Printer Friendly

Infinitival copular complement clauses in English: explaining the predominance of passive matrix verbs (1).


This article first provides empirical evidence from the 100-million-word British National corpus in support of earlier less well documented observations that infinitival copular complements occur much more often with passive matrix clauses (e.g. Towns were thought to be organic or haphazard creations until recently) than with active matrices (e.g. One might even have though him to be angry), and then addresses the pending question of why this should be so. A previous explanation has invoked the different word order of the passive construction versus that shared by the active construction and finite complements, combined with the purported markedness of nonfinite constructions. The alternative explanation presented here is based on the fact that though finite complements and infinitives with active matrices share the same word order, their typical information structure is different, in that the subjects of the former can introduce "new" referents, while "givenness" seems almost a necessary condition for the subjects of the latter. It is claimed that the typical information structure of active matrices plus infinitives makes the construction a dispreferred one, because it leads to redundancy in referential continuity and causes conflicts between candidates for sentence topic/theme status, disrupting the thematic progression of the text.

1. Introduction

In the wake of Postal's (1974) book On Raising, which was itself triggered off by work by Lees (1960), Rosenbaum (1967), and Chomsky (1973), a number of studies appeared in the linguistic mainstream of the day that addressed the issue of the supposed difference in meaning between infinitival and finite complement clauses after verbs like believe, consider, find, prove, show, etc. (e.g. Borkin 1974, 1984; Riddle 1975; Steever 1977; Maxwell 1984; for a discussion see Mair 1990:196-200; and Noel 1997). (2)

(1) a. The recent series of disasters from the Herald of Free Enterprise onwards, has led us to believe that Britain is particularly unsafe (A3B 84).

b. We cannot claim these life stories to be "representative" in the strict social scientific sense, but we do believe them to be "valid" (AP7 284).

(2) a. Mortensen (1933a) considered that O. clavigera is possibly distinct from Ophiomitrella cordifera, pointing out that the difference in shapes of the oral shield and disk spinclets could warrant specific distinction (H79 1394).

b. Correct me if I'm wrong, but I never saw the likes of "Doyler" sneaking off the pitch for a quickie, unlike, say, Jimbo Morrison or the boys in Led Zep who considered such interludes to be an amusing challenge to their versatility as artistes (ED7 1733).

(3) a. Many currants are now grown in Australia, but I find that the best are the tiny Greek vorzittzas, which are sweet and plump (ABB 1405).

b. This radical critique of workplace organization under modern capitalism finds technology to be developed and applied in ways consistent with the dominant relations of production (HTF 1444).

(4) a. "I aim to prove that privacy is a thing of the past," says Dame Edna. "I prod and pry in all the little nooks and crannies but it is all done in a caring way" (CH1 4407).

b. The newly published Annual Report on last season's excavations by the British School at Rome at San Vincenzo al Volturno, Molise, in Central Italy, gives details of the remains of the apse of the ninth-century Carolingian abbey church of San Vincenzo Maggiore, proving the church to be the largest of its date south of the Alps (CKU 873).

(5) a. Many surveys show that industrialists are unsure about the specific skill demands of the jobs they offer and appear to be only certain that what they need is not more trained labour but access to cheaper raw materials (F9E 1575).

b. The weakness of glass fibres brings us to the question of Griffith cracks and it also brings us back to professor Inglis, whom we left in Chapter 2 worrying about why ships broke in two at sea when simple calculation showed them to be amply strong enough (CEG 186).

(6) a. Why do some people think that faith is necessarily troubled by doubt but that knowledge is not? (C8V 397).

b. BCAR Chairman Steve Challis welcomed delegates from museums and groups all over the UK to the meeting, explaining that he thought BCAR to be "probably one of the smallest groups in the BAPC" (CGL 1252).

There is no received opinion on the purported semantics of the choice between infinitives and that clauses in such sentences. Elsewhere I have provided evidence in support of the thesis that this choice is not semantically but pragmatically motivated, accounting for it in terms of information structure and discourse organization (Noel 1997). The present study addresses the observation (made by, e.g., Postal 1974: 305; Bolinger 1977: 129; and Quirk 1965, albeit on the basis of little evidence) that the infinitival complement more often has a passive matrix than an active one (compare the [c] examples below with the [b] examples above) and argues that this imbalance can be explained along the same lines. The studies mentioned above on the "meaning" of infinitival complements vs. that of that clauses did not address this problem, which indeed cannot be accounted for with semantic conjecture.

(1) c. 1 in 500 Londoners are believed to be infected (A00 22).

(2) c. Although the Victorian era is considered to be the great age of plant collectors, the ancient Egyptians recorded plant collecting expeditions in their hieroglyphics (A0G 130).

