Introduction

1Some reference grammar textbooks, including textbooks intended for students and language learners, rule out the use of object + TO-infinitive clause as the complementation of a perception verb in the active ([1]), all the while strongly recommending TO-infinitivals – instead of the use of bare infinitive clauses – in the passive.

(1) But his expression sapped her courage, she saw him to be defensive and it was hard to find speech. (STRATHY)

2TO-infinitive constructions are thus either excluded – in both prescriptive and descriptive grammars, as well as on grammar forums and websites – or restricted to some perception verbs only. In the latter case, it is often said that the sentence does not convey the meaning of sensory perception, but that of mental judgement or inference. We are thus faced with a twofold norm: a syntactic norm, as to the complementation patterns that are judged to be correct, and a semantic norm, as to the proper interpretation to give to the sentences when they are considered to be grammatically correct.

3This study begins by reviewing the usages prescribed or described in both contemporary and older grammars, and shows that the so-called descriptive grammars can also be prescriptive in some ways. It focuses on utterances from three traditional varieties of English – British, American, and Canadian – as well as from Internet English, which were collected from the BNC, COCA, STRATHY and GloWbE corpora. The aim is to identify which perception verbs take this type of complement, while determining whether the sequence “perception verb + object + TO-infinitival” has a wider range of use in one of these varieties in particular. An additional goal is to see whether discourse types or modes – written or spoken – as well as registers – formal or informal – are relevant to the usage of TO-infinitivals in the complementation of perception verbs. The occurrences from the webpages in the GloWbE are also used to accomplish this goal, as well as to compare the three traditional varieties with a more recent one, Internet English.

4The main objective of this investigation is to provide an account for the fact that, contrary to what some grammars postulate, TO-infinitive clauses can be – and are – used in the complementation of perception verbs in the active. In order to do so, the study puts forward a semantic explanation as to the (in-)compatibility of perception verbs with TO-infinitivals, partly based on the types of these verbs. It thus seeks to answer the underlying question of whether the norms or usages mentioned in prescriptive or descriptive grammars influence the way speakers use such constructions, while confronting these norms and usages to evidence provided by attested examples.

1. TO-clause constructions as complements of perception verbs: Grammatical norms or usages prescribed or described

(2) She hung her head, feeling tears prickle in her eyes. (BNC)

(3) At the Tiki Theater Village, simulating a primitive lifestyle, I could smell meat roasting over open fires. (COCA)

6Similar conclusions are reached in some descriptive grammars, which, in addition, try to account for the difference of meaning between the two standard complements licensed by perception verbs (Adamczewski and Delmas 1998; Larreya and Rivière 2010; Souesme 1992).

7The aforementioned grammars and websites do not exclude TO-infinitivals; they simply do not mention them as possible complements for perception verbs. However, a few grammars explicitly rule out the use of a TO-infinitive clause after perception verbs. Among them are reference grammars (Aarts, Chalker, and Weiner 2014; McArthur 2003) as well as grammar forums and websites (BBC World Service; WordReference.): “Both see and hear, and also notice and watch and other verbs of perception, can be followed by object + bare infinitive (i.e. without to) or by object + verb-ing.” (BBC World Service)

8It appears that the date of publication of the grammar is of no relevance to either position – excluding TO-infinitive constructions as complements of perception verbs or disregarding them as possible complements. Both positions can be found in both contemporary and older grammars, as shown by the references cited above. It should also be noted that almost all of them indicate that TO-infinitive clauses are obligatory in the passive, as in (4); “the verb is used with a (pro)noun and a bare infinitive. Note that a to-infinitive is needed in the passive.” (Hornby 25)

(4) An Aeroflot jet landed in Shannon, Ireland, and a Russian was seen to deplane, hurry into the duty free shop, and whisper to the clerk, “I need protection”. (COCA)

