Direct compositionality (review)

The fourteenth installment of ‘Oxford studies in theoretical linguistics’ offers an impressive lineup of papers in formal semantics that are concerned with direct compositionality. This constraint requires not only that a semantic theory be compositional, but that syntactic and semantic composition be intimately linked in a certain way—each syntactic rule should correspond to a semantic rule, and anything that is recognized as a syntactic constituent should be assigned a meaning. Apart from an introduction, Direct compositionality contains eleven papers.

David Dowty’s ‘Compositionality as an empirical problem’ (23–101) is a lengthy but fairly accessible treatment of a variety of topics connected with compositionality. Essentially, it contains (i) a discussion of suitable desiderata for semantic theorizing, (ii) an a priori argument in terms of adaptive considerations in favor of certain desiderata, (iii) the tentative proposal that a context-free semantics—something quite close to a directly compositional semantics—might satisfy these desiderata, and (iv) some illustrations of this kind of semantics.

‘Direct compositionality on demand’ (102–31) by Chris Barker offers proof to the effect that the grammars in a certain proper subclass of the type ‘logical grammars’ are such that every derivation they can generate has a syntactically and semantically equivalent, directly compositional counterpart.

Chung-Chieh Shan’s ‘Linguistic side effects’ (132–63) attempts to come to terms with some challenges to compositionality (stemming from, for instance, binding and quantification) by exploiting an analogy between these and phenomena discussed in computer science. Shan makes use of two movement-governing control operators (‘shift’ and ‘reset’) in order to illustrate the utility of the analogy.

In ‘Type-shifting with semantic features: A unified perspective’ (164–90), Yoad Winter addresses type shifting, a central component of many directly compositional accounts. He argues that Partee and Rooth’s (1983) original account of type shifting should be amended in order to cover not only type mismatches, but also certain syntactic-semantic mismatches (such as those between morphosyntactic number and semantic number).

Pauline Jacobson, in ‘Direct compositionality and variable-free semantics: The case of “Principle B” effects’ (191–236), details a directly compositional account of ‘Principle B effects’, concerning why pronouns cannot be coindexed with a c-commanding NP within the same local domain. Relatedly, the article by Ivano Caponigro and Daphna Heller, ‘The non-concealed nature of free relatives: Implications for connectivity in specificational sentences’ (237–63), addresses connectivity effects in specificational sentences that pose problems for directly compositional approaches (that cannot rely on well-established analyses based on c-commanding). Drawing on data from several different languages, Caponigro and Heller present crosslinguistic evidence against the most promising nondirectly compositional alternative account of connectivity effects (the ‘question-answer’ approach) and suggest that the directly compositional approach to these effects might be worth pursuing. An opposite position is taken by Maribel Romero in ‘Connectivity in a unified analysis of specification subjects and concealed questions’ (264–305), in which Romero argues in favor of a nondirectly compositional analysis of connectivity effects in specificational sentences. Romero maintains that a variant of the (concealed) question-answer approach is the most promising one.

Rajesh Bhatt and Roumyana Pancheva’s ‘Degree quantifiers, position of merger effects with their restrictors, and conservativity’ (306–35) discusses data connected with degree quantifiers: for instance, the fact that the complements of degree heads always show up in an extraposed position, and the fact that the surface position of the complement of the degree head marks the scope of the entire degree expression. Although the discussion is not explicitly framed as a challenge to direct compositionality, the preferred solution makes use of techniques that are not available to the proponent of direct compositionality. (It posits syntactic rules that do not correspond to any semantic rules, such as the raising of the degree head.)

In ‘Two reconstruction puzzles’ (336–62), Yael Sharvit discusses which-interrogatives and...