Generate

And a generalisation

There is a simple, but very signifcant, generalisation:

Around the age of one and a half, a child learning English says or shows an understanding of a particular sort of syntactic structure, and shortly afterwards, between a week and three months later, asks or shows an understanding of a Wh-type question about the same sort of thing.

By showing an understanding, I mean responding in words in a way which is plausible and accurate – to a degree appropriate for a child of this age. By a Wh-type question, I mean a question beginning in English with a word like who, what or where. There are a number of other similar words, how, why, when, but who, what and where are the first to emerge. By the same sort of thing, I mean individual or physical identity or location.

All human languages have Wh-type questions. There are two main ways of asking them, with the Wh form at the beginning or on the left edge, or with the same form at the same point in the structure where an answer is expected. To a question like “Where is your friend?” Reasonable, interpretable answers would include; “Outside” or “My friend is outside.”  But “Outside is my friend” would be barely interpretable. Languages like English have the Wh form on the ledt. Languages like Japanese and the languages of China have the Wh form somewhere else. French has both. Languages like Kurmanji, the most widely spoken language of Kurdistan, have the different Wh forms in different positions. These are things which the language learner has to learn. They have to fall within the learnability space. But this is very obvious to child learners, with numerous clear examples to be heard everyday. It is not usually problematic.

The generalisation above holds across five children, including the two subjects here and three from an older study. It needs to be tested with children learning languages other than English, and particularly with children learning languages unrelated to English, perhaps including Arabic, Japanese, Turkish.

The generalisation above is significant because it is compatible only with a generative grammar, that is to say, a grammar which generates structures in either direction, to say them or to understand them. A first step towards a generative grammar was taken by an Indian scholar working about two and a half thousand years ago. But modern generative grammar seeks to capture grammatical knowledge as a whole, specifying not only what is said, but what is not said, and cannot be said in a way that other native speakers of the same language will understand. The goal, in other words, is to be as precise and as all-encompassing as possible.

Traditional grammars are silent on whether there was anything which was disallowed. They mostly spell out in detail the different forms of nouns, verbs, adjectives and adverbs, and constructions that differ most sharply with those of the writer’s first language. But this captures only a small amount of the grammatical knowledge of a normally developing three-year-old.

It is often said that a child has a specific problem with ‘expressive language’ as opposed to comprehension. But this may be an artifact of the test protocol and actually understate the problem.

Generative grammar is more ambitious than traditional grammars. This scholarship, now guided by an on-going world-wide tradition of research over the past 60 years, is what motivates the approach here. It should be noted that it represents an area of scholarly inquiry which predates sociology, economics and studies of gender by thousands of years.

This generative approach has guided my research, both practical and theoretical, for the past 49 years.  Possible Words Therapy, for example, is a strictly generative idea. Any attempt to reversion it in a non-generative framework, as by ReST for example, just does not make sense, and can’t be expected to work successfully, in the way described here. The claims made by ReST proponents are significantly less than the claims I make here and in Nunes (2002, 2006, 2009).

The various topics discussed here are not intended to promote a generative perspective or to belittle any other perspective. But neutrality on this issue may be not truly feasible, despite my best intentions otherwise.