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

248 (116)

CP 3 C TP [+ finite] 3 DPi T’ it 3 T FP 3 isk F VP 3 tk DP V’ 3 ti V DP tk me/I

The occurrence of the 1sg nominative I in this context could therefore be seen as further evidence that I patterns with the robust objective forms him, her, us, them in being favoured in asymmetrically c-commanded positions.29 8.6

Pronoun case in V-ing constructions As Abney (1987:212f) and Jespersen (1946:116) point out, Modern English V-ing constructions derive from two different sources: (a)

strictly nominal gerunds, which were originally marked with the suffix ung, and did not permit direct objects or adverbs; the subject of such a gerund always appeared in the genitive case (just like the possessive in a noun phrase)

29 As mentioned in Chapters 5 & 6, the assumption that the case constraints compete with a constraint relating pronoun form to asymmetric c-command allows us to capture the case similarities between focus pronouns in it BE sentences and pronouns in final conjuncts of coordinates. The approach proposed here thus has an advantage over the virus-based analysis proposed by Sobin (1997:336f), where the use of I in it BE sentences and the use of I in final conjuncts of coordinates are triggered by separate rules (i)-(ii).

(i)

The ‘it is I’ Rule If: it [Agrs is/was] [Prn +1, -pl, NOM]... 1 2 3 then: check NOM on 3. (Sobin 1997:337)

(ii) The ‘…and I…’ Rule If: …and [Prn +1, +sg, NOM]... 1 2 then: check NOM on 2. (Sobin 1997:336)

CHAPTER 8

(b)

249

strictly verbal present participle constructions marked with the suffix end(e)/-ind(e)

The phonological distinction between the nominal suffix -ung and the present participle suffix -end(e)/-ind(e) became neutralised during the Middle English period, when both of the suffixes started to be realised as -ing. The loss of the phonological distinction between the affixes “paved the way for the ‘mixing’ of the verbal properties of the participle and the nominal properties of the gerund” (Abney 1987:213; cf. also Jespersen 1946:116). In the mid-15th century we find the first evidence for ‘mixed’ gerunds, i.e. gerunds where a genitive subject cooccurs with a direct object (117). (117) You must excuse [my telling you] (Charles Dickens 1912[1865], Our mutual friend:28) [Jespersen 1946:148]

By the end of the 16th century, the gerund marks aspect and voice distinctions, so that we find examples of genitive subjects with perfective have (118) and with passives (119). According to Blume, a grammarian quoted by Jespersen (1946:110), the use of not with the gerund (120) also starts around Elizabethan times. (118) There is no record of [his ever having lost his temper] (Stacy Aumonier 1916, Olga Bardel, London:74) [Jespersen 1946:111]

(119) which … was the cause of [his being taken in the middle of the night out of his bed] (Laurence Sterne 1911, Tristram Shandy and A sentimental journey:1.110) [Jespersen 1946:115]

(120) knowing that [his not replying] would only infuriate the magistrate more (Charles Dickens, Oliver Twist:81) [Jespersen 1946:111] While pronominal subjects of gerunds in argument position (117)-(120) consistently surface in the genitive up until the 19th century, absolutive V-ing constructions only ever permitted objective or nominative subjects (121). (121) a. Maybe the dominie can clear it up, [him being a scholar] (James M. Barrie 1893, The little minister:118) [Jespersen 1946:49]

b. For, [he being dead], with him is beautie slaine (Shakespeare, Venus and Adonis:1019) [Jespersen 1934[1924]:128]

250

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

According to Jespersen (1946:45), absolutive V-ing constructions have their origin in translations from Latin, and are rare in Old English and Middle English. The subject of the construction was originally dative, in imitation of the Latin ablative, but with the loss of case distinctions in the (pro)nominal paradigms (cf. Chapter 1), nominative subjects became increasingly popular (Jespersen 1946:46). While noting that the nominative and dative appear to have been “used concurrently for some time” in the earlier texts, Jespersen (1946: 48f) argues that the current variation between nominative and objective forms in absolutive V-ing “is not a continuation of the old practice, but is due to the general dislocation of the feeling of cases”. In Present-Day English, the case of a pronoun appearing as the subject of a V-ing construction still depends in part on the position and function of the V-ing construction in the overall sentence. When the V-ing construction appears as the complement of a verb, a lone pronominal subject may surface either in the genitive (122) or in the objective case (123), but not in the nominative (124). (122) a. You won’t mind [my taking Blanche in to dinner] (George Bernard Shaw 1914, Misalliance, The dark lady, Fanny’s first play: 1.20) [Jespersen 1946:147]

b. take measures to prevent [our ever meeting again] (Charles Dickens 1900[1839], Nicholas Nickleby:726) [Jespersen 1946:110]

(123) a. you wouldn’t mind [me asking you about her] (Arnold Bennett 1924 [1923], Riceyman Steps:15) [Jespersen 1946:147]

b. to prevent [him making a fool of himself] (Anthony Hope [Hawkins] 1900, Father Stafford (6d.ed.):83) [Jespersen 1946:149]

c. You don’t mind [us having secrets]? (W.B. Maxwell [1928], We forget because we must:88) [Jespersen 1946:147]

d. She didn’t like [them taking notice of me] (W.B. Maxwell 1926, Fernande:196) [Jespersen 1946:147]

(124) a. * You won’t mind [I taking Blanche to dinner] b. * to prevent [he making a fool of himself] c. * She didn’t like [they taking notice of me]

CHAPTER 8

251

Although lone subject pronouns in object V-ing constructions never occur in the nominative (124), nominatives do sometimes appear when the subject is coordinated (125) (examples cited in Jespersen 1946: 135). (125) a. I recollect [[Pegotty and I] peeping out at them from my little window] (Charles Dickens 1897[1849-50], David Copperfield:21) b. Do you ever hear [[your mother and I] scrapping and fussing like that]? (Sinclair Lewis 1926[1925], Martin Arrowsmith:161) In absolutive V-ing constructions a pronominal subject may surface either in its nominative (126) or objective form (127), but never in the genitive (128). (126) a. But, my good Master Bates dying in two years after, and [I having few friends], my business began to fail (Jonathan Swift 1785, Works, Dublin:3.2) [Jespersen 1946:48]

b. For, [thou betraying me], I doe betray My nobler part (Shakespeare, Sonnet 151:5) [Jespersen 1946:50]

c. the little picture of Ashe and Lady Kiddy together - [he bending over her in his large, handsome geniality] and [she looking up] (Mrs. Humphrey Ward [1905], The marriage of William Ashe:36) [Jespersen 1946:46]

d. perch … [they being - like the wicked of the world, not afraid] (Izaak Walton 1653, The compleat angler:151) [Jespersen 1946:48]

(127) a. If I’d shut my mouth, absolutely they’d have used that to support their story [me being a very large holder of North Atlantics] (Arnold Bennett 1930, Imperial palace:260) [Jespersen 1946: 49]

b. But you see, [him being here, in the room] - I had to be careful (Arnold Bennett 1926, Lord Raingo:140) [Jespersen 1946:49]

c. why didn’t you say so before? and [us losing our time listening to your silliness]! (George Bernard Shaw 1916, Androcles and the lion:113) [Jespersen 1949[1927]:374]

(128) a. * [my having few friends], my business began to fail b. * But you see, [his being here, in the room] - I had to be careful c. * and [our losing our time listening to your silliness]!

252

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

When the V-ing construction appears as the subject of a clause or the complement of a preposition, the pronominal subject of the V-ing construction itself may appear either in the genitive (129)-(130), or in the objective (131)(132), or in the nominative case (133)-(134). (129) Examples illustrating the occurrence of genitive subject pronouns in V-ing constructions appearing as the subject of a clause a. [his being a prisoner here], renders it impossible (Charles Dickens 1890[1837f], Pickwick Papers: 498) [Jespersen 1946:101]

b. It cannot be wondered at that [their retiring all to sleep at so unusual an hour] should excite his curiosity (Henry Fielding 1782[1749], Tom Jones:3.71) [Jespersen 1946:128]

(130) Examples illustrating the occurrence of genitive subject pronouns in V-ing constructions appearing as the complement of a preposition a. without [my ever offering to pluck them] (Charles Lamb 1899, The essays of Elia:1.184) [Jespersen 1946:110]

b. it all depended on [your naturally liking me] (George Bernard Shaw 1898, Plays pleasant:272) [Jespersen 1946:101]

c. there could be no serious objection to [his doing formally what he might do virtually] (Thomas B. Macaulay, History of England:1.30) [Jespersen 1946:101]

d. the storm may be weathered without [our being, any of us, quite overcome] (Jane Austen [1811], Sense and sensibility:260) [Jespersen 1946:129]

e. the possibility of [their ever knowing what had happened] (George Eliot 1900[1859], Adam Bede:289) [Jespersen 1946:101]

(131) Example illustrating the occurrence of objective subject pronouns in V-ing constructions appearing as the subject of a clause [Him hanging around like this, just messing things up], don’t fit in anywheres that I can see. (Dashiell Hammett 1934, The thin man:252) [Jespersen 1946:139]

CHAPTER 8

253

(132) Examples illustrating the occurrence of objective subject pronouns in V-ing constructions appearing as the complement of a preposition a. to think of [me kissing Mr. H.] after all he’s done to me (Rudyard Kipling [1890], The light that failed:238) [Jespersen 1946:134]

b. There’s a talk of [him getting a knighthood shortly] (William Pett Ridge 1907, 96 Birnam Road:135) [Jespersen 1946:134]

c. you say nothing about [us calling] (Arnold Bennett 1912[1910], Clayhanger:2.79) [Jespersen 1946:134]

d. there could be no harm in [them walking together] (William Hazlitt [1823], Liber amoris:121) [Jespersen 1946:133]

(133) Examples illustrating the occurrence of nominative subject pronouns in V-ing constructions appearing as the subject of a clause a. [I having a great esteem for your honour and a better opinion of you than any of the quality], makes me acquaint you of an affair that I hope will oblige you to know (Addison et al. 1888, The spectator:394) [Jespersen 1946:138]

b. [They being her relations, too], made it so much the worse. (Jane Austen [1811], Sense and sensibility:120) [Jespersen 1946:138]

(134) Examples illustrating the occurrence of nominative subject pronouns in V-ing constructions appearing as the complement of a preposition a. Instead of [he converting the Zulus], the Zulu chief converted him (utterance overheard by Jespersen) [Jespersen (1934[1924]:141]

b. I should be his prisoner instead of [he being mine] (Arthur Conan Doyle, Strand Magazine, December 1894:571) [Jespersen 1934[1924]:141]

Abney (1987:167f) distinguishes argumental gerunds from participial V-ing constructions. Participial V-ing constructions, which include absolutive V-ing, have purely verbal characteristics, and are thus best analysed as containing only verbal projections. The occurrence of perfective have in absolutive V-ing constructions (135) suggests that participial V-ing is temporally independent and contains a TP-layer (cf. Wurmbrand 2001:100f).

254

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(135) and [we having fed them the instant they entered the room], they bowed and smiled (Henry Fielding 1762, Works (2nd ed.):4.17) [Jespersen 1946:46]

At the same time, the absence of consistent nominative case marking on pronominal subjects of absolutive V-ing indicates that participial -ING fails to project the CP-layer required for nominative Pos-Case checking (cf. Section 2.2.2). I will therefore assume that V-ing constructions in absolutive position have the structure given in (136). (136)

TP The structure and Pos-Case properties of 3 T’ absolutive V-ing constructions (participial) DPi he/him 3 T FP 3 F vP 1 3 v’ vl F DPj [obj] 1 her 3 Vk v ti v’ betraying 3 VP v [obj] tl 3 V’ tj 5 tk

In the proposed analysis, an object of the verb will check objective Pos-Case in [Spec,vP], and will also receive objective Arg-Case. In the absence of a CP-layer, the subject will be unable to undergo Pos-Case checking, and will instead be influenced by the Def-Case constraint, which calls for objective forms in all positions not covered by Pos-Case. Since the subject is the highest argument of the verb, it will also receive nominative Arg-Case.30 The variation between nominative and objective subjects in absolutive V-ing constructions could thus be argued to arise from competition between the Arg-Case constraint and the Def-Case constraint. However, since Arg-Case would appear to be more influential than Def-Case in most varieties of Present-Day English (cf. 30 Compare Jespersen’s (1934[1924]:128) observation that the nominative came to be used in non-finite clauses because of the association between subjects and nominative case. Jespersen (1934[1924]:128) points out that similar developments can be found in a range of IndoEuropean languages, and notes that “In English the nominative has prevailed in the standard language”.

CHAPTER 8

255

Section 9.2), the interaction of Arg-Case and Def-Case alone cannot account for increasing popularity of objective pronoun forms in this context. This suggests that the form of subject pronouns absolutive V-ing constructions is at least partly influenced by non-case factors, most notably the trend towards invariant me, him, her, us, them. According to Abney (1987:223f), V-ing constructions in argument position are DPs, and the case of the subject in a V-ing gerund depends on the adjunction site for the abstract morpheme -ING, which adds the categorical feature [+ N] to the projection it attaches to. In V-ing gerunds with genitive subjects (Poss-ing gerunds), -ING adjoins to VP and turns it into an NP. Since this NP is dominated by a complete DP-layer, the subject of the gerund will be able to raise to [Spec,DP] before Spell-Out, and check genitive case there. In V-ing gerunds with objective or nominative subjects (Acc-ing gerunds), -ING adjoins to IP and turns it into a DP.31 As the construction does not contain a D, the subject remains within IP and is not able to check genitive case. As discussed in Section 2.2.3, Abney’s (1987) approach to argumental gerunds needs to be modified if it is to fit in with current assumptions about phrase structure.32 Since -ING behaves like a functional head, it should be assigned a consistent category and complementation. The ready occurrence of perfective have in Poss-ing (135a) as well as Acc-ing gerunds (137b), suggests that both Poss-ing and Acc-ing gerunds are temporally independent and contain a TPlayer. (137) a. There is no record of [his ever having lost his temper]. b. There is no record of [him ever having lost his temper]. I will therefore follow Kate Kearns (p.c.) in assuming that -ING has the category Num in argumental V-ing constructions, and always takes a TP as its complement. The difference between Poss-ing and Acc-ing arises from the optional projection of a DP-layer, which is present in Poss-ing gerunds (138), but absent in Acc-ing gerunds (139)-(140). In the proposed analysis, both subject and object pronouns in a Poss-ing gerund will be able to enter into Pos-Case checking. The subject will check genitive Pos-Case in [Spec,DP], and the object will check objective Pos-Case in [Spec,vP]. Since the subject is the highest argument of the verb, it will also receive nominative Arg-Case, while the object will be linked to objective Arg31

Note that Abney’s (1987) IP is equivalent to TP in the approach adopted in this book. I would like to thank Kate Kearns (p.c.) for drawing my attention to the issues discussed here and for suggesting the subsequent analysis.

32

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

256

Case. As discussed in detail in Chapter 2, the selection of genitive rather than nominative pronoun subjects in Poss-ing gerunds indicates that Pos-Case overrides Arg-Case in Present-Day English. (138)

DP 3 D’ The structure and Pos-Case properties of DPi [gen] my 3 Poss-ing gerunds NumP D [gen] 2 -ING TP 2 T’ tj 2 T FP 3 F vP 1 3 v’ vl F DPj [obj] 1 him 2 ti v’ Vk v telling 3 v [obj] VP 5 tl tj t k

When -ING fails to project a DP-layer, the subject of the V-ing construction will be unable to check genitive Pos-Case, and an Acc-ing gerund results. Since T requires the presence of a CP-layer to be able to enter into Pos-Case checking, the subject of an Acc-ing gerund will also be unable to check nominative Pos-Case.33 When the Acc-ing gerund appears as the subject of a clause or the complement of a preposition, the subject of the gerund will remain within the TP throughout the derivation (139). Since the subject is unable to check Pos-Case within the gerund, its surface form will be influenced by the Arg-Case constraint, which requires the subject to be nominative, and the DefCase constraint, which calls for objective forms. As with absolutive V-ing constructions, the occurrence of both nominative and objective subjects in gerunds that appear as the subject of a clause or complement of a preposition (131)(134), could be argued to arise from the interaction of Arg-Case, Def-Case, and the trend towards invariant me, him, her, us, them in strong pronoun contexts.

33 As discussed in Section 2.2.2, T will check nominative Pos-Case on a noun phrase in [Spec,TP], if C is [+ finite], and objective Pos-Case if C is [- finite] and filled by for.

CHAPTER 8

257

When the Acc-ing construction appears as the complement of a verb, the subject will be able to raise out of the gerund, into the specifier of the vP projected by the matrix verb (140). Since the subject of the gerund occupies [Spec,vP] in the matrix clause at Spell-Out, it will receive objective Pos-Case as well as nominative Arg-Case. Unlike the competition between Arg-Case and Def-Case, the competition between Pos-Case and Arg-Case does not appear to lead to variation. As discussed above, lone pronominal subjects of gerunds in verbal complement position may only surface in the objective or genitive case, which can be seen as further evidence that Pos-Case overrides Arg-Case in Present-Day English. (140)

FP 3 F vP (139) NumP 1 3 2 vl F DPi [obj] v’ -ING TP 1 me 3 tmatrix subj v’ 2 Vn v T’ mind 2 DPi me/I 2 v [obj] VP T FP tl 2 6 V NumP tellingk himj ti tl tj tk tn 2 -ING TP 6 The surface position of the subject pronoun

in Acc-ing gerunds appearing as the subject of a clause or complement of a preposition

ti tellingk himj ti tj tk

Objective Pos-Case checking on a pronoun that functions as the subject of an Acc-ing construction in verbal complement position

The possible occurrence of nominative pronoun forms in coordinated subjects of Acc-ing gerunds in post-verbal position, could be seen as evidence that conjoined pronouns may remain within the gerund throughout the derivation, even when the gerund appears as the complement of a verb.34 If a coordinated subject appears in a gerund-internal position at Spell-Out, the conjoined pronouns will have the same case status as subjects of absolutive gerunds and gerunds appearing as the subject of a clause or complement of a preposition. We would therefore expect them to surface either in the nominative or in the objective case. 34 As discussed in Section 8.3, the information-structure properties of coordinates are different from the information-structure properties of lone pronouns, so we might expect coordinates to be able to remain within the gerund while lone pronouns have to raise to [Spec,vP].

258

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

As can be seen from (141)-(142), Poss-ing and Acc-ing gerunds may appear in virtually identical environments within the same texts. This suggests that the structural analyses associated with Acc-ing and Poss-ing gerunds are simultaneously available to speakers of Modern English, and do not necessarily have different semantic properties. (141) Who ever heard of [them eating an owl or a fox], madam, or [their sitting down and taking a crow to pick] (William M. Thackeray 1901[1853], The Newcomes:2) [Jespersen 1946:146]

(142) I should not mind [their talking about me] (1.267) I should not mind [them saying that] (1.269) (William Black [1873], The princes of Thule:1.267,269) [Jespersen 1946:147]

8.7

Pronoun case in to-infinitives Like V-ing constructions, to-infinitives with overt subjects may occur either in argument position or as an absolutive or independent constituent. What sets to-infinitives apart from V-ing constructions in Present-Day English, is their ability to contain the overt complementizer for.35 This complementizer seems to occur only in to-infinitives that function as the argument of a verbal predicate, and when for is present, a (lone) pronominal subject always surfaces in the objective case, regardless of whether the infinitive appears as the complement of a verb (143) or the subject of a finite clause (144). When the complementizer is absent and the to-infinitive appears as the complement of a verb, (lone) subject pronouns are obligatorily objective (145). In absolutive and independent to-inifinitives, on the other hand, a lone subject pronoun may surface either in its objective or its nominative form (146)-(147).

35

As Harris & Campbell (1995:62) point out, for started out as a preposition preceding a noun phrase that belonged to the matrix clause and controlled the empty subject of a following nonfinite clause (i). (i)

[it is bet for me] [to sleen my self than ben defouled thus] “It is better for me to slay myself than to be violated thus” (Chaucer; cited from Ebert 1978:12) [Harris & Campbell 1995:62]

During the Early Modern English period, the noun phrase complement of for came to be reanalysed as the subject of the non-finite clause, and for was reanalysed as a complementizer. This reanalysis of the preposition meant that for could now also occur with to-infinitives appearing as the subject of a finite clause (144).

CHAPTER 8

259

(143) a. I hardly know in what language you would choose [for me to reply] (Arthur W. Pinero 1895, The benefit of the doubt:235) [Jespersen 1946:301]

b. She wouldn’t like [for him to know anything] (Compton Mackenzie 1913-14, Sinister Street:1103) [Jespersen 1946:301]

c. I couldn’t bear [for us not to be friends] (Hugh Walpole 1928, The silver thorn:107) [Jespersen 1946:301]

(144) a. [for me to dispute that] would be all as one, as for you to dispute the management of a pack of dogs (Henry Fielding 1782[1749], Tom Jones:4.95) [Jespersen 1946:312]

b. [For him to win the race] would be unexpected. c. [For vs to leuy power Proportionate to th’enemy] is all impossible (Shakespeare, Richard II:II.ii.123) [Jespersen 1946:311]

(145) a. our neighbours did take [me to be a very goodly man] (John Bunyan 1907, Grace abounding, etc.:14) [Jespersen 1946:282]

b. I believed [thee to be too solemn] (Alfred Tennyson 1894, Poetical works:806) [Jespersen 1946:282]

c. I judged [him to be about sixty years of age] (George Gissing 1914, The house of cobwebs:48) [Jespersen 1946:282]

d. Poirot motioned [her to sit down] (Agatha Christie, Murder on the Orient Express:149) [Jespersen 1946:292]

e. I know [them to bee as true bred cowards as euer turn’d backe] (Shakespeare, Henry IV, Part 1:I.ii.205) [Jespersen 1946:283]

(146) ‘Oh - [me to come asking him for death] and [him to give me back my life instead].’ (R. Lehmann 1968 [1953], The echoing grove: 319) [Erdmann 1978:71]

(147) a. And [I to sigh for her, to watch for her, to pray for her] (Shakespeare, Love’s labour’s lost:III.202) [Jespersen & Haislund 1949:239]

b. we divided it: [he to speak to the Spaniards] and I to the English (Daniel Defoe 1719, Robinson Crusoe (1883 Facsimile ed.):2.194) [Jespersen 1946:322]

In Present-Day English, to-infinitives with an overt subject rarely appear without for in sentential subject postion, but Jespersen (1934[1924], 1946) pro-

260

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

vides a few examples from earlier texts, which suggest that a lone subject pronoun once tended to surface in the nominative case in these contexts (148). (148) a. [I to beare this, that neuer knew but better], is some burthen (Shakespeare, Timon of Athens:IV.iii.266) [Jespersen 1946:307]

b. [Thow to lye by our moder] is to muche shame for vs to suffre (Thomas Malory 1889[1485], Morte d’Arthur:453) [Jespersen 1946:307]

c. [She to be his], were hardly less absurd Than that he took her name into his mouth (Robert Browning 1896, Poetical works) [Jespersen 1934[1924]:130]

Since the subject of a to-infinitive is always the highest argument of the verb, the Arg-Case constraint would predict that all infinitive subjects should consistently surface in the nominative case, regardless of the syntactic position of the infinitive and the presence versus absence of the complementizer for. This suggests that the occurrence of objective subjects in to-infinitives must be due to influence from the Pos-Case and Def-Case constraint. As Wurmbrand (2001) demonstrates in great detail, there are good reasons to assume that not all infinitives project the same number of functional layers above the verb phrase.36 To-infinitives introduced by the complementizer for are clearly CPs with the subject in [Spec,TP] at Spell-Out (149). To36

Wurmbrand (2001:100f) suggests that the presence of a TP-layer can be motivated only for infinitives that permit independent temporal reference (i). If we follow Wurmbrand’s proposal, the subject in infinitives without independent tense properties (ii) will occupy the specifier of a functional head lower than T. Since neither T nor any of the lower functional heads above vP are able to check Pos-Case in the absence of CP, the lack of a TP layer in to-infinitives of the kind illustrated in (ii), will not have any bearing on the case of the subject in the approach proposed here. (i)

Examples illustrating the temporal independence of the infinitive following expect a. We expected [him to laugh immediately]. b. We expect [him to have finished the painting by this time tomorrow]. c. We expect [him to be perfect for the part in a few years’ time].

(ii) Examples illustrating the lack of independent temporal reference in to-infinitives following the verb consider a. We consider [him to be a fool]. b. We consider [that he has been a fool for far too long]. c. * We consider [him to have been a fool for too long]. d. We consider [him to be perfect for the part]. e. We consider [that he’ll be perfect for the part in a few years’ time]. f. * We consider [you to be perfect for the part in a few years’ time].

CHAPTER 8

261

infinitives without an overt complementizer are generally analysed as lacking a CP-layer (150). (149)

CP 3 C [- finite] TP for 3 T’ DPi [obj] him 3 FP T [obj] 6 to laughk ti tk

(150)

TP 3 DPi T’ he/him 3 T FP 6 to laughk ti tk

As discussed in Section 2.2.2, the infinitival marker to is best analysed as occupying a position lower than T at Spell-Out, because unlike a finite auxiliary, to must follow not if not is to be interpreted as negating the whole clause (cf. Law 2000:172-177; Wurmbrand 2001:114). The obligatory selection of objective pronoun forms in this context suggests that the presence of the complementizer for endows T with the ability to check objective Pos-Case on a pronoun in its specifier (cf. Section 2.2.2 for a possible alternative analysis of Pos-Case checking involving for). Since T is only able to enter into Pos-Case checking when C is present, the absence of a CP-layer means that the subject of the infinitive will be unable to check Pos-Case unless it raises to a Pos-Case position in the matrix clause before Spell-Out. A suitable matrix position is only available when the to-infinitive appears as the complement of a verb (151). In this context, a (lone) subject pronoun of the infinitive must raise to [Spec,vP] in the matrix clause, and check objective Pos-Case there. Since PosCase overrides Arg-Case in Present-Day English, the the subject will invariably surface in its objective case form. (151)

FP 3 Case checking on the subject of the F vP to-infinitive in the sentence 1 3 vl F DPi [obj] v’ We expected him to laugh. 1 him 3 (twe = trace of the matrix subject) twe v’ Vn v expected 3 VP v [obj] 3 tl V TP 3 tn DP T’ 6 ti to laughk ti tk

262

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Pos-Case checking is not possible for the subjects of absolutive and independent to-infinitives, or to-infinitives in subject position. This means that the surface form of the subject in these contexts will be determined by the interaction of the Def-Case constraint, which calls for objective forms, and the ArgCase constraint, which requires the subject to be nominative. As we will see in Section 9.2, the preference for nominative subject pronouns after fronted auxiliaries in most Present-Day varieties of English suggests that for most speakers of English, Arg-Case outweighs the Def-Case constraint. We would therefore expect to find mainly nominative subjects in independent, absolutive, and subject to-infinitives. The data presented in existing studies indicate that such a preference for nominative subjects can indeed be found in Early Modern English texts (147)-(148). Objective subjects occur primarily in more recent texts (146), which could be seen as evidence that the surface form of subjects in toinfinitives is increasingly influenced by non-case factors, such as the trend towards invariant me, him, her, us, them. 8.8

Pronoun case in small clauses The distribution of pronoun case forms in small clauses is fairly similar to the distribution of pronoun forms in to-infinitives lacking an overt complementizer. Like to-infinitives without for, small clauses rarely appear as the subject of a verb or complement of a preposition (152)-(153). When they do, the subject of the small clause tends to surface in the objective case, although lone unmodified nominatives may occur in subject small clauses. (152) [[He in the front seat] and [she in the back seat]] would be safer. (Andrew Radford, p.c.) [Aarts 1992:185]

(153) a. [Him free] poses a greater risk than [him behind bars]. (Aarts 1992:184)

b. She was distressed at the thought of [him alone in New York] (Kayne 1984a:161n.28)

c. I would have given half I had in the world for [him back again] (Daniel Defoe) [Aarts 1992:43]

In small clauses appearing as the complement of a verb, a lone unmodified subject pronoun will invariably surface in the objective form (154). However, when the subject pronoun appears with a modifier or as part of a coordinate, it may appear in the nominative case (155).

CHAPTER 8

263

(154) a. would you have [me perjure myself]? (Oliver Goldsmith 1889, Globe ed.:623) [Jespersen 1946:287]

b. They consider [him a fool]. c. She had gone of her own free will. Let [her find her own way back] (William Somerset Maugham 1934, Altogether:1518) [Jespersen 1946:477]

d. They saw [us leave]. e. His inconsiderate behaviour made [them angry]. (155) a. mischeefe which may make [this island Thine owne for ever, and [I thy Caliban for aye thy foote-licker]] (Shakespeare, The tempest:IV.i.217) [Jespersen & Haislund 1949:237]

b. Christians fall makes [[Faithful and he] go lovingly together] (John Bunyan 1678, The pilgrim’s progress (1st ed.):85) [Jespersen & Haislund 1949:237]

While lone pronominal subjects always take the objective form when the small clause appears as the complement of a matrix verb, nominatives occur quite readily in independent37 and absolutive small clauses.38 As can be seen from the examples in (156)-(162), the case of the subject does not seem to depend on category of the small clause predicate. (156) Objective subjects in independent and absolutive small clauses with a nominal predicate (examples from Jespersen 1949[1927]:374) a. that I suld live to be ca’d sae, and [me a born servant o’ the house o’ Tillietudlem!] (Walter Scott 1906[1816], Old Mortality:59) b. It would be a wild, presumptious thing, and [him a grand minister!] (James M. Barrie 1893, The little minister:257)

37 Independent small clauses with strongly emotive overtones (i) are generally referred to as Mad Magazine sentences in the literature (cf. Akmajian 1984; Siegel 1987; Schütze 1997).

(i) 38

[Him wear a tuxedo?!] He doesn’t even own a clean shirt. (Akmajian 1984:3)

Jespersen (1934[1924]:128) points out that the subjects of absolutive small clauses originally appeared in an oblique case, just like the subjects of absolutive V-ing constructions, but this oblique case was eventually replaced with the nominative.

264

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(157) Nominative subjects in independent and absolutive small clauses with a nominal predicate a. Who, I rob? [I a theefe]? Not I (Shakespeare, Henry IV, Part 1:I.ii.153) [Jespersen 1946:445]

b. I fail to hold and move One man - and [he my cousin], and [he my friend] (Elizabeth Barrett Browning [1856], Aurora Leigh:154) [Jespersen & Haislund 1949:240]

c. [He a republican]! He scorned the name. [He an enemy of our beloved church]! He esteemed and honoured it. (William M. Thackeray 1901[1853], The Newcomes:802) [Jespersen 1949[1927]:373]

d. [She a beauty]! I should as soon call her mother a wit (Jane Austen 1894[1813], Pride and prejudice:333) [Jespersen & Haislund 1949:239]

(158) Objective subjects in independent and absolutive small clauses with an adjectival predicate (examples from Jespersen & Haislund 1949:239) a. [Me married]? I don’t think (Compton Mackenzie 1922[1912], Carnival:264)

b. …admitted the said Joan to the ranks of Venerable and Blessed. [Me venerable]! (George Bernard Shaw 1924, Saint Joan:110) (159) Nominative subjects in independent and absolutive small clauses with an adjectival predicate a. How can ye chant, ye little birds, And [I sae weary fu’o’ care!] (Robert Burns 1896, Centenary ed.:3.124) [Jespersen 1949[1927]:374]

b. John had seen Glory on the racecourse in Drake’s company: [he proud and triumphant], [she bright and gay and happy] (Hall Caine 1897, The Christian:382) [Jespersen 1946:47]

c. [they dead], two men only would remain (Anthony Hope 1894, The prisoner of Zenda:227) [Jespersen 1946:57]

CHAPTER 8

265

(160) Nominative and objective subjects in independent and absolutive small clauses with a prepositional predicate a. A dead man, and [I by]! (Richard B. Sheridan, Dramatic works:333) [Jespersen 1949[1927]:374]

b. to bolt with the daughter of an old friend and [she only just out of the schoolroom] (William Somerset Maugham, Plays:4.289) [Jespersen & Haislund 1949:240]

c. We sat in the pub, [she at tomato juice] and [me with a brown ale] (Alan Sillitoe 1972[1970], A start in life:151) [Erdmann 1978:69] d. I’m not going to have any woman rummaging about my house, and [me in bed] (Arnold Bennett 1913[1911], The card:188) [Jespersen 1949[1927]:374]

(161) Nominative subjects in independent and absolutive small clauses with a verbal predicate a. [I say anything disrespectful of Dr. Kenn?] Heaven forbid! (George Eliot) [Jespersen 1946:328]

b. The Queen implored pardon … ‘[She ask my pardon], poor woman!’ cried Charles; ‘I ask hers with all my heart.’ (Thomas B. Macaulay, History of England:2.11) [Jespersen 1946:329]

c. Why! They don’t come here to dine you know, they only make believe to dine. [They dine here], Law bless you! (William M. Thackeray [1848-50], The history of Pendennis:2.130) [Jespersen & Haislund 1949:239]

(162) Objective subjects in independent and absolutive small clauses with a verbal predicate (examples (a) and (b) are cited in Jespersen & Haislund 1949:239) a. did you dance with her? - [Me dance]! says Mr. Barnes (William M. Thackeray 1901[1853], The Newcomes:171)

b. What! And [me be left all afternoon by myself]? (Arnold Bennett 1912[1902], Anna of the five towns:175)

c. [Him wear a tuxedo]?! He doesn’t even own a clean shirt. (Akmajian 1984:3)

d. What! [Her call me up]?! Never. (Akmajian 1984:3) e. What! [Us read that trash novel by tomorrow]?! (Akmajian 1984:3)

266

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Stowell (1981, 1995) argues that small clauses are (extended) projections of lexical heads, and that the subject of a small clause occupies the highest specifier position within that projection (see also Akmajian 1984:4-7). Bowers (1993:595f), on the other hand, suggests that small clauses involve a special functional head associated with predication (Pred). In Bower’s analysis, the subject of a small clause appears in [Spec,Pred], while the maximal projection of the lexical predicate appears as the complement of Pred. According to Bowers (1993:596f) the lexical realisation of the head Pred, is the particle as, which occurs in small clauses selected by verbs such as regard and view (163). (163) a. b. c. d.

They regard/view [him as an idiot]. They regard [him as crazy]. They view [him as beyond the pale]. They regard [him as without scruples].

One important property of the small clauses in (163) is that they involve individual-level predication. Ladusaw (1994) and Raposo & Uriagereka (1995: 185f) propose that individual-level predicates occur in categorical statements, whereas stage-level predicates appear in thetic statements.39 According to Raposo & Uriagereka (1995:186f), the difference between categorical and thetic statements is that categorical statements are about a subject (and thus have the effect of ‘topicalising’ the subject), whereas thetic statements are about the predicate. Raposo & Uriagereka (1995:186f) suggest that the difference between categorical and thetic small clauses should be reflected in the syntax. In their approach, categorical small clauses contain a functional head that enters into agreement with the subject of the small clause, while thetic small clauses contain a functional head that enters into agreement with the predicate of the small clause. While the element as could be seen as evidence for the existence of a functional head associated with categorical small clauses, there is little overt evidence in English for the presence of a separate functional head in thetic small clauses. I will therefore assume that only categorical small clauses contain the functional head Pred, with the subject in [Spec,PredP] (164). Thetic small clauses simply consist of the (extended) projection of the lexical head, with the subject in the highest specifier position (165). Categorical small clauses thus follow the small clause structure proposed by Bowers (1993), while 39 I would like to thank Kate Kearns (p.c.) for drawing my attention the correlations between the stage-level/individual-level distinction and the distinction between thetic and categorical statements.

CHAPTER 8

267

thetic small clauses have the small clause structure proposed by Stowell (1981, 1995). (164)

PredP 3 DP Pred’ subject 3 Pred XP

The basic structure of categorical small clauses (i.e. small clauses with an individual-level/property predicate)

(165)

XP 3 DP X’ subject 2 X …

The basic structure of thetic small clauses (i.e. small clauses with a stage-level predicate)

The proposed structural distinction between categorical and thetic small clauses captures the generalisation proposed by Kate Kearns (p.c.), that the presence of Pred is characteristic of categorical/property predication and is incompatible with thetic structures. The structures proposed in (164) and (165) also capture Raposo & Uriagereka’s generalisation that in categorical small clauses the subject has scope over the whole of the predicate, and thus resembles a topicalised constituent, whereas in thetic small clauses, the predicate takes precedence over the subject in that it gives the small clause its category, and is directly selected by the verb (1995:182f). As Raposo & Uriagereka (1995:181,192) point out, many adjectives may function either as individual-level predicates or as stage-level predicates. Depending on its interpretation, an adjectival small clause may therefore either have the structure in (166), or the structure in (167). (166)

PredP 3 DP Pred’ he/him 3 Pred AP (as) 5 crazy

Proposed structure of a small clause with an individual-level adjectival predicate (= categorical)

(167)

AP 3 DP A’ I/me 5 drunk

Proposed structure of a small clause with a stage-level adjectival predicate (= thetic)

Prepositional predicates usually have a stage-level interpretation, which means that prepositional small clauses will tend to have the structure in (168). However, when the predicate of a prepositional small clause expresses a property, the small clause will have the structure in (169).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

268 (168)

(169)

PP 3 DP P’ I 3 P DP at 6 the tiller-lines

PredP 3 DP Pred’ he/him 3 Pred PP (as) 6 beyond the pale without scruples

Proposed structure for a small clause with a stage-level prepositional predicate (= thetic)

Proposed structure for a small clause with an individual-level prepositional predicate (= categorical)

Since nominal predicates generally denote individual-level properties, a nominal small clause will usually be categorical and thus contain a PredP (170). I am assuming that the nominal projection selected by Pred in small clauses is NumP rather than NP. Nominal small clause predicates typically include the indefinite article a, but may also contain a possessive DP (e.g. my). In Section 2.2.3, I argued that cardinal determiners such as a, six, many belong to the category Num, and take NP complements (171). This means that a nominal small clause such as She a beauty! must contain at least a NumP-layer. Further evidence for the presence of a NumP-layer in nominal small clauses comes from the possible occurrence of possessive DPs in the nominal predicate. As discussed in Section 2.2.3, possessive DPs are most plausibly analysed as being base-generated in [Spec,NumP]. If we assume that predicative noun phrases lack a DP-layer, then a possessive DP appearing in a small clause predicate must occupy [Spec,NumP] at Spell-Out (172). (170)

PredP 3 DP Pred’ he/him 3 Pred NumP (as) 5 my friend a fool

(171)

NumP 2 Num NP a 5 beauty

(172)

NumP 2 DP Num’ my 5 friend

Verbal small clauses usually involve thetic predication, and therefore lack a PredP-layer. They may however contain a number of verb-related functional heads. In Section 2.2.1, I argued that the external argument of a causative/agentive verb is base-generated in [Spec,vP]. I also drew attention to the distributional differences between lone pronouns and full noun phrases in Vparticle constructions (173), which can be seen as evidence that pronominal objects raise out of their VP-internal base-position before Spell-Out.

CHAPTER 8

(173) a. b.

269

Betsy threw out her boyfriend. Betsy threw her boyfriend out.

c. * Betsy threw out him. d. Betsy threw him out.

As discussed in Section 2.2.1, I am assuming that the word order in (173d) arises from overt movement of the object pronoun to [Spec,vP], and overt movement of the lexical verb to the head of a functional projection dominating vP. This suggests that the category of a small clause with a causative/agentive verbal predicate will have to be at least vP, and most probably FP (174).40 (174)

FP 3 The structure of the (Mad Magazine) small clause F’ Me throw him out! (cf. Akmajian 1984:4-7) DPi Me 3 F vP 1 3 v’ vl F DPj [obj] 1 him 3 ti v’ Vk v throw 3 TelP v [obj] 3 tl Tel VP out 3 V’ tj 5 tk

The presence of the functional layers within verbal small clauses readily accounts for the possible occurrence of adverbs such as completely (175), and also correctly predicts that lone pronominal objects in a basic verbal small clause will consistently surface in the objective case required by the Pos-Case constraint. 40 Compare Hale & Keyser’s (1998:77fn.4) observation that “the object of the causative verb make is an extended projection of the verbal head”, because sentential negation is possible (i) and because we invariably get the infinitive marker to in the passive (ii). The occurrence of to in the passive but not active version highlights the similarity between to-infinitives and verbal small clauses (cf. Wurmbrand 2001)

(i) We made [John not bake the cake]. (ii) John was made [to bake the cake].

270

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(175) John saw [Mary completely destroy her car]. (Cardinaletti & Guasti 1995:16)

There is however one type of verbal small clause that lacks a vP-layer, and fails to provide the opportunity for objective Pos-Case checking within the small clause: the identificational small clause (176). (176) God saue the king! although I be not hee; And yet, Amen, if Heauen doe think [him mee] (Shakespeare, Richard II:IV.i.174) [Jespersen & Haislund 1949:251]

Although identificational small clauses may look like nominal small clauses at first glance, the second noun phrase in an identificational small clause is referential rather than predicative. The small clause in (176) does not attribute a property to him, but expresses the putative identity between him and mee. Identificational small clauses thus have a thetic rather than categorical interpretation. I will therefore assume that identificational small clauses are VPs headed by an empty counterpart of identificational be (177). (177)

VP 3 DP V’ him 3 V DP mee

The structure of identificational small clauses (= thetic)

The absence of a vP-layer in identificational small clauses means that neither the higher nor the lower argument will be able to check objective Pos-Case within the small clause. As a result, the lower argument in identificational small clauses will be more susceptible to non-case influences than the lower argument of other verbal small clauses. Thus, the use of the nominative he in (178) could be argued to arise from the need to distinguish the identificational small clause from the superficially parallel construction in (179), where the pronouns me and him are objects of wish rather than arguments of an identificational small clause predicate (cf. Jespersen & Haislund 1949:255). (178) And were I any thing but what I am, I would wish [me onley he] (Shakespeare, Coriolanus:I.i.236) [Jespersen & Haislund 1949:255]

(179) If I could wish for anybody, I would wish me only him.

CHAPTER 8

271

While the syntactic differences between basic verbal and identificational small clauses have a bearing on the case properties of any lower arguments in the small clause, the only small clause-internal syntactic factor that could potentially influence the case of small clause subjects is the presence versus absence of the functional head Pred.41 If Pred was analysed as an agreementrelated functional head able to check Pos-Case on a DP in its specifier, we would predict that the subject of a categorical small clause will consistently surface in the Pos-Case checked by Pred, provided we assume that the subject DP raises to [Spec,PredP] from a position within the (extended) projection of the small clause predicate.42 However, there is little evidence for agreement between Pred and the subject of a categorical small clause in English, and the pronoun case data presented above do not provide any evidence for case distinctions between the subjects of categorical and thetic small clauses. This suggests that Pred is unable to check Pos-Case on a DP in its specifier. The surface form of a pronominal subject in any independent or absolutive small clause will therefore be constrained by Def-Case rather than Pos-Case requirements.43 Since all small clause subjects function as the highest argument of a lexical predicate, they will also be linked to nominative Arg-Case. As mentioned in sections 8.6 and 8.7, the relative strength of the Arg-Case and Def-Case constraints in Present-Day English would lead us to expect that nominative subjects should be more strongly favoured in absolutive position than they are in Present-Day English. The ready occurrence of objective subjects in absolutive small clauses could thus be seen as evidence for the trend towards invariant me, him, her, us, them. As can be seen from (180) the subject of a small clause that appears as the complement of a matrix verb will be able to raise to [Spec,vP] in the matrix clause, and check objective Pos-Case there. The possible occurrence of nomi41 Compare Raposo & Uriagereka’s (1995:186f) proposal that the case properties of the different functional heads they propose for thetic and categorical small clauses are responsible for the case differences between categorical and thetic small clauses in Irish. 42 As discussed in Section 2.2.1, arguments are only able to check Pos-Case if their surface position differs from their θ-position. 43 Compare Abney’s (1987:191fn.52) and Schütze’s (1997:52-55) suggestion that the subject of an absolutive small clause appears in the objective default case, and Akmajian’s (1984:3) and Siegel’s (1987:70) proposal that the subject of a Mad Magazine sentence will surface in the objective case, because nominative case is assigned only to noun phrases “preceding a [+ Tense] auxiliary or verb”. The drawback of these analyses is that they are unable to account for the occurrence of nominative subjects in independent and absolutive small clauses. In the approach proposed here, on the other hand, the variation between nominative and objective pronoun forms in this context falls out from the competition between the Arg-Case constraint and the Def-Case constraint.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

272

native forms in this context when the subject of a small clause is coordinated or modified (cf. the examples in (155)), could be seen as evidence that coordinated pronouns and modified pronouns may remain within the small clause and need not raise to [Spec,vP] of the matrix verb. (180)

FP 3 F vP 1 3 v’ vl F DPi [obj] 1 him 3 tsubject v’ Vn v consider 3 VP v [obj] 3 tl V PredP 6 tn ti a fool

8.9

Pronoun case in gapping constructions Pronominal objects in gapped clauses consistently surface in their objective form regardless of whether the gapped clause contains an overt auxiliary (181)-(182) or not (183).44 (181) a. I despise you as heartily as you can __ me (Henry Fielding 1762, Works (2nd ed.):3.534) [Jespersen 1949[1927]:250]

b. Doesn’t it give you a funny feeling …? It does __ me. (Arnold Bennett 1909[1908], Old wives’ tale:1.166) [Jespersen 1949[1927]:249]

c. she ha’s deceued her father, and may __ thee (Shakespeare, Othello:I.iii.294) [Jespersen 1949[1927]:250]

(182) This sword hat ended him, so shall it __ thee (Shakespeare, Henry IV, Part 1:V.iii.9) [Jespersen 1949[1927]:250] (183) That would be a motive for [her murdering him], not [he __ her] (Berkeley Vane Mystery:208) [Jespersen 1946:136]

The case of subject pronouns in gapped clauses depends at least to some extent on the nature of the gapped constituent. When the gapped clause con44 Gapped clauses containing an overt auxiliary verb are generally referred to as pseudogapping in the literature (cf. Lasnik 1995; Johnson 1996:3; Kennedy & Merchant 2000:121f).

CHAPTER 8

273

tains an overt auxiliary, pronominal subjects appear to be obligatorily nominative, no matter whether they precede (184) or follow the auxiliary (185)-(186), and regardless of whether they receive contrastive focus (185) or not (186). (184) as hee tendreth virtue, so he will __ you (John Lyly, Campaspe, in Manly, Specimens of Pre-Shakespearean Drama:278) [Jespersen 1949[1927]:249]

(185) Well, you can do me a good turn, and so can I __ you (Mark Twain 1887[1883], Life on the Mississippi:168) [Jespersen 1949[1927]:250]

(186) We follow’d then our lord, our soueragine king; So should we __ you (Shakespeare, Richard III:I.iii.148) [Jespersen 1949[1927]:250] When the gapped clause contains neither a finite auxiliary, nor a lexical verb, the subject may surface either in the nominative (187) or in the objective case (188). Nominative subjects are not confined to clauses where the gapped verbal complex is finite (187), but also occur quite readily in gapped gerunds (189). While the occurrence of a nominative subject in a gapped gerund may coincide with a nominative in the initial conjunct of the gapping construction (189a), it need not (189c). (187) a. A: I’ll bring the salad. B: And [I __ the wine]. (Chao 1987:20) b. John is fond of them, and [they ___ of him] (Chao 1987:21) (188) Why couldn’t he be my age or [me __ his] (John Fowles 1971[1963], The collector:193) [Erdmann 1978:68]

(189) a. they separated, he driving back to the Temple, and [she __ to her own house] (Arthur Conan Doyle 1893-1905, Adventures of Sherlock Holmes: 36) [Jespersen 1946:47]

b. My mistress being dead, and [I __ once more alone], I had to look out for a new place (Charlotte Brontë 1867[1852], Villette:37) [Jespersen 1946:47]

c. That would be a motive for [her murdering him], not [he __ her] (Berkeley Vane Mystery:208) [Jespersen 1946:136]

274

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Johnson (1996:25f), following Siegel (1987:53-56,71n.2), draws attention to the difference between clauses that lack a finite verb but contain a nonfinite lexical verb (190), and sentences that contain neither a finite nor a nonfinite verb form (191) (cf. also (187)-(189)). According to Johnson (1996:25f), the use of a nominative subject is more degraded when the clause contains a nonfinite main verb (190a), than when no verb is present (190a). (190) a. ?? We can’t eat caviar and he ___ eat beans. b. We can’t eat caviar and him ___ eat beans. (191) a. b.

WE like CAVIAR, and HE ____ BEANS. WE like CAVIAR, and HIM ____ BEANS.

Since the missing predicate in a gapped sentence is reconstructed at a semantic level of representation (cf. Chao 1987:65-74), Arg-Case will always apply to both subjects and objects in gapping constructions. The availability of Pos-Case checking, on the other hand, depends on the surface position of the subject and object at Spell-Out. In so-called pseudo-gapped sentences, where a finite auxiliary verb is present but the lexical verb has been gapped, the subject pronoun is most plausibly analysed as occupying [Spec,TP] at Spell-Out (cf. Kennedy & Merchant 2000:22).45 As discussed in Section 8.1, I am assuming that finite C and T can combine to check nominative Pos-Case on a DP in [Spec,TP] or on a DP in [Spec,CP]. I am also assuming that Pos-Case checking is limited to one specifier position per head or combination of heads in any given derivation. Since T-to-C raising endows a finite C with the ability to check nominative Pos-Case on a DP in [Spec,CP], a DP in [Spec,TP] will only be able to check nominative Pos-Case if the finite verb has not undergone T-toC raising (cf. Sections 9.1 & 9.2). This means that the subject of a pseudogapped clause will only be able to check nominative Pos-Case in sentences such as (192), where the finite auxiliary occupies T at Spell-Out (194). In sentences like (193), where the auxiliary has undergone overt raising to C (195), T will be unable to check Pos-Case on the subject pronoun in [Spec,TP], and the pronoun will be instead receive objective Def-Case as well as nominative ArgCase (cf. Section 9.2).

45 The proposed analysis also fits in with Lasnik’s (1995) assumption that the subject of a pseudo-gapped clause occupies [Spec,AgrSP] at Spell-Out (cf. Section 2.2 for a discussion of case checking with and without Agr projections).

CHAPTER 8

275

(192) as hee tendreth virtue, so he will __ you (John Lyly, Campaspe, in Manly, Specimens of Pre-Shakespearean Drama:278) [Jespersen 1949[1927]:249]

(193) I shall always love you … So shall I __ you (Compton Mackenzie 1922[1912], Carnival:177) [Jespersen 1949[1927]:250]

In Present-Day English, pronominal subjects of pseudo-gapped sentences like (193) consistently surface in the nominative case. This suggests that the ArgCase constraint is stronger than the Def-Case constraint in Present-Day English. It also indicates that lone unmodified pronouns in [Spec,TP] are weak and therefore not subject to influence from any additional non-case constraints on pronoun form (cf. Chapter 9). The apparent obligatoriness of nominative pronoun forms after a fronted auxiliary in both contrastive (185) & (193) and noncontrastive contexts (186) could be seen as supporting evidence for Cardinaletti & Starke’s (1999:163) claim that weak pronouns are able to bear contrastive focus. (194)

CP (195) CP 2 ei XP C’ XP C’ so 3 so wo C TP C TP [+ finite] 3 2 ei DPi [nom] T’ Tk C DPi T’ he 3 1 [+ finite] [no Pos-Case] 1 Vj T I T … T [nom] … 1 shall tk Vk T will

Pseudo-gapping is often analysed as involving a combination of object movement and deletion of VP at PF. Jayaseelan (1990:64-73) and Kennedy & Merchant (2000:121f) assume that the object of a pseudo-gapped clause is right-adjoined to VP before the VP undergoes deletion at PF, while Lasnik (1995) proposes that the object raises to [Spec,AgrOP], which corresponds to [Spec,vP] in the approach adopted here. The right-adjunction analysis would predict that the pronominal object of a pseudo-gapped clause is unable to check Pos-Case, and is instead influenced by objective Def-Case as well as objective Arg-Case. Since a right-adjoined pronominal object does not occupy a surface position associated with the licensing of weak pronouns (cf. Chapter 3), we might also expect its surface

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

276

form to be influenced by non-case factors such as Relative Positional Coding and the trend towards invariant strong forms (cf. Chapter 6). The object-raising analysis (196) would predict that the pronominal object of a pseudo-gapped clause receives objective Pos-Case and Arg-Case (cf. Lasnik 1995).46 Since [Spec,vP] is a licensing position for weak pronouns in English, the pronominal object in (196) could be either weak or strong, and is thus not necessarily influenced by non-case constraints on pronoun form. A pronominal object is also likely to raise to [Spec,vP] before Spell-Out in gapped clauses that contain a nonfinite lexical verb but lack a finite verb (197). The object in gapped clauses like (197) will thus receive objective Pos-Case as well as objective Arg-Case.47 (196)

CP (197) FP 6 2 as you can vP DPi F’ 3 he 3 DPi [obj] v’ F vP me 3 1 3 v’ vl F DPj [obj] v’ tsubject 3 1 them 2 VP Vk v ti v’ v [obj] 2 attack 3 ti V’ v [obj] VP 4 4 tl V tj tk despise

The structure of the pseudo-gapped clause in the sentence I despise you as heartily [as you can me]

The structure of the gapped clause in the sentence She would defend their arguments and [he attack them].

Since there is no evidence for the presence of T at Spell-Out, the subject of a gapped clause such as (197) will be linked to nominative Arg-Case, but will receive objective Def-Case rather than nominative Pos-Case (cf. Chao 1987: 100f; Siegel 1987; Johnson 1996; Kayne 2000:169). As discussed above, ArgCase appears to be more influential than Def-Case in Present-Day English. The variation between nominative and objective subjects in this construction (190) & (197) is thus unlikely to arise from competition between the case constraints 46 Following the convention adopted in Kennedy & Merchant (2000), constituents deleted at PF are indicated with strikethrough in the tree diagram. 47 The structure proposed in (197) is identical to the structure proposed for verbal small clauses (cf. Section 8.8). This similarity between verbal small clauses and gapped clauses like (197) was already noted by Siegel (1987), who argued that gapped clauses containing an overt nonfinite lexical verb have the same syntactic properties as Mad Magazine sentences.

CHAPTER 8

277

alone, but points to additional influence from non-case factors such as the trend towards invariant me, him, her, us, them in strong pronoun contexts. Following Johnson (1996:24,38f), I will assume that subject and object pronouns occupy their θ-positions in gapped clauses containing neither an overt auxiliary nor an overt lexical verb (198).48 Since Pos-Case checking is confined to arguments that have raised out of their θ-positions by Spell-Out, neither the subject nor the object pronoun in (198) is able to check Pos-Case. This means that the subject will receive nominative Arg-Case and objective Def-Case, while the object will receive objective Arg-Case and objective DefCase. The ready use of objective subjects in gapped clauses of this kind, suggests that the surface form of the subject pronoun is further influenced by the general trend towards invariant me, him, her, us, them. (198)

vP The structure of the gapped clause in the sentence 3 DP v’ That would be a motive for [her murdering him], he/him 3 not [he/him __ her] v VP 3 DP V’ her 4 V tk

As mentioned above, Johnson (1996:25f) reports a greater preference for nominative subjects in gapped clauses without an overt verb (199) than in gapped clauses with an overt lexical verb (200). (199) a. b.

WE like CAVIAR, and HE ____ BEANS. WE like CAVIAR, and HIM ____ BEANS.

(200) a. ?? We can’t eat caviar and he ___ eat beans. b. We can’t eat caviar and him ___ eat beans. This case difference is unexpected even if we assume that the case constraints compete with the trend towards invariant strong forms, because the subject would be influenced by Arg-Case, Def-Case, and the Invariant Strong Form constraint in both (199) and (200). However, there is another non-case factor that could be responsible for the greater occurrence of nominatives in gapped 48 I follow Johnson (1996:24) in assuming that the gapped verb has undergone ATB-movement to a functional projection above vP.

278

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

clauses lacking an overt verb (199). In Section 8.8, I argued that the use of a nominative subject in identificational small clauses facilitates the identification of the first noun phrase as the subject of the small clause. Similarly, the use of a nominative subject in gapped clauses without an overt verb will facilitate the identification of the first noun phrase as the subject of a gapped clause.49 In gapped clauses containing an overt lexical verb (200), on the other hand, the subject status of the noun phrase preceding the lexical verb is already indicated by its position relative to the lexical verb, so nominative case is not required for disambiguation purposes. 8.10 Pronoun case in bare argument ellipsis Pronouns in constructions involving bare argument ellipsis (also known as ‘stripping’) generally surface in the objective case in Present-Day English, no matter whether they are interpreted as the subject of a clause (201a-c) or the object of a verb (201d). Nominatives appear (marginally) possible in stripped clauses with an overt subject (201c), but not in stripped clauses with an overt object (201d). (201) a. Yes, I know. And [me] too. (Doris Lessing 1972[1969], The four-gated city:603) [Erdmann 1978:68]

b. Other men might, but not [them]. (Frances Noyes Hart 1929, The Bellamy trial (American ed.):161) [Jespersen & Haislund 1949:275]

c. John might have wanted to leave, but not [me/?I]. (i.e. I didn’t want to leave)

d. Sarah saw Harry at the library, but not [me/*I]. (i.e. Sarah didn’t see me)

According to Moltmann (1992:228), “Bare Argument Ellipsis is a construction in which the coordinator seems to coordinate a single argument with a clause”. Moltmann (1992:22,228) proposes that bare argument ellipsis involves three-dimensional tree structures, where the bare argument has exactly the same syntactic and case status as the associated noun phrase in the preceding clause. While this approach correctly predicts that a pronominal object should obligatorily surface in its objective form in stripped clauses (199d), it is unable

49

Compare Sigursson’s (2003:243) suggestion that morphological case is designed “to make a difference” and renders DPs “more discernable”.

CHAPTER 8

279

to account for the clear preference for objective rather than nominative case forms in stripped clauses with an overt subject (201a-c). Chao (1987:34f) suggests that stripping constructions have a surface representation very similar to the structure of gapping constructions (cf. also Lobeck 1995:27f). In both types of constructions, (the heads of) at least some verbal projections are absent at Spell-Out. However, gapped clauses usually contain at least two constituents (e.g. the subject and the object), whereas stripped clauses contain only one. As discussed in Section 8.9, Johnson (1996) and Kayne (2000) argue that gapped clauses lack certain functional projections (especially TP). Johnson (1996) also proposes that gapping involves ATBmovement of the verb. To account for the absence of overt constituents other than the bare argument in stripping, we could combine Johnson’s (1996) ATBmovement approach to gapping with the VP deletion analysis proposed for pseudo-gapping constructions by Kennedy & Merchant (2000), Jayaseelan (1990), and Lasnik (1995) (see Section 8.9). Stripped clauses with an overt subject (201a-c) could then be analysed as in (202),50 while stripped clauses with an overt object (201d) could be analysed as in (203).51 (202)

vP ei v’ DPsubject I/me 3 v VP 3 (DPobject) V’ 4 V tk

(203)

VP ei DPobject V’ I/me 4 V ti

The analysis proposed in (202)-(203) predicts that overt subjects and objects in a stripped clause will be unable to enter into Pos-Case checking, because all stripped clauses lack a TP-layer,52 and stripped clauses which contain only an overt object also lack a vP-layer. Since LF reconstruction is obligatory for any 50

Although I have offered only examples where no DP object is present in the first conjunct of the stripping construction, bare argument ellipsis with an overt subject is also possible when the first conjunct of the construction contains a DP object. I have therefore included an optional deleted object DP in the tree diagram. 51 I am assuming that in stripping constructions, not has a status similar to that of and and but, and does not form part of the stripped clause (cf. Sections 5.1.3, 8.11, and 8.12 for further discussion of and, but, and not). 52 Cf. Chao’s (1987:100f) suggestion that the absence of the head of TP means that case cannot be checked at Spell-Out.

280

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

predicates that are missing at Spell-Out (cf. Chao 1987:65-74), subjects will still receive nominative Arg-Case, and objects will be linked to objective ArgCase. The preference for objective pronoun forms in stripped clauses with overt subjects indicates that nominative Arg-Case competes not only with objective Def-Case, but also with the trend towards invariant me, him, her, us, them in strong pronoun contexts. 8.11 Pronouns following but, save, except Jespersen & Haislund (1949:228-230) observe that but, save, and except are often followed by a nominative pronoun form, regardless of whether the pronoun is interpreted as a subject (204) or the object of a verb (205). According to Jespersen & Haislund (1949:229), save (and the earlier sauf), only rarely takes the objective case. However, but and except appear quite readily with an objective pronoun form, even when the pronoun is interpreted as the subject of a finite clause (206) (the examples given in (204)-(206) are all cited in Jespersen & Haislund 1949:228-230). (204) a. who sent it but I? (Redford, ‘Wyt and science’, in Manly’s Specimens I:421ff verse 712)

b. he dreamt that all were drowned save he (Hugh Walpole 1918, The green mirror:389)

c. No one heard save she (Arnold Bennett 1912[1910], Clayhanger:1.293) d. And everybody is to know him except I? (George Meredith 1893[1881], The tragic comedians:28)

(205) a. my father had no childe but I (Shakespeare, As you like it:I.ii.18) b. That I kisse aught but he (Shakespeare, Cymbeline:II.iii.153) c. Earth hath swallowed all my hopes but she (Shakespeare, Romeo and Juliet:I.ii.14)

d. What stayes had I but they (Shakespeare, Richard III:II.ii.76) (206) a. Nobody comes here but him for a long time now (John Galsworthy 1918, Five tales:29)

b. I suppose everybody knows but us (Somerset Maugham, Plays, volume 4:107)

c. Perhaps any woman would, except me (Thomas Hardy 1892[1891], Tess of the D’Urbervilles:101)

CHAPTER 8

281

The case variation after but, save, and except is similar to the case variation found in coordinates (cf. Section 5.1). However, the structures involving but, save, and except differ from coordinates with and in that the conjunction and second conjunct may be separated from the initial conjunct.53 What is more, there is usually no plausible ellipsis/conjunction reduction analysis for but, save, except constructions, even though such an analysis is potentially available for coordinates joined by and. Moltmann (1992:373,376) proposes that exception phrases headed by but, save, except are PPs base-generated in an adjoined position to a noun phrase, and may be subsequently extraposed to clause-final position (207). (207)

TP wo Exception structure with a TP PPi subject interpretation ei 2 T P DP DPjsubject 3 1 but me D NumP T … everybody 2 In exception constructions with an object NumP PP interpretation, the PP will be extraposed 4 ti from a noun phrase in [Spec,VP/vP]

Prepositions like but, save, and except do not assign any θ-role to a following noun phrase, which means that neither the Pos-Case nor the Arg-Case constraint impose any case requirements on a pronoun that appears in an exception structure. The choice of pronoun forms in this environment should therefore be determined entirely by Def-Case, and other, non-case related constraints. While the Def-Case constraint predicts that all pronouns in surface positions not covered by Pos-Case should appear in their objective case form, the Relative Positional Coding constraints call for the robust form of a pronoun in asymmetrically c-commanded positions within a phrase, which may explain why we tend to find instances of I after but, save, except. In exception structures where the PP appears adjacent to the noun phrase (208), another possible source of nominative pronoun forms could be a reanalysis of the preposition as a coordinate conjunction. This kind of reanalysis appears to be quite common, and can be seen in the history of the coordinating conjunction and: According to the OED, and derives from the Latin preposition ante ‘before’, and as Gelderen (1997:190-193) observes, and sometimes behaved like a preposition in earlier periods of English (especially in Old Eng53 Interestingly, Old English and displayed many of the properties characteristic of but, save and except today.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

282

lish).54 Similarly, the preposition with appears to be treated as a coordinating conjunction in some contexts and triggers plural agreement with the verb. If save has the status of a coordinating conjunction in examples like (208), the pronoun following save will be linked to nominative Arg-Case, and will also check nominative Pos-Case (209), provided we assume that coordinates are transparent to case (cf. Section 5.1). (208) a. All, save I, were at rest (Mary Shelley 1818, Frankenstein:143) [Jespersen & Haislund 1949:230]

b. Ther every wight save he … was with the leoun frete (Chaucer, Canterbury Tales in Skeat’s six-volume edition, Group B:473) [Jespersen & Haislund 1949:229]

(209)

CP ei C [+finite] TP qp ConjP [nom] T’ 3 3 T [nom] … QP [nom] Conj’ All 3 Conj DP [nom] save I but except

ConjP analysis of an exception phrase adjacent to a subject noun phrase (for a more detailed discussion of the quantifier all, see Section 8.14.3)

The case properties of exception constructions involving but, save, except are quite similar to the case properties of than, as, like comparatives/equatives (cf. Section 5.4). Both types of constructions involve focus prepositions, which do not take arguments, and are unable to check Pos-Case. And in both constructions the pronoun is asymmetrically c-commanded by the preposition, which may explain why the robust 1sg nominative form I is favoured in both contexts. However, comparatives and exception constructions do differ in certain respects. Than, as, like comparatives/equatives can be given an alternative analysis where the following noun phrase is treated as the subject of an ellipted finite clause. Such an ellipsis analysis is not available for exception constructions. On the other hand, but/except/save + pronoun can appear immediately after the associated argument in the matrix clause, and may thus give rise to surface strings that are reminiscent of coordinates. Such a coordinate analysis is considerably less plausible for comparatives and equatives. 54

I would like to thank Kate Kearns (p.c.) for drawing my attention to the etymology of and.

CHAPTER 8

283

8.12 Pronouns following not Jespersen (1946:445) observes that pronouns following not tend to surface in their objective form in emphatic contexts (210). Erdmann (1978:69) reports a similar trend for pronouns after not in contrastive constructions such as (211).55 (210) you were all in the same room together, were not you? ‘No indeed! not us.’ (Jane Austen [1811], Sense and sensibility:269) [Jespersen 1946:446] (211) ‘Your son stole it, not me.’ (Alan Sillitoe 1969[1967], A tree on fire:233) [Erdmann 1978:69]

In contrastive contexts, nominative pronoun forms are most likely to occur after not when the whole construction appears as part of a coordinate preceding the finite verb (212). While the occurrence of the nominative we in (212a) could be argued to be due to agreement between the pronoun and the following verb, the example in (212b) indicates that the use of nominatives after not does not depend on the presence of verb agreement. (212) a. Heauen, and not wee, haue safely fought to day (Shakespeare Henry IV, Part 2:IV.ii.121) [Jespersen & Haislund 1949:140]

b. God, and not wee, hath… (Quarto version of the above quote) [Jespersen & Haislund 1949:140]

In emphatic contexts, nominatives are most common when the antecedent is a nominative pronoun (213).56 However, we also find examples where the antecedent of a nominative pronoun following not, is a proper noun (214).

55 Example (211) bears a strong surface resemblance to bare argument ellipsis (cf. Section 8.9). However, in the absence of the (clausal) conjunction but before not, the pronoun is more likely to be analysed as a simple phrase than as the remnant of ellipsis. The pronoun in (211) would thus appear to have more in common with pronouns in exception constructions (cf. Section 8.11) than with pronouns in stripped clauses. 56 In examples such as (213a), not + pronoun would appear to have a status similar to rightdislocated pronouns (cf. Section 8.3). However, as the punctuation in (213b-c) indicates, the sequence not + pronoun in this context is most likely to be preceded by a noticeable pause. This suggests that not + pronoun generally forms an independent constituent in emphatic contexts.

284

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(213) a. But I didn’t follow his advice, not I (John Galsworthy, Man of property:103) [Visser 1963:56]

b. Were I a Steam-engine, wouldst thou take the trouble to tell lies of me? Not thou! (Thomas Carlyle [1839], Sartor resartus:169) [Jespersen 1946:446]

c. They wouldn’t touch us … Not they (John Galsworthy 1916, The Freelands:255) [Jespersen 1946:446]

(214) Meg don’t know what he likes. Not she! (Charles Dickens 1892, Christmas books:30) [Jespersen 1946:446]

If we assume that not resembles a preposition or conjunction in the above examples, pronouns following not could be analysed either as remnants of extensive ellipsis,57 or as DP complements of not. Any ellipsis approach would predict that a pronoun following not is influenced by Arg-Case, because the missing predicate would be present at Semantic Form. Pronouns interpreted as the subject of an ellipted finite clause could potentially be argued to check nominative Pos-Case with C and T (cf. Section 8.4). However, initial not never cooccurs with a finite verb, and the construction is temporally dependent on the preceding clause. This suggests that constituents following not lack a TP-layer (cf. the proposed structure for stripped clauses in Section 8.10). If no TP-layer is present in a not + pronoun construction, a pronoun interpreted as the subject will be unable to check nominative Pos-Case, and will instead be influenced by the Def-Case constraint, which calls for objective pronoun forms. Since Arg-Case tends to override Def-Case requirements in Present-Day English (cf. Sections 8.9 and 9.2), an ellipsis approach to not + pronoun constructions can readily account for the occurrence of nominative pronoun forms after not in (213)-(214). The use of objective pronoun forms in examples like (210)-(211) is more difficult to predict in an ellipsis approach, and could only result from the interaction of ArgCase and Def-Case with non-case constraints such as the trend towards invariant me, him, her, us, them. If a pronoun following not is analysed as a simple DP, rather than the remnant of ellipsis, the predictions of the three case constraints will depend on whether not + pronoun appears as part of a coordinate (215), or as an independent constituent (216). 57 Compare Jespersen’s (1946:445) observation that expressions like not he are the equivalent of he won’t, he isn’t, with the not negating the unexpressed predicate rather than the pronoun itself.

CHAPTER 8

285

(215) Heauen, and not wee, haue safely fought to day (Shakespeare Henry IV, Part 2:IV.ii.121) [Jespersen & Haislund 1949:140]

(216) a. ‘Your son stole it, not me.’ (Alan Sillitoe 1969[1967], A tree on fire:233) [Erdmann 1978:69]

b. We shan’t hang upon any misunderstanding. Not us (H.G. Wells 1909, Ann Veronica:338) [Jespersen 1946:446]

Assuming that coordinates are transparent to outside case influences (cf. Section 5.1.3), a pronoun following not in a subject coordinate (215) will be linked to nominative Arg-Case and will also check nominative Pos-Case if the subject coordinate occupies [Spec,TP] of a finite clause at Spell-Out. When not + pronoun form an independent constituent (217), the pronoun will receive neither Arg-Case nor Pos-Case, but will instead be influenced by the Def-Case constraint, which calls for objective pronoun forms in all contexts not covered by Pos-Case. (217)

XP 3 X DP not g D I

The syntactic status of a pronoun following not when the construction appears as an idependent constituent

By treating a pronoun following not as a simple DP, we can thus account both for the occurrence of nominative pronoun forms in sentences like (215), and for the use of objective pronoun forms in (216). The nominative forms I, they, and also thou, could be argued to occur after not in (213) because the Relative Positional Coding (RPC) constraints require asymmetrically c-commanded elements to be phonologically robust (cf. Section 6.2). I, they, and thou all contain a diphthong, which renders them more robust than the nominative forms he, she, we, and the objective forms me and thee. We would thus predict that I, they, and thou should be more likely to occur as the complement of not than the less robust forms he, she, we, me, and thee. The data cited in existing studies would seem to confirm this prediction, in that most of the examples of nominatives after not seem to involve the ro-

286

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

bust 1sg form I (cf. Jespersen 1946, Jespersen & Haislund 1949, Visser 1963: 56).58 While influence from the RPC constraints can account for the use of I, thou, and they after not, the occurrence of the gracile nominative he in (218) is entirely unexpected in a DP approach to pronouns after not, even though it is readily captured in an ellipsis approach. This suggests that at least some speakers analyse pronouns after not as remnants of extensive ellipsis rather than DP complements. (218) “Mr Ji-jimmy’s friend will have something worth having to put in his article now,” said Cyril very much later indeed. “Not he!” said Robert sleepily. (Edith Nesbit 1959[1906], The Story of the Amulet:ix) [Denison 1996:291f]

8.13 Pronoun case after only When only + pronoun appears in a canonical argument position, the pronoun always surfaces in the same case as a corresponding lone pronoun (219). When the combination of only + pronoun appears in right-dislocated position, on the other hand, the pronoun will tend to surface in the objective case, no matter what its argument status (220). (219) [Only I] had remembered to bring my notes. (220) Nobody thought much of him, [only me] (P.H. Johnson, 1971[1968], Catherine Carter:102) [Erdmann 1978:69]

Bayer (1999) suggests that focus particles like only may project their own Particle Phrase (PrtP), which either dominates the following DP, or is contained within the DP. When only takes scope over the whole DP, as in the pronoun examples we are concerned with here, it is best analysed as the head of a PrtP dominating DP (221)-(222). The choice of pronoun forms after only suggests that the phrase projected by only is transparent to outside case influences, just like ConjP. The three case constraints will thus make the following predictions for pronoun case after only: A pronoun following only will receive nominative Arg-Case if it is the highest argument of a predicate, and objective ArgCase if it is a lower argument of a verb. The pronoun will also check nomina58 Denison (1996:289-292) suggests that in 19th and early 20th century English, 3ps pronoun may have been more likely to surface in the nominative after not than 1ps pronouns, but he fails to provide convincing evidence in support of this hypothesis.

CHAPTER 8

287

tive Pos-Case requirements when the PrtP appears in [Spec,TP] of a finite clause at Spell-Out (221), and objective Pos-Case when the PrtP occupies [Spec,vP] (222). (221)

CP (222) vP wo wo TP PrtP [obj] v’ C [+ finite] wo 3 2 T’ Prt [obj] DP [obj] v [obj] … PrtP [nom] 3 2 only me Prt [nom] DP [nom] T [nom] … only I

The distribution of only + pronoun in V-particle constructions suggests that a pronoun does not need to raise to [Spec,vP], and indeed preferentially remains in VP-internal position if it is modified by only (223) (cf. Section 2.2.1 for more discussion on V-particle constructions). (223) a. ?? They threw [only me] out. b. They threw out [only me]. When an object pronoun following only remains in [Spec,VP], as in (223b), it will be subject to the Def-Case constraint, which calls for objective forms in all positions not covered by Pos-Case, and the Arg-Case constraint, which requires objects of verbs to surface in their objective form. When only + pronoun appears in right-dislocated position, the pronoun will be unable to check either Pos-Case or Arg-Case, and will only be influenced by the Def-Case constraint, which calls for an objective case form. The inapplicability of Pos-Case and Arg-Case renders the pronoun particularly susceptible to non-case influences, such as the trend towards invariant strong forms (all objective), and Relative Positional Coding constraints, which call for I, him, her, us, them in asymmetrically c-commanded position. 8.14 The case of modified pronouns As mentioned in Chapter 3, one important difference between weak and strong pronouns is the ability of strong pronouns to take modifiers. Just like coordinated pronouns, modified pronouns may exhibit case variation even when they are subject to Pos-Case demands (cf. Denison 1996:294f). This indicates that non-case factors, such as Relative Positional Coding and the trend towards invariant strong forms, have a powerful influence over the surface form of strong pronouns in Modern English. The nature of the modifiers and

288

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

their position relative to the strong pronoun also provide important clues to the internal structure of the DP projected by a strong pronoun. 8.14.1 Pronouns modified by adjectives While all non-neuter pronouns may follow adjectives in English (224), only 1pl and 2pl pronouns are able to precede a modifying adjective (225). The absence of 3pl pronouns in this context is reminiscent of the dispreference for they/them in 3pl-NP constructions (cf. Section 5.2), and may well be due to the fact that the determiner the already provides a 3ps interpretation when used with a following adjective (the uneducated, the poor) (cf. Abney 1987). (224) a. Poor little I / me (Jespersen & Haislund 1949:134) b. Good old you! (Quirk et al. 1985:352) Lucky you! (you = 2sg or 2pl) c. Silly him / her! d. Poor Druids! and Poor us! (Fox 2.147) [Jespersen & Haislund 1949:277f] e. ‘The French haven’t any of our inhibitions about dealing with witnesses.’ ‘Lucky them, my lord.’ (Dorothy L. Sayers 1934, The nine tailors:225) [Jespersen & Haislund 1949:278]

(225) a. we/us uneducated/poor b. you British (you = 2pl) When a pronoun precedes the modifying adjective (225), the adjective is usually interpreted as a restriction on the set of referents picked out by the pronoun (cf. Jespersen 1949[1927]:102). The case form of the pronoun tends to match that of unmodified lone pronouns in the same position. However, as with pronouns modified by a following noun phrase (cf. Section 5.2), we also find instances of us in subject position, and we in objective contexts. Pronouns premodified by an adjective most frequently occur as independent nonsentential constituents (224), but we do find some examples where a premodified pronoun appears as the object of a verb or preposition (226). In both contexts, pronouns tend to appear in their objective case forms in this context, although Jespersen & Haislund (1949:134) seem to suggest that the 1sg nominative I is marginally possible in independent nonsentential constituents (224a).

CHAPTER 8

289

(226) a. Such a lot of love and learning confronting [[poor me], who am so eager to lap it all in comfort] (Ivy Compton-Burnett 1971[1933], More women than men:74) [Erdmann 1978:78]

b. That you will here with [poor us] still remain (George Villiers 1671, The rehearsal:83) [Jespersen & Haislund 1949:134]

Abney (1987:322-334) argues that prenominal adjectives are best analysed as taking the following noun as their complement (227). If we assume that this is the only possible analysis for prenominal adjectives, then the occurrence of adjectives before pronouns in phrases like lucky you and silly me could be seen as evidence that strong pronouns are base-generated in N (cf. Cardinaletti 1994:202-205), and may appear in N at Spell-Out (228). An advantage of the analysis proposed by Cardinaletti (1994) is that it leads us to expect that strong pronouns may be reanalysed as common nouns to yield phrases like a whole new you, an older wiser me, and is it [a he] or [a she]. (227)

DP 3 D NumP the 3 Num AP three 3 A NP hungry caterpillars

(228)

DP 3 D NumP 3 Num AP 3 AA NP lucky them

Since the pronoun in (228) does not appear in D at Spell-Out, we would predict that it will not be as strongly influenced by Pos-Case and Arg-Case, because it is usually only the head of the highest layer of a nominal projection that is directly affected by the different case constraints. If Pos-Case and ArgCase are unable to percolate down to the head of NP, then a pronoun preceded by an adjective will be predicted to surface in the objective form required by the Def-Case constraint. If Pos-Case and Arg-Case are assumed to be able to percolate to the head of NP, then the surface form of the pronoun will depend on the argument structure status of the whole noun phrase, and the position of the DP at Spell-Out. Any analysis where the strong pronoun is assumed to occupy D at SpellOut (cf. Postal 1969; Abney 1987:284; and Longobardi 1994:635) would make the same case predictions as a case percolation account for (228). If a strong

290

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

pronoun always raises from N to D before Spell-Out,59 an adjective preceding a pronoun is most plausibly analysed as occupying [Spec,DP] (229).60 (229)

DP 3 AP D’ poor 3 D NumP 1 5 ti Ni D us

The syntactic status of a pronoun following an adjective, if we assume that strong pronouns raise to D before Spell-Out

All of the analyses discussed above would correctly predict that the pronoun should surface in the objective form when the DP appears as the object of a verb or preposition (226), or as an independent constituent (224). Although the use of premodified pronouns as objects is generally quite marginal, they may appear both before and after the particle in a V-particle construction (230). (230) a. You wouldn’t throw [poor little me] out! b. You wouldn’t throw out [poor little me]! If we assume that a premodified pronoun occupies D at Spell-Out, or that ArgCase and Pos-Case can percolate down to N, then the pronoun will receive objective Arg-Case and Pos-Case when it is an object that has raised to [Spec,vP] (230a), and objective Arg-Case and Def-Case when it is an object in [Spec,VP] (230b) or the complement of a preposition.

59 Longobardi (1994:628f) argues that in English, proper nouns do not undergo overt raising to D, because when a proper noun appears with adjectival modifier, the adjective always precedes rather than follows the noun (i).

(i) a. [Old John] came in. b. * [Johni old ti] came in. However, if we assume that strong pronouns appear in D even when they are premodified by adjectives interpretation, then it seems just as plausible that proper nouns preceded by adjectives should appear in D at Spell-Out. The ungrammaticality of (ia) could then be due to the same semantic factors that rule out the occurrence of singular and 3pl pronouns with postmodifying adjectives (*I old, *he old,* she old,* they old). 60 In bare phrase structure theory, a head will only be able to project a specifier if it also takes a complement (cf. Hale & Keyser 1998). The presence of a premodifying adjective in [Spec,DP] will thus require the concomitant presence of an empty NumP. As discussed in Chapter 3, I assume that the presence of a NumP (and NP) within the DP sets strong pronouns apart from weak pronouns in English.

CHAPTER 8

291

When premodified pronouns form nonsentential constituents, their form will be influenced primarily by the Def-Case constraint, which calls for objective pronoun forms in all positions not covered by Pos-Case. The absence of any other case influences might also lead us to expect that such pronouns may be particularly susceptible to non-case influences such as the trend towards invariant me, him, her, us, them. In an analysis where the pronoun occupies N at Spell-Out and Arg-Case and Pos-Case are unable to percolate down, premodified pronouns in object and prepositional complement positions will have the same case properties as independent premodified pronouns in the other analyses. A pronoun followed by an adjective will check nominative Pos-Case when the whole DP appears in [Spec,TP] of a finite clause at Spell-Out (231), and objective Pos-Case, when it appears in [Spec,vP] (232). The adjective following the pronoun could either be analysed as occupying the head of an APlayer between NumP and NP, as in diagrams (231) and (232), or it could be treated as an adjunct to NumP (which would make its status similar to modifying PPs and restrictive relatives, and arguably also modifying NPs). (231)

CP ei TP C [+ finite] wo DP [nom] T’ ei 3 NumP T [nom] … D [nom] we 2 Num AP 5 uneducated

(232)

vP wo DP [obj] v’ 3 3 D [obj] NumP v [obj] … us 2 Num AP 5 uneducated

The word order in V-particle constructions suggests that pronouns followed by an adjective need not raise to [Spec,vP] before Spell-Out (233). If the DP headed by the pronoun remains in [Spec,VP], the surface form of the pronoun will be constrained by Arg-Case and Def-Case, but not by Pos-Case. (233) a. Society has always tried to shut [us underprivileged] out. b. Society has always tried to shut out [us underprivileged]. 8.14.2 Pronouns followed by a numeral In most of the examples listed by Jespersen & Haislund (1949:141) the pronoun bears the same case as it would if it was unmodified. Thus, objects of prepositions (234) and subjects of to-infinitives and small clauses (235) surface

292

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

in their objective forms,61 while subjects of finite clauses appear in the nominative (236). However we also find instances of objective pronoun forms when the pronoun appears as the subject of a finite clause (237). (234) a. I am boy to [them all three], but all they three…could not be man to me (Shakespeare Henry V:III.ii.30) [Jespersen & Haislund 1949:141] b. the fayrest of [them thre] (William Caxton [1481], Reynard the fox:83) [Jespersen & Haislund 1949:141]

(235) a. I want [us three] to meet, you and she and I (Oppenheim, Pawns count:88) [Jespersen & Haislund 1949:141]

b. I want to see [them two] meet (George Bernard Shaw 1907, John Bull’s other island:225) [Jespersen & Haislund 1949:141]

(236) And in our sight [they three] were taken vp (Shakespeare, Comedy of errors:I.i.111) [Jespersen & Haislund 1949:141]

(237) [Us two] will lead the way (William Pett Ridge [1907], Name of Garland:138) [Jespersen & Haislund 1949:276]

When a pronoun modified by a numeral appears in right-dislocation, it may surface in its nominative form, even if its antecedent is the (objective) subject of a small clause following the verb let (238). (238) a. let us not be ashamed, [we two], but only very proud (Alfred Sutro 1912, Five little plays:96) [Jespersen & Haislund 1949:238]

b. to persuade Richard to let us go alone - [we three], you know (J.D. Beresford, Mount. Moon:6) [Jespersen & Haislund 1949:238]

As discussed in Section 2.2.3, I am following Ritter (1991:50-58) and Lobeck (1995:80) in assuming that numerals head the functional projection NumP, which intervenes between DP and NP. When a pronoun is modified by a numeral, the pronoun always precedes the numeral (239). This suggests that pronouns modified by numerals occupy D at Spell-Out.

61 For a more detailed discussion of pronoun case in to-infinitives and small clauses, see Sections 8.7 and 8.8, respectively.

CHAPTER 8

(239) a.

we three

293

b. * three we

Since case always percolates to the head of the DP in a case checking position, the pronoun in D will be influenced by Arg-Case if the DP functions as the argument of a predicate. The Arg-Case constraint predicts that any subject will surface in the nominative, whereas the object of a verb or preposition will surface in the objective case. If the whole DP appears in a Pos-Case position at Spell-Out, the pronoun will also be subject to Pos-Case requirements (240)(241). (240)

CP ei TP C [+ finite] wo DP [nom] T’ ei 3 D [nom] NumP T [nom] … we three

(241)

vP wo DP [obj] v’ 3 3 D [obj] NumP v [obj] … us three

Unlike lone object pronouns, pronouns modified by a numeral may either precede or follow the particle in a V-particle constructions (242), which suggests that a pronoun modified by a numeral may remain in [Spec,VP] at Spell-Out. (242) a. You wouldn’t throw [us three] out, would you? b. You wouldn’t throw out [us three], would you? When a pronoun modified by a numeral appears in [Spec,VP], or as the complement of a preposition, it will be influenced by Def-Case rather than PosCase, and might therefore be more susceptible to non-case influences such as the trend towards invariant us and them and the preference for we and they in asymmetrically c-commanding positions. Jespersen & Haislund (1949:141f) note that them + numeral “is now generally avoided [in favour of these/those/the+numeral], because them is felt to be like the vg [= vulgar] use of them as an adjunct (them boys)”. Speakers also seem to favour those in contexts where 3pl is modified by a restrictive relative clause (cf. Section 8.14.7). As discussed in Section 5.2, the case differences between 1pl and 3pl pronouns in pronoun-NP constructions, and the preference for 3pl demonstratives rather than personal pronoun forms in this context suggest that modified 1pl and 3pl pronouns occupy different syntactic positions: while 1pl pronouns head the whole construction, 3pl pronouns/demonstratives appear in the specifier of the DP. Such an analysis could also account for trend towards 3pl demonstratives before numerals and restrictive relatives.

294

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

8.14.3 Pronouns associated with the quantifiers all and both When all or both follow a pronoun appearing as the subject of a finite clause, the pronoun generally surfaces in the nominative case (243). A pronoun followed by all or both which appears as the object of a preposition will usually surface in its objective form (244). (243) We all have our loyalties. (Quirk et al. 1985:353) (244) I am boy to them all three (Shakespeare Henry V:III.ii.30) [Jespersen & Haislund 1949:141]

Unlike lone unmodified pronouns, personal pronouns modified by a numeral, noun, or prepositional phrase, may follow rather than precede the quantifiers all and both (245)-(247). (245) [Both us parents] were invited to the meeting. (246) I am boy to them all three, but [all they three] … could not be man to me (Shakespeare Henry V:III.ii.30) [Jespersen & Haislund 1949:141] (247) all the foure brethern, and [all theym of theyr companye] arayed themselfe (Caxton, Aymon:78) [Mustanoja 1960:129f] The examples in (245)-(247) suggest that modified pronouns preceded by all may surface either in the nominative or in the objective case when they appear as the subject of a finite clause. It is important to bear in mind, though, that (247) is an example from Middle English, and also differs from (246) in that the quantifier-pronoun complex forms the second conjunct of a coordinate rather than appearing by itself. When a pronoun associated with a quantifier appears as the subject of a V-ing construction, it may surface either in the genitive case (248) or in its nominative/objective form (249). (248) a. Em’ly was confused by [our all observing her] (Charles Dickens 1897[1849-50], David Copperfield: 133) [Jespersen 1946:129]

b. assurances of [their both being alive] (Jane Austen 1897[1814], Mansfield Park:29) [Jespersen 1946:129]

CHAPTER 8

295

(249) Can I count on [you all holding your tongues]? (William Somerset Maugham 1934, Altogether (Collected stories):1485) [Jespersen 1946:135]

Potsdam (1998:88-92), following Sportiche (1988), suggests that noun phrases associated with the quantifiers all and both are base-generated as a complement of the quantifier. Word order facts indicate that full noun phrases (250) and modified pronouns (251) may remain in post-Q position, but lone pronouns must raise past Q before Spell-Out. According to Potsdam (1998: 92), the lowest surface position available to lone pronouns associated with a quantifier is [Spec,QP] (252).62 (250)

QP (251) QP 3 3 Q DP Q DP all 6 all 6 them three both my brothers

(252)

theym of theyr companye us postgrads

QP 3 Q DPi you 3 Q DP all ti both

62 As Liz Pearce (p.c.) points out, the pronoun does not necessarily form a constituent with the quantifier at Spell-Out in (248)-(249). When perfective have is added to the V-ing construction, it may follow the quantifier (i), as would be expected if the pronoun and quantifier formed a constituent, but it may also intervene between the pronoun and the quantifier (ii).

