Merger DEMO : Purchase from www.A-PDF.com to remove the watermark An Introduction to the Study of Morphology Vit Bubeník S.ovn«v«c( \iäiskm.pit sign.:... Přír. č. -t 5 full text research abstracts of all titles monthly updates LINCOM webshop 9nn,,9 H t , www.lmcom-europa.com r 2003 (2nd printing) LINCOM EURÓPA Published by LINCOM GmbH 2003 (2nd printing) 1st printing: 1999 Ail correspondence concerning LINCOM Coursebooks in Linguistics should be addressed to: LINCOM GmbH Freibadstr. 3 D-81543Muenchen LINCOM.EUROPA@t-online.de http://home.t-online.de/home/LINCOM.EUROPA www.lincom-europa.com webshop: lincom at All rights reserved, including the rights of translation into any foreign language. No part of this book may be reproduced in any way without the permission of the publisher. Printed in E.C. Printed on chlorine-free paper Die Deutsche Bibliothek - CIP Cataloguing-in-Publication-Data A catalogue record for this publication is available from Die Deutsche Bibliothek (http://www.ddb.de) PREFACE This introductory textbook to the study of linguistic morphology is based on four previous versions of a manuscript entitled An Introduction to the Study of Morphology. They were published in a mimeographed form by Memorial University of Newfoundland (St. John's, Canada) in 1978, 1982, 1986 and 1997, and were used at the third-year level in the Department of Linguistics. Its current version is designed for use as a second- or third-year university level introductory textbook to linguistic morphology. Before taking this course, students should have previously completed one or two introductory courses to the whole discipline of linguistics at their first or second year at the university. Its argumentation is built around the major turning points in the recent history of morphology linked with European and American scholars such as C. Ilockett, P. If. Matthews, J. Bybee, W. Dressier, A. Spencer, A. Carstairs-McCarthy, M. Aronoff, and others. Its primary data are taken from representative Indo-European (English, German, Spanish, Latin, Greek, Russian, Sanskrit), Afro-Asiatic (Hebrew, Arabic, Berber) and several other languages (Turkish, Chinese, Algonkian and others). The book consists of ten chapters explicating fundamental principles of morphology by means of (numbered) examples. All chapters (with the exception of the last one) arc equipped with a number of pertinent exercises often arranged in the order of increasing difficulty. Its contents are as follows: 1. Introduction 2. Grammatical Units (words, morphemes, clitics) 3. Paradigmatic and Syntagmatic Relations 4. Inflectional and Derivational Morphology 5. Inflectional Categories Associated with Nominal Elements 6. Inflectional Categories Associated with Verbal Elements 7. Morphosyntactic Properties and their Exponents 8. Morpheme and Allomorph 9. Derivational Morphology (derivation and compounding) 10. Theoretical Models of Morphology For pedagogical purposes it is necessary to deal with subject matters in individual chapters as consisting of several units (indicated by subheadings). Recommended Readings at the end of each chapter should provide further ammunition to both instructors and students of this course. II AN INTRODUCTION TO THE STUDY OK MORPHOLOGY During my twenty years of introducing the subject of linguistic morphology to third-year students of linguistics, languages, psychology, anthropology, sociology and other disciplines of Humanities and Social Sciences I benefitted enormously from various comments and suggestions made on the intermediate versions of the present textbook by my colleagues and students. At this point I want to acknowledge advice of and many helpful comments by the following scholars: Dr. A. Bartoněk (University of Brno), Dr. A. Erhart (University of Bmo), Dr. J. Hewson (Memorial University of Newfoundland), Dr. B. Joseph (State University of Ohio), Dr. Stanislav Segerl (University of California at Los Angeles), Dr. K. Strunk (University of Munich), Dr. H. Paddock (Memorial University of Newfoundland), Dr. H. Petersmann (University of Heidelberg), Dr. L. Zgusta (University of Illinois). Many of my students during the 80's and 90's made a number of observations and suggestions on the style of the four previous versions, the clarity of their exposé and the level of difficulty of some of the exercises: Julie Brittain, Audrey Dawe, Barbara O'Dea, Kathy Francis, Margot French, Bernard Kavanagh, Angela Kotsopoulos, Dorothy Liberakis, Christa Lietz, Snezana Milovanovich, Sarah Rose, Donna Starks, Margot Stuart, and others. Many thanks for focusing my attention on the student point of view in composing this textbook. And finally, I am grateful to three graduate students who formatted the fourth edition (1997) of the manuscript: Henry Muzale, Natasha Squires and Valeri Vassiliev. My special thanks arc due to my research assistant Lawrence Greening who has been involved in editing, final text formatting, indexing and preparing a camera-ready copy for publication by Lincom Europa. St. John's, April 1999 Vit Bubeník Department of Linguistics Memorial University of Newfoundland CONTENTS Preface Preliminaries Chapter 1: Introduction 1.1 Language and its Units 1.2 Units and Rules 1.3 Language and its Symbolic Aspect 1.4 [conic Tendency of Language EXERCISES Chapter 2: Grammatical Units 2.1 The Word 2.1.1 Identification and Definition 2.1.2 Phonological, Grammatical and Lexical Words 2.1.3 Internal Cohesion of the Word 2.1.4 Phonological Correlations 2.2 fhe Morpheme 2.2.1 Identification and Definition 2.2.2 Segmentability of Words 2.2.3 Allomorphs 2.3 Analysis into Roots, Stems and Affixes 2.4 Clitics 2.5 Basic Approaches to Morphology 2.5.1 Item and Arrangement Model 2.5.2 Word and Paradigm Model 2.5.3 Item and Process Model EXERCISES Chapter 3: Paradigmatic and Syntagmatic Relations 3.1 The Notion of Distribution 3.2 Paradigmatics and Syntagmatics 3.3 Markedness EXERCISES 1 3 4 5 10 12 12 12 13 14 15 16 16 17 19 21 23 24 24 25 27 29 35 35 37 40 46 IV AN INTRODUCTION TO THK STUDY OF MORPHOLOGY CONTENTS v Chapter 4: Inflectional and Derivational Morphology 52 4.1 The Scope of Inflection and Derivation 52 4.2 Some Universal Tendencies of Inflection and Derivation 54 4.3 Analysis of Inflections 57 EXERCISES 65 Chapter 5: Inflectional Categories Associated with [Nominal Elements 70 5.1 Primary Nominal Categories 70 5.1.1 Nouns and Adjectives 70 5.1.2 Pronouns 73 5.2 Secondary Nominal Categories 79 5.2.1 Gender 79 5.2.2 Number 85 5.2.3 Case 88 5.2.4 Alignment 95 EXERCISES 98 Chapter 6: Inflectional Categories Associated with Verbal Elements 106 6.1 Verb as a Primary Grammatical Category 106 6.2 Quasi-Nominal Categories of the Verb: Infinitive and Participle 107 6.3 Secondary Grammatical Categories Associated with Verbal Elements 111 6.3.1 Person and Deixis 111 6.3.2 Tense 115 6.3.3 Aspect 116 6.3.4 Mood 120 6.3.5 Voice 125 EXERCISES 130 Chapter 7: Morphosyntactic Properties and their Exponents 139 7.1 Cumulative versus Agglutinative Exponcnce 139 7.2 Fused, Extended and Overlapping Exponence 142 EXERCISES 146 Chapter 8: Morpheme and Allomorph 148 8.1 The Alternation of Aliomorphs 148 8.2 Morphological vs. Phonological Conditioning of Aliomorphs 150 8.3 Turkish Vowel Harmony 152 8.4 Morphonology 159 EXERCISES 164 Chapter 9: Derivational Morphology 166 9.1 Theory of Word Formation 166 9.2 Derivation versus Compounding 168 9.3.1 Prefixation 170 9.3.2 Suffixation 173 9.4 Compounding 175 9.4.1 Coordinate Compounds 176 9.4.2 Determinative Compounds 178 9.4.3 Possessive Compounds 179 9.4.4 Syntactic Compounds 180 9.5 Noun Derivation in Arabic 181 EXERCISES 186 Chapter 10: Theoretical Models of Morphology 188 10.1 Morphology and Formal Syntax 188 10.2 Morphology and Generative Phonology 191 10.3 Morphology in Functional Grammar 194 10.4 Natural Morphology 197 10.4.1 Universals 197 10.4.2 Typology 198 10.4.3 System-Dependence 199 10.4.4 Paradigmatic Structure 199 10.4.5 Morphological and Phonological Naturalness 200 References and Select Bibliography 203 Index of Languages 208 Subject Index PRELIMINARIES Vll PRELIMINARIES Morphology in this book will be defined as thai subdiscipline of linguistics whose subject matter is (i) grammatical units (morphemes and lexemes) and (ii) grammatical categories. The latter are traditionally divided into primary grammatical categories (i.e., 'parts of speech' such as nouns, verbs, pronouns, adjectives, adverbs) and secondary grammatical categories (such as nominal categories of gender, number and case, and verbal categories of person, number, tense, mood, aspect and voice). Morphemes are traditionally defined as the smallest meaningful elements in a language. In the seventies the transformational-generative view of morphology as a section of syntax with its emphasis on relational aspects of language led to a neglect of the study of grammatical units and categories qua forms. However, it should be made clear that all the above mentioned grammatical units and categories can be studied most legitimately in three manners: morphological (or 'formal'), functional, and syntactic (or 'positional'). Any attempts to disregard formal aspects of language by overemphasizing functional or syntactic aspects are detrimental. Inspection of various introductory books on linguistics will reveal another aspect of the current neglect of morphology. Given the fact that the English morphological system is rather poor compared with that of, say, Spanish or Latin, these books concentrate on the phonemic aspect of morphology (phonological conditioning of allomorphs). Of course, it is important to discuss such facts as the allomorphy of the 3rd Sg Prcs /s/~/z/~/oz/ in English (in he walks, loves and poaches), this, however, should not detract our attention from the morphological aspects of the categories of person and number in Spanish, which display six different inflectional forms for three persons and two numbers (amo, amas, ama, amamos, amais, aman). Thus for Spanish, our task will be to account for accentual shift (amo - amamos) in terms of morphological categories such as stem and thematic vowel (and phonological categories such as penultimate syllable). Furthermore, it is necessary to consider any linguistic structure as possessing two aspects, namely syntagmatic and paradigmatic. It is the latter aspect which was completely discarded by transformational-generative grammar, but which nevertheless is a proper domain of morphology. In the following chapters we will spend a lot of time on analyzing and constructing paradigmatic sets for the above mentioned grammatical units and categories. This approach to morphology is known as the Word and Paradigm Model (cf. Hockett 1954, Robins 1959) and this model is especially suitable for the analysis of inflectional languages which are morphologically complicated in that they do not always display a one-to-one relationship between morpheme and sememe (polysemy and polymorphy). The other morpheme-based approach, known as Item and Arrangement Model, is suitable to the analysis of agglutinating and polysynthelic .1 Phrase structure rules DF.F.P STRUCTURE SYNTAX Transformational rules SURFACE STRUCTURE Morphological and phonological rules String of sounds Fig. 0.1 An earlier Transformalional Model of language languages. In these languages the segmentation of words docs not present any major problems, since the morphemes and sememes are mostly in one-to-one relationship. It should be mentioned that the earlier Transformational Model of language did not make any provision for the formal study of primary and secondary grammatical categories. These entities were taken for granted and the emphasis was laid on the study of transformational processes. Morphology was thus viewed only as a 'surface syntactic information', as shown in Figure 0.1. In the eighties, with de-emphasis on the transformational component the place was made for meaning-based approaches to morphology. Linguists returned to a more traditional concept of morphology as a study whose domain is the relation between meaning (semantics) and form (morphology proper). Among the earlier studies along these lines, .1. Byhee's Morphology (1985) has the lasting merit of freeing the morphological theorizing from genetic and areal biases (her hypotheses about inflectional morphology are based on a sample of fifty languages). In the eighties another approach to morphology gained prominence under the title of Natural Morphology in imitation of the title Natural Phonology (Hooper's An Introduction to Natural Generative Grammar; 1976). It was developed in Germany and Austria by W. Dressier and his co-workers, and is available in the collection of their articles entitled Leitmotifs in Natural Morphology (1987). Dressier operates with several explanatory principles (universals, typology, vin AN INTRODUCTION TO THE STUDY OF MORPHOLOGY PRELIMINARIES IX system-dependency, paradigmatic structure and naturalness). The relationship between expression and meaning (Saussure's signifiant and signifie) remains the main concern. In addition, Dressier emphasizes the role of linguistic types as mediating between universal principles and language-particular behavior (universal principles of naturalness vs. system-dependent naturalness). One of the central concerns is the nature and organization of inflectional classes (the 'conjugations' and 'declensions' familiar from the traditional descriptions of many languages). The influence of these ideas changed the study of formal syntax which in the eighties avoided the treatment of purely morphological phenomena and focused instead on the so-called interface questions such as the relation between morphology and syntax or that between morphology and phonology. To follow this change of mind one may consult Jensen (1990), Spencer (1991), Carstairs-McCarthy (1992), Aronoff (1993). Aronoffs pragmatic title, Morphology by Itself, marks the complete tum-about in the attitude of Generative Grammar towards morphology in that the latter is now considered not merely as an appendage of syntax and phonology; rather the author insists that linguistic theory must allow a separate and autonomous morphological component. The reader of this manual might be surprised by the wealth of data included. This has been done on purpose, since I share Bybee's conviction (1985) that morphological universals cannot be fruitfully investigated unless we are willing to examine parallel areas of the grammars of individual languages. Morphology, of course, represents the biggest challenge to universalists' hypotheses since it is precisely here where languages differ most. Thus an important aspect of any course in morphology should be a practical and theoretical experience of analyzing phenomena which are foreign to English. Previous knowledge of the languages to be discussed is not presupposed, but the author hopes that this course will foster interest in their study. Given the recent history of morphology, it is no surprise that there are only a few textbooks introducing linguistic morphology. The studies quoted above are not suitable for a second or a third year university course. Among earlier studies Matthews' Morphology (1974) has the merit of having been unique in pursuing word-based morphology independently of the generative concerns of the seventies. More recently, Bauer (1988) attempted a synthesis in the light of the influence of Natural Morphology on the field. Bauer's monograph provides both the general background to a number of morphological studies and various details of several theoretical approaches. Neither Matthews (1974) nor Bauer (1988) contain any exercises which are essential to further progress in this field. RECOMMENDED READINGS Aronoff, Mark. 1993. Morphology by Itself. Stems and Inflexional Classes. Cambridge, Mass.: MIT Press. Bauer, Laurie. 1988. Introducing Linguistic Morphology. Edinburgh: Edinburgh University Press. Bybee, Joan. 1985. Morphology: A Study of the Relation Between Meaning and Form. Amsterdam: Benjamins. Carstairs-McCarthy, Andrew. 1992. Current Morphology. London: Routledge. Dressier, Wolfgang U., W. Mayerthaler, O. Panagl & W. U. Wurzel. eds. 1987. Leitmotifs in Natural Morphology. Amsterdam: Benjamins. Hockett, Charles F. 1954. "Two models of grammatical description". Word 10.210-231. Hooper, Joan B. 1976. An Introduction to Natural Generative Phonology. New York: Academic Press. Jensen, J. 1990. Morphology. Amsterdam: Benjamins. Matthews, Peter H. 1974. Morphology: An Introduction to the Theory of Word-Structure. Cambridge: Cambridge University Press. Robins, Robert H. 1959. "In defense of WP". Transactions of the Philological Society 57.116-144. Spencer, Andrew. 1991. Morphological Theory: An Introduction to Word Structure in Generative Grammar. Oxford: Blackwell. CHAPTER ONE INTRODUCTION 1.1 Language and Us Units Human language is a particular kind of sign system which bridges two areas of the nonlinguistic universe: non-linguistic real (or imagined) world, i.e. the things we talk about, on the one side, and physical speech sounds produced by human speech organs, on the other. Put differently, language is a mechanism that connects meaning with sound. Various linguistic schools differ in the number of language levels (subsystems) they posit. Even the number of units assigned by various linguistic schools to each linguistic level is far from being agreed upon. Since the purpose of this book is not to argue for any particular linguistic school, we will simply enumerate and briefly characterize the concepts which appear in most European and American writings. Most linguists, no matter of what persuasion, recognize the following units: distinctive features, (allo)phones, phonemes, morphophonemics. (allo)morphs, morphemes, lexemes (words), (allo)semes and sememes. The first three may be called phonological units; morphs, morphemes and lexemes may be called grammatical units; sememes represent 'semological' or commonly semantic units. (1) Language Levels (Subsystems) Units The phoneme has been defined as a family (class) of sounds in a given language that function as one and to which the speakers react as one sound. The members of this class are (allo)phones, which occur in mutually exclusive phonetic environments, and which share at least one phonetic feature. Phonetic features are building blocks of phones (e.g., /g/ is a 'bundle' of closure, velarily and voice). Two phones are said to be in contrast if they occupy analogous slots in two different morphemes or lexemes, i.e., if they occur in paradigmatic distribution, such 'is fine vs. vine. On the other hand, this opposition does not necessarily hold on the morphophonemic level, e.g., knife vs. knive-s. Here the allomorphs /najf/ - /najv/ belong to the same morpheme {najl} and the same lexeme knife whereas /fajn/ and /'vaja' are two different morphemes {fajn} and {vajn} and two different lexemes fine and vine. Thus allomorphs are not only held together by morpho-phonemes, implemented by phonemes, but they arc also linked to the same semantic unit: sememe. Morphemes are the universal units of grammatical analysis and they are established on (i) phonology (ii) morphology (hi) lexicology distinctive features, phones, phonemes morphs, morphemes lexemes (iv) semantics ('scmology') sememes 2 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY a semantic and distributional basis. For instance, go and wen-(t) are usually grouped together into one morpheme {go} because both mean "go", and distributionally they behave in exactly the same way as sleep and slep-(t). However, there is no regular morphophonemic tie between the former pair whereas there is one in the latter case in the sense that there are more examples of the alternation lil - lei as in weep and wep-(t); consequently, /go/ and /wen/ should not belong to the same morpheme {go}. Here we witness that two different morphemes {go} and {wen) can represent the same semantic unit. This fairly well-known phenomenon, neglected by earlier theoretical treatments of morphology, is called suppletion or polymorphy. The opposite phenomenon is called polysemy. These phenomena are shown in Figure 1.1. For instance, in English the morpheme {s} (= Isl ~ Izl ~ hzl) represents the 3rd Pcrs Sg of verbs, and the possessive and plural on nouns. In Arabic the same discontinuous morpheme /i-a/ may represent the singular in kitdb "book" and plural in kilab "dogs" (singular kalb). In other words, morphology and semantics are independent of each other even if they were collapsed in many introductory textbooks to linguistics. What is of particular interest in the study of morphology is the nature of the link-up between morpheme and sememe in the linguistic sign; it may be one-to-one but also two-to-one or one-to-two. This is illustrated in Figure 1.2. It should be emphasized that systematic confrontation of morphemes and sememes (the smallest elements of the semantic content of language) was done mostly by structuralist linguistic schools, whereas it was neglected by generativists, who concentrated more on relational aspects of language and tended to disregard units in favor of rules. Also it should be mentioned that the background for distinguishing morphology from semantics was provided a long time ago by the work of linguists dealing with typology of languages. In one type of language, commonly denoted as agglutinating (e.g., Turkish) each sememe is expressed by a separate morpheme, while in another type, called inflectional (e.g., Latin), one morpheme can express more than one sememe. Consider the inflectional forms of the word for "man" in Latin and Turkish given in (2): (2) Latin vir "man" vir -1 Sg/Gen Spelling Turkish adam "man" adam - 0 - in Sg Gen Spelling adamm vir - orum Pl/Gen. adam - lar - in adamlarm PI Gen In the Latin form vir-l, -("expresses two sememes (grammatical meanings) namely the singular and the genitive case; -drum expresses the plural and the genitive case. Here the relationship between morphology and 'semology' is one-to-two. On the other hand, in Turkish, each sememe is expressed by a separate morpheme: -0 (zero) expresses the singular, -lar the plural and -in the genitive case. The relationship between morphology and 'semology' is here one-to-one. INTRODUCTION MORPHEME] MORPHEME2 SEMEML MORPHEME Fig. 1.1 Polymorphy and polysemy SEMEMEj SEMEME, M Fig. 1.2 Morpheme and sememe in linguistic sign The distinctive features of sound have been studied extensively since Trubetzkoy's and Jakobson's pioneer work in the thirties. They are relatively easy to study because they are only a few (between twelve to seventeen in most languages). The distinctive features of meaning arc parallel to phonetic distinctive features, but they are much more numerous and consequently much more difficult to study. Nevertheless, much has been accomplished at the level of semantics by so-called componential analysis limited to a few areas of lexicon, such as kinship terminology, animals, colors, etc.; there are many more semantic areas which are notoriously difficult to decompose into their semantic features. Consider, for instance, how the semantic features [ t male], [ i female], and [-(-young] combine with generic meanings of animal species, as shown in (3). (3) Generic Meaning Male Female Young horse stallion marc foal goose gander goose gosling dog dog bitch puppy cat tom-cat cat kitten man man woman child The independence of morphology and semantics becomes quite clear in that the same form can represent two meanings: generic and male (dog, man), generic and female (goose, cat). 1.2 Units and Rules It should be kept in mind that linguistic units cannot exist in language without rules governing their distribution. Both units and rules (or 'items' and their 'arrangements') arc equally important in any serious attempt to describe the functioning of language. Their mutual 4 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INTRODUCTION 5 relationship is of a complementary nature, i.e., it is misleading and detrimental to try to order them hierarchically, or to over-inflate either the entilative component (unit) or process component (rule) in linguistic descriptions. The study of phonotactic rules (constraints on phonological sequences) is a domain of phonology; the study of syntactic rules (lexotactics or rules governing distribution of words in sentences) is a domain of syntax. In morphology we will be dealing with moiphotactics. rules of word formation. Derivational morphology (derivation proper and compounding) is currently treated with a strong bias towards morphophonemics; it will be shown that the semantic aspects of word formation are equally important and interesting. 1.3 Language and its Symbolic Aspect We may start this section by examining one of the many problem-ridden definitions of language (Wardhaugh 1972:3): A language is a system of arbitrary vocal symbols used for human communicatic In view of our discussion above it is preferable to view language as a 'system of (sub)systcms' (with 'levels' such as phonology, morphology, lexology, semantics). The above definition makes no provision for the societal and cultural aspects of language. The term vocal in the definition over-emphasizes the fact that the primary medium of language is sound and that writing is only a secondary representation of the primary speech. Let us now examine the remaining term arbitrary symbols which brings us back to the Saussurean concept of the linguistic sign. According to Saussurc the linguistic sign is made up of signifier and signified: signe = signifiant i signifie. ft may be remarked that the Saussurean dichotomy continues a respectable tradition of semantics starting in Ancient Greece with the Stoics that had an identical dichotomy anuatvov /semaTnon/ plus oriuouvouxvov /semainomenon/ (trnucavtiv /semainein/ "signify"). The basic assumption here is the word (i.e., the basic unit of syntax and semantics) as a linguistic sign composed of two parts: the form of the word (signifier) and what is meant (signified), or its meaning (concept). It will be shown in chapters dealing with inflectional morphology that the form of a word must be distinguished from its inflected ('accidental') forms which the word assumes when it functions in the sentence. It must also be mentioned that this terminology can be confusing since the 'form' of a word (signifier) could be taken to 'signify' both the 'concept' (mental image) and the 'thing' itself (referent). As is well-known, there exists extensive scholastic literature bearing on the relationship between 'concepts' and 'things', but all this is only of marginal interest to linguists. However, we have to keep in mind that the domain of linguistic meaning docs not include the referent. Obviously, we can deal with 'things' themselves only by means of 'concepts', as expressed by the scholastic dictum voces significant mcdiantibus conceptibus "words signify by means of concepts". Hence the line between form (signifier) and referent (thing) in the famous 'scmiotic' triangle reproduced in Figure 1.3 is only dotted. Concept Word Referent Fig. 1.3 Semiotic triangle The relationship which holds between words (as units of linguistic meaning) and things (i.e., their referents) is the relationship of reference. Linguistically, words can be viewed as forms signifying concepts, extralinguistically (i.e., referentially) as linguistic signs referring to, or naming, cxtralinguistic things. In explaining the nature of the sign, Saussure states that it is arbitrary in that one signified will have different signifiers in different languages, and almost all these signiliers were 'chosen' arbitrarily. Linguistic signs or symbols have to be learned when one acquires one's language, since they are based on a learned conventional relation; in most cases, the names we give to things arc conventional, not of natural origin. However, (here arc two oilier types of linguistic signs (as defined by linguists working in semiotics), namely icons (literally 'pictures') and indexes which have to be defined referentially. Icons express mainly formal, factual similarity between the meaning and the form; in icons, there is physical resemblance between the shape of the sign and its referent (here, the line between form and referent is solid rather than dotted). Onomatopoeic words like bang, thump, roar, etc. are examples from English for this phenomenon of direct representational connection between a word and something in the 'real' world. As is well known, all languages possess highly iconic words by which speakers try to imitate the sounds of nature. Indexes express mainly factual, existential contiguity between meaning and form. The indexical features of language include relational concepts of place and time such as here - there, now - then, I - vou - he, this - that. Their reference is multiple (e.g., you can theoretically refer to millions of addresses) and only other linguistic elements in discourse can disambiguate their meaning. 1.4 Iconic Tendency of Language Onomatopoeic words are only one subcategory of icons, those sometimes called images. Lingmsts working in semiotics (the study or signs and sign systems) distinguish two more subclasses, namely diagrams and metaphors. Diagrams are characterized by a similarity between form and meaning that is constituted by the relations of their parts. A classical example 6 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INTRODUCTION 7 of an icon of relation is proportional analogy. An infantile speaker of language who creates a new form *brung (instead of brought) completes the following diagram (or proportion): (4) rmg bring rung X Metaphor is the semantic transfer through a similarity of sense perception and is based on a perception of a functional resemblance between two objects. Its full discussion belongs to semantics. Thus linguistic signs may be classified and hierarchized in the manner shown in Figure 1.4. The insistence of many linguists on the iconic tendency of language (whereby semantic sameness is reflected in formal sameness) may come as a surprise to many students of linguistics who have learned about the arbitrariness of language vs. the iconicity of certain systems of animal communication. Linguists claim, for instance, that a bee dance is iconic (rather than arbitrary) since it directly represents its subject matter (i.e., there is a direct connection between the dance itself and the source of nectar in the number and direction of the gyrations); on the other hand, it is assumed that there is almost never any connection between linguistic sign and referent the only counter-examples being onomatopoeic words (= 'images'). However, in morphology it is comparatively very easy to find iconic correlates between linguistic signs and their referents. For instance, in English, Latin and Arabic the positive, comparative and superlative degrees of adjectives show a gradual increase in their morphological 'flesh' corresponding to the increase (or decrease) on the part oflbeir referent, as shown in (S). (5) English high high-er high-est Latin alt-us alt-ior alt-issimus Arabic kabir Vakbar ?al-?akbar Positive Comparative Superlative "deep, high" "big, great" In Latin the comparative is longer than the positive and the superlative is longer than the comparative not only by sound count but also by syllable count (2 - 3 - 4); in English both comparative and superlative have the same number of syllables but the relationship holds in sound count (3-5-6) taking the diphthong as two sounds. Also, the Arabic elative (Arabic does not distinguish the comparative and superlative inflectionally; the superlative takes the definite article ?al- while the comparative does not) is longer only by sound count than the positive. However, it is possible to find languages which show less iconicity in this respect or even where this relationship does not hold. Greek, German and Czech may be used to exemplify these two possibilities, as illustrated in (6). sign symbol image diagram metaphor Fig. 1.4 Classification of linguistic signs (6) Greek German Czech hupsčlós hoch /ho:x/ vysoký Positive hupsělóleros holier /ho:3r/ vyšší Comparative hupsělótatos hóchst /ho:xst/ nejvyšši Superlative "high" "high" "high" In German the relationship holds in sound count (3 - 4 - 5), but not in syllabic count (I - 2 -1); in Czech the comparative is shorter by both syllable and sound count than its positive, bul the superlative is longer than the comparative; and in Greek the comparative and superlative have the same number of both sounds and syllables (but both are longer than the positive). Another frequently used example of iconicity is reduplication of the linguistic sign which indicates material increase on the part of referent. The subject of reduplication is practically inexhaustible with classical examples coming from languages where this linguistic strategy is grammatically less marginal than in English. For instance, in Malay orang means "man" and orang-orang means "people"; here the multiplication on the semantic side has its counterpart in the reduplication of the linguistic sign. It may be said that this type of pluralizalion is more iconic than the usual pluralization by means of grammatical morphology found in a variety of languages such as Indo-European or Semitic. There are situations where we are dealing with repetitive actions on the referential side; one could say that nothing would be more appropriate than reduplication on the linguistic side. Hence "daily" is expressed most 'naturally' (i.e., iconically) by reduplicating in many languages, for instance, Hebrew ydm-ydm, Hindi din-elm, Malay hari-huri (cf. English day by day). It seems that we have to assume that various languages would rank differently on the scale of iconicity. Latin proved to be more iconic in comparison than Czech, and Malay more iconic in pluralization than English. Of course, to make an exhaustive statement is next to impossible because it would mean contrasting the full systemic potentials of two languages and linguists are still far from being able to do that. Nevertheless, the conclusion of this section should be rather 8 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY simple; earlier statements such as "any search for ... iconicity in language will reveal language to be almost entirely noniconic ' (Wardhaugh 1972:25) will not stand up to the cross-language evidence as indicated above. Wardhaugh argued that "the English number system proceeds as follows: one, two, three, four, ... ten ... thousand, and so on, not one, one-one, one-one-one, one-one-one-one, ... and so on. Four is not four times as long as one". However, if we look at writing systems of various languages (add 'written' symbols to the above definitions of language), it will appear that this is a rather misleading argumentation. For instance, in the Arabic writing system four I (four strokes) is four times as long as one \ (one stroke); in Akkadian cuneiform five wedges is five times as long as one wedge), cf. Figure 1.5; similarly in Roman, Chinese and Japanese writing systems. Currently, the area of writing is not considered as secondary to linguistics; writing as the study of graphic signs (and their systems) is not essentially different from the linguistic study of vocal signs (and their systems) - both are simply subdisciplines of semiotics. Jt is of interest to note that the clearest examples of iconicity come from grammatical morphology whereas lexical morphology is based largely on unmotivated arbitrary signs (see however 9.4 for iconicity of compounds). Grammatical morphemes are easily diagrammatizablc (see the various diagrams displaying grammatical categories in this book) and they occupy fixed positions within the word (the classical example being BASE + Derivational Suffix + Inflectional suffix in Indo-European languages). Furthermore, grammatical morphemes differ from lexical morphemes by a restricted use of the sound units. For instance, in the Hebrew consonantal pattern the inflectional and derivational suffixes are realized by nasals (m, n), glides (w.j), dental stop (t) and glottal sounds (h, ?) whereas the lexical morphemes are built from 19 obstruents. In the Russian consonantal pattern only four obstruents (v, /, s, x) of the twenty-four function in inflectional suffixes. In English, the inflectional suffixes are represented by alveolar stops (t, d), alveolar fricatives (s, z), and their combination (si). YYYYYYTW 1 2 3 4 5 < « «< 9 10 20 30 Fig. 1,5 Writing numerals in Akkadian cuneiform INTRODUCTION RECOMMENDED READINGS Brekle, Herbert E. 1972. Semantik. Eine Einführung in die sprachwissenschaftliche Bedeutungslehre. München: Fink. Lyons, John. 1977. Semantics. Volume 1. Cambridge: Cambridge University Press. Ogden. Charles K. & I. A. Richards. 1946. The Meaning of Meaning. 8lh Edition. London: Routledge & Kegan Paul. (First edition 1923.) Peirce, Charles S. 1955. Philosophical Writings of Peirce ed. by Justus Buchler. New York: Dover. de Saussure, Ferdinand. 1955. Cours de linguistique generale. 5,h Edition. Paris: Payot. (English translation by Wade Baskin, Course in General Linguistics. New York: Philosophical Library, 1959). Wardhaugh, Ronald. 1972. Introduction to Linguistics. New York: McGraw-Hill. 4 10 AN INTRODUCTION TO T HE STUDY OH MORPHOLOGY EXERCISES I Analyze the paradigm of the verb "to be" in Latin, Spanish and French from the viewpoint of polymorphy (suppletion), polysemy and allomorphy. Identify first the root, stem and personal suffixes where possible. French data have to be phonemicized. Latin Spanish French Present Sg 1 sum soy suis Indicative 2 es cres es 3 est es est PI 1 sumus somos sommes 2 estis sois é tes 3 sunt son sont Present Sg 1 si m sea sois Subjunctive 2 sis seas sois 3 sit sea soit PI 1 simus seamos soyons 2 sltis scais soyez 3 sint sean soicnt INTRODUCTION 3 Discuss the paradigm of the verb "to go" in Latin, Spanish and French from the pom, of view olymorphy (section, polysemy and aUomorphy. Identify fir,, the root, stem and personal suffixes where possible. French data have to be phonemic.zed. Present Sg Indicative PI imperfect Indicative PI Latin Spanish French 1 eo voy vais 2 is vas vas 3 it va va 1 Imus vamos allons 2 ilis vais allez 3 cunt van vonl 1 lb am iba allais 2 ibäs ibas allais 3 íbat iba allah 1 íbämus ibamos all ions 2 Tbätis ibais allicz 3 íbant iban allaicnt 2. Analyze the paradigm of the verb "to be" in Sanskrit, Avestan (Old Persian), and Farsi (Modern Persian) from the poinl of view of polymorphy (suppletion) and allomorphy. As in Exercise 1, identify first the root and personal suffixes. Note: The forms of the 1SI Sg and the 2lni PI Imperfect in Avestan are not documented. Sanskrit Avestan Farsi Present Sg 1 ásm i áhmi hástam Indicative 2 asi áhi hastT 3 ásti ásti hást PI 1 smáh máhi hástim 2 sthá stá hástld 3 sánti hänti hástand Imperfect Sg 1 äsam 9 búdam Indicative 2 äsíh ás budí 3 äsit äs búd PI 1 äsm a ähmä budím 2 asta 7 budid 3 äsan hón búdand i GRAMMATICAL UNITS 13 CHAPTER TWO GRAMMATICAL UNITS 2.1 The Word 2.1.1 Identification and Definition This chapter will deal wilh the identification and definition of lexical units, their structure, and their relationship to smaller grammatical units, namely morphemes (2.2) and roots, stems and affixes (2.3). The attempts at defining the two primary units of grammatical analysis, the word and the morpheme, are essentially of a circular nature, since we must presuppose the knowledge of the morpheme if we want to define the word, and vice versa. Furthermore, we have to keep in mind that lexical and morphological analysis of a language is intimately connected with its syntactic and phonological analysis, i.e., the lexical units are governed by the syntactic rules when they are combined in sentences and morphological units cannot be discussed without paying due attention to their phonological substance. Conversely, working in the realm of phonology or syntax, it would be impossible to state a number of significant generalizations without reference to the notion of the word. It is customary to include both lexical decomposition of sentences into words, and morphological decomposition of words into morphemes under the label of grammatical analysis. The word is simply the unit par excellence of the grammatical analysis as a final point in syntax and as a starting point in morphology. The centrality of the concept of word will be even more obvious if we think of the grammatical analysis as the central linguistic activity flanked by discourse analysis (decomposition of texts into sentences), and phonological analysis (decomposition of morphemes into phonemes and words into syllables), as in Figure 2.1. Word thus appears to be a unit intermediate in rank between the sentence and the morpheme. It is worth mentioning that this unit is recognized by the conventions of various writing systems (with some notable exceptions such as that of Sanskrit). Furthermore, traditional grammatical lore deals with word forms, not with morphemes. Paying due attention to all the three aspects of wordhood (i.e., phonological, grammatical and semantic) we may adopt a 'traditional' definition of the word: word may be defined as the union of a particular meaning with a particular complex of sounds capable of a particular grammatical employment. This definition may strike us as somewhat pedantic but it is important to realize that neither one of these three aspects should be overemphasized or omitted if one wants to shorten this definition. This error was committed even by Leonard Bloomfield in his influential book Language (1935). His definition of word excels in Paninian brevity: the word according to Bloomfield was a "minimum free form". The terms 'free' and 'bound' are more commonly used in classifying morphemes (typically, the grammatical ones such as -s in the plural are bound). Discourse Analysis Grammatical Analysis o, Phonological Analysis Fig. 2.1 Linguistic decomposition Looking at the higher level, forms which occur as sentences are free forms; a free form which consists entirely of two or more lesser free forms (e.g.. poor John) is a phrase; and, finally, a free form which is not a phrase is a word. According lo Bloomfield (1935:178), "a word is a free form which docs not consist entirely of (two or more) lesser free forms; in brief, a word is a minimum free form". All the units of the ranks lower than words (i.e., morphemes and phonemes) are then bound since they never occur alone as sentences. Bloomfield himself was aware of various difficulties connected with this definition. For instance, the articles the and a, though rarely spoken alone, play the same part in the F.nglish language as the forms this and that, which freely occur as sentences. Hence the traditional classification of the articles as words. Bloomfield was preoccupied more with words under their phonological aspect than their grammatical and lexical properties. These terms will be discussed in the next section. 2.1.2 Phonological, Grammatical and Lexical Words As we saw in the preceding section, the term word turned out to be surprisingly problematic when it came to its definition. We may wish to approach the whole problem from a different angle and say that the word is a basic unit at the level of lexical analysis (lcxology) called lexeme. Lexeme can be thought of as an abstract unit which occurs in different inflectional forms; e.g., the lexeme WRITE can be realized according to the morphosynlactic rules as writes (3rJ Sg Present), write (all other persons Present), wrote (ihe past lense o( write) and written (the past participle of write). Ihe four inflectional forms of the lexeme WRITE may in their turn be referred to as four different grammatical words. The fact that these four grammatical words have to be realized at the phonological and orthographical level of the language brings us to the phonological and orthographical aspects of the word. For instance, the phonological word /rajts/ 14 AN INTRODUCTION TO TilE STUDY OF MORPHOLOGY GRAMMATICAL UNITS 15 and its orthographic counterpart writes represent a particular grammatical word which can be categorized as the third person singular present indicative. Needless to say, the relationship between phonological and grammatical words need not be one-to-one. The phonological word IxmiI represents two different grammatical words: the present tense of run and the past participle of run (i.e., two different inflectional forms of the lexeme RUN); the phonological word /hit/ represents three grammatical words (plus the past tense of hit), etc. 2.1.3 Internal Cohesion of the Word One of the characteristics of the word is that it tends to be internally stable, i.e., that the morphemes constituting a complex lexical item occur in a fixed order. The internal cohesion of the word contrasts with its positional mobility in the framework of the sentence. For instance, we may topicalizc the sentence mobility between the grammatical morphemes of the plural and past is actually quite common (and meaningful), as shown in (4). (4) ah - yor take - PROGRESSIVE "they were taking" lar PLURAL di PAST ah - yor - du - lar take - PROGRESSIVE - PAST - PLURAL "they were taking" (1) His carelessness was astonishing 12 3 4 by beginning with the complement astonishing: (2) Astonishing was his carelessness 4 3 12 The permutation 4 3 12 yields an acceptable English sentence. However, the three morphemes of carelessness (care + less + ness) cannot be permuted. Similarly in Latin the sentence of three words may be permuted by topicalization in at least four ways, as shown in (3). (3) Ignis camera coquet 1 2 3 Ignis coquet camem 1 3 2 Carnem ignis coquet 2 1 3 Carnem coquet ignis 2 3 1 "The fire cooks the meat" However, the two morphemes of either the nouns ignis and carnem or the verb coquet cannot be permuted (*isign or *etcoqu arc impossible). It appears that the positional mobility of meaningful elements is found at the level of phrase (and sentence), whereas the lower level of word shows the internal positional mobility only exceptionally. For instance, in the sequence of several inflectional suffixes variations sometimes may occur. In Turkish the phenomenon of positional For further discussion see 4.2. 