(3) c. A recent survey by the Consumers Association (CA) showed that a third of the tools hired from 32 randomly selected shops were found to be faulty when examined by experts (A70 401).

(4) c. Some of the social and economic assumptions on which Beveridge and other pioneers had based their welfare schemes had been proved to be over-optimistic (A66 431).

(5) c. Masculine sexuality is' shown to be complex and unstably implicated within the whole social domain (A6D 935).

(6) c. Fish is' thought by many to be "brain-food" (B7D 1652).

Before turning to an explanation, however, the present article will first establish the size of the proportion of passive to active matrices for about 60 verbs that display the alternation between that clauses and infinitival copular complements, because to date the extent of this lopsided proportion has not been properly documented. Mair's (1990) very thorough and encompassing study of Infinitival Complement Clauses in English, which includes a section on "The textual and structural factors responsible for the skewed relationship between actives and passives," is based on evidence from the corpus of the Survey of English Usage, which at the time Mair collected his data comprised 895,000 words (Mair 1990: 13), but "[t]he only passive matrix verb attested often enough in the corpus to warrant any definite conclusions is say" (Mair 1990: 183), of which there were 24 occurrences. The data that will be presented here are taken from the more than one hundred times larger 100-million-word British National Corpus (BNC) (Burnard 1995).

2. The predominance of passive matrix verbs with infinitival copular complements

The verbs that were selected for investigation are the sixty-odd "B-element R-triggers" (i.e. verbs of the type of believe that "trigger raising," or, in more theory-neutral terms, that display variation between that clauses and to infinitival complements) that Postal (1974:297-317) lists in his chapter on "The scope of raising in clause domains." Here they are in alphabetical order: (3)

(7) acknowledge, admit, affirm, allege, ascertain, assume, believe, certify, concede, consider, declare, decree, deduce, deem, demonstrate, determine, discern, disclose, discover, establish, estimate, feel, figure, find, gather, grant, guarantee, guess, hold, hypothesize, imagine, intuit, judge, know, note, posit, presume, presuppose, proclaim, prove, reckon, recognize, recollect, remember, report, repute, reveal, rule, rumor, say, show, specify, state, stipulate, suppose, surmise, take, think, understand, verily, wager

In order to determine the frequency of passive verbs with infinitival complements the BNC was queried for occurrences of the past participle form of the verb either immediately followed by to be (i.e. Ved to be, e.g. [8]) or separated from to be by a by phrase with a one-, two-, three-, or four-word NP (i.e. Ved by to be, Ved by--to be, Ved by--to be, and Ved by--to be, e.g. [9]). To determine the frequency of active verbs with infinitival complements the corpus was queried for the base form, the -s form, the -ing form, and the preterite and/or past participle form of the verb followed by either a one-, two-, three-, or four-word NP and to be (i.e. V--to be, V--to be, V--to be, and V--to be, e.g. [10]) or by to be only (i.e. V to be, to include occurrences in relative clauses like [11], and other cases in which the subject of the infinitive is fronted, like [12]).

(8) At a time when scientific advance was seen as universally beneficial, the nuclear industry was judged to be at the cutting edge of technological endeavour (AN9 631).

(9) Part or all of the aedeagus is held by some authors to be of a secondary, non-appendicular nature but others consider it to be formed by the division of the gonapophyses of the 9th abdominal segment, the two median halves fusing during development to form the penis while the lateral halves constitute the parameres (EVW 754).

(10) Although puzzled, he hesitated to investigate more closely for fear his driver should imagine his interest to be prurient (B20 1699).

(11) Given (or old) information is that knowledge which the speaker assumes to be in the consciousness of the addressee at the time of the utterance (FRL 1193).

(12) This I subsequently discovered to be false (B78 2194).

The results of these queries are displayed in Tables 1 to 3. Table 1 lists the verbs alphabetically; Table 2 lists only those verbs for which the queries produced over a hundred occurrences, starting with the verb that has the highest proportion of passives to actives, and ending with the verb that has the highest proportion of actives to passives; Table 3 lists the remaining verbs according to the absolute number of attested passives in descending order.

The following observations can be made. A first one, which is of minor importance, is that a fair number of the verbs mentioned by Postal--who repeatedly refers to "[his] own speech" (e.g. Postal 1974: 305, 309, 312)--were not attested at all in the BNC with infinitival copular complements. This may point to differences between American and British English and/or to certain peculiarities of Postal's idiolect.

