UD for Thai 
Tokenization and Word Segmentation
- The Thai writing system does not have the concept of orthographic words. Longer text segments are occasionally
separated by whitespace but most words are not delimited this way. Therefore, in the annotation,
SpaceAfter=Nois used even between letters. Multiword tokens are not used.
Morphology
Tags
- The Thai treebank currently uses only 15 of the 17 universal POS categories, including particles (PART). There are no interjections and unknown words.
- Ten words are currently annotated as auxiliaries (AUX):
- The copula เป็น pen “is”.
- Perfect มา maː.
- Perfect แล้ว læ̂ːw.
- Perfect ได้รับ dạîrạb.
- Past เคย gʰeːy.
- Progressive กำลัง kãːlạŋ.
- Future จะ ča.
- Passive ถูก tʰuːk.
- Necessitative ต้อง tɔ̂ːŋ “must”.
- Potential ได้ dạî “can”.
Syntax
Core Arguments, Oblique Arguments and Adjuncts
- Nominal subject (nsubj) is a bare noun phrase (without adposition).
- A subordinate clause may serve as the subject and is labeled csubj.
- Direct object (obj) is a bare noun phrase.
- Nominal objects of some verbs alternate with finite clausal complements, which are labeled ccomp.
Non-verbal Clauses
- The default copula is เป็น pen “is”.
Relations Overview
- Similarly to Chinese, Thai uses classifiers (clf).
- The following relation subtypes are used in Thai:
- nsubj:pass for nominal subjects of passive verbs
- csubj:pass for clausal subjects of passive verbs
- aux:pass for passive auxiliaries
- acl:relcl for relative clauses
- compound:prt for verbal particles
- nmod:poss for possessive/genitive modifiers
- obl:poss for possessive modifiers when the parent is a verb (nominalized, but still tagged VERB)
- obl:tmod for temporal modifiers (a nominal modifies a clause)
- flat:name to connect parts of a personal name
- cc:preconj for the first part of a multi-word coordinating conjunction (both)
- The following main types are not used alone and must be subtyped: flat
- The following relation types are not used in Thai at all: expl, orphan
Treebanks
There are 2 Thai UD treebanks: