ON THE PASSIVE OF INFINITIVAL PERCEPTION VERB COMPLEMENTS Renaat Declerck This paper is concerned with Infinitival Perception Verb Complements swim in I saw John swim. While the derivation of sentences like these has remained controversial in transformational grammar (see Declerck (1981b) for a discussion and critique of five different analyses), all grammars apparently agree that structures with IPVC’s can be passivized as shown in the following examples: 1. (henceforth: IPVC’s) like John to your house yesterday. (by Bill) to come to your house yesterday. (2) (a) Mary heard Tom sing a song every hour. (b) Tom was heard (by Mary) to sing a song every hour. (1) (a) Bill (b) John saw John come was seen The claim that (1, b) and (2,b) are the passive counterparts of (1, a) and (2,a) has been made both in traditional grammar (see e.g., Jespersen (1961:315), Quirk et al. (1972:841)) and in transformational grammar (see e.g. Huddleston (1971), Emonds (1976), Horiguchi (1978). In the latter theory the concomitant claim is made that the to which occurs in the passive construction but not overtly in its active counterpart must also be present in the underlying structure of the active IPVC’s but is obligatorily deleted there. In fact it is not hard to understand why the view that (l,b) and (2,b) are the passive versions of (l,a) and (l,b) has been so persistent: since IPVC’s (bare infinitives) can follow active perception verbs but no passive ones (cf. (3, a-b)), while to-IPVC’s can follow passive perception verbs but no active ones (cf. (3, c-d), it is, indeed, tempting to look upon (3,d) as the passive counterpart of (3, a) . (3) (a) Someone saw John write a letter. (b) *John was seen write a letter. (c) *Someone (d) John was saw seen John to write a letter. to write a letter. The assumption that (3, d) is the passive of (3, a) is not even challenged in Kirsner (1977), although it is pointed out there that the two sentences are not quite synonymous. Kirsner attempts to explain this semantic difference on the basis of pragmatic considerations, without questioning the fundamental voice relation between the sentences (see below). It goes without saying that treating (3, d) as the passive of (3, a) has some consequences for the analysis of IPVC’s in transformational grammar. To give only one example: since it is widely accepted in present-day TG that transformations cannot create structure, 1 the fact that to shows up in (3, d) entails that the active structure from which (3, d) is derived must also contain to (or at least a category from which to can be derived). If one assumes that this active structure is (3, a) it follows that, contrary to what is claimed by Gee (1975) and Akmajian (1977), IPVC’s must be S’s in underlying structure, since 27 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 either be the lexicalization of COMP (as claimed in Huddleston (1971: and suggested in Gee (1977:480)) or be derived from AUX (as is more often assumed--see e.g. Bach (1977:625), Rouveret & Vergnaud (1980)). Both COMP and AUX are nodes that require an S above them. This and other illustrations make clear that the question of whether (3, d) is really the passive of (3, a) is of some theoretical importance. to must 183) (or S) 2. It has been pointed out by Kirsner (1977-173) that passives like (4, a-b) differ from &dquo;their formally corresponding actives&dquo; (5, a-b) in suggesting that the perception process was accidental and that the event perceived was not intended by its agent to be witnessed: (4) (a) John was seen (by Sam) to cross the street. The President was heard to utter an expletive. (5) (a) Sam saw John cross the street. (b) We heard the President utter an expletive. (b) Kirsner remarks that &dquo;the complements of the active sentences describe events which the perceivers could well plan to perceive and which are intended by their agents to be perceived (... ). By contrast, the infinitive phrases of the passive sentences describe events whose perception is fortuitous and which are not performed in order to be witnessed&dquo; (p. 174). According to Kirsner this semantic difference follows pragmatically from the fact that agentive perception verbs like watch and listen to cannot occur in sentences like (4, a-b): &dquo;if the explicitly agentive sensory verbs are (for whatever reason) excluded from passives, it is understandable that those sensory verbs that do occur there would be taken as referring to unplanned perceptions of unanticipated and sometimes ’furtive’ events&dquo; (p. 175). Two questions suggest themselves here: (a) does Kirsner’s explanation satisfactorily account for the semantic distinction observed, and (b) does it follow from this distinction that the existence of a voice relation between the relevant pairs of sentences should be questioned? In my opinion both questions should be answered negatively. The reason is that the semantic distinction in question occurs whenever a sentence involving a perception verb is passivized. For example: Mary had seen John the day before. (b~ Mary heard the children. (7) (a) John had been seen (by Mary) the day before. (6) (a) (b) The children were heard (by Mary). can refer both to ’accidental’ and to ’planned’ perception; the other hand, suggest ’accidental’ perception only: they are far less likely to be used than (6, a-b) if Mary made an intentional effort to see John or to hear the children. This is exactly the same difference as distinguishes (4, a-b) from (5, a-b) and as can be observed when the perception verb is fol- Sentences (6, a-b) (7, a-b), on lowed by a participial complement (henceforth: PPVC): 28 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 (8) (a) (b) (9) (a) (b) Mary had seen John crossing the street the day before. Mary heard the children playing in the garden. John had been seen (by Mary) crossing the street. The children had been heard (by Mary) playing in the garden. Again, the active sentences (8, a-b) are compatible with both planned and accidental perception whereas (9, a-b) suggest accidental perception only. It follows that this semantic difference