For several decades, Chomskyan linguistics and Darwinian evolution have been seen by most scholars as antithetical. A handful of linguists (Pinker, Bloom, Newmeyer) have tried to reconcile the two frameworks. However, their proposals satisfy neither side, and fail to show why one particular set of universals evolved rather than any other.

What follows is a synthesis of Chomskyan and Darwinian thinking consistent with a broad range of facts which shows why syntax takes the shape it does.

Some two million years ago a structureless protolanguage developed spontaneously under conditions in which extractive foraging and predator avoidance strategies played major roles. Currently popular theories relating language origin to social intelligence treat hominids entirely out of context, extrapolating from current conditions among apes.

However, one aspect of social intelligence is implicated in the subsequent development of syntax: reciprocal altruism, which probably predated protolanguage by millions of years. Reciprocal altruism cannot work if one can cheat successfully (accumulate favors without returning them). Cheaters can only be detected if animals maintain a calculus of who did what to whom. This in turn requires the creation of abstract categories (Agent, Theme, Goal) and the casting of event memories into an argument structure frame.

Some time (perhaps 1.5 million years) after protolanguage developed, cheater detection was adapted to give protolanguage a basic phrasal- clausal structure. Delay was due to patterns of brain development. Protolanguage began in brains of at most 6-800 ccs. Simple one-way messages went from the lexical store to motor areas: nothing was assembled before dispatch.

Language requires assembly of complex structures prior to utterance; this in turn requires that representations remain coherent throughout. But coherence requires (a) large assemblies of neurons firing synchronously (b) fast and efficient transmission between prefrontal cortex (home of verbs and grammatical morphemes) and temporal cortex (home of nouns and adjectives).

These conditions were satisfied by brain enlargement and growth of the arcuate fasciculus--developments supported by long growth-curve activities such as precision throwing, rather than by protolanguage, which would have remained unaffected until a coherence threshold was achieved. Crossing that threshold made it possible to superimpose argument structure on protolanguage output. Construction and parsing of long, complex sentences, however, remained problematic.

But the existence of partially syntacticized language created new selective pressures favoring variations in brain development that supported government and case-marking by means of grammatical morphemes, as well as algorithms defined over argument structure for the identification of null elements. This selection took place in the period between the emergence of h.s.sapiens and the cultural explosion of forty thousand years ago, to yield language as we know it today.