adults, as previously described, are still in progress. These studies will eventually provide us with information not only about the precise competences of the various groups of signers and their acquisition, but also about the relation of these competences to age at first exposure to the language and to details of input as well. At present, I can offer little insight into why the learning process leads in consistent directions. Nevertheless, the evidence suggests that it does, and that it may, under certain circumstances, change the language.
According to our linguistic analyses, signed languages (at least American Sign Language) have the same kind of analytic character as spoken languages. First, ASL marks the same kinds of semantic distinctions marked by many spoken languages. More importantly, ASL has the same kind of formal apparatus for marking these distinctions: a limited number of parameters along which a limited number of discrete values are signaled, and a shell-like, hierarchically organized combinatorial system governing the permissible ways in which these discrete units may co-occur. Moreover, this analytic kind of organization appears in ASL despite the very clear iconic base for these forms, and despite the resulting potentially different options the visual-gestural modality makes available.
Our linguistic analyses are supported by our studies of the acquisition of ASL: Young deaf children appear to enter the system making precious little use of the iconic or analogue possibilities; rather, they doggedly perform their own formal analyses, morpheme by morpheme, over a period of several years.
Finally, and more speculatively, it appears that certain of these learners perform this analysis on an input that is not itself fully analyzed, creating morphological structure where their parents had mastered predominantly frozen, unanalyzed forms. This finding, along with others in the literature on language evolution and acquisition, suggests that the explanation of such structural regularities may lie in an understanding of the learning process itself.
This research was supported in part by a grant from the Research Board, University of Illinois, by NSF grant ♯BNS-76-12866 to the Salk Institute for Biological Studies, and by PHS grant ♯MH-15828 to the Center for Human Information Processing, University of California, San Diego. All figures were drafted by Ted Supalla; Figs. 1, 2, 5, 6, 7, 8, 9, 10 are reproduced, with permission, from Supalla (in press). I would