Abstract
Abstract
Where morphology is largely absent from the written record, derivation has to be inferred from evidence. Old Chinese is such a case: its affixes are reconstructed rather than observed, and its script writes whole syllables, so no affix is written on its own. This study asks whether a native analyst’s classification can serve as evidence, and for which part of derivation. It takes the yìshēng 亦聲 (“also phonetic”) label of the Shuōwén jiězì (100 CE), which marks compounds whose phonetic component also carries meaning. Labelled characters are compared with ordinary phonetic compounds on the same phonetics, using Middle Chinese readings, Old Chinese reconstructions and a blind semantic coding. The label tracks relatedness of meaning. Among related pairs it shows a moderate preference for the same word or a minimal derivative, notably pairs that differ only in the tone reflecting the suffix *-s. It says nothing detectable about which member is derived. The sound effects are weaker in the labels most likely to be Xu Shen’s own, whereas the contrast in meaning is not. It follows that a native classification evidences lexical relatedness and formal closeness, and that where morphology is covert, relatedness, formal relation and direction must be diagnosed separately.