ebook img

Categorial morphology PDF

129 Pages·1985·7.768 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Categorial morphology

RIJKSUNIVERSITEIT TE GRONINGEN CATEGORIAL MORPHOLOGY PROEFSCHRIFT ter verkrijging van bet doctoraat in de Letteren aan de Rijksuniversiteit t!;O' Groningen op gezag van de Rector Magnificus Dr. E. Bleumink in het openbaar te verdedigen op donderdag 20 december 1984 des namiddags te 4.00 uur door JACOB HOEKSEMA geboren te Groningen 1984 DRUKKERIJ VAN DENDEREN B.V. GRONINGEN PROMOTORES: Prof. Dr. A. Sassen Prof. Dr. J. F. A. K. van Benthem ACKNOWLEDGEMENTS. Writing a dissertation can be a lonesome experience, To get it finished, one has to stay away from friends and social activities as much as possible. Yet one never walks that road alone. It is my pleasure to acknowledge here my gratitude to all those people who helped me along the way, To Albert Sassen, my promoter, I am indebted for his detailed criticisms of earlier versions of this study, but above all for his unceasing support and guidance, I will certainly miss those lovely garden parties at his house in Haren. To my other promoter, Johan van Benthern, I am likewise indebted for valuable criticisms, but even more for his stimulating presence, which kindled my enthusiasm for semantics and helped create the sudden semantic wave in Groningen, It has been a pleasant experience to study and to do research in the department of Dutch Linguistics and Literature at the Rijksuniversiteit Gro ningen, I wish to thank all my teachers and fellow students for creating a most agreeable environment, Thank you Theo van den Hoek, Car Hoppenbrouwers, Theo Janssen, Gerrit Brummel, Machteld van Royen, Nanne Streekstra, Wiet ske Wiersema, Sjaak de Meij, Eric Reuland and Frank Reny for teaching me linguistics -- it took your combined forces to teach me the little I know. A spe'cial word of thanks goes to Frans Zwarts and Ron van Zonneveld. They have been invaluable teachers, colleagues and friends, Frans' work in syntax, later semantics, has always been something like a North Star to me. Ron was responsible for talking me into morphology. Our joint- collabora tion on an introduction to morphology started my interest in the field and much of what I know about it dates back to that period, Ron's work in mor phology has led me' to consider categorial grammar as a viable framework to study word structure in, My interest in categorial grammar was reinforced by two visitors from Amherst, Deirdre Wheeler and Michael Flynn, I am glad· to have met them, I have always enjoyed and Often profited from the numerous conversa tions I had with the other junior researchers in Groningen, Crit Cremers, Jan Jullens, Siemon Reker, Nico van der Zee, and, at other departments, Jan van Eijck, Alice ter Meulen, Elias Thijsse, Jan de Vuyst, Christine ter Mars and Leonie de Smet, I thank Jan van Eijck for his conunents on the first three chapters. For checking my English, I thank Steph Robinson, Rich Janda and Eleanor Sapp, All remaining errors are my fault, I'm afraid. The research reported here was supported by the Foundation for Linguis tic Research, which is funded by the Netherlands Organization for the Advan cement of Pure Research (ZWO), grant no. 17-24-04. I greatly enjoyed working for ZWO and the freedom I was allowed to pursue my various linguistic inter ests. I thank my relatives and friends for putting up with me and providing moral support, Without them, I don't know whether I would have made it. Offset Van Denderen, Groningen PREFACE. In what follows, I present an account of certain problems of morphologi cal analysis within a theoretical framewoTk that derives its inspiration from recent studies of the lexicon in generative grammar, as well as from the semantic theories of Richard Montague and his followers. The latter influence will be most evident in the choice of categorial grammar as the model of syntactic description and the use of formal interpre tations for lexical expressions. The influence of generative studies is equally evident from the choice of the issues to be addressed here, viz, the role of the compositionality principle in word-formation, the analysis of synthetic compounds and the in terpretation of nominalisations, all of which figure quite prominently in generative studies of the lexicon, such as Williams (1981), Roeper and Siegel (1978) and Chomsky (1970). The starting point of this book was the controversy about the proper analysis of synthetic compounds. Are they, as their name suggest$, really compounds, or phrasal derivations, or do they constitute a type of word formation of their own? All possible pc;isitions have been defended in the literature, and fOr each of the proposals, sensible arguments have been ad vanced. A deeper understanding of the issues involved could only be arrived at, I concluded, if the questions to be answered were formulated in an expli cit and precise way. For instance, it has been argued, most forcefully by Botha, that synthetic compounds should be