Quizgecko Grammar (2002) PDF

1 Preliminaries Geoffrey K. Pullum Rodney Huddleston 1 The aim of this book 2 2 Prescriptivism, tradition, and the justification of grammars 5 2.1 Prescriptive and descriptive approaches: goals and coverage 5 2.2 Disagreement between descriptivist and prescriptivist work 6 3 Speech and writing 11 3.1 The representation of English pronunciation 13 3.1.1 Rhotic and non-rhotic accents 13 3.1.2 An accent-neutral phonological representation 14 3.2 Pronunciation and spelling 17 4 Theoretical framework 18 4.1 Description and theory 18 4.2 Basic concepts in syntax 20 4.2.1 Constituent structure 20 4.2.2 Syntactic categories 21 4.2.3 Grammatical constructions and functions 23 4.3 Morphology, inflectional and lexical 26 4.4 Defining grammatical concepts 28 5 Semantics, pragmatics, and meaning relations 33 5.1 Truth conditions and entailment 34 5.2 Non-truth-conditional aspects of sentence meaning 36 5.3 Pragmatics and conversational implicatures 37 5.4 Pragmatic presupposition 40 1 2 1 The aim of this book This book is a description of the grammar of modern Standard English, providing a detailed account of the principles governing the construction of English words, phrases, clauses, and sentences. To be more specific, we give a synchronic, descriptive grammar of general-purpose, present-day, international Standard English. Synchronic versus diachronic description A synchronic description of a language is a snapshot of it at one point in time, the opposite of a diachronic or historical account. English has a rich history going back over a millennium, but it is not the aim of this book to detail it. We include only a few notes on historical points of interest that will assist the reader to understand the present state of the language. Of course, at any given moment English speakers with birthdates spread over about a century are alive, so the idea of English as it is on one particular day is a fiction: the English used today was learned by some speakers at the end of the twentieth cen- tury and by others near the beginning. But our practice will be to illustrate relevant points mainly with examples of use of the language taken from prose produced since the mid twentieth century. Examples from earlier periods are used only when particularly apposite quotations are available for a point on which the language has not subse- quently changed. Wherever grammatical change has clearly occurred, our aim will be not to describe the evolutionary process but rather to describe the current state of the language. Description versus prescription Our aim is to describe and not prescribe: we outline and illustrate the principles that govern the construction of words and sentences in the present-day language without recommending or condemning particular usage choices. Although this book may be (and we certainly hope it will be) of use in helping the user decide how to phrase things, it is not designed as a style guide or a usage manual. We report that sentences of some types are now widely found and used, but we will not advise you to use them. We state that sentences of some types are seldom encountered, or that usage manuals or language columnists or language teachers recommend against them, or that some form of words is normally found only in informal style or, conversely, is limited to rather formal style, but we will not tell you that you should avoid them or otherwise make recommendations about how you should speak or write. Rather, this book offers a description of the context common to all such decisions: the linguistic system itself. § 1 The aim of this book 3 General-purpose versus special-purpose We exclude from consideration what we refer to as special-purpose varieties of the language. Newspaper headlines, road signs, notices, and the like have their own special styles of abbreviation (Man bites dog, arrested; EXIT ONLY THIS LANE), and we do not provide a full treatment of the possibilities. Likewise, we do not provide a description of any special notations (chemical formulae, telephone numbers, email addresses) or of the special language found in poetry, heraldic descriptions, scientific works, chemical compound naming, computer jargon, mathematical proofs, etc. To some small extent there may be idiosyncratic grammatical patterns found in such areas, but we generally set them aside, avoiding complicated digressions about usages found within only a very narrow range of discourse. Present-day English versus earlier stages Modern English is generally defined by historians of English to be the English used from 1776 onwards. The recent part of the latter period (say, since the Second World War) can be called Present-day English. Linguistic changes have occurred in the grammar of English during the Modern English period, and even during the last half-century. Our central aim is to describe Present-day English in its standard form. This means, for example, that we treat the pronoun system as not containing a contrast between familiar and respectful 2nd person pronouns: the contrast between thou and you has been lost, and we do not mention thou in this grammar. Of course, this does not mean that people who use thou (actors in period plays, people addressing God in prayers, or Quakers who have retained the older usage) are making a mistake; but they are not using the general-purpose standard Present-day English described in this book. Grammar versus other components A grammar of a language describes the principles or rules governing the form and meaning of words, phrases, clauses, and sentences. As such, it interacts with other com- ponents of a complete description: the phonology (covering the sound system), the graphology (the writing system: spelling and punctuation), the dictionary or lexicon, and the semantics. Phonology and graphology do not receive attention in their own right here, but both have to be treated explicitly in the course of our description of inflection in Ch. 18 (we introduce the concepts that we will draw on in §3 of this chapter), and Ch. 20 deals with one aspect of the writing system in providing an outline account of the important system of punctuation. A lexicon for a language deals with the vocabulary: it brings together information about the pronunciation, spelling, meaning, and grammatical properties of the lexical items – the words, and the items with special meanings that consist of more than one word, the idioms. The study of conventional linguistic meaning is known as semantics. We take this to cut across the division between grammar and lexicon. That is, we distinguish between lexical semantics, which dictionaries cover, and grammatical semantics. Our account of grammatical meaning will be quite informal, but will distinguish between semantics (dealing with the meaning of sentences or words as determined by the language system itself) and pragmatics (which has to do with the use and interpretation of sentences 4 Chapter 1 Preliminaries as used in particular contexts); an introduction to these and other concepts used in describing meaning is given in §5 of this chapter. A grammar itself is divisible into two components, syntax and morphology. Syntax is concerned with the way words combine to form phrases, clauses, and sentences, while morphology deals with the formation of words. This division gives special prominence to the word, a unit which is also of major importance in the lexicon, the phonology and the graphology. Standard versus non-standard Perhaps the most subtle concept we have to rely on is the one that picks out the partic- ular variety of Present-day English we describe, which we call Standard English. Briefly (for we will return to the topic below), we are describing the kind of English that is widely accepted in the countries of the world where English is the language of gov- ernment, education, broadcasting, news publishing, entertainment, and other public discourse. In a large number of countries (now running into scores), including some where most of the people have other languages as their first language, English is used for most printed books, magazines, newspapers, and public notices; for most radio and televi- sion broadcasting; for many or most film scripts, plays, poetry, and other literary art; for speeches, lectures, political addresses, proclamations, official ceremonies, advertise- ments, and other general announcements. In these countries there is a high degree of consensus about the appropriate variety of English to use. The consensus is confirmed by the decisions of broadcasting authorities about the kind of English that will be used for public information announcements, newscasts, commentaries to broadcasts of na- tional events such as state funerals, and so on. It is confirmed by the writing found in magazines, newspapers, novels, and non-fiction books; by the editing and correcting that is done by the publishers of these; and by the way writers for the most part accept such editing and correcting of their work. This is not to say that controversy cannot arise about points of grammar or usage. There is much dispute, and that is precisely the subject matter for prescriptive usage man- uals. Nonetheless, the controversy about particular points stands out against a backdrop of remarkably widespread agreement about how sentences should be constructed for such purposes as publication, political communication, or government broadcasting. This widespread agreement defines what we are calling Standard English. National versus international Finally, we note that this book is not intended to promote any particular country’s variety of Standard English as a norm; it is to apply internationally. English is the single most important language in the world, being the official or de facto language of the United Kingdom, the United States of America, Canada, Australia, New Zealand, South Africa, and dozens of others, and being the lingua franca of the Internet. Many varieties of English are spoken around the world – from lectures in graduate schools in Holland to parliamentary proceedings in Papua New Guinea – but interestingly the vast majority of the variation lies in pronunciation and vocabulary. The num- ber of differences in grammar between different varieties of Standard English is very § 2 Prescriptivism, tradition, and justification of grammars 5 small indeed relative to the full range of syntactic constructions and morphological word-forms. Nevertheless, there undoubtedly are differences of this kind that need to be noted. For example, the use of the verb do following an auxiliary verb, as in % I’m not sure that I’ll go, but I may do is not found in American English, and conversely the past participle verb-form gotten, as in % I’ve just gotten a new car, is distinctively American. We use the symbol ‘% ’ to mark constructions or forms that are restricted to some dialect or dialects in this way. The regional dialects of Standard English in the world today can be divided into two large families with regional and historical affinities. One contains standard edu- cated Southern British English, henceforth abbreviated BrE, together with a variety of related dialects, including most of the varieties of English in Great Britain, Australia, New Zealand, South Africa, and most other places in the British Commonwealth. The second dialect family we will refer to as American English, henceforth AmE – it contains the dialects of the United States, Canada, and associated territories, from Hawaii and Alaska to eastern Canada. 2 Prescriptivism, tradition, and the justification of grammars The topic of prescriptivism and its relation to the long tradition of English grammatical scholarship needs some further discussion if the basis of our work, and its relation to other contributions to the field, is to be properly understood. It relates to the issue of how the statements of a grammar are justified: what the support for a claimed grammatical statement might be. 2.1 Prescriptive and descriptive approaches: goals and coverage The distinction between the prescriptive and descriptive approaches to grammar is often explained by saying that prescriptivists want to tell you how you ought to speak and write, while descriptivists want to tell you how people actually do speak and write. This does bring out the major difference between the two approaches: it is a difference in goals. However, it is something of an oversimplification, because writing a descriptive grammar in practice involves a fair amount of idealisation: we need to abstract away from the errors that people make, especially in speech (this point is taken up again in §3 below). In addition, it glosses over some significant differences between the kinds of works prescriptivists and descriptivists characteristically produce. Differences in content The basic difference in goals between prescriptive and descriptive works goes hand in hand with a striking difference in topics treated. The subject matters overlap, but many topics dealt with by prescriptive works find no place in a descriptive grammar, and some topics that must be treated in a descriptive grammar are universally ignored by prescriptive works. The advice of prescriptivists is supplied in works of a type we will refer to as usage manuals. They are almost invariably arranged in the style of a dictionary, containing an 6 Chapter 1 Preliminaries alphabetically arranged series of entries on topics where the issue of what is correct or acceptable is not altogether straightforward. In the first few pages of one usage manual we find entries on abacus (should the plural be abaci ?), abbreviations (which ones are acceptable in formal writing?), abdomen (is the stress on the second syllable or the first?), abduction (how does it differ in meaning from kidnapping?), and so on. These points concern inflection, formal writing, pronunciation, and meaning, respectively, and on all of them a degree of variation and occasional uncertainty is encountered even among expert users of English. Not all of them would belong in a grammatical description. For example, our grammar does cover the plural of abacus (Ch. 18, §4.1.6), but it does not list abbreviations, or phonological topics like the placement of stress in English words, or lexical semantic topics like the distinction between abduction and kidnapping. These we take to be in the province of lexicon – matters for a dictionary rather than a grammar. Usage manuals also give a great deal of attention to matters of style and effective ex- pression that lie beyond the range of grammar as we understand it. Thus one prescriptive usage dictionary warns that explore every avenue is a tired cliché (and adds that it makes little sense, since exploration suggests a more challenging environment than an avenue); that the phrase in this day and age ‘should be avoided at all costs’; that circling round is tautologous (one can only circle by going round) and thus should not be used; and so on. Whether or not one thinks these are good pieces of advice, we do not take them to fall within the realm of grammar. A sentence like In this day and age one must circle round and explore every avenue may be loaded with careworn verbiage, or it may even be arrant nonsense, but there is absolutely nothing grammatically wrong with it. There are also topics in a descriptive grammar that are uniformly ignored by prescrip- tivists. These include the most salient and well-known principles of syntax. Prescriptive works tend to be highly selective, dealing only with points on which people make mis- takes (or what are commonly thought to be mistakes). They would never supply, for example, the grammatically important information that determinatives like the and a precede the noun they are associated with (the house, not ∗house the),1 or that modal auxiliaries like can and must are disallowed in infinitival clauses (∗I’d like to can swim is ungrammatical), or that in subordinate interrogative clauses the interrogative element comes at the front (so we get She asked what we needed, not ∗She asked we needed what). Native speakers never get these things wrong, so no advice is needed. 2.2 Disagreement between descriptivist and prescriptivist work Although descriptive grammars and prescriptive usage manuals differ in the range of topics they treat, there is no reason in principle why they should not agree on what they say about the topics they both treat. The fact they do not is interesting. There are several reasons for the lack of agreement. We deal with three of them here: (a) the basis in personal taste of some prescriptivist writers’ judgements; (b) the confusion of informality with ungrammaticality; and (c) certain invalid arguments sometimes appealed to by prescriptivists. These are extraneous features of prescriptive writing about language rather than inherent ones, and all three of them are less prevalent now than they were 1 Throughout this book we use an asterisk to indicate that what follows is ungrammatical. § 2.2 Disagreement – descriptivist vs prescriptivist 7 in the past. But older prescriptive works have exemplified them, and a few still do; their influence lingers on in the English-speaking educational world. (a) Taste tyranny Some prescriptivist works present rules that have no basis in the way the language is actually used by the majority of its native speakers, and are not even claimed to have any such basis – as though the manual-writer’s own judgements of taste took precedence over those of any other speaker of the language. They expect all speakers to agree with their judgements, no matter what the facts of language use might show. For example, one usage manual, discussing why it is (supposedly) incorrect to say You need a driving instructor who you have confidence in, states that ‘The accusative whom is necessary with the preposition in, though whom is a word strangely shunned by most English people.’ We take the implication to be that English people should not shun this word, since the writer (who is English) does not. But we are inclined to ask what grounds there could be for saying that whom is ‘necessary’ if most English people (or speakers of the English language) would avoid it. The same book objects to centre (a)round, calling it incorrect, although ‘probably more frequently used than the correct centre on’. Again, we wonder how centre (a)round can be determined to be incorrect in English if it is indeed more commonly used by English speakers than what is allegedly correct. The boundary would appear to have been drawn in the wrong place. Prescriptive works instantiating this kind of aesthetic authoritarianism provide no answer to such obvious questions. They simply assert that grammar dictates things, with- out supporting their claim from evidence. The basis for the recommendations offered appears to lie in the writer’s taste: the writer quoted above simply does not like to see who used where it is understood as the object of a preposition, and personally hates the expression centre around. What is going on here is a universalising of one person’s taste, a demand that everyone should agree with it and conform to it. The descriptivist view would be that when most speakers use a form that our grammar says is incorrect, there is at least a prima facie case that it is the grammar that is wrong, not the speakers. And indeed, even in the work just quoted we find the remark that ‘Alright is common, and may in time become normal’, an acknowledgement that the language may change over time, and what begins as an isolated variant on a pattern may eventually become the new pattern. The descriptive grammarian will always adopt a stance of something more like this sort, thus making evidence relevant to the matter at hand. If what is involved were a matter of taste, all evidence would be beside the point. But under the descriptive viewpoint, grammar is not a matter of taste, nor of aesthetics. This is not to say that the expression of personal aesthetic judgements is without utility. The writer of a book on usage might be someone famous for brilliant use of the language, someone eminently worthy of being followed in matters of taste and literary style. It might be very useful to have a compendium of such a person’s pref- erences and recommendations, and very sensible for a less expert writer to follow the recommendations of an acknowledged master of the writer’s craft (assuming such rec- ommendations do reliably accord with the master’s practice). All we are pointing out is that where the author of an authoritarian usage manual departs from recommendations that agree with the way most people use the language, prescriptivist and descriptivist 8 Chapter 1 Preliminaries accounts will necessarily disagree. The authoritarian prescriptivist whose recommen- dations are out of step with the usage of others is at liberty to declare that they are in error and should change their ways; the descriptivist under the same circumstances will assume that it is precisely the constant features in the usage of the overwhelming majority that define what is grammatical in the contemporary language, and will judge the prescriptivist to be expressing an idiosyncratic opinion concerning how the language ought to be. (b) Confusing informal style with ungrammaticality It has been a common assumption of prescriptivists that only formal style is grammat- ically correct. The quotation about whom given above is representative of this view, for whom can be a marker of relatively formal style, being commonly replaced by who in informal style (see Ch. 5, §16.2.3, for a detailed account of the use of these two forms). There are two related points to be made here. The first is that it is important to distinguish between the two contrasts illustrated in the following pairs: i a. It is clear whom they had in mind. b. It’s clear who they had in mind. ii a. Kim and I saw the accident. b. ! Kim and me saw the accident. In [i], both versions belong to Standard English, with [a] somewhat formal, and [b] neutral or slightly informal. There is no difference in grammaticality. But in [ii], the [a] version is standard, the [b] version non-standard; we use the ‘! ’ symbol to mark a construction or form as ungrammatical in Standard English but grammatical in a non-standard dialect. Construction [iib] will be heard in the speech of speakers of di- alects that have a different rule for case inflection of pronouns: they use the accusative forms (me, him, her, us, them) whenever the pronoun is coordinated. Standard English does not. A common view in the prescriptivist tradition is that uses of who like [1ib] are not grammatically correct but are nevertheless ‘sanctioned by usage’. For example, Fowler, one of the most influential prescriptivists of the twentieth century, wrote: ‘The in- terrogative who is often used in talk where grammar demands whom, as in Who did you hear that from? No further defence than “colloquial” is needed for this.’ This im- plies a dichotomy between ‘talk’ and ‘grammar’ that we reject. The standard language embraces a range of styles, from formal through neutral to informal. A satisfactory grammar must describe them all. It is not that formal style keeps to the rules and infor- mal style departs from them; rather, formal and informal styles have partially different rules. (c) Spurious external justifications Prescriptive grammarians have frequently backed up their pronouncements with appeals to entirely extraneous considerations. Some older prescriptive grammars, for example, give evidence of relying on rules that would be better suited to the description of classical languages like Latin than to Present-day English. Consider, for example, the difference between the uses of accusative and nominative forms of the personal pronouns seen in: a. It is I. b. It’s me. With who and whom in [1i] we saw a construction where an accusative form was associ- ated with relatively formal style. In , however, it is the sentence with the nominative § 2.2 Disagreement – descriptivist vs prescriptivist 9 form I that belongs to (very) formal style, while accusative me is neutral or informal (again, see Ch. 5, §16.2.1 for a fuller description of the facts). Confusing informality with ungrammaticality again, a strong prescriptivist tradition says that only [2a] is grammat- ical. The accusative me is claimed to be the case of the direct object, as in It hurt me, but in the noun phrase after the verb is a predicative complement. In Latin, predicative complements take nominative, the same case as the subject. An assumption is being made that English grammar too requires nominative case for predicative complements. Use of the accusative me is regarded as a departure from the rules of grammar. The mistake here, of course, is to assume that what holds in Latin grammar has to hold for English. English grammar differs on innumerable points from Latin gram- mar; there is no reason in principle why the assignment of case to predicative comple- ments should not be one of them. After all, English is very different from Latin with respect to case: the nominative–accusative contrast applies to only a handful of pro- nouns (rather than to the full class of nouns, as in Latin). The right way to describe the present situation in Standard English (unlike Latin) is that with the pronouns that have a nominative–accusative case distinction, the choice between the cases for a pred- icative complement noun phrase varies according to the style level: the nominative is noticeably formal, the accusative is more or less neutral and always used in informal contexts. Another kind of illegitimate argument is based on analogy between one area of gram- mar and another. Consider yet another construction where there is variation between nominative and accusative forms of pronouns: a. They invited me to lunch. b. % They invited my partner and I to lunch. The ‘% ’ symbol is again used to mark the [b] example as typically used by some speakers of Standard English but not others, though this time it is not a matter of regional variation. The status of the construction in [b] differs from that of It’s me, which is undisputedly normal in informal use, and from that of !Me and Kim saw her leave, which is unquestionably non-standard. What is different is that examples like [b] are regularly used by a significant proportion of speakers of Standard English, and not generally thought by ordinary speakers to be non-standard; they pass unnoticed in broadcast speech all the time. Prescriptivists, however, condemn the use illustrated by [3b], insisting that the ‘cor- rect’ form is They invited my partner and me to lunch. And here again they seek to justify their claim that [3b] is ungrammatical by an implicit analogy, this time with other situ- ations found in English, such as the example seen in [a]. In [a] the pronoun functions by itself as direct object of the verb and invariably appears in accusative case. What is different in [b] is that the direct object of the verb has the form of a coordination, not a single pronoun. Prescriptivists commonly take it for granted that this difference is irrelevant to case assignment. They argue that because we have an accusative in [a] we should also have an accusative in [b], so the nominative I is ungrammatical. But why should we simply assume that the grammatical rules for case assignment cannot differentiate between a coordinated and a non-coordinated pronoun? As it hap- pens, there is another place in English grammar where the rules are sensitive to this distinction – for virtually all speakers, not just some of them: a. I don’t know if you’re eligible. b. ∗ I don’t know if she and you’re eligible. 10 Chapter 1 Preliminaries The sequence you are can be reduced to you’re in [a], where you is subject, but not in [b], where the subject has the form of a coordination of pronouns. This shows us not only that a rule of English could apply differently to pronouns and coordinated pronouns, but that one rule actually does. If that is so, then a rule could likewise dis- tinguish between [3a] and [3b]. The argument from analogy is illegitimate. Whether [3b] is treated as correct Standard English or not (a matter that we take up in Ch. 5, §16.2.2), it cannot be successfully argued to be incorrect simply by virtue of the analogy with [3a]. The claim that [1ib] (It’s clear who they had in mind) is ungrammatical is supported by the same kind of analogical reasoning. In They had me in mind, we have accusative me, so it is assumed that the grammar likewise requires accusative whom. The assump- tion here is that the rules of case assignment are not sensitive to the difference in the position of the pronoun (after the verb for me, at the beginning of the clause for who), or to the difference between interrogative and personal pronouns. There is, however, no basis for assuming that the rules of grammar cannot make reference to such dif- ferences: the grammar of English could assign case to clause-initial and non-clause- initial pronouns, or to interrogative and non-interrogative pronouns, in slightly different ways.2 We should stress that not all prescriptive grammarians exhibit the shortcomings we have just catalogued – universalising taste judgements, confusing informality with ungrammaticality, citing spurious external justifications, and arguing from spurious analogies. There are usage manuals that are accurate in their understanding of the facts, clear-sighted in their attitudes towards usage trends, and useful in their recommenda- tions; such books can be an enormous help to a writer. But the good prescriptive manuals respect a crucial tenet: that their criterion should always be the use of the standard language by its native speakers. As we have said, to some extent good usage manuals go far beyond grammar into style, rhetoric, and communication, giving advice about which expressions are over- used clichés, or fail to make their intended point, or are unintentionally ambiguous, or perpetuate an unfortunate malapropism, or any of a large number of other matters that lie beyond the scope of this book. But when it comes to points of grammar, the only legitimate basis for an absolute judgement of incorrectness in a usage manual is that what is being rejected is not in the standard language. The aspects of some prescriptivist works that we have discussed illustrate ways in which those works let their users down. Where being ungrammatical is confused with merely being informal, there is a danger that the student of English will not be taught how to speak in a normal informal way, but will sound stilted and unnatural, like an inexpert reader reading something out from a book. And where analogies are used uncritically to predict grammatical properties, or Latin principles are taken to guarantee correct use of English, the user is simply being misled. 2 A further type of invalid argument that falls under the present heading confuses grammar with logic. This is illustrated in the remarkably widespread but completely fallacious claim that non-standard ! I didn’t see nobody is intrinsically inferior to standard I didn’t see anybody because the two negatives cancel each other out. We discuss this issue in Ch. 9, §6.2. § 3 Speech and writing 11 The stipulations of incorrectness that will be genuinely useful to the student are those about what is actually not found in the standard language, particularly with re- spect to features widely recognised as characteristic of some definitely non-standard dialect. And in that case evidence from use of Standard English by the people who speak it and write it every day will show that it is not regularly used, which means prescriptive and descriptive accounts will not be in conflict, for evidence from use of the language is exactly what is relied upon by descriptive grammars such as we present here. The evidence we use comes from several sources: our own intuitions as native speakers of the language; the reactions of other native speakers we consult when we are in doubt; data from computer corpora (machine-readable bodies of naturally occurring text),3 and data presented in dictionaries and other scholarly work on grammar. We alternate between the different sources and cross-check them against each other, since intuitions can be misleading and texts can contain errors. Issues of interpretation often arise. But always, under the descriptive approach, claims about grammar will depend upon evidence. 3 Speech and writing There are significant and interesting differences between spoken and written language, but we do not regard written English as a different language from spoken English. In general, we aim to describe both the written standard variety that is encountered in contemporary newspapers, magazines, and books and the spoken standard variety that is heard on radio and television programmes in English-speaking countries. ‘Speaker’ and ‘utterance’ as medium-neutral terms Most of what we say will apply equally to the spoken and written varieties of the language. As there is no non-technical term covering both one who utters a sentence in speech and one who writes a sentence, we will follow the widespread practice in linguistics of extending the ordinary sense of ‘speaker’ so as to subsume ‘writer’ – a practice that reflects the fact that speech is in important respects more basic than writing.4 We likewise take ‘utterance’ to be neutral between the mediums, so that we will refer to both spoken and written utterances. Practical bias towards written English Despite our neutrality between speech and writing in principle, there are at least three reasons why the reader may perceive something of a bias in this work towards data from 3 The computer corpora that we have made use of are the Brown corpus of a million words of American English; the Lancaster/Oslo/Bergen (LOB) corpus of British English; the Australian Corpus of English (ACE); and the Wall Street Journal corpus distributed by the Association for Computational Linguistics. The British National Corpus (BNC) was only released to scholars working outside the UK after the book was in final draft. We have also drawn on a variety of other sources, including collections of our own from sources such as magazines, newspapers, plays, books, and film scripts. 4 Since our discussion of sentences will very often make reference to the way they are used we will have very frequent occasion to talk of speakers, and in order to avoid repeatedly using the term ‘speaker’ we will often simply use the 1st person pronoun I. Given that the book has joint authorship this pronoun could not be used in reference to any specific person, and hence is available as a convenient variant of ‘the speaker’. 12 Chapter 1 Preliminaries written English. To the extent that it is present, it stems from practical considerations rather than matters of principle. We will discuss here the three factors motivating the choices we have made. Citation of forms and examples First, we normally follow the usual practice in grammars of citing words or sentences in their written form. This is mainly a matter of practical convenience: it is much more straightforward typographically, and more widely accessible to readers, to supply exam- ples in this form. In certain cases – as, for example, in describing the inflectional forms of verbs and nouns in Ch. 18 – it is necessary to indicate the pronunciation, and for this purpose we use the system of transcription described in §3.1.2 below. Representations in written form are given in italics, while phonological representations are enclosed in obliques. Accessibility of print sources Second, we make frequent use of genuinely attested examples (often shortened or other- wise modified in ways not relevant to the point at issue), and it is significantly easier to obtain access to suitable large collections, or corpora, of written data in a conveniently archived and readily searchable form than it is for speech. Error rates in speech Third, and most importantly, it must be acknowledged that the error content of spoken material is higher than that of written material. Those who have listened to tape record- ings of spontaneous conversation are likely to have been struck by the high incidence of hesitation noises, false starts, self-corrections, repetitions, and other dysfluencies found in the speech of many people. It is not hard to see why speech contains a higher number of errors than writing. The rapid production of speech (quite often several words per second) leaves little time for reflection on construction choices or planning of sentence structure, so that at normal conversational pace slip-ups of the kind mentioned are very common. As a result, what speakers actually come out with reflects only imperfectly the system that defines the spoken version of the language. Hardly noticed by the listener, and often compensated for by virtually unconscious repair strategies on the part of the speaker, these sporadic interruptions and imperfections in speech production are inher- ently outside the purview of the grammarian (the discipline of psycholinguistics studies them in order to learn about the planning, production, and perception of speech). They therefore have to be screened out through judicious decision-making by a skilled na- tive speaker of the language before grammatical description is attempted. The original speaker is not always available for the tedious editing task, and so someone else has to interpret the transcript and remove the apparent errors, which means that misunder- standings can result (word sequences that were actually due to slips might be wrongly taken to represent grammatical facts). Written English has the advantage that its slow rate of composition has generally allowed time and opportunity for nearly all these slips and failures of execution to be screened out by the actual author of the sentence. This provides a practical reason for us to show a preference for it when selecting illustrative examples: we have very good reason to believe that what ultimately gets printed corresponds fairly closely to what the writer intended to say. § 3.1 The representation of English pronunciation 13 The nature of the written medium and the slower sentence-planning environment permits the construction of longer sentences than typically occur in speech, but we take this to be a matter of degree, not a matter of written English instantiating new possibilities that are completely absent from the spoken language. The basic point of most written material is that people who are ordinary native speakers of the language should read it and understand it, so the pressure will always be in the direction of keeping it fairly close to the language in which (ignoring the speech errors referred to above) ordinary people talk to each other. Thus while we acknowledge a tendency for the exemplification in this grammar to be biased towards written English, we assume that the goal of providing a description that is neutral between spoken and written English is not an unreasonable one. Sharp divergences between the syntax of speech and the syntax of writing, as opposed to differences that exist between styles within either the spoken or the written language, are rare to the point of non-existence. 3.1 The representation of English pronunciation This section provides an introduction to the system of representation we use in this book in those cases where it is necessary to indicate the pronunciation of words or word sequences. Developing a system that will be readily usable by non-specialists is by no means a trivial enterprise; English has a remarkably complex vowel system compared to most other languages, and one of the most complex patterns of fit between sound and spelling found in any language. Taken together, these facts raise some significant and unavoidable difficulties even if only one variety of English is under consideration. But an additional problem is that English is a global language with something like 400 million native speakers pronouncing the language in many different ways: pronunciation differs across the world more than any other aspect of the language. 3.1.1 Rhotic and non-rhotic accents We will use the term accent for varieties of a language distinguished by pronunciation, opposing it to dialect, which applies to varieties distinguished by grammar or vocabulary. The most important accent distinction in English concerns the sound we represent as /r/. Most speakers in the BrE family of dialects have a non-rhotic accent: here /r/ occurs in pre-vocalic position, i.e. when immediately preceding a vowel, as in run or area, but not in post-vocalic position, after the vowel of a syllable. For example, in a non-rhotic accent there is no /r/ in any of the words in (as pronounced in isolation): i a. mar, bear, floor, stir, actor b. care, hire, bore, sure, cure ii a. hard, torque, term, burn b. hammered The words in [i] all end in a vowel sound, while those in [ii] end in a vowel followed by just one consonant sound; note that the letter e at the end of the words in [ib] and of torque in [iia], and also that before the d in [iib] are ‘silent’ – i.e. there is no vowel in this position in the spoken form. In many of the non-rhotic accents such pairs of words as mar and ma, floor and flaw, or torque and talk are pronounced the same. A non-rhotic accent is thus one which lacks post-vocalic /r/. 14 Chapter 1 Preliminaries Most speakers in the AmE family of dialects, by contrast, have a rhotic accent, where there is no such restriction on the distribution of /r/: all the words in are pronounced with an /r/ sound after the (final) vowel, or (in the case of stir and term) with a rhotacised (‘r -coloured’) vowel sound, a coalescence of /r/ with the vowel.5 The English spelling system reflects the pronunciation of rhotic accents: in non-rhotic accents post-vocalic /r/ has been lost as a result of a historical change that took place after the writing system became standardised. Linking and intrusive /r/ A further difference between non-rhotic and rhotic accents is seen in the pronunciation of such words and word sequences as those given in , where we use the symbol ‘·’ to mark grammatical boundaries within a word (in these examples, between base and suffix): i a. marr·ing, sur·est, soar·ing b. the fear of death ii a. saw·ing, thaw·ing b. the idea of death In non-rhotic accents the words in [ia] are all pronounced with /r/: the dropping of post-vocalic /r/ in the words mar, sure, soar does not apply here because the addition of a suffix beginning with a vowel makes the /r/ at the end of the base pre-vocalic. Similarly the word sequence [ib] is usually pronounced with an /r/ at the end of fear because the initial vowel of the next word makes it pre-vocalic. The /r/ in pronunciations of [2i] in non-rhotic accents is called a linking /r/. Within a word, as in [ia], linking /r/ is obligatory; in word boundary position, as in [ib], the /r/ is optional though strongly preferred in most styles of speech. In [ii], where there is no r in the spelling, an /r/ pronounced at the end of the bases saw· and thaw· or of the word idea is called an intrusive /r/. Word-boundary intrusive /r/ in the pronunciation of sequences like [iib] is very common; word-internal intrusive /r/ in words like those in [iia] is much less common and quite widely disapproved of. Rhotic accents do not have intrusive /r/ at all: they maintain a sharp distinction between [2i] and [ii], with /r/ appearing only in the former. And although they pronounce /r/ in the forms in [i], this is not linking /r/, since the bases mar, sure, soar, and fear have /r/ in these accents even when not followed by a vowel. 