An Introduction to English Morphology: Words and Their Structure Andrew Carstairs-McCarthy Edinburgh University Press 01 pages i-viii prelims 18/10/01 3:42 pm Page i An Introduction to English Morphology 01 pages i-viii prelims 18/10/01 3:42 pm Page ii Edinburgh Textbooks on the English Language General Editor Heinz Giegerich, Professor of English Linguistics (University of Edinburgh) Editorial Board Laurie Bauer (University of Wellington) Derek Britton (University of Edinburgh) Olga Fischer (University of Amsterdam) Norman Macleod (University of Edinburgh) Donka Minkova (UCLA) Katie Wales (University of Leeds) Anthony Warner (University of York)      An Introduction to English Syntax Jim Miller An Introduction to English Phonology April McMahon An Introduction to English Morphology Andrew Carstairs-McCarthy 01 pages i-viii prelims 18/10/01 3:42 pm Page iii An Introduction to English Morphology Words and Their Structure Andrew Carstairs-McCarthy Edinburgh University Press 01 pages i-viii prelims 18/10/01 3:42 pm Page iv To Jeremy © Andrew Carstairs-McCarthy, 2002 Edinburgh University Press Ltd 22 George Square, Edinburgh Typeset in Janson by Norman Tilley Graphics and printed and bound in Great Britain by MPG Books Ltd, Bodmin A CIP Record for this book is available from the British Library ISBN 0 7486 1327 7 (hardback) ISBN 0 7486 1326 9 (paperback) The right of Andrew Carstairs-McCarthy to be identified as author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. Contents Acknowledgements viii 1 Introduction 1 Recommendations for reading 3 2 Words, sentences and dictionaries 4 2.1 Words as meaningful building-blocks of language 4 2.2 Words as types and words as tokens 5 2.3 Words with predictable meanings 6 2.4 Non-words with unpredictable meanings 9 2.5 Conclusion: words versus lexical items 12 Exercises 13 Recommendations for reading 14 3 A word and its parts: roots, affixes and their shapes 16 3.1 Taking words apart 16 3.2 Kinds of morpheme: bound versus free 18 3.3 Kinds of morpheme: root, affix, combining form 20 3.4 Morphemes and their allomorphs 21 3.5 Identifying morphemes independently of meaning 23 3.6 Conclusion: ways of classifying word-parts 26 Exercises 27 Recommendations for reading 27 4 A word and its forms: inflection 28 4.1 Words and grammar: lexemes, word forms and grammatical words 28 4.2 Regular and irregular inflection 31 4.3 Forms of nouns 34 4.4 Forms of pronouns and determiners 38 4.5 Forms of verbs 39 01 pages i-viii prelims 18/10/01 3:42 pm Page vi vi AN INTRODUCTION TO ENGLISH MORPHOLOGY 4.6 Forms of adjectives 40 4.7 Conclusion and summary 42 Exercises 42 Recommendations for reading 43 5 A word and its relatives: derivation 44 5.1 Relationships between lexemes 44 5.2 Word classes and conversion 45 5.3 Adverbs derived from adjectives 48 5.4 Nouns derived from nouns 49 5.5 Nouns derived from members of other word classes 50 5.6 Adjectives derived from adjectives 52 5.7 Adjectives derived from members of other word classes 53 5.8 Verbs derived from verbs 54 5.9 Verbs derived from member of other word classes 55 5.10 Conclusion: generality and idiosyncrasy 56 Exercises 57 Recommendations for reading 58 6 Compound words, blends and phrasal words 59 6.1 Compounds versus phrases 59 6.2 Compound verbs 60 6.3 Compound adjectives 61 6.4 Compound nouns 61 6.5 Headed and headless compounds 64 6.6 Blends and acronyms 65 6.7 Compounds containing bound combining forms 66 6.8 Phrasal words 67 6.9 Conclusion 68 Exercises 68 Recommendations for reading 69 7 A word and its structure 71 7.1 Meaning and structure 71 7.2 Affixes as heads 71 7.3 More elaborate word forms: multiple affixation 72 7.4 More elaborate word forms: compounds within compounds 76 7.5 Apparent mismatches between meaning and structure 79 7.6 Conclusion: structure as guide but not straitjacket 82 Exercises 83 Recommendations for reading 84 01 pages i-viii prelims 18/10/01 3:42 pm Page vii CONTENTS vii 8 Productivity 85 8.1 Introduction: kinds of productivity 85 8.2 Productivity in shape: formal generality and regularity 85 8.3 Productivity in meaning: semantic regularity 88 8.4 Semantic blocking 91 8.5 Productivity in compounding 93 8.6 Measuring productivity: the significance of neologisms 95 8.7 Conclusion: ‘productivity’ in syntax 97 Exercises 98 Recommendations for reading 99 9 The historical sources of English word formation 100 9.1 Introduction 100 9.2 Germanic, Romance and Greek vocabulary 100 9.3 The rarity of borrowed inflectional morphology 102 9.4 The reduction in inflectional morphology 104 9.5 Characteristics of Germanic and non-Germanic derivation 106 9.6 Fashions in morphology 108 9.7 Conclusion: history and structure 110 Exercises 111 Recommendations for reading 113 10 Conclusion: words in English and in languages generally 114 10.1 A puzzle: disentangling lexemes, word forms and lexical items 114 10.2 Lexemes and lexical items: possible reasons for their overlap in English 115 10.3 Lexemes and lexical items: the situation outside English 116 10.4 Lexemes and word forms: the situation outside English 118 Recommendations for reading 119 Discussion of the exercises 120 Glossary 141 References 148 Index 150 Acknowledgements I would like to thank Heinz Giegerich for inviting me to write this book, and him and Laurie Bauer for useful comments on a draft version. I must admit that, when I set out to write what is intended as an introductory text on an extremely well-described language, I did not expect to learn anything new myself; but I have enjoyed discovering and rediscovering both new and old questions that arise from the study of morphology and its interaction with syntax and the lexicon, even if I cannot claim to have provided any conclusive new answers. The Library of the University of Canterbury has, as always, been efficient in supplying research material. I would also like to thank my partner Jeremy Carstairs-McCarthy for constant support and help. viii 02 pages 001-152 18/10/01 3:43 pm Page 1 1 Introduction The term ‘word’ is part of everyone’s vocabulary. We all think we understand what words are. What’s more, we are right to think this, at some level. In this book I will not suggest that our ordinary notion of the word needs to be replaced with something radically different. Rather, I want to show how our ordinary notion can be made more precise. This will involve teasing apart the bundle of ingredients that go to make up the notion, showing how these ingredients interact, and introducing ways of talking about each one separately. After reading this book, you will still go on using the term ‘word’ in talking about language, both in everyday conversation and in more formal contexts, such as literary criticism or English language study; but I hope that, in these more formal contexts, you will talk about words more confidently, knowing exactly which ingredients of the notion you have in mind at any one time, and able where necessary to use appropriate terminology in order to make your meaning absolutely clear. This is a textbook for students of the English language or of English literature, not primarily for students of linguistics. Nevertheless, what I say will be consistent with mainstream linguistic views on word- structure, so any readers who go on to more advanced linguistics will not encounter too many inconsistencies. A good way of teasing apart the ingredients in the notion ‘word’ is by explicitly contrasting them. Here are the contrasts that we will be looking at, and the chapters where they will be discussed: • words as units of meaning versus units of sentence structure (Chapters 2, 6, 7) • words as pronounceable entities (‘word forms’) versus more abstract entities (sets of word forms) (Chapters 3, 4, 5) • inflectionally related word forms (forms of the same ‘word’) versus deriva- tionally related words (different ‘words’ with a shared base) (Chapters 4, 5) 1 02 pages 001-152 18/10/01 3:43 pm Page 2 2 AN INTRODUCTION TO ENGLISH MORPHOLOGY • the distinction between compound words and phrases (Chapters 6, 7) • the relationship between the internal structure of a word and its mean- ing (Chapter 7) • productive versus unproductive word-forming processes (Chapter 8) • historical reasons for some of the contemporary divisions within English morphology, especially Germanic versus Romance word- formation processes (Chapter 9). These various contrasts impact on one another in various ways. For example, if one takes the view that the distinction between compound words and phrases is unimportant, or is even perhaps a bogus distinction fundamentally, this will have a considerable effect on how one views the word as a unit of sentence-structure. Linguistic scholars who specialise in the study of words (so-called ‘morphologists’) devote considerable effort to working out the implications of different ways of formulating these distinctions, as they strive to discover the best way (that is, the most illuminating way, or the way that seems to accord most accurately with people’s implicit knowledge of their native languages). We will not be exploring the technical ramifications of these efforts in this book. Never- theless, I will need to ensure that the way I draw the distinctions here yields a coherent overall picture, and some cross-referencing between chapters will