跨語族視角下的粵語標調習慣省思 Reflections on Cantonese Tone Marking Conventions from a Cross-Family Perspective

(An English translation of this short essay is appended following this Chinese version.)

廖漢波 2026年9月9日星期三

上個學期，我在香港大學教了一門文學碩士（MA）課程《The History and Structure of Cantonese》（粵語的歷史與結構）。這個學期，我又開始在香港教育大學教一門本科課程《Chinese Phonetics and Phonology》（漢語語音學），當中的「漢語」主要涵蓋現代漢語（普通話）和粵語。因為教學需要，課堂上自然會深度涉及粵語的聲調教學。

我本身是壯語（儂佒/德靖）母語者，長期從事仡偙語（Kra-Dai，即壯侗/侗台）和嶺南漢語的研究，對跨語族的聲調標記方式累積了一些心得。其中最讓我至今到感到不順手的，就是粵語研究圈子的數字標調，不管是用IPA（國際音標）還是粵拼。坦白說，我覺得這是粵語研究和應用圈自娛自樂、不太理會圈外學界慣用法的結果。

粵拼以及用IPA給粵語標調類的常用做法，是用1、2、3、4、5、6來標調，順序是：1陰平、2陰上、3陰去、4陽平、5陽上、6陽去，即先來三個陰調，再來三個陽調。這種排法，對只熟悉粵拼的人來說或許自然，但對熟悉壯侗語、苗瑤語、越南語等其他相關的亞洲聲調語言學圈子的人來說，一開始真的很難適應。因為這些學界非常一致地採用單數為陰調、雙數為陽調（即1、3、5為陰調，2、4、6為陽調）的做法，這樣分類在研究上呈現出一致性和分類方便性，且利於方言間甚至語言間的比較。然而，粵拼把三個陰調排完再排三個陽調，等於打散了陰陽對應的邏輯，導致只熟悉粵拼標調法的人，第一次接觸其他語言的標調方式時，總要花一段時間才能轉換過來。

更麻煩的是教學層面。粵拼那種「三個高調先讀完，再讀三個低調」的教學順序，實際上削弱了聲調成對高低對比的聽覺訓練。尤其是，陰上（調值25）與陽上（調值23）、陰去（調值33）與陽去（調值22），調值走向本來就非常接近，非母語者很難區分，將這兩對聲調分別拆分到前三和後三調類，教學上就令非母語者無法直接連讀對比，體會其細微差異。如果參照壯文的聲調教學順序——即1陰平、2陽平、3陰上、4陽上、5陰去、6陽去——以陰陽交替的方式呈現，對比就會立刻浮現，學習效率高得多。事實上，就連很多粵語母語者，自己也常常分不清陰上和陽上，這正好說明了問題所在。

值得一提的是，Robert Bauer教授的粵語系列研究，比如在他1996年發表的論文〈Identifying the Tai Substratum in Cantonese〉中，對粵語聲調的標記方式其實與壯侗語圈的習慣一致，可見並非做不到，只是粵拼的設計選擇走了另一條路。

另外，我對粵語入聲的調值標調方式也不以為然。粵語圈習慣把入聲調值標成2、3、5——請注意，這裡的問題不在於用數字，而在於只用一個數字來代表整個調值。一個調值本質上是一個聲調輪廓（contour），有起點就有終點，邏輯上用兩個數字才能完整標示，例如高平調應標為55，中平調為33，低平調為22，更加複雜的調值甚至帶特殊的發聲態則用213、453、24ʔ、33ʔ等。只用一個數字（2、3、5），等於把調值輪廓抹掉了，這在語音描寫上是不夠精確的。粵語圈之所以這樣做，大概是認為入聲音節比舒聲音節短，用一個數字就夠了。但再短的音節，也有自己的調值高低變化，就該標出起點和終點，聲調的時長問題用文字補充說明即可，不必在標調符號上另搞一套。總之，粵語圈習慣做法做會造成幾個問題：

