Differential Possessor Expression: Are Pair-Wise Comparisons Ever Required?

Accounts of clause-level phenomena such as inversion and obviation (Aissen 1997, 1999) make crucial reference to the relative ranking of two arguments, subject and direct object, on a person/animacy/discourse hierarchy. Are such pair-wise comparisons ever required within the NP? O'Connor (1999a,b) proposed an analysis of NP-internal case-marking of possessors in Northern Pomo that paralleled Aissen's analyses of clause-level phenomena. An OT analysis employing harmonic alignment and local conjunction yielded universal constraints on pairings of possessor and possessum, ranked on a hierarchy of NP form (pronoun, proper, common). This paper proposes an alternative analysis of the Northern Pomo data, one that does not require universal constraints on the NP forms of pairs of possessors and possessa. In this reanalysis, I find evidence of another parallel between the clausal and NP levels of structure: absence of case-marking on possessors of kinship nouns in Northern Pomo is explainable as a result of functional uniqueness: kinship stems display an obligatory pronominal prefix that acts as an incorporated pronoun, fulfilling the possessor argument requirement of the kinship stem and blocking the possibility of a case-marked possessor inside the possessive NP. This parallels clause-level subject pronominal incorporation in Chichewa (Bresnan & Mchombo 1986). Finally, pair-wise comparisons are shown to figure in yet another kind of differential possessor expression: the phenomenon of "genitive promotion" as described in Nez Perce (Rude 1986b).