Prepositional Phrases and Prepositional Verbs: A Study in Grammatical Function 9783110802368, 9789027976161

Citation preview

JANUA LINGUARUM STUDIA MEMORIAE N I C O L A I VAN WIJK D E D I C A T A edenda curai C. H. VAN SCHOONEVELD Indiana University

Series Minor, 161

Prepositional Phrases and Prepositional Verbs A Study in Grammatical Function

Torben Vestergaard University

of

Aarhus

Mouton Publishers The Hague • Paris • New York

ISBN 90-279-7616-3 © Copyright 1977 Mouton Publishers, The Hague No part of this book may be translated or reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publishers Printed in the Netherlands

PREFACE

The word 'function' in the subtitle of this study refers not to 'the social function of language' but to 'the internal function of elements within the clause', i.e. to such concepts as Time, Location, Beneficient, Experiencer, etc. (for the two uses of the term 'function' in linguistics, see Halliday, 1973: 29). The original source of inspiration behind the work was Chomsky's Aspects (1965), and it was only after I had spent some time on problems connected with the description of adverbials in general that my attention was caught by functional ideas such as those set forth in Fillmore's 'The case for case' (1968) or Halliday's 'Notes on transitivity and theme in English' (1967-8). If in 2.1 I appear to be indulging in Chomsky-baiting, I would like to make amends for it here, however. Lyons's (1970: 9) statement that 'Every other "school" of linguistics at the present time tends to define its position in relation to Chomsky's views' certainly covers this study. This book would not have been possible without the help and advice of the following friends and colleagues: Professor Knud S^rensen, my supervisor, who has taken a very active interest in the project from the very beginning and through its various stages of completion. Thanks to his keen and penetrating criticisms the number of inaccuracies and inconsistencies has been considerably reduced. Professor S^rensen has also provided some particularly interesting examples from his own collections. Professor Randolph Quirk, University College, London, who permitted me to use the extensive material of the survey of English Usage, and permitted me to work in the congenial and stimulating atmosphere created by the staff of the Survey Research Room. Dr. John Anderson, Edinburgh, who helped me clarify my ideas on case relations.

vi

Frede 0stergaard, my one-time room mate, who has patiently answered my often monotonous questions and suggestions during innumerable and protracted tea and coffee breaks. My informants Michael Black, Gordon Campbell, Brian Donnelly, Philip Edmonds, A1 Jones, Neil Keeble, Anna Rutherford, Kathleen Thaysen, and last but not least, Shirley Larsen, who has also read the penultimate version of the manuscript and pruned it of some of its worst mannerisms. Signe Frits, who in the midst of her manifold secretarial duties found time to type the manuscript. I extend sincere gratitude to all those who have given me their help and comments. I may doubtless come to regret that I did not follow their advice more often. Aarhus, July 1976

TORBEN VESTERGAARD

CONTENTS

Preface

v

List of symbols and abbreviations List of tables

xi

xii

1. Introduction 1 1.1 Pretheoretical considerations 1 1.2 Prepositional verbs and related phenomena 1.2.1 Multi-word verbs 3 1.2.2 Multi-word prepositions 5 1.3 Sources of data 6 1.3.1 The corpus 8 Notes 10

2

2. Object and Adjunct 11 2.1 Syntactic structure: category and configuration 12 2.1.1 Chomsky's Aspects 12 2.1.2 Discussion 15 2.1.3 Lakoffs treatment of adverbials 20 2.1.4 Discussion 23 2.2 Syntactic function 24 2.2.1 Modalities and circumstantials: cases and participants 2.2.2 Modalities, circumstantials and free adjuncts 30 2.2.2.1 The abstract group 30 2.2.2.2 The concrete group 32 2.2.3 Cases, participants and bound adjuncts 33 2.2.3.1 Central participants 36 2.2.3.2 Marginal participants 38 Notes 42 3. Criteria 45 3.1 Functional criteria

45

VU1

3.2 Structural criteria 46 3.2.1 Criteria for (P-)N 47 3.2.2 Criteria for V-P 55 3.3 Framework for the description of corpus data Notes 62 4. Non-role Playing Prepositional Phrases 64 4.1 Syntactic features 64 4.1.1 Disjuncts 65 4.1.2 Conjuncts 66 4.1.3 Viewpoint/aspectual adjuncts 66 4.1.4 Intensifies 68 4.1.5 Circumstantial adjuncts 69 4.2 The corpus 70 Notes 71 5. Circumstantials I: The Abstract 5.1 Syntactic features 72 5.1.1 Time 74 5.1.2 Duration 76 5.1.2.1 Introductory remarks 5.1.2.2 Syntactic features 76 5.1.3 Frequency 78 5.1.4 Reason 80 5.1.4.1 Introductory remarks 5.1.4.2 Syntactic features 81 5.1.5 Epithet 83 5.1.5.1 Introductory remarks 5.1.5.2 Syntactic features 83 5.1.6 Manner 85 5.1.7 Measurement 87 5.2 The corpus 88 Notes 91

Group

72

76

80 83

6. Circumstantials II: The Concrete Group 6.1 Syntactic features 93 6.1.1 Free Location 95 6.1.1.1 Introductory remarks 95 6.1.1.2 Syntactic features 96 6.1.2 Instrument/Means 102 6.1.2.1 Introductory remarks 102 6.1.2.2 Syntactic features 103

93

ix 6.1.3 Having 107 6.1.3.1 Introductory remarks 107 6.1.3.2 Syntactic features 108 6.1.4 Comitative 111 6.1.4.1 Introductory remarks 111 6.1.4.2 Syntactic features 112 6.1.5 Beneficient 115 6.1.5.1 Introductory remarks 115 6.1.5.2 Syntactic features 116 6.2 The corpus 119 Notes 122 7. Bound Adjuncts I: Marginal Participants 7.1 Syntactic features 125 7.1.1 Location 128 7.1.1.1 Introductory remarks 128 7.1.1.2 Syntactic features 129 7.1.2 Path 134 7.1.2.1 Introductory remarks 134 7.1.2.2 Syntactic features 135 7.1.3 Direction 138 7.1.3.1 Introductory remarks 138 7.1.3.2 Syntactic features 139 7.1.4 Purpose 144 7.1.4.1 Introductory remarks 144 7.1.4.2 Syntactic features 144 7.1.5 Dative and Interlocutor 148 7.1.5.1 Introductory remarks 148 7.1.5.2 Syntactic features 149 7.1.6 Equative 154 7.1.6.1 Introductory remarks 154 7.1.6.2 Syntactic features 155 7.1.7 Objective 158 7.1.7.1 Introductory remarks 158 7.1.7.2 Syntactic features 160 7.1.8 Phenomenon 164 7.1.8.1 Introductory remarks 164 7.1.8.2 Syntactic features 165 7.2 The corpus 170 Notes 175

125

X

8. Bound Adjuncts II: Central Participants 8.1 Syntactic features 179 8.1.1 External Causer 180 8.1.1.1 Introductory remarks 180 8.1.1.2 Syntactic features 181 8.1.2 Agentive 183 8.1.2.1 Introductory remarks 183 8.1.2.2 Syntactic features 184 8.1.3 Experiencer 186 8.1.3.1 Introductory remarks 186 8.1.3.2 Syntactic features 187 8.1.4 Affected 190 8.1.4.1 Introductory remarks 190 8.1.4.2 Syntactic features 191 8.2 The corpus 194 Notes 195 Bibliography Index

203

197

179

LIST OF SYMBOLS AND ABBREVIATIONS

Adj

adjective

Adv

adverb(ial)

Aux

auxiliary or do

Det

determiner

impS

imperative (form of) S

intS

interrogative (form of) S

N

noun/nominal

Nadj

adjective/adjectival form related to N

Nadv

adverb(ial form) related to N

Nrel

relative pronoun with N as antecedent

negS

negated (form of) S

nomV

nominalisation of V

P

preposition

[P-N]rel

relative pro-adverbial with P-N as antecedent

[P-N]wh

non-relative wh-iorm substituting for P-N

proN

(anaphoric) pronoun substituting for N

pro [P-N]

other pro-adverbials substituting for P-N

pro [(Sb-)V]

pro form substituting for (Sb-)V

pro VP

Verb Phrase pro form

S

sentence minus the P-N under investigation

Sb

subject

Tns

tense morpheme

V

auxiliary

V

verb

Ven

past participle of V

Vinf

infinitive of V

xii present participle of V transitive verb interrogative pronoun is related to, is not related to entails, does not entail

Ving Vtr wh =, * &

coordinating conjunction

In addition, the standard symbols of transformational grammar are used, particularly in 2.1.

Identification of Examples (cf. below p. 9) CAPS

T. E. B. Howarth, Culture, Anarchy and the Public Schools (London: Cassell, 1969) The Guardian Weekly Costs and Revenue of National Newspapers (London: HMS, 1970) John le Carré, A Small Town in Germany (Pan Books, 1969) Gerald Durrell, My Family and Other Animals (Penguin Books, 1959)

GW CRNN JCSG FOA

LIST OF TABLES Table Table Table Table Table Table Table Table Table Table Table Table

1.1 4.1 4.2 5.1 5.2 6.1 6.2 7.1 7.2 7.3 8.1 8.2

8 65 70 73 90 94 120 127 173 174 180 194

1. INTRODUCTION

1.1

PRETHEORETICAL CONSIDERATIONS

In the present study we shall be concerned with sentences in which an intransitive verb is accompanied by one or more prepositional phrases. In particular we shall examine the differences in the relationship between the verb and the prepositional phrases in examples like (1) [....] it is always in this sense that we speak of "infinite" specific heats, etc. (83.3.23-1) 1 The most frequently noted aspect of this difference is the fact that the complement of of may occur as the subject of the passive version of (1), whereas the complement of in may not. Thus (l')a. is a paraphrase of (1) but ( l ' ) b . is not: (l')a- "infinite" specific heats, etc. are always spoken of in this sense, b. *this sense is always spoken in of "infinite" specific heats, etc. Another frequently invoked criterion is the substitutability of a transitive verb for the intransitive verb+preposition, e.g. discuss/mention for speak o f , whereas this possibility does not exist in the case of speak in. And this point naturally leads to the observation that the relation between speak of and infinite heats is parallel to that between a transitive verb and its object, whereas the relation between speak and in this sense is the relation normally described as verb - adjunct. 2 Examples like speak of are commonly referred to as prepositional verbs (e.g. Quirk et al, 1972: 811 ff). It is clear, however, that there is no clear-cut border-line between prepositional verb + object on the one hand and verb + prepositional adjunct on the other. Even a cursory glance at the examples of (2) will show that the two shade gradually into one another (cf. Fairclough, 1965; Carvell & Svartvik, 1969; Mitchell, 1958):

2 (2) a. b. c. d. e. f. g.

It's hard to come by a new job. Fred will pay for the tickets. The watch-man fumbled for his keys. The procession arrived at the Town Hall. Ed is painting with my water colours. George appeared on the appointed day. Mike will disagree in any case.

In addition to the type already mentioned there are of course also (a) 'double object' prepositional verbs as well as (b) double object transitive-prepositional verbs, e.g. (a) speak to sbdy about sth; (b) provide sbdy with sth, provide sth for sbdy (see Corder, 1968). Of these only type (a) will be dealt with here. The reason for this is merely practical. If not only cases of intransitive verb+prepositional phrase(s) but also cases of transitive verb+object+prepositional phrase(s) were to be considered, it would have meant an overwhelming increase in the data-collecting phase of the work. Moreover, I believe that the transitive-prepositional type offers no special methodological problems. The purpose of this study will be to investigate the relations holding between the elements verb (V), preposition (P), and nominal (N) in the italicised parts of such examples as (1) and (2), with special emphasis on cases where V and P are fused to form 'a single transitive verb' (Zandvoort, 1962: 201). Any sequence containing the elements V-P-N will be a candidate for consideration, regardless of order, and for this reason the term CLUSTER will be used (rather than string or sequence). Thus (1) will be said to contain two V-P-N clusters, speak-in-this sense, and speak-of'infinite" specific heats, etc. When, in cases where the relation between V-P is closer than that between P-N, we speak of V and P alone, they will be referred to collectively as a (V-P) COMBINATION.

1.2 PREPOSITIONAL VERBS AND RELATED PHENOMENA

The prepositional verb is by no means the only example of that process 'at the border of syntax and morphology' (Bolinger, 1971: 111-12) whereby two (or more) items unite into one new, internally complex item. Examples are found within all three lexical classes 'verb', 'preposition', and 'noun'. Multiword nouns are known as compounds, but since the internal structure of N offers no — theoretical or practical — problems in this context, I shall have no more to say about nominal compounds. This is not the case with complex verbs and prepositions, since obviously one's conception of the distinction between multi-word lexical items and free syntactic construction is crucial

3 when it comes to deciding whether some or any of the examples in (3)-(4) contain examples of V-P-N clusters. (3) a. b. c. d. (4) a. b. c.

His conduct gave rise to some misunderstanding. These words may give offence to some people. The cave gave shelter to the wanderer. The old lady gave her money to the poor. The cat stayed in front of the gate. The cat sat to the left of the gate. The cat stayed behind the bars of the gate.

1.2.1 Multi-word verbs According to their constituents multi-word verbs apart from prepositional verbs can be divided into three main groups, (i) verb+adverbial particle ('phrasal verb'), (ii) verb+object, (iii) verb+prepositional phrase: (i) (ii) (iii)

Our provisions are running out. These words may give offence. John has fallen in love.

In principle, all three types may function as V in a V-P-N cluster (we're running out of provisions, these words may give offence to some people, John has fallen in love with Susan), and I shall thus have no use for the term 'phrasal prepositional' for items like run out of (Mitchell, 1958: 106; Quirk et al, 1972: 811), since they will be seen to be no more than one particular type of V-P combination in which the exponent of V is itself a multi-word verb. I shall deal first with types (ii)-(iii), and then return to the phrasal verb, which offers special problems. As the examples of (3) are intended to show, 'there may well be gradience here rather than a clear-cut dichotomy' 3 (Huddleston, 1971: 86), and I shall arbitrarily regard the criterion that Huddleston (op cit) offers as his criterion (b) as diagnostic: The N of a multi-word verb of type (ii) may not be the focus of a pseudo-cleft construction (cf. below p. 50) This places give shelter only just outside the class of multi-word verbs of type (ii), and give offence inside it:

4 (5) a. (? ) What the cave gave to the wanderer was shelter, b. *What the show gave to its audience was offence. The same criterion can be applied to type (iii), where it will distinguish between e.g. arrive at a settlement, and come to an understanding: (6) a. What they arrived at was a provisional settlement. b. *What they came to was a provisional understanding. I shall rely on this criterion although it is somewhat over-inclusive, and in at least one case (below p. 157) forces the interpretation 'multi-word verb' upon a cluster which the descriptive apparatus established in chaps 2-3 is perfectly capable of handling. Whereas the only problem with types (ii) and (iii) is that of deciding whether a particular string is sufficiently fused to deserve the name 'multiword verb', the problem with verb+adverbial particle is that of distinguishing it from verb+preposition. Quite a large number of particles, referred to variously as 'prepositional adverbs' (Bolinger, 1971: 26) or 'adverb-preposition words' (Sroka, 1972: 37), may function now as prepositions, now as adverbial particles, e.g. along, down, in, o f f , on, through, up.4 The clear cases of prepositional or adverbial function can be distinguished by the fact that adverbial particles may follow an object (and if the object is pronominal, usually do), whereas prepositions may not (cf. Bolinger, 1971: 10-11). Thus up is prepositional in (7)a. and adverbial in (7)b.: (7) a. he lived up the hill f *he lived it up b. he looked up the information = he looked it up In some cases, however, the particle may take on an intermediate function, in which it belongs partly with the verb (as an adverbial particle), partly with the object (as a preposition). Bolinger (1971: 26 ff.) names this function 'adprep' and illustrates it by means of the triply ambiguous (8), where off may be interpreted as an adverbial particle (a), as a pure preposition (b), and as an adprep (c): (8) She swept off (a) She swept (b) She swept (c) She swept

the stage. it off. (cleaned it) off it. (did her sweeping somewhere not on the stage) off it. (departed from it majestically)

Off in (c) is as much a constituent of the phrase sweep off as of off the stage,

5 and although the relative position of particle and pronoun 'argues that the particle is a preposition' (op cit p. 27), the adverbial nature of adpreps is brought out by another test: (9) a. b. (10) a. b.

*He ran towards, pell-mell, the first hill he saw. He ran, pell-mell, towards the first hill he saw. He ran up, pell-mell, the first hill he saw. He ran, pell-mell, up the first hill he saw.

Towards can only be interpreted as a pure preposition and therefore it may not be separated from its complement. Up may in addition be an adprep and tolerate the position of adverbs characteristic of adverbial particles (cf. he looked up carefully all the words he didn't know). In this study particles in adprep function are included among prepositions although their dual nature will force us to give them separate treatment with some of the diagnostics used (see below 6.1.1.1).

1.2.2 Multi-word prepositions There are two types of multi-word prepositions, (i) a three-word type of the pattern Pl-N-P 2 , e.g. in front of, by means of, (ii) a two-word type, in which the first item is an adverb/conjunction/verb, etc.: ahead of; out of; because of; owing to; due to. Here too there is a gradient from completely fused types to free syntactic constructions (for type (i) cf. Quirk & Mulholland, 1964). At the closed end of the gradient we have such type (i) strings as by dint of, in lieu of, where the exponent of N never functions as an independent lexical item; closer to the open end we have strings like for the benefit of, in the interest of, which share some properties with free syntactic constructions (cf. for his benefit, in his interest) but not all (note e.g. invariability of article, *for a benefit of). Since this type, like the multi-word verbs of types (ii)-(iii), involves a nominal element, I shall regard the independence of the nominal as diagnostic and use the same criterion. Accordingly in the interest of will be seen to be somewhat more tightly fused than for the benefit o f , although neither of them is a free syntactic construction like e.g. for the welfare of: (11) a. he worked in the interest of the public f b. *what he worked in was the interest of the public. (12) a. he worked for the benefit of the public f b. ? *what he worked for was the benefit of the public.

6 (13) a. he worked for the welfare of the public = b. what he worked for was the welfare of the public. 1 have dealt with type (ii) elsewhere (Vestergaard, 1973 a) and shall only give a brief survey here. Its members fall into two subtypes (a) those where the first element may function both as the P element of a phrasal verb and as a premodifier of the second element: (14) he (stared [out) of the window] (b) others (owing to, etc.). Both subtypes should be distinguished from constructions where a particle is simply a premodifier of a preposition (they live OVER IN Canada), or where it is simply the P element of a phrasal verb (he PULLED UP at the traffic lights). For subtype (a) the divisibility criterion (see below p. 59) is regarded as diagnostic; i.e. the two elements of a multi-word preposition of this type may not be separated by an adverb. The examples of (15) illustrate the gradient: (15) a. *The scouts rode ahead courageously of the army b. ? the steamer throbbed away noisily from the coast c. he travelled back hurriedly from Paris Owing, no doubt, to the varied class membership of the first element in type (b) sequences, it is hard to establish one single criterion differentiating between multi-word items and free constructions, and all I will offer at this point is a list of the items of this type occurring in the corpus of the present study: according to, allowing for, because of, depending on, except for. In addition to these, I know of one sequence which definitely occurs at P in V-P-N clusters, but which does not fit into any of the types dealt with so far, viz. as far as, cf. we walked to the station, we walked as far as the station.

1.3

SOURCES OF DATA

Linguistic data may be elicited in three ways: (a) from a text corpus, (b) through introspection, (c) from native informants. 5 Each method has a number of advantages and disadvantages, and for a full description of a linguistic phenomenon all three should be used. Corpus data are of course indispensable in descriptions of actual usage, i.e. descriptions concerned with such questions as 'how often does this particular element occur in that position/with that intonation, e t c . ? ' a r e there

7 stylistic differences in the incidence of element A in position B? The usefulness of corpus data is particularly apparent if the linguist is concerned with registers that he is not familiar with, and, of course, if he is not a native speaker of the language he is studying. Similarly, corpus data may also provide him with information about linguistic facts which he might simply have failed to think of, had he relied on introspection alone. The main disadvantages of corpus data are (i) that they provide information only about the overt features of a given phenomenon (e.g. the position, exponence, and intonation of an element) but not about its full potential (its covert features), (ii) If the description is corpus bound, the linguist may have to go on indefinitely looking for an example illustrating a particular point. Methods (b), introspection, and (c), informant tests - formal and informal, 6 are closely related. Ideally, any example that the linguist has thought up himself should be checked with native informants, no matter whether the language under inspection is his native language or not: The linguist's Sprachgefühl is likely to become somewhat distorted after long periods of immersion in a problem. On the other hand, the use of informants presupposes introspection on the part of the linguist, since his questions to them will usually have the form 'are sentences A and B similar or dissimilar to each other? ', 'is X a possible form in your language? ', 'are A and B related to each other in the same way as C and D ? ' . Before the linguist can ask questions like these, he must have formed ideas about the similarity, etc. of the utterances concerned. Accordingly, while data obtained through introspection and information tests lack the advantages of corpus data, the two techniques in combination make up for the disadvantages of corpus data: (i) Information about (aspects of) covert features is obtained through performing certain operations on the original example and then consulting the informants, (ii) the corpus data can be easily, and infinitely, supplemented by making up new examples and presenting them to the informants. In this study all three methods are used. Each of the chapters IV-VIII is divided into two main sections. In the first of these the grammatical potentials of the types of V-P-N cluster under consideration are discussed, and here informant controlled introspection is the main source of information, although textual examples are used for illustration whenever convenient. The second section of each of the chapters is concerned with usage, or more precisely, with the incidence of various positional and exponential types among the types of cluster dealt with in the chapter. In those sections only corpus data are used. It should perhaps be pointed out at this stage that my use of introspection and information tests does not meet the ideal requirement set up above: I have consulted the informants only where I had reason to distrust my own

8 intuition. Thus I have not bothered my informants with e.g. the relative acceptability of (l')a.-b. above, but I have consulted them about e.g. (11)(13).

1.3.1 The corpus The corpus data stem from a corpus of about 200,000 words of spoken or written text. It is based in part on the files of the Survey of English Usage, in part on collections of my own. The distribution of the corpus over various registers is tabulated below. WRITTEN (i)

informative a. learned science arts b. persuasive c. press d. official

(ii)

8a.3 8b.1 CAPS GW CRNN

imaginative novels

JCSG FOA

SPOKEN (i)

recorded non-surreptitiously radio discussions

(ii)

5b. 16

rzcorded surreptitiously conversations

s.lc.ll s.lc.2 s.lb.6

Table 1.1 Examples from the Survey corpus are identified by means of the number of the relevant slip in the Survey files. Each Survey text consists of about 5,000 words. The sources of the texts are not detectable from the code numbers, but the Survey files will be open to serious students for inspection. The spoken texts are transcribed in a notation explained in Crystal & Quirk (1964).

9 Since the only prosodic information relevant in this study (cf. below 3.2.1) is the boundaries of the tone groups and pauses, I only retain the following prosodic symbols: (a) (b) (c) (d)

boundary of tone group: mani=t onset of a tone group: /man pause equivalent to the speaker's rhythm unit: manpause shorter than - : man.

Segmental features are given in their normal orthographic form, if there is one, if not a phonemic transcription (IPA) is resorted to. The part of the corpus compiled by myself consists of the following texts: CAPS GW CRNN JCSG FOA

T.E.B. Howarth, Culture, Anarchy and the Public Schools (London: Cassell, 1969) 35,000 words The Guardian Weekly, Jan. 31, 1970 40,000 " Costs and Revenue of National Newspapers (London: HMS, 1970) 15,000 " John le Carré, A Small Town in Germany (Pan Books, 1969) pp. 5-105 40,000 " Gerald Durrell, My Family and Other Animals (Penguin Books, 1959), pp. 9-96 36,000 "

Examples from this part of the corpus are identified by the abbreviation followed by the page number, or, in the case of GW, date, page number and column. Thus (CRNN 12) after an example means that it comes from Costs and Revenue of National Newspapers, p. 12; and (GW 31.1.70, 21:4) means that the origin is The Guardian Weekly, 31. Jan. 1970, p. 21, column 4. When examples not occurring in the corpus are cited, the full source is given in footnotes. In several cases I have italicised the relevant parts of an example; italics belonging to the original are marked by [ital] after the (last) italicised word. Interpolations in the original are surrounded by square brackets [ ]. Thus (16) might be rendered as (16'): (16) It is a sobering, though not altogether fanciful, speculation that the Commissioners may conclude that the trouble with these particular schools is that they are too good and are therefore in urgent need of adjustment to enable them to conform more nearly to the average. (CAPS 16)

10 (16') [. . .] they [sc the public schools] are too [ital] good and therefore in urgent need of adjustment to enable them to conform more nearly to the average. (CAPS 16)

NOTES 1. For the identification of the examples, see below 1.3.1. 2. This formulation actually begs an important question: By saying that there is a relation on the one hand between speak of and infinite heats, and on the other between speak and in this sense, I have been assuming, too, that there is a direct relation between speak and of but not between speak and in. Although this is a frequently propounded hypothesis (cf. e.g. Bach, 1967; Chomsky, 1965: 101, 191; Huddleston, 1971: 95; Robinson, 1970: 59-60; Vasiliu, 1968) it has never, to my knowledge, been subjected to empirical verification. Nor shall I attempt to demonstrate it explicitly. Note, however, that it is only in the later chapters (7 and 8) of this study - i.e. those dealing with combinations similar to speak of - that the criteria dealing specifically with the relation between V and P (cf. 3.2.2.1) systematically show anything but negative reactions. Which implies that it is only with these combinations that anything interesting can be said about the relation between V and P. 3. For the use of the concept of gradience in linguistics, see Quirk (1965) and Carvell & Svartvik (1969). 4. For this reason Jespersen (1924: 88-9) proposed re-naming Preposition and Adverbial Particle as 'transitive particle' and 'intransitive particle' respectively. 5. For a fuller discussion of the advantages and disadvantages of the three methods, see Greenbaum (1969: 8-14). 6. The formalised use of informant tests is described by Greenbaum & Quirk (1970). For this study only informal tests have been used. Partly because the group of informants available to me was neither large nor homogeneous enough for the purposes of formal experiments; partly because formal experiments tend to attract a certain amount of attention to the reaction of the informants in itself, and this aspect does not form part of the subject matter of this book.

2. OBJECT AND ADJUNCT

In order to make clear what is actually meant by the observation that some prepositional complements are object-like and that some prepositional phrases are adverbial (above 1.1), we shall have to examine in some detail what are the characteristics of the constituents 'object' and 'adjunct', and, more specifically, whether, in terms of syntactic structure, 1 there are different kinds of object and adjunct. If such differences can be found, they might help us to understand why prepositional phrases are felt to be functionally different in the way they are (cf. above, 1.1 [2j a. - g.). These problems form the topic of 2.1. In a number of fairly recent publications 2 it has been claimed that syntactic (constituent) structure should be seen as a reflection of something else. That is, behind the relations between constituents that can be seen in syntactic structure are some 'deeper' relations which, although they are expressed through syntactic structure, are not directly observable in it. To take a standard example: (1) a. Peter bought a car from John b. John sold a car to Peter A constituent analysis would be at a loss to explain that both a. and b. express the same relation between the constituents Peter, John, and a car. In order to state this fact one might introduce FUNCTIONAL labels saying that in both cases a car is an object being transferred, that Peter is the goal of the transference and that John is the source. The relevance of such a conception of grammar to the problem of describing the relations within V-P-N clusters lies in the possibility of assigning different functional labels to prepositional phrases that are intuitively felt to differ in their relations to V, thereby explaining the basis for the intuition. Syntactic function in this sense will form the topic of 2.2.

12 2.1

SYNTACTIC STRUCTURE: CATEGORY AND CONFIGURATION

2.1.1 Chomsky's ASPECTS In this section we shall examine the Transformational - Generative model of grammar with a view to its applicability to the task outlined in 1.1. The discussion will focus on chap. 2 of Chomsky's Aspects (1965), a well-known, explicit, and reasonably detailed example of a constituency grammar (in the sense that its basic concept is constituent structure). 3 Chomsky (1965: 64-74) compares the two phrase markers (2) and (3) (p. 65 and 69, respectively), and observes that (3) is redundant.

(3)

the

boy

The redundance is due to the fact that (in deep structure) the functions Subject, Predicate, Main Verb, Object are represented directly in the configurations of the categories NP, VP, V, NP. Thus "Subject o f ' can be defined simply as 'the relation holding between the NP of a sentence NP Aux VP and the whole sentence', and ' "Object o f ' as the relation between the

13 NP of a VP of the form V NP and the whole VP, etc.'. Accordingly, all the functional information necessary for the interpretation of a sentence should be represented in base structures solely by categorial symbols and their configuration (cf. also p. 117). Another important point is that the shape of a phrase marker should reveal how closely two constituents are related. (Since our primary interest is in the relations between the verb and other constituents, I shall discuss only these relations). It will be seen readily that as far as the relation between the verb and the nominal constituents is concerned, this requirement is in fact fulfilled by (2): The subject-verb relation is not as close as the object-verb relation, in so far as the choice of a particular verb determines only the class of the subject ([+ animate], [+ abstract], etc.), but both the presence or absence of the object (transitive/intransitive verbs) and, if present, its class (cf. Hudson, 1967: 231). The type of relation that is stated in terms of the features of one of the constituents involved (in practice, the nominal constituent) Chomsky refers to as a 'selectional relation' (p. 113), and the type of relation that refers to the very presence or absence of a constituent is called a relation of 'strict subcategorisation' (p. 95). Chomsky now notes (i) that a verb is strictly subcategorised only in respect of constituents directly dominated by VP, and (ii) that all constituents dominated by VP are relevant to the strict sub-categorisation of verbs (p. 96). Together the two conditions are referred to (p. 105) as the 'strictly local subcategorisation principle'. Consider now (4) (adapted from Hudson, 1967): (4) To my surprise, John has spent ten pounds on the car since Christmas. Clearly, the relation between the verb and the three prep.-phrases is not of the same type. Particularly close is the relation between spent and the o«-phrase. In fact, spent in this type of constructions demands the presence (at least implicitly) of an ow-phrase. This is seen most clearly in the fact that if one hears the sentence John (has) spent ten pounds, it is always meaningful to ask what on?. It will appear that the relation between spend and the on-phrase is that of subcategorisation, and according to condition (i) of 'the strictly local subcategorisation principle' we can now determine the constituency of the onphrase as being dominated by VP. Since Christmas is related to the aspectual form of the verb (Hudson, loc. cit.) rather than to the semantic content of it (cf. he has spent ten pounds - since when? and *he spent ten pounds - since when?). For this relation to

14 come out in a phrase marker, we need a node intermediate between S and VP, directly dominating both Aux and VP, and to which the temporal since Christmas may be attached. This is the node referred to as Predicate-Phrase in Aspects (p. 102, passim). Finally, the 'attitudinal disjunct' (Greenbaum, 1969) to my surprise. It is clear that there are no ties between it and any particular constituent of the clause. Rather, it says something about the speaker's attitude to the content of the clause. In a constituency grammar this can be shown by having it appear outside the clause, so to speak, directly dominated by S.4 It follows that the phrase marker (4') will be assigned to (4) (irrelevant details omitted): (4') Adv I to my surprise has

V spent

NP ten pounds

PP on the car

since Christmas

Within the Aspects model, then, there is, in principle, only one way of introducing objects, and three ways of introducing prep .-phrases. Further, observe that as far as constituency is concerned the prep.-phrases most intimately connected with the verb have exactly the same status as the object: both are directly dominated by VP. Below, 2.1.2, I shall discuss whether these three possibilities are sufficient to account for the range of prep.-phrases actually occurring. But it may be in order to point out, already at this stage, that if the model is found inadequate, such inadequacy might be remedied by postulating new intermediate nodes, as was the case with Predicate Phrase. This is a very dangerous course of action, however. Whereas the term 'Predicate' is well established and independently motivated, such further intermediate nodes as one might wish to posit in order to account for more types of prep.-phrases than the three accommodated by the model would be entirely arbitrary.

15 2.1.2 Discussion Considerations similar to those motivating phrase marker (4') lead Chomsky to propose the following base rules (1965: 102): (5) i. ii.

S

NPPredicate-Phrase

Predicate-Phrase ->• Aux VP (Place) (Time) 5

iii. VP ->

be Predicate (NP) (Prep-Phrase) (Prep-Phrase) (Manner) V Adj S' (like) Predicate-Nominal

iv. Prep-Phrase -

Direction Duration Place Frequency etc.

With the change in (5)i. provided for in fn. 4 above, these rules generate (4'), except that the node dominating Since Christmas will be labelled Time rather than Adv. The rest of this section will be devoted to an examination of the functional types of prep.-phrase accommodated by (5), that is, 'Sentence Adverbial' [the revised form of (5)i.], 'Verb Phrase Complements' [(5)ii.], and 'Verbal Complements' [(5)iii.]. 6 As the present study is concerned especially with those cases where the link between the verb and the prep.-phrase is particularly close, I shall deal rather summarily with Sentence Adverbials, and concentrate the discussion on Verb Phrase and Verbal Complements. Examples of these are: to my astonishment/relief/ surprise, etc.; in all courtesy/fairness/honesty, etc.; in general, without doubt; in addition, in other words, in spite of that, on top of that. From among the several criteria distinguishing Sentence Adverbials from other types (cf. Greenbaum, 1969: 18-23, 41-44, 113-27) I cite only one (cf. further below 4.1.1-2): SENTENCE ADVERBIALS. 7

(i) they may not be the focus of clause negation: (6)a. *he did not pay for the car to my relief, but he did pay for it to my astonishment

16 b. *he but c. *he but

did not pay for the car in general, het did pay for it in particular did not pay for the car in spite of that, he did pay for it on top of that

If we think of the object-like functions and adverbial functions of prepositional phrases as the poles of a cline, it is obvious that the Sentence Adverbials will be situated at the extreme adverbial end of the cline. In this respect a phrase marker like (4') is quite suggestive, since it shows clearly that there are no direct relations between the Sentence Adverbial and the verb. VERB PHRASE COMPLEMENTS and VERBAL COMPLEMENTS. Whereas it is reasonably easy to isolate Sentence Adverbials on the one hand from Verb Phrase and Verbal Complements on the other, it is far from easy to decide when a prepositional phrase is a Verb Phrase Complement, and when a Verbal Complement. There are a number of criteria available, but none of them are applicable to all cases (for discussion of the criteria, see below, chap. 3). As 'Place' is the only type appearing in both functions according to (5), the problem might reduce to deciding when a locational phrase is to be introduced by (5)ii., and when by (5)iii.-iv. Test (ii) takes care of this problem:

(ii) A verb without its Verbal Complements may not be the focus of a do/happen what sentence (cf. N. 17, below p. 54). This accounts for the following: (7)a. b. (8)a. b.

What are you doing in my room? — I'm working. What is he doing on the corner? — He's waiting (for the bus). *What are you doing in my room? — I'm living, *What is he doing on the corner? — He's sitting.

(In what follows prepositional phrases inside the Verb Phrase will be referred to as 'bound', those outside VP as 'free'). (7)-(8) illustrate the necessity for postulating two sources for locational phrases. A small class of verbs (sit, stay, remain, etc.) takes (explicit or implicit) bound locational phrases, whereas a large class takes free locational phrases, though the locational phrase is never part of the necessary context. The latter class, of which the former is a subclass, is coextensive with the class of 'activity verbs' of Greenbaum (1970: 83-84) (cook, run, ask, weep,

17

talk, etc.); and it is also characterised by its ability to take manner adverbials and to occur in the progressive aspect. There is also a third possibility, however, and this is very hard to accommodate within the Aspects framework: A large class of verbs, the 'stative verbs of inert perception and cognition' (Quirk et al. , 1972: 96) does not normally collocate with locational phrases in the present, but may do so in the past tense (cf. Lees, 1960a: 12; Fillmore, 1968: 26 fn. 34): (9)a. *Peter knows Greek in London b. Peter knew Greek in London (b. meaning 'at the time when he was in London, he knew Greek'). The distinction comes out clearly with appeal to, which may be either 'activity' or 'stative' (example adapted from Greenbaum, 1970): (10)a. Mr. Heath appeals to M Pompidou b. Mr. Heath appeals to M Pompidou in Paris (10)a. may mean either 'arouses favourable emotions in' or 'makes earnest request to'; owing to the presence of the locational phrase, b. only has the latter meaning. But note that what we have been doing here is to subclassify verbs with respect to their freedom of occurrence in the context [-Place], and according to the strictly local subcategorisation principle (above, p. 13) all contexts with respect to which a verb is subclassified are constituents of the VP. This would force us to conclude that ALL locational phrases belong in the VP, but then constituency would no longer reveal the varying degrees of cohesion between the verb and the 'place' constituent (above, p. 13), nor would there be any way of accounting for the differences in acceptability between the sentences in (7)-(9). I know of no way of getting out of this difficulty within the Aspects model. If (ii) is criterial of verb phrase constituency, directional phrases are always Verbal Complements: (11) *What did the steamer do away from the coast? - It throbbed. On the other hand, by criterion (ii) Frequency phrases always seem to fall outside the VP, even Chomsky's own example (Aspects, p. 102): (12) What does he do three times a week?

- he wins.

18 Unlike what (5)iv. predicts, Durational phrases, even with such verbs as wait, continue, are Verb Phrase Complements (free adjuncts) according to the do/happen what test: (13) What did he do I ^ happened

next) five hours? - He waited.8

The class of verbs that may take Duration as a free adjunct comprises the activity verbs and process verbs of Quirk et al. (1972: 95). Another large class of verbs, transitional event verbs and momentary verbs (op. cit.: 95-96), never collocates with durational phrases, unless the meaning of the verb is 'unfulfilled/repeated action': (14)a. *For two days the boat arrived. b. The boat has been arriving for two days. Finally I wish to discuss the status of manner adverbials, which Chomsky regards as particularly important in verb subcategorisation (Aspects, p. 103), and which consequently must be regarded as typical verbal complements. As the following example shows, they do not have the same status as directional and bound locational phrases, but they do behave like (free) durational phrases: (15)a.

1. What did he do in despair? - he jumped out of the window. 2. *What did he do out of the window? - he jumped in despair. b. 1. What he did with complete equanimity was sit in my chair. 2. *What he did in my chair was sit with equanimity. c. 1. What he did with patience was wait for three hours. 2. What he did for three hours was wait with patience.9

On the other hand, although 'manner' is not as closely related to the verb as the bound phrases, it is certainly more closely related to it than 'time': there are obvious, and well known, compatibility restrictions between the verb and 'manner', but none between the verb and 'time'. The situation is that mentioned above p. 14. In order for the relative cohesion between the verb and the 'time', 'manner', 'durational' and 'bound' elements to be reflected in their constituency, it would be necessary to create a new node. In addition to the cases we have been considering, Chomsky (1965: 103-5) mentions the following examples of verbal complements: 'he aimed (the gun) at John', 'he argued with John (about politicsf, 'he talked about Greece',

19 'he ran after John, 'he decided on a new course of action', 'they argued against the proposal, 'they agreed on a new course of action', 'everyone looks up to John', 'they are working at this job'. The do-what and do-so tests show that the italicised phrases certainly are verbal complements, and we thus have a class of verbal complements comprising examples like the above, as well as directional and bound locational phrases. Obviously, this class is far too heterogeneous for the purposes of the present study. In the first place all the above examples passivise quite freely (John doesn't like being argued with, etc.), whereas directional and bound locational phrases will passivise only in rather special circumstances (this island has only been sailed to once, my new hat has been sat on). Second, take pronominalisation. The prepositional complements in the argue with group pronominalise freely. So do the nominals of directional and bound locational phrases (when I found my chair somebody else was sitting in it), but these phrases also have special adverbial pro-forms {there, etc.). For further differences between the various types of bound adjunct, see below chap. 7. Above (1.1) it was observed that some prepositional phrases were like objects and others like adverbials in their relation to the verb, and that there is a gradual transition from one type to the other. From the above discussion it will appear that only the rough outlines of this gradient between objectlike and adverbial functions can be described within the Aspects model, and, in fact, that any grammar will be enormously complicated if the degrees of cohesion between verb and prep.-phrase is to be revealed in constituency. First, since only one type of object is provided for, all prep.-phrases have to be dealt with along with adverbials. Second, there is a limit to the amount of information that can be carried by constituency alone, without having recourse to purely ad hoc devices (cf. above, p. 14). In our case, the Verb Phrase Complement - Verbal Complement dichotomy breaks down into at least a five-way distinction: time - frequency - manner, duration, free location - direction, bound location - argue with, etc. and for this to come out in a phrase marker, the Predicate Phrase would have to be enriched by three new nodes. Moreover, there is no reason to believe that we can stop at this. As more types of adverbials are taken into account (instrument, reason, extent, etc.), we may very well need to invent even more nodes in order to account for their different syntactic potentials, and as was noted above (p. 14) there seems to be no independent motivation for this step. For these reasons, then, one may conclude that the amount of information necessary to account for the full range of prepositional phrases cannot be conveyed by constituency alone.

20 2.1.3 Lakoffs treatment of adverbials Through a series of arguments different from the ones employed above, Lakoff (1968; 1970a: 154-58; 1970b: 156-87) has arrived at the same conclusion, namely that certain facts about the behaviour of adverbials cannot be accounted for within the Aspects framework. TIME ADVERBIALS.

Lakoff (1968: 156) discusses sentences like the following:

(16) Goldwater won in 1964, but it didn't happen in 1968. There is no doubt that the italicised it is an anaphoric pronoun referring to Goldwater won. In a grammar of the Aspects type, however, there is no single symbol dominating only Goldwater won (cf. above 2.1.1 [5]), and accordingly, there is no antecedent for the pronoun to refer to. To remedy this situation Lakoff tentatively proposes that the underlying structure of (16) should be something like (16'):

LOCATIONAL ADVERBIALS. Exactly the same argument is adduced to show that, in deep structure, locational adverbials are outside the clause they modify:

(17) Goldwater won in the West, but it didn't happen in the East. where again, it can only refer to Goldwater won. Another argument appears in (1970b), where the following examples are considered: (18)a. b. (19)a. b.

do you beat your wife in the yard? I don't beat my wife in the yard is it in the yard that you beat your wife? it is not in the yard that I beat my wife

21 Said with neutral intonation the two sentences of (18) are synonymous with the corresponding sentences of (19). Since, in questions and negation, it is usually the highest VP that is questioned/negated, we would expect (18)a.b. to have the meanings of (20)a.b., if the Aspects derivation of them was correct. (20)a. is it beat your wife that you do in the yard? b. *it is in the yard that I don't beat my wife If locational adverbials are regarded as derived from complex underlying sentences like (21), the reading of (19)a.b. is automatically explained:

I beat my wife

? takes place

in the yard

The negation and question will now correctly apply to the VP containing in the yard, rather than to the VP, beat my wife, of the embedded sentence. 10 FREQUENCY a n d REASON ADVERBIALS.

The question/negation argument

shows that these, too, originate in the 'higher' predicate of a complex deep

structure (do you beat your wife often?, do you beat your wife because you don't like her?). MANNER ADVERBIALS. The question/negation argument also applies to manner adverbials: (22)a. b. (23)a. b. (24)a. b.

do you beat your wife enthusiastically? I don't beat my wife enthusiastically is it with enthusiasm that you beat your wife? it is not with enthusiasm that I beat my wife is it beat your wife that you do with enthusiasm? *it is with enthusiasm that I don't beat my wife

This might lead us to postulate deep structures for manner adverbials similar to those of (16') and (21). But Lakoff (1970b: 156-59) further notes that at least some manner adverbials share co-occurrence restrictions with the

22 corresponding adjectives: (25)a. 1. The tailor fitted me carefully. 2. The tailor was careful in fitting me. b. 1. *The suit fitted me carefully 2. *The suit was careful in fitting me. These and similar examples show that sentences with manner adverbials and the related constructions with adjective+sentential complement have identical deep structures, since, obviously, the ungrammatically of (25)b.l.-2. is due to the same restriction. For this reason Lakoff postulates a deep structure like (26) for (25)a. and other sentences containing manner adverbials:11

If a deep structure like this is assigned to (22)a.-b., their interpretation will correctly be expected to be that of (23)a.-b., since, in both cases, the topmost VP, to which the negation/question applies will be enthusiastic. INSTRUMENTAL ADVERBS. In (1970b: 171) Lakoff notes that the question/ negation tests apply to instrumental adverbs in the same way as to locational adverbs, etc.:

(27)a. do you beat your wife with a whip? b. I don't beat my wife with a whip And accordingly sentences containing instrumental adverbials should be derived from complex deep structures. In (1968) Lakoff further notes 12 that sentences with instrumental adverbials share so many features (e.g. co-occurrence restrictions) with the corresponding sentences containing use, that they can profitably be derived from a common deep structure source, and, moreover, that this common source must be pretty close to the deep structure that the Aspects model would postulate for the use sentence:

23 (28)

S NP

PredP ,VP

Frank V used

a whip

P

NP

for

S Frank beat his wife

(irrelevant details omitted)

2.1.4 Discussion Lakoff s analysis of adverbials can be seen as an attempt to remedy some of the shortcomings of Aspects, while at the same time preserving the constituency approach. Lakoff s revisions are motivated in the first place by the demand of descriptive adequacy: In the cases I have illustrated the phrase structure rules of Aspects simply fail to describe how e.g. interrogative/negative sentences containing certain adverbials are understood the way they are. There is also at least one theoretical point in which Lakoff s analysis should be considered superior to Chomsky's: As was noted above (p. 13) Chomsky imposes the demand upon himself that all the necessary syntactic information about a given sentence should be shown in a phrase marker consisting exclusively of a configuration of categorial symbols. His use of such terms as Place, Time, Manner, Direction, etc. must be looked upon as a failure to meet this demand (cf. fn. 5), but in Lakoff s analysis there is no place, and no need, for such 'pseudocategories'. Thus Time, Location, Frequency, and Reason are characterised uniquely by the fact that they are introduced under the rightmost NP in the configuration (29) NP S

V ?

NP

24 where the difference between them comes out unambiguously in the choice of the abstract verb that I have marked '? \ 1 3 In like manner, Manner adverbials are characterised by their derivation from the Adj constituent in the configuration

And Instrumental adverbials are derived from the object NP of the superordinate clause in configurations like (28). I wish to raise two points about this analysis of adverbials. First, an empirical question: The deep structures introducing Time, Location, Frequency, and Reason adverbials differ from each other only in one formative, whereas they differ from those introducing Manner and Instrumental adverbials both in formatives and configuration. Even if we grant the correctness of Lakoffs observations, there is, in principle, no a priori reason why this should be so. The question that we shall have to investigate is the following: are Time, Locational, Frequency, and Reason adverbials similar to each other and different from Manner and Instrumental adverbials in terms of their relations to the verb, as suggested by Lakoffs analysis? Second, a theoretical point: in Lakoffs analysis each adverbial is defined uniquely in terms of a particular configuration and/or a particular formative. If this analysis is extended to all adverbials, the difference between this type of grammar and a functional grammar (cf. 2.2), in which each type of constituent is uniquely characterised by functional labels like 'Manner', 'Instrument', etc., seems to reduce to a notational difference. 14 Note further that phrase markers (28)-(30) are redundant in a way analogous to the redundancy Chomsky wished to avoid (cf. above, p. 12): adverbials are characterised both by the configurations by which they are introduced and by formatives which, as I have argued, are substantially equivalent to functional labels.

2.2

SYNTACTIC FUNCTION

It is an extremely interesting feature of the recent history of linguistics that in the latter half of the 1960's several scholars, independently, arrived at the conclusion that an adequate grammar would have to provide a direct characterisation of functions within the clause. That is, the function of the

25 items is not derivable from constituency as argued by Chomsky. Or to be more precise, there are syntactic phenomena which will remain unexplained if we content ourselves with the functions (subject and object) that can be derived from constituency. In a functional grammar the 'deep' structure of a sentence is regarded as consisting of a predicate (verb) plus a number of roles (cases/participants) associated with it (Fillmore, 1968; Danes, 1968; Halliday, 1967-8; Anderson, 1971). Huddleston (1971: 64) points out that in grammars of this kind 'the distinction between prepositional phrases and NP's will not be of real significance as such', since, obviously, the same function may be expressed now through Prep-phrases and now through NP's (they went round the world/they circumnavigated the world [example from Fairclough, 1965: 111]). The same point is made repeatedly in Fillmore (1968), where it is emphasised that we cannot expect a direct correspondence between deep cases and their surface realisations as subjects, objects, or various prep-phrases/surface case-forms.

2.2.1 Modalities and circumstantials: cases and participants There is an interesting similarity between Fillmore's and Halliday's proposals for a functional grammar, in that both of them employ a distinction between functions which are particularly closely associated with the verb (Fillmore's cases, Halliday's participants), and those where the association with the verb is more free (Fillmore's modalities, Halliday's circumstantials). An examination of this distinction will be of some importance for this study, since it will be reasonable to expect prepositional verbs to occur only with the closely associated functions. It should be emphasised, however, that whereas most writers on grammatical function operate on a very high level of abstraction (cf. Anderson's (1971) reduction of the number of cases to three or four), the task outlined in 1.1 is concerned with a purely surface phenomenon, and therefore it is not to be expected that any existing model of grammatical function is directly applicable to this task. A complete theory of functional grammar should (a) give unambiguous definitions of the functions (and classes of functions) it employs in language independent, semantic terms (Chapin, 1972: 651). This means that the general theory cannot define the distinction between e.g. closely and loosely associated functions by means of a syntactic operation of any particular language, say English, (b) On the other hand, if a particular function is postulated for a particular language, it should be 'detectable contextually — including syntactically' (Cruse, 1973: 15). Unfortunately, neither Fillmore nor Halliday satisfies these demands, and to a large extent we are forced

26 to rely on interpretations of their examples Fillmore (1968: 23) tells us about modalities that they 'will include such modalities on the sentence-as-a-whole as negation, tense, mood, and aspect'. In addition it is stated that modalities are likely to comprise (functions realised as) 'certain temporal adverbs' (ibid). The only additional information about modalities to be found is the highly suggestive (in the context of the present study) remarks that (a) the locational-directional distinction may be determined by constituent structure (p. 25), and (b) the distinction between the locational phrases in examples like (31)a.-b. may point to a distinction between locational phrases which are modalities ([31]b.) and those which are not: (31)a. John keeps his car in the garage. b. John washes his car in the garage. Similarly, since the only relevant constituent structure provided by Fillmore is the distinction between modalities and cases, we must understand his remark about direction and location to mean that the directional phrase of (32) is a case, whereas the locational phrase is a modality: (32) We will come back in the last chapter to a closer consideration of what could conceivably [....] be done [....] (CAPS 67) If such examples can be taken as typical of the distinction between modality and case (but Fillmore indicates that he is not quite sure about this), it will appear that, in English, the distinction is identical with the free-bound adjunct distinction drawn above (p. 16), and accordingly the do what test would be diagnostic: (33)a. 1. *What does John do in the garage? — Keep a car. 2. What does John do in the garage? - Wash his car. b. 1. *What we will do to a closer consideration is come back in the last chapter. 2. What we will do in the last chapter is come back to a closer consideration. Unfortunately things are not as simple as that. Among the cases Fillmore introduces 'Instrumental' (1968: 24), and although the validity of Fillmore's concept of instrumentality has been seriously challenged (cf. Huddleston, 1970), there is one type of function which everybody is prepared to regard as instrumental, namely an element which is 'causally involved in the action

27 or state identified by the verb' (Fillmore, loc. cit.), and at the same time presupposes an animate agent (cf. Lakoff, 1968; Huddleston, op. cit.). This type is exemplified by the italicised phrases in (34)-(35): (34)a. b. c. (35)

The key opened the door. John opened the door with the key. John used the key to open the door. John shot at Peter with Tom's gun.

But when this type of instrumental element is realised as a prepositional phrase, it is always a free adjunct: (36)a. What did John do with the key? - He opened the door. b. What did John do with the gun? - He shot at Peter. In an earlier paper, published as Fillmore (1969 b), Fillmore provides two precise criteria for the modality — case distinction: (a) 'The category Proposition [under which cases are introduced - TV] includes the verb and all those nominal elements which are relevant to the subclassification of verbs' (i.e. in the Chomskian sense), (b) 'Roughly speaking, all adverbial elements capable of becoming subjects or objects are introduced in the expansion of Proposition' (p. 366). 15 The first of these plays no role in the later (1968) version, since in that conception of grammar 'no distinction is needed between "strict subcategorisation features" and at least the highest level of "selectional features" ' (Fillmore, 1968: 52; cf. also Robinson, 1970: 70-71, and my arguments above, 2.1.2, that Chomsky's notion of subcategorisation is untenable anyway). The second criterion is not employed either in Fillmore (1968), presumably because it is too language specific, but as things stand, the ease with which instrumental elements subject- and objectivise is the only available explanation as to why they should be included among the cases rather than e.g. free locational adjuncts, to which they bear a close syntactic resemblance (see further below 6.2). 16 Halliday's distinction between circumstantials and participants corresponds to, but is not identical with, the Fillmorian distinction between modalities and cases, and the suggestion that the criterion for the distinction should be freedom of occurrence as subject and object recurs in Halliday (1967-68, part III: 179, 181): The relation between participants and processes (verbs) is accounted for by the transitivity system of the language, and 'participants . . . [are] syntactically, those functions having in general . . . nominal realisations'. It is not quite clear, however, whether the decision to leave out elements realised by prepositional phrases is due to a self-imposed method-

28 ological limitation, or whether it is an essential part of the theory (cf. op. cit., part III: 180-81). Hudson (1967: 242-43) states quite clearly that 'nuclear adjuncts' occurring with e.g. pay for, wait for, sit on, eat from are to be accounted for within the transitivity system. Since, in any case, a criterion relying on the surface form of the elements (as subjects or objects) has to be hedged in by qualifications such as 'roughly speaking' or 'in general', I shall not pay further attention to it here, but turn to a more fruitful criterion suggested by Halliday (1967-68, part III: 196), namely the possibility of recognising clause types (and participant types associated with them) 'by reference to the preferred form taken by the related identifying clause having just the verb as the identifier' (i.e. a pseudo-cleft clause). It appears that each of the 'central' participant roles identified by Halliday (see below 2.2.3.1) is characterised by its capability of occurring at N in at least one of the schemata: (37)a. b. c. d.

What What What What

happened to N' was that Pro-N' V N' did to N" was V Pro-N" N' did was V N' made N" do was V

(e.g. John walked out on Mary - What John did to Mary was walk out on her.) - with the addition that for some elements (in particular, direct objects in double object clauses) with should be permitted instead of to in b. In principle (37)a.-d. constitute a clear and objective criterion for a distinction between certain types of functions, and in the following I shall use them with the modification suggested by Cruse (1973), who, in order to avoid unclarities, reformulates a criterion of this type in terms of statements of entailment. Thus: (38) John walked out on Mary entails a. John did something to Mary b. Something happened to Mary but not c. Mary did something Although significant distinctions can be based upon entailment relations building on (37) (see below chap. 8), they are not particularly helpful in a study of prepositional verbs, in that the functions with which prepositional verbs most frequently occur (Halliday's 'beneficient', 'range' and 'phenomenon') would be grouped along with much more marginal functions such as reason or manner: In none of the examples (39)-(42) will happen, do and make tests reveal anything about the nature of the italicised nominals' associ-

29 ation with the process: (39) (40) (41) (42)

I have thought of several improvements. Smith deals in cars. She cooked for her father. The adults emigrate from their neighbourhood owing to pressure from the young.

In fact, only in (42) does an entailment involving happen,/do¡make

apply:

(42') Pressure from the young makes the adults emigrate from the neighbourhood. 17 The conclusion that the owing-to phrase of (42) is more closely related to its verb than are the phrases of (39)-(41) would, in the first place, be highly counter-intuitive, and, secondly, it is not corrobated by the do what test (what they do owing to pressure is emigrate - *what Smith does in cars is deal). Therefore we cannot rely on these criteria to isolate the types of functions with which prepositional verbs occur from other, more marginal functions, and the most generally applicable criterion still seems to be the do what test. This test can now be reformulated into a statement of entailment. We shall say that an element is a participant if the relation (43) Subject-V-P-N entails a. Subject did something P-N b. Something happened P-N does not hold. Thus, of the italicised elements in (39)-(42) only those of (39)-(40) are true participants, since it is only (41)-(42) that exhibit entailment relations of the pattern of (43): she did something for her father, something happens/they do something owing to pressure from the young. (43) will establish a two-way distinction between true participants and other elements. But a functional point of view forces us to recognise (at least in surface structure) yet a third kind of element, exemplified by: (44) Mike will disagree in any case. (45) To my surprise, John has paid for the car. These are characterised by being related to

30 (44') Mike will disagree. This is true in any case. (45') To my surprise it is true that John has paid for the car. As the near-paraphrases show, elements like these do not perform a function IN the clause in which they occur. Rather, they are comments ABOUT its truth-value, likelihood, desirability, etc., indications of the sense in which it should be understood or limitations on its applicability. This type of element I shall refer to as 'non-role playing', and no more will be said about it here. Below 3.1 the diagnostics used will be formalised, and a brief survey will be given in chap. 4.

2.2.2 Modalities, circumstantials and free adjuncts In this section I give a survey of the functions which, although role playing, do not qualify as true participants. They will be dealt with in detail in chap. 5 - 6 , and no rigorous definitions will be attempted here. Most of the terms are taken over from traditional grammar and are, hopefully, self-explanatory; where this is not the case references and explanations are added. Free adjuncts fall into two clear-cut semantic groups, a concrete group and an abstract group, according to the exponent of N. Syntactically, the dividing line is less sharp, but still clear enough for the distinction to be valid. The cumulative effect of the criteria applied in chaps 5 - 6 demonstrates that the members of the concrete group have much more in common with true participants than do the members of the abstract group (see pp. 73, 94). 2.2.2.1 The abstract group 1. TIME. (46) The growth of workers' militancy [....] was described by [....] the British Prime Minister [....] as "one of the big problems of democracy as we enter the 1970's," when he spoke in New York on Monday at the start of his visit to the US. (GW 31.1.70, 24: 1) 2. DURATION.

