Despite the impressive advances made possible by neural networks, current NLP systems are still far from displaying the learning abilities of humans in many languages. This project aims to improve language modeling for low-resource morphologically rich languages, taking inspiration from child language acquisition insights.
Among other methodologies, an artificial language learning paradigm will be used to simulate the learning of typologically diverse languages and evaluate the effect of known properties of child-directed language on the acquisition of morphology and other language aspects.
Other possible research directions include: the design of better input segmentation methods; language acquisition inspired curriculum learning; and leveraging existing language resources (like dictionaries or morphological analyzers) to boost the learning process in very low-resource settings.
This PhD position offers a unique opportunity to acquire valuable research experience in an international environment: You will be part of the
Computational Linguistics group (@
GroNLP), which is part of the Centre for Language and Cognition of the University of Groningen (CLCG).
Main requirement: A Master’s degree in computational linguistics, artificial intelligence, computer science, information science, or related area.