Much more relevant for the issue I want to address is the observation that the above-mentioned intuition that matrix verbs of infinitival copular complements are more often passive than active is basically correct: 74% of all the attested verb complement combinations have passive matrices. This is something that needs to be explained. On the other hand, 74% is also the median value, which means that for one in two of the verbs investigated the percentage value of passives is lower. Active matrix verbs should therefore not be cast aside as rare occurrences. Restricting ourselves to the 23 verbs for which the total number of attested infinitival complements exceeds one hundred (Table 2), there are even five verbs that more often have an active than a passive matrix. If there is a general principle that accounts for the high occurrence of passives, these deviations from this principle will need to be explained as well.

3. Explaining the predominance of passive matrix verbs

Mair (1990: 178) has already dismissed the explanation offered by Bolinger (1977: 129) that "[t]hese verbs express opinions and viewpoints concerning which the speaker, when he wants to sound impressive, would rather shift responsibility to some unnamed--and hence remote and powerful agent," which passive matrix verbs allow them to do, saying that "to trace back the emergence and spread of a grammatical construction solely to the speaker's psychological stance does seem far-fetched." Without question the popularity of many of the (BE) Ved to be patterns, particularly in genres like news texts and scientific discourse, can be related to the fact that these patterns have become grammaticalized as lexicogrammatical paradigmatic options available in systems of "evidentiality" (cf. Chafe and Nichols 1986; Thompson and Mulac 1991; Traugott 1995; Verhagen 1996), but how did the option become available in the first place? (4)

Mair himself suggests that "the higher frequency of passive matrix verbs is mainly due to the fact that, unlike their active analogues, such passive constructions are a useful means of redistributing sentence information" (1990: 180); that is, passive matrices allow one to move the subject of the complement clause to sentence-initial position, the unmarked position for information that is contextually "given" (as argued in Bolinger 1952; Chafe 1976, 1987; Firbas 1967, 1992; Givon 1983; Halliday 1967, 1985: chapter 3; Kuno 1972; Prince 1981), whereas

[i]f the matrix verb is in the active voice, the order of constituents is exactly the same, regardless of whether an infinitival or a finite complement clause is used. In most cases, therefore, the speaker or writer will choose the stylistically and structurally less marked that-clause (Mair 1990:181).

Compare the corpus example (13a) with the constructed alternatives (13b) and (13c). In (13a), with a passive matrix plus infinitive, the pronoun referring to Ian Spiro is in sentence-initial position; in both (13b) and (13c), with a finite complement and an active matrix plus infinitive respectively, it follows the matrix verb.

(13) a. Briton hunted after family is massacred By Shenai Raif

BRITON Ian Spiro was yesterday named as a suspect by Californian police hunting the killer of his wife and three children. He is thought to be armed with a handgun. The 46-year-old businessman has been missing since the bodies of his family were found at their luxury home in San Diego on Thursday. At first, police were not sure if he was also a victim or on the run from hitmen. But after a detailed examination of the house they said he was suspected of the shootings (K97 865).

b. Police think (that) he is armed with a handgun.

c. Police think him to be armed with a handgun.

So infinitives with passive matrices provide a textual alternative to finite complements, which infinitives with active matrices do not, and the latter have the additional disadvantage, says Mair, that they are more "marked" than their finite counterparts. The alternative explanation for the predominance of passive matrices that I would like to defend in this paper offers a corrective to Mair's account. I would like to argue that there is no need for this questionable deus ex machina of stylistic and structural markedness, whose impact it might be difficult to prove, and which, if markedness is defined on the basis of frequency, introduces the risk of making the explanation circular. (5)

Something Mair did not take into account is the typical information value of the subjects of infinitives with active matrices, which indeed he could not do because the evidence he presents is based on examples with say, which does not take an infinitival copular complement when it is active. Yes, subjects of passive matrix verbs with infinitives typically have "given" referents, whereas the subjects of that clauses more often introduce "new" referents, but what about the subjects of infinitives with active matrices? Using evidence based on examples with believe and prove, I have established elsewhere (Noel 1997, 1998) that they, too, are typically "given": in 90% of all cases with believe, and in 94% of all cases with prove. Subjects of that clauses after these verbs, on the other hand, introduce "new" referents in 36% and 30% of all cases respectively. Evidence based on examples with think further confirms this. (6) Of all occurrences of the THINK NP to be pattern (48 cases out of 67), 72% occur in contexts like (14)-(16), in which the NP takes up a referent from a preceding clause in the same sentence or from a previous sentence.

(14) This distinction can be put briefly by saying that whereas an appellate court has power to decide whether the decision under appeal was "right or wrong," a court exercising supervisory powers may only decide whether THE DECISION under review was "legal" or not. If THE DECISION is illegal it can be quashed; otherwise the court cannot (with one exception) intervene, even if it thinks THE DECISION to be wrong in some respect (EBM 94).

(15) Everywhere Joan went, RICHARD would appear, tipped off by mutual friends in Virgin, among whom he acquired the nickname Tag--after his constant request, "What are you doing tonight? Seeing Joan? Mind if I tag along?" Joan's friends did not think BRANSON to be a particularly eligible figure (FNX 344).

(16) Although the instruction books that come with new machines recommend certain ways of setting such machines up, I have always liked to experiment across the range of possible settings. On this field I found the best combination was to set the Silver Sabre at MINIMUM DISCRIMINATION AND MAXIMUM SENSITIVITY. (After quite a few months of use, I still think THESE to be the best settings for ploughed land.) (G30 742).

In each of these examples the constituents in caps are coreferential. Sometimes (in 11 cases out of 67 = 16%) the subject of the infinitive summarizes a whole state of affairs expressed in the previous discourse, as in (17).

(17) Many women of a previous generation, in which sex and reproduction were seen as inevitably linked, believed that sex stopped--or at all events should stop--with the "change of life." Some still think this to be the case (EW8 698).

There is only one case in which the subject of the infinitive introduces "new" information: (7)

(18) I baited a 6's hook to 61b b.s. line paternostered on a 1[OMEGA] oz bomb with half a lobworm and sent it out to the marker. The other rod I baited with sweetcorn. Almost immediately I experienced a series of knocks and pulls to the worm rod but nothing concrete to strike at. I was getting rather frustrated abortively striking at these plucks and finding my worm chewed up. [I] thought bream to be the culprits, but couldn't be sure because so far I had not seen any prime out in the swim to indicate they were, in fact present there.

The subjects of that clauses, on the other hand, take up a referent from the previous discourse in only 46 of the 100 examples. In ten cases they refer to a state of affairs expressed in the previous discourse. In 23 cases, however, they either introduce new referents or reintroduce one that has remained unmentioned for a couple of sentences. (8) Here are a few examples of the latter kind:

(19) The dream of making this world into a global market can only come about by perpetuating injustice. As the Uruguayan author Eduardo Galeano wrote: "Incapable of combating poverty, the dominant culture combats the poor and blesses the violence of power." That's why some people think that communication is needed--to convince us that it is better to "have" and not to "be," that happiness is to possess, to compete and to succeed no matter how. But that is domination, not communication (EBJ 30).

(20) Drinking every day is least common in the north, where men are more likely to drink a moderate 2 to 3 times a week. Job satisfaction is least important for Southern men and most important in the north, while Midlanders are not worried about having a career. Midlanders value owning their own home more than Southerners, who think that freedom is important (ECT 1274).

(21) Despite her long-standing affair and rumoured pending marriage to Commander Tim Laurence, Anne has managed to escape the sort of publicity that has dogged other female members of The Firm. Sexist cynics might suggest this is because she doesn't have Di's looks or Fergie's telephotogenic qualities. I prefer to think that Anne's secret is that if you act like royalty, the chance is you'll be treated like royalty (HAE 2935).

Communication in (19), freedom in (20), and Anne's secret in (21) are all mentioned for the first time in their respective paragraphs.

Table 4 summarizes the data for the three verbs, believe, prove, and think, when used in the active voice. In other words, also with active matrices givenness comes close to being a necessary condition for the subjects of infinitival complements, which is completely in line with the linguistic universal proposed by Givon (1990: 875-876; 1995: 36) with respect to the relation between finiteness and thematic coherence or continuity: "Clauses that involve higher referential continuity tend to receive less-finite marking." More important for the argument at hand, however, is that the typical information value of the subjects of infinitival complements leads to greater referential continuity (e.g. Givon 1993: 286-287) or more anaphoric grounding (e. g. Givon 1995: 350) of sentences with active matrices than is necessary for efficient communication, since two given elements open the sentences in that case, instead of just one with passive matrices. Compare the following corpus examples of passive matrices with their constructed active counterparts:

(22) a. Police launched a nationwide hunt for Ahrjinder Singh Dosanjh--known as Bobby--after he was viciously beaten and bundled into a van. Hours later a man believed to be East End boss Edward Clancy rang detectives accusing the Asian of taking his 38,000 [pounds sterling] Silver Shadow. Burly Clancy--the 6ft, 19st owner of a crane hire business--is thought to be one of five men involved in the ambush at Leamington Spa magistrates court, Warwicks. The five heavies pounced on Dosanjh, 21, as he walked through the court lobby after being bailed on an unrelated theft charge. He was dragged kicking and screaming to a van parked nearby. Witnesses could see the Asian getting a "severe beating" as the van raced off (CH2 6078).

b. They think Burly Clancy--the 6f, 19st owner of a crane hire business--to be one of five men involved in the ambush at Leamington Spa magistrates court, Warwicks.

(23) a. Pub chain trains for Europe Taylor Walker is benefiting from a customer care programme that it believes will enable it to compete with its European counterparts. Taylor Walker managing director David Longbottom said: "Unlike our European counterparts, in the UK we do not perceive bar work as having professional status. Bar work is thought to be transitory, undemanding and unskilled. We are trying to correct this with training programmes and staff recognition schemes" (A7F 695).

b. In the UK we think bar work to be transitory, undemanding and unskilled.

The (b) sentences, with active matrices, are more redundant referentially than their counterparts with passive matrices in (a), because both the active matrix clause subjects and the subjects of the infinitives take up referents from the previous sentences. Often such referential redundancy would put a nontopical or nonthematic given referent before a topical/thematic one, upsetting the "thematic progression" (cf. Danes 1974) of the text:

(24) a. Many bio-acoustic experts agree that the echolocation clicks are created by implosive movements of air in the nasal passages, but the exact process of sound production and projection is unknown--possibly, the echolocation waves pass through the melon. The manner in which the dolphin receives THE RETURNING ECHOES is also a mystery, but THEY are thought to be picked up by all parts of the body, to travel through the bones to THE HEAD. A significant proportion of THE DOLPHIN'S brain is thought to be used in processing the information produced by the echolocation system (ABC 331).

b. The manner in which the dolphin receives the returning echoes is also a mystery, but experts think them to be picked up by all parts of the body, to travel through the bones to the head. They think a significant proportion of the dolphin's brain to be used in processing the information produced by the echolocation system.

(25) a. We undertook a packed programme. Seven churches and a museum on one day devoted to Byzantine mosaics in Ravenna; Rimini; a unique medieval library in Cesena; and the perfect hilltop town of URBINO, full of fascinating treasures. In the 15th century THE TOWN and its palace were thought to be the most beautiful in all Italy. Even in incessant rain it still seemed lovely and, at least within its walls, remarkably unspoilt, though behind the medieval facades many houses have been relentlessly modernised. Walking around Urbino's quiet streets on a wet Sunday out of season was an evocative and fascinating experience, even though the churches and palace had closed early in the morning (AHK 1683).

b. In the 15th century people thought the town and its palace to be the most beautiful in all Italy.

If the extract in (24) had been about bio-acoustic experts rather than dolphins, the (b) sentences with active matrices, in which these experts are "topic" or "theme," would have fitted in quite naturally. The passive matrices in the original (a) version, however, help to ensure that most clauses receive a dolphin-related topic/theme and to avoid that the experts acquire theme status on a higher level (paragraph or text). In (25) the passive establishes the town of Urbino as the theme of the rest of the paragraph and prevents reference to nontopical/nonthematic people. In other words, the active matrix clause subject, whose sentence-initial position makes it topic/theme by default (see, e.g., Givon 1993: 47; Halliday 1985: 38), often conflicts with the given subject of the infinitive, itself a candidate to be topic/theme of the sentence. Passive matrices remove this conflict. An examination of a hundred occurrences of the BE thought to be pattern revealed that in 25 cases the suppressed "agent" (or "thinker") is a previously mentioned participant with a low topicality value (as in [24]; another example is [26], in which the "thinkers" can be assumed to be the mentioned relatives); in 75 cases the unexpressed "thinker" does not have specific reference but can be said to refer to people in general (as in [25]) or, especially in scientific or media texts, people who know (as in [27]).

(26) Anxious relatives raised the alarm when the men, who all live near Borth, were nearly five hours overdue. One man's father told coastguards that normally the group only travelled a few miles along the coast on the speedboat trips. Aberystwyth lifeboat secretary David Jenkins said one of the missing men was wearing a wetsuit but the other two were thought to be wearing shirts (HJ4 7031).


The idea that villages might have been planned, or at least regulated in their growth, is a relatively new one. Towns, which have been more intensively studied and for which there is generally more documentary and cartographic information, were thought to be organic or haphazard creations until recently. Exceptions were known, of course, like Salisbury, Winchelsea and Ludlow, where deliberately created towns were well documented and where the regular gridiron street pattern suggested town planning (H8U 873).

Sometimes, however, thematic progression makes it necessary that the active subject is mentioned, ensuring the continued existence of the active pattern:

(28) a. He suspected that, inside herself, she was totally bewildered and perhaps despaired of ever finding Resenence Jeopardy. Over the years, Dauntless had become used to loneliness, but he suspected that for Cleo it was a new condition which she was having trouble getting used to. He knew that she thought him to be a stiff and unimaginative person (GW2 3246).

b. ... He knew that he was thought to be a stiff and unimaginative person.

(29) a. At the sight of Amaranth Wilikins there was a deal of coughing, a shuffling of feet and a perceptible rise in the level of attention. Ron Barton, who had had a word with his editor at his regular table at the Savoy grill, had been told he could pay five thousand for an exclusive if he thought her to be worthwhile (HNK 1744).

b. ... if she was thought to be worthwhile.


As has been said, a decision was made to abandon high-order averaging by the instrument so as to obtain a full record of the variations in measurement. The author did not understand these and thought them perhaps to be due to changes in reflectance (HGX 1834).

b. ... The author did not understand these and they were thought perhaps to be due to changes in reflectance

Note that, upon reading the (b) sentences, one does not automatically connect the unexpressed "thinker" with the participant referred to by the active subject in the (a) sentences. Rather, one's first interpretation is likely to be that people in general are implicitly referred to, which may point to the grammaticalization referred to above. I would argue, however, that this grammaticalization is as much the result as the cause of the predominance of the passive pattern and that the process was brought about by the fact that the active pattern is usually a dispreferred pattern for information-structural reasons.

The fact that some verbs, like imagine, declare, and prove, more often occur in the active than in the passive (see Table 2) may be due to their lexical-semantic specificity, which might favor the mentioning of two participants, instead of just the one.

4. Conclusions

In this article I have presented empirical evidence showing that infinitival copular complements typically occur much more frequently with passive matrix clauses than with active matrices. An earlier attempt at explaining this predominance of passive matrices attributed it to the fact that infinitives with passive matrices order their participants differently from finite complements and that infinitives with active matrices do not offer this advantage, so that when the word order shared by the that clause and the infinitive with active matrix is required for reasons of information structure the former is the preferred, purportedly unmarked, choice. The supposed markedness of nonfinite structures has been questioned, however, and an alternative explanation might therefore be in order. I have established that not just infinitives with passive matrices but also infinitives with active matrices do information-structurally different things from that clauses. The subjects of that clauses often introduce "new" referents, whereas the subjects of infinitival complements typically take up previously mentioned referents, irrespective of whether their matrices are active or passive. I would posit, however, that the typical information structure of an active matrix plus infinitive normally makes it a dispreferred construction, because of the redundancy resulting from the "givenness" of both the matrix clause subject and the subject of the infinitive. In addition, the subject of the infinitive often competes with the matrix clause subject for "topic" or "theme" status.

University of Gent

Received 29 June 1998

Revised version received

21 September 1998


Bolinger, Dwight L. (1952). Linear modification. Publications of the Modern Language Association of America 67, 1117-1144.

--(1977). Meaning and Form. London: Longman.

Borkin, Ann (1974). Raising to Object Position. University of Michigan Papers in Linguistics 2, 1. Ann Arbor: University of Michigan.

--(1984). Problems in Form and Function. Norwood, NJ: Ablex.

Burnard, Lou (1995). Users Reference Guide for the British National Corpus. Oxford: Oxford University Computing Services.

Chafe, Wallace, L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and points of view. In Subject and Topic, Charles N. Li (ed.), 26 56. New York: Academic Press.

--(1987). Cognitive constraints on information flow. In Coherence and Grounding in Discourse, Russell S. Tomlin (ed.), 21-55. Amsterdam: Benjamins.

--; and Nichols, Johanna (eds.) (1986). Evidentiality: The Linguistic Coding of Epistemology. Norwood, NJ: Ablex.

Chomsky, Noam (1973). Conditions on transformations. In A Festschrift for Morris Halle, Stephen R. Anderson et al. (eds.). New York: Holt, Rinehart and Winston.

Danes, Frantisek (1974). Functional sentence perspective and the organization of the text. In Papers on Functional Sentence Perspective, Frantiasek Danes (ed.), 106-128. The Hague: Mouton.

Firbas, Jan (1967). On the interplay of means of functional sentence perspective. Proceedings of the International Congress of Linguists 10(2), 740-745.

--(1992). Functional Sentence Perspective in Written and Spoken Communication. Cambridge: Cambridge University Press.

Givon, Talmy (1983). Introduction. In Topic Continuity in Discourse: A Quantitative Cross-Language Study, Talmy Givon (ed.), 1-41. Amsterdam: Benjamins.

--(1990). Syntax: A Functional-Typological Introduction, vol. 2. Amsterdam: Benjamins.

--(1993). English Grammar: A Function-Based Introduction, vol. 2. Amsterdam: Benjamins. (1995). Functionalism and Grammar. Amsterdam: Benjamins.

Halliday, M. A. K. (1967). Notes on transitivity and theme in English, part 2. Journal of Linguistics 3, 199-244.

--(1985). An Introduction to Functional Grammar. London: Arnold.

Hudson, Richard A. (1971). English Complex Sentences: An Introduction to Systemic Grammar. Amsterdam: North-Holland.

Kuno, Susumo (1972). Functional sentence perspective: a case study from English and Japanese. Linguistic Inquiry 3(3), 269-320.

