My PhD research aims to explore the benefits of Deep Learning models in theoretical cognitive linguistics. In particular, I am using Convolutional Neural Networks to model the development of periphrastic do in Early Modern English by charting its relation to the modal auxiliaries will, shall, may, can and must. Rooted in the framework of connectionism, I seek to explore how deep learning models can be of use to 1) retrieve data and clean up large (noisy) corpora in a semi-automatic fashion, 2) model language use in a profoundly dynamic and flexible way, and 3) gain insights into how humans acquire and process language and thereby induce change in the language as a whole.
I am interested in a large variety of scientific disciplines, ranging from (historical) linguistics over cognitive science to computer science and artificial intelligence. My background in the humanities and computer science has provided me with the necessary skills to both understand and implement deep learning models myself, as well as reflect in a nuanced way on their strengths and weaknesses, assess their implications (and societal impact) and report my results to audiences with or without a background in AI.
In 2014 I obtained an MA in Linguistics from the University of Leuven, with a specialization in English historical linguistics. The year after I completed the advanced MA of AI (Speech and Language Technology) at the same university. As part of this MA, I worked as an intern for Nuance Communications (Aachen, Germany), building a language model for compounds in German and Dutch speech recognition. From August to November 2015, I developed a website and search tool for the Uitleenwoordenbank at the Meertens Insitute (Amsterdam, The Netherlands).
Statute & functions
Bijzonder academisch personeel
- postdoctoral researcher