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.
194<br />
gative languages because <strong>the</strong>ir C/T system has nominal or prepositional properties<br />
(cf. Johns 1992, Alexiadou 2001, Mahajan 1996 for such connections).<br />
(295) Obligatory Case Parameter: In <strong>the</strong> lexicon, one <strong>of</strong> v (absolutive) or T<br />
(nominative) has a phi-probe.<br />
Active loci are phases, since phasehood follows upon Agree. Thus in unaccusatives,<br />
vABS Agree leads to Transfer <strong>of</strong> <strong>the</strong> vP phase in ergative languages, TERG<br />
Agree <strong>of</strong> <strong>the</strong> TP phase in accusative ones. These are approximations, ins<strong>of</strong>ar as<br />
cyclicity may permit building up structure between <strong>the</strong> Agree <strong>of</strong> a head <strong>and</strong> its<br />
consequent Transfer, for instance <strong>the</strong> CP above an Agreeing TP in Chomsky<br />
(2001, 2008). For convenience, I will speak <strong>of</strong> <strong>the</strong> v-phase <strong>and</strong> <strong>of</strong> <strong>the</strong> C/T-phase as<br />
<strong>the</strong> phases due to <strong>the</strong> Agree <strong>of</strong> v <strong>and</strong> T. Potential Agree/Case loci are not phases as<br />
such, but o<strong>the</strong>r <strong>features</strong> make <strong>the</strong>m so, such as <strong>the</strong> EPP/Edge Feature <strong>of</strong> Chomsky<br />
(2008). This is st<strong>and</strong>ardly assumed at least for C/T.<br />
By <strong>the</strong> Obligatory Case Parameter, <strong>the</strong> lexicon provides one active locus per<br />
CP, v or T. This is adequate for unaccusatives. It is also all that unaccusatives can<br />
tolerate, if a probe must find a goal, since <strong>the</strong>y have only one DP. For transitives,<br />
ℜ (293) steps in to add a phi-probe to a potential Agree/Case locus in <strong>the</strong> derivation,<br />
activating it. This is <strong>the</strong> source <strong>of</strong> dependent Case. In an ergative system, <strong>the</strong><br />
derivation proceeds as in (296), diagrammed in (297):<br />
(296) ℜ in an ergative system:<br />
1. vABS-Agree with O <strong>and</strong> S, valuing [phi:] on v <strong>and</strong> [Case:] on <strong>the</strong>m.<br />
2. Transfer upon v-Agree <strong>of</strong> <strong>the</strong> complement <strong>of</strong> v, with convergence.<br />
3. The next phase is built up, including <strong>the</strong> EA at <strong>the</strong> vP edge.<br />
4. The potential Agree/Case locus T has no phi-probe, thus no T-EA Agree.<br />
5. Transfer <strong>of</strong> <strong>the</strong> C/T-phase, which crashes because <strong>of</strong> [Case:] on EA.<br />
6. ℜ adds [phi:] on <strong>the</strong> potential locus T <strong>of</strong> <strong>the</strong> failed numeration.<br />
7. The construction <strong>of</strong> <strong>the</strong> phase restarts from <strong>the</strong> new numeration.<br />
8. Agree between T <strong>and</strong> EA, valuation <strong>of</strong> <strong>the</strong>ir [phi:] <strong>and</strong> [Case:].<br />
9. Transfer <strong>of</strong> <strong>the</strong> C/T-phase, with convergence.<br />
(297) Derivation <strong>of</strong> dependent ergative<br />
a. v[φ:,κ:A] [… DP[φ:1PL,κ:]]] →Agree<br />
b. T[κ:E] [DP[φ:2PL,κ:] v[φ:1PL,κ:A] [… DP[φ:1PL,κ:A]]] →√Transfer,…<br />
c. T[κ:E] [DP[φ:2PL,κ:] v[φ:1PL,κ:A] [… DP[φ:1PL,κ:A]]] →*Transfer, ℜ<br />
d. T[φ:2PL,κ:E] [DP[φ:2PL,κ:E] v[φ:1PL,κ:A] [… DP[φ:1PL,κ:A]]] →√Transfer<br />
(Notation: κ [Case], φ [phi], strikethrough Transfer)<br />
It has not been specified how ℜ knows to add a phi-probe, nor to add it on <strong>the</strong><br />
potential Agree/Case locus T. It seems unnecessary. A potential Agree/Case locus<br />
is such by having an interpretable [Case] feature. If <strong>the</strong> phi-probe is added elsewhere,<br />
<strong>the</strong> [Case:] <strong>of</strong> a DP cannot be valued. How ℜ knows to choose a phi-probe