EESE 6/99


1 Cf. Herbst/Klotz (1998) for details of the valency dictionary project.
2 Cf. Halliday's (1973: 29) view of the language system as a "meaning potential, represented as a network of options, which are derived from a particular social function and are realized, in their turn, by structures whose elements relate directly to the meanings that are being expressed." Cf. also Sinclair's (1991) description of the open-choice principle as "a way of seeing language text as the result of a very large number of complex choices."
3 For a survey of dependency models see Hudson (1993).
4 A similar approach is taken by Seuren (1996: 16).
5 The majority of the examples used in this article is taken from one of the following sources: AC - accidential corpus (consisting of the following books: T. Cross: (1994) The Shining Sands. Artists in Newlyn and St Ives 1880-1930 , T. Cross (1995): Painting the Warmth of the Sun. St. Ives Artists 1939-1975. Tiverton/Cambridge: Westcountry Books/The Lutterworth Press, M. Gooding (1994): Patrick Heron, London: Phaidon), CC - the Cobuild corpus, DC - sentences from the Cobuild corpus contained in the English Valency Dictionary; QE - quoted examples. I would also like to thank David Heath for his assessment of example sentences and his valuable comments on an earlier version of this paper.
6 Cf. Matthews (1981).
7 For a slightly different use of the term semantic valency see Helbig (1992).
8 Cf. Allerton (1975) and also Matthews (1981: 125). Helbig (1992: 106) uses the terms definit fakultative and indefinit fakultative Ergänzungen to reflect the distinction between Allerton's contextual and indefinite deletion.
9 For the criterion of verbatim recoverability see CGEL (12.32).
10 For a survey cf. Herbst/Heath/Dederding (1980) and Helbig (1992).
11 Cf. Buysschaert (1982), Somers (1984) or Herbst (1987). 
12 Cf. Helbig/Schenkel (1968) and Herbst (1983 and 1987).
13 For possible modifications of the who-/what-question criterion see Herbst (1987: 36-9).
14 Cf. Seuren (1996: 1). See also Matthews (1993: 142-5).
15 Cf. also Brinker (1977) and Herbst/Heath/Dederding (1980). See also Welke (1995: 172-174).
16 For the distinction between psychological, grammatical and logical subject cf. also Halliday (1985: 31-33).
17 Cf. Matthews (1981: 101) graphic representation of the two principles, where bracketing indicates a constuent analysis and the arrows indicate depedency relations:
18 Cf. Allerton's (1982) view of semantic, valency and surface levels.
19 Up to a point, this description is redundant since the subject can be described in terms of morphological and positional criteria.
20 The main differences between the two approaches lie in the use of the term discontinuous modifier by Aarts/Aarts (1982) and the view of determiners as a functional rather than a formal category. The latter aspect is however irrelevant to the present analysis.
21 All statements concerning the position of elements refer to sentences which can be considered to be neutral with respect to their theme-rheme-structure.
22 A further argument for this analysis is presented by the fact there are cases where a predicator allows a passive with a noun phrase complement but not with a that-clause complement, for instance, although the subject is the same in each case.
23 Note that this kind of approach is in line with a view expressed by Welke (1995: 172): "Die Behandlung des Subjekts als Ergänzung des Verbs muß nun aber keineswegs eine Sonderstellung gegenüber den anderen Ergänzungen des Verbs (oder eines anderen Valenztrögers) ausschließen". At the same time, this model meets Welke's (1995: 172) objection that numbering complements implies "Ungleichrangigkeit".
24 Cf. Stein (1979: 94) for a discussion of such examples.
25 For a discussion of the passive in this respect cf. Stein (1979), Svartvik (1966), CGEL (3.67-78).
26 Cf. CGEL's (16.27) discussion of They have a nice house QE .
27 Emons (1974: 116) establishes two complement classes E1 and E2 on the basis of the operation of commutation.
28 Emons (1974: 137) complement class E5 is characterized by the fact that such complements assign certain attributes to the referents of the E1 or E2. (He seems happy QE. We think him a coward. QE) Number concord between the E1 or E2 and the respective E5 and restrictions on the passive are also mentioned as characteristics of this complement class.
29 Allerton (1982: 82-88) also lists a number of syntactic criteria which help to support the distinctions made, such as restrictions on "tough-movement" for objoids or the fact that with measure-objoids how much-questions and what-questions are both possible.
30 Furthermore, the dividing line between these categories is not always very clear. Thus one might argue that thousands of pounds in His house in Hampstead must have cost thousands of pounds QE, which is classified as a predicator complement by Aarts and Aarts (1988: 142) meets the semantic description of attributes in the same way as the subject attribute It is very expensive or the object attribute in He found it very expensive
31 Cf. CGEL (16.21-2 and 16.45-6).
32 In CGEL (16.55), for instance, ditransitive verbs are defined as verbs with two objects but it also pointed out: "Some ditransitive verbs have two passive analogues." Allerton (1982: 102) introduces a category indirect objoid.
33 Aarts/Aarts (1988: 139) also include cases where no prepositional substitute is possible (such as I don't think he can forgive her her rudeness QE) under the indirect object.
34 Cf. Huddleston (1984: 68): "The primary syntactic effect of the exposition transformation is to deprive the embedded clause of subject status, but because the subject is obligatory in most classes of clause, this must be accompanied by the insertion of a dummy element to fill the vacated subject function."
35 It must be pointed out that an analysis of seem as divalent in sentences such as It seems daft that we are pumping away a resource from a reservoir into a river to flow into the sea (DC) as well as in Pumping away a resource from a reservoir into a river to flow into the sea seems rather daft might entail classifying it as monovalent in It would seem that he has been wrong all the time(QE).