3.1.2 An accent-neutral phonological representation Where we need to give pronunciations of words or larger expressions, it would be incon- sistent with our goals to confine ourselves to one accent, but to attempt a complete listing of the pronunciations in each significant regional or other variety would be tedious. We therefore present here a unitary way of representing pronunciations for major BrE and AmE accents, whether rhotic or non-rhotic. For this purpose it is necessary to indicate more distinctions than would be needed in a system constructed for any one accent. In 5 The correlation between the rhotic vs non-rhotic accent distinction and that between the BrE and AmE family of dialects is not perfect. Ireland, Scotland, the west of England, and some English-speaking Caribbean countries have rhotic accents and yet belong to the BrE family, and, conversely, there are various non-rhotic accents within the United States, including some working-class northeastern varieties and some upper-class southeastern varieties. The term ‘rhotic’ derives from the Greek name of the letter r. § 3.1.2 An accent-neutral phonological representation 15 particular, since it cannot be determined from the pronunciation in a non-rhotic accent where post-vocalic /r/ would occur in a rhotic one (for example, southern British English has /tɔk/ for both torque and talk), post-vocalic /r/ will have to be shown in some way even though it is not pronounced in the non-rhotic accents. Other differences have to be dealt with similarly. The system we adopt is set out in , with illustrative examples in which the letter or letter sequence that symbolises the sound in question is underlined. Some notes on the system follow below. short vowels ɒ odd, lot, lost e get, fell, friend, endeavour æ gas, fat, pan i happy, pennies, maybe gut, much, done i kit, build, women ə alone, potato, stringent, sofa i wanted, luggage, buses ər lunar, driver, actor υ look, good, put long vowels ɑ spa, calm, father ɔ awe, dawn, caught, fall ɑr are, arm, spar ɔr or, corn, warn r err, bird, work, fur u ooze, blue, prune, brew, through i eel, sea, fiend, dream, machine diphthongs aυ owl, mouth, plough eər air, bare, pear ei aim, day, eight, grey oυ owe, go, dough, toe, goat ai I, right, fly, guy ɔi oil, boy iə idea υər poor, sure, dour iər ear, fear, pier, mere triphthongs aiər ire, pyre, choir aυər our consonants b boy, sobbing ŋ sing, drink, dinghy d day, address θ thigh d judge, giant, germ p pie ð this, although, bathe r rye, wrist f food, phonetics, if, off, rough s see, kiss, city, psychology g good, ghost, guide ʃ show, sure, charade, schmuck h hood t tall, pterodactyl j yes, fjord tʃ chin, watch k cat, chorus, kiss, brick, Iraqi v view, love, of l lie, all w wet m me, thumb, damn z zeal, peas n nigh, knife, gnaw, pneumatic measure, evasion, beige, rouge diacritics n syllabic /n/ (likewise for /l/, etc.) stressed syllable (a loof, sofa) 16 Chapter 1 Preliminaries Notes on the transcription system Post-vocalic /r/ This is represented by a superscript /r /. In rhotic AmE, it is pronounced as a sepa- rate /r/ consonant or coalesces with the preceding sound to give a rhotacised vowel. In non-rhotic BrE it is not pronounced at all – though, as we noted above, a word- final /r / will typically be pronounced in connected speech as a pre-vocalic linking /r/ when followed by a word beginning with a vowel. Pre-vocalic /r/ corresponds to an r in the spelling. We do not include intrusive /r/ in our representations, since it is pre- dictably present (between a low vowel or /ə/ and a following vowel) in those accents that have it. /i/, /ə/, and /i/ The unstressed vowel in the second syllable of orange, wanted, wishes, lozenge, etc., is a significant difficulty for an accent-neutral transcription. In BrE it is typically identical with the vowel of kit, which we represent as /i/; in most AmE and some Australian varieties it is usually identical with the second vowel of sofa, /ə/. Many of its occurrences are in the inflectional endings; but there is one inflectional suffix in English that contains /i/ in virtually all accents, namely ·ing, and there are suffixes containing a vowel that is /ə/ in all accents (e.g. ·en in written). Hence we need a third symbol for the vowel that varies between accents. We use /i/. This has been used by American phonologists as a phonetic symbol for a vowel slightly less front than /i/ and slightly higher than /ə/, so it is a good phonetic compromise, and visually suggests the /i/ of those BrE accents that have a minimal contrast between counted / kaυntid/ and countered / kaυntəd/. It should be kept in mind, however, that it is used here not with an exact phonetic value but rather as a cover symbol for either /i/ or /ə/ according to accent. /ɒ/ versus /ɑ/ For the vowel of pot, rock, not, etc., we use /ɒ/. Most varieties of AmE never have /ɒ/ phonetically in any context, so the American pronunciation can be derived simply by replacing our /ɒ/ by /ɑ/ everywhere. Hence there is no possibility of ambiguity. /oυ/ versus /əυ/ For the vowel of grow, go, dough, etc., we write /oυ/, in which the ‘o’ makes the phono- logical representation closer to the spelling; for most BrE speakers /əυ/ would be a phonetically more appropriate representation. /ɔ/ versus /ɑ/ BrE has distinct vowels in caught and calm: we represent them as /ɔ/ and /ɑ/ respectively. AmE standardly has the same vowel here, so for AmE the transcription /ɔ/ should be read as /ɑ/. /æ/ versus /ɑ/ Both BrE and AmE have distinct vowels in fat, /fæt/, and calm, /kɑm/, but there are a considerable number of words where most BrE accents have /ɑ/ while AmE (but also some accents within the BrE family) has /æ/. Very few of these arise in our exam- ples, however, so instead of introducing a third symbol we give separate BrE and AmE representations when necessary. § 3.2 Pronunciation and spelling 17 // versus /ə/ The opposition between // and /ə/ is a weak one in that there are very few word-pairs kept distinct solely by this vowel quality difference. It is absent in many AmE accents – those in which butt and but are pronounced alike in all contexts, in which just has the same pronunciation whether it means “merely” or “righteous”, and in which lust always rhymes with must regardless of stress. We show the distinction between these vowels here (it is generally clear in BrE), but for many Americans both vowels could be written as /ə/. /ju/ versus /u/ In many words that have /ju/ following an alveolar consonant in BrE, AmE has /u/. Thus new, tune, due are /nju/, /tjun/, /dju/ in BrE but usually /nu/, /tun/, /du/ in AmE. We write /ju/ in these cases; for AmE, ignore the /j/. Intervocalic /t/ We ignore the AmE voicing of intervocalic /t/, contrasting latter as /lætər / and ladder as /lædər / with the medial consonants distinguished as in BrE accents. 3.2 Pronunciation and spelling The relation between the sounds shown by our transcription and the ordinary English spelling of words is a complex one, and certain analytical concepts will help in keeping clear about the difference. Symbols and letters When we match up written and spoken forms we find that in the simplest cases one letter corresponds to one sound, or phoneme: in /in/, cat /kæt/, help /help/, stand /stænd/, and so on. But very often the match is more complex. For example, in teeth the two-letter sequence ee corresponds to the single phoneme /i/ and th to /θ/; in plateau the three- letter sequence eau corresponds to /oυ/ (a diphthong, analysed phonologically as a single phoneme); in through the last four letters correspond to the phoneme /u/. We will use symbol as a technical term for a unit of writing that corresponds to a phoneme, and we will refer to those symbols consisting of more than one letter as composite symbols.6 The letter e can form discontinuous composite vowel symbols with any of the letters a, e, i , o, u: a... e as in pane, e... e as in dene, i... e as in bite, o... e as in rode, and u... e as in cute. Vowels and consonants The categories vowel and consonant are defined in terms of speech. Vowels have unimpeded airflow through the throat and mouth, while consonants employ a sig- nificant constriction of the airflow somewhere in the oral tract (between the vocal cords and the lips). The terms can be applied to writing derivatively: a vowel symbol is a symbol representing a vowel sound, and a consonant symbol is a symbol representing a consonant sound. We will speak of a vowel letter or a consonant letter only in the case 6 ‘Digraph’ is widely used for a two-letter symbol and ‘trigraph’ is also found (though much less frequently) for a three-letter symbol, but there is no established term for a four-letter symbol, and no cover term for composite symbol. 18 Chapter 1 Preliminaries of non-composite symbols: a single letter constituting a whole symbol may be called a vowel letter if it is a vowel symbol or a consonant letter if it is a consonant symbol. Thus y is a vowel letter in fully (representing /i/); it is a consonant letter in yes (it represents /j/); and in boy it is just part of a complex vowel symbol (representing /ɔi/). Similarly, u is a vowel letter in fun (//), a consonant letter in quick (/w/), and part of a composite symbol in mouth (/aυ/).7 It should be noted, however, that r counts as a consonant letter even in non-rhotic accents, as shown by the rule of final consonant letter doubling in inflected forms dis- cussed in Ch. 18, §2.2.1: map/mapping, bat/batting, trek/trekking, pin/pinning, etc., are parallelled by mar/ marring, with r doubling like other consonant letters. Similarly, the e of the suffix ·ed counts as a vowel symbol even when no vowel is pronounced (e.g. it determines consonant doubling in forms like sipped /sipt/ and banned /bænd/). In both cases, of course, the spelling corresponds more closely to an earlier stage of the language than to the contemporary language. 4 Theoretical framework The primary goal of this grammar is to describe the grammatical principles of Present- day English rather than to defend or illustrate a theory of grammar. But the languages human beings use are too complex to be described except by means of a theory. In this section we clarify the relation between description and theory in this book, and outline some of our most important theoretical distinctions. 