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.

62<br />

Combinations <strong>of</strong> two 3rd person arguments behave differently. They will not<br />

<strong>the</strong>mselves be <strong>the</strong> focus here, but <strong>the</strong>y provide an important tool. Each 3rd person<br />

argument is inherently lexically ei<strong>the</strong>r animate or inanimate. Each is also ei<strong>the</strong>r<br />

proximate, reserved for <strong>the</strong> most 'prominent' (topical, central) 3 rd person, or obviative.<br />

Combinations <strong>of</strong> two 3 rd person EA <strong>and</strong> O are governed by <strong>the</strong> hierarchies 3<br />

proximate > 3 obviative <strong>and</strong> 3 animate > 3 inanimate. They have no bearing on <strong>the</strong><br />

agreement prefix, but <strong>the</strong>y are relevant for <strong>the</strong> suffixes, <strong>and</strong> unlike <strong>the</strong> 2 > 1 > 3<br />

hierarchy, in <strong>the</strong> conjunct as well as <strong>the</strong> independent order (Rhodes 1994: 432).<br />

Obviation <strong>and</strong> <strong>the</strong> 3↔3 interactions based on it are important here because <strong>the</strong>y<br />

are well established to be syntactic ra<strong>the</strong>r than morphological phenomena. Evidence<br />

for this is rehearsed below, but one striking result <strong>of</strong> Bruening (2001) for<br />

<strong>the</strong> Algonquian language Passamaquoddy may be mentioned immediately:<br />

3.OBVEA→3.PROXO but not 3.PROXEA→3.OBVO combinations allow O to<br />

scope over EA, an unambiguously syntactico-semantic property. This is not evidence<br />

for PH-interactions in syntax because obviation is not known to be a phifeature.<br />

However, clear PH-interactions will turn out to affect obviation, <strong>and</strong> thus<br />

syntax. 29<br />

The literature on Algonquian PH-interactions splits on whe<strong>the</strong>r <strong>the</strong>y are syntactic<br />

or morphological. To <strong>the</strong> former view belong Rhodes (1976, 1994) for<br />

Ojibwa, LeSourd (1976) for Fox, Béjar (2003) for Nishnaabemwin; to <strong>the</strong> latter,<br />

Anderson (1992) <strong>and</strong> Halle <strong>and</strong> Marantz (1993) for Potawatomi. Rhodes (1994)<br />

demonstrates that for some Ojibwa varieties, <strong>the</strong> PH-interactions seen in verbal<br />

morphology have syntactic correlates. Their winner is not only relevant for control<br />

<strong>of</strong> <strong>the</strong> agreement prefix <strong>and</strong> for suffixal morphology, but also <strong>the</strong> pivot <strong>of</strong> unambiguously<br />

syntactic phenomena like control. This evidence is reviewed next. Not all<br />

Ojibwa varieties have all <strong>the</strong> syntactic correlates discussed below, <strong>and</strong> o<strong>the</strong>r Algonquian<br />

languages differ as well. Thus ei<strong>the</strong>r a uniform syntactic system underlies<br />

<strong>the</strong>ir PH-interaction with <strong>the</strong> potential to have or not certain syntactic consequences,<br />

as in Béjar <strong>and</strong> Rezac (2009: 68), or for some languages <strong>the</strong> PHinteractions<br />

occur in syntax while for o<strong>the</strong>rs in morphology (see section 3.5). 30<br />

Among <strong>the</strong> syntactic phenomena <strong>of</strong> Ojibwa, some are not sensitive to PHinteraction.<br />

For instance, <strong>the</strong> shared argument in <strong>the</strong> constructions as stop (preverb)<br />

+ V-ing must be <strong>the</strong> EA, as in English, suggesting <strong>the</strong> same <strong>the</strong>matic structure.<br />

However, <strong>the</strong> phenomena resumed in (89) <strong>and</strong> taken up below refer to <strong>the</strong><br />

winner <strong>of</strong> <strong>the</strong> EA-O PH-interaction. Their pivot is <strong>the</strong> EA in 1/2EA→3O combinations,<br />

<strong>the</strong> O in 3EA→1/2O – in <strong>the</strong> independent order. Rhodes remarkable conclu-<br />

29 The 3.PROX > 3.OBV hierarchy is actually 3 topic > 3 nontopic, which determines <strong>the</strong> former<br />

whenever one <strong>of</strong> <strong>the</strong> arguments is proximate because <strong>the</strong> proximate must be <strong>the</strong> unique topic, but<br />

also encompasses 3.OBV↔3.OBV that may be direct or inverse (Dahlstrom 1986: 53f., Klaiman<br />

1992: 247; cf. Rhodes 1974: 211 note 5, Bruening 2001: 124).<br />

30 Rhodes (1994: 443) notes that some speakers pick <strong>the</strong> EA as <strong>the</strong> pivot <strong>of</strong> cross-clausal obviation<br />

<strong>and</strong>/or cross-clausal agreement, ra<strong>the</strong>r than <strong>the</strong> winner <strong>of</strong> EA-O PH interaction (some picking<br />

EA for one diagnostic but <strong>the</strong> PH winner for <strong>the</strong> o<strong>the</strong>r). Dahlstrom (1986) shows that crossclausal<br />

agreement in Plains Cree is not affected by <strong>the</strong> 3.PROX > 3.OBV hierarchy.

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

Saved successfully!

Ooh no, something went wrong!