Lees, Robert B. (1960). The Grammar of English Nominalizations. The Hague: Mouton.

Mair, Christian (1990). Infinitival Complement Clauses in English: A Study of Syntax in Discourse. Cambridge: Cambridge University Press.

Maxwell, Michael B. (1984). The subject and infinitival complementation in English. Unpublished doctoral dissertation, University of Washington.

Noel, Dirk (1997). The choice between infinitives and that-clauses after believe. English Language and Linguistics 1(2), 271-284.

--(1988). The proof of the pudding: is prove to be/that like believe to be/that? In English as a Human Language, Johan Van der Auwera et al. (eds.). Munich: Lincom Europa.

Postal, Paul M. (1974). On Raising. One Rule of English Grammar and Its Theoretical Implications. Cambridge, MA: MIT Press.

Prince, Ellen F. (1981). Toward a taxonomy of given-new information. In Radical Pragmatics, Peter Cole (ed.), 223-255. New York: Academic Press.

Quirk, Randolph (1965). Descriptive statement and serial relationship. Language 41, 205-217.

Riddle, Elisabeth (1975). Some pragmatic conditions on complementizer choice. In Papers from the Eleventh Regional Meeting, Robin E. Grossman et al. (eds.), 467-474. Chicago: Chicago Linguistic Society.

Rosenbaum, Peter S. (1967). The grammar of English predicate complement constructions. Unpublished doctoral dissertation, MIT.

Steever, Sanford B. (1977). Raising, meaning, and conversational implicature. In Papers from the Thirteenth Regional Meeting, W. A Beach et al. (eds.), 590-602. Chicago: Chicago Linguistic Society.

Thompson, Sandra A.; and Mulac, Anthony (1991). A quantitative perspective on the grammaticization of epistemic parentheticals in English. In Approaches to Grammaticalization, vol. 2, Elizabeth Closs Traugott and Bernd Heine (eds.), 313-329. Amsterdam: Benjamins.

Traugott, Elizabeth Closs (1995). Subjectification in grammaticalisation. In Subjectivity and Subjectivisation, Dieter Stein and Susan Wright (eds.), 31-54. Cambridge: Cambridge University Press.

--; and Heine, Bernd (1991). Introduction. In Approaches to Grammaticalization, vol. 1, Elizabeth Closs Traugott and Bernd Heine (eds.), 1-14. Amsterdam: Benjamins.

Verhagen, Arie (1996). Tekstsegmentatie, onderschikking en subjectiviteit. Gramma/TTT 5(3), 249-268.


(1.) The research reported on in this paper was made possible by the Research Fund of the University of Gent (Bijzonder Universitair Onderzoeksfonds contract no. 12052095). Correspondence address: Department of English, University of Gent, Rozier 44, B-9000 Gent, Belgium. E-mail:

(2.) All examples are taken from the first (1995) version of the British National Corpus, a licensed copy of which is available in the Department of English of the University of Gent. The corpus was queried using version 0.927 of the SARA client software. In the source code that follows each example the three positions before the space identify the excerpt it was extracted from (which can be looked up in Burnard 1995); the number following the space is the line number of the example within the excerpt.

(3.) This is not a complete list of all verbs that display the alternation. For additions to the list, see Mair (1990: 175-176), who also refers to a list by Hudson (1971: 369-372).

(4.) It falls outside the scope of this paper to argue that the term "grammaticalization" is correctly applied here. Grammaticalization processes do not themselves explain high frequency, however, since frequency is a necessary condition for such processes to occur in the first place (see Traugott and Heine 1991: 9).

(5.) The conventional approach that identifies nonfinite structures as marked is questioned by Givon (1995: 34).

(6.) 67 occurrences of THINK NP to be were compared with 100 instances of THINK that NP is/are. Both patterns were looked for with one-, two--, and three-word complement-clause subjects. The 100 instances of the THINK that pattern were a random reduction of 1455 occurrences in which the original proportion of one--, two--, and three-word subjects was maintained.

(7.) In the remaining seven cases the subjects are nonreferential, as in They would support the union if they thought it to be in their interest to do so (AC2 2296) and In the British Social Attitudes Survey of 1988, 90 per cent of people thought there to be less respect for teachers amongst parents and pupils than 10 years ago (AN5 1223).

(8.) Since the subjects of that clauses also often take up "given" referents, information structure cannot be the only factor at work in the choice between the finite and nonfinite patterns. For a discussion of another possible contributing factor, see Noel (1997).