(i)

… by [our all having observed her] …from [our both having addressed the same lady] … of [their both having been alive]

(ii) … by [our having all observed her] … from [our having both addressed the same lady] … of [their having both been alive] I assume that pronouns forming a constituent with a following quantifier (i) occupy [Spec,QP] at Spell-Out (252), while pronouns separated from a following quantifier by intervening projections (ii) have raised out of QP before Spell-Out. Their ability to form a constituent with an associated pronoun at Spell-Out sets all and both apart from each, which must float (iii)-(v). (iii) They all took a candle. They both took a candle. They each took a candle. (Quirk et al. 1985:353). (iv) We’ve contacted [them all] / [them both] / *[them each]. (iii)

[All us postgrads] are going to the conference. [Both us parents] were invited to the meeting. * [Each us witnesses] were interviewed separately.

296

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

When a pronoun appears within QP at Spell-Out (251)-(252), it will only be able to check Pos-Case if we assume that QP is entirely transparent to outside case influences. In order to establish whether QP is transparent to Pos-Case or not, we will need to compare the case preferences for pronouns that have raised out of QP with the case preferences for pronouns that clearly appear within QP. If QP is transparent to Pos-Case, we would expect a modified subject pronoun to be just as likely to surface in its nominative form when it appears after all (253) as when it has raised to [Spec,TP] by itself (254). (253) a. [All we postgrads] are going to the meeting. b. [All us postgrads] are going to the meeting. (254) a. [We postgrads] are all going to the meeting. b. [Us postgrads] are all going to the meeting. While I did not include examples like (253) in my survey, the use of we after all in (253a) strikes me as considerably less idiomatic than the use of we in (254a), where the pronoun has raised out of [Spec,QP] into [Spec,TP]. This would suggest that QP is not (entirely) transparent to Pos-Case influences. At the same time, a pronoun within QP might be expected be able to receive ArgCase, because it is the pronominal DP that functions as the argument of the predicate. As illustrated by (254), a pronoun associated with a quantifier need not be adjacent to this quantifier at Spell-Out. This suggests that pronouns are able to move out of [Spec,QP] before Spell-Out. If a pronoun moves to the specifier of an agreement-related functional head, its surface form will be constrained by Pos-Case as well as Arg-Case, because it will be in a direct spec-head relationship with the functional head. A pronoun that has moved from [Spec,QP] to [Spec,TP] in a finite clause will check nominative Pos-Case; a pronoun that has moved to [Spec,vP] will check objective Pos-Case; and a pronoun that has moved from [Spec,QP] to [Spec,DP] in a Poss-ing gerund will check genitive Pos-Case. 8.14.4 Pronouns followed by a PP Quirk et al. (1985:353) suggest that nominative pronoun forms may alternate with objective forms when a pronoun is modified by a PP (255).

CHAPTER 8

297

(255) a. we of the modern age b. us over here (annotated as ‘familiar’) (Quirk et al. 1985:353) Unfortunately Quirk et al. (1985) do not provide a context illustrating the position of the modified pronoun in the clause, but it seems likely that the annotation ‘familiar’ with (255b) refers to the use of us over here as the subject of a finite clause. Further examples cited in the literature suggest that pronouns modified by a PP may surface in the nominative case if they appear in apposition to a noun phrase (256), while objective forms occur when the pronoun + PP complex appears as the subject of a past-participle construction (257). (256) I heard one of my examiners - [he of the braided surtout] - whisper to his co-professor (Charlotte Brontë 1867[1852], Villette:386) [Jespersen & Haislund 1949:225f]

(257) the most maddening of masters [[him before me] always excepted] (Charlotte Brontë 1867[1852], Villette:343) [Jespersen 1946:49]

Rothstein (1988) observes that PPs modifying a noun or pronoun may be either ‘restrictive’ (255b) & (257), or ‘adjectival’ (255a) & (256). The paucity of examples in the literature makes it difficult to determine whether the nature of the modifying PP has any bearing on the case of the pronoun preceding it, but it is interesting to note that the nominative pronouns in (255a) and (256) appear with an ‘adjectival’ PP, while the objective pronouns in (255b) and (257) are modified by a ‘restrictive’ PP. According to Lobeck (1995:79) PPs modifying a noun are either adjoined to NumP or to NP. ‘Restrictive’ PPs, which can only occur within definite noun phrases (258), adjoin to NumP (260). ‘Adjectival’ PPs, which occur with both definite and indefinite noun phrase (259), are adjoined to NP (261). (258) a. He brought me [the/every glass on the table]. b. * He brought me [a/some glass on the table]. (259) a. He brought me [the/every ladybird with blue wings]. b. He brought me [a/some ladybird with blue wings].

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

298 (260)

DP 3 D NumP the 3 NumP PP 5 5 glass on the table

(261)

DP 3 D NumP (the) 3 Num NP a 3 NP PP ladybird 6 with blue wings

Since both restrictive and adjectival PPs appear as adjuncts to a projection below D, the difference between the two types of PP will not have any bearing on pronoun case predictions. The case of a pronoun modified by a PP will be influenced by the Arg-Case constraint whenever the whole DP functions as the argument of a predicate, and by Pos-Case, if the DP appears in a position covered by Pos-Case, such as [Spec,TP]. A pronoun modified by a PP will be influenced by nominative Arg-Case and Pos-Case when it functions as the (preverbal) subject of a finite clause. If the pronoun appears as the subject of an absolutive past-participle construction (257), it will receive nominative Arg-Case, but it will be unable to enter into Pos-Case checking, and will instead be influenced by objective Def-Case. The word order in V-particle constructions (262) suggests that object pronouns modified by a PP tend to remain in [Spec,VP]. (262) a.

I notice that you didn’t throw out [him with the expensive sunglasses].

b.?? I notice that you didn’t throw [him with the expensive sunglasses] out. This means that pronouns modified by a PP will be influenced by objective Def-Case rather than objective Pos-Case, when they appear as the object of a verb. The same applies to modified pronouns appearing as the complement of a preposition. An appositive pronoun modified by a PP (256) could be argued to inherit the case of the head it is adjoined to, through semantically licensed case agreement (cf. Sections 8.1, 8.2, and 9.7). The use of he before a modifying PP may be further reinforced by the preference for the gracile pronoun forms me, he, she, we, they in asymmetrically c-commanding positions. Since a pronoun in D asymmetrically c-commands any modifying PP, the Relative Positional Coding constraints will predict that the 3sg pronoun in (256) to surface in the gracile form he, rather than the robust form him (see Section 6.2 for a more detailed discussion of the gracile/robust distinction)

CHAPTER 8

299

The case-based approach outlined here correctly predicts the use of the nominative form we when (255a) appears the subject of a finite clause, and it can also capture the occurrence of the nominative he in (256). However, the three case constraints alone are unable to account for the use of objective pronoun forms in contexts where the pronoun modified by a PP functions as the subject of a finite (255b) or non-finite clause (257). Since the influence of ArgCase generally outweighs the influence of Def-Case in Present-Day English, we would expect all pronominal subjects to surface in their nominative form. The use of objective pronoun forms in subject position thus indicates that pronouns modified by a PP are influenced by additional non-case constraints, such as the trend towards invariant me, him, her, us, them. 8.14.5 Pronouns followed by a self-reflexive As noted by Jespersen & Haislund (1949:225), pronouns followed by a self-reflexive tend to surface in the nominative, regardless of their position in the sentence. Thus we find instances of nominatives followed by a self-reflexive in absolutive small clauses (263a) and in the focus of it-clefts where the relativised constituent in the clause is the object of a verb (263b). Instances of objective pronoun forms followed by a self-reflexive are comparatively rare, and appear to be confined to contexts traditionally associated with objective case (264). (263) a. And shall the figure of God’s Maiestie … Be iudg’d by subiect, and inferior breathe, And [he himself] not present? (Shakespeare, Richard II:IV.i.129) [Jespersen & Haislund 1949:239f]

b. It was [she herself] he wanted to see. It was [he himself] he wanted to see. (Jespersen & Haislund 1949:255) (264) No one goes in there without an invitation from [him himself] (W.B. Maxwell 1926, Gabrielle: 92) [Jespersen & Haislund 1949:171]

If we assume that strong pronouns are base-generated in N (cf. Chapter 3), then the cooccurrence of a pronoun and a reflexive in the same DP could be seen as the result of Copy Spell Out (cf. Grohmann 2003:112-116). The selfreflexive following a pronoun would thus appear in N at Spell-Out, while the pronoun appears in D (265).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

300 (265)

DP ei D NumP 1 3 N D Num NP ! hei N himselfi

Copy Spell-Out analysis of a pronoun modified by a self-reflexive

Under this analysis, we would expect pronouns followed by self-reflexives to be subject to the same case influences as an unmodified lone pronoun. That is, a pronoun modified by a self-reflexive will receive nominative Arg-Case if it functions as the highest argument of a predicate, and objective Arg-Case if it is a lower argument of a verb or preposition. If the DP containing the pronoun and self-reflexive appears in [Spec,TP], the pronoun will also be able to check nominative Pos-Case, and if the DP appears in [Spec,vP], the pronoun will check objective Pos-Case. In all other syntactic positions, the pronoun will be influenced by the Def-Case constraint, which calls for objective pronoun forms in Modern English. The interaction of the three case constraints can account for the occurrence of the objective him in (264) and also for the occurrence of the nominative he in (263a), provided we assume that Arg-Case outweighs the Def-Case constraint. However, the use of he and she in the focus of the it-clefts in (263b) cannot be captured in a purely case-based approach, because the only case influence on the pronoun is the Def-Case constraint, which calls for objective pronoun forms. This suggests that the surface form of pronouns followed by a self-reflexive is at least partly influenced by non-case factors. Since the pronoun asymmetrically c-commands the self-reflexive in (265), we could see the occurrence of he and she in (263b) as further evidence that he and she are associated with asymmetrically c-commanding positions. Another possible factor influencing the surface form of a pronoun followed by a corresponding reflexive is suggested by Jespersen & Haislund (1949:225), who comment that such a pronoun will “probably always have the nominative in order to avoid collocations like him himself, her herself”. 8.14.6 Pronouns followed by an appositive Kjellmer (1986:447) argues that pronouns will always surface in the same case as corresponding lone pronouns when they are modified by an appositive noun phrase (266).

CHAPTER 8

301

(266) a. [We, (who are) the parents of these children], think that … b. * [Us, (who are) the parents of these children], think that … However, while we might argue that the use of the nominative he in (267) arises from a general preference for nominative subjects in independent to-infinitives (cf. Section 8.7), we also find instances of nominative pronouns modified by an appositive in prepositional complement position (268), where lone pronouns generally surface in their objective form. (267) Sir Jee was taken aback. [He, the chairman of the borough Bench, and the leading philanthropist in the country, to be so spoken to!] (Arnold Bennett 1924, The grim smile of the five towns:145) [Jespersen 1946:329]

(268) Finally, I must speak of the frustrations of [we, the journalists] (The Observer, 12 August 1984) [Kjellmer 1986:448]

Appositive noun phrases modifying a pronoun are most plausibly analysed as DP-adjuncts, which means that they will have the same syntactic relationship with the pronoun as a non-restrictive relative clause (cf. Section 8.14.7). If we adopt the analysis of pronoun-NP constructions proposed in Section 5.2, the difference between appositive and restrictive noun phrases modifying the pronoun will be similar to the difference between non-restrictive and restrictive relative clauses. Non-restrictive relatives and appositive DPs adjoin to DP (269), while NPs and relatives with a restrictive interpretation adjoin to NumP or NP. (269)

DP ei DP DP we 6

The syntactic status of a pronoun modified by an appositive noun phrase

the journalists

Since the pronoun in (269) heads the DP modified by the appositive, a purely case-based approach would predict that it should surface in the same case form as an unmodified pronoun. Thus, we would expect pronouns modified by appositives to receive nominative Arg-Case when they occupy the highest position on the argument hierarchy of a predicate, and objective Arg-Case when they are lower arguments of a verb or preposition. When the pronoun + appositive appears in [Spec,TP], it will be influenced by nominative Pos-Case; when it appears in [Spec,vP], it will check objective Pos-Case; and when it appears in

302

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

[Spec,DP], it will check genitive Pos-Case. In any positions not covered by Pos-Case, the pronoun will be subject to the Def-Case constraint, which calls for objective pronoun forms. Given that Pos-Case generally outweighs other case constraints in Present-Day English (cf. Section 2.3), the preference for objective pronouns in gerunds such as (270) suggests that pronouns modified by an appositive noun phrase are unlikely to raise to [Spec,DP] (cf. Section 8.6). The use of objective pronoun forms in this context also points to the influence of additional noncase constraints, such as the trend towards invariant me, him, her, us, them, because objective Def-Case alone is usually unable to override nominative ArgCase. (270) What were the chances of [him, Sam, getting back to work]? (Arnold Bennett 1926, Lord Raingo:351) [Jespersen 1946:134]

8.14.7 Pronouns followed by a relative clause The case of a pronoun modified by a relative clause appears to be influenced primarily by the status of the pronoun in the matrix clause. However, some of the evidence presented in existing studies suggests that the surface form of the pronoun may also be influenced by the function of the relativised constituent in the clause. When both the pronoun and the relativised constituent in the clause function as the subject of a finite clause, the pronoun will generally surface in its nominative form, no matter whether the relative clause is restrictive (271)(272) or non-restrictive (273), and regardless of whether the relative clause is introduced by a wh-pronoun (271) & (273) or the complementizer that (272) (cf. Jespersen 1949[1927]:97-103; and Kjellmer 1986:447). (271) a. [He who can’t keep a penny] will never have many (Jespersen & Haislund 1949:133)

b. [We who are about him], have done our part (Charles Dickens 1887[1848], Dombey and son:389) [Jespersen 1949[1927]:103]

c. [They who cannot forget Gordon] must always be grateful to Tennyson (Andrew Lang 1904, Tennyson:189) [Jespersen 1949[1927]:98]

CHAPTER 8

303

(272) a. [he that fights and runs away] may live to fight another day (Jespersen 1949[1927]:97)

b. [we that are true louers], runne into strange capers (Shakespeare, As you like it:II.iv.54) [Jespersen 1949[1927]:102]

c. It was too true, though [they that say so] knew nothing of the matter (Daniel Defoe [1722], Journal of the plague year:43) [Jespersen 1949[1927]:98]

(273) a. he gave in .. [He, who had never looked strong nor well], looked ghastly now (Beerbohm, Seven men:21) [Jespersen 1949[1927]:99] b. But [she, who had never felt these mad, amazing impulses], could nevertheless only smile fearfully (Arnold Bennett 1909[1908], Old wives’ tale:1.23) [Jespersen 1949[1927]:100]

Objective pronoun forms are favoured when the relativised constituent in the clause is the object of a pied-piped preposition, and the pronoun itself appears as the complement of a preposition (274). (274) giuen to [them for whom it is prepared] (Authorised version of the Bible 1611 (1833 Facsimile ed.):Mark 10.40) [Jespersen 1949[1927]:192]

According to Jespersen & Haislund (1949:227) a case clash between the pronoun and the relativised constituent in the clause is usually resolved in favour of objective case forms. Thus, objective forms are favoured when the relativised constituent in the clause is the object of a verb or preposition, and the pronoun appears in postverbal subject position (275) or in apposition to the subject of the matrix clause (276). (275) Before her, in the arms of death, lay [him on whom her hopes of happiness seemed to have formed so firm a basis] (Percy Bysshe Shelley 1912[c1820], Prose works:1.96) [Visser 1963:248]

(276) a. thou didst say that Kallikrates - [him whom thou sawest dead] was thine ancestor (Rider Haggard 1896[1887], She:246) [Jespersen & Haislund 1949:226]

b. Our noble Arthur, [him Ye scarce can overpraise], will hear and know (Alfred Tennyson 1894, Poetical works (in one volume):370) [Jespersen & Haislund 1949:226]

304

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

A pronoun modified by a contact relative where the relativised constituent is the object of a verb (276b) & (277), is likely to surface in the objective form even when it appears as the subject of the matrix clause (277). And independent 1sg pronouns are likely to surface in the objective form me, even when the relativised constituent is the subject of a finite clause (278). (277) a. for learn this, Silius, Better to leave undone than by our deed Acquire too high a fame when [him we serve]’s away (Shakespeare, Antony and Cleopatra:III.i.13) [Jespersen & Haislund 1949:226]

b. [her I loue] now Doth grace for grace, and loue for loue allow63 (Shakespeare, Romeo and Juliet:II.iii.86) [Jespersen & Haislund 1949:226]

(278) [Me that’s led such a quiet life]! (J. Wain 1969[1953], Hurry on down:101) [Erdmann 1978:69]

Jespersen & Haislund (1949:226) suggest that a case clash between nominative and objective case in a relative construction is rarely resolved in favour of the nominative. However, we do find instances of variation between nominative and objective forms, where the relativised constituent is the subject of a finite clause, but the pronoun modified by the relative is the object of a preposition (279) or verb (280). (279) a. Everything comes to [he who waits]. (Corby trouser-press advertisement, London Underground, 7 November 1994) [Wales 1996:96]

b. You think you’ll stop him giving it to [them that have a right to look to him]? (Mrs. Humphrey Ward 1892, David Grieve:3.226) [Jespersen 1946:149]

(280) Praise [him that got thee], [shee that gaue thee sucke] (Shakespeare, Troilus and Cressida:II.iii.252) [Jespersen & Haislund 1949:227]

Topicalised object pronouns are particularly likely to surface in the nominative case when they are modified by a relative clause (cf. Section 8.1), no matter whether the relativised constituent is the subject of a finite clause (281), the object of a verb or preposition (282), or a possessive (283).

63 According to Jespersen & Haislund (1949:226), the oldest quarto has she whom in this sentence.

CHAPTER 8

305

(281) [She, who had been the bane of his life]i … he treated ti with the respect a good son might offer a kind mother (Charlotte Brontë 1867[1852], Villette, London:378) [Jespersen & Haislund 1949:225]

(282) a. [She whom thine eie shall like]i, thy heart shall haue ti (Christopher Marlowe, Doctor Faustus:594) [Jespersen & Haislund 1949:224]

b. [She in whom I might have inspired a dearer love]i, I had taught ti to be my sister (Charles Dickens 1897[1849-50], David Copperfield: 775) [Jespersen & Haislund 1949:224f]

(283) [She, whose happiness you most desire]i, you choose ti to be your victim (Robert Louis Stevenson 1894[1881], Virginibus puerisque:31) [Jespersen & Haislund 1949:225]

Restrictive relatives modifying a pronoun are most plausibly analysed as NP- or NumP-adjuncts (284) & (291), while non-restrictive relatives are best treated as DP-adjuncts (285) (cf. Section 9.7). (284)

DP (285) 3 D NumP 3 she Num NP ei NP CP 4 6

DP ei DP CP she 6

who had been the bane of his life

whom thine eie shall like

Since the pronoun heads the DP in (284)-(285), and also (291), it will receive nominative Arg-Case if it functions as the highest argument of a matrix predicate, and objective Arg-Case otherwise. When the DP appears in the specifier of an appropriate agreement-related functional head at Spell-Out, the pronoun will also be able to enter into Pos-Case checking. Subject pronouns modified by a relative clause will check nominative Pos-Case with T if they appear in [Spec,TP] at Spell-Out. As can be seen from (286) object pronouns modified by a relative clause generally follow the particle in V-particle constructions, and are thus most plausibly analysed as occupying [Spec,VP] rather than [Spec,vP] at Spell-Out. (286) a. She has thrown out [him who has always supported her]. b.*? She has thrown [him who as always supported her] out.

306

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

This means that an object pronoun modified by a relative clause will be unable to check objective Pos-Case, and will instead be influenced by the Def-Case constraint, which calls for objective forms in all contexts not covered by the Pos-Case constraint. A purely case-based approach would thus predict that pronouns modified by a relative clause should surface in the nominative case when they appear as the (preverbal) subject of a finite clause, and in the objective case when they appear as the complement of a verb or preposition. The occurrence of examples like (287) indicates that unlike lone subject pronouns, subject pronouns modified by a relative clause do not have to raise to [Spec,TP] before Spell-Out. (287) Before her, in the arms of death, lay [him on whom her hopes of happiness seemed to have formed so firm a basis] (Percy Bysshe Shelley 1912[c1820], Prose works:1.96) [Visser 1963:248]

Since the lexical verb is unable to raise to T or C in Present-Day English, the subject pronoun in (287) is best analysed as occupying its base-position in [Spec,VP].64 As discussed in Chapter 2, a DP can only enter into Pos-Case checking if its surface position is different from its θ-position. A subject that remains in [Spec,VP] or [Spec,vP], where it receives its θ-role, will therefore be unable to check Pos-Case. This means that its surface form will be influenced by the Arg-Case constraint, which requires subjects to be nominative, and the Def-Case constraint, which calls for objective forms. We could thus account for the use of him in (287) by assuming that the Def-Case constraint exceptionally overrides the Arg-Case constraint in this instance.65 Alternatively, we might argue that the use of the objective form is due to Arg-Case agreement between the pronoun and the relativised constituent in the clause (cf. Bianchi 2000:59; Alexiadou et al. 2000:3).66 Since the wh-pronoun in (287) 64 As Liz Pearce (p.c.) points out, lie is most plausibly treated as an unaccusative verb, which means that its highest argument is base-generated in [Spec,VP] rather than [Spec,vP] (see Section 2.2.1 for further discussion). 65 As we will see in Section 9.2, Def-Case appears to have been more influential than Arg-Case in Early Modern English. Since archaisms are often a feature of poetic language it is not entirely implausible that a poet might imitate a case pattern characteristic of earlier stages of the language. 66 As Delahunty (1982:214-217) points out, wh-pronouns and operators in relative clauses are variables that must be bound at a semantic level of representation. The binding relationship between a wh-pronoun/operator and its antecedent effectively links the antecedent to a position on the argument hierarchy of a predicate in the relative clause, which makes it plausible to assume that the antecedent could surface in the Arg-Case associated with the open position in the relative clause (cf. Section 5.3 for further discussion).

CHAPTER 8

307

functions as the object of a preposition, the Arg-Case inherited by its antecedent would be objective. Although the preverbal subject pronouns in (288) could also be argued to inherit objective Arg-Case through agreement with the relativised constituent in the clause, mere Arg-Case agreement would not seem to be sufficient to account for the occurrence of objective pronoun forms in preverbal position. (288) a. [him whom thou sawest dead] - was thine ancestor (Rider Haggard 1896[1887], She:246) [Visser 1963:248]

b. [Them she lived with] would have killed her for a hat-pin (George Bernard Shaw [1912], Pygmalion:III.iii) [Visser 1963:248]

Since Pos-Case generally overrides Arg-Case in Present-Day English (see Section 2.3), we would expect subject pronouns in [Spec,TP] to surface in their nominative form even when they receive objective Arg-Case through Arg-Case agreement. This suggests that the use of him and them in (288) is at least partly due to non-case factors such as the trend towards invariant him and them. The occurrence of nominative pronoun forms in sentences like (289) could be analysed as the result of Arg-Case agreement between the personal pronoun and the wh-subject, but no purely case-based approach can predict the use of nominatives when both the pronoun and the relativised constituent in the clause function as the object of a verb or preposition (290). (289) Everything comes to [he who waits]. (Corby trouser-press advertisement, London Underground, 7 November 1994) [Wales 1996:96]

(290) [She whom thine eie shall like], thy heart shall haue (Christopher Marlowe, Doctor Faustus:594) [Jespersen & Haislund 1949:224]

This suggests that the selection of the nominative rather than objective form when a pronoun is modified by a relative clause is at least partly due to factors other than case. As can be seen from the tree diagrams in (284) & (291), the pronoun asymmetrically c-commands any restrictive relative clause modifying it. The occurrence of non-1sg nominative forms in this context could thus be seen as further evidence that he, she, we, they are favoured in asymmetrically c-commanding positions. The preference for me over I in (292) would seem to provide further evidence that the 1sg objective form me patterns with the non1sg nominatives in this respect.

308

(291)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

DP The syntactic status of a pronoun modified by a ei D NumP restrictive relative (in a NumP-adjunct analysis) she ei NumP CP 4 6 whom thine eie shall like

(292) [Me that’s led such a quiet life]! (J. Wain 1969[1953], Hurry on down:101) [Erdmann 1978:69]

8.15 Summary and conclusions The pronoun case data discussed in this chapter offer further insights into the factors that lead to pronoun case variation in strong pronoun contexts. Differences in pronoun case choice generally correlate with differences in syntactic status, and variation often occurs in configurations that can be given more than one structural analysis. While alternative structural analyses and/or parameter settings could plausibly account for some of the pronoun case trends reported in existing studies, many of the data cited are not easily captured in a purely case-based approach. The pronoun case trends discussed in this chapter indicate that the structural position of a pronoun at Spell-Out has an important bearing on the case form it surfaces in. The data also point to a decline in the importance of grammatical relations (i.e. subject, object) in case marking. The argument structure status of a pronoun (or structure containing a pronoun) clearly still has some bearing on pronoun case choice, but only unmodified, phonologically reducible lone pronouns in finite declarative clauses consistently surface in the case form corresponding to their grammatical relation in the sentence. In Chapter 10, I will argue that the importance of Positional Case and Relative Positional Coding in Modern English are both by-products of the shift to the exclusive positional licensing of arguments during the Middle English period. The inadequacy of purely case-based approaches when it comes to accounting for the distribution of pronoun forms in strong pronoun contexts should not be entirely surprising. After all, strong pronouns tend to appear in positions that are to some extent separated from the agreement-related functional heads associated with Pos-Case checking (C and T for subjects, v for objects). Thus, pronoun case often varies when the pronoun occupies a surface position other than [Spec,TP], [Spec,vP], or [Spec,NumP]. For pronouns occupying [Spec,CP], case variation occurs primarily when C is unable to check Pos-Case on its specifier, i.e. when T has failed to raise to C, and an overt con-

CHAPTER 8

309

stituent intervenes between C and T at Spell-Out. Thus, we find case variation with topicalised pronouns and left-dislocated pronouns, as well as pronouns in the focus position of it-clefts (cf. Chapter 5) and pronouns following focus prepositions. Pronoun case variation also occurs when a pronoun is embedded in a construction that lacks the relevant agreement-related functional head required for Pos-Case checking. Thus we find case variation in the subject position of absolutive/independent V-ing constructions, absolutive/independent to-infinitives, and absolutive/independent small clauses. However, pronoun case variation does not only arise in contexts where Pos-Case fails to apply. The case patterns found with modified and conjoined pronouns indicate that pronouns in these contexts are influenced by Pos-Case requirements, but nevertheless exhibit case variation. This suggests that the distribution of strong personal pronoun forms in Modern English is heavily influenced by non-case factors such as Relative Positional Coding and the trend towards invariant strong forms. As we will see in Chapter 9, the distribution of wh-pronoun forms provides further evidence for the interaction between case and non-case factors in determining the case morphology of pronouns in Modern English.

CHAPTER 9 THE DISTRIBUTION OF WH-PRONOUN FORMS IN MODERN ENGLISH

9.0 Introduction Probably the most advanced development in the pronoun case system of Modern English is the spread of the nominative wh-form who to contexts that used to require the objective form whom (cf. Sigley 1997:67f; Lasnik & Sobin 2000). According to Jespersen & Haislund (1949:241ff), the use of who as the object of a verb or stranded preposition is extremely common from the (late) Middle English period onwards. However, the extension of who did not immediately apply to all objective wh-contexts. As Traugott (1972:125ff) points out, questions like (1a) and (1b) are attested as early as 1500 (cf. also Mustanoja 1960:181), but the use of who instead of whom in relatives such as (2a) and (2b) is a more recent development (cf. Householder 1986:149-156). Thus, the same author may use who in interrogatives questioning the object of a verb or stranded preposition, but still favour whom in a corresponding relative, as in Table 9.1 (Klima 1964:3f; Traugott 1972:127; Householder 1986:149). (1) a. Who did you see? (who = object of the verb see) [Traugott 1972:125] b. Who did you talk to? (who = object of the stranded preposition to) (2) a. Could you identify the man [who you’d seen outside the building]? b. I saw the man [who you told me about]. [Traugott 1972:126] matrix questions Who could she see? Who did he speak with? With whom did he speak?

embedded questions He knew [who she could see]. He knew [who he spoke with]. He knew [with whom he spoke].

headed relative clauses The leader [whom she could see] left. The leader [whom he spoke with] left. The leader [with whom he spoke] left. Table 9.1. The case forms of interrogative versus relative wh-pronouns in an intermediate style proposed by Klima (1964:3fn.3) and Householder (1986:149)

CHAPTER 9

311

According to Householder (1986:152), the distribution of who and whom in Shakespeare’s works indicates that Shakespeare was a speaker of the intermediate dialect illustrated in Table 9.1. The use of wh-pronouns in headed relatives is a comparatively recent development (cf. Section 9.7). In Middle English, headed relatives tended to be introduced by the complementizers ðe and ðæt (cf. Allen 1980:202-206; and Gelderen 2004:81-87). Wh-pronouns first appeared in non-subject relatives, and Shakespeare still exhibits a clear preference for that in restrictive subject relatives. In Present-Day English, restrictive subject relatives may be introduced either by a wh-pronoun or by that. When the relativised constituent is the object of a verb or a stranded preposition, on the other hand, the relative clause tends to occur without an overt relative marker (3) (cf. Sigley 1997:273 for detailed figures from the Wellington Corpus of Written New Zealand English). (3) a. Could you identify the man [you’d seen outside the building]. b. I saw the man [you told me about]. The different histories of wh-pronouns in relative clauses and questions are likely to have contributed to the case differences between relative and interrogative wh-pronouns reported by Klima (1964), Traugott (1972), and Householder (1986). The alternation between who and whom in questions and relatives contrasts sharply with the obligatory occurrence of the genitive whose whenever the wh-pronoun modifies a noun in a noun phrase (4). (4) a. [Whose/*who/*whom novel] did they publish? b. I don’t know [[whose/*who/*whom novel] they published]. c. the writer [[whose/*who/*whom novel] they published] In the approach proposed here, the consistent selection of whose in (4a-c) falls out from the convergence of Positional Case (Pos-Case) and Argument Case (Arg-Case) requirements (cf. Section 2.3). The Arg-Case constraint requires the wh-pronoun to surface in its genitive form because it is not the highest structural argument of the noun novel.1 For DPs that occupy [Spec,DP] at 1 The highest structural argument of a noun is referential, and is usually θ-bound by a nounrelated functional category. Since θ-binding does not have any effect on the feature specification of the remaining arguments, any structural argument resulting from the projection of the TELIC or AGENTIVE quale of the noun will bear the feature [+ higher], even when the referential argument is not structurally realised (cf. Section 2.1.1 for further discussion).

312

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Spell-Out, the genitive form imposed by the Arg-Case constraint is further reinforced by genitive Pos-Case checking with D (5).2 (5)

DP 3 DPi [gen] D’ whose 3 D [gen] NumP 6 ti novel

Genitive Pos-Case checking on a wh-phrase in [Spec,DP]

Since this analysis can account for the distribution of whose in the various whconstructions discussed here, the remainder of this chapter will focus on the distribution of who and whom. As mentioned above, variation between who and whom first occurred in interrogative contexts. I will therefore look at pronoun case trends in different types of wh-questions in Sections 9.1 to 9.5, before considering the distribution of wh-forms in free and headed relatives in Sections 9.6 and 9.7. 9.1

wh-pronouns in matrix questions Both traditional grammars and empirical studies of wh-forms in Modern English agree that who is obligatory in matrix interrogatives questioning the subject of a finite clause (6). (6) Who/*whom won the race? However, there is a noticeable divergence between prescriptive norms and actual usage in wh-interrogatives questioning the object of a verb or preposition. While prescriptive grammars tend to dictate the use of the objective wh-form whom whenever the wh-pronoun functions as the object of a verb (7a) or preposition (7b), naturally occurring instances of whom in matrix questions are extremely rare in Present-Day English (cf. Klima 1964:1-4; Householder 1986: 155f; Lasnik & Sobin 2000:343-347). (7) a. Whom did you see? b. Whom did he speak with? / With whom did he arrive? 2

As mentioned in Section 2.2.3, I am assuming that definite noun phrases contain both a NumP- and a DP-layer, in addition to the NP core. Since a DP is only able to check Pos-Case when its surface position differs from its θ-position, the wh-pronoun must have raised to [Spec,DP] from a position within NumP, in order to be able to check genitive Pos-Case with D.

CHAPTER 9

313

As Lasnik & Sobin (2000:353,356) point out, the interrogative pronoun most readily surfaces as whom when it occurs with a pied-piped preposition (8), and when it immediately follows a verb or preposition in multiple wh-questions (9). (8) Who does this shop belong to? I mean [to whom does this shop belong]? (George Bernard Shaw 1935, Too true to be good:154) [Jespersen 1946:484] (9) a. Who saw whom? b. Who spoke with whom? c. Who considers whom (to be) underpaid? (Lasnik & Sobin 2000:353,356) Who is clearly favoured in simple matrix interrogatives questioning the object of a verb (10) or the object of a stranded preposition (8) & (11). (10) who can he mean by that? (Richard B. Sheridan, Dramatic works:48) [Jespersen & Haislund 1949:242]

(11) a. Who does it come from? … Do you know who it is from? (Oliver Goldsmith 1889, Globe ed.:668) [Jespersen 1946:484]

b. Who have you smil’d with? (John Keats 1900, The complete works:5.180) [Jespersen 1946:484]

As discussed in Chapter 2, all structural arguments of a predicate are affected by the Arg-Case constraint, which requires the overt case-form of an argument to be compatible with its relative position on the argument hierarchy. The Arg-Case constraint makes the following case-predictions for wh-pronouns in matrix questions: When the questioned constituent is the subject of a finite clause, the wh-pronoun must surface in the nominative form who (12), because the subject is the highest structural argument of verb. When the questioned constituent is the object of a verb or preposition (and thus not the highest argument of a predicate), the wh-pronoun must surface in the objective form whom (13)-(14), because the objective is the only case available to lower arguments of [- N] predicates. (12) Who won the race? (13) Whom did you see? (14) Whom did she speak with? / With whom did she speak?

314

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The Arg-Case constraint thus predicts that the distribution of wh-forms in matrix questions will always follow the usage prescribed in traditional grammars. It is unable to account for the general trend towards who wherever the wh-pronoun is not immediately preceded by a verb or preposition, as in (8), (10) & (11). The obvious relevance of surface position to the case of wh-pronouns points to the involvement of the Pos-Case constraint and the Default Case constraint (Def-Case), both of which make an explicit link between case and surface position. Pos-Case restricts the case forms of arguments appearing in the specifier of certain agreement-related functional heads at Spell-Out. Def-Case requires overtly-case marked noun phrases to surface in their objective form when they are unable to enter into Pos-Case checking. Since Def-Case predicts the occurrence of the objective form whom wherever it applies, we will have to rely on Pos-Case to account for the popularity of initial who in matrix questions if we want to maintain a purely case-based approach to the distribution of wh-forms. Sentence-initial wh-phrases are generally assumed to occupy [Spec,CP] at Spell-Out. In matrix interrogatives questioning the object of a verb or preposition, the wh-pronoun is always followed by a finite auxiliary (e.g. did), which is assumed to have undergone T-to-C raising (15). In subject questions, no Tto-C raising takes place, and C remains empty (16) (cf. Rizzi 1996:63-69). (15)

CP 3 C’ DPi who 3 C TP 1 3 T’ Tk C DP did you 6 tk……ti….

(16)

CP 2 DPi C’ who 2 C TP 2 DP T’ ti 2 T …

If we want to account for the occurrence of who in (15) and (16) in terms of Pos-Case requirements, we will have to argue that finite C is able to check nominative case on a DP in [Spec,CP]. The association between finite C and nominative case checking was already noted in Chapter 2, where finite C was assumed to endow T with the ability to check nominative Pos-Case on a DP in [Spec,TP]. If we assume that nominative case checking always involves a collaboration between finite C and T, we would predict that an argument DP in [Spec,CP] should be able check nominative Pos-Case when T has raised to C

CHAPTER 9

315

(17).3 The consistent selection of who in subject questions indicates that finite C also acquires the ability to check nominative case on its specifier, when no overt constituent intervenes between C and T at Spell-Out (18).4 (17)

CP 3 DPi [nom] C’ who 3 C [nom] …ti… 1 Tk C [+ finite] did

(18)

CP 3 DPi [nom] C’ who 3 C [nom] TP [+ finite] 3 ! DP T’ 3 ! ti z----- T …

Further support for this analysis comes from the case of subject pronouns that follow the finite auxiliary in matrix questions and other types of clauses that involve T-to-C raising. 9.2

The case of subject pronouns after fronted auxiliaries As noted in Section 1.4, 1pl subject pronouns sometimes surface in their objective form (u)s when they follow a fronted auxiliary in Early Modern English questions (19), exhortatives (20), and sentences containing a preposed adverbial (21). (19) Say, where shall’s lay him (Shakespeare, Cymbeline:IV.ii.233) [Gelderen 1997:68]

(20) Shall’s geld him. (Francis Beaumont and John Fletcher 1905[1607-11], ed. Glover and Waller:1.139) [Jespersen & Haislund 1949:256]

(21) hens must vs flee (The Towneley Plays, ed. England, EETS 1897:31) [Jespersen & Haislund 1949:256]

3

Compare Jespersen & Haislund’s (1949:243) suggestion that “the tendency to replace whom by who” is particularly strong in (matrix) questions, because the wh-pronoun is immediately followed by the verb. 4 Following suggestions by Bobaljik (1994:1f,7f) and Bošković (2000:75), we could argue that the surface adjacency between C and T leads to a PF merger of the features contained in these two heads, and thus endows C with the ability to check nominative Pos-Case. (PF = Phonological Form)

316

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

A similar trend can be found in varieties of English spoken in the south-west of England (cf. Ihalainen 1991:106; 1994:230f). According to Ihalainen (1991: 106), 1pl and 3pl pronouns consistently surface in their nominative forms we and they when they appear as the (preverbal) subject of a declarative sentence in Somerset English. In questions, on the other hand, objective forms such as us and ’em “are almost the rule” for subject pronouns (22). Thomas Hardy’s novels, which are set in the south-west of England, contain examples of objective 1pl and 3pl forms in exactly the same context (23). (22) a. Didn’s get it? (Ihalainen 1994:231) b. We don’t know, do us? (Wales 1996:91) (23) a. Let’s look into Warren’s, shall us, neighbours? (Thomas Hardy 1906, Far from the madding crowd:438) [Jespersen & Haislund 1949:257]

b. They move his soul; don’t ’em, father? (Thomas Hardy, Under the greenwood tree:70) [Jespersen & Haislund 1949:257]

As discussed in Section 2.2.2, nominative Pos-Case checking always involves a collaboration between finite C and T in the approach proposed here. In Section 9.1, I argued that T-to-C raising endows a finite C with the ability to check nominative Pos-Case on a DP in [Spec,CP]. If we assume that Pos-Case checking is limited to one specifier position per head or combination of heads in any given derivation,5 then T-to-C raising will simultaneously result in the loss of nominative Pos-Case checking between T and a DP in [Spec,TP] (24).6 5

The proposed approach to case checking is more restrictive than the approach outlined in Chomsky (1995:286), which explicitly allows for the possibility that a head is involved in multiple instances of case checking. 6 For a similar proposal, see Gelderen (1997). Gelderen (1997:67f) argues that the occurrence of objective rather than nominative subject pronouns in (19)-(20), is due to the absence of an overt spec-head relationship between the subject pronoun and the finite verb. She also suggests that a finite verb that has raised to C will enter into feature-checking with the wh-pronoun in [Spec,CP] (1997:71). There are, however, two important differences between Gelderen’s approach and my analysis: (a) Gelderen (1997:67,71) assumes that the verb in C only checks verbal agreement with a whpronoun in [Spec,CP], whereas I am arguing that T-to-C raising endows C with the ability to check nominative Pos-Case on its specifier. (b) In Gelderen’s approach, the verb in C checks the case of the subject pronoun under government (1997:67f), whereas in my analysis, a subject pronoun in [Spec,TP] is unable to enter into any case checking relationship with C or T once T has raised to C.

CHAPTER 9

(24)

317

CP ei The consequences of T-to-C raising for DP [nom] C’ Pos-Case checking within CP and TP who wo C [nom] TP 1 ei DP T’ Tk C did [+ finite] [no Pos-case] 2 you T … tk

It is important to note that any raising of T to C will lead to the loss of PosCase checking between T and [Spec,TP], even when no actual case checking takes place between C and [Spec,CP]. A subject pronoun that follows a fronted finite auxiliary at Spell-Out will always be unable to check Pos-Case, no matter whether [Spec,CP] is (a)

filled with a DP (24),

(b)

filled with an XP that may or may not be able to enter into Pos-Case checking (25),7 or

(c)

completely absent at Spell-Out (26).8 (25)

CP ei XP C’ where ei how C TP hens 2 2 Tk C DP T’ shall [+ finite] us 2 must T … tk

7

(26)

CP ei C TP 2 2 C DP T’ Tk shall [+ finite] us 2 don’t ’em T … tk

Note that at least some of the fronted elements included in this group could be analysed as ‘bare-NP adverbs’ (cf. Larson 1985:612f). If we adopt a ‘bare-NP adverb’ analysis for these initial constituents, and follow Larson (1985:605f,620) and Przepiórkowski (1998) in assuming that adverbial noun phrases should be treated as optional structural arguments of the verb, then they could be argued to check nominative Pos-Case with C, like the wh-pronoun in (24). 8 I am assuming that an agreement-related functional head will only be involved in Pos-Case checking if its specifier is occupied by an argument noun phrase at Spell-Out which has raised out of its θ-position (cf. Section 2.3). If no suitable noun phrase occupies the specifier position, the Pos-Case constraint is not applicable, and has no bearing on the convergence of the derivation. The analysis proposed here thus differs from a minimalist approach to case and agreement checking, where T must be involved in case checking in order for the derivation to converge. Thanks to Liz Pearce (p.c.) for drawing my attention to this issue.

318

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Since a pronoun in [Spec,TP] is unable to check Pos-Case when T appears in C at Spell-Out, the surface form of a subject pronoun that follows a fronted finite auxiliary will be influenced by Def-Case and Arg-Case. The ArgCase constraint predicts that the subject pronoun should be nominative, because it is the highest argument of the predicate, whereas Def-Case requires the pronoun to surface in its objective form. The occurrence of us and ’em after the fronted auxiliaries in (19)-(23) could be captured by assuming that the Def-Case constraint is more influential than the Arg-Case constraint in Early Modern English and South-Western English. In most varieties of Present-Day English, on the other hand, Arg-Case would appear to outweigh Def-Case, and lone subject pronouns consistently surface in their nominative case forms, even when they follow a fronted auxiliary (27). (27) a. Who did she talk to? b. Where shall we put him? c. They move your soul, don’t they? 9.3

wh-pronouns in echo questions As noted in Section 9.1, the objective wh-form whom is most likely to occur after a lexical verb or preposition in matrix questions. However, this does not mean that whom is obligatory in these contexts. According to Gelderen (1997:81f), the use of who after verbs and prepositions is particularly common in echo questions (28)-(29). This preference for who is evident both in echo questions asking for clarification (28a) & (29a), and in echo questions expressing surprise and disbelief (28b) & (29b). (28) a. I’ve seen her. - Seen who? b. I saw Elvis yesterday. - You saw who? (29) a. What do you think of him? - “Think of who?” inquired Mrs. Squeers; who (as she often remarked) was no grammarian, thank Heaven (Charles Dickens 1900[1839], Nicholas Nickleby:62) [Jespersen 1946:484f]

b. What did you do with my lecture notes? - I gave them to Jason. You gave them to who? The occurrence of the nominative who after verbs and prepositions is difficult to account for in a purely case-based approach. None of the three case

CHAPTER 9

319

constraints introduced in Chapter 2 is able to predict the occurrence of a nominative form in this environment. The Arg-Case constraint predicts that any object of a verb or preposition should surface in the objective case. The Pos-Case constraint predicts that any wh-pronoun in [Spec,vP] will check objective case. The Def-Case constraint predicts that any wh-pronoun in a position not covered by Pos-Case will surface in the objective form whom. The popularity of who in echo questions thus suggests that the distribution of wh-forms is at least partly influenced by factors other than case. Foremost among these factors appears to be a tendency towards the use of who in all wh-positions (cf. Gelderen 1997: 81f; Lasnik & Sobin 2000). If we assume that the surface form of a pronoun is determined by the interaction of all constraints that apply in a given context, then we would expect the influence of non-case factors to be particularly pronounced in positions where case is less influential. The pronoun data discussed in Chapter 2 and Sections 9.1-9.2, indicate that Arg-Case and Def-Case are both weaker than Pos-Case in Present-Day English. Pronouns that appear in positions covered only by Arg-Case and/or Def-Case, should therefore be more susceptible to non-case influences than pronouns in Pos-Case positions. As mentioned in Section 2.2.4, prepositional phrases do not contain any agreement-related functional head that could check Pos-Case on the object of the preposition. Any object of a preposition that appears within its PP at Spell-Out, will therefore be affected only by Arg-Case and Def-Case requirements, and may be influenced by the general tendency towards who. The ready occurrence of who after verbs as well as prepositions in echo questions, suggests that a wh-pronoun in post-verbal position is just as susceptible to non-case influences as the object of a preposition. This similarity between wh-objects of verbs and prepositions would seem to indicate that postverbal wh-pronouns do not occupy a Pos-Case position in echo questions. In Section 2.2.1, I argued that any (non-deictic) pronominal object of a verb raises to [Spec,vP] before Spell-Out, and checks objective Pos-Case in this position. The primary evidence for the overt movement of pronominal objects came from V-particle constructions: while full noun phrases can appear both before and after the particle, lone pronouns are generally confined to pre-particle position (30) (examples adapted from Johnson 1991:593f). (30) a. Betsy threw out her boyfriend. b. Betsy threw her boyfriend out.

c. * Betsy threw out him. d. Betsy threw him out.

320

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

If we assume that the particle heads its own phrase (TelP) which intervenes between vP nd VP at Spell-Out, then any phrase that precedes the particle must have raised at least as high as [Spec,vP].9 Phrases that follow the particle, on the other hand, will occupy a VP-internal position, and will be unable to check objective Pos-Case. As (31) illustrates, wh-pronouns in echo questions tend to appear after the particle in a V-particle construction.10 This suggests that whpronouns do not undergo obligatory movement to [Spec,vP] in echo questions, and are thus able to appear in a surface position not covered by Pos-Case (32).11 (31) a. She’s thrown him out. - Thrown out who? b. We looked up Rob and Kim on the way back. - You looked up who? (32)

FP 3 F vP 1 3 v’ vl F e 1 3 Vk v v TelP 3 thrown tl Tel VP out 2 DP V’ who 4 tk

9.4

wh-pronouns in embedded questions The distribution of wh-forms in embedded questions strongly resembles the distribution of who and whom in matrix interrogatives. Thus, who appears to be obligatory when the questioned constituent is the subject of a finite clause (33), and readily occurs as the fronted object of a verb (34) or stranded preposition (35). 9

As mentioned in Section 2.2.1, I am assuming that Tel is unable to project a specifier, which means that TelP cannot serve as a possible landing site for object movement. 10 Thrown who out? seems possible in (31a), but Thrown out who? feels more natural. Liz Pearce (p.c.) notes that this word order preference extends to multiple echo questions. Thus, (ii) is a more plausible echo question than (i). (i) Who threw whom out? [list/information question] (ii) ? Who threw out who? [echo/contrast question] 11

Compare Gelderen’s (1997:82) suggestion that whom is more likely to appear when the whpronoun moves through [Spec,AgrOP] before Spell-Out.

CHAPTER 9

321

(33) I don’t know [who / *whom won the race]. (34) We want to know [who you’ve got in this house, at present] (Charles Dickens 1890[1837-38], Pickwick Papers:101) [Jespersen 1946:496]

(35) a. now I see [who he laughed at] (Ben Jonson, the Mermaid Series:1.17) [Jespersen & Haislund 1949:242]

b. Who does it come from? … Do you know [who it is from]? (Oliver Goldsmith 1889, Globe ed.:668) [Jespersen 1946:484]

c. how can he tell [who it was intended for]? (George Bernard Shaw 1898, Plays pleasant:84) [Jespersen & Haislund 1949:242]

Whom is most likely to appear immediately after a preposition, either in a piedpiping construction (36), or when the PP appears in situ in a multiple wh-question (37). (36) a. If you did know [to whom I gave the ring], If you did know, [for whom I gave the ring] (Shakespeare, The merchant of Venice:V.i.193) [Jespersen 1946:484]

b. it is only a question of [with whom I shall do so] (Grant Allen 1895, The woman who did:81) [Jespersen 1949[1927]:48]

(37) to know … [who was smitten with whom at Vienna] (William M. Thackeray 1901[1853], The Newcomes:586) [Jespersen 1946:494]

While matrix wh-questions with whom in initial position appear to be extremely rare in Modern English, initial whom does occur in some of the embedded wh-interrogatives cited in existing studies. Jespersen (1949[1927], 1946) offers examples of initial whom in embedded questions with preposition stranding (38), and in embedded questions involving identificational be (39).12 (38) the thought of [whom it hath recourse to] (John Milton, Areopagitica:1) [Jespersen 1949[1927]:48]

12 As noted in Sections 8.5, identificational be takes two individual arguments. Since the embedded questions in (39) target the lower of the two arguments, the wh-pronoun is followed by the subject rather than the verb.

322

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(39) a. then I know [whom you are] (Hall Caine 1897, The Christian:422) [Jespersen 1946:483f]

b. She did not know [whom this strange young man might be] (Hugh Walpole [1913], Fortitude:138) [Jespersen 1946:484]

Arg-Case predictions for wh-case in embedded questions are identical to Arg-Case predictions for wh-case in matrix questions: When the questioned constituent is the subject, and thus the highest argument of the verb, the whpronoun will surface in the nominative form who (40). When the questioned constituent is the object of a verb or preposition, the wh-pronoun will take the objective form whom (41). The same goes for the lower argument of identificational be (42a), although we could argue that the coreferentiality between the two arguments in an identificational sentence could lead to nominative ArgCase agreement between the higher and the lower argument (42b).13 (40) I don’t know [who won the race]. (41) a. Find out [whom she likes]! b. Do you know [with whom he left]? c. Do you know [whom he left with]? (42) a. I know [whom he is]. b. I know [who he is]. Unlike the Arg-Case constraint, the Pos-Case constraint makes quite different predictions for the distribution of wh-forms in matrix and in embedded questions. The main difference between matrix questions and embedded questions is the relative order of wh-pronoun, finite verb, and subject. While the wh-pronoun is always followed by a finite auxiliary in a matrix interrogative questioning the lower argument of a verb or preposition (43), the subject intervenes between the wh-pronoun and the finite verb in a corresponding embedded question (44). Only subject questions exhibit the same word order in matrix and embedded contexts (45).

13 As discussed in Section 8.5, the present-day preference for who in identificational questions is more plausibly analysed as arising from the general trend towards invariant who. If the use of who was due to Arg-Case agreement with the higher argument of be, we also would expect to find a preference for nominative pronoun forms after be in all declarative identificational sentences. Since there is little evidence for such Arg-Case agreement in Present-Day English, the use of who in (42b) is likely to be triggered by something else.

CHAPTER 9

323

(43) Who did you see? Who are you? Who did you talk to? (44) a. Tell them [who you saw]! b. I know [who you are]. c. I know [who you talked to]. (45) a. Who won the race? b. Do you know [who won the race]? In Section 9.1, I argued that the preference for initial who in matrix questions arises from nominative Pos-Case checking between finite C and the whpronoun in [Spec,CP]. Finite C can acquire the ability to check nominative Pos-Case on its specifier through T-to-C raising (in matrix interrogatives questioning the object of a verb or preposition), or through surface adjacency between C and T (in subject questions). As (46) illustrates, no overt constituent intervenes between C and T at Spell-Out in embedded subject questions, which means that finite C is able to check nominative Pos-Case on the wh-pronoun in [Spec,CP]. The case status of the wh-pronoun in embedded subject questions is thus identical to the case status of the wh-pronoun in matrix subject questions. (46)

CP 3

Nominative Pos-Case checking in embedded

DPi [nom] C’ and matrix subject questions who 3 C [nom] TP [+ finite] 3 ! DP T’ 3 ! ti z------ T FP 6 wonk ti tk the race

In embedded interrogatives questioning the object of a verb or preposition, C is unable to check nominative Pos-Case on its specifier, because the subject of the clause intervenes between C and T at Spell-Out (47). The same goes for embedded interrogatives questioning the lower argument of identificational be (48). Since the wh-pronoun in such embedded questions does not occupy a Pos-Case position at Spell-Out, its surface form will be constrained by DefCase (as well as Arg-Case). The examples of embedded questions discussed in this chapter so far all involve a finite verb. However, embedded interrogatives questioning the object of a verb or preposition may also be non-finite (49). Although no overt constituent intervenes between C and T at Spell-Out in these non-finite questions,

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

324

C is still unable to check nominative Pos-Case on the wh-pronoun in [Spec,CP], because it lacks finiteness. As discussed in Section 2.2.2, non-finite C can only participate in Pos-Case checking when it is filled with the complementizer for at Spell-Out. (47)

CP (48) CP 3 3 DPj C’ DPj C’ who(m) 3 who(m) 3 C TP C TP [+ finite] 2 [+ finite] 2 T’ ! DPi T’ ! DPi ! you 2 ! you 2 z---=- T FP z---=-- T FP 6 arek 5 sawk ti tk tj tk ti tk tj talkedk ti tk to tj

The Pos-Case status of the wh-pronoun in embedded interrogatives questioning the object of a verb or preposition

(49)

The Pos-Case status of the wh-pronoun in an embedded identificational question (cf. Section 8.5 for further discussion)

CP The structure of the non-finite embedded questions 3 DPi C’ in the sentences They’ll tell you [who to call] and who 3 She didn’t know [who to turn to] C TP [- finite] 3 PRO T’ 3 to call ti to turn to ti

Since the wh-pronouns in (47)-(49) occupy a surface position not covered by Pos-Case, they are subject to Def-Case and Arg-Case requirements. Both the Def-Case and the Arg-Case constraint call for initial whom in finite and non-finite embedded interrogatives questioning the object of a verb or stranded preposition. They can thus account for the occurrence of whom in sentences like (38)-(39), but are unable to predict the even greater popularity of who in the same contexts (cf. (34)-(35), (44) & (49)). In Section 9.3, I suggested that the occurrence of who after a verb or preposition in echo-questions is due to the influence of factors other than case, such as the preference for invariant who in all wh-positions. This tendency towards who will be particularly pronounced in positions not covered by Pos-Case, because Arg-Case and Def-Case tend to be weaker than Pos-Case in Present-Day English. The variation between initial

CHAPTER 9

325

who and whom in interrogatives questioning the object of a verb or stranded preposition, could thus be analysed as the result of competition between the demands of Arg-Case and Def-Case on the one hand, and the Invariant whform constraint on the other.14 9.5

wh-pronouns in sluicing constructions The most common examples of sluicing cited in the literature involve wh-pronouns that appear either before or after a preposition. Lasnik & Sobin (2000:345f) point out that the nominative who is virtually obligatory when the wh-pronoun precedes the preposition in a sluiced matrix clause (50). When the wh-pronoun follows the preposition, on the other hand, both whom and who occur in the data reported in existing studies (50)-(51) (cf. Lobeck 1995:54 and Merchant 2001:92n.3 for similar observations). As the examples in (52) illustrate, instances of who before and after prepositions also occur in embedded sluiced questions.15 (50) a. she fell in love … - How do you know - who with - where is she - ? … But in love - with whom? (Farnol [1913], The amateur gentleman:274) b. You can get married if you wish. - Who to? - To whom? Oh, anyone (George Bernard Shaw 1935, Too true to be good:154) (51) a. hee is in loue, With who? (Shakespeare, Much ado about nothing:I.i.214) b. yeeld thee, theefe. - To who? to thee? (Shakespeare, Cymbeline:IV.ii.76) c. A history! of who? (Laurence Sterne 1911[1759-67], Tristram Shandy and A sentimental journey: 1.76)

14

The proposed approach closely resembles Klima’s (1964) analysis of the distribution of whforms in different styles of English. Klima (1964:13-17) suggests that the use of the nominative who in interrogatives and relatives where the wh-pronoun is the object of a verb or stranded preposition, is due to a delay in the application of case-marking until after the pronoun has undergone ‘wh-attachment’, which corresponds to wh-movement in current approaches. Since Klima assumes that objective case is assigned under adjacency to V or P, pronouns that have undergone wh-movement will fail to be assigned objective case and surface in their unmarked form. For the personal wh-pronoun, the unmarked form is assumed to be who (Klima 1964:12). 15 The examples in (50)-(52) are all cited in Jespersen (1946:484f). As Diane Massam (p.c.) points out, it is important to note that the examples in (50) are taken from 20th century texts, while the examples in (51) date back to the 16th-18th century. (50) and (51) can thus not be seen as evidence for synchronic variation between who and whom after prepositions in sluiced questions.

326

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(52) a. I heard yesterday that she’d run away. I wasn’t told [who with] (Stephen McKenna 1918, Sonia married:200)

b. I am going to Gretna Green, and if you cannot guess [with who], I shall think you a simpleton (Jane Austen 1894[1813], Pride and prejudice:356)

In embedded sluicing constructions questioning the subject of a finite clause or the object of a verb, the nominative who is clearly favoured in Present-Day English (53)-(54). However, whom seems to be marginally possible, at least in some (earlier) varieties of English (55). (53) Somebody left, but we don’t know [who]. (Chao 1987:104) (54) John likes some girl but I don’t know [who]. (Frazier & Clifton 1998:500) (55) Some one was close behind, I knew not [whom]. (Robert Louis Stevenson [1882], Treasure Island:171) [Jespersen & Haislund 1949:223]

Merchant (2001:40-60) presents a range of arguments in favour of treating sluiced questions as CPs headed by an empty C specified for the features [+ wh] and [+ Q]. The feature [+ Q] captures the fact that sluicing can only occur in questions, and the feature [+ wh] ensures that sluiced questions always contain an overt wh-phrase in [Spec,CP] (cf. also Lobeck 1995:50; and Frazier & Clifton 1998:501). Most recent approaches to sluicing agree that the TP complement of C is phonologically empty in sluiced questions, but there is considerably less agreement on the syntactic content of the TP at Spell-Out. Chao (1987:158), Lobeck (1995:60), and Chung et al. (1995:242) assume that the wh-phrase enters the derivation as [Spec,CP], and the TP complement of C is base-generated as an empty category, which lacks internal structure. In support of this analysis, Lobeck (1995:58-60) draws attention to the absence of verbal agreement morphology on sluiced constituents in languages like Bavarian German. For example, in full embedded questions, the Bavarian German 2sg agreement marker st may show up on both the finite verb in T, and on the fronted wh-phrase introducing the embedded clause (56). In a corresponding sluiced question, on the other hand, the occurrence of -st on the wh-phrase is ungrammatical (57).

CHAPTER 9

327

(56) Du woiddst doch kumma, owa mia wissn ned [wann-st (du) you wanted come, but we know not when-2sg you kumma woidd-st]. come wanted-2sg “You wanted to come, but we don’t know when you wanted to come.” (Lobeck 1995:58)

(57) Du woiddst doch kumma, owa mia wissn ned [wann (*-st)]. you wanted come, but we know not when “You wanted to come, but we don’t know when.” (Lobeck 1995:59) If we assume that overt agreement marking in C arises from a syntactic relationship between C and T, then the lack of verbal agreement morphology in sluiced questions will fall out naturally from the assumption that T is absent at Spell-Out. A rather different approach to ellipsis, and sluicing in particular, is advocated by Merchant (2001). Merchant (2001:60) argues that sluicing involves PF deletion licensed by feature checking between T and C. In Merchant’s approach, the full internal structure of the TP is present in the surface syntax but remains unpronounced, because the PF features of the various heads have been deleted. The absence of agreement morphology and overt complementizers in C is assumed to arise from prosodic constraints rather than a lack of syntactic structure at Spell-Out (Merchant 2001: 72f, 80-83, 230f). Although the two approaches differ quite markedly in their assumptions about the syntactic properties of sluiced constituents, the differences in the surface syntax turn out to have little bearing on the predictions of the three case constraints proposed in Chapter 2. As we have seen in preceding sections, finite C can acquire the ability to check nominative Pos-Case on its specifier through T-to-C raising, or through a PF merger under surface adjacency with T. Neither of these will be possible if T is absent at Spell-Out, or if the PF features of T have been deleted. Both an empty TP approach and a PF deletion approach to sluicing will thus predict that C should be unable to check nominative Pos-Case in sluiced questions, which means that the wh-pronoun in [Spec, CP] will be subject to Def-Case rather than Pos-Case. The consistent selection of whose in sluiced possessives (58) suggests that the wh-pronoun is able to check genitive Pos-Case in this context. (58) Somebody’s car is parked on the lawn, but we don’t know [whose / * who / * whom]. (Merchant 2001:43)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

328

As mentioned in Section 2.2.3 and Section 9.1, a DP can only check genitive Pos-Case if it appears in [Spec,DP] at Spell-Out and does not receive its θ-role in this position. If we assume that whose refers to the owner of the car, the genitive wh-pronoun in (58) occupies the specifier of the DP in [Spec,CP] at Spell-Out, but receives its θ-role from its trace in [Spec,NumP], which is either present in the overt syntax, or reconstructed at LF (59). (59)

CP wo DP C’ 3 2 D’ C TP DPi [gen] whose 3 D [gen] NumP See Lobeck (1995:85-96) for arguments supporting 3 the presence of a fully articulated NumP-layer in ( ti [θ-role] ) noun phrases involving NP ellipsis

Since the predicate of any sluiced question must be present at Semantic Form to ensure the desired interpretation of the sluiced constituent (cf. Chao 1987:65-74; Lobeck 1995:32-35; Chung et al. 1995:263f), wh-pronouns in sluicing constructions will be affected by Arg-Case requirements, no matter whether we adopt a PF-deletion approach to sluicing, or assume that TP lacks internal structure at Spell-Out. While the Def-Case constraint calls for the objective form whom wherever it applies, the Arg-Case of a pronoun depends on its status on the argument hierarchy. In (53) and (55), the wh-pronoun is the highest argument of the predicate in the sluiced question, and would thus be required to take the nominative form who by the Arg-Case constraint. The marginal occurrence of the objective form whom in sentences like (55) could be seen as evidence that the Def-Case constraint may outweigh the Arg-Case constraint (in certain varieties of English). The general trend towards the nominative who in sluicing constructions is more problematic for a purely case-based approach. While the occurrence of who in sluiced subject questions like (53) is predicted by the Arg-Case constraint, neither Arg-Case nor Def-Case could predict the use of who in sluiced questions where the wh-pronoun is the object of a reconstructed verb or preposition, as in (54). The distribution of wh-forms in sluiced questions thus provides further evidence for competition between the case constraints and a tendency towards invariant who. An additional non-case factor appears to influence the choice of whforms in sluicing constructions where the sluiced constituent is the object of a

CHAPTER 9

329

preposition. As can be seen from (60), the preposition may either precede the wh-pronoun (pied-piping), or it may follow the pronoun (swiping).16 (60) a. Mary left with someone, but we don’t know [with who(m)]. b. Mary left with someone, but we don’t know [who with]. (Lobeck 1995:60)

When the preposition is pied-piped with the wh-pronoun (60a), both who and whom occur, but when the preposition is swiped (60b), only who seems possible (cf. Lasnik & Sobin 2000:345f). This suggests that the position of the whpronoun relative to the preposition has some bearing on its surface form in sluiced questions. The structure of sluicing constructions involving a pied-piped preposition (60a) follows the structure of the relevant part of full questions, with the whpronoun contained in a PP that occupies [Spec,CP] (61) (cf. Lobeck 1995:60f). The syntactic structure of sluices involving swiping (60b) is more problematic. In full questions, stranded prepositions are generally assumed to remain within TP when the wh-pronoun raises to [Spec,CP]. However, such an analysis would not appear to be available for swiped prepositions in sluiced questions, no matter whether we adopt an empty category or PF-deletion approach to ellipsis. If a sluiced wh-phrase is directly inserted into [Spec,CP] without prior movement, and the TP complement is base-generated as an empty category, a stranded preposition will not be able to occupy a TP-internal position in sluicing constructions. If the wh-phrase is assumed to have undergone movement from within TP, but the phonological contents of TP are deleted at PF, then a preposition stranded within TP will remain unpronounced (cf. Merchant 2001: 91-107). Lobeck (1995:61f), following a proposal by Riemsdijk (1978), therefore suggests that a wh-pronoun preceding the preposition in a sluiced question occupies [Spec,PP] at Spell-Out (62). (61)

CP ei PP C’ 2 2 P DP C TP [+wh, +Q] with g D who(m)

16

(62)

CP wo PP C’ 2 2 P’ C TP DPi who 2 [+wh, +Q] P DP with ti

The term ‘swiping’ was coined by Merchant (2001:123) and stands for sluiced wh-phrase inversion with prepositions in Northern Germanic.

330

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

While only wh-pronouns seem to undergo PP-internal raising in Present Day English, Pintzuk (1996:392,295) argues that raising from [Comp,P] to [Spec, PP] is responsible for the occurrence of personal pronouns before an associated preposition in Late Old English prose texts (63). As Kate Kearns (p.c.) points out, raising to [Spec,PP] also appears to be available as a poetic device (64), and is a possible source for the frozen expression the whole world over. (63) hwæðer hiera mehte maran fultum [him to] geteon which 3pl.GEN might greater help 3sgM.DAT to draw “…which of them might draw greater help to him.” (Orosius 78.33-79.1) [Pintzuk 1996:392]

(64) …and all her hair / in one long yellow string I wound / three times [her little throat around] (Robert Browning [1836], ‘Porphyria’s Lover’) As discussed in Section 2.2.4, objects of prepositions are unable to check Pos-Case inside the PP, because P is not an agreement related head, and there is little justification for positing additional functional projections within the prepositional phrase. Since the wh-pronoun appears within the PP both in sluiced questions where the preposition is pied-piped (61), and in sluiced questions with P-stranding (62), we will not be able to appeal to differences in the applicability of the three case constraints when trying to account for the occurrence of whom in post-P but not pre-P position. In sluicing constructions, all wh-objects of prepositions will be influenced by Arg-Case and Def-Case (both of which call for whom in this context), regardless of whether they appear in pre-P or post-P position. I have been arguing that the ready occurrence of who in embedded and sluiced questions is largely due to a general trend towards invariant who. If this trend was the only non-case influence to compete with Arg-Case and Def-Case in sluicing constructions, we would expect to find the same degree of variation between who and whom when the wh-pronoun precedes an associated preposition as when it follows the preposition. The interaction of the case constraints with the trend towards invariant who is thus unable to account for the formal differences between pre-P and post-P wh-objects in sluiced questions. I propose that these differences in form are directly linked to differences in structural position, and provide further evidence for Relative Positional Coding. Whom has a greater phonological complexity than who, because it contains a coda consonant as well as the vocalic nucleus. The Relative Positional Coding (RPC) constraints will therefore call for the use of the robust form whom when the wh-pronoun appears in the asymmetrically c-commanded complement posi-

CHAPTER 9

331

tion at Spell-Out (61), but will demand the gracile form who in the asymmetrically c-commanding [Spec,PP] position (62) (cf. Section 6.2). As a result, speakers for whom the RPC constraints outweigh the Invariant wh-form constraint will favour whom after P, but who in pre-P position. 9.6

wh-pronouns in free relatives In Old English and Middle English, wh-pronouns were largely confined to questions. The first wh-pronouns to appear in relative clauses formed part of the combination swa hwa swa ‘so who so’, which was used to introduce free relatives (Jespersen 1949[1927]:116; Allen 1980:114).17 During the Middle English period, the initial swa disappeared from this construction, and the second swa was often replaced with that or ðe, which could in turn be omitted, leaving only the wh-pronoun (Allen 1980:207ff). At the same time, the adverb ever started to be used in free relatives without its temporal meaning. The increasing association of ever with the wh-phrase gradually lead to its reanalysis as part of the wh-pronoun (Allen 1980:209ff,387). Allen (1980:114-121) distinguishes two types of free relatives, which have different case and preposition-stranding properties (cf. also Pintzuk 2000:53f). Argument relatives function as the argument of a matrix predicate, while leftdislocated free relatives are associated with a ‘returning’ (or ‘intrusive’) pronoun in the matrix clause.18 In Old and Middle English, the case of the whphrase in an argument relative was always determined by the function of the relative in the matrix clause, even when it disagreed with the function of the wh-phrase within the relative (cf. Allen 1980:114f,208f; Bresnan & Grimshaw 1978:375; Pintzuk 2000:53). For example, the wh-pronoun wam functions as the subject of the relative clause in (65), but the relative itself functions as the object of the matrix preposition to. Since the matrix case always wins out over the relative-internal case, the wh-pronoun in (65) surfaces in the objective form wam, rather than the nominative form hwa. When the wh-phrase functioned as the object of a preposition within an argument relative, preposition stranding appears to have been obligatory (66) (cf. Allen 1980:145n.14).

17 As Allen (1980:115) points out, the second swa in this construction could be separated from the wh-pronoun in Old English, which suggests that it is best analysed as a complementiser selected for by the swa that introduces the wh-phrase in a free relative. 18 See Demirdache (1997:198) for a discussion of the exact nature of the matrix pronoun associated with the dislocated constituent in left-dislocation structures.

332

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

beoð (65) Ðe holi gost … hine dealeð to [wam him the holy ghost … 3sgM.ACC gives to wh.OBJ 3sgM.OBJ is lofue] pleasing “The holy ghost … gives it to whomever is pleasing to him.” (Layamon (Otho) 9081) [Allen 1980:208]

(66) Ac seid to [hwam he wið spekeð], hwi sholde ich but says to wh.OBJ 3sgM.NOM with speaks how should 1sg.NOM him luuien? 3sgM.OBJ love “But says to everyone he speaks with, ‘Why should I love him?’” (Old English Homilies 2,XXIX,183.32) [Allen 1980:209]

In left-dislocated free relatives, on the other hand, prepositions were obligatorily pied-piped (67), and the case of the wh-phrase always corresponded to its function within the relative clause (68).19 (67) Ðat is min red, [wið quam ðu is findes]i, ðat that is my advice with wh.OBJ 2sg.NOM them find that be dead hei 3sgM.NOM be dead “this is my advice: whoever you find them with, he should be killed” (Genesis & Exodus 1768) [Allen 1980:209]

(68) [Hwam ich biteche ðat bred ðat ich on wyne wete]i, the bread that 1sg.NOM in wine wet wh.OBJ 1sg.NOM give hei me schal bitraye 3sgM.NOM 1sg.OBJ shall betray “whoever I give the bread that I wet in wine, he shall betray me” (Jesus MS Passion 103) [Allen 1980:209]

While free relatives may still appear either as an argument of a matrix predicate or as a left-dislocated constituent in Modern English, preposition stranding is now possible in both types of constructions (cf. the Modern English translation in example (67)), and the distribution of wh case forms in free relatives is no longer as clear-cut as in Old and Middle English. As we will see 19 In all of the examples involving left-dislocated free relatives, the free relative is coindexed with the intrusive/returning pronoun in the matrix clause.

CHAPTER 9

333

in Sections 9.6.1 and 9.6.2, the form of the wh-pronoun in a Modern English argument relative may meet either matrix or relative-internal case requirements. What is more, both argument relatives and left-dislocated free relatives may be introduced by wh-pronouns whose case corresponds neither to the function of the wh-phrase within the relative clause nor to the function of the relative in the matrix clause. In Modern English free relatives, the complex wh-forms whoever, whomever, whoso(ever), and whomsoever tend to be favoured over the simplex forms who and whom (cf. Jespersen 1949[1927]:62; Baker 1995:210f), except when the relative clause involves VP ellipsis (69) or Null Complement Anaphora (70). VP ellipsis typically occurs after modal verbs, while Null Complement Anaphora occurs with lexical verbs such as like, please, and choose (cf. Chao 1987:104-176; and Lobeck 1995:141-191). (69) She had the art of pleasing [whom she would] (Charlotte Brontë 1867[1852], Villette:122) [Jespersen 1949[1927]:63]

(70) courting [whom she pleased] and ignoring all others20 (Theodore Dreiser 1918, Free, and other stories:63) [Jespersen 1949[1927]:63]

Since the syntactic properties of free relatives with VP ellipsis (VPE) or Null Complement Anaphora (NCA) differ from the properties of other free relatives, I will first discuss the case trends and predictions for free relatives without ellipsis, and then take a closer look at VPE and NCA relatives in Section 9.6.2. 9.6.1 Free relatives introduced by complex and simplex wh-pronouns The most common complex wh-forms to occur in the Modern English examples collected by Jespersen (1949[1927]) are the nominative whoever and the objective form whomsoever. While the nominative is readily tolerated in a wide range of contexts, the objective form is largely confined to positions immediately following a verb or preposition. Since pied-piping of prepositions remains ungrammatical in argument relatives,21 the wh-pronoun introducing the argument relative will only follow a verb or preposition if the whole relative clause appears as the complement of a verb or preposition (71)-(72). While the use of whom(so)ever in this context 20

Note that please is synonymous with want, like, and choose in free relatives involving Null Complement Anaphora. 21 The only possible instances of pied-piping in argument relatives occur when the relative clause itself appears as the complement of a preposition, and the preposition associated with the wh-phrase is able to ‘merge’ with the matrix preposition (see examples (82) & (88) below).

334

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

may meet both matrix and relative-internal case requirements (71), instances of whom(so)ever are also attested in argument relatives where the function of the relative pronoun within the relative clause would lead us to expect a nominative (72).22 (71) with Martin and Leora, or with [whomsoever he could persuade to come] (Sinclair Lewis 1926[1925], Martin Arrowsmith:400) [Jespersen 1949[1927]:58]

(72) a. power to summon [whomsoever might throw light upon the events] (newspaper article, 1919) [Jespersen 1949[1927]:57] b. winking over her shoulder at [whomever would watch her comedy] (Compton Mackenzie 1913-14, Sinister street:894) [Jespersen 1949[1927]:57]

The occurrence of whom(so)ever in examples like the ones given in (72) would appear to suggest that the case of wh-pronouns in argument relatives is still determined by the function of the relative clause in the matrix sentence. However, wh-pronouns introducing argument relatives in object and prepositional complement position just as readily take the nominative form whoever in Modern English, especially when the wh-pronoun functions as the subject of the relative (73). (73) Rhonda dances with [whoever asks her to dance] (Baker 1995:210) As we might expect, the nominative whoever is obligatory when both the relative clause and the wh-pronoun function as the subject of a finite clause (74). However, we also find instances of whoever which meet neither matrix nor relative-internal case requirements, especially in fronted object relatives (75). (74) [Whoever has an ambition to be heard in a crowd] must press, and squeeze, and thrust, and climb, with indefatigable pains … (Jonathan Swift 1909 [1704], A tale of a tub, The battle of the books and other satires:43)

(75) And generally, [who euer the King fauours], The Cardinall instantly will finde imployment (Shakespeare, Henry VIII:II.i.47) [Jespersen 1949[1927]:58] 22

(72b) is the only example of whomever cited in Jespersen (1949[1927]).

CHAPTER 9

335

In left-dislocated free relatives, the whom(so)ever is most likely to occur when the relativised constituent is the object of a preposition, and the preposition in question is pied-piped with the wh-phrase (76). Who(so)ever is favoured when the wh-pronoun functions as the subject of the left-dislocated relative (77), and also when it is the object of a verb or stranded preposition (78). (76) Giue me also this power, that [on whomsoeuer I lay hands]i, heei may receiue the holy Ghost (The authorized version of the Bible 1611 (1833 Facsimile ed.):Acts 8.19) [Jespersen 1949[1927]:72]

(77) [whosoeuer shall smite thee on thy right cheeke]i, turne to himi the other also … And [whosoeuer shall compell thee to goe a mile]j, go with himj twaine. (The authorized version of the Bible 1611 (1833 Facsimile ed.):Matthew 5.39) [Jespersen 1949[1927]:71f]

(78) a. [whoever I give the bread that I wet in wine]i, hei shall betray me b. this is my advice: [whoever you find them with]i, hei should be killed (Allen’s 1980:209 translations of examples (68) and (67)) As mentioned earlier, who and whom occur quite readily in free relatives containing VP ellipsis or Null Complement Anaphora, but are disfavoured in non-ellipsis contexts. Virtually all of Jespersen’s (1949[1927]) examples of who and whom in free relatives without ellipsis come from poems, Shakespeare plays, and the 1611 version of the Bible. These examples suggest that the nominative form who is the only option in free relatives without ellipsis when both the relativised constituent and the relative clause itself function as the subject of a finite clause (79). Who seems equally obligatory when the wh-pronoun functions as the subject of a left-dislocated free relative (80). (79) [Who steales my purse], steales trash (Shakespeare, Othello:III.iii.157) [Jespersen 1949[1927]:56]

(80) a. [who pretendeth to God]i, God attendeth to himi (William Caxton) [Steinki 1932:45-6, cited in Householder 1986:151]

b. [who tels me true]i, though in his tale lye death, I heare himi as he flatter’d (Shakespeare, Antony and Cleopatra:I.ii.102) [Jespersen 1949[1927]:71]

336

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

The objective form whom occurs most readily when the relative clause appears as the complement of a matrix preposition and the wh-pronoun functions as the object of a verb (81) or preposition (82). (81) a. Eternity so spent in worship paid To [whom we hate] (John Milton, Paradise lost:2.249) [Jespersen 1949[1927]:58]

b. the souls of [whom thou lovest] (Percy Bysshe Shelley 1904, Poetical works:260) [Jespersen 1949[1927]:58]

(82) Therefore hath hee mercie on whom he will have mercy (The authorised version of the Bible 1611 (1833 Facsimile ed.):Romans 9.18) [Jespersen 1949[1927]:56]

As can be seen from the example in (82), the preposition associated with the wh-pronoun need not be stranded in argument relatives where both the whpronoun and the relative function as the object of a preposition, provided the two prepositions are identical. I will argue below that the presence of just one preposition in the surface string suggests that free relatives of this type are prepositional rather than nominal. In (81) and (82), the occurrence of the objective form whom fits in with both matrix and relative-internal case requirements. However, Jespersen (1949 [1927]:57) also offers examples of whom in argument relatives where the whpronoun functions as the object of a verb, but the relative clause itself appears as the subject of a finite clause (83). (83) [Whom I marry] shall be noble (Elizabeth Barrett Browning [1856], Aurora Leigh:231) [Jespersen 1949[1927]:57]

Allen (1980:116-120) argues that the case differences between argument and left-dislocated wh-relatives in Old English (OE) and Middle English (ME) fall out from differences in the syntactic status of the wh-pronoun. In OE and ME argument relatives, the wh-phrase is influenced only by matrix case requirements, because it is base-generated as a nominal projection outside the relative clause (Allen 1980:117; cf. also Bresnan & Grimshaw 1978; Larson 1987). This analysis also accounts for the absence of pied-piping in argument relatives. In OE and ME left-dislocated free relatives, the case of the wh-phrase is determined solely by its function in the relative clause, because it is basegenerated within the relative clause (Allen 1980:120) The case trends summarised above indicate that there is no longer such a clear distinction between argument relatives and left-dislocated free relatives in

CHAPTER 9

337

Modern English. While the wh-pronoun introducing an argument relative may surface in the case form corresponding only to the function of the relative in the matrix clause (72), we also find argument relatives where the wh-form corresponds to the function of the wh-pronoun in the relative, but violates matrix case requirements (83). This suggests that in Modern English, the wh-pronoun has a closer relationship with the relative clause in argument relatives than it did in Old English and Middle English (cf. also Jespersen 1949[1927]:57f). Alexiadou et al. (2000:22f) propose that free relatives are CP complements of a phonetically zero D. In such an approach, the wh-pronoun will occupy a Pos-Case position only when it functions as the subject of the clause (84). Since no overt constituent intervenes between C and T at Spell-Out, finite C inherits the ability to check nominative Pos-Case on its specifier from T. When the wh-pronoun is the object of a verb or stranded preposition, C and T will be separated by the subject of the clause at Spell-Out (85). As a result, the wh-pronoun will not occupy a Pos-Case position, and will instead be influenced by the Def-Case constraint. (84)

DP (85) DP 3 3 D CP D CP e 3 e 3 DPj C’ DPi [nom] C’ who 3 whom 3 TP C TP C [nom] [+ finite] 3 [+ finite] 3 ! DP T’ ! DPi T’ ! I 3 ! ti 3 z__________ T FP z____=____ T FP 6 5 ti marry tj ti steals my purse

Any wh-pronoun that functions as the argument of a predicate in the relative clause will be influenced by Arg-Case requirements associated with this relative-internal predicate. Since the wh-pronoun does not have any direct relationship to a matrix predicate in Alexiadou et al.’s (2000) analysis, we would expect no Arg-Case influence from the matrix clause (I am assuming that the Arg-Case constraint always applies to the head of an argument DP). The analysis proposed by Alexiadou et al. (2000:22f) correctly predicts the distribution of simplex wh-pronouns in the examples cited earlier. However, their approach is unable to offer an explanation for the occurrence of the objective form whomsoever in subject relatives that appear as the object of a preposition (86), because the presence of the empty D in (85) prevents the wh-

338

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

pronoun from entering into a syntactic or semantic relationship with the matrix clause. (86) She sat by the fire conversing with [whomsoever approached her] (Henry James [1877], The American:1.267) [Jespersen 1949[1927]:56]

One way to establish a link between the matrix clause and the wh-pronoun would be to assume that the free relative DP is headed by a null pronominal element (pro), which is identified through φ-feature (and case) agreement with the wh-pronoun (87) (cf. McCreight 1988:107f; Suñer 1984; Grosu 1996: 289ff). (87)

DP pro-head analysis of free relatives with DP status ei D CP ei proi [φ] DPj C’ wh-pronouni 3 [φ] C TP [φ] 6 tj steals my purse I marry tj you find them with tj

The analysis in (87) would not only predict that a wh-pronoun introducing an argument relative in Modern English may be influenced by matrix as well as relative-internal case requirements, but could also explain why preposition stranding is obligatory in argument relatives unless the preposition is also needed in the matrix clause: Observations by Grosu (1996:289ff) suggest that agreement between pro and the wh-pronoun is only possible if the wh-pronoun enters into spec-head agreement with C. This means that the wh-pronoun must head the wh-phrase in [Spec,CP] in order to be identified with pro. If the whpronoun is embedded inside a PP, it will not be in a spec-head relationship with C, and pro will remain unidentified, causing the derivation to crash. In free relatives of the type illustrated in (88), pied-piping of the preposition is possible, because the CP is dominated by a PP headed by a pro-P rather than a DP headed by pro (89) (cf. Grosu 1996:259,289 for a similar proposal). (88) beare … all the rest To whom they are directed (Shakespeare, Henry IV, Part 1:III.iv.3) [Jespersen 1949[1927]:55f]

CHAPTER 9

(89)

339

PP ei pro-head analysis of free relatives with PP status P CP pro-Pi ei [P] PPj C’ 2 3 P DP C TP 6 toi whom [P] [P] they are directed tj

The pro-P in (89) is licensed through feature agreement with the preposition to. Such agreement between the pro-P and the overt preposition is possible because to heads the PP in [Spec,CP], and is thus able to transmit its P(reposition) features to C through spec-head agreement. The assumption that free relatives are headed by an empty pronominal element clearly allows us to capture a number of properties characteristic of Modern English argument relatives. However, even case agreement between pro and the wh-pronoun would not be sufficient to predict the occurrence of whomsoever in sentences like (86). As discussed in Section 2.2.4, prepositions are unable to check Pos-Case. Any pro heading a free relative in prepositional complement position would thus receive objective Arg-Case and Def-Case, but no Pos-Case. The wh-pronoun, on the other hand, will check nominative PosCase, because it appears as the specifier of a finite C, and no overt constituent intervenes between C and T at Spell-Out (90). Since Pos-Case generally overrides Arg-Case and Def-Case in Present-Day English (cf. Section 2.3), the whpronoun would be expected to surface in the nominative form whoever rather than the objective whom(so)ever. As (90) illustrates, [Spec,CP] is a nominative Pos-Case position in free relatives with a wh-subject. If we want to prevent the wh-pronoun in sentences like (86) from checking nominative Pos-Case, we will have to assume that it occupies a position other than [Spec,CP] at Spell-Out. Kayne (1994:125,154 n.13) argues that the head of the wh-phrase in a free relative undergoes headmovement to the CP-external D which heads the relative construction. In this analysis, the wh-pronoun introducing a free relative will check Pos-Case only if the free relative itself occupies a Pos-Case position. When the free relative appears as the complement of a preposition (91), the wh-pronoun will be subject only to Arg-Case and Def-Case requirements, because prepositions are unable to check Pos-Case.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

340 (90)

PP (91) PP 2 3 P DP P DP with 2 with ei D CP D CP 1 3 proi 3 DPj [nom] C’ Dj D DPi C’ whoeveri 3 whomsoever g 3 D C TP C [nom] TP [+ finite] 2 tj [+ finite] 3 ! DP T’ ti approached her ! tj 2 z_________ T FP 6 tj approached her

As mentioned earlier, I am assuming that the Arg-Case constraint always applies to the head of an argument DP. In (91), the wh-pronoun heads the DP that dominates the relative clause, and the trace of the wh-pronoun heads the CPinternal DP. The form of the wh-pronoun will therefore be influenced by matrix as well as relative-internal Arg-Case requirements.23 If the matrix ArgCase wins out, the wh-pronoun will surface in the objective form whom(so)ever, because the free relative is the lowest argument of the preposition with. The selection of whom(so)ever is further supported by the Def-Case constraint. If the relative-internal Arg-Case wins out, the wh-pronoun will surface in the nominative form whoever, because the wh-phrase is the highest argument of the verb approach. The choice of the nominative form is further supported by the trend towards invariant who in all wh-contexts. Free relatives of the type illustrated in (92) can also be analysed as PPs headed by an overt preposition rather than pro-P. In such an approach, the preposition raises from the wh-phrase in [Spec,CP] to the head of the PP dominating the whole structure (93). (92) beare … all the rest To whom they are directed (Shakespeare, Henry IV, Part 1:III.iv.3) [Jespersen 1949[1927]:55f]

23 See Vogel (2001, 2003) for suggestions on how the interaction between matrix and relativeinternal case influences could be modelled in an optimality-theoretic approach. A detailed discussion of the competition between matrix and relative-internal case in free relatives in a number of different languages can be found in McCreight (1988:85-109).

CHAPTER 9

(93)

341

PP ei P-raising analysis of free relatives with PP status P CP 1 ei C’ Pi P PPj to 2 6 P DP they are directed tj ti whom

The word order found in V-particle constructions suggests that free relatives tend not to raise to [Spec,vP] when they are objects of a verb (94). A whpronoun introducing a free relative that functions as the object of a matrix verb is thus unlikely to be influenced by matrix Pos-Case. When a free relative appears in [Spec,TP] of the matrix clause, on the other hand, a wh-pronoun heading the relative clause will be able to check nominative Pos-Case with the matrix T, provided the matrix clause is finite (95). (94) a. ?? We’ll throw [whoever bothers the barman] out. b. We’ll throw out [whoever bothers the barman]. (95)

CP wo C TP [+ finite] qp DP [nom] T’ wo 3 CP T [nom] … D [nom] 1 wo Dj D DPi C’ qp whoever g D C TP qp tj [+ finite]

ti approached her / I marry ti / you find them with ti

Since Pos-Case generally overrides Arg-Case in Present-Day English, the analysis in (95) would lead us to predict that only nominative wh-forms should be able to occur in free relatives that appear as the subject of a finite clause. The occurrence of whom in examples like (96) is thus unexpected. (96) [Whom I marry] shall be noble (Elizabeth Barrett Browning [1856], Aurora Leigh:231) [Jespersen 1949[1927]:57]

One way to account for the coexistence of (96) and (97) would be to assume that only complex wh-forms are able to undergo head-movement to D.

342

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(97) She sat by the fire conversing with [whomsoever approached her] (Henry James [1877], The American:1.267) [Jespersen 1949[1927]:56]

If the simplex wh-pronoun occupies [Spec,CP] at Spell-Out in sentences like (96), it will not be affected by matrix Pos-Case requirements, and will be predicted to surface in the objective form that corresponds to its relative-internal argument status. Such an approach would predict that the form of simplex whpronouns will be determined primarily by their function within the relative clause, while the form of complex wh-pronouns will be influenced by both matrix and relative-internal case requirements. Alternatively, we could argue that any wh-pronoun introducing a free relative may be analysed as occupying either D or [Spec,CP] in Present-Day English. This approach would lead us to expect that both simplex and complex wh-pronouns may be influenced by matrix as well as relative-internal case requirements. Whatever structural analysis we adopt, a purely case-based account will always remain unable to account for the occurrence of whoever in positions where both matrix and relative-internal case requirements would lead us to expect an objective pronoun form (98). (98) And generally, [who euer the King fauours], The Cardinall instantly will finde imployment (Shakespeare, Henry VIII:II.i.47) [Jespersen 1949[1927]:58]

This suggests that the distribution of wh-forms in free relatives is at least partly influenced by the trend towards invariant who(ever) in all wh-contexts. As mentioned above, the tendency towards invariant whoever appears to be particularly pronounced in left-dislocated free relatives, no matter whether the relative is coreferent with the subject (99) or object of the matrix clause (100). (99)

this is my advice: [whoever you find them with]i, hei should be killed (Allen’s 1980:209 Modern English translation of example (65))

(100) [Whoever you found them with]i, you should kill himi. (example provided by Liz Pearce)

If we assume that left-dislocated relatives have the same structure as argument relatives, this greater susceptibility to non-case influences will fall out naturally. Since left-dislocated relatives do not function as arguments of a matrix predicate (cf. Section 8.2), the wh-pronoun will not be influenced by matrix

CHAPTER 9

343

Pos-Case or Arg-Case, but only by relative-internal Arg-Case and either relative-internal Def-Case (if we assume that the wh-pronoun occupies [Spec,CP] at Spell-Out) or matrix Def-Case (if we treat the wh-pronoun as the head of the relative construction). 9.6.2 Free relatives with VP ellipsis and Null Complement Anaphora VP ellipsis (VPE) and Null Complement Anaphora (NCA) are only possible in argument relatives. The missing predicate generally corresponds to the matrix predicate, and the function of the wh-phrase within the relative corresponds to the function of the relative in the matrix clause. For example, both the relative clause and the wh-pronoun function as the object of the verb please in (101a), and both the relative clause and the wh-pronoun function as the object of the verb marry in (102a). When the wh-pronoun and the relative function as the object of a preposition in a VPE (101b) or NCA construction (102b), only one instance of the preposition appears in the surface string, always in a position immediately preceding the wh-pronoun (the examples in (101)-(102) are all cited in Jespersen 1949[1927]:63). (101) Argument relatives with VP ellipsis (VPE) a. She had the art of pleasing [whom she would] (Charlotte Brontë 1867[1852], Villette:122)

b. waiting till I could fling my shoes at whom I would (Hope, King’s M.:13)

(102) Argument relatives with Null Complement Anaphora (NCA) a. Constance may marry [whom she pleases] (Oliver Goldsmith 1889, Globe ed.:676)

b. I’ll go with whom I choose (John Galsworthy, A family man and other plays:292)

While Null Complement Anaphora is confined to relatives that function as the object of a verb or preposition (102), VP ellipsis is also possible in subject relatives, provided that the relative clause follows a matrix verb. The matrix verb in question may be identical to the missing verb in the relative (103), but it need not be. In (104) the missing constituent in the relative corresponds to the following predicate (be clever) rather than the preceding verb let.

344

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(103) I will say my say to the end, mock at it [who may]24 (John Ruskin 1904[1867], Time and tide:94) [Jespersen 1949[1927]:63]

(104) Be good, sweet maid, and let [who can] be clever (Charles Kingsley 1856, ‘A farewell. To C.E.G.’) [Jespersen 1949[1927]:56]

VPE relatives following let have exceptional case properties. In the VPE and NCA examples presented in (101)-(103), the case form of the wh-pronoun meets both matrix and relative-internal case requirements, and corresponds to the case found on lone personal pronouns in the same contexts. In examples like (104), on the other hand, the obligatory use of the nominative who violates matrix Pos-Case requirements. Since the wh-pronoun precedes a finite auxiliary within the relative clause, it will check nominative Pos-Case within the relative clause. However, the position of the relative in the matrix clause would lead us to expect an objective pronoun form. As can be seen from (105), a lone pronoun occupying the same position as the relative in (104) will alway appears surface in the objective form rather than the nominative.25 (105) Let them / * they be clever. While the case form of the wh-pronoun in (102) is at least predictable from the position and function of the wh-pronoun within the relative clause, Jespersen (1949[1927]:63) also provides examples of argument relatives with VP ellipsis (106) and Null Complement Anaphora (107) where the wh-form meets neither matrix nor relative-internal case requirements. (106) Let the patent be bought by who it will (Daniel Defoe 1890, The complete gentleman, London:45)

(107) I marry [who I please] (Compton Mackenzie 1922[1912], Carnival:355)

24 Note that the relative clause serves as the subject of the verb mock in this example. As Liz Pearce (p.c.) points out, the surface order is most likely to have resulted from the raising of the VP mock at it past the free relative. 25 Nominatives do occur when a pronoun following let appears in a coordinate (i).

(i) let [he and I] say good night together (Charles Dickens 1900[1839], Nicholas Nickleby:366) [Jespersen & Haislund 1949:237] This suggests that coordinates and free relatives share certain syntactic properties which affect the case status of pronouns contained in them (cf. Section 5.1 for a more detailed discussion of pronoun case in coordinates).

CHAPTER 9

345

Chao (1987:104ff) and Lobeck (1995:32-35,141-150) argue that the missing constituent in VP ellipsis constructions is base-generated as an empty category, which is interpreted through reconstruction at a semantic level of representation (108)-(109). Chao (1987:104ff) proposes a similar analysis for Null Complement Anaphora (110).26 (108)

CP 3

wh-phrase

C’ 2 C TP 2 DP T’ 3 I T VP 1 3 FP Vi T V e will ti

The base structure of the clause in free relatives involving VP ellipsis (VPE relatives) where the relativised constituent is the object of a V or P

(110)

(109)

CP 2 C TP 3 DP T’ 3 who T VP 1 2 V FP Vk T can tk e

The base structure of the clause in VPE relatives where the relativised constituent is a subject

CP 3

wh-phrase

C’ The base structure of the clause in free relatives with Null Complement Anaphora (NCA relatives) 2 C TP 2 T’ DPi she 2 T VP 3 DP V’ 3 ti V FP pleases e

26 Not everybody agrees that ellipsis involves the base-generation of empty categories followed by LF reconstruction. Merchant (2001:70-73,230f) equates ellipsis with PF deletion, and argues that the internal syntactic structure of the missing constituents is present throughout the derivation, even though it remains unpronounced. As mentioned earlier, the differences between an empty category approach and a PF deletion analysis have little bearing on the predictions of the constraints proposed in this book. In order to reduce the complexity of the tree diagrams, I have therefore decided to focus on the empty category approach in my discussion of VP ellipsis and Null Complement Anaphora.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

346

Since the missing verb phrase in VP ellipsis constructions may be agentive/causative, I am assuming that the category of the base-generated empty constituent is FP rather than VP, where F is the functional category an agentive/causative transitive verb moves to before Spell-Out (cf. Section 2.2.1). I am also assuming that Null Complement Anaphora verbs subcategorise for a to-infinitive lacking a TP-layer (cf. Wurmbrand 2001 for supporting arguments). I have decided to label the ellipted to-infinitive as a functional projection (FP), but the case predictions would be the same if to was assigned the category V, as suggested by Law (2000:175-177). Since the FP in both constructions is base-generated as an empty category, the wh-phrase has to be inserted directly into (the higher) [Spec,CP] if it functions as the object of a verb or preposition (108) & (110). When the wh-phrase functions as the subject of a VPE relative, it will be base-generated in [Spec,TP] (109). The exact case status of the wh-pronoun and its position at Spell-Out depends on the function of the wh-pronoun in the relative clause, and on our basic analysis of free relatives (cf. Section 9.6.1). When the wh-pronoun functions as the subject or as the object of a verb, the relative clause will be dominated by a DP. Depending on our analysis of free relatives, this DP will either be headed by pro, which needs to be identified through φ-feature agreement with the wh-pronoun (111), or by the wh-pronoun itself (112). (111)

DP ei D CP ei proi [φ] DP C’ wh-pronouni 3 [φ] C TP [φ] 5 I will e she pleases e

(112)

DP ei D CP 1 2 Di D DP C’ wh-pronoun ! 2 D C TP ti 2 I will e she pleases e

The surface structure of VPE and NCA relatives dominated by a DP, in a pro-head approach to free relatives (109), and in a wh-head approach (110) (cf. Section 9.6.1)

As discussed in Section 9.6.1, the wh-phrases can occur in ellipsis constructions because the missing predicate and the wh-chain are present at Semantic Form (cf. Chao 1987:148-155; Lobeck 1995:32-35). Wh-pronouns in VPE and NCA relatives will therefore always be subject to Arg-Case requirements related to the missing predicate. Both the pro-head approach (111) and the wh-head approach (112) predict that the form of the wh-pronoun in subject and object relatives involving ellipsis should also be influenced by matrix Arg-

CHAPTER 9

347

Case, because the free relative is headed either by the wh-pronoun itself (112), or by a pro that is identified with the wh-pronoun (111). Since the argument status of the free relative in the matrix clause is identical to the argument status of the wh-pronoun in VPE and NCA relatives, matrix and relative-internal ArgCase will always reinforce each other. When the wh-pronoun and the free relative function as the highest argument (= subject) of a predicate, the wh-pronoun will be linked to nominative Arg-Case. When the wh-pronoun and the free relative function as the object of a verb, the wh-pronoun will be linked to objective Arg-Case. For NCA relatives and VPE relatives where the wh-pronoun functions as the object of a verb, both the pro-head and the wh-head approach predict that the wh-pronoun will be unable to check relative-internal Pos-Case. In a prohead analysis, the wh-pronoun appears in [Spec,CP] at Spell-Out, but C is unable to check nominative Pos-Case, because the subject intervenes between C and T at Spell-Out (113)-(114). Since the wh-pronoun appears in a CP-internal position not covered by Pos-Case at Spell-Out, it will be influenced by relativeinternal Def-Case requirements. (113)

DP (114) DP 3 3 D CP D CP proi 3 proi 3 DP C’ DP C’ wh-pronouni 2 wh-pronouni 2 C TP C TP [+ finite] 2 [+ finite] 2 T’ ! DP T’ ! DPj ! I 2 ! she 2 z-=- T VP z__=___ T VP 4 4 5 tj pleases e willk tk e

Tree diagrams illustrating the lack of surface adjacency between C and T in a pro-head analysis of VPE relatives (113) and NCA relatives (114) where the wh-pronoun functions as the object of a verb

In a wh-head analysis (112), the wh-pronoun heads the DP dominating the relative CP. As the head of the DP, the wh-pronoun is immune to relativeinternal Pos-Case and Def-Case influences, but it could be influenced by matrix Pos-Case and Def-Case requirements. The word order in V-particle constructions suggests that object relatives involving VPE or NCA will generally remain in VP-internal position (115). Assuming that VPE and NCA object relatives are unable to raise to [Spec,vP], the wh-pronoun heading the relative

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

348

will be influenced by matrix Def-Case, but not Pos-Case (116). (115) a.*? Throw [who you must] out. b. Throw out [who you must].

c.*? I’ll throw [who I want] out. d. I’ll throw out [who I want].

(116)

FP The syntactic status of a wh-pronoun introducing 2 F vP a VPE or NCA relative that functions as the object 1 2 of a verb, if we assume that the wh-pronoun heads vl F e v’ the relative construction (wh-head approach) 1 3 v TelP Vk v throw tl 3 Tel VP out wo DP V’ ei 4 D CP tk 1 3 DP C’ Di D 3 who ! D you must e I want e ti

Null Complement Anaphora only seems possible in relatives where the wh-pronoun functions as the object of a verb or preposition, but VP ellipsis also occurs when the relativised constituent is a subject. VPE relatives with wh-subjects always follow a verb or verbal projection. When a VPE relative functions as the subject of a finite clause (117), it will appear after the verb and any complements at Spell-Out. (117) I will say my say to the end, mock at it [who may] (John Ruskin 1904[1867], Time and tide:94) [Jespersen 1949[1927]:63]

This surface order is most likely to result from raising the verbal projection mock at it past the free relative. If we adopt a wh-head approach to free relatives and assume that the relative has raised to [Spec,TP] (with the VP in a specifier position within the C-system), the wh-pronoun will be able to check Pos-Case with the matrix T, and is therefore predicted to surface in the nominative form who. If the relative remains in [Spec,vP], the wh-pronoun will be influenced by the Def-Case constraint, which calls for the objective form whom. However, the wh-pronoun will also be subject to relative-internal and matrix Arg-Case requirements, both of which call for the nominative who in

CHAPTER 9

349

this context, because both the wh-pronoun and the free relative itself function as the subject of a clause. Since matrix Def-Case is unlikely to override the joint demands of matrix and relative-internal Arg-Case, the wh-pronoun would again be predicted to surface as who. The function of the wh-phrase in a VPE and NCA relatives always corresponds to the function of the relative in the matrix clause. In (118), the whphrase functions as the subject of the VPE relative, which suggest that the free relative itself functions as the subject of a small clause following let. (118) Be good, sweet maid, and let [who can] be clever (Charles Kingsley 1856, ‘A farewell. To C.E.G.’) [Jespersen 1949[1927]:56]

The only potential matrix Pos-Case position available in sentences like (118) would be a [Spec,vP] position associated with let. However, if free relatives do not raise to [Spec,vP] when they function as the object of a verb (115)-(116), it would seem unlikely that VPE relatives that function as the subject of a small clause would be able to raise to the [Spec,vP] position in the matrix clause. The clear preference for who in sentences like (118) certainly suggests that the whpronoun is unable to check objective Pos-Case in VPE relatives functioning as the subject of a small clause (119) (cf. Section 8.8 for a more detailed discussion of pronoun case in small clauses). (119)

VP wh-head analysis of a VPE relative functioning wo as the subject of a small clause V VP let qp DP V’ ei 2 D CP V AP 1 3 be clever DPi C’ Dj D 3 who g D C TP 3 tj DP T’ 5 ti cank tk e

Since VPE subject relatives do not appear in Pos-Case positions at Spell-Out, the wh-head approach will predict that the form of the wh-pronoun should be influenced by matrix Def-Case, which calls for the objective form whom, as well as matrix and relative-internal Arg-Case, which call for the nominative who. Given that matrix Def-Case is unlikely to override the combination of

350

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

matrix and relative-internal Arg-Case, we would expect who to be strongly favoured, which does indeed seem to be the case. Unlike the wh-head approach outlined in (119), a pro-head approach predicts that a wh-pronoun functioning as the subject of a VPE relative checks nominative Pos-Case within the relative clause, because the wh-pronoun occupies [Spec,CP], and finite C is adjacent to T at Spell-Out (120). Since the wh-pronoun occupies a CP-internal position at Spell-Out, it will not be influenced by matrix Pos-Case or Def-Case in a pro-head analysis. (120)

DP 3 The Pos-Case predictions of a pro-head approach for D CP VPE relatives where the wh-pronoun functions as proi 3 the subject of the relative clause DPj [nom] C’ 3 whoi C [nom] TP [+ finite] 3 ! DP T’ 3 ! tj z-_---- T VP 4 4 tk e cank

When the wh-pronoun functions as the object of a preposition in a VPE or NCA relative, the preposition is always pied-piped (121) (cf. also (101b), (102b), and (106)). (121) a. M Paul, then, might dance with whom he would (Charlotte Brontë 1867[1852], Villette:135) [Jespersen 1949[1927]:63]

b. I’ll dance with whom I please. This obligatory pied-piping, and the absence of an additional matrix preposition, suggests that the relative clause is dominated by a PP rather than DP when the relative pronoun follows a preposition (cf. Section 9.6.1). If we assume that free relatives are always headed by an empty pronominal constituent, VPE and NCA relatives involving a preposition will be headed by a pro-P (122). If we assume that the head of the phrase in [Spec,CP] undergoes head-movement to head the free relative construction at Spell-Out, prepositional VPE and NCA relatives will be headed by the preposition itself (123).

CHAPTER 9

351

(122)

PP (123) PP ei 3 P CP P CP 1 3 pro-Pi ei [P] PP C’ Pi P PP C’ 2 with 1 2 2 P DP C TP P DP C TP withi whom [P] 5 ti whom 5 [P] he would e he would e I please e I please e

Both analyses predict that the wh-pronoun will be influenced by Arg-Case requirements associated with the preposition, and also by relative-internal DefCase, because objects of prepositions are unable to check Pos-Case within the prepositional phrase. The wh-pronoun will not be influenced by any matrix case requirements, because it appears in a CP-internal position that has no link to the matrix clause. Since both the Arg-Case and the Def-Case constraint require objects of prepositions to surface in their objective form, the occurrence of the nominative who in examples like (124) suggests that the distribution of wh-forms in VPE and NCA relatives is also influenced by the trend towards invariant who. (124) Let the patent be bought by who it will (Daniel Defoe 1890, The complete gentleman:45) [Jespersen 1949[1927]:63]

The distribution of wh-forms in the VPE and NCA relatives discussed so far can be captured equally well in a wh-head approach as in a pro-head approach. However, Jespersen’s (1949[1927]:63) discussion of free relatives involving ellipsis also includes examples like (125), whose occurrence is predictable in a pro-head approach (126), but impossible if we assume that all free relatives must be headed by an overt element (i.e. either a wh-pronoun or a preposition). (125) a. I shall accept [whose company I choose] (Thomas Hardy 1912[1878], The return of the native, London:313)

b. they are all at liberty to be the recipients of [whose smiles they please] (Charlotte Brontë [1847], Jane Eyre:222)

352 (126)

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

DP ei pro-head analysis of an NCA relative D CP with a genitive wh-pronoun proi wo [φ] DP C’ 3 2 DPj D’ C TP whose 5 [φ] 6 they please e tj smilesi [φ]

As discussed in Section 2.2.3 and Section 9.1, genitive DPs are assumed to be base-generated in [Spec,NumP], and check genitive Pos-Case in [Spec,DP]. This means that a genitive DP will always occupy a specifier position within the noun phrase. When the genitive DP is a relative pronoun, the noun phrase containing the genitive DP will itself occupy a specifier position, namely [Spec,CP]. A wh-head analysis of (125) would therefore have to involve illicit head-movement out of the specifier of a specifier. The possible occurrence of NCA relatives of the type illustrated in (125) could be thus be seen as evidence that free relatives do not necessarily have to have an overt head. It is however important to bear in mind that such NCA relatives are rare, and may be influenced by Latin constructions (cf. Jespersen 1949[1927]:63). 9.7

wh-pronouns in headed relatives The use of wh-pronouns in headed relatives is a comparatively recent development (cf. Gelderen 2004:81-88). In Old English, headed relatives were introduced by the complementizer ðe and/or a form of the demonstrative se (Allen 1980:75-91). Se was a definite determiner as well as a demonstrative in Old English, and its full paradigm of case forms is given in Section 1.1. When a headed relative was introduced by se alone (127), the demonstrative generally surfaced in the case corresponding to its function in the relative clause (Allen 1980:84). (127) Ða man ofsloh ðes Caseres gerefan then one killed the emperor’s reeve.SG.MASC.ACC [se was Labenius gehaten] DEM.SG.MASC.NOM was Labenius called “Then they killed the king’s reeve, who was called Labenius.” (Peterborough Chronicle Prologue) [Allen 1980:83]

In relatives introduced by se ðe, on the other hand, the demonstrative could surface either in the case associated with its function in the relative clause

CHAPTER 9

353

(128), or with the case assigned to the head noun in the matrix clause (Allen 1980:86ff). According to Allen (1980:87f), the demonstrative introducing the relative clause tended to attract the case of the head noun only if the head noun bore dative or accusative case (129)-(130). (128) Swa swa Aaron wæs, se arwurda bisceop, as Aaron was, the.SG.MASC.NOMworthy bishop [ðone ðe God sylf geceas] DEM.SG.MASC.ACC that God self chose “As Aaron was, the worthy bishop, whom God himself chose” (Ælfric Lives XXIX.190) [Allen 1980:87]

(129) Ne he bið Iudeum anum seald, ac he bid eallum not he is Jews.DAT alone.DAT given, but he is all.DAT ðeodum, [ðam ðe on God gelyfan willað] people.DAT DEM.PL.DAT that in God believe will “He is not given to the Jews alone, but to all people who will believe in God” (Vercelli V.182) [Allen 1980:87] (130) ic wat wytodlice ðæt ge secað ðone 1sg.NOM know truly that 2pl.NOM seek the.SG.MASC.ACC hælend [ðone ðe on rode ahangen wæs] saviour DEM.SG.MASC.ACC that on cross hung was “I know truly that you seek the Saviour, who was hung on the cross.” (St. Matthew 1766 (XXVIII.5)) [Allen 1980:87]

By the Early Middle English period, the demonstrative pronoun had disappeared from the construction, and the complementizers ðe and ðæt had become the favoured relative markers in headed relatives (Allen 1980:202-206).27 Research by Allen (1980:197ff) suggests that wh-pronouns started to appear in headed relatives during the Middle English period, and were at first used mainly after prepositions (131), and in genitive contexts (132) (see also Gelderen 2004:87).

27 Gelderen (2004:81-87), presents evidence which suggests that ðæt originally started out as a demonstrative but replaced ðe as a complementizer in the 13th century, before wh-pronouns started to be used in headed relatives.

354

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

we beoð (131) And alle ðeos weren min eldre [of wan and all these were my ancestors of wh.OBJ 1pl.NOM are ispronge] descended “And all these were my ancestors, of whom we are descended” (Layamon (Caligula) 25081) [Allen 1980:200]

(132) Eadi is his spuse, [hwas meiðhad is unwemmet] blessed is his spouse wh.GEN maidenhood is untouched “Blessed is his spouse, whose maidenhood is untouched” (Hali Meidenhad 578) [Allen 1980:200]

While the use of wh-pronouns in headed non-subject relatives had become quite common by the middle of the 13th century, headed subject relatives introduced by a wh-pronoun remained extremely rare up to the Early Modern English period (Allen 1980:200,202; Householder 1986:151). In Middle English, the case of the wh-pronoun in a headed relative was generally determined by its function within the relative clause. Since the use of wh-pronouns in headed relatives was largely confined to objective and genitive contexts, the objective hwæm ‘whom’ and the genitive hwæs ‘whose’ predominate in headed wh-relatives in Middle English (Allen 1980:199; Jespersen 1949[1927]:80; Mustanoja 1960:200f; Householder 1986:151; Gelderen 1997:73,78). According to Householder (1986:151,159,162), the nominative wh-form who first appeared in non-restrictive subject relatives such as (133), but subsequently also spread to restrictive subject relatives (134). (133) by the grace of God, [who have you ever in his keeping] (Paston letters: common closing formula) [Mustanoja 1960:200]

(134) Above all, they constantly attended those committees of senators [who are silent in the house and loud in the coffee-house] (Jonathan Swift 1909[1704], A tale of a tub, The battle of the books and other satires:54)

Interestingly, in some of the 15th century Paston letters, the subject relative in the closing formula appears with the objective form whom (135) rather than the nominative who (133). (135) by the grace of God, [Whom have yow in Hys kepyng] (Paston letters III 238 [1478]) [Mustanoja 1960:200n.1]

CHAPTER 9

355

While the choice of whom in (135) could be due to its status as an invariant whform for headed relatives, as suggested by Gelderen (1997:78), it is also possible that the use of the objective form is due to case agreement with the head noun God, which functions as the object of the preposition of. Shakespeare still exhibits a clear preference for that in restrictive subject relatives, but Ball (1996:248-250) reports a sharp increase in the use of who in this context from the 18th century onwards. Virtually all of the British and American authors in Householder’s (1986:159) corpus of 20th century mystery writers favour who over that in restrictive subject relatives, and Householder (1986:159) notes a similar preference for who in the Brown University Corpus of 20th century American English texts, and in a corpus of spoken British English compiled by Randolph Quirk in the 1960s. In the Wellington Corpus of Written New Zealand English who occurs in 97.7% of 1316 unembedded restrictive subject relatives with a human antecedent, but that is slightly favoured over which when the antecedent is inanimate (Sigley 1997:273). 9.7.1 The extension of who to objective contexts Given the comparatively late emergence of headed subject relatives introduced by who, it is not surprising that authors who favour who in interrogatives questioning the object of a verb or preposition still retain the objective form whom in corresponding headed relatives (cf. Section 9.1). In Present-Day English, overt wh-pronouns occur primarily in non-restrictive relatives. In nonrestrictive relatives, an overt wh-pronoun is favoured not only when the relativised constituent is the subject of the clause (133), but also when it is the object of a verb (136), or a stranded or pied-piped preposition (137)-(138) (cf. Sigley 1997:273). (136) a. To Canada, [whom we love and prize] (Alfred Tennyson 1894, Poetical works:575) [Jespersen 1949[1927]:121]

b. the boy, [who they called Xury] (Daniel Defoe 1719, Robinson Crusoe, (1883 Facsimile ed.):25) [Jespersen & Haislund 1949:243]

(137) Even John Arnold, [whom I confided in] … has proved an execrable villain (Richardson, Pamela:1.144) [Visser 1963:402] (138) But thou, [to whom my jewels trifles are] (Shakespeare, Sonnets:48.5) [Jespersen 1949[1927]:192]

356

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

In corresponding restrictive relatives, the use of a wh-form is obligatory only after a pied-piped preposition (139), and when an adverbial clause precedes the subject of the relative (140) (cf. Sigley 1997:273). (139) a. a tutor [for whom you have absolute reverence] (John Ruskin 1904[1864], Sesame and lilies:124) [Jespersen 1949[1927]:192]

b. I had nobody [to whom I could in confidence commit the secrecy of my circumstances to, and could depend upon for their secrecy] (Daniel Defoe [1722], Moll Flanders:108) [Jespersen 1949[1927]:193]

(140) She is the daughter of a lady, [who, when she was a beautiful girl herself, and I was very many years younger, I loved very dearly] (Charles Dickens 1900[1839], Nicholas Nickleby:566) [Jespersen & Haislund 1949:243]

As (139b) illustrates, pied-piping may cooccur with preposition stranding in headed relatives. The occurrence of two instances of the same preposition in the relative could be seen as supporting evidence for an analysis of movement as feature-copying (cf. Chomsky 1995:251ff). If we assume that movement involves the copying rather than displacement of a constituent to a new position, the cooccurrence of the pied-piped and stranded to in (139b) could be argued to arise from the spelling out of both copies of the preposition. Wh-pronouns are fairly common in restrictive subject relatives (134), but when the relativised constituent is the object of a verb or stranded preposition, the wh-pronoun is preferentially omitted, especially when the antecedent is human (141)-(142) (cf. Householder 1986:158; Sigley 1997:273). (141) Mr. Clutterbuck, you are speaking of the man [Ø I love]. (P.G. Wodehouse 1992[1956], French leave:205)

(142) No doubt he reminds him of the horrible men [Ø he used to go about London with in his younger days] (P.G. Wodehouse 1929, Summer lightening:25) [Visser 1963:404]

As can be seen from the examples in (136)-(137), (140), and also (143), the wh-object of a verb or stranded preposition in a headed relative may surface either in the objective form whom or in the nominative who, although who tends to be favoured in Present-Day English.

CHAPTER 9

357

(143) a. a man [whom no pure-minded girl should be allowed to know] and [whom no chaste woman should sit in the same room with] (Oscar Wilde [1891], The picture of Dorian Gray:154) [Jespersen 1949[1927]:189]

b. the man [who we last saw on the platform] (H. Seton Merriman [H.S. Scott] 1905[1896], The sowers:21) [Jespersen & Haislund 1949:243]

When the wh-pronoun occurs after a pied-piped preposition (138)-(139), on the other hand, the objective form whom is strongly favoured. Sigley’s (1997:273) search of the Wellington Corpus of New Zealand English yielded 57 headed relatives with a human antecedent and pied-piped preposition, and in every one of them the wh-pronoun surfaces in the objective form whom. Similarly, all of the examples collected by Jespersen (1949[1927]) involve whom, and Householder’s (1986:158) survey of Modern English texts and corpora yielded many examples of whom in headed relatives involving pied-piping, but not a single instance of who. 9.7.2 The extension of whom to nominative contexts Although most existing studies of who and whom have focused on the use of the nominative form who in objective contexts, we also find instances of the objective whom where we might expect who. Jespersen (1934[1924]:117, 348-350), Kayne (1984:2-6,19n.10), Radford (1988:575f), Sigley (1997:68f, 273ff), and Lasnik & Sobin (2000:345,356f) all point out that both who and whom occur in complex headed relatives where the wh-pronoun functions as the subject of an embedded clause, no matter whether the embedded clause is finite (144)-(145) or non-finite (146).28 The occurrence of who and whom is more common when the complex relative is non-restrictive (145) & (146a) rather than restrictive (144), because the wh-pronoun is frequently omitted in complex restrictive relatives such as We feed children [we think are hungry] (cf. Jespersen 1934[1924]:349f). (144) a. We feed children [whoi we think [ti are hungry]]. b. We feed children [whomi we think [ti are hungry]]. (Jespersen 1934[1924]:348) 28 Jespersen (1949[1927]:196) terms this kind of construction ‘relative concatenation’, and notes that it tends to occur with verbs like say, hear, fear. According to Jespersen (1949[1927]: 198), authors such as Benjamin Franklin alternate freely between who and whom in headed relatives of this type.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

358

(145) a. except of course your esteemed mother, [whoi we all agree [ti is perfect]] (Arnold Bennett 1922, Mr. Prohack:258) [Jespersen 1949[1927]:201]

b. There was a man, too, [whomi she had only just time to realize [ti was the doctor, not the undertaker]] (Compton Mackenzie 1922[1912], Carnival:7) [Jespersen 1949[1927]:198]

(146) a. Prince William [whoi everyone expected [ti to run amok in the abbey]] (Wellington Corpus of Written New Zealand English:C14 062) [Sigley 1997:275]

b. they murdered all they met [whomi they supposed [ti to be gentlemen]] (Charlotte M. Yonge, Kings of England:125) [Jespersen 1949[1927]:87f]

This variation between who and whom in complex relative constructions contrasts with the obligatory selection of who in simple headed relatives where the wh-pronoun functions as the subject of a finite clause, no matter whether the relative is restrictive (147) or non-restrictive (148). Jespersen (1949[1927]: 199f) and Lasnik & Sobin (2000:345n.4) observe that who is also strongly favoured when the wh-subject in a simple headed relative is followed by a parenthetical (149) (see also Kayne’s 1984:3,19n.10 observations about parentheticals and case in headed relatives). (147) a. We feed children [who are hungry]. b. * We feed children [whom are hungry]. (Jespersen 1934[1924]:349) (148) a.

deserted by all the company except the uncle of young Nightingale, [who loved his bottle as well as Western himself] (Henry Fielding 1981[1749], The history of Tom Jones:631)

b. * except the uncle of young Nightingale, [whom loved his bottle as well as Western himself] (149) a.

There was one H-, [who, I learned, in after days, was seen expiating some maturer offence in the hulks] (Charles Lamb 1899[1823,1833], The essays of Elia:1.25) [Jespersen 1949[1927]:200]

b.*? There was one H-, [whom, I learned, in after days, was seen expiating some maturer offence in the hulks]

CHAPTER 9

359

Although example (148a) is taken from an 18th century text, it is also perfectly grammatical in Present-Day English. (148b) is clearly ungrammatical in Present-Day English. Similarly, (149a) would seem to be clearly preferable to (149b) in Present-Day English, although both have a rather literary, archaic flavour, which may affect our grammaticality judgments. Since it is impossible to obtain grammaticality judgments from 18th century speakers, we can’t ever know for certain whether sentences like (148b) & (149b) would have been equally ungrammatical in 18th and 19th century English (thanks to Diane Massam for drawing my attention to this issue). The ready occurrence of examples such as (148a) & (149a), and apparent absence of examples like (148b) & (149b) from 18th-19th century texts, does however suggest that whom was at least disfavoured in these contexts. It is also worth noting that the extension of whom to embedded subject relatives appears to be largely confined to written texts in Present-Day English. If a subject wh-pronoun preceding a parenthetical had the same case status as the fronted wh-subject of an embedded relative clause, the literary flavour of (149) should actually predispose present-day speakers towards favouring whom over who. The present-day preference for who over whom before the parenthetical in (149) would thus seem to provide quite strong evidence that the case status of the wh-pronoun in sentences like (149) is different from the case status of the fronted wh-subject in examples like (150). (150) Martin had obtained a flashlight and was desperately searching for Hanna [whomi he thought [ti was in her cabin below]] (Wellington Corpus of Written New Zealand English:F19 112) [Sigley 1997:68]

9.7.3. Implications for formal analyses of headed relatives Headed relatives differ from free relatives in that the wh-pronoun introducing a headed relative quite clearly occupies a CP-internal positition at Spell-Out, and does not seem to be able to undergo head-movement out of the relative CP. This would lead us to expect that the distribution of wh-forms in headed relatives should be identical to the distribution of wh-forms in embedded questions (cf. Section 9.4), because the wh-pronoun appears in [Spec,CP] in both types of wh-construction. When the wh-pronoun functions as the subject of a simple headed relative, it will be linked to nominative Arg-Case, and will also be able to check nominative Pos-Case, because it heads the wh-phrase in [Spec,CP], and no overt constituent intervenes between finite C and T at

360

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Spell-Out. The case constraints thus correctly predict that the wh-pronoun will always surface in the nominative form who in simple subject relatives.29 As discussed above, who is also clearly favoured in headed subject relatives where the wh-pronoun is followed by a parenthetical (151). (151) John, [who, as I found out later, had already won the race twice] Since finite C cannot inherit the ability to check nominative Pos-Case if it is separated from T by an overt constituent at Spell-Out, the parenthetical in relatives like (151) is best analysed as a C’- rather than TP-adjunct (152). As mentioned in Section 9.1, I am assuming that the ability to check nominative PosCase can be passed on to C through a PF merger with T under adjacency. Unlike Bobaljik (1994:2), who argues that intervening adjuncts have no bearing on the kind of adjacency required for a PF merger between two nodes, I am assuming that C is only able to inherit the relevant Pos-Case features from T when absolutely no overt constituent intervenes between C and T at Spell-Out. If the parenthetical in (151) was treated as a TP-adjunct, it would prevent C from acquiring the ability to check nominative Pos-Case on a DP in [Spec,CP], and the strong preference for who in this context would be puzzling. As a C’adjunct, on the other hand, the parenthetical will not intervene between C and T at Spell-Out, and is also unable to interfere with spec-head agreement between C and the wh-pronoun in [Spec, CP]. The proposed approach could be seen as a rough minimalist equivalent of Taglicht’s (1998) HPSG analysis of parentheticals, which assumes that a parenthetical node “is always a noninitial 29 Law (2000:182) argues that relative pronouns do not occupy [Spec,CP] at Spell-Out, but instead appear in a position adjoined to TP (in finite clauses) or VP (in to infinitives). In such an approach, only an assumption of syntactic equivalence between adjuncts and specifiers of TP will allow us to account for the consistent selection of the nominative who in simple headed subject relatives, and still retain the generalisation that all wh-phrases occupy the same surface position in headed relatives. Since a wh-pronoun is only able to check nominative Pos-Case if it occupies [Spec,TP] or [Spec,CP] at Spell-Out, the wh-subject of a simple headed relative must be analysed as occupying [Spec, TP], if it is to receive nominative Pos-Case in Law’s analysis. If we want to argue that all wh-phrases occupy the same surface position in headed relatives, we will therefore have to assume that the TP-adjoined position occupied by whpronouns in finite clauses is equivalent to [Spec,TP]. For relative pronouns that function as the object of a verb or stranded preposition, or as the subject of an embedded clause, Law’s analysis will yield the same case predictions as the [Spec,CP] analysis adopted here, provided we assume that a given agreement-related functional head can only check Pos-Case once (cf. Section 9.2). This will ensure that the wh-pronoun is unable to check Pos-Case in relatives that already contain an overt subject even if a TP-adjunct is treated as syntactically equivalent to a specifier of TP (cf. Kayne 1994:16f; Law 2000:194fn.11).

CHAPTER 9

361

and nonfinal daughter of its mother” (196,206), “for which the grammar specifies no function in relation to any sister node” (195,205). In headed subject relatives where the wh-phrase functions as the subject of an embedded clause, by contrast, the wh-pronoun will be unable to check nominative Pos-Case, because the subject of the higher clause intervenes between C and T at Spell-Out (153). Since the wh-pronoun in (153) does not occupy a Pos-Case position at Spell-Out, it will receive objective Def-Case as well as the nominative Arg-Case which encodes its status as the highest structural argument of the embedded predicate. (152)

CP (153) CP ei 3 C’ DPj C’ DPi [nom] who ei who(m) 3 CP C’ C TP [+ finite] 3 6 3 as I found out later C [nom] TP ! DPi T’ [+ finite] 2 ! they 2 ! DP T’ z---=-- T FP ! ti 2 6 ti thought [tj had z----- T VP hadj 6 won the race] already tj wonk ti tk the race twice

ti knew [tj to have won the race]

In some earlier discussions of wh-case in complex subject relatives, the use of whom was argued to arise from objective case assignment to the whpronoun by the higher verb (i.e. thought or knew in (153)) in the course of the derivation (cf. Kayne 1984:5f,19n.11; Radford 1988:575f). Such an analysis is ruled out in the approach proposed here, because neither Arg-Case nor PosCase is designed to encode intermediate syntactic relations between a head and a DP that does not function as the argument of this head. As discussed in Chapter 2, Pos-Case encodes the surface relation between an overt DP and an agreement-related functional head, and Arg-Case only applies to arguments of a predicate. Since the wh-pronoun in (153) neither functions as an argument of the higher verb (thought, knew), nor appears in an associated [Spec,vP] position at Spell-Out, the case properties of the higher verb can have no bearing on the case form the wh-pronoun surfaces in. When the wh-pronoun functions as the object of a verb or stranded preposition in a headed relative, it will be linked to objective Arg-Case, and it will again be unable to check Pos-Case, because the subject of the relative clause intervenes between C and T at Spell-Out (154). In headed relatives

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

362

where the preposition is pied-piped with the wh-pronoun, the wh-pronoun is likewise unable to check Pos-Case, because it appears as the complement of the preposition at Spell-Out (155), and prepositions are unable to check PosCase (cf. Section 2.2.4). (154)

CP (155) CP 3 ei DPj C’ PPj C’ who(m) 3 2 2 C TP P DP C TP [+ finite] 3 in who(m) [+ finite] 2 T’ DPi T’ ! DPi 2 she 2 ! they z--=-- T FP T FP 6 6 calledk ti tk tj Xury confidedk ti tk in tj

confidedk ti tk tj

The absence of Pos-Case checking on the wh-pronoun in (154) and (155) means that any wh-pronoun that functions as the object of a verb or preposition in a headed relative is influenced by the Arg-Case and Def-Case constraints, both of which call for the objective form whom in this context. The occurrence of who in non-subject relatives would seem to suggest that the case constraints may be overridden by the trend towards invariant who. The particular preference for whom after prepositions could be argued to provide evidence for the influence of Relative Positional Coding, which calls for the robust form whom in the asymmetrically c-commanded complement position and the gracile who in the asymmetrically c-commanding specifier position (cf. Section 9.5 for a similar trend in sluiced questions). However, nothing in the analysis presented so far could explain why the tendency towards who should be weaker in headed relatives than in embedded questions.30 30 The analysis proposed by Lasnik & Sobin (2000) faces the same problem. Lasnik & Sobin (2000) argue that who is the basic form of the wh-pronoun, which can check either nominative (NOM) or objective case (ACC). The form whom is assumed to contain an additional ACC feature, which is associated with the suffix -m and has to be checked independently of the case feature of the stem who (Lasnik & Sobin 2000:354). The ACC feature of the suffix must be checked either by the Basic ‘whom’ Rule, which licenses whom after a verb or preposition, or by the Extended ‘whom’ Rule, which licenses whom in any contexts where a subject noun phrase intervenes between the wh-pronoun and the nearest theta-role assigning verb. Since the Extended ‘whom’ Rule predicts the use of whom in any type of wh-construction where the whpronoun functions as the subject of an embedded clause, or as the object of a verb or stranded preposition, Lasnik & Sobin’s (2000) approach would predict that whom should be equally favoured in non-subject wh-questions and non-subject relatives.

CHAPTER 9

363

If we assume that wh-pronouns in headed relatives are subject to the same clause-internal case influences as wh-pronouns in embedded questions, then we must look to the relation between the wh-pronoun and the matrix clause for an explanation of the case differences. McCreight (1988:74f) suggests that case agreement between the head of a relative and the relative pronoun can lead to multiple case assignment to the relative pronoun: “a relative pronoun may be assigned case within its clause and then also acquire the case originally assigned to its antecedent in the main clause” (McCreight 1988:17f). As mentioned at the start of Section 9.7, case agreement between the relative pronoun and its antecedent was common in Old English relatives introduced by se ðe (129)-(130). It is also attested in Latin, Ancient Greek, and Old High German (cf. Bianchi 2000:58). In the approach proposed here, the only case a relative pronoun could inherit from its antecedent is the Arg-Case of the antecedent. Since the Pos-Case and Def-Case status of a DP is primarily determined by its position relative to certain agreement-related functional heads at Spell-Out, Pos-Case and Def-Case cannot be transmitted from (the head of) a DP in one position to (the head of) a DP in another. If we assume that wh-pronouns are able to inherit the Arg-Case of their antecedent, the case status of wh-pronouns in headed relatives will be similar to that of whpronouns introducing a free relative in a pro-head analysis of free relatives (cf. Section 9.6.1). When a headed relative modifies the object of a verb or preposition, the wh-pronoun will inherit objective Arg-Case from its antecedent, and when the headed relative modifies a subject, the wh-pronoun will inherit nominative Arg-Case. Since Pos-Case generally overrides Arg-Case in Present-Day English (cf. Section 2.3), the influence of matrix Arg-Case requirements is most likely to be evident when the wh-pronoun is unable to check nominative Pos-Case, i.e. in headed relatives where the wh-pronoun functions as the object of a verb or preposition (154)-(155), and also in headed relatives where the wh-phrase functions as the subject of an embedded clause (153). The greater tolerance of whom in headed relatives than in embedded questions can thus be argued to arise from the reinforcement of relative-internal Arg-Case and Def-Case through objective matrix case.31 The influence of objective matrix case would 31

In headed relatives where the wh-phrase functions as the subject of an embedded clause, the objective Arg-Case assigned to the antecedent of the wh-pronoun will reinforce relativeinternal Def-Case. In headed relatives where the wh-pronoun functions as the object of a verb or preposition, the objective matrix case will reinforce relative-internal Def-Case and Arg-Case requirements. The occurrence of matrix Arg-Case on relative pronouns that function as the subject of a simple headed relative in Old English (129)-(130) suggests that Pos-Case was less influential in earlier periods of English than it is now (cf. Chapter 10 for further discussion).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

364

be predicted to outweigh the influence of nominative matrix case, because the wh-pronoun can inherit objective Arg-Case from a wider range of antecedents: both objects of verbs and objects of prepositions receive objective Arg-Case, but nominative Arg-Case only appears on subjects. The case data discussed in Section 9.7.2 indicates that the Arg-Case assigned to the antecedent of the wh-pronoun has a similar influence on the distribution of wh-forms in restrictive headed relatives as in non-restrictive headed relatives. This suggests that the differences between restrictive and nonrestrictive relative clauses are largely irrelevant to case agreement between the wh-pronoun and its antecedent. One way to capture the wh-case similarities between restrictive and non-restrictive relatives would be to assume that the syntactic relationship between the wh-pronoun and its antecedent at Spell-Out is the same in restrictive relatives as in non-restrictive relatives (cf. Kayne 1994: 110f). Such an approach would allow us to restrict case agreement between a wh-pronoun and its antecedent to a single syntactic configuration. Kayne (1994:87-90,110f) and Bianchi (2000:61-69) both advocate an approach to headed relatives that is reminiscent of the syntactic analysis adopted for free relatives in Section 9.6.1. The relative CP is immediately dominated by a DP, and the wh-pronoun in [Spec,CP] is assumed to be governed by the head of this DP. The nominal projection ‘heading’ the relative is base-generated as the complement of an overt or covert wh-pronoun, and raises to the specifier of the wh-phrase before Spell-Out (156). (156)

DP Approach to headed relatives advocated by ei D CP Kayne (1994) and Bianchi (2000) 32 the wo C’ DPi 3 ei D’ we thought [ti would leave early] NumPj 5 2 three guests D NumP whom tj

If both restrictive and non-restrictive relatives have the structure outlined in (156), then the transmission of matrix Arg-Case to the wh-pronoun in a headed relative can be argued to depend on a government relation between the D heading the relative structure, and the wh-pronoun in [Spec,CP] (cf. the pro-head 32 Kayne (1994:90f) and Bianchi (2000:62) assume that the nominal projection base-generated as the complement of the wh-pronoun is NP, but if we adopt the approach to noun phrase structure outlined in Section 2.2.3, the category of the complement must be NumP.

