Catenatives or complex VP the debate about specific verbs in English
by
Jan Niehues
Paper submitted in partial fulfillment of the requirements for a graded credit for the course “Problems of English Grammar” in Summer Term 2005 Submission Date: October 10, 2005 Approved by: Prof. Dr. Jürgen Handke Philipps University Marburg
2
Contents 1 Introduction......................................................................................................... 3 2 Catenative Verbs................................................................................................. 4
2.1 Definitions..........................................................................................................4 2.2 Auxiliaries, modals, modifier, operator, quasi-modal........................................6 2.3 From finite to non-finite VP...............................................................................7 2.4 Simple vs. complex verb phrases....................................................................... 7 3 Simple and complex catenatives.........................................................................9
3.1 Simple catenatives............................................................................................10 3.2 Complex catenatives........................................................................................ 11 4 Classes and classification..................................................................................13
4.1 Criteria for classification..................................................................................13 4.2 Syntactical or semantic grouping..................................................................... 13 5 Conclusion / Summary......................................................................................15 6 References.......................................................................................................... 16
3
1
Introduction
Most linguists agree that there is a particular feature of certain verbs like e.g. want , begin, try or seem that sets them apart from other verbs: their ability to be combined into chains of verbs, to 'catenate' (Lat. catena: chain). (1) I don't want to have to be forced to begin to try to make more money. (Palmer 1987: 172) The term usually used for these verbs is 'catenative verbs'. There remains some form of disagreement about almost every aspect of these verbs, however, ranging from the question which verbs actually are catenative, to the problem of how to analyse or categorise them. Some linguists even question the need to define a class of catenative verbs in the first place. Huddleston concedes that: "This is one of the most difficult areas of English grammar and despite a great deal of intensive study over the last twenty years there remains much disagreement over the most basic aspects of the analysis."(Huddleston 1997: 209)
The treatment of the grammatical phenomena is further complicated by the fact that linguists tend to introduce their own categories or descriptions for existing categories. This is particularly evident in the argument of modals vs. auxiliaries vs. operators. An approach differing from that traditionally taken by grammarians is that presented by Dieter Mindt who bases his observations on the analysis of a corpus of actual language. This paper will try to give an overview of the theories concerning catenative verbs,
their
relation
to
the
auxiliaries
and
their
features
of
clause
complementation. Due to the scope of work that has been published, only the main approaches will be considered. The field of semantics in particular would merit a much closer look on the effects of sentence taxis 1.
1
And indeed there has been a rather fervid debate about this between Huddleston on the one side and Matthiesen and Martin on the other, conducted largely in the 'Occasional papers for systemic Linguistics'.
4
2
Catenative Verbs
Here, the main definitions of catenative verbs will be presented, noting their different approaches to certain features and their treatment of the auxiliaries.
2.1
Definitions
Richard Hudson defines catenatives as "verbs that combine with a following non finite verb". He includes "verbs like get, keep, start, help as well as the traditional auxiliary verbs". (Hudson 2002) Sample sentences given by him include: (2) a) She was/got chosen for the job. b) She was/kept talking. Huddleston and Pullum state that a catenative is present in "most cases where a non-finite clause is an internal complement of a verb". They illustrate this by giving cases of non-catenative complements: predicative complements (3a), objects (3b) and PP complements (3c). (3) a) Kim seemed a keen student. b) Kim began the journey. c) Kim hoped for a successful outcome. (Huddleston 2005: 215) Gramley and Pätzold agree by defining verbs which are followed by nonfinite verb forms but which are not operators as catenative verbs. (Gramley & Pätzold 1992: 132) Palmer defines catenatives as verbs that combine with a full verb into verb phrases of theoretically unlimited length. (cf. 1) Although he applies the term 'complex phrase', he contrasts complex phrases utilising catenative verbs against examples such as: (4) I bought the boat to sail the world. (Palmer 1987: 172pp.) In (4), there is hardly any semantic relationship between the clauses, whereas catenatives usually imply some semantic restriction on the following verb. Palmer
5
sees a much tighter semantic and syntactic relationship, similar to auxiliary verbs, exemplified by the impossibility of certain constructions: (5) a) *He kept to talk. b) *He has talking.
(Palmer 1987: 172pp.)
He specifically excludes infinitives of purpose and of result, only the usage in (6a) being catenative. (6) a) I promise to make you happy. b) I promise, to make you happy. (Palmer 1987: 206) Palmer rejects the approach of analysing the subordinate clause as a nominal that is the object of the catenative verb, thereby setting the catenatives alike to transitive verbs. While this analysis may be applicable to certain constructions, it is by no means a valid description of sentences like (7a). This is obvious from the fact that a subordinate clause can appear with verbs that do not allow an object. (7) a) He decided to go. b) *He decided the plan. (Palmer 1987: 212) The easiest method of distinguishing between catenatives and full verbs are the TNP tests, namely tense, negation and passivisation. While a catenative can be marked for tense and negation simultaneously with its main verb, a full verb only allows this once. This is shown by the following: (8) a) have remembered seeing... b) remembered having seen... c) have seen... (9) a) have not remembered seeing... b) remembered not having seen... c) have not seen... [INT1] The passivisation test helps to distinguish catenatives from transitive verbs with an object complement. While a transitive verb with a direct object can be passivised easily, this is not true for catenatives. (10)a) The girl liked working. b) Working was liked by the girl. (11)a) The girl kept (on) workig. b) *Working was kept on by the girl. [INT1]
6
2.2
Auxiliaries, modals, modifier, operator, quasi-modal
Often, the definition of a separate class of catenative verbs requires some redefinitions and further distinctions to be made amongst the auxiliaries. These are of course different with each approach taken towards the catenatives. Hudson's decision to include the traditional auxiliary verbs amongst the catenatives leads him to establishing a class of non-catenative operators, allowing him to distinguish between catenative and non-catenative uses of be and have. While the use in (2) is termed catenative, that in (12) is not. (Hudson 2002) (12)She is ready. - Is she ready? Despite the fact that "there is no clear line between auxiliaries, catenatives and other verbs that may have subordination" (Palmer 1987: 29), Palmer dismisses the suggestion to treat all auxiliaries as full verbs. He argues that, although here "[t]he TNP tests are ... rather inconclusive" (Palmer 1987: 31), a distinction can be made on semantic grounds. While the auxiliaries do not cause semantic restrictions on the choice of subjects, restrictions may occur based on the first full verb. He contrasts: (13)a) The water may run down the street. b) *The water intended to run down the street. Thus, the auxiliaries are only modifiers to the full verb which is the main verb of the verb phrase. (Palmer 1987: 31) Huddleston, in his 2005 collaboration with Pullum, revised his position on the auxiliaries that had been criticised by Palmer (Palmer 1987: 28) stating that "auxiliaries, when used as markers of tense, aspect, mood or voice, are catenative verbs, entering into the simple catenative construction", generally taking raised subjects. (Huddleston 2005: 219) Already in 1997 he had explained his position of "not applying the term auxiliary to what [he is] calling the operator class". (Huddleston 1997: 143) In their 'Comprehensive Grammar of the English Language', Quirk et al. argue for a gradient of modality, ranging from the central modals consisting of one verb phrase (can, may, shall &c.) to main verbs with non-finite clauses that are
7
analysed as two verb phrases (e.g. hope with to-infinitive). The catenatives occupy a position between the semi-auxiliaries (have to, be about to ...) and the main verbs. (Quirk 1985: 137) While they resemble the auxiliaries in that most of the catenatives share the independence of subject, they are "nearer to main verb constructions than are semi-auxiliaries, patterning entirely like main verbs in taking do-support". (Quirk 1985: 146)
2.3
From finite to non-finite VP
There are four basic non-finite forms of main verbs: the bare infinitive, the toinfinitive, the -en and the-ing form (the latter summarising gerunds and present participles). These occur in simple as well as in complex catenative verb phrases. A rather isolated position towards the to-infinitive is taken by Richard Hudson who suggests the analysis of to as a non-finite auxiliary verb on its own. (Hudson 1998: 68)
2.4
Simple vs. complex verb phrases
A rather straightforward definition for the distinction between simple and complex VP is given by Quirk et al. who define the finite verb phrase as being simple when it consists of one word and complex when it consists of more. (Quirk 1985: 151) According to Palmer, the three criteria that can be used to distinguish between simple and complex verb phrases are: tense, negation and passivisation; called the TNP tests. In a simple verb phrase both tense and negation can occur only once and a simple phrase can be passivised without problems. Sequences of such phrases are then complex verb phrases. (Palmer 1987: 28) He summarises that: "Phrases involving primary auxiliaries are fairly clearly simple, while those with catenatives are complex (though not all pass all the tests). Phrases with modals lie somewhere between the two, sharing characteristics of both simple and complex phrases." (Palmer 1987: 28)
This is not completely congruent with the more traditional approach of applying the term 'complex verb group' to the combination of one to three
