My research focuses on computational models that explain how infants learn their native language. This involves analyzing spoken language, doing computer simulations on speech data, and testing predictions with human learners. Below is a brief description of the projects I am involved in:
The effect of motherese on infant language development: A computational and experimental perspective - University of Pennsylvania (NWO Rubicon grant)
I currently work with Dan Swingley at the Penn Infant Language Center. Together we are discovering how 'motherese' (speech addressed to infants) helps infants to learn some of the essential properties of their native language. We use a combination of computational modeling and behavioral experiments. The project is funded by the Netherlands Organisation for Scientific Research (NWO).
The induction of phonotactics for speech segmentation - Utrecht Institute of Linguistics OTS, Utrecht University
My PhD dissertation dealt with the computational modeling of infants' learning of phonotactics, and the use of phonotactics for the detection of word boundaries in continuous speech. The work was part of René Kager's NWO Vici project Phonotactic Constraints for Speech Segmentation: The Case of Second Language Acquisition (which also involved Natalie Boll-Avetisyan, Tom Lentz and Diana Apoussidou). Some of the work is ongoing.
Faceted search and data-centric information retrieval - Institute for Logic, Language and Computation, University of Amsterdam
I have been working in Information Retrieval, specifically on faceted search - the automatic generation of facets and facet values from structured data sets (such as the IMDb movie database) with the purpose of optimally guiding searchers toward relevant information. This work was in collaboration with Jaap Kamps and Marijn Koolen (as part of the NWO project EfFoRT - Effective Focused Retrieval Techniques). The results of our participation in the INEX 2011 Data Centric Track are described here.