CHAPTER 9

365

analysis of free relatives presented in Section 9.6.1). However, the assumption that the antecedent of the wh-pronoun is generated as a complement of the whpronoun itself is rather problematic, especially when the the antecedent is a proper noun or a personal pronoun.33 Proper nouns and pronouns are clearly DPs rather than NPs or NumPs, and personal pronouns are generally assumed to occupy D at Spell-Out (cf. Chapter 3 and Section 8.14 for further discussion). This means that we will either have to assume that the nominal projection base-generated as the complement of the wh-pronoun may have the category DP rather than NumP (157), or that pronouns can undergo head-movement out of the specifier of the wh-phrase to the head of the DP dominating the headed relative (158). (157)

DP (158) DP 3 wo D CP D CP wo wo 1 DPi C’ Nk D DPi C’ 2 ei she 3 ei D’ we thought [ti would NumPj D’ we thought [ti would DPj she 2 leave early] 2 1 leave early] D DP Num NP D NumP 4 whom tj whom tj tk

Both of these options would weaken our model of phrase structure and movement. If we want to assume that pronouns modified by a headed relative are base-generated as DPs, we will have to allow D to take DP complements as well as NumP complements. If we want to argue that personal pronouns can undergo head-movement out of the specifier of the wh-phrase, we will have to provide an account of how the trace of the pronoun can be licensed when it appears within the specifier of the wh-phrase. As discussed in Section 9.6, it is normally only the head of the phrase in [Spec,CP] that can raise to a higher head position, and in (158), this head is whom, rather than she. The drawbacks of the approach advocated by Kayne (1994) and Bianchi (2000) suggest that it might be preferable to assume that case agreement between a wh-pronoun and its antecedent is licensed by an interpretive rather than a syntactic relation (cf. Bianchi 2000:59; Alexiadou et al. 2000:3). This would allow us to adopt an adjunction analysis of headed relatives where the antecedent of the wh-pronoun is base-generated outside the relative CP, and the syn33 See Borsley (1997), Alexiadou et al. (2000:16-20), and Platzack (2000:276f,286f) for further discussion of Kayne’s (1994) analysis of headed relatives.

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

366

tactic relationship between the head of the antecedent phrase and the wh-pronoun is not one of strict head government. Unlike Kayne’s (1994:110f) approach, an adjunction analysis of headed relatives is able to capture the scopal differences between restrictive and non-restrictive relatives in the surface syntax (cf. McCreight 1988:73-83, Alexiadou et al. 2000:5,9): A relative clause adjoined to NP (159) will necessarily have a restrictive interpretation, and a relative clause adjoined to DP (160) will always be non-restrictive (cf. Demirdache 1991, Alexiadou et al. 2000:5). A relative clause adjoined to NumP will be restrictive if the antecedent is definite and thus projects a DP-layer (161), but it will be non-restrictive if the antecedent is indefinite and fails to project a DP-layer (162).34 (159)

NumP 3 Num NP 3 threei NP CP guests 6 C’ DPj whomi 6

(160)

DP 3 DP CP Johni 3 DPj C’ whomi 6

we thought [tj would leave early]

we thought [tj would leave early]

When the relative clause is adjoined to NP, its interpretation will always be restrictive (cf. Lobeck 1995:184).

(161)

DP 3 D NumP thei ei NumP CP 5 3 three guests DPj C’ whomi 6

When the relative clause is adjoined to DP, its interpretation will always be non-restrictive.

(162)

NumP ei NumP CP 2 3 C’ Num NP DPj threei guests whomi 6

we thought [tj would leave early]

we thought [tj would leave early]

When the relative clause is adjoined to NumP, its interpretation will be restrictive when the antecedent is definite (161), but non-restrictive when the antecedent is indefinite (162). 34

The interpretive relation between the wh-pronoun and its antecedent is indicated by coindexation. If we want to argue that the wh-pronoun may be influenced by matrix Arg-Case even when its antecedent fails to project a DP-layer, we will have to assume that Arg-Case is linked not to D (cf. Section 9.6.1), but to the highest head in the noun phrase. For definite noun phrases, this head is D (160) & (162), for indefinite noun phrases, it is Num (159) & (161).

CHAPTER 9

9.8

367

Summary and conclusions The case trends summarised in this chapter suggest that the variability of wh-pronoun case in Present-Day English resembles the case variability found with personal pronouns in strong pronoun contexts. The distribution of who and whom provides further evidence for the interaction between the three case constraints and Relative Positional Coding, and points to the existence of an Invariant wh-form constraint, which requires personal wh-pronouns to surface as who or whoever. Like strong personal pronouns, wh-phrases tend not to appear in canonical argument positions at Spell-Out. As a result, wh-pronouns are either entirely unable to check Pos-Case, or are subject to different Pos-Case requirements than weak personal pronouns. Possessive wh-pronouns are a notable exception, because they occupy [Spec,DP] at Spell-Out and thus always check genitive Pos-Case, just like possessive personal pronouns (see Section 9.1). In Chapter 2, I argued that an argument DP checks nominative Pos-Case if it occupies [Spec,TP] at Spell-Out in a finite clause, and its surface position is different from its θ-position. To be able to check objective Pos-Case, an argument DP must appear either in [Spec,vP], or in [Spec,TP] of a non-finite clause introduced by for. The position of wh-pronouns and free relatives in Vparticle constructions suggests that wh-pronouns are unlikely to head a DP in [Spec,vP] (cf. Sections 9.3 & 9.6). The occurrence of wh-constructions in [Spec,TP] of a non-finite clause introduced by for seems to be equally marginal. Wh-pronouns are thus generally unable to check objective Pos-Case. If we assume that wh-pronouns must either raise to [Spec,CP] before Spell-Out or be base-generated this position (cf. Sections 9.5 & 9.6.1 on sluiced questions and relative clauses involving VP ellipsis or Null Complement Anaphora), only a wh-pronoun introducing a free relative could conceivably appear in [Spec,TP] of a finite clause and check nominative Pos-Case in this position (163). As can be seen from the tree diagram in (163), the whpronoun will only be able to enter into nominative Pos-Case checking with the matrix T if it is analysed as heading the free relative DP at Spell-Out, i.e. if we adopt a wh-head approach (cf. Section 9.6.1). The distribution of wh-forms in questions and relative clauses suggests that wh-pronouns are also able to check nominative Pos-Case when they head the wh-phrase in [Spec,CP] of a finite clause and T has raised to C (164) or no overt constituent intervenes between C and T at Spell-Out (165). Nominative Pos-Case checking between finite C and a wh-pronoun in [Spec,CP] is impossible when an overt constituent intervenes between C and T at Spell-Out (166), and when the wh-pronoun fails to head the wh-phrase (167).

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

368 (163)

CP ei The Pos-Case predictions of a wh-head approach C TP to free relatives in [Spec,TP] of a finite clause [+ finite] wo DPk [nom] T’ ei 3 CP T [nom] …tk… D [nom] 1 3 DPi C’ Dj D whoever ! 3 D ti approached her I marry ti tj you find them with ti

(164)

CP 3

DPi [nom] C’ who 3 …ti… C [nom] 1 Tk C [+ finite] did The Pos-Case status of a wh-pronoun in matrix interrogatives questioning the object of a V or stranded P (cf. Section 9.1)

(165)

CP 3

DPi [nom] C’ who 3 C [nom] TP [+ finite] 2 ! DP T’ ! ti 2 z------ T … The Pos-Case status of a wh-pronoun in subject questions (cf. Sections 9.1 & 9.4), simple headed subject relatives (cf. Section 9.7.3), and free subject relatives if we assume that the relative DP is headed by pro (cf. Section 9.6.1)

(166)

CP (167) CP 3 wo C’ PPj C’ DPj who(m) 3 2 3 C TP P DP C … [+ finite] 2 in who(m) [+ finite] ! DPi T’ ! you 2 z---=- T …tj…

No Pos-Case checking is possible when an overt constituent intervenes between C and T at Spell-Out (166), or when the wh-pronoun appears within a pied-piped PP (167).

When a wh-pronoun fails to check Pos-Case, as in (166) and (167), its surface form will be determined by Arg-Case and Def-Case. The Def-Case constraint calls for whom in all positions not covered by Pos-Case. The ArgCase constraint stipulates that the wh-pronoun will surface as who when it functions as the subject of a clause, and as whom if it functions as the object of

CHAPTER 9

369

a verb or preposition. In relative clauses (cf. Sections 9.6-9.7), the surface form of the wh-pronoun is influenced not only by CP-internal Arg-Case requirements, but also by the Arg-Case assigned to its antecedent (in headed relatives) or to the whole relative clause (in free relatives). The interaction of the three case constraints correctly predicts that speakers of Present-Day English should show an overwhelming preference for who over whom in matrix questions (168a) and in simple subject relatives (168b), but may nevertheless opt for whom in headed relatives where the wh-pronoun functions as the subject of an embedded clause (168c). (168) a. Who is hungry? / Who do you love? / Who did you give it to? b. We feed children [who are hungry]. c. We feed children [whom we think are hungry]. However, the distribution of wh-forms reported in existing studies is not entirely predictable from the interaction of Arg-Case, Pos-Case, and Def-Case. The occurrence of the nominative who in contexts where all applicable case constraints call for the objective form whom points towards a tendency towards invariant who in all wh-contexts. The distribution of wh-forms in sluicing constructions suggests that the relative position of a wh-pronoun within a construction also has a bearing on its case form. When the wh-pronoun appears as the specifier of a phrase such as PP or CP, and thus asymmetrically c-commands the remaining constituents of the phrase (cf. Section 6.2), it will tend to surface as who (169). When the wh-pronoun appears as the complement of a head, it will tend to surface as whom (170). (169)

XP 2 DP X’ who 2 X …

(170)

XP 2 X DP g D whom

The data presented in this chapter thus provide further evidence that pronoun case forms in Present-Day English not only reflect the structural case of a pronoun, but also code its syntactic position in a construction and serve to mark its morphosyntactic status. In the final chapter of this book, I will look at historical evidence, which suggests that the reanalysis of pronoun case forms as markers of structural position and morphosyntactic strength was triggered by a shift in the licensing of structural arguments during the Middle English period.

CHAPTER 10 SPECULATIONS AND CONCLUSIONS

10.0 Introduction Linguistic variation is often a sign of change in progress. The pronoun case trends identified in this study point to an increasing influence of structural position on pronoun case choice. I would like to propose that the current variation in pronoun case is the result of a shift to exclusive positional licensing of structural arguments during the Middle English period (cf. Kiparsky 1997). This diachronic change in the structural licensing of arguments is the focus of Section 10.1. Section 10.2 argues that the loss of morphological licensing and lexical case marking strengthened the correlation between structural position and pronoun case, and thus contributed to the development of Positional Case checking. Section 10.3 concludes the chapter and the book with a look at the history of the English pronoun system. I will demonstrate that pronouns appeared in both weak/clitic and strong pronoun contexts throughout the history of English, and I will argue that the morphological divergence between the two pronoun series in Modern English was triggered by the loss of morphological licensing and the rise of Positional Case. 10.1 Morphological and positional licensing in the history of English Kiparsky (1997:460-463,482) proposes that the status of a structural argument on the argument hierarchy of the main predicate in the clause, may be signalled in either of the following ways: (a)

by case morphology on the argument itself and/or agreement morphology on the predicate (morphological licensing), e.g. German

(b)

by the surface position of the structural argument in the clause (positional licensing), e.g. Icelandic, Modern English, Dutch, Swedish

(c)

by both the morphological marking on the argument and/or predicate and the surface position of the structural argument in the clause (morphological and positional licensing), e.g. Grissons Swiss German

CHAPTER 10

(d)

371

by either the morphological marking on the argument and/or predicate or the surface position of the structural argument in the clause (morphological or positional licensing), e.g. Old English

Since the status of an argument in the argument structure of the main clausal predicate must always be signalled in some way, the loss of case and agreement morphology in a language will inevitably lead to rigid constraints on the order of direct nominal arguments, even though a rich inflectional system does not necessarily correlate with free word order (Kiparsky 1997:461). In Old English, the status of an argument on the argument hierarchy of a verbal predicate was reflected partly in the overt case morphology found on the argument, and partly in its surface position. In the absence of lexical case marking, the highest argument on the argument hierarchy surfaced in the nominative case form, and the lowest argument on the hierarchy surfaced in the accusative case. Although the word order in Old English was considerably freer than in Middle English and Modern English, the highest argument on the argument hierarchy was generally mapped into the highest argument position projected in the surface syntax.1 This mapping between argument structure and surface position appears to have been obligatory when one of the arguments on the argument hierarchy was lexically case-marked by the verb (cf. Kiparsky 1997:479f; Allen 1995:96-157). As demonstrated in Section 2.1.1, lexical casemarking of an argument can affect the case assigned to the remaining arguments of a predicate, especially when the lexically case-marked argument occupies the highest position on the argument hierarchy (1). Since the highest argument on the argument hierarchy (λx) is prespecified for the structural feature [+ higher] in the lexical entry of the verb, it will surface in its dative form rather than in the nominative case form that usually identifies the highest argument on the argument hierarchy.2 The nominative case will instead be linked to the next argument on the argument hierarchy (λy), which receives the feature [- higher] by default. As a result, the case marking of the arguments in the clause no longer reflects the relative positions of the arguments on the argument hierarchy. This means that the status of the different arguments on the argument hierarchy can only be signalled through

1

See Allen (1995:32-50,417,424) and Pintzuk (1995 & 1996), and Kiparsky (1997:469-472) for more detailed discussions of word order patterns in Old English. 2 See Sections 1.3 & 2.1.1 for evidence that x does indeed occupy a higher argument structure position than y in the semantic representation of verbs such as lician.

372

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

positional licensing, which maps the highest argument on the hierarchy into the highest argument position available in the surface syntax (2).3 (1) The effect of lexical feature-specification on structural feature and case assignment for the experiencer verb lician in Old English4 him gelicade hire þeawas 3sgM.DAT liked her virtues.NOM/ACC “He liked her virtues / Her virtues pleased him” (COE Chron D (Classen-Harm) 1067.1.35) [Allen 1995:142] lician [x LIKE y] λx λy [+ higher] [- higher] [+ lower]

[- lower]

! !

! !

DAT [+ higher] [+ lower]

NOM [ ]

(Semantic Form) (structural arguments at TS) (structural features specified in the lexical entry of lician) (structural features added by default) (structural features encoding the position of the arguments in the argument hierarchy) (argument cases available in Old English, and their structural features)

(2) The mapping from argument structure to surface position in positional licensing lician [x λx

!

LIKE

highest position (e.g. [Spec,TP])

3

y] λy

!

lower position

(Semantic Form) (structural arguments at TS) (argument positions available in the surface syntax)

Research by Allen (1995:96-121) suggests that the (generally pronominal) dative experiencer of verbs such as lician, usually preceded the target of emotion in the surface syntax, and was unable to cooccur with topicalised noun phrases (cf. Section 1.3). It thus appears that the surface order of arguments in sentences involving a dative experiencer in Old English did indeed reflect the relative positions of the arguments on the argument hierarchy. 4 The approach to structural linking adopted here uses the feature system proposed by Wunderlich (1997), but is otherwise compatible with the approach put forward by Kiparsky (1997). For a detailed discussion of case assignment through structural linking see Section 2.1.1.

CHAPTER 10

373

As mentioned in Section 1.1, a series of phonological changes at the end of the Old English period led to the loss of the distinction between nominative and accusative case inflections on nouns and adjectives (cf. Allen 1995:163ff). Coupled with the comparative poverty of verbal agreement morphology5 and the gradual levelling of the determiner towards þe in the singular and þa in the plural (cf. Allen 1995:171f,190f), this neutralisation between nominative and accusative case forms effectively eliminated the possibility of licensing arguments through morphology rather than surface position (cf. Kiparsky 1997: 488). In the absence of an overt distinction between nominative and accusative case, the position of a noun phrase on the argument hierarchy will no longer be evident from its surface morphology, and speakers will tend to avoid marked word orders, where the highest structural argument is not mapped to the highest argument position in the syntax (cf. McFadden 2004:186). The loss of nominal case morphology thus led to the loss of morphological licensing and the exclusive reliance on positional licensing in Modern English, which is reflected in the comparatively fixed order of structural arguments in the surface syntax. 10.2 The rise of Positional Case As discussed in Section 10.1, the case form of an argument noun phrase in Old English depended not only on the position of the argument on the argument hierarchy, but also on the presence versus absence of pre-specified structural features in the lexical entry of the verb (1). The surface position of an argument, on the other hand, was determined solely on the basis of its position on the argument hierarchy (2), if it was influenced by the argument structure at all. Since lexical case assignment could affect the surface form of an argument, but had no direct bearing on its surface position, we find no one-to-one correspondence between case morphology and surface position in Old English: In clauses without lexical case marking, a noun phrase occupying the highest argument position available in the syntax, would tend to surface in the nominative case. In clauses with lexical case marking, on the other hand, a noun phrase occupying the highest argument position could surface either in a lexical case (usually dative) or in the nominative, depending on whether the structural fea5

As Kemenade (1987:203f) and Kiparsky (1997:488) point out, the phonological changes at the end of the Old English period did not have as devastating an effect on verb agreement as on case marking. However, verbs only ever exhibited subject agreement morphology in English, so objects would still have had to be positionally licensed (cf. Kiparsky 1997:488). Since mixed licensing systems appear to be rather unstable, morphological licensing through verb agreement was not really an option after the nominative-accusative case distinction was lost on nouns.

374

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

tures pre-specified in the lexical entry of the verb targeted the highest argument or an argument lower down on the argument hierarchy. The loss of morphological licensing and the gradual demise of lexical case marking during the Middle English period, would both have served to strengthen the correlation between case and surface position for pronouns, which, unlike nouns and adjectives, had retained a nominative-accusative distinction (cf. Section 1.2). When arguments can only be licensed positionally and predicates are no longer able to assign lexical case, both the (canonical) surface position and the case of an argument will be determined solely by its position on the argument hierarchy. An argument occupying a canonical argument position in the surface syntax will therefore always appear in the case form associated with a particular position on the argument hierarchy. The distribution of argument noun phrases in Modern English suggests that topics tend to surface in higher structural positions than arguments contributing new information to the utterance (cf. Section 2.2.1). I will therefore assume that an argument with new information focus may remain in its base position, while arguments interpreted as pure topics will generally raise out of their base position. For the highest argument of an agentive/causative verb, the base position is [Spec,vP], and for the highest argument of an unaccusative verb or identificational be, the base position is [Spec,VP] (cf. Section 2.2.1). When the highest argument of a verb is a pure topic, it will generally raise to [Spec,TP]. Lower arguments of a verb will tend to remain within the VP if they receive a rhematic interpretation, and raise to [Spec,vP] if they are interpreted as pure topics. By the end of the Middle English period, the only nominal elements to exhibit distinct nominative and objective case forms were pronouns. Because of their inherent topic properties, pronouns tend to be licensed in [Spec,TP] if they occupy the highest position on the argument hierarchy, and in [Spec,vP] if they appear lower down on the argument hierarchy. As we will see in Section 10.3.1, pronominal subjects already appear to have been confined to [Spec,TP] in Old English. Pronominal objects of verbs and prepositions could originally raise to a position preceding the finite verb in Old English, but started to lose this ability at the end of the Old English period (cf. Kemenade 1987:189-196; Allen 1995:34). Since preverbal object pronouns appear to have disappeared from the English language during the 14th century (cf. Kemenade 1987:195), it is likely that weak object pronouns obligatorily occurred in [Spec,vP] by the 15th century. Once the last vestiges of lexical case assignment had disappeared from the English language during the 15th century, the case form of a pronoun would

CHAPTER 10

375

have been determined solely by its position on the argument hierarchy. As a result, nominative case marking would have started to correlate with the appearance of a pronoun in [Spec,TP], while objective case marking would have become associated with [Spec,vP]. It therefore seems plausible that the case marking of the pronoun was reanalysed as a reflex of its surface position. The correlation between nominative case marking and the vestiges of subject agreement morphology found on finite verbs is likely to have contributed to the reanalysis of agreement-related functional heads as checkers of Positional Case. In the following section, I propose that the rise of Positional Case checking paved the way for the morphological divergence of the weak and strong pronoun series in Modern English, and also contributed to the emergence of Relative Positional Coding. 10.3 The divergence of the weak and strong pronoun series in English As I will demonstrate in Section 10.3.1, pronouns already appeared in both weak/clitic and strong pronoun contexts during the Old English period. However, case variation in strong pronoun contexts appears to be found only from Late Middle English onwards (cf. Chapters 5, 8 and 9). In Section 10.3.2, I will argue that the case differences between weak and strong pronouns arose only after the rise of positional licensing and Positional Case, because before then, the case of all types of pronouns and noun phrases was determined solely on the basis of their argument structure status, and did not depend on their surface position. 10.3.1 The distribution of personal pronouns in Old English Evidence presented in Kemenade (1987:109-112) and Pintzuk (1995: 242f; 1996:389) suggests that the status of lone subject pronouns in Old English was similar to the status of weak subject pronouns in Present-Day English. Pronominal subjects generally preceded the finite verb (3), even though full noun phrase subjects could appear either before (4) or after the verb (5) in main clauses involving a topicalised object. (3) ælc yfel he mæg don each evil 3sgM.NOM can do “He can do each evil” (Wulfstan 4.62) [Pintzuk 1995:242] (4) Feawa manna crist sylf gefullode few men Christ self baptized “Christ baptized few men himself” (Ælfric Catholic Homilies III, 214) [Allen 1995:44]

376

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

ælmihtiga on ðam (5) Tyn beboda awrat se ten commandments wrote the.SG.NOM Almighty on the.DAT twam tabelum two tablets “The Almighty wrote ten commandments on the two tablets” (Ælfric Catholic Homilies XII, 255) [Allen 1995:42]

As can be seen from the examples in (6)-(9), subject pronouns consistently followed the finite verb in direct questions (6), V-initial declaratives and imperatives (7), narrative-advancing clauses with an adverb in initial position (8), and some clauses with a negated V (9). The initial adverb in narrativeadvancing clauses with this word order is most commonly þa/þonne ‘then’, but other adverbs also appear (Pintzuk 1995:255n.16). (6) hwi sceole we oþres mannes niman why should 1pl.NOM other man’s take “Why should we take those of another man?” (Ælfric Lives of Saints 24.188) [Pintzuk 1995:242]

(7) a. hæfdon hi hiora onfange had 3pl.NOM/ACC 3pl.GEN sponsored “they had sponsored them” (Chronicle A 86.28-29 (894)) [Pintzuk 1995:242] b. beo ðu on ofeste be 2sg.NOM in haste “Be quick.’ (Beowulf (3rd ed.) 386) [Pintzuk 1995:243] (8) þa ge-mette he sceaðan then met 3sgM.NOM robbers “then he met robbers” (Ælfric Lives of Saints 31.151) [Pintzuk 1995:243; 1996:381]

(9) ne furðon an ban næfde he mid oðrum not even one bone not-had 3sgM.NOM with others “He didn’t have even one bone joined to the others” (Ælfric Lives of Saints 23.496) [Pintzuk 1995: 243]

Since none of the constructions in (6)-(9) cooccur with an overt complementizer, the most plausible analysis of the surface order in (6)-(9) is that the finite verb has moved to C past the subject pronoun (Pintzuk 1995: 243f). The data in (5)-(9) would thus seem to indicate that lone subject pronouns already had to

CHAPTER 10

377

be licensed in a position between C and T in Old English (10), just like weak subject pronouns in Modern English (for further discussion see Pintzuk 1995: 243; 1996:386-389,395f). As Pintzuk (1995:255n.18) notes, “for clauses with the verb in COMP it is impossible to determine the underlying position of INFL”. I have therefore omitted the trace of T and the internal structure of T’ in (10). (10)

CP 3 The surface position of lone subject pronouns in AdvP C’ Old English (cf. Pintzuk 1995:243; 1996:381) þa 3 C TP 4 3 T’ T C DPi 4 he 6 Vk T ti sceaðan tk ge-mette

If we assume that only pronominal subjects had to undergo movement to [Spec,TP] in Old English, the distributional differences between pronominal and full noun phrase subjects in main clauses containing a topicalised object (3)-(5), could be accounted for by assuming that the noun phrase subject in sentences such as (11) occupies its base position within vP. (11) eow sceolon deor abitan 2pl.OBJ shall beasts devour “beasts shall devour you” (Ælfric Lives of Saints 24.35) [Pintzuk 1995:242] Lone pronouns functioning as the object of a verb or preposition appear to have had a freer distribution in Old English than in Modern English. Pronominal objects of verbs could surface in a variety of positions within the extended projection of the verb (cf. Pintzuk 1996:390f; Allen 1995:33f). Like full noun phrase objects, object pronouns generally preceded any nonfinite verbs and unambiguously sentence-final finite verbs (12). (12) gif he hit him ðonne sellan mæge if 3sgM.NOM 3sgN.NOM/ACC 3pl.DAT then give can “if he can give it to them then” (Pastoral Care 44.323.24) [Koopman 1992:57]

When the finite verb appeared in medial position, a pronominal object could either follow the finite verb (13), as was common for noun phrase objects, or it

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

378

could appear in preverbal position (14) (cf. Allen 1995:32-50 and Pintzuk 1995 for further discussion). (13) a. þæt anig oðer man sceolde hire had on sættan that any other man should 3sgF.OBL veil give “…that any other man should give her the veil…” (Ælfric Lives of Saints 31.583-584) [Pintzuk 1996:391]

b. ac ða burhware noldon þæs færes him getyðian but the burghers not-would the passage 3pl.DAT permit “But the burghers would not permit them the passage…” (Ælfric Lives of Saints 25.444-445) [Pintzuk 1996:391]

(14) a. þæt þa Deniscan him ne mehton þæs ripes so-that the Danes 3pl.DAT not could the harvest forwiernan refuse “…so that the Danes could not refuse them the harvest.” (Chronicle A 89.10 (896)) [Pintzuk 1996:390]

b. and hi man mæg wenian wundorlice to gefeohte and 3pl.NOM/ACC one may tame wonderfully for battle “And one may tame them wonderfully for battle.” (Ælfric Lives of Saints 31.71) [Pintzuk 1996:390]

The occurrence of object pronouns before finite verbs in medial position (14), has often been taken as evidence for the clitic-hood of (unstressed) object pronouns in Old English (cf. Kemenade 1987:108-141; Pintzuk 1996). However, although pronominal objects were much more likely to occur in preverbal position than full noun phrase objects (cf. Allen 1995:33), full noun phrase objects appear to have been able to occur before the finite verb (15). This suggests that the raised object pronouns may have been weak rather than clitic. (15) a. Ic

þæm godan

sceal for his modþræce madmas

1sg.NOM the.DAT good.DAT shall for his daring

treasures.ACC

beodan offer

“I shall offer the good man treasures on account of his daring” (Beowulf 384) [McLaughlin 1983:11]

CHAPTER 10

b.

379

And þyssere mægðe God sealde and gesette æ and this.DAT people.DAT God gave and set law “And God established and gave a law to this people” (Ælfric Anglo-Saxon p.24.16) [Allen 1995:44]

Pronominal objects of prepositions most commonly occurred after the preposition, just like full noun phrases (16), but they could also raise to a position preceding the preposition (17), an option that does not seem to have been available to full noun phrases (cf. Kemenade 1987:114ff; Koopman 1992:54; Pintzuk 1996:391f). (16) a. he hæfde ge-axod be ðæs hælendes wundrum learned about the saviour’s miracles 3sgM.NOM had “He had learned about the saviour’s miracles…” (Ælfric Lives of Saints 24.86) [Pintzuk 1996:391]

b. ic hæbbe gehyred be ðe 1sg.NOM have heard about 2sg.OBJ “I have heard about you…” (Ælfric Lives of Saints 24.90) [Pintzuk 1996:391]

(17) a. ða hyrdas ða spræcon him betweonan the shepherds then spoke 3pl.DAT between “the shepherds then spoke among themselves” (Ælfric Catholic Homilies i.2.40.1) [Koopman 1992:54]

b. & him man nam syððan frið wið and 3sgM.DAT one made afterwards peace with “And afterwards they made peace with him…” (Chronicle A 126.7 (993)) [Pintzuk 1996:392]

One way to analyse the distributional differences between nominal and pronominal objects of prepositions would be to argue that pronominal objects of prepositions may either be deficient (i.e. weak or clitic) or strong. If they are strong, they will follow the preposition, but if they are deficient, they must raise to a higher position to be licensed. The data in (16)-(17) could thus be seen as evidence that pronouns could occur both in weak/clitic positions and in strong pronoun contexts. Further evidence that not all pronouns were deficient in Old English comes from the ready occurrence of pronouns in coordinates (18). As discussed in Chapter 5, deficient pronouns are usually unable appear in coordination, which suggests that us, eow, and him are all strong in (18). In

380

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

(19), the 1pl pronoun is not only coordinated but also modified by a relative clause, which indicates even more clearly that us is strong rather than weak. (18) a. ði læs ðe hit ne genihtsumige [us and eow] lest that it not suffice 1pl.OBJ and 2pl.OBJ “lest it is not sufficient for us and you” (Ælfric Catholic Homilies ii.44.327.16) [Koopman 1992:61]

b. and þæt he [him and his geferan] bigleofan and that 3sgM.NOM 3sgM.DAT and his companions food ðenian wolde serve would “and that he would serve him and his companions food” (Ælfric Catholic Homilies ii.9.78.198) [Koopman 1992:61]

(19) eallum mannum to steore, [eallum folce þa ðe geo all.DAT persons.DAT to direction all.DAT people then that once wæs, ge [us þe nu sindan]] was and 1pl.OBJ that now are “all persons to direction, all people then that once were and us that now are” (Ælfric Wulstan 188.120) [Brooklyn Corpus 2000]6 As Pintzuk (1996:387f) points out, raised pronominal objects of verbs and prepositions tend to be unstressed in Beowulf, whereas pronouns appearing after prepositions are usually stressed. The clitic-like behaviour of unstressed lone pronouns, paired with the ready occurrence of pronouns in coordination and with modifiers, would suggest that we are already dealing with two pronoun series in Old English, which are identical in form but differ in their distribution and prosodic properties. Such an analysis of the Old English pronoun system was already proposed by Foulet (1936:64), who relates later changes in the distribution of pronoun case forms to the already existing dichotomy between weak/clitic and strong pronouns. 10.3.2 The emergence of Relative Positional Coding and the trend towards invariant strong pronoun forms As discussed in Chapter 3, strong pronouns tend to be in some way separated from the agreement-related functional heads associated with Positional Agreement and Positional Case. This means that the case form of a 6

Many thanks to Susan Pintzuk for granting me access to the Brooklyn-Geneva-AmsterdamHelsinki Parsed Corpus of Old English.

CHAPTER 10

381

strong pronoun could generally not have been interpreted as encoding a direct spec-head relationship between the pronoun and an agreement-related functional head. At the same time, the loss of morphological licensing would have dramatically reduced the motivation for signalling the argument structure status of a pronoun through case morphology. Alternations between nominative and objective forms in strong pronoun contexts, which would have correlated with the argument structure status of the pronoun rather than its surface position, would thus have lost much of their original case significance by the start of the Early Modern English period. Aronoff (1976), Lapointe (1985:63f), Pinker (1984:177-208,360f), and Williams (1997:577-582) have all drawn attention to the importance of blocking in the acquisition and evolution of morphological paradigms. The Blocking Principle and Pinker’s (1984) Uniqueness/Unique Entry Principle stipulate that a difference in form must correlate with a difference in meaning. That is, different forms must encode different semantic/grammatical features. If the original meaning difference between two distinct forms in a pronominal paradigm is lost, the Blocking Principle will predict either of the following developments: (a)

one of the forms is eliminated from the paradigm

(b)

the different forms are reanalysed as encoding a new meaning difference

In view of the importance of positional licensing in Modern English, and the close correlation between the surface form and surface position of weak pronouns, we might expect that any new coding function acquired by strong pronoun forms would be related to positional relationships in the surface syntax.7 As I have tried to demonstrate in this book, the distribution of strong pronoun forms in Present-Day English provides evidence for developments in both of the directions predicted by the Blocking Principle: (a)

Speakers tend to favour the use of invariant pronoun forms in all strong pronoun context (me, him, her, us, them, and who, whoever).

(b)

Strong pronoun forms encode not only case, but also asymmetric c-command relationships, with me, he, she, we, they, who favoured in initial/ asymmetrically c-commanding positions, and I, him, her, us, them, whom favoured in final/c-commanded positions.

7

Cf. Sigursson’s (2005) observations that ‘case poor’ Germanic languages (i.e. languages where case morphology is confined to pronouns) tend to employ morphological case distinctions for a range of different purposes which suggest that case is best viewed as “an interpretation of various syntactic structures in morphological terms”.

382

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

In the constraint-based approach proposed here, the trend towards invariant forms (a) is captured in a set of Invariant Strong Form constraints (cf. Chapter 6). The additional coding function of strong pronoun forms (b) is captured in the two Relative Positional Coding constraints proposed in Chapter 6. In the absence of a detailed diachronic study, it is difficult to know exactly which factors led to the grouping of pronoun forms associated with Relative Positional Coding, but it seems likely that the asymmetric c-command relationship between nominative and objective Pos-Case positions (20) contributed to the association of the nominative strong pronoun forms he, she, we, they, who with c-commanding positions, and the objective strong pronoun forms him, her, us, them, whom with c-commanded positions. (20)

CP 3 The asymmetric c-command relationship between the basic nominative and objective C TP [+ finite] 3 Pos-Case positions in a finite clause DPi [nom] T’ 3 FP T [nom] 3 F vP 3 v’ DPj [obj] 3 v’ ti 3 v [obj] … tj …

As discussed in Chapter 6, the grouping of me with the non-1sg nominatives, and I with the non-1sg objective forms, would appear be motivated by phonological factors. Thus me bears a close phonological resemblance to he, she, we, who, (and to a lesser extent they), while I (and also they) shares certain properties with him, her, us, them, whom. The classification of pronoun forms for the purposes of Relative Positional Coding may have been further reinforced by the emphasis on certain pronoun case configurations in prescriptive texts. For example, prescriptivists often criticise the use of me in initial conjuncts of subject coordinates, and instead advocate the use of I in final conjunct position (21). Similarly, prescriptive corrections to sluiced questions involving wh-pronouns associated with a preposition, may have reinforced the classification of who as a gracile form, and whom as robust. As can be seen from the passage in (22), sluiced questions involving who plus preposition stranding are frequently corrected to versions where the pronoun appears as whom after a pied-piped preposition.

CHAPTER 10

383

(21) When a country cousin remarks that ‘Me and Tom’ have done so and so, it is necessary for the town cousin, in the interests of polite speech and general gentility, to reply that the same thing was once done by ‘Tom and I’. (The Press, 8 October 1887) (22) ‘… that gaby Mary Ramsbottom has got herself engaged.’ … ‘Who to?’ demanded Tommy. ‘You mean “to whom”. The preposition governs the objective case,’ corrected her James Douglas McTear … who himself wrote English better than he spoke it. (Jerome K. Jerome [1904], Tommy & Co:71) [Jespersen & Haislund 1949:242]

The frequent juxtaposition of me and X (23) and X and I (24) is likely to have reinforced the analysis of me as a gracile form used in asymmetrically c-commanding positions, and I as a robust form used in c-commanded positions (cf. Section 5.1). Juxtaposing PPs where who appears in specifier position (25) with PPs where whom appears in complement position (26) likewise serves to reinforce the association between pronoun form and asymmetrical ccommand relationships (cf. Section 9.5 for further discussion). (23)

ConjP (24) ConjP (25) PP (26) PP 2 2 2 2 P’ P DP DP Conj’ DP Conj’ DPi me 2 X 2 who 2 to g Conj DP Conj DP P DP D whom and X and I to ti

Despite such prescriptive reinforcements of Relative Positional Coding, it is the trend towards invariant strong forms that would appear to have the greatest bearing on the distribution of strong pronoun forms in Present-Day English. As discussed in Chapters 5 and 6, many speakers exhibit a clear preference for the objective non-1sg forms him, her, us, them in all of the strong pronoun contexts tested in the empirical survey. Only the distribution of 1sg forms is still quite strongly influenced by Relative Positional Coding. If the distribution of English pronoun forms continues to develop along the lines suggested by the data presented in this book, it thus seems likely that the English pronoun system will eventually come to resemble the pronoun systems found in Romance languages, where case distinctions in the weak/clitic paradigm contrast with an invariant strong pronoun series.

REFERENCES Aarts, Bas. 1992. Small Clauses in English: The Nonverbal Types. [Topics in English Linguistics 8] Berlin: Mouton de Gruyter. Abney, Steven Paul. 1987. The English Noun Phrase in its Sentential Aspect. PhD thesis, MIT. Akmajian, Adrian. 1970. “On Deriving Cleft Sentences from Pseudo-Cleft Sentences”. Linguistic Inquiry 1. 149-168. Akmajian, Adrian. 1984. “Sentence Types and the Form-Function Fit”. Natural Language and Linguistic Theory 2. 1-23. Alexiadou, Artemis, and Chris Wilder (eds). 1998. Possessors, Predicates and Movement in the Determiner Phrase. [Linguistik Aktuell/Linguistics Today, 22] Amsterdam: John Benjamins. Alexiadou, Artemis, Paul Law, André Meinunger, and Chris Wilder (eds). 2000. The Syntax of Relative Clauses. [Linguistik Aktuell/Linguistics Today 32]. Amsterdam: John Benjamins. Allan, Keith. 1987. “Hierarchies and the Choice of Left Conjuncts (With Particular Attention to English)”. Journal of Linguistics 23. 51-77. Allen, Cynthia L. 1980. Topics in Diachronic English Syntax. [Outstanding Dissertations in Linguistics] New York: Garland. Allen, Cynthia L. 1995. Case Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Clarendon Press. Ambar, Manuela. 1999. “Aspects of the Syntax of Focus in Portuguese”. The Grammar of Focus, Georges Rebuschi and Laurice Tuller (eds), 23-53. [Linguistik Aktuell/ Linguistics Today 24] Amsterdam: John Benjamins. Anagnostopoulou, Elena. 1997. “Clitic Left Dislocation and Contrastive Left Dislocation”. Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk, and Frans Zwarts (eds), 151-192. [Linguistik Aktuell/Linguistics Today, 14] Amsterdam: John Benjamins. Angermeyer, Philipp S. and John Victor Singler. 2003. “The Case for Politeness: Pronoun Variation in Co-ordinate NPs in Object Position in English”. Language Variation and Change 15. 171-209. Anttila, Arto. 1995. “Deriving Variation from Grammar: A Study of Finnish Genitives”. [Rutgers Optimality Archive # 63-0000] Arnold, Mark D. 1996. “A Unified Analysis of P-Stranding, ECM, and that-Deletion, and the Subsequent Loss of Verb Movement in English”. Proceedings of the North East Linguistic Society (NELS) 26, Kiyomi Kusumoto (ed.), 1-15. Amherst, MA: GLSA, University of Massachusetts. Aronoff, Mark. 1976. Word Formation in Generative Grammar. [Linguistic Inquiry Monographs 1] Cambridge, MA: MIT Press. Baker, C.L. 1995. English syntax (2nd ed.). Cambridge, MA: MIT Press. Ball, Catherine N. 1994. “Relative Pronouns in it-Clefts: The Last Seven Centuries”. Language Variation and Change 6. 179-200.

REFERENCES

385

Banerji, Sharbani. 2003. “Focus, Word Order and Scope Effects in Bangla”. Folia Linguistica XXXVII. 35-75. Bartlett, Christopher. 1992. “Regional Variation in New Zealand English: The Case of Southland”. New Zealand English Newsletter 6. 5-15. Barton, Ellen L. 1990. Nonsentential Constituents: A Theory of Grammatical Structure and Pragmatic Interpretation. [Pragmatics & Beyond, New Series 2] Amsterdam: John Benjamins. Bayer, Josef. 1999. “Bound Focus or How Can Association with Focus Be Achieved without Going Semantically Astray?”. The Grammar of Focus, Georges Rebuschi and Laurice Tuller (eds), 55-82. [Linguistik Aktuell/Linguistics Today 24] Amsterdam: John Benjamins. Behaghel, Otto. 1909/10. “Beziehungen zwischen Umfang und Reihenfolge von Satzgliedern”. Indogermanische Forschungen 25. 110-142. Benveniste, Émile. 1966. “Structure des relations de personne dans le verbe”. Problèmes de linguistique générale, 225-236. Paris: Gallimard. Benveniste, Émile. 1974. “L’antonyme et le pronom en français moderne”. Problèmes de linguistique générale II, 197-214. Paris: Gallimard. Bianchi, Valentia. 2000. “Some Issues in the Syntax of Relative Determiners”. In Alexiadou, Law, Meinunger & Wilder 2000. 53-81. Bittner, Maria and Ken Hale. 1996. “Ergativity: Towards a Theory of a Heterogeneous Class”. Linguistic Inquiry 27. 531-604. Bobaljik, Jonathan. 1994. “What Does Adjacency Do?”. The Morphology-Syntax Connection, Heidi Harley and Colin Phillips (eds), 1-32. [MIT Working Papers in Linguistics 22] Cambridge, MA: MITWPL. Bock, J. Kathryn. 1982. “Toward a Cognitive Psychology of Syntax: Information Processing Contributions to Sentence Formulation”. Psychological Review 89. 1-47. Boersma, Paul and Bruce Hayes. 2001. “Empirical Tests of the Gradual Learning Algorithm”. Linguistic Inquiry 32. 45-86. Bolinger, Dwight. 1971. The Phrasal Verb in English. Cambridge, MA: Harvard University Press. Boyland, Joyce Tang. 2001. “Hypercorrect Pronoun Case in English? Cognitive Processes that Account for Pronoun Usage”. Frequency and the Emergence of Linguistic Structure, Joan Bybee and Paul Hopper (eds), 383-404. [Typological studies in language 45] Amsterdam: John Benjamins. Bowers, John. 1993. “The Syntax of Predication”. Linguistic Inquiry 24. 519-656. Bowers, John. 2002. “Transitivity”. Linguistic Inquiry 33. 183-224. Borsley, Robert D. 1997. “Relative Clauses and the Theory of Phrase Structure”. Linguistic Inquiry 28. 629-647. Bošković, Željko. 2000. “Sometimes in [Spec,CP], Sometimes in Situ”. Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, Roger Martin David Michaels, and Juan Uriagereka (eds), 53-87. Cambridge, MA: MIT Press. Bresnan, Joan. 1994. “Locative Inversion and the Architecture of Universal Grammar”. Language 70. 72-131. Bresnan, Joan and Jane Grimshaw. 1978. “The Syntax of Free Relatives in English”. Linguistic Inquiry 9. 331-391. Brousseau, Anne-Marie and Elizabeth Ritter. 1991. “A Non-Unified Analysis of Agentive Verbs”. Proceedings of the 10th West Coast Conference on Formal Linguistics (WCCFL 10), 53-64. Stanford, CA: CSLI Publications.

386

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Browning, M.A. 1996. “CP Recursion and that-t Effects”. Linguistic Inquiry 27. 237-255. Büring, Daniel. 2001. “Let’s Phrase It! Focus, Word Order, and Prosodic Phrasing in German Double Object Constructions”. Competition in Syntax, Gereon Müller and Wolfgang Sternefeld (eds), 69-105. [Studies in Generative Grammar 49] Berlin: Mouton de Gruyter. Burzio, Luigi. 1986. Italian Syntax. Dordrecht: Reidel. Burzio, Luigi. 2000. “Anatomy of a Generalization”. Arguments and Case: Explaining Burzio’s Generalization, Eric Reuland (ed.), 195-240. [Linguistik Aktuell/Linguistics Today 34] Amsterdam: John Benjamins. Cardinaletti, Anna. 1991. “On Pronoun Movement: The Italian Dative loro”. Probus 3. 127153. Cardinaletti, Anna. 1994. “On the Internal Structure of Pronominal DPs”. The Linguistic Review 11. 195-219. Cardinaletti, Anna. 1999. “Pronouns in Germanic and Romance Languages: An Overview”. Clitics in the Languages of Europe, Henk van Riemsdijk (ed.), 33-82. [Empirical Approaches to Language Typology 20, Eurotyp 5] Berlin: Mouton de Gruyter. Cardinaletti, Anna and Maria Teresa Guasti (eds). 1995. Syntax and Semantics 28: Small Clauses. San Diego: Academic Press. Cardinaletti, Anna and Michal Starke. 1995. “The Tripartition of Pronouns and its Acquisition: Principle B Puzzles are Ambiguity Problems”. NELS 25, Volume 2: Papers from the Workshop on Language Acquisition and Language Change, Jill N. Beckman (ed.), 1-12. Amherst, MA: GLSA, University of Massachusetts. Cardinaletti, Anna and Michal Starke. 1996. “Deficient Pronouns: A View from Germanic”. Studies in comparative Germanic syntax II, Höskuldur Thráinsson, Samuel D. Epstein, and S. Peter (eds.), 21-65. Dordrecht: Kluwer. Cardinaletti, Anna and Michal Starke. 1999. “The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns”. Clitics in the Languages of Europe, Henk van Riemsdijk (ed.), 145-233. [Empirical Approaches to Language Typology 20, Eurotyp 5] Berlin: Mouton de Gruyter. Chao, Wynn. 1987. On Ellipsis. PhD thesis, University of Massachusetts, Amherst. Chomsky, Noam. 1975. The Logical Structure of Linguistic Theory. Chicago: University of Chicago Press. Chomsky, Noam. 1993. “A Minimalist Program for Linguistic Theory”. The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger. Ken Hale and Samuel Jay Keyser (eds), 1-52. [Current Studies in Linguistics 24] Cambridge, MA: MIT Press. Chomsky, Noam. 1995. “Chapter 4: Categories and Transformations”. The Minimalist Program, 219-394. [Current Studies in Linguistics 28] Cambridge, MA: MIT Press. Chomsky, Noam. 2000. “Minimalist Inquiries”. Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, Roger Martin David Michaels, and Juan Uriagereka (eds), 89155. Cambridge, MA: MIT Press. Chomsky, Noam. 2001. “Beyond Explanatory Adequacy”. Ms, MIT. Chung, Sandra, William Ladusaw, and James McCloskey. 1995. “Sluicing and Logical Form”. Natural Language Semantics 3. 239-282. Collins, Chris. 1997. Local Economy. [Linguistic Inquiry Monographs 29] Cambridge, MA: MIT Press.

