Our paper reporting on our initial work on developing a paraphrase ranking model for question interpretation in a virtual patient dialogue system, which we presented at the 10th Workshop on Innovative Use of NLP for Building Educational Applications (BEA10 at NAACL HLT 2015), is now on-line.

The OpenCCG tool ccg2jsgf, an extension developed for Knexus Research Corporation for compiling an OpenCCG grammar into a context-free grammar in the Java Speech Grammar Format used by the Sphinx speech recognizer, has been released open source. See the full README in the download for details.

The survey that Rajakrishnan Rajkumar and I wrote on surface realization and recent connections with psycholinguistics is (finally!) out in corrected form.

Better late than never! Our chapter on evaluating prosody in speech synthesis with eye tracking is finally out in the book Natural Language Generation in Interactive Systems. (With Rajakrishnan Rajkumar, Kiwako Ito and Shari R. Speer.)

Somehow managed to win the NLG Idol contest at INLG-14, presenting in favor of the renewed relevance of Neumann and van Noord’s (1992) paper on ambiguity avoidance (slides).

My paper reporting on initial steps towards inducing broad coverage Combinatory Categorial Grammars for surface realization (and potentially also parsing) from a corpus of dependencies, to appear at INLG-14, is now on-line.

My paper with Manjuan Duan on using parsers to avoid “vicious” (!) ambiguities in the output of a surface realizer, to appear at ACL 2014, is now on-line; click on Papers for the link.

It seems I’ve been elected to a two-year term as Secretary of NAACL, the North American Chapter of the Association for Computational Linguistics. One of my main duties will be to conduct the annual elections for new officers and board members, including my successor. Until then, wish me luck on keeping NAACL running smoothly!

I’m delighted to announce that I’ll have the opportunity to continue my sabbatical work with James Curran and James Constable at U Sydney on “closing the loop” between parsing and generation with NSF funding. The big idea is to use automatically generated disambiguating paraphrases together with crowd-sourcing to gather “silver-standard” data for enhancing parser training. Joining me on the project here are Manjuan Duan and David Howcroft. See my Projects page for more info.

A belated congratulations to Scott Martin on his new job at Nuance!