The Prepositional Passive in English: A semantic-syntactic analysis, with a lexicon of prepositional verbs 3-484-10363-9

Table of contents :
Foreword......Page 11
1 Traditional definition......Page 13
2 Survey of past treatment......Page 17
3 Present analysis......Page 31
1 Procedure......Page 36
2 Subject notions......Page 38
3 Subject notion and object eligibility......Page 51
1 Procedure......Page 55
2 Object notions......Page 56
3 Object notion and subject eligibility......Page 96
1 Semantic-syntactic configuration......Page 102
2 Lexical choice......Page 109
3 Contextualization......Page 117
4 Acceptability in practice......Page 126
1 Towards a grammar of the PPP......Page 128
2 The PPP and the passive proper......Page 130
3 The PPP and the active......Page 131
4 The PPP and linguistic theory......Page 135
Appendix: Corpus correlation between surface preposition and semantic function......Page 138
A Lexicon of Prepositional Verbs......Page 141
References......Page 216

Citation preview

Linguistische Arbeiten

81

Herausgegeben von Herbert E. Brekle, Hans Jürgen Heringer, Christian Rohrer, Heinz Vater und Otmar Werner

Elizabeth Couper-Kuhlen

The Prepositional Passive in English A semantic-syntactic analysis, with a lexicon of prepositional verbs

Max Niemeyer Verlag Tübingen 1979

CIP-Kurztitelaufnahme der Deutschen Bibliothek Couper-Kuhlen, Elizabeth: The prepositional passive in English : a semant.-syntact. analysis, with a lexicon of prepositional verbs / Elizabeth Couper-Kuhlen. — Tübingen : Niemeyer, 1979. (Linguistische Arbeiten ; 8l) ISBN 3-484-10363-9

ISBN 3-484-10363-9

ISSN 0344-6727

© Max Niemeyer Verlag Tübingen 1979 Alle Rechte vorbehalten. Ohne ausdrückliche Genehmigung des Verlages ist es auch nicht gestattet, dieses Buch oder Teile daraus auf photomechanischem Wege zu vervielfältigen. Printed in Germany. Druck: fotokop wilhelm weihert KG, Darmstadt.

ACKNCWI£DGEMENTS

The present study is a revised version of my doctoral dissertation, presented to the Philosophische Fakultät III of the university of Freiburg i. Br. in June 1977. I am indebted to Ann Kogerson, Elizabeth Roone and Valerie Rousse for serving as informants/ as well as to H. Vater for valuable comnents on the original version. Above all, however, my thanks go to A. Schöpf, who persuaded me to tackle the problem and provided unquestioning support during the course of my research. Without his help and that of family, friends and colleagues, this work would not have been possible. Frankfurt April 1979

E. C.-K.

CONTENTS

Foreword I

xi

Approaches to the PPP - Past and Present 1 Traditional definition

1

2 Survey of past treatment 2 .1 Traditional grammar

5 5

2.2 Modern linguistic theories 2.3 Conclusions

3 Present analysis 3.1 Corpus 3.2 Types of constraint 3.2.1 Semantic-syntactic 3 . 2 . 2 Lexical, contextual 3.3 Empiricism vs theory

II

19 19 2O 2O 22 23

The Semantics of Subject 1 Procedure

24

2 Subject notions

26

2.1 2.2 2.3 2.4 2.5

III

11 18

Agentive Experiencer Instrument Location Objective/Patient

26 31 34 35 36

3 Subject notion and object eligibility

39

The Semantics of Object 1 Procedure 2 Object notions

43 44

2.1 2.2 2.3 2.4 2.5 2.6 2.7

Agentive Benefactive Comitative Direction Experiencer Extent Goal 2 . 7 . 1 Basic Goal 2.7.2 Goal as ending-point 2 . 7 . 3 Goal as target 2 . 7 . 4 Goal as direction 2 . 7 . 5 Goal as objective 2.8 Instrument 2.9 Location 2.10 Manner 2.11 Path 2.12 Objective/Patient 2.12.1 Concrete Patient 2.12.2 Metaphorical Patient and Objective

44 44 45 47 48 48 49 SO 55 56 57 58 59 61 72 73 74 74 76

viii

2.13 2.14 2.15 2.16 2.17

Reason Replacive Source Stimulus Time

3 Object notion and subject eligibility IV PPP Acceptability 1 Semantic-syntactic configuration 2 Lexical choice 2.1 2.2 2.3 2.4

Idiom Infrequent use Metaphorical use Complex structure

3 Contextualization 3.1 Tense, modality, aspect 3.2 Information focus

4 Acceptability in practice

78 79 8O 82 83

84 9o 97 98 1OO 1OO 1O2

1O5 1O5 11O

114

V Granmar and Theory of the ?PP 1 Towards a grartmar of the PPP 2 The PPP and the passive proper 3 The PPP and the active 4 The PPP and linguistic theory

116 118 119 123

Appendix: Corpus correlation between surface preposition and semantic function

126

A Lexicon of Prepositional Verbs References

129 204

ABBREVIATIONS, SYMBOLS AND TYPOGRAPHICAL CONVENTIONS

PPP

prepositional passive

NP VP Aux V

noun phrase verb phrase auxiliary verb

K Ag Bas Goal Ben Com Dir E-p Goal Exp Ext Instr Loc Man Obj Pat Reas Rep Sour Stim Targ Goal Vol

case Agentive Basic Goal Benefactive Comitative Direction Ending-point Goal Experiencer Extent Instrument Location Manner Objective Patient Reason Replacive Source Stimulus Target Goal Volitive

A.E. B.E.

American English British English

U Fr Ger Lat

upper-class usage French German Latin

G N

given new

p np

passivizable non-passivizable

* ?

unacceptable of doubtful acceptability, or informant unable to decide stressed word equivalent in meaning to active example passive example

= ( ( ( ,

) )' a) , , . J

,

" . }

, . ..

}

paraphrases different contextualizations alternate readings

Examples are numbered anew in each chapter. Cross-references can be found in the lexicon.

FOREWORD

The discovery of the so-called prepositional passive in English is an occasion of surprise and delight for most foreign students of English. The majority are unaware that such a granmatically unorthodox structure is not only tolerated but used freely by native speakers of English. In fact, the prepositional passive lends a good deal of relish to ordinary speech. Take, for example, the added effect produced by passivizations such as: How many times have I told you that I don't like being spat at? Near the cart was a cow waiting patiently to be rained on This lamppost is smelled at by all the dogs who go by

The foreign learner's delight at the creative power behind this structure, however, may soon give way to despair as the veritable jungle of seemingly contradictory facts (for, need it be added, not all combinations of verb + preposition + noun allow the passive), and at the quasi-total lack of rules for use of the structure. Consider, for instance, the following fron the standpoint of an unsuspecting foreign learner: Someone slept in this bed Someone slept near this bed

Who but the happy few (or native speakers) would guess that This bed was slept in is acceptable English, whereas This bed was slept near is at best a highly suspect, if not completely unacceptable sentence in the English language? The present study is intended to fill the gap in English grammar which the prepositional passive has formed until now. It provides the foreign student or teacher of English (as well as other interested linguists) with a more theoretical survey of the aspects involved in the prepositional passive and ipso facto with guidelines for its use, in addition to containing a rather complete survey of prepositional verbs in English, illustrated in sample contexts and classified as basically passivizable or not. Traditional grammarians, to the extent that they were aware of the phenomenon, described the prepositional passive in terms of 'cohesion1 — a rather hazy concept when applied to grammar, as there are no independent, reliable criteria

xii

for determining its presence or absence. More modern approaches to grammar, however, hardly represent an improvement over the traditional cohesionists. Neither transformational generative grammar nor case grammar has so far devised a set of rules which could adequately explain the phenomenon. This is partly due to the fact that the necessary fieldwork, close study of a corpus of material, has been neglected until now. But if all previous approaches have shared in a common lack of knowledge, they have also shared in a cotmon methodological shortcoming: the prepositional passive has been viewed in terms of one (usually syntactic) dimension. Unidimensionality is undoubtedly admirable, indeed desirable in many areas of scientific investigation. It is doubtful, however, whether it has any real application with regard to the description of natural language. The present study does not provide a formalized answer to the problem of the prepositional passive. Instead, it contains a detailed analysis of the phenomenon, based on an extensive corpus of data, in the light of modern linguistic theory. This is a first necessary step in the direction of developing a formalized system for the production and comprehension of English sentences. In fact, the results may even contribute to determining exactly what kind of system must be developed in order to account for the empirical data.

APPROACHES TO THE PPP - PAST AND PRESENT

1

Traditionell definition

1.1 The first question to which we must address ourselves is the one inevitably asked by laymen and all but a handful of professionals: What is the prepositional passive? According to a traditional handbook of English grartmar (Scheurweghs 1959:16): Some verbs are used with a preposition and a noun in such a way that they resemble a transitive verb with its direct object; the noun is called the prepositional object,.. The prepositional object can become the subject of a verb in the passive to which the preposition remains attached.

Scheurweghs1 examples include: Alcuin's counsel was acted upon immediately3 A theatrical licence was not applied for. . .Any allegation against me can be disposed of, Two soldiers can be fired on in the darkness, etc. 1.2 This description of the prepositional passive (henceforth PPP) is typical of that given by most traditional graninarians. But in fact it makes a number of tacit assumptions which are by no means proven fact: a) The PP? is a deviant of the passive proper. Thus, so this view of the PPP goes, many (transitive) verbs combine with nouns directly and their direct objects can become the subjects of passive sentences: . Verb T

-f

However, some (intransitive) verbs combine with a preposition and a noun and these prepositional objects can also become the subjects of passive sentences: »

Verb +

Prep

b) Both the PPP and the passive proper are, according to this definition, purely syntactic (indeed, surface syntactic) phenomena. The terms verb, noun, preposition refer to surface syntactic categories and the passive is described as an operation which modifies the position of these categories in surface syntactic structure.

c) The PPP,

as well as the passive proper, is the converse of the active.

That is, it can be derived simply by interchanging the nouns in the active. This definition assumes that every passive sentence has an active equivalent; priority is thus accorded the active. 1.3 These assumptions, however, appear overhasty, if not unjustified in light of the fact that: a) The deviancy claim makes two further assumptions, namely: (i) the PPP differs from the passive proper only in the presence of the preposition, and (ii) the PPP is subject to the same constraints as the passive proper. However, (i) is based on superficial and imperfect knowledge of the phenomenon, and (ii) is vacuous, since the constraints operant on the passive itself are scarcely known. In fact, assumption (ii) can be readily disproved by attempting to passivize all occurrences of verb + preposition + noun which do not disqualify due to known constraints on the passive proper. In contrast to the small set of "transitive1 verb + noun combinations which do not passivize, there are any number of verb + preposition + noun combinations which do not allow the noun to become the subject of a passive sentence. b) As even Jespersen (1924:165) was well aware, it is possible to distinguish a "syntactic" and a "notional" passive: Whether a verb is syntactically active or passive depends on its form alone; but the same idea may be expressed sometimes by an active, sometimes by a passive form.

That is, Jespersen continues, there are syntactic passives which are also notional passives, such as Jill is loved by Jack. However, some syntactic actives may be notionally passive, such as fhe book sells well. Jespersen does not elaborate on what criteria determine notional passivity. His primary concern is syntactic, as evidenced by the fact that he ultimately restricts the term passive to a syntactic configuration [calling other candidates "instances of notional passive unexpressed in form" (1924:166)]. However, one of the most important insights to come from recent developments in linguistic theory is that semantic considerations cannot be categorically excluded at a syntactic level without loss of descriptive adequacy. There have recently been a number of proposals which make a systematic attempt to incorporate sernan-

tic notions into the treatment of the passive.

Regardless of the

merits or demerits of these various theories, they all reflect the realization that many linguistic phenomena can only be insufficiently explained on syntactic grounds alone. Indeed, the PPP itself is precisely one of these phenomena. If we apply the syntactic criterion of 'transitivity1 (or presence of a direct object) in order to account for passivization, we must deny its existence altogether. And if we acknowledge its existence as an exception, we cannot account for the fact that only some verb + preposition -I- noun combinations can be passivized, nor can we account for which ones passivize and which ones do not.

c) A certain amount of evidence has emerged in the recent past which suggests that viewing the passive as the converse of the active does not do full justice to the linguistic data. According to Langacker/Mmro fl 975:798), for instance, passives are in at least two non-related languages basically impersonal constructions, i. e. they express facts about an object but do not necessarily relate these facts to an agent (1975:821). Even in English, it is doubtful whether all so-called 'truncated1 passives, i. e. those without a by-phrase, can be adequately derived from active sentences with unspecified agents (cf Freidin 1975: 386f, Hainan 1976:2Of). 2

1

Cholodovic and Mel'cuk (cf Pilch 1975:lO4f) extend the term passive to all configurations of patient in subject position and agent in object position (as well as to those in which one or the other is lacking), thus including mcthinks and This house is now renting as well as the usual Jack was killed by John (see also Helbig 1 9 7 5 : 2 7 1 f f ) . In Fillmore's case grammar (1968b,1971a) the passive is associated with the selection of a certain deep case as surface subject. Gary and Keenan (1977:95ff) propose a similar approach to the passive within the framework of relational grammar.

2

Cf also Bresnan 1978:2Of, who proposes to represent the passive as a oneplace quantified relation of the form 3 χ (χ eats y).

1.4

These considerations have influenced the present study in a number of

ways. We too shall define the PPP syntactically as a construction of the form: NP +

can must will 0

be get

,3 en + Prep (-toy NP)'

•

However, our study will concentrate especially on the semantics of this syntactic construction. Furthermore, we have made a conscious effort during the course of the investigation to treat questions of PPP relation to the passive proper and to the active as unresolved problems. Once the PPP is better understood, we can turn our attention to sane of the more basic issues involved, e. g.: a) How does the PPP relate to the passive proper? Do the restrictions known to operate on the passive proper also apply to it? What additional constraints apply? b) How does the PPP relate to the active? Should it be considered a derivative of the active? What grounds do we have for treating it as a converse of the active? c) What consequences does a theory of the PPP have for modern linguistic theory in general?

3

Although the i>y-phrase is optional, its potential occurrence has the effect of excluding so-called 'statal passives' from this definition of the PPP. Jespersen (1931:98) illustrates these with His bills are paid ('so he owes nothing n o w ' ) vs His bills are paid ('regularly every month 1 ). Only the latter 'actional' type will be considered here.

2

Survey of past treatment

The problems connected with the PPP can best be demonstrated by a critical analysis of how the phenomenon has been handled in gramnar until now. 2.1

Traditional gramnar

2.1.1

O. Jespersen (1928:273) advocates a distinction between the "adverbial"

and the "prepositional" nature of the particle in combinations of the form verb + particle + object. For example, in I couldn't get in a word, in can be shown to be an "adverb", whereas in I couldn't get in the box, in is a "preposition". According to Jespersen, the distinction is one of "sound... and... meaning". "Sound" undoubtedly refers to the fact that get in in the former example would

4 be stressed get in, whereas in the latter the stress would be get in. How "meaning" is to serve as a criterion, however, is somewhat less clear, - nor does Jespersen elucidate the matter more than to say "The meaning is undoubtedly adverbial in..." (1928:274). In cases where "meaning" is not decisive, he continues, word order is to serve as a criterion. Thus: ( 1 ) a. The river b. The river c.*The river d.*The river (2)

a. I b.*I c. I d. I

cannot cannot cannot cannot

passes passes passes passes pass pass pass pass

by a small village by it a small village by it by

by the matter by it without the matter by it by without

without making a protest ... without ... ...

The particle by in (2) is an "adverb", as it can occur after the object (and indeed must, if the object is a pronoun). In (1) by is a "prepositional" particle. It cannot be placed after the object (even if the latter is a pronoun) and is "felt as more closely connected with the preceding verb than with the following object" (1928:276). Regardless of whether one accepts the sonewhat intuitive criterion of "meaning" (actually word order is a much surer test), Jespersen's distinction between "preposition" and "adverb" is a basic one, which is upheld in most subsequent grammars (albeit under different terminology: prepositional verb or verb + preposition vs phrasal verb). Jespersen's treatment of this type of structure with regard to passivization is, The difference is especially noticeable in the passive: Wot a word could be got In The box could not be got in For a more detailed description, see Strang (1962, 1969:178).