第一，數字2、3、5容易跟調類編號混淆；

第二，壯侗語、苗瑤語、越南語的入聲在時長上也一樣比舒聲音節的聲調短，但這些語言的學界仍然用雙數字完整標示入聲調值，可見時長不應成為簡化標調的理由，粵語另闢蹊徑的做法其實沒有必要；

第三，粵語本身就有長陰入（所謂下陰入，調值常用數字3）和短陰入（所謂上陰入，調值常用數字5）的對立，它們時長也是一長一短，但都用單個數字3、5標註，如何區分它們在時長上的差異？

第四，粵語圈處理遇到高升變調的入聲字時，又要標成35，又產生了入聲調值格式不一致的問題；

第五，不言而喻，這種做法無形中造成了粵語研究圈與其他語言學圈子的隔閡。

所以我個人認為，粵語入聲調值老老實實標回55、33、22這些完整兩個數字的調值才是合理的，至於時長問題，一開始用文字說明就好。

不過，說到底，這些都只是我個人的想法。粵語圈子的數字標調習慣早已約定俗成，深入民間與教學體系，我無意也無力改變什麼。只是作為一個跨語族的研究者，看到不同學術圈之間因為標調習慣不同而產生不必要的溝通成本，心裡總有些感慨，順手記下來罷了。

Reflections on Cantonese Tone Marking Conventions from a Cross-Family Perspective

Liao Hanbo, Wednesday, 9 September 2026

Last semester, I taught a Master of Arts (MA) course at the University of Hong Kong entitled The History and Structure of Cantonese. This semester, I have begun teaching an undergraduate course at the Education University of Hong Kong entitled Chinese Phonetics and Phonology, in which "Chinese" primarily covers Modern Chinese (Standard Mandarin/Putonghua) and Cantonese. Given the teaching requirements, the classes naturally involve in-depth instruction on Cantonese tones.

I am a native speaker of Zhuang (Nong Iaang / Dejing), and have long been engaged in research on Kra-Dai (i.e., Zhuang-Dong / Kam-Tai) and Lingnan Chinese languages. Over the years, I have accumulated some insights into tone marking conventions across language families. What has remained most awkward for me to this day is the use of numerical tone marking within Cantonese studies circles — whether in IPA or in Jyutping. To be frank, I consider this a case of the Cantonese studies and applied linguistics circles amusing themselves without much regard for conventions prevailing in other academic circles.

The common practice in Jyutping, as well as in IPA tone marking for Cantonese, is to use 1, 2, 3, 4, 5, and 6 to denote tones, in the order: 1 = yīn píng (Upper Level), 2 = yīn shǎng (Upper Rising), 3 = yīn qù (Upper Departing), 4 = yáng píng (Lower Level), 5 = yáng shǎng (Lower Rising), 6 = yáng qù (Lower Departing) — that is, all three Upper tones first, followed by all three Lower tones. This ordering may feel natural to those familiar only with Jyutping, but for those acquainted with other tone-language circles in Asia, such as Kra-Dai, Hmong-Mien, and Vietnamese, it is quite difficult to adapt to at first. Those circles have very consistently adopted the convention of odd numbers for Upper (yin) tones and even numbers for Lower (yang) tones (i.e., 1, 3, 5 for Upper; 2, 4, 6 for Lower), which offers consistency and convenience for classification, and facilitates comparison across dialects and even across languages. By contrast, Jyutping places all three Upper tones before all three Lower tones, thereby disrupting the yin-yang pairing logic. As a result, those who are only familiar with Jyutping tone marking invariably need some time to adjust when they first encounter tone marking systems used in other languages.