analysed as phrasal derivations, since only that analysis would be compatible with a compositional interpreta tion of these items. To see whether such an argument holds water, it is necessary to spell out a formal procedure to interpret lexical expressions and to formulate, within that theory of interpretation, the principle of composi tional"ity, Such a theory is readily available in Montague grammar and has already been set to work in a number of publications on word-formation. The syntactic framework that goes with it, categorial grammar, is not really neces sary for the purposes of semantics and indeed some linguists have combined transformational grammar with the type of modeltheoretic interpretation that one finds in Montague grammar (e.g. Mccloskey 1979), or context-free phrase structure grammar (cf, Gazdar 1982), Nevertheless it appears to me that categorial syntax has a number of properties which make it particularly well opposed to compositional - semantics of some of the affixes involved in suited for the purposes of lexical analysis, For one, it is less redundant (Dutch) synthetic compounding, to wit, the nominalisation markers ge- and than phrase structure granunars in several respects. One only needs to speci -ing. I focussed on these affixes because they are interesting in several fy the semantic domains for the basic, or atomic, categories, and all of the respects: they are sensitive to aspect, have a different distribution and other domains of interpretation can be predicted by rule, For instance, if different meanings, The semantics given for these items attempts to account the expressions of category A are interpreted in the set X, and the expres for all of this at once. sions of category Bare interpreted in the set Y, then the expressions of the A large part of today's morphology is concerned with the relation be derived category A/B are interpreted as members of the set of all functions tween word structure and phonology. The relation between word structure and from Y into X. Also, there is no need for a long list of phrase structure word meaning, although generally considered important, receives less atten rules. Only a few very simple combination schemata are needed. In addition tion, Often, the meaning of a word is used as a pretheoretical guide to to this, categorial grammar has a very simple and straightforward way to cate morphological structure, For instance, if ticklish is analysed as tickle + gorise affixes: the category of an affix is completely determined by the cate ish, while rubbish is not parsed as rub + ish, that is due to the semantics gories of the bases it attaches to and the derived words that it creates. of these words, Besides this fairly trivial aspect, there is often remark These properties of categorial grammar make it, in my opinion, a really able little interest in the meaning of complex words as an object of formal attractive framework to do morphology in. linguistics (as opposed to informal observations), If this book contributes The structure of this book is as follows: first, the general design of a to bringing work in modeltheoretic semantics and categorial grammar under categorial approach to morphology is sketched (chapter one) and some of the the attention of those who study those elusive objects called 'words', then notational conventions to be used later on are introduced. Then I discuss it will have served its purpose well. the status of the compositionality principle within this categorial theory and discuss some of the objections that have been raised against it in the liter ature (chapter two), In chapter three, I outline an analysis of composition along categorial lines and introduce a distinction b.etween nominal compounds and relational noun compounds, which turns out to be useful for the analysis of synthetic compounds as well. In chapter four, the recent literature on synthetic compounds is reviewed, while in chapter five a categorial analysis of these troublesome items is given, which is nonuniform in the sense that different types of synthetic compounds are treated differently. Hence the traditional name "synthetic compound" is not, in fact, theoretically justi fied on the account to be presented here. In all cases, the compound analysis of synthetic compounds as argued for by linguists such as Selkirk and Moortgat, is vindicated, even though a phrasal derivation approach, as pro ~osed by Botha and others, is not ruled out on theoretical grounds, since I want to allow for a greater amount of interaction between syntax and morphology than the strictly word-based morphology of Aronoff (1976) and his followers admits. The final chapter, chapter six, is devoted to the lexical - as TABLE OF CONTENTS. Page Page 2.5. More on head operations 56 CHAPTER ONE: THEORY OF THE LEXICON 2.6. Concluding remarks 62 1.1. In the beginning 1 Notes 65 1.2 What is the lexicon ? 2 1.3. Actual and potential words 4 1.4. The mental lexicon 7 CHAPTER THREE: COMPOSITION. 