26.11.2012 Views

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

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

179<br />

These observations do not mean that phi-agreement, Case, <strong>and</strong> A-movement<br />

cannot affect LF or PF, only that <strong>the</strong>y do not reduce to <strong>the</strong>ir requirements. They<br />

can clearly be displayed at PF, <strong>and</strong> A-movement at least creates new scopal <strong>and</strong><br />

binding positions at LF, as for each o<strong>the</strong>r in (272)a <strong>and</strong> (273)a. Yet binding is not<br />

<strong>the</strong> reason for A-movement, which occurs independently <strong>of</strong> it (Chomsky 2001:<br />

32). Without A-movement, English phi-agreement <strong>and</strong> structural Case even seem<br />

to lack all detectable interpretive correlates, such as <strong>the</strong> capacity to create anaphora<br />

binding configurations in (272)b <strong>and</strong> (273)b (discussed in chapter 1).<br />

(272) a. Some linguists-NOMi seem.PLi to <strong>the</strong>mj / each o<strong>the</strong>ri [ti to have been<br />

given good job <strong>of</strong>fers].<br />

b. There seem.PLi to <strong>the</strong>mj / *each o<strong>the</strong>ri) [to have been some linguists-<br />

NOM given good job <strong>of</strong>fers].<br />

(273) a. The DA proved [two men-ACCi to have been at <strong>the</strong> scene] during<br />

<strong>the</strong>irj / each o<strong>the</strong>ri's trials.<br />

b. The DA proved [<strong>the</strong>re to have been two men-ACCi at <strong>the</strong> scene] during<br />

<strong>the</strong>irj / *each o<strong>the</strong>ri's trials.<br />

(Lasnik 1999: 138, 18, adapted)<br />

<strong>Phi</strong>-agreement, Case <strong>and</strong> A-movement thus seem to have no motivation in PF<br />

or LF requirements. Chomsky (e.g. 2000a: 113, 118f.) concludes that <strong>the</strong>y are<br />

driven by purely syntactic <strong>features</strong> <strong>of</strong> lexical items, <strong>features</strong> that are illegible to<br />

both <strong>the</strong> interfacing systems, PF <strong>and</strong> LF, in violation <strong>of</strong> <strong>the</strong> Interpretability Condition.<br />

Illegibility to PF/LF may be spoken <strong>of</strong> as uninterpretability to <strong>the</strong>m, a term<br />

particularly suited for illegibility to LF. Syntactic <strong>features</strong> that are illegible to an<br />

interfacing system must be eliminated prior to it to meet Full Interpretation. To<br />

this need is attributed <strong>the</strong> core mechanism <strong>of</strong> syntax beyond pure Merge: <strong>the</strong> operation<br />

Agree, which relates an uninterpretable feature to an interpretable one, <strong>and</strong><br />

removes it. Illegible <strong>features</strong> <strong>and</strong> <strong>the</strong>ir Agree relationships correspond to syntactic<br />

licensing in o<strong>the</strong>r frameworks, but <strong>the</strong>ir grounds are sought in <strong>the</strong> Full Interpretation<br />

requirement <strong>of</strong> interfacing systems. Syntactic construction proceeds so long<br />

as <strong>the</strong>re is material to assemble, but its output may not be legible.<br />

In <strong>the</strong> implementation <strong>of</strong> <strong>the</strong>se ideas, uninterpretable <strong>features</strong> are construed as<br />

feature types without values, for instance [number:] on T. They are present on<br />

lexical items as <strong>the</strong>y enter syntax, <strong>and</strong> constitute a locus <strong>of</strong> its parametrization.<br />

They are legible nei<strong>the</strong>r to LF nor PF, for <strong>the</strong>re is nei<strong>the</strong>r realization nor interpretation<br />

for valueless [number:]. The syntactic operation Agree provides a value by<br />

relating an uninterpretable feature to an interpretable one on ano<strong>the</strong>r element, such<br />

as <strong>the</strong> interpretable [number:plural] on a noun. The newly-valued [number:plural]<br />

is legible to PF. LF however still does not find it interpretable on T. To satisfy Full<br />

Interpretation, <strong>the</strong> valuation <strong>of</strong> a feature is accompanied by its deletion. Deletion<br />

occurs upon <strong>the</strong> Transfer (Spell-out) to LF. Thus Agree-cum-valuation has as consequence<br />

Transfer <strong>of</strong> <strong>the</strong> object containing <strong>the</strong> just-valued feature. If at Transfer a<br />

syntactic object meets Full Interpretation, it converges; o<strong>the</strong>rwise it crashes.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!