Get our free extension to see links to code for papers anywhere online!

Chrome logo  Add to Chrome

Firefox logo Add to Firefox


Cognate-aware morphological segmentation for multilingual neural translation

Aug 31, 2018
Stig-Arne Grönroos, Sami Virpioja, Mikko Kurimo


Share this with someone who'll enjoy it:


This article describes the Aalto University entry to the WMT18 News Translation Shared Task. We participate in the multilingual subtrack with a system trained under the constrained condition to translate from English to both Finnish and Estonian. The system is based on the Transformer model. We focus on improving the consistency of morphological segmentation for words that are similar orthographically, semantically, and distributionally; such words include etymological cognates, loan words, and proper names. For this, we introduce Cognate Morfessor, a multilingual variant of the Morfessor method. We show that our approach improves the translation quality particularly for Estonian, which has less resources for training the translation model.

* To appear in WMT18 


   Access Paper Source



Share this with someone who'll enjoy it: