It is obvious that there is structure in language. Talking and understanding are not just about words, but about the structure into which the words go.It seems obvious that there are words. But what are they exactly? And what else is there? There are words, composed of parts. And the words and the parts of words of words are ordered with respect to each other. And there are meanings in the words and their parts and different meanings according to how they are ordered and the situations in which they are used. How is the structure here assembled? In what order?
- Is the meaning conceived first, some desired intention, with the words strung together fully formed to express that intention?
- Or is the structure defined first, with the words fitted into the structure according to the intention?
- Or are the meanings separate from the structures and the intentions?
- Or is the building in parts, with various sub-assemblies?
All of these ideas have been entertained.
It is obvious that this structure is learned. But how does the learning process work? How does it reach a predictable conclusion, what we call a ‘grammar’, despite an infinite variety of input evidence, much of it not fully grammatical when adults fail to complete their sentences or accidentally make what they would recognise as grammatical mistakes if others made them. This is not obvious at all. It is not even obvious whether to call it ‘learning’.
For those (many) linguists who think that there are questions worth asking here, it may seem that the answer is in what is now known as the ‘architecture’ of language, that is to say, its most essential parts.
The first formulation of the questions here was by Chomsky (1965). He proposed a distinction between two criteria, what he called ‘descriptive adequacy’ – describing what does, and just as importantly, what does not happen in a language, and ‘explanatory adequacy’ – doing this in such a way that it can be learnt by children in a given time frame (from birth to around the age of ten) from limited evidence varying from one child’s experience to the next child’s. And he proposed a dedicated ‘Language Acquisition Device’.
By the framework assumed here and the proposal here, following some aspects of Chomsky’s current work, there is an architecture by which the meanings and intentions are built step by step, or in what are now generally known as ‘phases’. For example, in English “Who is he doing?” is a question and “He is seeing who?” is not a question but an expression of some surprise. In the first, who is displaced from one point in the structure to another. In the second it stays put. The structural difference is an abstract question element on the left edge which draws who left wards, but only in the case of a true question.
The elements of this architecture, defining the core of what the language learner has to learn, are universal, but the ways they are brought into play are language-specific, and thus forcibly part of what Marlys Macken (1995) called the learnability space.