Retour vers liste

Détail de la contribution

Auteur: Amy Rose DEAL

Short and long scrambling in Nez Perce

Abstract/Résumé: Nez Perce is a language with extremely free word order. From a minimalist perspective, this is naturally understood as involving a fixed underlying structure subject to various phrasal movements -- scrambling. Nez Perce shows scrambling both within and across clauses. As is typical, long scrambling behaves like A’ movement: it shows weak crossover and superiority effects. Yet these effects are strikingly absent in short scrambling. (1) 'isii-ne_1 pewewpexciix 'ipnim_1 lawtiwaama? who-OBJ are.hitting his friends? Who_1 are his_1 friends hitting t_1? (2) 'ituu-ne 'isii-nm peepe? what-OBJ who-ERG ate What did who eat? (3) 'isii-nm 'ituu-ne peepe? who-ERG what-OBJ ate? Who ate what? Similar facts in Hindi (Mahajan 1990), German (Wiltschko 1997) and Warlpiri (Legate 2002) have been given a simple explanation. In short scrambling, the object may be initially A-moved past the subject. This allows the object to bind into the subject, obviating WCO, and leaves the object relatively local to A’ positions in the left periphery, giving apparent superiority violations. WCO and superiority reappear in long scrambling because A-movement is clause-bounded. This explanation leads us to expect that the WCO and superiority obviations characteristic of short scrambling will correlate with other diagnostics of A-movement, such as case and agreement. Yet agreement in particular poses a curious challenge. Nez Perce has a set of C heads which agree with local persons within the CP. C may probe both a 1st and a 2nd person argument when the former is underlyingly superior to the latter, (4). When 2nd is underlyingly superior, however, C may not access 1st, (5). (4) ke-m-ex kaa cewcewteetu, C-2-1 then call When I call you, (5) ke-m-(*ex) kaa cewcewteetu, C-2-(*1) then call When you call me, This suggests that C probes for [participant,addressee] -- features which are only partially matched by 1st (so probing continues) but fully matched by 2nd (so probing stops). Crucially, C may never access a representation in which the object is higher than the subject, even when the object appears to the subject’s left: (6) Ke-m-(*ex) kaa 'iin-e(=cim) 'ee cewcewine, C-2-(*1) then 1SG-OBJ(=only) you called When you called (only) me, It is difficult to account for this fact derivationally. A-scrambling feeds binding (and thus LF) and word order (and thus PF); it must happen in the narrow syntax. Movement simply does not create copies that C can probe. I propose that the restriction on agreement in (6) results from the deactivation of the object upon A-scrambling. Nez Perce allows its DPs to phi-Agree no more than twice. By the time the object has agreed in situ and agreed with the scrambling head, it presents no further activation potential and cannot be a goal for Agree with C.