is a function of passivizing perception verbs in general and therefore provides no ground for not treating (4,a-b) as the passives of (5, a-b) as long as (7, a-b) are treated as the passives of (6, a-b) and (9, a-b) as the passives of (8, a-b). Of course, this does not mean that no other reasons could be found for denying the voice relation in the case of IPVC’s only (see be- low). As to Kirsner’s explanation of the semantic difference between (4, a-b) and (5, a-b), we can point out two things. First, the observation that agentive verbs like watch and listen to can occur in (5, a-b) but not in (4, a-b) could explain the semantic difference only if it were equally impossible to use these verbs in (7, a-b) and (9, a-b), since the same semantic distinction holds between these and their active counterparts. In fact, however, such passive sentences are not excluded: (10) (a) John had been watched (by Mary) the day before. (b) The children were listened to (by Mary). (c) John had been watched crossing the street the day before. (d) ’ The spy was listened to being interrogated by Mr. Philps. A second objection to Kirsner’s explanation is that the readings ’accidental’and ’unplanned’ are not excluded by explicitly agentive perception verbs like watch and listen to. However, this claim is essential to Kirsner’s reasoning, which runs as follows- (a) watch and listen to imply that the perception is agentive, i. e. , that (b) that the perceiver is responsible for the perception; it follows that watch and listen to will favour the inference that the perceived event was anticipated, hence that the perception was planned’ (c) passive sentences involving to-IPVC’s cannot involve watch or listen two (d) it follows that those perception verbs that do occur in such passive sentences will be taken as referring to unplanned perceptions of unanticipated events. Apart from the fact that (dB would follow only weakly from (a)-(c), it is not correct that (b) follows from (a): the intentionality implied in the agentive perception verbs concerns the continuation of the perception process rather than the expectation of the perception. What distinguishes watching from seeing is that, once the perception has started (and this beginning itself may be quite accidental and unexpected), the perceiver makes a conscious effort to protract the perception process: he keeps his eyes fixed on the event that caught his attention. Thus, (11) does not imply that we expected a car to come to our house, 29 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 but it does (11) imply that we continued We watched the car looking driving at it up to our once we had caught sight of it. house. Summarizing this discussion of Kirsner (1977), we can conclude that (a) passive structures involving perception verbs in general differ from their corresponding actives in that they suggest accidental and unplanned perception; (b) Kirsner’s pragmatic explanation of this is unsatisfactory: and (c) the semantic difference in question provides no ground for questioning the assumption that John was seen to come is the passive of They saw John come, since the same difference holds for pairs like John was seen versus They saw John which must be related by passivization if such a transformation exists. It would seem, then, that there is no reason for challenging the traditional belief that (3, d) is the passive of (3, a). Yet in this paper I wish to defend the claim that this conclusion is not correct. I will point out a number of facts that in my opinion are incompatible with the view that there is a voice relation between (3, a) and (3, d). In doing so I will also offer an explanation for several of the observations, e. g., for the fact (noted by Kirsner) that agentive perception verbs like watch, look at and listen to cannot occur in pattern (3,d). 3. Before starting the argumentation proper, it is necessary to point out basic characteristics of PVC’s. A first remark concerns the distinction between PPVC’s and IPVC’s: some I saw John (12) (a) (b) I saw working. John work. The difference between these sentences is that the PPVC in (12, a) expresses progressive (durative) aspect while the IPVC in (12,b) does not. That is, the PPVC refers to John’s being in the middle of the durative activity of working, while the IPVC refers to the same activity as a &dquo;single unanalysable whole&dquo; (Comrie (1976:3) without referring to its internal temporal structure (beginning, middle or end). The claim that the difference between PPVC’s and IPVC’s is merely one of aspect was formulated by traditional grammarians like Poutsma, Kruisinga and Jespersen and has rarely been challenged since though there have been exceptions to this, see e.g. Akmajian (1977). In Declerck ( 1981a) a number of seeming counterarguments against the claim are refuted and a great deal of syntactic and semantic evidence is advanced in support of it. A second point that needs going into is the relation between IPVC’s (bare -- infinitives) and to-IPVC’s: (13) ra) I saw him open the door. (b) When the figure came nearer, I saw it to be a woman. This is a more difficult question. Most linguists seem to have adopted the view summari7ed in (among others) Kirsner & Thompson (1976) that IPVC’s and 30 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 PPVC’s communicate a direct (physical) perception of an event whereas toIPVC’s and that-PVC’s communicate an indirect report about or a deducing of situation. For example: (14) (a) (b) (c) (d) I I I I a observed Mary knit a sweater. observed Mary knitting a sweater. observed Mary to be knitting a sweater. observed that Mary was knitting a sweater. Sentences (14, c-d) can be used in a context in which I observed nothing but knitting-wool and needles or an unfinished sweater, but (14,a-b) require that I actually perceived with my eyes the activity of Mary knitting a sweater. However, Spears (1977) points out that the labels ’direct perception’ and ’indirect perception’ are inadequate insofar as (a) direct perception can also be involved when a to-IPVC or that-PVC is used (as is the case in (14, c-d) where the deducing is based on some direct perception of knitting-wool, needles, etc.), and (b) some degree of inference (deducing) can also be involved when an IPVC or PPVC is used. For example, in (15) I saw John walking home. what I saw was no more than that John was walking- the indication home is the result of an inference on my part which is based on knowledge that has nothing to do with the perception proper, e. g., knowledge of where John’s home is. Spears therefore proposes the following criteria to distinguish between IPVC’s/ PPVC’s, to-IPVC’s and that-PVC’s (all of which can imply direct perception as well as inference): (a) With IPVC’s and PPVC’s direct perception necessarily plays a role in the formulation of the contents of the PVC. (b) With to-IPVC’s the subject (perceiver) necessarily has a role in the formulation of the PVC. (c) With that-PVC’s the subject need have no role in the formulation of the PVC. That is, the object of the perception may be someone else’s report about a situation or event, as in I hear that John has been ill for some time. However, these criteria help us to separate the extremes of the different types of PVC’s only. Thus, criterion (a) can make clear the difference between (16,a) and (16,b) (since the expression of anteriority in (16,b) presupposes some ’indirect’ perception), but direct perception seems to be equally relevant to the formulation of the content of the PVC in (17,b) as in (17, a). Similarly, criterion (c) can distinguish between (18,a) and (18,b) but is useless in cases where the that-PVC does not represent an external report, as in (17, c). In fact, none of the criteria (a)-(c) will be of much use when it comes to defining the precise difference between (17,a-c).2 31 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 I noticed John paint(ing) the house. I noticed the house to have been painted. I observed John work(ing) in the garden. I observed John to be working in the garden. (c) I observed that John was working in the garden. (a) I heard it raining. (b) I heard from John that it was raining. (16) (a) (b) (17) (a) (b) (18) opinion it must be possible to describe the different meanings of in terms of the traditional labels ’direct’ and ’indirect’ perception if we manage to define these in an accurate way. Spears objects to the distinction, saying that inference (which is taken as criterial for indirect perception) can also play a role in IPVC’s/PPVC’s (which are taken to express direct perception), as in (15). However, the kind of inference referred to here is not the same as is assumed to be typical of that-PVC’s. Compare: In my (17, a-c) (19) (a) (b) I I can see John can see John is walk(ing) towards the beach. walking towards the beach. ( 19, a) not all the elements of the IPVC/PPVC are the result of direct perception : the element towards the beach is added on the basis of contextual knowledge. Sentence (19, a) thus implies inference in the sense that the speaker not only refers to what he actually perceives but also includes the interpretation he assigns to this in his report of the event perceived--an interpretation which is partly determined by his knowledge of the world. It should be noted that this kind of inference is not typical of IPVC’s/PPVC’s only, not even of PVC’s in general only, but may be involved whenever a predication is made on the basis of direct perception. Thus, it also plays a role in independent clauses like In (20) Look ! John is walking towards the beach. This same kind of inference, which consists in an immediate and unconscious addition of interpretive elements to what is actually perceived, also shows up in (19,b), but here there is a second kind of inference that plays a role as well. Sentence (19,b) suggests something like ’Looking at the situation (event) I come to the conclusion that John is walking towards the beach.’I That is, using a that-PVC the speaker expresses more explicitly that he is summing up the situation perceived. The that-PVC in (19,b) represents a kind of ’reported’ or ’indirect’ version of the IPVC/PPVC in (19. a), in the same way as (21~ represents a ’reported’ (21) The use or ’indirect’ speech version of (20): He says that John is of that therefore walking implies some towards the beach. reflection, some distance necessary to cognitive process up the situation and formulate a conclusion. This kind of is not implied by the use of IPVC’s/PPVC’s, as in (19, a). sum 32 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 There are several syntactic and semantic facts that corroborate this conclusion: a. Because that-PVC’s involve a more conscious process of logical deduction they make a stronger assertion as to the truthfulness of their contents than IPVC’s and PPVC’s. This is clear from examples like (22,a-d), where (22,d) is much less felicitous than (22,b) because the truth of the content of the PVC is explicitly questioned by the perceiver: (22) (a) John has seen us fighting. (b) John has seen that we were fighting. (c) John wonders if he has seen us fighting. *?John wonders (d) if he has seen that we were fighting. Notice also that (22, e) does not share the relative unacceptability of (22,d) bethe conclusion expressed in the that-PVC is no longer ascribed to John and is therefore not questioned by him any more: cause I wonder if John has (22) (e) The following seen that we were sentences further illustrate these (23) (a) John heard us fighting. points: calling. John heard that we were calling. (c) John wondered if he heard us calling. (d) *John wondered if he heard that we were calling. (e) I wondered if John heard that we were calling. (b) b. That-PVC’s can contain perfect forms expressing anteriority, while IPVC’s and PPVC’s cannot: (24) (a) (b) The I saw that the house had been repainted. *I saw the house have/having been repainted. is clearly that, while anteriority cannot be directly perceived, it is to infer to the occurrence of a past activity on the basis of results that reason possible have remained perceptible. c. IPVC’s and PPVC’s cannot that-PVC’s can: (25) (a) (b) (26) (a) (b) The reason