“Crash vs. Yield”: On the Conflict Asymmetry in Syntax and Phonology

Recent advances in Optimality Theoretic syntax reveal that the basic analytic concepts and formal machinery developed in OT phonology, such as a notion of markedness, the nature of constraints (functional/typological/cognitive grounding) and constraint interaction, harmonic alignment and constraint conjunction, and edge alignment, can be fruitfully applied to the domain of syntax, and they provide a fundamental explanation for various syntactic problems, particularly those that lie at the boundary between syntax and morphology. In recent work for instance, a classical notion of syntactic markedness familiar in the structuralist tradition (e.g. Jakobson 1932 [1984]) and typological work (e.g. Greenberg 1963, 1966) has been reconsidered in light of the OT approach to a typology of pronominal systems (e.g. Bresnan In Press b, Forthcoming a), systematic gaps in morphological paradigms as observed in the English negative auxiliaries (Bresnan In Press a, Forthcoming b,c), and so-called ‘blocking effects’ where lexical categories preempt phrasal counterparts (Bresnan’s work cited above; see also Sells (1998a) for blocking). Bresnan’s work on typological variation in pronominal systems shows that cross-linguistic patterns of variation in morphosyntactic inventories follow from an interaction of context-free and context-sensitive faithfulness/markedness constraints in much the same way as that in phonology, predicting an asymmetric cross-linguistic distribution of some marked forms (e.g. reduced pronominals), positional neutralization, and allotactic distribution of marked and unmarked forms, and further predicting certain distributional patterns to not exist in languages (e.g. languages with only reduced pronominals, or those that reserve reduced pronominals only for non-core arguments). Harmonic alignment and constraint conjunction have been extended to account for intricate facts about word order variation and freezing (Lee 1999), case marking (Aissen 1999, Forthcoming, Donohue 1999, Sharma 1999), pronominal incorporation (Arstein 1998, Donohue and Lee 2000), argument-function mapping (Asudeh 1999, Morimoto 1999), and possessive constructions (O’Connor 1999a,b), all of which exhibit sensitivities to prominence hierarchies such as syntactic rank, semantic roles, person, animacy, and definiteness. Harmonic alignment of these prominence hierarchies and constraint ranking together provide the formal means of modeling language-particular markedness