4.1 Description and theory The problem with attempting to describe English without having a theory of grammar is that the language is too big to be described without bringing things together under generalisations, and without a theory there are no generalisations. It does not take much reflection to see that there is no definite length limit to sentences in English. Sentences 100 words long, or longer, are commonly encountered (especially in writing, for written sentences are on average longer than spoken ones). And, given any sentence, it is always easy to see how it could have been made even longer: an adjective like good could be replaced by very good, or a verb like exceed could be supplied with a preceding adverb to make something like dramatically exceed, or a noun like tree could be replaced by tall tree, or the words I think could be added at the beginning of a whole declarative clause, or the words and that’s what I told the police could be added at the end, and so on through an endless series of different ways in which almost any grammatical sentence of English could be lengthened without the result being something that is recognisably not English. The importance of the fact that English sentences can be constructed to be as long as might be necessary to express some meaning is that it makes the sentences of English impossible to encapsulate in a list. The number of sentences that have been spoken or 7 It will be clear, then, that we do not follow the traditional practice of simply dividing the alphabet into five vowels (a, e, i , o, u) and twenty-one consonants: we will see that the traditional classification does not provide a satisfactory basis for describing the spelling alternations in English morphology. § 4.1 Description and theory 19 written so far is already astronomically vast, new ones are being produced every second around the world by hundreds of millions of people, and no matter what the information storage resources available, the problem is that there would be no way to decide where to end the list. An alternative to listing sentences is therefore needed. To describe the sentences that belong to English we have to provide a general account of their structure that makes their form follow from general statements, not about particular sentences but about sentences of English quite generally. We need to bring together the principles that sentences all conform to, so that we can use those principles to appreciate the structure of new sentences as they are encountered, and see how new ones can be constructed. This means developing a theory of the ways in which sentences can be put together by combining words. This book is an attempt to summarise and illustrate as much as possible of what has so far been determined about the ways in which sentences can be constructed in English, and it presupposes a theory that classifies the words of the dictionary and specifies ways in which they are combined to form sentences. We emphasise, however, that it is not the aim of this book to convince the reader of the merits of the theory for general linguistic description. Quite the reverse, in a sense: wherever it is possible to make a factual point overshadow a general theoretical point, we attempt to do that; whenever a theoretical digression would fail to illuminate further facts about English, we curtail the digression; if ever the facts at hand can be presented in a way that is neutral between competing theoretical frameworks, we try to present them that way. However, a significant amount of space is devoted here to arguing carefully that the particular analysis we have decided to adopt, within the framework of theory we assume, is the right analysis. What we mean by that is that even someone with a different idea about how to design a theory of syntax would have to come to a conclusion tantamount to ours if they considered all the facts. It is necessary for us to provide arguments concerning specific grammatical analyses in this book because, although this grammar is descriptive like the great traditional grammars that have been published in the past, it is not traditional in accepting past claims and analyses. We depart from the tradition of English grammar at many points, sometimes quite sharply. For example, in this book the reader will find nothing of ‘noun clauses’, ‘adjective clauses’, or ‘adverb clauses’, because that traditional distinction in subordinate clause classification does not divide things satisfactorily and we have abandoned it. The reader will likewise find nothing of the traditional distinction between since as a preposition (I haven’t seen them since Easter), since as an adverb (I haven’t seen them since), and since as a subordinating conjunction (I haven’t seen them since they went overseas), because we have concluded that this multiplication of categories for a single word with a single meaning makes no sense; we claim that since belongs to the same category (preposition) in all of its occurrences. On these and many other aspects of syntactic analysis we depart from traditional analyses (we draw attention to the major cases of this kind in Ch. 2). At such points we provide detailed arguments to convince the reader that we have broken with a mistaken tradition, and – we hope – made the correct decision about how to replace it. The reader will therefore find much more discussion of grammatical concepts and much more syntactic argumentation than is usually found in grammars of English. It 20 Chapter 1 Preliminaries is supplied, however, not to establish some wider theoretical point applying to other languages, but simply to persuade the reader that our description is sound. While the application of grammatical theories to the full range of human languages is an important matter within linguistics, it is not the purpose of this book to develop that point. Detailed technical or descriptive discussions that can be skipped by non-specialists without loss of continuity have been set off in smaller type with a shaded background. 4.2 Basic concepts in syntax Three essential concepts figure in the theory we use to describe English syntax in this grammar. Each is very simple to grasp, but together they permit extremely broad and powerful theories to be constructed for indefinitely large collections of sentences. We express them tersely in. i Sentences have parts, which may themselves have parts. ii The parts of sentences belong to a limited range of types. iii The parts have specific roles or functions within the larger parts they belong to. The idea that sentences have parts which themselves may have parts, i.e. that larger stretches of material in a sentence are made up by putting together smaller stretches, is the basis of ‘constituent structure’ analysis. The idea that the parts fall into a limited range of types that we can name and refer to when giving a grammatical description is the root of the concept of ‘syntactic categories’. And the idea that the parts also have specific roles or functions, or special slots that they fill in the larger parts they belong to, is the idea of ‘grammatical functions’. The next three subsections are devoted to explaining these three fundamental ideas. 4.2.1 Constituent structure Sentences contain parts called constituents. Those constituents often have constituents themselves, and those are made up from still shorter constituents, and so on. This hierarchical composition of wholes from parts is called constituent structure. Consider a simple one-clause sentence like A bird hit the car. It is divisible in the first instance into two parts, a bird (the subject) and hit the car (the predicate). The phrase a bird is itself made up of smaller parts, a and bird; so is hit the car, which we divide into hit and the car; and finally the car also has two parts, the and car. This structure can be represented as in. a bird hit the car Such representations of the constituent structure are called trees or tree-diagrams (though the trees are upside down, with the root at the top and the ends of the smallest § 4.2.2 Syntactic categories 21 branches at the bottom). The words are the smallest constituents, and the points closer to the root where branches join identify the larger constituents. A bird, for example, is identified as a constituent because this word sequence can be traced via the branches to a single point in the tree; similarly with the car and hit the car. The sequence bird hit, on the other hand, is not a constituent, as there is no point in the tree that leads down branches to just these two words and no others. The parts of the sentence shown at the first level down, a bird and hit the car are said to be the immediate constituents of the sentence; similarly, hit and the car are the immediate constituents of hit the car. The words are the ultimate constituents of the sentence. The evidence that this is the correct analysis of the sentence comes from the whole of the rest of the grammar, all of which provides, by virtue of the coherence of the description it gives, the evidence that the lines of separation have been drawn in the right place. We can give an illustrative example of how other parts of the grammar can provide supportive evidence by considering where we can insert an adverb such as apparently (indicating that what the rest of the sentence asserts appears to be true). A rough account of where English grammar permits it to be positioned (at least in clauses as simple as our example) is that it can be anywhere in the clause it mod- ifies, provided it does not interrupt a constituent. This is illustrated in , where the grammatical [a] examples conform to this rule, and the ungrammatical [b] ones do not: i a. Apparently a bird hit the car. b. ∗An apparently bird hit the car. ii a. A bird apparently hit the car. b. ∗A bird hit apparently the car. iii a. A bird hit the car, apparently. b. ∗A bird hit the apparently car. The five words of our example sentence permit six different logically possible placements for apparently that are between words (before any of the five words, or after the last one), but only three are permissible. Breaking the sentence into constituents in exactly the way we have done, we are able to make a general statement about where an adverb like apparently (a ‘modal’ adverb) can be positioned in it: such an adverb must not interrupt a constituent of the clause. Hence [ib] above is disallowed because it would interrupt the constituent a bird; [iib] is disallowed because it would interrupt hit the car; and [iiib] is disallowed because it would interrupt the car. Inspecting the diagram in , we see that each of these uninterruptible sequences is a constituent smaller than the whole sentence. The full support for a decision in grammatical description consists of confirmation from hundreds of mutually supportive pieces of evidence of many kinds, this being only one very simple example. 4.2.2 Syntactic categories Diagram shows just the hierarchical part–whole relationships in the sentence. This is only the starting-point for a description, identifying the constituents that have to be described. The next step is to classify these constituents, to say what syntactic category they belong to. For words, these syntactic categories correspond to what are traditionally 22 Chapter 1 Preliminaries called the ‘parts of speech’, and most of the categories for larger constituents are based on the ones for words. Where we need to refer to just the categories that have words as members, we will call them lexical categories. Lexical categories Any theory of syntax of the general sort we provide, and most types of dictionary, must include a list of the lexical categories or parts of speech assumed. For nearly all theories and nearly all dictionaries, noun, verb, adjective, and adverb will be among them, these being terms that have a history going back to the grammar of Classical Latin and Classical Greek some 2,000 years ago, but they are apparently applicable to almost all human languages. Our complete list is given, with some illustrations of membership, in : category label examples i noun N tree, pig, sugar, hatred, union, Picasso, London ii verb V do, fly, melt, think, damage, give, have, be, must iii adjective Adj good, nice, big, easy, ugly, helpful, reddish, fond iv adverb Adv obviously, easily, helpfully, frankly, soon, so, too v preposition Prep of, to, by, into, between, over, since, toward(s ) vi determinative D the, this, that, a(n), some, all, every, each vii subordinator that, for, to, whether, if viii coordinator and, or, but, nor ix interjection ah, damn, gosh, hey, oh, ooh, ouch, whoa, wow This scheme differs in several respects from the classification familiar from traditional grammar. Our determinatives are traditionally subsumed under the adjective category: they are said to be ‘limiting adjectives’ as distinct from the ‘descriptive adjectives’ illus- trated in [4iii] – though some traditional grammars do recognise the articles the and a(n) as a distinct part of speech. We also take subordinators and coordinators to be distinct categories, not subclasses of the traditional conjunction category. Conversely, we regard pronouns as a subclass of nouns, not a distinct primary category. Our reasons for departing from the traditional analysis are given in the relevant chapters. Phrasal categories Constituents containing more than one word (more specifically, containing a central and most important word augmented by appropriate accompanying words that elaborate its contribution to the sentence) are called phrases, and are assigned to phrasal categories.8 The lexical categories have corresponding phrase types that are in a sense expansions of them. A phrase consisting of a noun and the constituents that go with it most closely is a nominal; a nominal plus a determinative makes a noun phrase; a verb and its various complements makes up a verb phrase; a noun phrase and a verb phrase make up a clause; and so on. The full list of phrasal categories we employ in this book, together with our abbreviatory labels for them and an example phrase of each type, is given in.9 8 There are circumstances in which phrases may consist of a single word: see the discussion of ‘singulary branching’ in §4.2.3. 9 The term ‘sentence’ does not figure here. As will be explained more fully in Ch. 2, §1, a sentence in our terms is typically either a main clause or a coordination of main clauses. § 4.2.3 Grammatical constructions and functions 23 category label example i clause Clause she saw something in there ii verb phrase VP saw something in there iii noun phrase NP this clear case of dedication to duty iv nominal Nom clear case of dedication to duty v adjective phrase AdjP very eager for further news vi adverb phrase AdvP quite separately from this issue vii preposition phrase PP right out of the area viii determinative phrase DP almost every We can represent the structure of sentences in more detail than is done in a diagram like if we show the category to which each constituent belongs, as in. Clause NP VP D N V NP D N a bird hit the car 4.2.3 Grammatical constructions and functions The third central theoretical idea we must introduce is that constituents always have particular roles to play in the constructions, the larger units, that they belong to. We call these roles grammatical functions. In our example sentence the phrases a bird and the car belong to the same category, NP, but they have different functions, subject and object respectively. They belong to the same category because they are alike in their internal structure (both have a noun as the major element), but they have different functions because they stand in different relations to the verb. The opposite type of situation is illustrated in such a pair as: a. His guilt was obvious. b. That he was guilty was obvious. Here the underlined constituents have the same function (subject) but belong to different categories (NP and clause respectively). They have the same function because they stand in the same relation to the predicate, and they belong to different categories because the first is centred on a noun (guilt) while the second is centred, ultimately, on a verb (was). We say that the subject is realised by an NP in [a], by a clause in [b]. 24 Chapter 1 Preliminaries Heads and dependents There is a set of functions that to a large extent apply in the same way within all phrasal categories. The first division we make is that between the head and the various depen- dents that can combine with it. The head, normally obligatory, plays the primary role in determining the distribution of the phrase, i.e. whereabouts in sentence structure it can occur. Note, then, that while his guilt and that he was guilty can both function as subject they differ in other aspects of their distribution – we can have, for example, The news that he was guilty was devastating, but not ∗The news his guilt was devastating (we need a preposition: The news of his guilt was devastating), and this difference is attributable to the fact that the head of the former is a noun while the (ultimate) head of the latter is a verb. Dependents, often optional, are syntactically subordinate elements. The term ‘depen- dent’ reflects the fact that in any given construction what kinds of dependent are permit- ted depends on the head. For example, too (with the sense “excessively”) can function as dependent to an adjective or adverb (too careful, too carefully), but not to a noun or verb (∗their too extravagance, ∗You shouldn’t too worry). Similarly sufficiently can function as dependent to an adjective, adverb, or verb, but not to a noun (sufficiently good, sufficiently often, practised sufficiently, ∗sufficiently reason). Predicate and predicator as special cases of the head function Within this framework, what is traditionally called the predicate is a special case of the head function: the predicate is the head of a clause. Similarly, the term predicator is commonly used for the function of the verb itself, i.e. for the head of a verb phrase. We will retain the traditional terms, which indicate the characteristic semantic role of the element concerned, but it should be kept in mind that they are particular kinds of head. Subtypes of dependent Dependent is a very general function, and for many purposes we need to distinguish different subtypes of dependent according to their more specific relation to the head. At the first level of subdivision we distinguish complements, modifiers, and determiners, illustrated here in NP structure: i the photographs of their dog that they had brought with them [complement] ii the photographs of their dog that they had brought with them [modifier] iii the photographs of their dog that they had brought with them [determiner] In these examples, of their dog complements the head noun photographs; that they had brought with them modifies the head nominal photographs of their dog; and the determines the head nominal photographs of their dog that they had brought with them. At the next level we distinguish different kinds of complement, such as subject (the photographs are excellent), object (He destroyed the photographs), predicative (these are excellent photographs), and so on. A head element is said to govern its complements. The determiner function is found only in the structure of the NP, whereas comple- ments and modifiers occur quite generally. Note that the function ‘determiner’ is distinct from the lexical category ‘determinative’ (D). These need to be distinguished for the same § 4.2.3 Grammatical constructions and functions 25 reason as we distinguish subject and NP. Thus although this functions as determiner in this height, it functions as modifier in the structure of an AdjP in examples like She is about this tall. Conversely, while the determiner function is realised by a determinative in a doctor, it is realised by a genitive NP in my neighbour’s doctor.10 Non-headed constructions Although the functions of head and dependent apply to a very wide range of construc- tions, we must also allow for non-headed constructions, as in: i She bought [a hamburger, some chips and a glass of milk]. [coordination] ii A storm damaged – or so I’m told – the roof of their house. [supplementation] The underlined NPs in [i] are of equal syntactic status: we cannot say that one is head and the others dependents. Each of them has the same function within the bracketed construction, that of coordinate. In [ii] the underlined constituent is what we call a supplement: instead of being integrated into the constituent structure of the sentence as a dependent or coordinate, it is loosely attached, set off from the rest in speech by separate intonational phrasing and in writing by punctuation. Note that it interrupts the sentence at a point where a dependent could not occur, between the predicator and the object: compare [3iib] above.11 These two types of non-headed construction are described in Ch. 15. Diagrammatic representation of functions Functions, we have said, are essentially relational concepts: to specify the function of a constituent is to say what its relation is to the construction containing it.12 One way to capture this would be to write the name of the function on the line (branch) of the diagram joining the constituent to the construction. The first level in the structure of our model sentence might then look as in. Clause Subject Predicate NP VP In more complex cases, though, diagram design becomes a problem, and we have found it preferable to present the functional labels separated from the category labels by a colon, and written above them in diagrams. In this format the analysis of our earlier example sentence looks as in. 10 In other works ‘determiner’ is often used as a category term. The corresponding function is then sometimes called ‘specifier’, sometimes called ‘determinative’, and sometimes not clearly distinguished from the category term. 11 It must be emphasised, therefore, that [3iib] was marked as ungrammatical with the understanding that apparently is integrated into the structure (as indicated by the absence of any punctuation). If apparently were set apart as a supplement, the sentence would not be ungrammatical – but it would be a different sentence from [3iib]. 12 ‘Grammatical relation’ is indeed commonly used as an alternative term to ‘grammatical function’. 26 Chapter 1 Preliminaries Clause Subject: Predicate: NP VP Det: Head: Predicator: Object: D N V NP Det: Head: D N a bird hit the car (Note that we use ‘Det’ as the abbreviation for the function ‘determiner’, and ‘D’ for the category ‘determinative’.) Singulary branching We have said that dependents are often optional, and this implies that we can have a head on its own, without any dependents. Compare, for example: i Some children were playing in the park. ii Children were playing in the park. The underlined expressions are NPs functioning as subject of the clause: children is the head, determined by some in [i], but standing alone in [ii]. The relevant parts of the structure are thus as in. a. NP b. NP Det: Head: Head: D N N some children children In [b] there is a single branch descending from the category label NP, and this part of the tree-diagram is said to exhibit singulary branching, in contrast to the binary branching of [a]. 4.3 Morphology, inflectional and lexical A grammar, we have said, is divided into two major components, syntax and mor- phology. This division follows from the special status of the word as a basic linguistic unit, with syntax dealing with the combination of words to make sentences, and mor- phology with the form of words themselves. In some respects the formation of words is comparable to the formation of larger units, but in others it is significantly differ- ent, and it is these differences that motivate dividing the grammar into two separate components. § 4.3 Morphology, inflectional and lexical 27 Words, lexemes, and inflection The term ‘word’, as used in traditional grammar, has two rather different senses. We can approach the difference by asking how many distinct words there are in, for example: You are working hard, but your sister is working even harder. It is clear that the third and ninth words are not distinct: they are tokens (instances) of the same word. But what about hard and harder: are these the same word or different words? The answer depends on what you mean by ‘word’. In one sense they are obviously different: harder has a suffix that is missing from hard. This enables it to occur in constructions like that of Your sister works harder than you, where it could not be replaced by hard; and conversely hard could not be replaced by harder in Your sister works very hard. So from a syntactic point of view they are different words. But there’s another sense in which they are traditionally said to be ‘different forms of the same word’. The perspective this time is that of the dictionary, which would have just one entry, labelled hard. The same applies to are and is in : syntactically these are different words, but lexically (i.e. as far as the dictionary is concerned) they are the same. In order to avoid possible misunderstanding we will restrict the term word to the syntactically-oriented sense, so that hard and harder are different words, and likewise are and is. For the more abstract, lexically-oriented sense we will use the term lexeme. Hard and harder are then forms of the same lexeme, as are are and is. In many cases it makes no difference whether we take a syntactic or a lexical perspec- tive. Lexemes such as the and and are invariable, i.e. there is only one word corresponding to each. Also invariable are lexemes like efficiently: although more efficiently is in some respects like harder, it is not a single word, but a sequence of two, and hence efficiently and more efficiently are not forms of a single lexeme. Variable lexemes, by contrast, are those which have two or more forms. Where we need to make clear that we are con- sidering an item as a lexeme, not a word, we will represent it in bold italics. Hard, for example, represents the lexeme which has hard and harder – and also hardest – as its forms.13 Similarly are and is, along with be, been, being, etc., are forms of the lexeme be. In example , then, we have two occurrences of the lexeme hard, but only one of the word hard, and of course just one of the word harder. A variable lexeme is thus a word-sized lexical item considered in abstraction from grammatical properties that vary depending on the syntactic construction in which it appears. The variation found in variable lexemes is known, more specifically, as inflection, and the various forms are called inflectional forms of the lexeme. For the most part, inflectional categories apply to large sets of lexemes. Almost all verb lexemes, for example, inflect for tense (e.g. preterite took vs present tense take), most nouns inflect for number (e.g. singular dog vs plural dogs), many adjectives one or two syllables in length inflect for grade (e.g. plain old vs comparative older vs superlative oldest). The inflectional contrast of nominative case vs accusative case (e.g. we vs us), however, applies to just a handful of pronoun lexemes. 13 We minimise the use of bold type for lexemes, because in many cases it would simply distract. If we are simply listing adjective lexemes that can occur in the construction They are difficult to please, for example, we will generally list them as ‘difficult, easy, hard, impossible, tough’, etc., rather than ‘difficult, easy, hard, impossible, tough’, etc.; the fact that easy is inflectionally variable and difficult invariable has no relevance in that context. 28 Chapter 1 Preliminaries Inflectional morphology and lexical word-formation The distinction between words and lexemes provides the basis for the division of mor- phology into two branches: inflectional morphology and lexical word-formation. Inflectional morphology deals with the inflectional forms of variable lexemes. It has something of the character of an appendix to the syntax, the major component of the grammar. Syntax tells us when a lexeme may or must carry a certain inflec- tional property, while inflectional morphology tells us what form it takes when it carries that inflectional property. For example, a rule of syntax stipulates that a verb in con- struction with the perfect auxilary have must carry the past participle inflection (as in They have killed it, She had rung the bell), while inflectional morphology describes how the past participles of verbs are formed from the lexical base: killed is formed from the base kill by adding the suffix ·ed, rung from ring by changing the vowel, and so on. Lexical word-formation, by contrast, is related to the dictionary. It describes the processes by which new lexical bases are formed and the structure of complex lexical bases, those composed of more than one morphological element. The traditional term is simply ‘word-formation’: we add ‘lexical’ to exclude the formation of words by inflectional processes. The three major processes involved in lexical word-formation are the following: i compounding: forming a new base by combining two bases ii derivation: forming a new base by adding an affix to an existing base iii conversion: forming a new base using the pronunciation/spelling of a base of related meaning in some other category An example like blackbird illustrates compounding: it is formed by combining two smaller bases, black and bird. Efficiently illustrates derivation: an affix (the suffix ·ly) is added to an adjective base (efficient) to form an adverb. Another example, this time not involving a change from one category to another, is the derivation of inefficient by adding the prefix in· to the same base. And conversion is illustrated by the underlined verb in I managed to elbow my way to the front. The base elbow is primarily a noun (having the singular form elbow and the plural form elbows) denoting a part of the body. The verb base elbow (the base of the lexeme whose forms are elbow, elbows, elbowed, elbowing) is formed from the noun by conversion – the shape of the noun is simply borrowed to make a verb of related meaning. 4.4 Defining grammatical concepts A grammatical description of a language inevitably draws on a large repertoire of gram- matical terms and concepts – noun, verb, preterite, imperative, subject, object, and countless more. A question arises concerning how these concepts are to be explained and defined. Traditional grammar’s notional definitions It is useful to begin by considering the kind of definition familiar from dictionaries and traditional school grammars, which are known as notional definitions, i.e. they are based on the meaning of the expressions being classified, not on their grammatical § 4.4 Defining grammatical concepts 29 properties. These are typical examples: i noun: the name of a person, place, or thing ii preterite: a tense expressing past action or state [notional definitions] iii imperative: a clause expressing a command To determine whether a word is a noun, for example, one asks what it means or denotes; to determine the tense of a verb one asks in what time period it locates the action or state expressed by the verb; and so on. Such definitions have long been criticised by linguists. Indeed, it takes only a moment or two’s reflection to see that they do not provide satisfactory criteria for determining the correct classification of words or verb-forms or clauses. Take first the definition of preterite, and consider such examples as the following: i a. The finals started yesterday. b. You said the finals started tomorrow. ii a. I gave them his address. b. I regret giving them his address. In [i] we find started associated with past time in [a] but with future time in [b], as indicated by the temporal modifiers yesterday and tomorrow respectively. The started of [ia] thus satisfies the definition for preterite tense, while that of [ib] clearly does not. Nevertheless, everyone agrees that started in [ib] is a preterite form: this represents a different use of the same form as we have in [ia], not a different form, for the phenomenon is quite general, applying to all verbs, not just start. The opposite kind of problem arises in [ii]. Here the [a] and [b] versions are alike not in the form of the verb, but in the time of the associated event, which is located in the past. Both verbs therefore satisfy the definition of preterite tense, but while gave is certainly a preterite form, giving is not. The notional definition thus gives the wrong results

Quizgecko Grammar (2002) PDF

Document Details

Tags

Related

Summary

Full Transcript

Upgrade to continue