normally involve stative predicates, but *I saw John be~ing) ill that day. I saw that John was ill that day. *I observed the house need(ing) painting. I observed that the house needed painting. for this is that stative predicates refer to such abstract 33 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 ’things’ as and states of affairs which cannot be perceived infer on the basis of perception. relations, conditions but which one can d. Explicitly agentive perception verbs PPVC’s but not by that-PVC’s: (27) (a) (b) (28) (a) (b) can be followed directly by IPVC’s/ We all watched them leave/leaving without saying a word. *We all watched that they left/were leaving without saying I looked at Sam make/making up the fire. *I looked (at it) that Sam a word. made/was making up the fire. can explain this from the fact that these predicates necessarily refer to a durative (ongoing) perception and are therefore inappropriate for summing up the situation perceived. Since ongoing perception must necessarily be direct perception, predicates expressing this kind of perception cannot at the same time express the inference that is typical of that-PVC’s. It should be noted that the same explanation accounts for the fact that IPVC’s/PPVC’s can depend on progressive perception verb forms while that-PVC’s cannot: We on in!) We’re seeing Apollo 19 take/taking off. from Kirsner & Thompson (1976:221)) (b) *We’re seeing that Apollo 19 takes/is taking off. (30) (a) We were observing the children playing outside. (b) *We were observing that the children were playing. (29) (a) (Come (example e. Unlike the perception expressed by IPVC’s/PPVC’s, the perception expressed by that-PVC’s cannot be durative. Compare the following sentences (in which the duration adverbial should be taken to bear on the perception verb rather than on the PVC): (31) (a) (b) From 8 to 9 I watched the moon rise/rising over the hill. *From 8 to 9 I saw that the moon rose/was rising over the hill. Unlike (3 1, a), (31,b) is incoherent because it combines the idea of ’arrival at a conclusion’ with the idea of ’duration of one hour,’ thus yielding only two possible interpretations: either it took me an hour to infer from what I saw that the moon was rising over the hill or I reached this conclusion again and again over a period of one hour. Both interpretations are rather implausible, given our knowledge of the world. It may be noted that (31,b) becomes much better if a plural subject is used, as in (31, c). The reason is that a repetitive interpretation (one after another, many people came to the conclusion that... ) is not inconsistent with the duration adverbial. (31) (c) From 8 to 9 many the hill. people observed that the moon was 34 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 rising over f. IPVC’s/PPVC’s cannot normally contain a negation, but that-PVC’ss can: (32) (a) John noticed that Mary did not say a word. (b) *John noticed Mary not say a word. (33) (a) I heard that he was not singing any more. (b~ *I heard him not singing any more. The reason for this infelicity of negative IPVC’s/PPVC’s is that the perception of the non-occurrence of an event is less direct than the perception of its occurrence : to perceive that something is not happening is a kind of conclusion that is partly based on the expectation that the event in question would or might happen. IPVC’s and PPVC’s, which express direct perception, are not very suitable for expressing this kind of indirect perception, but that-PVC’s, which imply some logical deducing, are. g. Up to now I have linked up the idea of ’reported’ perception with that-PVC’s only, but it is clear that this idea is compatible with the use of other complementizers as well. Thus, next to (34, a) we can have sentences like (34,b-e): I have I have I have I have (e) I have (34) (a) (b) (c) (d) opened the door. opened the door. seen when John opened the door. seen where John opened a door. seen why John opened the door. seen seen that John how John All these sentences imply some logical deducing, but the conclusions arrived at are different. In (34, a) the conclusion concerns the very fact that John opened the door; in each of the sentences (34,b-e) it concerns some aspect of the situation that is linked up with an adverbial meaning (manner, place, time, reason). IPVC’s and PPVC’s, which do not imply inference, do not by themselves signal any of these meanings they just refer to the perception of a situation without expressing any of the inferences that can be made from it. However, precisely because of this neutrality it is possible to use IPVC’s and PPVC’s in contexts inviting any of the meanings signalled by the complementizers of (34,a-e~. For example, the IPVC him swim will be given the same interpretation as a thatPVC in (35, a) and the same interpretation as a how-PVC in (35,b): Is it true that John had a swim this morning B. Yes, I saw him swim. I believe John has injured his arm again. Did you (35) (a) A. (b) see him swim yesterday? A sentence like (37, a) but (37,b) can only be can be used an answer to reply (36,a): as a to any of the questions 35 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 in (36, a-d), walking in the park? walking? (c~ Do you know where John was walking? (d) Do you know when John was walking in the park? (37) (a) Yes, I’ve seen him walk(ing). (b) Yes, I’ve seen that he was walking. Do you know that John was (b) Do you know how John was (36) (a) Similarly, (38, a) will be interpreted as roughly synonymous with (38,b), (39, a) will be semantically related to (39,b), but (40, a) will be given the meaning as (40,b): and same (38) (a) When you hear him speak English you get the impression that he is a foreigner. (b) When you hear how he speaks English you get the impression that he is a foreigner. (39) (a) You should have seen them play against Liverpool. (b) You should have seen how they played against Liverpool. (40) (a) I saw the train leave before John reached the platform. (b) I saw that the train left before John reached the platform. In Dutch, IPVC’s can yield the interpretation of