Phi-features and the Modular Architecture of - UMR 7023 - CNRS
Phi-features and the Modular Architecture of - UMR 7023 - CNRS
Phi-features and the Modular Architecture of - UMR 7023 - CNRS
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
192<br />
c. TNOM v S d. T vABS S Unaccusative<br />
(292) Obligatory Case Parameter (descriptive, see (295)): T, v are potential<br />
Agree/Case loci. If <strong>the</strong>re is only Agree/Case relation active, it is<br />
parametrically: v (vABS, ergative system) / T (TNOM, accusative system).<br />
The Obligatory Case Parameter states which potential Agree/Case locus is active<br />
when <strong>the</strong>re is only one DP, in unaccusatives, <strong>and</strong> it will be later implemented<br />
as a lexical property <strong>of</strong> T/v. The o<strong>the</strong>r Agree/Case relationships, <strong>the</strong> accusative<br />
<strong>and</strong> <strong>the</strong> ergative, must be limited to <strong>the</strong> remaining argument <strong>of</strong> transitives. Their<br />
distribution could be stipulated in <strong>the</strong> lexicon by specifying a phi-probe on <strong>the</strong><br />
transitive but not unaccusative v in nominative-accusative systems, <strong>and</strong> on T selecting<br />
a transitive v in ergative-absolutive ones (Albizu <strong>and</strong> Fernádez 2006).<br />
However, this does not capture <strong>the</strong> parallelism between two. Both are relations<br />
that surface only when <strong>the</strong>re remains a DP after <strong>the</strong> obligatory relation has been<br />
used up. The proposals <strong>of</strong> Marantz (2000), Bittner <strong>and</strong> Hale (1996), <strong>and</strong> Laka<br />
(2000) aim to do that. They group <strong>the</strong> ergative <strong>and</strong> accusative as dependent Case,<br />
which becomes automatically available if <strong>the</strong>re remains a DP after <strong>the</strong> obligatory<br />
Case has been discharged by a distinct DP, its Case-competitor, independently <strong>of</strong><br />
any selectional relationships. The pattern <strong>of</strong> PCC repair in unaccusatives is just<br />
such a last-resort emergence <strong>of</strong> ergative <strong>and</strong> accusative when <strong>the</strong> obligatory relation<br />
fails to license a DP in <strong>the</strong> PCC. 139<br />
Dependent Case dem<strong>and</strong>s an extension <strong>of</strong> syntax beyond ordinary<br />
Agree/Move dependencies, because it does not have a uniform path-<strong>the</strong>oretic relationship<br />
to <strong>the</strong> Case-competitor or <strong>the</strong> obligatory Case on which it depends. In<br />
(291), <strong>the</strong> ergative c-comm<strong>and</strong>s <strong>the</strong> absolutive, but <strong>the</strong> accusative is c-comm<strong>and</strong>ed<br />
by <strong>the</strong> nominative. Only one <strong>of</strong> <strong>the</strong>m can easily be made to depend on a given<br />
obligatory Agree/Case relationships while obeying <strong>the</strong> Path Condition (263). The<br />
ergative EA is on <strong>the</strong> path from <strong>the</strong> C/T/v system to O <strong>and</strong> can be made to depend<br />
on a relationship between <strong>the</strong>m (Bobaljik <strong>and</strong> Branigan 2006). The accusative O is<br />
below <strong>the</strong> path from C/T/v to EA, so its dependence on <strong>the</strong>ir relation would be<br />
countercyclic (Harley 1995: 4.3 note 104). Instead, <strong>the</strong> <strong>the</strong>ory <strong>of</strong> dependent Case<br />
proposes that distribution <strong>of</strong> <strong>the</strong> ergative <strong>and</strong> accusative is 'global'. They depend<br />
on <strong>the</strong> Agree/Case relation <strong>of</strong> ano<strong>the</strong>r argument in <strong>the</strong> same CP, without a fixed<br />
path-<strong>the</strong>oretic relationship to it. The same conclusion has been drawn about PCC<br />
repairs in section 5.3, <strong>and</strong> <strong>the</strong> same mechanism will be useful for <strong>the</strong>m.<br />
139 O<strong>the</strong>r instances <strong>of</strong> <strong>the</strong> emergence <strong>of</strong> dependent Case without selectional relationships will be<br />
seen in section 5.6. See Harley (1995: 4.2) for discussion <strong>of</strong> <strong>the</strong> o<strong>the</strong>r original justification <strong>of</strong> dependent<br />
Case: <strong>the</strong> insensitivity <strong>of</strong> nominative to structural positions, so that it falls to <strong>the</strong> highest<br />
Caseless DP in <strong>the</strong> domain <strong>of</strong> T while <strong>the</strong> next lower one gets accusative (cf. Yip, Maling <strong>and</strong><br />
Jackend<strong>of</strong>f 1987, Maling 1993). It loses its force with <strong>the</strong> replacement <strong>of</strong> spec -head checking by<br />
Agree. Distinct from dependent Case is default Case, accusative in English, nominative in Icel<strong>and</strong>ic,<br />
absolutive in Basque, assigned at realization in configurations independently licensed for<br />
<strong>the</strong> Case Filter (Schütze 2001, Quinn 2005; cf. Bittner <strong>and</strong> Hale's 1996 structural oblique).