EVIDENCE FROM CHARACTERS

Semantic-phonetic compound characters

Although Chinese writing cannot directly tell us how Chinese words sounded in the past, there is one kind of relational information that we can nonetheless get from characters. Such relational information comes from semantic-phonetic compound characters.

The phonetic component in semantic-phonetic compound characters serves to indicate the pronunciation of the character. To serve this function well, the pronunciation of the phonetic component should be completely identical to that of the whole character. But very often we find difference in pronunciation between the two. The following cases suffice to illustrate the phenomenon:

Pronunciation of Phonetic Component	Pronunciation of Whole Character	Pronunciation of Whole Character
非 fei	悲 bei	排 pai
甫 fu	浦 pu	葡 pu
方 fang	旁 pang	彷 pang
登 deng	澄 cheng	橙 cheng
周 zhou	调 tiao/diao	雕 diao
占 zhan	点 dian	店 dian
甘 gan	钳 qian	箝 qian
句 ju	狗 gou	够 gou

Should these apparent exceptions undermine our faith in the ability of phonetic component to indicate pronunciation?

Despite appearance to the contrary, our answer is ‘no’. First of all, the differences between the first and the second and third column are systematic. As we can see, only the initials of the syllables are different. Furthermore, there is consistency in what the differences are. We therefore should be confident that the use of the phonetic component in these characters was indeed based on the Rebus principle and not used arbitrarily.

Give the fact that the sounds that differ here are produced at parts of the mouth in close proximity, we can assume that they are closely related. For example [f] and [p] [b] are all produced with the lip. [d] and [ch]/[zh] are produced at the neighboring regions of the alveolar and palate respectively. Furthermore, given what we know about the common sound changes that can relate these different sounds, we can be confident that at the time these characters were created, the phonetic component did indeed serve to indicate the pronunciation of the whole characters perfectly. It is due to later sound changes to either the pronunciation of the phonetic component or the whole character that created the observed discrepancies in pronunciation. We can therefore use the following assumption in our reconstruction:

If two characters share the same phonetic component, they must have had the same pronunciation (not necessarily tones) when the characters were created.

Scribe errors:

Another kind of evidence from characters we can use for reconstruction is the errors that scribes made in the past. They are the same kind of mistakes that native speakers most commonly make today, homophonic substitution, namely the substitution of the right character with a wrong character pronounced the same way.

For example, the following mistakes were observed in historical documents:

Correct character	Modern Mandarin pronunciation	Wrong character	Modern Mandarin pronunciation
鹉	wu	母+鸟	mu
匍	pu	扶	fu

The above errors tell us two things:

1. To allow homophonic substitution, 武 and 母 must have been pronounced the same way once, despite the fact that in modern Mandarin they are pronounced as wu and mu respectively. This can explain why many phonetic components having the same initials do not seem to work. For example, 文 (wen) in 悯 (min) and 门(men) in 闻 (wen).

2. The substitution of匍 (pu) by 扶(fu) tells us that they were at one time pronounced the same way. This will explain why many phonetic components having these initialsdo not work anymore. For example, 非 (fei) in 悲 (bei).