Gisa rauh syntactic categories Their identification and description in linguistic theories (oxford surveys in syntax and morphology) oxford university press, USA (2010)

Oxford Surveys in Syntax and Morphology provides overviews of the major approaches to subjects and questions at the centre of linguistic research in morphology and syntax. The volumes are accessible, critical, and up to date. Individually and collectively they aim to reveal the field’s intellectual history and theoretical diversity. Each book published in the series will characteristically contain: (1) a brief historical overview of relevant research in the subject; (2) a critical presentation of approaches from relevant (but usually seen as competing) theoretical perspectives to the phenomena and issues at hand, including an objective evaluation of the strengths and weaknesses of each approach to the central problems and issues; (3) a balanced account of the current issues, problems, and opportunities relating to the topic, showing the degree of consensus or otherwise in each case. The volumes will thus provide researchers and graduate students concerned with syntax, morphology, and related aspects of semantics with a vital source of information and reference.

Syntactic Categories

general editor : Robert D Van Valin, Jr, Heinrich-Heine University and the University at Buffalo, State University of New York

advisory editors : Guglielmo Cinque, University of Venice; Daniel Everett, Illinois State University; Adele Goldberg, Princeton University; Kees Hengeveld, University of Amsterdam; Caroline Heycock, University of Edinburgh; David Pesetsky, MIT; Ian Roberts, University of Cambridge; Masayoshi Shibatani, Rice University; Andrew Spencer, University of Essex; Tom Wasow, Stanford University

4 Computational Approaches to Syntax and Morphology

Brian Roark and Richard Sproat

5 Constituent Structure (Second edition)

Andrew Carnie

6 Processing Syntax and Morphology: A Neurocognitive Perspective

Ina Bornkessel-Schlesewsky and Matthias Schlesewsky

7 Syntactic Categories

Gisa Rauh

in preparation

The Acquisition of Syntax and Morphology

Shanley Allen and Heike Behrens

The Phonology–Morphology Interface

Great Clarendon Street, Oxford ox2 6dp

Oxford University Press is a department of the University of Oxford.

It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in

Oxford New York

Auckland Cape Town Dar es Salaam Hong Kong Karachi

Kuala Lumpur Madrid Melbourne Mexico City Nairobi

New Delhi Shanghai Taipei Toronto

With oYces in

Argentina Austria Brazil Chile Czech Republic France Greece

Guatemala Hungary Italy Japan Poland Portugal Singapore

South Korea Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trade mark of Oxford University Press

in the UK and in certain other countries

Published in the United States

by Oxford University Press Inc., New York

# Gisa Rauh 2010

The moral rights of the author have been asserted

Database right Oxford University Press (maker)

All rights reserved No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press,

or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above

You must not circulate this book in any other binding or cover

and you must impose the same condition on any acquirer

British Library Cataloguing in Publication Data

Data available

Library of Congress Cataloging in Publication Data

Library of Congress Control Number: 2010922496

Typeset by SPI Publisher Services, Pondicherry, India

Printed in Great Britain

1 1 Categories and categorization in linguistics 1

1 2 Subject matter, aims, and outline 8

2 The traditional parts of speech 13

2 2 The grammar of Dionysius Thrax 14

2 3 The parts of speech in the Te´khne¯ grammatike¯ 17

2 4 The traditional parts of speech in selected grammars of the

2 5 Parts of speech as semantic categories 24

2 6 Problems with the traditional parts of speech 28

3 The American Structuralists’ approach: Syntactic categories as

3 2 Language as an object of scientific study 34

3 3 Towards a description of the structures of languages 36

3 4 Identifying syntactic categories 41

3 5 Inconsistencies and problems concerning the identification of

4 Syntactic categories in early Generative Grammar 54

4 2 Aspects of the Chomskyan theory of language 57

4 3 On the organization of a Generative Grammar: two examples 60

4 4 Syntactic categories and universal grammar 68

4 5 Reducing the set of categories 73

4 6 Universal vs language-specific syntactic categories 76

5 Categories, features, and projections 82

5 2 A modular theory of grammar: the Theory of Principles and

5 2.1 General theoretical considerations 84

5 2.2 On determining the D-structure of sentences 87

5 2.3 On deriving S-structure, LF, and PF 89

5 2.4 Some important questions 92

5 3.1 The origin of X-bar theory and the identification of

5 3.2 Towards a motivation of the categorial features 96

5 3.3 Including minor categories 98

5 3.4 Identifying functional categories 102

5 4 Lexical and functional categories and the licensing of

5 4.1 Licensing complements and specifiers of lexical heads 108

5 4.2 Licensing specifiers of functional heads 110

5 4.3 Licensing complements of functional heads vs.

licensing functional heads in extended projections of

5 4.4 A modified u-theory and the licensing of functional

heads and adjuncts in projections of lexical heads 115

5 5 Morphosyntax and syntactic categories 120

5 5.1 Morphosyntax, word order, and the splitting of IP 120

5 5.2 Splitting CP and yet more functional categories 124

5 5.3 Consequences for the identification and description of

5 6 Features, checking, and syntactic categories 129

5 6.3 Syntactic categories in the MP 136

5 7 Feature-based representations of syntactic categories in the

6 Syntactic categories, functional features, and feature structures 150

6 2 A lexical-functional approach to language 153

6 2.1 Considering language universal and language

6 2.2 On generating annotated c-structures 155

6 3 Syntactic categories in LFG 161

6 3.1 Functional features and the description of lexical and

6 3.2 Conditions on c-structures and their impact on the

description of syntactic categories 165

6 4.2 On deriving feature structures of type word in the lexicon 175

6 4.3 Syntactic principles and rules 178

6 4.4 Instead of transformations 183

6 5 Syntactic categories in HPSG 186

6 5.1 Feature structures and the description of syntactic

6 5.2 Some examples as illustration 190

6 6 Feature structures and syntactic categories in LFG and HPSG 196

7 Notional approaches to syntactic categories and Cognitive

7 3.2 Grammatical constructions: symbolization,

categorization, and integration 220

7 3.3 Prototypical and non-prototypical component and

7 4 Towards an identification of syntactic categories in Cognitive

7 4.1 Basic and non-basic grammatical categories 237

7 4.2 Grammatical constructions and distribution 250

7 5 Some problems in the identification of syntactic categories 257

8 A notional-feature basis for syntactic categories in a Localist

8 2 Notional features and the description of word classes 268

8 2.1 Basic word classes or ‘primary categories’ 269

8 2.2 Intermediate primary categories 271

8 3 Subcategorizations and redundancies in the lexicon 279

8 3.1 Categorial selection of complements and by

8 3.2 Functor features specifying valencies 282

8 4 Building syntactic structures 288

8 4.2 Building surface structures 290

8 5 Syntactic categories in a Localist Case Grammar 298

8 5.1 General assumptions, inconsistencies, and problems 299

8 5.2 Extensions and revisions 305

8 6 Notional approaches to syntactic categories and the question

9 Syntactic categories and language typology 322

9 2 Investigating linguistic categories in language typology 325

9 2.1 Parts of speech or ‘word classes’ 325

9 2.2 Syntactic categories or parts of speech? 332

9 2.3 Lexical and syntactic categories 339

9 3 A grammatical model for language typology: Dik’s

9 4.2 Logical structures and semantic representations 369

9 4.4 Syntactic vs lexical categories in RRG 378

9 5 The special situation of language typology 383

10 Syntactic categories and parts of speech: Two types of

Oxford Surveys in Syntax and Morphology provides overviews of themajor approaches to subjects and questions at the centre of linguisticresearch in morphology and syntax The volumes are accessible, crit-ical, and up to date Individually and collectively they aim to reveal thefield’s intellectual history and theoretical diversity Each book pub-lished in the series will characteristically contain: (1) a brief historicaloverview of relevant research in the subject; (2) a critical presentation

of approaches from relevant (but usually seen as competing) ical perspectives to the phenomena and issues at hand, including anobjective evaluation of the strengths and weaknesses of each approach

theoret-to the central problems and issues; (3) a balanced account of thecurrent issues, problems, and opportunities relating to the topic,showing the degree of consensus or otherwise in each case Thevolumes will thus provide researchers and graduate students con-cerned with syntax, morphology, and related aspects of semanticswith a vital source of information and reference

In Syntactic Categories: Their Identification and Description in guistic Theories, Professor Gisa Rauh surveys a wide range of linguistictheories, investigating and discussing their identification and descrip-tion of syntactic categories – the building blocks of sentences and theunits of grammatical analysis Her critical examination offers a sys-tematic account of syntactic categories, provides insights into thefunctioning of various grammatical models, and deals with formal aswell as functional theories of language and with language typology

Lin-Robert D Van Valin, Jr

General Editor University at Buffalo, The State University of New York Heinrich Heine University,

Du¨sseldorf

This book owes its existence to a number of people who have directly

or indirectly contributed to the making of it The first person to bementioned is Robert D Van Valin, Jr, the general editor of the series,who decided on its topic and its author I am grateful to him for giving

me the opportunity to write this book It was quite a challenge but arewarding one Next I would like to thank Hans Thilo Tappe, whosewide-ranging and at the same time detailed knowledge of linguistictheories made him the best discussion partner I could have hadconcerning all linguistic matters He has accompanied every section

of this book with valuable comments and suggestions which withoutdoubt have contributed to its quality

My thanks also go to Leah Bauke, Stephanie Goethe, and the bers of various advanced linguistics classes in the English department

mem-at the University of Wuppertal who studied selected chapters of thisbook and in particular ensured that my discussions were clear andconvincing The comments they made were very valuable The bookhas also profited from the expertise of Helga Helmich and above allColin Foskett Thank you for your support For the preparation of thefinal version of the manuscript, my thanks go to Matthias Naumannand in particular to Boris Ku¨hne It was a relief to know these matterswere in good hands

I would also like to thank John Davey, the consultant editor for tics at Oxford University Press, for his friendliness, empathy, patience,support, and helpful advice while I was preparing and writing this book

linguis-I am extremely grateful to the rheumatologist Helmut E Stierle, whoimmediately realized what the problem was when my right handpainfully and mysteriously refused its services Without his successfultherapy I would not have been able to physically finish this book.Last but not least, there are two men to whom I am indebted andwho deserve my warmest thanks The first is my husband, Klaus Peters

He never complained that I had to spend what must have seemed aninterminable length of time at my desk In fact, he encouraged andsupported me whenever necessary and wherever possible His interest

in my work has been a constant source of the energy needed

to complete a book like this in addition to my duties as a professor

The second man I must thank is my former academic teacher, the lateThomas Jefferson Gardner of the University of Go¨ttingen He wasmuch more to his students than just a professor of linguistics, and it

is unfortunate that he did not live to see this book published I dedicate

it to the honour of his memory

Gisa Rauh University of Wuppertal

CHL Computational System of the Human Language

f-description Functional Description

PredP Predicate Phrase

srv Subject-Raising Verb

Introduction

1.1 Categories and categorization in linguistics

This book is concerned with syntactic categories and thus with aspeciWc type of categorization of linguistic items It seems thereforeappropriate to start the introductory chapter with some remarks oncategories and categorization, both in general and speciWcally inlinguistics

Categorizing is a fundamental aspect of how humans process reality.The formation of categories gives structure to the enormous amount ofsensory input Items that share properties are combined to formgroups, and it is these groups that deWne categories Because of theshared properties of their members, categories enable us to formulategeneralizations In the human perception of reality, the formation ofcategories thus serves cognitive economy Insights and statements nolonger refer to individual items, but can be generalized to whole groupswhich are subsumed under categories

Because of their generalizing nature, categories are indispensable inany area of scientiWc study Any science will quite naturally systematizeits object of study This means, above all, forming categories anddescribing the relations between them, thereby providing a structureand thus yielding insights into the object of study When formingcategories, it is necessary to establish the basis for their formationand also what purpose they serve, thus showing what generalizationsfollow from the formation of categories and how these generalizationsyield insights into the object of study

However, when considering what linguists say about categories, no clearpicture emerges What is immediately obvious is a variety of termin-ology, including the terms ‘parts of speech’, ‘word classes’, ‘form classes’,

‘lexical categories’, ‘grammatical categories’, and ‘syntactic categories’.What often remains unclear is whether these terms refer to diVerentkinds of categories or whether they are more or less synonymous

That the latter is the case with regard to some of these terms is claimed

by Haspelmath, among others He starts his 2001 article ‘Word Classesand Parts of Speech’ by listing ‘the ten traditional categories’ (ibid

16538), which he calls ‘word classes’, namely Noun, Verb, Adjective,Adverb, Pronoun, Preposition/Adposition, Conjunction, Numeral,Article, and Interjection Haspelmath continues:

Besides the term word class, the older term part of speech (Latin pars orationis) is still often used, although it is now quite opaque (originally it referred to sentence constituents) The term word class was introduced in the Wrst half of the twentieth century by structuralist linguistics Another roughly equivalent term, common especially in Chomskyan linguistics is ‘syntactic category’ (although technically this refers not only to lexical categories such as nouns and verbs, but also to phrasal categories such as noun phrases and verb phrases).

(Haspelmath 2001: 16539)According to this statement, Haspelmath considers the terms ‘wordclass’, ‘part of speech’, ‘syntactic category’, and ‘lexical category’ to beequivalent or at least ‘roughly equivalent’, and thus to be synonyms ornear-synonyms

As far as the terms ‘part of speech’ and ‘word class’ are concerned,linguists are divided over whether these are in fact synonyms, asclaimed by Haspelmath, or not According to BloomWeld, to whomHaspelmath implicitly refers, since it was probably BloomWeld whointroduced the term ‘word class’ (cf 1914: 108), this is not the case.BloomWeld says of the relationship between parts of speech and wordclasses: ‘The maximum word-classes of a language are the parts ofspeech of that language’ (1926: 160); and: ‘The term parts of speech istraditionally applied to the most inclusive and fundamental word-classes of a language’ (1933: 196) From this, it follows that the sets ofparts of speech and word classes are not identical Apart from the parts

of speech, there are other ‘less inclusive’ and ‘less fundamental’ wordclasses However, Sasse (1993) shares Haspelmath’s view, and refers to aremark on parts of speech by Lyons (1977), which runs as follows: ‘Wewill assume that every word-lexeme is assigned, in the analysis of anylanguage-system to one, and only one, such class’ (ibid 423) Sasse’scomment on this is to say ‘we will call such classes word classes’, adding

‘word classes are lexical categories’ (1993: 648)

This shows that Sasse not only agrees with Haspelmath’s claim that

‘part of speech’ and ‘word class’ are synonyms, but that, in addition, heshares his view that ‘word class’ and ‘lexical category’ are synonyms

Aaronson and Ferres (1984) also share this latter view and deWne ‘lexicalcategory’ as ‘the traditional lexical category or form class of words asdeWned in dictionaries (e.g noun, verb)’ (ibid 21) However, otherlinguists interpret ‘lexical category’ in a diVerent way, namely as asubset of Haspelmath’s list: e.g noun, verb, and adjective (e.g Chomsky

1981: 48; Baker 2003); noun, verb, adjective, and preposition ition) (e.g Chomsky 1986a: 160; Chomsky 1986b: 2; Haegeman 1994:

(postpos-146; Wunderlich 1996: 2); or noun, verb, adjective, adverb, and ition (e.g Radford 1997: 37 f; Haegeman and Gue´ron 1999: 58) Theselinguists contrast lexical categories with functional categories, whichinclude, among others, complementizers, determiners, pronouns, andauxiliaries (cf e.g Radford 1997b: 45) The term ‘word-level category’ isused for what Haspelmath calls ‘lexical category’ in order to distinguishthis category type from phrasal categories (e.g Radford 1988: 56).The synonymy or near-synonymy of ‘lexical category’ (¼ ‘part ofspeech’) and ‘form class’, as is assumed by Aaronson and Ferres (seeabove), or of ‘part of speech’ and ‘syntactic category’, as put forward byHaspelmath, is something that BloomWeld would not have accepted For

prepos-it was above all BloomWeld – later followed by others, including Lyons(1977: 423 f.) – who took great care to distinguish between parts of speechand form-classes, the latter comprising morphological categories on theone hand, and syntactic categories on the other This broad classiWcation

is a result of the structuralists’ view that morphology and syntax form acontinuum and that there is in principle no diVerence between these twodomains In accordance with this view, BloomWeld identiWes ‘number-aYxes’ and ‘Wnite verb expressions’ as examples of morphological andsyntactic form-classes (e.g 1926: 159) and provides the following charac-terization of a syntactic form-class in a discussion of syntax and syntacticconstructions: ‘All the forms which can Wll a given position [in a con-struction] thereby constitute a form-class’ (1933: 185) Lyons’s character-ization is rather similar, arguing in favour of a deWnition of ‘form-class’

‘in terms of syntactic equivalence’ (1977: 424), and stating:

[T]wo forms, fiand fj, are members of the same form-class Fx if and only if they are intersubstitutable (i.e have the same distribution) throughout the sentences of the language.

(ibid.)

Anderson (e.g 1997) and other linguists would agree with this deWnition

of syntactic categories as a subset of form-classes, distinguishing

them from the parts of speech By contrast, Sasse, and also math, not only considers syntactic categories and parts of speech as

Haspel-‘roughly equivalent’, but apparently as ‘fully equivalent’, for he states:

‘The analysis of syntactic categories was familiar to the traditionalgrammarians under the title parts of speech’ (1993: 646) The sameapplies to Croft, who originally described noun, verb, and adjective as

‘syntactic categories’ (1984, 1990, 1991), but later gives the very samedescription of the very same categories for ‘parts of speech’ (2000,

2001, 2005)

The examples selected here, to which others could be added, serve toillustrate that the terms employed for linguistic categories do not give aclear picture On the one hand, diVerent category labels are used forthe same set of items, and on the other, the same category label refers

to diVerent sets of items As is often the case, the confusion is not only

a terminological matter For if there were a clear deWnition for each use

of a term, which allowed items and category labels to be assigned toeach other, it would be a simple matter to exchange one category labelfor another and still know what it refers to However, it is precisely herethat one of the problems of linguistic categorization is to be found For

it is the exception rather than the rule to Wnd a clear explanation of thebasis on which categories are established, the kind of generalizationthis categorization expresses, and the insights into language which arethereby gained

What is especially striking is the lack of a recognizable and clearlydeWned basis for the categorization of parts of speech, the mosttraditional and most frequently discussed linguistic categories Theirnames and typical representatives are familiar to anyone concernedwith language, and the parts of speech roughly correspond to Haspel-math’s list It is because of this lack of a recognizable and clearlydeWned basis that it is not at all clear what generalizations thesecategorizations are intended to express and what insights into languagecan be expected

It was above all in the linguistic approach known as structuralismthat the basis for categorizing the parts of speech inherited fromancient Greece was criticized as problematic This was because thetraditional criteria for category membership applied to, and by andlarge applicable to, European languages were not suitable for the iden-tiWcation of analogous categories in non-European languages such asthe languages of Native Americans in North America (cf Boas 1911).Important criticism was Wnally voiced by Gleason (1965: 114 V.) On the

one hand, he demonstrates that diVerent criteria are employed fordiVerent parts of speech – for adverbs and nouns, for example,namely the criterion of modiWcation for the former (‘An adverb is aword that modiWes a verb, adjective or another adverb’, ibid 115) andthe criterion of reference as well as the criterion of inXection for thelatter (‘A noun is a name of a person, place or thing’ and ‘A noun is aword which forms a plural by adding -s or the equivalent’, ibid.) Onthe other hand, he shows that items that are traditionally assigned to

a part of speech by no means fulWl all these criteria Thus, althoughthe noun table denotes a thing and is inXected for plural by adding -s,the noun handshake only fulWls the criterion of inXection anddenotes an action rather than a thing; and cattle does not fulWl theinXection criterion and denotes a set of items rather than an indi-vidual thing Based on this criticism, Gleason demands that parts ofspeech be identiWed and described on the basis of a uniform set ofcriteria

Since then, numerous attempts have been made to comply with thisdemand Yet this has failed to lead to a greater degree of clarity, as thereare now competing proposals, each oVering a uniform basis butincompatible with the others Thus, on the one hand there are pro-posals that advocate a notional basis for the description of parts ofspeech (e.g Lyons 1966), and on the other hand there are approachesthat favour a syntactic basis (e.g Emonds 1987) There is also aproposal claiming that parts of speech express discourse functions(Hopper and Thompson 1984) and another which instead identiWesthem as expressions of propositional acts (e.g Croft 2000) As a result,

it has become even less clear what kinds of insights are to be gainedinto language by the identiWcation of parts of speech, ultimately raisingdoubts about the point of identifying parts of speech at all

Others as well as Gleason have noted that the traditional criteria forassigning linguistic items to a speciWc part of speech are not fulWlled

by all the items that are usually included as members of this part ofspeech, and this has given rise to new questions instead of providinganswers to old ones One consequence is for the internal structure ofparts of speech and thus for the structure of linguistic categories

in general Until the 1970s, it was taken for granted that categoriesshould be deWned following the classical, Aristotelian approach,i.e referring to a Wxed set of necessary and suYcient features thatare exhibited by all the members of a category, with categorymembership determined on the basis of such a Wxed set of features

Fillmore (1975) compares such a procedure to checking features on a

‘checklist’

The results of various experiments led the cognitive psychologistRosch (e.g 1973a,b, 1977a,b, 1978; Rosch and Mervis 1975) to postulate adiVerent basis for the human categorization of the real world, and thusfor the formation of cognitive categories According to her, cognitivecategories do not correspond to a set of entities featuring identicalcriterial properties Instead they exhibit a prototype structure Thisimplies that such categories have a core, represented by the prototype,which displays the characteristic properties of this category, and is thusits best exemplar Other members of the category are arranged aroundthis prototype and associated with it by ‘family resemblance’ This termused by Rosch and Mervis (cf 1975: 575 V.) is based on Wittgenstein’sstatements on categorial structures and the relationship between mem-bers of a category (1953 [2001]: 27ef.), statements which had previouslyreceived little attention ‘Family resemblance’ here expresses the ideathat entities share several or only one property with each other and/orwith the prototype, but that they do not share all the relevant proper-ties The prototype is commonly seen as occupying a central position,and entities that are located in its vicinity share properties with thisprototype, while entities that are located at a greater distance from theprototype share properties with those category members that are closer

to the prototype, and so forth (cf e.g Givo´n 1989: 39) The mostextreme case within such a prototype approach is that two entities

do not share any property and exhibit membership of the samecategory only by sharing properties with other entities of that category

It is also the case that categories of this kind do not have clearly deWnedboundaries Transitions between categories are smooth, the resultbeing that it is perfectly justiWable to claim that entities at the adjacentperipheries of two categories can be members of either of these twocategories It is above all the prototypes that are responsible for per-ceiving prototype categories as distinct ones

The assumption of such a type of categorization for linguisticcategories was Wrst suggested by LakoV (1987: 58 V.) and then by Taylor(1989 [1995]) It is above all Taylor who discusses the categorial struc-ture of ‘grammatical categories’ in greater detail, which according tohim include ‘word classes – the traditional parts of speech’ and ‘syn-tactic categories like noun phrase’ (1989 [1995]: 183) He takes up theproblem mentioned by Gleason (see above), namely that not all theitems that are assigned to grammatical categories such as noun or

adverb exhibit all the properties associated with these categories, and

he proposes a solution to this problem by assuming that these ies have a prototype structure Thus, grammatical categories have ‘aprototype structure, with central members sharing a range of bothsyntactic and semantic attributes’ (ibid 196) Following prototypetheory, other category members can then share properties either withthe central members or with one another, but in no case do they have

categor-to exhibit all the properties of the central members

This view of the structure of linguistic categories has now becomewidespread, but it raises new questions, since Taylor subsumes both

‘word classes’ or ‘parts of speech’ and ‘syntactic categories such asnoun phrase’ under the label ‘grammatical categories’ This align-ment, together with Taylor’s comments, clearly indicates that he fol-lows Haspelmath in not distinguishing between word classes andsyntactic categories that are represented by words rather than byphrases If, on the other hand, such a distinction is made, an importantquestion arises regarding whether or not the assumption of a proto-type structure for categories, which seems at Wrst glance convincing forthe traditional parts of speech, is also appropriate for syntactic cat-egories Can it be concluded that the formation of linguistic categories

is made on the basis of prototypes because this corresponds to humancategorization, as shown by Rosch’s experiments?

Although this book is expressly devoted to syntactic categoriesrather than to the parts of speech, the fuzziness and uncertaintiesfound with regard to the latter are nevertheless of special relevancefor a number of reasons On the one hand, as indicated above, somelinguists assume that the terms ‘part of speech’ and ‘syntactic cat-egory’ are almost synonymous, i.e that the traditional parts of speechand syntactic categories are more or less the same On the other hand,various attempts have been made to describe parts of speech on asyntactic basis and thus as syntactic categories Finally, as will beshown in what follows, even when expressly dealing with syntacticcategories rather than with parts of speech, the names of the latter(‘noun’, ‘verb’, ‘adjective’, etc.) are used repeatedly This creates theimpression that it is in fact the familiar parts of speech that aremeant, and ultimately no distinction is drawn or even recognizablebetween syntactic categories and parts of speech For this reason, abook which is concerned with syntactic categories will also have todeal with parts of speech in order to show that parts of speech are notidentical to syntactic categories, and perhaps also to show why parts

of speech play such a dominant role in the categorization of linguisticitems.

Last but not least, issues concerning the ways in which linguisticitems are categorized are relevant to the subject matter of the presentbook, since one question to be answered is whether syntactic categor-ies exhibit a classical, Aristotelian structure with the members of acategory sharing all the necessary and suYcient properties of thiscategory, or whether a more appropriate description is based on aprototype structure with a best exemplar displaying all the relevantproperties of the category to which other, less ‘suited’ members of thecategory are associated via family resemblance

However, what is much more important in a book on syntacticcategories is above all an answer to the question of what syntacticcategories are and what insights into language as an object of scien-tiWc study are to be gained by their identiWcation and description.1.2 Subject matter, aims, and outline

The subject matter of this book is syntactic categories and theiridentiWcation and description in linguistic theories Syntactic categor-ies are part of the syntactic description of languages, and the syntacticdescription of languages is concerned with the structure of sentences.Sentences are complex syntactic objects that are composed of lesscomplex syntactic objects Syntactic tests, which are part and parcel

of any introduction to syntax, show that in addition to a phoneticallymanifested linear order, sentences exhibit a hierarchical structure.Linear order and hierarchical structure together represent the essentialaspects of the structure of sentences This overall structure derivesfrom the positioning of linguistic items in sentences, something thatcannot be changed at will, as well as from the relationships betweenthese items

What can in principle be changed, however, are the possibilities forlinguistic items to occur in given positions This means that it isgenerally not the case that a given position can be occupied by justone particular item Instead, items in given positions are interchange-able, and where this is possible, the items form a syntactic category.This is the basis for the deWnition of a syntactic category, which Wrst ofall refers to individual languages

A syntactic category is the set of linguistic items that can occupy thesame positions in the structures of the sentences of a given language

This is an extensional deWnition which exclusively refers to the sion of a category, i.e to the set of its members, and for which it issuYcient simply to list these members In contrast, an intensionaldeWnition is based on the set of those properties of linguistic itemswhich enable them to be placed in the same positions in the sentencestructures of a language The latter deWnition is the more demandingand the more revealing of the two, since it requires both an identiWca-tion of the members of a syntactic category and a description of theproperties that are relevant for category membership Finally, such adescription provides insights into which units can occur in givenpositions of sentence structures and which cannot, as well as into theproperties that are responsible for this Accordingly, it can be stated forwhich properties items must be speciWed in order to be able to occupygiven positions in sentence structures.

exten-Linguistic items whose membership of syntactic categories is to bedetermined can be of various kinds To begin with, they can be words.However, they can also be larger or smaller units, that is, phrases ormorphemes While these are normally units that are phoneticallyrealized, phonetically empty units that are solely represented by sem-antic and syntactic features can likewise assume positions in (abstract)sentence structures, thus acquiring the status of syntactic categories.Just what units are identiWed and described as syntactic categories candepend on the particular theoretical framework It will therefore be amajor concern of this book to establish what syntactic categories areidentiWed and described in the theories under consideration, above allanalysing how this is done Naturally, the identiWcation and descrip-tion of words as members of syntactic categories will play a central role,since no theory can dispense with such a description of words, whichare the basic units of sentences

It is the primary aim of this book to provide an overview of theidentiWcation and description of syntactic categories in various lin-guistic theories This above all entails establishing the means by whichsyntactic categories are described in particular theories, i.e what thecriteria are on which the categorial descriptions are based Yet it would

be a mistake to assume that such an overview could limit itself to justgiving the category labels or categorial descriptions that are presented

in the various theories under the heading ‘syntactic categories’ It isessential to any description of a genuinely syntactic category thatitems assigned to it on the basis of their description actually fulWl therequirements of being a syntactic category, i.e such a categorial

description should determine the possible positions of the categorymembers in sentence structures It is therefore necessary to investigatewhether, and if so how, in a given model of grammar based on aparticular theory, linguistic items are placed in those positions insentence structures that are appropriate, and how this is reXected intheir categorial description This requires a discussion of the basicassumptions of the theories presented, as well as of the essential aspects

of the models of grammar, in order to show how sentence structuresare built within these theories and what role is played by the categorialdescriptions

This approach leads to another aim of the present book, which is theevaluation of the theories under consideration with regard to theirability, or at least their potential, to describe syntactic categories, i.e todescribe category membership in such a way that by applying themeans provided by the particular grammar the items categorized areplaced in appropriate positions This also includes the question of theappropriateness of descriptions to categorize not only central linguisticitems, in particular items from the set of words, but ideally all of them

A critical discussion of the various theories as well as of the categorialdescriptions based on these theories is intended to sharpen the reader’sunderstanding of what exactly syntactic categories are, what they areexpected to do, what status they have in a grammar and how theyshould be designed in order to fulWl their task

In addition, based on the observation made in 1.1 that parts of speechand syntactic categories are often equated and thus even mistaken forone another, this book not only aims to point out that the traditionalparts of speech and their descriptions have little to do with syntacticcategories, but also to develop an explanation as to why they repeatedlyplay a role in the context of the description of syntactic categories Thisnecessitates determining the precise status of parts of speech

The traditional parts of speech are often treated as identical withsyntactic categories, and they inXuence the description of syntacticcategories even where this is not the case For this reason, chapter 2 will

be devoted to them It will examine the context in which the traditionalparts of speech were developed and the role the properties assigned tothem played in their original theoretical and grammatical environ-ment This will be done by presenting the Te´khne¯ grammatike¯ ofDionysius Thrax, dating from the Wrst century BC, and the parts ofspeech it describes, since this is generally assumed to be the origin

of the traditional parts of speech Chapter 2 will also introduce the

17th-century Port Royal grammar of Antoine Arnauld and ClaudeLancelot and the description of the parts of speech in this framework,which is likewise based on ideas dating back to ancient Greece, although

a diVerent perspective is taken, thus leading to a diVerent approach.What is pointed out in both cases is that the descriptions of the parts ofspeech are entirely independent of any syntactic properties

Chapter 3 will give an overview of how American Structuralistscriticized traditional grammar and its description of language, includ-ing its basis for a categorization of words as parts of speech Following

an outline of what Structuralists considered an appropriate description

of language, the approaches put forward by Harris (1951) and Fries(1952) will be presented and examined as two examples of how syn-tactic categories are identiWed and described as distributional classes.The fourth chapter starts with an introduction to Chomsky’s earlyviews on an adequate linguistic theory that includes language acquisi-tion It then presents his Wrst two generative models of grammar, i.e.Chomsky (1957) and Chomsky (1965), as well as the descriptions ofsyntactic categories they provide The critical evaluation in this chapterwill also consider alternative proposals put forward, in particular those

of Generative Semanticists

Chapter 5 will continue the discussion of Chomsky’s views onlinguistic theory and the models of grammar developed within hisframework, including a description of syntactic categories Specialconsideration will be given to the Theory of Principles and Parametersand the Minimalist Program Besides Chomsky’s own approaches (e.g

1981a, 1986a, 1993, 1995), a large number of complementary proposalswill be considered which above all tackle the issue of categorization.The sixth chapter will look at two approaches to the description ofgrammar that are competitors to Chomskyan views although they arealso considered to belong to the formally-oriented theories of lan-guage These are Bresnan’s Lexical-Functional Grammar (LFG)(1982a,b, 2001) on the one hand and on the other hand Head-DrivenPhrase Structure Grammar (HPSG), developed chieXy by Pollard andSag (1987, 1994) Here too, the basic theoretical assumptions and themodels of grammar are considered Wrst, before analysing and discuss-ing how syntactic categories are identiWed and described within theseframeworks The results of this analysis are then compared with thoseobtained from the discussions in earlier chapters

Chapters 7 and 8 will focus on linguistic approaches which, unlikethose previously considered, expressly advocate a notional basis for

syntactic or grammatical categories Some general introductory remarkswill precede a discussion of Langacker’s Cognitive Grammar (e.g 1987,

