It appears that you have Javascript disabled.
Please consider enabling Javascript for this page to see the visualizations.
UD for Cantonese 
Tokenization and Word Segmentation
- Cantonese is written using Chinese characters. Whitespaces are not used in this writing system, hence there are no orthographic words
(except for occasional intrusion of words in Roman and other alphabets). Syntactic words (nodes in the tree structure) have been
identified following language-internal criteria. In the CoNLL-U file,
SpaceAfter=Nois used at almost every token, including at the end of the sentence.
Morphology
Tags
This is an overview only. For more detailed discussion and examples, see the list of Cantonese POS tags.
- All 17 universal POS categories are relevant for Cantonese, including particles (PART). However, the categories SYM and X do not occur in the current data.
- TODO: Auxiliaries.
Features
- At present only one language-specific feature is used, NounType
=Clf, to distinguish classifiers from ordinary nouns.
Syntax
This is an overview only. For more detailed discussion and examples, see the list of Cantonese relations.
Relations Overview
- The following relation subtypes are used in Cantonese:
- nsubj:pass for nominal subjects of passive verbs
- nsubj:periph for postposed nominal subjects
- obj:periph for preposed objects
- obl:agent for agents of passive verbs
- obl:patient for objects raised to preverbal position and marked by 將 zoeng1
- obl:tmod for temporal oblique modifiers
- advmod:df for noun phrases headed by classifiers and specifying duration or frequency of event
- advcl:coverb for coverb phrases
- mark:adv for the manner adverbializer 噉 gam2
- mark:rel for the adjectival, relativizer, and nominalizer 嘅 ge3
- case:loc for postpositional localizers
- compound:dir for directional verb compounds
- compound:ext for extent and descriptive verb compounds
- compound:quant for verb-quantifier compounds
- compound:vo for verb-object compounds
- compound:vv for verb-verb compounds
- clf:det for classifiers used as determiners
- discourse:sp for sentence-final particles
- The following relation types are not used in Cantonese at all: expl, fixed
Instruction: Give criteria for identifying core arguments (subjects and objects), and describe the range of copula constructions in nonverbal clauses.
Treebanks
There is 1 Cantonese UD treebank: