5. Allophones

My resource mentioned only consonantal allophones for Toisan, some of were touched upon in the section on consonants. A common allophone in Margaret’s speech is the alternation between [w] and [v], the former of which is a phoneme in Cantonese. These segments are in free alternation of one another and are indicative of the influence of Cantonese on the speech of Taishan. The examples are numbers (43) and (44) on the tape:

5.1 Vowel Allophones

Mentioned two sections ago, the segments [e] and [E] as well as [u] and [√] seem to be in perfect contrast to one another. This should hint at allophony. In the section on vowels, I chose to classify these segments as discreet phonemes, but here I am changing course and argue that the underlying vowels for the pairs above are [e] and [u].

Rather than discreet allophones, however, it seems that these phonemes alternate over a continuum. I give three examples from the section on vowels to illustrate this assertion. These examples are not ordered according to the tape.

It seems that continuum for [e] may move as far as [Œ] (the first articulation of (50)). With a filled and unvoiced coda spoken very quickly, the vowel seems to be [Œ]. With an open coda, it is [e]. With a nasal coda, it is somewhere in between. The alternation may indeed be more discreet than I have presented it to be, but with the range of data I have observed, the alternation appears to be relative to the nature of the coda.

5.2 Allotones

After looking over the data collected in this project, I feel that the two rising tones Toisan may actually be alltones of a single underlying rising tone: [35]. The examples below show distinct phonetic qualities and both words show a very similar contour (although I failed to notice length in example (40) below, as for all others of this tone).

The allotonal split seems to be across syllables with voiced or unvoiced onsets. The vowel seems long in both cases. For syllables with unvoiced onsets, creaky tone is applied. Possibly because of creakiness, the long vowel appears “broken” by a glottal stop [?]. The syllable with a voiced onset lacks overt creakiness, but in certain contexts this tone seems to be creaky and low, or even falling (see (29) below and the last word of (54) in the next section).

When discussing her native intuitions, Margaret expressed some preference towards considering example (29) (“crooked,” [me˘³⁵]) as having a double, rather than long, vowel: [mee³⁵]. Margaret also felt the word to be a single syllable, in spite of considering its vowel to be doubled. This evidence may give insight into why the vowels of syllables such as (40) appear to be broken, but I don’t have enough data or knowledge of metrics to posit a credible hypothesis on this subject.