Educated (and not so educated) guesses about how to read Sinographs

« previous post | next post »

Here is a painting that is being exhibited in Taipei now:

(click on painting to embiggen)

This is Taiwanese artist Chang Wan-chuan's (張萬傳) 1937 painting, "Landscape of Gulangyu Island".  It shows the international character of China's Amoy (Xiamen) at the time (note, for example, the English in the advertising and a poster for a Shirley Temple movie). Some of that character survived the depredations of subsequent decades and makes Xiamen a place still worth visiting.

Chang's painting is part of an exhibition titled "Worldward: The Transformative Force of Art in Taiwan's New Cultural Movement" which is being shown (through November 28) at The Taipei Fine Arts Museum.  This show features many important representative works of Taiwanese artists during Japanese rule.

The works on exhibit are good examples of how Taiwan wasn't just water buffaloes back then but had its own artistic achievements and sophistication. (But, yes, the show does have a few paintings of water buffaloes, for those looking for that sort of thing.) Although many of the artists represented in the exhibition lived long lives, the death in 1947 of the painter Tân Têng-pho (陳澄波) serves as a reminder of the shadow that 2-28 (the February 28 Incident) cast over Taiwan.

For further information, see the show's newsletter (in English as well as Mandarin). And for more on how one hundred years ago in many ways China was just as well if not better set up to handle Western travelers than it was even into the early 1990s, see this remarkable time capsule: China 1921: The Travel Guide.

Now, to turn to the linguistics of the matter, notice that below his signature, the artist has written "Amoy Kolonsu" (鼓浪嶼 POJ Kó·-lōng-sū).  The character for "island" 嶼 is pronounced yŭ in MSM, but sū in Southern Min (SM).  This word is a good example of the heterogenous sources of Sinitic topolectal lexicons.

This discrepancy is born out by the spatial and temporal distribution of the pronunciation of 嶼:

(Zhengzhang): /*ljaʔ/


Historical linguists, not knowing what to make of this phenomenon (multiple lexical sources for one and the same Sinograph) came up with this ingenious pseudo-explanation:

Youbian dubian (simplified Chinese有边读边traditional Chinese有邊讀邊pinyinyǒu biān dú biānlit. 'read the side if any'), or dubanbian (读半边讀半邊dú bàn biān; 'read the half'), is a rule of thumb people use to pronounce a Chinese character when they do not know its exact pronunciation. A longer version is '' (yǒu biān dú biān, méi biān dú zhōngjiān; lit. "read the side if any; read the middle part if there is no side").

Around 90% of Chinese characters are phono-semantic compounds that consist of two parts: a semantic part (often the radical) that suggests a general meaning (e.g. the part  [shell] usually indicates that a character concerns commerce, as people used shells as currency in ancient times), and a phonetic part which shows how the character is or was pronounced (e.g. the part  (pinyinhuáng) usually indicates that a character is pronounced huáng in Mandarin Chinese).

The phonetic part represents the exact or almost-exact pronunciation of the character when the character was first created; characters sharing the same phonetic part had identical or similar readings. Linguists rely heavily on this fact to reconstruct the sounds of ancient Chinese. However, over time, the reading of a character may be no longer the one indicated by the phonetic part due to sound change and general vagueness.

When one encounters such a two-part character and does not know its exact pronunciation, one may take one of the parts as the phonetic indicator. For example, reading  (pinyin: yì) as zhǐ because its "side"  is pronounced as such. Some of this kind of "folk reading" have become acceptable over time – listed in dictionaries as alternative pronunciations, or simply become the common reading. For example, people read the character  ting in 西門町 (Ximending) as if it were  ding. It has been called a "phenomenon of analogy", and is observed in as early as the Song Dynasty.


There are countless mysteries about the Chinese writing system yet to be explained.

Selected readings

[Thanks to Mark Swofford and Chau Wu]


  1. Chris Button said,

    November 16, 2021 @ 8:52 pm

    I’m curious what you mean by “pseudo explanation”? The expected reflex from Middle Chinese times is xù, but yǔ has become the accepted pronunciation based on people guessing at the pronunciation in more recent times. Isn’t that a reasonable explanation?

  2. Victor Mair said,

    November 17, 2021 @ 10:06 am

    @Chris Button

    It's such a gigantic can of worms that I'm prompted to write a separate post on this mentality. I'll probably do so within a few days, and it will be called something like "Morphemes without characters".

    Stay tuned.

  3. Terry K. said,

    November 17, 2021 @ 10:43 am

    It strikes me that it doesn't mention anything about how connecting the written character with one's spoken vocabulary, with help from context clues to meaning, helps a person figure out a character's pronunciation. Which I would think would be the case, though probably complicated by differing topolects. I admit, I can imagine the possibility that a reader of Chinese would take the phonophore as indicating the pronunciation in MSM, and wrongly assume that the difference between the phonophore and the word in one's head is a case of the standard language vs one's own idiolect, rather than a case of the phonophore being an imperfect match in the current standard pronuciation. (My thoughts from the perspective of one who doesn't speak or read Chinese, but has been reading this blog for many years.)

  4. Victor Mair said,

    November 17, 2021 @ 12:12 pm

    @Terry K.

    Thank you for being a faithful and thoughtful reader of Language Log.

  5. David Marjanović said,

    November 17, 2021 @ 5:13 pm

    Stay tuned.

    Looking forward to it!

    rather than a case of the phonophore being an imperfect match in the current standard pronuciation

    Well, very imperfect matches are pretty common; things like guo vs. hua aren't rare (and for the tones of MSM I don't know if the phonophores work better than guessing).

RSS feed for comments on this post