The problem becomes even more pronounced at the pedagogical level. The Jyutping teaching sequence — "read all three high tones first, then all three low tones" — actually weakens auditory training on the paired high-low contrasts. In particular, Upper Rising (tone value 25) and Lower Rising (tone value 23), as well as Upper Departing (tone value 33) and Lower Departing (tone value 22), have quite similar tonal contours. Non-native speakers find them very difficult to distinguish. By splitting these two pairs across the first three and last three tone categories, the pedagogical approach denies non-native learners the opportunity to read them in immediate succession for direct contrast, making it harder for them to perceive the subtle differences. If, by contrast, one follows the Zhuang tone teaching sequence — i.e., 1 = yīn píng (Upper Level), 2 = yáng píng (Lower Level), 3 = yīn shǎng (Upper Rising), 4 = yáng shǎng (Lower Rising), 5 = yīn qù (Upper Departing), 6 = yáng qù (Lower Departing) — and presents them in alternating yin-yang order, the contrasts become immediately apparent and learning efficiency improves considerably. In fact, even many native Cantonese speakers often cannot distinguish Upper Rising from Lower Rising, which precisely illustrates the problem.

It is worth noting that Professor Robert Bauer's Cantonese research, such as his 1996 paper "Identifying the Tai Substratum in Cantonese", employs tone marking conventions consistent with those of the Kra-Dai circles. This shows that it is not impossible to do otherwise; it is simply that the design choices of Jyutping have taken a different path.

Furthermore, I am not convinced by the way Cantonese checked-tone (entering tone) values are marked. Cantonese circles conventionally mark checked-tone values as 2, 3, and 5 — please note that the issue here is not the use of numerals per se, but the use of a single numeral to represent an entire tone value. A tone value is essentially a tonal contour, with a starting point and an ending point. Logically, two numerals are needed to represent it completely; for example, a high-level tone should be represented as 55, a mid-level tone as 33, and a low-level tone as 22. For more complex tone values, or those involving special phonation types, representations such as 213, 453, 24ʔ, 33ʔ, etc., may be used. Using a single numeral (2, 3, or 5) effectively erases the tonal contour, which is insufficiently precise in phonetic description. The Cantonese circles presumably do this because they consider checked syllables to be shorter than unchecked (open) syllables, and believe a single numeral suffices. However, no matter how short a syllable is, its tone still has pitch movement, and the starting and ending points should be indicated. The issue of tonal duration can be supplemented with textual explanation; there is no need to devise a separate marking system. This practice creates several problems:

First, the numerals 2, 3, and 5 are easily confused with tone category numbers.

Second, in Kra-Dai, Hmong-Mien, and Vietnamese, checked tones are likewise shorter in duration than unchecked tones, yet those scholarly circles still use two-numeral representations for checked-tone values. This shows that duration should not be a justification for simplifying tone marking, and the Cantonese practice of taking a different path is really unnecessary.

Third, Cantonese itself has a contrast between the long checked entering tone (so-called Lower Entering, commonly represented by the numeral 3) and the short checked entering tone (so-called Upper Entering, commonly represented by the numeral 5), which differ from each other in duration as well. Yet both are marked with single numerals, 3 and 5 respectively — how then can their durational difference be distinguished?

Fourth, when Cantonese circles encounter checked-tone characters that undergo rising-tone changed tones, they mark them as 35, which creates an inconsistency in the format of checked-tone marking.

Fifth, and needless to say, this practice has inadvertently created a divide between Cantonese studies circles and other linguistic circles.

For these reasons, I personally believe that the more reasonable approach is to mark Cantonese checked-tone values properly as complete two-numeral tone values — 55, 33, 22 — and to explain the duration issue in text from the outset.

That said, ultimately these are only my personal views. The numerical tone marking conventions of Cantonese circles have long been established and are deeply embedded in both popular usage and pedagogical systems. I have neither the intention nor the power to change anything. I merely write down these reflections in passing, as a researcher working across language families who has observed the unnecessary communication costs arising from differing tone marking conventions between academic circles, and who feels a certain sense of regret.