How is structure built and interpreted?
Alcohol is used, in our culture, for a lot of things. Lubricating social occasions, antifreeze, preserving foods. But alcohol is not what it is used for. To understand why alcohol has the uses it has, we need to understand its structure. The reason alcohol can have these various uses is its chemical structure. I think that this is a good analogy for language. Language is used for many things: honing our thoughts, interacting with other people, creating literature, planning our actions. But it can be used for these things because of its structure: in each utterance we make, every time we think through a problem in our heads using language, we make use of a rich set of relations and connections between the basic units of language. What are the basic units, and what are the relations?
My latest attempt to get at what is going on is work stemming from my 2013 book, A Syntax of Substance. There I argued for a radical separation between the systems that create syntactic relations from one kind of basic unit (uncategorised root-like units), and those that identify how the structures created interface with systems of meaning and of sound via another kind of basic unit (functional category labels). I argued that Chomsky’s Merge operation can create unary branching structures, opening up this new architecture (following the lead of work by Brody in the 2000s). This perspective leads to the rejection of ‘functional’ categories as separate heads in the syntactic representation: rather they are just labels for structures built up out of basic units of sound and meaning. So the basic units are of two kinds: syntactically uncategorised elements that are at the heart of structure building, and category labels that link structures to their meanings and pronunciations. The relations are just those allowed by a very bare, binary, Merge operation.