Representational Complexity and Memory Retrieval in Language Comprehension a Dissertation Submitted to the Department of Linguistics and the Committee on Graduate Studies of Stanford University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

How an object or event is described depends upon numerous factors that have been identified by prior research in linguistics, pragmatics, and psychology. What this prior research has not established, however, is what consequences the resulting choice of linguistic form has on subsequent comprehension processes. This dissertation begins with the assertion that the quantity of information encoded in a linguistic expression has an impact on the efficiency of subsequent retrieval processes that target the corresponding mental representation. In particular, I hypothesize that information processed during the encoding phase of comprehension facilitates the retrieval process. Consequently, more informative linguistic expressions are predicted to enhance memory retrieval. A series of self-paced reading tasks show these predictions to be true in a variety of filler-gap constructions with a variety of phrase-types in English, e.g. argument wh-phrases, temporal adjunct phrases, indefinites, and definites. In all these experiments, reading times are faster when the memory representation associated with a more informative filler-phrase is targeted. Crucially, these “informativity effects” consistently appear at likely retrieval sites. Moreover, in nearly all of the experiments, there is no effect of informativity in the word regions prior to the retrieval site. This fact argues in favor of treating the informativity effects as intrinsically linked to memory retrieval. The underlying relationship between informativity and retrieval is also explored in depth within this dissertation. That is, I address the question of why greater syntactic and semantic complexity facilitates retrieval. There are multiple, plausible reasons for why informativity attenuates retrievability, including encoding time differences and differences in plausibility. In the end, I conclude that this relationship reflects what happens cognitively during the encoding process and the retrieval process. Specifically, the process of building a more complex syntactic representation requires repeatedly accessing that representation