<<

ON PHRASE COMPLEMENTATION*

SHUJI CHIBA Tsuda College

Those verbs which allow or require the present subjunctive in the embedded clause may generally be expected, from their semantic properties, to take a transitive VP complement. However, this is not always the case. Besides transitive VP complements, some of them will be found to take other kinds of infinitival complements; we will find that these present subjunctive verbs can be divided into several different types as to their subcategorization properties. This paper explores these syntactic properties of some of those subjunctive- taking verbs, especially in the framework of the Government and Binding theory.

1. INTRODUCTION. Subcategorization properties of verbs may be expected to be completely predicted by the semantic characteristics of those verbs. However, since we lack, at present, the complete description of the semantic characteristics of verbs, we cannot predict their subcategorization properties to the smallest detail. For example, as Williams (1974: 38) points out, 'the three verbs decree, order, and demand are similar semantically, but their complement struc- tures differ radically,' as shown in 1: {decreed (1) a. John *ordered that Bill leave.1 demanded} {*decreed

b. John *ordered of Bill that he leave. demanded} * This is a revised version of sections 1-7 of my paper, Chiba 1984 and is based on my talk at the annual meeting of the English Linguistic Society of Japan which was held at Tsukuba University in November, 1984. I am grateful to those people who gave me invaluable comments on my talk there. I would also like to thank Akira Ota, whose insightful comments helped shape this paper. Of course, the responsibility for remaining inadequacies is entirely my own. 1There is a dialect variation concerning the possibility of the subjunctive clause after the verb order, as we will see more closely in section 4.

-81- 82 ENGLISH , VOLUME 2 (1985)

(*decreed c. John ordered Bill to leave. *demanded}

{*decreed d. John *ordered to leave. demanded} The purpose of this paper is to investigate, in the framework of the Government and Binding theory (henceforth the GB-theory), the subcategorization properties of some of those verbs which have generally been considered to be transitive VP complementation verbs. Especially our attention will be drawn to those verbs which allow or require the present subjunctive in the embedded clause, as in 1a above. These verbs, it will be shown, do not behave in the same way as to the syntactic structures which they can take, although they are the same in that they take the subjunctive clause. The verbs which we take up in this paper are the following: advise, ask, beg, command, demand, dictate, insist, order, request, require, and urge. (In section 8, we will consider the case of force, although it usually does not allow the present subjunctive clause.) In the following sections we will pay attention to some of the important subcategorization properties of these verbs and try to describe these properties in the framework of the GB-theory.

2. SOME IMPORTANT GENERAL PRINCIPLES IN THE GB-THEORY. Before proceeding directly to this task of ours, it would be appropriate to briefly mention here some of the important general principles of the GB-theory which we will appeal to in the following sections, when we try to describe the subcategorization properties of those verbs mentioned above. In the GB-theory it is assumed that the properties and the distribu- tion of empty categories (i.e. PRO and trace) can be accounted for by the theories of government, binding, and Case. The theory of govern- ment contains the Empty Category Principle (ECP): (2) ECP: nonpronominal empty categories must be properly governed By way of illustration, consider the following examples: (3) a. Bill was believed [s t to be competent] b. *Bill was wanted [s'[s t to be competent]] In 3a, the trace t is properly governed by the main verb, while in 3b, t is ON TRANSITIVE VERB PHRASE COMPLEMENTATION 83 not so governed because of the boundary S'. (Note that believe is one of those verbs which can delete the S' category by S'-deletion.) Thus, the ill-formedness of 3b can be ascribed to the ECP. Next, as a second general principle in the GB-theory, let us consider the Binding Theory, especially its principles A and B: (4) Binding Theory (A) An anaphor is bound in its governing category (B) A pronominal is free in its governing category To see how principle A can account for the ill-formedness of relevant examples, compare the following two sentences: (5) a. It was believed that John is competent. b. *John was believed [s' (that) [s t is competent]] Principle A requires t in 5b, which is an anaphor, to find its antecedent in its governing category, the embedded S in this case. This requirement, however, is not satisfied; hence principle A of the Binding Theory is violated. (You will find that 5b violates the ECP also.) PRO, a pronominal without a phonetic matrix, is considered a pronominal anaphor. From this dual characteristics of PRO, we eventually get the following derivative principle: (6) PRO is ungoverned Since PRO is governed in 7b-c (by the main verb in the former case, and by the agreement element AGR in the embedded S in the latter case), both of the sentences are ungrammatical: (7) a. We wanted [s'[s PRO to be competent]] b. *We believed [s PRO to be competent] c. *We believed [s'[s PRO is competent]] On the other hand, 7a is grammatical, because PRO is not governed. The last general principle we consider here is the Case Filter, which is stated as follows: (8) The Case Filter *NP, where NP has phonetic content, but no Case The Case Filter has the effect of blocking a sentence which contains an NP with a phonetic matrix which, however, has no Case. Consider, for example, the following sentences: (9) a. John expected [s Bill to help Mary] b. *John tried [s'[s Bill to help Mary]] In 9a, Bill can be assigned the objective Case by the main verb expected, because the category S' has been deleted by S'-deletion. In 9b, however, Bill will be left with no Case assigned to it, since it 84 ENGLISH LINGUISTICS, VOLUME 2 (1985) does not appear in a position to which Case can be assigned. (Note that try is not an S'-deletion verb.) So the sentence is blocked by the Case Filter. With these general principles as our guidelines, let us now turn to the description of those subjunctive verbs which we mentioned above.

3. THE VERBS Demand, Dictate, AND Insist. In this section, let us investigate the syntactic characteristics of the verbs demand, dictate, and insist. First, let us consider the following examples: {a) demanded (10) *John b) dictated Tom to leave immediately. c) insisted} As these examples show, demand, dictate, and insist cannot appear in this sentence structure. Next consider the following examples: {a) demanded (11) *Tom was b) dictated to leave immediately. c) insisted} The fact that the sentences in 11 are ungrammatical implies that the S-structures in 12a, b cannot be appropriate for these verbs: (12) a. Tom was Ven [NPt] [s' PRO to leave immediately] b. Tom was Yen [s t to leave immediately] c. Tom was Yen [s' [s t to leave immediately]] The reason for this is that t in 12a, b and PRO in 12a satisfy the general principles shown in section 2; if the S-structure of 11a-c is either 12a or 12b, the sentences in 11 would wrongly be predicted grammatical. On the other hand, the S-structure in 12c would correctly predict the ill-formedness of these sentences. That is, t in 12c is not properly governed, which is a violation of the ECP; cf. 3b in section 2. This structure would also go well with the fact that 10a, b, c, which are repeated below as 13a, b, c, respectively, with their presumed S-structure, are ungrammatical: {a) demanded (13) *John b) dictated [s'[s Tom to leave immediately]] c) insisted} That is, in 13a-c, the embedded subject Tom cannot be assigned a proper Case, thus a violation of the Case Filter; cf. 9b in section 2. This structure also seems to explain the fact that the sentences in 14 are ungrammatical: ON TRANSITIVE VERB PHRASE COMPLEMENTATION 85

{a) demand (14) *Who did John b) dictate [s'[s t to leave immediately]]? c) insist} Note that, in 14a-c, t must be properly governed, according to the ECP, which, however, is not satisfied in this structure; thus 14a-c are ungrammatical.2 So far, so good. However, consider the following sentences in 15 with their supposed structure shown in 16: {a) demanded (15) John b) *dictated to leave immediately. c) *insisted} {a) demanded (16) John b) dictated [s'[s RPO to leave immediately]] c) insisted} Supposing, as we did above, that the embedded structure [s'[s...]] is correct for these verbs, it should predict that all of these sentences are grammatical, since PRO in this structure satisfies the principle in 6 in section 2. But, in reality, only the a-sentence is grammatical. To be especially noticed here is that we cannot regard the S-structure of sentences like 15b, c as the one in 17: {a) dictated (17) John b) insisted}[NPPROi] [s'PROi to leave immediately] A structure like this is generally prohibited, because here the first PRO is governed, which is a violation of the general principle concerning PRO. Given these properties of demand, dictate, and insist, it would be natural to assign the structure [s/[s...]] only to demand among these three, and to regard the other two as those verbs which do not occur in any infinitival constructions. This conclusion would also be supported by the following examples: {a) demanded (18) *Max is the man John b) dictated to leave immediately. c) insisted}

2 We tentatively disregard the possibility of positing another t in the COMP position of the lower S', since, with that hypothesis also, we eventually arrive at the same conclusion. For further details, see Chomsky (1981: ch. 5) and Chiba (1984: §9). 86 ENGLISH LINGUISTICS, VOLUME 2 (1985)

{a) demanded (19) *John b) dictated the prisoner to be transferred to the c) insisted} prison hospital. {a) demanded (20) *The prisoner was b) dictated to be transferred to c) insisted} the prison hospital.3 As we saw above, these three verbs do not occur in the same syntactic constructions, though they are semantically similar in that they are much stronger expressions of a desire than ask, order, or advise, and that they involve a notion of power or authority on the part of the subject, as Riddle (1975: 473) notes.

4. THE VERBSCommand ANDOrder. As contrasted with the case of demand (as well as dictate and insist), command and order make the following sentences grammatical: {a) commanded (21) John b) ordered} Tom to leave immediately. {a) commanded (22) John b) ordered} them to transfer the prisoner to the prison hospital. {a) commanded (23) John b) ordered} the prisoner to be transferred to the prison hospital.4 The result of application of wh-movement to Tom in 21 and to them 3 Some of my informants, however, accept the demand version in such examples as the following: {a) demanded (i) She was reading a book, one of Merrin's that she'd b) *dictated c) *insisted} to be delivered to the house. {a) demand (ii) Which book did John b) *dictate to be read in a week? c) *insist} This suggests that, at least for such speakers, demand is one of those verbs which (in part) allow S'-deletion. 4 Note that, concerning the possibility of a 'small clause', these verbs show the same distribution as shown above: {*demanded/*dictated/*insisted (i) John commanded/ordered} the prisoner transferred to the prison hospital. ON TRANSITIVE VERB PHRASE COMPLEMENTATION 87 in 22 are, respectively, 24 and 25: {a) command (24) Who did John b) order} to leave immediately? {a) commanded (25) Max is the man John b) ordered} to transfer the prisoner to the prison hospital. If them in 22 is moved by NP-movement (i.e. passivization in this case), we get the following sentences: {a) commanded (26) They were b) ordered} to transfer the prisoner to the prison hospital. Given these properties of command and order (contrasted with those of demand), we can suggest for these verbs the S-structure in which the embedded structure is [s...] (which is the result of the application of S'-deletion to [s' [s...]]), following the case of the verb believe, as mentioned in section 2 concerning sentence 3a. If we assume this structure, those NPs which directly follow these verbs, whether they are lexical NPs, as in 21-23, or they are traces, as in 24-26, would be properly governed by these verbs. Therefore, in the former case, those NPs can be assigned Case properly, and in the latter case, the traces satisfy the ECP. In the S-structure of 26a-b, the NP trace would be bound in its governing category, i.e. the outer S, thus also satisfying the binding principle A. The S-structure proposed above for command and order seems to be also valid when we try to explain the fact that sentences such as 21a-b become ungrammatical when the embedded subject is PRO, as illustrated by the examples in 27 with their S-structure shown in 28: {a) commanded (27) *John b) ordered} to leave immediately. (28) *John {commanded/ordered} [s PRO to leave immediately] In 28, PRO is governed by the main verb, violating the principle concerning PRO. One might, however, ask whether this is the only S-structure for these verbs that is consistent with their syntactic behaviors we have observed so far. In fact, it seems that the structure NP S' is more appropriate at least for those cases in which the NP which directly follows command and order is a human noun (or the trace of it). This is because it seems that, in these cases, those NPs (or their traces) should be given a θ-role (thematic role) as the object of 88 ENGLISH LINGUISTICS, VOLUME 2 (1985) command or order (rather than as the subject of the embedded VP). If so, the S-structure of 21a-b, for example, will be something like the following:

(29) John {a) commanded b) ordered} [NP Tomi] [S' PROi to leave immedi- ately] This is also true with, at least, 22a-b, 24a-b, 25a-b, and 26a-b. This structure, it should be clear, can account for the fact that those sentences are grammatical, since the structure satisfies the relevant general principles shown in section 2. Here, we have two different S-structures for command and order which satisfy those general principles. It means that, correspondingly, we have two different D-structures (and LFs (=Logical Forms)) for these verbs, according to the projection principle, which stipulates that 'representations at each syntactic level (i.e., LF, and D-and S-structure) are projected from the , in that they observe the sub categorization properties of lexical items.' (Chomsky (1981: 29)) That is to say, we assume here a dual lexical categorization for command and order. One might point out that this assumption of dual categorization runs counter to the assumption of uniformity of lexical entry which Chomsky 1981 seems to consider to be generally preferable. (For example, see his comments on this matter in Chomsky (1981: 9, 105-6, 310).) Generally preferable as the latter assumption may be, it, of course, must be considered as an empirical hypothesis which can be maintained to the extent that it can account for the relevant empirical facts. In fact, it seems that there are some amounts of evidence which support the assumption of dual categorization suggested above for the verbs command and order. Palmer (1974: 197) includes command on the list of those verbs which he says 'seem to be simultaneously members of both the want and persuade class.' Other verbs on this list are allow, cause, enable, forbid, order, and permit. Consider the examples in 30, which he gives to show this fact: (30) a. I ordered the chauffeur to fetch the car. b. The chauffeur was ordered to fetch the car. c. I ordered the car to be fetched by the chauffeur. As Palmer (1974: 183) says: 'The passivization of the subordinate clause [as in 30c] ON TRANSITIVE VERB PHRASE COMPLEMENTATION 89

suggests that the chauffeur is not the object of ordered, and want could be substituted for order in both sentences [i.e. 30a and 30c]. Semantically, too, order seems to function in two ways. Either orders are given to someone to do something or they are given that certain things shall be done.' Therefore, the following S-structures would both be appropriate to sentence 30a: (31) a. I ordered [s the chauffeur to fetch the car] b. I ordered [NPthe chauffeurs] [s' PROi to fetch the car] On the other hand, the S-structures for 30b, c are, respectively, 32a, b: (32) a. The chauffeurs was ordered [NPti] [s' PROi to fetch the car] b. I ordered [s the car to be fetched by the chauffeur] Jespersen (1961: 285-93) suggests a similar syntactic analysis for these verbs. He mentions that 'verbs of permission and command offer some theoretical difficulties.' (p. 285) Concerning the construc- tion for verbs such as command, order, and charge, he states that it 'is not always clearly to be analyzed as V O (S2I) [which corresponds to our structure V S], but may be V Oi O (I) [which corresponds to our structure V NP S'].' (p. 288) He further says: 'In some cases, however, the infinitival nexus is clearly a direct object: V O (SI); thus especially with a passive: He ordered (commanded) the prisoner to be shot Doyle NP 1895 he ordered the cripple to be carried out again. [...] He ordered a taxi to come round at once: here two interpretations are possible, either S V O (S2 I) or S V O (12(I)) with the infin. as .' (p. 288) As it may be expected from the ambiguous character, illustrated above, of these verbs, judgments seem to vary among native speakers of English concerning the acceptability of sentences such as the following: (33) a. On the sixth day he {commanded/ordered} Smith to be transferred to the prison hospital. b. The prisoner was {commanded/ordered} to be trans- ferred to the prison hospital. Similarly, while some linguists regard the following sentence as ungrammatical: (34) I ordered Bob to be removed from the room,5

5 This is example 83b in Postal (1974:319 -20). Note also his grammaticalexample 90 ENGLISHLINGUISTICS, VOLUME 2 (1985) others consider grammatical such examples as the following: (35) a. He ordered John to be sacked. b. John ordered Mary to be examined by the doctor. c. They ordered him to be shot.6 Furthermore, there seem to be idiolectal variations concerning well- formedness of 'double passives' such as the following: (36) a. The table was ordered to be removed. b. The looting was commanded to be stopped. c. The flag was ordered to be hoisted. d. The report was ordered to be printed by Hansard.' Before closing this section, let us note again that there seems to be a dialect variation concerning the acceptability of sentences such as the following: (37) a. (*)John ordered that Tom leave immediately. b. (*)Joseph ordered that all of us read the book. As we pointed out in fn. 1, there are some speakers who would consider such sentences ungrammatical. To be noticed about this matter is that, even for those speakers, such sentences as the following, in which the embedded clause is a passive sentence, seem to be acceptable: (38) a. Joseph ordered that the book be read by all of us. b. On the sixth day he ordered that Smith be transferred to the prison hospital. Furthermore, it seems that, at least for some of those speakers, the accusative with infinitive version of 37a-b, on the one hand, and that of 38a-b, on the other, differ in their acceptability, as shown in 39: (39) a. John ordered Tom to leave immediately. b. *John ordered the book to be read by all of us. If the reaction of these speakers to these related sentences is generally as it is indicated here, one might suppose that this is an example of those cases in which the acceptability of a sentence is influenced by whether or not an alternative construction is possible.8 In order to show that this is really the case, we need 83d, 'I ordered the chair (to be) removed from the room.' 6 Examples 35a, b, c are those by Huddleston (1971:169), Culi cover (1982:276), and Harris (1982:104, n. 14), respectively. 7 36a, b are regarded as ungrammatical by Jespersen (1961: 317) and Harris (1982: 288), respectively. 36c, d are from Zandvoort (1975: 20) and Scheurweghs (1959: 230), respectively. 8 For another example, see Chomsky (1981: 230, fn. 68). ON TRANSITIVE VERB PHRASE COMPLEMENTATION 91 more empirical research.

5. THE VERBS Request AND Require. As we saw in the preceding section, grammaticality judgments can vary about those sentences in which such verbs as command and order are followed by infinitival complements, especially when passivization is involved. What is noteworthy about the case of request and require is that there are idiolectal variations observed even with sentences such as the following: {a) requested Tom to leave immediately. (40) John b) required} To each of these sentences, there seem to be some who would not accept it. For example, among the seventeen native speakers I asked, nine persons marked only 40a with an asterisk or a question mark, and, similarly, 40b was so marked by six persons. Judging from the definitions of request and require and the examples given in English dictionaries, there seem to be no semantic problem with the sentences in 40a-b. Shown below are the reactions of some group of my informants to 40a-b and some other similar examples: {a) requested (41) John b) *required} Tom to leave immediately.

{a) requested Tom to kill Max. (42) John b) *required}

{a) requested them to stop making such a noise. (43) John b) *required} One might conclude from this that such speakers allow neither the structure NP S' nor the structure [s...] to the verb require. However, they seem to accept both versions of such a sentence as the following:

{a) requested to leave immediately. (44) Tom was b) required} Furthermore, at least when the embedded subject is a nonhuman noun, the structure [s...] seems to be acceptable for both request and require, as the following examples suggest: {a) requested the book to be read in a week. (45) John b) required} 92 ENGLISH LINGUISTICS, VOLUME 2 (1985)

{a) requested (46) The book was b) required to be read in a week.9

{a) request to be read in a week? (47) Which book did John b) require} Although we need further empirical research on the possible structures concerning request and require for these speakers, let us tentatively posit for request the S-structures NP S' and [s...], and for require the structure [s...] (and in part the structure NP S'), with the condition that the embedded subject of the S-structure [s...] not be an agentive noun. According to some other informants, both request and require seem to allow the structure NP S', as suggested by the following examples: {a) requested (48) John b) required} Tom to leave immediately.

