Current Trends in Linguistic Theory 9788436263220, 8436263227

Introducción a los últimos avances en los principales modelos lingüísticos actuales, es decir, la Gramática Generativa,

1,089 135 10MB

Spanish Pages 316 [314] Year 2012

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Current Trends in Linguistic Theory
 9788436263220, 8436263227

Table of contents :
Current.p1
Current.p2
Current.p3
Current.p4
Current.p5
Current.p6
Current.p7
Current.p8
Current.p9
Current.p10
Current.p11
Current.p12
Current.p13
Current.p14
Current.p15
Current.p16
Current.p17
Current.p18
Current.p19
Current.p20
Current.p21
Current.p22
Current.p23
Current.p24
Current.p25
Current.p26
Current.p27
Current.p28
Current.p29
Current.p30
Current.p31
Current.p32
Current.p33
Current.p34
Current.p35
Current.p36
Current.p37
Current.p38
Current.p39
Current.p40
Current.p41
Current.p42
Current.p43
Current.p44
Current.p45
Current.p46
Current.p47
Current.p48
Current.p49
Current.p50
Current.p51
Current.p52
Current.p53
Current.p54
Current.p55
Current.p56
Current.p57
Current.p58
Current.p59
Current.p60
Current.p61
Current.p62
Current.p63
Current.p64
Current.p65
Current.p66
Current.p67
Current.p68
Current.p69
Current.p70
Current.p71
Current.p72
Current.p73
Current.p74
Current.p75
Current.p76
Current.p77
Current.p78
Current.p79
Current.p80
Current.p81
Current.p82
Current.p83
Current.p84
Current.p85
Current.p86
Current.p87
Current.p88
Current.p89
Current.p90
Current.p91
Current.p92
Current.p93
Current.p94
Current.p95
Current.p96
Current.p97
Current.p98
Current.p99
Current.p100
Current.p101
Current.p102
Current.p103
Current.p104
Current.p105
Current.p106
Current.p107
Current.p108
Current.p109
Current.p110
Current.p111
Current.p112
Current.p113
Current.p114
Current.p115
Current.p116
Current.p117
Current.p118
Current.p119
Current.p120
Current.p121
Current.p122
Current.p123
Current.p124
Current.p125
Current.p126
Current.p127
Current.p128
Current.p129
Current.p130
Current.p131
Current.p132
Current.p133
Current.p134
Current.p135
Current.p136
Current.p137
Current.p138
Current.p139
Current.p140
Current.p141
Current.p142
Current.p143
Current.p144
Current.p145
Current.p146
Current.p147
Current.p148
Current.p149
Current.p150
Current.p151
Current.p152
Current.p153
Current.p154
Current.p155
Current.p156
Current.p157
Current.p158
Current.p159
Current.p160
Current.p161
Current.p162
Current.p163
Current.p164
Current.p165
Current.p166
Current.p167
Current.p168
Current.p169
Current.p170
Current.p171
Current.p172
Current.p173
Current.p174
Current.p175
Current.p176
Current.p177
Current.p178
Current.p179
Current.p180
Current.p181
Current.p182
Current.p183
Current.p184
Current.p185
Current.p186
Current.p187
Current.p188
Current.p189
Current.p190
Current.p191
Current.p192
Current.p193
Current.p194
Current.p195
Current.p196
Current.p197
Current.p198
Current.p199
Current.p200
Current.p201
Current.p202
Current.p203
Current.p204
Current.p205
Current.p206
Current.p207
Current.p208
Current.p209
Current.p210
Current.p211
Current.p212
Current.p213
Current.p214
Current.p215
Current.p216
Current.p217
Current.p218
Current.p219
Current.p220
Current.p221
Current.p222
Current.p223
Current.p224
Current.p225
Current.p226
Current.p227
Current.p228
Current.p229
Current.p230
Current.p231
Current.p232
Current.p233
Current.p234
Current.p235
Current.p236
Current.p237
Current.p238
Current.p239
Current.p240
Current.p241
Current.p242
Current.p243
Current.p244
Current.p245
Current.p246
Current.p247
Current.p248
Current.p249
Current.p250
Current.p251
Current.p252
Current.p253
Current.p254
Current.p255
Current.p256
Current.p257
Current.p258
Current.p259
Current.p260
Current.p261
Current.p262
Current.p263
Current.p264
Current.p265
Current.p266
Current.p267
Current.p268
Current.p269
Current.p270
Current.p271
Current.p272
Current.p273
Current.p274
Current.p275
Current.p276
Current.p277
Current.p278
Current.p279
Current.p280
Current.p281
Current.p282
Current.p283
Current.p284
Current.p285
Current.p286
Current.p287
Current.p288
Current.p289
Current.p290
Current.p291
Current.p292
Current.p293
Current.p294
Current.p295
Current.p296
Current.p297
Current.p298
Current.p299
Current.p300
Current.p301
Current.p302
Current.p303
Current.p304
Current.p305
Current.p306
Current.p307
Current.p308
Current.p309
Current.p310
Current.p311
Current.p312
Current.p313
Current.p314
Current.p315
Current.p316
Current.p317
Current.p318
Current.p319
Current.p320

Citation preview

46506UD01A01

46506UD01A01

46506UD01A01

UNIVERSIDAD NACIONAL DE EDUCACIÓN A DISTANCIA

Unidad Didáctica

Unidades Didácticas Cuadernos de la UNED Aula Abierta Estudios de la UNED Actas y Congresos Estudios de Educación a Distancia Educación Permanente Varia Herramientas

Current Trends in Linguistic Theory es una introducción a los últimos avances en los principales modelos lingüísticos actuales, es decir, la Gramática Generativa, la Gramática del Papel y la Referencia, la Lingüística Sistémico-Funcional y la Gramática Cognitiva. Si bien este libro está diseñado para los alumnos de la asignatura Modelos Teóricos Descriptivos de la Lengua Inglesa de la Licenciatura de Filología Inglesa de la UNED, también puede ser de gran utilidad para los estudiantes de cualquier universidad presencial, tanto para aquellos que se encuentren cursando la licenciatura como para los que deseen emprender su labor investigadora en alguno de los modelos aquí tratados. El libro está dividido en cuatro capítulos, cada uno de ellos correspondiente a uno de los modelos teóricos descritos. La progresión de ideas en cada capítulo es de lo introductorio a lo más complejo y específico, de tal forma que los estudiantes puedan familiarizarse con los principales conceptos de cada modelo. Todos los capítulos incluyen no sólo un resumen de los principales postulados teóricos sino también una serie de ejercicios para poner en práctica los conocimientos adquiridos, así como algunas revistas científicas y páginas web de interés para aquellos que deseen adentrarse más en los últimos avances en los modelos teóricos recogidos en esta obra. Los principales objetivos de Current Trends in Linguistic Theory son ayudar a los estudiantes a entender algunas de las contribuciones más relevantes en Lingüística en la actualidad, así como su impacto en los estudios del lenguaje y animar a los estudiantes a explorar el fascinante mundo de la Lingüística.

Filología Inglesa

CURRENT TRENDS IN LINGUISTIC THEORY

Colecciones de la UNED

Ricardo Mairal Usón M.ª Ángeles Escobar Álvarez M.ª Sandra Peña Cervel Eva Samaniego Fernández

U.D.

Ricardo Mairal es doctor por la Universidad de Zaragoza y catedrático de Lengua y Lingüística Inglesa en el Departamento de Filologías Extranjeras de la UNED. Su actividad docente y su investigación se han centrado en las asignaturas de gramática inglesa, semántica, lexicología y sintaxis. Desde hace varios años participa como investigador principal o como parte integrante en diferentes proyectos de investigación financiados por entidades públicas y es autor de numerosas publicaciones.

CURRENT TRENDS IN LINGUISTIC THEORY Ricardo Mairal Usón M.ª Ángeles Escobar Álvarez M.ª Sandra Peña Cervel Eva Samaniego Fernández

M.ª Ángeles Escobar es doctora en Lingüística por la Universidad de Utrecht (Países Bajos), licenciada en Filología Inglesa y maestra con la especialidad de didáctica de las lenguas extranjeras por la Universidad Complutense de Madrid. Ha recibido numerosas becas de investigación en el extranjero para hacer estudios de lingüística teórica y de adquisición del lenguaje. Imparte asignaturas en la licenciatura y doctorado de Filología Inglesa y en la diplomatura de Turismo de la UNED. M.ª Sandra Peña Cervel se doctoró en Filología Inglesa en la Universidad de La Rioja. Imparte docencia en la licenciatura y doctorado de Filología Inglesa de la UNED. Su investigación, centrada en la organización del conocimiento en forma de modelos cognitivos idealizados, se ha plasmado en un nutrido número de publicaciones y en una participación activa en congresos científicos nacionales e internacionales. Es parte integrante de varios proyectos de investigación financiados por entidades públicas. Eva Samaniego Fernández es licenciada y doctora en Filología Inglesa, Máster en Traducción e Interpretación, traductora titulada por el Institute of Linguists e Intérprete Jurado de inglés. Ha formado parte de varios proyectos de investigación financiados por entidades públicas. Ha participado como ponente en numerosos congresos nacionales e internacionales sobre lingüística y sobre traducción. Actualmente imparte docencia en la UNED.

UNIVERSIDAD NACIONAL DE EDUCACIÓN A DISTANCIA

Ricardo Mairal Usón M. Ángeles Escobar Álvarez M.a Sandra Peña Cervel Eva Samaniego Fernández a

CURRENT TRENDS IN LINGUISTIC THEORY

Unidades Didácticas de Modelos Teóricos Descriptivos de la Lengua Inglesa

UNIVERSIDAD NACIONAL DE EDUCACIÓN A DISTANCIA

CURRENT TRENDS IN LINGUISTIC THEORY

Quedan rigurosamente prohibidas, sin la autorización escrita de los titulares del Copyright, bajo las sanciones establecidas en las leyes, la reproducción total o parcial de esta obra por cualquier medio o procedimiento, comprendidos la reprografía y el tratamiento informático, y la distribución de ejemplares de ella mediante alquiler o préstamos públicos.

© UNIVERSIDAD NACIONAL DE EDUCACIÓN A DISTANCIA - Madrid, 2011

www.uned.es/publicaciones © Ricardo Mairal Usón, M.a Ángeles Escobar Álvarez, M.a Sandra Peña Cervel, Eva Samaniego Fernández

ISBN electrónico: 978-84-362-6322-0 Edición digital: octubre de 2011

ÍNDICE

INTRODUCTION.........................................................................................

9

Chapter 1. An overview of Generative Grammar ..............................

13

Chapter 2. An overview of Role and Reference Grammar ................

97

Chapter 3. An overview of Systemic (Functional) Linguistics.......... 177 Chapter 4. An overview of Cognitive Linguistics............................... 229

INTRODUCTION

The present book aims to be a practical introduction to current trends in linguistics. It sets out to give a critical survey of some of the most important approaches within the field. The book is thus designed to provide a solid theoretical survey and, more specifically, to serve as a set coursebook for undergraduate students undertaking «Modelos Teóricos Descriptivos de la Lengua Inglesa». It is essential to remind students that the actual distribution of the chapters for the UNED ‘Pruebas Presenciales’ in January/February, May/June and September, as well as the general instructions and recommendations for the course have to be consulted in the ‘Guía Didáctica’ and also in the ‘Orientaciones Metodológicas’, which must necessarily be read by all students undertaking the above mentioned course. The book aims to enable students to develop their understanding of the issues dealt with, to be familiar with the associated terminology (metalanguage) used within each paradigm, and also to apply the models themselves with the help of practical exercises. In this way, the book attempts to provide a stimulating introduction to a limited range of theoretical approaches to linguistics. Each of the chapters surveys a major paradigm of the discipline. They are designed to be self-standing, so that students with a specific focus can quickly find the descriptions that are of most interest to them. Furthermore, this relative separation between the chapters allows students to study the paradigms independently. However, conceptual links between chapters do exist and the book has been structured so that it can function as a coursebook.

10

CURRENT TRENDS IN LINGUISTIC THEORY

The progression of ideas in each chapter is from the introductory (presenting the main issues within each approach) to the more complex, as the students become more accustomed to the terminology and concepts. Clarity has been a major consideration, so each chapter follows a similar format of introduction, main text (which includes the main topics together with discussion and research points), evaluative summaries of the aspects dealt with, examples, suggested activities, bibliography and/or suggestions for further reading, web pages, journals, etc. Obviously, this textbook has had to be selective. The models covered have been chosen because of their strong impact and influence on linguistics and, in the case of the authors, because they are particularly representative of the approaches in each chapter. Much as it is to be regretted, exclusion of other truly worthy material has been due to space constraints, to the fact that this textbook has been designed and conceived for a specific course with a limited length and also to the focus of the book, which is to give a clear and relatively brief introduction and overview of a number of theoretical approaches to linguistics. All these factors have had a considerable impact on the final choice of theories, the primary works covered, the range of suggested readings and also the extension of the textbook. Suggestions for reading and further reading have been designed to encourage students to go to the primary texts rather than publications on very specific aspects. The bibliography for each chapter also tries to follow up ideas that have been raised in it and to investigate the research that has been or is being carried out. An attempt has also been made to include works that are readily available, as well as useful websites, where up-todate information on each linguistic approach (conferences, publications, organizations, etc.) can be found. We have also included a list of some of the most relevant journals in the field. It is hoped that reading in those areas or theories which are of greatest interest to the individual student will be encouraged. Our hope is thus that this book will help students understand some of the most relevant contributions to linguistic theory in recent times as well as their impact on language studies, and also to encourage students to develop an interest in the fascinating world of linguistics. The actual distribution of the book is as follows: • Chapter 1, «An overview of Generative Grammar», attempts to provide an accessible introduction to generative grammar through a general coverage of the major approaches currently employed by linguists active in the field. The chapter begins with

INTRODUCTION

11

an introductory section dealing with the spirit of the theory and continues with other four sections in its main area of research: syntax. The topics under study are: thematic structure, the distinction between lexical and functional categories and the syntax of clauses and transformations. The final section is designed to introduce Noam Chomsky’s Minimalist Program which is primarily concerned with the most important theoretical developments within the Principles and Parameters approach. • Chapter 2, «An overview of Role and Reference Grammar», introduces the RRG model. First a historical introduction to the model is provided, as well as its relation to other linguistic models; methodological issues which clarify the basic premises of RRG are also included in the first section of this chapter. Section 2 deals with the internal structure of the lexicon component, with particular emphasis on (i) the criteria used to determine verb classes; (ii) the inventory of logical structures; (iii) macrorole assignment. Section 3, concerned with the relational aspects of the theory, presents the basic units of analysis of the Layered Structure of the Clause. Finally, section 4 describes the rudiments of the linking algorithm, or the set of operations, which account for the systematic relationships between the syntactic and the semantic components. • Chapter 3, An overview of Systemic (Functional) Linguistics», deals with the SFL (Systemic Functional Linguistics) model. The systemic-functional approach to language implies that language use is functional, semantic, contextual and semiotic. It places the function of language as fundamental (what language does, and how it does it). It takes the text rather than the sentence as its object, and defines its scope by reference to usage rather than grammaticality. Thus, Section 1 defines the model and introduces its basic tenets. Section 2 deals with the origins of the model as well as its relationship with other linguistic models. The most fundamental concepts are explained in Section 3, which builds basically upon the work of M.A.K. Halliday, including aspects such as the functional semantic approach to language, the social and cognitive perspective, language as a semiotic system and the issues of choice, metafunctions and stratification, language and context (context of culture and context of situation) and the related concepts of field, tenor and mode, and also coherence and cohesion. Finally, a review of the impact of SFL on other areas such as Text Linguistics or Critical Discourse Analysis is included.

12

CURRENT TRENDS IN LINGUISTIC THEORY

• Chapter 4, «An overview of Cognitive Linguistics», deals with the main essentials of Cognitive Linguistics. It starts out from the distinction between Cognitive Linguistics and Cognitive Grammar and focuses on the former on the grounds that it encompasses the latter. The main emphasis is placed on the description of the different Idealized Cognitive Models put forward within this framework. Propositional models, metaphor, metonymy, and image-schemas are studied and some cognitive continua are postulated, especially in connection with metaphor and metonymy. The main ideas underlying this chapter are that language emerges from bodily experience, that linguistic knowledge and encyclopedic real world knowledge should not be differentiated, and that the limits between categories are fuzzy. The authors: Ricardo Mairal Usón M.a Ángeles Escobar Álvarez M.a Sandra Peña Cervel Eva Samaniego Fernández

CHAPTER

1

AN OVERVIEW OF GENERATIVE GRAMMAR M.a Ángeles Escobar Álvarez UNED

INTRODUCTION This first chapter aims to provide a short introduction to Generative Grammar under a recent approach. Generative Grammar may be defined as a linguistic theory that attempts to give response to the native speaker’s ability to produce language. Originally developed by Noam Chomsky, it has many followers who have adopted it as their theoretical framework to provide explanatory adequacy to grammars of different languages. The generative linguist abandons traditional linguistic models that merely describe external linguistic phenomena and adopts hypotheses about how native speakers have knowledge of language. In this sense, their main interest relies on the main characteristics of internal language beyond external linguistic generalisations. In other words, the linguist inquires the internal principles that are responsible for linguistic data of different languages. The generative linguist’s ultimate goal is then to develop a theory of Universal Grammar (UG, in the linguistic theory), on the assumption that a theory of grammar should provide us with the tools needed to describe any natural language. In principle, children can acquire any natural language as their native language. The explanatory hypothesis is that human beings have a genetically endowed language faculty by which they incorporate a set of principles of UG. That is, the language faculty allows children to develop a grammar of any natural language on the basis of suitable linguistic experience of the language. As Radford (1997) puts it: «Experience of a particular language L (examples of words, phrases and sentences in L which the child hears produced by native speakers of L) serves as input to principles of UG which are an inherent part of the child’s language faculty, and UG then provides the child with an algorithm for developing a grammar of L». Next we include an outline of the theoretical issues that will be covered in this chapter.

OUTLINE 1. Methodological underpinnings 2. Competence and language acquisition 2.1. Insufficient external stimuli 2.2. UG and parameters 3. Lexical and functional categories, subcategorization and constituent structure 3.1. Syntactic and morphological evidence 3.2. Lexical categories 3.3. Constituent structure and subcategorization 3.4. Functional categories 3.4.1. English auxiliaries and infinitival «to» 3.4.2. Determiners 3.4.3. Complementizers 4. The Syntax of clauses 4.1. Xbar syntax: a brief historical perspective 4.2. Structural relations 4.3. Xbar theory, subcategorization and reanalysis of sentences 4.4. Case theory 4.5. Binding theory 4.6. Transformational syntax and movement 4.6.1 Negation as a functional head 4.6.2 Question formation 4.6.2.1. Yes/No questions 4.6.2.2. Wh-questions 4.7. A-movement 4.7.1. Passives 4.7.2. Thematic roles 4.7.3. The double object construction 4.7.4. Unaccusatives 4.7.5. Raising 5. Recent modifications under the Minimalist Program 5.1. Movement and checking theory 5.2. Interpretable and uninterpretable features 5.3. Checking other grammatical features 5.4. Bare phrase structure 6. Suggested activities 7. References

1. METHODOLOGICAL UNDERPINNINGS Language is difficult to define, but we may understand it as an intellectual activity that makes use of mental tools to produce speech throughout general structures. Ideally, Language needs to provide answers to: a) particular phenomena; b) learnability; and c) universality. Linguistics is the science of Language where linguists attempt to provide theories of Language: it must be fully explicit and to be explicit, it must be formal - i.e. make use only of theoretical constructs which have definable formal properties. The use of formal apparatus (involving a certain amount of technical terminology) may seem confusing at first to the beginner, but as in any other serious field of enquiry (e.g. molecular biology) no real progress can be made unless we try to construct formal models of the phenomena we are studying. It would clearly be irrational to accept the use of formalism in one field of enquiry (e.g. molecular biology) while rejecting it in another (e.g. linguistics)». (Andrew Radford 1997:.99).

Noam Chomsky’s put forward a particular theory of internal language, as the initial state of human language. The hypothesis of the initial state or Universal Grammar (UG) implies that the study of any particular external language (e.g. Chinese, Italian or English) is an instance of such a grammar. For example, when studying the English language, the generative linguist is more interested in giving an account of general principles of UG apparent in that particular language. For instance when considering data as in (1), the general principle of UG that anaphors corefer with its antecedent is confirmed: (1)

a.

John washes himself

b.

Mary washes herself

20

CURRENT TRENDS IN LINGUISTIC THEORY

Sentences in (1) are grammatical English sentences and are accepted by any native speaker of English. Sentences in (2), however, are ungrammatical or not accepted by native speakers. For convention a star (*) in front of the sentence indicates its ungrammaticality. (2)

a.

*Herself washes Mary

b.

*Himself washes John

When considering the ungrammatical sentences in (2), the linguist corroborates the general claim that any sentence has a complex hierarchical categorical constituent structure. The ungrammaticality of the sentences in (2) shows that subjects need to hierarchical dominate object anaphors. This can be seen as a syntactic constraint straightforwardly explained in structural terms. In addition the linguist, when theorizing, cannot gather the infinite list of all possible sentences in a language. On the other hand, some linguistic evidence may be useful to analyse a particular phenomenon. The linguist needs to provide the data with explanatory adequacy, i.e. to explain why the native speaker accepts some sentences and rejects some others (e.g. 1 versus 2). To further provide support to the hypothesis, independent evidence should be found. The syntactic constraint mentioned before, for example, may imply that anaphors need to be coreferential with its closest nominal antecedent and so predicts the grammaticality contrast illustrated by the minimal pair in (3): (3)

a.

John wishes that Mary looks at herself in the picture

b.

*Mary wishes that John looks at herself in the picture

The ungrammaticality of the sentence (3b) indicates that the anaphor «herself» cannot corefer with their farthest nominal phrase «Mary», independently from the fact that it shares the same morphological features of gender (feminine) and number (+sing) with it. The sentence (3a) acts as the grammatical counterpart with respect to the sentence (3b). The minimal pair in 3 also serves to establish a difference between anaphors, on the one hand, and pronouns, on the other hand, given the fact that pronouns, in contrast with anaphors can corefer with their farthest antecedent, and in no way can they corefer with their closest antecedent, as illustrated by the minimal pair in (4). Recall that the star in front of a sentence means that the sentence is ungrammatical or not accepted by native speakers. (4)

a.

*John wishes that Mary looks at her in the picture (her=Mary)

b.

Mary wishes that John looks at her in the picture (her=Mary)

AN OVERVIEW OF GENERATIVE GRAMMAR

21

In effect, speakers do not accept that the pronoun «her» is coreferential with the nominal phrase «Mary» in (4a), while they accept that the pronoun «her» is coreferential with the farthest nominal phrase «Mary» in (4b), hence its grammaticality. The amount of principles of UG considered with respect to particular linguistic evidence of a language constitute the grammar of that language. Notice, however, that not all principles of UG have to surface in all grammars. In the following section, we study how UG accounts for language diversity through the notion of parameter setting. The linguistic research initiated by Noam Chomsky goes parallel with other cognitive sciences. He claims in favour of UG which produces a unique computational system of the human mind, generating a number of strings of words derived from certain morphological and syntactic structures. This account may also explain language variation. Languages do not differ with respect to the computational system, but with respect to specific morphological properties included in the lexicon of each particular language. The generative grammar abandons the traditional idea that language variation responds to different grammatical systems as the set of rules specifying each particular language. According to this alternative approach, we cannot speak of different systems of rules to explain apparent contradictory data found in the variation of languages. Rather, a set of specific linguistic parameters allowed by UG characterizes the particular grammar of a language. What traditionally has been known as grammatical constructions turns to be taxonomic epiphenomena that respond to particular structures containing properties that respond to the interaction of principles and parameters. In this sense, we may speak of a derivational component equal for all languages in which a language L sets a system of grammar allowed by UG parameterised for that particular language. In this respect, Noam Chomsky points out: I will assume the familiar Extended Standard Theory (EST) framework, understood in the sense of the Principles and Parameters approach. We distinguish the lexicon from the computational system of the language, the syntax in a broad sense (including phonology). Assume that the syntax provides three fundamental levels of representation, each constituting an «interface» of the grammatical system with some other systems of the mind/brain: Deep structure, Phonetic Form (PF), and Logical Form (LF). (Chomsky: 1996:130)

Each lexical unit in the lexicon contains a consistent system of features according to some phonetic and syntactic properties that determine its

22

CURRENT TRENDS IN LINGUISTIC THEORY

sound, meaning and syntactic roles through more general principles parameterised for each language. For example, the English verb «hit» is specified for its vowel quality, for the properties proper of an action verb, as well as for the requirement of taking an object since it is a transitive verb and accordingly forms part of verbal phrase (VP). In this theory, we may only speak of a single internal language and languages differ from each other with respect to features parameterised in the lexicon. The task of language acquisition is reduced to acquisition of those particular features.

SUMMARY At the end of this section, you should have a clear idea of the distinction between Universal Grammar and the grammar of a particular language. You should also have understood the concept of «grammaticality» and how to analyse linguistic phenomena through minimal pairs. Language diversity should also be understood in terms of parameters and universal principles.

NOW YOU ARE READY TO DO EXERCISES 1, 2, 3, 4 AND 5

2. COMPETENCE AND LANGUAGE ACQUISITION According to generative grammar, linguistic competence is the set of rules that conform the speaker’s knowledge of language, thanks to which it is possible to learn and produce language. The English speaker’s competence explains the ability to render grammaticality judgements with respect to an unlimited number of English sentences. It also allows the generation of new strings of language according to English grammar. La competencia (competence) se opone a la actuación (performance), definida ésta por el conjunto de las restricciones que se ejercen sobre la competencia para limitar su uso: la actuación da cuenta de las diversas utilizaciones de la lengua en los actos de habla. Se distingue una competencia universal, constituida por reglas innatas que sustentan las gramáticas de todas las lenguas, y una competencia particular, constituida por las reglas específicas de una lengua, aprendidas merced al entorno lingüístico. (Dubois y otros:1994,competencia, p 119)

AN OVERVIEW OF GENERATIVE GRAMMAR

23

Our ability to speak a language is based partly on the innate principles and parameters available in UG, partly on the triggering experience of exposure to a specific language. On the basis of these components we develop a grammar of one (or more) specific languages: the core grammar of such a language. Schematically we can represent the generative view of language acquisition as follows Triggering UG Core grammar experience (with parameters) Language X « Language X (Haegeman 1994:16)

Turning back to the example «John washes himself», we showed that the anaphor «himself» necessarily refers to the nominal phrase «John». Now we argue that this is due to the fact that there is an internal rule of binding that applies to anaphors. The English child will not have to learn such a rule because it is part of her linguistic competence: what she only needs to learn is that «himself» belongs to the category of anaphors in her language. Furthermore, even if a homogeneous speech community existed, we would not expect its linguistic system to be a «pure case». Rather, all sorts of accidents of history would have contaminated the system, as in the properties of (roughly) Romance versus Germanic origin in the lexicon of English. The proper topic of inquiry, then, should be a theory of the initial state that abstracts from such accidents, no trivial matter. For working purposes (and nothing more than that), we may make a rough and tentative distinction between the core of a language and its periphery, where the core consists of what we tentatively assume to be pure instantiations of UG and the periphery consists of marked exceptions (irregular verbs, etc). Note that the periphery will also exhibit properties of UG, although less transparently. A reasonable approach would be to focus attention on the core system, putting aside phenomena that result from historical accident, dialect mixture, personal idiosyncrasies, and the like. As in any empirical inquiry, theory-internal considerations enter into the effort to pursue this course, and we expect further distinctions to be necessary. (Chomsky, 1996: 20)

2.1. Insufficient external stimuli As we saw, knowledge of language may be seen as an internal ability the speaker has, although he is unaware of it. It is formally represented by UG with parameters for each specific language. The grammar of a language contains all the possibilities to generate unlimited possible sentences providing each derivation with their syntax, semantics and phonetics. The linguist’s task is reduced to formalize the theory of that language.

24

CURRENT TRENDS IN LINGUISTIC THEORY

The linguist can take specific external data to develop such a theory. However this might be insufficient. To offer a feasible characterization of any production of language, the linguist should also study the speaker’s competence, by analysing the way the speaker himself assesses his grammar. For example, English speakers know unconsciously that the sentence in (5) is grammatical because they accept it, as opposed to the ungrammatical sentence in (6): (5)

Pall looks at her because he loves her

(6)

*Pall looks at her because loves her

It is only the task of the linguist to explain the speaker’s grammaticality judgements with respect to the sentences (5) and (6). At first sight, the linguist may postulate that the sentence (6) is ungrammatical because the embedded clause lacks a subject. However this is not a universal rule, because other languages (Spanish) allow embedded sentences without overt subjects, cf. (6) versus (7): (7)

Pall la mira porque la quiere

The Spanish sentence in (7) is indeed grammatical and therefore we find the first difference between a language like English and a language like Spanish. The question is then whether the linguist needs to exclusively look at one particular language to provide a consistent theory of that language. From the generative perspective, the linguist needs to look at both UG and the particular grammar of a language. By doing so, the linguist shows how the principles of UG are fixed for each particular parameter. In the generative work, English has been characterised as part of the obligatory subject parameter, whereas Spanish belongs together with other languages (Italian, Portuguese, and many more) to the null subject parameter. Work in generative linguistics is therefore by definition comparative. Generative linguists often do not focus on individual languages at all: they will use any human language to determine the general principles of UG and the choices it allows. Data from a dialect spoken by only a couple of hundred people are just as important as data from a language spoken by millions of people. Both languages are human languages and are learnt in the same way. (Haegeman, 1992, p.18)

2.2. UG and parameters In addition, the linguist needs to explain how each language is acquired. The null hypothesis is that the child living in an English community will

AN OVERVIEW OF GENERATIVE GRAMMAR

25

be exposed to sufficient data to fix her own obligatory subject parameter. In this way, grammar should offer explanatory adequacy for particular linguistic phenomena rather than being merely descriptive. The problem of language acquisition cannot be solved by assuming pure imitation of stimuli, since the child will never be exposed to all possible sentences of her language. On the other hand, she needs to be exposed to those crucial examples, which will lead her to fix her own language parameter. The linguist’s task is to identify those crucial examples, which actually trigger language acquisition. Let us suppose that the child hears the following sentences: (8)

a.

I like the boy who I see in that picture

b.

I like the boy that I see in that picture

c.

I like the boy I see in that picture

According to the sentences in (8), the child may think that the relative pronoun «who» may freely alternate with «that» and may also be omitted in relative clauses with a human antecedent. However, this hypothesis would be incorrect since she will never hear sentences like the one in (9), because it is ungrammatical and adults never produce ungrammatical sentences: (9)

I like the boy speaks Japanese

The sentence in (9) is ungrammatical because in English the relative pronoun cannot be omitted with subject relative clauses regardless the human antecedent. It is difficult to see how the child acquires this knowledge without assuming that she does not start from scratch but from UG which allows her to make this difference: object relative clauses allow the three options that/who/- whereas subject relative clauses do not. Obviously, the adult does not have to explain this since he may not be aware of it either. It is also tenable to assume that the child will only hear grammatical sentences and therefore will always be exposed to positive evidence. This goes against the idea that children learn from negative feedback or uncontradictory data. Rather, the child will always be exposed to grammatical sentences and from here she will fix her own parameter. We assume that the system described by UG is a real component of the mind/brain, put to use in the complex circumstances of ordinary life. The validity of this assumption is hardly in question. To reject it would be to assume either (1) that non-homogeneous (conflicting) data are required for language acquisition, or (2) that the mind/brain does indeed have the system described by UG, but it is not used in language

26

CURRENT TRENDS IN LINGUISTIC THEORY

acquisition. Neither assumption is remotely plausible. Rejecting them, we accept the approach just outlined as a reasonable approach to the truth about humans, and a likely prerequisite to any serious inquiry into the complex and chaotic phenomenal world. (Chomsky, 1996:19)

Since the child does not learn from conflicting data, the problem of language acquisition is reduced to fixing parameters allowed by UG. Recall Haegeman’s schema above, where UG acts as a filter between language experience and the actual grammar of a particular language. In fact UG was proposed by Chomsky to explain how children acquire language, which is a complex task, at such an early age and with such speed and efficiency. The term «universal» suggests that this theory is used to explain language acquisition in general (i.e., it’s not language specific). It would take a lifetime to learn all the rules and different possible ways words can be used, so making language an inefficient way of communicating, therefore evolutionary would not have stood the test of time. However, language is perhaps one of the things that make us human, and is indeed a universal phenomenon amongst humans, in its many forms. So, how does a child, by the age of four, have grammatical rules in place, which it could not have been possible to be learnt through Skinnerian type conditioning, in that time? This is where UG comes in; it is proposed that UG is an innate, unconscious ability present at birth, knowledge of grammar. This is not suggesting that a child does not make grammatical errors, as we all know, children do, but it seems that they only make irregular type errors, such as «he holded» instead of «he held», so somehow they have the ability to accept these rules and apply them. It can be argued, by people such as Skinner, that these rules are simply learnt through conditioning, but how can this be so? As the child never hears anyone make these mistakes, so does not learn them that way, this is the poverty of the stimulus theory i.e. there is not enough, or indeed any, of this kind of stimulus to learn from. With UG set in place at birth, the child is able to take on whichever language it is exposed to, as all languages have common elements and are inter- translatable.

SUMMARY If we want to provide our theory of language with explanatory adequacy we need to assume UG. By adopting linguistic competence as the core grammar, language acquisition is the task to learn those marked options that characterise a particular parameter.

AN OVERVIEW OF GENERATIVE GRAMMAR

27

1. We cannot count with all possible linguistic data of a particular language to formalize a grammar. In addition, if we only pay attention to linguistic performance, we will sometimes obtain contradictory data caused from other factors such as slips of the tongue, insecurity, nervousness, etc. 2. Linguistic competence is unlimited and in principle we may learn any particular language. Third, the speaker may produce new sentences, never heard before, and be sure that such sentences are grammatical in the grammar of his language.

NOW YOU ARE READY TO DO EXERCISES 6, 7 and 8

3. LEXICAL AND FUNCTIONAL CATEGORIES, SUBCATEGORIZATION AND CONSTITUENT STRUCTURE The lexicon constitutes the basis for all grammatical relations. In fact, this grammar level needs to satisfy «external» constraints of the interface relation with other levels of grammatical representation. The constituent structure and the so-called subcategorization give account of the organization of syntactic categories reflected in the structure of the semantic universe. In this chapter, we look at the main categories found in the English lexicon and describe them considering their structure and how they relate to each other. The lexicon of English as of any language is composed of words and lexical units that define it in a particular way. Each word can be classified into categories such as nouns, verbs, prepositions, adjectives, and adverbs. Whereas we may find a limit set of categories, new words are freely incorporated as long as they are created. In order to classify each lexical unit we may take advantage of features. For example a word like «pen» is a lexical unit with the features [+N] and [+sing]. A word like «ugly» is another lexical unit with the exclusive feature [+Adj] and no number feature in this language. Likewise, there are other elements, which although cannot be created as freely as we saw before, are part of constituent structure and are subject to subcategorization requirements. They are devoid of lexical content and since they exclusively play a grammatical role, they are interpreted as functional categories. For example, the conjunction «which» cannot

28

CURRENT TRENDS IN LINGUISTIC THEORY

represent any lexical meaning with respect to the external world but is however required to form relative or interrogative sentences. This particular grammatical role is already present in the lexicon, where the word «which» is related to the feature [+Q]. In the following pages, we will describe both lexical and functional categories in turn. In addition, we will discuss some English evidence showing the importance of functional categories in the English syntax.

3.1. Syntactic and morphological evidence When we say that a word belongs to a grammatical category, we mean that it shares with other words a number of properties so that it acts in a similar way with respect to grammar. As we will see these properties are mainly related to morphological and syntactic evidence. For example, words like «girl, horse, leg, principle, house, work etc.» belong to the grammatical category «noun» and as such they all share certain grammatical properties: e.g. they have a plural form (through the suffix +s), they can be modified by the definite article «the», etc. Similarly, words like «hear, believe, love, admire, hit, seem, etc.» belong to the grammatical category «verb» and as such they all share certain grammatical properties: e.g. they can bear morphological features when they are finite, or take the progressive +ing suffix, or appear after preposition «to» when they act as nonfinite. Likewise, words like «short, bright, dark, fast, etc.» belong to the grammatical category of «adjective» and as such they share a number of grammatical properties in common (e.g. they can take the comparative +er suffix), and may be modified by the adverb «very». Words like «really, hardly, always, often, never,etc.» belong to the grammatical category of «adverb» and share certain grammatical properties (not least the fact that they all end in the suffix «+ly»). Finally, words like «in, on, up, down, over, across, between, etc.» belong to the grammatical category «preposition» and they share some grammatical properties in common (e.g. they can be intensified by a word like «right» or «straight»). At first sight, the generative grammar might not have been interested in the lexicon, since its main focus was initially transfomational rules, i.e., those deriving complete sentences ranging from exclamatory to whquestions. However, during the past twenty five years of generative grammar studies, the lexicon has assumed an increasingly important role. In fact many phenomena seem to be better explained by rules associated with particular lexical entries or sets of entries (see Chomsky 1981, 1996;

AN OVERVIEW OF GENERATIVE GRAMMAR

29

Jackendoff, 1977, 1978, Hale & Keyser, 2002, among many others). In this section, we exclusively focus on some morphological and syntactic evidence discussed in the literature for words to belong to one or another category and briefly discuss some specific configurations of lexical entries to address the question of how many arguments can appear with a word.

3.2. Lexical categories In the first place, the relevant morphological evidence supporting the existence of lexical categories comes from the inflectional and derivational properties of words. In turn, inflectional properties relate to the fact that some words may take or end in the same suffix. Derivational properties relate to the process by which a word can be used to form a different kind of word by the addition of another morpheme (e.g by adding the suffix «ize» to the noun «television» we can form the verb «televize»). Although English has a highly impoverished system of inflectional morphology, there are none the less two major categories of word which have distinctive inflectional properties –namely «nouns» and «verbs». We can identify the class of nouns in terms of the fact that they typically inflect for «number», and thus we have (potentially) distinct «singular» and «plural» forms cf pairs such as dog/dogs, man/men, ox/oxen, etc. Accordingly, we can differentiate a noun like «fool» from an adjective like «foolish» by virtue of the fact that only (regular) nouns like «fool» not adjectives like «foolish» can carry the noun plural inflection «+s», cf. e.g. (i) They are fools [noun]/*foolishes [adjective] (Radford, 1997: 38).

English verbs may regularly inflect for «tense» and thus we have (potentially) distinct «present» and «past» verb forms cf. «want/wanted», «record/recorded», watch/watched». Present English verbs also have a defective inflection form for «person», and thus we have verbal forms such as «I/you/we/they record» versus «He/she/it records». Syntactic distribution is the other piece of evidence for words to belong to specific categories. For example, it is well known that adjectives in English precede nouns. So, the only possible slot to be filled with an adjective in the example (1) below precedes the noun and cannot follow it, as illustrated by the grammaticality contrast of (1a) versus (1b): (1)

The ...... man ....... left the room a.

The sad man left the room

b.

*The man sad left the room

30

CURRENT TRENDS IN LINGUISTIC THEORY

We also know that only verbs may follow modal verbs: (2)

a.

We can speak aloud

b.

You must be silent

In addition, we can differentiate a finite present verb like «watch» from a nonfinite verb form taking the suffix «ing» like «watching» by virtue of the fact that only the latter may be combined with the auxiliary «be» to form the progressive, as shown by the grammaticality contrast of the following sentences: (3)

a.

John is watching the football

b.

*John is watch the football

If UG is responsible for the mechanisms that lead words to systematically belong to one particular category, irregular forms clearly present a problem for our idea of language acquisition. In fact, findings in language acquisition reseaarch show that the English child simultaneously overegularize stems and their irregular forms, producing «went, wented, wenting» as well as «go, goed and going». The null hypothesis is that the child simply does not recognize that «go» and «went» even belong to the same paradigm. Pinker (1996) argues that some conjunction of phonetic and semantic similarity is involved; the phonetic dissimilarity of «go» and «went» may at first prevent the child from combining them into a single paradigm, requiring a correspondingly greater amount of semantic evidence before the child can do so. After this, all these irregular forms will occur in the Child lexicon as marked lexical entries.

