ABSTRACT
Early Immediate Constituents (EIC), Discourse Locality Theory (DLT), and Retrieval Interference Theory (RIT) make clear predictions about processing difficulties associated with definite versus indefinite direct object NPs in Hindi (see pages 40, 42, 52). EIC predicts that definiteness marking should have no effect on processing difficulty, since pragmatics plays no role in processing per se. DLT predicts that indefinites will be easier to process than definites, because indefinites introduce a new discourse referent whereas definites presuppose an old/familiar one; DLT thus predicts an increase in storage cost but no change in integration cost. RIT predicts that there will be greater processing difficulty in retrieving a dative marked NP at a verb when a definite-marked direct object NP (as opposed to a bare NP) intervenes between the dative NP and the verb. This is due to increased similarity (morphophonemic identity) of such NPs with superficially similar (more precisely, morphophonemically identical) case marking.
