Phrase Structure As a Representation of “Primitive” Grammatical Relations

Abstract

One of the defining properties of the narrowly Chomskian approach to syntax’ over the years has been its commitment to the idea that grammatical relations such as subject and object are not primitives of grammatical theory. Rather, these notions, inherited from traditional grammar, are to be understood in terms of more basic syntactic relationships, in particular phrase structure configurations. This view gives phrase structure a kind of primacy in Chomskian theory that it does not have in other frameworks. In this conceptually oriented paper, I want to reexamine this distinctive claim, reviewing the old controversy between Chomskian theory and Relational Grammar (RG) and related frameworks that that claim was part of. In the course of this, I will consider what the ongoing legacy of these debates is in the “post-RG” syntactic theories of the 2000s. This issue is roughly equivalent to the question of what the characteristic phrase structure representations of Chomskian theory actually mean within the network of assumptions that they are now embedded in. My basic claim is that the meaning of the old slogan has changed somewhat over time, as ideas about phrase structure have changed, in ways that have not been fully realized. In particular, the Chomskian notion of phrase structure has come partially unhinged from its origins as a representation of basic constituency facts. As a result, it is now fair to say that phrase structure is essentially a representation of grammatical function relationships, not fundamentally different in kind from the representations posited by the Relational Grammarians and others. However, I will argue that Chomskian phrase structure is a particularly good representation of grammatical relations. In particular, it is superior to relational nets or the f-structures of Lexical Functional Grammar (LFG) in that it captures certain basic properties of linguistic prominence that are not captured by these more general and flexible-looking representational schemes. Thus, while the meaning of the claim that grammatical relations are derived from phrase structure has shifted over time, that claim is still meaningful. Finally, I will ask why it is that phrase structure is such a successful representation, showing how this relates to lexical semantic work on the decomposition of verbs into more basic components.