{a) requested Tom to kill Max. (49) John b) required} {a) requested (50) John b) required} them to stop making such a noise. They also accept 44a-b. However, for those speakers, only require seems to allow the S-structure [s...], as illustrated by the following examples: {a) *requested (51) John b) required} the book to be read in a week. {a) *requested to be read in a week. (52) The book was b) required} To complicate matters still further, the wh-moved version of 51a- b, i.e. 47a-b above, seem to be both acceptable to those speakers. Interestingly enough, however, this seems not to be an exceptional case. As far as my informant check is concerned, similar cases have been observed several times. First, out of seven of my informants who reject 53a, five persons accept 53b:

9 See also the following example: (i) His income is not required to be included in the return of his parent. (A Standard Sample of Present-DayEdited American English, for Use with Digital Computers,1680E1H24) Note, however, that Wood (1981:228) considers the following example ungram- matical: (ii) The money is required to be paid by Saturday next. ON TRANSITIVE VERB PHRASE COMPLEMENTATION 93

(53) a. *John requested Tom to leave immediately. b. Who did John request to leave immediately? Secondly, out of six informants who consider 54a marginal or ungrammatical, five persons find 54b grammatical or not so bad: (54) a. *John required Tom to leave immediately. b. Who did John require to leave immediately? Finally, let us consider a rather 'extreme' case including such verbs as beg, ask, advise, and demand:10 {a) *begged b) *asked c) *advised the book to be read in a week. (55) John d) *demanded e) *insisted f) ordered} {a) *begged b) ?asked c) ?advised to be read in a week. (56) The book was d) *demanded e) *insisted f) ordered} {a) beg b) ask c) advise to be read in a week? (57) Which book did John d) demand e) *insist f) order} As these examples show, insist and order are, respectively, con- stantly ungrammatical and constantly grammatical. However, for other verbs there is a change observed in their acceptability, according as the sentence patterns change from 55 to 57. Note especially the case of demand. The S-structure [s...], which seems not to be allowable to this verb, as far as 55d and 56d are concerned, seems to become necessary to account for the fact that 57d is grammatical.11

10 We added insist and order for the sake of comparison. 11 For another case of similar observations, see Quirk et al. (1972: 900) and Postal (1974: 310). 94 ENGLISH LINGUISTICS, VOLUME 2 (1985)

Although we do not know the reason why this should happen, there seems to be some interesting linguistic property involved here which should be given further consideration. Next let us pay attention to another difference between require and request which is illustrated by the following examples: {a) requested to leave immediately. (58) John b) *required} This suggests that only request can be followed by the S' complement at S-structure, as shown in 59: (59) Johni requested [s'[s PROi to leave immediately]] Notice that the structure [s'[s...]] is necessary here in order for PRO to satisfy the principle in 6 in section 2. On the other hand, as we saw above, there are those speakers who accept sentences such as 45a, 46a, and 47a, which we repeat here as examples 60a, b, c, respectively: (60) a. John requested the book to be read in a week. b. The book was requested to be read in a week. c. Which book did John request to be read in a week? As we have already shown, it is necessary to posit the structure [s...], not the structure [s-[s...]], for these sentences. Thus, it seems that request requires both complements [s'[s...]] and [s...]. This situation reminds us of the case of the verb want, which shows the following behavior: (61) a. John wants [PRO to win] b. John wants [Bill to win] c. *John was wanted [t to win] To explain this behavior of want, which might appear inconsistent concerning the effect of the general principles given in section 2, Chomsky (1981: ch. 2) proposes a for-infinitival complement, as in 62: (62) John wants [s' for [s Bill to win]] This is supported by the fact that sentences like 'John wants for Bill to win' are possible in some dialects and that the following examples are generally grammatical: (63) a. John wants very much for Bill to win. b. What John wants is for Bill to win. The embedded structure in 61a-c, then, will be labelled S', not S. For further details, see Chomsky (1981: ch. 2, p. 252, p. 298). Following this analysis for want, one might suggest a similar analysis of infinitival complements of request for those speakers ON TRANSITIVE VERB PHRASE COMPLEMENTATION 95 mentioned above, considering also the fact that the following sentence seems acceptable for them: (64) What John requested was for Tom to leave immediately. However, there is a problem to this analysis. That is, this analysis should suggest that 60b is ungrammatical, which it is not. Cf. 61c. For want of any good idea to explain this situation, we must leave this question open. Returning to example 58b and other relevant examples, we can assume that require undergoes S'-deletion. Thus, the structure of 58b, at the level where the principle in 6 in section 2 applies, is something like the following: (65) *John required [s PRO to leave immediately] This suggests that PRO generally cannot occur as the embedded subject of the infinitival complement of require. This, in turn, suggests that the S-structure of sentence 66 is not something like 67a but something like 67b: (66) Teachers require to be warned against shouting. 12 (67) a. Teachers; require [s PROi to be warned against shouting] b. Teachersi require [s ti to be warned against shouting] That is, we regard require in 66 as one of raising verbs such as happen, seem, appear, prove, and threaten. Compare the two sentences in each of the following pairs: (68) a. John proved Bill to be the culprit. b. John proved to be the culprit. (69) a. The killer threatened to murder me if I didn't obey.13 b. There threatens to be famine in Bulgaria.14 The S-structure representations of these sentences would be approx- imately the following: (70) a. John proved [s Bill to be the culprit] b. Johni proved [s ti to be the culprit] (71) a. The killeri threatened [s' PROi to murder me if I didn't obey] b. Therei threatens [s ti to be famine in Bulgaria] In 70b and 71b, the matrix subject has been raised to that position, which is not a θ-position, from the position occupied by t. Likewise,