(9.) See note 4.
Table 1. Proportion of passive to active matrix verbs
(alphabetical listing)

Verb n passives % passives n actives % actives

acknowledge 54 79.41 14 20.59
admit 19 90.48 2 9.52
affirm 0 -- 0 --
allege 101 97.12 3 2.88
ascertain 1 100.00 0 --
assume 403 73.41 146 26.59
believe 916 53.32 802 46.68
certify 1 33.33 2 66.67
concede 4 100.00 0 --
consider 1086 54.55 905 45.45
declare 57 26.03 162 73.97
decree 5 100.00 0 --
deduce 1 33.33 2 66.67
deem 409 88.91 51 11.09
demonstrate 20 76.92 6 23.08
determine 5 62.50 3 37.50
discern 0 -- 1 100.00
disclose 1 100.00 0 --
discover 37 59.68 25 40.32
establish 1 20.00 4 80.00
estimate 212 87.97 29 12.03
feel 235 48.25 252 51.75
figure 0 -- 1 100.00
find 1033 69.52 453 30.48
gather 0 0
grant 0 -- 1 100.00
guarantee 31 96.88 1 3.13
guess 0 -- 17 100.00
hold 417 83.90 80 16.10
hypothesize 1 100.00 0 --
imagine 9 8.74 94 91.26
intuit 0 0
judge 141 66.82 70 33.18
know 564 59.00 392 41.00
note 6 85.71 1 14.29
posit 0 -- 1 100.00
presume 98 86.73 15 13.27
presuppose 1 100.00 0 --
proclaim 4 12.90 27 87.10
prove 56 36.36 98 63.64
reckon 113 85.61 19 14.39
recognize 48 76.19 15 23.81
recollect 0 0
remember 0 -- 4 100.00
report 391 97.26 11 2.74
repute 98 100.00 0 --
reveal 20 23.81 64 76.19
rule 2 40.00 3 60.00
rumour 85 100.00 0 --
say 1945 100.00 0 --
show 584 66.67 292 33.33
specify 4 80.00 1 20.00
state 37 74.00 13 26.00
stipulate 0 0
suppose 1935 96.27 75 3.73
surmise 0 -- 2 100.00
take 237 42.93 315 57.07
think 1382 92.69 109 7.31
understand 161 75.23 53 24.77
verify 0 0
wager 0 0
Total 12971 73.67 4636 26.33
Average 61.90 38.10
Median 73.70 26.30

Table 2. Proportion of passive to active matrix verbs for the verbs
with more than 100 occurrences (listed in descending order for the
passives, ascending order for the actives)

Verb % passives % actives

say 100.00
report 97.26 2.74
allege 97.12 2.88
suppose 96.27 3.73
think 92.69 7.31
deem 88.91 11.09
estimate 87.97 12.03
presume 86.73 13.27
reckon 85.61 14.39
hold 83.90 16.10
understand 75.23 24.77
assume 73.41 26.59
find 69.52 30.48
judge 66.82 33.18
show 66.67 33.33
know 59.00 41.00
consider 54.55 45.45
believe 53.32 46.68
feel 48.25 51.75
take 42.93 57.07
prove 36.36 63.64
declare 26.03 73.97
imagine 8.74 91.26
Total 73.83 26.17
Average 69.45 30.55
Median 73.41 26.59

Table 3. Number of passive and active matrix verbs for those verbs
of which there are less than 100 attested occurrences, listed in
descending order as per absolute number of passives

Verb n passives n actives

repute 98 0
rumour 85 0
acknowledge 54 14
recognize 48 15
discover 37 25
state 37 13
guarantee 31 1
reveal 20 64
demonstrate 20 6
admit 19 2
note 6 1
determine 5 3
decree 5 0
proclaim 4 27
specify 4 1
concede 4 0
rule 2 3
establish 1 4
certify 1 2
deduce 1 2
ascertain 1 0
disclose 1 0
hypothesize 1 0
presuppose 1 0
guess 0 17
remember 0 4
surmise 0 2
discern 0 1
figure 0 1
grant 0 1
posit 0 1
affirm 0 0
gather 0 0
intuit 0 0
recollect 0 0
stipulate 0 0
verify 0 0
wager 0 0

Table 4. Percentages of infinitives with "given" and that clauses
with "new" subjects

 Given infinitival New that clause
Verb subjects (%) subjects (%)

believe 90 36
prove 94 30
think 88 23

Average 91 30
COPYRIGHT 1998 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Noel, Dirk
Publication:Linguistics: an interdisciplinary journal of the language sciences
Article Type:Report
Geographic Code:4EUUK
Date:Dec 1, 1998
Previous Article:Geert E. Booij, Christian Lehmann, Joachim Mugdan, and Stavros Skopeteas, editors: Morphologie/Morphology. Ein internationales Handbuch zur Flexion...
Next Article:Suppletion, natural morphology, and diagrammaticity *.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters