UD for Chukchi 
Tokenisation and Word Segmentation
Tokens in the corpus are defined as sequences of non-whitespace or punctuation characters separated by whitespace.
Underneath tokens are syntactic words. These are words which take part in dependency relations.
A single token may be split into more than one word based on the following criteria:
- Emphatic clitics are split (notably -ъм and -а)
- In non-absolutive noun phrases, attributive modifiers (nouns, adjectives and numerals) which have been incorporated are split. The splits are based on the provided gloss.
Incorporation into lexical verbs does not involve splitting in the basic representation, sentences are annotated according to their surface syntax, e.g. a verb with an incorporated object is annotated as intransitive following its agreement morphology.
In the enhanced representation, the argument structure is recovered, incorporated items are given nodes.
Morphology
Tags
The following are described by Dunn as particles, but are annotated here as AUX
because they provide the clause with tense-aspect-modality features.
- ӄырым:
- ванэван (also ванэ):
The following are copulae:
- итык:
- вак:
Features
Syntax
- Core arguments are marked on the verb. Subjects of intransitive verbs have agreement markers with
subj, e.g.Person[subj]=1. Subjects of transitive verbs have agreement markers withagent, e.g.Person[agent]=1. Objects are marked withobj, e.g.Person[obj]=3.
The following relation subtypes are used in the Chukchi data:
- acl:attr Adnominal clause with attributive morphology
- acl:relat Adnominal clause with relational morphology
- advmod:emph Used for discourse “particles” such as -ым and -а/-э
- aux:neg Negative auxiliary
- flat:foreign Used for chunks of text that are in Russian.
- flat:name Proper noun chunks without internal structure.
- nmod:attr Nominal modifier of a noun phrase with attributive morphology
- nmod:poss Nominal modifier of a noun phrase with possessive morphology
- nmod:relat A noun phrase modifying another noun phrase with relational meaning, see Dunn (1999: §8.7)
- parataxis:rep Parataxis where the “dependent” is reported speech.
Treebanks
There is 1 Chukchi UD treebank: