Constraints on pseudogapping constructions with do

Constraints on pseudogapping constructions with doThis paper is part of a wider research project into the ways in which do functions in a variety of anaphoric structures, and the interaction between its particular semantics and its anaphoric potential. Here we focus on a subset of structures involving what generative syntacticians have named “pseudogapping” (cf. Levin 1986). The structures we are interested in are clauses in which a form of the verb do – sometimes in combination with an auxiliary – stands for a lexical verb in a preceding clause, as in:(1) PeopleinGreecedrinkmoreouzothantheydobrandy.(Levin1986:16)The constituent to the right of do, usually called a “remnant” (e.g. Lasnik 1999), semantically contrasts with a constituent in the antecedent clause.Linguists have very different acceptability judgements with respect to pseudogapping constructions (see Lasnik 1999: 152-53). The only way to circumvent this difficulty is to rely on corpus data. We have started searching the British National Corpus for relevant occurrences and have so far collected numerous instances where the remnant is a PP:(2) Such gestures appealed to many Romanians, as they did to Western observers anxious to see Ceaucescu as a “patriot” and reformer.(3) WeneverhadhuskiesonBirdIsland,butwedidontheotherbases.It is useful to note that the prepositional remnant in (2) repudiates an argument of appealed; whereas that in (3) contrasts with an adjunct to have. Now the very existence of (2)-(3) may appear to clash with the grammaticality judgments of a panel of informants we have consulted:(4) *JohnwenttoGermanyandMarydidtoAustria.Go (to) and other verbs such as leave (for) or live (in) take locative complements that behave more like arguments than like adjuncts. What (2) and (4) have in common is that the repudiated PP is an argument; but there is a relevant difference: the verb in (2) is (indirect) transitive. What (3) and (4) have in common is that the repudiated element is a locative complement; in (4), however, it is an argument while in (3) it is an adjunct. This suggests the following hypothesis: pseudogapping is blocked with a PP when the latter repudiates an argument of an intransitive verb.Although (3) and (4) seem to involve bona fide pseudogapping (cf. Lasnik 1999: 163), they are very close to Verb Phrase Ellipsis. Compare with:(5) There was a boy at school who was deaf, but he could read lips so we didn't have toshout. But I do with Mr Frost. He's always saying “What say, boy?”.The only difference here is that the element with which the remnant contrasts is not mentioned in the antecedent clause, but it is retrievable from the context: “with that boy”. If that difference turns out to be only secondary, there may actually be no clear-cut boundary between pseudogapping and certain forms of VPE.So far, we have found no reason to take issue with linguists who looked at pseudogapping independent of which auxiliary was used: the semantics of do has not proved to be a relevant parameter. Where its semantics does come in is in connection with examples like:(6) I called CCDO and told him I was going to post [the signs] around the neighborhood.He agreed to take some and canvas one section while I did the other.Here, it is difficult to decide if we are dealing with pseudogapping do or with lexical do used as a “general” verb in phrases such as do the living-room.ReferencesLasnik, H. 1999. “A Note on pseudogapping”. In Minimalist Analysis. Malden, Mass:Blackwell, pp. 151-74.Levin, N. 1986. Main Verb Ellipsis in Spoken English. New York: Garland.Souesme, J.-C. 1985. Do something et ses diverses réalisations en anglais contemporain.Unpublished PhD thesis, Université Paris 7.