6

however, somewhat less satisfactory. The PPP construction, he asserts, may be called a "correlate" of the prepositional particle construction mentioned above, as "in both we see that the particle has greater cohesion with the verb than with what (in the active) is the object either of the particle alone (preposition) or of the whole phrase" (1928:313). This statement is confusing at the least, if not misleading. The temptation is great on the part of the reader to expect a correlation between prepositional particle and the possibility of passivization, as in both constructions there is, according to uespersen, a high degree of cohesion between verb and particle. This is, however, not necessarily the case, as the following passivization test demonstrates: ( 1 ) ' * A small village is passed by ... ( 2 ) ' The matter cannot be passed by

...

In (1)'i>z/ would be classified as a "preposition" and yet a passive is not possible, whereas in (2)' by is an "adverb" and yet passivization is acceptable. Although it nay be argued that uespersen never intended to assert that there is such a correlation, the £act is that his statements, because they are cursory and vague, do not clearly exclude this interpretation. In short, in dealing with the PPP it is essential to remember that there are two separate and independent issues involved: (a) the distinction between phrasal and prepositional verb, and (b) the possibility of a passive construction. The latter is of interest only for prepositional verbs, as all phrasal verbs, provided they occur with an object at all, can normally be passivized: (3) He looked the word up ( 3 ) ' The word was looked up ... (4) He took the picture down ( 4 ) ' The picture was taken down ... (5) He turned the light out ( 5 ) ' The light was turned out ...

2.1.2 H. Poutsma (1926:31ff) gives a more detailed and in many ways more satisfactory treatment of the subject. Although some of his explanations may prove to be insufficient on closer look, modern linguistic research on the passive has in fact hardly advanced beyond a reformulation of many of his observations. Like Jespersen, Poutsma also distinguishes between adverb and preposition, with the important difference, however, that for Poutsma "prepositional object" and "adverbial adjunct" are subcategories of the structure verb + immovable

particle + noun. Thus, Poutsma separates phrasal verbs from prepositional verbs at the outset. The existence of two subcategories for the latter makes clear the fact that only some and not all prepositional verbs allow passivization. For Poutsma, a prepositional object (as opposed to an adverbial 'adjunct') is characterized by the following three criteria: (i)

(ii) (iii)

the modifier is felt to be a "necessary complement" to the verb, e. g. He looked at the sky the preposition is "distinctly vague in meaning", e. g. This does not tally with your former statement the preposition is "intimately connected with the governing verb", forming a kind of unit, as can be seen from simplex equivalents in the same or in some cognate language, e. g. to impose upon a person = 'to deceive a person'

In contrast, there is, according to Poutsma, little coalescing between an adverbial 'adjunct' and the verb: (6) (7)

He lay on the floor He chuckled for pleasure

However, there are exceptions, Poutsma points out. The sentence (8)

Late at night he arrived at his destination

satisfies the above criteria - and yet the verb-modifier cannot be considered a prepositional object, because "the thing for which the noun in question stands is not felt to be a secondary participant in the action expressed by the verb" (1926:31). Furthermore, sentences such as (9)

He has slept in this bed

do not seem to satisfy condition (iii) that the preposition be intimately connected with the governing verb, nor condition (ii) that the preposition be vague in meaning. Yet, according to Poutsma, the verb-modifier here should be considered a prepositional object, as substantiated by the following two facts:

Note that Poutsma does not distinguish here between adverbials which must appear after the verb ('complements') and adverbials which may be left out ( ' a d j u n c t s ' ) . The latter term covers both varieties.

(a)

the verb + preposition is equivalent to a transitive verb: He has slept in this bed = (Dutch) Hij heeft dit bed beslapen

(b)

the noun connected with the preposition can be made the' subject of a passive sentence: This bed has been slept in

Note that the former observation demonstrates that sleep in α bed fulfils criterion (iii) for prepositional objects. The latter, however, does little more than reveal Poutsma's perhaps unconscious motivation for considering verbmodifiers of this sort as occupying an "intermediate position" (1926:35). It enables him to state as a general rule: Prepositional objects may be made the subject of passive sentences; adverbial 'adjuncts' may not. It is doubtful, however, whether equivalence to a simplex in the same or in a related language is an adequate criterion for determining what is a prepositional object and therefore whether passivization is possible or not. Poutsma's example impose upon = 'deceive1, for instance, demonstrates one of the problems involved: the predicate impose upon can have several meanings, including 'deceive' (archaic in modern-day English) and 'take advantage of. fact that a simplex equivalent exists for impose upon (in its

If the 1

'deceive meaning)

is taken as a criterion for a prepositional object and consequently for passivizability, does this mean that impose upon in its 'take advantage of meaning is not passivizable because 'take advantage of happens not to be a simplex? (Actually a passive with impose upon in this sense is quite ccntnon.) Furthermore, how is equivalence to be determined in the first place? How equivalent must the predicates be? And how is the fact to be accounted for that many predicates,

such as stand for = 'support1, do with = 'tolerate', etc. have simplex

equivalents but cannot be passivized? (10) This candidate stands for racial tolerance (lO)'*Racial tolerance is stood for by this candidate (11) I can't do with him and his insolence ( l l ) ' * H e (and his insolence) can't be done with

Of course what Poutsma may be getting at is that the existence of a lexical verb which is syntactically "transitive 1 corresponds to deep semantic transitivity. That passivizability depends to a great extent on the latter is precisely what this investigation will show.

As for Poutsma's tacit claim that prepositional objects can be passivized and adverbial 'adjuncts' cannot, the matter does not seem quite as simple if we take a closer look at some of his examples. The sentence ( 1 2 ) This does not tally with your former statement

- a sentence in which, according to Poutsma, the verb-modifier is a prepositional object - does not have an acceptable passive in English: ( 1 2 ) ' * Y o u r former statement is not tallied with by this

On the other hand, the verb-modifier in the sample sentence for an adverbial 1 adjunct': (6)

He lay on the floor

can be made the subject of quite an adequate passive sentence in English: ( 6 ) " This floor should not be lain on — it h a s n ' t been swept in weeks

In fact, even Poutsrra's argumentation with respect to (8) must be doubted in light of the following acceptable passive (Kruisinga 1911, 1931:312): ( 1 3 ) ' The scene of their exploits was arrived at by way of Gibraltar

Poutsma in fact has fallen victim to some rather circular thinking. He makes a distinction between prepositional object and adverbial 'adjunct1 primarily in order to be able to explain which combinations of verb + preposition + noun can be passivized, yet with examples such as sleep in a bed he uses the fact that a passive is possible as justification for classifying bed as a prepositional object. One wonders why the distinction need be insisted on at all, if it is, as Poutsma himself admits, so "vague and floating" (1926:31), and if it must be abandoned or grossly distorted in order to handle an important part of the data, i. e. those sentences where the prepositional phrase has full locative meaning. Nevertheless Poutsma's treatment of the PPP - in spite of its inadequacies has been fundamental to all subsequent discussions of the topic within the framework of traditional grammar (e. g. Curme 1931:113f, Zandvoort 1957, 1969:53f, Scheurweghs 1959:18f, etc.).

10

2.1.3 J. Svartvik (1966:20f) takes a slightly different approach to the PPP. Rather than treating sentences such as (9) above as exceptions to the rule of 'close association', he proposes, to make passivizability the criterion for the distinction between a "prepositional verb", V + N: (14) She sent for the coat ( 1 4 ) ' The coat was sent for

and a "verb plus prepositional phrase", V + N: (15) She came with the coat (15)' *The coat was come with

Note that without being circular, this classification cannot account for when a combination of the form verb + preposition + noun is passivizableand when it is not. Svartvik notes incidentally that the acceptability of passives with prepositional verbs may vary considerably: Some ... collocations (like look at) are very close and can occur in the passive with few or no restrictions, but others (like go Into) will do so only under certain conditions ... I think a lot of that /'problem; can be gone into ... *The room was gone into at once The difference in acceptability between the two passives must, it seems, be accounted for by some concept like concrete/abstract subject or literal/metaphorical verb (compare The town/conclusion was arrived at) (1966:165).

However, this explanation does not take into account the fact that a sentence like: (16)'

This room must not be gone into (or else the floor will cave in)

is perfectly acceptable in English [see also (6)'above] . Once a large corpus is considered, it becomes obvious that not only does a concrete noun as subject not block the passive (as Svartvik would suggest), but an abstract noun as subject does not always permit it: (17)'*0bscurity was emerged from ... (18)'*111 health was continued in ...

This should suffice as evidence that an explanation merely in terms of "concrete/abstract subject" is insufficient. Furthermore, a literal verb does not always block the passive:

11

( 1 9 ) ' The trap was fallen into (by the unsuspecting bear) (2O)' This cage cannot be easily escaped from ... nor does a metaphorical verb always allow it: ( 2 1 ) ' * A pocket was dived into (by Tom) for money (22)'*Thieves were fallen among ...

Svartvik's hypothesis, although appealing in its simplicity, is not borne out by empirical evidence. 2.2

Modern linguistic theories

Since the advent of generative grammar, the PPP has assumed a slightly more important position (by virtue of its transformational nature), although descriptions of it and/or rules for its application have hardly become more adequate. 2.2.1 In early transformational theory (Chomsky 1955:7O2f) , structures of the form verb + preposition + noun were to be dealt with by making a basic distinction between intransitive verbs (e. g. John ran after lunch, or They hoped for three years) - in which the preposition is more closely connected to the noun following it - and complex transitive verbs (They ran after John or They hoped for peace) - in which the preposition is more closely connected to the verb. This distinction is of course none other than that proposed by the 'cohesionists' of traditional granmar. However, it is perhaps significant to note that certain verb + preposition combinations would have been treated as being related to transitive verbs, presumably on the grounds that the passive transformation can be applied to both of them. In other words, structures of the form verb + preposition + noun which passivize are said to contain a complex transitive verb; those which do not are said to contain an intransitive verb. This solution, just as that offered by traditional grammarians, however, does not in any way help determine which verb + preposition combinations are passivizable and when. The Syntactic Structures model (Chomsky 1957:42f) provides for a transformational derivation of the English passive from active sentences of the form NP^ - Aux - V - NP-. This structural description, however, automatically excludes prepositional verbs and, as Chomsky himself later admits (1965:104), a second transformation would be necessary to account for the PPP. See Kullmann (197O: l l f ) for a discussion of Chomsky (1955) as it relates to the PPP.

12

The Aspects model incorporates several important modifications in order to handle the PPP more efficiently (Chomsky 1965:101ff). Structures of the form verb + preposition + noun in which the prepositional phrase has no particular connection to the verb are said to contain verb phrase (VP) complements, whereas those in which the prepositional phrase is closely connected to the verb are said to contain verb (V) complements (introduced by a subcategorization of the verb). [The former, Chomsky points out, can be optionally placed at the beginning of the sentence, whereas the latter cannot (1965:101).] If a prepositional phrase is a VP complement, it has locative or temporal function and is external to the VP itself; if it is a V complement, it fulfils the function of direction, duration, place or frequency, etc. and is internal to the VP. To illustrate the basic difference between the two, Chomsky cites the following two sentences (1965:217, Note 2 7 ) : ( 2 3 ) John died in England ( 2 4 ) John stayed in England

The former can be permuted to: (23a)

In England, John died

but for the latter, a similar permutation is impossible: (24a) * In England, John stayed

The prepositional phrase in (23) can thus be said to be external to the VP itself and therefore a VP complement; that in (24) , internal to the VP and a V complement. Sentence (25) is ambiguous,according to Chomsky: (25)

John lived in England

since it could conceivably be interpreted either as (25i)

John remained alive ( = did not die) John really lived ( = began to enjoy life) (cf In England, John lived)

or as (25ii) John resided in England (cf '-In England, John lived)

In interpretation (25i) the prepositional phrase would be derived as a VP complement; in interpretation (25ii) as a V complement.

13

However, allowing for two different derivations of Prep-Phrase is not the only innovation with respect to the passive in the Aspects model. In addition, since it is considered desirable to avoid the significant meaning change brought about by transformations such as Negative, Interrogative, etc. applied to kernel sentences, the Passive transformation is now to be triggered automatically by the presence of a dummy symbol in deep structure, one of the possible realizations of Manner. (Which verbs can occur with Manner is to be specified in the

Sate directional prepositions may indeed co-occur with is but when they do, the notion of direction is incorporated in them. They express then resultative position [ cf Quirk et al (1972:314) ] , i.e. a position which results from a perfected action involving direction: B

across

John! } Igoes I

across the river

against

Johnj

against the wall

along3

John! 1 Igoes ι

along the road

beyond

Johnj

J

beyond the gate

down

John! \ Igoes J

down the street

past ^

Johnj } Igoes J

past the station

round

John] } Igoes I

round the corner

through

Johnj

up f

Johnj \ Igoes ι

j

1S

j

through the tunnel up the tree

See also Bennett ( 1 9 7 5 ) , whose independent analysis of prepositions has produced a classification in many ways similar to this one.

52 Implicit in a sentence such as John is across/down/up

... the street is

thus 'John has gone across/down/up ... the street ' · Other prepositions such as before,

behind, near, over, under etc.

nay express

direction or position (non-resultative) freely: C

after

John I 1 I goes >

after Paul

before

John { 1 l goes aoes >

before the station

behind

John ί I goes is John I l goes aoe s

1 J

behind the tree

1

between the lines

between

J

by *

John j | I goes J

by the fence *

near

John { } I goes >

near the bed

off

John ( I goes is { goes

off the grass

under

} ι ι [ >

John { } I goes >

over the mountains under the bridge

Directional (A, B) or directional/positional (C) prepositions are typically found in conjunction with verbs of locomotion in sentences with Basic Goal. Thus: (38) (39) (40) (41) (42)

We can run to the store in a matter of minutes If you jump into the boat it may capsize They could drive across the road only at great risk We sailed along the coast for miles before we encountered a bay He dodged behind a tree in order not to be seen

(43)

You mustn't go beyond the gate

(44) (45) (46) (47) (48)

by j his house every day past People don't normally jump down rabbit holes Almost anyone can jump over this fence It would be days before they could tunnel through the snow bank Many people don't realize how carefully firemen must go up this ladder

Dozens of school-children walk j

However, the presence of locomotion verb + directional preposition does not automatically imply the presence of Goal. Consider, for instance: (49) (50) (51)

We tramped (all) over the moors Hundreds of dancers were whirling (all) round the room Wild animals range (all) through these forests

7

John is after Paul is to be interpreted here not as 'John is chasing Paul 1 but as 'John comes after Paul 1 (in a series, for instance).

8

Off has been placed in this category based on the author's idiolect. In Bennett ( 1 9 7 5 : 7 2 f f ) , however, it is considered to be a basically directional preposition with resultative meaning if used positionally.

53

In these examples, the locomotion + direction predicates have an added component of meaning, namely 'pervasive meaning' [ cf Quirk et al (1972:314) ]. Generally speaking, if pervasive meaning is present, the activity is thought of as occurring everywhere or at various spots within a given area (as evidenced by the fact that all can often be added without affecting the meaning). There is no reference, however, to the activity extending beyond the limits of the area [see also Lindkvist (1972:35)] . Since it is precisely extension beyond boundaries which is essential to the notion of Goal, we shall assign Location (rather than Goal) to sentences such as these. In summary, if actions were to be symbolized pictorially, where we have

the entity in object position is Location. On the other hand, where we have or as in: (52) (53) (54)

We had to tramp over the moors to reach the catastrophe area They travelled round the world in 8 days John stumbled through the speech in fifteen minutes

the entity in object position is Goal. Only in this case does the locomotion involve a change of ultimate location or boundary-crossing. 2.7.1.2 The delimitation of Basic Goal in such concrete terms does not mean that it is necessarily restricted to literal predicates. Indeed a number of corpus sentences can be found in which a metaphorical version of Basic Goal is involved. Examples include: (55) (56) (57) (58) (59) (60)

We shouldn't rush into the decision The reader can jump over a number of chapters He was galloping through the work in order to meet the deadline They ought not to jump to conclusions People can easily get (a)round that difficulty Another woman came between them

These too must be considered in determining the eligibility of Basic Goal as P?P subject.