8
auxiliaries with a lexical verb or a complex verb (i.e. a lexical verb with a complement). [INT1] Instead of arguing for a clean cut distinction between cases involving one or two verb phrases, Quirk et al. argue the gradient from auxiliary to main verb implies a similar gradience in the analysis of verb phrases, resulting in catenatives having intermediate status between a single and a multiple verb phrase analysis. They suggest that (14) "consists syntactically of two verb phrases would have had and to have arrested , but semantically only of one". (Quirk 1985: 154p.) (14)...we would have had to have arrested you. Quirk et al. are accused by Huddleston of not being too precise on their point of modality and he suggests they use "one analysis in which auxiliaries a nd related expressions are treated as verbs taking non-finite complementation, so that a more comprehensive and systematic account could be given of the verb/adjective + nonfinite construction". (Huddleston1988 2: 350) Huddleston criticises Halliday's definition of verbal groups put forward in his 'Introduction to functional Grammar', where a sequence of words of the primary class verb constitutes a verbal group (the sample given is: didn't know how to spell ), as this definition implies, according to Huddleston, "that to and not are either verbs or not words". (Huddleston 19881: 172)
9
3
Simple and complex catenatives
Most linguists agree on four basic forms of catenative constructions that can be distinguished, based on the type of non-finite construction that is employed. The four basic types of catenatives as exemplified by Palmer: (15)a) bare infinitive b) to-infinitive c) -ing form d) -en form
He helped wash up. He wants to go to London. He keeps talking about it. He got shot in the riot. (Palmer 1987: 173)
The choice, which non-finite construction to use, is largely dependent on the catenative verb, while some allow variation others allow only one type of nonfinite construction. (Huddleston 1997: 210) When there is a choice of constructions, there is generally some difference in meaning. Palmer contrasts the -ing form and the to- infinitive, obviously carrying different meanings: (16)a) He was seen walking away. b) He was seen to walk away. (Palmer 1987: 189) Huddleston notes that only the bare infinitive and the to-infinitive "occur with significant numbers of catenative verbs", and additionally, clauses involving the -en form are invariably passive. (Huddleston 1997: 211) As with the VP, the catenative verbs can occur in simple or complex constructions. A different approach is taken by Mindt in his 'Empirical Grammar of the English Verb', who not only recognises the four classes mentioned above, but also defines catenative auxiliary constructions (e.g. is going to be, are allowed to do) and catenative adjective constructions(e.g. were able to escape). He states that the last two classes are far less frequent than the four 'central catenative constructions' and backs up his claim with data from his corpus analysis 1 (Mindt 1995: 285). His description of the function of modal verbs and catenative verbs is criticised by Kamphuis as being "in fact identical". (Kamphuis 1996: 89) The classification of catenative verbs proposed by Mindt is severely criticised by de Haan according to 1
This data, however, is somewhat problematic as he does not give any sources for his corpus, stating only that he "had access to more than 240 million words of English". (Mindt 1995: 6)
10
whom he "fails to justify the existence of the category of catenative verbs"; while he acknowledges that there are good reasons for wishing to distinguish such a category, he accuses Mindt of not presenting "any convincing arguments for this", and not being consistent in his argumentation. (de Haan 2002) One of Mindt's definitions is that "catenative verbs allow the overlap of two meanings within one verb phrase. This overlap cannot be achieved by modal verbs alone, because a verb phrase cannot contain more than one modal verb." (Mindt 1995: 469) This, in turn, is rejected by de Haan as well.
3.1
Simple catenatives
The identification of the subject in simple catenative sentences is usually unproblematic. Here the subject of the subordinate clause is identical with that of the main clause. Huddleston, however, notes that there are one or two exceptional verbs, giving say as an example: (17) Your mother said to meet her at two o'clock. (Huddleston 2005: 216) In (17), the subject cannot be determined syntactically, it has to be inferred from the context. Whereas Huddleston's 'Introduction to the Grammar of English' only distinguished between the verbs of the seem and expect classes (noting that most verbs belong to the latter) (Huddleston 1997: 212pp.), Huddleston and Pullum specify this by distinguishing between ordinary and raised subjects in simple catenative phrases. While an ordinary subject is in semantic relation to the verb, a raised subject belongs semantically to the subordinate clause. While (18a) contains an ordinary subject, the subject in (18b) is raised. A simple passivisation test illustrates the difference: While (18b) and (18d) have the same meaning, the passivisation of a sentence containing an ordinary subject will result in a change of meaning, e.g. from a) to c). (Huddleston 2005: 216) (18)a) Sara wanted to convince Ed. b) Sara seemed to convince Ed. c) Ed wanted to be convinced by Sara. d) Ed seemed to be convinced by Sara.
