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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

not cross this distance, just as in English -er affixation. In Irish however, subjects<br />

raise in <strong>the</strong> syntax to a position immediately c-comm<strong>and</strong>ed by <strong>the</strong> finite verb<br />

(Chung <strong>and</strong> McCloskey 1987: 227). Morphology <strong>the</strong>n realizes <strong>the</strong> subject as affix<br />

or strong pronoun under adjacency (Ackema <strong>and</strong> Neeleman 2003).<br />

The grouping in <strong>the</strong> Irish syn<strong>the</strong>tic-analytic alternation <strong>of</strong> syntactico-semantic<br />

inertness, sensitivity to arbitrary gaps, <strong>and</strong> limitation to small domains that may<br />

cross syntactic isl<strong>and</strong>s, is once more <strong>the</strong> full signature <strong>of</strong> morphology in a strong<br />

modular architecture. Beside such purely morphological alternations, o<strong>the</strong>rs exist<br />

that are purely syntactic: potentially visible to syntax <strong>and</strong> interpretation, blind to<br />

purely morphophonological information <strong>and</strong> <strong>of</strong>ten referring to a fuller panoply <strong>of</strong><br />

syntactic information that morphology does, with <strong>the</strong> potential to span phrasestructurally<br />

large domains. The French alternation <strong>of</strong> unfocussed clitics <strong>and</strong> focussed<br />

strong pronouns can be thought <strong>of</strong> in this way. A subtler French clitic-strong<br />

pronoun alternation is <strong>the</strong> subject <strong>of</strong> chapter 4. It is <strong>the</strong> repair <strong>of</strong> <strong>the</strong> Person Case<br />

Constraint that st<strong>and</strong>s in minimal contrast to both morphological opaque cliticization<br />

<strong>and</strong> syn<strong>the</strong>tic-analytic alternations on all aspects <strong>of</strong> <strong>the</strong> modular signature. 25<br />

25 Syntactic models <strong>of</strong> morphophonologically-governed syn<strong>the</strong>tic-analytic alternations <strong>of</strong>ten ensure<br />

a weaker modular signature by restricting <strong>the</strong> syntactically visible aspects or domains <strong>of</strong> realization:<br />

see Poser (1992), Embick (2000), Chomsky (2001). Richer models with less restricted<br />

paradigms <strong>of</strong> syn<strong>the</strong>tic <strong>and</strong> analytic constructions are Andrews (1990), Börjars et al. (2005),<br />

Sadler <strong>and</strong> Spencer (2000). Fully amodular interaction <strong>of</strong> syntactic, interpretive, <strong>and</strong> realizational<br />

constraints is proposed by Kiparsky (2005).<br />

The locality prediction <strong>of</strong> strong modularity would be nuanced if an analytic construction first<br />

spelled out according to morphophonological conditions <strong>and</strong> <strong>the</strong>n could be dissociated by syntactic<br />

movement <strong>of</strong> a part <strong>of</strong> its PF-content, as sometimes proposed for 'stylistic' rules or inert<br />

scrambling (Chomsky 2001). Consider <strong>the</strong> Latin perfect active. It alternates between <strong>the</strong> syn<strong>the</strong>tic<br />

for most verbs, laudā-v-ī 'praised-PERF-1SG = I have praised', <strong>and</strong> <strong>the</strong> analytic for an arbitrary<br />

set <strong>of</strong> 'deponent' verbs that are passive in perfect formation but active in syntax secūtus<br />

sum, 'followed I.am = I have followed (transitive)'. Embick (2000) entertains <strong>the</strong> analysis where<br />

<strong>the</strong> arbitrary property P <strong>of</strong> deponent roots blocks postsyntactic amalgamation <strong>of</strong> V with T+AGR.<br />

V is <strong>the</strong>n spelled out as <strong>the</strong> participle <strong>and</strong> may move away from T+AGR by semantically inert<br />

syntactic movement. He favors an alternative with early insertion <strong>of</strong> P into <strong>the</strong> syntax where it<br />

block syntactic head raising <strong>of</strong> VP to T, allowing <strong>the</strong> unraised VP to move in syntax prior to spellout,<br />

<strong>and</strong> <strong>the</strong>n be spelled out as <strong>the</strong> 'passive' participle separate from T+AGR. The third alternative,<br />

which does not weaken modularity, would allow any V to move away from T prior to spellout,<br />

amalgamate with T at spell-out if adjacent, <strong>and</strong> assume <strong>the</strong> 'passive' participle form o<strong>the</strong>rwise.<br />

Embicks rejects this because it would let any verb have <strong>the</strong> deponent analytic formation if<br />

moved away from T+AGR, not only those with P (*laudātus … sum 'I have praised (active)' beside<br />

laudāvī, p. 203). However, for ordinary transitives, <strong>the</strong> moved V would in fact not have any<br />

form to assume, since Latin lacks perfect active participles. The content <strong>of</strong> P is precisely to allow<br />

<strong>the</strong> perfect active vP <strong>of</strong> VP to spell out as perfect passive participle.<br />

Kiparsky (2005) describes clearer evidence for Embick's problem. The Sanskrit analytic perfect<br />

mostly occurs only when <strong>the</strong> syn<strong>the</strong>tic perfect is unavailable: (i) regularly when a morphological<br />

affix prevents <strong>the</strong> root from reduplicating; (ii) sometimes when a phonological reason does so;<br />

(iii) for a h<strong>and</strong>ful <strong>of</strong> roots where syn<strong>the</strong>tic <strong>and</strong> analytic perfects coexist. The two parts <strong>of</strong> <strong>the</strong> analytic<br />

perfect may occur widely-separated: prabhraṅcayām yo naguṣām cakāra 'who (yo) made<br />

(cakāra) Nagusha (naguṣām) fall headlong (prabhraṅcayām)' (Whitney 1896: §1071-2). This<br />

does suggest that <strong>the</strong> result <strong>of</strong> an analytic formation sensitive to morphophonology can subse-<br />

53

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

Saved successfully!

Ooh no, something went wrong!