where-PVC’s. Thus, (4 1, a) equivalent to (41,b), not (41, c). (In English, an adverbial of place like anywhere has to be added to the IPVC.) is Ik zie het niet liggen. (’I do not see it lie’) (b) Ik zie niet waar het ligt. (’I do not see where it lies’) (c) Ik zie niet dat het ligt. (’I do not see that it lies’) (41) (a) Summing up section 3, we can say that there is sufficient evidence, both syntactic and semantic, for claiming that the distinction between IPVC’s/ PPVC’s on the one hand and that-PVC’s on the other can be defined in terms of direct versus indirect (reported) perception. By direct perception we mean that the contents of the PVC represent the interpretation which the perceiver assigns immediately and largely unconsciously to the situation perceived. Indirect perception, on the other hand, means that the PVC reports a conclusion that is more consciously arrived at by the perceiver (or someone else) on the basis of perception. 4. Let us now direct our attention to to-IPVC’s. Consider the following examples: (42) (a) When the figure got nearer I perceived it to be a woman. (example from Poutsma (1929:794)) (b) I generally observe such men to retain a certain freshness. (id.) (c) In shadow the hue seemed black, but when illumined it was seen to be the darkest, warmest brown. (example from Kruisinga (1931:212)) 36 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 from these examples that to-IPVC’s should be related to IPVC’s/PPVC’s. As far as I can see there is no semantic difference whatever between (42, a-c) and (43, a-c): It is immediately clear that-IPVC’s, not to (43) (a) When the figure got nearer I perceived that it was a woman. (b) I generally observe that such men retain a certain freshness. (c) (... ) it was seen that it was the darkest, warmest brown. Moreover, all the semantic and syntactic characteristics that in connection with that-PVC’s are shared by the to-IPVC’s: we have observed a. Like that-PVC’s, to-IPVC’s involve some logical deduction and therefore assert the truth of their message more forcefully than IPVC’s and PPVC’s. For this reason, to-IPVC’s are as incoherent as that-PVC’s in contexts where their truth is questioned. Compare the following sentences, in which he refers back to John: explicitly (44) (a) John wondered if he had seen the body move /moving. (b) *John wondered if he had seen the body to be that of a boy. (c) *John wondered if he had seen that the body was that of a boy. can b. Like that-PVC’s, to-IPVC’s differ from IPVC’s/PPVC’s in that they involve a perfect verb form expressing anteriority. (45) (a) (b) the house to have been repainted. they turned into the welcome shade of the cool old hall that Mr. Lorry observed a great change to have come over the doctor. (example from Poutsma (1929:794)) I saw It was when Like that-PVC’s, and unlike IPVC’s/PPVC’s, _to-IVVC’s predicates, as in (42, a-c). c. stative can involve d. Like that-PVC’s, and unlike IPVC’s/PPVC’s, to-IPVC’s cannot occur after agentive perception verbs like watch, listen to, etc. , nor after progressive verb forms of other perception verbs. (46) (a) *We watched the figure to be a woman. (b) *We listened to him to be sleeping. (c) *John looked at the house to have been repainted. (47) (a) We’re seeing Apollo 19 take off. (= (29, a) above) (b) *We’re seeing the figure to be a woman. e. Like the perception expressed by that-PVC’s, and unlike that expressed by IPVC’s/PPVC’s, the perception expressed by to-TPVC’s is necessarily punctual (nondurative). Thus, if the duration adverbial is taken to bear on the perception verb, (48, a) is as incoherent as (48,b): 37 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 *From 8 to 9 I *From 8 to 9 I (48) (a) (b) can perceived the figure to be that of a woman. perceived that the figure was that of a woman. f. Like that-PVC’s, to-IPVC’s differ from involve a negator: IPVC’s/PPVC’s in that they (49) (a) I perceived the figure not to be that of a woman after all. (b) John noticed the house not to have been repainted. (c) Mary observed the man to be no foreigner. g. To-IPVC’s are interpreted as synonymous with that-PVC’s: unlike IPVC’s/PPVC’s they cannot yield the interpretations of how-PVC’s, whenPVC’s, etc. Thus, 150,b) cannot be interpreted as synonymous with (50,a), in spite of the context inviting a how-interpretation. Instead, the to-IPVC will yield the same interpretation as the corresponding that-PVC, so that its use in (50,b) is incoherent. (50) (a) (I have the impression that John has how he walked just now ~ (I have the impression that John has observe him to be walking just now ? injured his leg.) Did you ob- injured his leg.) *Did you serve (b) Similarly, (51,b) can yield the interpretation of (51, a), but (51, c) cannot: (51) (a) When you observe how he works you get the impression that he done this kind of job before. you observe him work(ing) you get the impression that... *When you observe him to (be) work(ing) you get ... has never (b) When (c) In sum, to-IPVC’s share all the syntactic and semantic characteristics that that-PVC’s from IPVC’s/PPVC’s (except perhaps the possibility of representing an external report--see, however footnote 2). We can conclude, then, that the meaning of PVC’s can be defined in terms of either direct or indirect perception in the sense defined above; IPVC’s and PPVC’s signal the former, to-IPVC’s and that-PVC’s express the latter.4 distinguish 5. The above observations on the respective meanings of the different types of PVC’s provide a good basis for returning to the question that has been raised at the beginning of this paper, viz. whether the widespread view that (3, d) (repeated here) is the passive counterpart of (3, a) (id.) is correct or not. (3) (a) Someone (d) John was saw John write seen to write In order to answer this question we IPVC in (3,d) has the semantic and does, (3, d) can be treated as a letter. letter. a simply have to ascertain whether the to- syntactic characteristics of IPVC’s. If it the passive of ~3,a). If it does not, but,on the 38 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 other hand, shares the characteristics of to-IPVC’s, it will be impossible to consider (3, d) as the passive of (3, a). Plainly, it does not make sense to assume a transformational relation between two structures that are as different, both syntactically and semantically, as IPVC’s and to-IPVC’s. When examining sentences like (3,d) we ascertain that their PVC’s share every one of the characteristics that distinguish to-IPVC’s from IPVC’s/PPVC’s: In sentences like a. not (3,d) the PVC’s are to-IPVC’s by form since to can- possibly be omitted: (52) *John was seen write a letter. b. As we have seen, to-IPVC’s involve some logical deduction and therefore assert the truthfulness of their contents more forcefully than IPVC’s. For this reason to-IPVC’s are less coherent than IPVC’s in contexts where their truth is explicitly questioned: (53) (a) John wondered if he had heard a bell ring. (b) *John wondered if he had heard the man to In a similar context, constructions like with to-IPVC’s: (53) (c) *John wondered if a (3,d) are as have hoarse voice. a incoherent as sentences bell had been heard by him to ring. Compare also: (54) (a) John doubts if he saw the body move. (b) *John doubts if the body was seen by him to move. (c) I doubt if the body was seen by John to move. Unlike is not incoherent because the truth of the conclusion exin the to-IPVC is no longer questioned by the person who arrived at it. (54, b), (54, c) pressed c. Like tences like (55) (a) (b) d. like (3, d) can IPVC’s, the to-IPVC *John saw the lawn have been mown. The lawn was seen by John to have been Like other to-IPVC’s, and unlike IPVC’s, can involve a stative predicate: (56) (a) *We (b) e. in general, and unlike contain perfect forms: to-IPVC’s (3, d) John saw in sen- mown. the to-IPVC in sentences John be in need of assistance. (by us) to be in need of assistance. was seen In sentences involving to-IPVC’s the perception verb cannot be agentive 39 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 (watch, look at... ) sentences like or progressive. We notice the same restriction in (3,d) :5 (57) (a) Bill watched John cross the street. (b) *John was watched by Bill to cross the street. (58) (a) The boy used to look at people do things. (b) *People used to be looked at by the boy to do things. (59) (a) We’re seeing Apollo 19 take off. (= (29,a)) (b) *Apollo 19 is being seen by us to take off. f. Like the perception expressed by to-IPVC’s, the perception expressed the PVC’s in sentences like (3,d) is not durative but punctual. Thus, if the duration adverbial in (60) is taken to bear on the perception verb (rather than on the to-IPVC), (60) can only have a repetitive interpretation, i.e. over a period of several hours it was noticed by different people (or by the same person at different times) that John walked to the hill. 6 by (60) can For several hours John was seen g. Like to-IPVC’s and unlike contain a negator - IPVC’s, to walk to the hill. the PVC’s in sentences like (3,d) (60) (a) *We saw only John not run away. (b) Only John was seen not to run away. h. Like to-IPVC’s in general, and unlike IPVC’s, the to-IPVC’s in such sentences as (3, d) can be interpreted as synonymous with only (i. e. not as synonymous with how-PVC’s, when-PVC’s, etc.). For example, whereas an IPVC can be interpreted as equivalent to a that-PVC, as in(61, a), or as equivalent to a how-PVC, as in (61,b), its putative passive can only be interpreted as equivalent to a that-PVC (as in (62, a) and is therefore incoherent when a how-interpretation is invited (as in 62,b)): that-PVC’s (61) (a) (I wonder whether John has ever mentioned me.) heard him utter my name ? (b) (John has a funny way of pronouncing my name. ) heard him pronounce it&dquo; (6?) (a) (I wonder whether John has ever mentioned me.) been heard to utter my name‘~ (b) (John has a funny way of pronouncing my name.) been heard to pronounce it-% Have you Have you Has he 1<Eas he ever ever ever ever Similarly, (63, a) (which uses an IPVC in a context inviting a how-interpretation is fine, but (63,b) (which uses a passive construction with a to-IPVC in the same context) is incoherent: 40 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 (63) (a) When (b) one hears him talk one gets the impression that he is of French birth. *When he is heard to talk one gets the impression that he is of French birth. Notice that (63, c) is again impeccable, because he is heard talking is one hears him talking (see also Declerck (1981d)): truly the passive of (63) (c) When he is heard talking, French birth. 6. one gets the impression that he is of Conclusion 6. 1. In this paper I have attempted to define the distinctions between the different types of PVC’s as precisely as possible. From this discussion it has become clear that constructions of the type (3,d) behave syntactically and semantically not like sentences of the type (3, a) but like sentences involving that-PVC’s or to-IPVC’s. Since assuming a voice relation between two sentences is always an attempt at accounting for semantic and syntactic similarities, we must conclude that (3, d) is not the passive counterpart of (3, a), hence that sentences with IPVC’s are not passivizable. (I will point out below that this conclusion concurs with the fact that other Germanic languages where toIPVC’s do not occur do not have a construction corresponding to (3, d) they do have an IPVC construction corresponding to (3, a). ) though 6.2. The above conclusion raises nature: a number of questions of a theoretical As long as (3,d) was considered the passive of (3, a) one had to assume also present in the underlying structure of (3, a). Since to must be generated under either AUX or COMP it followed that this could be considered as evidence for the claim (rejected by Gee (1975) and Akmajian (1977)) that IPVC’s are S’s in underlying structure. Now that the voice relation between (3, a) and (3, d) has been called in question, this is no longer a valid argument. (See, however, Declerck (1981b, 1981c) for other arguments to the same effect.) a. that to was b. If the hypothesis that I have defended is correct it means that to in sentences