The Big Mess Construction

There is a construction in English, exemplified by how long a bridge, which is so irregular that it has been named the Big Mess Construction, see Berman (1974). This paper first sketches its main characteristics (section 1) and a treatment of the internal structure of the noun phrase which serves as a background for the analysis (section 2). It then presents three ways in which the Big Mess Construction can be analysed; two of them are lexicalist and are shown to be implausible; the third is constructivist and is argued to be superior (section 3). In a next step, the discussion is extended to two other types of constructions. The first concerns the English adnominal reflexives, as in the children themselves, and is shown to require a constructivist analysis which is similar but not identical to the one for the Big Mess Construction (section 4). The second concerns the combination of such and what with the indefinite article, as in such a pleasure. In spite of its obvious resemblance with the Big Mess Construction this combination does not require a constructivist analysis; instead, it fits the lexicalist mould of most of the rest of HPSG (section 5). 1 The Big Mess Construction In English noun phrases the determiner canonically precedes the prenominal adjectives, both the lexical and the phrasal ones. (1) a. a big house b. a very big house (2) a. * big a house b. * very big a house A notable exception are the adjectival phrases which are introduced by as, so, too, how, this and that. When they occur in a nominal which contains the indefinite article, they precede the determiner (Huddleston and Pullum, 2002, 435). (3) a. It’s so good a bargain I can’t resist buying it. b. How serious a problem is it? (4) a. * It’s a so good bargain I can’t resist buying it. b. * A how serious problem is it? This construction, for which Berman (1974) coined the term Big Mess Construction, only ocurs in nominals with an indefinite article. It does not occur in nominals with another kind of determiner, as in (5a), nor in nominals without determiner, as in (5b). For their comments and suggestions for improvement I thank the anonymous reviewers of the HPSG-2007 programme committee, the attendants of the workshop on constructions and grammatical theory (Stanford, July 21 2007) and my colleagues at the Centre for Computational Linguistics in Leuven. (5) a. * How serious some problem is it? b. * They are so good bargains I can’t resist buying them. A further complication is provided by the APs which are introduced by more or less. They can either occur in the canonical position or in the exceptional one (Huddleston and Pullum, 2002, 435). (6) a. This is a more serious problem than the other. b. This is more serious a problem than the other. Also here, the exceptional position is only possible in combination with the indefinite article. What makes the Big Mess Construction interesting is not only its idiosyncracy and the descriptive challenges which it raises, but also the light which its treatment sheds on the issue of the trade-off between lexicalism and constructivism in formal grammar. To pave the way for the treatment I first present my analysis of the internal structure of the noun phrase (section 2). It deals with the canonical order, as exemplified by (1) and (6a). The exceptional order, as exemplified by (3) and (6b), is modeled in section 3. 2 The internal structure of the noun phrase My treatment of the internal structure of the noun phrase is based on two assumptions. First, that the noun is the head of the noun phrase and, second, that the prenominal dependents are functors, in the sense of Allegranza (1998) and Van Eynde (1998). Since the first assumption is controversial, given the fact that many authors treat the determiner as the head of the noun phrase (cf. Abney (1987), Hudson (1990) and Netter (1994)), and since the second assumption may be unfamiliar, I start with a defense of the former and a succinct presentation of the latter. 2.1 The head of the noun phrase To substantiate the claim that the noun is the head of the noun phrase adopt the commonly, though often tacitly, made assumption that a noun phrase shares its person, number, gender and case values with its head daughter. Of special relevance are, hence, the noun phrases in which the determiner has other values for these features than the noun, since they allow us to identify the head by simple observation. Here are some of such examples: (7) My neighbors are/*am rich. (8) a. What birds have/*has two wings and four legs? b. What comes/*come next? (9) a. A good many pages are/*is lost forever. b. A few pages are/*is still missing. Given the form of the finite verb in (7) the subject NP must be plural, which implies that its head daughter can be the third person plural neighbours, but not the first person singular my. A similar remark applies to the interrogative determiner in (8a). Given the form of the finite verb, the subject NP in (8a) must be plural, which meshes well with the assumption that the plural birds is the head, but not with the alternative assumption that the interrogative what is the head, since what is by itself singular, as shown by (8b). Further evidence is provided by the quantifying determiners in (9). Also here, the form of the finite verb demonstrates that the subject NPs are plural, and while this is perfectly compatible with the assumption that the plural pages is the head, it is at odds with the alternative assumption that the head is the quantifying a good many and a few, since these are both singular, as demonstrated by their compatibility with the indefinite article. To provide an example which turns on the case distinction I switch to Dutch. (10) Wiens whose paard horse heeft has hij he gestolen? stolen? ‘Whose horse did he steal?’ The fronted NP wiens paard ‘whose horse’ is the object of gestolen ‘stolen’ and, hence, accusative. This is compatible with the assumption that the non-genitive paard ‘horse’ is the head, but not with the alternative assumption that the genitive wiens ‘whose’ is the head. For more arguments in favor of the NP-hypothesis and against the DP-hypothesis, see Van Eynde (2006). 2.2 The prenominal dependents Turning now to the prenominal dependents the central assumption of the functor treatment is that specifiers and modifiers had better be treated along the same lines. The distinction between specifying determiners and modifiers goes back to Chomsky (1970) and is motivated a.o. by the fact that a head can take at most one specifier, whereas it can take any number of modifiers. Within the lexicalist HPSG framework this is reflected by the assumption that a noun lexically selects its specifier, but not its modifiers, see Pollard and Sag (1994) and Ginzburg and Sag (2000). The feature which models the selection of the specifier is a valence feature, called SPR, and its role in the analysis of the noun phrase is illustrated in (11).