Colloquium led by Léa Nash

October 20
3:30-5:00 pm
Pick 016
Université Paris 8 CNRS
Person Split and Pronominal Asymmetries

Many ergative languages exhibit splits based on the nominal type of main arguments. The most typical manifestation of this type of split, known as person split, involves marking 3rd person arguments (3PA) according to the ergative case-schema, while 1st and 2nd person arguments (1/2PA) follow nominative case-alignment (nominative case on all subjects). 

A number of questions arise with respect to the phenomenon:

(i) Is person split motivated by structural differences of clauses with 1/2PA and 3PA?  

(ii) Is person split a more superficial phenomenon that has no structural source? Is the ergative morphological marking incompatible with certain feature specifications of 1/2PA?

(iii) Can person split receive a unitary explanation in languages where it is manifested?

Addressing these questions in my investigation of person split in Georgian, I put forth an analysis for the mixed behaviour of 1/2PA which manifest both ergative and nominative properties when they function as transitive subjects in ergative environments. A close investigation of three types of 1/2PA—(i) 1/2P pronouns, (ii) constructions with 1/2P pronominal determiners of type we linguists (Postal 1969) dubbed 1/2Pron-N, and (iii) superficially standard nominals triggering person agreement on the verb, 1/2P-N — leads to the conclusion that person split in Georgian is best accounted in structural terms. It is conditioned by structural differences of nominal constituents rather than by structural asymmetries of clauses containing them. 1/2PA and 3PA are not structured alike: only 1/2PA are endowed with person features bundled in the category PARTICIPANT and projected in the specifier positions of their maximal projections. The two key properties – absence of case-marking on only 1/2PRON and uniform person-agreement on the verb triggered by the three types of 1/2PA — are directly correlated with the presence of PARTICIPANT. 

The most important conclusion of the talk is that the division of argument types into expressions referring to speech act participants and expressions that refer to any other type of event participants is not a sufficient condition to account for person-split. What determines the presence of case on the argument is not its semantic-pragmatic type or its discourse role but rather its structure: if PARTICIPANT and the nominal layer co-occur in the structure of a (pro)nominal argument, it will be marked for case.