A systematic comparison of methods for low-resource dependency parsing on genuinely low-resource languages
This addresses the challenge of building parsers for many under-resourced languages, but it is incremental as it compares existing strategies rather than introducing a new method.
The paper tackled the problem of dependency parsing for low-resource languages by systematically comparing methods like data augmentation, cross-lingual training, and transliteration, finding that data augmentation helps when only low-resource data is available, cross-lingual training complements it with related high-resource data, and transliteration aids with different writing systems.
Parsers are available for only a handful of the world's languages, since they require lots of training data. How far can we get with just a small amount of training data? We systematically compare a set of simple strategies for improving low-resource parsers: data augmentation, which has not been tested before; cross-lingual training; and transliteration. Experimenting on three typologically diverse low-resource languages---North Sámi, Galician, and Kazah---We find that (1) when only the low-resource treebank is available, data augmentation is very helpful; (2) when a related high-resource treebank is available, cross-lingual training is helpful and complements data augmentation; and (3) when the high-resource treebank uses a different writing system, transliteration into a shared orthographic spaces is also very helpful.