9Nevertheless, TO-infinitive constructions are also judged to be correct complements of perception verbs by other descriptive grammars, whether they be older (Poutsma 1928–1929; Curme 1931; Jespersen 1940) or more recent publications (Berland-Delépine 1974; Quirk et al. 1985; Huddleston and Pullum 2002; Dufaye and Khalifa 2006). They all postulate that TO-infinitive clauses can be licensed by perception verbs only under certain conditions:

10- TO-infinitivals are in most cases restricted to some perception verbs – only feel is concerned according to Quirk and Greenbaum (364) and Quirk et al. (1203). Other authors extend the list to see and perceive (Jespersen 1940; Zandvoort 1962), and observe (Poutsma 1928–1929; Curme 1931). Huddleston and Pullum’s descriptive grammar, one of the most recent reference grammars, provides a longer list of verbs (1237) – feel, hear, notice, observe, overhear and see – while at the same time adding restrictions on the range of use of TO-infinitivals:

• 2 In Huddleston and Pullum’s grammar (1236–37), [ii] refers to -ING constructions, as in We saw Kim l (...)

With hear and overhear [iv] is virtually excluded in the active (*We’d heard him to be an impostor) […]. Watch wholly excludes the to-infinitival, whether active or passive. […] notice is also marginal in [iv]. Smell is generally restricted to [ii], and hence is listed in Class 3 Cii below; it combines predominantly with burn (I can smell something burning).2

11- Jespersen (240–241) adds that the TO-infinitive construction is rare with perception verbs and that TO is often used with be, which is also underlined by Curme: “after […] feel, see, observe, find, we sometimes employ also the infinitive with to, indeed regularly […] if the infinitive is the copula be” (Curme 124).

12- As far as register is concerned, some linguists emphasise the idea that TO-infinitivals are used only in formal language (Berland-Delépine 193) or literary English (Zandvoort 1962).

13- Semantically, the combination of a perception verb with a TO-infinitival “does not indicate immediate perception but an inference” (Jespersen 280). This idea is taken up by Huddleston and Pullum (1236), who also note that with verbs of sensory perception, the construction under study “does not represent the primary sense: it is not a matter of sensory perception but of mental inference”.

An examination of the actual meaning expressed by the to infinitive after verbs of perception in the active voice shows therefore a necessary before/after relationship between the event of the verb of perception (which has shifted to evoke the notion of inference) and the event of the infinitive (which denotes the conclusion reached by means of the inference). This sequential representation, on the other hand, does not characterize the verbs of perception in their basic perceptual sense, where the relation is rather one of coincidence in time between the perceiving and the phenomenon being perceived.

15Interestingly enough, none of the aforementioned grammars, be they prescriptive or descriptive, considers the TO-infinitive construction as grammatical with watch, listen and look, namely agentive perception verbs; they are either excluded from this construction – this is the case for watch (Dixon 1991; Duffley 1992; Huddleston and Pullum 2002) – or never mentioned at all.

16A first conclusion that can be drawn from those inter-grammar comparisons is that the so-called descriptive grammars can also be prescriptive in some ways. They state, for instance, that the TO-infinitive structure is excluded or marginal with a particular verb (see, for example, how Huddleston and Pullum (2002) limit the use of TO-clauses with notice and smell), or they present incomplete descriptions of acceptable constructions for a specific verb (Quirk et al. (1985) disregard TO-infinitivals as an alternative construction for see in the active). It also appears that grammar websites, which can be considered as ‘non-academic’ grammars, are more prescriptive and far less complete than traditional or reference grammars.

17Whether attested utterances conform to the norms as well as to the conditions and restrictions listed above remains to be verified. This issue will be addressed in sections three and four of the analysis, while the following section is devoted to describing the process of collecting data.

