The Pragmatics of Constituent Order in English

 

In recent work, I have argued for the existence of sets of syntactic variants, or ‘alloforms’, of noncanonical-word-order constructions, defined by a shared discourse function but conditioned by their syntactic environment. For example, inversion and long passives share a functional constraint, but appear in complementary distribution syntactically, and thus may be considered alloforms of a single construction. At the same time, an analysis of inversion as an alloform of both preposing and postposing fully accounts for its distribution in discourse, without the need to stipulate an independent pragmatic constraint. Similar reasoning suggests alloform status for other pairs of noncanonical structures. But increasing numbers of such relationships raise a worrisome question: If, for example, long passives are an alloform of inversion, and inversion is an alloform of presentationals, what does that mean for the relationship between long passives and presentationals? In this talk, I confront the inevitable problem of tying it all together. Using patterns of both syntactic and functional distribution, I argue for a class of preposing structures that constitute one abstract construction with syntactically-conditioned alloforms, and a class of postposing structures that constitute another such construction, along with a class of argument-reversing structures constituting conditioned alloforms of both. I then consider the ramifications of this analysis for various approaches to the notion of a construction, and argue for a view of noncanonical constructions as functionally constrained structural templates for syntactically-conditioned alloforms.