Part-time Postdoc in Computational Modeling of Linguistic Variation

Research in computational linguistics at the RuG focuses on the computational analysis of language to further linguistic insight. We expect interested postdocs and graduate students to pursue research projects in close collaboration with ongoing research activities.

A good deal of work is focused on the parsing of Dutch. Relevant url's:

Question Answering for Dutch using Dependency Relations
Algorithms for Linguistic Processing

The important topic for this position is the application of distance measurements for modeling various phenomena in dialectology and language contact. For instance, phonetic and lexical distance measurements are used to model intelligibility between different but related languages. In dialectology, it has been shown that string edit distance can be a valuable tool for the modeling of gradual changes found between neighboring dialects. Additional factors such as social contact and geography have also been assessed. Relevant url's:

Informatiekunde: Dialectology
Wilbert Heeringa: Dialectology
Linguistic determinants of mutual intelligibility in Scandinavia

The project "Mutual Comprehensibility of Language Varieties in the Lowlands: Atttitudinal and Linguistic Determinants", funded by the Flemish-Dutch Committee (Flemish FWO, Dutch NWO), aims to identify factors determining how well Dutch varieties are understood by different Dutch speakers, esp. linguistic factors such as the relative distance between varieties), and the attitudes of interlocutors. The Universities of Leuven and Groningen and the Radboud University, Nijmegen collaborate in this research.

The University of Groningen subproject will develop and evaluate computational models suitable for assaying the comprehensibility of one variety for speakers of another. Although we shall attend to simple, symmetric measures of distance between varieties such as lexical (non-)overlap and pronunciation distance as realized by Levenshtein distance (edit-distance), we are aware that comprehensibility is often asymmetric: sometimes speakers of one variety understand speakers of another better than vice versa. For this reason we are particularly interested in asymmetric measures which seek to assess the complexity of mapping from one variety to another, in particular conditional entropy.

The perfect candidate for this position would be enthusiastic about applying techniques from computational linguistics to problems in language variation and language contact. This will involve software development and evaluation, but also intensive collaboration with non-computational linguists.