home edit page issue tracker

This page still pertains to UD version 1.


Universal part-of-speech tags and universal features in the Latvian data have been obtained by an automatic conversion of the Latvian Treebank morphological tags, also taking into account syntactic roles (to distinguish DET from PRON), lemmas and wordforms.

Lemmas from Latvian Treebank is used as-is except “words with spaces”, where spliting on whitespaces provides correct result in all known cases.

Currently no language specific tags or features are used.

Known discrepancies