be necessary for that. Each of Chapters 2 to 9 inclusive is provided with exercises. This is designed to make the book suitable for a course extending over about ten weeks. Relatively full discussions of the exercises are also provided at the end of the book. For those exercises that are open-ended (that is, ones for which there is no obvious ‘right’ answer), these discussions serve to illustrate and extend points made in the chapter. As befits a book aimed at students of English rather than linguistics students, references to the technical literature are kept to a minimum. However, the ‘Recommendations for reading’ at the end of each chapter contain some hints for any readers who would like to delve into this literature, as well as pointing towards more detailed treatments of English morphology in particular. Finally, I would like to encourage comments and criticisms. My choice of what to emphasise and what to leave out will inevitably not please everyone, nor will some of the details of what I say. I hope, how- ever, that even those who find things to disagree with in this book will also find it useful for its intended introductory purpose, whether as students, teachers or general readers. 02 pages 001-152 18/10/01 3:43 pm Page 3 INTRODUCTION 3 Recommendations for reading At the end of each chapter are recommendations for reading relating to the subject-matter of the chapter. Here I offer some comments on general works dealing with English or morphology or both. Of the available books on English morphology in particular, Bauer (1983) delves deepest into issues of linguistic theory (although a now somewhat dated version of it), and offers useful discussion and case- studies of fashions in derivational morphology. Marchand (1969) is factually encyclopedic. Adams (1973) concentrates on compounding (the subject-matter of our Chapter 6) and conversion (discussed here in Chapter 5), but says relatively little about derivation (covered here in Chapter 5). There is no book that deals adequately with morphology in general linguistic terms and that also takes into account fully up-to-date versions of syntactic and phonological theory. Bauer (1988) is a clear introduc- tory text. The main strength of Matthews (1991) is its terminological precision. Carstairs-McCarthy (1992) is aimed at readers whose know- ledge of linguistics is at advanced undergraduate level or beyond. Spencer (1991) covers much ground, and may be said to bridge the gap between Bauer and Carstairs-McCarthy. 2 Words, sentences and dictionaries 2.1 Words as meaningful building-blocks of language We think of words as the basic units of language. When a baby begins to speak, the way the excited mother reports what has happened is: ‘Sally (or Tommy) has said her (or his) first word!’ We would be surprised at a mother who described little Tommy’s or Sally’s first utterance as a sentence. Sentences come later, we are inclined to feel, when words are strung together meaningfully. That is not to say that a sentence must always consist of more than one word. One-word commands such as ‘Go!’ or ‘Sit!’, although they crop up relatively seldom in everyday con- versation or reading, are not in any way odd or un-English. Nevertheless, learning to talk in early childhood seems to be a matter of putting words together, not of taking sentences apart. There is a clear sense, then, in which words seem to be the building- blocks of language. Even as adults, there are quite a few circumstances in which we use single words outside the context of any actual or recon- structable sentence. Here are some examples: • warning shouts, such as ‘Fire!’ • conventional commands, such as ‘Lights!’, Camera!’, ‘Action!’ • items on shopping lists, such as ‘carrots’, ‘cheese’, ‘eggs’. It is clear also that words on their own, outside sentences, can be sorted and classified in various ways. A comprehensive classification of English words according to meaning is a thesaurus, such as Roget’s Thesaurus. But the kind of conventional classification that we are likely to refer to most often is a dictionary, in which words are listed according to their spelling in alphabetical order. Given that English spelling is so erratic, a common reason for looking up a word in an English dictionary is to check how to spell it. But another very common reason is to check what it means. In fact, that is what a dictionary entry basically consists of: an association of a word, alphabeti- cally listed, with a definition of what it means, and perhaps also some 4 02 pages 001-152 18/10/01 3:43 pm Page 5 WORDS , SENTENCES AND DICTIONARIES 5 information about grammar (the word class or part of