The modularity of syntax and three types of strong/weak alternations

old_uid5742
titleThe modularity of syntax and three types of strong/weak alternations
start_date2008/12/01
schedule10h-12h
onlineno
location_info/, salle 159
summaryUninterpretable occurrences of phi-features ("agreement", broadly speaking) are explored as window onto the modular architecture of language, particularly the character of syntax as a module distinct and autonomous of morphophonology (realization) and interpretation. Current models question aspects of this modular architecture and the place of uninterpretable phi in it: some relegate uninterpretable phi wholly to the realizational component(s) outside syntax; some rather reduce the scope of the former and so shift much allomorphy to syntax; and some identify the syntactic and interpretive components, leaving no place for uninterpretable elements. Our exploration suggests that the classical conception is on the right track, deriving significant generalizations about uninterpretable phi phenomena in language. Some of them have the signature expected of morphophonology: syntactico-semantic inertness and arbitrariness, and a restriction to word-like domains. Here belong some "opaque" cliticization and agreement and some analytic-synthetic alternations. By the same criteria, other, superficially similar phenomena referring to uninterpretable phi prove to belong to syntax. Evidence comes particularly from person-governed interactions between arguments, such as the me-lui (Person Case) constraint and its "repairs". Thus, the modular separation of syntax and morphophonology properly individuates two classes of uninterpretable phi phenomena. The tools provided by the syntactic class of phenomena also furnish a new tool to address the relationship of "interpretable" phi-features on arguments and their actual interpretation. Systematic mismatches between these two modules surface in fossil (grammaticalized) phi-specifications, metonymy, impostors, and fake indexicals. Syntax thus seems autonomous of the expected properties of the interpretive component as well as of the realizational one, constituting the classical autonomous computational module linking the two.
responsiblesAroui