The morphological and syntactic facts from Georgian create a unique puzzle for the study of sentence processing. The word order is characterized by considerable freedom and case marking is not uni-directionally associated with θ-roles. This article presents a grammatical account of Georgian case marking and a study on incremental sentences processing. The empirical findings show that case is indeed a more reliable cue than word order in processing clauses with thematically ambiguous arguments. Furthermore, the obtained data suggest an asymmetry between dative experiencers and dative actors, such that only the revision of the thematic properties of the latter is associated with high processing cost.
PRELIMINARIESThat the human sentence processing mechanism makes use of all available cues for grammatical function assignment and structure building in incremental sentence processing is a very natural assumption that has been confirmed in a number of experimental studies. MacWhinney et al. (1984) were among the first showing this by demonstrating that overt case marking, morphological agreement information and syntactic position are used to different degrees in different languages in a sentence interpretation experiment.Early online-studies concerned with the effect of explicit morphological marking used morphological information that is (relatively) unambiguous in nature. For example, Krems (1984) found increased reading times for German sentences beginning with an NP unambiguously marked for the accusative case that almost always marks direct objects only (as compared to sentences beginning with a nominative noun phrase), a result later confirmed by Hemforth (1993) and Fanselow et al. (1999a) Unambiguous case marking need not, however, indicate the grammatical function and/or structural position of the NP bearing it in an unambiguous way. While subjects are marked with the nominative case and indirect (second) objects with