What does ‘also phonetic’ mark? The Shuōwén’s yìshēng label, homophony and derivation in Old Chinese

Language is often a map of how people think, but for ancient languages, that map is frequently incomplete. In many modern tongues, we can see how new words are built from old ones because the changes are written down: a prefix is added, a suffix is attached, or a sound shifts in a predictable way. But for Old Chinese, spoken over two thousand years ago, the written record is different. The script used at the time did not break words down into smaller parts like prefixes or suffixes; instead, it wrote whole syllables as single blocks. This means that the tiny grammatical markers that created new words have largely vanished from the text, leaving modern scholars to guess at the rules of word formation based on reconstructed sounds rather than direct observation. The question becomes: how can we know if two ancient words are related if the written evidence of their connection is missing?

For centuries, scholars have looked to a specific feature of the ancient Chinese dictionary, the Shuōwén jiězì, compiled around the year 100 CE, for answers. This dictionary contains a special label, yìshēng, which translates to "also phonetic." It appears next to certain characters that are built from two parts: one part that suggests the meaning and another that suggests the sound. Usually, the sound part is just a sound helper. But when the yìshēng label is present, it indicates that the sound part is doing double duty: it is providing the sound and contributing to the meaning. For example, a character might be built from a component meaning "guest" and a component meaning "to lay out the dead," with the label suggesting that the "guest" part is also the sound source. The big question has always been what this label actually tells us. Does it simply mean the two words share a meaning? Does it mean they are the same word written differently? Or does it mark a specific grammatical process where one word was derived from another?

A recent study by Yusen Wu and Weiyi Qu set out to answer this by treating the ancient label not as a mystical insight, but as data that could be tested against the rest of the language. The researchers gathered every instance of this label in the dictionary and compared them to similar characters that did not carry the label. They looked at how the sounds of these characters changed over time, using both the recorded pronunciations from a later period and modern reconstructions of the ancient sounds. They also had a computer program, acting as a blind observer, read the definitions of the characters to determine if the meanings were truly connected, without knowing which characters had the special label. This approach allowed them to see if the label was actually picking out a specific pattern of word relationships that other characters in the same sound groups did not share.

The results were clear and surprisingly specific. The study found that the label is a very strong indicator that two words are related in meaning. When the researchers looked at the characters with the label, they were far more likely to have a clear semantic connection to their sound source than the ordinary characters in the same groups. However, the label does not tell us the direction of that relationship. It does not say which word came first or which one was derived from the other. It simply confirms that the two words belong to the same family.

More importantly, the study revealed that the label is not just about meaning; it also tracks a very specific type of sound relationship. Among the words that are related in meaning, the ones with the label are much more likely to be either identical in sound to their source or to differ only by a specific tone change that corresponds to an ancient suffix. This suggests that the ancient analyst who created the label was often identifying cases where a word had been slightly modified—perhaps by adding a suffix to change a verb into a noun, or to create a new sense of the same word—and then given a new written character to reflect that change. The label marks these "minimal derivatives" or cases where a single word has been split into two written forms.

The researchers also discovered that the label's reliability varies depending on which version of the dictionary is being read. The version most likely to contain the original judgments of the dictionary's author, Xu Shen, shows a weaker connection to sound than later versions. In the later versions, the label appears more often on characters that are identical in sound to their source, suggesting that later editors may have applied the label more loosely, perhaps focusing more on the sound match than the deeper meaning. This finding helps separate the original insights of the ancient scholar from the additions of later scribes.

Ultimately, the study concludes that the yìshēng label is a valuable tool for understanding how ancient Chinese words were related, but it is not a complete guide to how they were formed. It successfully identifies when two words are part of the same family and often points to the specific type of sound change that linked them, particularly those involving a tone shift. However, it does not reveal the direction of the derivation, leaving scholars to determine which word came first using other methods. The research demonstrates that even without overt grammatical markers, a native analyst's classification can reveal deep patterns of lexical relatedness, provided it is tested against a broad and independent set of evidence. The label is not a magic key to the entire history of word formation, but it is a precise marker of a specific kind of connection: the moment a word was recognized as being built from another, and written down with a new face.