Retour vers liste

Détail de la contribution

Auteur: Michelle SHEEHAN

Towards a parameter hierarchy for alignment

Abstract/Résumé: Following the format in Roberts (2012), this paper presents an attempt to characterize the parameter hierarchy governing case/agreement alignment. It has long been known that there is no single ‘ergativity parameter’ regulating alignment in transitive clauses (Moravcsik 1978, Dixon 1994). While split-ergativity (whereby a language is accusative in some contexts and ergative in others at the clausal level) may not exist (cf. Coon & Preminger 2012), various different alignments are fairly uncontroversially attested: morphological ergativity (Anderson 1976), split-S and fluid-S systems (Dixon 1994: 73-8, Comrie 2011, Deal 2012), syntactic ergativity (Dixon 1994, Manning 1996), which can be subdivided into High ABS and Low ABS (Aldridge 2004, 2008, Coon, Mateo Pedro & Preminger 2011, Legate 2008). The proposed alignment hierarchy provides a new perspective on these patterns, building on the insight that ERG is a theta-related case/Case (Woolford 2006). We show that this hierarchy not only provides a coherent minimal description of attested alignments but also explains certain important gaps and one-way implications, notably the nonexistence of languages which ERG-mark only unergative subjects. It also explains another much discussed alignment asymmetry: the fact that apparently no language has ergative agreement and accusative case alignment, though the reverse is possible (Anderson 1977, Corbett 2006). This follows because, according to the proposed alignment hierarchy, a language with ergative case morphology can be underlyingly accusative (ERG can be quirky) whereas the reverse is ruled out. We further argue that the same kind of parameter hierarchy should be extended to cover alignment in ditransitives (cf. Dryer 1986, Baker 1988, Haspelmath 2005, Malchukov et al. 2010). Assuming, following Aoun & Li (1989), that goals are base generated above themes, the ‘ergative’ pattern inside VP is one where goals receive a theta-related case (DAT), and themes get structural ACC by agreeing with v. Once again, we propose that the theme must raise past the goal in such instances to be accessible to the probe v. The ‘accusative’ pattern is instantiated in secundative languages in which the goal gets structural ACC (cf. Yoruba, Atoyebi et al. 2010). Crucially, both alignment hierarchies have the same basic structure: Is X present? Is X always present? Is X associated with EPP? etc. Furthermore, it can be shown that much of the ordering of parameters can be attributed to extra-linguistic factors. As such, the hierarchy itself can be seen as emergent from the process of language acquisition.