1991a, 2003), which has a functional rather than a formal orientation,including his description of grammatical categories This will befollowed by a consideration of how syntactic categories can bedescribed in this context Attention will then turn to Anderson’sLocalist Case Grammar (e.g 1991, 1997, 2006a, 2007) and the waysyntactic categories are described here (chapter 8) Unlike Langacker’sapproach, Anderson’s theoretical framework is classiWed as a formalone The chapter will conclude with a discussion of whether a classical

or a prototypical structure would be appropriate within notionallybased categorial descriptions

The ninth chapter will be devoted to an analysis of the identiWcationand description of syntactic categories within the framework of lan-guage-typological research After an evaluation of various approachesthat show a trend in typological research, the discussion will Wrst of allcentre on Functional Grammar as developed in particular by Dik(e.g 1978, 1983, 1997), and on the description of syntactic categories

it provides This will be followed by a critical evaluation of Foley andVan Valin’s Role and Reference Grammar (Foley and Van Valin 1984;Van Valin 1993, 1995, 2005, 2008; Van Valin and LaPolla 1997), including

a discussion of how this framework tries to distinguish between tactic and lexical categories

syn-Chapter 10 will begin by taking up and summarizing developments

in the description of syntactic categories as found in the approachesconsidered in this book, with further discussion of the basis of theseapproaches This chapter will also return to the question of the status

of parts of speech, presenting an analysis that distinguishes themclearly from syntactic categories but explains their role in connectionwith the categorization of linguistic items

1997b), or Cognitive Grammar (e.g Langacker 1987), to mention just

a few, also make use of these categories

The fact that different approaches often use the same label todesignate different categories is to begin with unimportant, but willprove to be very significant in the course of this book It is alsoremarkable that, despite their often-attested shortcomings for thedescription of non-Indo-European languages, these traditional parts

of speech are repeatedly used here too to serve this purpose (cf e.g.Schachter 1985; Hengeveld 1992; Croft 2000; Wierzbicka 2000).1 What

is especially interesting in this context is the frequent tacit assumptionthat word classes like noun, verb, adverb, etc are given units whoseexistence in the European languages is beyond doubt and that only twoquestions are of any relevance: first, what are the properties that

1 At this point it is of no importance that the definitions vary It should only be noted that the traditional parts of speech play an essential role for the identification of word classes.

identify these categories and second, do these categories occur in thelanguage under investigation or not It is significant for the presentdiscussion that the traditional parts of speech are also typicallyidentified and described as ‘syntactic categories’ (cf e.g Emonds

1987; Zimmermann 1990) although they were not introduced into thedescription of grammar as such, as will be seen in the following.These observations are sufficient reason to start an investigation ofsyntactic categories in various grammatical theories with a consider-ation of how the traditional parts of speech were originally seen Indoing this, it is important to examine them not in isolation but in theirgrammatical context in order to determine their original role Section

2.2 therefore starts by examining the background to the development ofthe grammar which introduced the parts of speech and founded thetradition that has lasted two thousand years This is followed by adescription of this grammar Section 2.3 presents the parts of speechidentified in this grammar and explains the basis of their descriptionwithin it In section 2.4, two late 19th-century grammars are used asexamples to illustrate the fact that about two thousand years later, more

or less the same parts of speech are identified and described in more orless the same way One of the grammars chosen deals with Romancelanguages and the other with English, and it is demonstrated thatsuccessful identification and description of the parts of speech depends

on the characteristics of the particular language or languages This led

to the introduction of other criteria for the description of word classes,

as section 2.5 explains The chapter finishes with a discussion of theproblems related to the identification of the traditional parts of speech

2.2 The grammar of Dionysius Thrax

The identification and description of parts of speech such as noun, verb,adverb, conjunction, etc goes back to grammatical descriptions whichwere developed by philosophers and philologists in Greece before the firstcentury BC, and summed up in the grammar of Dionysius Thrax (170-90BC), the Te´khne¯ grammatike¯ This grammar was written at the end of thesecond and the beginning of the first century BC2 and is considered thefirst comprehensive and systematic grammatical description published

2 Di Benedetto (1958-59, 2000) presents doubts about the authorship of the Te´khne¯ and about when it was written According to Frede, however, these doubts have been removed:

‘nowadays the text is generally accepted as genuine’ (1977: 52) This view is taken by Wouters (1979: 36) and Erbse (1980) as well See also the discussion in Robins (1993: 42 ff.).

in the western world A student of Aristarchus (217-145 BC), DionysiusThrax was born in Alexandria but because of political unrest later fled toRhodes, where he taught He was one of the Alexandrians, and was active

in the Greek colony of Alexandria, where in the third century BC thefamous library had been established It was in connection with the estab-lishment of this library that intensive philological activities had begun inAlexandria One of the aims was to collect, restore, and systematizethe great literary works of the past in order to make them accessible forcontemporaries and to preserve them for future generations This requiredcareful work on the original papyruses and led to the development oftechniques for editing, which included explanations and interpretations ofthe texts and thus promoted the development of literary criticism

In general, the great Alexandrians who developed these techniqueswere themselves writers and thus had a special relationship to the textsthey edited For an understanding of the Te´khne¯ grammatike¯, it isimportant to know that its grammatical descriptions grew out of lan-guage studies which aimed at producing critical editions of texts Theprincipal aim was to preserve the purity and perfection of the language

of writers from the past It was not until later, when the development of

a grammar became an end in itself and grammars were used as gogical instruments, that the supposed perfect language of these writerswas set as the norm for everyday language use In order to make thepronunciation of the texts easier, these philologists introduced stressmarks and punctuation3 in their editions, subsequently identifying andformulating rules.4 For the formulation of stress rules it was necessary toidentify and describe syllables, which in turn required an identification

peda-of the sounds peda-of the language In order to understand and interpret thetexts it was important to identify the words which constituted the textsand to determine their meaning And in order to determine the mean-ing of words it was often necessary to take into account their origin andtheir history, which led to the development of etymology

The study of words was a major concern of Aristophanes (257-180BC), the fourth librarian at the library of Alexandria,5 whose Lexeis

3 Pfeiffer (1968: 180) remarks that the early Ptolemaic papyruses did not have any stress marks and that Aristophanes was the first to place them in Homeric and other texts.

4 According to Pfeiffer (1968: 219), it was Aristarchus (217-145 BC) who in his mentaries gave reasons for his practice of stress placement and punctuation and thus provided the basis for the formulation of rules.

com-5 The librarians before Aristophanes were Zenodot, Apollonius Rhodius, and thenes (cf Pfeiffer 1968: 153 ff.).

Eratos-established him as a lexicographer It was also Aristophanes who, whencollecting and describing words, discovered recurrent patterns inGreek declensions and set up general rules for their formation Accord-ing to him, the regularities discovered were based on the principle of

ƺªÆ (‘analogy’) This principle was recognized by his studentAristarchus as a general principle of language, i.e of stress placementand morphology as well as of interpretation, and it provided animportant basis for the grammatical descriptions in the Te´khne¯grammatike¯, the grammar of Aristarchus’ own student Dionysius.The following brief characterization of this grammar demonstratesthat the aims of these philologists in their work on editing texts andthe insights into language they gained in this context strongly influ-enced the approach adopted by Dionysius

The Te´khne¯ grammatike¯ starts with a general characterization of agrammar and lists its constituent parts:6

Grammar is an experimental knowledge ( ØæÆ) of the usages of guage as generally current among poets and prose writers It is divided into six parts:

lan-1 Trained reading with due regard to Prosody.

2 Explanation according to poetical figures.

3 Ready statement of dialectal peculiarities and allusions.

4 Discovery of Etymology.

5 An accurate account of analogies.

6 Criticism of poetical productions, which is the noblest part of matic art.

gram-(T Davidson 1874: 326 f.)

In accordance with the elements of a grammar presented above, thefirst part of the Te´khne¯, which comprises sections 2 to 11, is concernedwith matters that relate to reading aloud and thus to aspects of thedescription of language relevant to its pronunciation: ‘ON READING’,

‘ON TONE’, ‘ON PUNCTUATION’, ‘WHEREIN DOES THE FULL STOP DIFFER FROM THE COMMA?’, ‘ON RHAPSODY’, ‘ON ELEMENTS’(LETTERS’), ‘ON SYLLABLES’, ‘ON LONG SYLLABLES’, ON SHORT SYLLABLES’, ‘ON COMMON SYLLABLES’ (T Davidson

1874: 327 ff.) A second and final part, sections 12 to 20, deals withwords and their categorization on the basis of ‘[a]n accurate account

of analogies’ (ibid 326) In this part, following a definition of the unit

6 The following quotations are taken from the translation by T Davidson (1874), which

is here preferred to Kemp’s translation (1987).

‘word’, definitions and characterizations of eight parts of speech arepresented which will be looked at in detail in the next section of thischapter It should be noted here that syntax is not part of thegrammar of Dionysius Thrax since in the Greek grammar of thefirst century BC syntax did not play any role In fact, it did notbecome a concern of grammatical description until three hundredyears later in the grammar of Apollonius Dyscolus, which was afurther development of Dionysius’ work.7 Without any doubt then,the parts of speech defined in the Te´khne¯ grammatike¯ cannot repre-sent ‘syntactic’ categories This, however, does not mean that they

do not represent ‘grammatical’ categories, as will be seen in thefollowing

2.3 The parts of speech in the Te´khne¯ grammatike¯

In his grammar, Dionysius Thrax describes eight parts of speech: noun,verb, participle, article, pronoun, preposition, adverb, and conjunc-tion.8 In order to give the reader of this book as accurate an impression

as possible of the definitions and characterizations of these eight parts

of speech, these are quoted here in his own words (as translated by

T Davidson 1874):

1 A Noun is a declinable part of speech, signifying something either concrete

or abstract (concrete, as stone; abstract, as education); common or proper (common, as man, horse; proper, as Socrates, Plato) It has five accidents: genders, species, forms, numbers, and cases (ibid 331)

2 A Verb is an indeclinable word, indicating time, person and number, and showing activity or passivity The verb has eight accidents: Moods,

7 This does not mean that syntax was altogether unknown at that time Marcus Terentius Varro (116-27 BC), a contemporary of Dionysius, devoted books 14 to 25 of his grammar De Lingua Latina to syntax Of the preceding books, six are concerned with etymology (2-7) and six with morphology (8-13) Of the 25 books, only some – books 5 to

10 – have survived (cf e.g D J Taylor 2000) The completely different structure of this grammar as well as the fact that Varro – unlike Dionysius – distinguishes only four parts of speech (words with case: nouns, including pronouns and adjectives; words with tense: verbs; words with both case and tense: participles; and words without case or tense: particles) strongly suggests that there was no influence in either direction.

8 It is said that it was Aristarchus, Dionysius’ teacher, who was the first to identify eight parts of speech (e.g Kemp 1987: 171) According to Frede (1977: 55), this claim was first made by Quintilian (book I, chapter 4, paragraph 20), who assigns the identification of eight parts of speech to Aristarchus Frede points out, however, that it is by no means certain that the parts of speech identified by Aristarchus are identical to those listed by Dionysius since no work on parts of speech by Aristarchus exists.

Dispositions (voices!), Species, Forms, Numbers, Tenses, Persons, gations (ibid 335)

Conju-3 A Participle is a word partaking of the nature both of nouns and verbs It has all the accidents which belong to nouns as well as those which belong to verbs, except mood and person (ibid 336)

4 An Article is a declinable part of speech prefixed or subjoined to the various cases of nouns It has three accidents: Gender, Number, and Case (ibid.)

5 A Pronoun is a word assumed instead of a noun, and indicating definite persons It has six accidents: Person, Gender, Number, Case, Form, and Species (ibid.)

6 A Preposition is a word placed before any of the parts of speech, both in Composition and in Syntax.9 (ibid 337)

7 An Adverb is an indeclinable part of speech, said of a verb or added to a verb Of the Adverbs, some are Simple, and others Compound – Some are indicative of time Some indicate manner ; some, quality ; some, quantity ; some, number ; some, place Some Adverbs signify a wish ; some express horror ; some, denial or negation (ibid 337 f.)

8 A Conjunction is a word binding together a thought in order and filling up the hiatuses of speech Of conjunctions, some are copulative, some dis- junctive, some conjunctive, some præter-conjunctive, some causative, some dubitative, some conclusive, and some expletive (ibid 338)

In the text of Dionysius’ grammar, these definitions are followed byfurther information which, supported by examples, explains the termsintroduced and describes the phenomena, making use of further cat-egorial specifications For example, the definition of ‘Noun’ is followed

by an explanation of terms such as Genders, Species, Forms, Numbers,and Cases With the help of examples, Dionysius then states whichgenders, species, forms, etc have to be distinguished Thus, threecategories of Gender[s] are named: ‘masculine’, ‘feminine’, and ‘neuter’(cf T Davidson 1874: 331) ‘[P]rimitive’ and ‘derivative’ are distin-guished as Species of nouns and verbs (cf ibid 331 and 335), and

‘simple, compound and super-compound’ as Forms (cf ibid.) Thecategory of Number[s] is divided into ‘singular, dual and plural’ (ibid

332), and that of Case[s] into ‘the right’ (‘nominative’), ‘the generic’(‘possessive’), ‘the dative’, ‘the accusative’, and ‘the vocative’ (cf ibid

333) The verbal category of Mood[s] is divided into ‘Indicative,

9 The term ‘syntax’ at this point is an interpretation by the translator Kemp, in his 1987 translation, uses the term ‘combination’.

Imperative, Optative, Subjunctive and Infinitive’ (ibid 335), and that

of Disposition[s] into ‘Activity, Passivity and Mediality’ (ibid.) ThreeTenses are distinguished – ‘Present, Past, Future’ (ibid 335) – with afurther distinction for ‘Past’ into ‘Imperfect, Perfect, Pluperfect andAorist’ (ibid.)

The definitions and characterizations given here demonstrate that adivision into parts of speech is first and foremost based on morpho-logical properties (‘declinable’ (cf e.g T Davidson 1874: 331), ‘in-declinable’ (cf e.g ibid 334)) and – where possible – complemented

by semantic ones This holds for nouns and verbs In both casesinflectional properties (declinable, indeclinable) are mentioned first,followed by semantic properties Additional specifications then includeinflectional properties (e.g gender, number, case, mood, tense) as well

as properties of derivation and compounding (species, form).10 Apartfrom their relationship to other parts of speech, participles, articles,and pronouns are exclusively characterized on the basis of morpho-logical properties, primarily on those of inflection In those caseswhere no specification of inflectional properties is possible, as is thecase with prepositions, adverbs, and conjunctions, the characterization

is either based solely on semantic properties (conjunctions), or on veryrudimentary syntactic properties (prepositions ¼ ‘pre-positioned’).The importance of morphological properties is seen again in thecharacterization of adverbs, where an extensive semantic specification

of various types of adverbs is preceded by the statements that an adverb

is an indeclinable part of speech and that some adverbs are simple andsome are compound

To summarize, it can be noted that the parts of speech introduced inthe Te´khne¯ grammatike¯ are primarily defined on the basis of inflec-tional properties.11 This explains why adjectives do not occur in the listprovided by Dionysius: in Greek they exhibit the same inflectionalproperties as nouns do and therefore cannot be distinguished fromnouns on this basis They are therefore included in the category ofnouns, which altogether comprises 24 classes Wherever possible, inaddition to inflectional properties, further – analogous – morphologicaland semantic properties are identified Where necessary, these are

10 No fundamental distinction is made here between inflection and derivation, as is illustrated by the mixed lists of instances of both.

11 This also holds for Varro’s grammar However, the four parts of speech he lists are exclusively identified on the basis of inflectional properties Neither derivational nor semantic properties are also considered (cf D J Taylor 2000: 457).

replaced by other properties Since Greek is a highly inflecting guage, it comes as no surprise that analogy as a grammatical principlewas first recognized in connection with the visible forms of inflection.

lan-In this sense, the parts of speech defined here do represent grammaticalcategories Their definition is in accordance with what Dionysius andothers considered the guiding principle of grammar: analogy Analogy

is here first seen in inflectional forms and subsequently – but to a muchlesser degree – in derivational affixes and semantic properties, and only

in the context of prepositions in rudimentary syntactic properties

2.4 The traditional parts of speech in selected grammars

of the nineteenth century

The grammar of Dionysius Thrax initiated a tradition which mined the structure and the content of grammars, and not only ofpedagogical grammars, up to the nineteenth and even the early twen-tieth century The enormous influence of his grammar was greatlyassisted by its translation into Latin and its careful application toclassical Latin in Remnius Palaemon’s Ars Grammatica in the firstcentury AD Although this work has not survived, it strongly influ-enced the development of the theory of grammar in Roman times,especially the much better known and very influential grammar Insti-tutiones Grammaticae by Priscianus (sixth century AD; cf Arens 1955[1969]: 32) Remnius Palaemon borrowed Dionysius’ terminology andgrammatical descriptions and applied them to Latin Like Dionysius,

deter-he identified eight parts of speech, replacing articles, which do notoccur in Latin, by interjections Except for this difference, the sameparts of speech are identified on the same basis, with adjectives andnouns again not distinguished, as in Latin too they exhibit identicalinflectional properties

The successful transfer of the grammatical descriptions in the Te´khne¯grammatike¯ to Latin encouraged their application to other languages aswell As a result, all the European languages were eventually described

on this basis and the Te´khne¯ grammatike¯ determined their grammaticaldescription for many centuries The identification and characterization

of the parts of speech in these languages followed Dionysius’ method,although due to language-specific differences, a different emphasis wasoften placed on the various properties This ultimately influenced laterapproaches to grammar and the categorization of words, as can be seen

in the following brief consideration of two examples of grammars

written in the nineteenth century The first one deals with Romance uages and the second with English It should be pointed out in advancethat certain deviations from the traditional Greek approach are a result

lang-of the inclusion lang-of syntax Beginning with the grammars lang-of ApolloniusDyscolus and Priscianus,12 syntax formed the third part of a grammar(after phonetics and accidence) and described the relationshipsbetween words in a sentence as well as their grammatical functions.The first example is the Grammatik der romanischen Sprachen byDiez (1836-38), which deals with Italian, Spanish, Portuguese, Proven-c¸al, and Walachian, and thus with languages closely related to Latin.This grammar consists of four books, which are concerned withphonetics (‘Lautlehre’), inflection (‘Wortbiegungslehre’), word-formation(‘Wortbildungslehre’), and syntax (‘Syntax’) Except for the article,which is described in the fourth book as an element that accompanies

a noun,13 thus pointing out a syntactic property, all the parts of speechare described with respect to their morphological properties To beginwith, the inflecting parts of speech are identified These are nouns andadjectives – now distinguished on the basis of their different syntacticproperties – numerals and pronouns, all of which are subject todeclension, and the verb as a part of speech that is subject to conjuga-tion The third book lists the derivational properties of these parts ofspeech, and the fourth book gives their syntactic properties Diez alsoidentifies adverbs, prepositions, conjunctions, and interjections, clas-sifying them as particles The fact that these non-inflecting and non-declining parts of speech are introduced in the third book on the basis

of their derivational properties confirms the view that Diez – followingDionysius – defines the parts of speech primarily on morphological

12 In addition to shorter pieces on pronouns, conjunctions, and adverbs, etc., Apollonius Dyscolus presented four books on syntax (cf Apollonius Dyscolus) In his writings on syntax, his aim is to show that language is rule based and that it is the task of syntax to put together words in a pre-determined order to form sentences He claims that syntax as well

as the other areas of grammar is based on the principle of analogy Blank (2000: 411) calls Apollonius’ works ‘masterpieces of ‘‘analogical’’ grammar’ (cf also Blank 1994) Thus, Apollonius continued the tradition of the Te´khne¯ grammatike¯ This is also evident from the fact that he took over the eight parts of speech identified by Dionysius without providing a characterization of his own Apollonius’ account of syntax, which relates to Greek, was applied to Latin by Priscianus in his grammar Institutiones Grammaticae (cf ‘Priscianus’ in Wissowa 1954: 2328-48).

13 This exception is interesting considering the fact that the languages concerned are descended from Latin Since Latin does not have an article, no such part of speech is described in grammars of this language.