Research
Research  My research is based on the belief that language is fundamentally an artefact of human cognitive processes, and is most accurately characterised by dynamic theories of these processes as instantiated by computational models. I have conducted work on incremental, principle-based parsing mechanisms, exploring the issues modularity, distributed processing, and cross-linguistic processing universals.

In collaboration with Patrick Sturt, I have also done work on monotonic parsing and reanalysis for English and Japanese. My current research focusses upon the role of statistical mechanisms in human sentence processing. There is substantial evidence that people exploit frequency-bases biases in resolving local ambiguity. I have collaborated with Steffan Corley in developing a model of human lexical category disambiguation, based on POS tagging theories. I am also interested in using probabilistic frameworks to reason about the optimal behaviour of the human sentence parser, and am currently developing the Informativity Model in collaboration with Martin Pickering and Nick Chater. In related work, I am working on a critique of maximal likelihood characterisations of human syntactic disambiguation.  

I am also interested in statistical techniques for applied language procesing problems, and am collaborating with Paola Merlo on the development of backed-off estimation techniques for multiple PP attachment disambiguation.  

Activities Beyond my local research, teaching and supervising duties, I am also a co-founder of the AMLaP conference, and the recently established AMLaP committee.