36 Note that the discontinuous complements as outlined here contain as one of their constituents a pro-form referring to the other constituent and that their structure thus differs from that of discontinuous modifiers (as introduced by Aarts and Aarts 1988) or the discontinuous noun phrases listed in CGEL (18.39). The latter would also be treated as discontinuous complements in a valency account. One might consider expanding the notion of discontinuous complement to account for cleft sentences where, like in the cases discussed here, the it refers to a clause which is positioned after another complement: It was there that I elected to paint my first Newlyn picture AC. Such a complement could then be described as [it + relative clause].
37 This analysis is different from the treatment in government-binding theory, where the term expletive is used to refer to "elements in NP positions which are not arguments and to which no theta role is assigned" (Haegeman, 1991: 53). By considering such cases as discontinuous subjects, the it is also attributed complement status (or the status of being part of a complement) and thus no distinction has to be made between subjects that belong to the valency structure of a verb and subjects that do not.
38 The X is to represent the type of clause.
39 This kind of discontinuous complement is different in structure from that identified in 3.5.1, where the anticipatory element it could be seen as a kind of pro-form of the postponed element.
40 The bare infinitive should probably not be totally ruled out in such cases. CGEL (16.52) mentions a passive ("normally with a to-infinitive"). Corpus searches produce examples of the type: Very ordinary building materials have been made look interesting. Her parents didn't really want her go. The Test, which Gabba spectators had seen fluctuate so much over five days ..., ... the political changes we've already seen happen, which, however, were rated as unacceptable by some native speakers. Cf. also Jespersen (1914: 10).
41 Cf. 3.9 for the category ADV.
42 Jespersen (1914: 10) uses the notion of complex object to cover cases such as They called their boy Tom. He made his wife happy, which would be considered to be trivalent in a valency account. 
43 The complement status of such phrases is revealed by the who/what-question test, for example.
44 In CGEL (16.5) two analyses for She looked after her son QE are discussed, one as SVA, one as SVO (with look after as V). However, it is pointed out that her son is considered -"despite the fact that the passive is frequently possible" - the complement of the preposition after (a prepositional object) but not as the direct object of a verb look after. The potentiality of adverbial insertion (Many people looked disdainfully at the picture
45 Cf. Herbst (1996). For a critical discussion of the concept of prepositional verbs see also Lamprecht (1983: 63-68). Cf. Götz/Herbst (1989).
46 This is different in the case of a marginal complement category that has been labelled DESCRIPTION in the English Valency Dictionary to account for cases such as: In Gaelic the word "corimaeilia" means "hill of harmony". DC Sir Brian will be speaking about 'Education and Training Teachers at Warwick: The Next Decade'. DC When we label experiences as "being cold" or "being tired," the remedies seem naturally also to fall into appropriate categories and do not need to be rediscovered for each instance. DC
47 One might consider including the possibility of antecedent as in the definition of [QUOTE] to cover cases such as: (50d) AC For as he wrote in 1953, 'How can one paint the wholeness of anything without the true experience of one's complete self? ...'.
48 Without such an analysis it would be very difficult to distinguish between [QUOTE]-complements and comment clauses: How else, it was asked, could the most oppressive dictatorial regime in history be overcome, except by another dictatorship? CC.
49 Examples such as ask show that the link established between such clauses and that-clause complementation in CGEL (15.54) has to be treated with some caution. Examples of verbs where [SENTENCE] but not [QUOTE] occurs are: allow, assume, consider, decide, expect; etc., offer has [SENTENCE] and [QUOTE] but no that-clause; arrange takes [that-CL] but not [QUOTE] or [SENTENCE], for example.
50 It could be argued that a [there]p should be introduced to cover cases such as It seems to me there may be a few lessons learned by many seaside resorts. QE However, these could also be treated as combinations of [there]a and [V-ed CL], Compare also Haegeman's (1991: 54-55) treatment of expletive there.
51 Cf. 3.5.1. Note that Haegeman (1991) includes these uses of it under expletives, whereas CGEL (10.26b) states: "The prop subject it here should be distinguished from the anticipatory it of sentences like It was a great pleasure to see you again (...), where it has cataphoric reference to a postponed clausal subject."
52 Cf. Heringer (1967) and Emons's (1974: 118-122) account of the discussion of Witterungsimpersonalia in German linguistics. Compare also Matthews (1981: 103-5) discussion of the subject status of it in this context. See also Allerton (1982: 44): "... verbs like rain have no requirement for a subject in their valency (they are, in other words, zero-valent) and a lexically full subject is prohibited (except in figurative and poetic use); the surface syntax of English, however, requires the expression of a subject in every declarative sentence. The empty pronoun it fills the compulsory surface subject slot, when there is no valency subject ...". For zerovalent verbs see also Allerton (1982: 94-5).
53 Cf. 2.4.2. Compare also Haegeman (1991: 55).
54 Compare Lamprecht (1983: 152-3).
55 One might consider whether question forms, which are an important part of Allerton's (1982) classification of verb elaborators should also be seen as separate complement classes. This applies in particular to Allerton's (1982: 88) class "measure objoid" which takes what and how much questions.
56 Thus Emons (1974: 121) takes it as a realisation of E1; similarly, [it]a can be seen as a special, lexically restricted type of [NP]a in the present account for which [NP: it] would also be a possible notation.
57 For a detailed study of this phenomenon cf. Klotz (forthcoming 1999).