In 2005 I became Assistant Professor of Language and Cognition at the Artificial Intelligence department. Language research within artificial intelligence is similar to work done within computational linguistics and within linguistics proper, but it also differs in the scope of the research. For example, like computational linguists, I use computational tools to search for patterns in large corpora with an aim to extract data that could be used in natural language applications. Currently I'm using this very method to look at antonymy and contrast. But the work doesn't stop there. I am also interested in how speakers react to extracted data, and what cognitive processes are involved, how antonymy, canonical and non-canonical, are recognized, and what their categorization by humans tells us about the nature of their representation in the mind. Similarly, I have done work on the use of the Dutch reflexives zich and zichzelf, looking at very large corpora to try to predict the distribution of the two reflexive forms for verbs that can take both as a reflexive argument. This work was initially inspired by a need to choose comparable reflexive verbs for a child language experiment about pronouns. This is therefore another example where in my work I draw on methods from both computational and general linguistics, to answer a diverse set of questions.
In 2006 I started working on a VENI-grant project awarded by NWO, the Netherlands Organization for Scientific Research. My project is entitled "The contribution of contrast in context". You can read more here. PhD student Anna Lobanova is also working on topics closely related to this project. Currently we are looking at what evidence exists for making distinctions between contrast relations, and we recently presented a paper on this topic at IwCS 8. We are also working on automatic antonym extraction and antonym evaluation. The goal of this work is to use the automatically extracted antonyms to try to automatically identify (unmarked) contrastive relations, and to account for the contrastive interpretation.
My interest in contrast developed from my fascination with inference in discourse and dialogue. Most of my work is within the fields of computational semantics and pragmatics, and as a methodology I prefer to use corpus data to test formal theories. In addition to contrast I do work on presupposition theory (the topic of my dissertation), conversational implicatures, and ellipsis, in particular VP ellipsis.
Additionally I am very interested in optimality theory (OT) approaches to semantics and pragmatics. I organized the Stockholm workshop in "Variations in Optimality Theory". I am also very interested in first language acquisition, and have been working on explaining the delay of principle B effect in children's acquisition of pronouns, together with Petra Hendriks, Erik-Jan Smits, and more recently, Jacolien van Rij. Our explanation of the delay is embedded in Bi-directional OT.
My interest in Ellipsis is still strong, and together with Johan Bos I annotated the Wall Street Journal Corpus for VP-ellipsis. A paper detailing this work will appear in Language Resources and Evaluation and the stand-off annotations are available here. We are also planning a shared task on VP Ellipsis for the next SEMEVAL 2010.
Petra Hendriks I had earlier organized the ESSLLI Workshop on Cross-modular Approaches to Ellipsis.
I also organized two workshops in Stockholm, one on "Contrast, Information Structure and Intonation" which was held in Stockholm October 28-30, 2005, and one on "Variation within Optimality Theory", from which there is a very nice proceedings.