speech that the word belongs to) and its pronunciation. Here, for example, is a specimen dictionary entry for the word month, based on the entry given in the Concise Oxford Dictionary (6th edition): month noun. Any of twelve portions into which the year is divided. It seems, then, that a word is not just a building-block of sentences: it is a building-block with a meaning that is unpredictable, or at least suffi- ciently unpredictable that learners of English, and even sometimes native speakers, may need to consult a dictionary in order to discover it. We may be tempted to think that this constitutes everything that needs to be said about words: they are units of language which are basic in two senses, both 1. in that they have meanings that are unpredictable and so must be listed in dictionaries and 2. in that they are the building-blocks out of which phrases and sentences are formed. However, if that were all that needed to be said, this would be a very short book – much shorter than it actually is! So in what respects do 1. and 2. jointly fall short as a characterisation of words and their behavi- our? A large part of the answer lies in the fact that there are units of language that have characteristic 1. but not 2., and vice versa. Sections 2.3 and 2.4 are devoted to demonstrating this. First, though, we will deal in Section 2.2 with a distinction which, though important, is independent of the distinctions that apply to words in particular. 2.2 Words as types and words as tokens How many words are there in the following sentence? (1) Mary goes to Edinburgh next week, and she intends going to Washington next month. If we take as a guide the English spelling convention of placing a space between each word, the answer seems clearly to be fourteen. But there is also a sense in which there are fewer than fourteen words in the sentence, because two of them (the words to and next) are repeated. In this sense, the third word is the same as the eleventh, and the fifth word is the same as the thirteenth, so there are only twelve words in the sentence. Let us say that the third and the eleventh word of the sentence at (1) are distinct 02 pages 001-152 18/10/01 3:43 pm Page 6 6 AN INTRODUCTION TO ENGLISH MORPHOLOGY tokens of a single type, and likewise the fifth and thirteenth word. (In much the same way, one can say that two performances of the same tune, or two copies of the same book, are distinct tokens of one type.) The type–token distinction is relevant to the notion ‘word’ in this way. Sentences (spoken or written) may be said to be composed of word- tokens, but it is clearly not word-tokens that are listed in dictionaries. It would be absurd to suggest that each occurrence of the word next in (1) merits a separate dictionary entry. Words as listed in dictionaries entries are, at one level, types, not tokens – even though, at another level, one may talk of distinct tokens of the same dictionary entry, inasmuch as the entry for month in one copy of the Concise Oxford Dictionary is a different token from the entry for month in another copy. Is it enough, then, to say that characterisation 2. (words as building- blocks) relates to word-tokens and characterisation 1. (words as mean- ingful units) relates to word-types? Again, if that were all there was to it, this book could be quite short. The term word would be ambiguous between a ‘type’ interpretation and a ‘token’ interpretation; but the ambiguity would be just the same as is exhibited by many other terms not specifically related to language, such as tune: a tune I heard this morning may be ‘the same’ as one I heard yesterday (i.e. they may be instances of the same type), but the two tokens that I have heard of it are distinct. However, the relationship between words as building-blocks and as meaningful units is not so simple as that, as we shall see. So, while it is important to be alert to type–token ambiguity when talking about words, recognising this sort of ambiguity is by no means all there is to sorting out how characteristics 1. and 2. diverge. 2.3 Words with predictable meanings Do any words have meanings that are predictable – that is, meanings that can be worked out on the basis of the sounds or combinations of sounds that make them up? (I consciously say ‘sounds’ rather than ‘letters’ because writing is secondary to speech: every normal human learns to speak, but it is only in the last century or so that a substantial proportion of the world’s population has learned to read and write.) The answer is certainly ‘yes’, but not necessarily for reasons that immediately come to mind. It is true that there are some words whose sound seems to reflect their meaning fairly directly. These include so-called onomatopoeic words, such as words for animal cries: bow-wow, miaow, cheep, cock-a-doodle-doo. But even here convention plays a large part. Onomatopoeic words are not the same in all languages; for example, a cock-crow in German is 02 pages 001-152 18/10/01 3:43 pm Page 7 WORDS , SENTENCES AND DICTIONARIES 7 kikeriki, and a dog’s bark in French is ouah ouah (pronounced roughly ‘wah wah’). There are also sets of words in which some similarity in sound (say, in the cluster of consonants at the beginning) seems to reflect a vague similarity in meaning, such as smoothness or wetness or both in the set of words slip, slop, slurp, slide, slither, sleek, slick, slaver, slug. A technical term for this situation is sound symbolism. But in sound symbolism, quite apart from the role of convention, the sound–meaning relationship is even less direct than in onomatopoeia. The fact that a word begins with sl- does not guarantee that it has anything to do with smoothness or wetness (consider slave, slit, slow), and conversely there are many words that relate to smoothness and wetness but do not begin with sl-. The idea that some words have meanings that are ‘natural’ or pre- dictable in this way is really a leftover from childhood. Young children who have been exposed to only one language are often perplexed when they encounter a foreign language for the first time. ‘Aren’t cat and dog obviously the right words for those animals?’, an English-speaking child may think; ‘Why, then, do French people insist on calling them chat and chien?’ Pretty soon, of course, everyone comes to realise that, in every language including their own, the associations between most words and their meanings are purely conventional. After all, if that were not so, the vocabularies of languages could not differ as much as they do. Even in onomatopoeia and sound symbolism this conventionality is still at work, so that people who know no English are unlikely to predict the meaning of cock-a-doodle-doo or bow-wow any more accurately than they can predict the meaning of cat or dog. What kinds of word do have predictable meanings, then? The answer is: any words that are composed of independently identifiable parts, where the meaning of the parts is sufficient to determine the meaning of the whole word. Here is an example. Most readers of this book have probably never encountered the word dioecious (also spelled diecious), a botanical term meaning ‘having male and female flowers on separate plants’. (It contrasts with monoecious, meaning ‘having male and female flowers, or unisexual flowers, on the same plant’.) If you had been asked the meaning of the word dioecious before today, you would probably have had to look it up in the dictionary. Consider now sentence (2): (2) Ginkgo trees reproduce dioeciously. To work out what this sentence means, do you now need to look up dioeciously in a dictionary? It is, after all, another word that you are encountering here for the first time! Yet, knowing the meaning of dioecious, you will agree (I take it) that a dictionary is unnecessary. You 02 pages 001-152 18/10/01 3:43 pm Page 8 8 AN INTRODUCTION TO ENGLISH MORPHOLOGY can confidently predict that (2) means ‘Ginkgo trees reproduce by means of male and female flowers on separate plants’. Your confidence is based on the fact that, knowing English, you know that the suffix -ly has a consistent meaning, so that Xly means ‘in an X fashion’, for any adjective X. Perhaps up to now you had not realised that you know this; but that merely reflects the fact that one’s knowledge of one’s native language is implicit, not explicit – at least until aspects of it are made explicit through schooling. Dioeciously is an example of a word that, although not brand new (it may even be listed in some dictionaries), could just as well be brand new so far as most readers of this book are concerned. The fact that you could nevertheless understand it (once you had learned the meaning of dioecious, that is) suggests that you should have no difficulty using and understanding many words that really are brand new – words that no one has ever used before. It is easy to show that that is correct. Here are three sentences containing words that, so far as I know, had never been used by anyone before my use of them today, in the year 2000: (3) Vice-President Gore is likely to use deliberately un-Clintonish electioneering tactics. (4) It will be interesting to see how quickly President Putin de- Yeltsinises the Russian government. (5) The current emphasis on rehabilitative goals in judicial punishment may give rise to an antirehabilitationist reaction among people who place more weight on retribution and deterrence. You will have no difficulty interpreting these sentences. Un-Clintonish tactics are tactics unlike those that President Clinton would use, and a de-Yeltsinised government is one purged of the influence of Boris Yeltsin. The word antirehabilitationist may strike you as ugly or cumber- some, but its meaning is likewise clear. In fact, it is virtually inevitable that words with predictable meanings should exist, given that English vocabulary changes over time. If one examines words that first came into use in the twentieth century, one will certainly encounter some that appear from nowhere, so to speak, with meanings that are unguessable from their shape, such as jazz or gizmo. The vast majority, however, are words whose meanings, if not strictly predictable, are at any rate motiv- ated in the sense that they can be reliably guessed by someone who encounters them for the first time in an appropriate context. Examples are computer or quadraphonic or gentrification, all of which have meanings that are sufficiently unpredictable to require listing in any up-to-date dictionary, but none of which would have been totally opaque to an adult English-speaker encountering them when they were first used. 02 pages 001-152 18/10/01 3:43 pm Page 9 WORDS , SENTENCES AND DICTIONARIES 9 What these examples show is that one of the characteristics suggested in Section 2.1 as applicable to all words – that they have meanings that are unpredictable and so must be listed in dictionaries – is not after all totally general. If dioecious and rehabilitation are listed, then dioeciously and antirehabilitationist do not need to be listed as well, at least not if semantic unpredictability is the criterion. And a novel word such as un- Clintonish is perfectly understandable even though the base from which it is formed is a proper name (Clinton) and hence will not be listed in most dictionaries. The link between wordhood, semantic unpredict- ability and dictionary listing is thus less close than you may at first have thought. In Exercise 1 at the end of this chapter you will find further examples of words whose meanings are predictable, alongside words of similar shape whose meaning certainly cannot be guessed. Is it, then, that the common view of words as basic semantic building- blocks of language is simply wrong? That would be too sweeping. What examples such as computer illustrate is that a word’s meaning may be motivated (a computer is certainly used, among other things, for com- puting, that is for performing calculations) but nevertheless idiosyn- cratic (it is not the case, in the early twenty-first century, that anyone or anything that performs calculations can be called a computer). In some instances a word’s original motivation is totally obscured by its pronun- ciation but can still be glimpsed from its spelling, as with cupboard and handkerchief. It is as if words are intrinsically prone to drift semantically, and in particular to acquire meanings that are more specialised than one would predict if one had never encountered them before. Why this should be is a large question, still not fully answered, involving the study of linguistic semantics, of language change, and of how knowledge about words is acquired and stored in the brain. For present purposes, what matters is to be aware that not every word can be listed in a dictionary, even in the fullest dictionary imaginable. 2.4 Non-words with unpredictable meanings In Section 2.3 we saw that it is possible for a linguistic item to be a basic building-block of syntax – that is, an item that is clearly not itself a sentence or a phrase – and yet to have a meaning that is predictable. We saw, in other words, that characteristic 2. does not necessarily entail characteristic 1. In this section we will see that characteristic 1. does not necessarily entail characteristic 2.: that is, something that is clearly larger than a word (being composed of two or more words) may nevertheless have a meaning that is not entirely predictable from the meanings of the words that compose it. 02 pages 001-152 18/10/01 3:43 pm Page 10 10 AN INTRODUCTION TO ENGLISH MORPHOLOGY Consider these two sentences from the point of view of a learner of English who is familiar with the usual meanings of the words expenditure, note and tab: (6) I keep notes on all my expenditure. (7) I keep tabs on all my expenditure. Will this learner be able to interpret both these sentences accurately? The answer, surely, is no. Sentence (6) presents no problem; the learner should be able to interpret it correctly as meaning ‘I write down a record of everything I spend’. But faced with sentence (7), on the basis of the usual meaning of tab, the learner is likely to be puzzled. Does it mean something like ‘I attach small flaps to all the notes and coins that I spend’? Or perhaps ‘I tear off small pieces from the paper money that I spend, and keep them’? Neither interpretation makes much sense! Native speakers of English, however, will have no difficulty with (7). They will instinctively interpret keep tabs on as a single unit, meaning ‘pay close attention to’ or ‘monitor carefully’. Thus, keep tabs on, although it consists of three words, functions as a single unit semantically, its meaning not being predictable from that of these three words individually. In tech- nical terms, keep tabs on is an idiom. Even though it is not a word, it will appear in any dictionary that takes seriously the task of listing semantic idiosyncrasies, probably under the headword tab. Idioms are enormously various in length, structure and function. Keep tabs on behaves rather like a verb, as do take a shine to ‘become attracted to’, raise Cain ‘create a disturbance’, have a chip on one’s shoulder ‘be resent- ful’, and kick the bucket ‘die’. Many idioms behave more like nouns, as the following pair of sentences illustrates: (8) The interrogation took a long time because the suspect kept intro- ducing irrelevant arguments. (9) The interrogation took a long time because the suspect kept intro- ducing red herrings. Again, a learner of English might be puzzled by (9): did the suspect keep pulling fish from his pocket? A native speaker, however, will know that red herring is an idiom meaning ‘irrelevant argument’, so that (8) and (9) mean the same thing. Other noun-like idioms are white elephant ‘unwanted object’, dark horse ‘competitor whose strength is unknown’, Aunt Sally ‘target of mockery’. In most of the idioms that we have looked at so far, all the individual words (tabs, shine, bucket, elephant etc.) have a literal or non-idiomatic meaning in other contexts. Even in raise Cain, the fact that Cain is spelled with a capital letter hints at a reference to the elder son of Adam, 02 pages 001-152 18/10/01 3:43 pm Page 11 WORDS , SENTENCES AND DICTIONARIES 11 who, according to biblical legend, murdered his younger brother Abel. However, there are also words that never occur except in an idiomatic context. Consider these examples: (10) My aunt took pains to get the answer right. (11) My aunt took part in the conversation. (12) My aunt took offence at the suggestion. (13) My aunt took umbrage at the suggestion. (10), (11) and (12), take pains, take part and take offence all deserve to be called idioms, because they are multi-word items whose meaning is not fully predictable from their component words. (To a learner of English, (11) might seem to imply that my aunt was present during only part of the conversation, and (12) might suggest that she committed an offence.) If so, then presumably we should say the same of (13), containing the phrase take umbrage at. The difference between (13) and the others, how- ever, is that umbrage does not appear anywhere except in this phrase (in my usage, at least). This restriction means that it would not really be sufficient for a dictionary to list umbrage as a noun meaning something like ‘annoyance’; rather, what needs to be listed is the whole phrase. Similarly, the word cahoots exists only in the phrase in cahoots with ‘in collusion with’, and it is the whole phrase which deserves to be lexically listed, as an idiom. Akin to idioms, but distinguishable from them, are phrases in which individual words have collocationally restricted meanings. Consider the following phrases: (14) white wine (15) white coffee (16) white noise (17) white man Semantically, these phrases are by no means totally idiosyncratic: they denote a kind of wine, coffee, noise and man, respectively. Nevertheless, in a broad sense they may count as idiomatic, because the meaning that white has in them is not its usual meaning; rather, when collocated with wine, coffee, noise and man respectively, it has the meanings ‘yellow’, ‘brown (with milk)’ (at least in British usage), ‘containing many fre- quencies with about equal amplitude’, and ‘belonging to an ethnic group whose members’ skin colour is typically pinkish or pale brown’. If a typical idiom is a phrase, then a word with a collocationally restricted meaning is smaller than a typical idiom. That provokes the question whether there are linguistic items with unpredictable mean- ings that are larger than phrases – specifically, that constitute whole