Memory Retrieval in Dependency Resolution Beyond the Syntactic Domain
Language is structured in a way that requires the establishment of linguistic dependency relations. This involves the storage and retrieval of linguistic information in memory during language processing. Linguistic rules determine which dependency relations are possible and which are not. Studying the real-time deployment of these rules has provided important insights into the retrieval mechanism underlying language processing, which is generally assumed to be a cue-based, content-addressable memory mechanism.
The majority of existing research focused on the deployment of syntactic rules in intra-sentential dependency relations. However, dependencies are not limited to the syntactic domain: relations in discourse are essential for language production and comprehension as well. This dissertation focuses on the resolution of linguistic dependencies that span across multiple sentences, and the impact of structural constraints at discourse level. In a series of experiments on the resolution of cross-sentential anaphoric pronouns and additive presuppositions, this dissertation shows that the retrieval of information from memory is not restricted to the syntactic level but also exists at a much larger scale: discourse. The relevance of structural constraints at discourse level challenges the architecture of existing theories of language processing, and this dissertation advocates a general cue-based memory retrieval mechanism, by assuming that not the resolution process itself, but the relevant constraints can be domain-specific.
In sum, this dissertation provides an important window into the memory retrieval process underlying the resolution of dependencies in discourse, and in this way contributes to a more complete picture of language processing.