3.3. Constituent structure and subcategorization In the first studies of the lexicon (Chomsky 1970, Emonds 1976, Stowell 1981) only four categories, the major lexical categories N, V, A, and P could project a particular structure, in the sense of having different projectionlevels in terms of which a Specifier position (spec) and a Complement position (cmp) could be distinguished: (4) N2 Spec

V2 N1 N0

Spec Cmp

A2 V1 V0

Spec Cmp

P2 A1

Spec

P1

A0

Cmp

P0

Cmp

However, some years later the constituent structure depicted for the lexical categories above was extended to functional categories, as we will

AN OVERVIEW OF GENERATIVE GRAMMAR

31

discuss in the following section. Before dealing with them we must deal with the notions of complements and specifiers. A complement is a phrase that a lexical category takes or selects. Which complements are taken by a particular verb, for instance, is an arbitrary property of that verb. In what follows, we illustrate this with some Verb Phrase (VP) examples1: (5)

VP examples a)

died/ *died the corpse/ *died to Peter about politics

b)

relied on John/ *relied / *relied from John / *relied to Peter

c)

dismembered the corpse/ *dismembered

d)

talked (to Mary) (about languages) / *talked that English is easy

e)

read (the paper) (to your mother) / read that the economy is poor

f)

supplied the Iraquis (with arms) / *supplies

g)

told Anne (that it is late)

h)

revealed (to Peter) that problems exist/ revealed the answer (to Peter)

The verb died in (5a) cannot have any complements; the verb relied in (5b) must have a Prepositional Phrase (PP) as complement; the verb dismembered in (5d) must take a Noun Phrase (NP) as complement; the verb talked in (5e) can take an optional PP complement with to as the head and/or an optional PP complement where the prepositional head is about, etc. These complement selection requirements can be represented in subcategorization frames, as shown next, where the square brackets delimit the phrase and the environment bar indicates the position of the lexical head. While required complements are listed, optional complements are enclosed in parentheses. In the cases where a complement with a particular head is subcategorized for, the head is listed as a feature on the complement (see rely and talk).

1 As pointed out in section one, a star (*) before a word, phrase or sentence indicates ungrammaticality. Any grammar with explanatory adequacy must not only generate all grammatical sentences in the language under study but also correctly rule out all ungrammatical sentences.

32

CURRENT TRENDS IN LINGUISTIC THEORY

(6)

Subcategorization frames Die, V, [_] Rely, V [_ PP[on] ] Rememeber, V [ _NP] Talk,V, [_(PP[to] (PP[about] ]

Adjectives, nouns, and prepositions also subcategorize for their complements, as indicated by the following examples: (7)

(8)

(9)

AP examples a)

blue / *blue that Mary will wear

b)

afraid (of darkness) / *afraid to this section

c)

orthogonal (to this question)

d)

ambivalent ((to Peter) about his feelings)

e)

certain (that Mary is going out with John)

f)

insistent (to his husband) (that they will leave)

NP examples a)

team (of students)

b)

individual

c)

book (about languages / to English)

d)

generosity (to mankind)

e)

dislike of beasts

f)

ambivalence ((to Peter) about his feelings)

g)

rumor (that he will be sacked)

h)

message (to Peter) (about his application)

PP examples a)

about [the meeting]

b)

before [he resigned]

c)

from [every part of the country]

d)

[hurry] up

We can generalize that the lexical categories (V, N, A, P) in English: • subcategorize for their complements; • precede their complements in the phrase; and, • co-occur with other constituents.

AN OVERVIEW OF GENERATIVE GRAMMAR

33

Finally, as we will see, heads and phrases are not only parts of phrases. For example, NPs can be preceded by other words or even phrases like: the, no, some, every, Mary’s , my father’s. APs can be preceded by degree words such as: very, extremely, rather, quite. These items differ from complements because they precede the lexical category and they are not subcategorized for. They are called specifiers.

3.4. Functional categories Chomsky (1986b) puts forward a similar constituent structure for functional categories such as I (=Inflection) and C (=Complementizer), where Inflectional Phrase (IP) replaces S (=sentence) and Complementizer Phrase (CP) replaces S’. In this system, the more usual notation is then XP,X’ and X instead of X2, X1 and X0: (10)

CP Spec

C' C Spec

IP. I' I

VP

Node C is the category of complementizers like «that, for, and if» and its Specifier is the landing site for wh-movement, as we will see in the following section 3. I (=Inflection) is the category of verbal inflection and, in English, of modal auxiliaries; the subject of the sentence occupies the Specifier of IP. Interestingly the functional projection I subcategorizes for the lexical category V. So, it is through subcategorization in the lexicon that functional categories relate with lexical categories and this will be projected in the syntax. We will return to this in section 3. Finally, Chomsky’s proposal of functional categorioes such as CP and IP has been followed for many other minor categories and inflectional elements. Abney (1987), for example, proposes that determiners and degree elements (D and Deg) are also heads in the X’-theory:

34

CURRENT TRENDS IN LINGUISTIC THEORY

(11) a.

b.

DP

Spec

D'

DegP

Spec

D

NP

Spec

Deg' Deg

N'

AP

Spec

A'

N

A

Another influential proposal is found in Pollock (1989) on the basis of languages like French where the verb is inflected for tense and agreement features. According to this proposal the elements under I: Tense (TP) and Agreement (AgrP) are realized as separate functional categories, each heading their own projection: (12)

TP T' T

AgrP Agr' Agr

VP

The hypothesis is that T subcategorizes for AgrP as its complement and Agr for VP in this order. Other functional heads that have been proposed for the verbal system are given in (12): (12)

Asp for aspectual markers or auxiliaries (Tenny 1987) Neg or Pol for negative and affirmative markers (Pollock 1989) Mod for modal auxiliaries (Ouhalla 1991) Agr-S for subject agreement (Chomsky 1988) Agr-O for object agreement (Chomsky 1988) Voice or Pass for passive morphology (Ouhalla 1991)

For the nominal system, Hale & Keyser (1991) have a head for Case, K(ase), above D as illustrated in (13). Others like Ritter (1991) puts forward a functional category NumP above NP.

AN OVERVIEW OF GENERATIVE GRAMMAR

(13)

35

KP K' K

DP D' D

NP

Although the introduction of so many new heads seems to be a radical innovation of the theory of grammar, the existence of functional categories has always been recognized in one way or another. The distinction between lexical and functional heads may be made in different contexts under different names. Some examples follow: (14) open class items

-

closed class items

content words

-

function words

lexical formatives

-

grammatical formatives

major categories

-

minor categories

Almost all functional heads that have been proposed correspond directly or indirectly to the so called grammatical categories of traditional grammar. Each major part of speech (noun, verb, and adjective) has its own characeteristic properties, expressed by the inflectional affixes and closed characteristic properties, expressed by the inflectional affixes and closedclass particles which co-occur with it (Lyons 1968, Schachter 1985): (15) part of speech:

grammatical category:

noun

case, definiteness, number, gender

verb

mood, tense, polarity, aspect, voice

adjective

degree

What is really new about today’s functional heads is the way they are assumed to project syntactically. This can most clearly be seen in the case of noun phrases. In fact, there has been a change in the way they are analysed. Consider the following two syntactic trees:

36

CURRENT TRENDS IN LINGUISTIC THEORY

(16) a.

b.

NP DP

N'

D' D

DP D' D

N

NP N' N

In the classical analysis of the noun phrase (16a), the determiner is not a head of the noun phrase but its specifier (sister of N’) . In the modern analysis of the noun phrase (16b), the determiner is the head of the noun phrase, taking NP as its complement. The same difference can be seen in the analysis of degree-elements, auxiliaries, negation, and complementizers. Functional heads shouldn’t therefore be seen as an exclusive peculiarity of the generative view on syntactic structure. The study of the properties of lexical and functional categories and of their differences is a matter of general linguistic concern, independent of the theoretical framework. This is also evident from the fact that the lexical-functional dichotomy plays a role in such diverse areas as language variation, acquistion, agrammatism, and language production. Although the lexical-functional dichotomy can be approached from many different perspectives, syntactic questions like: what distinguishes lexical and functional categories from each other? Why is it that a lexical category has its own set of functional categories, occurring in a specific order? or Is it possible to distinguish natural classes of functional categories with more or less properties? have always been addressed from the generative perspective. For some references see Zwarts (1992), Abney (1987), Fukui & Speas (1985), Grimshaw (1991), Van Riemsdijk (1990).

3.4.1. English auxiliaries and infinitival «to» So far, we have argued in favour of the existence of functional categories. In this section we look at auxiliaries and infinitival «to» in English following Radford’s (1997) description of functional categories. As traditional grammar has shown, auxiliaries behave in a very different manner with respect to verbs. Auxiliaries then belong to a different category. Auxiliaries typically take a verb expression as their complement, and have

AN OVERVIEW OF GENERATIVE GRAMMAR

37

the semantic function of marking grammaticasl properties associated with the relevant verb, such as tense, aspect, voice, mood or modality. Examples of auxiliaries that take verbal complements in brackets are: «have» traditionally considered to be a perfective auxiliary as in (17a), «be» an imperfective/progressive auxiliary as in (17b,c), «do» a dummy (i.e. meaningless) auxiliary as in (17d), and the modal auxiliaries «can/could/may/might/will/would/shall/should/must» as in (17e-i) (Radford 1997, (30 a-i), p50): (17)

a.

He has/had [gone]

b.

She is/was [staying home]

c.

He is/was [seen regularly by the doctor]

d.

He really does/did [say a lot]

e.

You can/could [help]

f.

They may/might [come back]

g.

He will/would [get upset]

h.

I shall/should [return]

i.

You must [finish your assignment]

The main differences between auxiliaries and main verbs is that the former but not the latter undergo some syntactic processes such as: i) inversion in questions as illustrated by the examples in (18); ii) they can be directly negated by a following «not» usually contracted down onto the auxiliary in the form of «n’t», as in examples in (19); and, iii) appear in sentence-final tags, as in examples in (20): (18)

(19)

(20)

a.

Can you read French?

b.

Do you live in Spain?

c.

Are you calling abroad?

a.

Mary cannot/can’t swim

b.

He does not / doesn’t live in Spain

c.

He is not/ isn’t listening

d.

We have/ haven’t read the paper today

a.

The doctor hasn’t come yet, has he?

b.

They won’t lose, will they?

c.

You aren’t working, are you?

d.

He can’t speak English, can he?

38

CURRENT TRENDS IN LINGUISTIC THEORY

Note that main verbs cannot, on the other hand, undergo inversion, be directly negated or appear in sentence-final tags, as illustrated by the ungrammaticality of the examples in (21), (22) and (23) respectively: (21)

(22)

(23)

a.

*Read you French?

b.

*Live you in Spain?

c.

*Calling you abroad?

a.

*Mary swims not

b.

*He lives not in Spain

c.

*He is listening not

a.

*The doctor hasn’t come yet, comes he?

b.

*They won’t lose, lose they?

c.

*You aren’t working, work you?

A second type of functional category is English infinitival «to» which also takes an infinitive as complement. Typical examples of infinitival «to» taken from Radford (1997: 37) are illustrated in (24) below: (24)

a.

I wonder whether to [go home]

b.

Many people want the government to [change course]

c.

We don’t intend to [surrender]

Radford (1997) addresses the question of the categorial status of this infinitival «to». He gives compelling reasons for assuming that infinitival «to» belongs to a different category from prepositional «to». The proposal then is that while infinitival «to» is a functional category, prepositional «to» is a lexical category. First of all, Radford (1997) points out that prepositional «to», like other prepositions, takes a determiner phrase (DP) as its complement in brackets in the examples of (25) below: (25)

a.

He stayed to [the end of the film]

b.

Why don’t you come to [the point]?

c.

He went to [the police]

In the examples of (25), prepositional ‘to’ has intrinsic semantic content (e.g. it means something like «as far as»). Infinitival «to», on the other hand, seems to be a dummy (i.e. meaningless) functor with no intrinsic semantic content. In fact, the preposition «to» can often be modified by intensifiers like «straight/right» ( a characteristic property of prepositions) - cf. (Radford 1997: 39)

AN OVERVIEW OF GENERATIVE GRAMMAR

(26)

a.

He stayed right to the end of the film

b.

Why don’t you come straight to the point?

c.

He went straight to the police

39

As expected, the dummy functional category «to» cannot be modified by those intensifiers: (27)

a.

*I wonder whether straight/right to [go home]

b.

*Many people want the government straight/right to [change course]

c.

*We don’t intend straight/right to [surrender]

A famous argument in the generative grammar defending a functional category slot for infinitival «to» derives from the phenomenon of the «wanna contraction». As is well known, in American English the verb «want» can be contracted with «to» in affirmative sentences, as illustrated by the pair in (28): (28)

a.

I want to go

b.

I wanna (want+to) go

However if there is an embedded subject followed by the infinitive particle that takes an infinitive like any other auxiliary, the particle «to» can no longer contract down onto the main verb «want» as illustrated by the following minimal pair: (29)

a.

I want you to [come]

b.

*I wanna you come

In the example (29a), «to» patterns like any auxiliary we discussed before in taking an infinitive and following a subject. The raising question is whether it belongs to the same type of functional category. Chomsky (1981, p.18) proposes the functional category INFL (I) or inflection for both finite auxiliaries and infinitival «to». The general idea behind the I label is that finite auxiliaries inflect for tense/agreement, and infinitival «to» serves much the same function in English as infinitive inflections in languages like Italian which have overtly inflected infinitives. As Radford (1997) argues, we can then say that an auxiliary like «should» is a finite I/Infl, whereas the particle «to» is an infinitival I/INFL.

3.4.2. Determiners In this section we will provide more arguments for the idea that the functional categories I and D constitute a natural class of categories.

40

CURRENT TRENDS IN LINGUISTIC THEORY

That the functional heads in (30) have something in common is one of the central claims of Abney (1987): (30)

IP

DP

I'

D'

I

VP

D

NP

Abney suggests that what I and D have in common is their «determinerlike» status that may be characterised according to the following properties: a) they can carry grammatical features, b) they can be pronominal, and c) they may carry referential force. In the first place, pronouns may be classified as functional categories of the Determiner type by virtue of the fact that they lack descriptive content but carry grammatical features such as case. Secondly, whereas a noun like «boys» denotes a specific type of people, a personal pronoun like «they» denotes no specific type of entity, but has to have its reference determined from the linguistic context. Fnally, the fact that pronouns necessarily refer to other entities explains their referential force. As is well known, personal pronouns simply encode sets of person, number, gender and case features - as represented in the table (31) below: (31) I we you he she it they me us you him her it them mine ours yours his her its theirs

PERSON +1 +1 +2 +3 +3 +3 +3 +1 +1 +2 +3 +3 +3 +3 +1 +1 +2 +3 +3 +3 +3

NUMBER +sing +plur +/-sing +sing +sing +sing +plur +sing +plur +/-sing +sing +sing +sing +plur +sing +plur +/-sing +sing +sing +sing +plur

GENDER ——————————————————————————————————————————-

CASE +Nominative +Nominative +Nominative +Nominative +Nominative +Nominative +Nominative +Acusative +Acusative +Acusative +Acusative +Acusative +Acusative +Acusative +Genitive +Genitive +Genitive +Genitive +Genitive +Genitive +Genitive

AN OVERVIEW OF GENERATIVE GRAMMAR

41

Possessive pronouns: «my,our, your, his, her, their» also lack descriptive content and necessarily make reference to a second personal entity by virtue of their referential force. In addition, we may also argue for the existence of empty determiners which explains the interpretation of mass nouns in the absence of a lexically filled D. As is well known, mass nouns are interpreted as indefinite and this may straightforwardly follow from such an assumption. I just want to comment on the issue of DPs. The issue is whether Nouns or Determiners ‘project’; that is, which element is the head of the constituent. And the argument used to be that Nouns project, thus the constituent was an NP, and the D a specifier. However, think of some of the examples that Radford provides in his Minimalist Syntax book (2004), Ch. 4.10. In the example «Italians like Opera», you would have two NPs, ‘Italians’, and ‘Opera’. However, it was noted that ‘The Italians like the Opera’ was also an acceptable sentence of English and that the D was an optional element. You need to account for this fact and you are, therefore, forced to say that ‘the Italians’, with an overt determiner, has the same status as the ‘bare’ ‘Italians’. The structure of the latter would be something like [DP [D’ [D 0 NP [N’ [N Italians]]] ]]. In addition, consider the following sentences: «Italians and [the majority of Mediterraneans] like opera.» «Italians like [opera] and [the finer things in life]». The fact that the bare elements ‘Italians’ and ‘Opera’ can be coordinated with the fully fledged DPs «the majority of Mediterraneans» and «the finer things in life» (understood here as elements headed by a D=the) indicates that the former have to be analysed as DPs too, since only similar kinds of categories can be coordinated. Thus, ‘Italians’ and ‘Opera’ are analysed as N preceded by an empty D, i.e. DPs. The conclusion is then that functional D has a grammatical feature «+def» which explains why bare elements are interpreted as generic as other fully fledged DPs. In sum, there are a number of grammatical properties along with empirical evidence that make Determiners be functional categories, playing an important role in the syntax-semantics interface.

3.4.3. Complementizers As we already argued, Complementizers are part of the closed class of functional categories. Following Radford (1997) we will next illustrate English examples of complementizers (abbreviated to COMP in earlier work and to C in more recent work). Consider the following examples:

42

CURRENT TRENDS IN LINGUISTIC THEORY

(32)

a.

I think [that you can go]

b.

I doubt [if you can come]

c.

I’m eager [for you to complete the course]

Each of the bracketed clauses in (32) is a complement clause, in that it functions as the complement of the word immediately preceding it (think/doubt/eager); the italicized word which introduces each clause is the complementizer. As we mentioned before, complementizers play an important grammatical role. In the examples above they serve to introduce finite and non finite clauses (32a,b and 32c, in turn). Complementizers in structures like (32) have three grammatical functions: i) they serve to introduce the complement of some other word (think/doubt/eager); ii) they indicate the type of clause they introduce, i.e. finite (containing an inflected verb or an auxiliary verb) or non finite or infinitival (containing infinitival «to») and iii) they mark the illocutionary force (i.e. semantic/pragmatic function) of the clause they introduce. In particular, «if» introduces an interrogative clause, whereas «that/for» introduce other types of clause:e.g. «that» typically introduces a declarative/statement-making clause. As Radford (1977) also argues, there are significant differences between complementizers and other apparently similar words. To illustrate, the complementizer «for» and the preposition «for» differ in that the preposition «for» has intrinsic semantic content and as expected can be intensified by «straight/right», whereas the complementizer «for» is a dummy functor and can never be so intensified. Compare the contrast of grammaticality of the following examples: (33)

a.

The car headed for/straight for the motorway

b.

*I’m eager [straight for you to complete the course]

In addition, the syntax for the preposition «for» and the complementizer «for» is also different. Whereas the complemetizer «for» can introduce a complete sentence, the preposition «for» cannot: (34)

a.

For him to speak in favour of the victim was very difficult

b.

*For him was very difficult

Complementizers are different from other functional categories like I or D discussed before in that they cannot be related to a subject case position. In fact the complementizer «for» always introduces infinitival sentences with an embedded subject that cannot appear in the nominative case, as illustrated by the following minimal pair:

AN OVERVIEW OF GENERATIVE GRAMMAR

(35)

a.

Her parents are anxious for her to marry

b.

*Her parents are anxious for she to marry

43

We may then conclude that complementizer «for» serves to introduce complement clauses. This may be derived from the fact that «for» is a preposition in the lexicon and prepositions cannot assign nominative case. This is supported from the fact that other complementizers, which cannot have any preposition counterpart, may, on the other hand, introduce finite clauses. This is the case of «if» and «that», as illustrated by the following examples: (36)

a.

I wonder if she will get marry once

b.

You must realise that she will get marry once

Note that neither of these can intorduce non-finite or infinitival clauses: (37)

a.

*I wonder if her to get marry once

b.

*You must realise that her get marry once

Following Radford (1997), there are a number of reasons for rejecting the possibility of analysing complementizer «if» as a wh-adverb of some kind. In addition to the above illustrated fact that «if» cannot introduce a non-finite clause in (37a) in contrast to typical wh-adverbs as illustrated by the following examples: (38)

a.

I wonder when to go

b.

I wonder where to go

c.

I wonder whether to go

Unlike wh-adverbs, complementizer «if» cannot follow a preposition, which explains another contrast of grammaticality in the following examples: (39)

a.

I’m not sure about when/where/whether she will marry

b.

