A learnability space
What is built where
What the language learner has to learn is what is built where. In the framework here, speech and language are learnt in a mental space. This is what Marlys Macken in 1995 characterised as ‘the learnability space’. Exactly what defines this space remains the topic of intense, ongoing research and argument. It has to be large enough to encompass the known variation across different languages, but not so large as to encompass variations which do not in fact occur. This space may SEEM to be factored into two components:
- The various categories, what are traditionally known as ‘parts of speech’ or in a clearer, more modern way as ‘syntactic categories’ – nouns, verbs, adjectives, and so on, and different sorts of speech sound like vowels and consonants defined on particular features, and the assembly of these elements into syllables and words, according to what is known as the ‘phonotactics‘ of the language;
- The different ways the categories are defined.
But this apparent distinction can’t be real, or acquisition could not progress reliably to a common result or ‘end-state‘. There is an obvious circularity, constituting one aspect of a logical problem. By the proposal here, the circularity is resolved by what Martina Wiltschko (2014) calls ‘the Universal Spine Hypothesis’ or USH. By the USH, children are born knowing that language is organised by a decision tree into which categories are fitted. The child effectively retraces the process by which the USH evolved, but at an enormously accelerated tempo, over less than two years, rather than the millions of years of human evolution, thanks to the strategic advantage of having fully competent guides to learn from.
One significant variation across languages is whether particular elements are pronounced. Another is when, in the course of the derivation, they become such that they can be pronounced. This must be a significant factor in speech and language acquisition. And the formation of words and what Chomsky in 1965 regarded as the grammar are all part of one derivational sequence.
But children can easily misinterpret what they are hearing, or cannot implement what they do hear, or don’t have, or don’t yet have, the mental tools for a correct analysis. The USH isn’t a magic bullet.
By the proposal here, what the language learner has to learn involves both speech and language, that is to say how words are built – by what is known as ‘morphology‘, how they are put together – by what is known as ‘syntax‘, how the speech sounds or ‘phonemes’ are formed – by what is known as ‘phonetics’, and how system of sounds is organised – by what is known as ‘phonology‘, by the way these things work together through the ‘architecture‘.
The characteristic errors in child speech are by failures to assemble the sound structures in full in the right positions. These failures can occur at any of a considerable number of points, not an infinite number, but enough in combination to allow the enormous variety of incompetences which actually occur. Listing the commonest of these errors as ‘processes‘ has the effect of hugely understating the the scale of what can go wrong in children’s speech.