UD_Cantonese-HK
Treebank card from UD release 2.18: metadata and texts come from UD_Cantonese-HK/README.md; size is counted by streaming the .conllu files and checked against the release stats.xml. Generated by tools/treebanks build; do not edit by hand. Papers: Papers: UD_Cantonese-HK; seeds: seeds/; converter: convert.md.
| Field | Value |
|---|---|
| UD name | UD_Cantonese-HK |
| Language | Cantonese (yue) |
| Family | Sino-Tibetan |
| Treebank code | yue_hk |
| UD page | universaldependencies.org/treebanks/yue hk/index.html |
| License | CC BY-SA 4.0 |
| Genres | spoken |
| Text in data | yes |
| Parallel | hk |
| In UD since | UD v2.1 |
| Authors | Kim Gerdes; John Lee; Herman Leung; Tak-sum Wong |
| Contact | tswong-c@my.cityu.edu.hk; jsylee@cityu.edu.hk |
| Development | outside the UD repo (elsewhere): fixes go to the source or the converter |
| Documentation status | partial |
| Data source | manual |
Annotation layers
| Layer | Origin |
|---|---|
| Lemmas | automatic |
| UPOS | manual native |
| XPOS | not available |
| Features | manual native |
| Relations | manual native |
UD scale: manual native — annotated by hand directly in UD; converted from manual — converted from manual annotation in another scheme; converted with corrections — converted and partly corrected by hand; automatic with corrections — automatic, with partial manual corrections; automatic — automatic; not available — the layer is absent.
Size
| Split | Sentences | Tokens | Words | Multiword tokens | Empty nodes |
|---|---|---|---|---|---|
| test | 1 004 | 13 918 | 13 918 | 0 | 0 |
| total | 1 004 | 13 918 | 13 918 | 0 | 0 |
Tokens are surface tokens (Spanish vámonos is one), words are syntactic words (vamos + nos); multiword tokens are those split into several words; empty nodes are the N.M nodes of the enhanced graph. Splits train, dev are absent from the release.
Summary
A Cantonese treebank (in Traditional Chinese characters) of film subtitles and of legislative proceedings of Hong Kong, parallel with the Chinese-HK treebank.
Introduction
The Cantonese-HK UD treebank was manually annotated by Tak-sum Wong and Herman H. M. Leung at City University of Hong Kong, by finely transcribing three films shooted by students from the School of Creative Media as well as part of the section of the Election of President of the council meeting of the Legislative Council of the HKSAR of the P. R. China on 12th October, 2016 commencing at 2:13 p.m. The data are in Traditional Chinese. These trees form a parallel treebank with those in Chinese-HK.
ORIGIN
#send_id = 1 to 410
- Title: Missing days / 小時光
- Source: youtube.com/watch?v=1qSMiw0vhzU
- Duration: 30 minutes
- Fluency: The dialogues are mostly prepared.
#send_id = 411 to 547
- Title: Tempo in Temple / 廟眾樂樂
- Source: youtube.com/watch?v=8e8Lqd6grTE
- Duration: 16 minutes
- Fluency: This is an interview. Most sentences are not prepared and thus contain a lot of disfluencies. A lot of reparandums are tagged.
#send_id = 548 to 650
- Title: What day is today / 今日星期幾
- Source: youtube.com/watch?v=bBGwxIDiZ o
- Duration: 31 minutes
- Fluency: The dialogues are mostly prepared. This film contains a lot of dead air.
#send_id = 651 to 1004
- Title: Election of President: The council meeting of the Legislative Council of HKSAR on 12th October, 2016(14:13:49 ~ 14:47:54)
- Source: legco.gov.hk/yr16-17/chinese/counmtg/hansard/cm20161012-translate-c.pdf#nameddest=eop https://webcast.legco.gov.hk/public/zh-hk/SearchResult?MeetingID=M16100003
- Duration: 34 minutes 5 seconds
- Fluency: This is an discission. Most sentences are not prepared and thus contain a lot of disfluencies. A lot of reparandums are tagged.
DATA
Since the Chinese-HK counterpart contains less than a myraid tokens, to make it comparable with the Mandarin counterpart, only the test set is provided.
BASIC STATISTICS
- Tree count: 1004
- Word count: 13918
- Token count: 13918
- Dep. relations: 48 of which 17 language specific
- POS tags: 15
- Category=value feature pairs: 1
FORM and LEMMA
Since Cantonese is not an official written language, there is no standard written form. In this treebank, the FORM field adopts the popular written form while we have tried our best to find different characters attested in dictioanries or other studies for the LEMMA field to discern the homophones, in order to faciliate computer processing.
POS TAGGING
When determining the POS, one usually considers both the "morphological evidence", i.e., the linguistic form of the word, as well as the "distributional evidence", i.e., its syntactic use in the sentence. Usually, these two kinds of evidence should agree; in some cases, however, the POS of some word is tentatively alterned to fullfil the special purpose of the author.
Take an example in English, consider the word "pen" in the sentence "I pen a letter". Morphological evidence suggests the word "pen" should be tagged as an noun (NN), reflecting its normal usage. Distributional evidence suggests it should be tagged as a verb, since this word is in a typical syntactic position of verb in a SVO language.
When these two kinds of evidence contradict one another, the morphological evidence prevails.
This page in Markdown → https://md.mova.elamur.ai/free/en/docs/treebanks/yue/hk/