2. Data collection procedure

19The research focused only on certain perception verbs, which were selected according to the following criteria:

20- the criterion of experience, which is widely used by psychologists, linguists and lexicologists. It shows that the concept of perception is expressed by basic verbs, such as see, look, watch, hear, listen, smell, taste, and feel, which refer to the fundamental structure of human perception – sight, hearing, smell, taste, touch and proprioception;

21- the criterion of register, which allows for both selecting verbs that are considered to be representative of everyday language and excluding more specialised terms. The verbs thus selected constitute the fundamental core of the lexicon used by “the ordinary man” (Austin 1962), whereas the terms excluded are either technical or scientific (e.g. investigate, scan, palpate), formal or informal (e.g. contemplate, regard, stink), literary (e.g. behold, espy) or even archaic (e.g. harken);

22- the criterion of frequency, which, for instance, permits to exclude verbs of limited or less common use (e.g. discern, overhear and hark). It appears that see and look are more frequently used than the other visual verbs;

23- finally, the criterion of distribution, which allows for the exclusion of verbs licensing only nominal or prepositional object complements, such as gape, gawk, inspect and scrutinize.

24The verbs thus selected for the corpus search represent the five/six sensory modalities, whether they be non-agentive or agentive (see Table 1). It appears that they are all mentioned in the grammars cited in the preceding section. Only discern (listed in Poutsma 1928–1929) was excluded as it does not satisfy the selecting criteria.

25Semantically speaking, these verbs behave slightly differently. See, hear, notice and perceive refer to non-agentive perception, whereas look, watch and listen belong to the field of agentive perception; as for observe, it can be used to express both agentive and non-agentive visual perception. It should be noted that notice and perceive are not restricted to one particular perceptual modality, though they are often used for sight and hearing.

27The current study now sets out to accomplish two different goals concerning the comparison of the use and frequency of TO-infinitivals in the complementation of perception verbs: first, to determine if these TO-constructions are linked to a specific variety of English and then, to see if the discourse situation and register types influence the use of such complements.

3. TO-clause constructions: Usages linked to specific varieties, modes and registers

28The question that comes to the fore is whether the use and distribution of TO-complement constructions with perception verbs can be accounted for by the variety of English in which they are used, and/or by the register in which they can be found. This section primarily focuses on the frequency of TO-complement clauses in three traditional varieties of English: British, American and Canadian. It will end with an analysis of register types and modes of communication in these varieties.

29I have chosen to study British and American English primarily because the distribution of the variants they show is often explained using sociolinguistic criteria (see, for instance, Wierzbicka’s explanation (1988) with respect to the use of periphrastic verbal phrases such as have a V, take a V and give a V). In addition to comparing British and American English, I wanted to compare two varieties whose speakers share a similar culture. This was made possible via a comparison of American and Canadian English. Finally, the GloWbE is a useful tool giving a precise idea of the development and the characteristics of Internet English, which can be considered as a new variety of English.

32The BNC, the COCA and the STRATHY all contain spoken and written sections, whereas the GloWbE is composed of words from webpages, either Internet websites or blogs. I have chosen to work on these three main modes of communication – spoken and written English, and webpages – as they correspond to different types of registers: more formal for written English, especially in academic papers listed in the three corpora; less formal, and sometimes informal, for spoken English. As for webpages, both formal and informal registers are concerned as websites and blogs can be of varied quality.

33Tables 4–7 present a comparison of the modes of communication and show that in the three varieties of English under study, TO-complements of perception verbs are the most frequent in webpages (54.76% of the 2,717 total occurrences; see Table 4) and the most infrequent in spoken English (5.56%). A closer look at their distribution in each variety brings to our attention noteworthy differences (see Tables 5–7). Indeed, if this conclusion is also valid for British and Canadian English, it is not so for American English, where the majority of TO-complement clauses are found in written form (49.6%). It could be inferred that British native speakers, as well as Canadian ones to a certain extent, are more careful and respectful of norms and usages in traditional discourse situations – fiction, prose, academic papers, and so on – than they are in less conventional ones, such as Internet websites and blogs. On the contrary, as American English shows little difference between the written mode and Internet English, this might suggest that Americans are less sensitive to, or aware of, norms or so-called restrictions on language whatever the communication context. However, another plausible explanation could be that the constructions under study have already entered written discourse, which could be the sign of a current change. Finally, TO-clauses appear to be quite rare in spoken language, but far more frequent in less conventional media, such as Internet websites and blogs, which favour more informal speech as well. A more in-depth study would be helpful to precisely determine which percentage of Internet pages belong to formal and informal speech, as Internet websites can show either type of register.

34Another conclusion that can be drawn is that TO-complement clauses are more frequent in present-day language, which could be another sign of an ongoing change in the complementation of perception verbs. Another clue with respect to this current change appears when looking at non-standard structures for agentive verbs, such as listen to / look at / watch + NP/pronoun + TO V. Indeed, TO-infinitivals are usually excluded from the complementation of this type of verb. Though the occurrences found in the corpora are few – especially compared to the occurrences of bare infinitive clauses and -ING clauses (see Tables 8–9) –, it can be noted that they are favoured in informal language, namely spoken English, as well as in less conventional modes of communication, that is Internet websites (see Tables 4–7):

35- out of the 14 occurrences of listen to NP/pronoun TO V, 10 appear in American English and 9 of them are from websites;

36- watch NP/pronoun TO V occurs only 25 times, mainly in Internet English – this seems in conformity with the various mentions in grammars of the impossibility for watch to take a TO-infinitive construction;

37- look at NP/pronoun TO V was found most frequently in American and Internet English. This is by far the most controversial structure, and the variety and mode of communication in which it appears are quite revealing. Indeed, they seem to indicate that norms and standards are less significant in non-conventional modes of communication, regardless of the variety of English, as well as much disregarded in American English.

38Therefore, it can be argued that the Internet plays a leading role in the current changes in the complementation of perception verbs, and especially in the increasing occurrence of TO-infinitive constructions. Indeed, if we compare the results obtained with the two standard constructions, bare infinitives and -ING clauses, it appears that both of them are most frequently found in written and spoken English (see Tables 10–11). Only the verbs observe, listen and perceive have a majority of their -ING complements found in Internet English (51.5%, 52.96% and 55.70% respectively) while bare infinitives are the most frequent structures in webpages only with listen and notice (50.11% and 64.98% respectively). For instance, perceive, which shows the highest percentage with TO-constructions, also has the majority of -ING occurrences in Internet websites (55.70%) but not that of bare infinitives (47.06%). The same remark applies to observe, the percentages of whose occurrences are similar to those of perceive (51.5% with -ING constructions and 47.36% with bare infinitives). As for listen, the highest percentages of the occurrences of the three variants are all found on Internet websites: 64.29% of TO-constructions, 52.96% of -ING-clauses and 50.11% of bare infinitives. In those cases, it could be concluded that the use of a TO-construction derives from the speaker’s choice, which seems to support the hypothesis of the role played by the Internet in the evolution of the English language. It indeed looks as if the influence of the Internet is already perceived in the USA, as if Internet language had already begun to appear in conventional discourse types – especially written – and registers.

39Non-agentive perception verbs could also be subject to current changes. This issue will be addressed more particularly in the fourth and final section, which is devoted to the usages of TO-infinitivals in the complementation of perception verbs with respect to verb types, according to semantic considerations.

4. TO-clause constructions and semantic considerations: Usages linked to verb types

40The aim of this final section is to determine whether the search results meet the conditions and restrictions identified in prescriptive and descriptive grammars. To do so, a semantic viewpoint will be adopted to account for the seemingly paradoxical association of perception verbs with TO-clauses.

(5) When the likes of Costco, Lowes, Home Depot, and the list goes on and on start to merchandise their own SMU’s in this category […] you know they can smell money to be made and hence the race to the bottom of the barrel begins for not only the manufacturers doing business in the “middle” but the category as well and one would assume the end consumer ends up the winner, right? (GloWbE – Canada)

(6) One of the things I was struck by in listening to him to preach in particular, is that he is very empathic towards the pain and suffering in his congregation. (GloWbE – US)

(7) There are still negotiations going on between the Clinton folks and the Obama folks about the convention, still not resolved about a roll call vote, but if there is a roll call vote for Hillary Clinton, watch it to occur in the morning. (COCA)

42If these findings came as no surprise with smell and taste, which do not frequently take standard clausal constructions either, the same cannot be said of agentive watch, which is quite frequent with bare infinitive and -ING clauses (see Tables 8–9). One possible explanation may lie in semantics, rather than in syntax.

43Indeed, TO-infinitivals are more frequent with non-agentive perception verbs, perceive showing the highest proportion of these clauses (2.086%), immediately followed by feel (0.699%). Perceive can be conceived as a kind of hyperonym of all perception verbs, but it also expresses mental perception as it can be paraphrased as “grasp with the mind”. It consequently is in semantic keeping with the inferential reading that TO-infinitivals bring to the whole sentence. As for feel, though it can express both agentive and non-agentive perception, it always functions on a non-agentive mode when taking a TO-complement construction, denoting proprioception, impression, and even belief or another cognitive state. However, observe, with TO-complement clauses representing 0.067% of all its occurrences, does not seem to fit the bill. Indeed, observe has both an agentive and a non-agentive meaning, and in its former sense, it is semantically close to watch. But contrary to watch, observe implies a cognitive dimension: a mental process is always involved in observing, which is often carried out in order to learn something. It appears that it is this deductive process that matters most in sentences including observe, as in (8–9).

(8) When I asked you, ‘Did you ever observe them to argue or fight,’ did you tell me about the Christmas Eve argument? (COCA)

(9) Let me tell you, your reality is not reality. Only you choose and observe it to be your reality. (GloWbE – US)

44As is underlined in various grammars, a perception verb licensing a TO-complement clause undergoes a lexical shift and the whole sentence takes on an inferential meaning, which is quite obvious and frequent with non-agentive verbs. In such constructions, TO indicates that there is no coincidence between the perceiving process and the state of affairs described in the TO-clause. Indeed, the process denoted by the infinitive clause is not the perceived act itself; it can only be inferred from the perceptual data obtained thanks to the perceiving act. This state of affairs is therefore conceived as subsequent to the perceiving act:

A perceived event is not in the same relation to the act of perceiving it as an inferred event is to the act of inferring it. The former exists throughout the process of perception; the latter, however, has no existence before being conceived by the mind, and exists only as a conclusion which arises as a result of, after, the mental operation of inferring that the state of affairs it describes is true. (Duffley 33)

45This leads Duffley to conclude that in such constructions, there is a sense of afterness conveyed by TO: “this use of the infinitive involves a final interception of to” (Duffley 36). It can therefore be argued that TO discursively serves to point out the gap between perception and conclusion or inference (see also Cotte 1982) because of its before/after dimension. With perception verbs, the gap comes to be interpreted as a temporal gap between perception and conclusion, as well as a gap between perceiving subject and cognitive subject. This gap is, in a way, bridged by the inferential process which takes place thanks to a perceptual-data basis and which leads to a cognitive, conceptual conclusion. TO can be considered as a kind of discursive indicator or pointer showing the way that leads from perception to cognition; as such, it has an iconic dimension, mirroring the mental path the perceiver-experiencer takes.

46The inferential reading of the sentence goes together with the lexical shift undergone by the main verb, which is itself produced by the semantic content of the TO-clause. Indeed, the state of affairs it denotes is often expressed by stative verbs or processes, which constitute the great majority of processes in the TO-infinitivals under study. Among them can be found stative lexical be (1), as stated in grammars, but also have (10), other stative verbs (11), as well as perfect (12) and passive (13) verb phrases. The nature of the infinitive predicate is in keeping with the inferential reading of the sentence, as a perceiving act can only grasp a dynamic process, an event unfolding. Since in TO-clauses the process is stative, it cannot be grasped by a physical act, but can only derive from a mental process.