54 2.7.1.3 The distinction between location vs ultimate change of location or boundary-crossing is not only useful in identifying concrete Basic Goal. It is also essential in determining PPP acceptability. Thus, virtually all' the examples above with Basic Goal in object position are passivizable: (38)' (39)' (4O)' (41)'

The store can be run to in a matter of minutes ... If the boat is jumped into ... it may capsize The road could be driven across only at great risk ... The coast was sailed along for miles ... before a bay was encountered ( 4 2 ) ' A tree was dodged behind ... in order not to be seen ( 4 3 ) ' The gate mustn't be gone beyond ... ( 4 4 ) ' His house is walkedir^

jby dozens of school-children every day

(45)' (46)' (47)' (48)'

Rabbit holes aren't normally jumped down ... This fence can be jumped over by almost anyone It would be days before the snow bank could be tunnelled through .. Many people don't realize how carefully this ladder must be gone up... ( 5 2 ) ' The moors had to be tramped over ... to reach the catastrophe area ( 5 3 ) ' The world was travelled cound ... in 8 days ( 5 4 ) ' The speech was stumbled through ... in fifteen minutes Metaphorical varieties of Basic Goal are likewise generally compatible with the passive in subject position. Thus: (55)' The decision shouldn't be rushed into ... ( 5 6 ) ' A number of chapters can be jumped over ... ( 5 7 ) ' The work was being galloped through ... in order to meet the deadline (58)' Conclusions ought not to be jumped to ... ( 5 9 ) ' That difficulty can be easily got (a)round ... (6O)' They were come between by another woman On the other hand, if sane of these same examples are manipulated in order to express, say, motion + direction (or resultative position) rather than locomotion + direction, then the notion in object position becomes one of Location and a passive is no longer possible: (4O)' The road could be driven across only at great risk ... (61)' "The road could be played across only at great risk ... (41)' The coast was sailed along for miles ... ( 6 2 ) " *The coast was often sunk along by ships (43)' The gate mustn't be gone beyond ... (63)' *The gate mustn't be played beyond ... (64)' The street was being raced down ... (65)' *The street was being played down ... (66)' The street was often gone up ... (67)' *The street was often played up ...

55

Likewise, the objects in examples (49), (5O) and (51) with pervasive meaning are Locations and correlate in other ways with the passive (see III.2.9). It is thus important to distinguish Basic Goal fron Location in determining passivizability. Whereas the former is almost unconditionally eligible for subject position in the passive, the latter, as we shall see, often cannot beccme a PPP subject. 2.7.2

Goal as ending-point

Lindkvist (1950:178) notes the existence of an 'ending-point relation1 with ocmpleted-action or perfective verbs in English: " ... as the verb denotes or implies that a movement has taken place or is to take place, the preposition also indicates where the ending-point of that movement lies". In terms of our classification, this notion is to be found with verbs expressing loccmotion in conjunction with prepositions expressing position. The latter include: at

John L l Fgoes J

a

in

John L S l "goes '

in the bath

on

John L \ Γgoes >

on the couch

about

John { . .} I (goes)>

about the garden

above

John f , .} L (goes)>

above the trees

among

John {

among the flowers

}

t the office

Among the corpus examples illustrative of ending-point Goal as understood here are: (68) (69) (70) (71) (72) (73) (74)

We didn't arrive at the airport until early morning You can sit at John's desk if you get tired They moved in the house three weeks ago If you stand on a stool, you can reach it He got on the horse gingerly You roust lie on this couch carefully No one had ever landed on the moon before

(All of the above predicates are to be understood as implying movement, i.e. sit at = 'sit down at', lie on = 'lie down on1, etc.) The prepositions about, above, among may occasionally occur with go or a similar verb [The dog ran about the garden, The pilot flew above the clouds , They went among the poor], but they express position nevertheless, namely the position the agent is in after having moved or while moving.

56

Metaphorical examples include: (75) (76)

They never arrived at a solution We must come to a decision immediately

With the exception of (68) this group yields for the most part acceptable passives: (68)' (69)" (7O) ' (71)' (72)' (73)' (74)' (75)' (76)'

?The airport was not arrived at until early morning ... John's desk can be sat at ... if you get tired The house was moved in three weeks ago . . . If a stool is stood on ... it can be reached The horse was got on gingerly ... This couch must be lain on carefully ... The moon had never been landed on before ... A solution was never arrived at ... A decision must be come to immediately ...

However, sll of these sentences behave with respect to the passive in a way quite similar to sentences with Location objects. That is, as we shall see shortly, the active object can be construed as instrumental in, or affected by the action (cf III. 2. 9). Airport· in (68)', on the other hand, cannot be easily interpreted this way and that is why it is not an acceptable PPP subject. Now compare the following PPP, documented in Kruisinga (1911; 1931 :312) : ( 7 7 ) ' The scene of their exploits was arrived at by way of Gibraltar

Here scene of their exploits lends itself more easily to an instrumental interpretation (cf III. 2. 9. 1.2) and a passive is acceptable. We must conclude therefore that ending-point Goal is eligible for PPP subject position only under the same or similar conditions as Location. 2.7.3

Goal as target

The notion basic to this type of Goal is that of target, something towards which a movement is directed but with which physical contact is not necessarily made [see also Lindkvist (1950:187)] . Notice that in this category locomotion is not necessarily involved. Corpus examples include: (78) (79) (80) (81) (82) (83)

He aimed at the target but missed People pointed at him The man snatched at the rope desperately A man With a .big stick came at me They fired on the crowd ruthlessly We made for the open sea

57

and metaphorically: (84) (85) (86)

The lawyer pointed to their poverty as an extenuating circumstance People have guessed at her age for years He worked at the job diligently

All of the above sentences are passivizable: (78)' (79)' (8O)' (81)' (82)' (83)' (84)' (85)' (86)'

The target was aimed at ... but missed He was pointed at by people The rope was snatched at ... desperately I was come at by a man with a big stick The crowd was fired on ruthlessly ... The open sea was made for ... Their poverty was pointed to ... as an extenuating circumstance Her age has been guessed at ... for years The job was worked at ... diligently

Goal-Target is thus eligible as PPP subject.

2.7.4

Goal as direction

Similar to Goal-Target, Goal as direction is associated with an action directed towards someone or something. However, in contrast to Goal-Target, the categories locomotion, notion and non-motion are wholly irrelevant for Goal as direction. (This is how it differs from the notion of Direction proper; see III.2.4). The activities associated with this type of Goal include seeing, hearing, talking, laughing and gesticulating, i. e. metaphorical movement at best. Typical corpus examples are: (87) (88) (89) (90) (91) (92) (93) (94) (95)

If a dog barked at John, he became frightened She was constantly whispering in his ear Someone was peering at him from behind the curtain He could skim over the book in a second People can't see through lined envelopes You may glance (a)round the room from the doorsill but don't enter it All the neighborhood dogs who go by smell at this lamppost People didn't laugh at his jokes any more His parents have always frowned on his activities

or metaphorically: (96) (97) (98)

He inquired into the matter immediately Someone should look into her affairs People will see through his game some day

All of these have acceptable passives: ( 8 7 ) ' If John was barked at by a dog, he became frightened (88)' His ear was constantly being whispered in ... ( 8 9 ) " He was being peered at ... from behind the curtain

58 (9O)' (91)' (92)' (93)' (94)' (95)' (96)' (97)' (98)'

The book could be skimmed over in a second ... Lined envelopes can't be seen through ··· The room may be glanced (a)round from the doorsill but not entered. This lamppost is smelled at by all the neighborhood dogs who go by His jokes were not laughed at any more ... His activities have always been frowned on by his parents The matter was inquired into immediately ... Her affairs should be looked into ... His game will be seen through some day ...

Goal-Direction thus also meets the conditions for PPP subjecthood. 2.7.5

Goal as objective

Another major interpretation of Goal involves an objective which is to be got, found or caught. This category may co-occur with locomotion but need not necessarily do so: (99) (100) (101) (102)

Hundreds of policemen were scouring after him They advertised for a manageress immediately We prayed for rain but to no avail He was gradually gaining on the other runners in the race

Metaphorical examples include: (103)

You have to study for exams

All have acceptable passives: (99)' (1OO)' (101)' (1O2)' (1O3)'

He was being scoured after by hundreds of policemen A manageress was advertised for immediately ... Rain was prayed for ... but to no avail The other runners in the race were gradually being gained on Exams have to be studied for ...

...

Goal-Objective is thus likewise eligible for subject position in the PPP.

2.7.6

In conclusion, the sentences above present overwhelming evidence for the

compatibility of Basic Goal, Goal-Target, Goal-Direction and Goal-Objective with the passive. Ending-point Goal is an exception to the extent that its partly locational nature brings other considerations into play. It must therefore be considered only conditionally eligible as PP? subject. All other Goal types are unconditionally eligible.

59

2.8

Instrument

Defined originally by Fillmore as "the case of the inanimate force or object causally involved in the action or state identified by the verb" (1968b:24), Instrument (Instr) was later modified to "the case of the immediate cause of an event, or, in the case of a psychological predicator, the "stimulus', the thing reacted to" (197la:42). Collapsing a stimulus and an instrument notion into one case was obviously motivated by considerations of economy and stringency within the case grammar model. However, as the two are distinct notions which correlate differently with the PPP, we shall treat them separately in this analysis.

2.8.1 The Instrument notion takes a number of different forms in our corpus. For instance, it appears as 'material' or 'medium' in: A

(1O4) (105)

They traded in furs and skins He worked in leather

and as metaphorical medium in: (106) (107)

People rarely specialize in medieval history here You may encounter rejection if you innovate in externals

On the other hand, it appears as 'tool', i.e. an object or device whose nature makes it appropriate for a certain use, in B

(1O8) If you look through a telescope, you will see Jupiter's moons (109) You can juggle with these balls (110) You can write with this pen

and metaphorically: (111) (112)

You can't go by appearances No one can possibly live by this ru2e

However, Instrument may also be more generally 'means', i.e. an entity which is involved in achieving a certain purpose: C

(113) (114) (115)

Someone may fool with these matches (and cause a fire) Out-of-town visitors frequently board with John's aunt He should consult with a lawyer immediately

Finally, Instrument may appear as 'accidental cause', i.e. an object not used volitionally in achieving a purpose, but merely accidentally involved in bringing about an event. Thus, the objects in the following also have Instrument status:

60 (116) (117)

People will stumble over this chair if i t ' s not moved Someone will slip on that banana peel sooner or later

2.8.2 Passivizing sentences with these various kinds of Instrument produces different results. On the one hand, we find acceptable passives in B, C and D: (1O8)' (1O9)' (HO)' (113)' (114)' (115)' (116)' (117)'

If a telescope is looked through ... Jupiter's moons will be seen These balls can be juggled with ... This pen can be written with ... These matches may be fooled with ... (and cause a f i r e ) John's aunt is often boarded with by out-of-town visitors A lawyer should be consulted with immediately ... This chair will be stumbled over ... if it's not moved That banana peel will be slipped on ... sooner or later

The 'tool', 'means' and 'accidental cause1 interpretations of Instrument appear to be generally suitable for subject position in the PPP. On the other hand, in group A, concrete medium is not particularly compatible with the passive: ( 1 O 4 ) ' *Furs and skins were traded in ( 1 O 5 ) ' ?Leather was worked in ...

Although this might appear to be grounds for assuming that two different semantic categories are involved (Instrument and Medium), the constraint on the latter can be explained with reference to the Objectification' process which all PPP subjects undergo. We shall discuss this process shortly (see III.2.9.2). Note that metaphorical medium is more acceptable as PPP subject: (1O6)' Medieval history is rarely specialized in ... (1O7)' Rejection may be encountered if externals are innovated in ...

Thus we may conclude that Instrument in general is eligible for PPP subject position. 2.8.3 It has been suggested that superficial instrumental adverbs such as the one in Seymour sliced the salami with a knife can be related in deep structure to Seymour used a knife to slice the salami (G. Lakoff 1968:6f). Note that the presence of Instrument, as we understand it here, cannot be perfectly detected by use of a paraphrase technique based on this deep structure relation. Although it is acceptable to paraphrase sentences with Instrument-Tool as 1O

Instrument as 'tool 1 , however, is especially sensitive to lexical considerations; see I V . 2 . 4 .

61

follows: (lO8a) (109a) (llOa)

If you use a telescope to look (through) ... You can use these balls to juggle (with) You can use this pen to write (with)

sentences with Instrument-Means often require a laxer interpretation of the verb

use: (113a) (114a) (115a)

Someone may 'use' these matches to fool with Out-of-town visitors frequently 'use' John's aunt to board with He should 'use' a lawyer to consult with

and those with Instrument-Accidental Cause would be quite unheard of with use: (116a) *Someone used the chair to stumble over (117a) *Someone used a banana peel to slip on

A use ... to paraphrase is thus not completely reliable as a means of

identifying

Instrument and therefore a potential PPP subject.

2.9

Location

The locative case is described in Fillmore (1968b:25) as the one "which identifies the location or spatial orientation of the state or action identified by the verb". For our purposes, it is helpful to view Location (Loc) in contrast to Goal. Whereas the latter is closely identified with movement and implies boundary-crossing as well, Location is associated either with no movement at all or with movement only within the boundaries of a given area. It thus identifies the place at which, or the area within which, an activity or event occurs. Just as Goal objects in our corpus occur typically in a configuration with loccniotion verb + direction preposition, so Location objects occur typically with (A) non-notion verb + position preposition, or (B) motion verb + position preposition. The following examples illustrate these two varieties: A

(118) (119) (120) (121) (122)

You can nestle among these cushions John was lying in the water Susan waited on the doorstep The Queen resides at this palace A half dozen men are always dangling about her

B

(123) (124) (125) (126) (127)

Lambs frolicked beyond the gate People were constantly nosing about my desk Fido trembled under the table You can luxuriate in this bathtub for hours We had to root among piles of old paper

62

Location may also occur of course with a locomotion verb + position preposition. In this case, pervasive meaning is present (see p. 53 above). C

(49) (51) (128) (129)

We tramped (all) over the moors Wild animals range (all) through these forests They traipsed (all) round the department store for hours Someone was sneaking (all) about the ship

2.9.1 Given these different types of Location objects, the next step is to determine how suitable each is as the subject of a PPP. Passivization in A and B above produces sentences of varying degrees of acceptability: A

B

(118)' These cushions can be nestled among ... (119)'*The water was being lain in ... (12O)'*The doorsteps were waited on ... (121)" This palace is resided at by the Queen ( 1 2 2 ) ' She is always being dangled about by a half dozen men (123)'*The gate was frolicked beyond ... ( 1 2 4 ) ' My desk was constantly being nosed about ... (125)'*The table was trembled under ... (126)' This bathtub can be luxuriated in for hours ... ( 1 2 7 ) ' Piles of old paper had to be rooted among ...

The sentences in group C, on the other hand, all have acceptable passives: C

(49)' (51)' (128)' (129)'

The moors were tramped over ... These forests are ranged through by wild animals The department store was traipsed round ... for hours The ship was being sneaked about ...

It should be obvious from this small sample that a criterion merely in terms of Location is too general to use in accounting for PPP acceptability. Just as with Instrument, there appear to be a number of subcategories involved. These will be described in the following sections. 2.9.1.1 Examples (122), (128), (129) etc. represent a first sub-group of Location. In these sentences in particular, there is an accompanying notion of goaldirected behavior. Broadly speaking, we might say that the locations have a goal-like connotation without, however, losing their positional nature. This contrasts sharply with sentences such as: (130) (131) (132)

A group of teen-age boys was lolling about the door He mooned about the house and gardens for weeks She messed about the flowers

Lolling, mooning and messing are not activities directed towards places or people; consequently their locations do not have goal connotations. The data suggest that

63

only those locations which have a goal connotation are capable of assuming subject position in a PPP. Thus, compare (122)', (128)', and (129)' to: (13O)' ?The door was being lolled about by a group of teen-age boys (131y ?The house and gardens were mooned about for weeks ... (132)' *The flowers were messed about ...

Note that the notion of goal-directedness cannot be simply equated with that of volition, lolling, mooning and messing imply just as much an act of the will as do dangling, traipsing and sneaking. They differ, however, in the degree to which they are directed towards a location. We can conclude that locative expressions in sentences which lack a notion of goal-directedness do not fulfil the conditions for subjecthcod and are therefore unlikely to become PPP subjects. 2.9.1.2 Examples (118), (121) and (126) represent a second, larger subclass of Location. In these sentences and similar ones, the locations are accompanied by a notion of instrumentality. That is, they are considered to be serving a particular purpose. Lindkvist (1950) attests to the reality of such a category in his analysis of prepositions. There the locative meaning of a preposition in conjunction with instrumentality is handled under a separate heading, namely: .at

location close to or within a body, surface or area thought of as being used to serve a certain purpose (1950:164)

in

enclosure within a body, surface or area thought of as being used to serve a certain purpose (195O:lOl)

on

location in connection with a body or surface thought of as being used to serve a certain purpose (1950:249)

Locations likely to be thought of this way, according to Lindkvist, include the following (illustrated with examples from our corpus) : 11

Documented in our corpus but not included in this list are Lindkvist's categories (a) places of employment: He served on the committee for years-, (b) covering objects: you shouldn't travel in new clothes; (c) instrument: you must practice on the kettle drums for years before becoming accomplished. Only lexical considerations would argue in favor of assigning Loc here (in and on are 'locative 1 prepositions). In terras of semantic notions, it is more likely that Ben and Instr are involved in (a) and (b,c) respectively.

64

i.

human habitations or dwellings (1950:101; 167;249) (133)

ii.

buildings or places as centers of activity (1950:103;169;253) (134)

iii.

The poor frequently deal at this shop

means of conveyance (195O:11O;256) (135)

iv.

No one can live in this house any longer

No more than two people must sit in this cart at once

'supporting' objects (1950:112) or resting-places (1950:261) (136)

People allergic to doghairs shouldn't sleep on this mattress

Properly contextualized, corpus sentences with locative-instrumental expressions such as these passivize readily: (133)' (134)' (135)' (136)'

This house can't be lived in any longer ... This shop is frequently dealt at by the poor This cart must not be sat in by more than two people at once This mattress shouldn't be slept on by people allergic to doghairs

Note that the instrumental connotation which may accompany locative expressions is of two sorts and reflects the "tool1 vs 'means' varieties of Instrument proper (III.2.8). An instrument-tool connotation is illustrated by (133), (135) and (136), in which it is the nature of the location itself which makes it suitable or unsuitable as an instrument in a given action. An instrument-means connotation is illustrated by (134), in which it is use of the location which serves an instrumental purpose. In one case, focus is on the ultimate purpose resulting from use of the instrument; in the other, on the nature of the instrument itself. However, both types of instrumental connotation have in conroon the fact that a location is 'useful1 in carrying out an action or in bringing about an event. Even in sentences with metaphorical location as object, the location can often be construed instrumentally, or as serving a particular purpose. Thus: (137) John's brother presided at the marriage ceremony (137)' The marriage ceremony was presided at by John's brother (138)

Runners from over 5O different countries will participate in the race (138)' The race will be participated in by runners from over 5O different countries

65 (139) How can people exist in such wretched conditions? (139)' How can such wretched conditions be existed in? (140)

We can

trust in him believe

(14O)' He can be

trusted believed

in

Based on the evidence provided by sentences such as these, we can conclude that a notion of instrumental location fulfils the conditions for subjecthcod in a PPP.

2.9.1.3 Sentences (124), (127), (49), (51), etc. illustrate a third major subclass of Location. In these and many similar examples, the locations are affected in one way or another by the activity in question. Consider, for instance: (141) (142) (143)

If you continually jump on the i>ed, the springs will wear out The hen sat on the egg until it hatched People were constantly stamping about the room above

These sentences refer to situations in which the state or condition of the location is changed, or is capable of being changed by the activity, i. e. jumping, sitting, or stamping. Similarly, in the situations referred to by (124), (127), (49) and (51), the location is capable of being changed or affected by nosing, rooting, tramping and ranging. Solinger would undoubtedly refer to locations such as these as "true patients", which he defines as being "genuinely affected by the action of the verb" (1974:67). His explanation of the phenomenon is as follows: The semantic contrast is between true patients as defined and things or persons that are merely located with reference to others or to existence itself, i.e., presence on the scene, or absence. (1974:68)

We shall call locations which are affected by the action affected As such, they correlate highly with PPP acceptability. Thus:

locations.

(141)' If the bed is continually jumped on ... the springs will wear out ( 1 4 2 ) ' The egg was sat on ... until it hatched ( 1 4 3 ) ' The room above was constantly being stamped about ...

As these examples illustrate, the effect on the location may or may not be of a detrimental nature. Where damage or harm results, the question "What happened to ?' is likely to be appropriate, as well as the paraphrase "What some-

66

12 one did to vas ' [see also Chafe (1970:100)] . Thus, these can serve as aids in determining when a connotation of detrimental effect is present and consequently when passivization is possible: (144a) (144b) (144)'

What happened to your hat? What someone did to my hat was sit My hat was sat on

on it

Note, however, that the paraphrase technique fails manifestly in the case of non-detrimentally affected locations: (143a) (143b) (143)'

?What was constantly happening to the room above? ?What someone was constantly doing to the room above was stamping about it The room above was constantly being stamped about

(145a) (145b) (145)'

?What happened to this bed? ?What Thomas Jefferson did to this bed was sleep in it This bed was slept in by Thomas Jefferson

It would therefore be mistaken to restrict an effect connotation in locative expressions to cases in which a question with happen or a paraphrase with χ do to y is possible. Actually, the decision as to whether a location is likely to be affected by an activity or not is one which in the last analysis depends on extralinguistic knowledge. Experience has taught us that certain spatial relations between locations and activities are likely to lead to sane kind of effect and that others are not. For instance: (146) Someone slept in this bed (146)' This bed was slept in ... (147) Someone slept on this bed (147)' This bed was slept on ... but

(148) Someone slept by this bed (148)' *This bed was slept by ... (149) Someone slept near this bed (149)' *This bed was slept near ...

In our understanding of the world, a location cannot be affected by an activity which takes place merely in its vicinity. We are of course dealing here with extremely concrete notions of location and effect. This is not to insinuate, however, that only locations affected literally 12

Note that a get auxiliary is not infrequent with PPPs in which the location is detrimentally affected: My new hat got sat on [cf R. Lakoff (1971) for a discussion of the so-called get passive.]

67

are suitable PPP subjects. Consider, for instance, the metaphorical locations

in: (150) A triumvirate ruled over the empire for many years ( I S O ) ' The empire was ruled over for many years by a triumvirate (151) We can lean on John in time of need (151)' John can be leaned on in time of need

Perhaps a sort of metaphorical effect can be posited for the metaphorical locations in these sentences. In conclusion, we can state that a notion of affected location also fulfils the conditions for subjecthood in PPPs.

2.9.2

We have thus determined that there are three subcategories of Location

capable of becoming PP? subjects, to be termed Location-Goal, Location-Instrument and Location-Effect. Note that all three types of Location have one feature in camon: namely, they are objects. If we view the PPPs in which they occur as resulting from the passivization of active sentences, then we can say that Locations which became PPP subjects undergo a process of objcertification. That is, they lose their orientational character and become objects. For example, in (144)

John sat on my hat

the locative expression could very easily be intended as the spatial orientation of John's sitting, i.e. on the hat and not next to it or in front of it. On the other hand, in the same sentence passivized: (144)' My hat was sat on ... hat can only be interpreted as an object (which is affected in sane way). A nunber of characteristics testify to the objectified status of locative PPP subjects. For one, the entity in the subject position of a concrete locative PPP is normally given concrete interpretation. Compare, for instance: (152) People settled in the country (152)' The country was settled in ...

Whereas in the active version, country could very well refer to an abstract notion whose opposite is city, in the passive country must be understood as referring to a concrete nation such as France, Germany, etc. The more abstract notion of country as open space cannot appear in the subject position of a concrete locative PPP.

68

Furthermore, if the entity in object position is incapable of taking on concrete meaning, a passive will often be blocked. Compare, for instance: (153) They slept in the open air ( 1 5 3 ) ' *The open air was slept in ...

The fact that it is virtually impossible for open air· to be construed as a concrete object (and therefore capable of being affected or used in sane way) accounts for the basic non-passivizability of the predicate sleep in the open

air. Second, a locative PPP subject is interpreted holistically. The term holistic has been used by Anderson (1971:389) and others to refer to an object being affected as a whole by the action described by the verb. One of the classic examples of holistic vs partitive interpretation is: (154a) (154b)

John smeared paint on the wall John smeared the wall with paint

Whereas in (154a) it would be possible to state without contradiction 'John smeared paint on the wall but most of the wall didn't have any paint on it', the same cannot be said of (154b): '*John smeared the wall with paint but most of the wall didn't have any paint on it 1 . Notice that a passive subject also requires holistic interpretation: ( 1 5 4 ) ' The wall was smeared with paint by John (154a)'*The wall was smeared with paint by John but most of the wall didn't have any paint on it

The same holistic interpretation can be observed in PPP subjects. Consider, for instance: (155) We can rest on the top of the mountain ( 1 5 5 ) ' The top of the mountain can be rested on

(155) could quite conceivably correspond to a mental image of: resting-spots However, if a passive such as (155)'were to be used, it would call forth a mental image of:

—

resting-spot

That is, the top of the mountain would be construed in such a way that it could conceivably be used as a whole in the resting.

69

Third, a locative PPP subject is a discrete, separable entity. Fillmore (1970:127f) demonstrates what is meant by separability in discussing changeof-state vs surface-contact verbs. In his words, if we say John hit the top of the twarge^ the word top is merely referring to a portion of the surface area of the twarge; however, if we say John broke the top of the twarge, top is intuitively understood as a separable part of the twarge. This type of separability is also a feature of locative PPP subjects. Thus, in a sentence such as (156)

You can swim in the water

water might conceivably designate an unlimited expanse. But in (156)' The water can be swum in water clearly means a 'piece of water1, as demonstrated by the fact that the article this (or that ) can be substituted without a change of meaning: (156.1)' This (That) water can be swum in The features separability, holistic interpretation and concreteness are not necessarily independent of each other, and may even be necessary concomitants of objects. However, they add extra weight to the theory that locative PPP subjects undergo a process of objectification. Note, too, that the same type of objectification can be demonstrated for Goal and Instrument PPP subjects. Thus, whereas John leaned against the wall could conceivably mean he leaned against a portion of the wall, the sentence The wall was leaned against ... can only mean the wall as a detachable whole was used as a support. And we find a constraint in a sentence like *The ground was fallen to because ground does not refer to a detachable object capable of taking on holistic interpretation. Furthermore, objectification explains why sentences with literal Median such as *Leather was worked in are unlikely in the passive: the medium as an abstraction cannot be interpreted as a concrete object. We shall return to a discussion of this and other features which Location, Goal and Instrument have in common presently.

2.9.3 We must now attempt to clarify how Location-Goal, Location-Instrument and Location-Effect relate to each other in determining the suitability of locative expressions for subject position in PPPs. Note first that they do not mutually exclude each other. Sentence (135) above, for instance, illustrates a case in which an object if used instrumentally will be affected detrimentally:

70 i.e. the cart will break.

Contextualized sonewhat differently, the same predi-

cate could even illustrate a case in which a detrimentally affected object has been misused in some way: (135.1)' This cart was sat in by more than two people at once (now it's broken)

Furthermore, if an activity is directed towards a location as goal, as in (129) above, there is likely to be some resulting effect. Thus, it cannot be claimed that these are necessarily independent varieties of Location. However, it does appear that the presence of at least one of the three connotations is a necessary precondition for an acceptable PPP. Consider, for instance, our examples from above: (119) (120) (123) (125)

John was Lying in the water Susan waited on the doorstep Lambs frolicked beyond the gate Fido trembled under the table

In none of these is the activity directed toward the location in the sense of (122), (128) and (129) above. Therefore, a goal connotation is absent. Furthermore, the location is not likely to be affected by the activity in question. Water is not affected if someone lies in it, nor is a doorstep if someone waits on it, nor a gate if animals frolic beyond it, nor (normally) a table if something trembles under it. Of course, if some kind of effect can be made plausible, then a passive too becomes more likely. Thus, if we imagine, say, a large dog and a low table, then it might indeed be possible for the table to be affected by the activity and the following passive would be more conceivable: (125.1)' [We wondered why the table was vibrating so strongly; ?Then we realized that it was being trembled under by a very large dog

In general, however, the locations in sentences (119), (12O), (123) and (125) do not lend themselves easily to an affected interpretation. Nor is it really likely that the locations could be interpreted as instrumentally involved in the action. A table can hardly be used to tremble under, nor will trembling under it serve any ultimate purpose. The same can be said of a gate and frolicking or a doorstep and waiting. With water and lying there is admittedly room for imagination. For instance, a doctor who has just added a medicinal fluid to a tub of water might conceivably say to a patient suffering from 13

Example (136) illustrates yet another variation: here use of an object instrumentally brings detriment to the user (and not the object).

71

arthritis: ( 1 1 9 . 1 ) ' ?Now if this water is lain in for several hours, the pain will disappear

In general, however, the locations as contextualized in (119),(120),(123) and (125) cannot be given an instrumental interpretation any more than a goal or effect interpretation, and it is for this reason, we maintain, that a passive is blocked.

2.9.4 (a)

Suimarizing the discussion so far, we have seen that: Location as a notional category is not specific enough per se to permit accurate predictions of PPP acceptability.

(b)

At least one additional condition must be fulfilled in order for a location to become a PPP subject: the location must be thought of as being a goal towards which the action is directed, or as being instrunentally involved in the action, or as being affected by the action (or any combination of these three).

These findings, it will be seen, have a number of consequences for grammatical analysis, and for case grammar in particular. Considered from a case point of view, it should be obvious that the PPP data, especially with respect to Location, make a number of changes in the model compelling. If standard case grammar could account for the PPP, we would expect certain cases to be compatible with the passive, i.e. eligible for subjectivization in a passive sentence, and certain others to be incompatible, i.e. ineligible for passive subjectivization. However, our findings indicate that Location does not behave consistently with respect to the passive and that its behavior can only be explained if

it

is subcategorized. Notice too that the subcategories necessary to account for the data, Location-Goal, Location-Instrument and Location-Effect, are (?coincidentally) not very distant relatives of the case notions Goal, Instrument and Patient proper. The case model nust therefore be modified if it is to be useful in accounting for the PPP. The alternatives which present themselves are as follows: 1) The locative case could simply be subdivided, say, in Location-Location, Location-Goal, Location-Instrument and Location-Patient. Only the latter three would be eligible for passive subjectivization. However, this subdivision would have no external motivation (to our knowledge) and, as formilated, would make it look suspiciously like four different cases are actually involved.

72

2} Locative expressions having a goal, instrument or patient connotation in the active could be assigned to the Goal, Instrument or Patient case respectively (and would then become eligible for subjectivization as such). This alternative is somewhat more attractive, although it makes no allowance for non-determination in the active. Part of the 'meaning' of a sentence such as Someone sat on my hat is surely that it is unclear whether the hat is a location or a patient. (In the passive My hat IMS sat on, of course, there is no longer any doubt.) But according to this alternative, it would be necessary to decide in the active either for Location or for, say, Patient and the non-determined nature of the locative expression would be lost. 3) A process of case-shifting could be posited by which a Location in the active would take on characteristics associated with the Goal, Instrument or Patient case if it were to assume subject position in a passive sentence. (Only those Locations capable of doing so would be eligible for passive subjectivization.) This alternative has the advantage of allowing for semantic non-determination, although the process of case-shifting would have to be worked out in more detail. [Lambert's (1969) proposals to handle metaphorical phenomena in English involve case features being passed on to lexical items but case-shifting as such is rejected (1969:308).] 4) Active sentences with Location objects capable of having a goal, instrument or patient connotation could be considered ambiguous and assigned two different semantic structures, one consisting of [+Ag +Loc], the other of [+Ag +Goal] or [-tAg +lnstr] or [+Ag +Pat]. In the active either might be present. Thus, a sentence such as Someone sat on my hat could arise as a comment on where the sitting took place or as a comment on the effect to the hat. If only Goal, Instr and Pat (but not LOG) were allowed to become passive subjects, then this would explain why the PPP My hat was sat on can only be given an effect interpretation. This alternative has the advantage of allowing for semantic ambiguity in the active while explaining the non-ambiguity of the passive, and we shall adopt it for the purposes of this analysis. Of course, in a comprehensive grammar of the English language, various of the other alternatives might require re-consideration. 2.1O

Manner

Manner (Man), as the term suggests, describes the way something happens or is done. Examples include:

73 (157) (158) (159)

John spoke in riddles They coincided in opinion We differ in taste

(160)

The train was going at 5O mph

and

Itonner appears to be totally incompatible with the passive in subject position: (157)' (158)' (159)' (16O)'

*Riddles were spoken in ... *Opinion was coincided in ... Taste is differed in ... *5O mph was gone at ...

