In his brief "Notes on Dating the Cantonese Dentilabialization of Middle Chinese Gutturals" in JAOS 98.1 (1978), William Boltz provided data to support his proposal of both the approximate time and the sequence of dentilabialization of the Middle Chinese (MC) initials /k^h/ and /X/, qi mu ·Ë ¥À and xiao mu ¾å ¥À respectively, in (Standard) Cantonese. The source of his data was a text compiled ca. 1750 in Macao called the Ao men ji lue ¿D ªù °O ²¤ (Sketch Notes on Macao), which contains a Sino-Portuguese glossary with 395 entries. Boltz¹ and his predecessors² who studied the glossary correctly identified the dialect used in the phoneticization of the Portuguese words to have been a Yue³ dialect, and had proceeded to romanize the transcription using Cantonese. What is important to realize is that the dialect used in the transcription was not merely any Yue dialect, for although the choice of the standard dialect, Cantonese, is sufficiently close to aid in 'reconstructing' the original Portuguese words, it is not precise enough for postulating certain historical sound changes in Cantonese. The latter task must assume that the two dialects undergo similar sound shifts.

The dialect used in the phoneticization was, in fact, Zhongshan, and it is precisely the difference in the phonological systems of Zhongshan and Cantonese which invalidates Boltz' tentative conclusions on the dating of Cantonese dentilabialization. The claim that the dialect was Zhongshan can be substantiated on several grounds. First of all, geographically, and probably historically, Macao was part of the Zhongshan dialect, even though politically, it has been occupied by the Portuguese since the sixteenth century. The description of "The Ho_ngshan or Macao dialect" by J. D. Ball⁴ before the turn of the century provides a second reason for considering the dialect of a century and a half earlier in Macao to have been Zhongshan as well. It might be noted here that "Ho_ngshan" and "Zhongshan" refer to the same district. The name of "Ho_ngshan" was changed to "Zhongshan" in honour of Dr. Sun Yat-sen, better known in China as Sun Zhong-shan. A third and final basis for suggesting Zhongshan is derived from the glossary itself. A special feature of Zhongshan is the loss of the distinction between labialized and non-labialized velar plosives (i.e., /k/ vs. /k^w/ and /k^h/ vs. /k^hw/) found in Cantonese. Ball's article shows a total loss of labialization in Macao in contrast to Shekki, the capital of Zhongshan district, where maintaining the distinction or merging the segments occurred in free variation. Both Y.R. Chao's⁵ data on the Shekki, or standard Zhongshan, dialect and the material collected by the writer on the same dialect demonstrate a preservation of the distinction only before the low vowels, /a/ and /Œ/ and before the final /Ik/. In the Sino-Portuguese glossary, both the syllables /k�/ and /k^w�/ in Cantonese were used to transcribe non-labialized segments in Portuguese. Thus, although the characters ­ô¡A­Ó¡A ºq (Cantonese /k�/) were used most frequently to represent Portuguese /ko/, the characters ªG in entry number 45 and ¹L in entry number 228 (Cantonese /k^w�/) were used for the same purpose, thereby suggesting a loss of labialization in the dialect used for the phoneticization⁶.

Having now established the dialect to have been Zhongshan, one difference noted by Ball between Cantonese and Zhongshan which is significant to our present discussion is that the dentilabial /f/ in Cantonese corresponds to laryngeal /h/ in Zhongshan before /u/ and /u@/, that is, before [U] and [u], and /f/ elsewhere⁷, with a few isolated cases of the zero or velar plosive initials. Chao's analysis of Zhongshan, based on data collected in 1929 and 1939, and the writer's field notes on the dialect show sound correspondences similar to those observed by Ball. In terms of the historical phonology of Zhongshan, the few words bearing MC /k^h/ which underwent deplosivization and subsequent dentilabialization came from the various he-kou ¦X ¤f, or closed, rhymes of the guo ªG rhyme group, and from the second division only of the closed rhymes of the xie ÃÉ rhyme group. With regards to MC /X/ initial, the dentilabialization process affected the closed rhymes of the guo ªG, jia °², zhi ¤î, zhen ¿² and dang ÌX rhyme groups. As a result of different phonological conditioning between Cantonese and Zhongshan, dentilabialization in the latter occurs in fewer rhymes, affecting only those in which the reflex is not a high back vowel. It is therefore a sound change of more limited scope than that found in Cantonese.

Given the recognition of the particular phonological conditioning in the shift in Zhongshan, a pattern emerges from the Chinese entries for which Portuguese words have been reconstructed. For simplicity, an English translation of the Chinese gloss is given in the table. "Zhongshan" and "Cantonese" refer to the modern dialects. The MC segments are the reconstructed initials for the underlined characters transcribed in the table for the two dialects.

Note first of all that entry numbers 175 and 375 served as the basis for Boltz' inference that the dentilabialization of MC /k^h/ and /X/ was not a simultaneous process in Cantonese, in that the sound shift had already affected MC /k^h/ by mid-eighteenth century, though not yet MC /X/. Boltz then made the tentative proposal that the change of the MC initials to dentilabials may have occurred during the middle decades of the eighteenth century. However, from our observations of the conditioning involved in Zhongshan dentilabialization, it is obvious that other factors are operating. Neither the date nor the simultaneity versus successivity of the process of dentilabialization in Cantonese can be gleaned from the particular examples provided by Boltz. His assumption that dentilabialization of number 175 would occur subsequently is not borne out in modern Zhongshan. Note, furthermore, that numbers 232, 318 and 370 are counter-evidences to his claim that dentilabialization had not yet touched MC fricative /X/ in the eighteenth century corpus. Our re-analysis clearly leads to the conclusion that the transcription of the glossary demonstrates in fact a sound change which had stabilized by mid-eighteenth century in Zhongshan, and a corresponding shift is likely to have had also occurred by that time in Cantonese and other Yue dialects. Since dentilabialization of the MC gutturals predated the Sino-Portuguese glossary, ca. 1750 would therefore serve at least as a terminus ad quem for the shift, as was originally proposed by Boltz, except that the 'rabbit' hopped into the scene.⁸ The rabbit can now be safely scuttled back out without much ado.

MARJORIE K.M. CHAN
UNIVERSITY OF WASHINGTON

8. Cf. Boltz, 1977, p.144, n.5. [BACK]

---

Return to Publications Page.
Copyright (c) 1997 Marjorie K.M. Chan. All rights reserved.
Created 2 December 1997.
URL: http://cohums.ohio-state.edu/chan9/articles/jaos-ipa.htm

---