12 Zandvoort (1975: 30). 13 Longman Dictionary of Contemporary English, s.v. threaten. 14 Postal (1974: 293). 96 ENGLISH LINGUISTICS, VOLUME 2 (1985)

67b can be considered the result of raising of the embedded subject, and this, we suppose, is the correct analysis for sentence 66. The same analysis would apply to sentences such as the following: (72) a. Irony requires to be used with care. b. One does not require to be specialist. c. A candidate requires to hold a certificate.15 With the raising analysis of require illustrated above, let us end this section and consider, in the next section, some other verbs which can be followed by an S' complement, as in the case of request.

6. THE VERBS Ask AND Beg. As we have seen, demand and request can be followed by an infinitival complement in which PRO is the subject. The same is true with ask and beg, as illustrated in examples 73a-b and their S-structure 74:

{a) asked to leave immediately. (73) John b) begged} (74) John {asked/begged} [s' PRO to leave immediately] These two verbs can also be followed by the embedded structure NP S', as in the following examples:

{a) asked Tom to leave immediately. (75) John b) begged} To be noticed about these sentences is that they can be ambiguous, at least in the case of ask, as to controller. Thus, the subject of the infinitive can be John (subject control) or Tom (object control). This is because ask and beg have a lexical property of taking the S' complement, as well as the NP S' complement. Besides NP S' and S' complements, these verbs seem to allow a for- infinitival complement. This should be compared with the case of demand and request, which do not allow it, as illustrated in the following examples:

15 72a is drawn from COD6, s.v. require and 72b-c are from Webster's Third New International Dictionary, s.v. require. Notice that there seems to be some dialectal variability as to well-formedness of sentences such as 72b-c. Thus, according to Wood (1981: 228), the following example is ungrammatical: (i) You require to have a University degree for a post of that kind. For some further comments about the acceptability of sentences like this, see also H. W Fowler, MEU2, s.v. require. ON TRANSITIVE VERB PHRASE COMPLEMENTATION 97

{a) asked b) begged for Tom to leave immediately. (76) John c) *requested d) *demanded} With regard to possibility of S'-deletion, the following examples suggest that ask and beg do not allow it: (77) a. *John {asked/begged} the book to be read in a week. b. *The book was {asked/begged}to be read by all of the students. However, further research on this matter is necessary before we can say anything more definite about it.

7. THE VERBS Urge AND Advise. That urge takes the NP S' comple- ment is suggested by the fact that the following pair of sentences are not cognitively synonymous, as pointed out by Quirk et al. (1972: 839): (78) a. I urged a doctor to examine him. b. I urged him to be examined by a doctor. The same thing is true with advise. These two verbs are also similar in that they cannot take the S' complement, as the following examples show: {a) urged to (be allowed to) leave immediately. (79) *John b) advised} Compare these ungrammatical sentences containing urge and advise with those grammatical ones containing ask and beg in 73. Then, is it possible for urge and advise to be followed by the embedded structure [s...] at S-structure? That this may probably not be possible is illustrated in the following examples: (80) a. *John {urged/advised} the book to be read in a week. b. *The book was {urged/advised} to be read in a week. c. *She was reading a book, one of Merrin's that she'd {urged/advised} to be delivered to the house. Based on the observation above, let us tentatively conclude that the only infinitival complement that is possible for urge and advise is NP S'.

8. THE VERBForce. Although the verb force does not generally allow the present subjunctive clause to occur with it, as shown in 81, it would not be out of place to take up the verb here and consider 98 ENGLISH LINGUISTICS, VOLUME 2 (1985) its subcategorization properties: (81) *John forced (that) Tom leave immediately. Since the appearance of the first comprehensive study, in the framework of transformational grammar, on English predicate complementation by Rosenbaum 1965, the verb force has generally been treated as one of those verbs which take only transitive VP complementation. The main reason for considering force a transitive VP complementation verb is the fact that the pair of sentences in 82 are not cognitively synonymous:16 (82) a. The detectives forced the DA to interrogate the butler. b. The detectives forced the butler to be interrogated by the DA. As another reason, one may cite some ungrammatical sentences with force which would be grammatical if the verb is replaced with object NP complementation verbs such as expect: {a) *forced there to be three men at the party. (83) John b) expected} Similarly, ungrammatical sentences such as those in 84 may be used to construct the hypothesis that force belongs only to transitive VP complementation verbs:17 (84) a. *I forced Sam's proposal to be rejected by the committee. b. *I forced the earthquake to destroy the city. c. *I forced the election to be a landslide. d. *I forced it to be obvious to everyone that the earth is round. Based on their observation of these ungrammatical sentences, Soames & Perlmutter (1979: 469-70) conclude that 'there are selectional restrictions on the surface objects of force; only those NPs that are capable of denoting animate beings can be objects of force.' Is this generalization of the selectional restriction on the surface objects of force generally true? The purpose of this section is to show that it is not completely true. First, Gruber (1976: 168), illustrating some of the characteristics of causative agentive verbs such as cause, make, coerce, induce, force, and drive, notes that the surface objects of these verbs do not have to be human, as shown in 85:

16Examples 82a-b are from Rosenbaum (1967:118). 17All of these examples are drawn from Soames & Perlmutter (1979:469 -70). ON TRANSITIVE VERB PHRASE COMPLEMENTATION 99

(85) John forced the roof to cave in.18 Next, Celce-Murcia & Larsen-Freeman (1983: 480) also cite a grammatical example such as the following: (86) He forced the door to be open. Following are some of the similar examples I have collected: (87) a. [T]he English language forces the noun to be either singular or plural. (McCawley (1981: 111)) b. The theory proposed here forces want to be a case of NOC, ... (LI 11 (1980), p. 215) c. [I]t will force certain changes to be gradual in the sense that ... (Lightfoot (1979: 143)) d. The king forced the growing demand for decentrali- sation to take an antimonarchical form. (Scheurweghs (1959: 228)) e. The same evidence would force a theory based on antecedency only to add an exception clause to the Network Filter, ... (Williams (1971: 77)) It would be interesting to consider here Chomsky's comment about the structure of the sentences containing force. Chomsky (1981: 147, n. 108) says that the following examples seem moderately acceptable: (88) a. They forced it to rain (by seeding the clouds). b. They forced better care to be taken of the orphans (by passing new laws). According to Chomsky, these examples are only derivatively generated (in the sense of Chomsky (1965: 227; 1972: 27f.)). As evidence supporting this conclusion, he points out that these

18He also mentions that the embedded verb occurring with the main verb force must be active, giving the following examples (Gruber (1976:170)): (i) a. John forced the ball to role into the hole. b. *John forced the ball to be a red color. c. *John forced the ball to weigh five pounds. A similar observation is made by Givon (1975:78-80). Notice, however, that he shows the following example with a question mark, not an asterisk: (ii) ?She forced the confusion to take place. Note also the followingexamples, which are drawn from Langacker(1975: 368): (iii) a. He forced the peg to enter the hole. b. The rock on his head forced Elvin to sink faster in the quicksand. These examples show, as he says (p. 368), that 'force does not always imply volition.' Especially notice that ia, iiia, and in part ii constitute counterexamples to the constraint on the object of force mentioned in the text. 100 ENGLISH LINGUISTICS, VOLUME 2 (1985) examples are resistant to further grammatical operations, as shown in 89: (89) a. *It was forced to rain. b. *Better care was forced to be taken of the orphans. Admitting that 88a-b are only moderately acceptable and that they are only derivatively generated, as Chomsky says, what about the sentences in 85-87? Are they also only moderately acceptable? If they are grammatical, they would suggest the necessity of postulating a structure such as 90, at least for these sentences: (90) It will force [s certain changes to be gradual] Although I have not examined yet whether or not the passive version of each of those sentences in 87a-e is acceptable, the presence of such passive examples as the following seems to suggest this possibility: (91) a. If the adverb placement transformation is forced to apply, … (Reibel & Schane (1969: 478))

b. In § 3 of Section 81 we saw that the Pronominalization Hypothesis is forced to posit an ad hoc constraint... (Soames & Perlmutter (1979: 358)) c. If the English Stress Rule were forced to respect the foot boundaries assigned earlier, ... (Hayes (1981: 159)) If sentences such as those in 87 and 91 are really well-formed sentences, it might become necessary to postulate for force a struc- ture such as the one in 90, even if partly, besides the structure NP S', which is in any way necessary for those sentences in which the object NP following force is a human noun.

9. CONCLUSION. In this paper, we have seen some of the important syntactic properties of subjunctive-taking verbs, espe- cially with respect to the types of infinitival complements those verbs take.19 Those verbs which we considered here, which are sometimes called 'speech-act verbs', share an important semantic property connected with giving direction to others to do or not to do something. This semantic property is usually closely con- nected with the syntactic properties represented by the structures of present subjunctive clauses and transitive VP complements.

19 For other interesting properties which we have not taken up in this paper, see Chiba 1981, 1982, 1983, 1984. ON TRANSITIVE VERB PHRASE COMPLEMENTATION 101

However, this close connection between these two properties does not always turn out the same for each of these verbs. We found that there are several different types of realization of this connection. We investigated these different types especially in the framework of the GB-theory. By our investigation, in this study, of some of the major transitive VP complement verbs and some other related ones, I hope we have deepened our understanding of the syntactic and semantic properties of these verbs. Although there are many problems left unsettled concerning the properties of these verbs, this paper, I hope, will have something to contribute to further studies on those problems.

REFERENCES

CELCE-MURCIA, MARIANNE and D. LARSEN-FREEMAN. 1983. The grammar book. Rowley, Mass.: Newbury House. CHIBA,SHUJI. 1981. On present subjunctives in present-day American English. The Tsuda Review 26.67-105. _. 1982. On three major aspects of present subjunctives in English. The Tsuda Review 27.1-26. _. 1983. On the position of preverbs in present subjunctive clauses in English. The Tsuda Review 28.107-35. _. 1984. On present subjunctives and transitive verb phrase complemen- tation in English. The Tsuda Review 29.33-77. CHOMSKY, NOAM. 1965. Aspects of the theory of . Cambridge, Mass.: The MIT Press. _. 1972. Studies on in . The Hague: Mouton. _. 1981. Lectures on government and binding. Dordrecht: Foris Publications. CULICOVER, PETER W. 1982. Syntax. 2nd ed. New York: Academic Press. GIVON, TALMY. 1975. Cause and control: on the semantics of interpersonal manipulation. Syntax and semantics, vol. 4, ed. by J. P. KIMBALL, 59-89. New York: Academic Press. GRUBER, JEFFREY S. 1976. Lexical structures in syntax and semantics. New York: North-Holland. HARRIS, ZELLIG. 1982. A grammar of English on mathematical principles. New York: John Wiley & Sons. HAYES, BRUCE. 1981. A metrical theory of stress rules. Reproduced by the Indiana University Linguistics Club. HUDDLESTON, RODNEY D. 1971. The sentence in written English: a syntactic study based on an analysis of scientific texts. Cambridge: Cambridge University Press. JESPERSEN, OTTO. 1961. A modern English grammar on historical principles. Part V. London: George Allen & Unwin, Ltd. 102 ENGLISH LINGUISTICS,VOLUME 2 (1985)

LANGACKER,RONALD W. 1975. Functional stratigraphy. Papers from the parasession on functionalism, ed. by R. E. Grossman et al., 351-97. Chicago: The Chicago Linguistic Society. LIGHTFOOT,DAVID W. 1979. Principles of diachronic syntax. Cambridge: Cambridge University Press. MCCAWLEY,JAMES D. 1981. Everything that linguists have always wanted to know about logic but were ashamed to ask. Chicago: The University of Chicago Press. PALMER,F. R. 1974. The English verb. London: Longman. POSTAL,PAUL M. 1974. On raising: one rule of English grammar and its theoretical implications. Cambridge, Mass.: The MIT Press. QUIRK, RANDOLPH et al. 1972. A grammar of contemporary English. London: Longman. REIBEL, DAVIDA. and S. A. SCHANE.1969 (eds.) Modern studies in English. Englewood Cliffs, New Jersey: Prentice-Hall. RIDDLE, ELIZABETH.1975. Some pragmatic conditions on complementizer choice. CLS 11.467-74. Chicago: The Chicago Linguistic Society. ROSENBAUM, PETER S. 1965. The grammar of English predicate comple- ment constructions. MIT dissertation. [Published, Cambridge, Mass.: The MIT Press, 1967.] SCHEURWEGHS, G. 1959. Present-day English syntax. London: Longmans. SOAMES, SCOTT and D. M. PERLMUTTER. 1979. Syntactic argumentation and the structure of English. Berkeley, California: University of California Press. WILLIAMS, EDWIN S. 1974. Rule ordering in syntax. MIT dissertation. WILLIAMS, GEORGE M., JR. 1971. Network of anaphora: An essay in the syntax of pronominalization. MIT dissertation. WOOD, FREDERICK T. 1981. Current English usage. Revised ed. Revised by R. H. Flavell and L. M. Flavell. London: The Macmillan Press, Ltd. ZANDVOORT, R. W. 1975. A handbook of English grammar. 7th ed. London: Longmans, Green & Co., Ltd.