*I’m not sure if she will marry

To conclude, «if» is a complementizer like «that» or «for» and as such serves the function of introducing specific types of clauses.

44

CURRENT TRENDS IN LINGUISTIC THEORY

SUMMARY In this chapter we have discussed the difference between lexical and functional categories. We have mainly addressed questions like: 1. What distinguishes lexical and functional categories from each other? 2. Why is it that a lexical category has its own set of functional categories, occurring in a specific order? or 3. Is it possible to distinguish natural classes of functional categories with more or less properties? The first question was answered on the basis of morphological and syntactic evidence. The second question was adressed under the perspective of constituent structure and subcategorization. According to this, all lexical entities project in the syntax through functional categories that enter into certain syntactic configurations, basically through the Spec-head and head-complement relations. Finally we have discussed three main natural functional categories in English: • Inflection (I), • Determiners (D) and • Complementizers (C).

NOW YOU ARE READY TO DO EXERCISES FROM 9 to 18

4. THE SYNTAX OF CLAUSES In the previous section we identified words with different syntactic classes. We also saw that one-word items can form part of larger syntactic combinations. They were grouped into phrase-level categories thanks to the X-bar schema. In this section, we extend the X-bar theory to clauses and we will focus on the development of the syntactic analysis of phrases and larger clauses. We will look at the internal structure of each constituent in the clause and consider all the possible word orders in English in all declarative, negative, and interrogative sentences. We will end up the section with the analysis of passives. At the end of this section you will be able to reformulate traditional grammatical notions such as subject, object or indirect object in more refined syntactic analyses. The sections devoted to syntax are designed to provide a data-motivated, stepwise introduction to the main tenets of the Government and Binding (GB) theory developed in the Principles and Parameters approach to language by Chomsky (1981, 1982, and 1986).

AN OVERVIEW OF GENERATIVE GRAMMAR

45

4.1. X-bar syntax: a brief historical perspective In the tradition of generative grammar, trees and phrase-markers have always been one of the most well-known technical mechanisms ever since the beginning of the theory in the late 1950s. As a matter of fact, the first thing that the generative grammar did was to define what syntactic structure means. Consider the sentence in (1): (1)

The students passed their exams

You may analyse sentence (1) as a simple linear string of words. You may describe the sentence as consisting of the words the, students, passed, their, exams in that order. However if that were all there was to syntax, we would not expect the rich work developed in this theory. To begin with, the statement that sentence (1) consists of a linear string of words misses several important generalizations about the internal structure of sentences and how these structures are represented in our minds. The Xbar theory claims that the words in sentence (1) are grouped into units (called constituents) and that these constituents are grouped into larger constituents, and so on until one gets a sentence. In section 2 we looked at the kinds of words that form sentences. We saw that the Government & Binding approach seeks to capture the similarities between different lexical and functional categories by assigning the same structures to them. Rather than having different phrase structure rules for VPs, NPs, IPs, CPs, etc, just the two following basic rules cover all the lexical and functional categories: i) XP → Specifier X’ ii) X’ → X Complements (=YP) In this section we will study the specific position occupied by those constituents within sentence structure and also the movement of such constituents within those limits. First, we will briefly discuss treediagrams. At the beginning of generative grammar, the representation of phrases and sentences in tree-diagrams implied what was known as flat structure. This is a kind of structure where all major elements appeared at the same level. See the configuration (2) for the sentence in (1):

46

CURRENT TRENDS IN LINGUISTIC THEORY

(2)

S NP

Aux

Det

N

The

studens

VP

have

V

NP

passed Det the

N exams

On a par with tree-diagrams or phrase-markers, labelled bracketing was and still is a means of representing the configuration of phrases, as shown next: (3)

[S [NP [D the] [N students]] [Aux [have] [VP [V passed] [NP [D the] [N exams]]]]

One of the most decisive aspects in the analysis of the tree-diagrams since the beginning was their hierarchical structure by which we could study the different levels that each constituent occupies in the structure. According to the Xbar theory, all major nodes within a sentence bore the same categorical type and all minor nodes or heads projected up to three levels. Therefore, we may speak of three projections, namely: the maximal projection XP or X», the intermediate projection X’ and the head or lexical projection X. (4)

XP/X'' Spec

X' X

Complement

Although the original idea that lexical heads project into maximal projections offer many advantages in the analysis of linguistic facts, the existence of intermediate projections such as Xbar has not always been easy to show.2 In fact, the two most relevant relations that have always been kept in mind along the years are internal to the hierarchical structure of (4). First, the one established between the head and the Specifier at a higher level, and the one established between the head and the complement at the same

2 In a more recent refined analysis Chomsky (1995) proposes that we should only speak of functional heads that project into maximal heads. Elements will either merge with the former or the latter.

AN OVERVIEW OF GENERATIVE GRAMMAR

47

level. The empirical fact widely attested in the syntactic theory is that specifiers and complements are not compulsory nodes on a phrase-marker, but if they occur, their positions must be naturally different from each other. Distinct syntactic and semantic properties are, in addition, found with them. For example, the English genitive «Jonh’s book» is analysed as in (5), where the head is the genitive marker «s», the possessor occupies the Specifier while the possessed occupies the complement position: (5)

[John [s [head]]] = [DP [NP John] [[D s] [NP [N head]]]

The structure in (5) correctly captures the mapping between the semantic notions of possessor and possessed and their syntactic distribution based on the hierarchical structure of (5), where the possessor occupies the subject position and the possessed the object position.

4.2. Structural relations As we have previously argued, there is an organization in each treediagram, where each constituent relates with one another in a very particular way. Here you will learn about some purely formal properties of trees. Before we could only give a vague definition of constituents. We then argued that their distribution in the sentence affects to their interpretation. Here we support this by looking at anaphors. Anaphors can only appear in certain positions in the geometry of the tree. A more accurate approach to anaphors will be provided in subsection 4.5, where we will focus on the binding theory, a different module of generative grammar. To begin with, we have to analyse the parts of the tree. Look at a very abstract tree drawing in (6): (6)

A B D

E

C F

G

H

I

The tree in (6) would be generated by the phrasal rules in (7).You may check the derivation of the tree by applying each of these. Next, we can describe the various parts of the tree. (7)

A→BC

48

CURRENT TRENDS IN LINGUISTIC THEORY

B→DEF C→GHI (8)

Different parts of the tree a) Branch: A line connecting two parts of a tree. b) Node: The end of a branch. c) Label: The name given to a node. d) Root node: the root with no line on top of it. e) Terminal node: Any node with no branch underneath it. f) Non-terminal node: Any node with a branch underneath it.

Given the tree structure at hand, all the notions in (8) refer to structural relations. In addition, we may also find an informal set of terms in the generative literature that usually refer to dominance. This other set of terms is based on the fact that syntactic trees look like family trees. As Castillo (2003) points out: According to horizontal-like relations, a node precedes another node if it is at the same level as the latter and to its left. Along with precedence or like-like relations, dominance or vertical relations constitute the other basic spatial connection in syntax, from which the technical terms of mother, daughter, and sister derive: a mother node immediately dominates two daughter nodes, which are sisters of each other.

In more formal terms, the notion of immediately dominance may be expressed as follows: (9)

Immediately dominance (Carnie 2002: p.70: 19) Node A immediately dominates B if there is no intervening node G that is dominated by A, but dominates B. (In other words, A is the first node that dominates B.)

In effect, in (6) A dominates all the other nodes in the tree, but it only immediately dominates B and C. It does not immediately dominate any of the other nodes because A and B intervene. Therefore we may express the relation of motherhood and sisterhood as follows: (10)

a)

A is the mother of B if A immediately dominates B;

b)

B is the daughter of A if B is immediately dominated by A;

c)

Two nodes that share the same mother are sisters.

AN OVERVIEW OF GENERATIVE GRAMMAR

49

Applying each relationship to our previous tree drawing in (4), now in (11): (4)

XP/X'' Spec

X' X

Complement

we will say that the Spec is daughter to the phrasal projection XP or X» and sister to the intermediate X’ whereas the complement is daughter to the intermediate projection X’, and sister to the head or lexical projection X or X0. As Castillo argues, … the principle or condition to which a mother node immediately dominates a maximum of two nodes is known as binary branching. Binary branching is a most welcome condition on the representation of phrases which is thought to be in accord with the human mind’s processing mechanism. From each node can originate either only one or two nodes, the occurrence of additional constituents being solved by creating a higher number of (bipartite) divisions within sentence structure.

Turning back to syntactic relations, and to be able to further understand how the position of one constituent may affect its interpretation, you should know the notion of c-command, which will be very important when we deal with binding in the next subsection. C-command is both formally and informally defined by Carnie (2002) as follows: (12)

Carnie (2002, p.75: 41 & 42) a)

C-command (informal) A node c-commands its sisters and all the daughters (and granddaughters, and great-granddaughters, etc.) of its sisters.

b)

C-command (formal) Node A c-commands node B if every branching node dominating A also dominates B, and neither A nor B dominate the other.

Look at the tree in (13). The node D only c-commands the nodes in the circle:

50

CURRENT TRENDS IN LINGUISTIC THEORY

(13)

A B

C D

E F

G

H

I

J

L

K

M

That is, D c-commands its sister (E) and all the nodes dominated by its sister ( F, G, H, I, J, K, L, M). Look now at the same tree without the separating line, and look at the nodes c-commanded by J: (14)

A B

C D

E F

H

G I

L

J

K

M

Node J c-commands only node K (its sister). In fact, you may check that it does not c-command the nodes F, H, I, L or M. Given all this, we may conclude that C-command is a relation that holds between sisters and aunts and nieces. It never holds between a mother and a daughter. Finally, you may re-define the traditional grammatical relations of subject, direct object, or indirect object as object of a preposition as follows: (15)

a)

Subject: NP daughter of S

b)

Direct Object: NP daughter of VP

c)

Object of a preposition: NP daughter of PP

Check each grammatical relation in the examples provided next: (16)

Subjects a. The spider frightened the child b. It snows

AN OVERVIEW OF GENERATIVE GRAMMAR

51

c. John looks tired d. The city was destroyed (17)

Objects a. John loves Mary b. I like you c. The man hit the dog

(18)

Objects of Preposition a. He dreamt about her b. They fought against the heat in Africa c. I gave the book to Peter

4.3. X-bar theory, subcategorization and reanalysis of sentences As we have just suggested, the traditional analysis of sentences consisted of the simple phrase rule S → NP VP. However the development of the Xbar theory has required a reanalysis of this rule in X-bar terms, its head, complement and specifier must be determined. None of the constituents are on the right side if this rule can be the head of a phrase because they are phrases themselves, not lexical items or words. In addition, most syntactic properties related to sentences have been captured by assuming the existence of functional categories such as CP or IP. To find out what the head of a sentence is and to what type of functional category CP or IP the sentence belongs, we need to look for evidence of subcategorization. For example, consider the following groups of data and pay special attention to the embedded sentences in each: Group 1 a)

The president thought that he would win the elections.

b)

The president thought that the elections would be won.

c)

*The president thought that the elections to win.

Group 2 a)

John arranged for his holidays to be ready.

b)

*John arranged for his holidays were ready.

c)

*John arranged for his holidays would be ready.

52

CURRENT TRENDS IN LINGUISTIC THEORY

Group 3 a)

Mary wondered whether they would be invited to the wedding.

b)

Mary wondered whether they were invited to the wedding.

c)

*Mary wondered they were invited to the wedding.

When the complementizer (C) is either that or whether, the sentence that follows is a regular finite sentence, and to cannot be present as shown by the sentences c in the Groups 1 and 3. In contrast, when the complementizer is for as used in the sentences of Group 2, to must be present and a finite or tensed verb is allowed in the following sentence. So, the complementizers that and whether subcategorize for a finite complement, whereas for requires a non-finite complement. Yet, the head of that complement still needs to be determined. We saw in section 2 that to must be present when the complementizer is for. We then conclude that to is the marker for non-finite clauses in English. Thus, for subcategorizes for a non-finite complement that must have to, so to must be the head. Further evidence that to is a head can be seen in Groups 4 and 5 below. Since to subcategorizes for the bare form of the following it, to must be a head. Group 4 a)

John arranged for his holidays to be ready.

b)

*John arranged for his holidays to were ready.

c)

*John arranged for his holidays to would be ready.

Group 5 a)

We would love for him to marry.

b)

*We would love for him to married.

c)

*We would love for him to marries.

Furthermore, we still need a category for to and for its counterpart in finite sentences. According to the Government and Binding approach we follow here, the tense and agreement features fill the same head position in finite sentences that to fills in non-finite sentences. The category is therefore called Inflection, or I for short. This means that a sentence may be an Inflection Phrase (IP).

AN OVERVIEW OF GENERATIVE GRAMMAR

53

Again, in terms of subcategorization frames, the lexical entries for the three complementizers and non-finite to can be illustrated as follows: that,

C,

[ _ IP[+fin]

]

for,

C,

[ _ IP[-fin]

]

whether,

C,

[ _ IP[+fin]

]

to,

C,

[ _ IP[+bare]

] (where bare includes infinitives and uninflected forms)

The head I[+fin] will never be filled by a lexical word in English, so it does not have a lexical entry. It always takes a VP as its complement just as non-finite to does. The subject NP is assumed to fill the specifier position in the IP. In fact, we will next explain that this is the position for NPs to be assigned Nominative Case.

4.4. Case theory In section 2, we provided evidence that English pronouns are morphologically marked with a particular case. In this section, you will learn about the basics of Case theory in the formal framework of Principles and Parameters approach where Case is expressed as a structural relation. In brief, it is assumed that all NPs have Case (called abstract case) that should match the morphological case that shows up on pronouns. In phrase structure terms: (19)

Nominative Case is assigned to the NP specifier of I [+fin]. IP NP I [+fin]

I' VP

Accusative case is assigned to the NP sister of V or P. The C[for] which is homophonous with the preposition for acts like P for Case assignment. Note that the subject of a non-finite clause could not receive Case from I [-fin] since only I [+fin] assigns Nominative Case.

54

CURRENT TRENDS IN LINGUISTIC THEORY

(20)

VP

PP

or

CP

P'

V V

or

NP

P

C' NP

C

IP [–fin]

for NP

I' I [–fin] VP to

(21)



Genitive Case is assigned to the specifier of D DP NP

D' D

What is the same about these positions that receive Case and the positions that assign Case? Chomsky observed that every maximal projection (=XP) that dominates the NP that receives Case also dominates the head that assigns it (if we do not count the IP that intervenes between the C[for] and the NP).

4.5. Binding theory While studying structural relations, there are other aspects of grammar that you will capture, namely the interpretation of NPs. This is the topic of the Binding theory, a different module of the Principles and Parameters approach. In this section, you will learn to identify three types of English NPs: referring expressions (r-expressions), anaphors and pronouns depending on how these NPs are interpreted in the sentence. Here too, you will observe that the hierarchy of the sentence is crucial for these matters. In the sentence: «Anne brought strawberries from the market», the information that you receive is that there are some strawberries in the market that Anne brought, in the context where you know who Anne is. Both «strawberries from the market» and «Anne» get their meaning by referring to objects in the known world. This kind of NPs are called referring expressions or r-expressions.

AN OVERVIEW OF GENERATIVE GRAMMAR

55

Now, in the sentence: «John hit himself on the head», John is an rexpression and gets its meaning from the context, but «himself» must refer back to «John». It cannot refer to Peter or Mary. NPs that obligatorily get their meaning from another NP in the sentence are called anaphors. We saw in section 3 and subsection 4.4 above that Case Theory determines whether a nominative pronoun, such as she or he, is used instead of an accusative pronoun, her or him, or a genitive pronoun like his. It is Binding Theory’s job to determine when a reflexive anaphor, for example, herself, is used instead of one of the pronouns, she or her, because it specifically deals with the syntactic structures on where the NP types can appear in a sentence. Binding is a general term for any of various syntactic strategies of linking or tying together the reference of nominals using anaphors. This is why, in its narrowest conception, binding involves reflexive constructions such as (22): (22)

a.

John looks at himself in the mirror

b.

Mary washes herself

Since the beginning of the 1980s some specific principles ruled out the possibility of finding pronouns instead of reflexives in the linguistic contexts of (22), capturing the fact that the corresponding sentences in (23) are ungrammatical, where the pronouns refer back to their most proximate antecedents, the NPs: John and Mary in (23a) and (23b) respectively. The notation employed to mean coreference is trough coindexation: (23)

a.

* Johni looks at himi in the mirror

b.

*Maryi washes heri

The binding principles that constitute the theory emerged from the structural relations found in each sentence. In fact one of the most relevant syntactic relations is c-command. In what follows, you may check how this relation affects the interpretation of NPs. Before we saw that English pronouns may vary with respect to their Case, ranging from nominative, genitive to accusative case: in turn: he, his, and him. Then, we did not raise the question of reflexive anaphors which like accusative pronouns usually appear in object position in English, or after a preposition. In this sense, consider the following sentences where pronouns and anaphors are in complementary distribution: (24)

a.

She/*herself sings

b.

Johni enjoyed himselfi/*himi at the party

56

CURRENT TRENDS IN LINGUISTIC THEORY

c.

Mollyi wrote a letter to herselfi

d.

Mollyi thought that Maryj hated herself*i/j / heri/*j

e.

Mollyi talked to Peter about herselfi

f.

Mollyi thought that John hated *herselfi/heri

Two main generalizations can be derived from this data, namely, (1) reflexive pronouns must corefer with some NP before them in the sentence; and (2) there is a locality condition for this coreference relationship. Go back to the examples again and check whether these two conditions hold. In effect, examples (24a-f) suggest that the antecedent, which is the NP that the reflexive corefers with, must be within the same minimal clause (S=IP) as the reflexive. The S-structure tree for (24b) is given in (25) to make the same/different clause distinction clearer. The NP following enjoyed (marked as NP-2) can be the reflexive himself because it is coreferent with the NP-1 John in the same IP. (25)

IP [NP1 Johni]

I' I

VP

V' V NP enjoyed [NP2 himselfi]

PP P' P at

NP DP

N'

the

N party

Before formulating the binding principles more precisely, we need to discuss the structural relation of c-command by the hand of the examples above. In the previous subsection we briefly discuss c-command. Now, we may adopt Reinhart´s (1976) definition of C (onstituent)-COMMAND which formally expresses the notion of ‘higher in the tree than’ as follows: (26)

a C-COMMANDS b iff a)

a does not dominate b, and

b)

the first branching node that dominates a also dominates b.

AN OVERVIEW OF GENERATIVE GRAMMAR

57

In this definition again (as in the others above) a and b stand for particular categories. For example, in tree (25) we can let a be NP-1 John and see if it C-commands NP-2 himself (=b). Clause (a) of the definition requires that NP-1 does not dominate NP-2. This is true because NP-1 is not directly above NP-2 in the same branch of the tree. Clause (b) requires that the first branching node that dominates NP-1, which is IP, also dominates NP-2. IP does dominate NP-2, so NP-1 C-commands NP-2. (Note that NP-1 also C-commands everything else under IP on the right branch.) A simple way to think of C-command is to start with a category, go up the tree one level to where it branches, then a C-commands everything down in the other branch. So, if the category you are concerned about (b) is in that other branch, a C-commands b. As you might have guessed, Ccommand is one of the conditions on binding. Returning now to the binding conditions, the official definition of binding simply adds coindexing to the C-command relation. Coindexing is marked in the tree via subscripts and indicates that the two NPs refer to the same entity. (27)

a BINDS b iff a)

a C-commands b, and

b)

a and b are coindexed.

Before proceeding with the discussion of Binding principles, we should introduce the concept of A(argument) Binding. As we saw in the tree (25), both the NP1 John and the NP2 himself are in argument positions within the same clause, and so in a more detailed analysis we conclude that John A-binds himself and this relation needs to hold locally, viz. in the same sentence. However, you will next see that this observation cannot be held given the fact that the NP1 Molly cannot be coreferent with the NP3 herself in sentence (24d). Consider the corresponding tree in (28) below.

58

CURRENT TRENDS IN LINGUISTIC THEORY

(28)

IP [NP1 Molly]

I'

I [+past]

VP V'

V

CP

thought

C' C

IP

[NP2 Mary]

I'

I [+past]

VP V'

V

NP

hated

[NP3 herself/her]

In this tree, NP-1 Molly and NP-2 Mary are both in argument position and they both C-command the object of hated, NP-3, and therefore are coindexed with it. Therefore, NP-3 is A-bound. Yet the coreference reading proper of A-binding is ruled out. So, A-binding alone does not explain all the options for filling NP-3. Why must we use her and not herself to refer back to Molly, while just the opposite is true with respect to Mary? We still need conditions to rule out the ungrammatical cases. The Principles of Binding Theory determine whether a pronoun or a reflexive anaphor is correct in a particular position. (29)

Principles of Binding Theory A. Anaphors (e.g. reflexives and reciprocals) must be A-bound in their smallest clauses. B. Pronouns must not be A-bound in their smallest clauses. C. Full NPs (also called denoting expressions or R (eferential)expressions) must not be A-bound.

• Principle A says that an anaphor can only be used when the position that A-binds it is within the same clause. In tree (25), NP-1 John is the subject and NP-2 himself the object so that the anaphor himself correctly appears bound by the subject. On the other hand, the

AN OVERVIEW OF GENERATIVE GRAMMAR

59

NP-1 Molly in tree (28) is too far away from the embedded anaphor herself in NP-3, embedded in a different clause with a different subject: the NP2 Mary. • Principle B says that a pronoun can only be used if it is not A-bound at all, or if its A-binder is far enough away. This is why her cannot be used in NP-3 to refer back to NP-2 Mary but her may refer back to NP-1 Molly in (28). • Finally, Principle C says that r-expressions may not be A-bound at all. This is to rule out repetition of full nominals. (30)

*Mollyi hit Mollyi.

(31)

*Mollyi thought that Maryk hated Mollyi/Maryk.

Defining this local domain that requires an anaphor and cannot have a coreferent pronoun has been problematic. We saw above that a basic generalization is that the antecedent and the anaphor must be in the same clause. This works for most cases, but there are a few exceptions in the literature (e.g., Sally is eager for herself to succeed (Black 1998)) where the anaphor and antecedent are not in the same clause. The local domain is therefore defined in terms of A-binding, since most anaphors that have antecedents are subjects within the same clause.

4.6. Transformational Syntax and Movement In the generative tradition, there has been a deep interest in explaining the different word order found with declarative, negative and interrogative sentences. At the beginning, it was the so-called transformational grammar that attempted to analyse all those structures. The term «transformation» implied that we should bear in mind two different structures: (1) the basic or deep structure, on which the transformation applied to provide (2) the derived or surface structure. The development of the transformational grammar resulted in the 1980s with a different theory, the so-called Move ALPHA theory, according to which constituents move along the sentence to respond to certain syntactic configurations. In the following paragraphs, we will see what these configurations are as we develop the analysis of negative sentences and question formation involving both Yes/No questions and wh-questions.

60

CURRENT TRENDS IN LINGUISTIC THEORY

4.6.1. Negation as a Functional Head In the GB model, many analyses have been given to account for all possible ways to express negation in English. For the purpose of this brief introduction, we will exclusively centre on the analysis of negation in finite sentences. Consider the following sentences: a)

John doesn’t play tennis well

b)

John would not go to the party, if he was sick

c)

*John plays not tennis well

d)

*John not would go to the party, if he was sick

One of the well known properties of English is that unlike other languages, lexical verbs cannot be followed by negation. Sentences c) and d) above are in fact impossible. The only possible way for finite sentences to include verbs with negation is by inserting auxiliaries preceding the negation, as illustrated by the well-formed sentences a) and b) above. In the Xbar-schema adopted here, the word «not» is assumed to fulfil the functional category Neg that projects into the maximal projection NegP. To capture the data above, NegP must appear between IP and VP. Once more, subcategorization is responsible for the selection of VP, and the lexical entry for the word «not» is as follows: not, Neg [ _ VP ]. Further we must assume that I [+fin] is filled with the auxiliary in the D-structure or canonical structure. The corresponding tree for the sentence a) is next exemplified: a)

IP NP

I' I [+fin] NegP does Neg' Neg not

VP V' V

NP

AdvP

play

tennis

well

AN OVERVIEW OF GENERATIVE GRAMMAR

61

As for the contraction of the auxiliary in the «doesn’t» form, we follow the so-called MOVE ALPHA developed in late 1980s and early 1990s and assume that Neg moves into I, by which the word «not» incorporates into the auxiliary: …

I' I [+fin] doesn't

NegP Neg' Neg



In the GB tradition, head movement has always been a good way to deal with the interaction between morphology and syntax. In the following section, we will analyse other cases of head and phrase movement in the analysis of question formation in English.

4.6.2. Question Formation So far we have analysed NPs in their canonical position. In the following two sections we will study sentences in which NPs are found in different positions from the ones they are assumed to be inserted in the tree. We will, in particular, study the movement of heads and phrases in two types of configurations: yes-no questions and wh-questions.

4.6.2.1. Yes/No Questions To form a Yes/No question in English, an auxiliary is moved in front of the subject, as in the change from (32a) to (32b). The distribution in (32c-f) shows that we need to be careful in how we formulate this movement rule to be sure that only the grammatical Yes/No questions are generated. For example, only one auxiliary can move (an auxiliary only, not a main verb), it has to be the first auxiliary, and that first auxiliary agrees with the subject and determines the form of the following verb. (32)

a.

Molly has rejected the offer.

b.

Has Molly rejected the offer?

c.

*Has rejected Molly the offer?

62

CURRENT TRENDS IN LINGUISTIC THEORY

d.

*Has Molly might reject the offer?

e.

*Might have Molly rejected the offer?

f.

Might Molly have rejected the offer?

Sentences like: (33)

Peter could have been calling Molly

show that auxiliaries come in a certain order and that they subcategorize for the form of the verb that comes after them. This means that auxiliaries must be heads themselves, each subcategorizing for a VP of a certain type. For example, could, might, shall, etc. are modals which may not be followed by another modal and require the verb that comes after them to be in the bare form. The lexical entry for could is given in (34). Similar entries could be given for the other modals and for the non-modal auxiliaries. could V[+aux /+modal] [ _VP[-modal /+bare] ]

(34)

The structure for (33) has at first sight four VPs, stacked one upon the other, as shown in (35). (35)

IP I'

NP Peter

I [+fin]

VP V'

V[+aux/+bare] VP [+bare] could

V' V[+aux/+bare] VP [+en] have

V' V[+aux/+en] been

VP [+ing] V' V[+in] calling

NP Molly

With a movement analysis, we can account for the stacking of VPs via subcategorization, since at the level of the lexicon, the subcategorization

AN OVERVIEW OF GENERATIVE GRAMMAR

63

is responsible for the restrictions on the categories that a verb can have as a complement. Interestingly, only the first verb (auxiliary) moves in the yes-no question (36), into the highest head position: (36)

Could Peter have been calling Molly?

There are two main questions that raise here. The first question is whether modals are like lexical verbs and as such as also inserted in V. The second question is which the initial position of auxiliaries in yes-no questions is. In the first place, in traditional grammar modals and auxiliaries are distinguished. Here too, we want to assume that in English modal verbs such as «could» or «must» are of category I whereas auxiliary verbs such as «have» and «be» are inserted in V. Then, in the previous sentence «Peter could have been calling Molly» the first verb which is the modal must be inserted in I followed by the rest of the stacking verbs. IP NP Peter

I' I

VP

could V' V have …

The fact that the modal auxiliary «could» appears to the left of the subject in the question (36), in what is usually known as auxiliary inversion, may be then analysed as an instance of verb movement. In this case the modal «could» moves to a position higher than the matrix subject suggesting that we are in front of a CP structure. As is standard practice, we will assume that the Complementizer head (C) contains a feature [+qu] that is exclusively active in questions. In effect, the derived structure after the movement of the auxiliary to the front is as in (37), where the arrow shows how the auxiliary moves to the Complementizer with the [+q] feature proper of questions, through I [+fin]:

64

CURRENT TRENDS IN LINGUISTIC THEORY

(37) CP C' C [+q]

IP

Could NP

I'

Peter I [+fin] --

VP V'

V[+aux/+bare] VP [+bare] have

been calling Mary

This double movement of V [+aux] to I [+fin] and to C [+q] is called HEAD MOVEMENT, since a head is moving to other head positions. This movement gives us the word order of a Yes/No question. In cases of auxiliaries like «be» in sentences like «he’s crying» we may assume that the «be» form «is» is inserted in V, although it may move into T in its way to C in a main yes-no question such as «Is he crying?» as illustrated by the following derived structure: CP C' C is

TP NP he

T' T

VP V' V [+aux]

VP V' V crying

AN OVERVIEW OF GENERATIVE GRAMMAR

65

Other different but related question is why the modal or auxiliary does not move to the front in English embedded questions, as illustrated by the following example: (38)

I wonder whether Molly has rejected the offer

On the assumption that clauses are CPs, we may assume that the embedded clause following wonder is a CP. Semantically, both main clauses Yes/No questions and embedded Yes/No questions have the same interpretation; one only requires a more direct answer than the other. Drawing on this parallel between main clauses and embedded clauses, we can posit that main clauses are also CPs, and that the CP and its head C for both main clause and embedded clause questions have the feature [+q]. This meant the derivation in (37), where, as the arrows indicate, after the V [+aux] moves to I, it keeps moving on to C [+q] to obtain the word order of Yes/No questions. While we want to assume that the highest V[+aux] moves to I in all cases, the movement of I to C [+q] only occurs in questions, since declaratives and other non-questions will have a C [-q]. As for how we may account for the fact that no Subject/Aux inversion occurs in these cases, the claim is that as tree in (39) shows, the embedded C [+q] position is filled with whether, and this position is no longer available for the auxiliary. Note, in addition that the main C head does not either contain any [+q] feature, since it is a declarative sentence:

66

CURRENT TRENDS IN LINGUISTIC THEORY

(39) CP [–q] C' CP [–q] NP I

IP I' I [+fin] VP V' V

C [+q]

wonder C' C [+q/–root] IP whether

NP

I'

Molly I [+fin] VP V' V [+aux] VP has

V' V

NP

rejected the offer

4.6.2.2. Wh-questions In the movement theory, wh-questions are understood as those questions which include a wh-word (who, what, whom, whose, when, where, how) in the front. Semantically, this type of questions ask for more information than a «yes» or «no» answer. Like with yes/no questions, in wh-questions auxiliaries also undergo head movement to the root C head also marked with the [+q] feature. The difference between Yes/No questions and wh-questions is that an additional

67

AN OVERVIEW OF GENERATIVE GRAMMAR

movement takes place in the latter type of questions: a [+wh] phrase moves to the front. For example: (40)

Which offer has Molly rejected?

Since there is an open specifier position in the CP [+q], the wh-phrase which offer can move to that position, as shown in the tree in (41). This movement is called Abar-movement, which is movement to a non-argument position. (41)

CP [+q] Spec C' CP [+q]

IP NP

I'

Molly

I [+fin]

VP V' V[+aux] VP [+en] has

V' V[+en] rejected

NP [+wh] D [+wh] N' which

N offer

4.7. A-movement In the last section, we have learnt the syntax of the two main types of questions in English: yes-no questions and content questions with initial wh-words. We argued that they undergo movement of the auxiliary or the wh-phrase to the front in each case. In this section we will see other type of phenomena that involve movement of NPs to the left. This is known as A(rgument)-movement: movement of arguments to other positions, like the subject position. The cases of A-movement to be considered include: passives, unaccusatives, and raising constructions.

68

CURRENT TRENDS IN LINGUISTIC THEORY

4.7.1

Passives

As is well known in all generative grammar approaches, passives involve movement to the subject position. Since the beginning, the theory captured the traditional observations about passive constructions: (1) Most transitive verbs have passive alternates; (2) No intransitive verbs have passive alternates; and (3) The ‘subject’ of a passive verb corresponds to the object of its transitive alternate. In addition the Move ALPHA theory attempts to explain why the active and passive sentences share the same thematic distribution of arguments, regardless their syntactic position in the sentence. Consider the following examples: (42)

a.

John introduced Mary to Peter.

b.

Mary was introduced to Peter by John.

There is a sharp contrast between the examples (42a) and (42b). While the thematic subject, the NP John is only present in the initial position in the first example (42a), the NP John, in the second example (42b) follows the preposition «by» and appears in the rightmost position of the sentence. In contrast, it is the thematic object, the NP Mary, which appears in the subject position.

4.7.2. Thematic roles In section 3 we argued that there is a subcategorization principle active in the lexicon, according to which each verb selects all its arguments specified with a certain thematic role: agent, theme, recipient, goal, instrument, etc. In particular the lexical entry of a verb like «introduce» is as follows: (43)

introduce, V [__ NP (PP [to])] introduce’

First of all, the lexical entry of «introduce» include two internal arguments but no external argument. Secondly, the semantic roles assigned for a verb like «introduce» are: Agent, theme and goal. These theta roles will be matched in the syntax level with the corresponding grammatical relations, viz: subject, direct object, and indirect object. As seen in section 3.3, internal arguments are sisters to the head, as shown in the following tree for the declarative sentence (42a).

AN OVERVIEW OF GENERATIVE GRAMMAR

69

IP

(44) NP Jhon

I' I [+past]

VP V' V

NP

PP [to]

introduced Mary

to Peter

According to the formalized view of thematic roles of the Uniformity of Theta Assignment Hypothesis (Baker 1988:46), there is a one-to-one correspondence between the position of the argument and its theta role. For example, the THEME role is always assigned to the direct object when it is present and the RECIPIENT role is assigned to the indirect object, etc. Since the assignment of theta roles takes place in the lexicon, the movement of an argument, say the object, into the subject position, does not affect the thematic roles already assigned. The semantic role stays with the original position rather than moving with the phrase because the semantic role is not part of the tree but part of the lexical subcategorization. The full lexical entry for the passive form introduced, including semantic roles, is: (45)

introduced V [+pass] [_ NP (PP [to]) (PP [by])] introduced

Note that no external argument is assigned by the passive verb. The canonical structure for the passive sentence (42b) is given in (46) below, where the passive auxiliary verb is inserted from the lexicon as the only head which subcategorizes for a VP [+pass]. In this canonical structure, the object NP Mary remains in its object position. In the resulting passive configuration, it is however moved into the subject position. This Amovement can take place since there is no semantic role linked to the subject position. A coindexed trace is left behind to maintain the mapping of the object to its semantic role.

70

CURRENT TRENDS IN LINGUISTIC THEORY

(46)

Canonical structure IP I' I [+past]

VP V' V' VP[+pass] was V' V' V [+pass] introduced

NP

PP [to]

PP [by]

Mary

to Peter

by John

So the derived structure looks like (47): (47)

Derived structure after A-movement IP NP

I'

Maryk

I

VP

wasi

V' V

VP[+pass]

ti

V' V' V [+pass]

introduced

NP

PP [to]

PP [by]

tk

to Peter

by John

The movement of either the NP Mary to the subject position or the V was into the Inflectional head leave behind two traces in their canonical positions, each bearing a different index. As is common in English, the auxiliary part of the verb also moves into the inflectional head I. The reason

AN OVERVIEW OF GENERATIVE GRAMMAR

71

why the thematic object moves into the subject position, that is into Spec, I, is to enter into an agreement configuration with the verb in I by which it receives Nominative case, so the thematic object, the NP Mary is, in fact, the syntactic subject. In the sentences above, we see the double object construction similar to the analysis of verbs like «John gives a present to Mary», among the simplest ones. In traditional analyses, both objects were analysed at the same level of the verb and as a consequence we find three branches coming down from V’ here. However, in this paper we have followed binary branching and that’s why a recursive V’ rule has been used.3 Chomsky (1995) analyses the double object construction within a VP-shell (that is with two Vs, a lexical v and a functional V) and he manages to include both NPs as verbal complements. The issue of the double object construction is briefly discussed next.

4.7.3. The double object construction In this section we want to discuss the double object construction exemplified in a sentence like «John gave Mary a book» because in the recent literature it has been analysed as an instance of A-movement (Den Dikken 1995). According to this analysis, the phrase «Mary» in the previous sentence contains an empty P and has been moved to the left of the verb to get Case. Other linguists have analysed the double object construction within a VP shell. For example Larson (1988) posits the following structure: VP V

VP NP2

V' V

NP1

This structure explains the word order in the cases under discussion on the assumption that the embedded verb raises to the empty verb to 3 As Haegeman (1994, footnote 143) points out, the binary branching hypothesis (Kayne, 1984) raises important questions such as which is the structure of double object patterns in examples such as «John gave Mary the money» . In order to be consistent with binary branching one has to assume that the first NP «Mary» following the verb is the complement, i.e. sister to V, whereas the second NP «the money» is an adjunct, i.e. sister of V’. However Larson (1988) proposes an alternative analysis which we will discuss in the following section.

72

CURRENT TRENDS IN LINGUISTIC THEORY

provide NP2 with Case. In addition, it may also explain the binding facts Shei gave eachj girl herj/*i ticket where the pronoun «her» can only be interpreted as bound by the C-commanding quantifier «each girl». In this sense, Larson’s (1988) analysis allows one to distinguish the two objects structurally independently of how they are assigned Case. Its more articulated structure in addition provides a way to distinguish objects with different theta-roles. In the previous case under discussion: «Mary» as benefactive and «her book» as theme.

4.7.4. Unaccusatives In the lexicon, transitives are distinguished from intransitives, since they are subcateogorized in a different fashion. Following the Unaccusative Hypothesis (Perlmutter 1978), a difference should also be established between verbs that do not select for an object like breath, die and break in (48a), (48b) and (48c) respectively: (48)

a.

Mary breathes well

b.

Mary died

c.

The window broke

d.

The child broke the window

In (48a), breath is assumed to be a regular intransitive verb with an AGENT subject. In contrast, Mary did not do anything to make herself die, so in (48b) the subject is assumed to have the THEME role. Similarly, in both (48c) and (48d) the window is what the breaking happened to so it fills the THEME role in both the unaccusative construction in (48c) and the transitive construction in (48d). The assumptions about semantic roles require that while verbs like breath have an AGENT in subject canonical position, verbs like die have an empty subject in their canonical position. Further, a verb like break has an optional AGENT. Movement of the THEME to subject position for the unaccusative verb die is exactly parallel to the passive movement, as shown in (49). sleep, V [ _ ]

die, V [ _ NP ] break, V [ _ NP ]

sleep’

die’

break’

AN OVERVIEW OF GENERATIVE GRAMMAR

(49)

73

a) canonical structure S-structure (before A-movement of the NP Mary) IP I'

I [+past]

VP V' V

NP

die Mary

b) derived structure (after A-movement of the NP Mary) IP NP Maryi

I' I [+past]

VP V' V

NP

die

ti

4.7.5. Raising Finally, we may analyse raising constructions as the third type of Amovement. Raising constructions contain the English verbs such as «seem» and «is likely». These predicates take either a finite or a non-finite clause complement and do not assign a semantic role to the external argument. The lack of a semantic role assigned to the subject position can be seen by the presence of the dummy it when there is a finite clause complement (50a). In the case of a non-finite clause complement, the subject of the lower clause must raise to the main clause subject position (50b-c). (50)

a.

It is likely that Mary wins.

b.

Mary is likely to win.

c.

*It is likely Mary to win.

74

CURRENT TRENDS IN LINGUISTIC THEORY

The trees of the canonical structure and the derived structure of the sentence (50a) are given in (51a) and (51b) below. IP

(51) a.

I' I [+fin] VP V' V [+aux] AP is

A' A

CP

likely

C' C

IP

that NP

I'

Mary I [+fin] VP V' V wins

b. Derived structure after insertion of expletive «it» IP NP It

I' I [+fin] VP isi V' V [+aux] AP ti

A' A

CP

likely

C' C

IP

that NP

I'

Mary I [+fin] VP V' V wins

AN OVERVIEW OF GENERATIVE GRAMMAR

75

For sentence (50a) to be grammatical we have inserted dummy «it» in the subject position in the derived structure in (51b). Contrast the trees in (51) for sentence (50a) with the trees provided in (52) for sentence (50b), where the subject of the non-finite clause raises to the main clause subject position. (52)

a) canonical structure IP I' I [+fin] VP V' V [+aux] AP is

A' A

IP [–fin]

likely NP

I'

Mary

I

VP

to

V' V win

b) derived structure after A-movement IP NP

I'

Maryk I [+fin] VP isi V' V [+aux] AP ti

A' A

IP [–fin]

likely NP

I'

tk

I

VP

to

V' V win

76

CURRENT TRENDS IN LINGUISTIC THEORY

There are more reasons than just a requirement for the main clause subject position to be phonetically filled. For the purpose of this introduction course, we may just argue that Case Theory provides the motivation for A-movement of the particular NP which moves in passives, unaccusatives, and in raising configurations. SUMMARY This section has been designed to provide an introduction to the main tenets of the Government and Binding (GB) theory of syntax developed for the Principles and Parameters approach to language by Chomsky (1981, 1982 1986). In particular, we have explored the basics of the X-bar schema and studied the structural relations which are crucial for both the Case theory and the Binding principles. Further we have analysed different instances of movement. In sum we hope you have learnt: (1) the basic properties of trees, (2) how to draw tree-diagrams according to the Xbar-schema, and (3) identify the position of each constituent in both their canonical and their derived structures after the movement of heads or phrases in negative or interrogative sentences and in A-movement configurations.

NOW YOU ARE READY TO DO EXERCISES FROM 19 to 43

5. RECENT MODIFICATIONS ACCORDING TO THE MINIMALIST PROGRAM Early generative grammar seemed fairly to give answer to two immediate problems reviving the concerns of a rich tradition represented by linguists like Otto Jespersen: (1) to find a way to account for the phenomena of particular languages («descriptive adequacy») and (2) to explain how knowledge of these facts arises in mind of the speaker-hearer («explanatory adequacy»). The efforts in giving answer to these two questions culminated in the Principles and Parameters model (see Chomsky 1981 for one formulation). The P&P approach maintains that the basic ideas in early generative grammar are misguided in principle – in particular, the idea that a language consists of rules for forming grammatical constructions (relative clauses, passives, etc.). The P&P approach held that languages have no rules in anything like the familiar sense, and no problems of theoretically significant grammatical constructions except as taxonomic artefacts. There are universal principles

AN OVERVIEW OF GENERATIVE GRAMMAR

77

and a finite array of options as to how they apply (parameters), but no language-particular rules and no grammatical constructions of the traditional sort within or across languages. For each particular language, the cognitive system, we assume, consists of a computational system CS and a lexicon. The lexicon specifies the elements that CS selects and integrates to form linguistic expressions – (Phonetic Form, PF; Logical Form, LF) pairings, we assume. The lexicon should provide just the information that is required for CS, without redundancy and in some optimal form, excluding whatever is predictable by principles of UG or properties of the language in question. Virtually all items of the lexicon belong to the substantive categories, which we will take to be noun, verb, adjective, and particle, putting aside many serious questions about their nature and interrelations. The other categories we will call functional (tense, complementizer, etc), a term tat need not be made more precise at the outset, and that we will refine as we proceed (Chomsky, 1995: 6).

In this context, language acquisition is interpreted as the process of fixing the parameters of the initial state in one of the ways that UG permits. A specific choice of parameter settings determines a language in the technical sense: an I-language in the sense of Chomsky, where I is understood as «internal», «individual», and «intentional». In order to retake the early problems of inquiry and also under the impact of new empirical material and theoretical ideas, Chomsky, in the middle of the 1990s, initiates a new program of linguistic research based on some minimalist assumptions. Under the name of the Minimalist Program, he proposes some modifications to get rid of the redundancies found in the development of concepts and principles such as D-structure; S-structure; government; the Projection Principle; the Theta-Criterion; other conditions held to apply at D- and S-structure; the Empty Category Principle; X-bar theory generally; the operation Move; the split-I hypothesis; and others. In the previous sections we did not deal with all these concepts and principles and, therefore, this section covers only the modifications proposed in relation to those discussed so far by us. First, we redefine the movement theory according to the feature-checking mechanism. Second, we discuss several ways of dealing with the interaction between morphology and syntax through the incorporation of features in the computational system, and finally, we deal with the basics of bare-phrase structure. As usual, the chapter ends with a section of exercises where the student can practise on these modifications.

78

CURRENT TRENDS IN LINGUISTIC THEORY

5.1. Movement and checking theory In section 4 we analysed the movement theory in the PP approach, according to which we considered three types of movement transformations, as provided in (1): (1)

a)

Head movement: Auxiliaries to the front in yes-no questions;

b)

Phrase movement: wh-phrases to the front in wh-questions;

c)

NP movement: passives, unaccusatives and raising.

Each type of movement was triggered by different syntactic constraints. Yet, the landing site of either the head or the phrase was related to a functional projection. In the case of auxiliaries moving to the front, we postulated the presence of two functional heads I and C to where the auxiliary moves following a locality constraint: movement to C through I. In the case of wh-phrases or NPs, they both move to the Specifier of a local functional projection: wh-phrases move to the closest Spec, C and NPs to the closest Spec, I. If this is the case, then there isn’t really a significant difference between the movements previously illustrated. We might unify them into a single rule motivated by a single constraint. This is in fact what the Minimalist Program attempts to do. The next step is then to define the exact constraint that motivates and forces the movement transformation in each case. In brief, all movement transformations can be reduced to the checking of a particular feature. To illustrate, when we discussed the movement of wh-phrases or auxiliaries in questions we already mentioned the presence of a [+qu] feature related to the functional projection CP. In the minimalist program, this feature is renamed as the [WH] feature which usually appears in the complementizer and attracts the wh-phrase with the same feature [WH]. Once the wh-phrase moves, the [WH] feature is checked. Checking of features can only occur in a local configuration. In the case of wh-movement. The checking is derived from a Spec-head configuration as illustrated in (2): CP

(2) NP

C'

N'

C

N [+WH] [+WH]

AN OVERVIEW OF GENERATIVE GRAMMAR

79

As already mentioned, the movement of the wh-phrase is forced by a constraint which in the Minimalist Program is called the Principle of Full Interpretation (Chomsky 1993, 1995). In a very simplified version Carnie (2002, p.316) defines it as follows: (3)

Full Interpretation (FI): Features must be checked in a local configuration

(4)

Local configuration [WH] features: Specifier/Head configuration

In a similar way we can capture the other cases of movement above. First, we have a second instance of Spec/Head configuration in the cases of NP movement. In section 4, we argued that the movement of NPs to Spec, I was due to Case reasons. In particular, we observed that NPs moved into that position received Nominative Case. Now we may reformulate this by assuming the presence of a [NOM] feature in I which attracts those NPs holding the same feature. The resulting feature checking can be illustrated as in (5). IP

(5) NP

I'

N'

I

N [+NOM]

[+NOM]

Finally, we can account for head-movement transformations in a similar way. In section 4, we also mentioned the presence of some features like [+past] or [+finite] defining the functional I position. Now, we may assume that verbs are incorporated in the syntactic tree with similar features. When the verb and I check their corresponding features against one another we may interpret the verb as past or finite, for example. The local configuration in these contexts is within the head itself and is called headto-head configuration. (6)

IP I' I [+finite]

VP V' V [+finite]

(checking configuration)

80

CURRENT TRENDS IN LINGUISTIC THEORY

In instances of yes-no questions or wh-movement we also argued for the presence of a [qu] feature. In a similar fashion we can assume that both I and C bear a similar [WH] feature which need to be checked one against the other. This will finally motivate the movement of auxiliaries to the front in those cases, also responding to a head-to-head checking configuration as next illustrated: (7)

CP C' C [+WH]

IP I' I [+WH]

(checking configuration)

Within this new approach, we have a very simplified transformational system more accord with economy of language: a wanted result for the Minimalist Program. Notice that now there is only one transformational movement rule (instead of three).

5.2 Interpretable and uninterpretable features In the previous section we have characterised a number of features that must be checked in an appropriate manner according to minimalist assumptions. In this section we will focus on the type of grammatical features that characterize the morphological properties of words. In addition we will retake the principle of Full Interpretation already defined in the previous section to distinguish between interpretable and uninterpretable features. Most of the arguments that follow are originally to be found in Radford (1997, chapter 5). According to minimalism, phonetic, grammatical and semantic properties of words can be described in terms of sets of features. The work to be done by the linguist is to find a grammar that accounts for all of them. The Minimalist Program puts forward a grammar with only two levels of representation: the Phonetic Form (PF) and the Logical Form (LF). By assumption it is in the nature of both PF and LF representations that each contains only phonetically or semantic interpretable features in turn, as required by the UG constraint of Full Interpretation under Radford’s (1997, p.171) adopted version:

81

AN OVERVIEW OF GENERATIVE GRAMMAR

If a derivation results in a PF representation of an expression which satisfies PFI (and hence contains only phonetically interpretable features), it is said to converge at PF; if the derivation results in an LF representation which satisfies PFI (and hence contains only semantically interpretable features), it is said to converge at LF. If both the PF and LF representations for some expressions satisfy PFI, the associated derivation is said to converge (and the relevant expression is grammatical). If the PF and/or LF representations of an expression violate PFI, the resulting derivation is said to crash (and the relevant expression is ungrammatical (cf. Radford, 1997: 171).

In addition to interpretable features, there are other grammatical features which are uninterpretable because they cannot be read off by the PF or LF components. Rather, they are exclusively responsible for certain syntactic operations proper of the computational system. But how do the word features enter into the computational system of the grammar? In the first place, lexical items (comprising sets of phonetic, semantic and grammatical features) are taken from the lexicon by an operation of selection. Next, constituents are combined together in a pairwise fashion to form a phrase structure tree by the process of merger. After spell out, both PF and LF operations are responsible for the process of phonetic and semantic features. All uninterpretable features are invisible after spell out, presumably because they have been checked properly in the syntactic derivation. The resulting model can be represented as follows: Interpretable features LF operation (8)

phrase formation

spell out PF operation Interpretable features

Leaving aside interpretable features by the PF or LF operation, we should define a bit more what we understand by uninterpretable grammatical features. As Radford (1997) points out, Grammatical features include number (singular/plural) features, since they play an obvious role in the syntax of agreement (cf. these/*this books). They also include gender (masculine/feminine/inanimate) features, since they play a role in the syntax of reflexive anaphors (cf. He/*She/*It turned himself into a giant). Likewise, they include person features, which play a role in the syntax of subject-verb agreement (cf. He/*I/*You likes syntax). They also include features which determine the morphological form of items – for example the case features of pronouns (cf. He/*Him likes me/*I), or the inflectional features of verbs

82

CURRENT TRENDS IN LINGUISTIC THEORY

(cf. He has gone/*go/*going). But they do not include features which have no morphological or syntactic correlate: for example although words such as calf, foal, kitten, puppy, etc. share the feature [young], this is a purely semantic feature which plays no role in any grammatical process, and so is not a grammatical feature.

In Chomsky (1995), interpretable features are relevant for LFinterpretation and include categorial such as [+V, +P, +A] and nominal phi-features [+fem, +masc, +sing, +plur]. They are not deleted or erased after they are checked because they are relevant to the interpretative component. Non-interpretable features are deleted and they involve the case features of NPs and verbs and the phi-features of verbs. There are a number of reasons behind the distinction. Some features (e.g. phi-features of NPs) are said to remain visible after checking and hence cannot be deleted. This is the reason an NP can move cyclically and provide the phi-features along the way (Chomsky 1995: 282f). In languages where subjects agree with both the finite verb and the past participle or predicative adjective such as French, this is evident. The reason for abandoning the V- and N-features of early Minimalism in favour of categorial, Case and phi-features is that the Extended Projection Principle effects (i.e. that clauses have structural subjects) are accounted for by means of a strong D-feature in I (Chomsky 1995: 232). This is necessary since Minimalist trees do not automatically project a Specifier position along with the introduction of bare phrase structure and Spec IP must somehow be present. Hence, the assumption that a Dfeature exists in I. Wh-movement is triggered by a D-feature in C rather than by a [wh]-feature. The argument that it is the D-features that trigger movement comes from expletives. Finally, if we assume that the principle of Full Interpretation adopted by Radford (1997) is correct, and PF or LF representations contain only semantically interpretable features, it follows that uninterpretable features must somehow be eliminated in the course of deriving each representation, for the derivation to converge at LF. In fact we previously suggested a mechanism for eliminating those features, namely by means of checking. This is the topic of the next section.

5.3 Checking other grammatical features As we argued before, the Spec-Head relation is the primary means of expressing checking relations in the Minimalist Program. Then we saw

AN OVERVIEW OF GENERATIVE GRAMMAR

83

that the [WH] feature of wh-phrases must enter into a checking relation with the [WH] feature of the matrix CP. The question we want to address here is against which functional projection other grammatical features enter into a similar checking relation. First of all, we want to address the question of where the number (singular/plural) features of nouns are checked. Next, we will address the question of the case features of nouns. Adopting Abney’s (1987) DP-hypothesis for nouns, the Minimalist Program assumes that NPs are part of a DP structure when they are internal arguments. This explains why the grammatical feature of number of nouns can be checked against a functional category, capturing the agreement facts illustrated in (9): (9)

(10)

a

*This books

b

These books DP D'

D N' These N books

In the partial derivation of (10) the number feature of determiners against nouns is checked in situ. As for the Case feature of Nouns, a Move operation is however in place. Within the set of assumptions of the Minimalist Program, we assume that the operation Move –later reformulated by Chomsky (1995, chapter 4) as Attract – must meet certain economy conditions. For our present discussion, we want to focus on the Minimal Link Condition (MLC) (Chomsky and Lasnik 1993; Chomsky 1994), inspired by Rizzi’s (1990) Relativized Minimality. The MLC forces the shortest move and is inviolable. That is, given several potential landing sites, the closest must be chosen or else the derivation crashes. Chomsky (1995) defines the operation Attract/Move as follows: «K attracts α only if there is no β, β closer to K than α, such that K attracts β». In this formulation, K is the target and α is the Case feature of Nouns. In brief, we will only account here for two types of Case feature to be checked in the derivation, namely, Nominative and Accusative.

84

CURRENT TRENDS IN LINGUISTIC THEORY

In the first place, we follow Chomsky (1995) in assuming that the functional projection in charge of the checking of the Nominative Case feature is Tense Phrase (TP), substituting old IP. In this revised version, the subject of the sentence in (11) base generated as an internal argument of the VP in the partial derivation (12a) is attracted by the closest matrix TP as illustrated by the final derivation in (12b). (11)

John kissed his wife TP

(12) a)

T' T

VP DP John

V' V

DP

kissed

his wife

TP

b) DP Johni

T' T+V kissed

V' ti

V' e

DP his wife

In the final derivation (12b), the NP John moves into the Spec of T and enters into a checking relation with the verb raised into T under SpecHead agreement. As a result the Nom feature which is uninterpretable is deleted and the derivation converges. Following Hale and Keyser (1993a, b) and Chomsky (1995), we can assume that transitive accusative verbs have the following configuration: (13)

vP DP (subject) v' v

VP V DP (object)

AN OVERVIEW OF GENERATIVE GRAMMAR

85

This structure allows the morphological demands of the transitive verb (V) to be differentiated from those of v. The general idea is that v is to the morphological shape of the Case of the object what Tense is to the morphological shape of the Case of the subject. The proposal is then that while the DP subject moves into Spec, T, the DP object moves into Spec, V to enter into a Spec-head relation with the Verb in v, and as a result the uninterpretable Accusative feature is deleted for the derivation to converge. Notice that this short move operation doesn’t violate the already discussed MLC proposed by the Minimalist Program, since the functional projection vP is closer to the DP object than TP.

5.4. Bare phrase structure Another modification brought to the theory that attempts to get rid of redundancies in the structure is the use of bare phrase structures. This attempts to simplify the system, another wanted result of the Minimalist Program. According to Chomsky, the bare phrase structure theory departs from conventional assumptions in several respects: in particular, categories are elementary constructions from properties of lexical items, satisfying the inclusiveness condition. In the bare phrase theory, there are no bar levels and no distinction between lexical items and «heads» projected from them. We thus dispense with such structures as (14a) with the usual interpretation: the, cat taken to be terminal lexical items and D +, N+ standing for whatever properties of these items are relevant to further computation (perhaps the categorical information D, N; Case; etc.) In place of (14a) we have only (14b): (14) a.

DP D+

NP

the

N' N+ cat

b.

the the

cat

86

CURRENT TRENDS IN LINGUISTIC THEORY

A consequence of this is that an item can be both an X or a XP. Radford (1997) puts an example of how the bare phrase theory will work for a sentence such as «She has gone». Within the framework put forward in the previous sections, such a sentence will have the grammatical structure (15) below: TP

(15) DP

T'

She T [3FSNom]

VP

has gone [Pres] [+n] [3SNom]

As Radford argues, there is a potential inconsistency in the structure of (15): both the category levels and the words themselves represent different grammatical features. For example the fact that the verb has belongs to the categorical status of an auxiliary is indicated by the label T attached to the terminal node carrying the word has, whereas the fact that has is present, third person singular, and assigns Nominative is indicated by the bracketed [Pres], [3SNom] head features carried by has. To eliminate this notational inconsistency, we should incorporate the categorical properties of words into the head-features of the relevant items. According to this modification, terminal nodes will no longer carry category levels. The same could be assumed for nonterminal nodes. In fact, Chomsky (1995) suggests that the only information which we need about the constituents represented by nonterminal nodes is what the head of each such constituent is. Adopting this minimalist view, the corresponding structure for (15) is as in (16), where both terminal and nonterminal nodes are represented by words, the former specified by their features: (16)

has She [3FSNom D]

has

has gone [Pres T] [V+n] [3SNom] [+n]

In the structure (16), both has gone and she has gone are projections of has. By getting rid of labels such as X/X’/XP we fulfil the economy

AN OVERVIEW OF GENERATIVE GRAMMAR

87

principle of the Minimalist Program which proscribes the use of superfluous symbols in structural representations. What remains important in the system are the hierarchical relations established in the tree. The structures stipulated in earlier versions can be reformulated in elementary terms satisfying minimalist conditions, with no objects beyond lexical features. SUMMARY What the Minimalist Program has brought to the practice of the current theory is to work with: • features and the way they are checked in the structure; • rather than with category levels. This is the way to reduce the complexity with which the old Xbar theory provided the computational system.

NOW YOU ARE READY TO DO EXERCISES 44 and 45

6. SUGGESTED ACTIVITIES The following exercises are intended as practice of the theoretical issues covered in the previous description of generative grammar. A. Methodological underpinnings 1. Answer the following questions 1. How would you define the language faculty? 2. What unites language acquisition with foreign language acquisition? 3. What is the difference between ‘language’ and ‘a language’? 2. What general principles can be derived from the comparison of the following sentences in English. Recall that the use of a star in front of a sentence signals ungrammaticality. (1) a. John lives in big cities and has experience with drivers b. *John lives in a big city because has experience with drivers (2) a. The man who lives next door is Japanese b. *The man lives next door is Japanese

88

CURRENT TRENDS IN LINGUISTIC THEORY

c. The man who you met is Japanese d. The man you met is Japanese 3. What is the difference between the pronoun «him» and the anaphor «himself» n the following examples? Do both pronouns «him» and anaphor «himself» refer to the same person «John» in each case? (1) John looks at himself in the mirror (2) John looks at him in the mirror 4. Say whether the following statements are true or false according to the generative perspective. When learning a language the child is expected… True 1. 2. 3. 4.

False

to know formal grammar to learn from uncontradictory data to start from an initial state to imitate any sentence

5. Which definition of Universal Grammar is correct: 1 «the system of principals, conditions, and rules that are elements or properties of all human languages…the essence of all human languages.» 2 «the grammar that describes all languages in the world» B. Competence and Language Acquisition 6. Give your reasons for your positive or negative answer to the following question: If children were born with basically a clean slate, as believed by «environmentalists» (those who believe that language is acquired solely through human interaction and that language is not preprogrammed), would there be consistency in human thought patterns or would communication be even possible? 7. Fill up the blanks with the words in italics below to make sense of the text: (a) evidence (b) acquisition (c) universal grammar (d) Chomsky (e) innate (f) ability (g) parameters (h) linguists (i) knowledge (j) infinite A strong piece of (1)……………. that supports linguistic knowledge being (2)…………. over the theory of humans acquiring linguistic (3)…………………. through interaction is the

AN OVERVIEW OF GENERATIVE GRAMMAR

89

fact that children, through a short period of time, have the (4)……………… to produce an (5)………….. number of sentences, as well as perceive and comprehend an infinite number of sentences. A child will go into the learning of a language with the ability to formulate general (6)…………………… which, exist in a particular language, on the basis of a few instances of relative usage. As (7) ……………… research they often find that children in the learning stage of language (8)…………………… do not make errors that go against the principles of (9)…………………. If environment was intended to provide us with language rather than to stimulate it, according to (10)………………., then the only possible communication would occur between identical twins who would have spent many waking moments together, and there would yet still be discrepancies. 8. Choose between a) or b) to describe Language according to a generative linguist: In short, «Language is a) not really something the child does; it is something that happens to the child placed in an appropriate environment, much as the child’s body grows and matures in a predetermined way when provided with appropriate nutrition and environmental stimulation. b) really something the child does; it is something that happens to the child placed in an appropriate environment, much as the child’s body grows and matures in a predetermined way when provided with appropriate nutrition and environmental stimulation. C. Lexical and functional categories, subcategorization and constituent structure 19. Define the lexical features related to words like: «car», «big», «come», «for» 10. What type of syntactic and morphological evidence can you give to support the idea that a word like «come» is a verb? 11. Draw the tree of a functional head X that projects one Specifier and takes a complement.

90

CURRENT TRENDS IN LINGUISTIC THEORY

12. Which are the main grammatical features that characterise pronouns? Which are the main properties that characterise Determiners as functional heads? 13. What is the main property that distinguishes between the Determiners «this» and «every»? 14. How can you show that the word «to» is a functional category and not a preposition in certain configurations? 15. Which are the three grammatical functions of the complementizer ¨for» in the following sentence: «For you to go there on your own would be unwise» ? 16. What is the main difference between the wh-adverb «whether» and Complementizer «if»? 17. What is the main claim of X’-theory? 18. Which are the two main syntactic configurations that operate in syntax? D. The syntax of clauses 19. In each case, only one of the three statements A, B or C is correct. Choose which one it is: (1) Syntactic structure means: a) A linear string of words; b) A system that has arranged words into units which form part of larger constituents; c) A method to study isolated words. (2) How could X-bar Theory be defined? a) as the study of relations between the words inside each constituent. b) as the study of unknown languages. c) as the study of sounds. (3) According to the Xbar theory, all minor nodes or heads project: a) only one level; b) two levels; c) up to three levels. (4) Specifiers and complements are: a) found at the same level in the structure. b) found at different levels in the structure. c) found at no level in the structure.

91

AN OVERVIEW OF GENERATIVE GRAMMAR

20. Apply each of the following rules and draw the corresponding trees: X → YZ Y → ABC Z → D 21. Show the configuration of the APs, VPs and PPs using binary branching in each corresponding Xbar schema of the following phrases: a) b) c) d)

very proud of her mother talk to Mary give a present to John sorry for her behaviour

22. Identify the different parts of the tree below: a) b) c) d) e)

nodes? labels? root nodes? terminal nodes? non-terminal nodes?

A” C” C’ C

A’ A

B” B’ B

23. Look at the following tree and say which nodes the node C ccommands: A B C

G D

E

H

I

F

24. In each case, only one definition of the three a, b, c is correct. Choose which one it is. (1) Nominative case in the generative approach may be defined: a) as the case usually held by pronouns b) as a structural relation between a head and its specifier c) as a morphological case that only appears with certain words

92

CURRENT TRENDS IN LINGUISTIC THEORY

(2) Accusative case may be defined in the generative approach: a) as a structural relation between the verb and its complement b) as an inherent case of nouns c) as a case of NPs (3) Genitive case in the generative approach may be defined as a) as a structural relation between a noun head and its specifier b) as a morphological case of prepositions c) as a case of PPs 25. Identify the r-expressions, anaphors and pronouns in the following examples according to the Binding Theory: (1) Mary left the room after the lecture (2) She didn’t wait for the bus (3) She drove her car home (4) She saw herself go out of the car in the mirror. 26. Look at the examples below and make the necessary changes to turn them into grammatical sentences: (1) *Herself bought the present. (2) *Johni enjoyed himi at the party. (3) *Mollyi wrote a letter to heri. (4) *Molly thought that John hated herself. 27. Tell which Binding Principles are violated in the following impossible sentences: (1) *Peter told Mary to love himself (2) *Johni sent a book to Johni (3) *Peteri looks at himi in the mirror (4) *Himself likes Peter 28. Explain why the following yes-no questions are impossible in English (1) *Has written John the book? (2) *Has John might write that letter? (3) *Might have Peter written that book? 29. Make the necessary changes to make grammatical the impossible sentences of the previous exercise.

AN OVERVIEW OF GENERATIVE GRAMMAR

93

30. Draw the derived structure after the insertion or the movement of the auxiliary to the front in the following examples: (1) Have you got money? (2) Will you come to the party? (3) Do you want to come? (4) Would you like to have more children? 31. Draw the tree-diagrams corresponding to the indirect interrogatives below: (1) John asked whether Mary had come to the party. (2) John doesn’t know if Peter has written that book. (3) He wonders whether his team will win the prize. 32. Identify the movement processes applying in the sequences below as only (i) head movement; (ii) phrase and head movement; and (iii) phrase movement. (1) When can you bring the money? (2) Have you checked the bill? (3) I wonder who wrote the book. (4) Who are you calling? 33. Provide the derived structures corresponding to the interrogatives below: (1) What will you talk about in the conference? (2) Who did you see in the party? 34. Provide the derived structure corresponding to the embedded interrogatives below: (1) He doesn’t know who the winner will be. (2) He wonders who you sent the letter to. 35. Identify the A-movement processes in the sequences below as (i) passive; (ii) unaccusative or (iii) raising. (1) John seems to be happy (2) The city was destroyed by the enemy (3) The patient will die of cancer (4) The bag broke 36. Specify all the arguments of the following verbs with their corresponding thematic roles: (1) buys; (2) gave; (3) breathes; (4) lives

94

CURRENT TRENDS IN LINGUISTIC THEORY

37. Draw the derived structure of the following passive sentences: (1) John was given the present (2) The book was written by him 38. Identify the syntactic subjects and the thematic roles of the NPs in the following sentences: (1) The cat was run over by the car. (2) John was introduced by Peter to Mary. (3) The house was built by that famous architect in 1957. 39. Explain why the following sentences are impossible in English: (1) *Molly arrived a letter. (2) *The bed was fallen on by the dust. 40. Explain why the following sentences are impossible and make the necessary changes to turn them into grammatical sentences: (1) *It is likely John to leave (2) *Bill to leave is likely 41. Draw the derived structure of the following sentence: (1) Bill is likely to fail 42. Explain why the NP Bill needs to move to the initial position in the previous sentence. 43. What is the difference (if there is one) between these two pairs of sentences? (1) a. The man hit the cat. b. The cat was hit by the man. (2) a. John seems to be ill. b. It seems that John is ill. E. Recent modifications according to the Minimalist Program 44. Draw the tree diagram (without applying the bare phrase theory) representing the grammatical structure of the following declarative sentence, and say how the features carried by the verb in this sentence are checked. (1) She kissed him 45. Draw the tree diagram (without applying the bare phrase theory) representing the grammatical structure of the following wh-question, and say how the features carried by the verb and the wh-phrase are checked. (1) Who did you visit?

AN OVERVIEW OF GENERATIVE GRAMMAR

95

7. REFERENCES ABNEY, S. (1987). The English Noun Phrase in its Sentential Aspect. PhD. Dissertation MIT. CASTILLO, C. (2003). English Syntax for Spanish Speakers: A Comparative Introduction. Peter Lang. Berlin. CHOMSKY, N. (1970). «Remarks on Nominalization», In R.Jacobs & P. Rosenbaum, eds., Readings in English Transformational Grammar. New York: Ginn & Co. CHOMSKY, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. CHOMSKY, N. (1986b). Barriers. Cambridge, Massachusetts: The MIT Press. CHOMSKY, N. (1988). «Some Notes on Economy of Derivation and Representation». In I.Laka and A.Mahajan, eds., Functional Heads and Clause Structure. MIT Working Papers 10. CHOMSKY, N. (1995). Repr. 1996. The Minimalist Program. The MIT Press. DEN DIKKEN, M. (1995). Particles: On the Syntax of Verb-particle, Triadic and Causative Constructions. Oxford/New York: Oxford University Press. EMONDS, J. (1976). A Transformational Approach to English Syntax. New York. Academic Press FUKUI, N & M. SPEAS (1985). Specifiers and Projections. MIT Working Papers in Linguistics 8, 128-172. GRIMSHAW, J. (1991). Extended Projections. Brandeis University. HAEGEMAN, L. (1992). Reprinted 1994. Introduction to Government and Binding Theory, Blackwell, Oxford. HALE, K. & J. KEYSER (1991). «On the Syntax of Argument Structure». Ms. MIT. HALE, K & J. KEYSER (2002). Prolegomenon to a Theory of Argument Structure. The Mit Press. JACKENDOFF, R. (1977). X-Syntax: A Study of Phrase Structure. Cambridge, Massachusetts: MIT Press. JACKENDOFF, R. (1990). Semantic Structures. MIT Press. KAYNE, R. (1984). Connectedness and Binary Branching, Dordrecht: Foris.

96

CURRENT TRENDS IN LINGUISTIC THEORY

KAYNE, R. (1993). The Antisymmetry of Syntax. MIT Press. LARSON, R.K. (1988). «On the Double Object Construction», Linguistic Inquiry, 19, 335-91. OUHALLA, J. (1991). Functional Categories and Parametric Variation. London: Routledge. PINKER, S. (1996). Language Development. The MIT Press. POLLOCK; J. –Y. (1989). «Verb Movement, UG and the Structure of IP». Linguistic Inquiry, 20, 365-424. RADFORD, A. (1997). Syntactic Theory and the Structure of English. A minimalist Approach. Cambridge University Press. Cambridge (UK). RADFORD, A. (2004). Minimalist Syntax: Exploring the Structure of English. Cambridge University Press. Cambridge (UK). RITTER, E. (1991). «Two Functional Categories in Noun Phrases: Evidence from Modern Hebrew». In S. Rothstein,ed. Perspectives on Phrase Structure: Heads and Licensing. Syntax and Semantics 25. New York: Academic Press. 37-62. STOWELL, T.A. (1981). Origins of Phrase Structure. Ph D. Dissertation MIT. TENNY, C. (1987). Grammaticalizing Aspect and Affectedness. Ph d dissertation MIT. VAN RIEMSDIJK, H. (1990) «Functional Prepositions». In H. Pinkster & I. Genee, eds. Unity in Diversity. Papers presented to Simon C. Dik on his 50th birthday. Dordrecht: Foris 229-242. ZWARTS, J. (1992). X’Syntax - X’Semantics. On the Interpretation of Functional and Lexical Heads. OTS PhD dissertation. Utrecht University.

CHAPTER

2

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR1 Ricardo Mairal Usón UNED

Francisco Cortés Rodríguez Universidad de La Laguna

1 In writing this chapter we namely follow the works of Van Valin (1993, 2000, 2005), Van Valin and La Polla (1997) for the four sections of this chapter. Besides, more peripherally we have used other sources which have played a more marginal or peripheral role: Butler (2003), Van Valin (2001), Mairal and Van Valin (2002). Moreover, we refer the reader to the RRG web site (http://linguistics.buffalo.edu/research/rrg.html), which includes a short introduction to the model together with a complete catalogue of papers which show an updated account of the present research. Then, this chapter is just a summary of the major ideas of RRG and consequently no claim for originality is made. In fact, some of the sections are adaptations from the sources cited above.

INTRODUCTION This chapter provides a short introduction to Role and Reference Grammar (hereafter RRG), one of the most relevant functional models on the linguistic scene today. For reasons of space, only a glimpse of the basic rudiments of the theory is given. This signifies that innovative aspects of the model such as the treatment of grammatical relations, the analysis of complex clauses, the theory of linking within the framework of complex structures, as well as the semantic and syntactic representation of noun phrases will not be discussed. This chapter is thus organized in the following way: Section 1 provides an historical introduction to the model and its relation to other linguistic models. Moreover, this section also addresses methodological issues which clarify the basic premises of RRG. Section 2 deals with the internal structure of the lexicon component, with particular emphasis on the following: (i) the criteria used to determine verb classes; (ii) the inventory of logical structures; (iii) macrorole assignment. Section 3, which is concerned with the relational aspects of the theory, presents the basic units of analysis of the Layered Structure of the Clause. Although this structure consists of the constituent, operator and information projection, only the first two projections will be discussed. Finally, section 4 succintly describes the rudiments of the linking algorithm, or the set of operations, which account for the systematic relationships between the syntactic and the semantic components. Since linguistic description is not always synonymous with clarity, we have also included exercises that will help students to better understand the different topics discussed in this chapter. The theoretical issues covered in this chapter are the following:

OUTLINE 1. Methodological premises 2. Lexical representation and the lexicon 2.1. Determining verb classes 2.2. Lexical representation: an inventory of Logical Structures 2.3. Thematic roles and macroroles 3. Syntactic representation: the Layered Structure of the Clause 3.1. The constituent projection: basic units of syntactic analysis 3.2. The operator projection 3.3. The syntactic inventory 4. The design of a linking algorithm 5. Final remarks 6. Suggested activities 7. References

1. METHODOLOGICAL PREMISES RRG conceives language as a system of communicative social action, and accordingly, it is fully committed to what Van Valin and La Polla (1997: 11) refer to as the communication-and-cognition perspective. This means that language, and more particularly, its morphosyntactic structures and grammatical rules, should be explained in relation to their semantic and communicative functions. Hence, RRG can then be classified as a member of the functional paradigm, as opposed to the formal paradigm.2 The debate between formal and functional models has been present throughout the history of linguistics over the past two hundred years3. Very briefly put, formal and functional models are based on radically different conceptions of the nature of language: basic features of the generativist model are its innateness, modularity, and psychological adequacy, whereas functional-cognitive models, which focus on function, meaning, and usage, are non-modular and experience-based. Functionalists centre their interest on function, meaning, and language use (cf. Langacker 1987, Dik 1997, Van Valin and LaPolla 1997, Cuenca and Hilferty 1999). They focus on function because they consider that function and meaning are factors that condition form, and not vice versa.

2 Recall that the formal paradigm is identified with Generative Linguistics as initially proposed in Chomsky (1957, 1965) (see chapter 1 in this volume and also the textbook proposed for the subject Sintaxis Inglesa). It is interesting to note that Generative Grammar has undergone different revisions over the years. Following Brucart (2002) and Jackendoff (2002), there are five major versions of the model; Chomsky (1957); the Aspects model, which provides the major foundations of the theory, some of which remain intact; The Extended Standard Version, which came out as a response to the hot debate between generative and interpretative semantics; Principles and Parameters, which constitutes one of the most solid proposals (cf. Chomsky, 1981); the Minimalist Progrram (cf. Chomsky, 1995, 2000). 3 For reasons of space we cannot go into the exact details of this controversy. Thus, we refer the reader to Butler (2003: chapter 1), Dik (1997: chapter 1), Mairal and Gil (2004, 2005), Newmeyer (1998) etc. for a complete discussion of this issue.

104

CURRENT TRENDS IN LINGUISTIC THEORY

They focus on meaning because they believe that grammar is structured by semantic and conceptual content, and is thus dependent upon it. They focus on use because they do not separate the study of language from its communicative function. As a result, their primary data source is actual speaker output and not introspection. In connection with this, in recent years a number of different linguistic paradigms have emerged under the cover term ‘functional’. Following Nichols (1984: 102-103), functional models could be classified in conservative, moderate and extreme, depending on how they hold with respect to the issues of the autonomy of syntax and the autonomy of grammar4, two of the hallmarks of the formalist tradition: The conservative type merely acknowledges the inadequacy of strict formalism or structuralism, without proposing a new analysis of structure…The moderate type not only points out the inadequacy of a formalist or structuralist analysis, but goes on to propose a functionalist analysis of structure and hence to replace or change inherited formal or structural accounts of structure. Extreme functionalism denies, in one way or another, the reality of structure qua structure. It may claim that rules are based entirely on function and hence there are no purely syntactic constraints; that structure is only coded function, or the like.

Conservative functional models assume a formal system of representation for grammatical structure, and propose a number of functional rules and principles that come to complement or enrich formal representations. In a way, they constitute an extension of formal theories. At the other end, extreme functional models reject the notion of structure and reduce all grammatical rules to discourse. Although moderate functional theories maintain the validity of the notion of structure and do not claim that all grammatical structure can be reduced to discourse, they depart from formal representations and develop their own view of grammatical analysis. RRG is an important example of this moderate view as can be observed in the following: RRG takes language to be a system of communicative social action, and accordingly, analyzing the communicative functions of grammatical structures plays a vital role in grammatical description and theory from this perspective . Language is a system, and grammar is a system in the traditional structuralist sense; what distinguishes the RRG conception is the conviction that grammatical structure can only be understood

4 See Croft (1995) and Butler (2003) for alternating taxonomies of the different functional models.

105

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

with reference to its semantic and communicative functions. Syntax is not autonomous. In terms of the abstract paradigmatic and syntagmatic relations that define a structural system, RRG is concerned not only with relations of cooccurrence and combination in strictly formal terms but also with semantic and pragmatic cooccurrence and combinatory relations. (Van Valin, 1993: 2)

RRG views the analysis of grammatical structure within the wider framework of the type of functions it fulfils. In other words, grammar is to a great extent determined by semantics and pragmatics. Accordingly, the theory is strongly committed to exploring the interaction of structure and function in language (cf. Van Valin 2001: 330ff). This means that syntax is no longer autonomous, but rather pragmatics and semantics are the all-powerful components of language which constrain syntax.5 Having defined RRG as a moderate functional theory, the next issue is to find out what led to the creation of RRG. Van Valin (2005: 1) writes: 1. What would linguistic theory look like if it were based on the analysis of Lakhota, Tagalog and Dyirbal, rather than on the analysis of English? 2. How can the interaction of syntax, semantics and pragmatics in different grammatical systems best be captured and explained?

From the first question it can be deduced that RRG has a very solid typological orientation, and seeks to develop a theory that can account for the linguistic properties of the world’s languages. Hence, one of the aims of RRG is to provide a linguistic framework that can serve to describe the set of linguistic properties of a wide range of languages. In connection with this, RRG maintains that a syntactic description in purely formal units fails to capture comparable expressions from different languages, and it does not reflect distinctions that every language makes. If we consider the instances in (1)-(4), the aim of RRG is to develop a framework which allows the description of word order, for example, in both configurational languages (languages whose constituents must to a greater or lesser degree follow a certain order) and non-configurational (languages whose constituents apparently have no predetermined pattern) (cf. Mairal and Gil 2003, fc): (1) The teacher (2) Waúspekhiye teacher

is

reading

ki the

a

book

wówapi book

wa a

(English) yawá read

(Lakota)

5 Dik’s (1997) Functional Grammar follows this line of research and postulates a functional moderate model like RRG, although with a much more semantic orientation. For a comparison of both models, see Butler (2003) and Mairal and Van Valin (2001).

106

CURRENT TRENDS IN LINGUISTIC THEORY

(3) Manjaha read

buju book

guru i teacher the

(Toba Batak)

(4) Uˇcitel’nica teacher

cˇ itaet read

knigu book

(Russian)

Knigu Book

cˇ itaet read

cˇ itaet Read

uˇcitel’nica teacher

uˇcitel’nica teacher knigu book (Van Valin 2001: 2)

As shown in (1), the canonical word-order pattern for English –the same as in Spanish, French, and Italian– is SVO, where the Subject precedes the Verb, and the Verb precedes the Object. In contrast, in Lakota (2), an Amerindian Sioux language, the Subject and Object precede the Verb, the same as in Japanese, Turkish, and Navajo, all of which thus have an SOV word order. Finally, Toba Batak, a Malayo-Polynesian language (3) has a VOS word order like Welsh, certain Amerindian languages, and Arabic dialects in which the Verb precedes the Object and Subject. In contrast, constituent order in Russian (4) can vary without affecting the grammaticality of the sentence since subject and object roles are marked by fusional case morphemes such as –a and –u. In this language, it is thus word form that indicates the relational structure of the utterance. In essence, RRG formulates a theory that can account for the syntactic representation of the world’s languages, independently of whether they are configurationally or non-configurationally coded6 (see section 3). Finally, a brief mention should be made of the architecture of the theory, as illustrated in Figure 1. It is important to bear in mind that RRG is a monostratal theory, which means that the semantic and the syntactic components are directly mapped without the intervention of abstract syntactic representations (e.g. deep structure in Generative Grammar).7 There are three main levels of representation: (i) a semantic representation that captures the meaning of linguistic expressions in terms of an inventory of logical structures (cf. section 2); (ii) a representation of the syntactic structure of sentences based on universally valid distinctions (cf. section 3); (iii) a representation of the information structure of the utterance, a topic that will not be covered in this chapter (cf. Van Valin 2005: chapter

6 We refer the reader to Mairal and Gil (fc) for an extensive discussion of word order from a universal perspective. 7 This contrasts with the derivational nature of the format employed in Generative Grammar.