67 1.5. The actual and potential lexicon: a formal sketch 11 3.1. Delimitation of the subject 67 1.5.1. Phonological representations 12 3.2. Compounds versus syntactic phrases 69 1.5.2, Categorial representations 13 3.3: Some aspects of the categorial analysis of compounds 73 1,5,3, Semantic representations 13 3,4, Relational nouns 80 1,6, Lexical rules 14 3.5. Adjective compounds 86 l. 6 .1. Rule ordering 14 3,5,l, Elative compounds 86 1.6.2. The format of lexical rules 16 3.5.2. Relational adjectival compounds 91 1,6,3, Examples of lexical rules 17 1.6,3.1. One-place operations 17 3.5.3. Other cases 93 3.6. Compounds without a functor-argument structure 93 1,6,3.2. Two-place operations 18 3.7. Summary of this chapter 97 1.6.3,3. N-place operations 20 Notes 98 1.6.4. Categorial granunar and subcategorisation 20 1.6.5. Phonological sensitivities 23 1,6,6, Function application and semantic sensitivities 25 CHAPTER FOUR: SYNTHETIC COMPOUNDS. 100 1.7. Interaction between syntax and morphology 27 4.1. Synthetic versus other compounds 100 1.8. Summary of this chapter 30 4,2, Roeper and Siegel's lexical transformation theory 103 Notes 32 4.2.1. Outline of the proposal 103 4,2,2, Evaluation 108 4.3. Botha's phrasal analysis of synthetic composition 112 CHAPTER TWO: THE COMPOSITIONALITY PRINCIPLE IN LEXICAL THEORY 34 4,3.1, Outline of Botha's theory 112 2,1, Introduction 34 4.3.2, Evaluation of Botha's theory 120 2.2. Prelimary remarks 35 4.4. The Allen-Meijs-Selkirk account of synthetic composition 126 2.3, Arguments against the compositionality principle 38 4.4.1. Meijs' theory 129 2.3.1. The holistic point of view 38 4,4,2, Discussion of Meijs' theory 130 2.3.2. Mismatch between structure and interpretation 41 4.4.3. Selkirk's theory 131 2.3.2.1. Lexical relatedness 45 4.4.4. Discussion of Selkirk's theory 134 2.3.2.2. The argumentation for the 'noncompositional' 4.5. Moortgat's theory 136 structures 47 Notes 140 2,3,2.~. A digression on Subjacency and the Atom Condition 52 ) 2.4. The principle of compositionality 55 CHAPTER ONE: THEORY OF THE LEXICON. Page 1,1, In the beginning. CHAPTER FIVE: A CATEGORIAL THEORY OF SYNTHETIC COMPOSITION, 142 From the days of yore, the main tasks of the linguist have been the writ 5,1. A classification of Dutch and English synthetic compounds 142 ing of granunar books and dictionaries. Both grammars and dictionaries reflect 5.2. Deverbal synthetic compounds 146 important aspects of language: dictionaries provide the elementary building 5.2.1. Nominalisation 147 blocks, and the grammars specify how these blocks are combined to make well 5.2.2. Excursus on verb interpretation 149 forrned utterances. 5.2.3. Towards a modular theory 154 The theory of the lexicon c·ombines these two aspects: on the one hand, 5.2.4. Back to nominalisation 158 it describes generalisations about the basic elements, on the other hand, it 5.2.S. Agentives 165 studies the ways in which these elements interact and form larger word struc 5.2.6. Participles 170 tures. The latter part of lexical theory has been called 'word grammar' or 5,3, Denominal synthetic compounds 174 'word syntax', which seems to be a good name for the _enterprise, 5,3,l, -er derivations 175 In geherative grammar, lexical theory is regaining some of its former 5.3.2. Adjectival denominal synthetic compounds 176 popularity, after a period of neglect, but it is still not at the core of cur 5.3.2.1. Two types of pseudo-compound constructions rent linguistic developments -- unlike syntax or phonology. Within the tradi in Dutch 181 tion of generative grammar, the main issues concerning lexical theory have 5,4. Summary of this chapter 188 been the delimitation of the lexicon (i.e. drawing the border lines between Notes 189 morphology and syntax, and be.tween morphology and phonology) and the determi ]nation of the internal structure of the lexicon. With respect to these ques CHAPTER SIX: SEMANTIC PROPERTIES OF DUTCH NOMINALISATIONS, 193 tions, the dissertations of Siegel (1974) and Aronoff (1976) have been rather 6.1. Introductory remarks 193 influential, as well as some articles published in the same period, e.g. Jack 6,2, The mass-count distinction 194 endoff (1975). Another issue that has attracted the interest of many investi~ 6,3, Verbal aspect 200 gators has been the pragmatics of word creation (notably the work of Downing 6,4, Nominalisations and aspectual categories 205 (1977) on composition and that of Clark and Clark (1979) on conversion), 6,5, A semantic account 212 I will not attempt to give anything like a full account of the develop 6,5,1, The semantics of verbal aspect 212 ments in lexical theory in these pages. The interested reader may consult Boo~ 6,5.2. Applications of the aspectual semantics 216 (1977, chapter 1) for a good account of lexical theory within generative gram 6.5.3, The semantics of ge- and -ing nominalisations 220 mar, Another useful overview is to be found in the introduction of Hoekstra, 6,6, Summary of this chapter 223 van der Hulst & Moortgat (1980), where the whole of lexical theory is sketched Notes 224 from the 'lexicalist' point of view. My own work on morphology, to be presented in this study, takes as a starting point these recent developments in generat ive morphology, as well as the more semantically oriented studies emanating from the work of the late Richard Montague. References 226 Although my main preoccupation in this study will be composition, especi ally synthetic composition, some matters of larger scope will be treated as well, including a general sketch of the lexicon as I see it, to which this are clearly derived from the phrases old maid, out of state and cold war, res chapter is devoted, pectively, do not represent regular or productive processes of word-formation, but, as I will argue in the following pages, there are clearly regular and pro 1.2. What is the lexicon ? ductive types of word-formation, especially in compounding, which involve phra- ses. It is by no means easy to state with precision what is meant by that Since generative grammar -- or, more precisely, some prominent representa well-known term 'lexicon', The reason for this difficulty seems to be that tives of that paradigm -- is concerned with the psychological reality of its there are several distinct conceptions connected with the term, theoretical constructs, a third, more psychological, conception of the lexicon One conception of the lexicon is that is the locus of all irregularity has arisen, If a grammar is a formal representation of the linguistic compe in language, or, to put it more bluntly, the junk yard of language, This tence of an idealised language user, then it is only natural to consider the view has been rather influential, I gather, and it is certainly reflected lexicon as the formal representation of that part of the competence which con in the position that was assigned tO the lexicon in early generative grammar, tains the mental dictionary. In the latter conception, the lexicon is related This particular view of the lexicon is much older, however, since it can al to the long-term memory of the language user. From this perspective, there is ready be found in Bloomfield's Language, where one reads (1933: 274): '~The little sense in the Bloomfieldian view of the lexicon, according to which lexicon is really an appendix of the grammar, a list of basic irregularities". there is a dichotomy between the rules of grammar, including the word-forma This usage is not followed here. If the lexicon is really just a set of tion rules, on the one side, and the irregularities and arbitrariness of the irregularities and arbitrary facts, then, surely, it would not make sense at lexicon, on the other side, The rules of word-formation are, as Aronoff has all to speak of a 11 lexical theory11, For a theory, one needs at least some reg put it, once-only rules, which are not applied every time a complex word is ulal;'ities. uttered, but only once, when the word enters the personal dictionary of the Instead, we will adopt another, also quite familiar, and, for our purposes, speaker, In this respect, the role of word-formation rules is different from more interesting, conception of the lexicon. By 'lexicon', we will understand that of syntactic rules, which are invoked for every utterance of a sentence that part of a grammar, which takes care of the vocabulary of the language, in (unless, of course, this sentence happens to be memorised, which is possible, cluding its rules of word-formation, This conception of the lexicon is in line but less likely). By listing complex words in his/her long-term memory, the with current work in generative grammar, especially in the so-called 11lexicalist11 language user attempts to balance ease of computation against memory load. tradition within that paradigm. From the psycholinguistic viewpoint, several questions become important By postulating that the lexicon is a distinct component of the grammar, it which are not usually considered by linguists, such as the manner in which is not claimed that it is autonomous, i.e. not depending on other parts of the lexical information is actually stored in the brain and retrieved for pro grammar. I take it that there is some amount of interaction between syntactic cessing. In section 1.4., I will briefly return to these issues. and morphological rules, and that a complete description of the set of (possible) The psychological notion of the lexicon is certainly quite useful, but, words of a language such as English or Dutch presupposes a syntactic description still, it is important to keep it ap~rt from the earlier one. Therefore, when of certain phrasal constituents. In this respect, I follow linguists such as talking about the psychological notion, I will use the term mental lexicon, Bloomfield, to mention just one prime representative of a view pervasive in much provided this is not already evident from the context, The 'linguistic lexi of the older literature, and reject Aronoff's (1976: 21) hypothesis that all reg con' is not claimed, to be sure, to be without psycholinguistic relevance - ular word-formation processes are strictly word-based. To be sure, one might ar on the contrary -- but it is an abstraction of the psychological notion, one gue that formations such as Bloomfield's old-maidish, or out-of-stater, a word I in which matters of storage and retrieval are simply not considered. recorded from an episode of Magnwn, p.i., or the formation cold-warrior, which 1.3. Actual and potential words. ble word if and only if it is in some rule-governed extension of the lexicon. Extensions of the lexicon, such as lexical borrowing, are not rule-governed in Besides the distinction between a mental and a 'linguistic' lexicon, there our sense, although, of course, it cannot be denied that there are regularities is another distinction which appears to be relevant. Several linguists have ar in borrowing as well. gued that one should distinguish between the actual lexicon, the collection of The set of all possible words will be called the potential lexicon, or, in all actual, or type-familiar (to use a term of Meijs 1975), words, and the po cases where there is no danger of confusion, simply the lexicon. The set of all tential lexicon, consisting of all potential, or type-familiar, words. Natural actual words is the actual lexicon. The actual lexicon is by definition a sub ly, the actual is always part of the potential, so the actual lexicon is a sub set of the potential lexicon, since each actual word can be derived from an ac structure of the potential lexicon. tual word in zero steps, The potential lexicon is the real subject of lexical The borderline between the actual and the mere potential is, like the bor theory. The actual lexicon is more interesting for lexicographers, although, to derline between the potential and the sheer impossible, often hard to draw. Nev be sure, a good knowledge of the actual lexicon is a prerequisite for an accu ertheless, these distinctions, though not always perfectly sharp, do seem to be rate reconstruction of the potential lexicon, important. Possible, nonactual words are defined as those words which can be The fact that it is often not at all easy to state which rules of word-for formed by means of established rules of word-formation from actual words (in a mation are productive in the lexicon of a certain period, makes the reconstruc finite number of steps, if infinite expressions are disallowed), but which, for tion of the potential lexicon a rather difficult task for many languages, The some reason, have not seen fit to achieve the status of recognised members of mere existence of a regularity in the lexicon does not suffice to prove that the vocabulary. For example, there is a very general rule of English, which says there is a productive rule which accounts for this regularity. Suppose, for in that from any two common nouns, we may form a new one by concatenating them. Now stance, that our actual lexicon contains 12 words, with a certain affix X. Must it would appear that there are many exceptions to this rule, if we consider just we invoke a rule of affixation to account for these 12 words ? The answer would item-familiar words. We may concatenate car and thief to form the item-familiar be negative, I take it, if all of these words turn out to be borrowings from, word car-thief, but not, for instance, boy and girl to form boy-girl, or girl say, classical Greek, for in that case the affixation rule in question would be boy, since these words do not exist (at least not in my speech), On;:the other more at place in the grammar of Greek. Only when native speakers start forming hand, these words might be formed (make up your own interpretation). And sure words with the same affix, rather than borrowing them, it makes sense to assume ly, it would be quite unilluminating to just list all exceptions to the rule. there is an actual rule of word-formation which- is responsible for the creation So we will claim that boy-girl and girl-boy are possible words of English, and of X-derivations, and hence, that there are possible, but not yet actual, words just happen to be nonactual at present. of the same type. It is important to keep in mind that the notion 'possible word' is defined One of the main questions regarding productivity is whether it is an abso in a strictly linguistic sense. If a word is not possible, according to the def lute notion, or just a matter of degree. Schultink (1961) maintains that produc inition given above, that does not mean that it will never be part of the actual tivity is an absolute notion, whereas Aronoff (1976) represents the view that lexicon. A language may acquire words in all sorts of ways, such as borrowing productivity can vary for any two rules of word-formation, and therefore must from other languages, clipping, acronym coining, or by simply adding new simplex be a matter of degree, words. Only one of these ways is vocabulary extension by application of the reg Productivity in our sense is an absolute notion, because it corresponds to ular rules of word-formation. And also, the rules of word-formation themselves the existence, or non-existence, of a rule. However, the other view of producti may change in the course of time, thereby making some impossible words possible, vity seems to make sense as well. When there are two morphological rules work:tng and some potential words impossible. in the same domain, we often note that one is more 11active11 than the other one, Rephrasing what was said above, we state that a non-actual word is a possi- in the sense that there are more actual words generated by one than by the other. Hence I do not see any point in arguing that OU'e of these interpretations is the have not changed in the mean time. This method, of course, has to be supple 11correct11 one: these are mutually compatible notions, which only need to be kept mented with the intuitions of the investigator and his informants, and with any apart. So, in order to avoid terminological confusion, I will reserve the term other evidence available, activity for the matter-of-degree notion of productivity, As Aronoff has convin If one wants to exclude diachronic reasoning, or if one does not trust the cingly argued, the activity of a word-formation rule may depend on the proper method, then one could use psycholinguistic tests, or word-formation tasks in ties of the base word. For example, Aronoff notes that the suffix -ness is more which subjects have to form new words with a specified meaning, etc. (cf. Cut productively attached to bases of the form X-ous than its rival -ity. In Dutch, ler 1980 and Aronoff & Schvaneveldt 1978, for examples of this type of psycho the diminutive affix -tje can be attached to nouns and adjectives, cf. hoer linguistic research). 'farmer', boertje 'small farmer', and geel 'yellow', geeltje 'little yellow one' Not only the origin of a morphological rule can be a mystery: its death (a banknote). However, the examples of de-adjectival diminutives are far less may be equally hard to establish. ·The problem is that morphological rules are numerous than denominal diminutives, and usually have idiosyncratic or special in some respects not unlike vampires: they are never quite as dead as they ap ised meanings. Hence we may conclude that diminutive formation is very active pear to be. A lexical rule may sometimes be awakened after a long period of in in the domain of nouns, and rather unactive in the domain of adjectives. If activity, or used in an analogical fashion, to create-nonce-formations. (For there are two rules for making comparable derivations, e.g. nominalisations, examples, cf. Schultink 1961, 1962.) We broach here the difficult question of then the output of the more active rule is usually considered more 'natural' or 'creative' versus 'noncreative' use of rules. For example, the word picnik, in likely than the output of the less active one, Activity is by its very defini the sense of -'-picture of a spoetnik' (listed in Bauer 1983), is most certainly tion connected with the actual lexicon, since in the potential lexicon, all lex an instance of creative word-formation, in which the inventor did more than sim ical are, by definition again, maximally active. ply apply the word-fotmation rules of his or her language. Many other -nik deri The important diachronic question of how it happens that a certain rule be vations, e.g. computeT'Ylik, or Freudnik, have&' similar_creative, ~r playful, comes productive, and, after a period of activity, burns out and dies, is out flavour, though, perhaps, to a lesser degree. Schultink wants to exclude crea side the scope of the present investigation. It is rather obvious that a rule tive, or, as he terms it, 'intentional', word-formation from the domain of pro comes into existence by a gradual process of generalisation and analogical ex ductive word-formation. However, creativity and intentionality are to my mind tension, and not by instantaneous creation or mutation -- unpredictable quantum also matters of degree, and sometimes hard to establish. In any event, these no jumps, found, apparently, in biological evolution, (For a detailed study .of the tions, if incorporated into a theory of the le-Xicon, would make it rather hard development of a suffix, -nik, in the late fifties and early sixties in English, to give an objective basis to the whole enterprise, and hence I will ignore them cf. Bauer 1983,) If this picture is reliable, then there is a fuzzy borderline here. Fortunately, the types of word-formation to be studied here will be mostly between productivity and nonproductivity, fairly common and regular, and the problems of intentionality are consequently How does one find out whether a lexical rule is productive ? The most ob less urgent. vious criterion is the following one: a lexical rule is no longer productive if and only if it does not generate new words in the actual lexicon. This crite rion is of a diachronic nature and indeed one straightforward way to check the 1.4. The mental lexicon, productivity of a given rule is to consult two fairly recent dictionaries, in order to see whether the most recent one contains more examples of the rule in Since the start of generative granunar in the late fifties, the so-called question than the other dictionary. Of course, one has to make sure that the two mentalistic view of linguistic theory has gained quite a few adherents. It is dictionaries are comparable, so one will usually use two editions of the same claimed that linguistic theory should provide precise granunars, which not only dictionary, provided, that is, that the scope or the editor of the dictionary describe a language correctly, but also can function as models of the mental capacity of human beings to speak and understand their language. In other words, stress pattern, if that is not predictable). Some non-terminal nodes should al- the grammars postulated by linguistic theory should not only be adequate as so function as exit states in case the word in question is properly contai- descriptions, or algorithms, but also be learnable, efficient to use, and so on. ned in another word (like war in wardrobe). This view is largely due to Chomsky, and has served to make the field of lin It is not hard to see why this is a very effective format fbr lexical or- guistics more appealing to psychologists and philosophers. ganisation. For i.n stance, for a word of ten letters, no more than ten decisions Without any doubt, the potential lexicon is a rather abstract characteri have to be made, in order to arrive at its lexical entry (i.e. the data struc sation o~ some vital aspects of the mental lexicon, and the grammar a rather ab ture referred to above), Contrast this with a search through a linearly ordered stract representation of the most important features of the linguistic compe list. It might take thousands of decisions (of the type: no, this is not the tence of the language user. As I mentioned above in section 1.2,, the linguistic one) before the right entry is found (unless, of course, the list is ordered lexicon usually abstracts from certain aspects of the mental lexicon having to in such a way as to speed up the Search) • do with retrieval of lexical information and its storage in long term memory. Nevertheless, this simple and attractive model of lexical storage and re For instance, it is known that the lexicon has a certain organisation of trieval is probably not correct, as Forster (1976) points out, since it pre its own, one which enable the language user to find a word in his long term mem dicts that language users can establish the nonexistence of a given word in the ory·very quickly. Facts about the retrieval of lexical items (also known as 'lex same time or less than they need to establish the existence of an equally long ical access') have been used to form hypotheses about this organisation. A very word. For instance, if one looks up the word atishnet in a leXical decision efficient way of retrieval is possible when the lexicon has the structure of a tree, one may stop after atis, for there are no English words beginning with lexical decision forest, that is, a set of trees of the following form: these four letters. For the existing word American, on the other hand, 8 deci sions are needed, since after seven letters we do not know whether the word is Figure 1: A lexical decision tree. America, American, or perhaps Americana, etc. So we predict that the nonexis A--C--H--E tence of atishnet is decided much faster than the existence of American. For / ster (1976: 260) informs us that so-called lexical-decision experiments, in p which one measures the time needed by a subject to classify a given letter se /~E quence as existent in his/her lexicon or as nonexistent, -show that "familiar A~T--L-A- words are classified in around 500 msec., but that nonwords require about 650 N- T- I--C msec.11 \ \s This implies that our simple tree model of the lexicon is not the correct representation. For much more discussion of the organisation of the mental lex icon from the viewpoint of experimental psychology, the reader is referred to E \ the above-mentioned paper by Forster, and further to Brown & MCNeill (1966), Fay & Cutler (1977), Bradley (1979), Glanzer & Ehrenreich (1979), Stanners, T--H-E Neiser, Hernon and Hall (1979), Hurford (1981), Cutler (1982). While the storage and retrieval has been central in the psycholinguistic Such trees can be seen as transition diagrams of deterministic finite-state auto literature on the lexicon, there has been very little attention paid to it in mata, the root being the initial state, and the terminal nodes being exit states linguistic studies. This may seem strange, since the structure of the mental leading to data structures in the information about the words is stored (such as: lexicon would appear to be part of linguistic competence, and not of linguis its categories, its meaning, and certain phonological properties, perhaps its tic performance, if we define the latter as the actual use of language in con-

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.