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.
may also lead to <strong>the</strong> (re-)Merge <strong>of</strong> <strong>the</strong> goal at <strong>the</strong> position <strong>of</strong> <strong>the</strong> probe, giving<br />
movement, but <strong>the</strong> details <strong>of</strong> how <strong>and</strong> when this occurs are not relevant here. 131<br />
181<br />
(276) Matching is a relation that holds <strong>of</strong> a probe P <strong>and</strong> a goal G. Not every<br />
matching pair induces Agree. To do so, G must (at least) be in <strong>the</strong> domain<br />
D(P) <strong>of</strong> P <strong>and</strong> satisfy locality conditions. The simplest assumptions<br />
for <strong>the</strong> probe-goal system are[:]<br />
a. Matching is feature identity [<strong>of</strong> type, not value, p. 124].<br />
b. D(P) is <strong>the</strong> sister <strong>of</strong> P.<br />
c. Locality reduces to "closest c-comm<strong>and</strong>."<br />
(Chomsky 2000a: 122)<br />
In this sketch, <strong>the</strong>re are many options <strong>and</strong> alternatives explored in related<br />
work. Those that concern <strong>the</strong> Agree/Case approach to <strong>the</strong> PCC are discussed in <strong>the</strong><br />
literature cited in section 5.2. The PCC is <strong>the</strong> failure <strong>of</strong> a [+person] DP to be licensed<br />
if it fails to Agree for [+person], due to an intervening dative. Let us adopt<br />
<strong>the</strong> following hypo<strong>the</strong>ses, diagrammed in (277):<br />
– Only [+person] DPs have [person] <strong>and</strong> [number], o<strong>the</strong>rs only [number].<br />
– [person]-bearing DPs need Agree for [person] to be Case-licensed.<br />
– Datives that create <strong>the</strong> PCC block Agree for [person] by having some [person]<br />
but not [number] specification visible to Agree; call this specification<br />
"+". It is left open whe<strong>the</strong>r datives need [Case].<br />
(277) {[π:], [#:], [κ:ACC]}v … {[π:+], ([κ:])}DAT … {([π:1]), [#:PL], [κ:]}DP<br />
(Notation: π person, number, κ Case)<br />
The failure <strong>of</strong> [person] Agree with a DP must lead to failure <strong>of</strong> Case assignment<br />
to it. It might be easiest to suppose that each <strong>of</strong> [person] <strong>and</strong> [number] on a<br />
DP comes with its own unvalued [Case]. Agree for [number] <strong>the</strong>n values <strong>the</strong><br />
[Case] associated with [number], but not with [person]. Better explored for <strong>the</strong><br />
PCC is Chomsky's (2000a, 2001) proposal (278) (Anagnostopoulou 2003: 274,<br />
Béjar <strong>and</strong> Rezac 2009: 46f.). 132<br />
131 <strong>Phi</strong>/Case valuation may go toge<strong>the</strong>r intrinsically (Frampton <strong>and</strong> Gutmann 1999), or because<br />
Agree maximizes valuations as soon as it relates two lexical items (Rezac 2004b). The [Case:]<br />
feature <strong>of</strong> DPs might be able to probe as well, depending on how <strong>the</strong> relationship <strong>of</strong> phi <strong>and</strong> Case<br />
is understood. However, [Case:] might never be in a position where it could find a goal. If a head<br />
can probe only within its maximal projection, nei<strong>the</strong>r N nor D can ever see into <strong>the</strong> clause, save<br />
for a bare D, cf. Chomsky (2000a: 125, 2001: 16). N might probe its CP/IP complement, but this<br />
fails if CPs are phases <strong>and</strong> Ns do not take IP (raising, ECM) complements because <strong>the</strong>y cannot<br />
assign Case, *John's appearance to leave (Chomsky 1986b: 3.5.2.5; cf. Abney 1987, Sichel<br />
2007). On a proper underst<strong>and</strong>ing <strong>of</strong> <strong>the</strong> phi-Case link depends <strong>the</strong> possibility that o<strong>the</strong>r probes<br />
than phi establish <strong>the</strong> Agree needed for Case assignment (Rezac 2004b).<br />
132 To underst<strong>and</strong> matters better, partial agreement with case-agreement mismatches seem a<br />
promising domain (Rezac 2003, 2004a: chapter 5). If datives that intervene in <strong>the</strong> PCC need Case