With one point seemingly worth preserving

In a novel way, Nunes (2002) postulated a dedicated learnability function or device which applied equally to syntax and to phonology, evaluating the evidence of what the child heard, the ‘primary linguistic input’, as this is known. Chomsky (1965) postulated a Language Acquisition Device’ or LAD applying only in relation to the possible variables of syntax, such as where the verb appears in the sentence, whether the subject is always pronounced, and so on. Nunes (2002) attempted to extend such a function to phonology. Such a function had to be based on one system shared by both areas of the grammar. It would determine whether the grammar allowed syllables without vowels as their heads, as there are in words like English little, but not in most other languages including French, Spanish, Italian, and Greek, most African languages spoken to the South of the Sahara, and so on. In the framework widely assumed in 2002 and still assumed by many linguists, these variables are thought to work in terms of ‘parameters’, working like switches, pointing one way or the other. By the proposal of Nunes (2002), this learning and learnability function was a critical genomic entity. By conceptual necessity, it had to be the last step in modern human evolution. Addressing the issue of multi-factoriality and co-morbidities in disorder, a small congenital defect in the specification of this function would be expected to have global effects. And as the last evolutionary step, this would be expected to be a relatively unstable aspect of modern human cognition. Learnability would be vulnerable and commonly compromised. Necessarily, for obvious biological reasons, this function had to be simple in some appropriate sense. Crucially, such a function explained the finiteness of learnability, as manifest in the  fact that native speakers of English can agree that “The chicken is ready to eat is ambiguous,  while “The chicken is ready for eating” is not.

The notion of such a dedicated function explains:

• Multifactoriality, specificities, and asymmetries of disorder;
• The finiteness of learnability.

But the notion of a dedicated function has at least these three major defects. It

• Implicitly exaggerates the commonality between different instances of disorder,
• Makes it hard to explain the specificities of familial disorders across generations and family locales, other than by the interface with anatomical, neurological, and other physical strucrtures;
• Has little, if anything, to say about the highly uneven and asymmetric distribution of phonological ‘processes’, with, for instance, calculator as KALTALAYTA only occurring towards the end of the acquisition period and only in a small set of words, and working in the opposite polarity from doggie as GOGI (see Cruttenden (1978)) and magnet as MAGNIK.

Accordingly, I now believe that the notion of such a function serves no good purpose.

But the underlying reasoning had one useful element, I believe. Forcibly, in a way that had not applied in the framework of Chomsky’s 1965 proposal, a function applying equally to syntax and to phonology, entailed the principle of branching at most two ways or what is technically known as ‘maximally binary branching’ in phonology as in syntax – with demonstrable benefits in speech pathology.