home edit page issue tracker

This page pertains to UD version 2.

It appears that you have Javascript disabled. Please consider enabling Javascript for this page to see the visualizations.

UD for Classical Chinese

Tokenization and Word Segmentation

There are neither spaces nor punctuations between words or sentences. Every word consists of a single character, except for several (proper) nouns.

Morphology

Tags

The predicate-object-final structure of very early Chinese texts had only three categories of words: predicate, object, and final. Here in our linguistic model we tentatively call them “verb” “noun” and “particle” respectively. Several words were specialised to be used as verbs, several as nouns, but most of them had been used in two or three categories around Zhou (周) dynasty.

At that era, we can observe very early modifier usages of verbs. Several verbs were specialised to be used as adverbial modifiers, afterwards caused adverbs. In between verbs and adverbs, auxiliary verbs were almost specialised to auxiliary uses, but incidentally used as verbs. Adjective usages of verbs were not specialised as adjectives at that era, on the other hand, some caused prepositions.

For POS-tagging of classical Chinese texts in UD, we use VERB ADV AUX ADP and SCONJ to fill UPOS field of each verb-origin word, following the overview of modifier usages mentioned above. For noun-origin words we use NOUN PROPN PRON NUM and ADV (noun-origin adverbs including 何), categorising them in rather nowadays point of view. For particle-origin words we use PART CCONJ and INTJ, keeping up with the guideline of UD v2. We rarely use SYM, and do not use ADJ DET PUNCT or X.

Features

NameType=Sur Giv Prs Nat Geo for PROPN.
Case=Loc Tem or NounType=Class for NOUN.
PronType=Prs with Person=1 2 3 or Reflex=Yes for personal PRON.
PronType=Dem for demonstrative PRON.
PronType=Int for interrogative PRON.
NumType=Ord for zodiac NUM.
Polarity=Neg or Degree=Pos Equ Sup Cmp for VERB and ADV.
AdvType=Deg Tim Cau with Aspect=Perf or Tense=Past Pres Fut for ADV.
Mood=Pot Nec Des or Voice=Pass for AUX.
VerbType=Cop for copular use of verb (its UPOS is changed into AUX).
VerbForm=Part for adjective use of VERB.
VerbForm=Conv for adverbial use of verb (its UPOS is changed into ADV).

Syntax

discourse:sp to annotate the final sentence particles in the predicate-object-final structure.
nsubj:pass to annotate passive subjects.
nsubj:outer and csubj:outer to annotate subjects for predicate clauses.
obl:tmod to annotate temporal oblique nominals.
obl:lmod to annotate locational oblique nominals.
compound:redup (left-to-right) to annotate reduplicated compounds.
flat:vv (left-to-right) to annotate serial verbs (rather exocentric).
flat:foreign (left-to-right) to annotate foreign words.

Treebanks

There is one Classical Chinese UD treebank:

UD_Classical_Chinese-Kyoto (implemented in UD-Kanbun and SuPar-Kanbun)

References

Koichi Yasuoka: Universal Dependencies Treebank of the Four Books in Classical Chinese, DADH2019: 10th International Conference of Digital Archives and Digital Humanities (December 2019), pp.20-28.