Taishanese Romanization

Romanizations are important for indicating Taishanese pronunciation. Many different romanization schemes have been used from the International Phonetic Alphabet to the Wade-Giles-like scheme in the DLI Basic Course or the Pinyin-like scheme in the Kaiping Character Dictionary 开平方音字典. My preferred transcription scheme is more or less a hybrid of the latter two. This romanization can be easily explained in terms of consonants, vowels and tones.

There are 20 consonants (previously discussed here and here). The table below organizes the consonants such that consonants produced with the lips are at the top, those produced with the tip of the tongue in the middle, and those produced with the back of the tongue at the bottom. (The last row is actually those produced with both the lips and back of the tongue simultaneously.)

b	p	m	f
d	t	n	lh	l
j	ch		s	y
g	k	ng	h
gw	kw			w

Most of these consonants should be straightforward for English speakers to pronounce, save the lateral fricative, lh. It’s the consonant in 四 lhi “four.”

The vowels (which I’ve touched on briefly) are at once simpler and maddeningly more complex. There are five basic vowels:

a

e

i

o

u

In some dialects, there is another schwa-like vowel quality, similar to the a in about or in the Cantonese word 乜 mat. This vowel is usually a predicable variant of one of the vowels above. For example, when occurring before t, this vowel is usually a variant of e, as in 的 et (for some accents). Likewise, when occurring before ng, this vowel is usually a variant of u, as in 東 ung (for some accents).

These vowels combine into six diphthongs, formed by appending i or u to the vowels above. (For the linguistics geeks, the front vowels pair with u, while the back vowels pair with i. The central(ish) vowel a pairs with both i and u.)

au	eu	iu
ai	oi	ui

There is, in fact, yet one more diphthong: ia. This diphthong corresponds to the vowel e, so in these accents, words like 車, 吃 and 聽 are pronounced chia, hiak and hiang. Furthermore, the diphthong eu then becomes a triphthong: iau.

The tricky part is that, although one can transcribe all Taishanese with this five-vowel framework, the vowels trade places (sort of like musical chairs) from one accent to another. The most famous of these vowel shifts is the I-E shift. Where some dialects have the i sound (as in 四 lhi or 心 lhim), other dialects have an ei or e sound (as in 四 lhei or 心 lhem). (Damn, ei is yet another diphthong!)

Rather than propose a single way to transcribe a word like 四—lhi or lhei?—I am simply proposing that these five simple vowels are able to accommodate the full range of vowel qualities, even across various accents.

Lastly, there are five basic tones. I’ve written about these at length here and here. The table below provides the tones, their description, and an example written both in Chinese and in this romanization scheme.

1	2	3	4	5
mid-level	high-level	low-level	low-falling	high-falling
偷	口	頭	厚	後
hau	hāu	häu	hạu	hàu

Diacritics are used to denote the tones. The macron, trema, underdot and grave accent respectively denote the high level tone, the low level tone, the low falling tone and the mid falling tone. The mid level tone is indicated by the lack of any diacritic.

These tones could all alternatively be written without any diacritic. Instead of a diacritic, one could use the tone number, so the five vowels could be written with a following number. I chose to write this number superscripted for my aesthetic.

hau1

hau2

hau3

hau4

hau5

In addition to these basic tones, there are changed tones. The mid and low level tones can become rising tones (mid rising and low rising), while the falling tones can become dipping tones (low dipping and mid dipping). Since these changed tones are derived from the base tones, I simply indicate the changed tone with the addition of an asterisk (*). Thus, a word like kwọi “this” undergoes the tone change to become kwọi* “here,” where the combination of the underdot and asterisk indicates a low-dipping tone. (These words could also be written as kwoi4 and kwoi4*.)

It’s a bit of a mess, but that’s my romanization scheme in a nutshell.