11
3.2
Complex catenatives
The same constructions as in (16) are also possible with an intervening noun phrase between the catenative and the following verb. (19)a) bare infinitive b) to-infinitive c) -ing form d) -en form
He helped them wash up. He wants them to go to London. He kept them talking a long time. He had the rioters shot. (Palmer 1987: 174)
The identification of subjects in complex catenative phrases is not as straightforward as it is in simple ones. Palmer shows the subject of the subordinate clause to be identical with the intervening noun phrase by contrasting sentences with reversed NP. (20)a) The farmers want the hens to lay eggs. b) *The hens want the farmers to lay eggs. (Palmer 1987: 178) The identity relations are not, however, the same with all verbs. It is argued by Palmer that for some verbs "The intervening noun phrase is both the object of the verb of the main clause and the subject of the subordinate", noting this difference on the example of: (21)a) I wanted the doctor to examine the boy. b) I persuaded the doctor to examine the boy. (Palmer 1987: 179) By application of the passivisation test it is shown that the doctor in (21) is indeed the object of persuaded but not of wanted: (22)a) The doctor was persuaded to examine the boy. b) *The doctor was wanted to examine the boy. c) I wanted the boy to be examined by the doctor. d) I persuaded the boy to be examined by the doctor. (Pal mer 1987: 179) Passivisation of the main clause shows that (22a) is possible but (22b) is not. Likewise, the passivisation of the subordinate clauses results in a difference of meaning between the two sentences. A further feature of complex catenative verbs is that they frequently occur with prepositions. Here, the choice of preposition is determined by the catenative verb. These constructions can be classified according to the same patterns as catenatives without prepositions. Thus, Palmer likens 23a) to 23b),
12
(23)a) I long for John to come. b) I want John to come. (Palmer 1987: 185pp) Analogous to the distinction between ordinary and raised subjects with simple catenatives, Huddleston distinguishes ordinary and raised objects in complex catenative constructions. While the raised object, although it belongs semantically to the catenative complement, is syntactically embedded in the superordinate clause, the ordinary object is related semantically to the verb of the superordinate clause. Again this can be tested by means of passivisation: The meaning of (24b) remains the same when passivised to (24c), thus it is a raised object, that of (24a) does not, (24a) therefore having an ordinary object. (24)a) We urged a specialist to examine Ed. b) We wanted a specialist to examine Ed. c) We urged Ed to be examined by a specialist. d) We wanted Ed to be examined by a specialist. (Huddleston 2005: 221) Some verbs allow construction with a PossP instead of a NP. While all PossP can be replaced by NP, not all verbs that take a NP allow the possessive construction. Huddleston notes that the difference between the two forms is stylistic, (25b) being the more formal one. (Huddleston 1997: 221) (25)a) Ed resented your father opening the mail. b) Ed resented your father's opening the mail.
13
4
Classes and classification
There is an obvious interest in grouping catenative verbs into distinct classes of verbs. This can be done according to a variety of criteria, both semantic and syntactic. An important decision that is to be made is whether to allow multiple class membership for certain verbs. Rather of theoretical importance is the decision, if verbs are allowed multiple class membership, whether to treat this as cases of homonymy or polysemy. Palmer acknowledges his decision as "often fairly arbitrary", judging on practical reasons on a case to case basis.
4.1
Criteria for classification
Two distinct approaches toward the classification of catenatives can be seen. One method is to compare the syntactic patterns exhibited by each construction, from the type of infinitive used to the patterning of intervening NP. Another approach is to define groups of related meanings together, e.g. verbs of report, perception, attitude or need.