(10) Now my sister lives in Dorset but I must admit that, aside from a traditional cream tea and a pretty decent portion of fish and chips, I have never noticed Dorset to have particular regional food specialities. (GloWbE – GB)

(11) If we take our task to be an exercise in intersubjective translation, in speaking for others and their point of view, our hubris will cause us no end of difficulties, moral and philosophical. And if we see it to lie in the formal analysis of social systems or cultural structures, statistically or logically conceived, we evade the issue of representation and experience altogether. (COCA)

(12) However, the big day might be seriously marred by a group that feels itself to have been left behind in Russia’s ostentatious embrace of global competition. (COCA)

(13) Both sufferers and non-sufferers were asked about their perception as to the cause of incontinence (table VI). In both groups one quarter perceived it to be caused by a medical condition, particularly more older male sufferers. (BNC)

47However, contrary to what is claimed in many grammars, an inferential reading is not the sole possible interpretation for such sentences. The verb can – albeit less frequently – retain its perceptual sense as an event is perceived, and not inferred. In sentences as (14–16), the infinitive verb is dynamic and refers to an ongoing process occurring in the same time-span as the perceiving phenomenon. This scenario is quite common with hear (14), far more so than with other perception verbs (e.g. listen and see in [15–16]).

(14) Harry hears Suzy to be saying that her flaws are genial ones. She carries on about a particular failed negotiation. (COCA)

(15) One of the things I was struck by in listening to him to preach in particular, is that he is very empathic towards the pain and suffering in his congregation. (GloWbE – US)

(16) It amazes me to see people to vote for someone that hasn’t done anything to restore this country. (GloWbE – US)

48The use of TO-infinitivals in such sentences can be explained by the fact that TO is iconic of a mental path leading from perception to cognition. In sentences in which an inferential reading prevails, the whole mental process from perceiving to inferring has been completed. In sentences such as (14–16), TO might be taken to indicate that the perception-cognition pathway exists, but that the perceiver is still at the perceptual input level; the perceptual data obtained in the perceiving act could lead to some inference, and thus to the conceptual level, but this is not what matters most in the sentence. This observation is in keeping with Duffley’s “before/afterness” hypothesis (19–21) with respect to the use of TO-infinitive clauses.

49It should be added that a majority of such sentences are found in websites; see, for instance, example (16) taken from ABC News. Here again, this might be a sign of current changes, occurring first in informal language and/or discourse situations. This trend could permeate formal language and TO-infinitivals with perceptual sense could then become standard constructions of perception verbs.

Conclusion

51In this article, it has been shown that the sentences in which perception verbs take TO-infinitive complements contradict most of the norms that are prescribed or described in grammars. The current study puts forward a semantic explanation accounting for the (in)frequency of active perception verbs with TO-infinitivals. I have also demonstrated that a meaning of sensory perception can sometimes be conveyed in such sentences, in which case the interpretation of mental judgement can be disputed – though this may be the most relevant description provided in grammars with respect to TO-infinitive constructions taken by perception verbs. It therefore appears that the TO-infinitive complementation pattern licensed by perception verbs raises the question of where to place the limits of the field of perception, as the sentences under study can be said to be situated at the frontier between perception and cognition.

52Though no significant difference has been found between British, American and Canadian varieties of English, noteworthy dissimilarities have been identified with respect to the register and mode variations; indeed, Internet websites provide a majority of occurrences. It should already be clear that the structures studied in the current paper, though of limited use in current-day speech, are undergoing changes: they are more and more widely used in Internet English, in which they contribute to a perceptual reading that is more frequent than in traditional written or spoken modes. It would be interesting to broaden this study so as to encompass other varieties of English, whether they be of native speakers – as in Australia and Ireland, for instance – or of non-native speakers – such as in India and Jamaica. Such a study would crucially determine whether the so-called non-standard TO-infinitive constructions of perception verbs are subject to the same changes in other varieties of English, among which so-called ‘non-standard’ ones. It would then perhaps be possible to verify the hypothesis that the current changes identified in this paper are a sign of the evolution of the English language worldwide.

List of Tables