like (3, d) and in to-IPVC’s in general is an item with a lexical content than has generally been assumed. Moreover, since this to has the same meaning as that, it seems unmotivated to derive to from AUX if that is seen as a lexicalization of COMP. Rather it will have to be assumed that both to and that are lexicalizations of COMP, the former in nonfinite PVC’s, the latter in finite ones. much fuller c. The analysis argued here is incompatible with the analysis of raising constructions that is suggested in Bach (1977:625). Bach (1977) interferes in the controversy between Chomsky (1973) and Postal (1974) with the suggestion 41 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 Chomsky’s non-raising analysis of (64, c) and Postal’s Subject-toObject raising analysis, there is a third alternative for deriving (64, c) from (64, a), viz. by passivizing (64, a) into (64,b) and deriving (64, c) from (64,b) by Subject-to-Subject raising: that besides these truths to be self-evident. *These truths to be self-evident is held by everyone. (c) These truths are held by everyone to be self-evident. (64) (a) Everyone holds (b) Pillinger (1980:79-81) proposes the Latin, where all three stages are same derivation for similar sentences in actually grammatical: (65) (a) Omnes credunt Gallos esse altissimos. (b) Gallos esse altissimos ab omnibus creditur. (passivization) (c) Galli esse altissimi ab omnibus creduntur. (Subject-to-Subject raising. Since sentences like (3, a) are often assumed to have undergone Subject-tosee also Object raising (cf. e.g. Huddleston (1971), Horiguchi (1978) Declerck (1981b)), one might think of extending the analysis suggested by Bach to sentences like ~3,d). However, this is incompatible with the conclusion I have reached in this paper since any analysis that assumes (3, d) to be (directly or indirectly) derived from (3, a) disregards the semantic and syntactic differ-- ences between the two sentences. d. in the The analysis I have defended raises the question potentialities of PVC constructions: (66) (a) (d) why there are gaps the street. the street. *I saw John to cross the street. John was seen to cross the street. I saw John (b) *John was (c) of cross seen cross The question of why (66, a) does not have a passive counterpart (66,b), while (66, d) does not have an active counterpart (66, c) is one that I cannot answer. All I can do is point out that in other languages, too, gaps like these appear to be present. The following sentences are the equivalents of (66, a-c) in Dutch and German: (67) (68) K. U. (a) Ik zag Jan de straat oversteken. (b) *Jan werd gezien de straat oversteken. (c) *Ik zag Jan de straat over te steken. (d) *Jan werd gezien de straat over te steken. (a) Ich sah Johann die Strasse uberqueren. (b) *Johann wurde die Strasse uberqueren geshen. (c) *Ich sah Johann die Strasse zu uberqueren. (d) *Johann wurde die Strasse zu uberqueren gesehen. Leuven, Campus Kortrijk, Belgium. 42 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 NOTES This 1 claim is in the framework of Chomsky (1980) where, rules, there is only one syntactic rule, viz. the transformational rule ’move A, where A is a category.’ Structure-building transformations have no place in the system. clearly incorporated apart from the base Criterion 2 (b) will even be inadequate for those speakers who do not reject the following example from Kruisinga (1931:208-209): (i) As he stood on the pavement by the fountain and watched them all -passing by--just like what he had heard the Lord Mayor’s show to be he heard one lady in a carriage say to her companion--"Look at that little boy. In this sentence the subject of the parenthetical clause involving the to -IPVC has no role in the formulation of the contents of the to -IPVC (i.e., the to -IPVC communicates an external report). A similar example is to be found in Poutsma (1926:427): (ii) I’ve never heard anyone to touch you. " (Poutsma adds: "Observe that This 3 to hear = to be told and to touch = to equal .") with the observation in Kirsner & explanation Thompson (1976: that negative IPVC’s/PPVC’s are relatively more acceptable if next to the implication of indirect perception direct perception is crucially involved as well, i.e. if some participant in the non-occurrence is "on the scene and observable. " Thus, (i) is better than (ii) because it does not deny that something could be seen: The director saw Olivier not trip over the footstool. (i) ? (ii) *Mary heard Bill not come home last night. concurs 234) This 4 does not alter the fact that slight differences may be observed in the -IPVC’s and that -PVC’s--differences which are due to the potentialities of to fact that, unlike the former, that -PVC’s have the superficial structure of a complementizer plus a finite clause. For example, the fact that there is no complementizer between the perception verb and a -IPVC, together with the to fact that the subject of the to -IPVC appears in the object form, entails that some degree of verb-object relation will mostly be taken to exist between the perception verb and the subject of the to -IPVC. (For this reason many transformationalists assume that the NP in question is raised from the complement clause to the matrix.) That this is so appears from the fact that, if possible, the subject of the to -IPVC must indicate the entity on the perception of which the inference is based. Compare: (i) (a) I observed the door to have been shut. (b) *I observed John to have shut the door. (ii) (a) We noticed the house to have been painted. (b) *We noticed John to have painted the house. -PVC’s That (iii) (a) (b) naturally fail to show this restriction: I observed that the door had been shut. I observed that John had shut the door. 43 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 -PVC’s is that there are Another difference between to -IPVC’s and that restrictions on the former that are not shared by the latter. For example, after an active perception verb to -IPVC’s cannot involve a dynamic predicate, unless it is in the progressive form: (iv) (a) *I observed John to work. (b) I observed John to