4.2
Syntactical or semantic grouping
Gramley and Pätzold distinguish "some thirty different classes of catenatives", grouped according to their meanings. They count 500 - 600 verbs that are involved. This high number is partly due to the fact that some verbs are counted twice or more as "multiple class membership is common". (Gramley & Pätzold 1992: 168) Palmer distinguishes three classes of catenative verbs according to the pattern of subject identification, one class of verbs without, and two classes that take an intervening NP, labelled the persuade and the want pattern respectively. (Palmer 1987: 179pp.) He admits that this is no ideal solution as there are several verbs that do not fit either of this patterns. Examples he gives for such complicated
14
cases include: order , believe, expect and promise. In the case of order , he states that the distinction between the two classes is simply not valid, believe can be handled if the concept of subject raising is applied. He summarises that: "There is a great deal of indeterminacy here; the best we can do is to state the facts that there are verbs that may occur with either construction, ie may or may not have the NP as the object of the main clause, though often with no clear distinction between the two, and that there are others such as BELIEVE that permit main clause passivization, even though semantically the NP is not the object of the main clause." (Palmer 1987: 183)
According to Palmer, the use of all possible criteria for the classification of catenatives (i.e. the type of non-finite form, the pattern of identity relation and the specific features of tense, phase, aspect and voice) would result in "a vast number of classes", making it preferable "simply to approach the problem lexically, to list the verbs and to state for each individually, its characteristics in terms of the criteria." (Palmer 1987: 187) Reducing the number of criteria would lead to verbs being either placed in multiple classes or borderline cases not really fitting anywhere. Palmer proposes to apply a semantic grouping of verbs that, actually, bears close correspondence with the syntactic categories. He introduces nine classes with a total of 31 subclasses. (Palmer 1987: 191pp)
15
5
Conclusion / Summary
There remain the different approaches of whether to apply strict categories in analyses, or to allow for some gradience between two points (most prominently concerning finiteness and modality). Also, how to assess the importance of syntax or semantics, as an analysis will frequently differ in results based on whether a syntactic or a semantic approach was favoured. Due to the fact that there is no agreement yet (nor is there likely to be) over even the broader points of the analysis of catenative verbs, nor are there definite solutions for a number of problems associated with them, certain implications arise. It is obvious that the topic of catenative verbs will not feature very prominently in language learning, thus beginning learners (i.e. in schools) will most likely never come into contact with the more detailed points of their analysis. It will be enough to note that some verbs can 'introduce some sub-clause', but teachers will not have to be able to explain catenatives in detail. Another factor is the presentation of information in dictionaries: Huddleston states that "there is no getting away from the fact that the lexical entries for verbs must specify which kinds of complements they take and, where more than one is involved, the semantic differences (if any)". (Huddleston 1997: 210p.) While traditional dictionaries mark all verbs as transitive, intransitive or both, Huddleston suggests that this distinction is not very comprehensive stating the need for syntactically adequate lexica giving specific information on valid complements where "the transitive/ intransitive contrast will have a good deal less of importance". (Huddleston 1997: 223) This would then merit a closer look at the concept of valeny and extensive valency deicionaries.
16
6
References
de Haan, Pieter. 2002. "Review article on An empirical grammar of the English verb: Modal verbs, by Dieter Mindt." In Journal of English Linguistics. 30. pp. 274280. http://www.let.kun.nl/~p.dehaan/reviews/Mindt_2000_review.php. Gramley, Stephan / Pätzold, Kurt-Michael. 1992. A Survey of Modern English. London and New York: Routledge. Halliday, M.A.K . / Matthiesen, M.I.M. 2005. An Introduction to Functional Grammar . London: Arnold. Huddleston, Rodney . 19881. "Constituency, multi-functionality and grammaticalization in Halliday's Functional Grammar." In Li nguistics. 24. pp.137-174. Huddleston, Rodney . 19882. "Review article on A comprehensive Grammar of the English Language, by Randolph Quirk, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. Index by David Crystal." In Language. 64. pp. 345-354. Huddleston, Rodney . 1997. Introduction to the Grammar of English. Cambridge et. al.: Cambridge University Press. Huddleston, Rodney / Pullum, Geoffrey. 2005. A Student's Introduction to English Grammar . Cambridge et. al.: Cambridge University Press. Hudson, Richard . 1998. English Grammar . London and New York.:Routledge. Hudson, Richard. 2002. Linguistics Association of Great Britain - Language Fact Sheet: Auxiliary Verbs. http://www.phon.ucl.ac.uk/home/dick/aux.htm. 1 Kamphuis, Vera. 1996. "Review article on An empirical grammar of the English verb: Modal verbs, by Dieter Mindt." In ICAME Journal. 20. pp. 86-90. Mindt, Dieter. 2000. An empirical Grammar of the English Verb System. Berlin: Cornelsen. Palmer, Frank Robert. 1987. The English Verb. London and New York: Longman. Quirk, Randolph / Greenbaum, Sidney / Leech, Geoffrey / Svartvik, Jan.1985. A comprehensive Grammar of the English Language. London and New York: Longman. 1
html document without page numbering, thus references are given to entire document
17
Internet Sources [INT1] www.linguistics-online.de; module: Formal Aspects of the Verb - catenatives, accessed: September 10, 2005.