2.1.4 Phonological Correlations Phonological criteria may be used for the segmentation of phonetic strings, since in many languages the word is phonologically marked in some way. Of course, the use of phonological criteria for segmenting the phonetic strings presupposes some knowledge of the phonological system of the examined language; on the other hand, the phonological analysis of the language in question can advance only after a sufficient number of word and morpheme boundaries have been established. This fact demonstrates nicely the interdependence of grammatical and phonological analyses in that linguists cannot hope to complete first the grammatical analysis and then move on to phonology, or vice versa. Several kinds of phonological evidence arc relevant in segmenting the phonetic strings in words. For instance, a great number of languages have the so-called word accent, which means that most words (with the exception of clitics, cf. 2.4) are accented on one and only one syllabic (the accent may be of dynamic or melodic nature). In French the accent can fall only on the last pronounced syllable of a word (with the exception of the reduced schwa fa|); knowing this one may conclude that there is a word boundary somewhere before the next unaccented vowel. Analyzing Czech data, where the accent falls on the initial syllable of the word, it may be assumed that the following phonetic string contains at least four words: (5) Janudefilpsaklackem. "John hit the dog with the stick" A more complicated situation arises in languages where the accent is fixed with reference to the end of the word by the length or weight of the last or the penultimate syllable. For instance, in Ancient Greek the accent cannot fall on the antepenultimate syllable if the ultima contains a long 16 ■i INTRODUCTION TO THE STUDY OF MORPHOLOGY GRAMMATICAL UNITS 17 vowel; in Latin, on the other hand, the antepenultimate syllable can be accented provided the penultimate syllable is light (i.e., open and containing a short vowel). Analyzing Latin data it may be assumed that the following phonetic string contains at least three words: (6) Amlkuslupumnekavit "The friend killed the wolf It may be of interest to consider orthographical correlates of the phonological criteria since some writing systems employ graphemes which display different forms for initial, medial and final position in the word. For instance, in Biblical Hebrew the fricatives [P \ 6 x f 9] do not occur initially unless the preceding word ends in a vowel and these two words are in close connection, e.g. bzZ /bafJel/ "Babel" but "?333 /bapa(3el/ "in Babel" and brCQI /uP3|Ja|$el/ "and in Babel"; the plosives in initial position are indicated by the dot placed inside the letter. The Arabic system of graphemes is made up of four allographs (or only two in some cases) used for the three positions mentioned above plus the allograph when the grapheme is to be realized in isolation. Students of Greek will be reminded of the letter o (sigma) which has the allograph c when occurring in final position in the word. 2.2 The Morpheme 2.2.1 Identification and Definition Morphemes are traditionally defined as the smallest meaningful elements in a language. Grammatical analysis (i.e., analysis dealing with meaningful elements) of any language has to stop here since the units of lower ranks, namely syllables and phonemes, are non-meaningful. It should be emphasized that defining the morpheme as the minimal unit of grammatical (i.e., meaningful) analysis is conditioned by some explicit or implicit reference to the word as a grammatical unit of next higher rank (2.1.1). Morphemes are simply the units of lowest rank out of which words are composed. To use as an example one of the longest words in English: anti + dis + establish + menl t- ari ■+ an + ism It may be said that each one of the seven constituting parts of this word is associated with a particular meaning and that we are dealing with seven morphemes. Each one of them has a particular distribution and also a particular phonological (and orthographical) shape. In the era of American structuralism it was customary to segment sentences transcribed phonetically into strings of morphemes and thus by-pass the level of word. This procedure was developed when working on languages of agglutinative and polysynthctic typologies; the results are strange for flective languages such as English. For instance, Fromkin and Rodman in their introductory textbook to linguistics (1974:103) proposed to analyze the sentence The boys tossed Mary's hat over the fence in this fashion: (7) The + boy + s i toss + ed + Mary + s +- hat + over + the + fence. The level of word is by-passed and we reach the level of sentence directly from the level of morpheme. Despite the fact that some words in this sentence consist of only one morpheme wc have to introduce word boundaries (#) to avoid the confusing of the lexical and grammatical morphemes. (8) The # boy i s # toss i ed # Mary + s # hat # over # the # fence. 2.2.2 Segmentability of Words It is important to realize that whether a word can be divided into smaller meaningful elements is only a matter of degree. Of course, typical examples from a number of flective languages are those which are determinate with respect to segmentation, where the identification of morpheme boundary simply consists of putting the plus sign between two segments: hat * s, walk -t ed, ete But there arc quite a few English nouns and verbs which cannot be analyzed in this way. por instancc, the irregular plurals men, geese, mice have to be analyzed in terms of process winch replaces the vowel occurring in the singular form lie, u, aw/ by another vowel which is appropriate in the plural form /r, i, aj/. If the vast majority of English nouns are plurali/ed segmcntally (i.e., by adding the morpheme -s), there arc some exceptional cases where ihe pluralization is a matter of morphological processes, such as x - ► e (man), u ► i (goose), aw -• aj (mouse), etc. Similarly, the vast majority of English verbs form their past tense segmenia][y by adding the morpheme led}. But there are about a hundred of the strong verbs which form their past tense irregularly by a morphological process which replaces the vowel occurring in the singular form by another vowel which is appropriate in the plural form: (9) aj * ow (ride - ► rode) u * ow (choose —* chose) aj * aw (find —♦ found) Thus in all these instances we are not dealing with the addition of segments but with replacing one segment by another one, hence the use of the arrow. Some linguists call these processes rather misleadingly process morphemes (confusing units and processes). Of course, it can be maintained that the addition of segments represents replacive processes in that the so-called zero morpheme (O) of the unmarked form is replaced by the real morpheme of the marked fonii; (10) 0 (hat -+ hats) (man -» men) 18 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY GRAMMATICAL UNITS 19 Intuitively, however, we feel that these two strategies (adding the segments and replacing one segment by another) are sufficiently distinct to justify the introduction of the terms external versus internal inflection. External inflection consists of adding segments whereas internal inflection is realized through the modification of the phonological shape by replacing one of the internal segments by another one. In English there are examples of both external and internal inflection taking place in one word: child - children (where /aj/ is replaced by hi and /ran/ is added) and keep - kept (where I'll is replaced by lei and HI is added). To demonstrate that internal inflection may be central to the nominal and verbal sub-systems we may look at Arabic, which distinguishes about twenty productive patterns of internal inflection. For instance, the plural of kitdb "book" is kutub.jabal "mountain" Jibdl, nasr "vulture", imsiir, etc. The replacive processes in Arabic arc more complicated than in English, as can be seen. Indeed, it is necessary to talk about the consonantal root ('trililterar) into which various vocalic patterns are interdigitatcd: (11) root K T B singular i ä plural u u J B L a a i ä NSR a 0 u a The replacive processes in the verbal system are equally complex, e.g., kataba "he wrote", kutiba "it was written", yaklubu "he writes",yuktabu "it is written": (12) K T B past active a a present active past passive u i present passive (-a in katab-a and -u myaktub-u are suffixes). Y K T B a li 11 u Finally, it is possible to find so-called suppletive words which arc indeterminate with respect to segmentation: go vs. went, bad vs. worse, French ceil "eye" vs. yeux, Berber tit "eye" vs. aln. Undoubtedly, went stands in the same grammatical relationship to go as walked is to walk, but whereas there is phonological resemblance between the members of the latter pair there is none whatever between went and go. Similar examples could be provided from any language. Students of Greek will be reminded of their difficulties with learning the heavily suppletive morphology of the aorist (for instance, elipon "I left" stands in the same grammatical relationship to leipd "I leave" as eidon "I saw" to hard "I see"). We have to conclude that in many languages there are words winch cannot be segmented into parts and that the morpheme does not always have to be an identifiable segment of the word. Still, we would maintain that all the above unsegmentable words enter into a proposition of grammatical equivalence with the segmentable words: (13) hat hats man men goose geese walk walked write wrote find found Clearly, in these equations we may replace the lexemes by arbitrary' symbols (using capital letters) and the exponents of the grammatical meaning (number, tense) by another set of arbitrary symbols (using small letters): (14) Ax Bx C_x Ay By Cy Pm R m Sm Pn Rn S n In (14), each word is analyzed into two components — its lexical and its grammatical meaning (singular = x, plural ~ y, present = m, past = n). All the nouns on the top line of the equation have the component x (singular), on the bottom line y (plural); all the verbs on the top have the component m (present), and on the bottom n (past). Thus it can be maintained, as Lyons (1968:183) puts it, that "the morpheme is not a segment of the word at all ... but merely its 'factorial function'". What is distributed in the word are sememes rather than moiphemes, or, its lexical and grammatical meaning. It is only when the lexical and grammatical meaning are matchable with distinct segments, i.e., when the word is segmentable into parts that these can be referred to as morphs. The word wrote, which cannot be segmented into two morphs, still represents the combination of two sememes: write (lexical meaning) + Past (grammatical meaning). On the other hand, the word walked is segmentable into two morphs walk + a/and, of course, these arc exponents of two sememes: walk (lexical meaning) + Past (grammatical meaning). It will be argued correctly that the morph l\l occurring in walked (or lál occurring in begged) are indicative of the phonic substance of the morpheme (d}. However, we should avoid segmenting wrote in this manner: write + ed, since here the grammatical meaning of the past is expressed by the process of replacing /aj/ with /ow/ (see further discussion in 8.2); sec Figure 2.2. 2.2.3 Allomorphs In the preceding section we noticed that the regular past tense morpheme |d \ can be realized by two different allomorphs III and Id! in two different contexts: the former after voiceless consonants and Id/ elsewhere. If we examined further examples (such as petted, padded) we would discover that there is a third allomorph /ad/ occurring after /t/ and Id/. This is summarized in (15). 20 INTRODUCTION TO Till: STUDY OF MORPHOI GRAMMATICAL UNITS 21 Semantics Morphology Past Suppletion Phonology /t/-/d/~/ad/ Replacing /aj/ with low/ went Fig. 2.2 Morphological units and processes (15) (i) /ad/ after HI, Id/ (ii) IV after voiceless consonants other than IV (iii) Idl elsewhere It is customary to call these three alternative representations of the same morpheme allomorphs. For instance, the regular past tense morpheme in English, which may be referred to as {d}, is regularly represented by three allomorphs /ad/, III and Id/. It is important to realize that all these three allomorphs are pbonologically conditioned, which simply means that the selection of one of them is determined by the phonological shape of the preceding segment: the voiceless consonant selects IV, the voiced one Idl, and in the case of the homorgamc III and Idl at the end the third allomorph /ad/ is selected. Using the same procedure we would be able to establish that the plural morpheme jz} is realized by three allomorphs, namely Izl, Is/ and /az/. (See the analysis under 8.1). These three allomorphs are again phonologically conditioned, in the sense that the latter is selected after sibilants and affricates, and the former two after voiced and voiceless segments, respectively. (The voiceless allomorph Is/ is not selected after voiceless sibilants and affricates since this is the environment where /az/ is selected). So far, all these alternations in the phonological shape of these two morphemes were explicable in purely phonological terms without reference to the notions of morphology. We may wish to add other pluralizing elements to the list of the three regular allomorphs of the plural morpheme: {en} as in oxen and brethren. {0} as in deer and sheep, {a} as in data and criteria, etc. (See their complete list in 8.2). Examining the suffix -en in oxen we obviously cannot say that /an' is a phonologically conditioned variant of the morpheme {z}. First of all, Izl and Inl are not phonologically similar enough (the only feature they have in common is voice); second, similar words such as box select the regular allomorph /az/. Similarly, it is equally easy to argue against to/ in data being an allomorph of the regular morpheme. When some morphs are distributed in this manner we have to acknowledge the fact that they are not conditioned phonologically but lexically, in the sense that the word ox selects the pluralizing morpheme {n( and the word datum keeps the Latin pluralizing morpheme -a Izl. See the more detailed discussion in 8.2. 2.3 Analysis into Roots, Stems and Affixes Inflectional and derivational morphemes (cf. 4.1) are traditionally classified by their position with regard to the root (or the base). If they precede the root they are called prefixes; if they follow the root suffixes; and if they are placed inside the root infixes. In the word cats, for instance, cat is the root and -s is an inflectional suffix. In the word careless, care is the root (or the derivational base) and -less is a derivational suffix. Since English does not have inflectional prefixes, wc may look at Arabic where in yaktubu "he writes" ya- is an inflectional prefix (meaning 3rd person), KTuB is the root (note that the root in Arabic is a discontinuous morpheme), and -u is an inflectional suffix (meaning singular and indicative). In the word bemoan, moan is the root (or the derivational base) and be- is a derivational prefix. To exemplify infixes we may look at Latin or Arabic. For instance, in Latin the morpheme -n- which appears in the present tango "I touch" is an infix; notice that the perfect of the same verb tetigi"! have touched" docs not show it. The root is then said to be discontinuous ta-n-g. In Arabic the infix -t- derives reflexive or passive forms from the transitive verbs: FaHiM "understand" vs. fiJF-t-aHaM "comprehend". The root FHM is again discontinuous: F-t-aHaM. 'Interdigitated' vocalic patterns in Arabic roots are sometimes called transfixes (double or triple infixes): safiR "poet" vs. suYaRa? "poets". Another type of a double affix is called circumfix; for instance in Berber the circumfix (-/ (prefix and suffix added simultaneously) derives feminine nouns from their masculine counterparts.- amdakid "friend" --> t-amdakul-t "friend (Fern)". A less fitting example would be the German passive participle, e.g. ge-schlag-en "hit" because here the prefix and suffix do not display the same morpheme. And finally there is also a so-called interfix seen in compounds such as English huntsman or German Tag-e-buch or Tag-es-buch "diary" (lit. day-INTERFIX-book). The interfix should not be confused with an infix which by definition splits the lexical root in two segments. Sanskrit and other flective languages such as Latin and Greek have additional kinds of affixes, which are added to the roots, and inflectional affixes are then added to the complex form. This additional affix is known as a thematic vowel and the resulting complex form as a stem. For instance, in the Latin accusative singular puellam, puell- is the root, the added vowel -a is a thematic vowel, and -m is the inflectional suffix marking the accusative singular. (It may be noted that a traditional school analysis keeps the ending -am unanalyzed). In the Latin form laudamus "we praise", laud- is the root, the added vowel -a is a thematic vowel and -mus is the inflectional suffix marking the Is' person plural. This is shown schematically in Figure 2.3. The stem can be formed even by prefixing stem-forming elements, for instance, in Arabic yankalibu "it is written", where Ka'J'iB is the root to which the stem-forming prefix n (accompanied by the transfix a-i) is added {ya is an inflectional prefix meaning 3,J person and u is an inflectional suffix meaning singular and indicative). This is shown in Fig. 2.4. From these examples it appears that there is 22 ' INTR°WCTION TO THE STUDY OF MORPHOLOGY (GRAMMATICAL) WORD Thematic Vowel -a Inflectional Suffix -mus Fig. 2.3 Grammatical word in Latin (GRAMMATICAL) WORD Inflectional Suffix Inflectional Derivational ROOT Prefix Prefix ya- n- K-T-B -u Fig. 2.4 Grammatical word in Arabic a universal tendency for derivational affixes (prefixes and suffixes) to occur closer to the root than inflectional affixes (prefixes and suffixes); derivational elements tend to be central and inflectional elements peripheral. The root and derivational affixes (if any) constitute the stem, and the inflectional affixes are prefixed or suffixed to it (see the discussion in 4.2 for some controversial examples). Figure 2.5 visualizes this. At this point we have to distinguish more clearly between inflectional and derivational affixes. Inflectional affixes are those that mark secondary grammatical categories: gender, number, case with nouns (cf. 5.2), and person, tense, aspect, mood, voice with verbs (cf. 6.3). Defining negatively, it may be said that derivational affixes are those that are non-inflectional. GRAMMATICAL UNITS 23 Inflectional Prefix Derivational Prefix ROOT Derivational Suffix STEM Inflectional Suffix (GRAMMATICAL) WORD Fig. 2.5 Inflectional and derivational affixes Derivational affixes have the potential to change the membership in the group of primary grammatical categories (see under 5.1 and 6.1). For instance, the addition of the derivational suffix -ic to the noun democrat results in an adjective; the addition of the derivational suffix -ize results in a verb. On the other hand, the verb democratize inflected for the past tense remains a verb, or an adjective inflected for gender, number and case to agree with its head noun (in Latin) remains an adjective. The distinction between inflection and derivation may be blurred in some cases. For instance, the nasal infix of Latin mentioned above is a derivational rather than inflectional affix even if it serves to mark the secondary grammatical category of aspect (i.e., its presence does not change the grammatical class — both the present tango "I touch" and the perfect teligi- "I have touched" are only grammatical forms of the same verb). Perhaps the best we can do is to recall the traditional grammatical theory according to which inflection was considered to be any change made in the form of a word to express its relation to other words in the sentence. Hence all the grammars of flcctive languages include lengthy sections describing the declensions of nouns, adjectives, and pronouns, and the conjugations of verbs, according to selected models of formations, called paradigms. Shorter sections would be devoted to the study of various derivational processes, by which new words are formed from existing words (or roots) — verbs from nouns, nouns from verbs, etc. (See the discussion under 4.1). 2.4 Clitics At this point a mention must be made of certain word classes (parts of speech) which traditionally are viewed as falling between full-fledged words and grammatical affixes. They are referred to by the term clitics or grammatical words (not to be confused with grammatical words discussed in 2.1.2). Full-fledged words (nouns, verbs, adjectives, and adverbs) carry lexical accent and their lexical meaning is of symbolic nature; on the other hand, adpositions (prepositions and postpositions), articles, particles and pronouns do not (usually) carry accent and their lexical meaning is of indexical nature (cf. 1.3). The absence of then own lexical accent makes them clitics in that they have to "lean" against full-fledged words (from Ancient Greek eykXlvou.kvci /enklinomena/ "(words) leaning against"). 24 AN INTRODUCTION TO THE STUDY 01: MORPHOLOGY GRAMMATICAL UNITS 25 For instance, the definite article in English is a proclitic the^mdn while its counterpart in Rumanian is an enclitic dm-ul. Upon closer examination we may classify the English article as more 'word-like' than the Rumanian one, because of its ability to be separated from its noun by an adjective: the good man; in Rumanian, the article behaves almost as a suffix in being always attached to the noun or the adjective (both phonologically and orthographically): dm-ul bun ~~ bun-ul om. Other languages may repeat the article in the noun phrase, e.g. Biblical Hebrew hd-'/S hat-tofi or Ancient Greek ho aner ho agathos (but also ho agathos aner). Typical examples of clitics are short pronominal forms as in Ancient Greek moi "me" and soi "you" (vs. full form emoi "to me" and soi "to you"); contrast dos mot "give me (it)" or "gimme (it)" with dos X emoi "give X to me"; similarly, areskei moi, lit. it pleases me "1 like it" vs. emoi areskei, lit. it pleases ME (not someone else) "1 (emphasized) like it". Their intermediate status between full words and affixes is also reflected in varying spelling conventions of various languages; for instance, French hyphenates pronominal clitics in the imperative donnez-nous-la "give her to us" whereas Spanish spells them together with their verb damelo "give me it" (not *da me Id). However, both French and Spanish spell their pronominal clitics as full words in preverbal position.ye le vols and lo veo "I sec him", respectively. On the other hand, Semitic languages spell their pronominal clitics always as clitics. The same is true of the conjunction "and" and various prepositions which are always spelled as proclitics in Arabic and Hebrew. On the Indo-European side, in Latin the conjunction que "and" is spelled as an enclitic (e.g. pater materque "father and mother") but not the other copulative conjunction: pater et mater. In the same language, the postposition cum "with" is spelled as an enclitic with pronouns: mecum "with me" but as a full word with nouns: cum putre "with the father". To express the intermediate status of clitics between full words and affixes it is customary lo place the equation sign (=) between the word and the clitic, e.g. the=man; morpheme boundary is specified by +; and word boundary by #, e.g. the^good#friend+s. As far as the accentual properties of clitics and affixes are concerned, it is usually claimed that, unlike full words, they do not possess any accent. Of course, there are all sorts of counter-examples (e.g. Ancient Greek logos=tis "a certain word" vs. logoi-tines "certain words"; Latin laud+o "I praise" but laud-d+mus; Ancient Greek paideu+tos "educated"; etc.) But, on the whole, the above statement may be used as a first approximation in their identification. 2.5 Basic Approaches to Morphology There are three basic approaches to morphology: Item and Arrangement Model (IA), Word and Paradigm Model (WP), and Item and Process Model (TP). 2.5.1 Item and Arrangement Model The Item and Arrangement Model is a purely linear model which seeks to split each word (more specifically, each phonological form of a word) into a number of independently functioning segments (morphemes). This model operates with (i) a set of morphemes (ii) a set of phonemes (iii) a relation of sequence. This model is successful in the description of agglutinative and polysyntlietic languages where the phenomena of suppletion and discontinuity (see under 2.5.2) are non-existent, or are only very marginal. For instance, Cree verbal forms niwapamaw "I see him" and niwdpamik " he sees me" would be analyzed as follows: (16) ni 1SG ni 1SG wap see wap see am 3SG am 3SG aw "1 see him" 1 » 3 ik "He sees me" 3 —> 1 (-ik ~ inversion marker) In this model morphemes and sememes are paired in an exhaustive one-to-one correspondence. In the past this model was overused to the detriment of the useful distinction between morphological and syntactic structure, i.e. word vs. phrase/clause/sentence. If transferred from the analysis of the agglutinative and polysyntlietic languages to that of fieclive languages, one may become oblivious of a crucial role played by the word in most grammatical theories. In practical terms, one has to operate with two types of boundaries: + morpheme boundary and # word boundary, in the analysis of inflecting languages. Contrast (17) Cree: ni + wap + am + ik Polysyntlietic English: he # see + s # me Inflecting It would be wrong to analyze English ■■ he + .sec + s »- me. 2.5.2 Word and Paradigm Model The Word and Paradigm Model is a hierarchical (vertical) model which assigns a central role to the word as well as to its constitutive elements (morphemes). This model is successful in the description of inflecting (flcctive) and introflccting languages where the phenomena of infixation and transfixation create so-called discontinuous morphemes. Examine the following set of data from Latin: (18) rumpó "1 break" relinquo "I leave" fundo "I pour" riipi "I have broken" reliqui "I have left" fudi "I have poured" 26 AN INTRODUCTION TO 1HH STUDY OP MORPHOLOGY In the left column all the forms have one thing in common vis-á-vis their counterparts in the right column: they display a nasal infix before the final consonant of the root. Contrast ru + m + p with rup etc. However, to acknowledge the fact that there is a nasal infix in rumpo in terms of sequencing (so far the IA model) is not enough. One has to specify its meaning and this can only be done by contrasting the forms in the left and right columns. The conclusion based on the examination of English glosses would be that the nasal infix marks non-perfect forms or, to express it in negative terms, its absence marks perfect forms. This is the essence of the WP model: to come up with a grammatical solution one has to juxtapose, or rather superimpose, two related forms: (19) aquam (fundoj I fudi 1 1 pour 'I have poured the water" Another aspect of the WP model which is totally absent from the 1A model is its preoccupation with irregular and suppletive morphology; suppletion is the phenomenon where totally different forms may belong to the same paradigm, e.g., English go - went, he - was (cf. 2.2.2). To stay with Latin verbal morphology examine additional data with non-perfect vs. perfect contrast: (20) rumpo "I break" tango "I touch" laudo "I praise" duco "I lead" ferro "I carry" rupi "I have broken" tetigi "I have touched" laudavl "I have praised" duxi "1 have led" tull "I have carried" GRAMMATICAL UMTS 27 The transfixed (interdigitated) grammatical morpheme i-d is discontinuous, and so is the lexical consonantal root K-T-B). To come up with its grammatical analysis one has to apply the WP model, i.e. one has to juxtapose, or superimpose, the singular and plural forms. The solution is that i-d marks singular in "book" and "donkey" but plural in "mountain"; hence the need to classify Arabic nouns into declensional paradigms (cf 2.2.2). 2.53 Item and Process Model While the IA and WP models are diametrically opposed (horizontal syntagmatic versus vertical paradigmatic model) the concerns of the Item and Process Model are endemic to both IA and WP models. Its main concern are the morphological processes undergone by both lexical and grammatical morphemes. For instance, when contrasting the plural formation processes of inflectional vs. agglutinating languages one may profitably draw on this model. Examine the following sets of Latin and Turkish data: (22) Latin amic-us amic-I pater patr-ěs cas-a cas-ae vic-us vic-i Turkish dost dost-lar baba baba-lar ev cv-lcr koy kóy-lcr "friend" - "friends" "father" - "fathers" "house" - "houses" "village" - "villages" in Latin one will notice morphophonemic change in the ***paler versus^; in Turkish the plural suffix Mar} undergoes the process of vowel harmony Mar/ - Mer/, C. 8... Morphological irregularities of inflecting languages are captured by allocating forms with similar irregularities to different morphological paradigms. In our data laudd-v-l which forms the perfect by attaching the suffix -v belongs to Conjugation I, whereas duk-s-i (spelled duxi) which forms the perfect by the suffix -i' belongs to Conjugation III. The verbs which form their perfect by removing the nasal infix from the root (te-tig-Twilh partial reduplication and vocalic change) belong also to Conjugation III. In the last example, however, there is no resemblance whatever between the root fer- and its counterpart in the perfect, tul-. Transfixation found in introflecting languages such as Arabic splits not only the verbal or nominal root but also creates discontinuous grammatical morphemes. Examine the following set of Arabic plural forms with their lexical roots capitalized: (21) KiTáB "book" KuTuB "books" J a B a L "mountain" J i B a L "mountains" H i M á R "donkey" H a M!R "donkeys" 28 AN INTRODUCTION 10 THE STUDY OF MORPHOL OGY RECOMMENDED READINGS Bloomfield, Leonard. 1935. Language. London; Allen and Unwin. (Revised edilion.) (Chaplers 13-16). Fromkin, Victoria & Robert Rodman. 1974. An Introduction to Language. New York: Holt, Rinehart and Winston. Gleason, Henry A. 1961. An Introduction to Descriptive Linguistics. New York: Holt, Rinehart and Winston. (Revised Edition.) (Chapters 5 and 6). Harris, Zelig S. 1951. Methods in Structural Linguistics. Chicago: University of Chicago Press. Hockctt, Charles F. 1958. A Course in Modern Linguistics. New York: The Macmillan Company. (Chapters 14, 15, 19, 20). Klavans, Judy. 1985. "The independence of syntax and phonology in cliticization". Language 61.95 120. Krámský, Jiří. 1969. 77ie Word as a Linguistic Unit. The Hague: Mouton. Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. (Sections 3.2 and 5.3). Nida, Eugene A. 1949. Morphology: A Descriptive Analysis of Words. 2nd Edition. Ann Arbor, Mich.: University of Michigan Press. GRAMMATICAL UNIT'S 29 EXERCISES (A) Identification of words 1. Latin is a language that permits many variations in word order. Positional mobility is therefore highly relevant to the location of word boundaries in Latin sentences. Use the accentual criteria discussed in 2.1.4 in isolating the words in the sentences below and giving their grammatical meanings. (1) Reginamagnafeminaerosamdedit. "The great queen gave a rose to the woman". (2) Feminaededitrosamreginamagna. "The great queen gave a rose to the woman". (3) Patcrsapiensfilioepistulammlsil. "The wise father sent a letter to the son". (4) Epistulamsapienspatermlsitfilio. "The wise father sent a letter to the son". (5) Magnusrexgladiumfiliodedit. "The great king gave a sword to the son". (6) Flliomisitepistulamrexsapiens. "The wise king sent a letter to the son". (7) Gladiummilesmagnusrcgidedit. "The great soldier gave a sword to the king". (8) Rexsapiensmagnaereginaerosamdedit. "The wise king gave a rose to the great queen". (9) SapientipatriepistulamfTliusmTsit. "The son sent a letter to the wise father". (10) Magnareglnarcglsapientigladiumdcdit. "The great queen gave a sword to the wise king". 2. Isolate the following words in the Russian sentences below and state their meaning. Your analysis should assign every phonological segment in Russian to a certain English word. There should be no residues. (a) here (b) friend (c) dog (d) mother (e) father (f) will come (g) (h) (i) (j) sick my tomorrow 30 AN INTRODUCTION TO 1 HR STUDY OF MORPHOLt GRAMMATICAL UNITS 31 (1) býlzdésJdrúg "(A) friend was here". Turkish (2) sAbákabylábAFná "(The) dog was sick". (1) Ankara ve Izmire gidccegim (3) drúgbýlbólen "(The) friend was sick". (lit) Ankara+NOM and = Izmir+DAT go+PROGR+lSG (4) SAbákgbylázdés1 "(The) dog was here". "I am going to Ankara and Izmir". (5) móJAtétspriďótzáftra "My father will come tomorrow". (6) mAjámátJbAlJná. "My mother is sick". (2) uzun yol uzun yollar (7) záftropridJótmAjamátJ "My mother will come tomorrow". (lit) long road long roads (8) mójAtétsbólJcn "My father is sick". "the long road" "the long roads" 3. Isolate the words in the following Czech sentences and state their meanings. Your analysis should assign every phonological segment of every sentence to some word. There should be no residues. Latin (3) Ancyram Smyrnam-que (lit) Ankara+ACC IzmiH ACC - and "I am going to Ankara and Izmir". co go-rlSG (4) (a) (b) (c) (1) (2) (3) (4) II. (a) (b) (c) here man cat (d) (e) tomorrow (d) mother (e) my (f) sick was biltuclov'ek kóčkabílanémocna člčVekbílnémoccn bílatukóčka father is sick (1) mujótecpříjdezítra (2) mojemátkajenémocna (3) zítrapríjdemojemátka (4) mujótecjenémocen "(A) man was here". "(The) cat was sick". "(The) man was sick" "(The) cat was here". "My father will come tomorrow". "My mother is sick". "My mother will come tomorrow" "My father is sick". The make-up of a word in Turkish differs crucially from its make-up in Latm. Discuss this statement with regard to (a) phonological and (b) grammatical criteria for wordhood. You can use the following data for (b): via longa "the long road" viae longac "the long roads" (B) Identification of morphemes and morphological processes One of the characteristic features of Swahili (and Bantu languages in general) is the existence of noun classes. There are specific singular and plural prefixes that occur with the nouns in each class, in the following sentences, two of these classes are included: (1) mtoto amefika (2) vitabu vitaanguka (3) mtu amelala (4) visu vitaanguka (5) mtoto anafika (6) vikapu vinaanguka (7) visu vimeanguka (8) watu wamelala (9) watoto watafika (10) kitabu kimeanguka (11) kisu kinaanguka (12) watoto wanalala "The child has arrived". "The books will fall". "The person has slept". "The knives will fall". "The child is arriving". "The baskets arc falling". "The knives have fallen". "The people have slept". "The children will arrive". "The book has fallen". "The knife is falling". "The children are sleeping" (a) Identify all the lexical and grammatical morphemes you can detect and specify their meaning. 32 AN INTRODUCTION TO THK STI IDY OI MORPHOLOGY GRAMMA LICAL UNITS (b) How is the verb constructed? That is, what kinds of morphemes arc strung together and in what order? (c) How would you say in Swahili: (13) The men are falling. (14) The books have arrived. (15) The children will sleep. (16) The basket will fall. 6. Describe the morphological process found in the following data from Classical Arabic. Try to specify its meaning(s). (1) kataba "he wrote" kattaba "he dictated" (2) šarufa "he was noble" šarrafa "he honored" (3) fahima "he understood" fahhama "he explained" (4) kabura "he was old" kabbara "he magnified" (5) šakka "he was doubtful" šakkaka "he filled him with doubt" (6) Vazza "he was strong" Vazzaza "he reinforced" (7) fasara "he discovered" fassara "he explained" (8) qáma "he got up" qawwama "he set upright" 7. Using the IP and WP models identify and construct the paradigms for the formation of plural of nouns (and adjectives) in Biblical Hebrew. Start by separating them into two groups: those which form their plural by a suffix vs. those which form their plural by a suffix accompanied by a morphological processes. Hebrew distinguishes two genders, masculine and feminine, marked by -0 and - umřít "have died" umřít umírat "be dying bit "beat" zabít "kill" zabít zabíjet "be killing" marked'? Or more marked? (10) Perfective Sg 3 kataba "he wrote" 2 katabta "you wrote" katabtu "1 wrote" Imperfective yaktubu "he writes" taktubu "you write ?aktubu "I write" ■= .he neater likelihood of morphological mother morphological criterion for margess is ^ ^ ^ !n.egulanty in uomarked forms. In Jj democratization ('deverbaP noun). It is more difficult to establish the derivational relationship, if any, between democrat and democracy. Here we have to account precisely for the allomorphy of IdzmskxaAl and /dgmakrasi/; these two alternants which arc held logether by the morphophonemic alternations /e - si and /a ~ a/; furthermore, democrat and democracy are also linked to the same semantic unit which is usually called a derivational base. I lowever, the problem with a derivational base lies in the degree of abstractness. Champions of abstract phonology do not hesitate to postulate an unaccented derivational base and a number of morphological rules which are necessary for the derivation of surface forms. To account for /domakrssi/ from underlying /'demakraet + i/ we would need vowel-reduction rules conditioned by the preceding or following stress (e —» 3 /— V and ae —* 3 / V—) and a morphophonemic rule with a morphological conditioning: / > si— derivational suffix -y (iii). Note that this rule is NOT conditioned phonologically, i.e., the change ( —► s has nothing to do with phonological properties of HI, since there are forms such as democratic. For this and other reasons this solution is unacceptable to many linguists and we may consider an alternative less abstract solution, namely two accented derivational bases: /denwkraet/ (—» democrat, democratic) /domakrst-/ (—» democratize). Those who are unwilling to write abstract phonological rules with morphological conditioning have to consider the alternation t ~ s in /damakret-/ ~ /dsmakres-/ 'exceptional' and rely for its 'explanation' on parallel paradigmatic sets: 54 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL AND DERIVATIONAL MORPHOLOGY 55 (3) democrat democratic aristocrat autocrat aristocratic autocratic democracy aristocracy autocracy Those synchronic linguists who do not hesitate to open the Oxford English Dictionary will discover that democrat and aristocrat are popular formations of the French Revolution (formed around 1790), whereas the other two forms are borrowed from Greek via Latin. In other words, the rule t ■ -» .s postulated above is nothing else but a historical palatalization of Old French (in those times, of course, it was a purely phonological rule) which has been putatively morpho-logizcd in Modern French. This, of course, is another way of saying that the palatalization rule of Old French is dead. If we try to avoid postulating an abstract derivational base, we may be inclined to postulate a derivational relationship between democrat and democratic — i.e., the former formed from the latter by clipping (which is most likely what happened historically). At this point, it would be preferable to leave the issue of morpheme-based or lexeme-based derivational morphology open; however, it is worth emphasizing that the construction of far-reaching systems of derivational rules is only one aspect (the dynamic aspect) of the whole problem; the other aspect has to do with paradigmatics (the static aspect) discussed in Chapter Three. Obviously, any speaker of English stores the paradigmatic set such as democrat, democratic and democracy in his/her memory — the controversy among various linguistic schools starts when we try to account for the 'inner' form of his/her grammar in terms of rules. 4.2 Some Universal Tendencies of Inflection and Derivation As mentioned under 2.3, there is a universal tendency for inflectional affixes to occur on the periphery of words, whereas denvational affixes occur closer to the root. The assumption is that only when derivational morphemes have been added to derive nouns, adjectives or verbs, can inflectional elements be subsequently added to grammaticalize the lexical content. For instance, in Latin the derivational base cogit- "think" might be followed by a verbal inflectional suffix such as -o (1" Pers Sg Pres Indie) cogito "I think": or by a derivational suffix -dtion yielding a deverbal noun cogitdtio "thinking". The latter suffix may be followed by any of the six case endings to grammaticalize the lexical content of "thinking": (4) cogitätiônis cogitätiönl cogitätiônem cogit-ätiön-is Genitive -I Dative -em Accusative There are some rare and doubtfu, counter-examples. Cons.der the following German inflection* and denvational forms of (das) Kind "child": inflectional P) (Plural of Kind) (Diminutive of Kind) (Plural of Kind-chen) Kind Kind-er Kind-chen Kind-er-chen "child" "children" "baby" 'babies' ,„istead of expected *Kind-chen-er (Base - D.m.nutivizmg Suffix - Plural Suffix) we obtain Zen (Base - P.ural Suffix - Dim.nutivizing Suffix). However, here ,t seems tba we are , „Clin- w, h a real plural suffix bu, with a stem-forming element (Stammoi,ungse,ement, „II w,,?become obvious ,f we compare the above forms with p.uralized and diminutivized l.miis o((der) Bruder "brother": ((.) Kind Kind-er Kind-chen Kind-er-chen "child" "children" "baby" "babies" Bruder Brüder Brüder-chen Brüder-chen "brother" "brothers" "little brother" "little brothers" W V observe that there is no morpheme boundary in *tirud-er (brud- is not a meaningful element in (iurman); the plural has been formed by means of a productive process of vowel change u * ii | v] (called mutation or umlaut); the diminutive has been formed by adding the usual diminutive suffix; and finally the diminutivized form has been pluralizcd by a 0-suffix (the formation of plural by 0-suffix is one of the possibilities of German, e.g., der Lehrer "teacher", die Lelirer "leaehers"; the article is inflected and shows unambiguously that the following noun is in the plural). Consequently, we may assume that the first morpheme boundary in Kinderchen is rather illusory and that the correct morpheme break-down should look as follows: (7) Kinder-chen-0 Base - Diminulivizing Suffix - Inflectional Suffix Note that this decision could be developed as an argument in favor of lexeme-based derivational morphology. The correctness of this solution (or more generally, universality of the sequence Base-I lerivation-Inflection) may be confirmed by looking at languages with higher index of synthesis ((ireenberg 1966) such as Czech or Latin: (8) German Czech Latin Bruder bratr f«ter Diminutive Brüder-chen bratf-Tk frater-cul-us Plural Bmder-chen-0 bratf-Ic-i frater-cul-i 56 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY IN FLECTIONAL AND DERIVATIONAL MORPHOLOGY 57 and c«„ i,„e, 0 ,ulr°l™ ""k"«*«-g"i», whileta„ o™, (9) Diminutive Plural German Czech der Bruder bratr das Brüder-chen-0 bratř-Ik-0 die Briider-chen-0 bratr-ic-i Latin fräter fräter-cul-us fräter-cul-i It might be of interest to ascertain what happens when we have the sequence of base and several inflectional or several derivational suffixes. In the sequence of several inflectional suffixes, variations sometimes occur. This phenomenon may be regarded as a minor piece of counter evidence to the internal cohesion of the word (discussed under 2.1.3), but it is significant that examples appear to come only from agglutinative languages such as Turkish. Thus, whereas in English it is impossible to say they *edlove (or in Old English hie *luf-on-od instead of luf-od-on), in Turkish it is quite common to find examples of positional mobility such as the following: (10) alr-yor-lar-dt take-PROGRESSIVE-PLURAL-PAST "they were taking" al-iyor-du-lar take-PROGRESSIVE-PAST-PLURAL al-ir-lar-di take-AORIST-PLURAL-PAST "they took" al-ir-di-lar take-AORIST-PAST-PLURA L In these Turkish verbal forms the suffixes marking tense (Past) and number (Plural) are positionally interchangeable, but they must follow the suffix marking verbal aspect (Progressive, Aorist). There are no similar examples of positional mobility in derivational affixation. Derivatives are internally stable because derivation changes the membership in the class of primary grammatical categories; consequently, it is not possible to say *democrat-ation-ize if we want to form a deverbal noun democrat-ize-ation. The same seems to be true of compounds, but admittedly to a lesser degree. In word composition, the respective order of elements may be variable in certain areas of the lexicon. For instance, in Standard English we say only twenty-five (five-and-twenly is dialectal and archaic), in German only fiinfundzwanxig but in Czech it is possible to say both: dvacetpět and pětadvacet. The explanation seems to be a loan influence (calquing) of German on Czech, i.e.,pět-a-dvacet "five-and-twenty" is modelled on the pattern of German fiinf-und-zwanzig whereas dvacet-pět is the native formation. Thus we may conclude 'In- mi lion by assuming that with the possible exception of copulative compounds such as tiii-l«'i numerals, the order of elements in compounds is stable. t 1 \nalysis of Inflections I he following grammatical categories may be inflected in English: nouns (for gender, number and case), adjectives (for comparison, i.e., comparative and superlative), adverbs (for ■ niiiparison), pronouns (for gender, number and case), and verbs (for person, number, tense, and l>.iiily mood). The three other verbal categories (aspect, mood and voice) arc realized syntactically l In means of auxiliaries). A striking characteristic of English, in comparison with other lndo-I iini|ieau languages such as Spanish, Latin or Russian, is its paucity of inflections. Excluding for .i « lulc adjectives, adverbs and pronouns, one may almost say that English manages with one iiilUvlional suffix, namely -s, which is used to mark all nominal and verbal categories (with the i -\i vpiion of tense, marked by -ed, present participle, marked by -ing, and past participle, marked ny.iilaily by -ed, and irregularly by -en and/or ablaut). Historically, however, even English was liifjily inflectional. Old English shows inflections for four cases (five with pronouns), three numbers (singular, dual and plural) and three genders with nouns, pronouns and adjectives; for ilncc persons with verbs (in the singular only) and pronouns; for tense and mood (subjunctive) mill verbs; and for strong and weak nouns, adjectives and verbs. Almost all of this morphology was lost dunng the Middle English penod (1150-1500). lo exemplify some of the above forms wc may look at strong and weak nominal declensions in Old English: (11) Strong Declensions Masculine Sg. Norn Acc Gen Dat stone stan stän staneš stäne PL Nom/Acc stänas Gen stana Oat stanům Neuter (short root) "ship" scip scip scipes scipe scipu scip a scipum Feminine (long root) (short root) (long root) "house" "gift" litis giefu hus gicfe fiiises giefe huse giefe hus giefa,-e husa giefena husum giefum "teaching" lär larc läre hire lära,-e lärena lärum (12) Weak Declensions Masculine Neuter Feminine "name" "eye" "sun" Sg. Nom nama cage sunne Acc naman ěage sunnan Gen naman ěagan sunnan 58 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY Dat naman PI. Norn -an Acc -an Gen -ena Dat -um eagan sunnan (all plurals) g-h). Compare, tor mstanee, the fClowmg forms: (13) Sg Nom Gen PI Nom Dat °ld English Strong stän stänes stänas stanům Weak nama naman naman nam urn Modem German Strong Weak Stein Steines Steine Steinen Name Namens Namen Namen It may be observed that -s in weak German Namens (Gen Sg) is analogical with strong Steines. The category strong (vs. weak) will be discussed in the following chapter (under Gender, 5.2.1) and in the meantime we may look at an even more formal classification of nouns based on the thematic vowel (cf. 2.3) as it exists in various archaic Indo-European languages such as Sanskrit, Greek or Latin. In the latter language, the nouns are classified into five declensions according to the five types of stems: a-, o-, C(onsonant)- and i-, u- and e-stems. The five Latin declensions are usually displayed in the following manner: (14) Five Latin Declensions I a-stems Sg Nom pueila Gen ae (< äi) Dat ae Acc am Abl ä PI Nom ae (< äi) Gen arum Dat/Abl Is Acc äs II o-stems servus (< os) rumor is T em e ěs um ibus ěs ] ö um (< om) ö I örum is ös III IV C-stems (-stems u-stcms ignis manus is i em e ěs tum ibus ěs/is us Ul um ü ÖS uum V e-stems res ei ei eni ě ěs ěrum "/ibus ěbus us ěs The traditional teaching of Latin was based on the INFLECTIONAL AND DERIVATIONAL MORPHOLOGY 59 ■ 'l>u.iliiig with the notion of thematic vowel. However, they can be segmented only at the price "I .i hi lain arbitrariness and inconsistency and only some of them are analyzable. Working with 11 ii older Latin form servos (occurring in Plautus) we may discover the following pattern for the imnialinii of the nominative, accusative and ablative: (15) Nom Acc Abl a-stems pueIl-a-0 -am -a-: (length) o-stems serv-o-s -o-m -o-: (length) ii-stems man-u-s -u-m -U-: (length) I Mil monosyllabic /-stems, e.g., vis, vim, vi "force" and e-stems fit this pattern, but the problem vM«iild be to account for the length in the nominative. Consequently, we may try to reformulate the traditional statement (-a marks the Nom Sg of a-stems and -as marks the Nom Sg of o-stems ■ it ) by identifying case markers with post-thematic elements: -0 would mark the Nom Sg of a-.1« i us and -s the Nom Sg of all other stems. Then, we would need some morphophoncmic rules in account for rumor "rumor" and honds "honor" from 'underlying' representations *rumor-s, *lionor-s cf. *dent-s > dens); -m would mark the Acc Sg everywhere (rumdrem would be derived linin * rumor-m); and the Abl Sg would be formed by the lengthening of the thematic vowel. As I.n as the thematic vowel is concerned, it might be tempting to identify it with markers of gender: • : feminine, -o masculine (and neuter), (/-stems are mostly masculine and e-stems feminine; mi fortunately, /-stems are both masculine and feminine. Whoever is interested in this analysis may proceed along these lines. Actually, this analysis was proposed at the end of the 19'h c. by the Ncogrammarians who maintained that a good many of the Latin inflections could be explained as being due to the coalescence of a once distinct morphology. The best we can do for Latin is to assume that its inflections are not synchronically analyzable into morphemes. Hence, undoubtedly, the traditional handling of Latin declensions hv memorizing and this is why classical grammarians did not establish morphophonemic rules, but merely patterns of formation (i.e., paradigms). This problem will be discussed again from a different angle in Chapter Seven; for the time being we can make a significant observation that I atin inflectional endings show a considerable degree of cumulation (or fusion) of significates. I atin is simply a typical example of a fusional linguistic type (other well-known examples are Sanskrit, Greek, Lithuanian, Russian, Czech, Polish, Serbian). Diametrically opposed are so-called agglutinating languages in which inflectional suffixes .lie typically composed of a sequence of morphemes with each morpheme representing one grammatical meaning. Turkish, or any other Altaic language, may be taken as an example of this linguistic type. Consider the data in (16) from Latin and Turkish. 60 AN INTRODUCTION TO THF ST, ,rw STUDY OF MORPHOLOGY 06) Sg Nom Acc Gen PI Norn Acc Gen Latin vir vir-um vir-T vir-i vir-os Turkish adam adam-i adam-in adam-lar adam-lar-i adain-lar-m The plural in Turkish is marked with the suffix -lar (the singular with the 0-suffix) and the case is marked with distinct suffixes added after the plural suffix. Assuming that the nominative (or rather absolutive, see under 5.2.3) is marked with the 0-suffix we may elaborate the following tripartite analysis for the whole Turkish paradigm: Number 0 0 0 lar lar (17) Stem Sg Norn adam Acc adam Gen adam PI Nom adam Acc adam Gen adam lar Case 0 1 in 0 1 m ^^^Z:^Z a,0nS lhCSe 'ineS Pr°Ved «° he P-ib.0 only for some (18) Masc Nom Acc Abl Fern Nom Acc Abl Root serv serv serv puell puell puell Gender o o o a a a Case s m ■■ (length) 0 m •' (length) This type of analysis would certainly be enormously difficult for the number and case of vir-; actually, we should analyze Latin endings for three categories: gender, number and case (Turkish, as a 'gcnderlcss' language has only number and case). For instance, we may entertain the idea that the plural is marked by the lengthening of the thematic vowel (the long thematic vowel in the singular was analyzed above as Gender Case) and we would end up with the following tripartite analysis of some Latin endings: INFLECTIONAL AND DERIVATIONAL MORPHOLOGY 61 ď') Root Gender Number Case Fern Sg Nom puell a 0 0 Gen puell a i Acc puell a 0 m Abl puell a 0 PI Nom puell a i Gen puell a rum Acc puell a s Mase Sg Nom serv 0 0 s Gen serv 0 0 T Acc serv 0 0 m Abl serv 0 0 PI Nom serv o I Gen serv 0 rum Acc serv 0 s li is obvious that all this is still far from Turkish determinacy with respect to segmentation into morphemes (one-to-one correspondence between morpheme and grammatical meaning). First, »e analyzed only some of the o- and n-stems. Second, many problems remain. For instance, the fcin Sg Gen shows a long thematic vowel; the Masc Sg Gen is homophonous with Masc PI Nom ; but in this type of analysis the former has underlying -o -0 -f the latter o-.-f (plus the fact that iu; have to write a phonological rule for o + T—» /". Furthermore, we still do not escape the issue of polysemy (multiple meaning) of grammatical morphemes. For instance, in the above tripartite analysis -s marks not only the nominative (Sg Masc) but also the accusative (PI) and genitive elsewhere. Thus we have to conclude that whereas Turkish is a typical example of an agglutinating language in that it shows a one-to-one correspondence between morphemes and grammatical meaning, Latin is fusional in this respect, even if there are some traces of agglutination. However, as with all typological distinctions, we are dealing with a continuum and it might be instructive to examine a language which may be classified as semi-agglutinative (or scmi-fusional). In contrast with Latin, Classical Arabic is more successful in keeping plural markers from fusing with gender markers. These are the instances referred to as broken plurals c g., rajul-un "man", pluralized ri/al-un (-« indicates nominative and -n corresponds to the indefinite article of English). Consider the following paradigmatic sets (Classical Arabic has only iluee cases and two genders): (20) Sg Nom Acc Gen Masculine rajulun "a man" rajulan rajulin Feminine ?imra'?atun "a woman" ?imra?atan ?imra?alin 62 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL AND DERIVATIONAL MORPHOLOGY 63 PI Nom Acc Gen rijalun rijälan rijälin ?imra?ätun ?imra?ätin ?imra?ätin Sg Nom mudarrisun "a teacher (M)" mudarrisatun "a teacher (F)" Acc mudarrisan mudarrisatan Gen mudarrisin mudarrisatin PI Nom mudarrisuna mudarrisatun Acc/Gen mudarrislna mudarrisatin r . Me,r cut examples of agglutination but these ,.......«<* in Classical ^^X^ ^Junusual fusion of lexical and .........,crbalanced by examples of luaoo. )ndo_Elir0pcan languages). It may ,.,„,„„,„,, morphemes (somewhat —ccn to f-b> ng ofthe 8cnQC, -...... ,lU1U-i- or feminine nouns but oy „„„,„„,,„„„ vs. mudarrisina in (21). Clear cut examples of agglutination come from the inflection of feminine nouns. Assuming that the masculine gender is marked with the 0-suffix (versus feminine -at) we may propose the following tripartite analysis of the endings of externally inflected nouns (length is marked with a colon (:)). Stem Gender dumber Case (+Tndef) Masc Sg Nom mudarri s 0 0 un Acc mudarri s 0 0 an Gen mudarris 0 0 in Fern Sg Nom mudarris at 0 un Acc mudarris at 0 an Gen mudarris at 0 in Masc PI Nom mudarris 0 una Acc mudarris 0 ina Gen mudarris 0 ina Fem PI Nom mudarris at un Acc mudarris at in Gen mudarris at in The case endings of masculine nouns inflected externally in the plural are not identical with those in the singular. However, they are identical if the noun is inflected internally (a-u —> i-a): (22) Root+Number Gender Case Masc Sg Nom RaJuL 0 un Acc RaJuL 0 an Gen RaJuL 0 in Masc PI Nom RiJäL 0 un Acc RiläL 0 an Gen RiJáL 0 in 64 INFLECTIONAL AND DERIVATIONAL MORPHOLOGY 65 RECOMMENDED READINGS Applegate, Joseph R. 1958. An Outline of the Structure of Shilha. New York: American Council of Learned Societies. Campbell, A. 1959. Old English Grammar. Oxford: Clarendon Press. Greenberg, Joseph. 1966. Language Universals with Special Reference to Feature Hierarchies. The Hague: Mouton. Kurylowicz. Jerzy. 1964. Inflectional Categories of Indo-European. Heidelberg: Carl Winter. Lewis, Geoffrey L. 1967. Turkish Grammar. Oxford: Clarendon Press. Moreland, Floyd L. & Rita M. Fleischer. 1973. Latin: An Intensive Coarse. Berkeley: University of California Press. Wright, W. 1896- 1898. A Grammar of the Arabic Language. Volumes 1 and 2. Cambridge: Cambridge University Press. EXERCISES Identity the following morphological categories in Tasclhit, the Berber dialect spoken in ■.oiilliwest Morocco as described by Applegate (1958): (.i) Verbal roots and their lexical meaning. (b) Derivational verbal affixes and their lexico-grammatical meaning, d'l Inflectional verbal affixes and their grammatical meaning, id) Describe morphological processes expressing the past tense, (e) Construct paradigmatic sets for (i) Subject affixes (ii) Direct object affixes (iii) Indirect object affixes Describe their distribution within words with respect to the roots in various inflectional categories. Use the following data. Notice that (he examples are in phonemic transcription (i.e., do not try to pronounce them): (1) tdit "you went" (2) dänt "they (F) went" (3) ifaiast "he gave it to him/her" (4) urasntfint "they (F) did not give it to them (M)" (5) tram I "ye (F) wanted" (6) umzyt "1 took him/it" (7) raiiamz "he will take me" (8) isfirtt "he stole her" (9) fanasntt "they (M) gave it to them (F)" (10) ndä "we went" (11) tsfirmt "ye (F) stole" (12) ramtfiy "I will give it to you (F)" (13) fiyasl "1 gave it to him/her" (14) tkayast "1 used to give it to him/her" (15) uramtifi "he did not give it to you (F)" (16) urakzriy "1 will not see you (M)" (17) izrai "he saw me" (18) rakmiamz "he will take you (F)" 66 AN INTRODUCTION TO THE STUDY OH MORL 'HOLOGY (19) raiitifa (20) urrastfin (21) tbnunt (22) tbnutl (23) raktbnuy (24) fantaunt (25) urraunttfint (26) itsfar (27) tsfirt (28) nird (29) tsirdmt (30) sirdn (31) rakuntizra (32) urkunizri (33) tzraytnt (34) itdii (35) tkant (36) ntamz (37) ifaiamt (38) fantast (39) riy (40) urttriy (41) rattizra (42) tbint (43) (tbit (44) rabiy (45) izrati (46) ifaiit (47) tfitast (48) tfamtasnt (49) dly (50) tdäm "he will give it to me" "they (M) will not give it to him/her" "they (F) used to build" "you built it" "I will build it for you (M)" "they (F) gave it to you (M PI)" "they (F) will not give it to ye (F PI)" "he used to steal" "you stole" "we were clean" "ye (F) washed" "they (M) washed" "he will see you (F PI)" "he did not see you (M PI)" "1 used to see them (F)" "he used to go" "they (F) used to give" "we used to take" "he gave it to you (F)" "they (F) gave it to him/her" "1 wanted" "1 did not want her" "he will see her" "they (F) used to cross" "you used to cross" "1 will cross" "you saw me" "he gave it to me" "you gave it to him/her" "ye (M) gave him/it to them (F)" "1 went" "ye (M) went" Segment the words of Latin, Classical Arab.c and Turkish given below lnto morphemes and dtscuss the correspondence between morphemes and grammatical meanmg. INFLECTIONAL AND DERIVATIONAL MORPHOLOGY 67 Norn. Acc. (ien. I'l Norn. Acc. (ien. Norn. Acc. Gen. Nom. Acc. (ien. PI Latin vir "man" virum viri vifi virös virörum femina "woman" feminam feminae feminae feminäs feminarum Classical Arabic Turkish rajulun raj u lan rajulin rijalun rijalan rijahn imra?alun imra?atan imra?atin imra?ätun imra?ätin imra'ratin adam adami adamm adamlar adam 1 an adamlarin kadin kadmi kadmm kadtnlar kadinlan kadmlartn , „ was shown m 4.3 that Latin is fusional with respect to inflection of nouns. However, the ct 1, suffixes of Latin are segments in a more abstract analysis operating w ons such as a short and long thematic vowei, zero suffix and morphophonermc rule, *«* *« ^ ^ ***** ^ m(15). Identify the following morphological categories of Classical Arabic: (a) Verbal roots and their lexical meaning. (b) Derivational affixes and processes and (heir lexico-grammalical meaning. (c) Construct paradigmatic sets of inflectional suffixes expressing subject and direct object. (d) Predict the following forms: "ye (M)" and "you (F)". Use the following data: (1) katabahu (2) fahimatha (3) iktatabta (4) iftahamtinl (5) kutiba (6) aktabnahu (7) afhamtunna (8) istafhamna (9) tafahamu "he wrole it" "she understood her" "you (M) subscribed" "you (F) comprehended me" "it was destined" "we dictated it" "ye (F) instructed" "they (F) inquired" "they (M) understood one another" 68 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY (10) takátabna (11) qutilat (12) taqatalná (13) istaqtalti (14) tadáxalti (15) adxaltunnahá (16) adxalatki (17) fahimaka (18) adxalnákunna (19) xarajtu (20) axrajnákum (21) istaxrajuhum "they (F) wrote to each other" "she was killed" "we fought with one another" "you (F) risked your life" "you (F) interfered" "ye (F) introduced her" "she introduced you (F)" "he understood you" "we introduced you (F PI)" "I left" "we dismissed you (M PI)" "they (M) exploited them (M)" Translate the following sentences into Classical Arabic: (22) "they (F) left" (23) "she was dismissed" (24) "we understood one another" (25) "ye (F) inquired" (26) "they (F) introduced you (M PI)" INFLECTIONAL AND DERIVATIONAL MORPHOLOGY Norn .\cc lustr Dat Abl (icil 1 .oc pi Norn Acc Instr Dat Abl (ien toe Á-stems "daughter' suta ám aya ayai áyás áyás áyám a-stems "son" sutas am žna aya át asya é C-stems "friend" suhrd am a ě as as i í-stems "fire" agnis im as as as as án as ais bhis ábhis ábhyas ěbhyas bhyas ábhyas ěbhyas bhyas ánam ánám ám ásu ésu su má ayě ěs es áu ayas in ibhis ibhyas ibhyas ínám isu H-stems "sun" bhanus um una avé ós os áu ávas ún ubhis ubhyas ubhyas únám usu 5. The following table displays the battery of 60 inflectional suffixes of Sanskrit as presented in various traditional grammars. Prove that their number may be reduced considerably in a more abstract analysis which operates with the following notions: (a) thematic vowel (b) suffix (c) morphophonemic rules of ablaut (gradation) Hint: Consider /e/ and 161 as diphthongs /ai/ and /an/ in the underlying form (IP model). IN I-LECTION AL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 71 CHAPTER FIVE INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 5.1 Primary Nominal Categories 5.1.1 Nouns and Adjectives Classical parts of speech have been limited to a few broad classes: noun, verb, adjective, adverb, pronoun, preposition, interjection, numeral, conjunction, and possibly article and particle. In traditional grammatical theory the parts of speech were defined in notional terms. Noun was defined as the name of any person, place, animal and thing (concrete nouns) and the name of any property, action and state (abstract noun). Adjective was defined as an attribute (property) of any person, place, animal or thing denoted by a noun. Properties of substances expressed by adjectives are of a static nature; on the other hand, properties of substances expressed by verbs are of a dynamic nature, or, more commonly, verbs denote actions, states and changes of state of the subject. (Discussion ofsubjectless sentences belongs to syntax). Grammatical redefinition of these three notional categories would establish a basic dichotomy of noun and verb. Noun (and adjective) would be defined as a primary grammatical category the domain of which includes sub-categories of gender, number and case (plus comparison with the adjective); the verb would be defined as a primary grammatical category the domain of which includes subcategories of person, number, tense, aspect, mood, and voice. It may be noted in passing that this dichotomy became the first axiom of Transformational Grammar (S - + NP VP). In other words, the definition of the major grammatical classes of noun and verb cannot be made independently of syntactic and logical considerations (noun-verb, subject-predicate, agent-action). Plato (429-347 B.C.), who was the first to distinguish explicitly between nouns and verbs, used the term pfjuii /rhema/ for the latter category — the word rherna means both 'verb' and 'predicate'. It is significant that in this classification verbs and adjectives were put together in the same class. This may seem surprising on grammatical grounds since adjectives have the same sub-categories of gender, number and case as nouns do, but, on notional grounds, it makes perfect sense since both verbs and adjectives may be characterized as features on substances (dynamic and static). In this connection, it is of interest to note that later Greek grammarians abandoned the Platonic dichotomy of noun - (verb - adjective) and replaced it by another dichotomy of (noun -adjective) - verb, where nouns and adjectives were brought together in one class. If the Platonic dichotomy was of a notional nature, the later dichotomy was of a grammatical nature (i.e., the emphasis was on nominal subcategories shared by nouns and adjectives). The tripartite distinction of nouns, adjectives and verbs, or the compromise between notional and grammatical criteria, was made later on in medieval times and since those times it has survived in our school grammars. The emphasis on a tripartite distinction was born in dealing with Classical languages (Latin and »i.. K i i\lncli are heavily flective and where the adjective shows the same sub-categories as the • miii .lues However, if we look elsewhere, we realize that this situation is far from being .mil f..il. for instance, in English and Turkish attributively used adjectives are not inflected for . nili-i .iiul number, whereas in Latin and French they are. Put differently, the nominal sub- iii |.....cs of gender, number and case are inherent in nouns, but they are only secondary in i.l|n in is; marking for nominal sub-categories with adjectives by agreement or congruence was l.iiiinil lo be only a matter of 'surface' grammar. Consider the following data: I I ) English a good man a good woman good men good women Turkish iyi bir adam iyi bir kadin iyi adamlar iyi kadmlar Latin vir bonus femina bona viri boni feminae bonae I he Latin adjective shows complete agreement with its noun in gender, number and case (vir I'i'iiiis Nominative, viri bom = Genitive Sg or Nominative PI, etc.). On the other hand, we ' .iiiiinl say *good-s men in English or *iyi-ler adamlar in Turkish. Neither will the case be ■.Inivvn in the latter two languages, thus versus Latin virorum bonorum (Genitive Plural) we find l-iif.hsh good men's and Turkish iyi adam-lar-m. Of course, in both English and Turkish, .nl|cclivcs can be nominalizcd. For instance, young is an adjective used as a nominal in the I ni'.lish sentence: What can one expect from the young?. Observe, however, that we still cannot |ilm.ilize; 'from the youngs is not grammatical. On the other hand, in Turkish, nominalizcd .uljcclives behave like nouns in that they may take the plural, case and personal suffixes after them, or the indefinite article bir before them; consider: bi/yiik "big, old", huyiik-ler-im "my ciders"; luista "ill", bir hasta "a sick person"; gene "young", genc-ler-in "ofthe young". If we take as the criterion of an adjective the permissibility of forming the comparative and Miperlative, nouns will be excluded. For instance, in English we may say The rich live on the bay ■ ■i The richest live on the bay, but nothing similar can be done with The man lives on the bay. In oilier words, although most adjectives are nominalizable (i.e., can be used as nouns), the converse r. not true (adjeclivals, such as bus in bus slop cannot form the comparative or superlative). Note .ilso that the 'noun-ness' ofthe adjective can be a matter of degree. From the discussion above il appears that adjectives in Turkish arc more noun-like than adjectives in English; we cannot say * I'he riches live on the bay, i.e. to treat the adjective rich as a noun in respect to pluralization, but wi; can in Turkish. However, even in Turkish, and perhaps in all languages, we have to draw a dividing line between noun and adjective when it comes to the specific adjectival category of comparison. Latin and Greek adjectives are not formally different from nouns. As was shown in (1), we cannot talk about specific adjectival inflection. The adjective, as far as the categories of gender, number and case are concerned, is inflected nominally. Nevertheless, the Germanic and Balto- 72 AN INTRODUCTION TO THE STI ,TDY OF MORPHOLOGY (2) Slavic families of the Indo-European phylum created what may be called a specific adjectival inflection. Let us consider Old English as a representative of the Germanic family. The so-called strong declension of adjectives displays endings which differ from those of nouns but which are almost identical with those of pronouns. Consider the following data: Noun (Masc) Strong "stone" stan stan stan-es stan-e Sg Nom Acc Gen Dat Instr Adjeclive (Masc) Strong "good" god gôd-ne gôd-es god-urn gôd-e Demonstrative Pronoun (Masc) "this" 6es 6is-ne 9iss-es 9iss-um Gys However, the so-called weak dec!en,i,.„ f ■■ dec.ens.on of adjectives is ident.cal with that of weak nouns: (3) Nniin t\/I„„„\ . .. Sg Nom Acc Gen Dat Noun (Masc) Weak "name" nama naman n am an naman Adjective (Masc) Weak "good" goda godan godan godan Baltic and Slavic languages have two types of adjectives: simple (or indefinite) and complex (or definite). The simple forms are inflected, with some exceptions, identically with nouns. Consider the following data from Lithuanian: (4) Sg Nom Gen Dat Acc Loc Instr Noun (Masc) "man" výras výro výrui výra výre výru Simple Adjective (Masc) "good" geras gero geram gera geram e gerii INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 73 ľ') Complex Adjective (Masc) Personal Pronoun (Masc) "good" "he" S g Nom gerasis jis Gen gerojo J° Dat gcrajam jam Acc gerajj ji Loc gerajame jame Instr genioju juo '• I .' Pronouns Expressed traditionally, pronouns do not 'name' persons, animals and things but 'replace' in (ihe Latin term pronomen was calqued on Greek dvxu)vu|Lia /antonymia/ "instead of the iimni"). On the one hand, pronouns resemble nouns in that they are inflected for number, case .mil. lo a limited degree, for gender; on the other hand, pronouns share the category of person with > i ihs, furthermore, pronouns are a small closed (grammatical) class, whereas nouns are a large upcii (lexical) class to which we may freely add new ones. I'o analyze personal pronouns we have to introduce the notion of deixis (derived from Greek Hi ii vnpi /deiknumi/ "1 point, indicate"). We may say that personal pronouns arc only one class ill ihe so-called deictic elements, which include also advcrbials of place: here and there ('close in ihe speaker' vs. 'not close to the speaker') and time: now and then ('at the time of speaking' ii. 'not at the time of speaking'). The deictic category of proximity is not too common with personal pronouns of the 3rd Pers but examples may be found in more 'exotic' languages. For instance, Hindi pronouns of the 3rd Pers show this contrast: ye 'he/she close to the speaker' vs. ho 'he/she not close to the speaker'. Obviously, the deictic category of proximity is irrelevant \\ itli pronouns of the Is' and 2"d Pers since the speaker and the addressee arc always (disregarding conversations on the telephone, etc.) in the same spatiotemporal situation. This circumstance may explain why marking for gender is rather unusual in the 2nd Person, and even less common with Hie lsl Person. The situation in English is typical of many languages. Here, the personal pronouns nf the 1" and 2nd Pers Sg are genderless but the pronoun of the 3lli Pers Sg shows gender distinctions (he, she, it). In French and Latin even their plural counterparts show gender distinctions, whereas in Germanic languages they do not. Pronominal forms in Germanic and Romance languages are contrasted in (6). 74 AN INTRODUCTION TO THE 5 STUDY OE MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 75 Latin French i'M Ľ Sg 2nd Sg is ii il ils -heb -hök ea eae elle elles Masc -hěba -höka id ea lem -hěbi -höki Examples of gender distinctions in the 2nd Pers rmv h„ r™.„A (7) 2nd Person Masc Fern 3,d Person Masc Fern Singular Vanta ?anti huwa hiya Plural Tantum ?antunna hum hunna In Modem Arabic dialects the gender distinctions in the plurai d.sappeared. The system of Syrian Arabic is as follows: system oi (8) 2nd Person Masc Fern 3,d Person Masc Fern Singular ?onte ?onti huwwe hiyye Plural ?ontu hsnrie The reasons for this simplification have to do with reference. In referring to "men and women" Classical Arabic had to use the masculine form hum (this form would be used even when referring to "a hundred women and one man"), whereas Modem Syrian Arabic uses the unmarked form Iwnne, which is not in conflict with natural gender. This clash between natural and grammatical gender contributed probably to the loss of dual pronominal forms. In Classical Arabic, these were formed from the masculine base: ?antuma "you (two)" and htimd "they (two)". Again, in referring to the couple of "a man and a woman" or to "two women", there was a clash between natural and grammatical gender. In these cases Modern Syrian Arabic uses plural forms ?3Mu and hanne. As an example of a language making gender distinction in the Is' Pers Sg we may mention Adcni Arabic where we find and "I (Masc)" vs. ani"\ (Fern)". "We" distinguishes gender in Spanish nosotros "we (Masc)" vs. nosotras "we (Fern)". There are also examples of gender distinction with possessive pronouns. For instance, Bedawye (Kushitic group of the Afro-Asiatic phylum) has three forms of "my" and "your" (realized as suffixes on nouns): I In- first form is unmarked for gender, the second form specifies that the L' (or 2"J) Pers i" i ...,-vsor is male, and the third form specifies that the possessor is female. ii may comes as a surprise to realize that truly genderless languages are genderless even in ii>. ii pronominal systems. Examples arc comparatively numerous and may be found among \Ii.ik , Uralic and Amerindian languages. Turkish (Altaic family) may be used as an example: 110) Singular Plural 2nd Person sen siz 3rd Person o onlar \ . shown in (7), Classical Arabic has two forms i il,,., poor' system with the 'rich' systems of Latm and Czech: n each of these lour slots. Or we may contrast 111) Masc Fem Neuter Latin 3,d Person Sg PI is ii ea eae id ea Czech 3rd Person Sg PI on oni ona ony ono ona Turkish 3rd Person Sg PI Reluming to gender-languages, il is of interest to observe that Latin and Czech show the typical Indo-European syncretism of Feminine Singular and Neuter Plural (cf. Neul PI verb-a "words" and Fem Sg femin-a "woman") even with personal pronouns. There are various problems with pronominal number distinctions in the 1'' Person. Whereas "three tables" is "table, + table, -i- table3" we cannot say that "we (- three of us)" is "I, + 1, + I3". I lie pronoun we covers basically two distinct groups which can be established on the basis of the speaker and addressee distinction: (i) the speaker-group (= I + he, + he, ...) but NOT' the listener group; (ii) the speaker-group (- I or I + he, + he, ...) AND the listener group (- you or you and he, f he,...) as well. In both cases, one or more 3rd Persons may or may not be included. It is of interest to note that some languages (most notably the Algonkian family) grammaticalize this distinction between two 76 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 77 meanings of "we". Thus Cree has two 1st Person plural forms which keep these two basic possibilities apart: (i) nilanan includes the speaker group but excludes the group being addressed: "I and he/they but not you." This form is called lsl Pers PI exclusive. (ii) kilananaw specifies that both the speaker group and the addressee-group are included: "I (and he/they) and you (and he/they)". This form is called Is' Pers PI inclusive. Nevertheless, there arc languages which form their plural personal pronouns by the simple pluralizing of their singular counterparts. As a classical example we may quote Chinese which does not pluralize nouns, but does pluralize pronouns. The Cantonese system is as follows: (12) ngaw "1" ngaw-day "we" nay "you" nay-day "you, ye" koei "he/she" koei-day "they" Even some Altaic languages could be analyzed similarly; for instance, Turkish, as shown in (13) and (14): (13) ben "I" sen "you" o "he/she" biz "we" siz "ye" onlar "they" He c th llatlon „ more complica(£d 3„ person pronoun |s piuraii7gd ^ na, suffix e.g., eoeuk "child", eocuk-lar "children") whereas the ,« and 2- Person show the pronominal plurahztng suffix -« which occurs typically with possessive suffixes- (14) ev ev-im ev-im-iz "house" "my house" "our house" (the possessor is in the singular) (the possessor is in the plural) realised 7 1 " " eXamP'e °f ' ""^^ ^ wh™ "°SSessi- P™--, are rea tzed as suffixes (bound morphemes), whereas Latin and many other Indo-European languages realize their possessive pronouns as attributive adjectives (free morphemes) which may be mflcued for gender, number and case (to show agreement). Let us consider the less common I urkish system first: 11 "i ) ev-im "my house" -in "your (Sg) house" -i "his/her house" ev-im-iz "our house" -in-iz "your (PI) house" -ler-i "their house" ev-ler-im "my houses" -in "your (Sg) houses" -i "his/her houses" ev-ler-im-iz "our houses" -in-iz "your (PI) houses" -ler-i "their houses" N„ ....cresting problem connected with marking for nominal and pronominal plural appears in ,l„ i"1 l'ers. Here the single form evleri has three meanings: (16) |(i) his/her houses (ii) their house (iii) their houses (i) (ev+Pl) (ii) (ev) (hi) (ev+Pl) Poss Sg Poss PI Poss PI (ev + ler)i (ev)ler ( i (ev + 1er) 1er + i I Ik- form in (iii) *ev+ler+ler+i is simplified into ev+ler+i. The curiosity of this system lies in tin- fact that the possessive suffix of the 3rd Pers PI ler+i "their" is anomalous if compared with ilic r1 and 2nd Person. In these persons the pluralizing suffix is added on the right (as usual with nouns): im + iz, whereas it is added on the left in the 3'd Pers: ler+i. Hence the homophony of 'his/her houses" {ev+ler+i) and "their house" (ev+ler+i instead of *evH i+ler). This clash does mil exist in other languages with similar systems of possessive suffixes. For instance, Persian (which is one of the few Indo-European languages with a system of possessive suffixes) admits a pluralizing suffix on the right even in the 3'd Person. This is shown in (17): (17) barädar-am barädar-es barädar-em-än barädar-es-än "my brother" "his brother" barädar-an-am barädar-än-es "my brothers" "his brothers" "our brother" baradar-an-em-an "our brothers" "their brother" baradar-an-es-an "their brothers- Systems of other Indo-European languages show the typical heteron^y of persona, pronouns (/ - **, ke - *ey) even in the system of possessive pronouns. Consider the Latin data in(18). (18) frätermeus "my brother" fräter noster "our brother" fratres met "my brothers" frätres nostri "our brothers" 78 AN INTRODUCTION TO The tT, ,r ° rHE STUDY OF MORPHOLOGY I" Latin the possessive pronoun has to „ wbr> ™** -p«y -d. L easLgender: -^ ,ts now «*-»■ Contrast the foll^ ^ of the lexeme (19) Latin frater meus frätris mel frätrés mel frätrum meôrum fräter noster frätris nostri frätres nostri frätrum noströrum Turkish kardes-jm kardes-im-in kardes-ler-im kardes-lcr-mi-in kardes-im-iz kardes-im-iz-in kardes-ler-im-iz kardcs-ler-im-iz-in English my brother my brother's my brothers my brothers' our brother our brother's our brothers our brothers' The difference between these two linguistic systems may be portrayed as shown in (20). (20) Latin LEXEME Gender POSS PRONOUN Gender Number Number Case Case Turkish ((LEXEME + Number) Poss + Numbcr) casc Interrogative, relative and indefinite pronouns exhibit morphological similarities in many languages. In the flective Indo-European languages we find two sets of interrogative and indefinite pronouns: the nominal set has typically only two forms: one used in inquiring about animate beings (who?) and another one used about inanimate objects (what?); the adjectival set differentiates gender in reference to animale beings. Latin forms are given in (21). (21) Nominal Set quis? "who?" Masc Fern Neut quid? -vhat?" Adjectival Set qui (quis)? \ quae? ( quod? J "which (one)?' The adjectival forms are used if we demand additional information about animate beings or objects which have already been mentioned. For instance, if the discussion is about "poets" we may demand additional information by asking "Which poet is the best (one)?" or in Latin Quipoetu est optimus? However, if we wanted to ask 'out of the blue' about the best poet (i.e., in the case when "poet" would not be given as the topic by the context), we could use the form quis: Quis est optimus poetal "Who is the best poet?". A full explanation of this phenomenon INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 79 in lull)'.-, in syntax and discourse analysis; here we have to be satisfied with observing ......plmlogical differences between nominal and adjectival interrogative pronouns. Russian data ••Hi.....other example of this constraint: I ' 'I Masc fern Neuter Nominal Set kto? "who?" cto? "what?" Adjectival Set kotóryj? ") kotóraja? > kotóroje? J "which (one)?" 111.11 use corresponds to Latin or English in (21): Kotoryj poet nailucsij? "Which poet is the best i "in I'" versus Kto nailucsij poet? "Who is the best poet?" ^ .' Secondary Nominal Categories '■ ' I tinnier H is customary to start a discussion of grammatical gender with non-linguistic considerations .mil pioceed from them to establishing a 'natural' semantic basis for gender in individual i.ini'iiagcs. It is clear that the non-linguistic universe can be classified in a variety of ways .ii i Hiding to various sets of properties. One of the major distinctions is undoubtedly that of living "i iiiiiinatc beings (human and animal beings) versus inanimate things. This is the essence of mil' of the most famous classifications of the universe as elaborated by the neo-Platonic philosopher Porphyry, shown in Figure 5.1. Inorganic things (Porphyry's minerals) arc inanimate and so are plants. However, (he ■ l.r.silicalion of animals (Porphyry's brutes) as animate beings depends crucially on the ii itcipi elation of the word animale. Etymologically this word means "provided with soul" (Latin .miiiia "soul"), but in modern linguistic terminology this term means simply "living". For iiisi.iiice, some semantic features of man are [+ammate], [-i human] and those of cat are | ' .miniate], [-human]. This type of specifying the lexis for semantic features is claimed to capture MK.nilicant universal properties of human languages. Many linguists entertain the idea that the vocabularies of all human languages can be analyzed in terms of a finite set of semantic features "i M-mantic components (such as [±animale], [icountable], [±male], etc.), which are themselves independent of the particular semantic structure of a given language. To quote Katz (1966:156) "semantic markers ... cannot be identified with the words or expressions of the language ... K.ilher, they are to be regarded as constructs of a linguistic theory." Currently, however, linguists .ne more interested in the fact that universal semantic features are intimately linked with i miLcptual structures of individual languages which are accessible to us through their moiphological systems. Let us ascertain whether we can use the semantic 'marker' animate in analyzing the grammatical system of Sumcrian (genetically isolated). The basic dichotomy of Siimerian lexis is usually referred to as a group of 'persons' versus a group of 'things' with the membership as shown in (23). 80 INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 81 substance human spirit mineral brute (23) PERSONS gods heroes human beings Fig- 5.1 Porphyry's classification of the universe THINGS things abstracts animals This dichotomy is imposed on lexis by grammar; for instance, only PERSONS may form their plural by the suffix -ene, THINGS or their determining adjectives have to be reduplicated. It might seem very strange that Sumerian grammar treats animals as THINGS; but, obviously, the Sumerian dichotomy PERSONS-THINGS is best explainable in terms of'soul' in that divine and human beings possess a soul whereas animals do not. Consequently, we may wish to use the pair animate - inanimate in its etymological meaning [+soul] vs. [-soul] for the Sumerian PERSONS -THINGS opposition. Obviously, we cannot say that the meaning of animacy in Sumerian and, say, English is co-extensive. It is of interest to note that groups of PERSONS [+animate collective] in Sumerian may be treated as THINGS [+inanimate]. Needless to say, attempts to work with the category of animacy defined independently of a particular language would be doomed a priori. It may come as a surprise that some grammatical systems do not reflect the 'natural' dichotomy of animate (human and non-human) beings into male-female. In other words, there is nothing in them which would correspond to the masculine-feminine dichotomy of many Indo- i mi >| ii-.ni i ir Afro-Asiatic languages. Examples of such grammatical systems can be found among Ui n. ( I inkish) or Amerindian (Cree) and Eskimo-Aleut languages (Inuktitut). Thus in Turkish 'h. i'' I Vis pronoun o is used in referring to both male and female beings and in translating from l n111.11 we have to use "he" or "she" (or "it") according to the context. Similarly, in Cree wfya ■ «|ui-.:.es all "he", "she", and "it". It is of interest to note that in genderless languages epicene i. >. mi", occur even in the most unexpected cases (epicene =■ common gender, e.g., cat in English ■ n iluioic both tom-cat or (.she-teat). In Turkish kardes may denote either "brother" or "sister"; >■• a Li Id "sister" unambiguously we have to use kiz "girl" in apposition: kizkardes (literally i. in.ile sibling"). Or in Sumerian the epicene lexeme dumu may denote either "son" or ■ l,iii)'hlei"; to refer to "daughter" unambiguously we have to use mi "woman" in apposition: ■ iimni mi (literally "female offspring"). An unusual example of this nature comes from Janjero il.ir.liinc language, Afro-Asiatic phylum) which has an epicene lexeme asu denoting cither in.in" oi "woman". These may be referred to unambiguously by using words adk "male" and "i.n/, "k-inalc" in apposition: adk asu "male human being" - "man" and mask asu "female human in in);" "woman". The phenomenon of common gender is not limited to 'exotic' languages; . |H. i nc lexemes may be found in all natural languages. However, what is of interest from the > n u point of anthropological linguistics, is the balance of epicenes and heteronyms (heteronymy i. 11 ii opposite phenomenon of the common gender; e.g., stallion and mare are heteronyms since ih. ..i- two words cannot be related morphologically). Heteronymy is much more common in the ■ oI domesticated animals (ram vs. ewe, gander vs. goose etc.); in the case of wild animals, v ,ii ii his languages use epicene lexemes plus productive derivational processes lion - lioness, tiger tiyjcss). Turkish indicates the gender of animals by erkek "male" and disi "female" as English .I.ii-, wilh pronouns he and she: (24) erkek ay i di§i ayi he-bear she-bear Similarly, Latin could specify the natural gender of animals denoted by epicene lexemes by was "male" and femina "female": (25) vulpěs mäs vulpěs femina lupus lupa or lupus fémina "fox" "vixen" or "she-fox" "wolf "she-wolf All the nouns of many Indo-European languages can be classified into three genders masculine, feminine and neuter in order to account for several grammatical phenomena: (i) occurrence of specific endings (suffixes) 82 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 83 (ii) adjectival concord (agreement) (iii) pronominal reference (iv) correspondence to natural gender Any discussion of gender without paying attention to all these four points would be misleading. The first point, occurrence of specific endings, is not particularly reliable. For instance, it is impossible to assign gender in Latin solely on the basis of this criterion. If we say that the noun dominus "lord" is grammatically masculine because it has a characteristic suffix -us, then we have to make some provisions to accommodate nouns like manus "hand" which is feminine (as shown by adjectival concord (ii): manus longa "a long hand". On the other hand, the noun agricola "peasant" is grammatically masculine even if it has the same inflectional suffix -a as femina "woman". As these few examples show, adjectival concord and pronominal reference are more reliable indicators of grammatical gender than nominal morphology (i). The criterion of adjectival concord is also a better indicator of grammatical gender than natural gender (iv) since this criterion is useless when it comes to inanimate things (concrete objects, parts of human body, etc.) (26) dominus "lord" agricola "peasant" femina "woman" manus "hand" I II+III Morphology Adjectival Concord Pronominal Reference bon-us bon-us bon-a long-a -us -a -a -us IV Natural Gender male male female Grammatical Gender masculine masculine feminine feminine For the purposes of Latin it was not necessaiy to deal with criteria (ii) and (iii) separately. However, in the Germanic languages the situation is different; here, adjectival concord is purely formal in being overridden by the morphological criterion (i). Thus in Old English the noun wifmann "woman" shows the same type of adjectival concord as its masculine counterpart mann "man": se goda wifmann "the good woman" and se goda mann "the good man". However, in pronominal reference the noun wifmann behaved as feminine in accordance with its denoted natural gender (it was necessary to say Gesdwest 6u hie? "Did you see her''" and not hine "him" when referring to a woman). What is the notional basis for gender assignment in Indo-European languages? As is well-known, the three grammatical genders masculine, feminine and neuter found in many Indo-European languages reflect the association established by the traditional grammar between natural gender (sex) and grammatical gender. However, as is equally well-known all Indo- I .....I« .in languages abound in disagreements between linguistic and non-linguistic worlds. I'mh i ibi.il examples come from Old English where vwjP'wife, woman" and maigden "maiden" " • i> muter (as are their German counterparts das Weib and das Mädchen); among body parts '....."Ihcast" and heafod "head" were neuter, but wamb "belly" and caxl "shoulder" were Ii unlink' i in the other hand, some natural feminines (nouns denoting female bemgs) were grammati-■ masculine: wifmann "woman" and nuegdenmann "virgin". Similar examples could be iniilii|ilied from all Indo-European languages. It is significant that diachronically these 'illogical' \ :ii ins were everywhere 'improved' on in essentially two ways. On the one hand, the language t"i( ml of the gender system or, on the other hand, it made the system more 'logical' by i .i.ililisliing more agreement between natural and grammatical gender. Standard English went in iln lust direction when its "illogical' system of gender started to break down by the end of the i id I I nglish period (11,h c.) and the result is the genderless typology of Modem English. The loss .