be working. (see also section 6.2.d.) Kirsner 5 and ( 1977) attempts to explain the difference in acceptability between (i, a) (i,b) (i) (a) John was seen to cross the street. (b) *John was watched to cross the street. from the fact that watch is an agentive verb: he claims that for a passive construction to be felicitous it is necessary that the logical subject (here, the perceiver) be more ’agent-like’ or ’potent’ than the logical object (John). According to Kirsner, using watch rather than see has the effect of suggesting more potency on the part of John (because an event that is sufficiently interesting to watch must involve change and therefore an agent capable of bringing that change about). The ’potency gap’ between the perceiver and John is therefore decreased in (i,b) and the sentence is consequently unacceptable. However, if this explanation were correct it would also hold for passive constructions involving PPVC’s, such as (ii,a-b). However, there is no difference in acceptability between these two sentences. (ii) (a) John (b) John was seen was crossing the street. crossing the street. watched It seems therefore better to trace back the unacceptability of (i,b) to the restriction (observed above) that agentive perception verbs cannot be followed by PVC’s expressing indirect perception (i.e., that -PVC’s and to -IPVC’s, but not PPVC’s). Kirsner (1977:175) 6 makes the same observation in connection with the following sentences : (i) (a) They heard Sue sing for three hours that evening. (b) Sue was heard to sing for three hours that evening. Kirsner remarks that "the passive [i,b] seems strange unless interpreted referring to a situation in which either different individuals or the same individuals at different times accidentally come across Sue singing. The durative interpretation possible in the active sentence [i, a], ’iteratively,’i. e. , as where the same audience listens for the whole three hours is disfavored. " Kirsner attempts to explain this from the fact that perception verbs like watch cannot be used in sentences such as (i,b): "If the agentive sensory verbs, 44 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 favoring durative interpretations, do not occur in passives, one might expect those verbs that do to be taken as nondurative. " We have seen, however, that neither of these characteristics (the impossibility of using agentive perception verbs and a necessarily punctual interpretation) is typical of passives such as -PVC’s and to (i,b) only. They are also found with that -IPVC’s in general and should therefore be related to the meaning of ’indirect’ perception which is typical of all these constructions. REFERENCES 1977. "The complement structure of perception verbs syntax framework. " In: P. Culicover, T. Wasow & A. Akmajian, eds. , Formal Syntax . New York, San Francisco, London: Akmajian, Adrian. in an autonomous Academic Press. Bach, Emmon. 1977. Review article of Postal (1974). Language 53:621-654. Chomsky, Noam. 1973. "Conditions on transformations. " In: S. R. Anderson . New York: Holt, & P. Kiparsky, eds., Festschrift for Morris Halle Rinehart & Winston. 1980. "On binding." Linguistic Inquiry 11:1-46. Comrie, Bernard. 1976. Aspect . Cambridge: Cambridge University Press. Declerck, Renaat. 1981a. "On the role of progressive aspect in nonfinite Glossa 15:83-114. perception verb complements. " 1981b. "The structure of infinitival perception verb complements in a transformational grammar. " to appear in: L. Tasmowski-De Ryck, ed., Problems in Syntax. 1981c. "On the derivation of Dutch bare infinitives after perception verbs. " to appear in Theoretical Linguistics . 1981d. "The triple origin of participial perception verb complements." MS. Emonds, Joseph. 1976. A Transformational Approach to English Syntax : . New York, San Root, Structure-preserving, and Local Transformations Francisco, London: Academic Press. Gee, James Paul. 1975. Perception, Intentionality, and Naked Infinitives: a Study in Linguistics and Philosophy . Ph.D. dissertation. Stanford Univer-----. ----- ----------------- ----- sity. 1977. "Comments on Akmajian (1977). " In: P. Culicover, A. Akmajian, eds. , Formal Syntax . New York, San Francisco, London: Academic Press. Horiguchi, Ikiko. 1978. Complementati on in English Syntax: a Generative Semantics Approach . Ph.D. dissertation. Georgetown University. Huddleston, Rodney. 1971. The Sentence in Written English . Cambridge: Cambridge University Press. Jespersen, O. 1961. A Modern English Grammar on Historical Principles . Vol. V . London & Copenhagen: Allen & Unwin. Kirsner, Robert S. 1977. "On the passive of sensory verb complement sentences. " Linguistic Inquiry 8:173-178. ----- and Sandra Thompson. 1976. "The role of pragmatic ----- T. Wasow & 45 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016 inference in semantics: a study of sensory verb complements in English. " Glossa 10:200-240. Kruisinga, E. 1931. A Handbook of Present-Day English. Part II: English Accidence and Syntax . Vol. 1. Groningen: P. Noordhoof. Pillinger, O. S. 1980. "The accusative and infinitive in Latin: a refractory complement clause. " Journal of Linguistics 16:55-83. Postal, Paul M. 1974. On Raising: One Rule of English Grammar and Its Theoretical Implications . Cambridge, Massachusetts & London: the M. T. T. Press. 1926. A Grammar of Late Modern English. Part 2: The Parts of Speech. Section 2. The Verb and the Particles . Groningen: P. Noordhoff. 1929. A Grammar of Late Modern English. Part 1: The Sentence. Second Half: The Composite Sentence . Groningen: P. Noordhoff. Quirk, R., S. Greenbaum, G. Leech & J. Svartvik. 1972. A Grammar of Poutsma, H. ----------- Contemporary English . London: Longmans. Rouveret, Alain & Jean-Roger Vergnaud. 1980. "Specifying reference to the subject: French causatives and conditions on representations." Linguistic Inquiry 11:97-202. Spears, A rthur K. 1977. The Semantics of English Complementation . Ph. D. dissertation. University of California, San Diego. 46 Downloaded from eng.sagepub.com at PENNSYLVANIA STATE UNIV on February 20, 2016
© Copyright 2026 Paperzz