Syntax facit saltum: Computation and the genotype and phenotype of language

Human language has long captured the imagination of biological researchers. However, the gulf separating ‘computation’, ‘biology’, and ‘language’ has been equally long-standing-in large measure resulting from the gap between linguistic and biological description: we do not expect to literally find a ‘passive grammar rule’ inside a person’s head. Similarly, we evidently do not find a corresponding passive rule-specific ‘dysphasia’ that destroys one’s ability to say, “This problem was solved by the student” while sparing, along with the rest of the grammar, the ability to say “The student solved this problem”. The puzzle is a classic biological one: how to bridge between a ‘genotype’ and a ‘phenotype’, in this case perhaps the most complex behavioral phenotype we know. Linguistics has cast natural languages’s intricate and ultimately behavioral ‘outer form’ or ‘phenotype’ at an abstract level far removed form language’s computational and biological ‘inner form’ or ‘genotype’. Thus, even though linguistic science’s successful program the past 40 years has resulted in perhaps the richest description of a human ‘genotype to phenotype’ mapping that we know of-the initial substrate for human language and how language develops in an individual-until recently, progress at a more ‘reductionist’ level has been far more halting. The aim of this article is to show how recent developments in linguistic theory dubbed the “minimalist program” (MP) [l, 21 might help bridge the biology-language ‘abstraction gap’, because the minimalist program shows that despite its apparent surface complexity, language’s core might in fact be much simpler than has previously been supposed. For neuroscientists pursuing clues left by the linguistic phenotypes’s ‘fault lines’ down to the level of the real genotype, as exemplified in the work of Gopnik and colleagues [3, 4, 51, or even the so-called ‘candidate gene approach’, this is a promising development. The minimalist program is eliminative in exactly the right sort of way, since it boils down all syntactic relations and rules, including transformations, to feature matching-thus tying linguistic theory more tightly than ever to accounts such as Gopnik’s model of Specific Language Impairment. More generally, the minimalist program serves as a case study for how a complex behavioral phenotype can emerge from the interactions of a much simpler ‘genotype’. In particular, the minimalist program posits that the human ‘syntactic engine’consists of just three components: (1) words, that is, word semantics and word morpho-phonology; (2) word features; and (3) a simple operation that glues together or merges words and word complexes. This article demonstrates how just these three components interact to yield many, perhaps all, of the ‘special design features’ of human syntax. By ‘design features’ we simply mean