Cantonese phonology

The standard pronunciation of Cantonese is that of Guangzhou, also known as Canton, the capital of Guangdong Province. Hong Kong Cantonese is related to the Guangzhou dialect, and the two diverge only slightly. Yue dialects in other parts of Guangdong and Guangxi provinces, such as Taishanese, may be considered divergent to a greater degree.

Cantonese syllables

A syllable generally corresponds to a word or character. Most syllables are etymologically associated with either standard Chinese characters or colloquial Cantonese characters. Modern linguists have discovered there are about 1,760 syllables being used in the entire Cantonese vocabulary, which cover the pronunciations of more than 10,000 Chinese characters. Therefore, the average number of homophonous characters per syllable is six. Phonetically speaking, a Cantonese syllable has only two parts – the sound and the tone.^[1]

Sounds

A Cantonese sound (or sound segment) usually consists of an initial (onset) and a final (rime). There are about 630 sounds in the Cantonese syllabary.

Some of these, such as /ɛː˨/ and /ei˨/ (欸), /poŋ˨/ (埲), /kʷeŋ˥/ (扃) are no longer common; some, such as /kʷek˥/ and /kʷʰek˥/ (隙), or /kʷaːŋ˧˥/ and /kɐŋ˧˥/ (梗), have traditionally had two equally correct pronunciations but are beginning to be pronounced with only one particular way by its speakers (and this usually happens because the unused pronunciation is almost unique to that word alone), thus making the unused sounds effectively disappear from the language; some, such as /kʷʰɔːk˧/ (擴), /pʰuːy˥/ (胚), /tsɵy˥/ (錐), /kaː˥/ (痂), have alternative nonstandard pronunciations which have become mainstream (as /kʷʰɔːŋ˧/, /puːy˥/, /jɵy˥/ and /kʰɛː˥/ respectively), again making some of the sounds disappear from the everyday use of the language; and yet others, such as /faːk˧/ (謋), /fɐŋ˩/ (揈), /tɐp˥/ (耷) have become popularly (but erroneously) believed to be made-up/borrowed words to represent sounds in modern vernacular Cantonese when they have in fact been retaining those sounds before these vernacular usages became popular.

On the other hand, there are new words circulating in Hong Kong which use combinations of sounds which had not appeared in Cantonese before, such as get1 (note: this is non standard usage as /ɛːt/ was never an accepted/valid final for sounds in Cantonese, though the final sound /ɛːt/ has appeared in vernacular Cantonese before this, /pʰɛːt˨/ - notably in describing the measure word of gooey or sticky substances such as mud, glue, chewing gum, etc.); the sound is borrowed from the English word get meaning "to understand".

Initial consonants

Initials (or onsets) refer to the 19 initial consonants which may occur at the beginning of a sound. Some sounds have no initials and they are said to have null initial. The following is the inventory for Cantonese as represented in IPA:

		Labial	Dental/ Alveolar		Palatal	Velar		Glottal
			plain	sibilant		plain	lab.
Nasal		m	n^3			ŋ^3
Stop	plain	p	t	t͡s		k	kʷ^1	(ʔ)^2
	aspirated	pʰ	tʰ	t͡sʰ		kʰ	kʷʰ^1
Fricative		f		s				h
Approximant			l		j^1		w^1

Note the aspiration contrast and the lack of voicing contrast for stops. The affricates are grouped with the stops for compactness in the chart.

1. Some linguists prefer to analyze /j/ and /w/ as part of finals to make them analogous to the /i/ and /u/ medials in Mandarin, especially in comparative phonological studies. However, since final-heads only appear with null initial, /k/ or /kʰ/, analyzing them as part of the initials greatly reduces the count of finals at the cost of adding only four initials.
2. Some linguists analyze a /ʔ/ (glottal stop) in place of the null initial when a vowel begins a sound.
3. In casual speech, many native speakers do not distinguish between /n/ and /l/, nor between /ŋ/ and the null initial.^[2] Usually they pronounce only /l/ and the null initial. See the discussion on phonological shift below.

The position of the coronals varies from dental to alveolar, with /t/ and /tʰ/ more likely to be dental. The position of the coronal affricates and sibilants /t͡s/, /t͡sʰ/, /s/ is alveolar and articulatory findings indicate they are not significantly palatalized by vowels.^[3][4] Historically, there was another series of alveolo-palatal sibilants as discussed below.

Vowels and terminals

Chart of monophthongs used in Cantonese, from Zee (1999:59)

Chart of diphthongs used in Cantonese, from Zee (1999:59)

Finals (or rimes) are the part of the sound after the initial. A final is typically composed of a main vowel (nucleus) and a terminal (coda).

A main vowel can be long or short, depending on vowel length. The vowel pairs [aː] and [ɐ], [ɛː] and [e], [ɔː] and [o] have corresponding formants on acoustic findings^[5][6] and are hence considered to be allophones of the same underlying vowels with different vowel length, while [œː] and [ɵ] are likewise treated by convention. Short vowels do not occur without a terminal.

A terminal can be an approximant consonant, a nasal consonant, or a stop consonant. The approximant /j/ is rounded after rounded vowels.^[7] Nasal consonants can occur as base syllables in their own right and these are known as syllabic nasals. The stop consonants (/p, t, k/) are unreleased ([p̚, t̚, k̚]).

The following chart lists all the finals in Cantonese as represented in IPA.^[8]

		Open		Mid						Close			∅-
		a-		e-		o-		ø-		i-	u-	y-
		Long	Short	Long	Short	Long	Short	Long	Short	Long	Long	Long
-∅		aː 隙		ɛː 些		ɔː 疏		œː 鋸		iː 詩	uː 夫	yː 鋸
Approximant	-j	aːi̯ 街	ɐi̯ 雞		ei̯ 你	ɔːy̯ 愛			ɵy̯ 水		uːy̯ 會
	-w	aːu̯ 教	ɐu̯ 夠	ɛːu̯^1			ou̯ 好			iːu̯ 了
Nasal	-m	aːm 衫	ɐm 深	ɛːm^1						iːm 雞			m̩ 唔
	-n	aːn 山	ɐn 新	ɛːn^1		ɔːn 謋			ɵn 准	iːn 見	uːn 歡	yːn 你
	-ŋ	aːŋ 衫	ɐŋ 疏	ɛːŋ 鏡	eŋ 敬	ɔːŋ 方	oŋ 風	œːŋ 傷					ŋ̩ 五
Stop	-p	aːp 插	ɐp 輯	ɛːp^1						iːp 接
	-t	aːt 達	ɐt 突	ɛːt^1		ɔːt 水		œːt^1	ɵt 出	iːt 結	uːt 沒	yːt 血
	-k	aːk 百	ɐk 北	ɛːk 夫	ek 亦	ɔːk 謋	ok 六	œːk 街

1. Finals [ɛːu̯],^[7] [ɛːm], [ɛːn], [ɛːp], [ɛːt] and [œːt]^[9] only appear in colloquial pronunciations of characters.^[10] They are absent from some analyses and romanization systems.

When the three checked tones are separated, the stop codas -p, -t and -k become allophones of the nasal codas -m, -n and -ŋ respectively, because they are in the complementary distribution in which the former three appear in the checked tones and the latter three appear in the non-checked tones.

Tones

Relative fundamental-frequency contours for six Cantonese tones with examples and Jyutping/Yale tone numbers (modified from Francis (2008))

Like other Chinese dialects, Cantonese uses tone contours to distinguish words, with the number of possible tones depending on the type of final. While Guangzhou Cantonese generally distinguishes between high-falling and high level tones, the two have merged in Hong Kong Cantonese, yielding a system of six different tones in syllables ending in a semi-vowel or nasal consonant. (Some of these have more than one realization, but such differences are not used to distinguish words.) In finals that end in a stop consonant, the number of tones is reduced to three; in Chinese descriptions, these "checked tones" are treated separately by diachronic convention, so that Cantonese is traditionally said to have nine tones. However, phonetically these are a conflation of tone and final consonant; the number of phonemic tones is six in Hong Kong and seven in Guangzhou.^[11]

Syllable type	Open syllables						Checked syllables
Tone name	dark flat (陰平)	dark rising (陰上)	dark departing (陰去)	light flat (陽平)	light rising (陽上)	light departing (陽去)	upper dark entering (上陰入)	lower dark entering (下陰入)	light entering (陽入)
Description	high level, high falling	medium rising	medium level	low falling, very low level	low rising	low level	high level	medium level	low level
Yale or Jyutping tone number	1	2	3	4	5	6	7 (or 1)	8 (or 3)	9 (or 6)
Example	詩	埲	試	時	市	輯	識	夫	食
Tone letter	siː˥, siː˥˧	siː˧˥	siː˧	siː˨˩, siː˩	siː˩˧	siː˨	sek˥	sɛːk˧	sek˨
IPA diacritic	síː, sîː	sǐː	sīː	si̖ː, sı̏ː	si̗ː	sìː	sék	sɛ̄ːk	sèk
Yale diacritic	sī, sì	sí	si	sìh	síh	sih	sīk	sek	sihk

For purposes of meters in Chinese poetry, the first and fourth tones are the "flat/level tones" (平聲), while the rest are the "oblique tones" (仄聲). This follows their regular evolution from the four tones of Middle Chinese.

The first tone can be either high level or high falling usually without affecting the meaning of the words being spoken. Most speakers are in general not consciously aware of when they use and when to use high level and high falling. In Hong Kong, most speakers have merged the high level and high falling tones. In Guangzhou, the high falling tone is disappearing as well, but is still prevalent among certain words, e.g. in traditional Yale Romanization with diacritics, sàam (high falling) means the number three 三, whereas sāam (high level) means shirt 衫.^[12]

The relative pitch of the tones varies with the speaker; consequently, descriptions vary from one sources to another. The difference between high and mid level tone (1 and 3) is about twice that between mid and low level (3 and 6): 60 Hz to 30 Hz. Low falling (4) starts at the same pitch as low level (6), but then drops; as is common with falling tones, it is shorter than the three level tones. The two rising tones, (2) and (5), both start at the level of (6), but rise to the level of (1) and (3), respectively.^[13]

The tone 3, 4, 5 and 6 are dipping in the last syllable when is an interrogative sentence or an exclamatory sentence. 真係? "really?" is pronounced [tsɐn˥ hɐi˨˨˥].

The numbers "394052786" when pronounced in Cantonese, will give the nine tones in order (Romanisation (Yale) saam1, gau2, sei3, ling4, ng5, yi6, chat7, baat8, luk9), thus giving a good mnemonic for remembering the nine tones.

Like other Yue dialects, Cantonese preserves an analog to the voicing distinction of Middle Chinese in the manner shown in the chart below.

Middle Chinese			Cantonese
Tone	Initial	Nucleus	Tone Name	Tone Contour	Tone Number
Level	voiceless		dark level	˥, ˥˧	1
	voiced		light level	˨˩, ˩	4
Rising	voiceless		dark rising	˧˥	2
	voiced		light rising	˩˧	5
Departing	voiceless		dark departing	˧	3
	voiced		light departing	˨	6
Entering	voiceless	Short	upper dark entering	˥	7 (1)
		Long	lower dark entering	˧	8 (3)
	voiced		light entering	˨	9 (6)

The distinction of voiced and voiceless consonants found in Middle Chinese was preserved by the distinction of tones in Cantonese. The difference in vowel length further caused the splitting of the dark entering tone, making Cantonese (as well as other Yue Chinese branches) one of the few Chinese varieties to have further split a tone after the voicing-related splitting of the Middle Chinese four tones.^[14][15]

Cantonese is special in the way that the vowel length can affect both the rime and the tone. Some linguists believe that the vowel length feature may have roots in the Old Chinese language.

There are also two changed tones, which add the diminutive-like meaning "that familiar example" to a standard word. For example, the word for "woman" in a modified tone means "daughter". They are comparable to the diminutive suffixes 儿 and 結 of Mandarin. In addition, modified tones are used in compounds, reduplications (擒擒青 kam4 kam4 cheng1 > kam4 kam2 cheng1"in a hurry") and direct address to family members (妹妹 mui6 mui6 > mui4 mui2 "sister").^[16] The two modified tones are high level, like tone 1, and mid rising, like tone 2, though for some people not as high as tone 2. The high level changed tone is more common for speakers with a high falling tone; for others, mid rising (or its variant realization) is the main changed tone, in which case it only operates on those syllables with a non-high level and non-mid rising tone (i.e. only tones 3, 4, 5 and 6 in Yale and Jyutping romanizations may have changed tones).^[17] However, in certain specific vocatives, the changed tone does indeed result in a high level tone (tone 1), including speakers without a phonemically distinct high falling tone.^[18]

Historical change

Like other languages, Cantonese is constantly undergoing sound change, processes where more and more native speakers of a language change the pronunciations of certain sounds.

One shift that affected Cantonese in the past was the loss of distinction between the alveolar and the alveolo-palatal (sometimes termed as postalveolar) sibilants, which occurred during the late 19th and early 20th centuries. This distinction was documented in many Cantonese dictionaries and pronunciation guides published prior to the 1950s but is no longer distinguished in any modern Cantonese dictionary.

Publications that documented this distinction include:

• Williams, S., A Tonic Dictionary of the Chinese Language in the Canton Dialect, 1856.
• Cowles, R., A Pocket Dictionary of Cantonese, 1914.
• Meyer, B. and Wempe, T., The Student's Cantonese-English Dictionary, 3rd edition, 1947.
• Chao, Y. Cantonese Primer, 1947.

The depalatalization of sibilants caused many words that were once distinct to sound the same. For comparison, this distinction is still made in modern Standard Mandarin, with most alveolo-palatal sibilants in Cantonese corresponding to the retroflex sibilants in Mandarin. For instance:

Sibilant Category	Character	Modern Cantonese	Pre-1950s Cantonese	Standard Mandarin
Unaspirated affricate	將	/tsœːŋ/ (alveolar)	/tsœːŋ/ (alveolar)	/tɕiɑŋ/ (alveolo-palatal)
	張		/tɕœːŋ/ (alveolo-palatal)	/tʂɑŋ/ (retroflex)
Aspirated affricate	槍	/tsʰœːŋ/ (alveolar)	/tsʰœːŋ/ (alveolar)	/tɕʰiɑŋ/ (alveolo-palatal)
	昌		/tɕʰœːŋ/ (alveolo-palatal)	/tʂʰɑŋ/ (retroflex)
Fricative	相	/sœːŋ/ (alveolar)	/sœːŋ/ (alveolar)	/ɕiɑŋ/ (alveolo-palatal)
	傷		/ɕœːŋ/ (alveolo-palatal)	/ʂɑŋ/ (retroflex)

Even though the aforementioned references observed the distinction, most of them also noted that the depalatalization phenomenon was already occurring at the time. Williams (1856) writes:

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 contrariwise.

Cowles (1914) adds:

"s" initial may be heard for "sh" initial and vice versa.

A vestige of this palatalization difference is sometimes reflected in the romanization scheme used to romanize Cantonese names in Hong Kong. For instance, many names will be spelled with sh even though the "sh sound" (/ɕ/) is no longer used to pronounce the word. Examples include the surname 石 (/sɛːk˨/), which is often romanized as Shek, and the names of places like Sha Tin (沙田; /saː˥ tʰiːn˩/).

The alveolo-palatal sibilants occur in complementary distribution with the retroflex sibilants in Mandarin, with the alveolo-palatal sibilants only occurring before /i/, or /y/. However, Mandarin also retains the medials, where /i/ and /y/ can occur, as can be seen in the examples above. Cantonese had lost its medials sometime ago in its history, reducing the ability for speakers to distinguish its sibilant initials.

In modern-day Hong Kong, many younger speakers are unable to distinguish between certain phoneme pairs such as /n/ vs. /l/ and merge one sound into another. Although that is often considered as substandard and is denounced as being "lazy sounds" (懶音), it is becoming more common and is influencing other Cantonese-speaking regions (see Hong Kong Cantonese.)

See also

• Proper Cantonese pronunciation

Notes

1. While most linguists state that Syllable = Sound + Tone, a few prefer to say that Tonal Syllable = Base Syllable + Tone. For the sake of simplicity, this article chooses to use the first equation.
2. Yip & Matthews (2001:3–4)
3. Zee, Eric (1996). "Phonological Changes in Hong Kong Cantonese". Current Issues In Language and Society. 3 (2): 192–198. doi:10.1080/13520529609615469.
4. Lee, W.-S.; Zee, E. (2010). "Articulatory characteristics of the coronal stop, affricate, and fricative in cantonese". Journal of Chinese Linguistics. 38 (2): 336–372.
5. Zee, Eric (2003), "Frequency Analysis of the Vowels in Cantonese from 50 Male and 50 Female Speakers" (PDF), Proceedings of the 15th International Congress of Phonetic Sciences: 1117–1120
6. Bauer & Benedict (1997:46–47)
7. Zee, Eric (1999), "An acoustical analysis of the diphthongs in Cantonese" (PDF), Proceedings of the 14th International Congress of Phonetic Sciences: 1101–1105
8. Bauer & Benedict (1997:49)
9. Bauer & Benedict (1997:78–79)
10. Bauer & Benedict (1997:60)
11. Bauer & Benedict (1997:119–120)
12. Guan (2000:474 and 530)
13. Jennie Lam Suk Yin, 2003, Confusion of tones in visually-impaired children using Cantonese braille (Archived by WebCite® at http://www.webcitation.org/6AK0HT0Vk[])
14. Norman (1988:216)
15. Ting (1996:150)
16. Matthews & Yip (2013, section 1.4.2)
17. Yu (2007:191)
18. Alan C.L. Yu. "Tonal Mapping in Cantonese Vocative Reduplication" (PDF). Retrieved 27 September 2014.

References

• Bauer, Robert S.; Benedict, Paul K. (1997), Modern Cantonese Phonology, Walter de Gruyter, ISBN 978-3-11-014893-0
• Francis, Alexander L. (2008), "Perceptual learning of Cantonese lexical tones by tone and non-tone language speakers", Journal of Phonetics, Elsevier, 36 (2): 268–294, doi:10.1016/j.wocn.2007.06.005
• Guan, Caihua (2000), English-Cantonese Dictionary: Cantonese in Yale Romanization, New-Asia - Yale-in-China Language Center, ISBN 978-962-201-970-6
• Matthews, Stephen; Yip, Virginia (2013), Cantonese: A Comprehensive Grammar, London: Routledge
• Norman, Jerry (1988), Chinese, Cambridge Language Surveys, Cambridge University Press, ISBN 0-521-22809-3
• Ting, Pan-Hsing (1996), "Tonal Evolution and Tonal Reconstruction in Chinese", in Huang, Cheng-teh James; Li, Yen-hui Audrey, New horizons in Chinese linguistics, Springer Science & Business Media, ISBN 0-7923-3867-7
• Yip, Virginia; Matthews, Stephen (2001), Basic Cantonese: A Grammar and Workbook, Routledge, ISBN 978-0415193849
• Yu, Alan C. L. (2007), "Understanding near mergers: the case of morphological tone change in Cantonese" (PDF), Phonology, Cambridge University Press, 24: 187–214, doi:10.1017/S0952675707001157
• Zee, Eric (1999), "Chinese (Hong Kong Cantonese)", Handbook of the International Phonetic Association: A guide to the use of the International Phonetic Alphabet, Cambridge: Cambridge University Press, ISBN 0-521-65236-7