advantages of human language are just the kind of cognitive phenomenon that natural selection is sensitive to.
Therefore our best hypothesis is that language has evolved incrementally in response to natural selection. This is the
position I will take here; I will therefore attempt to provide some parts of a plausible evolutionary scenario. Two basic
insights contribute to breaking the logja mposed by Cho msky's version of UG. First, Chapters 5 and 6 have begun to
decompose the language capacity into many semi-independent parts. It thus becomes possible to ask to what extent
they could have emerged either independently or in sequence.In this respect I will concur with Toulmin's incremental
story—with the major difference that the pieces added incrementally are specifically linguistic rather than general-
purpose.
The second insight that contributes to breaking the evolutionary impasse posed by a “grammar box” is Derek
Bickerton's proposal for two incremental steps in the evolution of language (a similar and somewhat more highly
structured proposal is offered byGivón1995). Althoughtheproposal tobedevelopedherediffers fro mBickerton'sin
many respects, he provides an excellent starting point for substantive discussion.
8.2 Bickerton's proposal and auxiliary assumptions
Inhis bookLanguage and Species(1990), Bickertonproposes thatthehuman capacityfor languageevolved intwostages.
His second stage is language as we know it—let's call it“modern language.”He calls thefirst stage“proto-language”;
for afirst approximation one can think of it as modern language minus syntax. His hypothesis is that for several
million years, hominids spoke only in a protolanguage, and that the development of modern language is perhaps as
recent as 50,000 years ago, with the appearance ofHomo sapiens.
What elevates Bickerton's story above mere speculation is his claim that protolanguage is still present in the modern
human brain. It surfaces when modern language is disrupted; examples are pidgin languages (Bickerton 1981),
“Genie,”the woman isolated from human contact from age three to thirteen (Curtiss 1977), and possibly agrammatic
aphasics. It also surfaces in situations when modern language has not developed: on one hand in early child language,
and theother handintheexperimentsinteachinglanguage toapes (Linden1974; Savage-Rumbaugh etal. 1998). Thus
evolution did not throw a Good Idea away; rather it built on it. (This story is reminiscent of Rod Brooks's (1986)
notion of a“subsumption architecture,”in which new, more refined systems are added on top of less articulated
existing ones.)
Bickerton (1990) still views the development from protolanguage to modern language as a single rather miraculous
leap (in Calvin and Bickerton 2000 he