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.
sion is that EA <strong>and</strong> O PH-interaction counts for <strong>the</strong>se syntactic diagnostics, but<br />
only when it counts for <strong>the</strong> morphology: for <strong>the</strong> 2 > 1 > 3 hierarchy in <strong>the</strong> independent<br />
order, but not in <strong>the</strong> conjunct order. In <strong>the</strong> independent order, <strong>the</strong>ir pivot<br />
is <strong>the</strong> argument that controls <strong>the</strong> prefix agreement; in <strong>the</strong> conjunct, where <strong>the</strong>re is<br />
no prefix agreement, it is always <strong>the</strong> EA. For 1↔2 interactions, <strong>the</strong> diagnostics are<br />
ei<strong>the</strong>r inapplicable or untested. Thus (89) establishes that <strong>the</strong> 1/2 > 3 subset <strong>of</strong> <strong>the</strong><br />
2 > 1 > 3 hierarchy governs syntactically visible PH interactions.<br />
(89) a. Ban on inanimates: The pivot cannot be notionally inanimate, even if<br />
grammatically animate like mtig 'tree', mshiimin 'apple', pwaagan 'pipe'.<br />
(Rhodes 1994: 333f.; see also Rhodes 1993: 144).<br />
b. Cross-clausal obviation: Only <strong>the</strong> pivot triggers obviation on <strong>the</strong><br />
subjects <strong>of</strong> adjunct clauses.<br />
(Rhodes 1994: 440f.)<br />
c. Cross-clausal agreement ('raising'): Only <strong>the</strong> pivot <strong>of</strong> a PH-interaction<br />
(<strong>and</strong> only if topic <strong>and</strong> animate) can control agreement on certain upstairs<br />
verbs like gkendank 'to know', as if it were <strong>the</strong>ir O. This argument's PHinteraction<br />
with <strong>the</strong> matrix EA has syntactic repercusions for <strong>the</strong> way O<br />
links to its site in <strong>the</strong> embedded clause.<br />
(Rhodes 1994: 338-40; Bruening 2001: 275-8)<br />
The ban on inanimates is demonstrated in (90). 31<br />
(90) a. *Wgii-bsikwaan mshiimin niw pwaagnan.<br />
The apple (mshiimin) struck <strong>the</strong> pipe-OBV (pwaagn-an).<br />
b. *Wgii-bsikaagoon aw pwaagan mshiimnan.<br />
The apple-obv (mshiimn-an) struck <strong>the</strong> pipe (pwaagan).<br />
c. Ngii-bsikaag aw mshiimin.<br />
The apple (mshiimin) struck me (n-) (independent).<br />
d. Mii-sh *gii-bsikwid / ?*gii-bsikaagyaan aw mshiimin.<br />
[Then] <strong>the</strong> apple (mshiimin) struck me (n-) (conjunct).<br />
(Rhodes 1993: 144f., Ottawa Ojibwa)<br />
The ban requires that <strong>the</strong> one <strong>of</strong> EA, O which is <strong>the</strong> pivot be animate. The<br />
constraint operates over notional ra<strong>the</strong>r than morphological animacy, so that nouns<br />
animate only in <strong>the</strong> morphology like mtig 'tree' <strong>and</strong> mshiimin 'apple' count as<br />
inanimate for it. This indicates that <strong>the</strong> ban is about syntax or interpretation ra<strong>the</strong>r<br />
than morphology. When EA <strong>and</strong> O are both inanimate, <strong>the</strong>re is no way to satisfy<br />
<strong>the</strong> ban. This is so in (90)a <strong>and</strong> (90)b, although <strong>the</strong>ir verbs would be fine for an<br />
EA that is both morphologically <strong>and</strong> interpretively animate. If <strong>the</strong> sentences are<br />
modified to make one <strong>of</strong> <strong>the</strong> arguments animate, <strong>the</strong>y are fine provided it turns out<br />
to be <strong>the</strong> pivot <strong>of</strong> <strong>the</strong> PH-interaction. In <strong>the</strong> independent order <strong>of</strong> (90)c, <strong>the</strong> object<br />
is replaced by 1 st person to give a 3INANEA→1O. Since <strong>the</strong> 1 st person O outranks<br />
<strong>the</strong> inanimate EA on <strong>the</strong> 2 > 1 > 3 hierarchy, it wins <strong>the</strong> PH-interaction. In <strong>the</strong><br />
31 I do not add glosses to Rhodes's translations; <strong>the</strong> discussion below should clarify <strong>the</strong> examples.<br />
63