107

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

3). Moreover, the semantic and the syntactic components are directly mapped in terms of a linking algorithm, which includes a set of rules that account for the syntax-semantics interface. Figure 1 illustrates the basic architecture of the theory.

Parser Syntactic Inventory

Lexicon

Linking Algorithm SEMANTIC REPRESENTATION

Discourse-Pragmatics

SYNTACTIC REPRESENTATION

Figure 1: The overall architecture of RRG (< Van Valin, 2005)

As shown in Figure 1, the architecture of RRG consists of (i) a lexicon component where predicates and their corresponding meaning representations are stored; (ii) a syntactic inventory which contains the inventory of syntactic templates or constructions that occur within the grammar of a specific language. These two components are mapped by means of a linking algorithm. Interestingly enough, the RRG linking system is bidirectional (see the double arrow), in that it maps both from syntax to semantics and from semantics to syntax. In this way it models aspects of what both the hearer and the speaker do in a communicative exchange (see section 4).

SUMMARY This section introduces some of the basic methodological premises of the RRG framework. You should have a clear idea of the following issues: • Formal and functional models. • Within functional models, you should be able to identify the three major trends: (i) conservative; (ii) moderate; (iii) extreme. • The status of RRG as a moderate functional model. • The notion of typological adequacy. • The monostratal nature of the theory.

108

CURRENT TRENDS IN LINGUISTIC THEORY

NOW YOU ARE READY TO DO EXERCISES 1, 2 and 3

2. SEMANTIC REPRESENTATION AND THE LEXICON In linguistics the lexicon is the technical term used to refer to the place where words and their corresponding semantic representations are stored. This means that the lexicon is like a mental dictionary. In recent decades, the lexicon has come to occupy a prominent place in linguistics because it has been shown that the morphosyntactic structure of a predicate can be predicted from its meaning. The following texts refer to this panlexicalist vision of language: The change from perceiving the lexicon as a wastebasket full of peripheral, irregular, and inelegant facts about language to perceiving it as a central component of grammar is due largely to the discovery that the lexicon is a highly structured repository of rules and principles that give it status and prominence previously accorded only to syntax. (Fellbaum, 1988: 3) Only a few years ago, it was conventional practice in both theoretical and computational linguistics textbooks to cover all that needed to be said regarding the lexicon in one quick chapter, before getting to the more interesting and more substantive topics of syntactic form and semantic interpretation. Such an impoverished coverage today would scarcely reflect the vibrancy of the field of lexical research or the central role played by lexical knowledge in linguistic theory and processing models. It is now standardly assumed by most linguistic frameworks (both computational and theoretical) that much of the structural information of a sentence is best encoded from a lexicalized perspective (Pustejovsky, 1995: 5)

Based on Figure 1, this section discusses the internal structure of the lexicon component and the types of semantic representation that can be used to codify meaning.

2.1. Determining verb classes A fundamental assumption in a functional theory such as RRG or Dik’s (1997) Functional Grammar is the teleological conception of linguistic expressions: sentences, as linguistically articulate utterances are

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

109

grammatical structures constructed with the purpose to convey meanings. From this perspective, sentences are linguistic expressions that speakers use to describe things that happen in any possible world (whether real or invented) and the entities involved in those happenings. Such happenings are called States of Affairs (SoA) in RRG, and can be classified according to the following content-related features: (a) The first type of SoA is Situations, typically static in nature, i.e. unchanging for its duration. Typical situations denote internal experiences (e.g. someone loving spaghetti), locations (as a shop being round the corner) or conditions of entities (such as grandpa being tired). (b) A second type, Actions, contrasts with situations in that they are dynamic in nature; i.e. a participant in the happening depicted does something, for example someone crying or the earth rotating. (c) In contrast to situations and actions, which do not necessarily convey that the happening will end, Processes involve a change that takes place over a bounded period of time. Thus, while nothing implies that the Earth’s rotating should end (that is why it is an action), a book falling on the floor necessarily will finish falling when it reaches the floor. Apart from having an implicit end, processes involve duration over a period of time (as when someone is learning a language), in contrast to the fourth type of situations called Events. (d) Events, which happen instantly, as when someone spots someone else, or when a bomb explodes. In the above classification there are various semantic differences: (1) an SoA may be static (situations) or not (actions, processes and events); (2) a non-stative SoAs can be atelic with no implicit terminal point (e.g. actions), or telic with a terminal point (e.g. processes and events); (3) an SoA can be either punctual (e.g. events) or have a duration (e.g. actions, processes and situations). Other distinguishing parameters are (4) the number of participants involved, and (5) whether the SoA happens spontaneously (all examples mentioned above are spontaneous) or whether it is induced, as in the following examples: a movie frightening someone (i.e. the movie caused someone to feel fear → causative situation), the workers blowing up a building (i.e. the workers caused the building to blow up → causative event) or a torpedo sinking the ship (i.e. a torpedo made the ship to sink → causative process).

110

CURRENT TRENDS IN LINGUISTIC THEORY

Crosslinguistically, an interesting feature of this classification is that speakers use these semantic distinctions and build meanings in sentences according to the types of SoAs described above. In fact, sentences must be considered as the lexical expression of SoAs, and some elements of the sentences (typically verbs, but also other possible types of words and expressions with a predicating function) express happenings of the types mentioned in the previous paragraph.8 This is the reason why RRG starts building the semantic representation of a sentence by selecting the relevant predicate together with its semantic description. We must now ask ourselves how predicate meaning should be described. RRG classifies verbs (and also other predicates) in lexical classes, according to their Aktionsart (German word for «form of action» or SoAs) or internal temporal properties9, following Vendler’s (1967) classification of verbs. RRG also uses a modified version of Dowty’s (1979) system of representation to express the different lexical classes. The following table shows types of SoAs and their corresponding Aktionsart type:

8 Nevertheless, Van Valin and LaPolla (1997: 86) point out that it is very important to distinguish what is in the state of affairs from what lexical items encode. Speakers may choose among different lexical items to express a same state of affairs. Sentences like «The boy broke/smashed a window with a stone», «A stone broke the window», «The window broke/shattered», etc are alternative expressions for a same state of affairs, that of a window (being) broken (by some entity or entities). 9 This classification contrasts with the more semantic oriented typologies as found in Levin (1993), Faber and Mairal (1999), etc.

111

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

State of Affair Type

Definition

Corresponding Aktionsart Type

Defining parameters

Examples

Situation

State Static, non-dynamic state of affairs, involving the location of a participant, temporally unbounded

Event

happens instantly, changes of state and changes of activities as well, inherent terminal point

Process

involves change and Accomplishment [-static],[+telic], melt,freeze, (intrans.)dry, [-punctual] takes place over time, learn bounded.

Action

Activity Dynamic in which a participant does something, no inherent terminal point

Achievement

[+static],[-telic], know,believe, [-punctual] have, be sick, love

[-static],[+telic], pop, explode, shatter, [+punctual] collapse

[-static], [-telic], march, walk, roll (intrans.), [-punctual] think, rain

Figure 2: Types of SoAs and their corresponding lexical classes (< Van Valin, 2005)

Vendler’s original classification of verbs divides them into: a) States denote static situations that are inherently atelic; English state verbs are know, have, hate and also predicates such as be tall, be happy, etc. b) Accomplishments express changes of state that are inherently telic (i.e. have a terminal point) and have duration. English accomplishment verbs are dry (intransitive version), learn, recover. c) Achievements are also telic and not stative, but in contrast with accomplishments, they do not take place over an extended period of time. In other words, they express momentaneous changes of state (e.g. burst, pop, in their intransitive uses). d) Activities are verbs that encode dynamic, non-telic SoAs, e.g. think, walk (intransitive use), push, drive. The following diagram shows the differences between the four classes in terms of the three features [±static], [±punctual] and [±telic]:

112

CURRENT TRENDS IN LINGUISTIC THEORY

States

[+static]

[-telic]

[-punctual]

Accomplishments

[-static]

[+telic]

[-punctual]

Achievements

[-static]

[+telic]

[+punctual]

Activities

[-static]

[-telic]

[-punctual]

Figure 3: Distinguishing parameters associated to lexical classes

Each of these four classes has a causative counterpart: (5) Jane is annoyed (state) → Her answers annoyed me (Causative State). (6) Water freezes below Oº C (accomplishment). → We must freeze this piece of meat (Causative Accomplishment). (7) The balloon popped (achievement) → The boy popped the balloon (Causative Achievement). (8) The ball was rolling down the slope (activity) → The boy rolled the ball down the slope (Causative Activity). The preceding examples show that in the same way as there is not a one-to-one correspondence between SoAs and verb classes, neither is there a one-to-one correlation between the meaning of a verb and its use in particular clauses. The expression or omission of certain arguments leads to different interpretations of the verb’s Aktionsart. In this regard, notice the difference between the intransitive (non-causative) and transitive (causative) uses of the verbs freeze, pop and roll. Consequently, a verb will have one basic Aktionsart interpretation for its classification in the lexicon, although other interpretations are possible in certain utterances. Another important case of Aktionsart alternation affects certain activity verbs when modified by definite NP objects (in the case of verbs of consumption and creation) or directional PPs (with verbs of movement), as in the following examples: (9) He drinks a lot → He drank only a glass of wine. (10) My daughter writes very well → E. Dickinson wrote some very obscure poems. (11) The warriors rode after the fleeing army → The warriors rode to the fortress. The second sentences in these pairs show the telic uses of some activity verbs, which are called Active Accomplishments. There are also Causative Active Accomplishment, e.g. The warriors rode the prisoners to the fortress.

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

113

Another verb class that has been added to the classification of Vendler, and used by Smith (1997) and Verkuyl (1993), among others, is that of Semelfactives. These are verbs such as bang, sneeze, flash, knock that encode punctual events not conducive to a result state. Some examples are: (12) Peter coughed. (13) The gate banged. (14) The traffic light is flashing. One important clue that helps to understand the difference between Achievements (punctual verbs that denote a result state) and Semelfactives is their respective behaviour when used with Progressive aspect in sentences with a singular subject. While in a sentence like the traffic light is flashing, the iterative interpretation of the semelfactive verb is the most natural, it is impossible to have the same kind of meaning in the case of achievements, as in the firecracker is bursting. In this sentence, the default interpretation of the progressive is the reference to a future time for the occurrence of the event. Progressive aspect with an iterative reading is only possible in the case of achievements when the subject has a plural form: Firecrackers are bursting. There are a number of syntactic and semantic tests for determining the class of a verb. However, it should be pointed out that none of these tests is fail safe in itself. Each test evaluates one specific semantic feature of the predicates. Thus, the results of all of these tests must be taken into account in order to accurately establish the type of predicate under scrutiny. The following is a description of how these tests can be applied to noncausative verbs: Test 1: The progressive. This test is an indicator of [-static], [punctual]. When the progressive means that a happening is ‘going on’ or ‘incomplete’, then it is not compatible either with states or with punctual verbs (i.e. achievements and semelfactives): (15) *He is being tall. (16) *The bomb is exploding. (17) *Chris is coughing. Again, note that (16) and (17) are possible in the progressive form, but in that case their meaning would be different since the progressive would either mark a future time reference (‘the bomb is exploding’ = «the bomb is about to explode’) or have an iterative reading (‘Chris is coughing’ =

114

CURRENT TRENDS IN LINGUISTIC THEORY

«Chris coughs several times». A reading such as «Chris is coughing once/one time», would be impossible because of the presence of the underlined adverbs). Thus, only Activities, Accomplishments and Active Accomplishments allow this test: (18) The water in the lake is freezing. (19) Aunt Mathilda is now drinking her tea. (20) The rider was galloping to the farm. Test 2: Dynamic adverbs: Some adverbs like dynamically, energetically or vigorously code dynamic actions, and are only compatible with predicates that are marked with the feature [+dynamic]. [-Static] predicates generally express a change of state or situation on the part of at least one participant. However, a further distinction in the [-static] group of verbs between [+dynamic] and [-dynamic] can be made. More specifically, activities are dynamic, whereas Accomplishments, Achievements and Semelfactives are not only [-static], but also [-dynamic]. In summary, this test is only compatible with Activities and with their telic variants, Active Accomplishments. Compare the following instances: (21) The competitors were running/singing energetically/vigorously/ actively. with (22) *Mary is actively/vigorously ill/tall. (23) *The lake is freezing energetically/vigorously. (24) *The door is bursting actively/energetically/actively. However, the use of dynamic adverbs that involve a controlling subject; e.g. carefully, deliberately, must be avoided since they are not only incompatible with [+static] and [-static, -dynamic] predicates, but also with some activity verbs whose meaning involve non-voluntary dynamism, as in the case of shiver and shake in The little puppy shivered violently/*deliberately and The building shook violently/*carefully during the earthquake. Test 3: Pace adverbs. This test does not apply to States, and its purpose is to ascertain whether a verb is punctual or not. Pace adverbs that involve some duration like quickly, slowly, rapidly can only co-occur with [-static] verbs that encode duration; i.e. Activities and Accomplishments, whereas they are blocked by Achievements and Semelfactives:

115

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

(25) Mary danced slowly/rapidly. (26) The water froze slowly/quickly. (27) ??The traffic light flashed slowly/quickly. (28) *The roof burst slowly/rapidly Again, it should be pointed out that there are certain pace adverbs whose meaning involves very short periods of time (e.g. instantly) and in some contexts are acceptable with punctual verbs (The roof burst instantly). Therefore, this type of adverb should be avoided when applying the test. Semelfactives show one further complication because they are sometimes compatible with adverbs like slowly, but only if there is an iterative interpretation in the sentence (e.g. The branch tapped slowly on the window). Again, if an adverb like once is added, the sentence becomes unacceptable (e,g, ??The branch tapped slowly on the window once). Test 4: X for an hour, spend an hour X-ing. The meaning of these expressions, which involve duration, further confirms the results of Test 3. This test shows that States, Accomplishments, Active Accomplishments and Activities have internal duration, whereas Achievements and Semelfactives do not. One exception is that States that refer to permanent attributes or properties do not usually co-occur with these expressions (e.g. *Mary was thin/tall for an hour). These four tests provide a provisional characterization of a verb’s Aktionsart. Test 1 makes a distinction between States, Achievements and Semelfactives on the one hand, and Activities, Accomplishments and Active Accomplishments on the other: Test 1, Progressive: [+static] [+punctual] Lexical Classes

YES

NO

Activities Accomplishments Active Accomplishments

States Achievements Semelfactives

Test 2 does not work with States, Achievements and Semelfactives, but segregates Accomplishments. It also isolates Activities as well as their telic variants, Active Accomplishments: Test 2, Dynamic adverbs

YES

NO

Lexical Classes

Activities Active Accomplishments

States Achievements Semelfactives Accomplishments

116

CURRENT TRENDS IN LINGUISTIC THEORY

Test 3 detects punctual verbs, e.g. Achievements and Semelfactives. However, as has been already pointed out, it is not relevant for States. Test 4 validates the results of Test3: Test 3, Pace adverbs: Test 4, Durative expressions [–punctual] Lexical Classes

YES

NO

Activities Accomplishments Active Accomplishments (States ← only for test 4)

Achievements Semelfactives

Test 5: X in an hour. Expressions like in an hour or in ten minutes usually refer to the termination point of the event. They express the duration as well as the end point of a happening. For example, if Matthew read the book in a couple of hours, he started reading the book a couple of hours ago, and finished reading it after that period of time. Usually expressions of this kind are readily acceptable for [+telic] predicates that involve some duration such as Accomplishments and Active Accomplishments. Achievements and Semelfactives also allow in-phrases if they refer to substantially short intervals, as in the gate banged in an instant or the balloon burst in the blink of an eye. When they are modified by in-phrases that express longer periods of time, they do not refer to the terminal point, but rather to a future reference for the onset of the event (e.g. the torpedo will explode in ten minutes), something that also happens with Activities (e.g. they will run in half an hour). Therefore, we can say that the test is only valid for Accomplishments and Active Accomplishments: Test 5, X in an hour: [+telic] [–punctual] Lexical Classes

YES

NO

Accomplishments Active Accomplishments

States Activities Achievements Semelfactives

Test 6: Stative modifier. This test applies especially to punctual predicates, Achievements and Semelfactives. Given the fact that Semelfactives do not involve a result state, they cannot be used adjectivally as stative modifiers (*a flashed light, *a coughed man). Achievements, on the other hand, behave in the same way as stative modifiers: a burst bubble, an exploded bomb, a shattered window.

117

AN OVERVIEW OF ROLE AND REFERENCE GRAMMAR

The following table extracted from Van Valin and La Polla (1997: 94) summarizes the behaviour of the different lexical classes with these tests. The * refers to the complications that may arise with some of the classes in reference to each of the tests above. Criterion 1. Occurs with progressive 2. Occurs with adverbs like vigorously, actively, etc. 3. Occurs with adverbs like quickly, slowly, etc. 4. Occurs with X for an hour, spend an hour Xing 5. Occurs with X in an hour 6. Can be used as stative modifier 7. Has causative paraphrase

State Achieve Acccmp Activity Active Accomp Seml No No

No* No

Yes No

Yes Yes

Yes Yes

No* No*

No

No*

Yes

Yes

Yes

No*

Yes*

No*

Irrelevant*

Yes

Irrelevant*

No*

No Yes

No* Yes

Yes Yes

No No

Yes Yes

No* No

No

No

No

No

No

No

Figure 4: Tests and verb classes (< Van Valin and LaPolla 1997: 94)

Test 7: Causative paraphrase. Even though it is not strictly speaking a test, causative paraphrasing is often helpful to determine whether a verb is inherently causative or not, as in the following sentences from Van Valin (2005: 38): (29) The dog caused the boy to be afraid (~ The dog frightened the boy). (30) The cat caused the balloon to pop (~ The cat popped the balloon). (31) The hot water caused the ice to melt (~ The hot water melted the ice). (32) The girl caused the ball to bounce around the floor (~ The girl bounced the ball around the floor). The one condition is that the paraphrases must have the same number of NPs as the original sentences, as is the case in examples (29)-(32), but not in (33): (33) The man made himself ride towards the hill (not ~ The man rode towards the hill) In the case of causative classes, the results of the seven tests are given in Figure 5, which has been taken from Van Valin (2005: 39):

118

CURRENT TRENDS IN LINGUISTIC THEORY

Class State Activity Achievement Semelfactive Accomplishment Active Accomplishment Causative state Causative activity Causative achievement Causative semelfactive Causative accomplishment Causative active accomplishment

Test 1

Test 2

Test 3

Test 4

Test 5

Test 6

Test 7

No Yes No* No* Yes Yes Yes* Yes No No* Yes Yes

No Yes No No* No Yes Yes* Yes Yes* Yes* Yes* Yes

No Yes No* No* Yes Yes No Yes No* No* Yes Yes

Yes* Yes No* No* Irrelev.* Irrelev.* Yes Yes No No* Irrelev.* Irrelev.*

No No No* No* Yes Yes No No No* No* Yes Yes

Yes No Yes No Yes Yes Yes Yes Yes No Yes Yes

No No No No No No Yes Yes Yes Yes Yes Yes

Figure 5: Verb classes and their corresponding tests ( 1st arg of pred´ (x, y) > 2nd arg of pred´ (x, y) > arg of pred´ (x) In accusative systems like English and Spanish, the highest ranking macrorole argument is selected as the default choice for ‘subject’, while in syntactically ergative systems, e.g. Dyirbal, Sama, the lowest ranking macrorole is the default choice. Both types of languages have constructions which allow a marked ‘subject’ selection: passive in accusative systems and anti-passive in ergative systems. In step 4, we select the appropriate syntactic template from the syntactic inventory (cf. section 3.3). Finally, step 5 is concerned with the assignment of the elements in the sentence. For example, if there is a wh-constituent then it will occupy the precore slot position etc. Let us now analyze two case studies, each exemplifying a different aspect of the algorithm: (109) What did Sandy give to Robin yesterday? (