Thus a Manner notion does not fulfil the conditions for subjecthood in a PPP.

2.11

Path

Following Fillroore (1971a:5Of) and Bennett (1975:18), we shall use the term Path to refer approximately to the route or itinerary taken in the course of movement. For instance: (161) (162)

John walked behind the parked car (on the way to the bus) We had to pass through a number of villages (on the way to the border)

In general, Path objects are eligible for subject position in PPPs if they fulfil conditions similar to those placed on Location. That is, Path must be interpretable as instrumental in, or likely to be affected by, the action in question. (The meaning of Path excludes the possibility of a goal interpretation.) In (162) the action of passing is likely to affect the Path object and a passive is possible: ( 1 6 2 ) ' A number of villages had to be passed through. . . ( o n the way to the border)

In (161), on the other hand, it is not particularly clear how the car could be affected by, or instrumental in, the walking; thus a passive is somewhat dubious: (161)' ?The parked car was walked behind ... (on the way to the bus·) But contextualized in a way which emphasizes its instrumental (and detrimental) character, the same predicate improves markedly in the passive: (161.1)' Cars parked on hills shouldn't be walked behind (they might start rolling backwards)

Thus, similar to Location, a Path notion fulfils the requirements for PPP subjecthood only under certain conditions.

74

2.12

Obj ective/Patient

As mentioned in connection with our discussion of Objective/Patient in subject position (II.2.5), Fillmore's definition of this notion is a curious mixture of concrete and abstract features. On the one hand, the case is to encompass concretely "things which are affected by the action or state identified by the verb" (1968b:25). Later, this part of the definition is expanded somewhat to "the entity which moves or which undergoes change" (Fillmore 1971 a:42). In this concrete sense, the term patient may seem the most appropriate denomination. On the other hand, the original notion is also to include quite abstractly "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" (1968b : 25); in Fillmore's later words, it has "wastebasket" function (1971a:42). The term objective is undoubtedly a reflection of this more abstract part of the notion. As unwieldy and dissatisfying as such a dual definition is, our corpus nevertheless presents a certain amount of evidence to support it. We shall attempt to delimit the two aspects of the notion first before discussing the eligibility of each as PPP subject.

2.12.1

Concrete Patient

In concrete terms the notion of Patient takes on characteristic shape when contrasted with the notions of Goal and Location.

2.12.1.1 Goal-Target, for instance, involves something towards which a movement is directed but with which physical contact is not necessarily made. A Patient notion, however, automatically implies that physical contact is made [see also Lindkvist (1950:184;266;329)] . Compare, for instance: (163) (164)

John aimed at the bull John prodded at the bull

In (163) movement is directed towards the target but contact does not necessarily result; therefore, a Goal function can be ascribed to bull. In (164), howe\«r, there is likely to be contact, and a patient function can be ascribed to bull. Other corpus examples in which the 'contact1 aspect of Patient is particularly evident include:

75 (165) (166) (167) (168) (169)

If someone bumps against this table, the vase on it will fall and break A car ran into John Someone was hammering at the door The children trod ruthlessly on her flowerbed A burglar has tampered with this lock

In all of these, physical contact is made with the object. In addition, there nay be a note of abruptness or collision, although this is not a necessary feature.

2.12.1.2 In addition to the requirement of movement into physical contact, a concrete Patient notion also presupposes that sane perceptible effect take place as a result of the contact. This is the essential way in which Patient differs from Location and Goal. For example, in (170)

Peter fell to the ground

physical contact is made with the Object', but there is normally no perceptible effect made on the ground by falling. We would thus consider the entity in (17O) to be Goal. On the other hand, in (171)

Some child or other was always clinging to her skirt

we would assign the object Patient function since a skirt can normally be expected to bear the effect of clinging. In this connection, one is raninded of the distinction Fillinore (1970:125) makes between surface-contact and change-of-state verbs. The former "assert the occurrence of some physical contact between two objects, but ... one cannot necessarily infer that the objects have undergone any essential change." These surface-contact verbs occur with a place noun-phrase, which may sometimes appear as a locative prepositional phrase: I slapped him on the leg. Changeof -state verbs, on the other hand, "assert of an object a change in time from one 'state1 to another", as their name implies, and occur with an object nounphrase: I broke his leg. However, although Fillmore would presumably treat I slapped his leg as containing a surface-contact verb, we maintain that since slapping leaves its mark, leg in this sentence should be considered Patient. The essential criterion for distinguishing Patient from Location or Goal in this interpretation is whether there is any perceptible effect at all from the action.

76

To summarize, a concrete Patient is characterized by the features of movement into physical contact and perceptible effect. Admittedly, the two are not wholly independent of each other, since under normal conditions physical contact is necessary in order to bring about a change of state. But whether independent or not, they serve as useful criteria in determining the presence of concrete Patient.

2.12.1.3 Sentences in which a concrete Patient function can be attributed to the entity in object position are under normal conditions passivizable. Consider, for instance: (164)' The bull was prodded at ... (165)' If this table is bumped against ... the vase on it will fall and break (166)' John was run into by a car (167)' The door was being hammered at ... (168)' Her flowerbed was ruthlessly trod on ... (169)' This lock has been tampered with ... (171)' Her skirt was always being clung to by some child or other

Concrete Patient is thus wholly eligible as PPP subject.

2.12.2 Metaphorical Patient and Objective 2.12.2.1 In addition to these relatively clear-cut cases, the corpus contains a number of sentences with predicates of a less concrete nature. Some may be linked metaphorically to predicates with concrete Patient: A

(173) (174) (175) (176)

We hit on an excellent plan They blundered on an important book They chanced upon an abandoned well We met with a number of difficulties

Here, we might speak of metaphorical Patient. In many cases, however, the predicates are abstract and the entities must simply be assigned to Objective by default: B

(177) (178) (179) (ISO) (181) (182) (183)

Someone told on John Her grandparents have always fawned on her John waited on us graciously They can depend on John We decided on the boat ' People insisted on his innocence He came by the fortune late in life

77

2.12.2.2 However, one small but rather closely-knit group of predicates can be detected for which the entity in object position expresses a definable abstract notion, namely topic or subject natter. Illustrative sentences include: C

(184) (185) (186) (187) (188)

Human beings often speculate about the future of the race They constantly complained about the lack of facilities You shouldn't harp on misfortunes One must reflect carefully upon statements to the press People ought not to wrangle over trifles

This group, because it can be clearly delimited, could be handled as a separate notion Topic. However, we shall treat it as a subcategory of Patient since it behaves in a similar fashion with respect to the passive. 2.12.2.3 Sentences in all of the above-mentioned categories passivize, allowing abstract Patient to become the subject of a PPP: A

(173)' (174)' (175)' (176)'

An excellent plan was hit upon ... An important book was blundered upon ... An abandoned well was chanced upon ... A number of difficulties were met with ...

B

(177)' (178)' (179)' (18O)' (181)' (182)' (183)'

John was told on ... She has always been fawned on by her grandparents We were waited on graciously by John John can be depended on ... The boat was decided on ... His innocence was insisted on ... The fortune was come by late in life ...

C

(184)' (185)' (186)' (187)' (188)'

The future of the race is often speculated about ... The lack of facilities was constantly complained about ... Misfortunes shouldn't be harped on ... Statements to the press must be reflected on carefully ... Trifles ought not to be wrangled over ...

2.12.2.4 Thus, we can conclude that abstract Patient, in fact Patient in general, is a notion which is quite compatible with the passive in subject position. One or two exceptions, however, must be noted. Compare (174) and (175) above to the following sentences: (189) We chanced upon John (189) *John was chanced upon ... (190) We met with John by chance (19O)'*John was met with by chance ...

78

and (166) above to the following: (191) Someone ran into John (=meet by chance) (191)'*John was run into... (192) Someone bumped into John (=meet by chance) (192)'*John was bumped into ...

It appears that accidentally meeting a person cannot be expressed in the passive, whereas accidentally 'meeting' a thing can. A number of explanations can be advanced to explain this phenomenon. Tuyn (1970:6O), for instance, notes that the active counterpart of He was met can only be 'going forth with the express purpose of seeing a person1 and not 'coming across a person casually, by chance". However, he makes the somewhat counterintuitive claim that this is because the former meaning expresses activity, whereas the latter does not [sic]. Nor does the theory that the non-volitional element is responsible for the constraint stand up to scrutiny, since non-volitional meet with an object, chance upon an object, etc. are all passivizable. A somewhat more likely explanation on semantic grounds might be that if two persons are involved in an accidental meeting, the action is symmetric or reciprocal and a passive is blocked for this reason (see III.2.3) 14 . However, if this is the case, it is not particularly clear why this kind of reciprocity should not be capable of principality, since there could quite conceivably be differing degrees of participation in an accidental meeting (e.g., one person could be looking the other way, etc.). In the end, it may be that the explanation is not to be sought in semantics at all, but rather in lexical choice. We shall return to a discussion of this aspect shortly (IV.2.3). For our purposes at the moment, it is sufficient to state that a Patient notion in general is quite eligible for subject position in a PPP.

2.13

Reason

Reason (Reas) can be ascribed to entities which stand in a because relation to the action or state expressed by the verb. In contrast to as we shall see, occurs exclusively with psychological verbs subject position; cf II.2.2), Reason appears most frequently gical change-of-state verbs (Objective in subject position). 14

Stimulus, which, (Experiencer in with non-psycholoFor instance:

That is, the two entities involved can be interchanged without affecting meaning, e.g. Peter chanced upon Paul = Paul chanced upon Peter.

79

(193) (194)

She blushed for shame He exploded with jealousy

However, it may also occur with Agentive, as in: (195) (196)

He screamed in anger He yelled with pain

Likewise in the sentence: (197)

The child was crying (=weeping) for more cake

the entity in object position does not indicate the goal of the action

(x do y in order to get z) but the reason instead: χ 'do' y because p. The latter element in this case would be the subordinate proposition χ wants z. As the following demonstrate, a Reason notion is not compatible with the passive in subject position: (193)'*Shame was blushed for ... (194)'*Jealousy was exploded with ... (195)'*Anger was screamed in ... (196)'*Pain was yelled with ... (197)'*More cake was being cried for ...

jReason thus does not meet the requirements for PPP subjecthood.

2.14

Replacive

J. B. Walmsley (1971:496) proposes Replacive (Rep) as the case "expounded typically by instead of". Other case markers,he continues, might include for or on behalf of. We shall use the term here to express the concept of substitution or exchange [see also Wood (1967:35)]. Only one example is documented in our corpus: (198)

John deputized creditably for Mr. Sims

but any number of sentences can be constructed with a for χ phrase in a Replacive function: (199) (200) (201)

[If you don't want to go7 John will go for you [If you don't know the answer; John will answer for you iSince Tom is injured./ John will run for him

In addition, there are a number of examples in which a metaphorical Replacive can be identified: (202)

We must account for this loss

00

(203) (204)

You can never atone for this fault John must answer for Bob (=be responsible for)

In the passive, examples with concrete Replacive are of varying acceptability: (198)' Mr. Sims was creditably deputized for ... (199)'*You will be gone for ... (2OO)'*You will be answered for ... (2Ol)'*He will be run for by John

This phenomenon may be due to lexical constraints; see IV.2.4. On the other hand, metaphorical Replacive appears to be generally acceptable in the subject position of a PPP: (2O2)' This loss must be accounted for ... (2O3)' This fault can never be atoned for ... (2O4)' Bob must be answered for by John

Although more data would be helpful, we can conclude temporarily - based on the acceptable concrete and metaphorical passives - that Replacive is in general eligible for PPP subject position. 2.15

Source

The counterpart of Goal, Source (Sour) is described by Fillmore (1971a:41) as earlier location, earlier state or earlier time point. Other grammarians have used the term to refer to 'starting point with verbs of motion1 (Frajzyngier 1975:353) or 'starting-point of a change of position' (Bennett 1975:18). 2.15.1 In our corpus Source is a decidedly less frequent phenomenon than Goal in object position. However, the following are representative of some of the instances documented: (205) (206) (207) (208) (209) (210) (211)

A number of unsuspecting people have fallen from this roof A goddess ascended from the waters More and more people emigrate from this country every year No one has ever escaped from this jail We departed from him with regret You must withdraw from society in order to appreciate it They haven't heard from Martha recently

In addition, there are a number of cases which could perhaps be treated as metaphorical Source: (212) (213)

No one has ever departed from these standards We can't really generalize from this report

81 (214) (215)

You must refrain from such behavior He will never recover from the loss

2.J5.2 Vflien passivized, nost of these examples with concrete Source are acceptable: (2O5)' (2O7)' (2O8)' (2O9)' (21O)' (211)'

This roof has been fallen from by a number of unsuspecting people This country is emigrated from ... more and more every year This jail has never been escaped from ... He was departed from ... with regret Society must be withdrawn from ... in order to be appreciated Martha hasn't been heard from recently ...

Based on this evidence alone, we can conclude that Source is compatible with the passive in subject position. Notice, however, that in most of these sentences, Source has an instrumental connotation. The roof is such that it is instrumental in the fall, the country such that it is instrumental in the emigration, the jail such that it is instrumental in the escape, etc. Only (209) and (211) do not lend themselves to such an interpretation, perhaps because the subjects are human beings rather than inanimate objects and, as metaphorical sources, less likely to take on instrumental character. However, it appears that either an instrumental or an affected interpretation is an essential condition which concrete (and non-metaphorical) Source objects must meet in order to become PPP subjects. Thus in (206) above, the waters are indeed the source of the action, but it is difficult to imagine them as instrumental, i.e. such that they could be 'used1 in ascending, or as affected, i.e. undergoing a change of state as a result of the action. A passive is thus of doubtful acceptability: (2o6)'?The waters were ascended from by a goddess On the other hand, in a sentence like (216) John rose from the chair so suddenly that it fell over (216)' The chair was risen from so suddenly that it fell over

the chair is clearly affected by the action and Source can consequently assume subject position. A stipulation similar to that for Location must consequently be made for concrete Source, i.e. an instrumental or an effect connotation must be present. (A goal connotation is excluded in the case of Source.) On the other hand, the acceptability of metaphorical Source in subject position does not appear to

82

depend on such a condition: (212)' (213)' (214)' (215)'

These standards have never been departed from ... This report can't really be generalized from ... Such behavior must be refrained from ... The loss will never be recovered from ...

Although (213)'has a slight instrumental connotation, it is doubtful whether PPP acceptability depends on such an interpretation in every case. In general, metaphorical Source is eligible for subject position in the passive. 2.15.3 Several examples in the corpus involve a coupling of Source and Goal, either in surface structure, as in: (216)

They lapsed from virtue into vice

or at a deep semantic level, as in (217) (218)

Several large ships ply between Glasgow and London People have traded between London and ports in the Mediterranean for centuries

These are all doubtful or unacceptable in the passive: ( 2 1 6 ) ' *Virtue was lapsed from ... into vice ( 2 1 7 ) ' ?Glasgow and London are plied between by several large ships (218)' *London and ports in the Mediterranean have been traded between ... for centuries

Whether because breaking up a pair leaves one member dangling (216) or because perhaps the relation is inherently symmetric (217, 218), we can conclude that Source + Goal as a pair, or separate (if the other member is present), are ineligible for subject position in the passive. 2.16

Stimulus

In his 1971 model, Fillmore introduces the term stimulus to account for " ... in the case of a psychological predicator — the thing reacted to" (1971a:42). Although he would treat this notion as a subcategory of Instrument, we shall discuss Stimulus (Stim) independently, as it behaves somewhat differently with respect to the PPP. One of the essential criteria for presence of Stimulus in object position is the co-occurrence of a psychological verb. That is, Experiencer must appear in subject position. The following corpus examples illustrate a Stimulus notion in object position:

03

(219) (220) (221) (222)

We could only wonder at his success They quailed at the mere prospect People frequently bridled at her remarks The whole family rejoiced in her success

It is important to distinguish Experiencer in subject position from Objective, which, according to FiUmore (1971a:42), accompanies a non-psychological verb indicating a change of state (e.g. die or grow). For instance, in (223) (224)

She sickened at the sight of slaughter He collapsed under the weight of the snow

since sicken and collapse involve changes of state, the entities she and he are Objective and we must assume that sight and weight fulfil a Reason rather than a Stimulus function. Precisely this distinction proves to be crucial in determining PPP acceptability. Whereas the Stimulus examples produce acceptable passives: (219)' (22ΟΓ (221)' (222)'

His success could only be wondered at The mere prospect was quailed at ... Her remarks were frequently bridled at ... Her success was rejoiced in by the whole family

those with Reason do not: (223)'*The sight of slaughter was sickened at ... (224)"*The weight of the snow was collapsed under

...

Based on this evidence, we can conclude that a Stimulus notion, in contrast to Reason, is eligible for subject position in PPPs. 2.17

Time

Although often considered to be merely a temporal variety of Location (Fillmore 1971a:41; Bennett 1975:15f), Time requires separate treatment if semantic notions are to be used in accounting for the PPP. Thus, although Location is conditionally eligible for subject position in a passive sentence, Time is wholly ineligible as PPP subject: (225) Birds breed in the spring (225)'*The spring is bred in by birds (226) Congress will recess on Thursday (226)''Thursday will be recessed on by Congress (227) The revellers unmask at midnight (227)'*Midnight is unmasked at by the revellers

We can conclude that Time does not fulfil the condition for subjecthood in a PPP.

84

3

Object notion and subject eligibility

3.1 Figure 2 presents a tabular sunroary of the correlation between semantic notions in object position and eligibility as PPP subject: Figure 2

CORRELATION OF SEMANTIC NOTION WITH ELIGIBILITY AS PPP SUBJECT

Eligible

Eligible on condition

Ineligible

Agentive Benefactive Comitative Direction

χ

Experiencer

X

Extent

X

Goal Basic Ending-po int Target Direction Objective Instrument Location Manner Path Patient Reason Replacive Source Stimulus Time

χ X X

05

In interpreting this table, severed, points should be recalled: a) The decision as to eligibility is based on whether at least one acceptable PPP can be found with the given notion in subject position. Thus, in cases of eligibility, the presence of a notion should be regarded as minimal fulfilment of the condition for subjecthcod in a PPP. Depending on the notion, there may be any number of additional conditions operant. When these additional conditions appear to be of a definable semantic nature, the notion has been entered in the column 'Eligible on condition1. Obviously, however, the conditions may be quite different from one notion to another. b) In addition to considerations of a semantic nature, there are a number of other constraints operant on the PPP, e.g. lexical, contextual,etc. These have been ignored completely here for methodological reasons but will be discussed once the semantic investigation is completed.

3.2 As apparent from Figure 2,seniantic notions in object position demonstrate different degrees of eligibility for subject position in passive sentences. Whereas some are wholly ineligible, others are wholly or partially eligible. It is thus obvious that role determination is crucial to passivizability. Vihether, for instance, an entity is simple Location, e.g. (normally) stand on the sidewalk, or a potentially instrumental or affected location, e.g. stand on the stooly affects the passivizability of the predicate: (228)'*The sidewalk was stood on ... ( 2 2 9 ) ' The stool was stood on ... (to reach the cookies)

Such constraints are even nore evident in sentences which are potentially ambiguous in the active. Consider, for instance: (23O)

If you look over the top of the mountain, you will see a red cloud

The preposition over in this sentence can have at least two different meanings. On the one hand, it can mean (i) 'directly above1 and top of the mountain would be Location. The cloud is consequently high in the sky. On the other hand, over can niean (ii) 'above and to the other side', in which case the cloud is more likely to be in the valley. Because looking above and over an object involves, metaphorically, crossing a boundary, it is more appropriate to consider top of the mountain in this sense as Goal. The passive, however, allows only Goal to become its subject:

86

(23Oi)'*If the top of the mountain is looked over ... will be seen (=directly above)

a red cloud

(230ii)' If the top of the mountain is looked over ... a red cloud will be seen (=above and to the other side)

That is, the passive has the effect of filtering, or disambiguating potential ambiguity. The filter function of the PPP can be observed in other instances as veil. Consider, for instance: (231)

Someone was hammering at the door

Here the choice is between Location and Goal; door can refer either to the place of the action (hammer where?) or to the target of the action (hammer at what?). The passive, however, allows only a Goal interpretation: (231i)' The door was being hammered at (=what?) (231ii)'*The door was being hammered at (=where?)

Likewise in (232)

The searchers cried for the child

we could attribute either a Goal-Objective or a Reason function to child. In the former, the searchers would be shouting; in the latter, weeping. But the passive filters out a Reason notion in subject position: (232i)' The child was cried for .... (=shout) (232ii)'*The child was cried for ... (=weep)

And finally, consider: (233)

He was gaining on them

Here the ambiguity is likewise rooted in the dual meaning of the verb. With the 'get closer to1 meaning of gain on, them would be interpreted as Goal-Objective. In its

"go faster than, get farther in advance of meaning, however, them would

be interpreted as the relative standard against which one's speed or progress is measured. That is, it would no longer be Goal, but depending on the treatment accorded comparatives, perhaps Agentive in a second proposition adjoined to the major comparative one. Regardless, the passive acts once again as a filter, allowing only Goal to assume subject position: (233i)' They were being gained on (=get closer to) (233ii)'*They were being gained on (=go faster than)

87

3.3 Summarizing our results so far, the semantic notions which appear in the object position of active corpus sentences fall naturally into three groups with respect to subject eligibility. A first group (A) consists of notions which are incompatible with the passive in subject position: Agentive, Direction, Experiencer, Extent, Manner, Reason, Time. A second group (B) is composed of those notions which are compatible with the passive in subject position: Benefactive, Goal (except Ending-point), Instrument, Patient, Replacive, Stimulus. And a third group (C), to which Conitative, Ending-point Goal, Location, Path and Source belong, represents those which are conditionally eligible for PPP subject position. Taking a closer look, however, it becomes apparent that the notions in group C involve different conditions for eligibility. Whereas Ending-point Goal, Location, Path and Source are subject to the same or similar conditions, Comitative is subject to other conditions. The stipulations which apply for Ending-point Goal, Location, Source and Path recall in many ways features which are inherent in other notions compatible with the passive. Thus, in order for Location to be eligible for subject position, a goal, instrument or effect connotation must be present; likewise, in order for Ending-point Goal, Path and Source to became eligible, an instrumental or effect connotation must be present. Placing these subcategories in the Eligible column, the list of notions compatible with the passive in subject position becomes: Eligible:

Benefactive Comitative-Principality Ending-point Goal-Instrument

Ending-point Goal-Effect Goal Instrument Location-Goal Location-Instrument Location-Effect Path-Instrument Path-Effect Patient Replacive Source-Instrument Source-Effect Stimulus

A good deal of the redundancy in this list can be avoided by re-distributing the Locative, Ending-point Goal, Path and Source subcategories as follows: LocationGoal to Goal, Location-Instrument to Instrument, Location-Effect to Patient, etc

This produces the following list of notions eligible for subject position in the passive: Eligible:

Benefactive Comitative-Principality Goal Instrument Patient Replacive Stimulus

Taking into consideration now the relative frequency of these various notions, it becomes possible to draw a number of conclusions concerning the semantic nature of PPP subjects. By far the nost frequent occurrences in subject position are Goal, Patient and Instrument. These notions can be considered the semantic core of PPP subjects. The remaining notions are of minor importance judging from frequency of occurrence. Based on these conclusions, the semantic domain of passive subjects extends somewhat further than has been thought to be the case until now. Bolinger (1974:67), for instance, asserts that "true patient" is the one and only semantic requirement for all types of passivization. Thus he states: "If the grammatical object in the active construction is not conceived as a true patient, there will be no corresponding passive". Notice that even if we were to ignore the minor occurrences of other notions, this theory of the passive would still conflict with the major findings of our investigation, namely that Goal, Patient and Instrument are all sufficient to fulfil the condition for subjecthood. This, of course, raises the question of whether Goal and Instrument should not be regarded as deriving from, or as dependent upon Patient. Does goal-directed activity perhaps automatically imply effect? Does instrumental use presuppose effect in one way or another? If so, it could be argued that goals and instruments are also patients, and a claim such as Bolinger's would seem more likely. It is interesting to note that of late Fillmore apparently does assume such dependency in the case of instrument and patient (1977:75). A patient is now described as "the thing which gets manipulated". Does this not amount, in simple terms, to saying that patients are objects used instrumentally? If so, there is a certain amount of precedence for conflating patient and instrument, if not patient and goal notions into one. However, there is a point at which the conflation of senantic notions brings

diminishing returns. For one, a global statement of the sort that only Patient can become P?P subject ignores the fact that other 'minor1 semantic notions such as Benefactive, Conitative, etc. have been documented in subject position. Further, it amalgamates three distinct, major semantic notions into a nebulous whole which it is difficult to be specific about. In the last analysis, conflation may lead to more confusion than clarity concerning the semantics of passive subjects. It is for these various reasons, and others which will become evident in following chapters, that we shall consider there to be three major and several minor semantic dimensions to PPP subjects.

IV

PPP ACCEPTABILITY

1

Semantic-syntactic configuration

Our analysis has consisted so far in a detailed investigation of the various semantic notions appearing in subject and object position in corpus sentences with a view to determining which are eligible for PPP object and subject positions respectively. As pointed out earlier, however, the results obtained so far do not yet enable us to predict whether a given active sentence will have an acceptable passive or not. It is the conbination of notions in a sentence which determines its semantic configuration and therefore whether it meets the semantic-syntactic conditions for acceptability as a PPP. It is thus to an integration of our findings in II and III that we turn our attention now. Figure 3 indicates in diagranmatic form how the various notions identified in subject and object position of active sentences combine with each other in the corpus examples cited. And at the same time it shows which of these configurations are basically passivizable, and which are not. In interpreting Figure 3, the following points should be noted: i)

A subject-object semantic configuration has been termed 'basically passivizable1 if one acceptable passive can be found for an active sentence with this configuration.

ii)

iii)

No distinction is made here between 'Eligible* and 'Eligible on condition1. Subcategories of Location, Ending-point Goal, Path and Source which are eligible as passive subject have been re-distributed to Goal, Patient or Instrument accordingly.

91

Figure 3

NOTIONAL CONFIGURATION (SUBJECT-OBJECT) AND PASSIVIZABILITY

ACTIVE

ACTIVE

SUBJECT

OBJECT

Agentive

Experiencer

Instrument •••&=

Location

Time

Objective

Basically passivizable Basically non-passivizable

Using Figure 3 it is possible to determine whether a given configuration meets the semantic conditions for passivization. For instance, an Agentive subject + Benefactive object configuration is basically passivizable [e.g. (1)', p. 45], whereas an Objective subject + Experiencer object is not [e.g. (22)', p. 48]. Likewise, an Experiencer subject + Stimulus object configuration is passivizable [e.g. (219)^ p. 83 ], but an Instrument subject + Goal object is not [e.g. (2o)', p. 34], etc.

92

In addition to illustrating which subject-object combinations are basically passivizable and which are not, Figure 3 also demonstrates why the eligibility of a subject notion or an object notion alone is not sufficient to predict passivizability. Although according to our findings in Ch. II (see Fig. 1, p. 39), for instance, Agentive subject is eligible for object position, Figure 3 shows that, were it to occur with, say, Direction object, no passive would be possible. The same applies for the combinations Agentive + Extent, Agentive + Manner, Agentive + Reason and Agentive + Time. Likewise, although according to our findings in Ch. Ill (Fig 2, p. 84), an Objective/Patient object is eligible for subject position in the passive, PPP acceptability cannot be predicted on the basis of its presence alone, since if Objective/Patient were to appear with Location or Instrument, a passive would not be acceptable. Figure 3 can also be interpreted as a statement of which notional combinations are typical of the active and the passive. Figure 4 shows active semantic configurations (based on corpus examples cited):

93

Figure 4

NOTIONAL CONFIGURATIONS OF ACTIVE

ACTIVE SUBJECT

ACTIVE OBJECT

Time

Notice that every notion but one (Agentive) is capable of appearing as active object, whereas active subject is more restricted semantically. Figure 5 shows passive semantic configurations:

94

Figure 5

NOTIONAL CONFIGURATIONS OF PASSIVE

Agentive > •Benefactive

PASSIVE SUBJECT «^

Location

Λ

PASSIVE OBJECT

Here the relation of more restricted vs less restricted is reversed. It is passive object which is more limited semantically than passive subject. We can conclude that, given an unordered set of semantic notions, one type of subject-object configuration characterizes active sentences, another passive sentences. These types of possible configurations constitute a semantic specification of active and passive respectively. There are of course a number of compatibility constraints operant. Thus, according to Fig. 4, Experiencer may appear either as active subject or active object. However, if it is chosen for the former, then it cannot re-appear in the latter. The same applies for Instrument. This type of compatibility constraint could conceivably be handled by the case grammar rule that only one instance of a case may occur in a simple sentence (Fillmore 1968b:21). However, there are other compatibility constraints involved. Figure 6 demonstrates these in active sentences (based on cited examples):

95

Figure 6

NOTIONAL COMPATIBILITY IN THE ACTIVE

ACTIVE

SUBJECT

ACTIVE

OBJECT

Agentive

Experiencer

Instrument

Location

Objective/Patient

Figure 6 indicates, for instance, that if Agentive is chosen as subject of an active sentence, Experiencer cannot be chosen as object. Or that if Experiencer appears as subject, only Stimulus may appear as object, etc. Now consider the compatibility patterns for the passive (based on cited examples):

96

Figure 7

NOTIONAL COMPATIBILITY IN THE PASSIVE

PASSIVE

SUBJECT

PASSIVE

OBJECT

Benefactive

Comitative

Goal

~~^^^

Agentive

Instrument

Objective/Patient

^^

^^- Experiencer

Replacive

Stimulus

Here there are two semantic axes, one formed by Agentive object and all notions eligible as passive subject except Stimulus, the other formed by ExperiencerStitnulus. Since neither Experiencer nor Stimulus enters into configuration with other notions, the two axes are parallel. If, as proposed in II.2.2., Experiencer in active subject position is treated as a sort of metaphorical extension of Agentive (and Stimulus as a metaphorical version of, say, Instrument), then the semantic characterization of the passive can be simplified to Agentive object and Benefactive, Comitative, Goal, Instrument, Objective/Patient or Replacive subject. A comparison of Figures 6 and 7 will show that there are no left-right configurations which are identical in active and passive. A PPP can thus be distinguished semantically fron the active on the basis of subject-object notional configuration alone. In a language model for the generation of PPPs, we would probably assume that a state or event is analyzed in terms of a number of (unordered) semantic

97

notions, and that this set is subsequently ordered, depending on its composition, into subject-object configuration. If an event were analyzed into, say, [+Ag, +Loc], there would be only one choice for ordering, namely: SUBJECT

OBJECT

Agentive

Location

This would be an active sentence. If, however, an event were to analyze into [+Ag, +Pat], there would be two ordering possibilities: SUBJECT

OBJECT

Agentive

Patient

Patient

Agentive

The former would be an active sentence, the latter a passive one. Presumably a speaker would choose one or the other based on textual considerations.(How the passive option relates to the active, i.e. whether it is derived or formed directly, will be a matter for consideration later.) This model should make clear that the first step in predicting PPP acceptability is to determine whether the given constellation of notions, i.e. the proposition, is basically passivizable or not. Once this has been accomplished, other considerations may cone into play. It is to these that we turn our attention now.

2

Lexical choice

As Bolinger (1974:58) points out, the PPP has traditionally been explained in terms of lexicalization. Indeed, the 'cohesionist1 theory that certain verb + preposition combinations adhere more closely than others amounted basically to attributing lexical status to verb + preposition collocations which are closely connected and calling those which are not 'free combinations'. Although we reject the idea that lexicalization is primarily responsible for the PPP and have shown that the factors which influence it are above all semantic, the data do indicate that lexical considerations may play a secondary role in determining PPP acceptability. This becomes obvious when lexical substitutions are made within one and the same sanantic framework. Consider, for instance, three prepositional varieties of sentence (1) below, all having the same semantic configuration, Experiencer + Stimulus:

90

(1)

Girls don't like him

(2) (3) (4)

Girls don't care for him Girls don't take to him Girls don't go for him

Since this notional configuration is documented in the passive and furthermore since (1)'is perfectly acceptable: (1)' He isn't liked by girls we might expect the same of ( 2 ) ' , ( 3 ) ' , and (4)'in the passive. However: ( 2 ) ' ?He isn't cared for by girls (3)' *He isn't taken to by girls ( 4 ) ' *He isn't gone for by girls

In order to account for this type of phenomenon it is necessary to appeal to the concept of lexical choice. That is, we must allow for the fact that, although the semantic configuration in sentences is identical, the choice of one lexical item rather than another may influence PPP acceptability. In practice, the effect of lexical choice on the PPP is often unpredictable. However, at least four systematic types of influence can be detected.

2.1

2

Idion

It is curious to note that most traditional grammarians who attempt to explain the PPP in terms of 'cohesion1 neglect to mention the fact that too much lexical cohesion will block the passive. In this respect, Poutsma (1926:33) is an exception when he notes, with reference to the PPP, that "idiom not seldom stands in the way of this practice".

It is for this reason that dictionaries such as the one at the conclusion of this study are of value. This is especially true for metaphorical and/or abstract notions. Thus, American informants were unable to make a reliable judgment about the passivizability of B.E. sicken for a disease, although they had no trouble predicting (accurately) that B.E. troll for pike is basically passivizable. His examples are quite misleading, however: The ship abounds in conveniences The book belongs to me Surely neither sentence can be said to be refractory to the passive because of idiomaticity.

99

The fact that idicmatic expressions do not lend themselves to passivization is amply documented in our corpus. Consider, for instance: (5) She's just crying for the moon ( 5 ) ' *The moon's just being cried for

...

(6) John beat about the bush ( 6 ) ' The bush was beaten about ... (7) You're barking up the wrong tree ( 7 ) ' *The wrong tree is being barked up ... (8) He was asking for trouble ( 8 ) ' 'Trouble was being asked for ...

The question may of course justifiably be raised as to what constitutes an idicmatic expression. For instance, it is just as likely that we would find a lexical entry for (i) run into difficulties as for (ii) run for one's life, And yet the former must be considered an example of metaphorical use (which in this case can be passivized): (i)

(9) Should we run into difficulties, we'll contact you ( 9 ) ' Should difficulties be run into ... you'll be contacted

whereas the latter must be termed idicmatic use (which cannot): (ii)

(1O) He ran for his life (lO)'*His life was run for ...

For the purpose of this study, idiomatic expressions will be those verb + preposition + noun collocations which permit little or no lexical or syntactic variation for a given verb meaning. Thus, for (i) above, not only are other nouns possible with the same verb meaning, e.g. run into danger, debt, etc. but the nouns may \ary in terms of modification and/or quantification. Thus: i. run into

any difficulties a few difficulties serious difficulties a number of serious difficulties etc.

whereas life in (ii) is highly restricted as to the type of variation it can undergo: ii. , run for

one's life *a life *any life , J .. , dear life *a dear life *good life

100

Defined this way, idionatic expressions can be said to systematically constrain passivization of prepositional verbs.

2.2

Infrequent use

At the other end of the lexical scale, a particular verb or verb + preposition collocation may be infrequent or unfamiliar in the active. If so, although the active may be tolerated, a passive is likely to be doubtful. Thus: (11) A certain Dr. Meyer disputed against tolerance (11)' ?Tolerance was disputed against by a certain Dr. Meyer (12) The whole family joyed in her success (12)' ?Her success was joyed in by the whole family (13) The hotel which we abode at has long since disappeared (13)' ?The hotel which was abode at has long since disappeared

Compare, however: ( 1 4 ) ' Tolerance was argued against ... (15)' Her success was rejoiced in ... (16)' The hotel which was stayed at ... A coiparison of this type of lexical constraint with the one above (IV.2.1) indicates that both too much and too little lexical 'predictability1 are incompatible with the PPP.

2.3

Metaphorical use

Numerous lexical predicates in the corpus may have either literal or metaphorical reference. For instance: (17) (18)

They broke into the house They broke into the sovereign

(19) (20)

John walked into the wall John walked into the pie

(21) (22)

She sat on the egg for three weeks She sat on the committee for three weeks

Likewise dive into water vs dive into a wallet, kick at a ball vs kick at ill treatment, run into something (=collide) vs tally) , etc.

nm into someone (=meet acciden-

In these cases, sentences with metaphorical predicates are often blocked in the passive:

101

(18)' *The sovereign was broken into ... ( 2 O ) ' *The pie was walked into ... ( 2 2 ) ' *The committee was sat on ...

A passive can only occur if the lexical verbs are understood literally: ( 1 7 ) ' The house was broken into ... ( 1 9 ) ' The wall was walked into ... ( 2 1 ) ' The egg was sat on ... for three weeks

This phenomenon can also be observed in synonymous pairs where one item is metaphorical and the other not: (23) We all (24) We all ( 2 3 ) ' *He was ( 2 4 ) ' He was

felt with/for him sympathized with him felt with/for by us all sympathized with by us all

(25) The passengers took to the lifeboats (26) The passengers fled to the lifeboats ( 2 5 ) ' *The lifeboats were taken to ... when ( 2 6 ) ' The lifeboats were fled to ... when

(when the torpedo hit) (when the torpedo hit) the torpedo hit the torpedo hit

Here too a passive with the metaphorical lexical item is blocked. Of course, there are well-known counter-examples to this rule of thumb. For instance: (27)'

That problem can be gone into thoroughly vs ( 2 8 ) " *The room was gone into at once fSvartvik 1966:1657

or

(29)'

The expected result was arrived at vs (3O)' *The splendid stadium was arrived at

fQuirk et al 1972:8o47

Here it appears that only the metaphorical verb passivizes. However, in terms of our criteria, these are not counter-examples at all, since it is quite possible to say: (28.1)' This room mustn't be gone into (the floor has just been waxed)

or ( 3 O . 1 ) " The scene of their exploits was arrived at by way of Gibraltar

That is, the predicates go into and arrive at in their literal meaning are basically passivizable too. Within the framework set up here, such examples merely demonstrate that at times both literal and metaphorical use of a predicate are possible in the passive.

102

2.4

Complex structure

The notion of complex lexical structure as it will be used here was introduced by Fillmore (1971 a:48) with respect to the verb swim. A sentence such as t May I swim in?, he maintains, requests simple permission to enter, i.e. focus is on the motion per se. A sentence such as May I swim in?, however, requests permission to enter in a particular way, i.e. swimming. Here focus is on the manner rather than the motion. Certain verbs have complex lexical structure (in this case swim: motion + manner), whereas others have simple lexical structure (e.g. come or go: motion). Ihe concept of ocmplex vs simple lexical structure explains a number of otherwise puzzling facts about the PPP. Consider, for instance, the following set of examples (hypothetical in part) containing prepositional verbs associated with Goal in object position:

(31) They

went walked ran raced skipped . across the street hopped zigzagged staggered tottered waddled

It is obvious that with respect to Goal, go has the most simple structure: it is expressive of the most direct relation between predicate and goal. On the other hand, verbs like totter or waddle are complex in that they do not express simply reaching a goal but put special enrphasis on the way the goal is reached: GO + manner. Now consider the same sentences in the passive:

(31)'

The street was

gone across walked across run across raced across ?skipped across ?hopped across ?zigzagged across ?staggered across ?tottered across Pwaddled across

In general, those verbs which are strongly suggestive of manner are less acceptable in the passive than those which indicate mere reaching of a goal.

103

The same phenomenon can be observed with respect to Benefactive objects. Consider, for instance:

(32)

A number of people

worked travelled purveyed designed wrote cleaned

for the company

Here too there is a gradient with respect to how simple or direct the relation is between predicate and benefactor. The most neutral benefactive relation to a company is perhaps that described by work, one of the more indirect surely clean (unless the company deals exclusively in maid service). And this fact is reflected in passive acceptability:

(32)'

The company was

worked for travelled for purveyed for ?designed for Pwritten for *cleaned for

The same phenomenon is arguably at work with Replacive objects as well, where only the lexical item expressive of the purest Replacive relation is acceptable in the passive (see also III.2.14):

(33)'

John was

deputized for *gone for *answered for *run for

And finally a similar type of lexical constraint can be observed with Instrument objects. Solinger (1974:75) also notes that only certain Instruments are acceptable as PPP subjects but attributes it to "predictability" and "custom of use". That is, he claims that certain lexical associations - e.g. "write with pen, plow with plow, eat with knife and fork" - are more likely to occur than others - e.g. "steal with burglar's tools, walk with cane, and open with can-opener", and that corresponding instrumental passives are unacceptable where such predictability is lacking: i. ii. iii. iv. v. vi. vii. viii.

My knife is being cut with right now ?My shears are being cut with right now My pen is being written with right now *My burglar's tools are being stolen with right now My plow is being plowed with right now *My cane is being walked with right now My knife and fork are being eaten with right now *My can-opener is being opened with right now

104

However, Bolinger's examples are somewhat invalidated by the fact that in at least three of the four cases of constraint, the corresponding active sentence is elliptical, i.e. the verb and preposition occur together only because the direct object has been omitted. Since object ellipsis is unusual with these verbs (although of course not with all verbs), the active sentences thanselves are hardly well-formed. Thus: (34) ?Someone is cutting with my shears right now (cutting what?) (35) ?Someone is stealing with my burglar's tools right now (stealing what?) (36) *Someone is opening with my can-opener right now (opening what?)

On the other hand, (37)

Someone is walking with my cane right now (*walking what?)

is not elliptical in the active, ana an explanation must be sought for its lack of acceptability in the passive. Consider now a selection of sentences with Instrument object from our corpus: (38) (39) -(4O) (41) (42) (43)

You can You can You can You can You can You can

write with this pen signal with this handkerchief eat with this knife and fork swim with this mask talk with this microphone walk with this stick

Notice that the criterion of simple vs complex lexical structure applies here as well. That is, in each case we can determine how simple or direct the relation is between the predicate and the instrument. For instance, the instrument relation between a pen and writing is more direct than that between a mask and swimming. Expressed another way, a pen is essential to writing in a way that a mask is not to swimming. (The former enables writing to take place; the

lat-

ter merely facilitates vision while swimming.) That is, whereas write with a pen expresses a simple USE relationship, swim with a mask or walk with a stick expresses a USE + manner relationship. Therefore, in the passive it is not surprising to find: (38)' (39)' (4O)' (41)' (42)' (43)'

This pen can be written with ... This handkerchief can be signalled with ... This knife and fork can be eaten with ... *This mask can be swum with ... *This microphone can be talked with ... *This stick can be walked with ...

Generally speaking, a predicate with simple lexical structure is more likely to passivize than one with complex structure [cf, however, IV. 4]. The lack of lexical predictability in this acceptable PPP constitutes counter-evidence to Bolinger's claim that low predictability correlates with nonpassivizability.

105

3

Contextualization

Our discussion so far has shown that it is above all the semantic-syntactic configuration of active sentences which determines whether they are basically passivizable or not, but that lexical choice may influence PPP acceptability as well. It became apparent in the course of corpus testing, however, that other factors often determine whether a given 'contextualization' of a basically passivizable semantic-syntactic configuration is likely to be acceptable or not. Svartvik (1966:21) too notes this phenomenon: In a poll ... the minimal same sequence differently to for help by the people industrial areas produced

sentence The girl was turned to ... and the contextualized The Prime Minister was turned suffering from depression in the north-eastern very different results.

Regardless of the fact that this particular example is ill-chosen, different contextualizations of basically passivizable predicates do vary in degree of acceptability. In this section we shall discuss sane of the factors involved in contextualization which appear to affect PPP acceptability - based on corpus testing and suggest tentatively how they might do so. Our observations must be considered provisional, however, since they are merely by-products of the semanticsyntactic analysis. They should be understood above all as pointing to aspects of the PPP which call for more thorough investigation. 3.1

Tense, modality, aspect

The semantic-syntactic approach we have developed here concerns, in case grammar terms, the prepositional component of sentences, i.e. the "tenseless set of relationships involving verbs and nouns" (Fillmore 1968b:23). In addition, however, according to Fillmore's model, sentences have a 'modality1 component, which includes "such modalities on the sentence-as-a-whole as negation, tense, mood, and aspect". For terminological reasons, we prefer to call the features to be discussed in this section, - i.e. tense, modality and aspect-part of the

5

In fact, two different meanings of turn to are involved here. In the first sentence, turn to would normally be understood as referring to a change of body position, whereas in the second it refers to an appeal. According to our findings, metaphorical turn to is basically passivizable, but literal turn to is generally not (unless a Goal or Instrument connotation is present; cf III.2.4) .

106 verbal/modal component of sentences. 6 As indicated at the outset of this investigation, quite a few active corpus sentences, passivized as such, yield doubtful, if not unacceptable, PPPs. Thus: ( 4 4 ) We relaxed in the lounge (44)'?The lounge was relaxed in (by us) (45) John walked on the floor (45)' ?The floor was walked on (by John)

Even deletion of the £>z/-phrases (which are optional) does not greatly improve the acceptability of these sentences. However, if different choices are made in the verbal/modal constituent, we obtain more acceptable PPPs: ( 4 4 . 1 ) ' The lounce can be relaxed in ... ( 4 5 . 1 ) ' The floor has been walked on ... Whether verbal/modal features alone can be made responsible for this apparent constraint, however, is debatable. If we assume that the prepositional content in (44) and (45) has not changed in (44.1)', and (45.1)', then the acceptability of (44.1)'and (45.1)'as opposed to (44)'and (45)'appears to depend solely on verbal/modal considerations. If, however, as we have proposed, the active version of (44.1)', i.e. (44.1)

You can relax in the lounge

is given two alternative semantic configurations, namely [+Ag +Loc] and [+Ag + Instr], only the latter of which is basically passivizable, then (44.1)'and (45.1)' have a different semantic configuration from (44)' and (45)', and this is responsible for differing acceptability. In this case, the verbal/modal changes may only accompany propositional change. In order to be consistent with our proposal in III.2.9.4, we shall opt for the latter view. This should be considered a provisional decision, however, which may need to be re-considered when more empirical evidence is available. Even if we adopt the view that PPP acceptability is not directly influenced by verbal/modal features, there are nevertheless correlations to be found between semantic configuration and verbal/modal choice. And these appear to be consistent enough to warrant mention, although our impressions are based more The question of whether these are features of the verb or of the sentence as a whole will be ignored here.

107

on casual observation than systematic study.(The corpus sentences were not tested for all verbal/modal options but only for the possibility of one.) In the main we shall limit ourselves to outlining what appear to be typical verbal/modal choices for propositions in the passive. A survey of the passive examples cited so far reveals at least three major patterns of verbal/modal choice in acceptable PPPs.

These may in general be

found with any type of passivizable proposition, although sane preferences do emerge.

3.1.1 The first pattern can be illustrated by the following examples: (46)' (47)' (48)' (49)' (5O)'

The world was travelled round ... in 8 days The street was raced down ... in a matter of minutes The house was moved in ... three weeks ago The crowd was fired on ... ruthlessly A manageress was advertised for ... immediately

What appears to characterize verbal/modal choice in these sentences is use of past tense to refer to actions completed in past time. Notice, however, that present or future time reference is also possible: ( 4 6 . 1 ) ' The world!. , . 1 travelled round ... lis bej.ngJ

in 8 days

( 4 7 . 1 ) ' The street { . \ raced down ... in a matter of minutes I is being> be ι \ moved in ... in three weeks { will is being > wi J. J_ hf* ι . . \ fired on ... ruthlessly { is being> (5O.1)' A manageress |,

.

j advertised for ... immediately

Thus it would be more appropriate to say that what characterizes verbal/modal choice in this group of PPPs is use of tense to refer to events which occur in time. As the examples illustrate, this type of verbality is often found in conjunction with propositions containing Goal. Notice, however, that although a chronological event takes place, it is presented fron the viewpoint of Goal, so that the resulting passive is actually a statement which identifies a goal whose reaching occurs in time: Object χ is a goal which

7

is being was will be

reached

These are to be considered above all valid for literal predicates; metaphorical and abstract predicates often allow more freedom in verbal/modal choice.

108 3.1.2 A second typical pattern of verbal/modal choice for PPPs can be illustrated by the following: (51)' (52)* (53)' (54)* (55)'

This house can't be lived in any longer ... This mattress shouldn't be slept on by people allergic to doghairs This bathtub can be luxuriated in for hours ... These matches may be fooled with ... and cause a fire This cart must not be sat in by more than two people at once

This group is characterized in general by the presence of modal auxiliaries, here specifically to express possibility, advisability, likelihood and obligation. Modality of this type is often chosen in PPPs with Instrument subject, as the examples illustrate. The resulting passive makes a statement about the way an instrument can, should, must, etc. be used (or not be used). That is, it characterizes an object in terms of its qualities: Object x has the quality of

being able to be having to be

used

3.1.3 A third major verbal/modal pattern is to be observed in the following: (56)' (57)' (58)' (59)' (6O)'

This bed has been slept in ... My new hat has been sat on ... This lock has been tampered with by a burglar Her flowerbed had been ruthlessly trod on ... The floor had been walked on ... (there were footprints across it)

These PPPs are typified by perfective aspect. The present perfect is used as a so-called "perfect of experience" (Zandvcort 1957; 4 1966:62), which is said to have the effect of describing an unusual or unexpected state into which an object has been put as a result of the action predicated by the verb (cf Schmole 1975:32f). This pattern is quite frequently found in PPPs with Patient subject. The resulting passive makes a statement which describes what kind of experience has affected the object, i.e. it characterizes an object by describing the state it is in: Object χ is in the state of having been affected

Notice that since the perfect of experience describes an out-of-the-ordinary state, these sentences would only be uttered in situations where there is visual evidence of the way the object has been affected. H. Tuyn (197O:62) makes an observable result of this sort the major criterion for the passivization of locative prepositional verbs: "If the result can be observed, the noun in an

109

adjunct of place can become the subject of a passive sentence ..." However, while observable result (or something similar) frequently accompanies this use of the perfect in English and therefore frequently accompanies PPPs with this verbal pattern, a glance at the corpus will suffice to show that it can by no means account fully for the passivization of all locative prepositional verbs, much less prepositional verbs in general.

3.1.4

In sumnary, a sampling of the corpus demonstrates that certain verbal/

modal choices, namely chronological use of tense, modal auxiliaries and the perfect of experience, typically accompany acceptable PPPs. These choices are by their very nature closely associated with three of the semantic notions we have determined to be basic to the PPP, namely Goal, Instrument and Patient respectively. However, they are by no means restricted to occurrences of the semantic notion they are respectively associated with. Thus, a Goal subject may appear with a modal auxiliary or with the perfect of experience: ( 6 1 ) ' The store can be run to in a matter of minutes ( 6 2 ) ' The snow bank has been tunnelled through ...

...

The only result is that the Goal subject may have an added connotation of instrumentality or effect. Similarly, an Instrument or Patient subject may occur with chronological tense:. ( 6 3 ) ' A chair was stumbled over ... ( 6 4 ) ' The egg was sat on ... until it

hatched

What distinguishes all three verbal/modal options from others is their appropriateness in statements which identify, qualify or characterize objects. In general, PPPs identify goals which are reached, qualify instruments as capable of being used, and/or characterize patients as being in an affected state. Note that these correlations between verbal/moaal choice and PPP acceptability have been observed in a large but (for this purpose) not necessarily representative corpus of sentences with prepositional verbs. To what extent the patterns noted are binding for PPPs carfiot be determined on the basis of our material. An isolated example such as (65)': (65)'

*My bed has been being slept in ...

vs

( 6 5 . 1 ) ' My bed is being slept in ... ( 6 5 . 2 ) ' My bed has been slept in ...

110

suggests that perhaps a few general constraints of a verbal/modal nature must be reckoned with, but no specific hypotheses can be advanced at this time.

3.2

Information focus

A second factor involved in the contextualization of PPPs relates to the by-phrase. Recall that semantically a passivizable proposition is a configura-r tion of at least two notions, one of which, we have said, assumes subject position in the passive and the other of which assumes object position. However, although the subject of a PPP always appears in surface structure, its object may be emitted. Thus we may find: (66.1)' John's proposal was voted for by fifteen out of the sixteen committee members or

( 6 6 . 2 ) ' John's proposal was voted for ...

Whether the object in a PPP is deleted or not depends to a large extent on the greater context within which a PPP is embedded. Compare, for instance, the following hypothetical contexts: A

In spite of the fact that one committee member was vehemently opposed to it, ( John's proposal was voted for (?John's proposal was voted for by fifteen out of the sixteen committee ( members

B

Although there was originally some doubt as to whether any kind of majority could be found, in the end ?John's proposal was voted for John's proposal was voted for by fifteen out of the sixteen committee members

Text A is concerned with whether the proposal passes or not, i.e. it is oriented towards Goal in an Agent-Goal configuration. Here a PPP without a £>z/-phrase is sufficient; mention of who is Agent, because it is evident from the context, is at best distracting. Text B, on the other hand, is concerned with determining who will vote for the proposal, i.e.

it is oriented towards Agent in the same

configuration. The £>z/-phrase in the passive must be present because the information it carries (namely, who voted in favor) is essential to the message. These examples demonstrate that a PPP which is acceptable in isolation may be doubtful or unacceptable in a given context. The importance of external context in determining what form a PPP takes (and indeed in determining whether

111 Q

it is likely to occur in the first place) can hardly be overestimated. However, this is an aspect of the PPP which our corpus, consisting uniquely of sentences in isolation, does not permit reliable investigation of. Our remarks will then of necessity relate only to the acceptability or non-acceptability of PPPs in isolation. Even isolated PPPs, however, provide evidence that some ultimately contextual constraint is responsible for varying acceptability. Consider, for instance: (66.1)

Fifteen out of the sixteen committee members voted for John's proposal ( 6 6 . 1 ) ' John's proposal was voted for by fifteen out of the sixteen committee members

(66.3) Robert and Tom voted for John's proposal ( 6 6 . 3 ) ' ?John's proposal was voted for by Robert and Tom (66.4) They voted for John's proposal ( 6 6 . 4 ) ' * J o h n ' s proposal was voted for by them

The constraints to be observed here relate not to lexical choice but rather to information focus, defined by Halliday as a system which "assigns to the information unit a structure in terms of the two functions 'given1 and 'new'" (1967b:2O4). 'New' refers to something which the speaker presents "as not being recowrable fron the preceding discourse", 'given' to what is not new. Thus information focus is ultimately determined on a discourse or textual level. However, given information can be characterized by the fact that (a) it is usually said with less stress and lower pitch than new information, and (b) it may be proncminalized (cf Chafe 197O:31). That is, even in isolation, the given-new structure of a sentence can be identified to the extent that pitch and ability to undergo pronominalization can be determined out of context. It is to this extent that we can investigate the influence of information focus on PPP acceptability in our corpus. Based on criteria relating to pitch and ability to proncminulize, the examples above can be assigned the following information structure: In (66.1) the phrase fifteen out of the sixteen committee members is undoubtedly new, since it would be said with high pitch and would not be subject 8

Halliday (1967b:216f), for instance, claims that the speaker selects the passive because it allows him to make something other than agent, theme ii.e. "what is being talked about" (1967b:212);. The agent can then either be omitted or carry information focus.

112

to proncminalization. In (66.3), on the other hand, the phrase Robert and Tom would be given, under normal circumstances, as evidenced by the fact that it can be prononinalized [see (66.4)]. If, however, we were to allow John's proposal in the same sentence to prononinalize instead: (66.5)

Robert and Tom voted for

it

then it would be assumed to be the given part of the message. In (66.4), they as a pronominal form is obviously given. Cotiparing this structure now to PPP acceptability, it appears that a passive is acceptable when the by-phrase contains new information fas in (66.1)'] but unacceptable if it contains given information [as in ( 6 6 . 4 ) ' ] . In (66.3) , if we assume that John's proposal (or it) is given as in (66.5), then the passive (66.3/5)' John's proposal/It was voted for by Robert and Tom

is acceptable. However, if (66.3) is equivalent to (66.4) in information structure, then a passive is unacceptable: (66.3)'*John's proposal was voted for by Robert and Tom

The same phenomenon can be noted throughout the corpus. Sentences with byphrases which contain the given part of the message are unacceptable in the passive, whereas those where the 2»/-phrase communicates new information are acceptable. Thus: ( 6 7 . 1 ) ' *The U.S. is lived in by Barbara ( 6 7 . 2 ) ' The U . S . is lived in by millions of immigrants (68.1)' *The roof was alighted on by the bird ( 6 8 . 2 ) ' The roof was alighted on by a host of migrant birds ( 6 9 . 1 ) ' *The stadium was being hovered over by the Army helicopter ( 6 9 . 2 ) ' The stadium was being hovered over by Army helicopters

A first conclusion concerning the influence of information focus on PPP acceptability is thus that if a £z/-phrase contains given information it will be unacceptable in the passive, whereas if it contains new information, it will be acceptable. Notice that sentences with the indefinite pronoun someone in subject position are not to be confused with those in which the pronoun has definite anaphoric reference. Thus,in: (70.1)

Someone slept in this bed

(70.2)

They slept in this bed

vs

113

someone in (7O.1) is not given information like they in (70.2). In (70.1) it is this bed which must be considered given, and the fact that someone slept in it, new: N (7O.1)

G

Someone slept in I this bed

Therefore, a passive such as ( 7 O . 1 ) ' This bed was slept in by someone

is quite conceivable. Although the i>j/-phrase contains a pronoun form, the indefinite someone is part of the new information in the sentence. On the other hand, in (70.2) they is given, under normal circumstances, and sleeping in the bed is new: (ΊΟ.2)

G N They I slept in this bed

Here a passive is not acceptable: (7O.2)'*This bed was slept in by them

Thus, the generalization above still holds that given information cannot occur in a £>t/-phrase in the passive. Notice that according to this analysis a PPP without a i>z/-phrase, in contrast to a PPP with one, is ambiguous with regard to information focus. If it stems from an active sentence with an indefinite pronominal subject such as (70.1), then in the passive it is the subject which is given, the predicate which is new:

G

N

( 7 O . 1 ) ' This bed | was slept in (by someone)

The speaker assumes that this bed is a piece of information which is recoverable from the context or the discourse situation (perhaps speaker and hearer are both looking at a bed with rumpled sheets). On the other hand, a PPP which comes fron an active sentence such as (70.2) with definite anaphoric reference has the following information content: N ( 7 0 . 2 ) ' [ T h i s bed was slept i n ) . . .

Here the speaker emits explicit reference to the given; his utterance consists wholly of new information. This might be the case if, say, an innkeeper were pointing out to a criminal inspector where a suspect (present in the conscience of both speaker and hearer) might have left traces - "This bed was slept in, that table was eaten at ..."

114

The passive is thus in this respect more restricted than the active in terras of what can appear in subject and object position. Whereas an active sentence can conceivably have either the structure given - new [as in (66.4)] or new - given [as in (66.1)] , the passive allows only the structure given - new with subject-object configuration [cf (66.1)']. If object is omitted, then the passive may have either new or given-new [(70.2)' or (70.1)']. The latter is probably more cannon. We have restricted our comments here to the case of neutral or non-con trastive PPPs. The influence of contrastive stress on PPP acceptability is a question which cannot be gone into in detail here. Note, however, that the constraints observed so far appear to relax in cases of contrastive stress. Thus, a contrastive PPP such as the following is conceivable: (7O.3)' This bed was slept in by HIM, not HER That is, a definite pronoun might occur in the by-phrase if it has contrastive stress. Furthermore, it might even be possible to have contrast!vely: ( 7 O . 4 ) ' No, THIS BED was slept in by Peter, not THAT ONE That is, it appears that the by-phrase can contain given material if contrastive stress is placed on the subject. In a neutral or normal situation, however, according to our observations, a PPP will generally be doubtful or unacceptable if an object is present and contains given information.

4

Acceptability in practice

Although semantic con figuration is undoubtedly the basic factor in PPP acceptability, the importance of lexical choice and contextualization must not be overlooked. In practice, acceptability or non-acceptability is often dependent not on one single factor but on the interaction of all three. Consider, for instance, the following:

115

(71.1) Something alive rustled through the undergrowth ( 7 1 . 1 ) ' ?The undergrowth was rustled through by something alive ( 7 1 . 2 ) Something alive was rustling through the undergrowth ( 7 1 . 2 ) ' The undergrowth was being rustled through by something alive

The acceptability of (71.2)' as opposed to (71.1)' actually depends on the favorable conjunction of semantic, lexical and contextual factors. The semantic configuration is one involving Agentive and Goal. The verb, however, has complex lexical structure: GO + manner. Normally, this combination is constrained in the passive. But notice that it is given extra motivation here by the presence of a fey-phrase with an Agentive which justifies use of the complex verb (and which contains new information). Emphasis in the sentence is thus placed on manner of reaching the goal rather than simply reaching the goal, and it is for this reason that imperfective rather than perfective aspect is called for. Use of the continuous form is thus almost essential if the PPP is to be acceptable. The constraint in (71.1)', however, can hardly be attributed to this one factor alone. Instead, it is the interaction of at least three different aspects which is ultimately responsible for acceptability or non-acceptability.

Cf also: ?The street was waddled across by a small fat duck vs The street was being waddled across by a small fat duck,or ?The street was zigzagged down by a drunken man vs The street was being zigzagged down by a drunken man, etc.

V

GRAMMAR AND THEORY OF THE PPP

1

Towards a grammar of the PPP

The semantic-syntactic, lexical and contextual aspects of the PPP discussed above comprise a grammar of the PPP, whose functioning can be described, subject to due reservation, approximately as follows: A situation is analyzed in terms of the semantic role which its participants play. Depending on its constitution, the set of semantic notions thus derived may be 'basically passivizable' or not. In the case of the former, two possible subject-object configurations are determined accordingly. The proposition is completed by the addition of verbal/modal components such as tense, aspect, etc These choices are based on what type of statement is to be made, which in turn is often influenced by the semantic roles in question. Subsequently, the semantic roles are filled with lexical items. If the passive is still an option, and if this option is chosen, the form of the PPP, i.e.

with or without &z/-phrase,

is determined according to information structure. [The decision as to whether the passive option is chosen or not is presumably made in consideration of what is being talked about and what is new information.] Thus, acceptable PPPs can be shown to result from a series of successive options, the most fundamental of which is semantic in nature. In diagrammatic form, the process might be illustrated as follows:

117

c

XI

0]

1 Φ

•Ρ -Ρ rH

Ι

! Ι ι

οι in

XI ft •Ρ Φ

/ /

/

C O

/

i

/ ;/

y O

ft

Ο ΜΗ φ Τ3

(1)

XI

f. tn χι

ΕΗ h) B) XI

Φ rH

a Φ

ΟΜ

ft Α Ο

S Qa g

j> s u

Cn *C

Φ

4J

4J

υ

dl

-P C -H -^

•o n

s§ W

rj

O 0)

ft

ft H

Ο)

a*

•P tn c ο ο

W

in c ο υ φ

«3 Λ

υ φ

Χ! ui

Ι

ο 4J υ Φ

01 C

T3 01

ω

4J

Ι-Η

3u

EH ^ 0) XI

Μ Ο Φ Φ ft χ ft