(47) /well I had t o f f I/had to work for six months you seetf /get me over this w i n t e r ^ (s.lb.6-3)

-- to

3. FREQUENCY. At this stage the term can be taken to be self-explanatory. Note that it covers both place- and time frequency:

31 (48) At intervals doors led from the corridor into the classrooms. (49) [....] the love-songs would become more and more doleful, while between each Larry would pause to inform whichever member of the family happened to be present that spring, for him, did not mean the beginning of a new year [....] (FOA 83-84) 4. REASON. I shall use this term in quite an extensive sense, having it cover 'reason' in the narrower sense of the term as well as 'conditional', 'adversative' and 'concessive' (this decision will be justified in 5.1.4). Here, I shall illustrate only reason and adversative: (50) Over-reliance on advertising revenue makes for an unhealthy economic situation both because of its effect on competitive strength as between papers and because of its cyclical variability, (CRNN 8) (51) Despite increased competition from other newspapers using modern selling techniques, this newspaper's share has only marginally decreased [....] (CRNN 38) 5. EPITHET. The term derives from Schreiber (1972) and covers the type of adjunct termed subject adjunct by Greenbaum (1969: 171-73) and Quirk et al. (1972: 465-69). 18 Lakoff (1970b: 157-59, 165-67) considers them (a type of) manner adverbials derived from a superordinate predicate (see above 2.1.3). When realised as adverbs, they are homonymous with manner adjuncts (cf. he foolishly answered all the questions, he answered all the questions foolishly), but as pointed out by Greenbaum (op.cit.), the two differ in that Epithets characterise (the disposition of) the subject, whereas manner adjuncts characterise the manner in which the process identified by the verb is carried out. The morphological similarity with manner adjuncts recurs when they are realised as prepositional phrases: the same prepositions are used, and in both types the exponent of N is frequently a derived (deverbal or deadjectival) noun. (a. = epithet, b. = manner): (52)a. Turner held the jacket towards him, but Gaunt drew back in distaste. (JCSG 85) b. Sometimes they [sc the horns] sounded in series [....] sometimes in unison [....] (JCSG 13) (53)a. It is impossible not to admire the industry and assiduity with which Dr Lambert and his team of ardent young researchers bend to their lugubrious labours [....] (CAPS 70)

32 b. [the most remarkable feature of this winter has been] the rapidity with which the snows have disappeared from the fells. (GW 31.1.70,12: 1) 6. MANNER. The class traditionally referred to by this term minus epithets. For examples, see (52)b.-(53)b. 7.

MEASUREMENT.

(54) [....] the proportion of homes with television having been rising, though at a diminishing rate, over the whole decade. (CRNN 6) (55) Industrial exports increased in 1969 by 13 per cent. (GW 31.1.70, 6 : 2 ) 2.2.2.2 The concrete group 1. F R E E LOCATION. The distinction between free and bound locational elements was established above 2.1.2, for some remarks about borderline cases, see below 7.1.1.1. No further exemplification will be needed at this stage.

2. INSTRUMENT/MEANS. I use these terms only in the narrow sense defined above p. 26f. Note that Means is grouped along with Instrument in spite of the fact that its N is abstract. (56) I don't think he forgot a word anyone said to him. He could listen with his eyes, Leo could. (JCSG 103) (57) [such transitions can be understood] only if we make some attempt to take account of their interactions, which is normally done by means of terms like ajvi. (8a.3.24-2) 3. HAVING. This term covers some prepositional phrases in with/without. The type bears some resemblance to Instrument as well as to Comitative, but can be distinguished from the former by its failure to have a use paraphrase, from the latter by the fact that nouns in this function need not share the selectional features (in Chomsky's sense) of the subject of the clause. (58) The family frowns disapprovingly, and then just like an old Hollywood movie, someone enters with an important message. (GW 31.1.70, 20:4) 4.

COMITATIVE.

This function is characterised (a) by a paraphrase relation-

33 ship which permits the subject of the clause and the N of the prep phrase to be coordinated (cf. Jespersen, 1924: 90); (b) by the possibility of replacing the prep phrase by N+help, accompany and others (Walmsley, 1971) — Tom went to London with Bill, Bill accompanied Tom to London. Although Walmsley (op. cit.) has produced convincing arguments that Comitative and Instrumental should be considered subtypes of one function,19 there are also good reasons to keep them distinct (cf. below 6.1.4). (59) /Mr Brown^. who/1 was . appeared on a television programme with the other d a y # (5b. 16.25) 5. BENEFICIENT. This function should be distinguished from Recipient (below 2.2.3.2), with which it shares some morphological and semantic features (the preposition for is typical of both, and both typically have animate nouns at N). Fillmore (1972) has argued that sentences with Beneficients should be derived from a complex source with an abstract GIVE and the (surface) Beneficient as Recipient. Thus: I GIVE you [Rec] (I write the letter) [Objective] -»• I write the letter for you. In surface structure the two functions are quite distinct, however, as can be seen from the facts that (a) they can co-occur in the same sentence (Corder, 1968: 18; Huddleston, 1971: 90): will you buy some flowers for my wife for me?, and (b) a sentence may be ambiguous between a Beneficient and a Recipient interpretation: he bought some books for we. 20 These examples also illustrate the semantic difference between them: The Beneficient benefits from the process (the buying is done ON HIS BEHALF), whereas the Recipient benefits from a particular participant in the process (the flowers in the former example, the books in the latter). 21 (60) '[....] I have no idea why he telephoned at this hour, nor why he should call a meeting [....]' 'All the same, I accepted for you. It seemed wise.' (JCSG 23)

2.2.3 Cases, Participants, and bound adjuncts Once it has been decided to regard a nominal's possibility of occurrence in entailment statements as critical of its function (above p. 28 f.), a controversial question in functional grammars is solved automatically, viz. the question of the status of the functional labels. In Fillmore (1968) the case labels had the same status as the categorial symbols of Chomsky (1965): case labels are dominated by Proposition), and in their turn they dominate K(asus)+NP.

34 An alternative approach regards functional labels as much closer to features (cf. Anderson, 1971; Fillmore, 1969a; Halliday, 1967-68; Huddleston 1970, 1971). Apart from differences of formalisation, which need not concern us here, the crucial difference between the two approaches lies in the fact that the latter but not the former allows functional labels to occur in bundles. 2 2 Now let us consider the function of the prisoners in (61) John marched the prisoners. This sentence entails both (a) something HAPPENED TO the prisoners and (b) the prisoners DID SOMETHING (cf. Cruse, 1973: 14). If happen and do something entailments are regarded as criterial of the roles Affected and Agentive, respectively, it follows that the prisoners must realise both these roles. 2 3 One more general point remains to be discussed: the one-instance-perclause principle of Fillmore (1968: 21, 1972: 248). This principle stipulates that 'there is in a single clause at most one noun-phrase [....] serving a given case role'. (1972: 248). Fillmore further points out that acceptance of this principle requires that we 'deal with apparent counter-examples either by showing that the putative identical case roles are in fact distinct, or by showing that the construction is better treated as an instance of clause embedding' (loc. cit.). Thus Fillmore would have to regard (61) as derived from a complex deep structure, since obviously both John and the prisoners are Agentive (both DO SOMETHING). A plausible deep structure would be something like (61'): (61') John [Agt] CAUSE (prisoners [Agt] march) [Obj] 2 4 An alternative analysis, inspired by (Cruse's, 1973, interpretation of) Halliday (1967-68), would argue that John combines the roles Agentive and External Causer ('John MADE the prisoners do something'), and that the prisoners combine the roles Agentive and Affected ('something HAPPENED TO/WAS DONE TO the prisoners'): (61") John [Agt-EC] marched the prisoners [Agt-Af] 2 5 The two analyses of (61) differ in two respects, (a) (61') contains no combinations of functions in the same nominal, and (b) only (61') observes the one-instance-per-clause principle. As for (a), this provides no basis for a decision between them, since the use of entailment relations forces us to recognise combinations of functions anyway. As for (b), although Fillmore insists

35 on the necessity of this principle, he has never told us why it is necessary. Moreover, the difference between the two analyses seems to be of a purely notational kind, and (6l")certainly has the advantage of not postulating a deep structure whose validity depends crucially on an abstract causative verb the legitimacy of which is not at all uncontroversial (cf. Partee, 1971: 7-8). Finally, there are some predicates for which the restriction on the number of occurrences of each function in a clause is extremely difficult to justify (Huddleston, 1970; 1971: 73), namely the symmetric predicates of Lakoff & Peters (1969), such as John resembles his father, Chomsky agrees with Descartes, and certain predicates 'expressing various temporal or spatial relations' (Huddleston, 1971: 73), such as X is to the right of/to the left of/ in front of Y, and X precedes/follows Y. As for the former type Fillmore (1972: 249) argues that the two nomináis involved should be regarded as representing distinct functions, since their order is significant. This he puts down to the fact that predicates of this type involve 'the judgment that certain properties observable in [the first noun-phrase] are relatable to properties attributable to [the second noun-phrase], with the second noun-phrase serving to identify a standard according to which the entity named by the first noun-phrase is assigned some sort of position'. (Fillmore's italics). Fillmore's observations are correct, but his conclusion does not necessarily follow. Fillmore's case grammar is concerned exclusively with that dimension of syntax which Halliday (1967-8) refers to as 'the experiential/cognitive component', to which the 'transitivity system' (above p. 27) relates. Halliday employs two other systems, the 'modal system', which need not concern us here, and the 'thematic system', which is concerned with 'all those choices involving the distribution of information in the clause' (1967-8, part I: 37), 'with the status of the elements [....] as components of a message: with the relation of what is being said to what has gone before in the discourse, and its internal organisation into an act of communication' (part II: 199). It now appears that the difference between the two nomináis associated with symmetric verbs like resemble and agree is precisely of the kind provided for by the thematic component. The order of the elements John - his father, Chomsky - Descartes is fixed because part of 'what has gone before' is the knowledge that his father and Descartes are older than John and Chomsky, respectively, and it is for this reason that the only natural 'organisation of the act of communication' is the one which regards his father and Descartes as the 'standard according to which the entity named by the first noun-phrase is assigned some sort of position'. To sum up: In the preceding pages I have argued (a) that a single nominal should be allowed to realise a bundle of functions, and (b) that the same

36 function should be allowed to occur more than once in a clause. In the two following subsections I shall define and exemplify the true participant functions which I consider necessary for a (surface, cf. above p. 25) description of English. For practical reasons the order of presentation differs from that of chaps 7-8 (and from the corresponding part of 2.2.2), where the sequence will be 'less closely involved' — 'more closely involved'. This is motivated by the fact that some of the less closely involved participants are defined by their failure to exhibit the relations of entailment of the more closely involved participants. Finally, note that what I propose to do here is to give instances of INDIVIDUAL functions. For this reason combinations of functions in the same nominal will be commented on only where required by the argument. 2.2.3.1 Central participants The four functions involved in this group all exhibit characteristic paraphrase relations, and although there is no one-to-one correspondence between the functions as defined here and those of Danes (1968) and Halliday (1967-8), I believe that they are, essentially, compatible with them. All four frequently, indeed normally, occur as subjects and objects, and in some cases the prepositional realisation must be regarded as derived (see further below chap. 8). 1. EXTERNAL CAUSER. This function was illustrated in (61) above. It is characterised by its relation of entailment to statements of the type 'X made Y V/something happen to/with Y'. EC always permits a do paraphrase (cf. Halliday, op. tit., part III: 184), but is notionally distinguished from Agentive in that it denotes something exerting a force through its mere presence and not through an internal energy source (cf. Cruse, 1973). Cruse's 'Initiative' can probably be regarded as a combination of Agt and EC in the same nominal (e.g. John in [61]). Fillmore (1972: 253-4) explicitly argues against the recognition of EC, which he prefers to regard as a subtype of Ins. I include it since it is recognisable by the make paraphrase. The type of EC exemplified in (61) where it combines with Agt never seems to be realised prepositionally. But outside this combination prepositional realisations are quite common. Note that EC can be realised only as subject or prepositionally, never as a non-prepositional object. (62)

[I took a bucket of water and threw it at the dog] When I did so, it fell over with the shock and lay whining at the memory of quieter days [....] (GW 31.1.70, 16: 3)

37 2. AGENTIVE. Clauses with an element in this function always entail 'X did something'. Cruse (1973) points out that this is a necessary, but not always sufficient criterion. For this reason he supplements it with the notional definition 'an action performed by an object which is regarded as using its own energy in carrying out the action' (21). Fillmore (1968: 33) states that 'If there is an A, it becomes the subject', and prepositional realisations of Agt in active sentences are certainly rare. In the only type recognised here, Agt combines with Com: (63) The printing unions can best express their concern for the continuation of a diversified newspaper industry by [....] joining with management in devising proposals to accelerate improvements in productivity. (CRNN 15) 3. EXPERIENCER. Danes (op. cit.) uses this term about one of the participants associated with an emotional verb {hate, like, love, please), and notes that the function can be isolated by the question form 'what is X's attitude to Y'. I shall use the term in a somewhat broader sense, including also verbs of cognition and perception. The term will now cover the instances which Fillmore (1972: 261-2) seems to have in mind, though its coverage will be narrower than that of Halliday's 'processer' (1967-8, part III: 193), which includes the subjects of he liked the play, he heard a noise, he believed the story as well as the Agt subject of he said he was coming. The typical realisation is non-prepositional, but prepositional realisations occur: (64) [a comprehensive school] which at one point he commends, in terms which will not appeal to every parent, as 'a school that never says no' [....] (CAPS 51) 4. AFFECTED. This function is taken over directly from Danes (1968: 58-9), an affected element can always occur at Y in statements of the type 'X did something to/with Y; something happened to Y'. This use of the term should not be confused with Halliday's (op. cit.). Fillmore nowhere recognises this function. In (1968) animate and inanimate 'undergoers of the action' are labelled 'Dative' and 'Objective', respectively; in (1972) all affected elements, and a great many others, are referred to as 'Objective' regardless of animacy. Prepositional realisations are quite frequent, and moreover, the function seems optionally to combine with any (prepositional) object of a verb involving physical contact (cf. below chaps 6-7). (65) [...] we/shouldn't be imposing on you in vacation t i m e ^ (s.lc.l 1-14)

38 2.2.3.2 Marginal participants This final group of functions is the most important one for a description of prepositional verbs. By far the majority of them take 'objects' in one of the marginal participant roles. The group is defined negatively as consisting of participants failing to satisfy any of the criteria for central participants. The importance of these functions for a study of prepositional verbs necessitates subtler distinctions than were made in the preceding sections. From a theoretical point of view there is no doubt that the system set up is far too detailed, and that several of them would be treated as one in a general theory of syntactic function. However, the system is designed first and foremost as 'a notional system for ad hoc description' (cf. Chapin 1972). The first demand that such a system must satisfy is that it should be intuitively convincing, i.e. the classes it postulates should be easy to recognise and doubtful cases should be relatively rare. On the other hand, it seems to be of some theoretical interest that for all the functions postulated here — except two, Path and Purpose — Source, Goal and Neutral subtypes stand out clearly. 2 6 1. LOCATION. This term accommodates part of the phenomena referred to as 'range' by Halliday (1967-8), part I: 58), e.g. he climbed THE MOUNTAIN, he jumped THE WALL. Locational elements in participant roles are restricted to verbs of movement and position. The Source (ablative), Goal (allative) and Neutral (locative) types are clearly distinguishable. Source: (66) /Nightingale said that he . he might want to get away Lower ffNetherhall you s e e ^ (s.lc. 11-18)

from.

Goal: (67) theff t w e n t y / f i f t h ^ t w e n t y # / s i x t h I flew to San (s.lb.6a-5)

Francisoff

Neutral: (68) The rock was first charted 400 years ago but it has only been landed on five times since then. (GW 31.1.70, 10: 1) 2. PATH. At a higher level of abstraction Path will be regarded as a subtype of Loc in which Goal and Neutral are combined (Anderson 1971: 170). At a more superficial level it is necessary as a descriptive convenience allowing us to handle cases like (69), in which more than one locational participant occur in the same clause:

39 (69) Now when the male comes in search of the female he must walk over the moss to the trapdoor [....] (FOA 77) 3. PURPOSE. On the one hand it is tempting to regard Purpose and Reason as subtypes of the same hyper-function; Reason would then be the Source type, and Purpose the Goal type, note also that both may be elicited by the question form why. On the other hand there are reasons for regarding it as a subtype of Objective (cf. below 7.1.4.1). In some cases there is a superficial similarity to Phenomenon-as-Goal elements with for at P (e.g. he called for his pipe). This latter type, however, lacks both the why question form and the paraphrase with an in order to clause characteristic of the former. (70) The fact that I had no money on me did not worry me in the slightest, for I would simply tell the man to call at the villa for payment the next day. (FOA 48-9) 4. DATIVE. For lack of a more suitable neutral term, I use the term in the sense of Halliday's beneficiary (1967-8, part I: 57), an element which benefits from/suffers from the lack of another participating element. Nouns in this function are typically, but not exclusively, personal. The Source (privative), Goal (recipient) and Neutral (possessive) types are illustrated in they robbed HIM of his money, they gave HIM the money, HE has the money. Source: (71) All individuals begged from each other, by pushing at the mouth of the other, and even biting him. 2 7 Goal: (72) [....] an advantage accrues to the paper which can operate at a high level of output [....] (CRNN 8) Neutral: (73) Any appointments board must keep notes on the people for whom it is trying to find jobs. The notes belong to the board [....] (GW 31.1.70, 12:4) 5. INTERLOCUTOR. There is a close parallelism between this and the preceding function. Both of them require the (implicit) presence of another participating element, Phenomenon and Objective, respectively, and both frequently occur with 'double object verbs' in the sense of Corder (1968). The distinction between them is reducible to the class of verb with which

40 they occur: Int (and Phenomenon) with verbs of communication and cognition ('psychological verbs'), Dat (and Objective) with other verbs. Thus, with psychological verbs the second element refers to the content of the communication/cognition, with 'dative verbs' it refers to the thing which the Dat participant benefits from/suffers from the lack of. Source: (74) '[....] What happened? ' 'It was just a fight. They say he asked for it, that's all.' 'How?' 'I don't know. They say he deserved it, see. I heard from my predecessor [....]' (JCSG 87) Goal: (75) 'I am not obliged to report to Siebkron; nor do I intend to. [....]' (JCSG 23) Neutral (76) he /had the guts to resign t h o u g h ^ when he/didn't agree with the government^2* (5b.16.31) 6. EQUATIVE. This function is postulated to take care of the participants associated with 'intensive' clauses (cf. Halliday, 1967-8, part I: 39-40). Traditionally, only copulative verbs are considered 'intensive', but there are quite a few prepositional verbs which are notionally equivalent to the copula type in that they state an identity between two participants or attribute qualities of one participant to the other. Fillmore (1968: 84) explicitly states that the type of phenomena that I have in mind constitutes a problem for his theory. In (1972) he regards the Source and Goal types as simple 'source' and 'goal'. As I pointed out above (p. 35) the Neutral type is the main stumbling block for the one-instance-per-clause principle. At any rate, there is no doubt that a more general theory of functional grammar than the one proposed here would regard. Eq and the two following functions as subtypes of one hyperfunction. Source (originative): (77) Once Spiro had taken charge he stuck to us like a burr. Within a few hours he had changed from a taxidriver to our champion [....] (FOA 32-33) Goal (translative): (78) [the sort of mob war] that/makes a protest against something^ how/ever u n j u s t ^ de/velop into an organised riot at a meeting^ (5b.16.1)

41 Neutral (essive) (79) In our second example, the expression which we, and Geach, wish to classify as a B 2 has a complexity which the corresponding expression of the first example lacks. It consists of a verb and an adjective ('is wise') instead of a verb alone ('smokes'). (8b.1.145-2) This term derives from Halliday ( 1 9 6 7 - 8 , part III: 1 9 3 ff). One way of defining its scope would be the statement (op. cit.: 193) that it occurs with verbs of 'reaction, perception, cognition and verbalization'. Another distinguishing feature of Phen (op. cit.: 194) is 'its ability to take the form not only of a "thing" (a phenomenon of experience) but also of a "meta-thing" - a fact or a report'. For the affinity between Phen and Objective (Halliday's 'range') see below 7 . 1 . 8 . 1 . 7 . PHENOMENON.

Source: (80) He was smiling, but only at his own stupidity. (JCSG 71) Goal: (81) We must now inquire what further information we may hope for from any detailed theory of a change of state that may be developed. (8a.3.22-l) Neutral: (82) Do the mass of the British people know or care enough about the present colonies [....] (GW 31.1.70, 11:1) 8. OBJECTIVE. Fillmore (1968: 25) describes his objective case as 'the semantically most neutral case, the case of anything representable by a noun whose role in the action or state identified by the verb is identified by the semantic interpretation of the verb itself; conceivably the concept should be limited to things which are affected by the action or state identified by the verb'. In (1972: 251) Fillmore states explicitly that this case is used as a 'wastebasket'. Unfortunately a wastebasket is necessary also in this study, but by redistributing its contents among three other functions (Af, Eq, Phen) I have imposed some limitations on its use. Similarly, the range of phenomena covered by Obj as employed here is more restricted than Halliday's (1967-8) 'range', which also accommodates bound Loc (above p. 38). On the other hand Danes's 'specification' (1968: 59-60), which is defined as occurring with some 'peripherally transitive' 'engagement Verbs', e.g. he plays tennis, he studies Greek, where the object approaches the status of a cognate object, only covers a subclass of Obj.

42 Source: (83)

'Did he live above his means? ' 'How should I know what means he has? If he lived on what he got here, I suppose he lived up to them'. (JCSG 68)

Goal: (84)

Neutral: (85)

[....] I stood and watched, fascinated by the gleaming barrels and stock, sniffing rapturously at the rich heavy scent of the gun-oil. (FOA 90) From the 1870's onwards, when British industry began to run out of steam [....], the improvement in living standards was secured by the great retail revolution [....] (GW 31.1.70, 22: 3)

NOTES 1. I use syntactic structure as distinct from syntactic function. The distinction will be returned to presently. 2. For references see below 2.2. 3. I am aware that Aspects is not the final version of Chomsky's view of grammar. But the revisions it has undergone do not seem relevant for the present issue. 4. The possibility of having adverbials directly dominated by S is not provided for in the rules given in Aspects (102, 107). But Chomsky's rule (52)i., = (5)i. below, may be extended readily to cover sentence adverbials without any conflict with the theory (cf. Burt 1971: 243) 5. Chomsky actually fails to observe his own principle that all relevant information should be displayed in configurations of categorial symbols by using such 'pseudocategories' (Fillmore, 1968: 16) as Place, Time, Manner, etc. It is difficult to see what these terms are if not functional labels, used to account for the fact that the exponents of these functions may be of a variety of categories, e.g. Adverb, Prep phrase, Clause. For the difficulty of giving configurational definitions of grammatical functions see further Huddleston (1971: 63 fn. 2), and cf. below 2.1.4. 6. There are extremely interesting parallels between Chomsky's analysis of adverbials and that of Diderichsen (1943, 1964, 1966a: 181-83). Diderichsen establishes three types of adverbials: nexus adverbials, free adverbials, content adverbials, and, substantially, these types coalesce with Chomsky's Sentence Adverbial, VP Complement and Verbal Complement. 7. Detailed treatments of sentence adverbials are found in Greenbaum (1969) and Hartvigson (1969). 8. The only exception to this that I am aware of is with the verb last, which, however, may be regarded as a lexicalisation of the function Duration (cf. below fn. 14). 9. In despair, with equanimity, with patience all belong to the type discussed under the label 'manner' by Chomsky. Below 2.2.2.1 I shall introduce a distinction between this type and manner proper. 10. Lakoff's analysis of locational adverbials applies only to FREE location, as will appear from a comparison of (18)a.b. and (i)a.b.:

43 (i)a. b.

Do you deposit your salary in the bank? I don't deposit my salary in the bank.

If these examples are said with neutral intonation, what is questioned/negated is not the locational phrase but the whole deposit my/your salary in the bank. This, of course, is another piece of evidence for the correctness of the distinction between free and bound location. 11. Enthusiastic, careful, etc. will accordingly have to be regarded as transitive adjectives with sentential objects. This class of adjectives is necessary anyway, as witness guilty (of), responsible (for), aware (of), etc. 12. It may be helpful to remember that of the three works by Lakoff which we are discussing, (1970b) was the first to be written — in 1965. 13. (29) is the deep structure given for Time and Location in Lakoff (1970a: 156). In (1970b) Frequency adverbials are introduced by means of often dominated by V in phrase markers like (29). In view of the fact that the exponent of Frequency may be many times, at frequent intervals, now and again, etc. there is no doubt that Frequency, too, must be dominated by the rightmost NP. 14. See Walmsley (1971) who observes that the use of (28) can be regarded as no more than a lexicalisation of the function Instrumental. 15. The qualification 'roughly speaking' is necessary to exclude the temporal phrase in (i) from the status of a Proposition constituent, which might otherwise be conferred upon it on the strength of (ii) (i) (ii)

By midnight he was fast asleep. Midnight found him fast asleep.

But it is safe to say that temporal elements do not TYPICALLY occur as subjects outside intensive clauses (e.g. tonight was fun). 16. Note, however, that objectivisation of instrumental elements is restricted to the verb use (and synonyms), which can be regarded as lexicalisations of the functional relationship (cf. above fn. 14). 17. It should be emphasised that this is not how Halliday uses the test. In his version it is employed only to identify the type of relationship between the subject/object/indirect object and the verb. 18. In this study 'subject adjunct' refers to a purely syntactic property shared by a number of functionally different adjuncts, see below p. 55. 19. Note in passing that in Hungarian the Instrumental and Comitative functions are expressed through a single case form (Tompa, 1968: 202), and although Finnish has two distinct case forms, they have been shown to be reducible to one on the grounds of complementary distribution (Sebeok, 1946: 8 esp. fn. 5). 20. In some languages the two functions are reported to have distinct case forms (Fretheim, 1969: 32). 21. Note that Halliday (1967-8, part I: 57) uses the term 'beneficiary' for the function which I refer to as Recipient (below p. 39): 'a beneficiary requires, in English, A PARTICIPATING ELEMENT TO BENEFIT FROM' (emphasis mine). 22. Elsewhere (Vestergaard 1973b) I have pointed out that sentences of the type he loaded the truck with hay present a serious problem for the 1968 version of Fillmore's case theory because of the fact that in that theory combinations of functions are not provided for. 23. If the position that functions occur in bundles is accepted, an extremely interesting

44 theoretical problem arises: what (if any) are the restrictions on the combinations of functions within one and the same nominal? What functions may combine, and what may not? Do languages differ typologically in this respect? Such questions are well beyond the scope of this study, but for some comments see below chaps 6 - 8 . 24. The notation derives from Huddleston (1971: 73). It is not clear to me whether the Fillmorian theory would regard the embedded clause as Objective, the label under which embedded clauses are normally introduced (1968: 28), or Factitive, 'the case of the object or being resulting from the action or state identified by the verb . . (op. cit. : 25). In a later paper (1972: 252) Fillmore does provide for the introduction of embedded clauses under 'factitive' (there referred to as goal). 25. This analysis is quite similar to Huddleston's (1971: 73). Huddleston, however, does not distinguish between Objective and Affected (for this distinction see below), and thus regards the prisoners as [Agt-Obj]. 26. This, of course, is the kind of observation motivating 'localistic' case theories (Anderson, 1971; Zsilka, 1967). Anderson (op.cit.: 120) proposes that Goal (allative) and Neutral (locative) should be considered variants of the same 'deep' case, and I have elsewhere produced some evidence in support of this (Vestergaard, 1974). I still believe that Anderson's proposal is correct, and below, chaps 6 - 7 , I shall present in passing some evidence that it applies not only to the concrete locational categories, but also to the non-local Goal-Neutral distinction. 27. S. A. Barnett, 'Instinct' and 'Intelligence' (Pelican Books) p. 72-73. 28. Corder (1968: 21, fn 1) explicitly considers this type 'the prepositional realization of the Comitative case'. In the system employed here, there seems to be no reason for having a special casual category to accommodate one of the participants associated with symmetric predicates. At any rate, the difference between the two participants occurring with these predicates can be regarded as thematic (cf. Huddleston, 1971: 75, 289-90). Note that the function which I have referred to as Com (above p. 32 f.) is quite different from the superficially similar prepositional complement of symmetric predicates. Thus (59), above p. 33, is related to (i), but (76) is not related to (ii): (i) (ii)

What I did (together) with him was appear on a television programme. *What he didn't do with the government was agree.

3. CRITERIA

We can now posit two distinct but, ex hypothesi (cf. above p. 25), related sets of criteria to be used in the analysis of V-P-N clusters. The first set, the functional criteria, is motivated by the observation that in some clusters N is object-like and by the discussion (2.2.3.1.-2) of what the functions typically realised by objects are. The second set, the structural criteria, is based in part on the discussion of constituent structure (2.1): we will expect the V-P of a prepositional verb to exhibit all the characteristics of a transitive verb, and the N to exhibit all the characteristics of object nominals. The criteria will be presented in order of increasing strength, and for each criterion some attempts will be made to explain why it has the effect that it has. It should be noted that a certain freedom must be granted the linguist applying transformational criteria, since in some cases the transformation is blocked for reasons irrelevant to the criterion as such. In those cases it is the linguist's privilege to remove the factors blocking the transformation. Fairclough (1965: 97) mentions two examples of this: neither they stared at each other nor Emma asked for her coat passivises, although, in general, both stare at and ask for do. In the cases mentioned the restriction at work is the one which Postal (1971) has later called the cross-over constraint, which blocks transformations involving the crossing of two coreferential NP's.

3 . 1 . FUNCTIONAL CRITERIA F. 1.

a. P-N- it is

li!" ue } I the case)

that S

b-S.Thisis

Ce'casel

P N

"

S-P-N t> that S d. S. This is

N ad j

(For symbols and conventions, see.above pp. xi-xii). This criterion is taken

46 over from Bartsch (1972: 22), and was used informally above (p. 29 f.). It is the weakest possible criterion, since it states merely that if P-N is not a role-playing constituent in the clause in which it occurs, we will not expect V-P to be a prepositional verb. We shall say that a cluster satisfies this criterion only if it satisfies all its conditions. The cluster hatch out to my disappointment of (1) is an example of a cluster not satisfying F. 1, since it obviously violates conditions c. and d.: (1)

To my acute disappointment [....] the babies hatched out during the night. (FOA 38) c. Disappointingly it is true that the babies hatched out during the night. d. The babies hatched out during the night. This is disappointing. a. b. c. d.

F. 2.

S-P-N t>

Sb do something P-N something happen P-N something take place P-N something occur P-N

This criterion states that for a V-P to be a prepositional verb the (P-)N must be a true participant rather than a circumstantial element (cf. above pp. 28-9). By this criterion in the House of Commons of (2) is a circumstantial element (although a role-playing one, cf. F. 1) and we will not expect to find a prepositional verb talk in. (About) it, on the other hand is a true participant, and talk about will be regarded as a possible prepositional verb. (2)

/I get u p f f and /talk about it in the House of Commons about the # H bomb and d e f e n c e ^ (5b.16.25)

entails a. I do something in the House of Commons.

does not entail a'. I do something about it.

3 . 2 STRUCTURAL CRITERIA

Within this set I distinguish between a. criteria relating to the adverbial/ nominal status of (P-)N, and b. criteria relating to the unitary status of V-P. Note that in some cases the distinction is bound to be arbitrary; passivisation (N. 16), for instance, could be listed in either group, since it tells us both

47 that N behaves as an object noun and that V-P behaves as a single transitive verb.

3.2.1 Criteria for (P-)N N.l. S-P-N = impS-P-N The diagnostic value of this criterion is noted by Greenbaum (1969: 112). In (3) it will distinguish between e.g. the clusters go ahead by the nature of things (which does not satisfy the .criterion) and go ahead in the procession (which does): (3)

[They] sat beside the Tills, who by the nature of things had gone ahead of them in the procession. (JCSG 37) a.l. *Go ahead of them by the nature of things! 2. *By the nature of things, go ahead of them! b. Go ahead of them in the procession!

Halliday (1967-8, part II: 221) notes that 'modal adjuncts' (such as by the nature of things) 'cannot be thematic if the mood is already marked thematically'; this, presumably, accounts for the ungrammaticality of (3)a.2., but not for a.l. I shall not go into the details of the restriction at work, but merely note that if 'imperative' as well as 'modal adjuncts' are looked upon as abstract predicates on the clause (Langendoen, 1969: 122-24), the facts of (3)a. can be accounted for in terms of limitations on the co-occurrence of such abstract predicates. N. 2. S-P-N = intS-P-N or intS-P-N' This criterion too derives from Greenbaum (1969: 24), who phrases it: 'The item must be able to serve as the focus of clause interrogation, as demonstrated by its ability to contrast with another focus in alternative interrogation'. If we return to (1) above, it will appear that the criterion distinguishes between the clusters hatch out to my disappointment and hatch out during the night: (l')a. *Did the babies hatch out during the night to your disappointment, or did they hatch out to your relief? b. Did the babies hatch out during the night, or did they hatch out during the day? The rationale behind this criterion is the same as for the preceding one:

48 a restriction on the co-occurrence of modalities (Halliday) / abstract predicates (Langendoen) on a clause. N.3. S-P-N = negS-P-N but S - { ^ y ' } To satisfy this criterion, a P-N 'must be able to serve as the focus of clause negation, as demonstrated by its ability to be contrasted with another focus in alternative negation' (Greenbaum, loc. cit.). The mechanism behind this criterion is the same as for the two preceding ones; applied to (1) it has the same effect as N.2: (l")a.

*The babies didn't hatch out (during the night) to my disappointment, but they did hatch out to my relief, b. The babies didn't hatch out during the night, but they did hatch out during the day.

(4) is a textual example satisfying N.3: (4)

Suddenly they were all peering in the same direction; not with alarm or revelation, but curiously, noticing an absence for the first time. (JCSG 30)

N.4. S-P-N f P - N # negS To satisfy this criterion P-N must not be able to occur initially in an independent tone unit when the clause is negated (cf. Greenbaum, loc. cit.). Note that unlike the preceding criteria this criterion draws not only on restrictions on the co-occurrence of P-N and (other) modalities but also on the thematic position of P-N. Of the three clusters in (3) above, only go ahead of them satisfies this criterion: (3')a. By the nature of things, the Tills didn't go ahead of them. b. In the procession, the Tills didn't go ahead of them. c. * Ahead of them, the Tills didn't go.

N.5. S-P-N f

a. [P-N]rel- S b. [P-N]wh-S-6e+Tns-P-N c. [P-N]wh-Aux+Tns-S ?

To satisfy this criterion P-N must not be replaceable by a single pro form (a pro-adverbial) in a. relative clauses, b. pseudo-cleft constructions, and c.

49 interrogative clauses.1 The diagnostic value of this criterion lies in the fact that in clusters not satisfying it, P-N behaves as a single, adverbial, constituent in respect of the processes in question, and, accordingly, V-P does not have the unitary character of a prepositional verb. In (5) I illustrate the criterion by three cases not satisfying it: (5)a.l.

Flies, heat-drugged, would crawl slowly on the walls [....] (FOA 58) = 2. the walls where flies would crawl

b.l.

[....] it often takes longer to book your metro seat than to get to Honolulu by plane. (GW 31.1.70, 6: 4) = 2. How you get to Honolulu is by plane c.l.

[....] the curves cut at a certain temperature [....] (8a.3.29-2) = 2. When do the curves cut?

N.6. S-P-N f S-pro[P-N] This criterion states that there must be no (anaphoric) pro-adverbial (apart from those mentioned in N.5) replacing P-N. The reason behind N.6 is the same as for N.5, and I shall merely give one example not satisfying it: (6) In this and the following chapter we comment on two aspects of efficiency [....] (CRNN 11) = b. Here we comment on two aspects of efficiency. N. 7. S-P-N = a. P-Nrel-S b. Nrel-S-P To satisfy this criterion N alone must be replaceable by a relative pronoun. The diagnostic power of this criterion rests on the facts that a. there is a process with respect to which N behaves as a single constituent, independent of P, b. the form replacing N is pro NOMINAL. The criterion will be regarded as satisfied if one of its conditions is satisfied. Of the two V-P-N clusters in (6), only one satisfies both conditions: (6')a.l. The aspects of efficiency on which we comment, in these chapters 2. the aspects of efficiency that we comment on in these chapters

50 b.l. the chapters in which we comment on efficiency 2. *the chapters that we comment in on efficiency Condition b. goes a step further than condition a., in that it indicates not only that N behaves as a nominal constituent in its own right, but also that the constituency break is V - P / N rather than V/P-N. For this reason, N.7.b., along with the other 'preposition stranding' criteria (cf. N.8-10),2 might as well have been included among the V-P criteria (below 3.2.2). I include them here in order to avoid having to deal twice with structurally related criteria. N. 8. S-P-N = what-S-P-6e+Tns-N For a cluster to satisfy this criterion N alone must be able to be the topic of a pseudo-cleft construction with the pronoun what as the identified element (Huddleston, 1971: 241-42). 3 Since there is no possibility of P-what, the criterion also involves preposition stranding. (7) is an example of positive reaction: (7) Professor Sternglass [....] emphasised that he was talking about weapons testing, not nuclear power stations. (GW 31.1.70, 9: 4) = a. What he was talking about was weapons testing, not nuclear power stations. N. 9. S-P-N = a. P-wh-N-Aux+Tns-S? b. wh-N-Aux+Tns S-P? To satisfy this criterion the N of a cluster must be able to occur as the topic of an adjectival w/z-interrogative clause. The criterion is satisfied as long as one of its conditions is satisfied; thus (8) will be an example of positive reaction: (8) He feels that it was unrealistic of the CBSO to apply for a bigger grant at this time [....] (GW 31.1.70, 14:4) = a. at what time did they apply for a bigger grant ? f b. *what time did they apply for a bigger grant at''. N. 10. S-P-N = a. P-wh-Aux+Tns-S b. wh-Aux+Tns-S-P This criterion is satisfied if N can be elicited by an interrogative pronoun. The w/z-question is one of the established tests for objects in the grammatical

51 tradition, and the mechanism behind the diagnostic value of this and the preceding criterion is the same as for N.7. Fairclough (1965: 102) uses both N.9 and N.10 (as well as the adverbial interrogative, cf. N.5.C.), and notes that positive reaction to N.10 'is usually predictive o f positive reaction to N.9, but not vice versa. Thus (9) which satisfies (both conditions of) N.10 also satisfies N.9, but (8) does not satisfy N.10: (9) What we may call the tree-watering approach to education derives initially from Rousseau's view [....] (CAPS 51) = a. From what does the tree-watering approach derive? = b. What does the tree-watering approach derive from? Note that echo questions do not count as satisfactory reactions to N.9-10. Thus although the following exchanges might occur, they will not be evidence that the two criteria are satisfied: (10) A. B. A. (11) A. B. A.

He talked in a subdued voice. In what voice did he talk? In a subdued voice. He talked in a subdued voice. What did he talk in? In a subdued voice.

In fact, as Fairclough (1965: 101) has pointed out, all echo questions are 'to some extent grammatically deviant, relying for their acceptability upon a particular context of mishearing'. N . l l . S-P-N = S-P-proN To satisfy this criterion, N must be replaceable by a pronoun (other than those mentioned by N.7, 8, 10). The rationale behind this criterion is the same as for the preceding ones. (12) I / d i d a p p l y # for /6i:a the Baker Fellowship - at / O x f o r d ^ (s. lb.6-78) = a. I did apply for it. N.

12. S-P-N f S- {J'

&-P-N

Criteria relying upon coordination have enjoyed considerable popularity in

52 studies of prepositional verbs. Since the type of coordination that has been used in the past is coordination between V-P and a transitive verb, the most natural place to discuss the principles behind and the problems connected with the use of coordination as a criterion is in 3.2.2 (cf. criterion V.3). This criterion is satisfied only if both conditions are met. (13) is negative to both: (13) Having made a joke he would smile in his beard with a sort of vulpine pleasure [....] (FOA 57) = a. He would smile in his beard, gently, but with a sort of vulpine pleasure. = b. He would smile in his beard, quiet, but with a sort of vulpine pleasure. N.

13. S-P-N f P-N-S

For this criterion to be satisfied, P-N must be incapable of occurring in clause initial position. One of the features most clearly distinguishing adverbials from other elements is their positional freedom, and this and the following criterion rest on this feature. (14) fails to satisfy N.13: (14) the aerials on the flat roof hung like surrealist trees against the new dark. (JCSG 24) = a. against the new dark the aerials on the flat roof hung like surrealist trees. Carvell & Svartvik (1969: 46) observe that fronting of P-N 'is almost always possible with emphatic intonation and stress'. For this reason only non-emphatic fronting should be included. To that end we may add the restriction to N.13 that the determiner of the fronted N must be the definite or indefinite article only, and that the fronted P-N must not carry emphatic stress or intonation. Thus (15) will not count as evidence that the cluster agree on a point fails to satisfy N.13: (15) On one point only they seemed to agree: the incident had taken place when the music was loudest. (JCSG 46) N. 14. Sb-(v-v-)V-P-N f Sb-(v-)P-N-(v-)v To satisfy this criterion P-N must be unacceptable in mid position (both Ml and M2, cf. Quirk et al., 1972: 426). (16) fails to satisfy the criterion:

53 (16) The River Thames Society [....] has been pressing for the past seven years for action concerning the installation of a Thames barrage or barrier [....] (GW 31.1.70, 2: 5) = a. The River Thames Society for the past seven years has been pressing for action concerning the installation of a Thames barrage or barrier. N. 15. V-P-N = N-be+Tns-Adj- {

^

) V i n g

} -P

To satisfy this criterion a cluster must be able to occur in the constructions exemplified in the b.-cases of (17)-(18): (17) a. b. (18) a. b.

It is difficult to come by a new job. A new job is difficult to come by. It is worth looking at his new book. His new book is worth looking at.

in which a (prepositional or non-prepositional) object of a constituent clause is 'promoted' to subject of the main clause — hence the name 'object raising' (Huddleston, 1971: 161 ff). What actually goes on, and what conditions there are on the transformation, is still only vaguely understood (for some discussion, see Lees, 1960b; Huddleston, loc.cit.; Postal, 1971: 27ff; Bolinger 1973; Lasnik & Fiengo, 1974). For present purposes the most important thing about the structural description of the transformation is that Adj must be evaluative, and that V must be able to take a subject which can be the Experiencer of the evaluation; cf. the dual constituency of it is (difficult |for you) to come by a new job]. For this reason the criterion is inapplicable in quite a few cases (e.g. the book consists of five chapters). Note finally that the formulation of N.15 allows for two constructions which do not exemplify object raising proper, but in which an infinitive/participial clause is (part of) the modifier of Adj: (19) Aubrey is rich and sympathetic enough to persevere with [...,] (GW 31.1.70, 18:2) (20) The room is unsuitable for working m. 4 N. 16. Sb-V-P-N = N-6e+Tns-Ven-P(-6j>-Sb) To satisfy the criterion N must be able to become the subject in a passive version of V-P-N. Their capability of occurring in the passive is the feature of prepositional verbs that has appeared the most remarkable to grammarians.

54 So striking has this feature seemed, that most traditional grammarians (e.g. Jespersen, 1894: 228-31; H. E. Palmer, 1928: 156; van der Gaaf, 1930; Brose, 1939; Körner, 1948; Marchand, 1951; Zandvoort, 1962: 53-55; Scheurweghs, 1961: 18; Schibsbye, 1970: 293-94) 5 and several structural and generative grammarians (e.g. Long, 1961: 117; Chomsky, 1965: 101, 104-5; Thomas, 1965: 125; Christophersen & Sandved, 1969: 224-25) fail to realise that passivisation is merely one of several criteria. F. Palmer (1965: 188-89) and Carvell & Svartvik (1969: 15-17) express a much more critical attitude to the applicability of passivisation as a criterion for prepositional verbs. Both works note that exclusive reliance upon this criterion will yield counter-intuitive results, since there are clear cases of prepositional verb + object which do not passivise (21), and equally clear cases of verb + adverbial prep-phrase which do (22): (21) a.l. 2. b.l. 2.

I didn't take to him *He wasn't taken to. Ice consists of water. *Water is consisted of (by ice).

(22) a.l. 2. b.l. 2.

They sat on the chair. The chair's been sat on. They had slept in the bed. The bed had been slept in.

(the a.-examples are Palmer's, the b.-examples are Carvell & Svartvik's). From this Palmer concludes, 'The test of a passive transformation does not then help to establish what is a prepositional verb as opposed to a verb + preposition' (1965: 188). As I hope to show below (chaps 6 - 7 ) , this statement is too crude. Although there certainly are a number of cases which we should wish to regard as prepositional verbs, if this concept is to be of any significance, and which do not passivise, the restrictions on passivisation are subject to certain regularities.

N.17.

S-P-N ±

b. f 1. what-happen/etc.+Tns-P-N-be+Tns-that-S I 2. proVP-/uzppe«/etc.+Tns-P-N

For a cluster to satisfy this criterion, P-N must be unacceptable outside the scope of a Verb Phrase pro form. The criterion is satisfied only if all four conditions are satisfied. (23) is negative to all of them:

55 (23) In the second interval he went out [....] (FOA 89) = a.l. What he did in the second interval was go out. 2. If he had to go out he might have done so in the interval. b.l. What happened in the interval was that he went out. 2. If he had to go out, why should it happen in the interval? If a cluster is negative only to the a. conditions of N.17, its P-N will be referred to as a 'subject adjunct' (cf. above p. 31 fn. 18, below 5.1.5.1, 6.1.1 ff.): (24) [the tortoise] would retreat into his shell with an indignant wheeze [....] (FOA 50) = a. What he would do with a wheeze was retreat into his shell. f b. *What would happen with a wheeze was that he would retreat into his shell. To my knowledge, the diagnostic effect of conditions a.l., b.l. was first observed by Klima (1962: 106). What happens is that the Verb Phrase including bound prepositional phrases is made the focus of a pseudo-cleft construction, and thus only non-VP elements may be mentioned in the subject clause. The do so test (a. 2) was first discussed by Lakoff and Ross (1966), cf. also S. Anderson (1968). Note that there is at least one case in which do what and do so do not agree about the scope of the Verb Phrase (cf. below p. 150).

3.2.2 Criteria for V-P All criteria listed in this section are designed to illustrate the observation that in a prepositional verb 'The preposition is intimately connected with the verb so as to form a kind of unit with it' (Poutsma, 1928: 257). V. 1. V-P-N + nomV-fee+Tns-P-N This criterion is satisfied if an intensive clause with a nominalisation of V as subject and P-N as complement is ungrammatical. The criterion rests on the hypothesis that the V-P of a prepositional verb must be too tightly fused for V to behave as a separate constituent under nominalisation. (25) is negative to this criterion, (26) is positive: (25) Things are moving in the right direction [....] (GW 31.1.70, 14: 3) = a. The movement is in the right direction.

56 (26) Turner's [voice] barely rose above a murmur. (JCSG 63) = a. *The rise (of his voice) was barely above a murmur. V. 2. Sb-V-P-N + pro [ (Sb-)V ]-fte+Tns-P-N This criterion differs from the preceding one only in that a. it has a proform where V-l has a nominalisation, b. the pro-form may also cover Sb. More clusters are acceptable in the structure defined by V.2 than in that defined by V.l (i.e. V.2 is the less inclusive). There is nothing surprising in this in view of the fact that nominalisation is a partially productive morphological process whereas pronominalisation is a fully productive syntactic process. Thus (26) which was positive to V.l is negative to V.2, and (27) which is positive to V.l is also positive to V.2: (26') If his voice rose it was barely above a murmur. (27) Recent developments [....] have concentrated on the replacement of type metal plates [....] (CRNN 18) f a. *The concentration has been on the replacement of type metal plates. f b. *If recent developments have concentrated it is on the replacement of type metal plates. V. 3. Sb-V-P-N = Sb-V-P-&-Vtr-N Sb-Vtr-&-V-P-N Condition: N has the same function in relation to Vtr as to V-P. To satisfy this criterion a V-P combination must be able to coordinate with a transitive verb so that they share an object N which has the same function in relation to both. Thus e.g. the crew were to land on and explore the moon is not evidence that land on satisfies V.3 since the moon is Loc in relation to land on but Phen in relation to explore. Coordination with Vtr is one of the established tests for prepositional verbs (cf. Brose, 1939: 29, 37, passim; Jespersen, 1927: 277; Kruisinga & Erades, 1947: 28; Korner, 1948: 85-93; Fairclough, 1965: 110-12 [with reservations]; Carvell & Svartvik, 1969: 43). The diagnostic value of coordination rests upon the well-known fact that there must be a certain amount of similarity between two coordinated elements, and in spite of the fact that there are important unresolved theoretical problems connected with coordination, 6 the criterion may be regarded as serviceable in practice. But there are also practical problems which make coordination a somewhat tricky criterion to work with, (a) It relies heavily on the

57 patience, imagination and good fortune of the linguist. A failure on his part to think of a verb that might suitably be coordinated with the V-P under inspection need not imply that such a verb does not exist, but merely that he did not have the good luck to think of it. (b) The acceptability of a coordination depends in large measure on the relative order of the three elements involved (i.e. N, Vtr, V-P). Thus, though the textual example (28)a. is fully acceptable, the related (28)b.-c. are not: (28) a. Rain followed the wind, but it was a warm rain that you could walk in and enjoy [....] (FOA 178) b. ? *You could walk in and enjoy the rain. c. *You could enjoy and walk in the rain. 7 For this reason it is necessary to apply the criterion in a rigid test frame, and I shall- use it only as stated in V.3, i.e. the object is nominal and follows the coordinated elements. Thus (28)a. is not evidence that walk in satisfies V.3. 8 V. 4. P not commutable with other P's V. 5. V not commutable with other V's Condition: the functional pattern of the resultant Sb-V-P-N is identical with the original. Of the criteria used in this study, the two commutability criteria provide the only test for the claim that V-P make up a semantic unit (see van der Gaaf, 1930; Korner, 1948: 49, 162, passim). Mitchell (1958) includes V.5 among his criteria, and Fairclough (1965: 104) makes use of both. Carvell & Svartvik (1969: 45) propose a somewhat different criterion which must also be motivated by the unitary character of the prepositional verb: 'possibility of deletion of p N2, without structural change'. That form of the criterion has the advantage of being extremely easy to apply: deletion is a process that can be performed mechanically. On the other hand it will lead to intuitively unsatisfactory results, since it will distinguish between lie above and refer to on the one hand and look at on the other, where the intuitively convincing analysis would say that look at and refer to have more in common with each other than either of them has with lie above'. (29) a. Emma was referring to the fact that . . . b. *Emma was referring. (Carvell & Svartvik, loc.cit.) (30) a. This entails one of the curves (2 say) lying entirely above the other [....] (8a.3.29-3) b. T h i s entails one of the curves lying.

58 (31) a. Turner did not look at him. (JCSG 81) b. Turner did not look. The condition 'without structural change' is important, though, and the condition imposed on V.4-5 can be regarded as a translation of it into the theoretical framework of the present study. Without it we should be forced to conclude that both the V and the P of e.g. he looked after the children are commutable, since we can say he gazed/ran/arrived after the children and he looked at/'for¡to¡between the children. (32) is a typical example of a cluster failing to satisfy V.4-5:

(32) Two dogs were

fighting eating sleeping barking

in behind nèar outside

the gateway.

V.6. Sb-(Adv-)V-P-N = Det-(Adv-)V+en-P-N Adjectivisation of a V-P cluster has been regarded as criterial by several scholars - e.g. Volbeda (1926: 146), Jespersen (1927: 315), Brose (1939), for whom they are simply passive constructions, Kòrner (1948: 103). Although traditional grammar and transformational grammar share the view that attributive particles are passivai, acceptance of passivisation (N.16) is certainly not predictive of acceptance of adjectivisation. The Adv of the formula is taken to refer to any adverbial or negative particle modifying V. In several cases adjectivisation of past participles is possible only if the verb is somehow modified. This, however, is a restriction on adjectivisation in general, and not a peculiarity of prepositional verbs: *a read book - a frequently read book (cf. Barkai, 1972; Jespersen, 1914: 337). The V-P of (33) is an example of a cluster which, although it passivises, does not adjectivise. (34) is a textual example satisfying V.6: (33) a. The problem has been struggled with by several scholars, b. *A frequently struggled with problem. (34) Roger was so overcome by this unlooked-for attention that he let out a volley of surprised barks. (FOA 47)

59 V.7. V-P-N f V-X-P-N Condition: X may not be an intensifier or a premodifier of P. This is the indivisibility criterion suggested by Bloomfield (1935: 180) in his treatment of 'forms which lie on the border between words and phrases'. (And prepositional verbs are, of course, a subclass of this type of phenomenon, cf. above 1.2.) If desirable, this criterion can be further refined so as to take into account the class-membership of the exponent of X. This has been done by Mitchell (1958), one of whose criteria for prepositional verbs is non-interpolability of (pro)nouns between V and P. In this form the criterion will exclude ask (somebody) for something but not look at something. Another variant of the criterion appears in Carvell & Svartvik (1969: 49) where only cases where X=Adverbial are registered. Now look at will be excluded too, as witness: (35) [....] it will be impossible for his critics to keep looking rather snootily at Mr Aldiss [....] (GW 31.1.70, 19:1) Obviously, the criterion in the form of V. 7 is extremely selective, and relatively few combinations will pass. But there are some: (36) a. He went for me furiously, b. *He went furiously for me. 9 Note that the construction defined in 1.2.1, in which the V itself is complex, are not regarded as examples of V-X-P. Thus (37) is not regarded as evidence that look for fails to satisfy V. 7: (37) We've the BSI standards to look out for. (GW 31.1.70, 16: 1)

3.3.

FRAMEWORK FOR THE DESCRIPTION OF CORPUS DATA

In the following five chapters V-P-N clusters will be dealt with according to the function of (P-)N as established in 2.2: non-role playing, abstract circumstantial, concrete circumstantial, marginal participant, central participant. Each chapter will be divided into two main sections; the first of these investigates the relation between the function of (P-)N and the syntactic features of the cluster (on the hypothesis that there is such a relation) by means of the criteria set up in 3.2.1-2. The second main section investigates the overt features of the clusters attested in the corpus, again on the hypothesis that the incidence of various features correlates with the function of (P-)N.

60 An extremely large number of specific hypotheses can be derived from this general hypothesis. For instance, although both abstract circumstantials and concrete circumstantials permit intervening material between V and P (cf. V.7 above), we will expect the incidence of the structure V-X-P-N to be higher among abstract than among concrete circumstantials (cf. tables 5.2 and 6.2, column g). Similarly, although both concrete circumstantials and marginal participants allow both initial and final prepositions in relative clauses (cf. N.7 above), we will expect the incidence of relative clauses with initial P to be relatively greater among concrete circumstantials than among marginal participants (cf. tables 6.2 and 7.2, columns j-j'). Below I list the positional and exponential features used in the analysis of the corpus. Comments on the relevance of any given feature are added only where the feature in question has not been discussed in 3.2.1-2 above. Of the features not utilised in the criteria some are left out because they are too difficult to control, some because they would have had the same effect as a criterion already used. The overt features fall into three groups: i. features indicative of adverbial status of P-N, ii. neutral, iii. features indicative of nominal status of N/unitary status of V-P. i. adverbial a. P-N is coordinated with an adjective or adverbial (cf. N.12). b. P-N is outside the scope of VP or (Sb-)V (i) pro forms (cf. N.17, V.l-2), (ii) ellipsis: The newspaper industry is inclined to blame the competition of television for people's time, the proportion of homes with television having been rising, though [it has risen] at a diminishing rate, over the whole decade. (CRNN 6) c. P-N occupies mid position (cf. N.14). d. P-N occupies initial position (cf. N.13) e. P-N is realised by an adverbial pro form (cf. N.5-6) f. N is realised by an adverbial pro form: /in the event f f that the /Ford does not come t h r o u g h ^ . and /in the e v e n t ^ that the /Nuffield says noff — /you k n o w ^ /where does one go from thereof (s.lb.6-62) g. V and P are separated by an adverbial (including other prepositional phrases), an intonation break, or a comma (cf. V.7). ii. neutral h. The cluster has the structure V-P-N with N being realised by a nominal.

61 iii. i. j. j'.

nominal N is a personal or indefinite pronoun (cf. N.l 1). N is a relative pronoun with preposed P (cf. N.7). N is a relative pronoun with postposed P (cf. N.7), note that j'. includes cases of zero relative and reduced relative clause: Los /Angeles^ is /no place to go t o ^ (s.lb.6a-14) k. N is the topic of an interrogative (or exclamative) clause with adjectival interrogative (exclamative) word and preposed P (cf. N.9). k'. N is the topic of an interrogative (or exclamative) clause with adjectival interrogative (exclamative) word and postposed P (cf. N.9): /what real obscenitiesff /Denzil Bachelor thinks o f f f it's /quite remarkable^ (5b.16.15) 1. N is the topic of a pseudo-cleft construction (cf. N.8). Note that the textually most frequent form is the one in which the identifier is represented by a demonstrative pronoun and the identified-identifier sequence is reversed: '[....] Meadowes is taking it [sc the minute] to the Ambassador.' 'I imagined that was what Siebkron was ringing about.'' (JCSG 23) m. N is the topic of an interrogative clause with nominal interrogative word and preposed P (cf. N.10). m'. N is the topic of an interrogative clause with nominal interrogative word and postposed P (cf. N.10). n. N is a reflexive pronoun. Reflexive pronouns at N are treated as distinct from other pronominal realisations of N (cf. i. above), because there is evidence that the function of (P-)N decides whether N pronominalises or reflexivises under coreference with Sb (cf. Jespersen, 1949: 165, and cf. John looked behind him - John looked at himself). And Quirk et al. (1972: 212) indicate that reflexivisation is a characteristic of prepositional verbs. The following textual examples support these suggestions: a. [....] a cicada whispered gently to itself. (FOA 65) b. Jenny Pargiter [....] looked stiffly ahead of her [....] (JCSG 37) o. N has undergone object raising (cf. N.l 5). p. N is the subject of the passive version of V-P (cf. N.l 6). q. P-N is the topic of a cleft sentence. On the face of it this would seem to be an adverbial feature, since it is a clear case of a transformation treating P-N as one constituent. Note however that there are several clearly 'adverbial' functions (the majority of the abstract circumstantials) which are unacceptable in the construction (except under contrastive stress): (*it was for a moment that he gazed at me, *it was at intervals that he paused, *it was in ecstasy that young men shouted, etc.) q'. N is the topic of a cleft sentence. This construction is clearly a 'nominal

62 feature' since it indicates that the constituent break in the cluster is V-P/N:

r.

s. t. u.

it's /just his methodsff that's / all that they do object t o f f (s.lb.6b-7) N occupies initial position: [....] the massive, depressed dome o f the f o o t tunnel. A Wren parody it looks like [....] (GW 3 1 . 1 . 7 0 , 1 5 : 1 ) V-P is coordinated with a transitive verb (cf. V.3). V-P is adjectivised (cf. V.6). V-P is nominalised. The inclusion o f this among the nominal features is a consequence o f V . l , which stipulates that in a prepositional verb V may not nominalise alone. 'Mickie, the Ghanian Chargé will need looking after. Keep h i m away from the Ambassadress.' (JCSG 2 9 ) 1 0

NOTES 1. The three structures, relative - pseudo-cleft - interrogative, are listed in that order because of an observation of Huddleston's (1971: 243) to the effect that there is a gradient between them (cf. Quirk, 1965). Below 5.1.1.ff. I shall present some evidence in support of this observation. 2. To my knowledge Volbeda (1926) is the first to point to the diagnostic power of preposition stranding and to its dependence on the function of the prepositional phrase (cf. also Jespersen, 1927: 185-95; Erades, 1957), and in historical studies on the process leading to the possibility of passivising V-P clusters (see van der Gaaf, 1930; Brose, 1939; Korner, 1948) preposition stranding is generally taken to be a contributing factor. Kruisinga (1927) alone seems to reject this criterion. On the basis of examples like (i) it was the sort of night you stopped out all night on. (ii) Also that the applause that came therefrom, when they did stop, had a certain perfunctory air, as of plaudits something else makes room for, and comes back again after. he concludes that it does not matter whether the function of the prep phrase 'is one of time or manner or circumstance, or what not' (p. 8), and he explicitly refuses to accept the relevance of the objection that these examples may be odd or marginal. All that matters is the fact that they have been attested. Kruisinga's position is a good example of the situation in which the linguist's uncritical acceptance of all his material prevents him from basing any conclusions on it. 3. In transformational grammar the pseudo-cleft construction with what as the identified element is frequently used as a criterion for objects (see e.g. Rosenbaum, 1967). 4. The diagnostic value of object raising is noted by Korner (1948): 66, 225). 5. On the other hand, the fact that prepositional verbs mark one end of a cline and the consequent need for a multiple criteria treatment is clearly recognised in Poutsma (1926: 33-5), Volbeda (1926), Jespersen (1927: 276). 6. The exact nature of this similarity need not concern us here. Chomsky (1957: 36)

63 aigues that the similarity is of a structural nature, whereas Fillmore (1968) and Dik (1968) have produced convincing evidence that it is a question of functional similarity. 7. In fact, the pattern of (28)a., in which the shared object of the Vtr and V-P is a relative pronoun, seems to form the most favoured context for coordination. The unacceptability of (28)c. is of course contributed to by the fact that and suggests not only coordination but also temporal sequence. 8. There is one frequently mentioned criterion which I have decided not to use: replaceability of V-P by a simplex Vtr. Poutsma (1928: 257) notes that in a prepositional verb the V and P frequently make up a unit 'which in the same or some cognate language may be expressed by a verb without a preposition'. And the criterion recurs in more or less modified form in van der Gaaf (1930), Körner (1948: 49-50), Mitchell (1958), Fairclough (1965: 110). Fairclough, however, observes that the criterion is 'difficult to control' (p. I l l ) : there are'cases where a simplex Vtrans may substitute for a V-P cluster, but where other criteria show that the cluster cannot be considered a prepositional verb, e.g. they went round/circumnavigated the world, he went to/visited Santiago. We may go a step further and claim that the objections against using the synonymity criterion are not merely practical. In the absence of a theory about what kind of semantic complexes may constitute a word, we shall have to look upon the existence or non-existence of a simplex word synonymous with a phrase as a mere coincidence. There is no way in which we can decide whether the lack of a one-word substitute for a V-P cluster is a systematic or accidental lexical gap. What is more, the synonymity test has all the weaknesses of the coordination test plus the additional disadvantage that synonymity may very well be a relation between two items whereas possibility for coordination is a relation between classes of items. 9. He went straight for me is not counter-evidence, since it is one of the cases excluded by the condition on the criterion. 10. Nominalisation of V-P is regarded as a criterion for prepositional verbs by Poutsma (1926: 34), Körner (1948: 225).

4. NON-ROLE PLAYING PREPOSITIONAL PHRASES

4.1

SYNTACTIC FEATURES

As is shown by the b-cases of (l)-(5), all the examples cited fall under the heading 'non-role playing prep phrase' according to F.l: (1) a. I said how do you do, politely, but to my surprise the bearded man rose to his feet [....] (FOA 74) b. The bearded man rose to his feet. This was surprising. (2) a. [....] it is [....] possible to realise regions such as EF, HI, LM and PO, which correspond to local [ital] but not absolute, minima [....] On the other hand regions such as FGH are found to correspond to maxima [ital] of g [....] (8a.3.28-2) b. On the other hand it is true that regions such as FGH are found to correspond to maxima of g. (3) a. [the assembly] could, in principle, remain in the corresponding thermodynamic state indefinitely, in spite of the fact that a state such as X is of a lower free energy. (8a.3.26-4) b. In principle it is true that the assembly could remain in the corresponding thermodynamic state indefinitely. (4) a. A distinction made by Quine [....] seems to correspond to some extent with the List II distinction [....] (8b. 1.154-1) b. To some extent it is true that Quine's distinction corresponds with the List II distinction. (5) a. /what you've got to do in each case is to. /look .at the /case in front of y o u ^ (5b.16.72) b. in each case it's true that you've got to look at the case in front of you. Each of the above examples represents a major subclass of non-role playing constituent: (1) disjunct, (2) conjunct, (3) viewpoint/aspectual adjunct, (4) intensifler, (5) circumstantial adjunct. Since the criteria of the preceding chapter were devised primarily to distinguish between the more cohesive

65 V-P-N clusters, I shall not attempt to apply all of them to the five subclasses of constituents treated in this chapter - in fact all of them react negatively to most of the criteria. However, in order to show that (a) even within this class there are degrees of cohesion, and (b) that one of the subclasses, circumstantial adjunct, bears some resemblance to free adjuncts (see chap. 5), nonrole playing constituents have been tested with respect to N.l (imperative), N.2 (alternative interrogation), N.3 (alternative negation), N.4 (P-N outside focus of negation), N.7 (relativisation), and N . l 4 (M position). The result is shown in table 4.1; the order of the criteria has been changed so as to give a more compact picture. Note that, for various reasons, some of the criteria do not apply to all of the examples (l)-(5). 1 Disjunct Conjunct Viewpoint Intensifier Circumstance

2

3

7

4

14

+ +/-

+

+

+/-

-

+/-

+/-

+/-

+/-

+/-

+

+

+

+

-

-

+/-

Table 4.1: features of non-role playing prep phrases

4.1.1.

Disjuncts.

Failure to satisfy criteria N.2, 3, 4 constitutes Greenbaum's (1969 : 24) definition of disjuncts. Their behaviour with respect to these criteria is illustrated above 3.2.1 exx (l')a-> 0") a -> (3')a., and no further exemplification will be needed here. As for their occurrence with imperatives, Greenbaum (1969: 85, 112) notes that style disjuncts may, but attitudinal disjuncts may not occur with imperatives (6), (7). As Schreiber (1972: 340) notes, an imperative occurring with a style disjunct always has the illocutionary force of a hortative, never a command. Therefore (6)b. is ungrammatical: (6) a. b. (7) a. b.

In confidence, accept the offer, *In confidence, kill him! *To my surprise, let's not talk about it. *To my surprise, rise!

All disjuncts fail to satisfy N.7:

66 ( l ' ) *My surprise, to which he rose to his feet, was great. Failure to satisfy N.14 (acceptability in M position) is illustrated by the textual example: (8)

4.1.2

[they] sat in their customary pew beside the Tills, who by the nature of things had gone ahead of them in the procession. (JCSG 37)

Conjuncts.

Resemble disjuncts in their negative reaction to N.2, 3, 4, but differ from them in that they may occur with imperatives, even commands (Greenbaum, 1969: 39-40). (9) I'm afraid that you've outstayed your welcome. In other words, get out! The N of a prepositional phrase functioning as a conjunct never relativises (N.7): (2') T h e hand on which regions such as FGH are found to correspond to maxima . . . . Some conjuncts occur readily in M position (fail to satisfy N.14), but there are individual restrictions; in (10), for instance, the conjunct as an example could not be moved to M position, but its synonym for example could: (10) [....] we cannot think, as we can in fig. 3(a), of two physically distinguishable phases in equilibrium at the transition temperature. As an example, behaviour of the type shown in fig. 3(b) occurs with the Weiss model [....] (8a.3.29-l)

4.1.3 Viewpoint/aspectual

adjuncts.

The terms are taken over from Jacobson (1964: 23). Greenbaum (1969: 163-67) employs the term 'aspectual adjunct', and Quirk et al. (1972: 42930) refer to the same type as 'viewpoint adjunct'. There seem to be collocational restrictions between individual items and classes of verbs, hence

67 +/— for N.l. For instance, none of the verbs of the 13 examples cited by Quirk et al. (loc. cit.) can occur in commands, but in some cases an imperative-as-command is possible: (11) Your translations have not been up to standard this term. You work harder in this respect! The reaction to N.2, 3, 7, 4 (in that order) is shown in (12), the failure to satisfy N.14 (acceptability in M position) appears from (3) above. (12)a. b. c. d.

Does this ever happen in practice, or only in theory? This can never happen in practice, but it can in theory. *The practice in which this could happen . . . In practice, this could never happen.

Thus the cluster happen in practice does not accept relativisation (' — ' for N.7), but there are some which do, notably if the P-N is [ this in) some respect(s), or if the P is the complex in respect of:1 (13) the only cases in respect of which our investigations have led to reliable conclusions are the following: Certain verbs of comparison2 have as a near-obligatory complement a prep phrase denoting a specification of the aspects in respect of which the comparison is being made. For instance: (14) [....] there is no evident justification for newspapers continuing to increase in size indefinitely [....] (CRNN 24) 3 Danes (1968) discusses a similar constituent in examples like John exceeds Peter in weight, and considers the prep phrase a role playing constituent referred to as 'respect'. A functionally similar constituent occurring with symmetric predicates (e.g. A resembles B in Y, A differs from B in Y, A is similar to B in Y) is dealt with by Lakoff & Peters (1969). These constituents bear a certain resemblance to viewpoint adjuncts, for instance they may take the complex preposition typical of those: A exceeds B in respect of Y, and they have the same lexical pro form, in this respect: A is tall. In this respect he differs from B. I shall follow a suggestion of Huddleston's (1971: 291), however, according to which the prep phrases in question are not considered constituents in

68 the clause at all, but heads of the NP(s) being compared. Thus: A's Y differs from B's. Similarly, increase of (14) will not be analysed as a two-place predicate with the arguments newspapers and (in) size, but rather as a one-place predicate with the argument size of newspapers. Examples analysed in this way have been regarded as V-P-N clusters only at a superficial level, and they have not been included in the corpus of the present study; They include contract in volume, differ in volume, move on in sophistication, relate in time and place, vary in extent.

4.1.4 Intensifiers. I use this concept in the same way as Greenbaum (1970) and Quirk et al. (1972). As appears from Table 4.1 intensifiers do not behave as a group with respect to any of the criteria to which they have been exposed. This is a very unsatisfactory situation, of course. In principle, one can imagine several explanations for this: (a) there is no linguistic reality behind the concept, (b) although the group does exist, several of its members behave in highly idiosyncratic ways, (c) the criteria applied are unsuitable for the purpose. There is certainly some truth in both (b) and (c). The idiosyncratic nature of several intensifiers is an established fact (Greenbaum, loc.cit.; Quirk et al., loc. cit.), and since none of the criteria used in this study have been set up with a view to analysing intensifiers, it is hardly surprising that they fail to define them. Of the intensifiers in the corpus there is even one which is not a nonrole playing constituent by F.l: (15) [....] all diplomatic heads would cooperate to the utmost with the Federal Authorities. (JCSG 34) The cluster of (15) is positive to all the criteria except N.7: (15) a. Cooperate with them to the utmost! b. Did they cooperate with them to the utmost or only to a certain extent? c. They didn't cooperate with them to the utmost, but only to a certain extent. d. T h e utmost to which they cooperated . . . . e. *To the utmost, they didn't cooperate with the Authorities. f. (?) All diplomatic heads would, to the utmost, cooperate with the Federal Authorities.

69 In fact relativisation seems possible only when the exponent of N is one of the lexical pro forms extent, degree: (16) These differences between the extent to which advertising revenue has increased [....] are clearly related to the fact that [....] (CRNN7) Since intensifies are well outside the central topic of this study, I shall not go on to exemplify both positive and negative reaction to all the criteria. In most cases this would involve exemplification by individual clusters for each criterion.

4.1.5 Circumstantial adjuncts. As a group circumstantial adjuncts are characterised by the fact (a) that their lexical pro form is in Det case, and (b) that the exponent of P may be the complex in the case of. (5')a.-d., (17)a.-b., and (18) show their behaviour with respect to the six criteria: a. In each case, look at the case in front of you. b. Do I have to look at the case in front of me in each case, or only in some cases? c. You don't have to look at the case in front of you in each case, but you do in some cases. (17) a. [....] the extreme case in which the interactions do not fall off with distance at all [....] (8a.3.244) b. In the extreme case, the interactions do not fall off with distance. (18) a. a B-expression contains at least one finite form of a verb in the indicative mood which does not, within the limits of the B-expression, form part of a complete sentence or clause; (8b. 1.147-2) (5')

Of the non-role playing constituents circumstantial adjuncts are obviously the ones bearing the closest resemblance to free adjuncts; for instance they have pro-adverbial question forms (N.5), normally when: (19) When do the interactions not fall off with distance? In the extreme case.

70 4 . 2 . THE CORPUS.

In the corpus clusters involving non-role playing prepositional phrases occur in the following structures (cf. above 3.3): b.

c.

d.

g.

h.

j.

P-N is outside the scope of Verb Phrase or (Sb-)V pro forms or ellipsis. [....] we cannot think, as we can in fig. 3(a), of two physically distinguishable phases in equilibrium at the transition temperature. (8a.3.28-3) P-N occupies mid position. What he did not admit was that a child's education should in any way depend on the wealth of its parents. (CAPS 61) P-N occupies initial position. In summary, in the last three years the industry has been making considerable efforts to bring its costs under control [....] (CRNN 12) V and P are separated by an adverbial, an intonation break or a comma. The trouble begins with taxation, according to the Frankfurter [ital]; (JCSG 26) The neutral type. A distinction made by Quine which seems to correspond to some extent with the List II distinction of these writers is the distinction [....] (8b.1.154-1) N is a relative pronoun with preposed P. [....] we must have some such situation as that shown in fig. 3(b), in which a single minimum in the free energy splits into two [....] (8a.3.28-3)

The incidence of the six features is tabulated in table 4.2: b

c

d

g

1

7 10 6 1 8

7 ,2 2 4 6

6 6 9

1 3

17 13 17 15 32

32

21

21

4

95

Disjunct Conjunct Viewpoint Intensifier Circumstance

2 5

2 1 3 2 1

Total

8

9

8.4

9.5

%

h

j

33.7 22.1 22.1 4.2

Total

100.0

71 Note in particular that the 'neutral' feature (h) is absent with disjuncts and conjuncts (cf. below 7.2), and that only one nominal feature (j.) is represented at all. The percentages for the three groups of features, adverbial, neutral, and nominal are as follows: i. ii. iii.

adverbial neutral nominal

73.7 22.1 4.2

Cf. the corresponding figures below 5.2, 6.2, 7.2, 8.2.

NOTES 1. The N respect and the P in respect of may be considered LEXICAL PRO FORMS of viewpoint adjuncts; this term should not be confused with Strang's 'generic substitute' (1962: 99), which covers e.g. one for countable nouns and do for verbs, i.e. grammatical rather than lexical pro forms. 2. That is, verbs involving comparison between more than one item as well as verbs comparing temporally distinct stages of the same items as in (14). 3. I have corrected an obvious misprint.

5. CIRCUMSTANTIALS I: THE ABSTRACT GROUP

5.1

SYNTACTIC FEATURES

The members of this group were exemplified above 2.2.2.1, and in this chapter I shall present a brief discussion of their syntactic features and of some problems connected with the the classification proposed. Although the (P) N's realising the functions to be dealt with here share some features with the N of a prepositional verb (notably relativisation (N.7), see table 5.1) they are obviously only distantly related to them, and, accordingly, a detailed analysis is well outside the scope of the present study. An example of each of the seven functions falling under the heading of chap. 5 is given in (l)a.-(7)a.; the b.-cases show that they are role-playing constituents by F.l. (which they satisfy), the c.-cases show that they are circumstantial rather than true participating elements (in the sense of Halliday, 1967-68) by F.2. (which they fail to satisfy). Tim

(1) a. [George] ransacked his own library and appeared on the appointed day [....] (FOA 57). b. *On the appointed day it is true that George appeared. 3 c. Something happened on the appointed day.

Dur

(2) a. [he] gazed at me reflectively for a moment [....] (FOA 48) b. *For a moment it is true that he gazed at me reflectively. 3 c. He did something for a moment.

Frq

(3) a. At intervals [....] he would pause suddenly [....] (FOA 221) b. *At intervals it is true that he would pause suddenly. D c. Something happened at intervals.

Res

(4) a. [....] the parity of esteem programme, because of its fundamental lack of grounding in the facts of human nature, leads rapidly to much confusion of language. (CAPS 43) b. Because of the programme's lack of grounding it is true that it leads to confusion of language.1 D c. Something happens because of the programme's lack of grounding.

73 Ep

Man

(5) a. b. D c. (6) a. b.

Mes

D c. (7) a. b. D c.

Young men shouted in ecstasy. (JCSG 91) *In ecstasy it is true that young men shouted. Young men did something in ecstasy. [....] an advantage accrues to the paper which can operate at a high level of output [....] (CRNN 8) *At a high level of output it is true that the paper can operate. The paper does something at a high level of output. [....] productivity per man hour rose at an average annual rate of5Vi per cent. (GW 31.1.70, 6: 1) *At an average annual rate of 514 per cent it is true that productivity per man hour rose. Something happened at an average annual rate of SVi per cent.

All abstract circumstantials may occur in imperative sentences 2 ' 3 (N.l), they may be the focus of alternative interrogation (N.2), and of alternative negation (N.3). Since these criteria have no differentiating effect within the group, they are not included in table 5.1. Similarly, the V. criteria are obviously not relevant either (there is no question of regarding V and P as one constituent), and they too are left out. The criteria with respect to which the abstract circumstantials are tested are the following: occurrence outside the scope of negation (N.4), relative, etc. pro-adverbial (N.5), anaphoric pro-adverbial (N.6), relative pronoun (N.7), pseudo-cleft with nominal pro form (N.8), adjectival interrogative (N.9), nominal interrogative (N.10), anaphoric pronoun (N.ll), coordination with adverb (N.12), I position of P-N (N-13), M position of P-N (N.14). The reactions to the criteria are set out in table 5.1. The order of the criteria is rearranged so as to provide a more compact picture. Note (a) that the variation in positive and negative reactions between the various functions is quite small and provides no basis for pronouncing on

Tim Dur Frq Res Ep Man Mes

7

5

9

4

+

+/-

+/-

-

+/+/-

+/+/-

+/-

+/+/-

+/-

-

+/-

+ +

+ + +

+/-

+/(-) +/-

+

+/-

+/-

+

+/-

+/-

+

6

11

+/-

+/-

Table 5.1

13

14

8

10

12

74 the existence or non-existence of a gradience; (b) the high proportion of cases with both positive and negative reactions (+/—) might at first sight seem somewhat disturbing, but as I shall point out in the treatment of the individual functions there is a regularity behind each of the seemingly irregular +/— entries.

5.1.1 Time In this and the following sections I shall discuss reactions of individual functions to individual features. In order to avoid repetitions of rather trivial details, I shall not exemplify each individual reaction, but focus on the more interesting cases. RELATIVE, PSEUDO-CLEFT AND INTERROGATIVE PRO FORMS (N.5). In relative clauses adverbial pro forms occur (i.e. negative reaction) subject to two conditions: (a) the exponent of N should be a temporal noun, year, month, night, moment, etc., or the lexical pro form time; (b) the exponent of P should involve 'absolute point of reference' (Lyons, 1969 : 299), i. e. it must be one of the prepositions at, by, in, on. (8)a. meets both conditions and (8)b. is a paraphrase of it. (9)a. only meets condition a. and is not related to (9)b.; (10)a. only meets condition b. and is not related to (10)b.; finally (ll)a. fails to meet both conditions and is not related to (ll)b.:

(8) = (9) f (10) f (11) f

a. b. a. b. a. b. a. b.

The day on which he returned . . . The day that/when/0 he returned . . . The day after which he returned . . . The day that/when/0 he returned . . . The war in which he was killed . . . T h e was that/when/0 he was killed . . . The revolution before which he left for Europe . . . *The revolution that/when/0 he left for Europe . . .

In pseudo-cleft and interrogative constructions on the other hand there are no exponential restrictions on the use of pro-adverbials (i.e. the reaction is always negative): (12) 'And during World War II you went to Greece? ' 'No, when I went to Greece was between the two wars.'

75 (13) lWhen did you go to Greece? ' - 'Between the two wars.' 4 ADJECTIVAL INTERROGATIVES (N.9). Normally the reaction to N.9 is positive,

but in the following cases adjectival interrogatives are not possible: (a) with calendric time expressions,5 in 1948, in May, on Sunday, at 5.45, etc., but in all these cases adjectival interrogatives are possible if the name of the calendar unit is substituted: in what year?, etc.; (b) church festivals and other institutionalised time expressions (note that N is not preceded by a determiner): at/before/after Easter/matins/lunch/twilight, again, adjectival interrogatives are possible if N is replaced by a hyponym, e.g. event,6 (c) cases in which N represents a sentence to be supplied from the linguistic or cultural context: (14) '[Shares will] go up again after Brussels, don't you worry.' 'Who are you kidding? The talks are as good as dead aren't they? [....r (JCSG 15) Here it is clear that the meaning of N is e.g. the end of the Brussels talks. As in the two preceding cases adjectival interrogatives are possible if e.g. event is substituted for N. ANAPHORA (N.6, 11). As (18) shows the anaphoric pro-adverbial, then, does not share the restrictions of the relative pro-adverbial. In (18) then incorporates a P of relative point of reference and a non-temporal N, but it may also stand for a preposition of absolute point of reference and a temporal noun: (15) I'm seeing him on Monday. Then we can discuss the details. With the directional prepositions before, after and with by there is also a hybrid pro form where N is one of the pro adverbials now, then, but where P is expressed: (16) a. they're expecting me at 3.30 but I'll be there before then. b. I'll wait in for you until 3.30, but don't come after then. c. they should be there by now.1 Kuroda (1969: 278-79) discusses examples like (17) T h a t was the day John told Mary he would disappear and he actually disappeared on it. [my italics]

76 and concludes that 'noun phrases in . . . a time adverbial prepositional phrase may . . . not [be] replaced by personal pronouns'. Counterexamples are easy to come by, however: (18) The first to be arrested was General Sudirgo [....] After him came General Mursjid [....] then came General Suadi [....] (GW 31.1.70,24:4-5) The correct generalisation seems to be that N in a temporal prepositional phrase pronominalises precisely in those cases in which adverbial relatives are not permitted; i.e. if either N is not a temporal noun or P is a preposition of relative point of reference. Thus (19)a. is grammatical, but b. is not: (19) a. It was a very cold winter, and several babies died during it. b. *It was a very cold winter, and several babies died in it.

5.1.2 Duration 5.1.2.1. Introductory remarks. There are two types of duration (cf. Leech, 1969: 129-34), (a) a locational type denoting extension over a certain period. This type has an 'instrumental' subtype denoting time spent in doing something: (20) a. [The programme] has been running for years. (JCSG 74) b. You can get there in two hours. (b) a directional type denoting duration from some specified point in time onwards, or from the present moment until some point in the future: (21) The rock was first charted 400 years ago but it has only been landed on five times since then. (GW 31.1.70, 10: 1) (22) You'll probably be all right by next week. Until then I'll call regularly. With both types S may denote either states [as in (20)a.] or a succession of events [as in (21)-(22)]. Only with the instrumental type [(20)b.] may S denote a single event. 5.1.2.2 Syntactic features RELATIVE, PSEUDO-CLEFT AND INTERROGATIVE PRO FORMS (N.5). The

77 adverbial pro form for duration is (for) how long. This form, however, never occurs as a relative, the adverbial relatives that sometimes do occur are that and 0 (pace Huddleston, 1971: 229, who claims that the way that/ *how is the only case in which a w/i-pro-adverbial is ungrammatical and that is not):

)

*when that the programme has been running seem to be

totally wasted. The restrictions on adverbial relativisation of durational phrases very much resemble those for time: the preposition must be the primary durational preposition for, and N must denote length of time. Since neither that nor how long can be the pro form for the focal element of a pseudo-cleft construction, N.5.b. is always satisfied by durational phrases. It thus appears that for those durational phrases that do no accept that as a relative proadverbial, adverbial relative and adverbial pseudo-cleft share a feature which is absent from adverbial interrogative, which is always possible: (21 ")a. *I remember the day how long/thathe's known this. b. *How long he's known this is since I've told him. c. How long has he known this - since the day I told him. The pure adverbial interrogative without for is possible with both types if S denotes a state. Type (a): How long has the programme been running? (for) two years. Type (b): How long will you be staying? - Until next week. If S denotes a succession of events for seems to be obligatory: *(For) how long did he turn up once a day? — For two months. *(For) how long will you call regularly? - Until you're all right. Unlike directional time phrases, durational phrases type (b) accept the hybrid P-when interrogatives also outside echo questions (cf. fn 4): (24) Since when have you known that? (25) Until when will you be staying? ADJECTIVAL INTERROGATIVES (N.9) are formed less readily than with time phrases. Thus there is no adjectival interrogative corresponding to the central form of type (a), in which P is for and N is a temporal noun. For instance there is no interrogative clause with for what moment eliciting the response We stood in silence for a moment. With directional durational phrases, which always involve a point in time, adjectival interrogatives are possible with the

78 same restrictions as in the case of time phrases. ANAPHORA (6, 11). Durational adjuncts have no pro form comparable to the temporal then (positive reaction to N.6). Note in general that there is no simple relationship between the two criteria concerned with adverbial (N.6) and nominal (N.ll) anaphoric pro forms, if there were, any function would have the same values for both criteria. The restrictions on pronominalisation of N in durational phrases seem to be the same as for Time. That is, in type (a) pronominalisation is not allowed if P is for, in and N is a temporal noun: *he stayed for it [sc. a week], *he finished the job in it [sc. an hour]. But with within, a preposition of relative point of reference, pronominalisation is possible:

(26) I didn't see him until one week later. Within it he had recovered completely. Pronominalisation does not seem to occur with the directional type. Instead, the anaphoric pro form that does occur is the 'hybrid' P-proAdv illustrated above (21)-(22). This is a characteristic of directional phrases in general whether temporal, durational, or spatial (cf. below 7.1.3). Duration is irregular, however, in the existence of a hybrid pro form even in the locational type, though only with the P for, and the pro-adverbial now, e.g. this will do for now, (cf. *he said that this would do for then).

5.1.3 Frequency The prepositional phrases realising frequency are virtually a closed set: with Adj frequency, at Adj intervals, on Adj occasions, from time to time, at a/the rate of N, between/at each N. (27) [....] the love-songs would become more doleful, while between each Larry would pause to inform whichever member of the family happened to be present that spring, for him, did not mean the beginning of a new year, but the death of the old one. (FOA 83-4) RELATIVE, PSEUDO-CLEFT, AND INTERROGATIVE PRO FORMS (N.5) R e l a t i v e

pro-adverbials (negative reaction) are possible only for the P-N on (Adj) occasions:

79 (28) The rare occasions that/when/on which he spoke to me . . . Adverbial pro forms in pseudo-cleft constructions are never possible (positive reaction): (29) a. *When he spoke to me was on rare occasions. b. *How often he spoke to me was on rare occasions. Whereas all Frq phrases may serve as a response to a how often question: (30) How often did he speak' to you? On rare occasions/at irregular intervals. It thus appears that, apart from the cases where N is occasions, relative and pseudo-cleft share a feature that is absent from interrogative: (31) a. *I'll never forget the intervals b.

when he spoke to me. how often

When How often he spoke to me was at rare intervals.

c. How often did he speak to you? — At rare intervals. P-N OUTSIDE FOCUS OF NEGATION (N.4) may deserve some comment. It is hardly surprising that one can talk about the time at which something didn't happen or the period over which a phenomenon has been absent. The frequency with which something does not happen is much harder to think of, and yet we may attest: on rare occasions, he didn't speak to me. Note, however, that e.g. the P-N of the cluster of (32) could not possibly occur outside the scope of negation (i.e. the cluster is positive to the criterion). ADJECTIVAL INTERROGATIVES (N.9) are possible only if N is frequency or rate, which may be considered lexical pro forms of the function:

(32)

At what rate With what frequency At a rate With a frequency

do nerve impulses occur? of 100 in a second.

80 5.1.4 Reason 5.1.4.1 Introductory remarks. Under this function I subsume four subfunctions, a. positive realised reason, b. positive hypothetical reason, c. negative realised reason, d. negative hypothetical reason: (33) a. b. c. d.