REFERENCES

387

Cormack, Annabel. 1999. “Without Specifiers”. Specifiers: Minimalist Approaches, David Adger, Susan Pintzuk, Bernadette Plunkett, and George Tsoulas (eds), 46-68. Oxford: Oxford University Press. Dalrymple, Mary and Ronald M. Kaplan. 2000. “Feature Indeterminacy and Feature Resolution. Language 76. 759-798. Deevy, Patricia L. 1998. “The Comprehension of English Subject-Verb Agreement”. Handout distributed at dissertation defense, University of Massachusetts, Amherst. Dehé, Nicole. 2002. Particle Verbs in English: Syntax, Information Structure and Intonation. [Linguistik Aktuell/Linguistics Today 59] Amsterdam: John Benjamins. Delahunty, Gerald P. 1982. Topics in the Syntax and Semantics of English Cleft Sentences. PhD thesis, University of California, Irvine. Demirdache, Hamida. 1991. Resumptive Chains in Restrictive Relatives, Appositives and Dislocation Structures. PhD thesis, MIT. Demirdache, Hamida. 1997. “Dislocation, Resumption and Weakest Crossover”. Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk, and Frans Zwarts (eds), 193-231. [Linguistik Aktuell/Linguistics Today, 14] Amsterdam: John Benjamins. Denison, David. 1993. English Historical Syntax: Verbal Constructions. [Longmans Linguistics Library] London: Longman. Denison, David. 1996. “The Case of the Unmarked Pronoun”. English Historical Linguistics 1994: Papers from the 8th International Conference on English Historical Linguistics (8. ICEHL, Edinburgh, 19-23 September 1994), Derek Britton (ed.), 287-299. [Current Issues in Linguistic Theory 135] Amsterdam: John Benjamins. Dogil, Grzegorz and Hans Christian Luschützky. 1990. “Notes on Sonority and Segmental Strength”. Rivista di Linguistica 2. 3-54. Downing, Laura J. 1998. “On the Prosodic Misalignment of Onsetless Syllables”. Natural Language and Linguistic Theory 16. 1-52. Ebert, Robert Peter. 1978. Historische Syntax des Deutschen. Stuttgart: Sammlung Metzler. Emonds, Joseph E. 1985. A Unified Theory of Syntactic Categories. [Studies in Generative Grammar 19] Dordrecht: Foris. Emonds, Joseph E. 1986. “Grammatically Deviant Prestige Constructions”. A Festschrift for Sol Saporta, Michael Brame, H. Contreras, and Frederick Newmeyer (eds), 92-129. Seattle: Noit Amrofer. Erdmann, Peter. 1978. “It’s I, It’s Me: A Case for Syntax”. Studia Anglica Posnaniensia 10. 67-80. Everett, Daniel L. 1996. Why There Are No Clitics: An Alternative Perspective on Pronominal Allomorphy. [SIL and UTA Publications in Linguistics 123] Dallas, TX: Summer Institute of Linguistics and Arlington, TX: University of Texas. Everett, Daniel L. and Keren Everett. 1984. “On the Relevance of Syllable Onsets to Stress Placement”. Linguistic Inquiry 15. 705-711. Foulet, L. 1935a. “L’extension de la forme oblique du pronom personnel en ancien français: Premier article”. Romania LXI. 257-315. Foulet, L. 1935b. “L’extension de la forme oblique du pronom personnel en ancien français: Deuxième article”. Romania LXI. 401-463. Foulet, L. 1936. “L’extension de la forme oblique du pronom personnel en ancien français: Troisième et dernier article”. Romania LXII. 27-91. Frazier, Lyn and Charles Clifton, Jr. 1998. “Comprehension of Sluiced Sentences”. Language and Cognitive Processes 13. 499-520.

388

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Galves, Charlotte. 1997. “La syntaxe pronominale du portugais brésilien et la typologie des pronoms”. Les pronoms: Morphologie, syntaxe et typologie, Anne Zribi-Hertz (ed.), 1134. [Sciences du langage] Paris: Presses Universitaires de Vincennes. Gelderen, Elly van. 1997. Verbal Agreement and the Grammar behind its ‘Breakdown’: Minimalist Feature Checking. [Linguistische Arbeiten 364] Tübingen: Max Niemeyer. Gelderen, Elly van. 2000. A History of English Reflexive Pronouns: Person, Self, and Interpretability. [Linguistik Aktuell/Linguistics Today 39] Amsterdam: John Benjamins. Gelderen, Elly van. 2004. Grammaticalization as Economy. [Linguistik Aktuell/Linguistics Today 71] Amsterdam: John Benjamins. Ghomeshi, Jila and Elizabeth Ritter. 1996. “Binding, Possessives, and the Structure of DP”. Proceedings of the North East Linguistic Society (NELS) 26, Kiyomi Kusumoto (ed.), 87-100. Amherst, MA: GLSA, University of Massachusetts. Gil, David. 1982. “Case Marking, Phonological Size, and Linear Order”. Syntax and Semantics 15: Studies in Transitivity, Paul Hopper and Sandra A. Thompson (eds), 117-141. New York: Academic Press. Giusti, Giuliana. 2001. “The Birth of a Functional Category: From Latin ILLE to the Romance Article and Personal Pronoun”. Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi, Guglielmo Cinque and Giampaolo Salvi (eds), 157-171. [North-Holland Linguistic Series, Linguistic Variations 59] Amsterdam: Elsevier. Goldwater, Sharon and Mark Johnson. 2003. “Learning OT Constraint Rankings Using a Maximum Entropy Model”. Ms, Brown University. Grohmann, Kleanthes K. 2003. Prolific Domains: On the Anti-Locality of Movement Dependencies. [Linguistik Aktuell/Linguistics Today 66] Amsterdam: John Benjamins. Grosu, Alexander. 1996. “The Proper Analysis of ‘Missing-P’ Free Relative Constructions”. Linguistic Inquiry 27. 257-293. Gundel, Jeanette K. 1977. “Where do Cleft Sentences Come from?” Language 53. 543-559. Guy, Gregory R. 1997. “Violable is Variable: Optimality Theory and Linguistic Variation”. Language Variation and Change 9. 333-347. Hale, Ken, and Samuel Jay Keyser. 1993. “On Argument Structure and the Lexical Expression of Syntactic Relations”. The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, Ken Hale and Samuel Jay Keyser (eds), 53-109. [Current Studies in Linguistics 24] Cambridge, MA: MIT Press. Hale, Ken, and Samuel Jay Keyser. 1998. “The Basic Elements of Argument Structure”. Papers from the UPenn/MIT Roundtable on Argument Structure and Aspect, Heidi Harley (ed.), 73-118. [MIT Working Papers in Linguistics 32] Cambridge, MA: MITWPL. Harley, Heidi and Elizabeth Ritter. 2002. “Structuring the Bundle: A Universal Morphosyntactic Feature Geometry”. Pronouns – Grammar and Representation, Horst J. Simon and Heike Wiese (eds), 23-39. [Linguistik Aktuell/Linguistics Today, 52] Amsterdam: John Benjamins. Harris, Alice C. and Lyle Campbell. 1995. Historical Syntax in Cross-Linguistic Perspective. [Cambridge Studies in Linguistics 74] Cambridge: Cambridge University Press. Harris, Martin. 1981. “It’s I, It’s Me: Further Reflections”. Studia Anglica Posnaniensia 13. 17-20. Hawkins, John A. 1994. A Performance Theory of Order and Constituency. [Cambridge Studies in Linguistics 73] Cambridge: Cambridge University Press.

REFERENCES

389

Hawkins, John A. 2004. Efficiency and Complexity in Grammars. Oxford: Oxford University Press. Hay, Jennifer. 2001. “Intrusive /r/: Clawring and Ploughring through New Zealand Vowels”. Paper presented at the 14th New Zealand Linguistic Society Conference (LINGSOC XIV), University of Canterbury, Christchurch, New Zealand. Hay, Jennifer and Daniel Schreier. 2004. “Reversing the Trajectory of Language Change: Subject Verb Agreement with BE in New Zealand English”. Language Variation and Change 16. 209-235. Hedberg, Nancy. 2000. “The Referential Status of Clefts”. Language 76. 891-920. Heggie, Lori. 1988. “A Unified Approach to Copular Sentences”. Proceedings of the 7th West Coast Conference on Formal Linguistics (WCCLF 7), Hagit Borer (ed.), 129-142. Stanford, CA: Stanford Linguistics Association & CSLI. Heim, Irene R. 1982. The Semantics of Definite and Indefinite Noun Phrases. PhD thesis, University of Massachusetts, Amherst. Henry, Alison. 1995. Belfast English and Standard English: Dialect Variation and Parameter Setting. Oxford: Oxford University Press. Higginbotham, James. 1985. “On Semantics”. Linguistic Inquiry 16. 547-593. Householder, Fred W. 1986. “Some Facts about Who and Whom”. Language Research 22. 149-167. Householder, Fred W. 1987. “Some Facts about Me and I”. Language Research 23. 163-184. Hudson, Richard. 1987. “Zwicky on Heads”. Journal of Linguistics 23. 109-132. Hudson, Richard. 1995. “Does English Really Have Case?”. Journal of Linguistics 31. 375192. Ihalainen, Ossi. 1991. “On Grammatical Diffusion in Somerset Folk Speech”. Dialects of English: Studies in Grammatical Variation, Peter Trudgill and J.K. Chambers (eds), 104-119. London: Longman. Ihalainen, Ossi. 1994. “The Dialects of England since 1776”. The Cambridge History of the English Language, Volume V: English in Britain and Overseas, Robert Burchfield (ed.), 197-274. Cambridge: Cambridge University Press. Jackendoff, Ray. 1977. X-Bar Syntax: A Study of Phrase Structure. [Linguistic Inquiry Monographs 2] Cambridge, MA: MIT Press. Jäger, Gerhard. n.d. “Maximum Entropy Models and Stochastic Optimality Theory”. Ms, University of Potsdam. [accessed in August 2004] [http://www.uni-bielefeld.de/lili/personen/gjaeger/mesga.pdf] Jäger, Gerhard. 2002. EvolOT: Software for Simulating Language Evolution Using Stochastic Optimality Theory. Program and user manual. [http://www.ling.uni-potsdam.de/~jaeger/evolOT] Jäger, Gerhard and Anette Rosenbach. 2003. “The Winner Takes it All – Almost: Cumulativity in Grammatical Variation”. Ms, University of Potsdam & University of Düsseldorf. Jayaseelan, K.A. 1990. “Incomplete VP Deletion and Gapping”. Linguistic Analysis 20. 64-81. Jayaseelan, K.A. 2001. “IP-Internal Topic and Focus Phrases”. Studia Linguistica 55. 39-75. Jakubowicz, Celia and Catherine Rigaut. 1997. “L’acquisition des clitiques nominatifs en français”. Les pronoms: Morphologie, syntaxe et typologie, Anne Zribi-Hertz (ed.), 5799. [Sciences du langage] Paris: Presses Universitaires de Vincennes. Jember, Gregory K., John C. Carrell, Robert P. Lundquist, Barbara M. Olds, Raymond P. Tripp Jr. (eds). 1975. English - Old English, Old English - English Dictionary. Boulder, CO: Westview Press.

390

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Jespersen, Otto. 1894. Progress in Language, with Special Reference to English. London: S. Sonnenschein. Jespersen, Otto. 1934[1924]. The Philosophy of Grammar. (1992 facsimile edition with a new introduction and index by James D. McCawley). Chicago: University of Chicago Press. Jespersen, Otto. 1946. A Modern English Grammar on Historical Principles, Part V: Syntax, Fourth Volume. London: Allen & Unwin. Jespersen, Otto. 1949[1927]. A Modern English Grammar on Historical Principles, Part III: Syntax, Second Volume. Copenhagen: Einar Munksgaard, and London: Allen & Unwin. Jespersen, Otto and Niels Haislund. 1949. A Modern English Grammar on Historical Principles, Part VII: Syntax. Copenhagen: Ejnar Munksgaard. Johannessen, Janne Bondi. 1993. Coordination: A Minimalist Approach. PhD thesis, University of Oslo. Johannessen, Janne Bondi. 1998. Coordination. [Oxford Studies in Comparative Syntax]. Oxford: Oxford University Press. Johnson, David E. and Shalom Lappin. 1999. Local Constraints vs. Economy. [Stanford Monographs in Linguistics] Stanford, CA: CSLI Publications. Johnson, Kyle. 1991. “Object Positions”. Natural Language and Linguistic Theory 9. 577-636. Johnson, Kyle. 1996. “In Search of the English Middle Field”. Ms, University of Massachusetts, Amherst. Jolly, Julia. 1993. “Preposition Assignment in English”. Advances in Role and Reference Grammar, Robert van Valin Jr. (ed.), 275-310. Amsterdam: John Benjamins. Jones, Michael Allan. 1988. “Cognate Objects and the Case-Filter”. Journal of Linguistics 24. 89-110. Kaplan, Ronald M. and John T. Maxwell III. 1995. “Constituent Coordination in LexicalFunctional Grammar”. Formal Issues in Lexical-Functional Grammar, Mary Dalrymple, Ronald M. Kaplan, John T. Maxwell III, and Annie Zaenen (eds), 199-210. [CSLI Lecture Notes 47] Stanford, CA: CSLI Publications. Kayne, Richard S. 1975. French Syntax: The Transformational Cycle. [Current Studies in Linguistics Series 6] Cambridge, MA: MIT Press. Kayne, Richard S. 1984. Connectedness and Binary Branching. [Studies in Generative Grammar 16] Dordrecht: Foris. Kayne, Richard S. 1994. The Antisymmetry of Syntax. [Linguistic Inquiry Monographs 25] Cambridge, MA: MIT Press. Kayne, Richard S. 2000. Parameters and Universals. [Oxford Studies in Comparative Syntax] Oxford: Oxford University Press. Kayne, Richard S. 2001. “Prepositions as Probes”. Ms, New York University. Kearns, Kate. 2000. Semantics. [Modern Linguistics Series] London: Macmillan, and New York: St. Martin’s Press. Keating, Patricia A. 1983. “Comments on the Jaw and Syllable Structure”. Journal of Phonetics 11. 401-406. Keenan, Ed L. 1997. “A Case Study in Language Evolution and Some Biological Parallels”. UCLA Cognitive Science seminar handout, also presented at the University of Canterbury, Christchurch, New Zealand. Kemenade, Ans van. 1987. Syntactic Case and Morphological Case in the History of English. Dordrecht: Foris. Kemenade, Ans van. 1997. “V2 and Embedded Topicalization in Old and Middle English”. Parameters of Morphosyntactic Change, Ans van Kemenade and Nigel Vincent (eds), 326-352. Cambridge: Cambridge University Press.

REFERENCES

391

Kennedy, Christopher and Jason Merchant. 2000. “Attributive Comparative Deletion”. Natural Language and Linguistic Theory 18. 89-146. Kenstowicz, Michael. 1994. Phonology in Generative Grammar. Oxford: Blackwell. Kiparsky, Paul. 1997. “The Rise of Positional Licensing”. Parameters of Morphosyntactic Change, Ans van Kemenade and Nigel Vincent (eds), 460-494. Cambridge: Cambridge University Press. Kiss, Katalin É. 1998. “Identificational Focus versus Information Focus”. Language 74. 245273. Kjellmer, Göran. 1986. “‘Us Anglos Are a Cut above the Field’: On Objective Pronouns in Nominative Contexts”. English Studies: A Journal of English Language and Literature 67. 445-449. Klima, Edward S. 1964. “Relatedness between Grammatical Systems”. Language 40. 1-20. Koopman, Willem F. 1992. “Old English Clitic Pronouns: Some Remarks”. Evidence for Old English: Material and Theoretical Bases for Reconstruction, Fran Colman (ed.), 44-87. [Edinburgh Studies in the English Language 2] Edinburgh: John Donald Publications. Ladusaw, William A. 1994. “Thetic and Categorical, Stage and Individual, Weak and Strong”. Proceedings from Semantics and Linguistic Theory (SALT) IV, 220-229. Ithaca, NY: Department of Modern Languages and Linguistics, Cornell University. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Lapointe, Steven. 1985. A Theory of Grammatical Agreement. [Outstanding Dissertations in Linguistics] New York: Garland. Larson, Richard K. 1985. “Bare-NP Adverbs”. Linguistic Inquiry 16. 595-621. Larson, Richard K. 1987. “‘Missing Prepositions’ and the Analysis of English Free Relative Clauses”. Linguistic Inquiry 18. 239-266. Larson, Richard K. 1988. “On the Double Object Construction”. Linguistic Inquiry 19. 335391. Lasnik, Howard. 1995. “A Note on Pseudogapping”. Papers on Minimalist Syntax, Rob Pensalfini and Hiroyuki Ura (eds), 143-163. [MIT Working Papers in Linguistics 27] Cambridge, MA: MITWPL. Lasnik, Howard and Nicholas Sobin. 2000. “The Who/Whom Puzzle: On the Preservation of an Archaic Feature”. Natural Language and Linguistic Theory 18. 343-371. Law, Paul. 2000. “On Relative Clauses and the DP/PP Adjunction Asymmetry”. In Alexiadou, Law, Meinunger & Wilder 2000. 161-199. Levin, Beth, and Malka Rappaport Hovav. 1995. Unaccusativity: At the Syntax-Lexical Semantics Interface. Cambridge, MA: MIT Press. Lobeck, Anne. 1995. Ellipsis: Functional Heads, Licensing, and Identification. Oxford: Oxford University Press. Longobardi, Giuseppe. 1994. “Reference and Proper Nouns”. Linguistic Inquiry 25. 609-665. Lumsden, John Stewart. 1987. Syntactic Features: Parametric Variation in the History of English. PhD thesis, MIT. McCreight, Katherine L. 1988. Multiple Case Assignments. PhD thesis, MIT. McFadden, Thomas. 2004. The Position of Morphological Case in the Derivation: A Study of the Syntax-Morphology Interface. PhD thesis, University of Pennsylvania. McLaughlin, John. 1983. Old English Syntax: A Handbook. [Sprachstrukturen, Reihe A, Historische Sprachstrukturen 4] Tübingen: Max Niemeyer.

392

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Meinunger, André. 1998. “A Monoclausal Structure for (Pseudo-) Cleft sentences”. Proceedings of the North East Linguistic Society (NELS) 28, Volume 1: Papers from the Main Sessions, Pius N. Tamanji and Kiyomi Kusumoto (eds), 283-298. Amherst, MA: GLSA, University of Massachusetts. Merchant, Jason. 2001. The Syntax of Silence: Sluicing, Islands, and the Theory of Ellipsis. [Oxford Studies in Theoretical Linguistics] Oxford: Oxford University Press. Mitchell, Bruce. 1985. Old English Syntax, Volume I: Concord, the Parts of Speech, and the Sentence. Oxford: Clarendon Press. Mohanan, K.P. 1998. “Unpacking Optimality Theory: Non-Monotonicity as Transformations, Rule Ordering and Constraint Ranking”. Paper presented at the ALS Conference, 4 July 1998, Brisbane. Moltmann, Friederike. 1992. Coordination and Comparatives. PhD thesis, MIT. Moore, Samuel. 1928. “Earliest Morphological Changes in Middle English”. Language 4. 238266. Munn, Alan. 1987. “Coordinate Structure and X-Bar Theory”. McGill Working Papers in Linguistics 4. 121-140. Munn, Alan Boag. 1993. Topics in the Syntax and Semantics of Coordinate Structures. PhD thesis, University of Maryland. Mustanoja, Tauno. 1960. A Middle English Syntax, Part I: Parts of Speech. [Mémoires de la Société Néophilologique de Helsinki XXIII] Helsinki: Société Néophilologique. Nagy, Naomi and Bill Reynolds. 1997. “Optimality Theory and Variable Word-Final Deletion in Faetar”. Language Variation and Change 9. 37-55. Nash, Léa. 1997. “La partition personelle dans les langues ergatives”. Les pronoms: Morphologie, syntaxe et typologie, Anne Zribi-Hertz (ed.), 129-149. [Sciences du langage] Paris: Presses Universitaires de Vincennes. Nicol, Fabrice. 2002. “Extended VP-Shells and the Verb-Particle Construction”. Verb-Particle Explorations, Nicole Dehé, Ray Jackendoff, Andrew McIntyre, and Silke Urban (eds), 165-190. [Interface explorations 1] Berlin & New York: Mouton de Gruyter. Ouhalla, Jamal. 1999. “Focus and Arabic Clefts”. The Grammar of Focus, Georges Rebuschi and Laurice Tuller (eds), 335-359. [Linguistik Aktuell/Linguistics Today 24] Amsterdam: John Benjamins. Panagiotidis, Phoevos. 2002a. Pronouns, Clitics and Empty Nouns: ‘Pronominality’ and Licensing in Syntax. [Linguistik Aktuell/Linguistics Today 46] Amsterdam: John Benjamins. Panagiotidis, Phoevos. 2002b. “Pronominal Nouns”. Pronouns – Grammar and Representation, Horst J. Simon and Heike Wiese (eds), 183-203. [Linguistik Aktuell/ Linguistics Today, 52] Amsterdam: John Benjamins. Parker, Frank, Kathryn Riley, and Charles Meyer. 1988. “Case Assignment and the Ordering of Constituents in Coordinate Constructions”. American Speech 63. 214-233. Payne, John and Rodney Huddleston. 2002. “Nouns and Noun Phrases”. The Cambridge Grammar of the English Language, Rodney Huddleston and Geoffrey K. Pullum (eds), 323-532. Cambridge: Cambridge University Press. Peterson, Peter G. 2004. “Coordination: Consequences of a Lexical-Functional Account”. Natural Language and Linguistic Theory 22. 643-679. Pinker, Steven. 1984. Language Learnability and Language Development. [Cognitive Science Series 7] Cambridge, MA: Harvard University Press. Pintzuk, Susan. 1995. “Variation and Change in Old English Clause Structure”. Language Variation and Change 7. 229-260.

REFERENCES

393

Pintzuk, Susan. 1996. “Cliticization in Old English”. Approaching Second: Second Position Clitics and Related Phenomena, Aaron L. Halpern and Arnold M. Zwicky (eds), 375409. [CSLI Lecture Notes 61] Stanford, CA: CSLI Publications. Platzack, Christer. 2000. “A Complement-of-N0 Account of Restrictive and Non-Restrictive Relatives: The Case of Swedish”. In Alexiadou, Law, Meinunger & Wilder 2000. 265308. Pollard, Carl and Ivan A. Sag. 1994. Head-Driven Phrase Structure Grammar. Chicago: University of Chicago Press. Postal, Paul M. 1969[1966]. “On So-Called ‘Pronouns’ in English”. Modern Studies in English: Readings in Transformational Grammar, David A. Reibel and Sanford A. Shane (eds), 201-224. Englewood Cliffs, NJ: Prentice-Hall. Potsdam, Eric. 1998. Syntactic Issues in the English Imperative. [Outstanding Dissertations in Linguistics] New York: Garland. Priestley, Joseph. 1761. The Rudiments of English Grammar. (Microfilm edition 1971). New York: Garland. Primus, Barbara. 1987. Grammatische Hierarchien: Eine Beschreibung und Erklärung von Regularitäten des Deutschen ohne grammatische Relationen. Munich: Wilhelm Fink. Primus, Barbara. 1991. “Hierarchiegesetze der Universalgrammatik ohne grammatische Relationen”. Det, Comp und Infl: Zur Syntax funktionaler Kategorien und grammatischer Kategorien, Susan Olsen and G. Fanselow (eds). Tübingen: Max Niemeyer. Prince, Alan and Paul Smolensky. 1993. Optimality Theory: Constraint Interaction in Generative Grammar. [Technical Reports of the Rutgers University Center for Cognitive Science 2] Piscataway, NJ: Rutgers University Center for Cognitive Science. Prince, Ellen F. 1998. “On the Limits of Syntax, with Reference to Left-Dislocation and Topicalization”. Peter W. Culicover and Louise McNally (eds), 281-302. Syntax and Semantics 29: The Limits of Syntax. San Diego: Academic Press. Przepiórkowski, Adam. 1998. “On Case Assignment and ‘Adjuncts as Complements’”. Lexical and Constructional Aspects of Linguistic Explanation, Gert Webelhuth, Jean-Pierre Koenig, and Andreas Kathol (eds), 231-245. [Studies in Constraint-Based Lexicalism] Stanford, CA: CSLI Publications. Pustejovsky, James. 1995. The Generative Lexicon. Cambridge, MA: MIT Press. Quattlebaum, Judith A. 1994. “A Study of Case Assignment in Coordinate Noun Phrases”. Language Quarterly 32. 131-147. Quinn, Heidi. 1995. Variation in NZE Syntax and Morphology: A Study of the Acceptance and Use of Grammatical Variants among Canterbury and West Coast Teenagers. MA thesis, University of Canterbury, Christchurch, New Zealand. Quinn, Heidi. 2005. “Testing and Modelling the Distribution of Pronoun Case Forms in English”. Ms, University of Canterbury, New Zealand. Quinn, Heidi and Sharbani Banerji. 2003. “Focus and the Position of the Object in PseudoGapping and V-Particle Constructions”. Paper presented at the New Zealand Linguistics Society (NZLS) Conference, Victoria University of Wellington, 5 September 2003. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Radford, Andrew. 1988. Transformational Grammar: A First Course. [Cambridge Textbooks in Linguistics] Cambridge: Cambridge University Press. Raposo, Eduardo and Juan Uriagereka. 1995. “Two Types of Small Clauses (Toward a Syntax of Theme/Rheme Relations)”. In Cardinaletti & Guasti 1995. 179-206.

394

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Reinhart, Tanya. 1976. The Syntactic Domain of Anaphora. PhD thesis, MIT. Reinhart, Tanya. 1981. “Definite NP Anaphora and C-Command Domains”. Linguistic Inquiry 12. 605-635. Riemsdijk, Henk van. 1978. A Case Study in Syntactic Markedness: The Binding Nature of Prepositional Phrases. Dordrecht: Foris. Riemsdijk, Henk van. 1997. “Left Dislocation”. Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk, and Frans Zwarts (eds), 1-10. [Linguistik Aktuell/Linguistics Today, 14] Amsterdam: John Benjamins. Rigter, B. 1980. “Determiners and the Syntax of Pronominals, Relativisation and ModifierShift”. The Semantics of Determiners, Johan van der Auwera (ed.), 135-155. Baltimore: University Park Press. Ritter, Elizabeth. 1991. “Two Functional Categories in Noun Phrases: Evidence from Modern Hebrew”. Syntax and Semantics 25: Perspectives on Phrase Structure: Heads and Licensing, Susan D. Rothstein (ed.), 37-62. San Diego: Academic Press. Ritter, Elizabeth. 1995. “On the Syntactic Category of Pronouns and Agreement”. Natural Language and Linguistic Theory 13. 405-443. Ritter, Elizabeth and Sara Rosen. 2000. “Event Structure and Ergativity”. Events as Grammatical Objects: The Converging Perspectives of Lexical Semantics and Syntax, Carol Tenny and James Pustejovsky (eds), 187-238. [CSLI Lecture Notes 100] Stanford, CA: CSLI Publications. Rizzi, Luigi. 1990. Relativized Minimality. [Linguistic Inquiry Monographs 16] Cambridge, MA: MIT Press. Rizzi, Luigi. 1996. “Residual Verb Second and the Wh-Criterion”. Parameters and Functional Heads: Essays in Comparative Syntax, Adriana Belletti and Luigi Rizzi (eds), 63-90. [Oxford Studies in Comparative Syntax] Oxford: Oxford University Press. Rizzi, Luigi. 1997. “The Fine Structure of the Left Periphery”. Elements of Grammar, Liliane Haegeman (ed.), 281-337. Dordrecht: Kluwer. Rodman, Robert. 1997[1974]. “On Left Dislocation”. Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk, and Frans Zwarts (eds), 31-54. [Linguistik Aktuell/Linguistics Today, 14] Amsterdam: John Benjamins. Ross, John Robert. 1986[1967]. Infinite Syntax!. [Language and Being] Norwood, NJ: Ablex Publishing Corporation. Rothstein. Susan. 1988. “Conservativity and Determiners”. Linguistics 26. 999-1019. Runner, Jeffrey T. 1998. Noun Phrase Licensing. [Outstanding Dissertations in Linguistics] New York: Garland. Sawai, Naomi Kei. 1997. An Event Structure Analysis of Derived Verbs in English. MA thesis, University of Calgary. Schmid, Tanja. 2001. “OT Accounts of Optionality: A Comparison of Global Ties and Neutralization”. Competition in Syntax, Gereon Müller and Wolfgang Sternefeld (eds), 283-319. [Studies in Generative Grammar 49] Berlin: Mouton de Gruyter. Schütze, Carson T. 1996. The Empirical Base of Linguistics: Grammaticality Judgments and Linguistic Methodology. Chicago: University of Chicago Press. Schütze, Carson T. 1997. INFL in Child and Adult Language: Agreement, Case and Licensing. PhD thesis, MIT. Schütze, Carson T. 2001. “On the Nature of Default Case”. Syntax 4. 205-238.

REFERENCES

395

Schwartz, Bonnie D. 1985. “Case and Conjunction”. Studies in Syntax, Gary Gilligan, Mohammad Mohammad, and Ian Roberts (eds), 161-186. [Southern California Occasional Papers in Linguistics 10] Los Angeles: Department of Linguistics, University of Southern California. Selkirk, Elisabeth O. 1980. The Phrase Phonology of English and French. [Outstanding Dissertations in Linguistics] New York: Garland. Selkirk, Elisabeth O. 1984. “On the Major Class Features and Syllable Theory”. Language, Sound, Structure: Studies in Phonology Presented to Morris Halle by his Teacher and Students, Mark Aronoff and Richard T. Oehrle (eds), 107-136. Cambridge, MA: MIT Press. Selkirk, Elisabeth O. 1995. “The Prosodic Structure of Function Words”. Papers in Optimality Theory, Jill Beckman, Laura Walsh Dickey, and Suzanne Urbanczyk. (eds), 439-459. [University of Massachusetts Occasional Papers 18]. Amherst, MA: GLSA, University of Massachusetts. Siegel, Muffy E.A. 1987. “Compositionality, Case and the Scope of Auxiliaries”. Linguistics and Philosophy 10. 53-75. Sigley, Robert J. 1997. Choosing Your Relatives: Relative Clauses in New Zealand English. PhD thesis, Victoria University, Wellington, New Zealand. Sigursson, Halldór Ármann. 2003. “Case: Abstract vs. Morphological”. New perspectives on case theory, Ellen Brandner and Heike Zinsmeister (eds), 223-265. [CSLI Lecture Notes 156] Stanford, CA: CSLI Publications. Sigursson, Halldór Ármann. 2005. “Germanic Nom/Acc”. Paper presented at Grammar In Focus (GIF) 19, Lund, 10-11 February 2005. Siloni, Tal. 1997. Noun Phrases and Nominalizations: The Syntax of DPs. [Studies in Natural Language and Linguistic Theory 40] Dordrecht: Kluwer. Sobin, Nicholas. 1997. “Agreement, Default Rules, and Grammatical Viruses”. Linguistic Inquiry 28. 318-343. Spinillo, Mariangela Galvão. 2004. Reconceptualising the English Determiner Class. PhD thesis, University College London. Sportiche, Dominique. 1988. “A Theory of Floating Quantifiers and its Corollaries for Constituent Structure”. Linguistic Inquiry 19. 425-449. Steinki, Johannes. 1932. Die Entwicklung der englischen Relativpronomina in spätmittelenglischer und frühneuenglischer Zeit. Inaugural-Dissertation, Breslau. Stowell, Tim. 1981. Origins of Phrase Structure. PhD thesis, MIT. Stowell, Tim. 1995. “Remarks on Clause Structure”. In Cardinaletti & Guasti 1995. 271-286. Sudbury, Andrea and Jen Hay. 2001. “Rhoticity and /r/-Sandhi: Some Data on /r/ in Early New Zealand English”. Paper presented at the 14th New Zealand Linguistic Society Conference (LINGSOC XIV), University of Canterbury, Christchurch, New Zealand. Suñer, Margarita. 1984. “Free Relatives and the Matching Parameter”. The Linguistic Review 3. 363-387. Taglicht, Josef. 1998. “Constraints on Intonational Phrasing in English”. Journal of Linguistics 34. 181-211. Toivonen, Ida. 2003. Non-Projecting Words: A Case Study of Swedish Verbal Particles. [Studies in Natural Language and Linguistic Theory 58] Dordrecht: Kluwer. Traugott, Elizabeth Closs. 1972. A History of English Syntax: A Transformational Approach to the History of English Sentence Structure. [Transatlantic Series in Linguistics] New York: Holt, Rinehart and Winston.

396

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Vallduví, Enric and Maria Vilkuna. 1998. “On Rheme and Kontrast”. Syntax and Semantics 29: The Limits of Syntax, Peter W. Culicover and Louise McNally (eds), 79-108. San Diego: Academic Press. Vat, Jan. 1997. “Left Dislocation, Connectedness and Reconstruction”. Materials on Left Dislocation, Elena Anagnostopoulou, Henk van Riemsdijk, and Frans Zwarts (eds), 6792. [Linguistik Aktuell/Linguistics Today, 14] Amsterdam: John Benjamins. Vikner, Sten. 1995. Verb Movement and Expletive Subjects in the Germanic Languages. [Oxford Studies in Comparative Syntax] Oxford: Oxford University Press. Visser, F. Th. 1963. An Historical Syntax of the English Language, Part 1: Syntactical Units with one Verb. Leiden: E. J. Brill. Vogel, Ralf. 2001. “Case Conflict in German Free Relative Constructions: An Optimality Theoretic Treatment”. Competition in Syntax, Gereon Müller and Wolfgang Sternefeld (eds), 341-375. [Studies in Generative Grammar 49] Berlin: Mouton de Gruyter. Vogel, Ralf. 2003. “Surface Matters: Case Conflict in Free Relative Constructions and Case Theory”. New perspectives on case theory, Ellen Brandner and Heike Zinsmeister (eds), 269-299. [CSLI Lecture Notes 156] Stanford, CA: CSLI Publications. Wales, Katie. 1996. Personal Pronouns in Present-Day English. Cambridge: Cambridge University Press. Wallis, John. 1765[1653]. Grammar of the English Language: With an Introductory Grammatico-Physical Treatise on Speech (Or on the Formation of all Speech Sounds). (1972 edition with translation and commentary by J.A. Kemp). London: Longman. Wasow, Thomas. 1997. “Remarks on Grammatical Weight”. Language Variation and Change 9. 81-105. Watanabe, Akira. 1996. Case Absorption and Wh-Agreement. [Studies in Natural Language and Linguistic Theory 37] Dordrecht: Kluwer. Wechsler, Stephen. 1995. The Semantic Basis of Argument Structure. [Dissertations in Linguistics] Stanford, CA: CSLI Publications. Williams, Edwin. 1997. “Blocking and Anaphora”. Linguistic Inquiry 28. 577-628. Woodworth, Nancy L. 1991. “Sound Symbolism in Proximal and Distal Forms”. Linguistics 29. 273-299. Wunderlich, Dieter. 1997. “Cause and the Structure of Verbs”. Linguistic Inquiry 28. 27-68. Wurmbrand, Susanne. 2001. Infinitives: Restructuring and Clause Structure. [Studies in Generative Grammar 55] Berlin: Mouton de Gruyter. Yip, Moira, Joan Maling, and Ray Jackendoff. 1987. “Case in Tiers”. Language 63. 217-250. Zoerner, Cyril Edward III. 1995. Coordination: The Syntax of &P. PhD thesis, University of California, Irvine. Zribi-Hertz, Anne and Liliane Mbolatianavalona. 1999. “Towards a Modular Theory of Linguistic Efficiency: Evidence from Malagasy Personal Pronouns”. Natural Language and Linguistic Theory 17. 161-218. Zubizarreta, Maria Luisa. 1998. Prosody, Focus, and Word Order. [Linguistic Inquiry Monographs 33] Cambridge, MA: MIT Press. Zubritskaya, Katya. 1997. “Mechanism of Sound Change in Optimality Theory”. Language Variation and Change 9. 121-148. Primary sources of OE and ME examples taken from secondary sources Ælfric Anglo-Saxon = Thorpe, Benjamin (ed.). 1844. The Homilies of the Anglo-Saxon Church. London: Ælfric Society. (Reprinted 1971. New York: Johnson Reprint Corp). Cited by volume, page, and line number.]

REFERENCES

397

Ælfric Catholic Homilies = Godden, Malcolm (ed.). 1979. Ælfric’s Catholic Homilies: The Second Series. [EETS s.s.5]. Cited by homily and line number. Ælfric Homilies = Pope, John (ed.). 1967. Homilies of Ælfric: A Supplementary Collection. [EETS 259,260]. Cited by homily and line number. Ælfric Lives of Saints = Skeat, Walter W. (ed.). 1881-1900. Ælfric’s Lives of Saints. [EETS 76,82,94,114]. London: Trübner. Cited by life and line number. Ælfric Wulstan = “Ælfric’s First and Second Letters to Wulstan”. In Fehr, B. 1914. Die Hirtenbriefe Ælfrics in altenglischer und lateinischer Fassung. [Bibliothek der angelsächsischen Prosa, IX] Hamburg: Verlag von Henri Grand. Cited by page and line number (line = line within text). Bede = Miller, Thomas (ed.). 1890-98. The Old English Version of Bede’s Ecclesiastical History of the English People. [EETS 95-6,110-11] London: Trübner. (Reprinted 195963. London: Oxford University Press). Beowulf = Klaeber, Fr. (ed.). 1922. Beowulf and The Fight at Finnsburg. Cited by line number. Beowulf (3rd ed.) = Klaeber, Fr. (ed.). 1950. Beowulf and The Fight at Finnsburg (3rd edition). Lexington, MA: D.C. Heath. Cited by line number. Chronicle A = Chronicle, MS. A. In Plummer, Charles (ed.). 1892. Two of the Saxon Chronicles Parallel. Oxford: Clarendon. Cited by page and line number (year in brackets). COE = Healy, Antoinette and Venezky, Richard. 1980. A Microfiche Concordance to Old English. Toronto: Centre for Medieval Studies, University of Toronto. Parker Chronicle = The Parker Chronicle, MS. A. In Plummer, C. (ed.) 1892-1899. Two of the Saxon Chronicles in Parallel, 2 vols. Oxford: Clarendon, Oxford. (Reissued 1952 by D. Whitelock). Pastoral Care = Sweet, Henry (ed.) 1871. King Alfred’s West-Saxon Version of Gregory’s Pastoral Care. [EETS 45, 50] London: Trübner. Cited by page and line number. Pecock Repressor = Babington, Ch. (ed.). 1860. Pecock’s The Repressor over Much Blaming of the Clergy. [RBMAS 19] London. Robert of Gloucester = Wright, W.A. (ed.). 1887. The Metrical Chronicle of Robert of Gloucester. [RBMAS 86] London. Wulfstan = Bethurum, Dorothy (ed.). 1957. The Homilies of Wulfstan. Oxford: Clarendon. Cited by homily and line number.

