Item

Abstract

One of the most basic assumptions regarding language comprehension is that it proceeds incrementally, i.e. by seeking to maximise the degree of interpretation computed with each new word that is encountered. This perspective has often been though to entail that the thematic interpretation of an argument may be derived from its grammatical function, for example, via a preference-strategy associating a subject with a Causer. In the present paper, we argue that these conclusions are not supported by experimental findings. Rather, as we show on the basis of a number of studies using event-related brain potentials (ERPs), the degree of meaning derived from a given sentence fragment (form) during online sentence comprehension differs as a function of the morphological informativeness of the sentential arguments. Finally, we present a new model of sentence comprehension, the Argument Dependency Model (ADM), which is designed to capture these differences in the way that meaning is computed incrementally. Specifically, the ADM assumes that argument processing proceeds via two alternative processing pathways, one of which is syntactic, and the other of which is thematic (interpretive) in nature. Whereas both pathways seek to establish hierarchical relations between arguments, only the thematic pathway allows for incremental interpretation to be maximised. Which of the two pathways is chosen during the comprehension of a given sentence crucially depends on the morphological informativeness (i.e. presence or absence of unambiguous morphological case marking) of the arguments. Consequences for language architecture arising from this model will be discussed.