Martin Kroon. (2016). Automated multilingual cognate recognition & transition-rule extraction [Master’s thesis, University of Groningen & University of Lorraine]. Scripties & stageverslagen Letteren, Rijksuniveriteit Groningen. https://arts.studenttheses.ub.rug.nl/id/eprint/36109
The task of automated cognate recognition is useful for many fields of linguistics, but only a handful of studies have dealt with multiple languages at once. In the research put forth in this thesis a system is developed and evaluated that can automatically recognize cognates throughout multiple languages using multilateral transition rules. An SVM is also tested on the same data. It was found that adding more languages in the equation of cognate recognition and using multilateral transition rules improves cognate recognition. The result of this thesis is a list of extracted cognate tuples, a list of multilateral transition rules and their probability, and a multilingual cognate-recognition system.