NAME INDEX

A Aarts, B. 262 Abney, S.P. 37, 51f, 82, 130, 228, 248f, 253ff, 271, 288f Akmajian, A. 75, 83, 139, 263, 265f, 269, 271 Alexiadou, A. 50, 306, 337, 365f Allan, K. 158, 164, 168 Allen, C.L. 7-11, 13-21, 24, 30f, 56, 311, 331f, 335f, 342, 352ff, 371-379 Ambar, M. 48 Anagnostopoulou, E. 215 Angermeyer, P.S. 2, 81, 96, 98, 102f, 167 Anttila, A. 184 Arnold, M.D. 55f Aronoff, M. 381 B Baker, C.L. 333f Ball, C.N. 140, 355 Banerji, S. 46 Bartlett, C. 63 Barton, E.L. 229, 232, 235 Bayer, J. 286 Behaghel, O. 157f Benveniste, É. 65-69, 132 Bianchi, V. 306, 363ff Bittner, M. 48 Bobaljik, J. 315, 360 Bock, J.K. 158, 169 Boersma, P. 184, 186, 198 Bolinger, D. 48 Boyland, J.T. 2, 102 Bowers, J. 45 Borsley, R.D. 365 Bošković, Ž. 315 Bresnan, J. 189, 331, 336 Brousseau, A.-M. 41 Browning, M.A. 210 Büring, D. 184

Burzio, L. 26, 43 C Campbell, L. 157f, 258 Cardinaletti, A. 1, 65f, 74f, 82, 130f, 228, 270, 275, 289 Carrell, J.C. see Jember, G.K. Chao, W. 86, 223f, 230, 273f, 276, 279f, 326, 345f Chomsky, N. 14, 26, 39ff, 43f, 48, 54, 78, 158, 316, 356 Chung, S. 224, 326, 328 Clifton, C. Jr. 326 Collins, C. 47, 241 Cormack, A. 48 D Dalrymple, M. 80 Deevy, P.L. 63 Dehé, N. 48 Delahunty, G.P. 82, 84, 139, 141, 306 Demirdache, H. 210, 214f, 331 Denison, D. 17, 24, 102, 125, 143, 174, 203, 229, 234, 286f Dogil, G. 159f, 163, 169 Downing, L.J. 170 E Ebert, R.P. 258 Emonds, J.E. 1f, 79f, 101, 139, 203, 234 Erdmann, P. 2, 83, 102, 139, 214, 229f, 232, 234f, 245f, 259, 265, 273, 278, 283, 289, 304, 308 Everett, D.L. 171, 245 Everett, K. 171 F Foulet, L. 65f, 243, 380 Frazier, L. 326

NAME INDEX

G Galves, C. 65 Gelderen, E. van 11ff, 18, 47, 55, 62f, 74, 76, 80-85, 102, 140f, 243, 245, 280, 311, 315-320, 352-355 Ghomeshi, J. 132 Gil, D. 157, 164 Giusti, G. 131, 133 Goldwater, S. 186 Greenbaum, S. see Quirk, R. Grimshaw, J. 331, 336 Grohmann, K.K. 299f Grosu, A. 338f Guasti, M.T. 270 Gundel, J.K. 82, 85, 246 Guy, G.R. 6, 186ff H Haislund, N. 2, 11f, 16, 21ff, 32, 82, 85f, 125, 133, 143, 168, 203, 205ff, 212215, 221, 225, 228f, 233ff, 239, 241, 243-246, 259, 265, 270, 278, 280, 282f, 285f, 288f, 291-294, 297, 299f, 302305, 307f, 310, 315f, 326, 344, 355f, 383 Hale, K. 41, 48, 269, 290 Harley, H. 132 Harris, A.C. 157f, 258 Harris, M. 80, 101, 139, 203, 218, 246 Hawkins, J.A. 156ff Hay, J. 63, 161 Hayes, B. 184, 186, 198 Hedberg, N. 82, 85, 139 Heggie, L. 82, 85 Heim, I.R. 227 Henry, A. 61-65, 81, 101 Higginbotham, J. 31, 33, 51 Householder, F.W. 2, 101f, 205ff, 214, 221, 226, 229, 234, 245, 310ff, 335, 354f, 357 Huddleston, R. 82 Hudson, R. 82, 101, 130f I Ihalainen, O. 316

399

J Jackendoff, R. 82 see also Yip, M. Jäger, G. 186, 189 Jayaseelan, K.A. 82, 84, 275, 279 Jakubowicz, C. 67 Jember, G.K. 11f Jespersen, O. 2, 10ff, 16, 21ff, 32, 37, 52, 69, 82, 85f, 125, 132f, 143, 157f, 168, 203, 205ff, 212-215, 217, 221, 225, 228f, 233-236, 239, 241, 243-246, 248254, 258ff, 263ff, 270, 272f, 275, 278, 280, 282-286, 288f, 291-295, 297, 299305, 307f, 310, 313, 315f, 318, 325f, 331, 333-344, 348-352, 354-359, 383 Johannessen, J.B. 2, 79f, 102, 122 Johnson, D.E. 39 Johnson, K. 28, 44ff, 272, 274, 276f, 279, 319 Johnson, M. 186 Jolly, J. 32, 56 Jones, M.A. 79, 101 K Kaplan, R. 79f Kayne, R.S. 49, 65-69, 82, 84, 154, 215, 221-225, 262, 276, 279, 339, 357f, 360f, 364ff Kearns, K. 30, 45f, 50, 52, 152, 255ff, 266f, 282, 330 Keating, P.A. 171 Keenan, E.L. 13 Kemenade, A. van 9-12, 16, 20, 373ff, 378f Kennedy, C. 272, 274ff, 279 Kenstowicz, M. 159f Keyser, S.J. 41, 269, 290 Kiparsky, P. 3, 7, 27ff, 370-373 Kiss, K.É. 84f Kjellmer, G. 20f, 95, 124ff, 203, 229, 300ff Klima, E.S. 2, 80, 101, 245f, 310ff, 325 Koopman, W.F. 377, 379f L Ladusaw, W. 266 see also Chung, S. Lakoff, G. 47, 63

400

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

Lapointe, S. 381 Lappin, S. 39 Larson, R.K. 27, 35f, 41, 317, 336 Lasnik, H. 1, 205f, 246, 248, 272, 274ff, 279, 310, 312f, 319, 325, 329, 357f, 362 Law, P. 49, 261, 346, 360 see also Alexiadou, A. Leech, G. see Quirk, R. Levin, B. 41, 47, 63 Lobeck, A. 50, 87, 223f, 230, 279, 292, 297, 325f, 328f, 345f, 366 Longobardi, G. 82, 130, 228, 289f Lumsden, J.S. 80, 101 Lundquist, R.P. see Jember, G.K. Luschützky, H. 159f, 163, 169 M Maling, J. see Yip, M. Massam, D. 56, 204, 219, 325, 359 Maxwell, J. 79f Mbolatianavalona, L. 65 McCloskey, J. see Chung, S. McCreight, K.L. 26, 338, 340, 363, 366 McFadden, T. 2, 59, 78, 200, 373 McLaughlin, J. 378 Meinunger, A. 82, 84 see also Alexiadou, A. Merchant, J. 87, 223, 230, 272, 274ff, 279, 325-329, 345 Meyer, C. see Parker, F. Mitchell, B. 9f, 18, 55 Mohanan, K.P. 6, 186ff Moltmann, F. 278, 281 Moore, S. 13 Munn, A. 80, 122 Mustanoja, T. 12, 14, 21f, 294, 310, 354 N Nagy, N. 184 Nash, L. 48 Nicol, F. 45 O Olds, B.M. see Jember, G.K. Ouhalla, J. 48

P Panagiotidis, P. 76, 132 Parker, F. 2, 96, 102 Payne, J. 82 Pearce, E. 30, 56, 74f, 131f, 161, 166, 170, 295, 306, 317, 320, 342, 344 Peterson, P.G. 1, 79f Pinker, S. 381 Pintzuk, S. 243, 330f Platzack, C. 365 Pollard, C. 26, 79f Postal, P.M. 82, 289 Potsdam, E. 49, 210, 214, 295 Priestley, J. 21, 23f Primus, B. 157 Prince, A. 180, 186 Prince, E.F. 214 Przepiórkowski, A. 27, 35f, 38, 317 Pustejovsky, J. 33f Q Quattlebaum, J.A. 2, 102 Quinn, H. 2, 46, 94 Quirk, R. 139, 229, 246, 288, 294-297, 355 R Radford, A. 45, 357, 361 Raposo, E. 266f, 271 Rappaport Hovav, M. 41, 47, 63 Reinhart, T. 153 Reynolds, B. 184 Riemsdijk, H. van 210, 215f, 329 Rigaut, C. 67 Rigter, B. 131 Riley, K. see Parker, F. Ritter, E. 41, 45, 50f, 132, 292 Rizzi, L. 84, 141, 208ff, 216, 222 Rodman, R. 214 Rosen, S. 45 Rosenbach, A. 186, 189 Ross, J.R. 214, 216 Rothstein. S. 297 Runner, J.T. 55 S Sag, I.A. 26, 79f Sawai, N.K. 45

NAME INDEX

Schmid, T. 184 Schreier, D. 63 Schütze, C.T. 14, 39, 59, 79, 98, 263, 271 Schwartz, B.D. 2, 102 Selkirk, E.O. 74f, 168, 170 Siegel, M.E.A. 263, 271, 274, 276 Sigley, R.J. 310f, 355-359 Sigursson, H.Á. 2, 78, 200, 278, 381 Siloni, T. 50 Singler, J. 2, 81, 96, 98, 102f, 167 Smolensky, P. 180, 186 Sobin, N. 1f, 63, 82, 87, 102, 135, 205f, 234, 246, 248, 310, 312f, 319, 325, 329, 357f, 362 Spinillo, M.G. 82, 130-131, 133 Sportiche, D. 295 Steinki, J. 335 Starke, M. 1, 65f, 74f, 275 Stowell, T. 266f Sudbury, A. 161 Suñer, M. 338 Svartvik, J. see Quirk, R. T Taglicht, J. 360f Toivonen, I. 48 Traugott, E.C. 243, 310f Tripp, R.P. Jr. see Jember, G.K. U Uriagereka, J. 266f, 271

401

V Vallduví, E. 48 Vat, J. 215 Vikner, S. 245 Vilkuna, M. 48 Visser, F.Th. 2, 140, 142, 219, 224, 233ff, 239, 243, 245f, 303, 306-307, 355f Vogel, R. 340 W Wales, K. 2, 21, 23f, 82, 95, 124ff, 133, 135, 139, 143, 203, 229, 234, 245f, 304, 307, 316 Wallis, J. 21, 23f Wasow, T. 156ff Watanabe, A. 48 Wechsler, S. 30 Wilder, C. 50 see also Alexiadou, A. Williams, E. 205f, 208, 381 Woodworth, N.L. 169f Wunderlich, D. 3, 27-31, 33ff, 51, 372 Wurmbrand, S. 49, 52, 253, 260f, 269, 346 Y Yip, M. 19 Z Zoerner, C.E. 2, 79f, 102, 122 Zribi-Hertz, A. 65 Zubizarreta, M.L. 54 Zubritskaya, K. 184

SUBJECT INDEX

A Acc-ing gerund 52, 255-258 accusative case (ACC) 8-20, 66f, 216, 352f, 371-374 see also objective case adjacency 210f, 315, 323-325, 327, 347, 360 adjective and AP 9, 13, 27f, 32f, 38, 51, 73, 75, 130, 145, 228, 267, 288-291, 373f adverb 243, 248, 269f, 331, 376f bare NP adverb 35f, 317 adverbial 139, 260, 356-358 agreement 4, 14, 17ff, 39ff, 58, 61-64, 66, 70, 76f, 81, 83, 131, 133, 140f, 203, 212, 266, 271, 298, 338f, 346 and argument licensing 370f, 373, 375 head-head agreement 223 historical changes 242-246 spec-head agreement 39ff, 66, 70, 76f, 128, 316, 338f, 360 subject-verb agreement 17ff, 61-64, 81, 234f, 242-246, 282f, 315, 326f variation 62ff, 245 agreement-related functional head 4, 3955, 58ff, 66, 69f, 73f, 76f, 128, 202, 204, 230, 271, 296, 308f, 314, 317, 319, 330, 361, 363 appositive 69, 297f, 300-303 argument 84f, 138-142, 211, 216, 218, 223f, 226f, 230ff, 235-242, 244248, 253f, 260, 262, 271, 282, 286, 293, 296, 305f, 321f argument relative 331-352 external argument 32, 41-45, 54, 268 internal argument 41-45, 54 referential argument 29, 31, 33-36, 38, 51, 57, 311 structural argument 3f, 7f, 27-38, 42, 51, 57-60, 63f, 148, 178, 201, 311,

313f, 317, 370-375 Argument (Arg-)Agreement 63f, 140f, 242, 244f Argument (Arg-)Case 3, 6f, 26f, 57-61, 64, 79, 82-87, 107, 122ff, 130f, 139142, 147-150, 175-178, 181-185, 191199, 201, 203f, 211f, 216ff, 224-232, 236-244, 254-257, 260ff, 271, 274-277, 280-293, 296-302, 311-314, 318f, 322325, 328, 330, 337-343, 346-351, 359369 argument hierarchy 27f, 31, 36, 53f, 204, 244, 306, 313, 328, 370-375, 381 argument position 13, 74, 204f, 208f, 249, 258, 286, 367, 371-374 argument structure 26-39, 41-47, 53-58, 231, 289, 308 adjectival predicate 27f, 31ff, 38, 267 nominal predicate 27f, 31-38, 131, 268, 311 prepositional predicate 3, 17-20, 26ff, 31ff, 38, 53-56, 226f, 267f verbal predicate 3, 17-20, 22f, 26-32, 35, 37f, 41-47, 64, 84, 235-242, 244-248, 254, 258, 260, 262, 267270, 313, 317, 321f, 370-375, 381 asymmetric c-command 5, 124, 139f, 145, 153-157, 163, 202ff, 213, 218, 228, 232f, 247f, 281, 285ff, 293, 298, 300, 307f, 330f, 362, 369, 381ff B bare argument ellipsis 68, 204, 231, 278ff, 283 Bavarian German 326f Belfast English 61-64, 81, 101 binding 166, 208f, 306 see also theta-binding Blocking Principle 381 Burzio’s Generalization 43

SUBJECT INDEX

C C(omplementizer) and CP 40-41f, 48f, 54f, 59, 64, 70-73, 77, 84f, 140f, 202, 208-211, 215f, 222-226, 243, 254, 256, 260f, 274f, 284, 287, 308f, 314, 316-318, 323, 326ff, 331, 337-343, 347-352, 359-369, 376f for 38, 49f, 54, 202, 258-261, 324, 367 that, ðæt, ðe 82-85, 88, 92, 133f, 136ff, 142, 208ff, 215-216, 30ff, 311, 352f, 355 c-command 34, 141, 153-154 see also asymmetric c-command case agreement 85, 140, 142, 216f, 238ff, 298, 306, 322, 338f, 363-366 case and structural position 2-7, 121, 145150, 203, 308f, 369-375, 381ff in the clause 2, 4, 17-20, 22, 25ff, 76f, 79, 82-87, 98f, 103-129, 148ff, 175ff, 181-185, 189-199, 203, 289, 363 within a construction 2f, 5, 80, 98f, 101-124, 127f, 130-133, 135, 139ff, 144-171, 175ff, 181-186, 189-199, 202ff, 329-331 case assignment 1-8, 25-64, 201 historical changes 6ff, 17-20, 370, 373ff case checking 27, 39ff, 46, 51-64, 316f see also Positional (Pos-)Case case clash 303f, 334-342, 352-355 case inheritance see case agreement case percolation 131, 289, 291, 293 case syncretism 8-17, 20-24 case transmission see case agreement case variation 1ff, 6, 63f, 66, 70-74, 77f, 82f, 101-155, 161ff, 165, 171-177, 179, 184-201, 204, 246-258, 263ff, 271-274, 276-281, 287f, 294-297, 304, 308f, 330, 356ff, 370 between pronouns 2, 6, 9-16, 20-25, 77f, 82f, 101-121, 124-155, 161ff, 171-177, 184ff, 189, 198f, 229f, 232, 246ff, 285f, 307f between speakers 2, 6, 77ff, 103-147, 149f, 163, 164f, 171-174, 180-185, 189-192, 204

403

in the speech of individuals 2, 6, 78f, 103, 108-121, 124-131, 135-138, 140, 144-147, 149-153, 161ff, 171175, 180, 184-193, 198-200, 204, 258 categorical predication 266ff, 271 clitic pronoun 1, 22, 65-67ff, 69f, 378 cloze test 87-96 complement of preposition (P, P comp) 3f, 79-82, 87-95, 97ff, 102-130, 133, 135, 144-147, 181-185, 189-199, 252f, 256ff, 262, 298, 301, 303, 306, 313, 318f, 333-336, 339, 353f, 369 complement of verb 250f, 257f, 258-263, 271f, 306, 313, 333f complexity 75f, 81, 124, 156-161, 163167, 169 conjoined pronouns see coordinate constraint 3-7, 57-64, 148-203, 248 overlapping probability distributions 184ff, 198 strict dominance 180-186, 194, 200 ties 184f constraint interaction 2f, 6f, 60-64, 79f, 82, 84f, 124, 147f, 163, 175ff, 197-201, 203ff, 232, 248, 254, 261f, 271, 275ff, 284-287, 300ff, 306-309, 318f, 328331, 339ff, 348ff, 367ff constraint ranking 180-187, 198 constraint violations 3, 57, 176f, 180-184, 186-189, 191-196, 199 constraint weighting 6, 80, 84, 124, 163, 186-200, 300ff, 306f, 318, 328, 331, 363 coordinate (ConjP) 1-7, 47, 62ff, 66, 68, 73ff, 78-81, 88-91, 100-124, 127f, 135, 144-155, 159-177, 179, 181200, 214, 218, 220f, 225-228, 232, 248, 251, 257, 262f, 272, 281ff, 285ff, 294, 309, 344, 367, 379f conjunct order 80f, 90f, 96ff, 103, 109ff, 114, 158, 164-167 conjunct position 80, 87-91, 98f, 101124, 127f, 135, 145, 147, 149-155, 158, 161-171, 175ff, 181-185, 189199, 225f, 228, 232, 248 conjunction mates 68, 80f, 87-91, 9599, 101-113, 158, 161-167

404

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

copular construction 43, 63f, 84f Copy Spell Out 299f cumulativity 186-200 D D and DP 33, 50-54, 59, 70, 73f, 76f, 130133, 145, 155, 202, 255f, 268, 288-308, 311f, 337-342, 346-350, 352, 364-367 data elicitation 87-100 dative case 8-20, 66f, 77, 157, 250, 353, 371ff see also objective case, oblique case Default (Def-)Agreement 62ff, 245 Default (Def-)Case 4, 7, 57, 59-64, 79f, 82, 84f, 123f, 130f, 139, 144, 147-150, 175-178, 181-185, 191f, 195, 201-205, 211f, 216ff, 222, 224, 227f, 231f, 236f, 241, 254-257, 260, 262, 271, 274-277, 280f, 284f, 287, 289ff, 293, 298ff, 306, 314, 318f, 323-330, 339f, 343, 347351, 361ff, 368f deficient pronoun 1, 65f, 69, 74-77, 379 definite noun phrase 50, 52, 297f, 366 deictic there construction 47f, 63f demonstrative 95, 125, 130-133, 293, 352f determiner 10f, 14, 50f, 268, 288, 373 Discourse Inference 142, 231f domain 154, 156f double-object construction 29f, 41f E Early Modern English (EModE) 20-24, 62, 86, 219, 224, 233f, 241, 243ff, 249, 258ff, 262, 306, 310f, 315-318, 335, 354 ellipsis 85ff, 144, 223-231, 282ff, 286, 328, 345 see also VP ellipsis empty category/head 84, 87, 204, 208, 210, 223, 230, 258, 270, 326-329, 337f, 345f empty operator 85, 141, 208f, 212, 217, 306 equative construction (as, like) 85, 143, 153, 282 exception structure (but, except, save) 67, 70f, 143, 153, 279, 280-283 experiencer construction 19f, 30f, 371f

F feature-copying 356 finite clause 4, 22, 24f, 40f, 48f, 60-64, 79-82, 85ff, 97, 113, 121-124, 145f, 178, 181, 205f, 210f, 223, 229-232, 273ff, 280, 282-287, 291-294, 296, 308, 316ff, 320-324, 326ff, 334, 337, 339ff, 344, 347-350, 357-362, 367f, 375-379 focus 46, 68, 75, 82-85, 136-142, 205209, 234, 242-248, 299, 309 contrastive focus 273, 275, 283 focus particle 205, 234, 286f focus preposition 55, 67, 70f, 143, 153, 204, 282, 309 new information focus see rheme Focus Phrase (FocP) 208 free relative 236-239, 331-352, 363f, 367f argument relative 331-352 left-dislocated free relative 331ff, 335ff, 342f pro-head approach 237-340, 346f, 350ff, 363f wh-head approach 239ff, 346-352, 367f French 66ff functional category/head 4, 31, 33, 38-55, 58ff, 66, 69f, 73f, 76f, 84, 122, 128, 202, 204, 208, 210, 215f, 221-224, 231, 237, 255, 260, 266-269, 271, 292, 311, 330, 346 G gapping 70f, 204, 231, 272-279 genitive case (GEN) 9-12, 15-19, 28, 35ff, 50-53, 58-61, 131, 201f, 248-252, 255258, 294ff, 302, 311f, 327f, 351-354, 367 see also oblique case German 28, 30, 216 gerund see V-ing construction government 39, 141, 223, 316, 366 gracile pronoun forms 5f, 139, 152-156, 158-163, 167-171, 175, 213, 218, 228, 286, 298, 331, 362, 382f Gradual Learning Algorithm (GLA) 186 grammatical relation 17-20, 308

SUBJECT INDEX

grammatical virus 1, 87, 248, 362 grammaticality judgment 359 H hypercorrection 1, 96 I identificational be 43, 68, 70, 72, 85, 139142, 204, 233-248, 270, 321-324, 374 imperative 22f, 243, 375 indefinite noun phrase 50, 52, 297f, 366 independent pronoun 69, 70f, 142, 204, 220, 228-233 individual-level predicate 266ff Indo-European languages 254 Information Structure 48, 132, 167, 227, 257 inherent case see lexical case interrogative see question Invariant Strong Form (INV) constraints 6ff, 148, 171-177, 179, 181-185, 190199, 201, 203f, 324, 331, 367, 382 invariant strong pronoun forms 2, 6-9, 6669, 107, 113f, 121, 147-150, 171-175, 203f, 212, 226, 240, 247, 255f, 262, 271, 277, 280, 284, 287, 291, 293, 299, 302, 307, 309, 314, 319, 322, 324f, 328, 330, 340, 342, 369, 381ff it BE sentence 68, 72, 139f, 153, 203f, 233, 242-248 it-cleft 1f, 68, 72, 75, 78f, 82-85, 92, 100, 133-142, 144f, 153, 233f, 299, 309 J judgment fatigue 97ff L Latin 250, 281 left-dislocation 67, 72, 153, 204, 214-218, 222, 309, 331ff lexical case 8, 17-20, 30f, 39, 55f, 370-374 licensing of structural arguments 7f, 2738, 61, 370-375, 381 locative inversion 47f, 63f Logical Form (LF) 84ff, 141, 279, 328 M Mad Magazine sentence 263, 269, 271,

405

276 Middle English (ME) 7f, 14-22, 24, 56, 61, 140f, 221, 225, 228, 242ff, 249f, 294, 308, 310f, 331f, 336f, 353f, 369ff, 374f Modern English (ModE) 1-9, 21, 24-64, 69-200, 203-315, 318-352, 355-371, 373ff, 377, 381ff modified pronoun 47, 62ff, 66, 69, 73ff, 130f, 153, 155, 205, 207, 212-215, 217, 221, 225, 228, 262f, 272, 287-309, 380 morphological case 1-17, 20-27, 57-70, 77-87, 101-185, 189-383 historical changes 6-17, 20-25, 140f, 203f, 219f, 224, 229, 242-247, 250, 262, 310f, 370, 373, 380-383 morphological licensing 7f, 61, 370-374, 381 movement 356, 364-365 ATB-movement 277, 279 head movement 339ff, 352, 359 DP movement 40f, 44ff, 48, 79f, 123, 227, 236f, 241, 257, 261, 268f, 271, 275f, 287, 290f, 293, 295f, 317, 319, 341 347f, 374, 377-380 T-to-C raising 64, 210, 238, 274, 308, 314, 316ff, 323f, 327 verb movement 40f, 44-48, 141, 237, 243, 276f, 279, 346, 376f VP movement 344, 348 multiple choice questions 87-96 N negation 49, 243, 249, 261, 279, 283-286, 376 constituent negation 49 sentential negation 49, 269 nominative case (NOM) 1-7, 8-24, 26, 2833, 36f, 40f, 48f, 53f, 57, 59-65, 67, 77, 79, 81-87, 98f, 102-147, 149-155, 157, 159-165, 167-179, 181-185, 189-207, 209-221, 223-230, 232-265, 270-288, 291-294, 296-308, 310-335, 337, 339342, 344, 347-364, 367ff, 371-375 non-finite clause 38f, 49f, 52, 57, 61, 202, 204, 211f, 220, 254, 258-262, 299, 323f, 357f, 367 nonsentential constituents 220, 230f, 288,

406

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

291 not 279, 283-286 noun (N) and noun phrase 9f, 13f, 17-20, 27f, 32-38, 44-48, 50-53, 60, 131ff, 145, 166, 201, 232, 234, 248, 268, 270, 289f, 295, 297-302, 305, 317, 328, 365f, 373-379 see also pronoun-NP constructions Null Complement Anaphora (NCA) 333, 335, 343-352, 367 numeral and NumP 50-53, 69, 131, 145, 228, 255, 268, 290-295, 297f, 301, 305, 311f, 328, 352, 364ff O object movement 40f, 44ff, 48, 79f, 123, 268f, 275f, 287, 290-291, 293, 319, 341 347f, 374 object of preposition (P, P comp) 3-6, 14, 17f, 26, 32, 38, 53-57, 60f, 79-85, 8795, 97ff, 102-130, 133-136, 140, 142, 145-147, 149f, 153, 164f, 181-185, 189-199, 205ff, 209, 211f, 214f, 218f, 221, 225, 228-231, 288-294, 303-307, 310-315, 318-326, 328-332, 334-346, 350-351, 353-357, 360-364, 368-369, 374, 377, 379f object of verb (O, object) 3-6, 17-24, 26, 29ff, 37, 40-47, 53-57, 60f, 79-85, 8793, 95, 97f, 102-130, 133-136, 140, 142, 145ff, 149f, 175ff, 181-185, 189199, 205-209, 211-217, 219, 221, 225, 228-231, 248f, 254ff, 268ff, 272, 274280, 286-291, 293, 298f, 303-307, 310315, 318-326, 328, 334-338, 341-348, 353, 355ff, 360-364, 368f, 374, 377ff objective case (OBJ) 1-7, 20-24, 26, 2832, 37f, 40f, 43, 49f, 53f, 58-65, 77, 79, 81-85, 87, 101-147, 149-155, 159-165, 167-179, 181-185, 189-204, 206, 209, 211f, 214, 216, 218, 220, 224-243, 245-265, 269-281, 283-308, 310-322, 324-342, 354-359, 361-365, 367ff, 375 oblique case (OBL) 9, 157, 263 Old English (OE) 6-20, 28, 30f, 55f, 142, 242f, 250, 281f, 330ff, 336f, 352f, 363, 371-380 only 205, 234, 286f

Optimality Theory (OT) 179-187, 193f, 198, 200, 340 P parenthetical 358-361 Particle Phrase (PrtP) 286f passive 17-20, 29f, 35f, 42f, 98, 249, 269 past-participle construction 297ff perfective have 249, 253ff, 260, 295 personal pronoun 1-26, 44-48, 57-156, 158-309, 315-318, 370, 374-383 1sg 11-14, 83, 95f, 98f, 102-124, 133147, 149-155, 158-163, 168-176, 181-186, 188-193, 195ff, 199f, 203, 214, 229f, 232f, 246ff, 285f, 288, 304, 307f, 381ff 1pl 11-14, 22, 62, 81f, 102-147, 149155, 158-163, 168-175, 177, 181186, 188-192, 194, 197-200, 214f, 228, 246ff, 285f, 288, 293, 296f, 299, 307, 315-318, 381f 2ps 11-14, 19-25, 175, 214f, 217, 285f, 288 3sgM 13, 15f, 102-124, 133-147, 149155, 158-163, 168-175, 177, 181186, 188-192, 194, 197-200, 206f, 213ff, 246ff, 285f, 297ff, 307, 381f 3sgF 11, 13, 15f, 24f, 102-124, 133147, 149-155, 158-163, 167-175, 177, 181-186, 188-194, 197-200, 206f, 213f, 217f, 229f, 232f, 246ff, 285f, 381f 3pl 3, 5f, 10f, 13, 15f, 24f, 81f, 95, 102-147, 149-155, 158-175, 177, 181-186, 188-192, 194, 197-200, 214, 226ff, 246ff, 285f, 288, 293, 316ff, 381f phi (φ-)features (number, person, gender) 14, 39ff, 50f, 82f, 229f, 234, 239f, 244, 338, 346 Phonological Form (PF) 2, 78 PF-deletion 223, 230, 275f, 279, 327ff, 345 PF-merger 315, 327, 360 phonological properties of pronoun forms 23f, 66, 74f, 142, 147, 158-163, 167171, 175, 285, 382 Positional (Pos-)Agreement 46, 62ff, 70,

SUBJECT INDEX

76f, 242, 244ff Positional (Pos-)Case 4, 6ff, 27, 57-64, 70, 79f, 82-87, 107, 122ff, 128, 130f, 139, 144, 147-150, 175-178, 181-185, 191-199, 201ff, 210ff, 216, 223-227, 230-233, 236-242, 254-257, 260ff, 269-293, 296, 298, 300ff, 305-309, 311-320, 322-330, 337-344, 347-352, 359-363, 367-370, 373ff, 380ff Positional Default Case see Default Case positional licensing 7f, 61, 308, 370-375, 381 Poss-ing gerund 37, 52f, 60f, 255f, 258, 296 possessive 26, 68, 70, 166, 219, 221, 225, 248, 268, 304f, 327f, 351-354, 367 Pred and PredP 266-269 preposition (P) and PP 3f, 17ff, 27f, 3033, 35, 37f, 45, 51, 53-58, 131, 149, 153, 155, 201, 226ff, 258, 267f, 281f, 284, 291, 294-299, 319, 325f, 328-345, 350f, 353-357, 368f, 383 ‘adjectival’ PP 297f focus preposition 55, 67, 70f, 143, 153, 204, 282, 309 ‘restrictive’ PP 297f pied-piping 303, 313, 321f, 329-333, 335, 350f, 355ff, 362, 382f pro-P 338f, 350f, 355ff stranding 55f, 206, 209, 211, 329, 331f, 335f, 341f, 360ff, 382f swiping 329ff prescriptive norm 94, 314, 382f Present-Day English 1-6, 24-64, 69-200, 203, 219f, 224, 227ff, 234f, 238f, 242, 250, 254f, 258-262, 271, 275-280, 284, 299, 307, 312, 318f, 322, 324, 326, 330, 355f, 359, 363, 367ff, 375, 381ff pro 338ff, 346f, 350ff probability of occurrence 184-189 processing 158 pronoun-NP construction 2, 62ff, 74f, 78f, 81f, 88f, 91f, 100, 124-133, 145, 149, 153, 221, 225, 228, 288, 291, 293-296, 301 property predicate see individual-level predicate prosodic properties of pronouns 66, 74ff,

407

220, 380 pseudo-gapping 272-276 pseudo-passive 55f Q qualia structure 33-36, 311 quantifier and QP 282, 294ff question 55f, 72f, 243, 310-331, 355, 376 echo question 318ff embedded question 209, 211, 236-240, 247, 320-327, 359, 361ff identificational question 235-240, 246f matrix question 55f, 238, 247, 312315, 320-323, 325, 368f metalinguistic question 235 multiple wh-question 236, 239f, 313, 320f sluiced question 153, 325-331, 361f, 367, 369, 382f questionnaire items 2-6, 78-83, 86-99, 109-113, 164, 166f quirky case see lexical case R reconstruction 86, 223, 274, 279f, 328 reflexive pronoun 12f, 23, 82-85, 88, 92, 131, 133, 299f relative clause 69, 72, 153, 205, 207ff, 212-215, 217, 221, 225, 302-308, 310f, 331-366, 380 contact relative 303f complex headed relative 357-361, 363366, 369 free relative 236-239, 331-352 headed relative 310f, 352-366, 368f non-restrictive relative 214f, 301f, 354f, 357ff, 364ff restrictive relative 131, 140f, 207ff, 213ff, 293, 301ff, 305, 311, 354358, 364ff Relative Positional Coding (RPC) 3, 5-8, 139, 148, 150-171, 175ff, 179, 182f, 201f, 204, 213, 218, 228, 281, 285ff, 298, 308f, 330f, 362, 367, 375, 382f RPC 1 3, 5, 154f, 163, 176f, 179, 181185, 191-200, 202 RPC 2 3, 5, 154, 162f, 176f, 179, 181-

408

THE DISTRIBUTION OF PRONOUN CASE FORMS IN ENGLISH

185, 191-200, 202, 204 rheme/new information focus 48, 167, 227, 230f, 241, 374 right-dislocation 67, 72f, 204, 219-228, 283, 287, 292 robust pronoun forms 5f, 139, 152-156, 158-163, 167-171, 175, 248, 282, 285ff, 298, 330, 362, 382f Romance languages 1, 66, 131, 133, 383 S self-reflexive 131, 299f Semantic Form (SF) 26-38, 64, 224, 230, 274, 284, 328, 346 sluicing 153, 325-331, 362, 367, 369, 382f small clause 31ff, 37f, 57, 63f, 69ff, 125, 153, 204, 206, 211ff, 215, 220f, 226f, 262-272, 291ff, 299, 309 adjectival small clause 32, 264, 267 categorical small clause 71, 266ff, 271 identificational small clause 270f, 278 nominal small clause 32f, 263f, 268 prepositional small clause 32, 265, 267f thetic small clause 71, 266-271 verbal small clause 31, 265, 268ff, 276 Specificity Principle 29, 43 stage-level predicate 266ff stigmatised forms 125, 293 stress 21f, 47f, 66, 74f, 143, 159, 170, 220, 380 stripping see bare argument ellipsis strong pronoun 1-8, 25, 47f, 62-156, 158177, 179, 181-186, 188-315, 318-370, 375, 379-383 structural case 1, 8, 17-20, 22, 26-64 structural features 27-33, 36, 43, 53f, 57, 371f structural linking 3, 27-39, 52ff, 57f, 371f subject 3-6, 22-25, 29-33, 57, 205, 208f, 214-217, 219-221, 223, 225f, 234ff, 252f, 256f, 262, 280-287, 297ff, 311, 331f, 334-350, 352-366, 368f, 377 after a fronted auxiliary 22, 64, 262, 275, 315-318 postverbal subject 19, 22f, 47f, 303, 306, 343f, 348, 376

subject of a finite clause (S) 4, 17-20, 22, 24, 26, 40f, 46, 48f, 60-64, 66f, 70, 79-99, 103-138, 140, 142, 145147, 149f, 175ff, 181-185, 189-199, 205f, 210f, 214f, 229-232, 258f, 262, 291-294, 296-299, 302-306, 312-315, 320-324, 326, 328 subject of a gapped clause 272ff, 276ff subject of a non-finite clause 37ff, 49f, 61, 69, 211f, 220, 258-262, 291ff, 309 subject of a small clause 32f, 37f, 69, 125, 153, 206, 211f, 215, 220f, 226f, 262-272, 278, 291ff, 309, 343f, 349f subject of a stripped clause 279f subject of a V-ing construction 37, 52f, 61, 153, 248-258, 309 survey results 95f, 98f, 102-147, 149-153, 161-167, 171-175, 180f, 185f, 189f, 192-200 individual speaker patterns 113-121, 126-130, 133, 135-137, 140, 144147, 151f, 161f. 165ff, 171-175, 181-186, 189-200 overall trends 95f, 98f, 102-113, 125f, 133f, 142f, 149-153, 161-167, 171, 175 swiping 329ff T T(ense) and TP 4, 31, 33, 40f, 46-50, 52ff, 59, 62ff, 70-73, 76f, 84, 86f, 122f, 142, 202, 210f, 216, 221-226, 230, 236-243, 253-257, 260f, 274ff, 279, 284f, 287, 291, 296, 300f, 305-309, 316ff, 323f, 326ff, 341, 346ff, 350, 359-362, 367f, 374f, 377 T-to-C raising 64, 210, 238, 274, 308, 314, 316ff, 323f, 327 Telicity Phrase (TelP) 45f, 320 than comparative 1f, 68, 70f, 75, 78f, 85ff, 100, 203, 282 there BE sentences 63f theta (θ-)binding 33, 36, 38, 51, 57 theta (θ-)identification 28, 31ff, 37f theta (θ-)position 3, 41-47, 51, 58, 66, 76, 223, 230, 271, 277, 306, 312, 317

SUBJECT INDEX

theta (θ-)role assignment/θ-marking 39, 41, 43f, 51, 54f, 85, 223, 226, 281, 306, 328 theta (θ-)structure (TS) 3, 27-38, 57 thetic predication 266-271 to-infinitive 31, 37f, 49f, 52, 61, 70, 204, 206, 211f, 258-262, 269, 291ff, 301, 346, 360 topic 48, 167, 227, 374 Topic Phrase (TopP) 208 topicalisation 153, 204-213, 215f, 222, 267, 304f, 309, 334, 372, 375f V v and vP 4, 40-50, 53f, 59, 62f, 70, 77, 79f, 85, 123, 139, 182, 192, 196, 202, 204, 223, 226f, 230f, 233, 236f, 261, 268-272, 276, 279, 281, 287, 290f, 296, 301, 305f, 308, 319f, 341, 347ff, 361, 363, 367, 374f, 377 V-ing construction 37, 52f, 70, 153, 204, 248-258, 294ff, 302 absolutive V-ing 249ff, 254f, 257, 263 Acc-ing gerund 52, 255-258 gapped gerund 273, 277 Poss-ing gerund 37, 52f, 60f, 255f, 258 variable rule analysis 186ff verb (V) and verb classes 3f, 17-20, 22f, 27-32, 35-38, 41ff, 47, 61-64, 201, 267-270, 333, 346 agentive/causative verb 41, 268f, 346, 374 ditransitive 29f, 41f

409

experiencer verb 19f, 30f, 371f expletive verb 84f identificational be 43, 68, 70, 72, 85, 139-142, 204, 233-248, 270, 321324, 374 intransitive 23, 47 transitive 17f, 40-47, 346 unaccusative 41f, 47, 306, 374 unergative 41f, 47 verb movement 40f, 44-48, 141, 237, 243, 276f, 279, 346, 376f verb (V-)particle construction 44-48, 123, 158, 268f, 287, 290f, 293, 298, 305, 319f, 341, 347f, 367 VP 40-47, 54-58, 123, 139ff, 181f, 223, 230f, 268ff, 275ff, 298, 305f, 320, 347f, 360 VP ellipsis (VPE) 49, 236ff, 333, 335, 343-351, 367 W weak pronoun 1, 6f, 24ff, 44-47, 62, 64ff, 69f, 74-77, 175, 275f, 287, 290, 367, 370, 374-381, 383 licensing 1, 25, 46f, 64, 66, 69f, 76f, 275f, 376-379 wh-pronoun 1f, 6f, 10f, 72f, 141f, 174, 203f, 310-369, 381ff complex 207, 333ff, 338-342 simplex 11, 14, 133f, 136ff, 140ff, 152-155, 159ff, 168-171, 207-213, 217, 210, 235-240, 246f, 302-307, 310-315, 318-333, 335-369, 382f

In the series Linguistik Aktuell/Linguistics Today the following titles have been published thus far or are scheduled for publication: 87 JULIEN, Marit: Nominal Phrases from a Scandinavian Perspective. Expected October 2005 86 COSTA, João and Maria Christina FIGUEIREDO SILVA (eds.): Studies on Agreement. Expected December 2005 85 MIKKELSEN, Line: Copular Clauses. Specification, predication and equation. viii, 212 pp. Expected September 2005 84 PAFEL, Jürgen: Quantifier Scope in German. Expected December 2005 83 SCHWEIKERT, Walter: The Order of Prepositional Phrases in the Structure of the Clause. xii, 338 pp. Expected September 2005 82 QUINN, Heidi: The Distribution of Pronoun Case Forms in English. 2005. xii, 409 pp. 81 FUSS, Eric: The Rise of Agreement. A formal approach to the syntax and grammaticalization of verbal inflection. xii, 323 pp. + index. Expected October 2005 80 BURKHARDT, Petra: The Syntax–Discourse Interface. Representing and interpreting dependency. viii, 256 pp. + index. Expected October 2005 79 SCHMID, Tanja: Infinitival Syntax. Infinitivus Pro Participio as a repair strategy. 2005. xiv, 251 pp. 78 DIKKEN, Marcel den and Christina M. TORTORA (eds.): The Function of Function Words and Functional Categories. viii, 285 pp. + index. Expected October 2005 77 ÖZTÜRK, Balkız: Case, Referentiality and Phrase Structure. 2005. x, 268 pp. 76 STAVROU, Melita and Arhonto TERZI (eds.): Advances in Greek Generative Syntax. In honor of Dimitra Theophanopoulou-Kontou. 2005. viii, 366 pp. 75 DI SCIULLO, Anna Maria (ed.): UG and External Systems. Language, brain and computation. 2005. xviii, 398 pp. 74 HEGGIE, Lorie and Francisco ORDÓÑEZ (eds.): Clitic and Affix Combinations. Theoretical perspectives. 2005. viii, 390 pp. 73 CARNIE, Andrew, Heidi HARLEY and Sheila Ann DOOLEY (eds.): Verb First. On the syntax of verb-initial languages. 2005. xiv, 434 pp. 72 FUSS, Eric and Carola TRIPS (eds.): Diachronic Clues to Synchronic Grammar. 2004. viii, 228 pp. 71 GELDEREN, Elly van: Grammaticalization as Economy. 2004. xvi, 320 pp. 70 AUSTIN, Jennifer R., Stefan ENGELBERG and Gisa RAUH (eds.): Adverbials. The interplay between meaning, context, and syntactic structure. 2004. x, 346 pp. 69 KISS, Katalin É. and Henk van RIEMSDIJK (eds.): Verb Clusters. A study of Hungarian, German and Dutch. 2004. vi, 514 pp. 68 BREUL, Carsten: Focus Structure in Generative Grammar. An integrated syntactic, semantic and intonational approach. 2004. x, 432 pp. 67 MIŠESKA TOMIĆ, Olga (ed.): Balkan Syntax and Semantics. 2004. xvi, 499 pp. 66 GROHMANN, Kleanthes K.: Prolific Domains. On the Anti-Locality of movement dependencies. 2003. xvi, 372 pp. 65 MANNINEN, Satu Helena: Small Phrase Layers. A study of Finnish Manner Adverbials. 2003. xii, 275 pp. 64 BOECKX, Cedric and Kleanthes K. GROHMANN (eds.): Multiple Wh-Fronting. 2003. x, 292 pp. 63 BOECKX, Cedric: Islands and Chains. Resumption as stranding. 2003. xii, 224 pp. 62 CARNIE, Andrew, Heidi HARLEY and MaryAnn WILLIE (eds.): Formal Approaches to Function in Grammar. In honor of Eloise Jelinek. 2003. xii, 378 pp. 61 SCHWABE, Kerstin and Susanne WINKLER (eds.): The Interfaces. Deriving and interpreting omitted structures. 2003. vi, 403 pp. 60 TRIPS, Carola: From OV to VO in Early Middle English. 2002. xiv, 359 pp. 59 DEHÉ, Nicole: Particle Verbs in English. Syntax, information structure and intonation. 2002. xii, 305 pp. 58 DI SCIULLO, Anna Maria (ed.): Asymmetry in Grammar. Volume 2: Morphology, phonology, acquisition. 2003. vi, 309 pp.

57 DI SCIULLO, Anna Maria (ed.): Asymmetry in Grammar. Volume 1: Syntax and semantics. 2003. vi, 405 pp. 56 COENE, Martine and Yves D’HULST (eds.): From NP to DP. Volume 2: The expression of possession in noun phrases. 2003. x, 295 pp. 55 COENE, Martine and Yves D’HULST (eds.): From NP to DP. Volume 1: The syntax and semantics of noun phrases. 2003. vi, 362 pp. 54 BAPTISTA, Marlyse: The Syntax of Cape Verdean Creole. The Sotavento varieties. 2003. xxii, 294 pp. (incl. CD-rom). 53 ZWART, C. Jan-Wouter and Werner ABRAHAM (eds.): Studies in Comparative Germanic Syntax. Proceedings from the 15th Workshop on Comparative Germanic Syntax (Groningen, May 26–27, 2000). 2002. xiv, 407 pp. 52 SIMON, Horst J. and Heike WIESE (eds.): Pronouns – Grammar and Representation. 2002. xii, 294 pp. 51 GERLACH, Birgit: Clitics between Syntax and Lexicon. 2002. xii, 282 pp. 50 STEINBACH, Markus: Middle Voice. A comparative study in the syntax-semantics interface of German. 2002. xii, 340 pp. 49 ALEXIADOU, Artemis (ed.): Theoretical Approaches to Universals. 2002. viii, 319 pp. 48 ALEXIADOU, Artemis, Elena ANAGNOSTOPOULOU, Sjef BARBIERS and Hans-Martin GÄRTNER (eds.): Dimensions of Movement. From features to remnants. 2002. vi, 345 pp. 47 BARBIERS, Sjef, Frits BEUKEMA and Wim van der WURFF (eds.): Modality and its Interaction with the Verbal System. 2002. x, 290 pp. 46 PANAGIOTIDIS, Phoevos: Pronouns, Clitics and Empty Nouns. ‘Pronominality’ and licensing in syntax. 2002. x, 214 pp. 45 ABRAHAM, Werner and C. Jan-Wouter ZWART (eds.): Issues in Formal German(ic) Typology. 2002. xviii, 336 pp. 44 TAYLAN, Eser Erguvanlı (ed.): The Verb in Turkish. 2002. xviii, 267 pp. 43 FEATHERSTON, Sam: Empty Categories in Sentence Processing. 2001. xvi, 279 pp. 42 ALEXIADOU, Artemis: Functional Structure in Nominals. Nominalization and ergativity. 2001. x, 233 pp. 41 ZELLER, Jochen: Particle Verbs and Local Domains. 2001. xii, 325 pp. 40 HOEKSEMA, Jack, Hotze RULLMANN, Víctor SÁNCHEZ-VALENCIA and Ton van der WOUDEN (eds.): Perspectives on Negation and Polarity Items. 2001. xii, 368 pp. 39 GELDEREN, Elly van: A History of English Reflexive Pronouns. Person, Self, and Interpretability. 2000. xiv, 279 pp. 38 MEINUNGER, Andre: Syntactic Aspects of Topic and Comment. 2000. xii, 247 pp. 37 LUTZ, Uli, Gereon MÜLLER and Arnim von STECHOW (eds.): Wh-Scope Marking. 2000. vi, 483 pp. 36 GERLACH, Birgit and Janet GRIJZENHOUT (eds.): Clitics in Phonology, Morphology and Syntax. 2001. xii, 441 pp. 35 HRÓARSDÓTTIR, Thorbjörg: Word Order Change in Icelandic. From OV to VO. 2001. xiv, 385 pp. 34 REULAND, Eric (ed.): Arguments and Case. Explaining Burzio’s Generalization. 2000. xii, 255 pp. 33 PUSKÁS, Genoveva: Word Order in Hungarian. The syntax of Ā-positions. 2000. xvi, 398 pp. 32 ALEXIADOU, Artemis, Paul LAW, Andre MEINUNGER and Chris WILDER (eds.): The Syntax of Relative Clauses. 2000. vi, 397 pp. 31 SVENONIUS, Peter (ed.): The Derivation of VO and OV. 2000. vi, 372 pp. 30 BEUKEMA, Frits and Marcel den DIKKEN (eds.): Clitic Phenomena in European Languages. 2000. x, 324 pp. 29 MIYAMOTO, Tadao: The Light Verb Construction in Japanese. The role of the verbal noun. 2000. xiv, 232 pp. 28 HERMANS, Ben and Marc van OOSTENDORP (eds.): The Derivational Residue in Phonological Optimality Theory. 2000. viii, 322 pp.

A complete list of titles in this series can be found on the publishers website, www.benjamins.com