Linguistics PhD candidates Marcus Berger and Alan Ke were awarded Rackham Predoctoral Fellowships for 2018-19.

The Rackham Predoctoral Fellowship supports outstanding doctoral students who have achieved candidacy and are actively working on dissertation research and writing. The Fellowship includes a graduate stipend and candidacy tuition for twelve months and is awarded to students who are expected to complete an outstanding dissertation in the year in which they hold the fellowship.

Marcus Berger was awarded a predoctoral fellowship in the Humanities graduate division, and Alan Ke was awarded in the Social Sciences graduate division.

Berger’s dissertation, The Syntax of Co-Reference in Bora, contributes to current linguistic theory by analyzing original fieldwork data collected from the understudied, endangered Amazonian language, Bora. The Bora language has around 2,500 speakers remaining, and is at a critical point for documentation. Based on original fieldwork, Berger’s work will contribute to linguistic theory by showing that the subjects of subordinate clauses in Bora are typologically unlike many widely spoken languages. Using this data, he points out ways in which current theory cannot account for Bora subordinate clauses, and suggest ways the theory can be revised to account for this data. The results of his fieldwork also include an annotated collection of primary Bora data, including stories and cultural traditions for use by researchers and future Bora generations.

Ke’s dissertation, Feature Retrieval in the Processing of Grammatical Illusions, addresses the cue-based retrieval model, which faces theoretical and empirical questions because, on the one hand, it is not able to encode long-distance syntactic relations such as c-command; on the other hand, it fails to predict the processing differences in grammatical illusions between subject-verb agreement and reflexive binding. To overcome these shortcomings, Ke’s dissertation develops a new computational model by modifying two core components of the cue-based retrieval model. First, he proposes an algorithm that encodes c-command and locality as features on the c-commandees, which can then serve as retrieval cues for memory retrieval. Furthermore, he proposes a theory of cue-based retrieval based on an independently motivated Minimalist (feature interpretability) distinction. He then constructs three distinct implementations examining the overall predictions using computational modeling. Finally, the validity of these three theories is experimentally tested in an eye-tracking and an ERP experiment.