This is part of archived UD v1 documentation. See http://universaldependencies.org/ for the current version.
home issue tracker

Morphology

The universal part-of-speech categories, features and values in the Slovenian UD Treebank have been obtained by conversion from the JOS morphosyntactic specifications. The majority of conversions was done automatically by mapping JOS morphosyntactic tags to universal POS tags and features, while in some cases also word forms, lemmas and/or dependency relations were taken into account.

The Slovenian UD tagset includes all universal POS tags and features except for Voice. In addition to that, the set of universal features has been extended with five additional features to either describe language-specific features (such as Gender[psor], Number[psor] and Variant) or preserve some finer-grained morphological information encoded in the original ssj500k treebank (such as Abbr, Foreign and NumForm).

The conversion principles for individual POS tags and features are detailed in the corresponding Slovenian guidelines and the original JOS morphosyntactic tags have been preserved in the language-specific FEATS column. Further improvements and manual revisions of this work are still a work in progress.