He stayed at home because of the rain. He will stay at home in case of rain. He went out in spite of the rain. He will go out regardless of the weather.

b. is traditionally referred to as 'conditional'; Noreen (1904: 227) has pointed out that in fact conditional is no more than hypothetical reason (note the absence of a determiner with N); c. is traditionally named adversative, it denotes a 'cause' which did not have its expected effect; d., concessive, the corresponding hypothetical relation. The relationship between reason and adversative may be formalised in the following way: i. 'v (S because of X) -*• 'v S in spite of X (e.g. not-(he stayed at home because of the rain) -*• he not-stayed at home in spite of the rain). ii. 'v. (S in spite of X) -* 'v S because of X (e.g. not-(he went out in spite of the rain) -* he not-went out because of the rain). Bartsch (1972: 113-15) oddly enough discusses this relationship only in connection with hypothetical reason, i.e. condition-concession. Thus: iii. 'V (S in case of X) * ^ S regardless of X (e.g. not-(I will stay at home in case of rain) -*• I will not-stay at home regardless of rain). iv. 'v (S regardless of X) -*• ^ S in case of X (e.g. not-(I will go out regardless of rain) -> I will not-go out in case of rain). Note that Reason and External Causer are merely names for distinct realisations of the same hyperfunction: (34) a. The sun dried the corn. b. The corn dried in the sun. c. The corn dried because of the sun. (For further comments see below 8.1.1.1).

81 5.1.4.2

Syntactic features

RELATIVE, PSEUDO-CLEFT, AND INTERROGATIVE PRO FORMS (N.5). Pro-

adverbials are possible only with positive (realised or hypothetical) reason, (33)a.,b., and there is the further restriction on adverbial relatives that P-N has to be the lexical pro form for reason :

(35) a. The reason

that why *for which

b. The rain j ¿ ¡ ^

he stayed at home is . . .

Qf which

J he stayed at home was . . .

(36) a. Why he stayed at home was for the following reason: because of *in spite of

b. Why he stayed at home was

the rain

(37) a. Why did he stay at home? - For the following reason: b. Why did he stay at home?

-

Because of *In spite of

the rain. 8

It thus appears that in respect of Reason pseudo-cleft and interrogative are more closely related to each other than to relative. ADJECTIVAL INTERROGATIVES (N.9) seem to be generally possible, except for the exponential constraint that N should be a countable noun ( * b e c a u s e

of/etc. what rain did he stay at home? ). But there seem to be no restrictions on the exponents of P. As for the ANAPHORIC PRO FORMS (N.6, adverbial, and N . l l , nominal) it seems that only positive, realised reason can be replaced by the pro form

therefore: (38) There was a heavy rain,

a. therefore (= because of it) b. therefore (= in spite of it)

he stayed at home he went out (The asterisk indicates that the interpretation in question is ruled out, not, of course, that the sentence is ungrammatical). On the other hand, the only cases where P+personal pronoun are excluded seem to belong to the positive

82 reason type: (a) if P-N is the lexical pro form for Det reason, (b) in some cases where P is at, on, and where N provides the only indication that P-N is a reason adjunct: (39) a. T h e y thought that this was a very good reason, and they all left for it. b. * John's advice was to leave early, and we all left at it. In all other cases personal pronouns occur freely: (40) a. [The necessary faculties] are more likely to be perverted by too much culture than to wither for the lack of it (CAPS 52) b. Although opposition may be unpleasant you shouldn't resort to too tough measures in the face of it. c. This is an important stimulus, and the female usually fails to lay eggs in the absence of it. (For in spite of it, because of it see (38)). I have n o t illustrated COORDINATION OF P-N WITH AN ADVERB (N.12) w i t h

Time, Duration, and Frequency, since there are quite large classes of -ly adverbs in these functions with which the prepositional phrases in question obviously coordinate. There is only a handful of adverbs with Reason function, all of which have the syntactic characteristics of conjuncts (the contrastive, illative, and inferential conjuncts of Greenbaum, 1969: 36-37). The reason phrases treated in this section may coordinate with some of these conjuncts (the fact that reason phrases do not coordinate with all contrastive, illative, and inferential conjuncts must be put down to a restriction on the conjuncts rather than on the Reason phrases): (41) a. It was a sunny day. Therefore and because of his doctor's advice, he walked to the station instead of taking the car. b. Sometimes the doors won't open. Then or in case of fire you can get out through the emergency exit. c. There was a heavy rain.

}and in spite of his

mother's advice, he went out without an umbrella. Reason phrases may also coordinate with Epithet adverbs (cf. below): (42) Foolishly and in spite of the great danger, he insisted on swimming across the river where the current is strongest.

83 5.1.5 Epithet 5.1.5.1 Introductory remarks The morphological similarity between Epithet and Manner was mentioned above 2.2.2.1. The close link between the adverbial and the subject is reflected in several syntactic facts of which I mention only two: (i) There are compatibility restrictions between Epithets and the subject (cf. Quirk et al., 1972: 467). 9 (ii) In the criteria involving sentence, predicate, or verb phrase pro forms, and where there is a choice between (a) pro forms in Sb+ c. Praschko had done something to him J

All clusters containing central participants are positive to the following criteria: N.l (occurrence in imperative clauses), N.2 (occurrence as focus of clause interrogation), N.3 (focus of clause negation), N.4 (occurrence out-

180

side scope of negation), N.6 (anaphoric pro-adverbial), N.12 (coordination with adverb), N.17 (occurrence outside scope of predicate pro form). Table 8.1 records the reactions to the following criteria: N.5 (adverbial pro form in relative, interrogative, and pseudo-cleft constructions), N.7 (nominal relative), N.8 (pseudo-cleft construction with nominal pro form), N.9 (adjectival interrogative), N.10 (nominal interrogative), N.ll (pronominalisation of-N), N.13 (I-position of P-N), N.14 (M-position of P-N), N.15 (object raising), N.16 (passive), V.l (P-N outside scope of nominalisation of S), V.2 (P-N outside scope of pronominalisation of S), V.3 (coordination with Vtr), V.4 (commutability of P), V.5 (commutability of V), V.6 (adjectivisation of V-P), V.7 (indivisibility of V-P). v 5 EC Agt Exp Aff

+/-

+ + +

7 +/-

+

+ +

8

9

10

+/-

+/-

+/-

+

+ + +

+

+ +

+

+

11 +/-

+ + +

13 +/-

+ + +

15

1

v

v

v

v

v

14

2

4

3

7

5

16

v

-

-

-

_

+/-

_ _

6

+/-

+/-

+/-

-

+

+/-

+

+/-

+

0

+ +

+/-

+/-

+/-

+/-

+/-

+/-

-

-

+/-

+/-

+/-

+/-

+/-

-

0 +

+

+

Table 8.1

The distribution of positive and negative reactions indicates that there is a cline between EC and Affected. Note that in general central participants are more loosely attached to V than marginal participants. This accords well with the observation (above p. 36) that prepositional realisations are a-typical for these functions (cf. further below p. 195).

8.1.1 External Causer 8.1.1.1 Introductory remarks. Since the type where EC and Agt combine in one item (the sergeant in the segeant marched the soldiers) does not seem to have prepositional realisations (cf. above p. 36), the type of EC we are left with is semantically quite similar to Instrument, in that both are elements causually involved, though not through any energy of their own. They are distinguished by the fact that Ins demands the (implicit) presence of an Agt which controls it, whereas EC is causal through its mere existence. This semantic distinction is accompanied by a syntactic distinction: Ins is realised as a free adjunct, EC as a bound adjunct. The similarity between the two functions is reflected in

181 the fact that in some cases EC may answer a how question (see below): How did the problem arise? - From the breakdown of established values. Before I go on to consider the reactions to the individual criteria, I wish to consider briefly a problem of analysis. In the curtains fluttered in the breeze one would hardly hesitate to call the breeze EC (the breeze made the curtains flutter); the sea glittered green in the sun-light seems a completely parallel example (the sunlight made . . .), but what, then, are we to do with the sea glittered green in the dark (owing, presumably, to some kind of phosphorescent micro-fauna), which, again, appears to be exactly parallel, but where the dark is definitely not EC. I have relied upon the N-makeSb-V paraphrase, which makes a distinction between the first two and the last examples, although, on the other hand, it overemphasises the difference between the second and the third example. 8.1.1.2 Syntactic features. RELATIVE, PSEUDO-CLEFT AND INTERROGATIVE PRO FORMS (N.5, 7, 8,

10). Reaction to N.5 .a. (adverbial relative) is always positive, whereas negative reaction to N.7 (nominal relative) occurs in the two types tremble with enthusiasm, flush for shame (e.g. *the enthusiasm with which he trembled was ill-advised)', and in a third type (flutter in the breeze) relativisation is quite awkward (? the breeze in which the curtains fluttered). Note that the reaction to relativisation cannot be traced back to the abstract-concrete distinction in N: the breakdown in values from which the problem arose . . . The majority of clusters are positive to N.5.b. (pseudo-cleft with adverbial pro form), even tremble with, etc. (*why he trembled was with enthusiasm) but a handful of clusters accept the pro form how: (5) How the country will gain is out of there being a coalition. Similarly, the problem arose from the breakdown . . . These clusters are also among those positive to N.8 (nominal pseudo-cleft): what the problem arose from . . . , whereas tremble with, etc. are negative: *what he trembled with . . . Negative to N.5.c. (adverbial interrogative word) are not only gain out of, arise from, die of (how? ), but also tremble with, etc. (why?). Only quite few are positive to N . 5 . C . , among them he staggered under the weight, the curtain fluttered in the breeze. PRONOMINALISATION OF N (N.ll). The clusters that do not allow relativisation or nominal interrogatives are, as expected, also negative to this criterion: *his rage was so great that he smouldered with it. But if N denotes an event

182 rather than a psychological state the result is not quite as bad: ? the blow

was so hard that I nearly fell over with it. And the flutter in the breeze type is questionable: ? there was a bright sun, and the sea glittered in it. POSITION OF P-N (N.13-14). Most clusters are negative to N.14 (i.e. they accept M-position of P-N) provided that the medial P-N is outweighed by a rather heavy postverbal part of the sentence: (6) The curtain, in the breeze, undulated *(gently). Note also that P-N is separated from the rest of the sentence by commas (cf. above p. 109). For the cluster die of N M-position of P-N is out of the question (i.e. positive reaction): *few people, of smallpox, die nowadays. Die of again is positive to N.13 (I-position of P-N), whereas most other clusters are negative (i.e. P-N may be fronted), even without the addition of extra postverbal elements: (7) From the breakdown of established values, new problems arose. OBJECT RAISING (N.15). By far the majority of clusters are negative to this criterion, and the only completely natural candidate for object raising seems

to be die of: Tuberculosis is an unpleasant disease to die of. PASSIVISATION (N.16). EC clusters are invariably negative. (PRO)NOMINALISATION OF S (V.l-2). Negative reactions to V.l (i.e. acceptability of P-N outside the scope of nominalisation of S) occur only if P is according to or under [(8)a.-b.], whereas all clusters, even the highly cohesive die of N, are negative to V.4 (are acceptable in the construction) [(9)a.-b.]: (8)a. The variation in prices is

f o r d i n g to I

b. The collapse of the organisation was

thesupply

under *with

the added load.

(9)a. His eyes kindled. But it was not with enthusiasm, b. When he died it was of tuberculosis. COORDINATION WITH Vtr. (V.3). Since EC elements can occur either as subjects or as prepositional complements, but not as non-prepositional objects (above p. 36), the identity-of-function condition is never met, and strictly speaking the criterion is therefore inapplicable. Note, however, that even if the condition is relaxed, putative coordinations are unacceptable: *he caught

183 and died of TB. INCOMMUTABILITY OF P AND V (V.4-5). Probably all combinations are negative to V.4 (i.e. P is commutable), and almost certainly all to V.5 (V is commutable): change/vary according to/with the supply, laugh/shout with/for joy, float/wave in/before the wind. Note that in many cases there are close collocational ties between P and N rather than between V and P, and accordingly the combinations in question cannot be regarded as prepositional verbs (cf. above 1.1, fn. 2): the organisation breaks down WITH the allocation of a couple of new tasks/ UNDER the weight of the extra work; I could have died FOR shame, she died FROM natural causes, he was ready to die WITH vexation, he died OF fever and starvation (cit. Schibsbye, 1970: 352-3). INDIVISIBILITY OF V-P (V.7). As expected, relatively loosely linked combinations like EC combinations are always separable. (10) is a textual example: (10) [sea-slugs are] dim, primitive beasts that just lie in one spot, rolling gently with the sea's swing [....] (FOA 63) ADJECTIVISATION OF V-P (V.6). Since positive reaction presupposes positive reaction to passivisation, the negative reaction is expected. PREPOSITION STRANDING. For the clusters that do not accept relativisation (or nominal interrogatives, cf. above) the question of the position of P in relative and interrogative clauses does not arise, of course. For the others the possibilities are as follows: All clusters accept both clause initial and postverbal position of P in relative clauses (the conditions out of which the problem arose/. . . that the problem arose out of), with the exception that the complex P according to is extremely awkward postverbally (*the factors that prices vary according to). In interrogative clauses both positions are possible too, and the general tendency towards postverbal P's in interrogative clauses makes a final according to somewhat more likely here (/ wonder what factors prices vary according to).

8.1.2. Agentive 8.1.2.1 Introductory remarks. Agentive is definitely a function for which prepositional realisations are

184 secondary, the primary realisation being as subject (cf. above p. 37). In fact the only type I shall discuss in this section is Agentive-as-Comitative, since the only other prepositional realisation of Agt that I am aware of, viz. the by phrase in passive sentences is outside the scope of this study. For this reason, the cases here dealt with as Agt form an extremely homogeneous group, which appears most clearly in the fact that the exponents of V are all 'symmetric predicates' in the sense of Lakoff & Peters (1969); the exponent of P is invariably with', and the exponents of N are human nouns (including personifications) or collectives like federation, government, union, etc. Agt/Com is distinct from 'pure Comitative' (cf. 6.2.1.4) in being a bound adjunct, from Obj/Com (cf. above p. 159, i.) in the availability of an N-dosomething paraphrase, and from Int/Com (cf. above p. 152 f.) in the exponence of V. Note the ambiguity of e.g. fight with: In World War II Britain fought with Germany [Agt/Com] vs. In World War II Britain fought with France [Com] 3 'Britain did something together with France'. 8.1.2.2 Syntactic features. RELATIVE, PSEUDO-CLEFT AND INTERROGATIVE PRO FORMS (N.5, 7, 8,

10). Agentive clusters are positive to all these criteria, i.e. there are no adverbial pro forms, and constructions with nominal pro forms are always possible. The reaction to N.8 (pseudo-cleft with nominal pro form) is problematic only in so far as the human pro form who cannot function in pseudocleft constructions (or as an independent relative, cf. above p. 112). Instead, one has to resort to the one : the ones I joined with were the wrong people. PRONOMINALISATION OF N ( N . l l ) .

N m a y always pronominalise: I've

been

struggling with him for years. All clusters are positive to both criteria, i.e. P-N may not occur in M- or I-position except under emphasis: POSITION OF P-N (N.13-14).

( l l ) a . *Dust, with paper, competed for room on the shelves, b. *With paper dust competed for room on the shelves. Recall that in similar circumstances 'pure Comitatives' are possible in both positions (cf. above p. 113). By definition the structural description is always met, and the reaction is always positive: he's not worth fighting with. OBJECT RAISING (N.15).

185 PASSIVISATION (N.16). Agentive is listed as negative to this criterion in table 8.1, in spite of the fact that some of the clusters in question actually passivise, e.g. I dislike being competed with. In cases like this, however, the Objective interpretation of N referred to in fn. 1 below comes into force. This situation is thus analogous with that encountered with Comitative (/ have never been eloped with before, see above p. 114). Note that join lexicalises the Agt-Obj distinction: join with N [Agt/Com], join N [Obj], and that join with, accordingly, does not passivise: Mr Thorpe was joined (*with) by Mr May hew.

(PRO)NOMINALISATION OF S (V.l-2). In spite of the homogeneous character of Agt combinations, the reactions to these criteria seem to reveal three degrees of cohesion, (a) combinations negative to both criteria, (b) combinations positive to V.l and negative to V.2, (c) combinations positive to both (recall that acceptability in the relevant constructions is registered as negative reaction): (12)a.l. The most ruthless fight(ing) was with the Unions. 2. Whenever we have had to fight, it has been with the Unions. b.l. T h e only real competition was with Mrs King. 2. The only time Miss Evert had to really compete, it was with Mrs King. c.l. *Mr Mayhew's joining was with Mr Thorpe. 2. *When Mr Mayhew joined, it was with Mr Thorpe A large class patterns like fight: e.g. battle, feud, struggle, shake hands. A smaller class, consisting of Latinate verbs in co-, follows the pattern of compete, e.g. collaborate, cooperate. I am not aware of any verbs following the pattern of join, although their existence cannot be ruled out a priori.1 COORDINATION WITH Vtr (V.3). Since the function Agt is probably never realised as a (non-prepositional) object, the situation in which a Vtr and a V-P share an object does not arise, and the criterion is inapplicable. INCOMMUTABILITY OF P AND V (V.4-5). In table 8.1 Agt is listed as positive to V.4 (i.e. P is incommutable), in spite of the fact that in quite a number of the clusters with could be replaced by against, e.g. battle/compete/'fight/ struggle. In such cases, however, 'with involves a reciprocal concept, while against expresses only one aim in the action, viz. in the direction of the preposition's complement'. (Schibsbye, 1 9 7 0 : 3 1 5 - 1 6 ) . This statement translates directly into the terminology of the present study: the N of e.g. fight

186

with an enemy is Com/Agt, 2 while that of e.g .fight against illiteracy is Obj (of the purpose variety). This also explains why, with the verbs in question, any N that can function in a w/f/z-phrase can also function in an againstphrase (fight against an enemy)-, if all w/ili-phrases are ambiguous between a Com/Agt and an Obj interpretation, substituting against for with is in fact nothing but resolving the ambiguity in the Obj direction. By the same token, every instance of against is not replaceable by with (Schibsbye, loc. cit.) since there is no Agt - Obj ambiguity in against: race against time, fight against illiteracy/disease/pollution. All V s occurring in Agt clusters are mutually commutable, i.e. negative reaction to V.5. INDIVISIBILITY OF V-P (V.7). The only combination that I am aware of that does not allow separation of V-P by an adverb is, not unexpectedly, cf. (12) above, join with:

(13)a. Mr Mayhew invariably joins with the wrong people, b. *Mr Mayhew joins invariably with the wrong people. All other combinations are negative, e.g. fight/compete ruthlessly

with.

ADJECTIVISATION OF V-P (V.6). The negative reaction to this criterion follows from the negative reaction to passivisation (N.16).

PREPOSITION STRANDING. Regardless of degree of cohesion (cf. V.l-2, above), both clause initial and postverbal position of P is possible in relative as well as interrogative clauses: the people with whom he was fighting/. . . he was fighting with; I wonder with whom he will join next time/. . . who he will join with . . .

8.1.3 Experiencer. 8.1.3.1 Introductory remarks. Above p. 153 it was noted that there is a highly restricted set of verbs with an Objective element in subject position and a prepositional Dative element, e.g. belong to. The existence of a parallel construction with Phen as subject and a prepositional Int was not discussed, although the term 'Interlocutor' suggests that there are no such cases. There are however psychological verbs (in the sense of 2.2.3.2 and 7.1.5.1) taking Phen as subject and a typically human noun as a prepositional complement (e.g. the book ap-

187 pealed to me), and this type will be dealt with here under the label Experiencer. There are two reasons for treating this type as distinct from Int, whereas the corresponding Dat type is not accorded separate status: (a) the number of verbs entering into the construction is vastly greater than the handful of verbs accepting the corresponding construction with Dat. (b) In a number of cases it seems reasonable to regard the prepositional realisation as secondary as compared with a primary non-prepositional realisation. And as has been noted repeatedly (above pp. 36, 180, 183 f.) this is characteristic of central participants. Although the 1. and 2. examples of (14) do not demonstrate any rank between the prepositional and non-prepositional Exp elements, the pairs do show that many Exp clusters have (near-)synonymous versions with Exp as subject: (14)a.l. The book appealed to me. 2. J liked the book. b.l. It hadn't occurred to me. 2.1 hadn't thought of it. c.l. Politics doesn't matter for him. 2. He doesn't care about politics. d.l. The idea seems useless to me. 2.1 find the idea useless. Postal (1971: 39 ff.) regards the primacy of the non-prepositional realisations of Exp as established, and proposes to account for the derived nature of examples like b.l., d.l., by means of a specific transformation, 'Psych Movement'. 8.1.3.2 Syntactic features. RELATIVE, PSEUDO-CLEFT AND INTERROGATIVE PRO FORMS (N.5, 7, 8,10).

Exp clusters are positive to all these criteria, i.e. there are no adverbial pro forms, and constructions with nominal pro forms are always possible. Since N is always human, N.8 (pseudo-cleft with nominal pro form) is applicable only with the pro form the one: The ones the idea seemed useless to were the habitual pessimists. PRONOMINALISATION OF N (N.ll). N may always pronominalise:

(15) [....] it /seems to me there may be a f e w ^ /lessons^ /learned^ by I many seaside resorts^ (5b.16.67)

188 POSITION OF P-N (N.13-14). In spite of the existence of examples like ( 1 6 ) ,

in which P-N is fronted, (16) Now to an institution believing [....] that there is great merit in being independent these carefully written-in and underlying compulsory powers might appear in the Earl of Longford's phrase 'sinister' [....] (CAPS 68) Exp is registered as positive to N.13 (I-pos of P-N), since fronting is possible only under emphasis. Note that in (16) the requirement (cf. above p. 52) that N should be a simple (definite or indefinite) nominal group is not fulfilled. N.14 (M-position of P-N) splits Exp clusters up into two groups: (17)a. Classical culture, for most people, doesn't count any more, b. *Seafood, with many people, doesn't agree. Other group a. members are: matter to/for N; look/taste, etc. Adj to N. Other group b. members: appeal to N; occur to N. Note that this distinction links up with another distinction which also indicates that the more cohesive combinations are those of group b.: only group a. clusters allow deletion of the entire P-N without shift of meaning, e.g. details don't matter (to me), but the idea hadn't occurred *(to me). OBJECT RAISING (N.15). Since the subject of Exp clusters is Phen, it cannot

be the experiencer of an evalutation (cf. above p. 53), and it follows that the criterion is inapplicable. PASSIVISATION (N.16). All combinations are negative. This supports the observation (above p. 166) that prepositional verbs, unlike non-prepositional transitive verbs, require agentive subjects for passivisation to take place. (PRO)NOMINALISATION OF S (V.l-2). All combinations are positive to V.l: *the sour taste of the beer was to Jim, *the appeal of the poem was to Roy. All members of group a. except appear to, seem to are negative to V.2, and members of group b. are positive: (18)a. If these things matter at all, it is only to a tiny minority, b. *Whenever good ideas occur, it is to Frank. COORDINATION WITH Vtr (V.3). The only case of positive reaction that I can

189 think of is this: a book that both disturbs and appeals to me. All others seem to be ungrammatical, even e.g. occur to and intrigue, which would in fact make quite good sense: *an idea occurred to and intrigued me. INCOMMUTABILITY OF P AND V (V.4-5). The negative reactions to V.4 (incommutability of P) come from verbs which can take either for or to, e.g. count/matter, seem/appear (for the slight difference in meaning of for and to with these verbs, see Schibsbye, 1970: 347-51). Note that the combinations in question are also classified as relatively incohesive by N.14 (Mposition of P-N). Positive reactions come from a. cases where to is the only possible exponent of P (appeal to, occur to), b. the unique combination agree with. As the above examples show, V is commutable in all cases where P is either for or to, and the only case of positive reaction to V.5 thus seems to be agree with. INDIVISIBILITY OF V-P (V.7). The majority of V(-P)'s taking Exp complements belong to the class of 'emotive verbs' (cf. Greenbaum, 1970: 70), which is characterised by a tendency to collocate with no other adverbial adjuncts than intensifiers (op. cit.), and for this reason the restriction laid down above, p. 59, that the intervening element must not be an intensifier has to be given up in the treatment of the combinations in question. Instead, they are tested for divisibility with respect to the closed-class items very much/well and the -ly adverb definitely. It appears that all combinations (except occur to, which is not an 'emotive verb') may be separated by very much/well, whereas -ly adverbs are unacceptable or at best marginal: politics matters very much/*definitely to me, it looks very much/*definitely to me as if it's going to rain; the poem appealed very much/*definitely to Andrew, ? *it occurred suddenly to me . . . ADJECTIVISATION OF V-P (V.6). The negative reaction to this criterion follows from the negative reaction to passivisation (N.16). PREPOSITION STRANDING. The two groups of Exp clusters established by means of the reaction to N.14 (M-position of P-N), above, are also distinguished by the position of P in relative clauses. In group a. the postverbal position is extremely awkward, whereas in group b. both positions are possible: (19)a.l. There are several people for whom classical culture does not count. 2. There are several people to whom politics does not matter.

190 b.l. There are several people with whom seafood does not agree. 2. There are several people to whom such an idea would never have occurred. (20)a.l. *There are several people whom classical culture does not count for. 2. *There are several people whom politics does not matter to. b.l. There are several whom seafood does not agree with. 2. There are several people whom such an idea would never have occurred to. All group a. combinations permit clause initial P in dependent interrogative clauses (tell me for whom classical culture counts nowadays), whereas one group b. combination, agree with, resembles Phen, and certain Eq and Obj combinations in refusing to have P and V separated in this construction: (21)a. *Tell me with whom lobster does not agree. b. Tell me to whom this idea first occurred. We have observed repeatedly that P is more likely to be final in interrogative than in relative clauses, and in fact it turns out that even the group a. combinations that were found not to permit clause final P in relative clauses are at least somewhat better with final P in interrogative clauses: (22)a. ? Tell me whom classical culture counts for nowadays, b. ? Tell me whom politics matters to anyway.

8.1.4

Affected.

8.1.4.1 Introductory remarks In the preceding chapters I have discussed several examples in which Affected combines with another, primary function (cf. e.g. above pp. 100, 106, 110, 118, 131, 141). Thus, though the N of e.g. the flower beds have been trodden on is clearly affected by the action denoted by V-P, it is classified as Loc/Aff rather than Aff alone. In this way the majority of cases in which the referent of N is affected are accounted for, and we are left with a comparatively limited number of cases in which Affected must be regarded as the primary function. The V-P combinations taking an N in the role Aff fall into two semantic groups: a. combinations meaning 'affect/influence (favourably/unfavourably)', b. combinations meaning 'do something to/something happen to' in which the

191 content of 'something' is specified. (23)a. ill-health tells against his prospects, the call for equality may impinge on our educational system, the wealth of a society bears upon its outlook. b. the light broke upon us, I shouldn't be imposing on you, his wife had walked out on him. The combination happen to itself will have to be included in group b., in spite of the fact that it differs from the other members in several respects (see below). Note that group a. combinations take Objective subjects, whereas those of group b. may be either Objective or Agentive. 8.1.4.2 Syntactic features RELATIVE, PSEUDO-CLEFT AND INTERROGATIVE PRO FORMS (N.5, 7, 8,

10). Aff clusters are positive to these criteria, i.e. there are no adverbial pro forms, and constructions with nominal pro forms are always possible: the nerves that this drug acts on, what it really acts on is the motor system, what does it act on? PRONOMINALISATION OF N (N.ll). N may always pronominalise:

(24) A. [I'll get the exam papers through quickly, once I've had them from you] B. and we /shouldn't be imposing on you in vacation t i m e ^ (s.lc.l 1-14) POSITION OF P-N (N.13-14). All clusters are positive to both criteria, i.e. P-N is possible neither in I-position (N.13) nor in M-position (N.14). OBJECT RAISING (N.15). All clusters that can meet the structural description of object raising are positive to the criterion: he is too easy to impose on, he is dangerous to walk out on. But the criterion is of course inapplicable to the combinations taking Objective subjects, e.g. act on, bear on, tell against/ on; happen to. PASSIVISATION (N.16). The semantic distinction between groups a. and b. , above, is reflected in the reaction to passivisation, in that group a. combinations (whose subjects are Objective) are always negative (*his prospects

are told against by ill-health), with impinge on as a marginal exception

192 (? the quality of the education offered is likely to be impinged on by the call for equality). Group b. combinations passivise if the subject is Agentive: he is being imposed on/walked out on by all his friends, but not if it is Objective: *we were broken upon by the sun, *he had been happened to by nothing. (PRO)NOMINALISATION OF S (V.l-2). All clusters are positive to b o t h criteria

(i.e. the constructions are unacceptable): *John's imposing was on Fred, *when he imposed it was on Fred. COORDINATION WITH Vtr (V.3). The majority of combinations are positive.

Group a. the drug acts on and permanently changes vital parts of the nervous system, the tax is likely to impinge on and perhaps ruin several small firms; group b. old friends pestered and imposed on him, I saw how he operated on and transformed people. The only combination that definitely does not enter into coordinations is happen to. INCOMMUTABILITY OF P AND V (V.4-5). The only combinations failing to satisfy V.4 that I am aware of are tell against¡on, and tell against, along with happen to, seems to be the only combination taking a P other than (upjon.3 Within the -(upjon combinations it is possible to distinguish two degrees of cohesion according to the possibility of deletion of P-N (cf. Carvell & Svartvik, 1969: 45, and see above p. 57). The less cohesive type (i.e. where deletion is possible) follows the pattern of walk out on and is completely open-ended, e.g. (25)a. Men whose wives have walked out or died on them . . . b. 'Was this a grenade? ' he asked with excitement and hope. [....] once I was interested myself in what for want of a better term they call news. But grenades had staled on me; 4 Similarly: close down on, disappear on, hold back on, etc. This type is further characterised by the fact that of the variants on/upon, the longer form is virtually excluded. 5 The more cohesive type makes up a lexically closed, but quite large, set. All group a. combinations seem to belong here as well as a number of group b. combinations. Examples: the tax will impinge *(on small firms), the wealth of a society bears *(on its outlook); the light broke *(on us), he rounded *(on me) furiously. In this type there is an option between on and upon.

193 The only combinations where V is incommutable (positive reaction to V.5) are those with a P other than (upjon: tell against, happen to. INDIVISIBILITY OF V-P (V.7). Group a. combinations, which are frequently non-event verbs, are positive to V.7 in so far as V-P may be separated by intensifies and no other adverbs: (26)a. The wealth of a society will bear heavily upon its outlook, b. T h e wealth of a society will bear indisputably upon its outlook. 6 In group b. it is necessary to distinguish between the more and the less cohesive type established above. The more cohesive type allows insertion of intensifiers (if they can occur with the V-P in question at all) and relatively lightweight adverbs like suddenly. (26)a.l. 2. b.l. 2.

I've been imposing heavily upon you this last few weeks. *I've been imposing ruthlessly upon you . . . He rounded suddenly on his colleague. ? *He rounded furiously on his colleague.

This difference in acceptability according to information value of the adverb does not seem to obtain among the less cohesive type: (27)a. He closed down suddenly on me. b. He closed down rudely on me. ADJECTIVISATION OF V-P (V.6). As will have appeared, Aff combinations are on about the same level of cohesion as Obj or perhaps even Phen combinations, but in spite of this the reaction to V.6 seems to be invariably negative. Recall however, that even among Obj combinations adjectivisation is the exception rather than the rule (above p. 161), and moreover, as will appear presently, no Aff combination seems to reach the extreme degree of cohesion of not permitting fronting of P in interrogative clauses, unlike some Obj and many Phen combinations (above pp. 163, 169 f.). PREPOSITION STRANDING. There are no combinations in which both front and end positions of P in relative clauses are not grammatical (e.g. the nerves that this drug acts on/on which . . .). Similarly, both positions are equally grammatical in interrogative clauses (I wonder who he's been imposing on now/on whom . . .). It thus appears that Aff combinations make up a much

194 more homogeneous class that Exp combinations, which range from not permitting final P's in relative clauses {matter to) to not permitting initial P's in interrogative clauses (agree with), cf. above pp. 189-90.

8.2 THE CORPUS In the corpus the following structures are attested (cf. above 3.3): d. P-N occupies initial position. With that one little bite his head had swollen up as though his brains were pregnant [....] (FOA 72) g. P is separated from V by an adverbial (including other prepositional phrases), an intonation break, or a comma. [....] a rather ancient grand piano on which dust competed for room with sheets of written or printed notes. (GW 31.1.70, 18: 3) h. The neutral structure, V-P-N. [....] a brilliantly successful comprehensive school in the North of England (which at one point he commends, in terms that will not appeal to every parent, as 'a school that never says no') (CAPS 51) i. N is a (personal) pronoun. [....] it /seems to me that you've got two . two things^ a :m. that you can . / d o with t h i s ^ (s.lc. 11-43) The distribution of the four structures among the four functions is shown in table 8.2. d

g 16 2 3 5

h

i

total

54 6 13 17

6 1 10 12

77 9 26 34

EC Agt Exp Aff

1

total

1

26

90

29

146

%

.7

17.8 61.6

19.9

100.0

Table 8.2 The three types of structures, i. adverbial (d., g.), ii. neutral (h), and iii. nominal (i.), each account for the following percentages, which should be compared with the corresponding values for marginal participants, concrete

195 free adjuncts, abstract free adjuncts, and non-role playing phrases: i. adverbial ii. neutral iii. nominal

18.5 61.6 19.9

(cf. table

8.2

(16.8) (63.3) (19.9) 7.2

(36.7) (50.9) (12.3) 6.2

(67.1) (29.6) ( 3.3) 5.2

(73.7) (22.1) ( 4.2) 4.2)

The tendency towards a decreasing proportion of adverbial structures and a higher proportion of neutral and nominal structures is now broken: the differences in the distribution of the structural types among central and marginal participants are negligible. This is not surprising in view of the fact that (a) EC and Agt have been found to be less cohesive than the majority of combinations belonging to the marginal participants group, and (b) that the maximum degree of cohesion displayed by some Eq and Obj combinations and a number of Phen combinations is reached by only one combination from the central participants group, the Exp combination agree with (cf. above p. 189 f.). These facts are of course special instances of the observation (above p. 180) that within the central participants group the degree of V-P cohesion is in general smaller than within the marginal participants group. And the explanation for this (as well as for the small size of the corpus of central participants) lies in the observation (above p. 36) that central participants, unlike all other functions, are typically realised as non-prepositional nominals (i.e. as subject or object).

NOTES 1. Note that the complement of side with, which might seem a plausible candidate, is neither Comitative nor Agentive, but simply Objective: some backbenchers sided with the Opposition *some backbenchers and the Opposition sided some backbenchers and the Opposition did something It should perhaps be pointed out that in the cases dealt with here as Agt, the withphrases are ambiguous between an Objective and an Agentive (cf. also Vestergaard, 1973b, and see below). 2. Actually it is ambiguous between a Com/Agt and an Obj interpretation, cf. fn. 1. 3. Note that happen to/upon does not constitute a case of commutability of P since the identity-of-function condition on the criterion (above p. 57) is not met. The N of happen to is Affected, that of happen upon is Objective: something must have happened to him - he happened upon an interesting example (cf. above p. 162). 4. Graham Greene, The Quiet American (Penquin Books, 1962), p. 16. 5. The walk out on type can be seen as a way of attaching an Affected element to a verb which is not inherently marked for occurrence in the surrounding [-Aff] in much

196 the same way as Having and Purpose (above 6.1.3.1, 7.1.4.1) were regarded as 'intrusive Objectives'. In fact it seems that an Aff on-phrase may be used with any (transitive or intransitive) verb, as long as the process it identifies can be seen as affecting the N from a subjective point of view (e.g. she disappeared - she disappeared on me, he closed the door - he closed the door on me). The preference for on to upon in this type is thus an exception to the first half of Schibsbye's statement (1970: 380) that 'an association of something subjective, an emotional attitude, may be linked with upon-phrases, while corresponding on-phrases have an objective association'. 6. But if the adverbial is separated intonationally or orthographically from the clause, the result may be acceptable if not pretty: The wealth of a society will bear - indisputably, though perhaps not favourably - on its outlook.

BIBLIOGRAPHY

Anderson, J. M. 1971 The Grammar of Case, London: Cambridge UP. 1973 An Essay Concerning Aspect, The Hague: Mouton. Anderson, S. 1968 "Pro-sentential forms and their implications for English sentence structure", in Mathematical Linguistics and Automatic Translation, The Computation Laboratory, Harvard University, report no. NSF-20 to the National Science Foundation, VI-l-VI-43. Bach, E. 1967 "Have and be in English syntax", Language 43, 462-85. Bach E. and R. T. Harms (eds.) 1968 Universals in Linguistic Theory, New York: Holt. Barkai, M. 1972 "On the shiftability of past participles in English", Linguistic Inquiry 3, 377-83. Bartsch, R. 1972 Adverbialsemantik, Frankfurt a. M. : Athenäum. Bloomfield, L. 1935 Language, London: Allen & Unwin. Bolinger, D. 1971 The Phrasal Verb in English, Cambridge, Mass.: Harvard UP. 1973 "*John's easiness to please", in G. Nickel (ed.), Special Issue of IRAL on the Occasion of Bertil Malmberg's 60'th Birthday, 17-28. Brose, B. 1939 Die englischen Passivkonstruktionen vom Typus 'I am told a story' und 'I am sent for', Würzburg-Aumühle: Triltsch. Burt, M. K. 1971 From Deep to Surface Structure, New York: Harper & Row. Carvell, H. T and J. Svartvik 1969 Computational Experiments in Grammatical Classification, The Hague: Mouton. Chapin, P. G. 1972 "Review of Stockwell et al., Integration of Transformational theories of English Syntax", Language 48, 645-67. Chomsky, N. 1957 Syntactic Structures, The Hague: Mouton. 1965 Aspects of the Theory of Syntax, Cambridge, Mass.: MIT Press. Christophersen, P. and A. O. Sandved 1969 An Advanced English Grammar, London: MacMillan. Corder, S. P. 1968 "Double-object verbs in English", Studio Anglica Posniensia 1, 15-28. Cruse, D. A. 1973 "Some thoughts on agentivity", Journal of Linguistics 9, 11-23.

198 Crystal, D. and R. Quirk 1964 Systems of Prosodic and Para-linguistic Features in English, The Hague: Mouton. Danes, F. 1968 "Some thoughts on the semantic structure of the sentence", Lingua 21, 55-69. Diderichsen, P. 1943 "Logische und topische Gliederung des germanischen Satzes", in Diderichsen (1966b), 52-63. 1964 "Saetningsleddene og deres stilling - tredive ar efter", in Diderichsen (1966b), 364-79. 1966a Elementaer dansk Grammatik, K^benhavn: Gyldendal. 1966b Helhed og Struktur. Selected Papers with Detailed English Summaries, K^benhavn: Gad. Dik, S. C. 1968 Coordination: Its Implications for the Theory of Linguistics, Amsterdam: North-Holland. Erades, P. A. 1957 "Points of modern English syntax XXXIII", English Studies 38, 139-42. Fairclough, N. L. 1965 Some English Phrasal Types: Studies in the Collocation of Lexical Items with Prepositions and Adverbs in a Corpus of Spoken and Written Present-Day English, M. A. thesis, University College London. Fillmore, C. J. 1968 "The case for case", in Bach & Harms (eds.), 1-88. 1969a "Types of lexical information", in Kiefer (ed.), 109-37. 1969b "Toward a modern theory of case", in Reibel & Schane (eds.), 361-75. 1972 "Some problems for case grammar", mimeo, distributed by OSCULD, Dept. of Linguistics, University of Gothenburg. Fillmore, C. J. and D. Terence Langendoen (eds.) 1971 Studies in Linguistic Semantics, New York: Holt. Fräser, Bruce 1974 The Verb-Particle Combination in English, Tokyo: Taishukan Publishing Company. Fretheim, T. 1969 "Case grammar and the preposition med in standard colloquial Norwegian", in Tilegnet Carl Hj. Borgstr^m, Oslo: Universitetsforlaget, 30-48. Gaaf, W. van der 1930 "The passive of a verb accompanied by a preposition", English Studies 12, 1-24. Greenbaum, S. 1969 Studies in English Adverbial Usage, London: Longman. 1970 Verb-Intensifier Collocations in English, The Hague: Mouton. Greenbaum, S. and R. Quirk 1970 Elicitation Experiments in English, London: Longman. Halliday, M. A. K. 1967-68 "Notes on transitivity and theme in English, I, II & III", Journal of Linguistics 3, 37-81 and 199-244; 4, 179-215. 1970 "Language structure and language function", in Lyons (ed.) 140-65. 1973 Explorations in the Functions of Language, London: Arnold.

199 Haitvigson, H. H. 1969

On the Intonation

and Position

of the So-Called

Sentence Modifiers

in

English, Odense: UP.

Present-Day Hornby, A . S. 1974

Oxford

Advanced

Learner's Dictionary

London:

of Current English,

Oxford UP. Huddleston, R. D. 1970

" S o m e remarks on case-grammar", Linguistic

1971

The Sentence

Inquiry

1, 501-11.

in Written English, London: Cambridge UP.

Hudson, R . A . 1967

"Constituency in a systemic description o f the English clause", Lingua

18,

225-50. Jacobs, R. A . and P. S. Rosenbaum (eds.) 1970

Readings in English Transformational

Grammar, Waltham, Mass.: Ginn.

Jacobson, S. 1964

Adverbial

Positions

Stockholm: A B Studentbok.

in English,

Jespersen, O. 1894

Progress in Language, London: Sonnenschein.

1914

A Modern

English Grammar II, London: Allen & Unwin (1st. ed., Heidel-

berg). 1924

The Philosophy

1927

A Modern

of Grammar,

London: Allen & Unwin.

English Grammar III,

London: Allen & Unwin (1st. ed.,

Heidelberg). 1949

A Modern

English Grammar

VII, Copenhagen: Munksgaard.

Kiefer, F. (ed.) 1969

Dordrecht: (Reidel) ( = Foundations

Studies in Syntax and Semantics, Language, supplementary

of

series, vol. 10)

Klima, E. S. 1962

"Structure at the lexical level and its implication for transfer grammar", in 1961 International and Applied

Conference

Language Analysis,

on Machine

Translation

of

Languages

London: HMSO, vol. I, 97-108.

Körner, R . 1948

Studier över syntaktisk nybildning Passivbildningerna,

i Svenskan I. De

prepositionella

Lund: Gleerupska Universitetsbokhandeln.

Kruisinga, E. 1927

" A note on final prepositions", English Studies 9, 6-8.

Kruisinga, E. and P. A . Erades 1947

An English Grammar vol. f , Groningen: N o o r d h o f f .

Kuroda, S.-Y. 1969

"English relativization and certain related problems", in Reibel & Schane (eds.), 264-87.

Lakoff, G. 1968

"Instrumental adverbs and the concept o f deep structure",

Foundations

of Language 4, 4-29. 1970a

"Pronominalization, negation, and the analysis o f adverbs", in Jacobs & Rosenbaum (eds.), 145-65.

1970b

Irregularity

in Syntax,

New Y o r k : Holt.

Lakoff, G. and S. Peters 1969

"Phrasal conjunction and symmetric predicates", in Reibel and Schane (eds.), 113-42.

200 Lakoff, G. and J. R. Ross 1966 'A criterion for verb phrase constituency", in Mathematical Linguistics and Automatic Translation, The Computation Laboratory, Harvard University, report no. NSF-17 to The National Science Foundation, II-l -11-11. Langendoen, D. T. 1969 The Study of Syntax, New York: Holt. Lasnik, H. and R. Fiengo 1974 "Complement object deletion", Linguistic Inquiry 5. 535-71. Leech, G. N. 1969 Towards a Semantic Description of English, London: Longman. Lees, R. B. 1960a The Grammar of English Nominalizations, The Hague: Mouton. 1960b "A multiply ambiguous adjectival construction in English", Language 36, 207-21. Long, R. B. 1961 The Sentence and its Parts, Chicago: UP. Lyons, J. 1969 Introduction to Theoretical Linguistics, London: Cambridge UP. 1970 Chomsky, Fontana, Modern Masters. 1970 (ed.) New Horizons in Linguistics, Pelican. Marchand, H. 1951 "The syntactical change from inflectional to word order system and some effects of this change on the relation "verb/object" in English", Anglia 10, 70-89. Mitchell, T. F. 1958 "Syntagmatic relations in linguistic analysis", Transactions of the Philological Society, 101-18. Nilsen, Don L. F. 1973 The Instrumental Case in English, The Hague: Mouton. Noreen, A. 1904 Vart Sprak vol. V, Lund: Gleerups. Palmer, F. R. 1965 A Linguistic Study of the English Verb, London: Longman. Palmer, H. E. 1928 A Grammar of Spoken English, Cambridge: Heffer. Partee, B. Hall 1971 "On the requirement that transformations preserve meaning", in Fillmore & Langendoen (eds.), 1-21. Postal, P. M. 1971 Cross-Over Phenomena, New York: Holt. Potter, S. 1960 "Referential prepositions", in Iser, W. and H. Schabram (eds.), Britannica, Festschrift fur Hermann M. Flasdieck, Heidelberg: Winter, 210-14. Poutsma, H. 1926 A Grammar of Late Modern English Part II section II, Groningen: Noordhoff. 1928 A Grammar of Late Modern English Part I section fl, Groningen: Noordhoff. Quirk, R. 1965 "Descriptive statement and serial relationship", Language 41, 205-17.

201 Quiik, R. and J. Mulholland 1964 "Complex prepositions and related sequences", English Studies 45, Supplement, 64-73. 1972 et al. Grammar of Contemporary English, London: Longman. Reibel, D. A. and S. A. Schane (eds.) 1969 Modern Studies in English, Englewood Cliffs, N. J.: Prentice-Hall. Robinson, J. J. 1970 "Case, category, and configuration", Journal of Linguistics 6, 57-80. Rosenbaum, P. S. 1967 The Grammar of English Predicate Complement Constructions, Cambridge, Mass.: MIT Press. Scheurweghs, G. P. 1961 Present-Day English Syntax, London: Longman. Schibsbye, K. 1970 A Modern English Grammar2, London: Oxford UP. Schreiber, P. A. 1972 "Style disjuncts and the performative analysis", Linguistic Inquiry 3, 321-47. Sebeok, T. A. 1946 Finnish and Hungarian Case Systems: Their Form and Function, Stockholm (= Acta Instituti Hungarici Universitatis Holmiensis, series B. Linguistica, 3). Smith, C. S. 1969 "Determiners and relative clauses in a generative Grammar of English", in Reibel & Schane (eds.) 247-63. Sroka, K. A. 1972 The Syntax of English Phrasal Verbs, The Hague: Mouton. Strang, B. M. H. 1962 Modern English Structure, London: Arnold. Thomas, O. 1965 Transformational Grammar and the Teacher of English, New York: Holt. Tompa, J. 1968 Ungarische Grammatik, The Hague: Mouton. Vasiliu, L. 1968 "Some methodological remarks concerning a semantics of prepositions", Revue Roumaine de Linguistique 13, 541-49. Vestergaard, T. 1973a "On the open-endedness of the formclass "preposition" in English", English Studies 54, 148-63. 1973b "A note on Objective, Instrumental, and Affected in English", Studia Linguistica 27, 85-89. 1974 "Location direction, and aspect", in O. Dahl (ed.), Papers from the First Scandinavian Conference of Linguistics, Goteborg: Dept. of Linguistics. Volbeda, R. 1926 "On the place of prepositions", English Studies 8, 143-52, 169-81. Vondrak, W. 1928 Vergleichende slavische Grammatik vol. II, Gottingen: Vandenhoeck & Ruprecht.

202 Walmsley, J. B. 1971 "The English comitative case and the concept of deep structure", Foundations of Language 7, 493-507. Zandvoort, R. W. 1962 A Handbook of English GrammarLondon: Longman. Zsilka, J. 1967 The System of Hungarian Sentence Patterns, Bloomington: Indiana University Publications.

INDEX Bold face references are to paragraphs, all others are to page numbers. Under each verb are listed only the prepositions with which it forms combinations taking bound complements. ablative, 38, 139 fn. , 140 about, 18, 41, 46, 50, 95 f. , 150 f. , 153, 159-63, 164-69, 170 above, 56 f. , 86, 95 f. , 125, 156 abstract, 13, 30, 32, 72 abstract verb, 24, 33, 35, 115 accept, 33, 93, 115 f. , 118 accompany, 33 according to, 6, 70, 86, 182 f. account, for, 105, 164, 166 ff. accrue, from, 141 ff. to, 39, 126, 148, 150 f. , 153 across, 95, 129, 134, 136 f. , 162 f. act, for, 145 on, 191-93 activity verb, 16-18 add, to, 148 f. , 151-53 add up, to, 156 f. adjectivisation of V-P, 58, 62, 95, 127, 133, 137, 142, 146, 151, 156, 161, 167 f. , 172, 180, 183, 186, 189, 193 adprep, 4 f. , 95 f. , 138 advance, 86 adverbial pro form, see pro-adverbial adversative, 31, 80 Affected, 34, 37, 41, 100 f. , 106 f. , 110, 114, 118, 131, 136, 141, 145, 179 f. , 8.1.4, 194 after, 19, 50 fn. , 58, 62, 74 ff. , 89, 142, 144-47, 159, 161 ff. , 172 against, 19, 52, 95, 128, 144-47, 161, 185 f. 191-93 Agent(ive), 27, 34, 36 f. , 96, 100, 103, 106 f. , 110-12, 114-18, 121, 131, 136, 141, 148 f. , 158 ff. , 166, 179 f. , 8.1.2, 188, 191 f. , 194 f. agree, 19, 168 about, 165 among, 153 on, 52, 126, 165 with, 35, 40, 112, 153, 188-90, 195 ahead o f , 5 f. , 61, 95, 159 aim, at, 18 allative, 38, 128, 139 fn. , 143

allow, for, 168 allowing for, 6 allude, to, 168 f. along, 95, 134 f. , 136 amid, 95 among, 95, 122, 131, 134, 153, 171 amount, to, 154, 156 ff. anaphora, adverbial, 19, 49, 73, 75, 78, 81, 84, 86, 88, 94, 98, 105, 109, 112, 116 f. , 119, 127, 131, 136, 140, 145, 149, 155, 160, 165, 170, 180 nominal, 19, 51, 73, 75 f. , 78, 81, 84 fn. , 86, 88, 94, 98, 105, 109, 112, 116 f. , 130 f. , 136, 140, 145, 149, 155, 160, 165, 180 f. , 184, 187, 191 see also pro-adverbial, pronoun Anderson, J. M. , 25, 34, 38 fn. , 134 fn. , 135, 138, 152 fn. Anderson, S. , 55 answer, for, 167, 169 appeal, for, 150 to, 17, 37, 149-54, 187-89, 194 appear, 2, 33, 72 into, 141 on, 114 to, 188 f. appertain, to, 153 apply, 50 , for, 51 approve, o f , 168 argue, 120, 168 about, 18, 167 f. against, 19 among, 153 for, 99, 167 with, 18, 19, 98, 153 arise, from, 181 f. out o f , 139, 183 around, 95 f. arrive, 18, 58 at, 2, 128, 133 f.

204 from, 142 f. in, 128 fn. , 133 f. as, 66, 155-58 as far as, 6 ask, 16, 168 for, 45, 59, 148 f. , 167, 171 from, 148 f. assemble, for, 145 as to, 68 fn. , 69 at, 2, 6, 18 f. , 31 f. , 41, 45, 49 f. , 57-61, 72-75, 78 f. , 82, 86-89, 93, 95-99, 119 I 122, 128-31, 133, 138 f. , 159-62, 164, 167-68 away from, 6, 17, 38, 131, 141 f. babble, about, 166 fn. Bach, E. , 1 fn. bang, at, 162 on, 162 bark, 58 Barkai, M. , 58 Bartsch, R. , 46, 80 batter, at, 162 on, 159, 162 battle, against, 185 with, 185 bear, on, 191-93 upon, 191-93 because o f , 5, 6, 31, 72, 80-82 before, 73 fn. , 74, 75 fn. , 89, 95, 162, 183 beg, for, 149 fn. , 150, 167 f. from, 39, 150 f. begin, 86, 89, 101 with, 70 behind, 58, 61, 95, 97 f. , 100 f. , 129 believe, in, 168, 172 belong, in, 128, 133 to, 39, 128, 133 f. , 151-54, 186 below, 95 bend, to, 31 beneath, 95, 128, 131 Beneficient, 28, 33, 93-96, 102, 6.1.5, 120 ff. , 148, 152 beside, 95

between, 31, 58, 74 f. , 78, 95, 134 f. , 159, .162 beyond, 129 bleat, 168 about, 88 Bloomfleld, L. , 59 boil down, to, 157 Bolinger, D. , 2, 4, 53, 96, 101 f. , 109, 117 bother, about, 167 , with, 167 borrow, from, 152 f. o f f , 152 bound, defined 16, 18, 26, 33, 55, 86, 97, 108, 122, 129, 174 f. , 179 f. , 184 break, 120 into, 141 f. out o f , 142 upon, 191 f. break down, under, 183 with, 183 Brose, B. , 50 f a , 53, 56, 58 build, 100, 102, 104 Burt, M. K. , 14 fn. buy, from, 151, 153 o f f , 152 by, 2, 32, 47-49, 66, 74 f. , 87 f. , 95, 103-5, 134 f. , 138, 159-63 by dint o f , 5 by means o f , 5, 32, 93, 102, 105, 107, 111, 119 by way o f , 136-38 calendric, 75, 99 call, 76 f. for, 39, 144 fn. on, 144 fn. , 151 upon, 150, 152 f. campaign, 147 against, 145 f. for, 89, 145 f. care, about, 41, 167, 169 for, 169 carry on, about, 166 Caivell, H. T. , 1, 3 fn. , 52, 54, 56 f. , 192 case, 2.2 passim one-instance-per-clause, 34 f. , 40, 111, 155 cater, for, 118 fn. , 120

205 cause, 34 chafe, at, 168 chance, upon, 162 change, according to, 183 from, 40, 156 into, 156 f. towards, 157 with, 183 Chapin, P. G. , 25, 38 Chomsky, N. , v, 1 fn. , 2.1 passim, 25, 27, 32, 54, 56 fn. , 83 fn. choose, between, 159, 162 Christophersen, P. , 54 circumstantial adjunct, 64 f. , 69 f. , 91 circumstantial functions, 25, 27, 46, 59 f. , 61, 72 f. , 85, 93 f. , 95, 104, 121 cleft sentence, 61, 89, 171 f. cline, 3, 62, 74, 95, 127, 180 close down, on, 192-93 cluster, defined 2 cognition, verbs of, 37, 40, 41, 164 cohesion/cohesive, 17-19, 64 f. , 93, 95, 101 f. , 106 f. , 109 f. , 114, 115, 118, 121, 128, 133, 137, 141, 143, 145 f. , 156, 158, 160, 165, 167, 169, 175, 182, 185 f. , 188 f. , 192 f. , 195 collaborate, with, 185 collapse, under, 182 with, 182 collect, for, 148, 152 collocate/collocation, 17, 66, 129, 142, 183, 189 combination, defined 2 come, 76, 89, 135 across, 162 f. after, 159, 162 before, 162 by, 2, 53, 159 ff. for, 144 from, 139 f. , 141, 170 in search o f , 145, 147 to, 140 upon, 162 come back, 50 fn. , 109, 110 for, 144 fn. , 145

in search o f , 144 fn. , 146 to, 26 come forward, with, 159 f. come to an understanding, 4 come up, 112 to, 109 Comitative, 32 f. , 40 fn. , 93 f. , 96, 102, 107 f. , 6.1.4, 116, 117 f. , 12022, 159, 184 ff. comment, on, 49 f. , 100, 167, 172 communication, verbs of, 40, 148, 153 commutability, of P, 57, 127, 137, 142 f. , 146 f. , 152 f. , 162,180, 183, 185 f . , 189, 192 f. of V, 57, 127, 137, 142 f. , 146 f. , 152 f. , 162, 180, 183, 185 f. , 189, 192 f. compare, with, 162 comparison, verbs of, 67 compete, 103-5, 107, 147 against, 185 for, 145, 147, 184, 194 with, 184 ff. , 194 complain, about, 150, 167 to, 150, 152 o f , 167 concentrate, on, 56, 167 concerning, 68 f. concessive, 31, 80 conditional, 31, 80 conjunct, 64 f. , 66, 82 consist, in, 156 ff. o f , 41, 53 f. , 126, 154, 155 fn. , 156 ff. constituency/constituent, 11 f. , 17, 19, 23, 25 f. , 50, 61, 83, 103 continue, 18, 101, 134 contract, 68 contribute, to, 148-51, 153 coo, 153 cook, 16, 29 for, 148, 152 cooperate, with, 68, 185 coordination, with adjective/adverb, 51 f. , 60, 73, 82, 85, 86, 88, 94, 126, 180

206 with Vtr, 56, 62, 95, 127, 133, 137, 142, 146, 151, 156, 161 f. , 168, 172, 180, 182, 185, 188 f. cope, with, 163 copula(tive), 40, 156 durational, 86 fn. temporal, 86 fn. Corder, S. P. , 2, 33, 39, 40 fn. correspond, to, 64, 66, 155 f. with, 70 count, for, 188 ff. on, 167 f. to, 189 cover, 102, 118 crawl, 49, 97 f. , 100 creep, through, 136 fn. under, 135 Cruse, D. A. , 25, 28, 34, 36 f. Crystal, D. , 8 curve, along, 134 cut, 49 dance, 93, 99, 113 f. Danes, F. , 25, 36 f. , 41, 67, 159 dangle, 100 Dative, 37, 39 f. , 116, 126 f. , 7.1.5, 158, 170, 173, 175, 186 f. deal, in, 29 with, 102, 163 f. decide, on, 19 decrease, 31, 115 defective verbs, 86 f. defer, to, 151 depend, on, 70, 171 upon, 159, 164 delve, into, 142 depending on, 6 derive, from, 51 despite, 31 detract, from, 152, 153 develop, into, 40, 155 ff. out o f , 155 f. towards, 155 f.

