Coreference happens when different expressions refer to the same entity, as in “Thomas said he shaved himself”, where he may or may not refer to Thomas, while himself co-refers with he. We will examine Cantonese’s unique coreference behaviour, an under-researched area. Cantonese is an under-resourced language and considered potentially vulnerable by UNESCO, rendering our project particularly timely. Using linguistic diagnostics, corpus- data analysis, and acceptability-judgement experiments, we will create open-access datasets. Theoretical modelling will explore sentence-structure, meaning, and context interactions, while computational tools test these models and build language resources. We will report our findings in journal articles and conference presentations.
The Leverhulme Trust
Faculty of Linguistics, Philology, and Phonetics, University of Oxford