Conway's Romanisation for Cantonese

This is my custom romanisation system for Cantonese, as spoken and perceived by me.

Features Wade–Giles-style pre-merger initials and length-indicative finals.

Initials (聲)

Where typesetting of the character ʻ U+02BB MODIFIER LETTER TURNED COMMA (used for aspirates) is inconvenient, the ASCII-only substitute ' (an apostrophe) may be used instead.

Labial

Plain dental/alveolar

Velar and glottal

Sibilant dental/alveolar and palatal

ts vs ch, tsʻ vs chʻ, and s vs sh

Modern Cantonese has merged the initials in the following pairs:

• 將 tsœng vs 張 chœng
• 槍 tsʻœng vs 昌 chʻœng
• 相 sœng vs 傷 shœng

As noted by Samuel Wells Williams in 《英華分韻撮要》 published in 1856 (咸豐丙辰年):

The initials ch and ts are constantly confounded, and some persons are absolutely unable to detect the difference, more frequently calling the words under ts as ch, than contrawise.

For academic purposes (such as compatibility with old-school things like 反切), I preserve the pre-merger initials in my romanisation system, using the initial according to 《分韻撮要》.

Nevertheless, in non-academic daily conversation, I recommend using what I call favoured post-merger initials, unless you want to sound very weird. In two decades of personal experience, I have perceived a tendency for people to favour specific initials depending on the vowel afterwards:

For example, the modern speaker will prefer

• [sɛːk̚] sek over [ɕɛːk̚] shek for 石
• [saːŋ tɕʰɔːy] saang chʻoi over [ɕaːŋ tsʰɔːy] shaang tsʻoi for 生菜.

Unfortunately, subtleties like this are not captured by any modern romanisation scheme that collapses the merged pairs.

Finals (韻)

Where typesetting of the characters œ U+0153 LATIN SMALL LIGATURE OE and ü U+00FC LATIN SMALL LETTER U WITH DIAERESIS is inconvenient, the ASCII-only substitutes oe and ue may be used instead.

[aː]

[ɐ]

[ɛː] and [e]

[iː] and [e]

[ɔː] and [o]

Cantonese has no R sound. The letter R only appears in or to represent the vowel [ɔː], and is never sounded as a consonant.

uː and o

[œ]

[ɵ]

[yː]

[m̩] and [ŋ̍]

Nine tones (九聲) with tone diagram

Tone 4 is also pronounced slightly falling, i.e. [fɐn˨˩] (21), especially if the syllable is drawn-out (for instance in operatic or theatrical dialogue). Tone 2 can start slightly lower, i.e. [fɐn˨˥] (25).

Cantonese preserves all ending consonants of Middle Chinese, and thus there is a correspondence between the nasals ([m], [n], [ŋ]) and the unreleased stops ([p̚], [t̚], [k̚]) of the entering tones (入聲). Pronunciations in olden rime dictionaries can only be understood in terms of this correspondence:

While the entering tones (入聲) 7, 8, 9 are the same pitch as tones 1, 3, 6 respectively, they should not be regarded as the same. In poetry, balancing 平聲 (tones 1 and 4) and 仄聲 (the rest) is highly desirable; if you count tone 7 (高陰入) as tone 1 (陰平), you will get tone patterns (平仄) wrong. None of Jyutping, Yale, and Sidney Lau make this distinction.

Tone patterns and literary situations aside though, tones in vernacular Cantonese are versatile, in that they can be vernacularised as different tones in particular instances. This, in my opinion, makes vernacular Cantonese probably the most inconsistent language in the world. Here are a few examples:

Compact reference tables

Initials (聲)

Finals (韻)

Nine tones (九聲)

Examples

1. 身體髮膚、受之父母、不敢毀傷、孝之始也。
   • (Pre-merger initials) shan1 tʻai2 faat8 foo1, shau6 chee1 foo6 mou5, pat7 kam2 wai2 shœng1, haau3 chee1 chʻee2 yaa5.
   • (Post-merger initials) san1 tʻai2 faat8 foo1, sau6 tsee1 foo6 mou5, pat7 kam2 wai2 sœng1, haau3 tsee1 tsʻee2 yaa5.
2. 天地玄黃、宇宙洪荒。日月盈昃、辰宿列張。
   • (Pre-merger initials) tʻeen1 tei6 yün4 wong4, yü5 chau6 hung4 fong1. yat9 yüt9 ying4 chak7, shan4 suk7 leet9 chœng1.
   • (Post-merger initials) tʻeen1 tei6 yün4 wong4, yü5 tsau6 hung4 fong1. yat9 yüt9 ying4 tsak7, san4 suk7 leet9 chœng1.
3. 夫質以代興、妍因俗易。
   • (Pre-merger initials) foo4 chat7 yee5 toi6 hing1, yeen4 yan1 tsuk9 yik9.
   • (Post-merger initials) foo4 tsat7 yee5 toi6 hing1, yeen4 yan1 chuk9 yik9.
4. 攜書劍、滯京華。
   • (Pre-merger initials) kʻwai4 shü1 keem3, chai6 king1 waa4.
   • (Post-merger initials) kʻwai4 shü1 keem3, tsai6 king1 waa4.
5. 日日醉涼州、笙歌卒未休。
   • (Pre-merger initials) yat9 yat9 ts_ü3 lœng4 chau1, shang1 kor1 ts_t7 mei6 yau1.
   • (Post-merger initials) yat9 yat9 ch_ü3 lœng4 tsau1, sang1 kor1 ch_t7 mei6 yau1.
6. 胡不念花園盟香。
   • (Pre-merger and post-merger initials are the same) woo4 pat7 neem6 faa1 yün2 mang4 hœng1?
7. 妾從無錯處、嘆我自招報應、怨句匹夫變性。
   • (Pre-merger initials) tsʻeep8 tsʻung4 mou4 tsʻor3 chʻü3, tʻaan3 ngor5 tsee6 cheeu1 pou3 ying3, yün3 k_ü3 pʻat7 foo1 peen3 sing3.
   • (Post-merger initials) tsʻeep8 chʻung4 mou4 chʻor3 chʻü3, tʻaan3 ngor5 tsee6 tseeu1 pou3 ying3, yün3 k_ü3 pʻat7 foo1 peen3 sing3.
8. 請勿靠近車門。
   • (Pre-merger initials) tsʻing2 mat9 kʻaau3 kan6 chʻe1 moon4.
   • (Post-merger initials) tsʻing2 mat9 kʻaau3 kan6 tsʻe1 moon4.
9. 下一站，九龍塘，乘客可以轉乘九廣東鐵。
   • (Pre-merger initials) haa6 yat7 chaam6, kau2 lung4 tʻong4, shing4 haak8 hor2 yee5 chün2 shing4 kau2 kwong2 tung1 tʻeet8.
   • (Post-merger initials) haa6 yat7 tsaam6, kau2 lung4 tʻong4, sing4 haak8 hor2 yee5 chün2 sing4 kau2 kwong2 tung1 tʻeet8.

Cite this page

Conway (2023). Conway's Romanisation for Cantonese. <https://yawnoc.github.io/cantonese/conway-romanisation> Accessed yyyy-mm-dd.