Diderichsen, P. , 15 fn. die, 76, 118 for, 183 from, 183 o f , 181 ff. with, 183 differ, from, 67 f. , 154 ff. dig, 106 Dik, S. C. , 56 fn. diminish, 89 dine, 112, 113 o f f , 162 Direction/directional, 17-19, 23, 26, 75 fn. , 76-78, 125, 127 f. , 135, 136 fn. , 7.1.3, 148, 155, 173 f. disagree, 2, 29, 167 about, 165 f. on, 165, 167 with, 148, 151, 166 disappear, 75, 111, 135 fn. from, 32 in the direction o f , 142 into, 141 on, 192 through, 137 disjunct, 14, 64 ff. diverge, from, 142 f. to, 142 divisibility, 6, 59 f. , 70, 89, 119-21, 127, 129, 134, 137, 143, 147, 153, 157 f. , 162 f. , 168 f. , 170 f. , 180, 183, 186, 189, 193 f. do, 34, 36, 46, 78, 86, 96, 103 fn. , 108, 111, 116, 118 fn. , 147, 156, 184 to, 34, 37, 100 do •so/-what, see Verb Phrase pro form down, 134, 136 draw together, for, 145 draw back, 31 dream, about, 168 o f , 167 f. drink, 106 drive, 134 by way o f , 137 past, 138 to, 137 droop, over, 128 drop in, for, 146 drowse, 98 dry, 80

207 due to, 5 Duration, 18 f. , 30, 72 f. , 5.1.2, 82, 86, 88, 90, 104 during, 47 f. , 74, 76 eat, 58 from, 28 echo questions, 51, 75, 77 ellipsis, 60, 70, 88, 119, 170 e/ope, 111, 114, 185 emerge, 110 f. , 120 for, 144 f. from, 110, 114, 142 f. into, 142 emigrate, from, 29 emotional verb, 37, 164, 189 entailment), 28 f. , 33, 34, 36, 45 f. , 64, 72, 93, 111, 118 fn. , 179 enter, 32 Epithet, 31 f. , 73, 82, 5.1.5, 87, 90, 96, 102, 108, 116 equal, 156 Equative, 40 f. , 126 f. , 139, 7.1.6, 158, 173, 190, 195 Erades, P. A. , 50 fn. , 56 escape, from, 139 f. , 142, 144 into, 142 event/event verbs, 18, 76 f. , 134, 193 exceed, 67 except for, 6 exclamative, 61 exist, 86 fn. expand, from, 157 into, 156 f. towards, 157 Experience^ 37, 53, 116, 166, 179 f. , 8.1.3, 194 f. expression, verbs of, 164 External Causer, 34 fn. , 36, 80, 107, 139, 159, 179 f. , 8.1.1, 194 f. Factitive, 34 fn. Fairclough, N. L. , 1, 25, 45, 51, 56 f. , 102 fn. fall, 88 for, 168 f. fall down, from, 140 fall in love, 3, 157 fall into arrears, with, 157 fall o f f , 69

fall over, with, 36, 182 fall short, o f , 157 f. fear, for, 169 feed, on, 159 f. , 164 fence, 103 fend, for, 120 fn. feud, with, 185 fiddle, with, 160 Fiengo, R. , 53 fight, 58, 147 about, 162 against, 145, 185 f. for, 126, 145 over, 159 ff. with, 184 ff. Fillmore, C. J. , v, 15 fn. , 22 passim, 56 fn. , 96, 102, 107 f. , 111, 116, 139, 144 finish, 78 fire, at, 160 ff. fit, into, 142 onto, 142 fix, on, 128 flinch, from, 142 f. float, 120 before, 183 in, 183 flock, to, 88 fn. flow, along, 136 flush, for, 181 flutter, in, 181 f. fly, to, 38, 96 follow, 35 for, 2, 18, 28-31, 33, 39, 41, 45, 51, 53, 58 f. , 61, 72, 76-78, 81 f. , 88, 93, 11519, 120, 126, 138 f. , 142-47, 148-50, 152, 159-64, 166-72, 181, 183, 187-90, 194 for the benefit o f , 5, 115, 117

208 for example, 66 for (the) lack o f , 82 for the sake o f , 144 forget, about, 167 form part, o f , 69 Fraser, B. , 166 fn. free, defined 16, 18, 26 f. , 30, 65, 69, 86, 88, 90, 108, 110 f. , 118, 122, 128, 172, 174, 180, 195 Frequency, 17, 19, 21, 23 f. , 30, 72 f. , 5.1.3, 82, 88, 89, 90 from, 28 f. , 32, 38-40, 51, 60, 78, 95, 106, 139-44, 148-53, 154-58, 170, 181-83 from below, 142 frown, over, 167 upon, 167 f. fumble, for, 2, 144 f. with, 159 functions, combinations of, 34 fn. , 36-38, 100, 106, 110, 114, 118, 136, 138, 180, 184 f. , 190 Gaaf, W. van der, 50 fn. , 54, 57 gain, out o f , 181 gaze, 168 after, 58 at, 61, 72, 168 for, 168 gesticulate, 107 get, 49, 76, 144 into, 135 out o f , 142 to, 105, 140 f. get away, from, 38 get hold, o f , 126, 162 ff. get in, by, 138 through, 135 get on, about, 153, 166 to, 153 get out, 66, 82 get rid, of, 163 f. get through, to, 140 giggle, over, 168

give, 33, 39, 115 to, 152 f. give offence, to, 3 give rise, to, 3 glance, 84 at, 134, 168 for, 168 glare, at, 168 for, 168 glitter, in, 181 f. go, 134, 141, 170 after, 162 before, 162 by way of, 136 for, 59, 102 fn. , 144 from, 60, 140 into, 73 fn. past, 136 round, 57 fn. through, 138 to, 33, 57 fn. , 61, 85, 103-5, 139, 140, 144 via, 136 go ahead, o f , 47 f. , 66, 159 go away, 107 f. go on, about, 166 go out, 55, 80-82 into, 104 f. go right, 116, 118 go up, 75, 87 into, 88 Goal, 11, 34 fn. , 2.2.3.2 passim, 125, 138 f. , 144, 148, 152 f. , 156, 159 gradience, see cline Greenbaum, S. , 15 fn. , 16, 31, 65 f. , 68, 82 grin, 164 groan, 152 grope, 106 Halliday, M. A. K. , v, 2.2 passim, 72, 101 fn. , 111, 113 fn. , 154 f. , 164 hang, 52 on, 128, 159, 161 hang around, 96 happen, 46, 67, 84, 96, 103 fn. , 116, 156 to, 34, 36 f. , 191-93 upon, 192 fn. happen-what, see Verb Phase pro forms

209 Hartvigson, H. H. , 15 fn. hatch out, 46-48 hate, 37 have, 39, 144 Having, 32, 93 f. , 96, 102, 6.1.3, 133 f. , 116, 120 f. , 144 head, for, 139 fn. , 142 f. towards, 143 hear, about, 150 from, 40, 150, 152 f. o f , 150, 167 help, 33 help out, with, 160 hide, 97 f. , 100 f. hint, at, 167 hold back, on, 192 hold on, to, 104 hope, for, 41, 149, 167 from, 149 Hornby, A. S. , 142 fn. , 147 fn. , 152 Huddleston, R. D. , 1 fn. , 3, 15 fn. , 2527, 33 f. , 40 fn. , 49 fn. , 53, 67, 77, 85, 87 fn. , 97, 100, 111 f. Hudson, R. A. , 13, 28 hum, 152 hunt, 147 for, 145 hurry, to, 84 idiom(aticity), 117 f. , 118 fn. , 128, 133, 137, 142 f. , 169 imperative, 47, 65-67, 73, 94, 126, 179 impinge, on, 191 f. impose, on, 37, 191-93 upon, 153, 193 in, 2, 6, 15-17, 18, 20, 26, 29, 31, 46-49, 54 f. , 57 f. , 61, 64, 66 f. , 70, 73-76, 78, 84-86, 88 fn. , 89, 95-101, 120, 122, 125, 128-133, 138, 155, 156 (fn.), 159, 161-63, 164, 167 f. , 170, 172, 174, 18183 in the absence o f , 82 in case o f , 80, 82 in the case o f , 69 in the direction o f , 142, 144

in the face o f , 82 in front o f , 3, 5, 35, 128, 133 in the interest o f , 5, 115 in lieu o f , 5 in the middle o f , 95, 98, 102, 128 in order to, 39, 144, 147 in preparation for, 144 in quest/search o f , 144, 146 f. in respect o f , 67 in spite o f , 15, 80-82, 88 in terms o f , 86 increase, 32, 67 f. , 88, 115, 118 indivisibility, see divisibility indulge, in, 167 f. inquire, 168 about, 167 inside, 95 insist, on, 82, 88, 168 Instrumental), 22 f. , 24, 26 f. , 32, 36, 76, 86, 93 f. , 96, 6.1.2, 107-11, 116, 120 f. , 135 f. , 150 fn. , 164, 166, 180 intensifies 64 f. , 68, 87 f. , 91, 157, 189, 193 intensive, 40, 55, 154 Interlocutor, 39 f. , 112, 126 f. , 138, 7.1.5, 164, 166 f. , 170, 173, 175, 184, 186 f. internal processes, verbs of, 168 interrogation/ interrogative, adjectival, 50, 61, 73, 75, 77-79, 81, 84 f. , 88 f. , 94, 99, 101 f. , 112 f. , 115, 117 f. , 126, 147 fn. adverbial, 39, 49, 51, 69, 74, 77, 79, 81, 84 f. , 87 f. , 94, 96 ff. , 103 ff. , 108 f. , 112, 116, 119, 127, 129 f. , 135, 138 ff. , 144 f. , 149, 155, 160. 165, 170, 180 f. , 184, 187, 191 clause, 47, 65, 67, 73, 94, 126, 179 nominal, 50 f. , 73, 94, 96 ff. , 101-5, 108 f. , 112, 115 f. , 118, 127, 129 f. , 135, 138 f. , 144 f. , 149, 154 f. , 160, 164 f. , 171, 180 f. , 184, 187, 190 f. , 193 f. into, 40, 138 f. , 141-43, 155-57 intrigue, 114 Jacobson, S. , 66 Jespersen, O. , 4 fn. , 33, 50 fn. , 54, 56, 58, 61, 101, 151, 161, 167 join, in, 159, 162 with, 37, 112, 184 ff. jump, out o f , 18, 141

210 keep on, 166 fn. kick, at, 159, 162 kindle, with, 182 kiss, 96 Klima, E. S. , 55 know, 17 about, 41, 167 f. o f , 167 f. Körner, R. , 50 fn. , 53 fn. , 54, 56, 57 f. , 62 fn. , 161, 167 Kruisinga, E. , 50 f a , 56 Kuroda, S.-Y. , 75, 86 Lakoff, G. , 2.1.3-4 passim, 27, 31, 35, 55, 67, 83 fn. , 103, 153, 184 land, at, 133 on, 38, 56, 76, 89, 128, 131-3 Langendoen, D. T. , 47 f. Lasnik, H. , 53 last, 18 fn. laugh, at, 167 f. for, 183 with, 183 lead, 31 to, 67, 72, 162 lean, against, 128 learn, about, 150, 167 f. from, 148, 150, 152 f. o f , 148, 150, 168 leave, 82, 170 for, 74, 75 fn. , 170 Leech, G. N. , 75 fn. , 135, 138, 140 Lees, R. B. , 53 lend, to, 152 f. lexical pro form, 67, 69, 74, 79, 82-88, 97, 130, 165 lexicalisation of function, 18, 24 fn. , 83 lie, above, 57 behind, 129 beneath, 128, 131 in, 108, 170 in the middle o f , 128 on, 132 under, 131 upon, 129 lie down, 128 like, 37, 62, 84 f. , 155-58, 171

listen, 32 to, 99 live, 4, 86, 89, 170 above, 86 among, 131, 134, 171 in, 6, 16, 128 f. , 131-33 o f f , 162 on, 159, 162 on top o f , 134 load, 34 fn. Location/Locative, 15 fn. , 16, 20 f. , 23 f. , 26, 76, 105, 128, 159 f. , 164, 190 bound, 18 f. , 21 fn. , 32, 38, 41, 56, 98 fn. , 100, 127, 7.1.1, 136, 154, 173 f. free, 19, 21 fn. , 27, 32, 93 f. , 95, 6.1.1, 106, 109 f. , 113, 115 f. , 120-22, 125, 128 f. , 130, 132, 135 f. , 138 long, for, 167 ff. Long, R. B. , 54 fn. look, 64 after, 58, 62, 161 ff. , 172 ahead o f , 61 at, 53, 57 f. , 59, 61, 69, 81 fn. , 85 f. , 134, 167 f. behind, 61, 98 between, 58 for, 58, 81 fn. , 167 f. , 172 into, 141 like, 62, 155 ff. , 171 to, 58, 188 f. up to, 19 look forward, to, 168 look out, for, 59 look up, 4, 5 lose confidence, in, 168 lose sight, o f , 165 f. love, 37 Lyons, J. , v, 74, 86 f a make, 28, 36, 142, 181 after, 143 for, 31, 143, 162 towards, 143 make efforts, 70 make mistakes, in, 162 make progress, 105 make sense, o f , 168

211 make up, 155 f. Manner, 15 fn. , 17-19, 21-23, 28, 31 f. , 50 fn. , 73, 81 fn. , 83 f. , 5.1.6, 87, 90, 104 march, towards, 140 Marchand, H. , 54 matter, for, 187-89 to, 188-90 Means, see Instrument Measurement, 32, 73, 5.1.7, 90, 104 meet, 97 Mitchell, T. F. , 1, 3, 57, 59 model, for, 145, 147 momentary verb, 18 move, 55, 122 away from, 142 from, 140 in the direction o f , 144 to, 140 via, 125 move on, 68 movement, verbs of, 38, 134 Mulholland, J. , 5, 95 fn. multi-word preposition, 5 verb, 3, 120 fn. , 157, 169 fn. mumble, 164 murder, for, 144 mutter, 152 near, 58, 95 negation, 15, 21 f. , 48, 65, 67, 73, 79, 85 f. , 94, 126, 179 f. Neutral, 2.2.3.2 passim, 125, 144, 152, 158 Nilsen, D. L. F. , 88 fn. , 111 nominalisation of V, see sentence, (pro)nominalisation of of V-P, 62, 172 non-role playing elements, 30, 59, 64, 87, 90 f. , 121, 157, 172, 174, 195 Noreen, A. , 80 noun, abstract, 13, 30, 32, 72, 175 animate, 13, 27, 33, 37, 174 concrete, 129 countable, 104 human/personal, 39, 112, 116, 121 f. , 148 f. , 175, 184, 186 f. locational, 139 f. proper, 99, 112, 117 temporal, 74, 76-78

object, to, 62, 167, 172 object raising, 53, 61, 94, 100, 106, 109 f. , 113 f. , 117 f. , 127, 131, 136, 140, 145, 149, 155, 160, 165, 171, 180, 182, 184, 188, 191 Objective, 34 fn. , 37, 39-41, 107, 110, 112, 114, 121, 125 ff. , 139, 146, 148, 15052, 155, 7.1.7, 164, 169, 173, 184 ff. , 190 ff. , 195 intrusive, 108, 144, 192 fn. observation, verbs of, 134, 148, 167 f. occur, 46, 79, 86 fn. , 119 in, 154 to, 179, 187-90 with, 66 o f , 29, 32, 41, 47 f. , 53 f. , 61, 126, 138, 154, 156-58, 159-64, 165-70, 181-83 o f f , 4, 152, 162 on, 2, 13, 16, 19, 28, 30, 37, 49, 50 fn. , 52, 54, 56, 64, 66, 72, 74-76, 78 f. , 82, 95-100, 102-7, 122, 126, 128-34, 138, 141, 151, 159-62, 164 f. , 167 f. , 171 f. , 179, 190-93 on behalf o f , 33, 117-19 on top o f , 15, 129, 131, 134 onto, 142 open, 93, 98 operate, 73, 85 on, 192 out o f , 5 f. , 18, 104, 139, 141-43, 155 f. , 181, 183 outside, 58, 95, 100 over, 39, 88, 95, 119, 128 f. , 134, 159-62, 166 ff. owing to, 5 f. , 29 paint, 2, 106 Palmer, F. , 54 Palmer, H. E. , 54 Partee, B. H. , 35 participant, 25, 27, defined 29, 30, 33, 36, 38, 46, 59 f. , 72, 88 fn. , 93, 125 ff. , 180, 194 participate, in, 162 particle, adverbial, 4, 96 pass, as, 157 passive, 1, 19, 45 f. , 53 f. , 58, 61, 94 f. , 100 f. , 106, 110, 114, 118, 127, 131 f. , 136, 141, 145, 149 f. , 154 ff. , 160, 165 f. , 171, 180, 182, 185, 188, 191 past, 134, 136, 138, 171 Path, 38, 86, 125, 127, 7.1.2, 138, 141, 144, 173 f.

212 pattern, with, 112, 159 pause, 31, 61, 72, 78 on, 171 pay, 15 for, 2, 28 f. , 159 ff. , 171 pay attention, to, 166 peer, 48 at, 134 in, 85 perception, verbs of, 37, 41 persevere, with, 53 persist, in, 162 Peters, S. , 35, 67, 153, 184 Phenomenon, 28, 39 f. , 56, 121, 126 f. , 139, 146, 148, 150 f. , 153, 155, 158, 7.1.8, 173, 186, 188, 190, 193, 195 phrasal verb, 3, 6, 155, 166 picnic, 102 place, see Location play, 41, 98 fn. , 114, 117, 120 for, 162 f. , 171 plead, with, 152-54 please, 37 point, 107 at, 104 f. point of reference, 135 absolute, 74, 75 fn. relative, 76, 78, 98 pore, overs 166 position of N, 62, 172, 188 position of P-N, initial, 52, 60, 70, 73, 88, 94, 100, 106, 109, 113, 117, 119, 127, 129, 131, 136, 140, 145, 149, 155, 160, 165, 170, 180, 182, 184, 188, 191, 194 mid, 52 f. , 60, 65 f. , 70, 73, 88, 94, 99, 105 f. , 109, 113, 117, 119, 126, 180, 182, 184, 188, 191 position, verbs of, 38 Postal, P. , 45, 53, 187 Potter, S. , 168 fn. Poutsma, H. , 54 fn. , 57 fn. , 62 fn. precede, 35 predicate pro form, see Verb Phrase pro form preposition stranding, 50, 61, 101, 106 f. , 110 f. , 113-15, 118 f. , 134, 143 f. , 147, 154, 158, 163 f. , 169, 183, 186, 189 f. , 193 f.

prepositional verb, 1, 25, 28 f. , 38, 40, 45 f. , 49, 52, 54 f. , 57, 59, 61 f. , 72, 95, 100, 133, 148, 166, 183, 188 press, for, 53, 88, 162, 164 prevail, on, 151 upon, 153 pro-adverbial, 60 complex, 103, 105, 108, 116, 135 hybrid, 60, 75, 77 f. , 87, 89, 136 fn. , 139, 170 indefinite, 119, 170 see also anaphora, interrogative, pseudocleft, relative proceed, with, 160 pronominalisation, of N, see anaphora, interrogative, pronoun, pseudo-cleft, relative of V, see sentence, (pro)nominalisation of pronoun, 121, 126 indefinite, 61 personal, 61, 76, 89, 120, 143, 171, 194 reflexive, 61, 120, 171 see also pronominalisation of N pronounce, on, 168 pseudo-cleft, 55 adverbial, 48 f. , 74, 77, 79, 81, 83 f. , 85, 87 f. , 94, 96 ff. , 103 ff. , 108 f. , 112, 116, 127, 129 f. , 135, 139, 144 f. , 149, 155, 160, 165, 170, 180 f. , 184, 187, 191 nominal, 3, 50, 61, 73, 94, 96 ff. , 103 ff. , 108 f. , 112, 116, 127, 129 f. , 136, 139, 144 f. , 149, 155, 160, 165, 171, 180 f. , 184, 187, 191 psychological verbs, 40, 148, 186 pull, at, 159, 162 pull in, 99 pull up, 6 Purpose, 38 f. , 126 f. , 138, 7.1.4, 150, 159, 173 purr, 152 push, in, 138 put up, with, 167 quail, at, 168 qualify, as, 156 f.

213 quarrel, about, 150 among, 153 with, 150, 153 quarter-stave, 102 f. Quirk, R. , 1, 3, 5, 8, 17, 31, 49 fn. , 52, 61, 66-68, 73 fn. , 81 fn. , 83, 85 f. , 95, 155 race, against, 186 rain, on, 141 rasp, at, 102 f. , 121, 162 reach, for, 138, 145 f. into, 138 reaction, verbs of, 41 read, like, 157 Reason, 21, 23, 28, 31, 39, 72 f. , 5.1.4, 85, 90 rebel, against, 75 fn. recover, 78 refer, as, 155 to, 57, 155, 167 f. , 171, 174 reflect, on, 165, 168 regardless o f , 80 rejoice, in, 168 relate, 68 to, 162 ff. relative, adverbial, 48 f. , 73 f. , 76-78, 81, 83, 85, 87, 94, 96 ff. , 103 ff. , 108 f. , 112, 116, 119, 127, 129 f. , 135, 129, 144 f. , 149, 155, 160, 165, 170, 180 f. , 184, 187, 191 nominal, 49, 61, 66-70, 72-74, 87, 89, 94, 96 ff. , 101, 103 ff. , 108 f. , 112, 114, 116, 118, 120, 122, 126, 135, 139, 144 f. , 149, 154 f. , 158, 160, 165, 169, 171, 180 f. , 184, 187, 190 f. , 193 f. rely, on, 167 remain, 16, 64 in, 125, 128, 132 report, 153 on, 168 to, 40, 151 f.

resemble, 35, 67, 156 resign, from, 142 resort, to, 82 respect, 67 respond, to, 168 rest, in, 132 on, 128 under, 133 result, in, 159 ff. retreat, 55 return, 74, 110 for, 144 ff. in search o f , 144 f. ride, 6, 102 f. , 105-7 along, 135 f. rhapsodise, 168 ring, 153, 168 about, 61 riot, 147 against, 145-47 for, 145 rip, at, 162 rise, 32, 60, 64 f. , 73, 87 f. above, 125, 139 f. , 156 from, 140 out o f , 139 to, 137, 141 via, 137 roar, 164 from, 141 rob, 39 Robinson, J. , 1 fn. , 27 role-playing elements, 30, 46, 72, 93 roll, with, 183 Ross, J. R. , 55 round (P), 57 fn. , 95, 99, 134 f. round (V), on, 192 f. run, 5, 16, 76 f. , 96, 134, after, 19, 58, 142 along, 134 from, 112 fn. to, 112 fn. up, 136 run o f f , 113, 115 run out, 3 o f , 159 f. , 163

214 run short, o f , 163 sail, to, 19 Sandved, A. O. , 54 say good-bye, to, 151 f. Scheurweghs, G. P. , 54 Schibsbye, K. , 54, 138 fn. , 168 fn. , 169, 183, 185 f. , 189, 192 fn. Schreibei, P. A. , 31, 65 see, through, 136 seem, 156 to, 187-89, 194 selection, 13, 27, 32, 107, 111 sell, 88 fn. to, 149 f. , 153 sentence adverbial, 14 fn. , 15 sentence, (pro)nominalisation of, 55 f. , 127, 132 f. , 137, 141 f. , 146, 150 f. , 156, 161, 167, 170, 180, 182, 185, 188, 192 separate, 96 set, about, 159 ff. set o f f , down, 88 for, 139 fn. shake hands, with, 179, 185 share, in, 162 shoot, 105 at, 27, 160 f. into, 139 to, 138 up, 136 shout, 61, 73, 84 about, 164, 167 at, 164 for, 183 to, 134 with, 183 side, with, 185 fn. sigh, for, 167 sign, for, 144 sin, against, 161 since, 77, 89 sing, 87, 117 f.

sing for, 144 o f , 168 sit, 16, 18 f. , 28, 54, 98, 130, 170 at, 131 in, 16, 18, 98, 129, 132 on, 19, 28, 54, 129, 131-33 under, 128 sit down, for, 144 f. sleep, 54, 58, 100 slip away, 135 fn. down, 136 smell, o f , 160 f. smile, 52 at, 41, 164 Smith, C. S. , 158 smoulder, with, 181 snap, at, 162 sniff, at, 159 f. sound, 31 like, 155 Source, 11, 2.2.3.2 passim, 125, 139, 144, 148, 152 f. , 156, 159 speak, 30, 85, 117, 119 f. , 153, 168 about, 170 at, 164 o f , 1, 89, 167 to, 79, 126, 134, 149, 151, 164, 170 spend, on, 13 spit, at, 139 towards, 139 split, into, 70 squat, 98 squeeze, past, 171, stagger, under, 181 stale, on, 192 stand, 77, 112 as, 156 f. in, 130 in front o f , 128, 130, 133 to the left o f , 134 under, 132 stand on one's head, for, 144

215 stand in, 118 stare, across, 136 at, 45, 139, 168 for, 168 into, 139 out o f , 6 start, with, 97 state, verbs of, 76 f. , 168 stative verbs, 17, 95 stay, 16 77 f. at, 80 f. , 128 away from, 131 in, 133 steal, from, 148, 150, 153 step, across, 137 stereotype, see idiom stink, o f , 159, 161 stoop, for, 146 stop, 50 fn. Strang, B. M. H. , 67 fn. stretch, across, 134 towards, 141 strike, against, 145 for, 144-46, 170 strive, after, 145-47 struggle, with, 58, 184 f. study, 41 for, 145 subcategorisation, 13, 17 f. , 27 subject adjunct, 31, defined 55, 102, 108, 135, 144 fn. subscribe, for, 145 to, 167 f. subsist, on, 162 substitute, 118 succeed, in, 103, 162 suffer, for, 126, 145, 147 Svartvik, J. , 1, 3 fn. , 52, 54, 56 f. , 192 sweep, 4 swell up,

with, 194 swim, to, 141 swing, 85 symmetric predicates, 35, 40 fn. , 67, 153, 184 take, from, 152 o f f , 152 to, 54, 168 ff. take account, o f , 32, 93, 119, 168 take action, 93, 113, 115, 119 take care, o f , 161 take place, 46, 86 fn. , 99 talk, 17, 65, 85, 153, 167 about, 18, 46, 50, 150 f. , 166, 168 o f , 167 to, 134, 148-52, 154, 166, 170 f. with, 152 f. taste, to, 188 teach, 96, 100, 101 tear, at, 162 telephone, 153 about, 150 to, 150 tell, against, 191-93 on, 192 temporal, see Time theme, 35, 40 fn. , 47, 113, 117 think, about, 168 f. o f , 29, 61, 70, 120, 167 ff. Thomas, O. , 54 throb, 6, 17 through, 103 fn. , 104 f. , 107, 134 f. , 136-38 Time, 14, 15 fn. , 18 f. , 23 f. , 26, 30, 50 fn. , 72, 5.1.1, 77 f. , 82, 85 f. , 90, 96-98, 121 to, 15, 17, 19, 26, 28 f. , 33, 36-40, 46-48, 54, 57 f. , 61 f. , 64 f. , 68, 70, 126, 128 f. , 133 f. , 138-44, 148-54, 155-58, 162 f. , 164, 166-72, 174, 186-90, 191-94 to above, 142 to the left/right o f , 3, 35, 95, 101, 110, 134 toil, after, 145 f.

216 walk, 57, 107, 108 f. , 134 f. away from, 141 in the direction o f , 144 over, 39 to, 82 up, 137 walk out, on, 28, 179, 192 Walmsley, J. B. , 33, 111 watch, for, 167 wave, before, 183 in, 183 weep, 16 whine, at, 168 whisper, 153 to, 61, 152 win, 17, 20 under, 86, 89, 95, 97, 102, 128 f. , 131, wish, 133, 135, 181-83 for, 167 ff. underneath, 95 with, 2, 18 f. , 27 f. , 31-33, 34 fn. , 35-37, until, 76 f. 40, 48, 52 f. , 55, 58, 78, 83-85, 87, 93, up, 4 f. , 95, 134 f. , 136 f. 95, 102 f. , 104-7, 108-11, 111-15, 119, up to, 19, 86 121, 148, 150-53, 157, 159-60, 162-64, upon, 95, 129, 150, 152 f. , 159, 162, 164, 165-67, 179, 181-86, 188-90, 194 f. 167 f. , 191-93 with a view to, 144 use, 22, 24 fn. , 27, 32, 102, 135 wither, 82 vary, 68, 87 within, 69, 78, 95 according to, 182 ff. without, 15, 32, 88, 108-11 from, 87 work, 16, 30, 53, 97, 113-16, 118, 120 to, 87 against, 145 with, 182 f. at, 19 Vasiliu, L. , 1 fn.. for, 145 Verb Phrase Complement, 15 f. , 18 f. worry, Verb Phrase pro form, 16, 18-20, 26, 28 f. , about, 165 54, 60, 70, 83, 88, 94, 98, 119, 127, wrestle, 132, 136 f. , 141, 146, 150, 156, 160 f. , about, 162 166, 180 over, 162 Verbal Complement, 15 f. , 18 f. write, 33, 85, 89, 153, 168 verbalisation, verbs of, 41, 134 about, 168 Vestergaard, T. , 6, 34 fn. , 38 fn. , 95 fn. , concerning, 169 135, 139 fn. , 142 fn. , 185 fn. to, 83, 151, 169 via, 125, 134, 136-38 viewpoint adjunct, 64 f. , 66 yield, Volbeda, R. , 50 fn. , 54 fn. , 58 to, 151 volition, 85, 144 f. Vondrak, W. , 135 fn. towards, 5, 139-43, 156 f. trail, over, 83 transitivity system, 27 travel, 110, 119, 134 by way o f , 138 to, 142 via, 136, 138 tread, on, 190 tremble, with, 179, 181 try, for, 159, 162 turn, 99 into, 139 turn up, 77 twinkle, with, 159

wait, 16, 18, 100 for, 28

Zandvoort, R. W. , 2, 54 Zsilka, J. , 38 fn.