•mill be only partial, as is exemplified by French, which modified the three-way gender ilr.iim lion of Latin into a two-way distinction (masculine - feminine). On the other hand, Slavic I.Mi|'.iiii|',cs and Modem Greek preserved the three genders of Indo-European and consequently had in introduce various alignments of the 'illogical' relationship between natural and grammatical cruder. In Czech the natural distinction male animate (human or animal) being vs. female animate lii'ing vs. 'neutral' animate being (i.e., functionally neutral young human or animal) is frequently p.iiallclcd by the three-way grammatical distinction. Consider some examples: 27) Masculine Feminine Neuter býk kráva tele "bull" "cow" "calf beran ovce jehně "ram" "ewe" "lamb" hřebec klisna hříbě "stallion" "mare" "foal" muž žena dítě "man" "woman" "child" Smaller adjustments on the 'illogical' gender system are numerous in dialects. For instance Yiddish reasoned the gender of "horse"; der Pferd is masculine in , ,ei-man keeps the less ' iogtcal' neuter das Pferd. However, we shou.d be careful no, to push this type of reasonmg too far. For mstance, it would be hard to argue that Yiddish reassigned the noun Lser "water" (which ,s neuter in Standard German das Waster) to -he category of feminine, 84 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 85 die Wasser, for some deeper notional reasons as in the case of the "horse". In this case the most likely explanation is the interference from the Slavic adstratal languages where voda "water" is feminine. To criticize 'illogical' gender systems of various languages from a universalist viewpoint that only nouns denoting animate beings could be legitimately subcatcgorized into masculine and feminine (and eventually neuter) would be very misleading. Linguists have to look at language specific reasons (language internal 'logic') for this subcategorization. It is exactly here that all kinds of perennial problems of linguistic semantics emerge. Let us consider some examples from Latin. The most general principles of gender assignment (not including nouns denoting animate beings) in Latin can be formulated as follows: Masculine: names of nations, rivers, winds and months. Feminine: names of trees, countries, islands and cities. These semantic principles override those of morphology. For example, the river Tibens "Tiber" is masculine, even if the morphologically identical noun (belonging to the same declinational pattern) turns "tower" is feminine. However, no rule is without exceptions; thus the river Mdtrona "Marne" is feminine, the reason obviously being its morphology, i.e. the suffix -a. What is the ultimate explanation of the 'inherent masculinity' of rivers, winds and months in Latin? To take another example: the city of "Rome" Roma is feminine and so is the city of "Corinth" Corinthus. In the second case semantic principles override again those of morphology (Cormthus is inflected identically with dominus "lord"). What is the ultimate explanation of the 'inherent femininity' of cities, trees, countries and islands? That these problems are not pseudoproblenis and consequently that they have their legitimate place in any coherent expose of linguistic systems maybe proven by examining another language. Let us compare principles of gender assignment in Hebrew with those of Latin to sec if there is something 'deeper' behind their seemingly illogical systems. The Hebrew principles are basically as follows: Masculine: names of nations, rivers, mountains, seas, winds, months and metals. Feminine: names of cities, countries, parts of human body occurring in pairs. It appears that there is far-reaching agreement between Latin and Hebrew which cannot be due to coincidence. Various explanations have been proposed ranging from mythological to more linguistically based reasoning. Thus if female beings living in trees make part of the mythology of a certain nation, it should not be suiprising to find names of trees being assigned feminine gender in the language of this nation. Countries and cities are felt to be naturally feminine for all kinds of reasons (mother and her children?), but it is worth mentioning that very frequently they are regarded as collectives (cf. 3.3 and 5.2.2). Evidence for collectives being somehow 'inherently feminine' may be found in a variety of languages (e.g., in Old English all the »••"•«« ii i c i n urns are feminine: burg "city", ceastcr "military' camp", selr "county"). But obviously 'i» n i, nothing universal on that phenomenon as can be proven by examining other languages, tu i .i im.m names of cities are neuter and in Czech they can belong to any gender: Montreal is 11 *'i ■ * iiluic, I'raha "Prague" is feminine, and Toronto is neuter; in Czech morphological criteria i .iillm (), -a, -o) override those of semantics. ' ' Xitiithcr I In distinction between singular and plural, found in many languages all over the world, is ill. iinr.i common manifestation of the category of number. A logical counterpart of this Uuľiiľ.iH category is the category of quantity which is based on the recognition ofunity and I • I ii ■ uliiy (i.e., persons, animals and objects can be enumerated as 'one' or 'many' - more than '•m i I lie distinction between 'one' and 'more than one', i.e. countability is far from being a • n.iii'.liilorward notion since many entities are simply not countable. It is usually claimed that muss nouns are not pluralizable, which might imply that entities denoted by them are not i niini.ilile. However, we have to keep in mind, that what is taken as a 'single object', 'many . .1»i.■ i is', 'group of objects' or an amorphous 'mass' depends to a considerable degree on the I. mi.iI and morphological make-up of individual languages. But there is no doubt that the ■.i iii.niiic categories countable, mass and collective have to figure in our descriptions of nulu idual languages even if the semantic categorization of the world varies from language to l.iiij'.uage. Indeed, languages differ drastically in their grammaticah/.alion of these semantic imiuTsals. Assuming that the notion of countability is probably a universal category of all human I.manages, we may place it under substance in the following hierarchy of semantic features shown m figure 5.2. It is fairly common to recategorize certain mass nouns as countable in certain i niiiexls, such as They drink three or four different wines at every meal. Also some nouns can be used both as mass or countable, for instance, Have some apple vs. Have an apple. I he category of collective (group of objects as opposed to many objects) is not gramtnaiic-.ili/cd in English, but it is grammaticalized in many other languages. For instance in Arabic i nllective nouns (nouns denoting natural groups of objects or beings) are a starting point in the derivation of singulative nouns (nouns denoting an individual unit of what its basic noun denotes i nllectively); the examples shown in (28) are from Syrian Arabic. (28) Collective ?amerkän "Americans" dsbbän "flies" bed "eggs" Singulative ?amerkän-i "an American" dabbän-e "a fly" bédd-a "an egg" We have to emphasize that collectives (except for ethnic collective.) are g—.ally singula,. is shown by their agreement as in the sentence b^abal (the verb m singular) ; (he noun n intern a! plural, "in the mountains live wolves". It may be said that ,n Arabic AN INTRODUCTION TO THE STUDY OF MORPHOLOGY substance animate inanimate concrete book milk abstract love human man non-human dog Fig. 5.2 Hierarchy of semantic features collectives behave as mass nouns in English. In the latter language observe that hunters tend to use words denoting animals as mass nouns: an elephant and a herd of elephant; a fish and a school offish. An interesting phenomenon in Arabic is the possibility of forming plurals of collectives. In a sense, we may talk about a double plural paraphrasable as "many/various groups", shown in Figure 5.3. Since even singulatives are pluralizable we may obtain two different plurals with many nouns. The system works as shown in (29) for Egyptian Arabic. (29) Singular Plural Singulativc samak-e "a fish" samak-at "many" Collective samak "(a school of) fish" Tasmak "(various types of) fish, fishes" Singulativc sagar-a "a tree" sagar-at "(a few) trees" Collective sagar "(a lot of) trees" ?aSgar "(different kinds of) trees" In similar cases the plural of a singulative (a count plural) stands in contrast to the plural of the underlying collective (which indicates abundance or variety and which is not used after numerals). In terms of morphology, the singulative sagar-a "a tree" is regularly related to the collective sugar "(a lot of) trees" or "trees" in general by the addition of the feminine suffix -a (or -e). This suffix is lengthened and -t added to form the so-called plural of paucity since forms such as sagaral "(a few) trees" occur most commonly with numerals from three to ten. The 'double plural', called plural of abundance, is formed by a base pattern change ('broken' plural): SaGaR —» aSGdR "(different kinds of) trees". INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 87 Internal plural (many individuals in a group) External plural (many groups) Fig. 5..1 Plural of collectives Another numerical category in Arabic is dual. In Classical Arabic it is formed by the suffix ■ mi (Egyptian -en) added to both masculine (rajul-dni "two men), and feminine and singularized • nlleclive nouns (zawj-at-dni) "two wives" and sajar-atdni "two Irees"). It possesses only one iiblu|iie form -a\ni (both genitive and accusative) versus two oblique forms in the singular (-in ri-niiive and -an accusative). The plural may display two or only one oblique form (depending mi whether it is formed internally or externally): (30) Singular "king" Norn malik-un Gen -in Acc -an Plural (internal) mulük-un -in -an Plural (external) "clerk" kätib-üna -Ina -Ina Dual malik-äni -ayni -ayni , h,s state of affairs is paralleled by other languages possessing the category ot dual (Inuktitu , Sanskrit Ancient Greek). In terms of universal principles ofmarkedness (cl. 3.3) one expects fewer forms in the marked categories of plural and dual. For instance in Ancient Greek masculine nouns distinguish five forms (with vocative) in the singular, four in the plural and only two ,n the most marked category of dual. This is shown in (31): AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CA1 ĽGORIES ASSOCIATED WITH NOMINAL ELEMENTS 89 (31 Ancient Greek masculine nouns (o-stcms) "man- Singular Nom ánthrôp-os Gen -ou Dat Acc Voc -oy -on -e Plural ánthrôp-oi -ôn -ois -ous -oi Dual anthröp-ö -ôin -öin -ö -ô Given the seeming universality of the notion of countability it may come as a surprise that some languages do not have a grammatical category of number. A notorious example is Chinese, where the distinction between singular and plural can be made, if necessary, by means of a numeral but it may equally well be left unexpressed. Other means of expressing plurality in languages without morphological plural include reduplication of the lexical item or of its attribute. For instance Malay pluralizes as shown below: (32) orang "man" pěrěmpuan "woman" orang-orang "people" perempuan-perempuan "women" However, it should be kept in mind that the morphological process of reduplication expresses not only numerical plurality but also other notions such as indelmiteness, intensity or distribution. Examples from the same language include lujoh orang "seven people" (not *lujoh orang-orang), lama-lama dahulu "long ago", mata "eye" but mata-mala "policeman". Reduplication of the attributive adjective is common in Sumerian: na "stone", gal "big" na-gal-gal "big stones". It is of interest to notice that only nouns denoting THINGS are pluralizable by reduplication; nouns denoting PERSONS have to be pluralized by the suffix -ene, e.g., lugal "king" lugal-ene "kings", cf. 5.2.1). 5.2.3 Case It was recognized by ancient granmiarians a long time ago that case is the most important of the inflectional categories of the noun. In a traditional display of cases, such as that familiar from textbooks of Latin, Greek, Old English or German, each case is given a label which suggests at least one of its semantic functions. Thus the nominative was the case associated with naming (or marking) the subject of the sentence, the dative was the case denoting the receiver or beneficiary of giving. Some linguists tend to disregard these traditional taxonomies as worthless; this attitude, however, is based on taking certain cases at their 'face value' (thus it is easy to show how 'illogical' the Latin case system is by singling out examples such as accusative of place, where Latin uses the same syntactic case which is appropriate with transitive verbs: Roinam - ■• I :,ce Rome" and Roinam ed "I go to Rome"). However, no traditional grammar maintains iii.ii Libelling cases by their principal semantico-syntactic functions (basic meaning) exhausts the ini.it meaning of various cases and all traditional grammars have to specify a whole array of nh.hli.iiy meanings in lengthy sections dealing with syntactic and semantic values of cases. !!• hue discussing attempts to specify the category of case in terms of its total meaning, let us ■ - .iiiuiii.- some of the most common semantico-syntactic properties (grammatical functions) of . .r.i . m Latin. The traditional paradigm with principal semantico-syntactic functions of mi.In idual cases is given in (33): Case Function dominus "lord" Nominative subject of a sentence domine Vocative name of addressee dominum Accusative direct object of transitive verb domini Genitive possession domino Dative indirect object dominö Ablative (i) point of departure (ii) instrument Starting with the genitive (the case of 'possession') it is obvious that this label is fitting in . samples such as domus met paths "my father's house" where "my father" is a possessor and the house" his possession. However, we will be in trouble if we try to use the label 'possessive' in . .ises such as amorpatris. This phrase is ambiguous and can be translated (i) "father's love" or (in "love towards father". (Transformational Grammar maintained that amor patris in meaning (i) is a transformation of pater amat "father loves [us]" and in meaning (ii) that of amdmus I'ittrem "we love [our] father"). Obviously, "my father's love toward me" and "my father's house" can hardly be said to be 'possessed' in the same way. Traditional grammars recognize this I,hi by referring to the case in (i) as subjective genitive ("the father loves") and the ease m (ii) •is objective genitive ("we love the father"); in other words, they recognize the different sources ol the genitive. To use another example, the genitive decern annorurn in a phrase puer decern annorum, lit. boy often years, "ten year old boy" cannot be labelled 'possessive'. Its value is -.imply descriptive of a certain age of the boy; the genitive expresses here an attribute of the substantive (note that Russian would use an adjective in this case). If we examined a sufficient number of examples, it would appear that there is a basic dichotomy between purely syntactic genitives and the ones with more semantic content; the former can be subdivided into the subjective and objective genitives, the latter into a number of subtypes such as possessive, descriptive, partitive, etc. The last mentioned subtype of the genitive, the partitive genitive, denotes totality from which a part is taken out, e.g., libra olei"a pound of oil". See Figure 5.4. Let us examine some of the functions of the ablative. Its name (ab-ldtus is the passive participle ol'd-ferro "take away") suggests that one of its semantic functions is local (or spatial) 90 AN INTRODUCTION TO THE STUDY OE MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 91 Genitive subjective objective possessive descriptive partitive fig. 5.4 Types of the genitive since this case could express the 'point of departure' (i.e., place front which): Rotnd exed "1 go from Rome" and domo abed "1 am leaving the house". !t should be mentioned that prepositionlcss constructions of this type were common only with proper names of cities and smaller islands; elsewhere it was necessary to use the preposition ex Italia eo "I go from Italy". However, in the following examples we can talk of the point of departure only metaphorically. (34) Venus love nata et Diona "Venus born from Jupiter and Dione" Natus loco nobilt "Bom from the noble family" This abstract nuance of meaning of the former concrete local meaning was strong enough to ancient grammarians to justify a label abldtivus originis "ablative of origin". This subtype of ablative could only co-occur with passive participles denoting "born from" (such as natus. genitus. ortus, satus). We have to introduce another label to describe examples such as vivo Athenis "I live in Athens" or node dormio "1 sleep during the night". Obviously, here we are not dealing with the 'point of departure' but with the spatiotemporal framework of utterances; hence, the traditional label abldtivus loci et temporis "ablative of place and time". However, what is morphologically the genitive case could function in the same way, e.g., Romac "in Rome", Corinth! "in Corinth" (historically, here we are dealing with the Old Latin locative in -i); the ablative has to be used with proper names of cities and smaller islands if they belong to the 3rd Declension or if they are morphologically plural, e.g., Athenae "Athens"). Totally different semantic values of the ablative can be observed in the following examples: (35) Vir summo ingenio "A man of great talent" ('ornibus tauri se tutantur "Bulls defend themselves with horns" I acrimat gaudio "1 le weeps from joy" 11 .ulitiiiiially, these are called abldtivus qualitdtis "ablative of quality", abldtivus instrument! in .iiuinciital" and abldtivus causae "ablative of cause." These (and) some others, e.g., i limitative "with" (as in Caesar omnibus uopitsproficiscitur... "Caesar goes with all the army...") ......upposed to be subcategories of the instrumental. We may summarize our findings regarding iIh uii-aiiing of the Latin ablative in Figure 5.5. In dealing with the functions expressed by the l .iim> ablative an interesting observation can be made. It appears that there is a dichotomy In iuivii the more 'abstract' (grammatical) functions (such as comparison, quality, cause, .n. niiipaniment instrumentality) and the more 'concrete' ('local') functions (direction, space, inn. ) This distinction, 'grammatical' vs. 'local', is fairly common in many treatments of case- ,\■.kins found in a variety of languages. Since Latin gravitates toward 'grammatical' functions dli. loss of the Old Latin locative was mentioned above) we may profit from examining another liiiic.iiage where the 'grammatical' and 'local' notions are more in balance. An ideal language for ihi'. purpose is Turkish, which has three 'grammatical' cases (Nominative, Accusative, Genitive) .mil three 'local' cases (Dative, Locative, Ablative). (36) Case ev "house" Nominative cv-i Accusative ev-in Genitive ev-e Dative ev-de Locative ev-den Ablative Function i) subject of a sentence ii) indefinite direct object definite direct object possessive i) indirect object ii) allativc; place to (whither) time at (when) place in (where) place from (whence) I lie system of 'local' oppositions in Turkish is very simple: eve "to the house", evde "in the house" evden "from the house". As in Latin these three cases may be used in a more abstract 'grammatical' sense such as the dative of purpose: Ktz cicek dermieg-e cikiyor "the girl is going mil to pick flowers"; locative of property: kahve rengin-de bir sapka "a hat of coffee-colour"; ablative of cause: aclik-tan bitkm "exhausted from hunger". It may be noted that in contradis-imction with Latin, nominative is a misnomer for Turkish and the term absolutive would be more appropriate. The reason is the peculiar syntactic behavior of the absolute (suffixless) form which can appear as both a subject of a sentence and an indefinite direct object of a verb, as shown in (37). 92 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY Ablative direction origin comparison space time quality instrument cause accompaniment Fig. 5.5 Types of the ablative (37) ev acrldi "the house was opened" cv aldim "I bought a house" The accusative case can only be used if the object is defined (i.e., the accusative marks the definite object of a verb): (38) evi aldim "I bought the house" Case and definiteness intersect in similar ways in other languages such as Hebrew, Persian and Spanish. Observe, for instance, that in Spanish one has to say Busco a mi professor "1 am looking for my professor" (definite) but Busco un professor "1 am looking for a professor" (indefinite). A more complex system of'local' oppositions exists in Finnish. Here the three-way 'local' opposition of Turkish: "to" - "in" - "from" is combined with the features exterior vs. interior. As in Latin, there arc also grammatical cases (Nominative, Genitive, Accusative and three specifically Finnish cases). The whole system consists of 15 cases (the record is probably held by Tabassaran from the North-East Caucasian family with 54 cases) which can be displayed as shown below for talo "house": (39) Grammatical Cases talo "house" talo-n talo-n talo Case Nominative Genitive Accusative J Accusative II Function subject of a sentence possessive direct object direct object (in certain modal clauses) \FLFCT10NAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 93 lalo-a talo-na lalo-ksi talo-n Partitive mass or part of a whole from which a part is taken out Essive state Translative change of state Instrumental instrument, means lalo-ssa talo-sta lalo-on talo-lla talo-lta talo-Ile 'Interior' 'Exterior' "in the house" "from (inside) the house" "into the house" "at/near the house" "from (outside) the house" "to/towards the house" ,,„,, are two more cases which arc not local cases (and which are better not classified as ii.iinniatical): talo-tta talo-inensa Abessive "without the house" Comitative "with the house/s" (ompared with Turkish the Finnish system of local cases is more precise in marking implicitly the contrast exterior vs. interior, shown in Figure 5.6. This, of course, is not to claim lli.it distinctions such as "from inside the house" and "from outside the house" cannot be made in I urkish. What is meant is that Turkish ablative ev-den "from the house" does not show any .i iiniiphological markers; the 'alignment of case marking' refers in a neutral way to nominative-»(i native, ergative-absolutivc (and other patterns such as double-oblique and active-inactive). I Im I,manages we studied in 5.2.3 (Latin, Russian, Turkish, bul not Finnish) arc of familiar .....nnulivc-accusative typology which assigns the same suffix to both the agent (subject of the 11.m •. 111 \ o verb) and the subject (of the intransitive verb). On the other hand, languages, such as limkiiiiit (Eskimo). Georgian, and Hindi, are of less familiar ergative-absolutivc typology which i .m^iis Ihe same suffix to the subject of the intransitive verb and the patient (object of the ii.iiiMlive verb) while the agent is marked by a special suffix (so-called ergative case). The .11n.ition in Latin vs. Hindi is as follows: (43) Agent Subject Patient Latin Nominative-accusative Alignment -us -urn Hindi Ergative-absolutivc Alignment =ne -ä •onsider Latin (44) and Hindi (45) equivalents of the two sentences: "The friend saw „,„1 "The horse came" (it may be observed that English displays neutral alignment morphological markers for any of the three semantic functions): (44) Amicus friendj NOM Agent Equ-us horse-iNOM Subject (45) Dóst=ně fricnd-ERG Agent equ-um horse t ACC Patient vidit see+PF.RF^3SG a horse" with no (Latin) venit come+PERF+3SG ghör-ä horse+ABS Patient dekh-ä see+PP (Hindi) 96 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY Ghör-ä horse FA BS Subject äy-ä come i PP present tense: (46) Dôst „hôrS ,_lu gnor-a dekhta hai feendiABS horse+ABS sce+PRT is "The friend sees a horse" (Hindi) If the object is definite (marked by the DAT/ACC postposition =kó), in tiie crgative tenses the verb is always in the unmarked ('masculine') form in -a irrespectively of the number of the object. This is shown in (47): (47) i. Dósl=ně ghor-ě^kó děkh-á friend=ERG horse+OBL-DAT/ACC see+PP "The friend saw the horse" (47) ii. Dósr=ně ghor-o-k5 dékh-á friend=ERG horse+OBL/PL=DAT/ACC scc+PP "The friend saw the horses" If, however, the object is indefinite, typical ergative agreement may be observed in the plural: (48) Dost-né ghór-ě děkh-ě fricnd-ERG horsc+PL see+PP/PL "The friend saw horses" INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 97 RECOMMENDED READINGS \ i n In mih, John M. 1971. The Grammar of Case. Cambridge: Cambridge University Press. A i ij fl< !!<_-, Joseph R. 1958. An Outline of the Structure ofShilha. New York: American Council ni I earned Societies. H. i. h. I ininon W. & Robert T. Hanns, ed. 1968. Universals in Linguistic Theory. New York: Holt, Kinehart and Winston. .....ipbell, A. 1959. Old English Grammar. Oxford: Clarendon Press. i n\u-ll, Mark W. 1964. A Reference Grammar of Syrian Arabic. Washington, D.C.: Georgetown I iniversity Press. i m nie. (icorge O. 1960. A Grammar of the German Language. New York: Frederick Ungar. I ii.ikonoff, Igor M. 1988. Afrasian Languages. Moscow: Nauka. I .ilkcnslein, Adam. 1949. Grammatik der Sprache Gudeas von Lugas. Roma: Pontificium Insiilutum Biblicuni. I illmore, Charles .1. 1968. "The case for case". Bach & Harms 1968.1 88. lodní, I. 1959. "The origin of grammatical gender". Lingua 7.1 ~41, 186-214. II iii i is, Alice C. & Lylc Campbell. 1995. Historical Syntax in Cross-Linguistic Perspective. ( ambridge: Cambridge University Press. I l.i\ r.inek, Bohuslav & Alois Jedlička. 1963. Česká mluvnice. Praha: Státní pedagogické nakladatelství. I l|clmslev, Louis. 1935. La categoric des cas. Etude de grammaire generale. Acta Jutlandica VII.l.xij-184and LX.2.viij 78. I. ikobson, Roman. 1936. "Beitrag zur allgemeinen Kasuslehre". Selected Writings II, 23-71. The Hague: Mouton. Joly, André. 1975. "Toward a theory of gender in Modem English". Studies in English Grammar cd. by André Joly, 229-287. Lille: Presses de l'Université de Lille. K.uz, Jerrold J. 1966. The Philosophy of Language. New York: Harper and Row. Kurylowicz, Jerzy. 1964. Inflectional Categories of Indo-European. Heidelberg: Carl Winter. 1 ewis, Geoffrey L. 1967. Turkish Grammar. Oxford: Clarendon Press. I ewis, M. B. 1968. Malay. London: The English Universities Press. I yons,John. 1977'. Semantics. Volume 2. Cambridge: Cambridge University Press. (Chapter 15). Moreland, Floyd L. & Rita M. Fleischer. 1973. Latin: An Intensive Course. Berkeley: University of California Press. Robins, Robert H. 1967. A Short History of Linguistics. London: Longmans. Kosén, Haiim B. 1977. Contemporary Hebrew. The Hague: Mouton. Schachter, Paul. 1985. "Parls-of-speech systems". Language Typology and Syntactic Description. Volume 1 ed. by T. Shopen, 3-61. Cambridge: Cambridge University Press. 98 AN INTRODUCTION TO TUR STUDY OF MORPHOI. OGY EXERCISES 1. Analyze the system of possessive suffixes in Biblical Hebrew. (a) How are the sex and the number of the possessor, and the grammatical gender and the number of the possessed object realized morphologically? Hint: Before answering this question construct FOUR paradigmatic sets of Hebrew possessive suffixes for the following categories: Possessed Sg - Possessor Sg Possessed PI - Possessor Sg Possessed Sg - Possessor PI Possessed PI - Possessor PI and specify their sex/gender distinctions in appropriate persons. Be careful with English "your" which is four-way ambiguous: Masc Sg, Masc PI, Fem Sg, Fern PI. Use Ihe following data: (1) malki "my king" (2) süsööehen "their (F) mares" (3) šírexä "your (M) song" (4) moläxoočhém "their (M) queens" (5) doßäri "my word" (6) malkäQÓ "his queen" (7) šírexém "your(M) songs" (8) süsäöäh "her mare" (9) maläxenü "our kings" (10) süsö6aMiä "her mares" (11) malkěnu "our king" (12) široxém "your (M) song" (13) dsßarxen "your (F) word" (14) safäráyix "your (F) book" (15) šíršxä "your (M) songs" (16) süsäGän "their (F) mare" (17) diprěxén "your (F) words" (18) malxěxém "your (M) kings" (19) moläxôoáyix "your (F) queens" (20) sifrexén "your (F) books" (21) sifrěx "your (F) books" INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 99 (22) doßärex "your (F) word" (23) süsöGäxä "your (M) mares" (24) malkö 'his king" (25) malkän "their (F) king" (26) sifräm "their (M) book" (27) daßäräh "her word" (28) malkoxém "your (M) king" (29) nvaläxöGäw "his queens" (30) sifraxen "your(F)book" (31) malkä9äm "their (M) queen" (32) moläxáy "my kings" 1 he basic forms of the above nouns and their plural forms are as follows: mélex "king" msläxím sěfer "book" sofärim malkä "queen" maläxöö Šír "song" širim däßär "word" daßärim süsä "mare" süsöö (b) Describe the distribution of morphophonemic variants of the root. (c) Comment on the Teak' in the system of the possessive suffixes. (d) Translate into Hebrew: (33) "our words" (34) "our song" (35) "their words" (36) "your (M) word" (37) "your (M) queens" (38) "our songs" Analyze the system of possessive affixes in Coptic. (a) How arc the sex, number and person of ihe possessor, and the gender and the number of the possessed realized morphologically? Note: Coptic distinguishes two genders: masculine, e.g., kas "bone" and feminine, e.g., ovhe "tooth". (b) Construct the paradigmatic set of Coptic possessive affixes. Use the following data: 100 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY (1) pajöt "my father" (2) pefkot "his basket" (3) pekkah "thy (M) earth (4) nefkot "his baskets" (5) penjöt "our father" (6) peuköhit "their fire" (7) tenk'ic "our hand" (8) nenjöt "our fathers" (9) tekbö "thy (M) tree" (10) nelinovhe "your teeth" (11) toubö "thy (F) tree" (12) tesmäu "her mother" (13) taape "my head" (14) nekehe "thy (M) cows" (15) nakol "my baskets" (16) nesojk "her breads" (17) teumäu "their mother" (18) tesk'ic "her hand" (19) tefehe "his cow" (20) teubö "their tree" (21) lesovhe "her tooth" (22) nouovhe "thy (F) teeth" (23) poujöt "thy (F) father" (24) nenbö "our trees" (25) petinlas "your tongue" (26) netinehe "your cows" (27) tetinbö "your tree" (28) poukot "thy (F) basket" (29) ncukas "their bones" (30) i nekovhe "thy (M) teeth" 3. Plural formation in German can be described in its interplay with grammatical gender (M F, N). There are several plural suffixes (-,, -en, -0) and the root can be umlauted (a -a, u ■ ■» it, o -> 6, au -* an). (a) Elaborate as many plural patterns as possible in the following data. (b) Reduce their number by disregarding 'exceptions' (single occurrences). INFLECTIONAL CATEGORIES ASSOCIATED Willi NOMINAL ELEMENTS 101 i, ) Endeavor to make significant general statements regarding the distribution of four pluralizing suffixes and umlaut, and their interplay with gender. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) ( Tag Bach Onkel Sohn Hand Otter Jahr Rand Vogel Löwe (11) Apparat .) Biss (13) Fluss (14) Trübsal (15) K-unst (16) Tafel (17) Endung (18) Ente (19) Schaf (20) Bild (21) Haus (22) Schloss (23) Fenster (24) Kloster (25) Auge (26) Ohr (27) Geist (28) Kessel (29) Balken (30) Graben (31) Lehrer (32) Vater (33) Bruder (34) Bruch (35) Mensch (36) Buckel "day" "brook" "uncle" "son" "hand" "otter" "year" "margin" "bird" "lion" "utensil" "bite" "river" "sorrow" "art" "tablet" "ending" "duck" "sheep" "picture" "house" "castle: lock" "window" "monastery" "eye" "ear" "ghost" "kettle" "beam" "ditch" "teacher" "father" "brother" "fraction" "man" "hump" Tage (M) Bäche (M) Onkel (M) Söhne (M) Hände (F) Otter (M) Jahre (N) Ränder (M) Vögel (M) Löwen (M) Apparate (M) Bisse (M) Flüsse (M) Trübsale (F) Künste (F) Tafeln (F) Endungen(F) Enten (F) Schafe (N) Bilder (N) Häuser (N) Schlösser (N) Fenster (N) Klöster (N) Augen (N) Ohren (N) Geister (M) Kessel (M) Balken (M) Gräben (M) Lehrer (M) Väter (M) Brüder (M) Brüche (M) Menschen (M) Buckel (M) 102 INFLECTIONAL CATEGORIES ASSOCIATED WITH NOMINAL ELEMENTS 103 (37) Büchse (3S) Eber (39) Einfahrt (40) Einfall (41) Glas (42) Graf (43) Gunst (44) Gut (45) Haar (46) Hacken (47) Haifisch (48) Hahn (49) Halle (50) Herr (51) Hirn (52) Holz (53) Floss (54) Floh (55) Fohlen (56) Flut (57) Jude (58) Jagd (59) Joch (60) Order "box" "boar" "entrance" "intrusion" "glass" "count" "favor" "merchandise" "hair" "heel" "shark" "rooster" "hall" "lord" "brain" "wood" "raft" "flea" "foal" "flood" "Jew" "hunt" "yoke" "command" Büchsen (F) Eber (M) Einfahrten (F) Einfälle (JV1) Gläser (N) Grafen (M) Giinste (F) Güter (N) Haare (N) Hacken (M) Haifische (M) Hähne (M) Hallen (F) Herren (M) Hirne (N) Hölzer (N) Flösse (N) Flöhe (M) Fohlen (N) Fluten (F) Juden (M) Jagden (F) Joche (N) Ordern (F) (1) al-rajulu wasixun (2) al-kalbu wasixun (3) al-mar?alu wasixatun (4) al-rijälu wasixüna (5) al-kiläbu wasixatun (6) al-mar?ätu wasixatun (7) al-rajulu daxala (8) al-kalbu daxala (9) al-mar?atu daxalat (10) al-rijälu daxalu "the man is dirty" "the dog is dirty" "the woman is dirty" "the men are dirty" "the dogs are dirty" "the women are dirty' "the man came in" "the dog came in" "the woman came in" "the men came in" 11 11 al-kilabu daxalal 11 ') afmarVatu daxalna "the dogs came in" "the women came in" i .i) Using the following data elaborate the rules of verbal agreement in Biblical Hebrew; do it separately for the future and the past: (1) tismor ha-?issa (2) yiqtalü hä-7äßö9 (3) bau, tifbhl (4) ben, tiqtöl (5) tilködnä han-näsim (6) yiftah hä-?äß (7) näsTm, tismörnä (8) 'Ponäsim, tiftshü (9) päeahä hab-ba6 (10) sämarü han-näsim (11) läxad hab-ben (12) qätslü hä-?anäsim "the woman will protect" "the fathers will kill" "daughter, you will open" "son, you will kill" "the women will catch" "the father will open" "women, you will protect" "men, you will open" "the daughter opened" "the women protected" "the son caught" "the men killed" (b) There are three phenomena in Hebrew verbal agreement which are unknown in the Indo-European languages. Identify them clearly using appropriate terminology. (c) Translate into Hebrew: (13) "sons (= bdmm), you will kill" (14) "the father protected" (15) "daughter, you will protect" (16) "the women opened" Berber (Taselhit) spoken in southwest Morocco (according to Applegate 1958): (1) asif(M) "river" isafh (2) tagzif(F) "palm tree" tigzaf (3) tahanut (F) "stove" tihuna (4) amdakul (M) "friend" imdukal (5) tamdakult (F) "friend" timdukal (6) agudid (M) "bird" igudad (7) agadir (M) "fortress" igudar (8) adrar (M) "mountain" idrarn 104 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED Wli'H NOMINAL ELĽMEN I S (a) Describe the formatjon of the plural in Berber. (b) What type of affix is used to mark the feminine gender? (c) How does marking for the gender relate to marking for the plural? 7. Elaborate the rules of adjectival agreement (in gender and number) governing the morphology of numerals and counted objects in Biblical Hebrew: (1) ?išsa (2) ríš (3) šanayim (4) šstayim (5) šälôš (6) šološä (7) Varbaí (8) ?arbävä (9) ?arbaí (10) ?arbäíä (11) hsmiššä (12) hämčš ?aha6 ?£chäô koläßim naším íärím koläßim bänöG moläxlni běsím YeŕŕônoS ?äßöe moläxö0 "one woman" "one man" "two dogs" "two women" "three cities" "three dogs" "four daughters" "four kings" "four eggs" "four pencils" "five fathers" "five queens" Translate into Hebrew: (a) (b) (c) (d) (e) (0 (g) (h) (i) (J) "three women" "one father" "two queens" "three books" "four bitches" "one egg" "two kings" "two pencils" "five women" "one pencil" Vocabulary: keleß "dog" ?äß fir "city" (Fem) malkä ba9 "daughter" sefer melek "king" kalbä "father" "queen" "book" (Mase) "bitch" besä egg Yipparon "pencil" I (escribe the semantic values of Turkish 'local' cases. Use the following data (from Lewis l'H>7): I ocative 11) su-da (.') Ramazan-da I \) ihtiyarlik-ta (1) yirmi ya§in-da (s) bu fikir-dc degilim "in the water" "in Ramadan" (the month of fasting) "in old age" "twenty years old" "1 am not of this opinion" Ablative (<>) sehir-den aynldi 17) pencere-den girdi (X) on-dan (9) Türkiye l.übnan-dan büyük-tür (10) naylon-dan (11) kom^ular-dan biri (12) bu elmalarl kac-tan aldln? "he departed from the city" "he entered by the window" "for that reason" "Turkey is bigger than Lebanon" "of nylon" "one of the neighbors" "at what price did you buy these apples?" Dative (13) mektubu Ali-ye gösterdim (14) Türkiye-ye döndüler (15) talebe imtihan-a hazirlaniyor (16) bu elmalari kac-a aldin? "I showed the letter to Ali" "they returned to Turkey" "the student is preparing for the examination" "what was the total amount you paid for these apples9" Vocabulary: yirmi bu degilim aynl-di dön-dü-ler "twenty" "this" "1 am not" "he departed" "they returned" alma "apple" al-dm "you bought" mektub "letter" goster-dim "1 showed" talebe "student" haztrlan-iyor "he/she is preparing" INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 107 CHAPTER SIX INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 6.1 Verb as a Primary Grammatical Category As established under 5.1.1, the verb could be defined as a primary grammatical category the domain of which includes secondary categories of person, number, tense, aspect, mood, and voice. It was also mentioned that Plato grouped verbs and adjectives together since he considered the most typical function of both being that of predication; as we put it, verbs express dynamic features on substances (nouns) and adjectives express static features on substances. Whereas both verbs and adjectives predicate, the most typical function of the noun is that of naming the subject of predication. In other words, the definition of these two primary grammatical categories cannot be made independently of syntactic and logical considerations — subject and noun and predicate and verb are simply indissolubly associated in traditional grammatical and logical theory. A full discussion of functional categories ofsubjecl and predicate would of course bring us far beyond the domain of morphology proper. Suffice it to say that there is a far reaching agreement between the categories of logic and grammar in simple declarative sentences such as John ran away where the individual person (substance) is an instigator of the action. Apparently, in all languages of the world, the individual person in such a sentence would be grammaticalizcd as a noun and its action as a verb. Thus the correspondence between grammatical and logical categories are shown in Figure 6.1. At this point, it is important to establish the distinction between morphological and syntactic predication. The latter concerns the predication of one word on another, in this case of verb on subject, such as in English and Latin: (1) John ran away. Joannes efiugit However, in the context which is suitable to pronominal substitution, such as in answering the question Quid fecit Joannes "What did John do?", the strategies of English and Latin will differ. Compare the answers to this question: (2) He ran away. Effugit. In this case, Latin displays morphological predication, which may be defined as a predication which takes place within the system of the Latin verb. The Latin verb, in contrast to Grammar Morphology Noun Verb Syntax Subject Predicate Logic Agent Action Fig. 6.1 Correspondence between grammatical and logical categories ili.ii of English, allows for morphological predication using the secondary grammatical categories "I person and number. With this particular verb it is possible to predicate a different grammatical . .iiegoiy of number in the same person (3rd): fugerunt "they ran away" or different grammatical i .m-gories of person in the same number (singular): /Sgi" "I ran away", fugisti "you ran away" etc. \ee further discussion under 6.3.1. (i.2 Quasi-Nominal Categories of the Verb: Infinitive and Participle Quasi-nominal categories, namely, infinitive and participle (called also infinite or non-lluUo forms) share properties of both nouns and verbs. Participles in heavily lleclive languages lu-have like adjectives, i.e., they can be inflected for the nominal categories of number, case and j-i-iiiler (to a limited degree) and similarly to adjectives, they can form the positive and superlative I in a limited degree). With verbs they share the categories of aspect (to a limited degree), and also i>l voice and mood. Infinitives are more abstract in that they can be inflected only for the nominal category of case (but not for gender and number), and with verbs they share the categories of .ispect and voice. In both infinitives and participles the essential verbal categories of person and tense are missing: hence the label non-finite forms (opposed to finite verbal forms). Participation in nominal and verbal categories may be schematized as shown in Figure 6.2. Since many of these nominal and verbal categories are not realized by synthetic morphology in English, we may use Latin and Greek as examples in these cases. Compare the contrasts of aspect and voice with the infinitive in these languages. (3) English (to) praise (to) have praised (to) be praised Latin laudäre laudävisse (Perfect) laudäri Greek epainein cpainésai (Aorist), epéynekénai (Perfect) epaineisthai (Passive) The perfect infinitive is realized synthetically, i.e., by means of inflections in Latm and Greek (in Greek there are two other infinitives in the passive: epěynethénai (Aonst) and cpěyněsthai (Perfect), cf. (6)) but by means of the grammatical auxiliary have in English. The passive infinitive is realized analytically by means of the grammatical auxiliary be in English. 108 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY All these languages allow for the aspectual contrast even in the passive infinitive. The English system may be portrayed as a double binary opposition: (4) non-Perfect Active (to) praise Passive (to) be praised Perfect (to) have praised (to) have been praised The Latin system may be diagrammed similarly and it is of interest to observe an analytic formation in the perfect passive infinitive (where English has two grammatical auxiliaries have and he): (5) Infectum Perfectum Active laudare laudavisse Passive laudari laudátum esse Greek distinguishes the perfective (Aorist) and retrospective (Perfect) infinitives (see more under 6.3.3) and realizes all these distinctions synthetically: (6) Active Passive Imperfective epainein epaineisthai Perfective epainésai epainethěnai Retrospective epěynekénai epěyněsthai A parallel situation exists in the participle in English and Greek. English possesses four forms {praising, being praised, having praised and having been praised) and Greek six forms: (7) Imperfective Perfective Active epainön epainesäs Passive epainoúmenos epainetheís Retrospective epéynekos epěyněménos Latin does not possess the aspectual contrast of perfectivity in its participial system: lauddns "praising" is an active participle and laudätus "praised" its passive counterpart. But Latin (and also Greek) has participles which have modal meaning. The so-called future participle (formed curiously from the passive base by the suffix -ür-us: laudd-t-ür-us) is used only with the auxiliary esse "to be" in phrases which imply 'volition' on the part of the speaker. They correspond to English phrases "going to", "be about to": scriptürus sum means "I intend to write" or "I am going to write". Höstes bellum illätün erant may be translated "The enemy were likely to make war". The so-called gerundive (formed by the suffix -nd-us from the stem) is a passive participle whose contextual meanings are classifiable as follows: INFI.ECTIONAI CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS Nominal Elements 109 number sender number case comparison east 6^ Quasi-Nominal Elements Participle (Verbal Adjective) aspect voice gender numl Verbal Elements Verb ber ease comparison aspect mood voice person number tense aspect mood xoice Fig. 6.2 Primary and secondary grammatical categories 110 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS (i) the action which will be done in the future (i.e., the temporal category of futurity); (ii) the action which should be done (i.e., the modal category of necessity); (iii) the action which is under way (i.e., the aspectual category of progessivity). The modal meaning is the most common; for instance nobis eundum est "we have to (ought to) go", memona nobis exercenda est "we have to train our memory", etc. In his libris legendis "by reading these books" the third meaning can be exemplified. Let us now examine the nominal categories of participles and infinitives. As mentioned above, inflection for gender may be limited in participles. Thus in Latin, the active participle is not inflected for three genders, whereas the passive participle, the future participle and the gerundive are: (8) Present Participle Passive Participle Future Participle Gerundive laudät-us laudätür-us laudand-us laudäns -a -a -um ■a urn Greek, on the other hand, inflects unfailingly all its six participles, even the present participle, for three genders: (9) Latin Greek epainon epainousa epainoOn As far as case and number are concerned, Latin and Greek participles show the same number of forms as adjectives. It is astonishing to realize that all the six participles of Greek can be inflected for three genders, three numbers (Sg, Dual, PI) and four cases. It remains to demonstrate that the infinitive can be inflected for case. In Latin the syntactic cases (nominative and accusative, called also direct cases) are formally identical with the usual form in -are. English may use either the infinitive or the verbal noun in -ing in the function of the subject: Errdre humanum est "To err is human" or "Erring is human". Similarly, English may use both forms in the function of the object: Incipid scrTbere "1 start to write" or "I start writing". On the other hand, neither Latin nor English can use the infinitive in instances such ars scribendV'sxX of writing". The genitive of the infinitive (and other semantic cases, dative and ablative, called also oblique cases) are formed from the verb base in -nd plus the endings of the n-stem masculine nouns: (ID) Nom laudare Gen laudandl Dat laudando Acc laudare - ad laudandum Abl laudando these oblique cases are called gerunds (verbal nouns). The dative of the infinitive is quite i.in, ii occurs in constructions such as non sum solvendo "1 cannot pay" (lit. I am not [up] to paving). On the other hand, the ablative is quite common in adverbial phrases such as iniurids U undo "by (from) sustaining the injustice" and defessus dicendo "tired by (from) talking". As r. well-known, English has the option of constructing the gerundial phrase verbally (as above) • ii nominally "by sustaining of the injustice". Latin has an active option (as above) or a passive option inmriisferendis, where the gerundive has to be used. It is of interest to note that Greek has nothing comparable with Latin infinitival inflection by means of the gerund in oblique cases, i iieck inflects its infinitive simply by inflecting the preposed neuter article to (gentive toil, dative toy, accusative _ nominative to). Greek aethes toil hupakoiiein "not used to obey" would be lianslated by the gerund in Latin: insuetus oboediendi. An example of the dative: Nikeson hi gen toy logidzesthai kalds "Win over wrath by correct reasoning." <> J Secondary Grammatical Categories Associated with Verbal Elements (>..V1 Person andDeixis The category of person is definable with reference to the notion of participation in the discourse: the first person is used by the speaker to refer to her/himself as a subject of discourse; the second person represents the listener when spoken to about her/himself; the third person is used to refer to the persons (or things) other than the speaker and addressee. Tesniere (1959) introduced the following terms for these distinctions into French structural linguistics: (11) 'ontif - subject of discourse 'antiontif - its antipode, i.e. addressee 'anontif = neither subject nor addressee, i.e. spoken about The first and the second person are the positive members of the category of person (in that they refer to participants in discourse), whereas the third person is a negative notion. It is of interest to note that in many languages there is no overt marking for the third person and its meaning is given by the absence of the markers for the first and second person. Consider the personal endings in Turkish: 112 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY 2) Sg 1 2 3 gel-iyor-um -sun -0 "I am coming" "you are coming" "he/she/it is coming" {-iyor = progressive aspect) Paradoxically, in English it is the negative member of the category of person which is marked overtly by -s, whereas the first and second persons are left unmarked. Typologically, other constellations of markers are possible: (i) markers in all persons, e.g., Latin; (h) no markers, e.g., Chinese, Japanese; (iii) first person unmarked, second and third marked. This is a very unusual pattern which may be found in Old Norse: ek kalla "I call" vs. Ou kallar and harm kallar. However, in Modern Norwegian the personal suffix -r appears in all persons. The third person is distinguished from the first and second persons in many respects. First of all, the speaker and the addressee are necessarily present in any discourse; they coincide if someone talks to himself. While the speaker and the addressee are always given, other persons and things to which reference is made may be absent in both time and space from the situation of discourse. In grammatical terms, the category of the third person may combine with categories such as definiteness (definite vs. indefinite) and proximity (proximate vs. remote). On the other hand, personal pronouns of the first and second person are necessarily only definite and typically only proximate (unless we consider situations such as speaking on the telephone; here, of course, the remoteness is of non-linguistic referential character). Normally, pronouns of first and second person refer to human beings (unless we consider anthropomorphized ov personified animals and things in the world of fairy-tales) whereas pronouns of the third person may refer to both inanimate things and animate beings (human and non-human). See Figure 6.3. Let us exemplify these notions. We maybe inclined to think that the contrast such as English he/she and;/ (third person definite) versus somebody and something (third person indefinite) is universal. Turkish (and other Altaic languages) cannot grammaticalize the distinction of sex in the third person definite (o refers to both a male or female being) but it has the distinction of definiteness: o "he/she" versus biri "someone". The same situation obtains in other languages, e.g. in Plains Cree (Algonkian family): wiya "he/she" versus awiyak "someone". Furthermore, in Turkish the category of definiteness has to be marked obligatorily with personal pronouns by the suffix -i (personal pronouns are definite), whereas the marking with nouns depends on whether the noun is definite or indefinite: (13) sen-i gor-dii-m you-ACC see-PAST-lSG "1 saw you" ev al-di-m house buy-PAST-lSG "1 bought a house" INFLECTIONAL CAT EGOR1ES ASSOCIATED WITH VERBAL ELEMENTS 113 Person + definite + proximate 4 human 'second' addressee Fig. 6.3 The category of person ± definite + proximate + human 'third' spoken about ev-i al-di-m house-ACC buy-PAST-lSG "I bought the house" The category of proximity plays an important role with demonstrative pronouns (this vs. ilntt). This category is obviously determined in relation to the speaker (subject of discourse). This .iiiii here are proximate and that and there are remote with respect to the speaker. It may be observed that other languages have a more complex three-way system of proximity. Let us liixtapose Latin, Turkish, and English systems of demonstrative pronouns: (dialectical) this that yon Latin Turkish hie bu iste su ille o English this that Latin hie and Turkish bu (= "this") indicate proximity to the speaker, isle and fit (= "that") -moteness from the speaker, and ille and o (- "that", dialectical "yon", cf. German jener) udicate remoteness from both the speaker and the addressee. A more subtle distinction connected with three-way systems has to do with the notion of old vs. new information (in functional sentence perspective). In Turkish the pronoun bu is used when referring to old information, whereas the pronoun su will he used when referring to new information introduced into msciousness of listeners by the speaker. I lence, bu has to be used in the phrase bu tekltf'ihis )ioposal" when referring to the aforesaid proposal, the proposal which has just been mentioned (old information), whereas if the speaker wants to introduce a new proposal he has to say su teklif re u wi CO I" 114 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY "the following proposal", this proposal which he is about to mention (new information). Similarly, in Ancient Greek hode ho logos means "the following word" whereas hoiitos ho logos means "the aforesaid word". The three members of the system arc: hoitlos "this", hode "this, that" ekeinos "that". Traditionally, person has been regarded as a category of the verb since in flective languages it is marked by the personal suffixes. Since English and spoken French are poor in this respect we may profit from examining richer morphological systems such as those of Latin and Turkish. (15) Latin Turkish Sg 1 am-o sever-im "I love" 2 -äs -sin "you love" 3 -at -0 "he/she loves' Latin and Turkish (and many other flective languages) in contrast with English and French do not need analytic specification of the subject since they rely on morphological predication. To say "1 love you" in Latin it is enough to say te amo, or in Turkish sem severim. If we specify the subject by using the independent pronoun ego (or ben in Turkish) the meaning of ego te amo would be different. The speaker in this case emphasizes that HE (or SHE) loves the addressee in contrast with someone who does not love (or hates, etc.) the addressee. This contrast, of course, may only be implied and not realized linguistically. Thus we may translate ego te amo either by using sentential stress "7 love you" or by a so-called clefted sentence "1 am the one who loves you". The latter version might be preferable when the contrast is realized linguistically as in ego te. amo non frdter tuus "It is 1 who loves you not your brother!". Given these differences in discourse strategies of analytic (English-type) and synthetic (Latin-type) languages, we may wonder whether there are similar differences in what is called underlying or deep structure. It seems that these differences are here non-existent since in both cases we have to postulate an abstract pronominal element PRO (determined with respect to person and number) which is the subject of the verb, as shown in Figure 6.4. It might be tempting to talk about a pronominal element 'replacing' the subject (or NP) to obtain more universal diagrams than the usual S —► NP+VP, as shown in Figure 6.5. However, this would be a controversial procedure. First of all, pronouns as deictic elements depend on other elements in discourse. The first person is used by the speaker to refer to her/himself as a subject of discourse, the second person to refer to the hearer when spoken to about her/himself; on the other hand, the third person is used to refer to persons or things other than the speaker and hearer. The replacement of the subject by pronouns (linguistic pronominalization), in the realm of what is spoken about is qualitatively different from assuming the role of the subject in discourse. To exemplify this statement, it is a normal procedure to pronominalize in cases such as John came > he came but in / came the first person (subject of discourse) cannot be replaced by the noun John. Of course, we may say /, John Smith, INFLECTION Latin AL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS English S 115 Latin Fig. 6.4 1st Sg derived by pronominalization § English S Fig. 6.5 Deep PRO testify... in testimonies, oaths, etc. but this represents another type of discourse. Thus, it is preferable to consider personal pronouns and other deictic elements as basic, not derived from a deep NP. Let us remind ourselves that pronouns are universal and that we cannot imagine a language without pronouns, whereas there arc languages without adjectives or articles. 6.3.2 Tense The term tense goes back to the Latin word lempus meaning "time" (Latin tempus is calqued on Greek khronos). Since antiquity, this term has been used for labelling lime-relations which arc expressed by systematic grammatical contrasts such as Latin lauddbat "he/she praised" vs. laudai "he/she praises", or English (/) loved vs. love. Jakobson (1957) characterized tense as a deictic category, (a shifter) which puts the narrated event in reference to the speech event. Since the lime of the utterance is always 'now', the tense of the narrated action (or event or state) can be either 'beforc-now' (past time) or 'after-now' (future time), or 'simultaneous-with-now' (present lime). Hence, the typical three-way analysis of tense which prevails in many traditional grammars of various languages: Present, Past, Future. Even some linguists (e.g. Jespersen in his Philosophy of Grammar, 1929) believed this trichotomy to be representative of the 'natural' division of time into 'present', 'past' and 'future'. It is also noteworthy that this trichotomy, shown in (16) is reflected nicely in the system of adverbs of time: 'now', 'before' and 'after'. 116 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 117 (16) before PAST now PRESENT after -► FUTURE Thus Jesperscn talks about the past as before-now and the future as after-now. The primary distmchons of the past and future are then subdivided by means of a secondary application of the notions 'before' (past) and 'after' (future): (17) PAST 'before' 'after' pluperfect pre-present -O PRESENT I FUTURE 'before' 'after' future perfect future The result is a seven term notional tense-system, which is suitable for the analysis of the relative aspect (or anteriority) in terms of its Pluperfect, Pre-Prcsent and Future Perfect. 6.3.3 Aspect It is fundamental to distinguish between tense and aspect. Both are concerned with time (both are designators in Jakobson's terms) but in different ways. Whereas, as pointed out above, tense is a deictic category which relates the time of the action (or event or state) to the time of utterance which is 'now', aspect is concerned with representing different positions of the subject within Event Time. Put differently, aspect is concerned with the internal temporal constituency of the event (situation-internal time), whereas tense, as we saw above, allocates event within the cover of Universal Time. Using vertical lines to represent the initial and final limits of an event, we may discern five positions within Event Time: (18) AjB............ Prospective Inceptive .......C..........................D|E Progressive Perfective Retrospective Position A represents prospective aspect (I will write, Turkish yaz-acag-im); B represents inceptive (Russian ja vy-pju "I will drink (= empty the glass)" ); C represents progressive or imperfective (I am writing, Turkish yaz-iyor-um); D represents perfective (Russian ja vy-pil "I drank (~ emptied the glass)" or Greek 'aorist' e-lu-s-a "I solved"); and E represents retrospective (traditional perfect: I have written or Greek ge-graph-a ). B, C, D positions represent immanent aspects (interior to the event), while A and E represent transcendent aspects (exterior to the event). It is of interest to observe that the term aspect is a translation of the Russian word vid (from videt' "see, view") and it was used for the first time in the analysis of Russian and other Slavic languages: soversennyj = perfective and nesoversennyj vid = imperfective aspect. The terms I" i hi live and imperfective should not be confused with perfection and infectum, terms used by \in icnl grammarians for similar notions referring to completion of action or process. Thus the I .iliii verbal system may be analyzed along the following lines: (19) Aspect Tense Infectum Perfectum Present anw amavi Past amabam amaveram Future amabo amavero There are three binary contrasts in this paradigm: (i) aspectual contrast: perfect vs. nonperfect. Marking for the perfect is -v (there are other types of marking for the same category, most notably -s, and reduplication); (ii) temporal contrast: present time vs. non-present time. Marking for non-present is -h in the non-perfect forms and -er in the perfect forms; (iii) temporal contrast: experienced (past) time vs. non-experienced (future) time. This contrast operates only for non-present time. The morphological marking is less consistent and may be hest demonstrated for the Is' and Conjugation. Consider the paradigm of the 1" Conjugation: (20) Past Future Sg i am-ä-b-a-m am-ä-b-6 2 -ä-s -i-s 3 -a-t -i-t PI 1 -ä-mus -i-mus 2 -ä-tis -i-tis 3 -a-nt -u-nt In the majority of forms (2"d and 3rd Sg, Is' and 2nd PI) the contrast Past vs. Future is identifiable by the contrast -a vs. -i. These contrastive vowels occur immediately after the marker for non-present time in the system of the non-perfect aspect. Traditional grammars talk rather about Imperfect and Future endings -as vs. -ii-; however, it is obvious that these are analyzablc. I'hus the whole Latin system of aspect and tense in terms of its morphological markers may be represented as shown in Figure 6.6 (the final -t marks the 3rd Sg). Let us examine some simple examples for the values enumerated above. 118 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 119 non-Present -b Present -0 am-a-t Present non-Present Past -a am-a-b-a-t Future -/ am-ä-b-i-t Past -a am-ä-v-er-a-t Future am-a-v-er-i-t Fig. 6.6 Latin system of aspect and tense (21) i. Domum aedificat "he builds/is building (his) house" ii Domum aedificavit "he built/has built (his) house" The first sentence with the imperfective verb aedificat suggests an incomplete event (somebody's building activity takes place in the very moment of the narrator's utterance). It may be best translated into English by the progressive form "is building". The event is simply in progress and it will last for some time after someone's utterance has come to an end. On the other hand, the second sentence with the perfect aedificavit suggests a completed event at the time of the utterance (somebody's building activity went to its end before the narrator's utterance). It is usually said that this form (called traditionally perfect) covers simultaneously the perfective aspect and the present time reference; in semantic terms, that it relates the present state to the past event. To demonstrate this point, we may consider the following pair of English sentences: (22) i. I have lost (Perfect) my wallet ii. I lost (Preterit) my wallet The first sentence suggests that my wallet is still lost, whereas the second one with the simple past (preterit) may or may not (depending on the context). This stipulation makes the perfect a marked form, as was argued in 3.3. Ii is well-known that in English the present perfect may not be used with specification of .i nine. Thus it is impossible to say */ have seen thai film yesterday; on the other hand, the |m i ii-ct in / have recently learned that Bill is leaving is acceptable, although recently refers to .Him' point of time in the past. However, the English type of incompatibility of the perfect with ml\ tibials of past time is far from being universal. For instance, in Spanish the perfect may co-Mi i hi with adverbs of past time: Gustavo Ferrdn ha muerto (Perfect) oyer... se ha eslrellado ■ linn-lie en los monies de nieve "Gustavo Ferrari died yesterday ... he crashed last night on the .iniw covered mountains" (Stevenson, 1970:62). Similarly, it is possible to say in German Gestern habe ich vie! gearbeitet (Perfect), but it is impossible to say in English *1 have worked (Perfect) much yesterday. This restriction on the corn in rrence of temporal adverbs and the perfect should be further investigated in a variety of I.manages. The contrast present time vs. non-present time may be exemplified with the following I .11 iii sentences: (23) Domum suam aedificat "He/she builds/is building (his/her) house" Domum suam aedificabat "He/she built/was building (his/her) house" Domum suam aedificabit "He/she will build (his/her) house" There is no space in an introductory book on morphology to discuss the wide variety of tense-aspect systems found in different languages. Nevertheless, we may be interested in examining briefly a more complicated system than that of Latin. Contrasted with Latin, the system of Ancient Greek exhibits an additional form, called aorist, which may be analyzed as a perfective aspect. Let us use the verb luo "solve" as an example: (24) Imperfective Perfective Non-past lu-6 (Present) lu-s-o (Future) Past e-lu-on (Imperfect) e-lu-s-a (Aorist) Retrospective le-lu-k-a (Perfect) e-le-lu-k-en (Pluperfect) Compared with the Latin system of two aspects and three tenses, the Greek paradigm has to be analyzed as consisting of three aspects: the imperfective, perfective and retrospective; and two tenses: non-past and past. The perfect is formed by partial reduplication; the aorist by the suffix -s. Temporal contrast non-past vs. past is marked morphologically by the opposition 0- vs. augment e- (plus different personal endings). It is surprising to see the future listed under the perfective aspect but this may be justified morphologically, since both the aorist and future use the suffix -s (of course, with different personal endings). On semantic grounds one may observe that the perfective non-past event must necessarily refer to the future. Greek -s performs very 120 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY much the same task as the Russian prefixes used to perfectivize the non-past tense and thus ret'ei to the future; compare Greek grdp-s-d "I will write" with Russian ja na-pis-it. The traditional term aorist is taken from Greek aoristos (meaning "unbounded, unlimited, unqualified"). We may best understand the meaning of the aorist vis-a-vis that of the perfect and the imperfect. In Ancient Greek the aorist denoted a simple past occurrence of the event where the subject of Kvent Time is in position D (perfective aspect). The perfect, on the other hand, denoted past events resulting in the present state where the subject is in position E (retrospective aspect). For instance, the perfect pepoieke touto could be translated "he has (already) done this" (the past event with present relevance), whereas the aorist epoiese touto means simply "he did this". If we want to express aspectual qualifications such as progressivity orhabituahty we have to use the imperfeel epoiei touto "he was doing this" or "he used to do this". Traditional grammars maintain that the aorist narrates the event whereas the imperfect describes it; more importantly, both arc Immanent aspects, whereas the perfect in viewing the event externally is classified as a Transcendent aspect. This is shown in Figure 6.7. It is of interest to observe that the tense-aspect system of Modem Greek is essentially the same (in terms of oppositions) even if perfect and future forms were replaced by analytical formations (the perfect is nowadays formed by means of the auxiliary exo "have" and the future by 6a "will"). We may use the verbpedzo "play" as an example: (25) Imperfective Present pedz-o Past epedz-a Future 6a pedz-o Perfective pek-s-o (modal form) epek-s-a 0a pek-s-o Retrospective ex-o pek-si ix-a pek-si 0a ex-o pek-si 6.3.4 Mood Traditional grammar distinguishes three main classes of sentences: statements, questions and commands. In terms of their grammatical structure, these are referred to as declarative, interrogative and jussive sentences. The term command covers requests, entreaties, demands, as well as commands in the narrower sense. To avoid confusing these various senses of command some linguists employ the term mand (e.g. Lyons, 1977:745). In many languages the difference between mands and statements is realized in terms of the grammatical category of mood. For example, the 2nd Pers Sg imperative form of the Latin verb lauddre "praise" is laudd and the 2nd Pers Sg of the present indicative is laudas. It may be observed that the form of the 2nd Pers imperative is a bare stem (= root laud + thematic vowel -a) with no overt indication of person or tense. In contradistinction with English, Latin marks the person in the plural: laudate "praise!" (2nd Pers PI Imp) vs. lauddtis "you praise" (2nd Pers PI Ind). Other languages, e.g. .Ancient Greek, indicate the person in the plural but show no difference between the indicative and the imperative; (26) displays the pertinent forms of the verb leipein "to leave": INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS Greek aspects 121 external view \|H imperfective .......C....... (Imperfect) perfective retrospective .............I>l (Aorist) (Perfect) Fig. 6.7 Greek aspects (26) Ancient Greek Indicative and Imperative 2"< Sg 2nd PI Indicative leipeis Imperative leipe lcipete It is no coincidence that the imperative forms carry no overt indication of tense; the past is ruled out by the fact that is impossible to command someone to carry out some course of action in the past. The only tense distinctions that may be expressed in the imperative are those of more immediate and more remote futurity. The formal contrast of the present vs. future imperative is found, for instance in Latin (-0 vs. -to) and Hindi (-0 vs. -gd): (27) Cras petito, dabitur. Nunc abi. (Plautus) "Tomorrow ask, it will be given. Now go away." The present imperative of pelere "ask" would be pete; its future form petito indicates that the compliance with the command is not expected immediately but in the future (tomorrow!. Similarly, in Hindi the future imperative will be used for those actions that arc to be done after some lapse of time. Contrast the following two sentences: (28) Bil dijie. "Give [me] the bill (right away)!" Bil dijiega "Give [me] the bill (after a while)!' (Hindi) The „raC„l imp*™™ would be .tWed by an imp.iia« cuswk* «I» »«* <° "»» "* "2L ,bc r»,ure in,- - * -> W * " ' 122 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY As far as the aspectual contrasts are concerned, these are possible in the imperative but are not particularly common. As in the indicative, Ancient Greek allows for a three-way aspectual contrast: Present (= imperfective) - Aorist (= perfective) - Perfect (- retrospective); cf the forms of the verb leipein 'leave' in the 2nd Sg: (29) Aspectual Contrasts in the Ancient Greek Imperative Indicative Imperative Present leipeis leipe (imperfective) "you leave/are leaving" "leave/be leaving!" Aorist (perfective) elipes "you left" lipe "leave!" Perfect léloipas (retrospective) "you have left" leloipe lit. have left! In Russian we also find the usual aspectual contrast imperfective - perfective in the imperative: (30) Imperfective pij vodku "drink vodka" Perfective vypij vodku "empty (this glass of) vodka" (lit. have drunk) As far as the category of person is concerned, it is implicit in the notion of commanding that the command is addressed to the person who is expected to carry it out. In other words, the subject of an imperative sentence normally refers to the addressee. However, many languages possess a third-person imperative which is typically used in a more polite style, since the third-person imperative typically requires an intermediary to transmit a command. Examples of the third-person imperative are available from Ancient Greek and Sanskrit; their forms of the verb "to carry" are contrasted in (31): (31) 3rd Person Imperatives in Ancient Greek and Sanskrit Ancient Greek Sanskrit 2nd Pers Sg pher-e bhar-a "carry" 3rd Pers Sg pher-eto bhar-atu "may he carry" Another piece of evidence that the subject of a jussive sentence containing an imperative does not have to coincide with the addressee is supplied by the passive imperative. Thus in Sanskrit we have a choice of constructing the command in the active or in the passive voice, the latter typically in a more polite style. Contrast: INFLECTIONAL CATEGORIES ASS ASSOCIATED WITH VERBAL ELEMENTS 123 I W) Mahyam imam (Acc) dchi "Give me her!" Mahyam iyam (Nom) dryatam "May she be given to me!" Subjunctive sentences are another subset of jussive sentences. The subjunctive in the main i l.iusc is most typically used to express wish and the subjunctive in this function is called uptulive. In English we use the j'-less form in the 3rd Pers Sg and in French the subjunctive (w illuml que): (13) Long live the Queen! Vive la repnblique! I ised with que in the 3,d Pers, the subjunctive expresses a demand: (34) Qu'il eenve. "May he write!" I ised with the negative particle in the 3rd Pers, the subjunctive expresses a polite prohibition: (35) Qu'ils ne le fassent pas. "May they not do it!" In the Is' Pers the subjunctive may express indignation: (36) Que je viennc a cctle heure? "That I would come at this hour?" Used with the negative particle in the V Pers, the subjunctive expresses a weak negative assertion. Contrast the following minimal pair of sentences: (37) Je ne sais rien (strong negative assertion) "1 know nothing." Je ne sache rien (weak negative assertion) "I know nothing." The subjunctive is used in a variety of subordinate clauses, whose full treatment belongs to syntax. Thus in French we have to use the subjunctive if the main clause contains the verb "to wish": 124 AN INTRODUCTION TO THF STUDY OF MORPHOLOGY (38) Je veux que vous le fassiez. "I want you to do it." The subjunctive has to be used in a variety of subordinate clauses (final, consecutive, conditional, causal) after their specific conjunction: (39) It est content que je le lui aie dit. "He is satisfied that 1 have told him." In terms of morphology, Romance languages present full-fledged sub-systems of the subjunctive that are typically organized as their indicative counterparts. Thus in Latin we find the contrasts past - non-past and perfectum - infectum in the subjunctive. Contrast the non-modal (indicative) and the modal (subjunctive) forms of the verb lauddrc 'praise' in the 3rd Pers Sg: (40) Latin Modal Forms Indicative Infectum Present laud-at Past -abat Future -ábit Perfectum -ävit -äverat -äverit Subjunctive Present laud-et Past _arct -avent -ävissel posses rtrTs,mi,ar;here the paraiiei,sm is even m°re c°m^ - ^ possesses the tuture subjunct.ve. In contradistinction with Latin the retrospective forms are the forms of the verb trabajar "work" in the 3rd Sg: (41) Spanish Modal Forms Indicative Present Past Future lmperfective trabaj-a -aba Retrospective na trabajado habia habrá " Subjunctive Present trabaj-e ^ast -ára/ase Future -are haya hubiera/ese INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 125 I lie past forms of the subjunctive are used most typically in hypothetical judgements (i.e., iIhim- judgements which are qualified in terms of possibility). In Latin, for example, one may i millast the real wish laudet "may he praise" (the present subjunctive) or lauddverit "may he i' praised" (the perfect subjunctive) with the wish which is not realizable (irrealis): I la (42) SI me laudaret, amicus metis esset (Imperfect) "If he praised me, he would be my friend" SI me laudavisset, amicus meus fuisset (Pluperfect) "If only he had praised me, he would have been my friend" second sentence strongly implies that "he did not praise me". 11 i 5 Voice The term voice (vox) was originally used by Roman grammarians in two senses: (i) in the ■.ciise of 'sound' (translating the Greek phóně "sound"), hence the terms 'vowel' (from Latin \onus vocális) and 'voice' (the effect of the vibration of the vocal cords); (ii) and in the sense of I he 'form' of a word as opposed to its 'meaning' (in this sense, the term has disappeared from modern usage). The term has developed a third sense, deriving ultimately from (ii), in which it refers to the active and passive forms of the verb. The Greek term for voice as a category of the verb was diathesis 'state', 'disposition', 'condition'. The two extreme positions in the state of affairs expressed by the predicate are 'acting upon someone' (the active voice) and 'being acted upon by someone' (the passive voice). The middle voice can be thought of as being intermediate between the primary opposition of active and passive. It is found most typically in reflexive sentences where the use of the middle voice indicates that the results of the action affect the agent. We may contrast the active voice in / am washing the baby with the middle voice in the reflexive sentence I am washing myself (called also 'pronominal' voice). In some languages the middle voice may also be used in a transitive sentence with an object that is distinct from the agent but which typically belongs to the agent. Thus in Ancient Greek we would use the middle voice in loúomai "I am washing myself (vs. the active voice in loúb tó téknon "I am washing the baby") but also in loúomai tón khitóna "I am washing (my) shirt". Here the implication of the middle voice is that the action is being earned out by the agent for his/her own interest: in our case, it affects an object possessed by the agent. Some modern languages have a similar construction, e.g., in French we would use the pronominal voice in both je me. lave "I am washing myself and je me lave unc chemise "I am washing (myself) a shirt"; similarly in Czech we would say umývám se "1 am washing myself and umývám si košili "1 am washing myself a shirt" (se is the accusative form and si is the dative form of the reflexive pronoun which can be used in any person). It is of interest to observe that in many languages the verbs of saying and perception (verba dtcendl el sentiendt) occur in middle voice (formally identical with the passive). These verbs are different from typical transitive verbs, such as kill, hit, etc., in that the 126 AN INTRODUCTION TO THE STUDY OH MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 127 agent is affected by the action (the speaker normally hears himself through total feedback; the agent of perceiving is rather an undcrgoer of perception). Thus in Latin we find (he middle voice in loquor "1 speak", hortor "I admonish" (called verba deponentia); in Ancient Greek in akrodmai "I listen", thedmai "1 watch", aisthdnomai "I perceive", etc. Verbs indicating change of state occur also typically in middle voice, e.g. Latin morior "I die", Sanskrit mriye "1 die"; Latin nascitur "he is born", Sanskrit jdyate "he is born"; Greek gignetai "it becomes". In tenns of morphology, voice differences in the verb may be expressed analytically by the auxiliary and the passive participle of the verb) or synthetically (special endings different from the active ones). English realizes the passive voice analytically with the auxiliary "be"; in German the auxiliary is werden "become" and in Hindi jdnd "go". (43) Das Buch wird geschrieben "The book is (being) written" (German) Yah kitab likhi gal this book written+FEM gone+FEM "This book is written" (Hindi) Latin and other archaic Indo-European languages may be used to exemplify synthetic passive morphology. (44) lists the forms of the 3'd Pers Sg in all tenses and moods for the verb lauddrc "praise". The forms of infectum are synthetic, the forms of perfectum analytic (formed by the auxiliary esse "be" + passive participle). (44) Latin Passive Forms: Indicative Present Imperfect Future Perfect Pluperfect Future Perfect laudätur laudäbätur laudäbitur laudätus est laudätus erat laudätus erit "he is praised" "he was praised" "he will be praised" "he has been praised" "he had been praised" "he will have been praised" Latin Passive Forms: Subjunctive Present Imperfect Perfect Pluperfect laudětur laudäretur laudätus sit laudätus esset "may he be praised" "might he be praised" "may he have been praised" "might he have been praised" It may be said that Latin neutralizes the contrast of voice in its participial system in that the active participle is imperfective and the passive participle is retrospective. However, the contrast >.l unit can be realized with the verbs which occur in the middle voice called verba deponentia ,1. poncnt verbs' (lit. verbs which 'lay aside' certain forms). Here the passive participle has also ilir meaning of the active retrospective participle. Contrast the participial forms of the transitive Mib laudilre "praise" and those of the deponent verb loquor "speak": (45) Active Infectum laudäns "praising" Perfectum Infectum Perfectum loquens locülus "speaking" "having spoken" Passive laudätus "praised" locutus "spoken" I his anomaly in the participial system was solved during a later development of Romance languages when the auxiliary "have" + passive participle in the meaning of the active ii iiospeclive participle was introduced. Hence French developed the analytical participial ■ \piession ayanl lone "having praised" while in Latin we cannot say *habens lauddtuin. In functional perspective (cf. under 10.3), the use of the passive has to do with different pii'sentations of the state of affairs designated by the predication. In the case of the active voice 11 ii- subject coincides with the agent (Ag); in the passive voice the subject coincides with the patient or goal (Go). In cither case the subject function is interpreted as marking the entity which is taken as the primary vantage point for presenting the stale of affairs: (46) Mary (Ag Subj) kissed John (Go Obj) John (Go Subj) was kissed by Maiy (Ag) In the passive sentence our attention is drawn to the goal which becomes the topical element ol the sentence (for details see 10.3). It is important to realize that not all languages have (he passive voice (e.g. Chadic languages, many languages in New Guinea, Zapotcc, Yidiji); those which have the passive voice do not have to realize the agentive phrase, e.g., in Classical Arabic the above passive sentence would be literally translated as, John was kissed, kissed-him Mary; furthermore, the passive is avoided, especially in colloquial speech, even in languages which have ,i fully productive basic passive (cf. Keenan 1985:248). Thus the most natural way of saying "He was killed yesterday" in Russian would be "They killed him yesterday": (47) Včera egó ubili Yesterday him killed "They killed him yesterday" 128 INTRODUCTION TO THE STUDY OF MORPHOLi INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 129 Finally, it is worth mentioning that in many languages even intransitive verbs are passivizabk (e.g., in Turkish, Latin, Sanskrit). For instance, in Sanskrit the verb "to go" may be found in both impersonal and personal passive constructions: (48) Mayä grämam gamyate I+LNSTR village + ACC go+PASS+3SG Maya gramo gamyate I+fNSTR village I- NOM go+PASS+3SG Both versions mean "I an, going to the village". Similarly, ,n Lat, passive of "to go" as in (49). atm we may use the impersonal (49) Sic itur ad astra Thus go+3SG+PASS to stars "This is the way (to go) to the stars." RECOMMENDED READINGS i .mimic, Bernard. 1976. Aspect: An Introduction to the Study of Verbal Aspect. Cambridge: ( ainbridge University Press. 1985. Tense. Cambridge: Cambridge University Press, i i.uulwin, William W. 1894/1965. A Greek Grammar. London: Macmillan. I Irwson, John & Vit Bubenik. 1997. Tense and Aspect in Indo-European Languages: Theory, Typology, Diachrony. Amsterdam: Benjamins. 11ii lie, Walter H. 1975. Time, Aspect and the Verb, Quebec: Presses de l'Universite Laval. I.iknbson, Roman. 1957. Shifters, Verbal Categories and the Russian Verb. Cambridge, Mass.: Harvard University Press. Irspersen, Otto. 1929. The Philosophy of Grammar. London: Allen and Unwin. Kcenan, Edrward L. 1985. "Passive in the world's languages". Language Typology and Syntactic Description ed. by T. Shopen, 243-281. Cambridge: Cambridge University Press, kurylowicz, Jerzy. 1964. Inflectional Categories of Indo-European. Heidelberg: Carl Winter. I cwis, Geoffrey L. 1967. Turkish Grammar. Oxford: Clarendon Press. I yons, John. 1977. Semantics. Volumes 1 and 2. Cambridge: Cambridge University Press. Matthews, Peter H. 1972. Inflectional Morphology: A Theoretical Study Based on Aspects of Latin Verb Conjugation. Cambridge: Cambridge University Press. _. 1974. Morphology: An Introduction to the Theory of Word-Structure. Cambridge: Cambridge University Press. Mcillet, Antoine & Joseph Vendryes. 1948. Tratte de grammaire comparee des langues classiques. Paris: Champion. Mitchell, Terence F. 1962. Colloquial Arabic. London: The English Universities Press. Moreland, Floyd L. & Rita M. Fleischer. 1973. Latin: An Intensive Course. Berkeley: University of California Press. Palmer, Frank R. \ 965.A Linguistic Study of the English Verb. London: Longmans. Schwyzer, Eduard. 1959. Griechische Grammatik. München: Beck. Stevenson, C. H. 1970. The Spanish Language Today. London: Hutchinson. I esniere, Luden. 1959. Elements de syntaxe structurale. Paris: Klincksieck. Tzermias, Paul. 1969. Neugriechische Grammatik. Bern/München: Francke. 130 PRODUCTION TO TUH S 1'UDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 131 EXERCISES (2) (3) (4) (5) (6) Presem Imperfect Aorist Future Perfect Pluperfect 2. Using the following data, analyze and describe the morphological make-up of the system of tense and aspect in Hindi. Note: The progressive forms are built on the past participle rah a of the verb rahna "remain". I. Verb hona "to be" m? hu "I am" ml tha "I was" ml huga "I will be" II. Verb calnä "to go" mě čaltá hü mě čaltá thä me čal rahä hü ms čal rahä thä me calä mě calä hü ml calä thä mě calü mě calügä "I go" "1 wenl" - "I used to go" "I am going" "I was going" "I went" "I have gone" "I had gone" "I may go" '1 will go" Analyze the whole Sanskrit system of aspect and tense m terms of ,ts mo^hologica. markers. Use the following data: (1) Present karomi "I make" (2) Imperfect akaravam "I made" - "I was making" (3) Aonst akarsam "I made" - "I have made" (4) Future kansyami "I will make" ("i I Perfect {(>) Pluperfect (/) Conditional čakara ačakram akarisyam "I have made" "I had made" ''I would make" leipö "I leave" éleipon "I left" ~ "was leaving" (1) élipon "1 left" (•'-) leipsö "1 will leave" H) léloipa "I have left" eleloípěn "I had left" hi I Analyze the aspectual system of Modern Hebrew. Consider the following forms: )ber "he/she has spoken, had spoken, spoke" Establish three basic aspectual categories. Describe their morphology. Identify the morphemes marking tense, aspect and mood in Spanish in the following data. Be as formal as possible. (1) trabaja "he/she works" (2) trabajaba "worked" (3) trabajó "worked" ~ "has worked' (4) ha trabajado "has worked" (5) habia trabajado "had worked" (6) trabajará "will work" (?) habrá trabajado "will have worked" (8) trabajaría "would work" (9) habria trabajado "would have worked" (10) trabaje "may he/she work" (11) trabaja "work!" (). Identify the morphemes marking tense, aspect and mood in Italian, and attempt to hierarchize them in a tree diagram. Note: The auxiliary avere "have" has /;- in some persons. (1) am a "he/she loves" (2) ami "may he/she love" (3) ama "love!" (4) am era "will love" (5) amerebbe "would love" (6) amava "loved" (V) ha amato "has loved" (8) aveva amato "had loved" 132 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 133 (9) avra amato "will have loved" (10) avrebbe amato "would have loved" Analyze the Farsi (Modern Persian) system of aspect, tense, and mood in terms of its morphological markers'. Analyze the whole Russian system of aspect, tense and mood in terms of its morphological (1) porsad "he/she asks" markers. Use the following data: (2) miporsad "is asking" (3) beporsad "may he/she ask" (1) nesü "I carry" (4) porsid "asked" (2) nes "carried" (5) miporsid "was asking" (3) nes by "would carry" (')) porside ast "was asked" (4) prinesü "will have brought" (7) miporside ast "has been asking" (5) prines "have brought" (8) miporside büd "had been asking" (6) prines by "would have brought" (9) porslde bäsad "may he/she have asked" ~ "if only he/she (7) nosii "am carrying" (10) porslde büd "had asked" (8) nosil "was carrying" (11) xvähad porsid "will ask" (9) biidu nosit' "will be carrying" (12) xvähad porside büd "will have asked" (10) nosil by "would be carrying" (11) prinosu "am bringing" (a) Provide traditional labels for all the forms, e.g. (6) Perfect (12) prinosil "was bringing" (b) Attempt to hierarchize the markers identified in the forms above (13) biidu prinosit' "will be bringing" organized binarily. (14) prinosil by "would be bringing" (e) There are two forms which are difficult to accommodate in the tree. explain why. (Hint: Use the theory of markedness outlined in 3.3). Analyze the whole Lithuanian system of aspect, tense and mood in terms of its morphological markers. Use the following data: hem and I o. Using the Word and Paradigm model, analyze the Kurdish (dialect of Suleimaniye) system of tense, aspect, and mood in terms of its morphological markers. Use the following data: (1) dirbu (2) dirbau (3) dirbdavau (4) dirbsiu (5) dirbciau (6) esü dirbes (7) buvaü dirbes (8) büdavau dirbes (9) büsiu dirbes (10) büciau dirbes (11) buvaü bcdirbzjs (12) büdavau bedirbtjs (13) büsiu bedirbijs "I work" - "am working" "worked" "used to work" "will work" "would work" "have worked" "had worked" "had worked (at intervals)' "will have worked" "would have worked" "was working" "used to be working" "will be working" (1) akawirn (2) bikawim (3) käwtim (4) akawtim (5) bikawtimäya (6) kawtüwim (7) kawtibim (8) k aw t ilium (9) (bi)kawtibäm (10) (bi)kawtibämäya ~ (bi)kawläbämäya "I fall" "if I fall"-"let me fall" "fell" "kept on falling" "would fall" "have fallen" "if 1 should have fallen" "had fallen" "would have fallen" - "if I should have fallen" *"1 would have had fallen" Consider the following data which will help you to analyze the above forms: 134 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY akáwi(t) "you fall" kawtü "fallen" abím "I am" (bi)bim "let me be" bum "I was" büwim "I have been" Answer the following questions: (a) Attempt to hierarchize the identified markers for tense, aspect, and mood in a tree diagram organized binarily. (b) Provide 'traditional' labels for (1) - (9), e.g. (6) = Perfect. (c) Comment on the absence of bi in (7) and its optionality in (9) and (10). Hint: Use the theory of markedness outlined in 3.3. (d) What is 'unexpected' (non-prototypical) on the sequence of markers in (5) and (10)? 11. Analyze the Ancient Greek system of aspect, tense and voice in terms of its morphological markers. Attempt to hierarchize these markers in a tree diagram. Use the following data: (1) ágei "he/she leads" (2) égen "led" (3) áksei "will lead" (4) égagen "has led" ~ "led" (5) ékhe(n) "has led" (6) ékhein "had led" (7) ágetai "leads for himself ~ "is (being) led" (8) égeto "led for himself - "was (being) led" (9) áksetai "will lead for himself (10) égágeto "has led for himself - "led for himself (11) ekhtai "has led for himself ~ "has been led" (12) ékhto "had led for himself ~ "had been led" (13) ekhthe "has been led" - "was led" (14) akhthésetai "will be led" 12. Using the Word and Paradigm model analyze the following Modern Greek system of tense, aspect, and voice in terms of its morphological markers. Note: Read carefully B. Comrie (1976), Aspect (Chapter 4) before you start working on this assignment. (1) féri (2) vlépi (3) 0a pléni (4) 9a vlépsi "he/she carries" "sees" "will be washing "will see" INFLECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 135 (5) ékleve ((>) égrapse (Aorist) 17) éxi dísi (K) ixe grápsi t')) íxe klépsi (10) 0a éxi grápsi (11) 0a íxe disi (12) plénete (13) 8a skotónete (14) 0adiOi (15) (e)vlepótane (16) skotóOike (Aor) (17) éxipli0í (18) ixe skotoôí (19) 0aíxcdi0í "stole" "has written" - "wrote" "has dressed" "had written" "had stolen" "will have written" "would have dressed" "is washed" "will be being killed" "will be dressed" "was seen" "has been killed" - "was killed" "has been washed" "has been killed" "would have been dressed" Answer the following questions (a) (b) (d) Hierarchize the identified markers for tense, aspect and voice in a tree diagram organized binarily. Provide 'traditional' labels for all the identified verb forms. Comment on the 'leak' in the system of tense and aspect. Translate (20) (29) into Modern Greek: (20) "he/she will be carrying" (21) "has stolen" (22) "was dressed" (23) "has seen" - "saw" (24) "was washed" (25) "will be seen" "would have stolen" "would have been killed" "would have seen" "has been washed" ~ "was washed" (26) (27) (28) (29) , 3 Analyze the whole Latin system of aspect, tense and mood in terms of its morphological markers as systematically as you can. Use the following data: (1) ducit "he/she leads" 136 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY INFI ECTIONAL CATEGORIES ASSOCIATED WITH VERBAL ELEMENTS 137 (2) duxerit "he/she will have led" 1 lanslatc the following relative clauses into Turkish. Use the participles and observe the rules (3) dücebat "led" .il vowel harmony: (4) dücat "may lead" (5) dücet "will lead" (II) "the man who ought to die" (6) düceret "would/might lead" (12) "the woman who is coming now" (7, düxit "has led" (13) "those who will come" (8) düxerit "may have led" (14) "the men who came/have come" (9) düxerat "had led" (15) "the women who ought to be loved" (10) düxisset "would/might have led" Vocabulary: Analyze the whole Turkish system of aspect, tense and mood in terms of its morphological "love" markers as systematically as you can. Use the following data: 61- "die" sev- ol- "be" adam "man" (1) geliyorum "I am coming" gel- "come" kadm "woman" (2) gelecektim "would come" vaz- "write" -lar (plural suffix) (3) geliyordum "was coming" (4) gelmeliyim "ought to come" 1 (, ( omrie (1976:52) notes that the perfect is an aspect in a sense different from the represen- (5) gelirim "come" tation of the internal temporal constitution of a situation "since it tells us nothing directly (6) gelmeliydim "ought to have come" about the situation in itself, but rather relates some state to a preceding situation". His reason (7) gelirdim "used to come" for writing a chapter on the perfect is that "given the traditional terminology in which the (8) geldim "came" perfect is listed as an aspect, it seems most convenient to deal with the perfect in a book on (9) gelecegim "will come" aspect". (10) geldiydim "had come" Comment on this dilemma. Hint: Distinguish carefully between perfect (as detmed traditionally) and perfective (as used in modern linguistics). 15. Analyze the Turkish system of participial forms in terms of its morphological markers. Use the following data: (1, yazan adam "the man who writes" (2) yaziyor olan adamlar "the men who are writing now" (3) yazirlar "those who usually write" (4) yazmis olan adamlar "the men who wrote/have written" (5) yazacak olan adam "the man who is about to write" (6) yaziliyor olan mektuplar "the letters that are being written" (7) yazihr olan mektup "the letter that is usually written" (8) yazilmrs olan mektup "the letter that was/has been written' (9) yazilacak olan mektuplar "the letters that will be written" (10) yazilmalt olan mektup "the letter that ought to be written" 17. Matthews (1974:139) illustrates the concept of formative ambiguity by means of the present indicative and present subjunctive in Spanish: Indicative Subjunctive Conjugation 1 compro T buy' conipre compras 'you buy' compres compra 'he/she buys' compre Conjugation 2 como T eat' coma comes 'you eat' comas come 'he/she acts' coma Then he wonders: "What is the point ... in saying that 'SI IBJUNCTIVE' is an element in sequence which is located in its allomorphs e or a specifically? Obviously we CAN say 38 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY so if we must. But the traditional view seems more revealing. Mood is a category of words as wholes, which is identified by the oppositions of whole stems or word-forms in the individual paradigm." Explain, as best you can, these two viewpoints. Hint: Think of the two bas.c approaches to the study of morphology: the Item and Arrangement model vs. the Word and Paradigm model. CHAPTER SEVEN MORPHOSYNTACTIC PROPERTIES AND THEIR EXPONENT'S Secondary grammatical categories, such as gender and number (see Chapter Five), and I»1-1 son, number, tense and aspect (see Chapter Six) are frequently referred to as 'morphosyntactic' i .ilegories. Their individual terms (such as Masculine, Singular, Third Person, Past, Imperfective) .iic called morphosyntactic properties since they are properties of the word which play roles in both morphology and syntax. In the framework of the Word and Paradigm model the elements which identify morphosyntactic properties are called exponents. For instance, in Moroccan Arabic t-suf-u "you see" the pielix I- is an exponent of the 2nd Pers and the suffix -u is an exponent of the Plural. In Latin, on the other hand, exponents of Person and Number are fused in a suffix which is not analyzablc for these two properties: (1) Moroccan Arabic Latin "you see" t-suf vid-es "ye see" t-suf-u vid-etis Examination of a sufficient number of typologically divergent languages enabled linguists to establish five types of exponence: (i) cumulative (ii) fused (originally separate) (iii) extended (iv) agglutinative (non-cumulative) (v) overlapping 7.1 Cumulative versus Agglutinative Exponenence The best examples of cumulative exponence can be found in Ancient or conservative Indo-European languages (Latin, Russian). If we examine the nominal paradigm of Latin o-stems it will become obvious that there is no exponent which could be said to identify consistently Plural versus Singular. 140 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHOSYNT ACT'IC PROPERTIES AND THEIR EXPONENTS 141 Morphology Semantics Polysemy -Gen Sg (o-stems) -Norn PI (o-stcms) - Dat Sg (0-stems) Polymorphy -Ö (o-stems) --ae (a-stems)--T (0-stems) - -Dat Sg Fig. 7.1 Polysemy and polymorphy in Latin (2) Case and Number in Latin (o-stems) Nominative Accusative Genitive Dative Ablative Singular serv-us "slave" serv-um serv-i serv-ö serv-ö Plural serv-I sen/-os serv-örum serv-is serv-ís In other words, in Latin, Number is 'fused' with Case in the sense that the inflectional suffixes mark the lexical item for a particular case and a particular number simultaneously. A result of this situation is polysemy of individual suffixes (when the same suffix marks different combinations of case and number) and polymorphy of syntactic functions (when the combination of a particular case and a particular number is marked by different suffixes in different declensions). See Figure 7.1. The situation in Turkish is diametrically opposed in that Number is not fused with Case. Both number and case are marked by their own exponents and in all instances it is possible to establish the boundary between them. Unlike in Latin, Turkish nominal suffixes are always segmentable and constant for all nouns; while Latin has five patterns of declension, Turkish has only one. This type of exponence is called agglutinative (or non-cumulative). (3) Case and Number in Turkish Singular Plural Nominative ev "house" ev-ler Accusative ev-i ev-ler-i Genitive ev-in ev-lcr-in Dative ev-e ev-ler-e Locative ev-de ev-ler-de Ablative ev-den ev-Ier-den Similarly, in Turkish verbal paradigms, Number is not fused with Person whereas in Latin ii r, I .el us contrast the following verbal forms from Turkish and Latin; (4) Turkish Latin Sg 1 gör-üyor-um "1 see" vid-e-o 2 -sun -ě-s 3 -0 -e-t PI 1 -uz -6-mus 2 -sunuz -ě-tis 3 -lar -e-nt "I see" In Turkish we may identify the plural morpheme -uz (in the Is' and 2"d person) and -lar in the i"11'crson. Surprisingly, the lsl Pers PI is not the expected *gör-üyor-um-uz; rigidly agglutinative loims, however, obtain in possessive pronouns as shown in (5). (5) ev-im "my house" ev-in "thy house ev-im-iz "our house" ev-in-i/ "your house" A similar morphological analysis is simply impossible for Latin (cf. 5.1.2). As with all typological distinctions, of course, we are speaking of a continuum. The following data from Moroccan and Syrian Arabic may demonstrate that Arabic occupies an intermediate position between Latin and Turkish on the scale of cumulative <-♦ agglutinative exponence: (6) Sg Moroccan Arabic Syrian Arabic 1 n-šuf "I see" šuf "1 see" 2 m t-šuf t-šuf f t-šuf-i t-šuf-i 3m i-šuf y-šúf f t-šuf t-šuf 1 n-šuf-u n-šuf 2 t-šuf-u t-šuf-u 3 i-šuf-u y-šuf-u Pl In Moroccan Arabic it is possible to identify separately Person („- - L',t- - 2"d, i- 3ra) and Number (-0 = Sg, -u = PI), and we may conclude that wc are dealing with agglutinative exponence. On the other hand, in Syrian Arabic in the 1st Pers wc are dealing with cumulative exponence (0- =V' + Sg, n- - lbl + PI). 142 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHOSYN 1 AC TIC PROPERTIES AND THEIR EXPONENTS 143 (7) Moroccan Arabic (Agglutinative Exponence) n-suf-0 n-sul-u 1" Sg (8) Syrian Arabic (Cumulative Exponence) 0-suf f\ 1" Sg PI n-suf Is' PI However, in both Moroccan and Syrian Arabic, Gender is expressed identically by two different strategies: suffixation in the 2nd Pers Sg (-0 vs. -i = Masc vs. Fern) and prefixation in the 3rd Pers Sg (y- or /- vs. t- = Masc vs. Fern). Thus in both Moroccan and Syrian Arabic marking for morphosyntactic properties of Person and Gender is of an agglutinative character. One-observes, however, polysemy of the form tsuf- 2nd + Masc or 3,J + Fern; this is a consequence of the fact that prefixation is used primarily for marking Person. 7.2 Fused, Extended and Overlapping Exponence It is more difficult to distinguish cumulative from fused exponence. We saw in (4) that in Latin Person and Number arc identified cumulatively; in Latin, it is impossible lo analyze further the suffixes appearing after the thematic vowel. Now, let us examine the marking for these two properties in Spanish. Consider the following data: (9) Spanish PI 1 2 3 PI 1 2 3 vivimos "we live" vivis viven llamämos "we call" llamais Hainan In Spanish viv-is "ye live" the ending -is identifies the form as Present Indicative + 2nd Pers PI. However, examining the rest of the same paradigm and the paradigm of the Is' Conjugation (llamd-is "you/ye call") we would predict a form *vivi-is. The same form can be established on the basis of an Indicative - Subjunctive contrast (*viviis - vivdis) but this form simply does not 'surface' since its occurrence would violate phonotactic rules of Spanish. (Spanish allows for vocalic clusters such as uo and ao but not for homorganic *ii). Consequently, we may treat the form vivis as a form resulting from underlying vivl + is by a regular phonological process in •i|i.tiush. Thus it may be said that in this case the resulting fused exponence is underlyingly •UCliifiiiative. The corresponding subjunctive form vivdis "may ye live" displays agglutinative ■ upmience at both levels (underlying and surface). These matters are surveyed in Figure 7.2. Another type of exponence-relationship is that of extended exponence. It is customarily h In red to as double marking. Classical examples are supplied by German plural forms im i living simultaneous use of the process of umlaut and the sufilxation. Consider the following i »-muH plural formations: (10) Tag "day" Tag-e "days" Vater "father" Väter "fathers" Mann "man" Männ-er "men" Fuss "foot" Fiiss-e "feet" I luii plural is indicated cither by the suffix (Tag-e) or by the process of umlaut (Köter) or by both [Manner, Fiiss-e). The latter strategy (combining morphological process and suffixation) is in .i sense 'redundant' as the later state of affairs present in Dutch and English may indicate. Thus I iij'.lish relies only on umlaut in distinguishing plural counterparts of man and/oo/ (men and feet, icpectively); their plural suffixes which caused umlaut in the root were lost during the pre-Old English period. Another example of extended exponence is available from Ancient Greek which double-inarks its past verbal categories by the augment and secondary endings (cf. 6.3.3); see some n-presentative examples in (11): (11) leip-ö "I leave" e-leip-o-n "I was leaving" le-loip-a "I have left" e-le-loip-e-n "I have left" I lie secondary suffix carried more 'weight' and the augment could be left out (especially in I lomeric Greek). And finally, linguists recognize a fifth type of exponence, called overlapping. Strictly speaking, this is not a new type of exponence but rather an 'interplay' or 'interdigitation' of two 2nd Pers PI indicative vivis Indicative -i cf. vivdis Subjunctive -d 2nJ pcrs pi .is 2"d Pers PI -is Fig. 7.2 Fused exponence in Spanish 144 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY extended exponences. In Ancient Greek the categories of aspect, tense and voice pattern in tin: fashion. Let us examine the following verbal forms: (12) Active Present Imperfect Perfect Pluperfect lu-ei "he solves" e-lu-e(n) "he was solving" le-lu-k-e(n) "he has solved" e-lc-lu-k-ci(n) "he had solved" Mediopassive Perfect le-lu-t-ai "it has been solved" Pluperfect e-le-lu-t-o "it had been solved" The active pluperfect shows the overlap of markers for tense, aspect and voice in the following fashion: MORPHOSYNT ACT1C PROPERTIES AND THEIR EXPONENTS 145 RECOMMENDED READINGS It.i.'rll. Charles E. 1966. "Linguistic typology". Five Inaugural Lectures ed. by P. D. Strevens, "> 49. London: Oxford University Press, i nwi-ll, Mark W. 1964. A Reference Grammar of Syrian Arabic. Washington, D.C.: Georgetown I 'Diversity Press. Il.ii n-ll. Richard S. 1965. A Short Reference Grammar of Moroccan Arabic. Washington, DC: (ieorgetown University Press. I i wis, Geoffrey L. 1967. Turkish Grammar. Oxford: Clarendon Press. M.mhcws, Peter H. 1974. Morphology: An Introduction to the Theory of Word-Structure. < ambridge: Cambridge University Press. (Chapter 8). (13) Active Pluperfect (Ancient Greek) The mediopassive pluperfect shows the overlap of markers for tense, aspect and voice in the following fashion: In (13), the perfect is marked by the suffix -k and two processes (reduplication of the root and the shortening of the root-vowel lit ~> lu). The suffix -k also marks the active voice vs. mediopassive -/ in (14); here the perfect is only double-marked by the two processes of reduplication and vocalic shortening. As mentioned above the category of tense (past) is also double-marked at both extremities of the word resulting in the line crisscrossing the lines with aspectual markers. Consequently, this state of affairs may fittingly be described by the term overlapping exponence. 146 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHOSYNTACTIC PROPERTIES AND THEIR EXPONENTS 147 EXERCISES 1. Identify and exemplify the following types of cxponence: (a) cumulative (b) fused (c) extended (d) agglutinative (e) overlapping 2. It is claimed that Sanskrit inflectional endings show a considerable degree of fusion of grammatical meanings whereas those of agglutinating languages are typically composed of a sequence of morphemes, with each morpheme corresponding to one meaning. Discuss this statement using the following data: Sanskrit Finnish Sg Nom grham "house" talo "house" Gen grhasya talon Loc grhe talossa (lncssive) Abl grhat talolta PI Nom grhani talot Gen grhanam talojen Loc grhesu taloissa (Inessive) Abl grhebhyas taloilta 3. Marking for the verbal categories of person and number in Syrian Arabic differs from that of Moroccan Arabic. Describe and explain the following data in terms of cumulative and agglutinative exponence: PI Syrian Arabic Moroccan Arabic 1 suf "I sec" nsuf "1 see" 2 (M) tsuf tsuf (F) tsufi tsufi 3 (M) ysuf isuf (F) tsuf tsuf 1 nsuf nsufu 2 tsufu tsufu 3 ysufu isufu Past Sg 1 1 katabt "I wrote" ktebt "1 wrote' 2 (M) katabt ktebi (F) katabti ktebti 3 (M) katab kteb (F) katbet ketbet Analyze several grammatical forms of the five aspectual categories of Ancient Greek (cf. ,,3.3) in terms of then extended and overlapping exponence. You should consult Matthews (1974:148 149) before working on this question. Present Imperfect Aorist Perfect Pluperfect lsl Sg lu-6 e-lu-on e-lu-sa le-lu-ka e-le-lu-k-cn Is' PI lu-omen e-lu-omen e-lu-sa-men le-lii-ka-men c-le-lu-ke-men 3rJ pj lu-ousi e-lu-on e-lu-sa-n le-lii-ka-si e-le-lu-ke-san 1 anguages are frequently classified into structural types of isolating, agglutinative, (lectwc/mflectional (the latter subdivided into inflected externally and internally). (a) Define these three types. (h) Demonstrate that this classification is ultimately based on the disttnct.on between morpheme and sememe. MORPHEME AND ALLOMORPH 149 CHAPTER EIGHT MORPHEME AND ALLOMORPH 8.1 The Alternation of Allomorphs It was mentioned in Chapter Two that a particular morpheme is quite often represented noi by the same morph but by different morphs in different contexts. These alternate representations of a particular morpheme are called allomorphs. One of the important tasks of morphology is to account for these allomorphic alternations. For instance, the plural morpheme in English, which is homophonous with the possessive noun suffix or the verb suffix for the 3'd Pers Sg Indicative is regularly represented by the allomorphs /s/, Izl and Iszl: (1) Plural Iszl glasses I si cats Izl dogs Possessive glass's cat's dog's 3rd Pers Sg (he)sneezes (he) meets (he) feeds These three allomorphs occur in three mutually exclusive environments. If the morph representing the noun morpheme with which the plural morpheme is combined ends with (i) a (strident) alveolar or alveo-palatal sibilant fricative Isl, Izl, Isl, Izl or a (strident) affricate HI, l]l, the plural morpheme is represented by /oz/; (ii) a voiceless consonant other than the strident Is/, Isl, HI, the plural morpheme is represented by Isl; (iii) elsewhere the plural morpheme is represented by Izl. It may be observed that the orthographical conventions of English distinguish only two of these three allomorphs, with -s or -'s representing both Isl and izl, and -es or -'s representing iszl. The question arises how to represent the morpheme underlying these three allomorphs, or, in other words, what is the phonic substance of the underlying morpheme? Obviously, we have a choice between Isl and Izl. Those linguists who favor the latter alternant rely on the distributional account; if the contexts (ii) and (in) as listed above arc brought within the scope of a more systematic statement it will appear that Izl occurs in more environments than ,'s/. Since all vowels of English are voiced, the environment of Isl includes only three voiceless stops /p, t, kl and two voiceless fricatives Ifl, IQI; on the other hand, the environment of Izl includes three voiced stops lb/, Id, Igl, two voiced fricatives /v, 161, three nasals /ml, In/, /ij/ two liquids IV, Irl, five tense vowels III, Id, lal, lol, lal and three diphthongs, /aj7, /aw/, /oj7. The allomorph Izl is ilr.li ihulionally the major variant (to use the Prague School term) of the plural (and the |iuv,cssivc and the 3rd Pers Sg) morpheme and may thus be favored as most convenient for the |ilniiiiiic representation of the underlying morpheme. I'lie process description (in terms of Generative Phonology) of this allomorphic variation \muld run along these lines. We have to postulate an epenthesis rule which inserts schwa lol lu-iwecn a stem final sibilant and the suffixed Izl. Sibilants (alveolar and alveo-palatal fricatives .mil affricates in English) comprise the set of sounds which are f+strident +corona!]. The in i urrence of Isl after voiceless consonants would be attributed to assimilation of Izl to the ......mil voicelessness of the preceding consonant, and of course no conditioning would occur after Miiccd non-strident consonants. Schematically: (2) Epenthesis Devoicing /glees + zl a [glassgzj /kast + z.l [ka;tsj /dag + z.l |dagzj ll may be noted that if we had chosen the minor variant Isl of the plural morpheme as the underlying form we would have to posit a voicing rule. This rule would voice Isl after a voiced i nnsonant or vowel: (3) /gkes + si /kact t- si /dag + si Epenthesis o Voicing z - z [glasssz] Jkaets] l.dagzj It may be argued that z is a less common sound than .v (notice that it is possible for a language not to include any voiced obstruents in its phonemic inventory) but, on the other hand, to postulate the final devoicing rule for English seems to be very natural given the universal constraint that no voiced consonant follows a voiceless one in the same syllable coda. Another reason why {zf is preferable to {sf has to do with words ending in sonoranls (n and /) such has hens, sins, and falls, ells. If we chose (sf for the representation of the plural morpheme, the voicing rule would have to be made sensitive to other words which do not voice their final .s after n or /; hens Ihenzl but hence /hens/; sins IsmzJ but since /sins/; falls Ifalzl but false /fals/; ells !e\z/ hut else /els/. There is another familiar case of allomorphic alternation in English which is similar to the above. The allomorphs of the regular past tense and past participle are /od/, HI and idl: /ad/ in petted, padded HI kicked, ... 150 AN INTRODUCTION TO THH STUDY OF MORPHOI OGY MOKIMll M: AND ALLOMORPH 151 lál begged,... The environments for these three allomorphs can be stated as follows: (i) /3d/ after !\i, lál; (ii) III after voiceless consonants other than /t/; (iii) lál elsewhere. The allomorph represented by a voiced consonant only (in this case the voiced alveolar stop lál) proves again to be a major variant. The process description of this allomorphic alternation has to work with two phonological rules similar to those of the plural morpheme, namely, the rule inserting schwa (epenthesis rule) and the rule of devoicing: (4) /pet i d/ /kik + d/ /beg + d/ Epenthesis 3 Devoicing - t [petsd] [kikt] [begd] Wc may conclude that {z} and {d} should represent plural and past tense morphemes in English. 8.2 Morphological vs. Phonological Conditioning of Allomorphs The alternation found in the suffixes of the plural and the past tense in English has thus far been attributed to the phonological shape of the preceding nominal or verbal stem. Such alternations, which are explicable on purely phonetic grounds without reference to the notions of morphology, are said to be phonologically conditioned. However, the distribution of some morphemes cannot be accounted for phonologically. In such a case, it is inevitable to list the specific set of lexemes with which each irregular alternate occurs. When some morphemes arc distributed in this manner, we have to say that they are conditioned lexically. Familiar examples may be found in English. The plural morpheme beside showing three phonologically conditioned allomorphs Izl, Is! and Iszl displays also several other pluralizing morphemes which may be classified into ten groups: (i) oxen, children, brethren (ii) deer, sheep . . ., bass, pike . . .; quail, grouse . . . (iii) geese, teeth, feet, lice, mice, men, women (iv) data, media, memoranda'memorandums, curricula'curriculums (v) radii, fungi (vi) cherubim/cherubs, seraphim/seraphs (vii) criteria, phenomena (viii) formulae, larvae crises, theses (x) indices/indexes Ii is obviously true that the plural of ox or goose has the same kind of meaning as the plural nl cut; in all these instances we are dealing with more than one individual. However, on the |i)iiniological side, we are dealing with different morphemes expressing the same meaning, namely plurality. Consequently, the equation in (5) holds semantically but not morphologically in dial different morphemes represent the same semantic unit. (5) Polymorphy: Sememe - Morpheme, Morpheme, Morpheme, cat cats ox oxen goose geese I lie nouns in (i) do not add Iszi and Izl but Isnl; in addition, the last two change their root vowel and child also adds -/• before -an. The nouns in (ii) have a 0-suffix. It is notable that the words in litis group are the names of edible domesticated and game animals (fish and birds). Of course, ilicie arc similar words with a regular plural: pigs, goats, pheasants, ducks. But it is of mteresl In note that some have both forms, the forms with the 0-suffix appearing in the dialect of hunters: ,i farmer who has ducks on his pond may go out hunting duck. In such cases wc arc not dealing wilh plural forms but rather with collectives (see 5.2.2). The nouns in (iii) exhibit a vowel change (umlaut) of various types: lul ' III, /aw/ —> la)/, Ixl --»Izl. The nouns in (iv) replace the (Latin) singular suffix -urn by the plural suffix -a is/. The nouns in (v) replace the (Latin) singular suffix us by the plural suffix /aj/ and the nouns in (vi) keep the (Hebrew) plural suffix /im/, beside (lie regular plural in Izl. There are various terminological problems connected with this slate of aflairs. If wc use the term allomorph indiscriminately for all the pluralizing suffixes Is/, Izl, Iszi, 'on/, I0i, /si, Am/, it will become impossible to state what the phonic substance of the plural morpheme is, since only the first tliree arc phonologically related. Consequently, we should prefer to talk about six different pluralizing morphemes and to keep the term allomorph for the phonologically conditioned variants. This admittedly is a somewhat pedantic insistence on Icrminology since the whole problem is a marginal area of English grammar. What is essential after all is the fact that we may identify phonologically the regular pluralizing morpheme of English as jz( and the residuum has to be considered as exceptional (regardless of whether we decide to label Isn/ etc. morphemes or allomorphs). However, the situation is more complicated when we deal with languages where it is impossible to identify the regular pluralizing morpheme. This state of affairs is found in 152 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHEME AND ALLOMORPH Plural languages which subcategorize their nouns into various inflectional classes. For instance, it is impossible to talk about regular pluralizing morpheme in Latin without specifying the declension (and gender and case) of the noun, such as in the nominative case o-stcms take -fif the noun is masculine and -a if the noun is neuter. Similarly, Arabic uses three plural suffixes: -in, -e and -at and their distribution can be stated only with reference to rather complicated morphological phenomena. The situation in Syrian Arabic can be outlined along the following lines: (i) -in is used with nouns denoting male human beings: mS'allem "teacher" -+ mS'allmm and most occupational nouns of the pattern C|aC2C2aC,: nazzdr "carpenter" ♦ nazzdrin; (ii) -e is used with nouns ending in the suffix -zi or -i. xadarzi "green-grocer" > xadarziyye; hardmi "thief ► haramiyye; also with many occupational nouns of the pattern C,aC;C2aC3: wra/"moneychanger" —» sarrdfe; (iii) -at is used with feminine derivatives: xal "(maternal) uncle" ■-» xale "aunt" xdldt and with singulativcs (see 5.2.2): zdz "chicken(s) zdze "a chicken" -» zdzdt "some chickens". There are other subgroups which we may omit at this point but what is of interest is the fact that the suffix -at is used with most loanwords. This might indicate that this suffix is the most productive of the three pluralizing morphemes: (6) babor "steamship" —» baborat ?amiral "admiral" —» ?amiralat bebe "baby" -» bebiyat tren "train" > trenat Returning to our theoretical discussion, we cannot call these three pluralizing suffixes of Syrian Arabic allotnorphs since they are not phonologically related. For similar reasons as in English we may keep the term allomorph for a phonologically conditioned variant of the morpheme -e which obtains after pharyngeals or r (e.g., bahhdr "sailor" -» bahhara "sailors"). Thus in both English and Arabic we are dealing with polymorphy of the plural meaning (one-to-many relationships between semantics and morphology), cf. Figures 8.1 and 8.2. 8.3 Turkish Vowel Harmony We may be interested in examining less familiar examples of phonological conditioning of allomorphs than those of the English plural and past tense morphemes. We saw above that this particular alternation was ultimately reducible to phonological assimilatory processes. In Turkish, the alternations found in allomorphs of case, plural, and possessive suffixes are also Semantics Morphology Phonology /z/ Is/ Fig. 8.1 Polymorphy and allomorphy in English Semantics plurai Morphology Phonology Id /a/ Fig. 8.2 Polymorphy and allomorphy in Arabic Fig. 8.3 Turkish Vowels 154 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHEME AND AUTOMORPH 155 reducible to assimilatory processes. Here the alternations affect vowels and they are frequently referred to as vowel harmony. For the purposes of the following discussion, we have to classify Turkish vowels by three pairs of familiar features: front vs. back (;, e, ii, 6 vs. ;. a, u, o) high vs. low (i, ii, /, u vs. e, 6, a, o) and unrounded vs. rounded (i. e, i, a vs. ii, o, u, o). We may portray this three-dimensional system as a cube in Figure 8.3. The first set of data in (7) contains all the case forms of Turkish nouns. (8) (7) Sg PI "house" "room" "eye" "friend" Nom ev oda göz dost Acc evi odayi gözü dostu Gen evin odani gözün dostun Dat eve odaya göze dosta Loc evde odada gözde dostta Abl evden odadan gözden dosttan Nom evler odalar gözler dostlar Acc evleri odalan gözleri dostlan Gen evlerin odalann gözlerin dostlann Dat evlere odalara gözlere dostlara Loc evlerde odalarda gözlerde dostlarda Abl evlerden odalardan gözlerden dostlardan I Case | The order of morphemes in this data is as follows: ROOT I PI J Case. These morphemes are realized by various morphs: the plural suffix is realized by two morphs /ler/ - /lar/; the accusative (and genitive) suffix is realized by four morphs showing the alternation I'll ~ III ~ lul ~ lu.1; the dative (locative and ablative) suffix is realized by two morphs showing the alternation lei ~ /a/; furthermore, if the plural morpheme intervenes between the lexical root and the case even the accusative (and genitive) is realized by only two morphs I'll ~ III. So far, the conditioning for this distribution can be expressed along these lines: (i) the plural suffix or the case (Dat, Loc, Abl) contains the vowel Id if the preceding vowel is front (lei, 161); elsewhere, it is /a/; (ii) the case (Acc, Gen) contains: the vowel I'll if the preceding vowel is low front Id, the vowel III if the preceding vowel is low back (/a/); the vowel /ii/ if the preceding vowel is low front rounded 161; the vowel lul if the preceding vowel is low back rounded lol. We need additional data to complete our analysis. Below are listed words which contain high vowels (only three cases will be necessary, the rest of the paradigm is predictable): Sg Nom Acc Dat name isim ism i isme "forehead" ahn aim alma "measure" 0I9Ü ölcüyü ölcüye 'fear" ^orku ^orkuw ^orku\a I I use forms show that our previous tentative conclusions were corr^ lnllDi (1) the plural suffix or the case (Dat, Loc, Abl) contain the vow6| /g/ ^preceding vowel is front (I'll, Id, /Ü/, lol); furthermore, we may simplify °urstaterlifjtunder (ii); (ii) the case (Acc, Gen) contains: the vowel IM if the preceding vow(]j.front unrounded (Iii, Id); the vowel III if the preceding vowel is back unroun^ ^ :/)--•- notice *alm ► aim; the vowel lul if the preceding vowel is front round^ jjtlic vowel lul if the preceding vowel is back rounded lul, lol. Let us use additional data for elaborating on the sequence of morp^ ^jlliis allomorphic alternation. The possessive suffixes meaning "my" and "our" can bt these forms as '.liown below: (9) evim evvmin evimiz. evimizin cvlcrim evlerimin evlerimiz evlerimizin odam odamm odamtz odannzln odalarim odalanmtn odalanmiz odalanmizin "my house" "of my house" "our house" "of our house" "my houses" "of my houses" "our houses" "of our houses" "my room" "of my room" "our room" "of our room" "my rooms" "of my rooms" "our rooms" "of our rooms" 156 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHEME AND ALLOMORPH Root + PI Case Case Poss Fig. 8.4 The sequence of morphemes in Turkish nouns Case Root Case i/e Case i/a Case u/a Case ii/e Fig. 8.5 Turkish four-way vowel harmony (Other forms are predictable thus goztim "my eye", gdzlerim "my eyes" etc.; dostumuz "our friend", etc.). The sequence of morphemes seen in (9) can be visualized in Figure 8.4. The forms which do not have a plural suffix intervening between the lexical root and the case/possessive suffix exhibit the type of allomorphy shown in Figure 8.5. This type of allomorphy is frequently referred to as four-way vowel harmony. If the plural suffix intervenes only two-way vowel harmony results as shown in Figure 8.6. We may wish to formalize these conclusions. First, we have to express the eight vocalic phonemes of Turkish as 'bundles' of three distinctive features; this is shown in Figure 8.7. ^ ^ As we saw above, the suffixes undergoing the four-way vowel (Poss Pl2) [ ^cc j harmony have to agree with the preceding vowel in two features, namely backness and roundness (the feature of height is irrelevant, i.e., it does not matter if the preceding vowel is i or e, the vowel of the suffix has to be i etc.). Thus in (10) we write the four rules of agreement in two features. Root PI ler PI lar Case i/c Poss i Case i/a Poss I Case i/c Poss i-— Case i/e Case l/a Poss I-Case I/a Fig. 8.6 Turkish two-way vowel harmony +high +back -round +high -back -round +high +baek t round -high i-back +round -high -back +round (10) (i, u, i, u) V [+high] Fig. 8.7 Turkish vowels (i) -back I -round J (Ü) -back 1 +roundJ (i, e) V back round 1 C„ + C„ (ü, Ö) V -back 1 C„ + C„ i round J 158 AN INTRODUCTION TO THli STUDY OF MORPHOLOGY (I) r+back J 1-round i O, a) V +back J C„ + C0 -round J 00 +back ] +roundi (", o) V r+back i C„ + C„ I +round j V 'high] a back j ß round J a back l C0 + C0 1\i round j The suffixes undergoing the two-way vowel harmony (P,, / £ i j Loc >) have to agree with the (11) (e,a) V i high '-round I * (e) ' [-back] (i, e, ii, o) V [-back] C0 + C„ (a) -» [+back] These two rules may be collapsed into a single (i, a, u, o) V l+back] C0 F C0 (12) V i high ! -round J [a back] gle rule given below: [aback] C„ + Q - S'0" W,*"d bt — •"«' '"*»' ™*.»> .o .he rules of vowel h™„y MORPHEME AND At.LOMORPH 15') (11 Arabic or French words ending in clear / [1]: e.g. mahsul "produce" forms its accusative rnahsulii, not according to the rules of vowel harmony which would give: *mahsulu; (n) Arabic words ending in k: idrak "perception" forms its accusative idraki, instead of the predicted * idraki; (hi) Arabic monosyllabic words with an a followed by two consonants (the second of which is a front consonant): harp "war" forms its accusative harhi, not *harbi. I line are exceptions to the rules of vowel harmony even among native Turkish words. For iii-.i.iik e, the following simple words contain both back and front vowels: clma "apple", kardef Imillicr". Examples can also be found among compound words: bu "this" + giin "day" -» bugiin luday" (bugiin can be heard sometimes), ba$ "head" + miifettis "inspector" —► hafmiifettis "chief iir.pivlor". Many loanwords do not show any effects of vowel harmony: mikrop (•■ French) inn-robe", feribol (< English) "ferry boat"/>w£ti/>as (< Greek) "bishop". On the other hand, many loanwords have been turkicized by undergoing vowel harmony: Arabic mumkin "possible" > luikish mumkin > miimkiin; French epaulette > Turkish apolet (see Lewis 1967 for more • i.iinplcs). H.4 Morphonology The term morphonology was proposed 70 years ago by Trubetzkoy (1929). For Trubelzkoy morphonology was a particular section of linguistic descriptions (distinct from phonology dealing uilh the system of phonemes and distinct from morphology dealing with the system of morphemes) which studies the morphological utilization of phonological differences (1929:85). In use Trubetzkoy's example, in the Russian words ruká "hand" and ručnój "manual" the iillomorphs /ruk/ and /ruč/ represent one morpheme. These two allomorphs are held together on ilie phonological side by the regular alternation A ~ c" (regular means that there are more instances ■.uch as óko "eye" óčnyj "ocular", kulak "fist" kuláčnyj "having to do with the list") and they are linked to the same semantic unit (the perusal of similar numerous examples would reveal that we .ne dealing with a productive process of the forming of derived adjectives from nouns). According to Trubelzkoy /ruk/ and /ruč/ represent one morpheme "which exists in linguistic consciousness ... in the form ruk/č, where kJi, is a complex unit". Thus morphophonemes, in contrast to phonemes which are of strictly unitary nature, are complex units of two (or more) phonemes capable of alternating in one and the same morpheme. That l\U and It! are two distinct phonemes of Russian can be shown easily by minimal pairs such as kumá "godmother" vs. čumá "plague". Furthermore, it is not the phonetic context of the derivational suffix -no/ny that changes' the !kl of ruká into the li! of ručnój, since Ik! is retained in oknu "window". This of course is not to say that all of the alternations k ~ c" arc of a morphophoncmic nature; for example, we are dealing with true phonological conditioning in the verbal set of skakát' "to spring": skakál "he used to spring" vs. skáčeš "you spring," skácet "he springs" (herc k ~*č before front vowel <•). A problem, however, arises if we examine other inflectional forms of ruká; for instance, the 160 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY MORPHEME AND ALLOMORPH 161 Dat Sg is [ruk'ej which has to be phonemicized as /ruke/ - here the [k] of [ruka] alternates with [k'] of [ruk'e] in the same phonetic environment of the front vowel e (k ■■* k' I -e). This is why some linguists would prefer to consider even the above alternation of ruka ~ rucnoj as phonologically conditioned. This assumption, however, necessitates the introduction of very abstract underlying representations such as /ruk + in + oj/ from which the correct phonetic form [rucnoj] has to be derived by means of morphophonemic rules, bypassing the phonemic level of representation. Thus more recently the classical phonemics Ikl = [k] ~ [k'] ~ [c] and morphophonemics (morphonology) {k} = Ikl ~ Icl have been collapsed in Generative Phonology, which named Trubetzkoy's morphophoneme systematic phoneme. The latter is a building unit of underlying representations. The underlying representation contains only pari of the information about the pronunciation of the morpheme stored in the lexicon, and the other aspects of pronunciation are determined by phonological rules which apply to morphemes of the language. Thus the pronunciation of/ruk + in + oj/ is determined by at least two phonological rules: the rule which palatalizes and affricates k before the front vowel i and the rule which subsequently deletes / after it has triggered the palatalization and affrication: (13) palatalization/affrication deletion phonetic output ruk - in - oj c 0 [rucnoj] Needless to say, this type of analysis cannot be considered as truly synchronic analysis in terms of'here and now'. The alternation of k ~ c in ruka - rucnoj is only a legacy of diachrony which cannot be explained phonetically except by reference to conditions which have become obsolete for many centuries; the derivational suffix -in was a productive suffix in prehistoric times (cf. Greekphegos "beech" and its adjectival derivative pheginos) but its high front vowel was reduced to schwa and subsequently lost in Modern Russian. To use an example from a Romance language. Arc we dealing with purely phonogical conditioning in the case of Italian amico "friend" [amiko] whose root is pronounced [amic] before the plural suffix -/'? There are hundreds of words undergoing the same change: medico "doctor", medici; civico "urban", civici, etc. However, Italians have no difficulty in pronouncing k before the same plural form ;: stomaco "stomach", stomachi [stomaki], antico "ancient", antichi [antiki]. Since Ikl and Icl are two distinct phonemes of Italian, how do we explain this alternation? The usual explanation has to do with the accent pattern -— the words which are accented on their penultimate syllable keep their velar consonants (k, g) unchanged before the plural suffix, whereas those accented on the antepenult show the effects of palatalization (c, J). Consider the following data: (1-1) Penultimate Stress fuoco "fire" fuochi antico "ancient" antichi lago "lake" laghi albcrgo "hotel" alberghi Antepenultimate Stress medico "doctor" medici monaco "monk" monaci magnifico "magnificent" magnifici I here arc, however, several exceptions to these accentual rules. The words amico, nemico . nuiiy", porco "pig" and Greco "Greek" although accented on the penult palatalize the velar i uiisoiiant; on the other hand, stomaco "stomach" and cdrico "load", although accented on the iiiiicpenult, do not palatalize their velar consonant before Another typical morphophonemic alternation exists between a voiceless consonant and its vim cd counterpart. The familiar examples come from English which has the alternation between I and Nl in the forms knife/knives, wife/wives, leaf/leaves etc. Here the morphophoneme {f( is n'.ili/cd as Ifl in the ailomorph /najf/ occurring in the singular, and as Nl in the allomorph /najv/ mi furring in the plural. In German Rat "advice" and Rad "wheel" are both represented |ihuncmically as /ra:t/ (as is well-known, in German voiced obstruents are devoiced syllable-Imally). However, this analysis fails to account for the fact that /ra:t/ "advice" alternates in the pluial with /re:ts/, while /ra:t/ "wheel" alternates with /re:dor/. In the second word t is a nniiphophoneme {t} which is realized as It/ in the singular but as id/ in the plural. This analysis i'. open to objections if we want to pay due attention to the facts of morphology. If we consider i In- whole paradigm of Rad "wheel" we have to conclude that the form with a final HI is a minor variant in that it occurs only in the Nom/Acc Sg whereas in the remaining six forms exhibiting Millixes -es, -e, -er, -em the vowel of the suffix does not allow for the devoicing. Even better examples would come from Slavic languages which have longer nominal paradigms. For instance, to insist that the morphophonemic representation of Russian dub "oak" is jdup) I Kussian has the same rule of devoicing voiced obstruents syllable-finally as German) is strange (•iven the fact that the remaining ten members of the inflectional paradigm contain b before a vowel (dubd Gen Sg, dubii Dat Sg, dubom Instr Sg, dube Loc Sg, duby Norn PI, etc). Thus we may favor process-analysis which would derive a phonetic representation [dup] from the morphological representation {dub} by means of the rule of devoicing b —» p before word or morpheme boundary. The fruitfulness of this approach may be demonstrated by working out a solution of more complicated data such as the following from Latin. Latin verbs belonging to the I"1 conjugation show two allomorphs of the verbal root: one with the voiced obstruent Igl in the infinitive and another one with the voiceless /kl in the passive participle: (15) agere "act" actus legere "read" lectus frangere "break" fräetus pangere "fix" partus 162 INTRODUCTION TO THE STUDY OF MORPH O El MORPHEME AND ALLOMORPH 163 fingere "create" pingere "paint" pungerc "hit" cingere "gird" fictus pictus pünetus cínctus form 1 account for the ^1 fTthat the form w,th /8/18 a major " - —p. .o t i with /k/ by denying ,t from underlying {agtus! We will . , final consonant g in the cluster »Jth , / ^ deV01cin« the root- kt. Thus the process descnption of these forms can be elaborated a.ong these hues (16) ag-tus Cluster Simplification Devoicing ^ Compensatory lengthening ä äktus frang-tus fmg-tus 0 0 k k fraktus fiklus cing-tus clnktus RECOMMENDED READINGS Anderson, Stephen R. 1985. Phonology in the Twentieth Century. Chicago: University of ( hieago Press. (I Ii i< mi field, Leonard. 1935. Language. London: Allen and Unwin. (Revised edition.) (Chapter I ')• I Messier, Wolfgang U. 1985. Morphonology: The Dynamics of Derivation. Ann Arbor: Karoma Publishers. i ilcison, H. A. 1961. An Introduction to Descriptive Linguistic. New York: Holt, Rinehart and Winston. (Revised edition.) I Im'kctt, Charles F. 1958. A Course in Modern Linguistics. New York: The Macmillan Company (Chapters 32-35). Ilwn.in, Larry M. 1975. Phonology: Theory and Analysis. New York: Holt, Rinehart and Winston. (Chapter 3). I cwis, Geoffrey L. 1967. Turkish Grammar. Oxford: Clarendon Press. M,nlinet, André. 1965. "De la morphonologie". La linguistique 1.16-31. Ii nbetzkoy, Nikolai S. 1929. "Sur la morphonologie". Travaux du cercle linguistique de Prague. 1.85-88. 1934. "Das morphonologische System der russichen Sprache". Travaux du cercle linguistique de Prague. Volume 5, Part 2. 164 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY' MORPHEME AND ALLOMORPH 165 EXERCISES 1. Describe the allomorphy in Sanskrit verbal forms whose stem contains a nasal infix (i„ th, present and the imperfect): .„/// "to join". The root is seen in the passive participle: yuk "joined". Present Sg 1 2 3 yunäjmi "I join" yunäksi yunäkti Imperfect äyunajam äyunak äyunak PI 1 2 3 yunj mas yunktha yujijanti ayupjma äyurjkta ayupjan 2. Describe the allomorphy seen in the nouns whose stem is formed by -an(i) in Vedic Sanskrit. Monosyllabic nouns are to be taken as a basis for your predictions regarding the location ol accent. "king" "soul" "eating" "voice" Sg Nom räjä ätmä adän väk Ace räjänam ätmänam adantarn väcam Instr räjjtä ätmänä adatä väcä Dat räjjic ätmäne adate väce Gen rajjias ätmänas adatäs väcäs Loc räjni ätmäni adati väci PI Nom räjänas ätmänas adäntas väcas Ace räjjias ätmänas adatäs väcäs Instr räjabhis ätmäbhis adäbhis vägbhis Dat räjabhyas ätmäbhyas adädbhyas vägbhyäs Gen räjjiäm ätmänäm adatäm väcäm Loc räjasu ätmäsu adätsu väksu (a) Start by identifying the roots, the stem-forming elements and the suffixes. (b) Specify the distribution of allomorphs of individual stems. (c) Apply the IP model, i.e., specify the phonological and morphological conditioning which account for the shape of the allomorphs of the stems. \ i u-scribe and try to explain as best as you can morphophonemic variants of the root in the following sets of Hebrew nominal forms: (1) näßl "prophet" (.') nojJi?i "my prophet" (I> ii3(3l?8xem "your (PI) prophet" (4) nsßiVtm "prophets" (5) naßWexem "your (PI) prophets (1) däßär "word" (2) daßäri "my word" 0) daßarxem "your (PI) word" (4) daßärim "words" (S) dißrexem "your (PI) words' III. (1) melex "king" (2) malkt "my king" (3) malksxem "your (PI) king' (4) msläxim "kings" (5) malxexem "your kings" I Describe and try to explain as best as you can morphophonemic variation in the following set of Latin verbal forms: lsl Sg Present l5' Sg Perfect (1) verto "turn" veil! (2) fodiö "dig" fodi (3) lundö "pour" fudi (4) rumpö "break" rupi (5) findö "split" fidi (6) scindö "split" scidT (7) defendö "protect" defend! (8) prehendö "grasp" prehendi DERIVATIONAL MORPHOLOGY 167 CHAPTER NINE DERIVATIONAL MORPHOLOGY 9.1 Theory of Word Formation Word formation may be defined as that branch of linguistics which "studies the patterns on which a language forms new lexical units, i.e. words" (Marchand, 1969:2). According i«> Marchand, word formation is concerned only with composites or complex lexemes (derivatives and compounds) and not with simple one-morpheme words which are the subject matter ol lexology and morphology. For instance, simple nouns and verbs such as brain, bird and make can be studied morphologically (as shown in Chapters Five and Six), but not derivationally since these words are not composites. On the other hand, their derivatives such as brain-y (formed by suffixation) or re-make (formed by prefixation) or compounds such as bird-brain may be studied derivationally for their motivation, semantic restriction, etc. For reasons given under 4.1 we cannot consider word formation either as a part of inflectional morphology or as a part of the lexicon as suggested by generalive-lexicalists such as Chomsky (1970) and Aronoff (1976). The most obvious counter-argument to the latter hypothesis comes from polysynthetic languages (such as Inuktitut or Ainu) where lexicalists have to store whole sentences in the lexicon in order to account for 'sentence-words'. An interesting article was devoted to this problem by M. Halle (1973) who entertains the following model of the coexistence of morphology and lexology; see Figure 9.1. However, a closer scrutiny of Halle's article will reveal that Halle did not distinguish clearly between inflectional and derivational morphology (p. 7, "the rules of word formation generate the inflected forms"). This confusion is, of course, a result of handling both inflectional and derivational morphology by means of an oversized transformational component (or, in other words, of handling morphology syntactically). Hence, Halle's attempt to accommodate both inflectional and derivational morphology in the dictionary: (i) the dictionary contains only (and all) fully inflected forms of the language, (ii) the dictionary must be organized into derivational paradigms. Thus, according to Halle, the lexical entries for WRITE and PLAY are organized along these lines: (1) Inflected Forms WRITE write, writes, wrote, writing, written derivatives rewrite write, writes, wrote, writing, written " writer writers I i.i ul Morphemes output Fig. 9.1 Word formation according to Halle (1973) PLAY play, plays, played, playing derivatives replay play, plays, played, playing " player players (Ibviously, the first proposal, which de facto enshrines the grammar in the lexicon, would be lui'lily impractical in the case of heavily flective languages where the number of inflections may i mi mlo the hundreds (versus 5 inflections of English, 11 inflections of French). Furthermore, the liuulioning of the 'exception filter' (corresponding essentially to the lists of exceptions in ii.ulilional grammars) is surrounded by uncertainty. The second assumption that the dictionary musí he organized into derivational paradigms is nothing new. Dictionaries of Sanskrit and Aiahic were organized in this fashion centuries ago. We may examine the lexical entry for barad lie or become cold" in any traditional dictionary of Arabic (i.e. any dictionary which has not ln-cii organized alphabetically in imitation of dictionaries of European languages): (2) barad "be or become cold" bard "coldness" barad "hail" burud "coldness, frigidity" barrad "refrigerator" band "cold" barrad "make cold, cool" tabrid "cooling" mubarrid "cooling, refreshing" ?abrad "enter upon the cold season" tabarrad "refresh oneself ?ibtarad "become cold" (Verb, Class I) (Verbal Noun, Class I) (Verbal Noun, Class I) (Verba! Noun, Class I) (Occupational Noun, see 9.5) (Adjective = Participle, Class 1) (Verb, Class II - Causative) (Verbal Noun, Class II) (Participete, Class II) (Verb, Class IV) (Verb, Class V) (Verb, Class VIII) Actually the first information given will be a vowel in the stem of the imperfecta: 168 INTRODUCTION TO THE STUDY OF MORPHOLOGY DERIVATIONAL MORPHOLOGY 169 (3) barada 4- Perfective (ü) Vowel in the stem of the Imperfective bard Verbal Noun This, of course, is far from including 'only (and all) fully inflected forms' in the dictionary, since all these forms will have to appear in the grammar. Thus the grammar of Arabic — not its lexicon — contains the information on deictic categories of person (1, 2 and 3) and the category ol number (Sg, Dual and PI). All this seems to be rather too obvious, but let us recall the facl thai the morphology of tense and aspect is a grammatical (nol lexical) issue in a variety of languages However, the latter (actually the Aktionsart, i.e., lexical aspect) has to be handled in tin-dictionary in the case of Slavic languages. Since Semitic languages collapse aspect and tense, both will appear in the dictionary: (4) barada Perfective (-brud-) The stem of the Imperfective/Non-Past Similarly, the dictionary of Russian will have to list the perfective counterpart of the verb bit "to beat" since its lexical meaning has changed: u-bit' 'To kill". 9.2 Derivation versus Compounding A derivative (derived or complex lexeme) is a lexeme whose stem is formed from a simpler stem (derivational base) by some kind of morphological modification (most commonly affixation). For instance, the English suffix -ic derives denominal adjectives as in democrat * democratic. A compound, on the other hand, is a lexeme whose stem is formed by combining two or more stems (which may be separated by an interfix (cf. 2.3) as in huntsman). For instance, blackbird is a compound lexeme whose stem is formed by combining the adjective black and the noun bird. According to Marchand (1969:11) the coining of new words proceeds by way of "combining linguistic elements on the basis of a detenninant/dcterminatum relationship called syntagma". In terms of semantics, the determinatum represents that member of the composite lexeme which is modified or rather 'determined' by the determinant. For instance, in the composite lexeme steamboat the basic word boat underwent a semantic restriction or determination by the determinant steam. Marchand uses the same categories for the analysis of derivatives where the determinatum is the derivational suffix; in his words, derivation is "the transposition of a word n i i lie role of determinant in a syntagma where the determinatum is a dependent morpheme" i|> I i). Whereas in the compound steamboat the determinatum was a noun boat, in the derivative a, amer the determinatum is a derivational suffix -er. On the other hand, derivational prefixes Iiiim- to be classified as determinants; anti- in antifascist determines the determinatum fascist. Traditionally, the area of word formation was treated as consisting of derivation and i (impounding. The former was subclassified according to whether the derivational affix was 1'ielixcd or suffixed in prefixation and suffixation, shown in Figure 9.2. However, as was shown by Marchand (1969:11), it is possible to regroup this traditional ui lienia by subsuming prefixation and compounding under one heading of expansion. Marchand ilc lines the expansion as "a combination AB in which B is a free morpheme (word) and which i'. .malysable on the basis of the formula AB - B". This means that AB (black-bird, counier-■ iihuk) belongs to the same lexical class to which B (bird, attack) belongs. Put differently, . .impounds and prefixed words share the 'expanding' of the free determinatum: (3) Compound Determinant free morpheme black Determinatum free morpheme bird Derivative bound morpheme free morpheme prefixed word fore see suffixed word free morpheme king bound morpheme dom Consequently, we may wish to keep the temt derivative only to derivatives formed by •.iillixation (deriving by bound morphemes), as shown in Figure 9.3. This reasoning may be supported on semantic grounds. As recognized by traditional grammarians, an important difference between prefixes and suffixes lies in the fact that the former have a distinct meaning of their own (even if they are not used as independent words), whereas the latter do not. Of course, there are exceptions on both sides. The prefixes a- [a] and be- appear to have no distinct meaning of their own and to serve only as means of transferring a word from one lexical category in another. The former derives predicative adjectives from intransitive verbs (lie is asleep) and 11 to latter derives transitive verbs from nouns, adjectives and verbs (bespectacled, belittle, bemoan). On the other hand, the meaning of suffixes is usually best described grammatically; many suffixes convert one part of speech into another, e.g. kind —* kindness. It is rather exceptional to find a suffix which modifies the lexical meaning of its determinant; for instance, llic suffix -ish added to adjectives denoting color changes their lexical meaning from "X" into "lather X": blue -* bluish. 170 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY Word Formation Compounding black-bird Prefixation Suffixation counter-attack king-dom Fig. 9.2 Word formation Word Formation Derivation Suffixation king-dom Prefixation counter-attack Compounding black-bird Fig. 9.3 Word formation according to Marchand (1969) 9.3.1 Prefixation Prefixes may be defined as bound morphemes which are preposed to free (or bound) morphemes. As mentioned under 9.2, they function as determinants of the words (or bound stems) to which they are prefixed. For instance, the adjective natural may serve as a deler-minatum in various derivatives such as un-natural, super-natural and counter-natural. The bound stem -fer may serve as a determinatum in re-fer, de-fer, andpre-fer. Prefixes un-, super-, counler-are taken from the list of English prefixes which may be studied for their origm and productivity in various grammars in English. The list of productive English prefixes, taken from Zandvoor, (1966:291-298), is reproduced in (6). As is well known, almost all productive ('living') English prefixes are of non-Germanic origin, with the exception of a- (in asleep), be-, fore-, mis- and un-;for- as in forget and with- as in withhold are usually not included since these words are synchronically unanalyzable. It can be observed that negative a-, auto-, hyper-, and mal- combine only with non-Germanic words, DERIVATIONAL MORPHOLOGY 171 Id) English prefixes: a- [a] adrift, asleep, awash, a-flicker a- [ei] amoral, asexual ante- anteroom, antediluvian anti- antichrist, anti-aircraft arch- archbishop, arch-enemy auto- automobile, autobiography be- bespectacled, besmear bi- bilingual, bisexual co- co-operate, co-education countcr- counter-attack, counteract de- decode, defrost, dehumanize dis- dishonour, disagree en-, em- embed, endanger, enslave ex- ex-premier, ex-service man extra- extraordinary, extra-mural fore- foreground, foreword, foresee hyper- hyper-critical, hyper-sensitive in-, im-, il-, ir- inaudible, impolite, illegible, irreligious inter- international, interschool mal- maladjustment, malodorous mis- mislead, misconduct non- non-payment, non-existent post- post-war, post-reformation pre- predate, pre-war (cf. premature) pro- pro-German (cf. propel) re- rebuild, refuel, rebirth (cf remain, remote, recover) semi- semivowel, semicircle sub- submarine, subway super- supermarket, superstructure trans- transalpine, transplant ultra- ultra-violet, ultra-conservative un- unhappy, unkind, unrest, undress, unearth whereas the rest combines with both Germanic and non-Germanic words. However, this type of ■.ludy of derivational morphology belongs rather to diachrony. Synchronically, linguists arc interested in semantic restrictions on combinability of various prefixes with various lexical i lasses (nouns, verbs, adjectives, adverbs). Why is it, for instance, that we may combine the 172 AN INTRODUCTION TO THE STUDY OF MORPHOLi DERIVATIONAL MORPHOLOGY 173 prefix re- with both free stems and bound stems (such as -fer and -mil) but the prefix fore- on I \ with free stems? Furthermore, we may combine sub- with -mil but not with -fer. To facilitate the study of these restrictions linguists construct various derivational paradigms. The following paradigm of the latinate verbs is taken from Aronoff (1976:12) and it includes only verbs which are slressed on the stem such as: refer, excluding verbs stressed on the prefix such as: suffer. (In the system of Sound Pattern of English by N. Chomsky and M. Halle this class is marked phonologically by the presence of a special boundary symbolized as -): (7) Derivational Paradigm of Latinate Verbs X=fer X-mit X=sume X-cei ve X=duce refer remit resume receive reduce defer demit deceive deduce prefer presume infer induce confer commit consume conceive conduce transfer transmit submit subsume transduce admit assume adduce permit perceive Aronoff uses this data to demonstrate that neither the prefix nor the stem has any 'fixed meaning'. Obviously, this statement is only half-true since we may establish a 'basic' meaning of some of these prefixes and bound stems. It seems reasonable to assume that the basic meaning of trans- is "across, beyond, through" (transfer "move from X to Y, transmit" (make) pass on/along", transduce "draw across"). Similarly, we may assume that the basic meaning of -fer is "give, bring, send" (confer "give, grant", transfer "send across", defer "delay, postpone"). We could proceed along these lines with certain other words from the chart. However, this analysis would be impossible in other cases when the meaning is too abstract; prefer "like better" is obviously difficult to analyze in this vein. Thus we have to assume that in some cases the meaning of the verb is associated with the whole lexeme. There is nothing particularly disturbing on this solution and English offers a host of similar examples of morphological opacity. To exemplify this phenomenon with native words we may try to analyze compounds with a determinatum -berry such as cranberry, strawberry, raspberry, blackberry-, blueberry, gooseberry. It is no problem to identify determinants in these compounds but their semantic analysis is another problem. Cran- does not occur independently in Standard English or in other compounds and blackberries can be green or red. Does anybody connect geese, and gooseberries or straw and strawberries! We have to simply acknowledge that cran- and similar bound morphemes are semi-morphemic elements since the usual definition of morpheme as a .....iningfiil element (or more precisely the element connected with a particular sememe) does not Imlil I lowever, they are morphemes as distributional elements. 'it.' Suffixalion Suffixes may be defined as bound morphemes which are poslposed to free morphemes. As nii'iiiioned under 9.2, they function as determinala of simple or composite (i.e. compound or d>us alive) free morphemes: king-dom, color-blindness, dis-agree-menl. Their origin and |iin ability. I lie latter method can be subdivided as follows: (d) The suffix is added to a Latin stem which closely resembles a word that exists is English: scient- scientist (cf. science). The suffix is added to a Latin (or Greek) stem which has no adopted English equivalent: lingu- -* lingual, ehron- -* chronic (but see crony "an old chum"). (e) A more traditional distinction would be simply (l) derivation by means of native suffixes and (ii) ilenvation by means of foreign suffixes — the latter being divided into suffixation on either unlive or foreign bases. (It may be noted that both approaches combine synchrony and diachrony). Another subdivision of suffixation is based on the resulting grammatical category; thus we may distinguish suffixes deriving nouns, adjectives, verbs and adverbs. The list of the productive English suffixes shown in (8) was adopted from Zandvoort (1966:299 322). (8) (i) Suffixes deriving nouns: (a) Personal and concrete non-personal nouns -ee addressee, employee -eer mountaineer, profiteer -er hunter, writer, rooster, boiler -ess hostess, murderess, actress, lioness v INTRODUCTION TO THE STUDY OF MORPIIOLOG DERIVATIONAL MORPHOLOGY 175 -ist violinist, copyist, loyalist -en wooden, earthen -ite Sybarite, Wagnerite -ese Chinese, Viennese -sler gangster, trickster, songster -esque Dantesque, picturesque -fold twofold, manifold (b) Diminutives -ful beautiful, cheerful -ian Dickcnsian, Shavian, Canadian -el(te) kitchenette, owlet, islet -ic emphatic, phonetic, classic, historic -ic/y Annie, Johnny, piggie, doggie -ical classical, historical -kin catkin, lambkin -ing amusing, charming -let booklet, leaflet, ringlet, piglet -ish Danish, Jewish, girlish -ling duckling, fledgeling, weakling, gosling -ive attractive, instructive -less endless, countless (c) Abstract and collective nouns -like childlike, heartlike -ly lovely, manly, deadly -age mil(c)age, orphanage, drainage, percentage -ous dangerous, mountainous -al approval, arrival -some troublesome, toothsome -(i)ana Shakesperiana, Newfoundlandiana -th fourth, sixth -ance furtherance, utterance -ward backward, forward, inward, outward -ation starvation, sedimentation -y noisy, catchy, empty -cy accuracy, diplomacy -dom dukedom, freedom, kingdom (iii) Suffixes deriving verbs -hood childhood, neighbourhood -head godhead, maidenhead -en blacken, darken, worsen -ing bedding, matting -fy certify, satisfy -ism despotism, Calvinism, Americanism -ize civilize, organize -itis appendicitis -ity identity, visibility (iv) Suffixes deriving adverbs -merit shipment, deferment -ness drunkenness, foolishness -ly greatly, namely -(e)ry rivalry, chemistry, nursery -ways lengthways, sideways -ship friendship, scholarship -wise lengthwise, sidewise (ii) Suffixes deriving adjectives: -able (ible) -al -an -ed breakable, eatable convertible, discernible cultural, musical Indian, Lutheran landed, wooded, blue-eyed '».4 Compounding In discussing the status of compounds, linguists usually rely on three criteria: the underlying i unapt stress and spelling. However, all these three criteria are notoriously unreliable as the IH'iusal of various treatments of compounding may demonstrate. For instance, H. ICoziol, the iiuthor of the Erst monograph on English word formation, published in 1937 in German, claims iliai the criterion of a compound is the psychological unity of a combination. Obviously, even ••vuiaclic groups such as the Holy Roman Catholic Church may function as psychological units 176 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY DERIVATIONAL MORPHOLOGY 177 and thus it is extremely difficult, if not impossible, to establish a clear cut distinction between a compound and a syntactic group. Stress has been used as a criterion by Bloomfield (1935.228): "Accordingly, wherever we hear lesser or least stress upon a word which would always show high stress in a phrase, we describe it as a compound member: ice-cream /'ajs^krijm/ is a compound, but ice cream /'ajs 'krijm/ is a phrase, although there is no denotative difference of meaning". The criterion of stress was rejected by Jespersen (1942:8.12): "If we stuck to the criterion of stress, we should have to refuse the name of compound to a large group of two-linked phrases that are generally called so, such as headmaster or stone wall". The spelling is, of course, the worst criterion, since some compounds are hyphenated, others are not, and others are spelled with no separation between the constituents, e.g. gold-tail, stone wall, blackbird. Perhaps the two-stresseil syntactic groups (stone wall, paper bag, etc.) should be excluded from word formation. However, it is of interest to note that many such combinations have developed forestress (e.g. boyfriend, manservant). Furthermore, they may be classified as compounds in languages which indulge in compounding more than English does. Thus Steinmauer "stone wall" is classified as a compound in German but stone wall is rather a syntactic group in English. Compounds are usually studied according to their membership in the parts of speech (as given by their detenninatum): (i) compound nouns (steamboat, blackbird) (ii) compound adjectives (color-blind, heart-breaking) and (iii) compounds verbs (outbid, overflow, undertake). There are also compound pronouns (myself) adverbs (somewhere), prepositions (into). conjunctions (whenever) and interjections (heigh-ho). However, in this book a different type ot classification will be adopted — that which was elaborated centuries ago by Hindu grammarians: (i) Coordinate compounds (ii) Determinative compounds (these can be of two types — subordinate or descriptive) (iii) Possessive compounds (iv) Syntactic compounds 9.4.1 Coordinate Compounds Some coordinate compounds are additive. In modern languages this relationship obtains most typically in numerals; for instance, fourteen is "four" + "ten" (we may note that on the phonological side these numerals may be realized with double stress /fortin/ or single final stress /fortin/). Hindu grammarians applied the term dvandva to this type of compound. This tenn translates literally "two" + "two" but it means "pair" or "couple". This 'illogicality' is explained by the fact that in Rigvedic compounds of this type (nearly always names of deities) each member of the compound is formally dual. For instance, mitrd-vdrund means "Mitra and Varuna" (= twin deities) and not, as morphology suggests, *"two Mitras and two Varunas" (singular Mitrds, Rigvedic dual Mitra). Similarly mdtard-pitard means "mother and father" even if morphologically we are dealing with two duals. More 'logical' compounds of this type appear in later post-Vedic documents, for instance, indra-vdyu "Indra and Vaya" where the first member of the * i impound assumes the form of the stem and the second member takes the dual from (singular I i/ni.v, plural Vdyu). It is also notable that at this stage the first member of the compound loses it', stress (compare English /fortin/ or /fortin/). More complicated examples to analyze are |ilniiili/cd coordinate compounds. They arc formed when the pair of groups is to be denoted; for instance, afdvdyas means "(the flock of) goats and (the flock of) sheep". Tllogically', they show tin- stem-form in the first-member and the plural form in the second-member (the accent is on the linal syllable of the second member avis "sleep"). There are other types of relationships which may be found in coordinate compounds. For instance, English bitter-sweet means roughly "sweet v\ itli an admixture or aftertaste of bitterness" (OED). Here we are not dealing with a pair but with n mixture of two properties (one being predominant). Similarly, in religious terminology God-niaii (German Gotlmensch, French Homme-Dieu, all of them calqued on Greek thednlhrdpos) denotes someone who is both God and man, i.e., a mixture of two properties one being I in-dominant). Religious terminology of Hinduism abounds in terms such as Hari-Hara (Vishnu-Mnva), Ardhandrisvara (Hermaphrodite Lord) and Sury'a-Candra (sun-moon). The deity known as I lari-I lara was represented with Shiva characteristics for the right side of the body (trident in Ins hand, snakes on his arms, etc.) and with Vishnu characteristics for the left side of the body b nnch shell in his hands, necklace of flowers, crown on head, and half of the traditional V mark nl Vishnu on the forehead). Ardhanarishvara shows the male characteristics of Shiva on the right side and the female ones of Parvati on the left (according to an ancient legend Shiva and Parvati mice engaged in such a violent sexual intercourse that they merged into one androgynous being). As is well known, the Greek Hermaphrodite combines male and female sexual features in a different way (here the myth tells us that the son of Hermes and Aphrodite grew together with the iivniph Salmacis while bathing in her fountain). Obviously, the 'notional compounding' differs lioni culture to culture. To take an example from a totally different area, scientific terminology abounds in terms such as russula cyanoxantha, which is a "milk-mushroom" whose cap is both "dark-blue" (Greek kudneos) and "yellow" (Greek xanthos). The mixture of these two colors in tins case does not yield green; the cap of this mushroom is basically dark blue with yellow spots. Similar examples could be multiplied from any scientific terminology. More interesting examples come from languages which do not have derivational morphology and have to rely totally on compounding. For instance, Chinese forms its abstract nouns by i (impounding two adjectives of exactly opposite meaning (antonyms): (9) ta-hsiao ch'ang-tuan yuan-chin kuei-chicn literally "big" + "little" "long" + "short" "far" + "near" "dear" + "cheap" "size" "length" "distance" "price" The mental process illustrated by these Chinese examples is quite different from Indo-Inropean coordinate compounds quoted above. English bitter-sweet docs not mean "taste" but 178 AN INTRODUCTION TO THK STUDY OF MORPHOLOGY DERIVATION AI MORPHOLOGY 179 "sweet with bitter aftertaste". On the other hand, these Chinese examples must be analyzed figuratively, more specifically as examples of metonymy (usually defined as a semantic transfer based on temporal or spatial contiguity). This derivational type is extremely rare in Indo-European languages; nevertheless, some examples may be found in Hindi or Persian. For instance, in the latter language dmad-o-rafl means literally "(he) came-and-(he) left" and figuratively "traffic". Similarly, áb-o-havá "water-and-air" means simply "weather". 9.4.2 Determinative Compounds There are basically two types of determinative compounds. A noun may be determined by an adjective as in blackbird. Similarly, adjectives may be determined by the stem of a noun as in color-blind or they may be determined by another adjective as in icy-cold. In traditional grammars the first type is called subordinate (dependent) compound, the second type descriptive compound. It is possible to classify subordinate compounds according to the grammatical meaning which the determinant has at the level of the underlying sentence. For instance, the following three compound adjectives are quite different at the level of the underlying sentence: heart-break-ing, easy-going and man-made. The first one refers to the object of the sentence: grief broke his heart ~ • heart-breaking grief, the second one is based on an adverbial phrase, and the third one refers to the subject of the underlying sentence: man made the hut > man-made hut. There are different relationships in combinations such as color-blind or grass-green. The first one is paraphrasable as "blind with regard to color", or possibly "he does not sec colors" whereas the other one is based on the comparison "as green as grass is green". This type of analysis was elaborated centuries ago in Ancient India for Sanskrit by Pánini and his followers. Since Sanskrit is heavily flective they described subordinate compounds (called lat-purusa, lit. his man) by identifying the case function the determinant would have at the level of the underlying sentence. Thus in the subordinate compound go-ghna "cow-slaying" the determinant "cow" is a direct object (accusative in Sanskrit) underlyingly (compare English heart-breaking). In the compound adjective agni-dagdhá "burnt with fire", the determinant "fire" is an instrumental in Sanskrit. The underlying sentence has to be constructed in the mediopassive voice: agníná dahyáte: "he is burnt with fire" or "he burns himself with fire" (where agnind is morphologically an instrumental). In the compound adjective go-ja "produced from cows" the determinant "cow" is an ablative at the level of the underlying sentence: góbhyasjáyate "it is produced from cows" (where góbhyas is morphologically an ablative). The most common type of subordinate compounds in Sanskrit are the nouns determined by the genitive (in the broadest sense). The compound noun such as rája-putrá "king's son" is analyzable as a nominal phrase rájňas putrús where rájňas is morphologically the genitive of origin. In historical perspective, in all these cases we are dealing with syntactic groups (phrases) becoming compounds. In Vedic Sanskrit it is still possible to find compounds whose determinants retain the case ending: abhayam-karú "producing security" (with accusative, not *abhaya-kara), šúnesita "driven by dogs" (with instrumental šúná, Nom svá), divo-já "produced from heaven" (with ablative divas, Noiu dyü),jäs-pati "lord of a family" (with genitive jas, Nom jä) and divi-ydj "worshipping in lii'.iven" (with locative divi). I V'scnptive compounds with the adjeclive as a determinant (the type blackbird) are rare in Sanskrit: krsna-sakuni "raven" (lit. "black" + "bird"). This type of determinative compounds is u-iy common in Germanic languages. In English, here belong combinations with adjectives denoting color, bluebird, redfish, blackboard; dimension: longboat, broadside, shortcake; taste: wu-cthread, sourdough; genealogical compounds with the determinant meaning "standing in the .ceond degree of ancestry or descent": grandfather; ethnic names: Englishman. Irishman. Specific morphological problems in compounding appear in all flective languages. They have in do with the fact that many flective languages became analytic during their history and Muisequently it is hard to use the terminology of the case grammar in the analysis of compounds. 1 Ins problem is rather marginal in English which has few compounds of the type craftsmen where \ icpresents the old genitive ending. However, in modern English this type is not productive, 1e all formations such as huntsman, kingsman etc. were formed in previous centuries. But in ( n i man this problem is more serious. On the one hand, we find parts of subordinate compounds -.mil as Volkskunde "folklore" vs. Völkerkunde "ethnography" (Gottesdienst "divine service" vs. i iollerdienst "idolatry"; Landeskunde "areal studies" vs. Landerkunde "regional geography (of dillcrcnt countries)" with the opposition (e)s vs. "er. This might imply that speakers of (ii-mian still identify these suffixes on the determinant of the compound with the case endings id the genitive singular vs. the genitive plural. However, these examples are isolated and it is iiiiire common to find -s- in environments where we would postulate a genitive plural: for instance, we find Bischofsmütze "mitre" but also Bischofskonferenz "bishops' conference" (not "Hischöfekonferenz). On the other hand, (-—)/■- occurs in environments where we would postulate ,i genitive singular: Kindermörder "murderer of a child", Hühnerkeule "chicken leg" etc. furthermore, -s- can appear with feminine determinants while it never appears in the declension id the feminine nouns: Wohnungsinhaher "tenant" but the form in the genitival group has a l) suffix: ein Inhaber der Wohnung. For these reasons various grammars of Gennan handle this particular morpheme as a 'boundary marker' (Fugenzeichen) and they have to go into lengthy descriptions of its phonological shape: e, (e)n, er, (e)s, ens. In Section 2.3 we called this morpheme an intcrfix 'i 4.3 Possessive Compounds Possessive compounds are analyzed as combinations with a compound determinant and a /cio determinatum. These are cases where the relationship between the two members of the i iiuipound does not provide the essential meaning and an external element must be added. For inslance, a birdbrain is not a "bird's brain" but rather a "stupid person", i.e. "someone having a Midbrain", On the other hand, humpback can denote both "a back with a hump" or (by metonymy) "a person with a humped back". Hence, adjectival derivatives from possessive 11 impounds such as humpbacked, palefaced, etc. For this reason, these compounds arc called 180 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY DERIVATIONAL MORPHOLOGY 181 exocentric, as opposed to endocentric compounds including coordinate and determinative compounds. Another common term coming from Päninian grammar is bahuvrihi meaning literally "someone whose rice is plentiful" (yasya vrihir bahur asti). In Sanskrit adjectival possessive compounds were formed from both descriptive and subordinate compounds and this process was accompanied by a shift of accent from the final member of the compound to the first. For instance, a descriptive compound brhad-ašvá "great horse" could be transformed into an adjectival possessive compound by the shift of accent brhád-ašva "possessing great horses". Similarly, a subordinate compound rdja-putrá "king's son (= prince)" could be transformed into an adjectival possessive compound rdjá-putra "having kings as sons". The most common possessive compounds are those with numerals as a determinant found in all Indo-European languages: Sanskrit astd-pad "eight-footed", Greek oktöpous "octopus", Latin bipes "two-footed". The accentual shift can be found elsewhere, most notably in Ancient Greek, where the pairs such as the following can be found: patro-ktanos "patricide" vs. patró-ktonos "killed by the father"; thěro-tróphos "feeding the beasts" vs. thěró-trophos "nourished by the beasts". However, the relationship between them is rather active vs. passive at the level of the underlying sentence (kteinei ton patera "he kills the father" > patro-ktanos vs. hupo toú patrós kteínetai "he is killed by the father" —»patró-ktonos). Possessive compounds are fairly common in Sanskrit poetic discourse which abounds in opaque formations such as vrksá-keša "whose trees are like hair" or "tree-haired" = "mountain" or tapó-dhana "whose wealth is penance" = "ascetic". 9.4.4 Syntactic Compounds Any syntactic group or phrase may have a meaning that is not the same as the sum of the meanings of its constituents. Lexicahzed prepositional groups are extremely common; here belong examples such as lady in waiting, maid of honor, man in the street, good-for-nothing, cat-o '-nine-tails, etc. In many cases the process of their lexicalization, as analyzed by Jespersen (1942:8.83), was a lengthy one. This is shown especially by the uncertainty as to the place of the plural morpheme in Early Modern English and it may often be attached to the whole combination instead of the determinatum. The following forms for sons-in-law from King Lear may exemplify this point: sonne in lawes, sonnes in law, sons in laws (also hyphenated Sons-in-Laws). Another common type is represented by derivations from a verbal phrase such as looker-on, hanger-on, listener-in,passer-by. Their pluralization by 'infixed' -s shows that we are not dealing with a compound. Another common type is represented by additive phrases such as bread and butter, soap-and-water, deaf-and-dumb and, of course, by (archaic and dialectal) numerals: five and twenty. It is of interest to note that German uses exclusively the latter type in forming its numerals (fünfundzwanzig) whereas Standard English uses exclusively coordinate twenty-five. As mentioned in 4.2, Czech has both additive pětadvacet and coordinate dvacetpět. 'LS Noun Derivation in Arabic It is often said that the absence of compounding is one of the typical features of word loiinalion in Semitic languages. This statement will not stand the scrutiny and it is really possible in lind rare examples of compounding. Some of them are based on foreign models. Thus Arabic hni mufatlis "chief inspector" or bds-kdtib "chief clerk" etc. are calqued on Turkish compounds h,i\ "head" + bakan "minister" = "prime minister" (synchronically, bds may be taken as a prefix in Arabic). Modern Hebrew has formations which have to be analyzed as determinative i ompounds; for instance, ram "high" + kol "voice" - "microphone"; ra?i "appearance" + dak "thin" = "microscope"; kam ( men) are used. Even less natural plural formation process would be to remove thc singular suffix (e.g. in Syrian Arabic zalum-e "man" -> zolm "men"); in Natural Morphologists' terminology, the plural formation by umlaut is non-iconic, that by subtraction counter-iconic: (9) "man" Plural Latin (vir) vir-i English (man) men Arabic (zalam-e) zslm Marked + iconic + non-iconic + counter-iconic Another famous example of counter-iconicity is the genitive plural form of feminine nouns m Russian. This one is suffixless (markerless) in spite of being double-marked vis-á-vis its nominative singular counterpart. Contrast the following Russian forms with their Latin equivalents which are constructionally iconic: 198 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY THEORETICAL MODELS OF MORPHOLOGY 199 (10) Sg Norn Sg Gen PI Nom PI Gen Latin schol-a -ae -ae -arum Russian skol-a -y -y -0 Both languages treat the Gen Sg and Nom PI as marked categories (marked corresponds to markerful) but the Russian Gen PI is anomalous in being markerless in spite of being double-marked (for number and case as shown nicely by the disyllabic Latin suffix -drum). 10.4.2 Typology According to Dressier (1985) linguistic types mediate between universal principles and language-particular issues. Setting aside the difficulty of defining clearly the five morphological types (isolating, fusional, introflexive, agglutinative, polysynthetic) the Natural Morphologists are faced with several typological dilemmas. One of them is why the fusional type should exist at all given the optimality of one-to-one relationship between morphology and semantics. Dressier compares the Latin and Turkish ablative plural forms "from our islands" in this respect: (11) i. insul-is island-ABL/PL nostr-Is our-ABL/PL (Latin) ada-lar-imiz-dan island-PL-our-ABL (Turkish) The Turkish version in (ii) is perfectly diagrammatic in displaying a one-to-one relationship between the signifiers and signifieds (-lar - PL, -nniz - "our", -dan - ABL), whereas in Latin the ablative and plural are realized by a single polysemous suffix -Ts. In addition the same suffix -is encodes also the dative plural (vs. Turkish -lar-a = -PL-DAT), and two other functions with other lexical items (accusative plural with (-stems, civ-is "citizens", and 2nd Sg with verbs, aud-is "you hear"). Hence the legitimate question why the fusional type is so wide-spread (csp. in the Indo-European phylum of language) and diachronically so persistent. Dressier maintains that this is so by two other criteria of naturalness, namely, word-size and indexicality. As far as the average length of words is concerned, Dressier (1985) claims that it is between two and three syllables (notice that this figure coincides with the optimal size of a prosodic foot in phonology). This being so the Latin example in (1 Li) is 'more natural' than the Turkish one in (11 .ii). In general terms one can say that the cumulative and fusional exponence is more economical than the agglutinative which results typically in four- and five-syllable words. On the dimension of indexicality, Latin also scores better than Turkish in that its polysemous suffix -is locates precisely the root insul- in the sense that no other formative may intervene between these two. In Turkish, on the other hand, the ablative suffix -dan locates its nominal root very vaguely in that nlher formatives (such as the plural or the possessive suffix) could intervene (ada-lar-dan "from islands", ada-miz-dan "from our island"). 10.4.3 System-Dependence In addition to typology, there is another mediating factor between univcrsals and language-specific phenomena, namely system-dependent naturalness. In simple terms, two languages may differ as to which is the dominant pattern for each language. As an example we may mention ihe plural formation in Arabic and English with both languages exhibiting internal and external mllection (cf. 2.2.2 and 4.3): (12) Internal External Arabic Singular Plural RaJuL RiJäl. FaLLäH FaLI.äHIn English Singular Plural MaN MeN peasant peasants Hut as we found out, most Arabic nouns exhibit internal inflection, whereas nearly all English nouns exhibit external inflection in their plural formalion. External inflection therefore constitutes .i system-defining structural property of English; its internal counterpart, not being system congruent, is open to elimination (we know from the history of English that there were many more internal plurals, such as hoc "book" - bee "books", which were replaced by external ones). Il is the other way in Arabic where external plurals are vulnerable to erosion (e.g. in Gulf Arabic one forms plurals of occupational nouns on the pattern xabbdz "baker" - xababiz "bakers, samindk "fisherman" - samamic "fishermen", vs. external plurals of literary Arabic xabbdz-ln, siimmdk-in). 10.4.4 Paradigmatic structure One of the salient features of the languages with fusional and cumulative exponence is the existence of inflection classes (traditional conjugations and declensions). Their very existence is somewhat embarrassing for universal principles of Natural Morphology in that it implies lack of uniformity in morphological expressions. Not all the instances of polymorphy (such as English \ oi -en for Nom PI; Latin -i~or -is for Gen Sg) are rationalizable synchronically; but some are .ii least partly motivated by extramorphological factors (phonological, lexical or semantic). To slay with nouns, their referential animacy with its lexically and semanlically determined characteristics may be used to rationalize polymorphous realizations of Case-Number aggregates. I oi instance, in Slavic languages the accusative singular (of masculine nouns) may be realized bv either -a or -0, with -a being appropriate with animate nouns and -0 with inanimate nouns (in addition both -a and -0 are polysemous in that -a marks also the Gen Sg and -0 the Masc Sg), as shown in Figure 10.5. 200 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY THEORETICAL MODELS OF MORPHOL« 201 "studení" "table" učenik-a stol-0 GEN [animate] ACC [inanimate] NOM Fig. 10.5 Animacy and Case in Slavic languages Another concern of the Natural Morphologists is what constitutes the unmarked inflection class and its complementary relationship to other inflection classes. To use Wurzel's examples, German nouns ending in vowels (other than schwa) are pluralized differently, i.e. belong to different complementary classes: Auto "car", Autos; Cello "cello" Celli; Schema "schema", Schemata; Firma "firm", Firmen. The class with Auto is the unmarked one because of the existence of alternative plural forms Cellos, Schemas, Firmas. Another concern is the 'size' of paradigms, handled by what is called paradigm economy, and the existence of macroparadigms. For instance, we saw in Chapter Four (4.3) that it is possible to a certain degree to reduce the five (or six) Latin declensions to two macroparadigms (one for thematic nouns, I & II, and one for athematic nouns, III-V) if one operates with notions such as short and long thematic vowel, zero suffix and morphophonemic rules. In more general terms, one has to shift the emphasis from the Word and Paradigm (WP) model to the Item and Process (IP) model (cf. 2.5). 10.4.5 Morphological and Phonological Naturalness One of the major dilemmas of the Natural Morphologists are the intercomponential conflicts; the achievement of naturalness in phonology may obstruct the achievement of naturalness in morphology, and vice versa. This issue was studied by Dressier (1985) who placed the phonological, morphonological and morphological rules on the cline of naturalness. For instance, the morphonological rule producing an alternation electrifkj ~ electrifsjity is claimed to be more natural than the rule producing fusion as in conclu[d] ~ conclufzjn; or the 'weak' supplelion seen in child ~ children is more natural than its 'strong' counteipart in Glasgow ~ Glaswegian or Halifax ~ Haligonian. A famous intercomponential naturalness conflict is the Germanic umlaut which loses its phonological motivation and results in irregular (unnatural) morphology. For instance, in the following set of German verb forms umlaut correlates neither with number nor with person: (13) Sg 1 2 3 PI 1 2 3 geb- geb-en geb-ct geb-en Umlaut gib-st eib-t PI. 202 AN INTRODUCTION TO THE STUDY OE MORPHOLOGY RECOMMENDED READINGS Carstairs-McCarthy, Andrew. 1992. Current Morphology. London: Routledge. Dik, Simon C. 1980. Studies in Functional Grammar. London: Academic Press. _. 1989. The Theory of Functional Grammar. Dordrecht: Eoris. Doncgan, Patricia J. & David Stampe 1979. "The study of Natural Phonology". Current Approaches to Phonological Theory ed. by D. A. Dinsen, 126 1 73. Bloomington: Indiana University Press. Dressier. Wolfgang U. 1985. Morphonology: The Dynamics of Derivation. Ann Arbor: Karoma Publishers. -, W. Mayerthaler, O. PanagI, & W. U. Wurzel, eds. 1987. Leitmotifs in Natural Morphology. Amsterdam: Benjamins. Harris, James W. 1969. Spanish Phonology. Cambridge, Mass.: MIT Press. -- 1975. "Morphology in generative grammar: Vowel alternations in Spanish verb forms". El espanoly la linguistica generativo-transfornuicional ed. by .1. Guitart & J. Roy. Barcelona: Ediciones 62, Peninsula. Haegeman, Liliane. 1991. Introduction to Government & Binding Theory. Oxford: Blackwell. Hooper, Joan B. 1976. An Introduction to Natural Generative Phonology. New York: Academic Press. Hudson, R. A. 1976. Arguments for a Non-Transformationul Grammar. Chicago: University of Chicago Press. Itkonen, Esa. 1976. "The use and misuse of the principle of axiomatics in linguistics". Lingua 38.185-220. Keyser, Samuel J. & Paul M. Postal. 1976. Beginning English Grammar. New York: Harper & Row. Kiparsky, Paul. 1968. "Tense and mood in Indo-European syntax". Foundations of Language 4.30-57. Li, Charles N. 1976. Subject and Topic. New York: Academic Press. Lightfoot, David. 1975. Natural Logic and the Greek Moods. The Hague: Mouton. Spencer, Andrew. 1991. Morphological Tlieory: An Introduction to Word Structure in Generative Grammar. Oxford: Blackwell. Steinberg, Dany D. 1975. "Chomsky: From formalism to mentalism and psychological invalidity". Glossa 9.83-117. Stockwell, Robert P. 1977. Foundations of Syntactic Theory. Englewood Cliffs: Prentice-IIall. REFERENCES AND SELECT BIBLIOGRAPHY Anderson, John M. 1971. 77ie Grammar of Case. Cambridge: Cambridge University Press. Anderson, Stephen R. 1985. "Typological distinctions in word formation". Language Typology and Syntactic Description. Volume 3 ed. by T. Shopen, 3 56. Cambridge: Cambridge University Press. _. 1985. "Inflectional morphology". Language Typology and Syntactic Description. Volume 3 ed. by T. Shopen, 150-201. Cambridge: Cambridge University Press. __ 1985. Phonology- in the Twentieth Century. Chicago: University of Chicago Press. Applegate, Joseph R. 1958. An Outline of the Structure ofShilha. New York: American Council of Learned Societies. Aionoff, Mark. 1976. Word Formation in Generative Grammar. Cambridge, Mass.: MIT Press. _. 1993. Morphology by Itself. Stems and Inflexional Classes. Cambndgc, Mass.: MIT Press. Bach, Emmon W. & Robert T. Harms, ed. 1968. Umversals in Linguistic Theory. New York: Holt, Rinehart and Winston. Bauer, Laurie. 1988. Introducing Linguistic Morphology. Edinburgh: Edinburgh f Iniversity Press. Bazell, Charles E. 1966. "Linguistic typology". Five Inaugural Lectures ed. by P. D. Strevens, 29-49. London: Oxford University Press. Bloomfield, Leonard. 1935. Language. London: Allen and Unwin. (Revised edition.) Brekle, Herbert E. 1972. Semantik. Eine Einführung in die sprachwissenschaftliche Bedeutungslehre. München: Fink. ___ 1978. "Reflections on the conditions for the coining, use and understanding of nominal compounds". Proceedings of the XlTh ICL (Vienna, 1977). Innsbruck. Hybee, Joan. 1985. Morphology: A Study of the Relation Between Meaning and Form. Amsterdam: Benjamins. Campbell, A. 1959. Old English Grammar. Oxford: Clarendon Press. Carstairs-McCarthy, Andrew. 1992. Current Morphology. London: Routledge. ('homsky, Noam. 1970. "Remarks on nominalizations". Readings in English Transformational Grammar ed. by R. Jacobs & P. Rosenbaum, 184 229. Waltham, Mass.: Ginn. _& Maurice Halle. 1968. Sound Pattern of English. New York: Harper & Row. Comrie, Bernard. 1976. Aspect: An Introduction to the Study of Verbal Aspect. Cambridge: Cambridge University Press. __ 1985. Tense. Cambridge: Cambridge University Press. ( owell, Mark W. 1964. A Reference Grammar of Syrian Arabic. Washington, DC: Georgetown University Press. ('urme, George O. 1960. A Grammar of the German Language. New York: Frederick Ungar. 204 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY REFERENCES AND SELECT BIBLIOGRAPHY 205 Diakonoff, Igor M. 1988. Afrasian Languages. Moscow: Nauka. Dik, Simon C. 1980. Studies in Functional Grammar. London: Academic Press. -- 1989. The Theory of Functional Grammar. Dordrecht: Fons. Donegan, Patricia J. & David Slampe 1979. "The study of Natural Phonology". Current Approaches to Phonological Theory ed. by D.A. Dinsen, 126 -173. Bloomington: Indiana University Press. Dressier, Wolfgang U. 1978. "Elements of a polycentristic theory of word formation". Proceedings of the XIfh LCL (Vienna, 1977). Innsbruck. -- 1985. Morphonology: The Dynamics of Derivation. Ann Arbor: Karoma Publishers. ----, W. Mayerthaler, O. Panagl & W. U. Wurzel, eds. 1987. Leitmotifs in Natural Morphology. Amsterdam: Benjamins. Erben, Johannes. 1975. Einführung in die deutsche Wortbildungslehre. Berlin: Schmidt. Falkenstein, Adam. 1949. Grammatik der Sprache Gudeas von Lagaš. Roma: Pontificium Institutům Biblicum. Fillmore, Charles J. 1968. "The case for case". Bach & Harms 1968.1 88. Fleischer, W. 1975. Wortbildung der deutschen Gegenwartssprache. Tübingen: Niemeyer. Fodor, I. 1959. "The origin of grammatical gender". Lingua 7.1-41, 186-214. Fromkin, Victoria & Robert Rodman. 1974. An Introduction to Language. New York: Holt, Rinehart and Winston. Gleason, Henry A. 1961. An Introduction to Descriptive Linguistic. New York: Holt, Rinehart and Winston. (Revised edition.) Goodwin, William W. 1894/1965. A Greek Grammar. London: Macmillan. Greenberg, Joseph. 1966. Language Universals with Special Reference to Feature Hierarchies. The Hague: Mouton. Haegeman, Liliane. 1991. Introduction to Government & Binding Theory. Oxford: Blackwell. Halle, Maurice. 1973. "Prolegomena to a theory of word formation". Linguistic Inquiry IV.3-16. Halliday, Michael A. K. 1967. "Notes on transitivity and theme in English, Part I". Journal of Linguistics 3.3 7-81. Harrell, Richard S. 1965. A Short Reference Grammar of Moroccan Arabic. Washington, DC. Georgetown University Press. Harris, Alice C. & Lyle Campbell. 1995. Historical Syntax in Cross -Linguistic Perspective. Cambridge: Cambridge University Press. Harris, James W. 1969. Spanish Phonology. Cambridge, Mass.: MIT Press. __ 1975. "Morphology in generative grammar: Vowel alternations in Spanish verb forms". El espaňoly la linguistica generativo-transformacionul ed. by J. Guitart & J. Roy. Barcelona: Ediciones 62, Peninsula. Harris, Zelig S. 1951. Methods in Structural Linguistics. Chicago: University of Chicago Press. Havránek, Bohuslav & Alois Jedlička. 1963. Česká mluvnice. Praha: Státní pedagogické nakladatelství. Ilewson, John & Vit Bubeník. 1997. Tense and Aspect in Indo-European Languages: Theory, Typology, Diachrony. Amsterdam: Benjamins. Ilirtle, WalterH. 1975. Time, Aspect and the Verb. Quebec: Presses de 1'Universilé Laval. 1 Ijelmslev, Louis. 1935. La calégorie des cas. Etude, de grammaire generale. Ada Jullandica VU.l.xij 184and IX.2.VÜJ-78. I lockert, Charles F. 1954. "Two models of grammatical description". Word 10.210 231. ___ 1958. A Course in Modern Linguistics. New York: The Macmillan Company. I looper, Joan B. 1976. An Introduction to Natural Generative Phonology. New York: Academic Press. I ludson, R. A. 1976. Arguments for a Non-Transformational Grammar. Chicago: University of Chicago Press. Ilyman, Larry M. 1975. Phonology: Theory and Analysis. New York: Holt, Rinehart and Winston. Ilkonen, Esa. 1976. "The use and misuse of the principle of axiomatics in linguistics". Lingua 38.185-220. Jakobson, Roman. 1936. "Beitrag zur allgemeinen Kasuslehre". Selected Writings 11, 23 71. The Hague: Mouton. __ 1957. Shifters, Verbal Categories and the Russian Verb. Cambridge, Mass.: Harvard University Press. Jensen, J. 1990. Morphology. Amsterdam: Benjamins. Jespersen, Otto. 1929. The Philosophy of Grammar. London: Allen and Unwin. 1942. A Modern English Grammar on Historical Principles. Part VI. Morphology. London: Allen and Unwin. Joly, André. 1975. "Toward a theory of gender in Modern English". Studies in English Grammar ed. by André Joly, 229 287. Lille: Presses de FUniversite de Lille. Katz, Jerrold J. 1966. The Philosophy of Language. New York: Harper and Row. Keenan, Edrward L. 1985. "Passive in the world's languages". Language Typology and Syntactic Description ed. by T. Shopen, 243-281. Cambridge: Cambridge University Press. Keyser, Samuel J. & Paul M. Postal. 1976. Beginning English Grammar. New York: Harper & Row. Kiparsky, Paul. 1968. "Tense and mood in Indo-European syntax". Foundations of Language 4.30-57. Klavans, Judy. 1985. "The independence of syntax and phonology in cliticization". Language 61.95-120. Koziol, Herbert. 1937. Handbuch der Englischen Wortbildungslehre. Heidelberg. Krámský, Jiří. 1969. The Word as a Linguistic Unit. The Hague: Mouton. Kuryiowicz, Jerzy. 1964. Inflectional Categories of Indo-European. Heidelberg: Carl Winter. I.adefoged, Peter. 1975. A Course in Phonetics. New York: Harcourt Brace Jovanovich. 206 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY REFERENCES AND SELECT BIBLIOGRAPHY 207 Lees, Robert B. 1963. The Grammar of English Nominalizations. Bloomington/The Hague: Indiana University Press. (2nd Printing.) Lewis, Geoffrey L. 1967. Turkish Grammar. Oxford: Clarendon Press. Lewis, M. B. 1968. Malay. London: The English Universities Press. Li, Charles N. 1976. Subject and Topic. New York: Academic Press. Lightfool, David. 1975. Natural Logic and the Greek Moods. The Hague: Mouton. Lipka, Leonhard 1975. "Prolegomena to 'Prolegomena' to a theory of word formation". The Transformational-Generative Paradigm and Modern Linguistic Theory ed. by E. F. K. Körner. Amsterdam: Benjamins. Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. __ 1977. Semantics. Volumes 1 and 2. Cambridge: Cambridge University Press. Macdonell, Arthur A. 1916. A Vedic Grammar for Students. Oxford: Oxford University Press. Marchand, Hans. 1969. The Categories and Types of Present-Day English Word Formation. 2nd Edition. Munich: Beck. Martinet, André. 1965. "De la morphonologic". La linguistique 1.16-31. Matthews, Peter FT. 1972. Inflectional Morphology: A Theoretical Study Based on Aspects of Latin Verb Conjugation. Cambridge: Cambridge University Press. __1974. Morphology: An Introduction to the Theory of Word-Structure. Cambridge: Cambridge University Press. Meillet, Antoine & Joseph Vendryes. 1948. Traité de grammaire comparée des langues classiques. Paris'. Champion. Mitchell. Terence F. 1962. Colloquial Arabic. London: The English Universities Press. Moreland, Floyd L. & Rita M. Fleischer. 1973. Latin: An Intensive Course. Berkeley: University of California Press. Nida, Eugene A. 1949. Morphology: A Descriptive Analysis of Words. 2nd Edition. Ann Arbor, Mich.: University of Michigan Press. Ogden, Charles K. & I. A. Richards. 1946. The Meaning of Meaning. 8lh Edition. London: Routledge & Kegan Paul. (First edition 1923.) Palmer, Frank R. 1965. /Í Linguistic Study of the English Verb. London: Longmans. Peirce, Charles S. 1955. Philosophical Writings of Peirce ed. by Justus Buchler. New York: Dover. Robins, Robert H. 1959. "In defense of WP". Transactions of the Philological Society 57.116-144. __1967. A Short History of Linguistics. London: Longmans. Rohrer, Christopher. 1974. "Some problems of word formation". Linguistische Arbeiten 14. Tubingen: Niemeyer. Rosen, Haiim B. 1977. Contemporaiy Hebrew. The Hague: Mouton. de Saussure, Ferdinand. 1955. Cours de linguistique generale. 5,h Edition. Paris: Payot. (English translation by Wade Baskin, Course in General Linguistics. New York: Philosophical Library, 1959.) Schachter, Paul. 1985. "Parts-of-speech systems". Language Typology and Syntactic Description. Volume 1 ed. by T. Shopen, 3-61. Cambridge: Cambridge University Press. Schwyzer, Eduard. 1959. Griechische Grammatik. München: Beck. Spencer, Andrew. 1991. Morphological Theory: An Introduction to Word Structure in Generative Grammar. Oxford: Blackwell. Steinberg, Dairy D. 1975. "Chomsky: From formalism to mentalism and psychological invalidity". Glossa 9.83-117. Stevenson, C. H. 1970. The Spanish Language Today. London: Hutchinson. Stockwell, Robert P. 1977. Foundations of Syntactic Theoiy. Englewood Cliffs: Prentice-Hall, lesnicre, Lucicn. 1959. Elements de syntaxc structurale. Paris: Klincksieck. I rubetzkoy, Nikolai S. 1929. "Sur la morphologie". Travaux du cercle linguistique de Prague. 1.85-85. __ 1934. "Das morphonologische System der russichen Sprache". Travaux du cercle linguistique de Prague. Volume 5, Part 2. __1939. Grundzüge der Phonologic Travaux du cercle linguistique de Prague 7. (English translation by C. A. M. Baltaxe, Berkeley/Los Angeles: University of California Press, 1969). Tzcrmias, Paul. 1969. Neugriechische Grammatik. Bern/München: Francke. Wardhaugh, Ronald. 1972. Introduction to Linguistics. New York: McGraw-Hill. Wright, W. 1896-1898. A Grammar of the Arabic Language. Volumes 1 and 2. Cambridge: Cambridge University Press, /.andvoort, Reinard W. 1966. A Handbook of English Grammar. Englewood Cliffs: Longmans. INDEX OF LANGUAGES 209 INDEX OF LANGUAGES Arabic (includes Classical Arabic) adjeclives, 6 allography, 16 consonantal root, 6 derivational paradigm, 167 169 dative, 6 grammatical word, 22 infixation, 21 internal inflection, 18, 26 markedness, 43 number, 85 87, 199 prefixation, 21 pronominal forms, 74 semi-agglutinative, 61-63 semi-fusional, 61-63 transfixation, 21, 26 vocalic palterns, 18 Arabic, Egyptian collective nouns, 40, 86 singularization, 40, 86 Arabic, Moroccan exponence, 139, 141-142 Arabic, Syrian conditioning of allomorphs, 152 153 derivation (of nouns), 181-184 exponence, 141-142 iconicity, 197 number, 85 pronominal forms, 74 Bedawye (Kushitic) pronominal forms, 74 Berber: circumfix, 21 suppletion, 18 Cantonese pronominal forms, 76 Chinese compounding, 177 number, 88 Cree (Algonkian) gender, 81 pronominal forms, 76 third person, 113 verbal forms, 25 Czech accent, 15 adjectives, 6 aspect, 43 diminutive nouns, 55-56 gender, 83, 85 positional mobility, 56 pronominal forms, 75 voice, 125 English adjectives, 6, 71 allomorphy, 19-21, 148 150 aspect, 118 119, 188-191 compounding, 169, 175-476 conditioning of allomorphs, 150-151, 153 demonstrative pronouns with regard to proximity and remoteness, 113 derivational paradigm, 172 determinative compounds, 178-179 expansion (in word formation), 169 Functional Grammar, 194-196 gender, 81 infinitive, 107-108 markers of person, 112 morphological and syntactic predication, 107 morphonology, 161 neutral alignment, 95 number, 85 participles, 108 positional mobility. 14 prefixation, 170 173 pronominal forms, 73 segmentability of words, 17-18 semantic features. 3 subjunctive, 123 suffixation (in word formation), 173-175 syntactic compounds, 180 third person, 113 voice, 126 Finnish case, 92-94 French accent, 15 clitics, 24 gender, 83 overlapping distribution, 36 pronominal forms, 73 subjunctive, 123-124 suppletion, 18 voice, 125 German adjectives, 6 aspect, 119 exponence, 143 determinative compounds, 179 diminutive nouns, 55-56 gender, 83-84, 85 morphonology, 161 passive (analytic), 126 pronominal forms, 73 stem-forming clement, 55 strong nominal declension, 58 umlaut, 55, 200-201 voice, 126 weak nominal declension, 58 Greek, Ancient accent, 15 adjectives, 6 allography, 16 aspect, 1 16, I 19 120 clitics, 24 demonstrative pronouns with regard to proximity and remoteness, 114 exponence, 143-144 imperative, 122 infinitive, 108 markedness, 43 middle voice, 125 number, 88 participles, 108 110 suppletion, 18 210 ITRODUCTION TO Tlili STUDY OF MORPHOLOGY INDEX OF LANGUAGES 211 voice, 125 126 Greek, Modern distributional inclusion, 36, 42 gender, 83 tense-aspect, 120 Hebrew clitics, 24 derivation (of nouns), 181 gender, 84 graphemes, 16 reduplication, 7 Hindi ergative-absolutive alignment, 95 imperative, 121 passive (analytic), 126 personal pronouns, 73 reduplication, 7 voice, 126 Italian morphonology, 160-161 Janjero (Kushitic) gender, 81 Latin accent, 16, 93 adjectives, 6, 71 allomorphy, 161 aspect, 117-119 case, 88-92 clitics, 24 declensions, 58-61 demonstrative pronouns with regard to proximity and remoteness, 113 diminutive nouns, 55-56 discontinuous morpheme, 25 exponence, 139-141 Functional Grammar, 195-196 fusional type, 59 gender, 81-84 grammatical word, 22 imperatives, 121 infinitives, 107-108 irrealis, 125 middle voice, 126 mood, 120, 124 morphological predication, 106 morphonology, 161-162 nasal infix, 23, 26 nominative-accusative alignment, 95 overlapping distribution, 37 participles, 108-111 passive (synthetic), 126 personal suffixes, 114 positional mobility, 14 possessive pronouns, 77 -78 pronominal forms, 75 sememes, 2 subjunctive, 124-125 suppletion, 26 typology, 198 verba deponentia, 127 voice, 126-127 Lithuanian adjectives, 72-73 Malay number, 88 reduplication, 7 Modern Norwegian personal suffixes, 112 ' >M English adjectives, 72 gender, 83, 85 strong nominal declension, 57 58 weak nominal declension, 57 58 I )ld Norse markers of person, 112 Persian possessive pronouns, 77 Kliman ian clitics, 24 K ussian animacy, 119 200 aspect, 116 avoidance of passive, 127 case, 94-95 iconicity, 198 imperative, 122 morphonology, 159 161 syncretism, 44 Sanskrit coordinate compounds, 176-178 determinative compounds, 178-179 imperative, 122 123 middle voice, 126 passive construction, 128 possessive compounds, 179 180 voice, 125-126 Slavic gender, 84 Spanish accent, 191-194 aspect, 119 case, 92 clitics, 24 exponence, 142-143 subjunctive, 124 Sumerian gender, 80 number, 88 Turkish adjectives, 71 agglutinating language, 116 aspect, 59-60 case, 91-92 definiteness and personal pronouns, 112 113 demonstrative pronouns with regard to proximity and remoteness, 113 exponence, 140-141 gender, 81 markers of person. 111 112 personal suffixes, 114 positional mobility, 14, 56 possessive pronouns, 76 77 pronominal forms, 75 sememes, 2 third person, 113 typology, 198 vowel harmony, 27, 152- 159 Yiddish gender, 83 GENERAL INDEX GENERAL INDEX ablative (= abldtivus), 89 causae, 90 instrumental. 90 loci et temporis, 90 originis, 90 (jualitdtis, 90 ablaut, 57 absolutive (case), 60, 91 abstract agreement, 188 abstract noun, 70, 174, 182 accent, 191 accusative, 88 of place, 88 action. 70 active. 125, 127 additive compound, 176 addressee, 75 adjectival concord, 82 inflection, 71 adjective, 70, 174 simple (indefinite) in Baltic, 72 complex (definite) in Baltic, 72 adposition, 194 adverb, 70, 175 affix, 12 affixation, 168 agent, 95, 125 agentive phrase, 127 agglutinating (language), vi, 59, 61 agglutinative typology, 61, 198 exponencc, 140 agreement (= congruence), 71,82 marker, 188 Aktionsart, 168 alignment, 95 active-inactive, 95 double-oblique, 95 ergative-absolutive, 95 neutral, 95 nominative-accusative, 95 allograph, 16 allomorph, 1, 19, 20, 148 allomorphy, vi allophone, 1 alloseme, 1 ambiguity, 137 analogy (see proportional) analytic (morphology), 107, 126 animacy, 94 animate (being), 79 anteriority (relative aspect), 116 antonymy, 177 aorist, 119-120 arbitrariness, 4 article, 70 aspect, 70, 107, 116, 168 immanent, 116, 120 imperfeclive, 116, 121 inceptive, 116 perfective, 116, 119, 121 progressive, 116 prospective, 116 relative (anteriority), 116 retrospective, 116 transcendent, 116, 120 assertion, 123 assimilatory processes, 152 miiibutc, 70 iiiigment, 143 luse, 21 liasie meaning, 172 beneficiary, 88 bmarism, 41 hound (morpheme), 13 boundary, 24 morpheme, 24 word, 24 ealquing, 56 ease, 70, 88, 110 ablative, 90, 91 absolutive, 91 accusative, 88 comitafive, 91 dative, 88, 91 ergative, 95 essive, 94 genitive, 89 instrumental, 91 locative, 91 nominative, 88, 91 oblique, 110 partitive, 89, 94 translative, 94 ease marking, 194 causal clause, 124 change of state, 70 circumfix, 21 clause, 124 causal, 124 conditional, 124 consecutive, 124 final, 124 subordinate, 124 clitics (= grammatical words). 23 closed class, 73, 174 collective noun, 40, 85 comitative, 91 command, 120 comparison, 70 compensatory lengthening, 162 complementary distribution, 35 relationship, 200 units, 35 complex (definite) adjective, 72 complex lexeme (-composite), 166 componential analysis, 3 composite (- complex lexeme), 166 compound, 166, 176 additive, 176 coordinate, 176 copulative, 57 descriptive, 178 determinative, 176, 178 endocenlric, 180 exocentric, 180 possessive, 176, 179 subordinate. 178 syntactic, 176 compounding, 168, 169 concept, 5, 175 concrete noun, 70, 173 conditional clause, 124 conditioning lexical, 21, 150 morphological, 53, 150 phonological, 20-21, 150 congruence (= agreement), 71 conjugation(s), 23 conjunction, 70 consecutive clause, 124 constructional iconicity, 197 contrastive units, 35 214 AN INTRODUCTION TO THE STUDY OF MORPHOLOGY GENERAL INDEX 215 coordinate compound, 176 copulative compound, 56 countability, 85 countable noun, 85 counter-iconic, 197 cumulation (of significates), 59 cumulative (exponence), 139 dative, 88 of purpose, 91 declarative (sentence), 120 declension(s), 23 strong, 57, 72 weak, 57, 72 deep structure, 114 definite (complex) adjective, 72 definiteness, 91,92, 112 degree of markedness, 42, 92 deictic elements (pronouns), 114 deixis, 73 demand, 123 demonstrative (pronoun), 113 dependent (subordinate) compound, 178 deponent verbs, 127 derivation, 52, 169 derivational affix, 23 base, 53, 168 morphology, 4, 54, 166, 168 paradigm, 166, 172 derivative, 166, 168 description, 120 descriptive compound, 178 designator, 116 determinant, 168, 170, 178-179 determination, 168 determinative compound, 176, 178 determinatum, 168, 179 diagram (subcategory of icons), 5 diagrammatic (relationship), 198 diathesis (= voice), 125 diminutive noun, 174, 184 discontinuous morpheme, 2, 25 discourse analysis, 12, 38 distinctive features, 1, 3 distribution, 35, 42 complementary, 35 overlapping, 36 distributional equivalence, 35 inclusion, 36-37, 42 double marking (- extended exponence), 143 dual, 86 elative (in Arabic), 6, 184 endocentric compound, 180 epicene lexeme, 81 equational predication, 94 equipollent (opposition), 41 ergative (case), 95 essive (case), 94 Event Time, 116 exception filter, 167 exclusive (in the lsl Pcrs PI), 76 exoccntric compound, 180 expansion (in word formation), 169, 186 exponence, 139 agglutinative, 140 cumulative, 139 extended (- double marking), 143 fused, 142, 143 overlapping, 143 underlyingly agglutinative, 143 exponent, 139 expression rule, 194 extended (exponence), 143 exterior (with local cases), 92 external inflection, 18 extrinsic (ordering of rules), 192 lemiiiine, 81-85, 181, 183 htial clause, 124 lust person, 111, 113, 115 lixcd meaning, 172 locus, 195 lureign base, 173 lonn, vi, 4, 5 formal syntax, 188-191 formative ambiguity, 137 functional categories, 106 functional Grammar, 194-196 fused (exponence), 142 143 fusion (of significates), 59, 66 lusional (typology), 61, 198 future tense, 115-116 future time, 115 116 gender, 70 grammatical, 74, 82 natural, 74, 82 Generative Phonology, 160, 191-194 genitive, 89 descriptive, 89 objective, 89 partitive, 89 possessive, 89 subjective, 89 syntactic, 80 gerund, 111 gerundial (verbal) noun, 182 gerundive, 108, 110 grammatical analysis, 16 grammatical categories, vi primary, vi, 23 secondary, vi, 23 grammatical gender, 74, 79 word, 13 grammatical words (^ clitics), 23 grapheme, 16 heteronym, 81 heteronymy, 44, 81 of personal pronouns, 77 hypostatic noun (in Arabic), 184 hypothetical (judgment), 125 icon, 5 iconicity, 197 scale of, 7 image, 5 immanent (aspect), I 16, 120 imperfect, 119, 121 impcrfective, 116, 120, 122 imperative, 121 third person, 122 impersonal passive, 128 inanimate (thing), 79 inceptive (aspect), 116 inclusive (in the 1*' Pers PI), 76 indefiniteness, 92, 112-113 index, 5 indexicality, 198 indicative, 124 indignation, 123 infectum, 117, 126 infinitive, 107 passive, 108 perfect passive, 108 perfective, 108 retrospective, 108 infix, 21 inflection (inflexion), 52 adjectival, 72 classes, 199 216 ITRODUCTION TO THE STUDY OF MORPHOLOGY GENERAL INDEX 217 external, 18 internal, 18 inflectional affix, 23, 188 inflectional (flective) language, 2, 167 Inflectional Phrase, 188 instrumental case, 91 noun. 183 intercomponential conflicts, 200 intcrfix, 21, 168, 179 interior (with local cases), 92 interjection, 70 internal cohesion, 14, 56 inflection, 18 plural, 40, 85 interrogative pronoun, 78 sentence, 120 introflecting (language), 26 introflexive (typology), 198 irrealis. 125 isolating (typology), 198 Item and Arrangement Model, vi, 24-25 Item and Process Model, vi, 27 jussive, 120 language definition of, 4 levels, 1 langue, 39 lexeme, vi, 1, 13 lexeme-based derivational morphology, 54 lexology, 13 lexotactics, 4, 38 linear character, 38 linguistic pronominalization, 114 linguistic sign, 4 listener, 111 local case, 92 93 locative, 91 of property, 183 noun,183 macroparadigm, 200 mands, 120 markedness, 40, 197 degree of, 42 markerful, 197 masculine, 82 mass noun, 85 meaning (linguistic), 5 metaphor, 5 metonymy, 178 middle voice, 125 126 minimal pair, 36 minimum free form, 13 mood, 70, 106, 120 morpheme-based derivational morphology, 54 morpheme, vi, 1-3, 12, 16-17 morphological conditioning, 53 identity (= syncretism), 44, 52 markedness, 40-41 predication, 106 morphology (definition), vi morphophoneme, 159 morphophonemic rule, 160 morphonology, 159 morphosyntactic categories, 139 properties, 139 morphotactics, 4, 38 mutation (= umlaut), 55, 143, 201 mutually exclusive environments, 148 naming (function of the nominative), 88 narrated event, 115 n.uive base, 173 natural gender, 74, 82 Natural (Generative) Phonology, vii, 191 197 Natural Morphology, vii, 197 naturalness, x, 197 negative assertion, 123 neuter, 83 new information, 113 uominalization, 71 nominative, 88, 91 non-cumulative (exponence), 140 142 non-experienced time, 117 non-finite (verb forms), 107 non-iconic, 197 non-present time, 11 7 notional tense-system, 116 nouns, 70, 106 abstract, 70, 174, 182 collective, 85, 174 concrete, 70, 173 countable, 85 diminutive, 174, 184 gerundial (- verbal), 182 hypostatic (in Arabic), 184 instrumental, 183 locative, 183 mass, 85 occupational. 183 personal, 173 singulative, 85, 183 verbal (= gerundial), 182 number, 70, 85, 106 objective genitive. 89 oblique case, 110 occupational noun, 183 old information, 113 onomatopoeic, 6 open class, 73 opposition equipollent, 41 privative, 4] optative, 123 orthographical word, 13 overlapping distribution, 36-37 overlapping exponence, 143 paradigm economy, 200 paradigmatic, vi relations, 37 39 structure, 197 parole, 39 participation in discourse, 11 1 participles, 70, 107 active, 110 future, 108, 100 gerundive, 110 passive, 110 present, 110 partitive, 94 parts ofspeech, 70 passive, 125-128 past tense, 115-116 past time, 115-116 patient, 95 penultimate syllable, vi perfect, 117-120 perfective (aspect), 116. 119 perfectum, 117, 126 person, 70, 106 personal noun, 173 passive, 128 218 INTRODUCTION TO TUH STUDY OF MORPHOF GENERAL INDEX 219 pronoun, 77 phone, 1 phoneme, 1 systematic, 160 phonological analysis, 13 conditioning, 53, 150 unit, 1 word, 13 phonotactic rule, 4 phonolactics, 38 plural, 85 broken (= internal), 40 of abundance, 86 of paucity, 86 plurality, 85 polite prohibition, 123 polymorphy (= suppletion), vi, 2, 140 polysemy, vi, 2, 61, 140 polysynthctic (language), vi, 198 popular formation, 54 positional mobility, 14, 56 possessive compound, 176, 179, 181 pronoun, 76 possessor, 76 pragmatic functions, 195 predicate, 106 predication, 106 morphological, 106 syntactic, 106 prefix, 21, 170 prefixation, 169 preposition, 70 present tense, 115 present time, 115, 117 privative (opposition), 41 process morpheme, 17 progressive (aspect), 42, 116 prohibition, 42, 116 pronominal reference, 82 pronominal voice, 125 pronouns, 70, 73 demonstrative, 113 indefinite, 78 interrogative, 78 possessive, 76 relative, 78 proportional analogy, 6, 201 prospective (aspect), 116 proximate, 112-113 proximity, 73, 112-113 quantity, 85 quasi-nominal (verb forms), 107 question, 120 real (wish), 125 receiver, 88 reduplication, 7, 88 reference, 5 pronominal, 82 referent, 4 reflexive sentence, 125 relative aspect (anteriority), 116 sentence, 125 pronoun, 78 remoteness, 112- 113 resultative (aspect), 42 retrospective (aspect), 116, 119 root, 12, 21 consonantal, 18 second person, 111, 113 secondary endings, 143 segmentability, 17 semantic components ( - features), 79 markedness, 40 -.< iiianlico-syntactic functions (= properties). S9 •iriiieme, 1-3 m-iiii agglutinative, 61 m-ini lusional, 61 vmi morphemic, 172 .ciiiioiics, 5, 8 sentence, 38 defied, 114 declarative, 120 inlcrrogative, 120 jussive, 122 reflexive, 125 .eiilcntial stress, 114 .InHer (tense), 115 '•il'ii ( signe), 4 m>:»ikiuitiian A. Danylenko & S. v.ikuli-iiko» "■ I milom-se S.-Y. Killillglcy* n ' kn|ilisi h R. Schulz & A. Eberle im I *j ik.uivelian) U.J. Luders I" kol.ul (Papuan) T.E. Dutton* 11 < .ijiiIii/Kvvini (noil-Pama-Nyungun) w MiGicgor* I ' luni (1 ungusic) A.L. Malchukov* Middle Egyptian R. Schul/ el al. I« Sun.kiil I). Killingley & S.-Y. I- dlmglcy* I - lili-mci Otomi (Olomangucan) Y. '" M»i)l |Algic)K.V Teelei 11 Mnhala .1 W. Oair & I Paolillo* I' I uinunaco (Carib, extinct) Sp. Gilden & S Meira mi Zulu S E. Hosch & G. Poulos* '. > i uniiiiicn (Bantu) Ahmed-Chamaga • h I iilu-luuan (Polynesian) R Hooper* .'' Kiiiiiima M I,. Render* '.' Itt-luruvsian A Ja. Suprun & U I lolcsclial 1.1 Muldiv ian/Divehi J.W. Gair & tl. I 'urn 1.1 liogun V. Plungian* <<' Corse M. Giacomo-Marcellesi* i.i. Ilulgiirc J. Feuillel* i.h Siniierian J.L. Hayes* i.'i llmiliiatesedtal. dial.)R. Bigalke* 'ii II Gullcgo J.A. Perez Flouza* M I'iiiia Bajo(Uto-Aztecan)Z. Estrada I .ui.iiidez* ' l kuldcras (Romani) L.N. Ii liciciikov & M.F. Heinschink ' I Al.i mmc (Ital. dial.) R. Bigalke* '' I hiisa Tibetan S. DeLancey in I ,..lin ilia Val Badia L. Craft'onara "' Soulclin (Basque dial.) (J J. Lüders nut i anlese (Guyanese Creole) H. Devonish 81 Akkadian Sh Izre'el 82 Canaano-Akkadian Sh. Izre'el * 83 Papiatnentu (Creole) S. Kouwenberg & E. Murray* 84 Bcrbice Dutch Creole S. Kouwenberg 85 Rabaul Creole German (Papua New Guinea) C Volker 86 Nalik(Austronesian)C. Volker 88 Nyulnyul (non-Pama-Nyungan) W McGregor* 89 Warrwa (non-Paina-Nyungan) W. McGregor* 92 Icari (Dargwa) NR. Sumbatova & R.O. Mutalov 93 Daur (Mongolic) Chaolu Wu (Ojiyedtn Chuluu)* 100 Bare (Arawak) Alexandra Y Aikhenvald* 101 Acadian French D. Jory & V. Motapunvane* 102 Polabian (Slavic) W. Suprun & 11. Doleschal 103 ChamlingK. Ebert* 104 Kodava (Dravidian) K. Ebert* 105 Romanes (Sinti) D. Holzinger* 106 Sepecides-Romani P Cech & M.F. Heinschink* 107 Roman (Roinani) D.W. Halwachs et. al. 109 Karachay (Turkic) Sr. Seegmiller* 111 Nivkh E. Gruzdeva 114 Hittite S Luraghi* 115 Lower Sorbian (Slavic) G. Spieß 116 Songhay R Nicolai & P. Zima* 117 Macedonian V.A. Friedman 119 Abkhaz SI. Chirikbo 120 Ainu J.C Malier 121 AdygheR Smeels 122 Tuki (Niger Kordofan) E. Biloa 123 Hindi Mahendra K Verma 124 Q'eqchi' (Mayan) J. DeChicchis 125 Czech L. Janda & Ch.E. Towusend 127 Modern Hebrew O. Schwarzwald 128 Turin Piedmontese D Ricca 129 Siciliano K. Bigalke* 130Ratahan N.P. Himmelmaim & J.U. Wolff" 131 El nähuatl de Tezcoco Valentin Peralla 133 Tsakhur W. Schulze* 1.35 Late Cornish I Wmffre* 136 Fyem D. Nettle* 137 Yingkarta A. Dencil* 138Jurruru A. Dench 139 Svan K. Tuite* 141 Evenki N. Bulatova & L. Grenoble 142 Modern Hebrew O. Schwarzwald 143 Old Armenian N. Kozintseva 145 Russian E Andrews 146 Uzbek I.D. Cirtautas 147 Georgian M. Cherchi 148 Serbo-Croatian S Kordic* 150 Azeri A. Bodrogligcti 151 Tagalug L. Shkarban 152 Central Breton I Wmffre* 153 Demotic St. Vinson 154 Polci R. Cosper 155 Bashkiri A. Bodrogligcti 158 Vogul T. Riese 159 Mandan (Siouan) Mauncio Mixco* 160 Upper Sorbian G Schaarsclunidt 161 Toura (Mande) Th. Bearth 162 West Greenlandic J.M. Sadock 165 Dagaare (Gur) A Bodomo 166 Yuchi M.S. Linn 167 Ik-limn J. Bobaljik ! 68 Apache W de Reuse 169 Modern Greek B.D. Joseph 170 Tol D.Holt* 171 Secret Language of Chinese Yanbin Qu 172Lummi (Salish) R. Demers 173 Khamnigan Mongol Juha Janhunen 174 Nepali Balthasar Bickel & J. Peterson 175 Comecrudn R.C. Troikc 176 Panamint (Central Numic, Uto-Aztecan) J McLaughlin 179 Toba FIE. ManelisKlein 180 Degema EE. Kari* 181 Kupcfio J. Hill 182 Cayuga FI.-J. Sasse l83Jaqaru MJ. Hardman 184 Madurese W. D Davis 185 Kamass A. Kiinnap 186 Enets A. Künnap 187Guajiro J.Alvarez 188 Kurdish G. Haig 189 Salar AM Dwyer 190 Esperanto Ch. Gledhill 191 Bonan Chen Nai-Xiong 192 Maipure (Arawak) Raoul Zamponi 193 Kiliwa (Siouan) M Mixco 199 Miluk Coos (Coosan) Anthony Grant 200 Karbardian (East Circassian) John Colarrusso 201 Irish Aidian Doyle 202 Qae Evelyn Todd 203 Bilua Evelyn Todd 204 Ket Edward J. Vajda 205 Finnish Borje Vähämäki 206 Ancashino Quechua S. Hornau Aguilar 207 Damana (Clubclia) Maria Tnllos Aniava* I' .ill uady published 208 Embera (Choco) Daniel Aguirre' 21)9 Hiligaj non / Ilonggo Walter L. Spitz 210 Lobire Moses Kwado-Kambou 211 Fering (Northlrisian, Germanic) Karen Ebert 212 Udmurt (Finno-Ugric) Erberhard Winkler 213 Ancient Greek Silvia Luraghi 214 Chiwere Siouan N Louanna Furbce & Jill D Davidson 215 Chuckchee (Paleosiberian) Alexander Volodin 216 Chiriguano Wolf Dietrich 217 Latvian Nicole Nau* 222 Tyvan Gregory Anderson 225 Slovenian Ch. Gnbble 227 Malayalam Rodney Moag 242 Modern Scots Alexander T. Bergs 251 Xakat Gregory Anderson* 252 Old Saxon James E. Calhey 254 Saho (East Cusliitic) Giorgio Banti 255 Udeghe (Tungus-Manchu) Albina H.Girlanova 256 Newari/Newar E. Austin Hale 257 Tyvan (Turkic) Gregory Anderson 258 Biri (Pama-Nyungan) Angela Terrili* 260 Ostyak (Uralic) lrina Nikolaeva 261 Lingala Michael Meeuwis* 262 Klallam Timothy Monller 263 Manchu Carsten Naeher 266 Chuj Judith Maxwell 267 Kaqchikel Judith Maxwell 268 Urak Lawoi' David Ifogan* 273 Bubbure Andrew Haruna 274 Romanian Cynthia M. Vakareliyska 275 Aragoncs Carlos lnchaurralde 276 Chagatay A. Bodrogligeti 277 Turkish A. Bodrogligeti 278 Islcrm Spanish Felice Coles 298 Gheg Pandeli Pani 31«) Nuu-chah-nulth (Nootka) T. Nakavama 301 Oneida C Abbott 302 Sapuan P. Jacq & P Sidwell* 303 Oi P. Jacq & P. Sidwcll 304 Talicng P Jacq & P. Sidwell 305 Ostyak I Nikolaeva 306 Ottoman A. Bodrogligeti 307 Factar Naomi Nagy 308 Choctow P. Kwatchka 3! I Juang Manideepa Patnaik 312 Karitiana L. Raccanello Storto 320 Kawesqar Oscar Aguilar F. 321 Turkish A. Bodrogligeti 322 Shanghai Sean Zhu 323 Santali Lukas Neukom 324 Karaj K. David Harrison 325 Pileni Ashild Naess 326 Echie Ozo-Mekuri Ndimele 327 Judeo-Arabic Benjamin Hary 328 Tobelo Gary Holton 329 Ogbronuagum E. Kari 330 Old Nubian Gerald M. Browne 331 Taiwanese Lilly L. Chen 332 Kiswahili Sakari B. Salone 333 Wolof Fallou Ngom 334 Karao Sheni Brainard 335 Japanese Yoshihiko Ikegami 336 East Fricsland Yaron Matras & Ciertrud Reershemius 337 Selayarese Hasan Basri 338 Old Church Slavonic Boris Gasparov 339 Malagasy Charles Randria-masimanana Languages of the World/Text Collections: 01 Even- Texts Andrej Malchukov 05 Palestinian Texts Kimary N. Sllahm 07 Tariana Texts (North Arawak) Alexandra Aikhenvald* 08 Chinook Jargon Zvjezdana Vrzic tW Western Apache Texts W.de Reuse 11 Camling -Texts Karen Hbert 12 Helmen - Texts Jonathan David Bobaljik 14 A Collection of Laz Spoken Texts (+CD-ROM) Silvia Kutscher & Nuran Sevim Gene* 15 Saho Texts Giorgo Banti 16 Mbay Texts John M. Keegan 17 Der Ostfränkischc Basisdialekt von Hetzles Klaus Geyer Languages of the World/Text Library: 01 Minhe Mangghuer Folktales Zlm Yongzhong, Wang Xian/heng, Keith Slater & Kevin Stuart 02 Xunhua Salar Folklore Ma Wie, Ma Jianzhoug & Kevin Stuart 03 Huzhu Mongghul Folklore Limusishiden & Kevin Stuart 04 Huzhu Folklore Selections I.iiiiusishideii & Kevin Stuart (eds.) 05 Die udischen Evangelien der Cebruder Be)anov (1893) Wolfgang Schulze 06 Anthology of Menominee Sayings Timothy Guile 07 Kawesqar Texts Oscar Aguilar F. Languages of the World/Dictionaries: 01 Minhe Mangghuer - English Dictionary Wang Xianzheng (Qinghai Medical College/, 7hu Yongzhong (Zhongchuan Junior Middle School), Keith Slater (Qinghai Junior Teachers' (W/t^e), & Kevin Stuart (University of California, Santa Barbara) 03 Dictionary of Mbay John Keegan* 05 Dictionary of Sango Bradford & Bradford 06 A Dictionary of NegerhoUanrfs Robin Sabino & Aiwe-Katrin Oramberg 07 Degema - English Dictionary Elhelbert Kari 08 Eudeve Dictionary David Shaul 09 A Short Bonan-English Dictionary Chen Nai-Xiong 10 A Short Dongsiang-English Dictionary Chen Nai-Xiong 11 A Short Mongour-Engiish Dictionary Chen Nai-Xiong 12 A Short East Yugour-English Dictionary Chcti Nai-Xiong H A Short Dagour-English Dictionary Chen Nai-Xiong 14 Tyvan dictionary Gregory Anderson 15 Xaka* dictionary Gregory Anderson 16 Nhaheun - French - English Lexicon Miche] Ferlus (ed. by P. Jacq & P- S\dwell) 21 Comparative West Bahnaric Dictionary P. Jacq & P. Sidwel] 22 Palestinian Arabic-English / English-Palestinian Arabic Dictionary Kimary Shahin 23 Loven (Jruq) Consolidated Lexicon Pascale Jacq & Paul Sidwell An Introduction to the Study of Morphology Vll 13ubenik Molilalia! University of Newfoundland I in Ii chapter (with the exception of the last one) is provided with pertinent exercices. Its iiniii are taken from languags the author has been researching over the last twenty years (1 iihn, Greek, Turkish, Arabic, Hebrew, Sanskrit, Russian). Its argumentation is built mound Ihe major turning points in the history of morphology linked with scholars such as HiMkott (1954), Matthews (1974), Bybee (1985), Dressier (1985), Bauer (1988), Spencer (Kiwi), Carstairs-McCarthy (1992) and Aronoff (1993). In the last chapter the author implicates a cognitively conceived subdiscipline of Morphology in its relation to Formal i.ynlax, Generative Phonology, Functional Grammar, so-called Natural Morphology, i /mversal Grammar, and Typology. Contents: Introduction, Grammatical Units, Paradigmatic and Syntagmatic Relations, Inllcctional and Derivational Morphology, Inflectional Categories Associated with Nominal I laments, Inflectional Categories Associated with Verbal Elements, Morphosyntactic I'lnperties and their Exponents, Morpheme and Allomorph, Derivational Morphology, I hcoretical Models of Morphology, References. IMIN ;) 89586 570 2. IINCOM Coursebooks in Linguistics 07. <« //Opp. USD 48/DM 72/£28. Structure and Interpretation in Natural Language Marc Authier & Lisa Reed r/iH Pennsylvania State University Hie central objective of this book is to present an integrated theory of the syntax-smuantics interface, one which combines the most recent advances in the generative homework with the basic tenets of model-theoretic semantics. The three opening chapters develop, in a step-by-step and highly accessible fashion, an approach to •.liucture and meaning in these terms. The remaining chapters show how this approach sheds light on three long-standing issues in formal grammar: the treatment of "syntactically-triggered" presuppositions, the lieatment of some notable exceptions to the generative binding conditions, and the issue id Ihe relative autonomy of syntax and semantics. With respect to the first issue, it is indued that a compositional treatment of syntactically-triggered presuppositions can be loiuiulated as a condition which ties presuppositional triggers to a specific class of syntactic configurations definable in terms of devices found in Minimalist syntax A subsequent chapter demonstrates that the empirical coverage of so-called Bare-Output Conditions in generative syntax can be increased if such conditions are made sensitive to lín; Iwo types of semantic information which have sometimes been recognized in model-llieoretic semantics; that is, extension expressions and implicature expressions. Finally, iimpirical evidence is adduced which supports the view that there are two distinct types of semantic constraints and that those which make reference to features of tree geometry can, under specific circumstances defined by representational Economy conditions, nvDiride those which do not. Audience: Linguists, philosophers, computational and psycho-linguists, cognitive scientists, advanced undergraduates, graduate students and researchers in these fields. ISDN 3 69586 603 2. I INCOM Studies in Theoretical Linguistics 14. .'Illpp USD 70/DM 112/£42. Introduction to Linguistic Field Methods Bert Vaux & Justin Cooper Harvard University LINCOM Coursebooks in Linguistics The present volume addresses the need for an up-to-date, accessible, and comprehensive introduction to the elicitation of linguistic data from native speaker informants. The material, following an introductory chapter surveying the general enterprise of field research, is organized into eight major areas of current linguistic and anthropological interest: Phonetics, Phonology, Morphology, Syntax, Semantics, Sociolinguistics/ Dialectology, Lexicography, and Folklore. The chapters are designed to be covered at a rate of one per week, based on a sixteen-week semester. Each chapter presents basic structures to be elicited, and provides cautionary tales drawn from the experiences of seasoned field workers who have attempted to elicit these structures. These, in turn, are followed by suggested readings and illustrative exercises for each chapter. Emphasis is placed not on developing a theory of field work, but rather on providing enlightening suggestions and entertaining anecdotes designed to guide students down their own personal path to linguistic discovery. ISBN 3 89586 198 7. LINCOM Coursebooks In Linguistics 01 Ca. 240 pp. USD 48 / DM 72 It 28. Coursebook in Feature Geometry John Newman Afassey University in this series: (M Bert Vaux & Justin Cooper W John Newman i)!> Fernando Garcia Murga (l/ Vit Bubenik 11 Christopher Gledhill 12 Paul Bennett Introduction to Linguistic Field Methods Coursebook in Feature Geometry El Significado: Una Introduction a la Semantica An Introduction to the Study of Morphology Fundamentals of French Syntax Semantics: an Introduction to Non-Lexical Aspects of Meaning The Coursebook in Feature Geometry is an undergraduate course introducing students to current phonology through a sustained use of the Feature Geometry framework. It is written as a coherent, accessible, and well-illustrated introduction to the key ideas of Feature Geometry, focusing on rules of assimilation. In its 20 units and 40 exercises, it takes the reader step-by-step through the representational devices of Feature Geometry. The Coursebook attempts to present the core ideas of Feature Geometry in a unified way, rather than attempting to incorporate the (considerable) debate concerning almost every aspect of the theory. The version of Feature Geometry underlying the Coursebook is basically that found in Sagey's The Representation of features in non-linear phonology (1990), revised in accordance with the claims of Lahiri and Evans' 1991 article on Palatalization and coronality. The author is Senior Lecturer in the Department of Linguistics and Second Language Teaching, Massey University, New Zealand. The author has a PhD in linguistics from the University of California at San Diego. ISBN 3 89586 102 2. LINCOM Coursebooks in Linguistics 02. 160pp. USD 38 / DM 62 / £ 25. Course discounts available!