Comment on Geoffrey Sampson, “A Chinese Phonological Enigma”

The author explores various ways to resolve the paradox: language change avoids creating excessive homophony, while sound changes in the history of Chinese have created a massive level of homophony. In this comment, I consider why homophones occur even though humans try to manifest one-to-one correspondence between form and meaning based on the CELEX lexical database of English, version 2.5 (1995) and the evolution of diatones in English, and its implication to Chinese. Table 1 shows the number of homophones classified according to the syllable number in columns and the number of words in a homophone set in rows. The total number of types of words classified according to the syllable number is also given in parentheses. We find that 11,980 or 22.8 % of 52,447 words are homophones in Present-day English. The number of homophones decreases as the number of words in a homophone set increases, and the percentage of homophones for each syllable number decreases as the syllable number increases. Table 1 suggests a threshold of homophones in English that can be tolerated. We also find that 4,743, or 70.2 % of 6,761 one syllable words and 4,509, or 27.3% of 18,564 two syllable words are homophones, and they form 77.2% of all homophones.

Table 1.

Number of homophones in English (based on Ogura and Wang 2006)

Reprinted by the permission from De Gruyter

Table 2 shows the average word frequencies of homophones and all words classified according to the syllable number. Table 3 gives the average word frequencies ranked from the most frequent to the least frequent in the CELEX database, whose frequency information is taken from the 17.9 million words corpus.

Table 2.

Average word frequencies of homophones and all words

Table 3.

Average word frequencies ranked from the most frequent to the least frequent in the CELEX database (based on Ogura and Wang 2006)

Reprinted by the permission from De Gruyter

We find that within each syllable number, average word frequency of homophones is higher than all words except the 6 and 7 syllable words. The average frequencies of homophones in 1 and 2 syllable words are in the rank of 1001-2000 and 4001-5000 respectively. We may state that homophones of 1 syllable and 2 syllables form the majority of homophones and they are in the top 5000 most frequent words. We further find that in the top most frequent 5008 words, 2,432, i.e., 48.6% of them are homophones, and in the rest 47,439 words, 9,548, or 20.1% are homophones.

Why do homophones occur even though humans try to manifest one-to-one correspondence between form and meaning? Zipf (1949) suggests the simultaneous minimization of the two opposing forces from listener and speaker for form and meaning associations. One form for all meanings is an ideal code for the speaker, while one-to-one correspondence between form and meaning is an ideal code for the listener. Zipf’s law, which states that word frequencies decay as a power law of its rank, is the outcome of form-meaning associations adopted for complying with listener and speaker needs. Arranging signals according to Zipf’s law is the optimal solution for maximizing the referential power under effort for the speaker constraint. Thus we may assume that humans try to manifest one meaning, one form to avoid creating homophones, but at the same time they try to maximize referential power under effort for the speaker constraint, using a portion of words frequently and even forming homophones with high frequency words.

Homophony is a desirable feature in communication system and likely results from ubiquitous pressure from efficient communication (Ferrer i Cancho and Ricard V. Solé 2003, Piantadosi et al. 2012). Mandarin Chinese has only about 410 distinct syllables, and there are over 80,000 Chinese characters. But actually, only 1,000 characters occur in 90% of spoken and written Chinese (Matthew...