UD for Evenki 
Evenki is a Northern Tungusic language spoken by the Evenki people across a vast area of eastern Siberia (Russia) and in Heilongjiang and Inner Mongolia (China). Together with Even and Negidal, it belongs to the Northern branch of the Tungusic family. Evenki is the most widely dispersed Tungusic language but is nevertheless endangered, with the number of active speakers declining rapidly.
Tokenization and Word Segmentation
- Words are delimited by whitespace.
- Punctuation is separated from adjacent words as separate tokens.
- There are no multiword tokens and no words with spaces in the current treebank.
- Bound morphemes (case, possessive, and person suffixes) are written together with their host, following the phonemic Latin transcription used in the reference grammar. Following the analytical practice of Tungusic descriptive grammar, no morpheme boundaries are introduced inside inflected word forms during tokenization.
Morphology
Tags
- Tags actually attested in the treebank include ADJ, ADP, ADV, AUX, CCONJ, DET, INTJ, NOUN, NUM, PART, PRON, PROPN, PUNCT, SCONJ, VERB, and X.
NOUNis used for common nouns;PROPNis reserved for proper names (personal names, place names).PARTis used for discourse particles and the negation particle.- The
DET/PRONdistinction follows the general UD principle: forms that modify a nominal head are taggedDET, forms that head their own phrase are taggedPRON.
Features
- Nominal features attested in the treebank:
Case(Nom, Abl, Acc, All, Dat, Loc),Number(Sing, Plur),Person(1, 2, 3),Poss,PronType(Dem, Prs),Reflex,Definite(Ind),NumType(Card). - Verbal features attested in the treebank:
VerbForm(Fin, Part, Conv),Mood(Ind, Imp, Cnd, Prp),Tense(Past, Fut),Aspect(Hab, Imp),Polarity. - Adjectival features attested in the treebank:
Degree(Pos). - Evenki-specific morphological information that does not correspond to any value in the universal feature inventory is encoded in the MISC column rather than FEATS, in accordance with UD conventions.
Syntax
- Evenki is a head-final language with SOV constituent order. The subject (
nsubj) precedes the object (obj); both precede the verb. - Core arguments are identified morphosyntactically:
nsubjis typically in the nominative (unmarked),objis typically accusative, andiobjis typically dative. - Oblique arguments and adjuncts are tagged
obl; case marking is realized as suffixes on the nominal head. - Attributive modifiers precede their nominal head. Genitive and other adnominal modifiers use
nmod; adjectives useamod. - Copular constructions attach a nonverbal predicate as the head, with the copula (when overt) attached via
cop. - No language-specific dependency subtypes are introduced in the current release.
Treebanks
There is 1 Evenki UD treebank: