SUD data • Version 2.14 Edit the page

In version 2.14 of SUD data, released in June 2024, there are 285 corpora:

• 3 corpora are maintained in the mSUD format (called Native mSUD). For these corpora, both mSUD and SUD versions are available in the tgz archive.
• 9 corpora are maintained in the SUD format (called Native SUD)
• 273 corpora are automatically converted to SUD from the corresponding UD data (version 2.14)

Hence, the full 2.14 release contains 282 SUD corpora, together with the mSUD version for three of them: Beja-NSC, Chinese-Beginner and Chinese-PatentChar. Note that UD 2.14 has 283 corpora but one corpus cannot be released in the SUD version, because of its CC license which contain the ND (NoDerivative) flags:

• UD_Portuguese-CINTIL → License: CC BY-NC-ND 4.0

Download all corpora

Download the full set of 285 SUD corpora: sud-treebanks-v2.14.tgz.

Native SUD and mSUD corpora

The 3 Native mSUD corpora:

The 9 Native SUD corpora:

Conversion from UD

• 273 corpora of SUD 2.14 are obtained by conversion from UD. The version of the data and tools used:
  • Input data: version 2.14 of UD corpora
  • Grew conversion rules system: tag v2.14 of the conversion system
  • Tools: grew version 1.16.0, grewlib version 1.16.0 and conll version 1.18.1

Access to each corpus

In the table below, for each corpus you can access to the Grew-match query system.