- “Vowel Length in Niuean” and Déjà Vu
Even though Niuean and Tongan, although closely related, are considered to be separate languages, when I read “Vowel Length in Niuean” (Rolle and Starks 2014) in the previous issue of this journal, it was clear that the authors’ treatment of the main topic of the article—the patterning of long vowels vs. double vowels—could easily have been applied to the same phenomenon in Tongan. I described this feature and others related to Tongan accent in Schütz 2001 (which is included in the references section of the authors’ article). Apparently the authors missed this section (2001:319–21), even though the heading, in boldfaced capital letters, is: “What accounts for the ‘double vowel’ and the varying accent patterns of diphthongs in certain positions?”
After noting that C. Maxwell Churchward (1953, unexpectedly missing from the references) recommended writing both double vowels and long vowels, even though there was no phonemic contrast, I proposed (2001:319):
Based on the taped data for this study … it seems that a penultimate long vowel in a word said in citation form or at the peak of the phonological phrase (the latter condition is assumed in the former) can have the phonetic form of a double vowel, because there is a pitch change marking the peak [emphasis added]. In the reading of the following example (Pulu and Pope 1979:4), the word ‘eikimaama is at the peak of the phonological phrase, and we hear stair-step intonation on the long vowel, written here as a double vowel.
This description is illustrated by an example of a Tongan phrase, with lines showing the pitch fall on the accented long vowel of the phonological peak, maama:
Compare this treatment with a summary of the authors’ conclusions (Rolle and Starks 2014:294–95):
We have argued that the distribution of long and double vowels is conditioned by the prosodic stress algorithm of the language: double vowels surface when primary stress falls on the second vowel of an identical vowel-vowel sequence, whereas long vowels surface elsewhere. … [L]ong vowels and double vowels are [End Page 308] consistently distinguished from one another by pitch [emphasis added] rather than duration, a main phonetic correlate of stress.1
There are three reasons that my conclusion is slightly more complicated than that of Rolle and Starks. First, I was dealing with the language in context, based mainly on Tupou Pulu’s reading of her own written material (Pulu and Pope 1979), but on other taped sources as well (Schütz 2001:307, fn. 1). This type of data, as opposed to citation forms and short sentences out of context, shows that in addition to their position in the phonological phrase, long vowels are also affected by speech style and speed. Next, I included diphthongs in the description, since a diphthong in that phrase-peak position behaves similarly to a long vowel. Finally, the section in question here was only part of a broader description, which included other aspects of Tongan accent as well, such as definitive accent. I concluded:
I suggest that the articulation and perception of both long vowels and potential diphthongs in penultimate position as sequences, not units, is dependent on two factors. First, the position in which rearticulation is most likely to occur is that of the peak of the phonological phrase. Next, a casual style and faster speed have the effect of reducing the double-vowel effect and—indeed—in shortening the long unit.
My only other comment on this particular topic is that, contrary to Rolle and Starks’ conclusions, I consider that a long vowel is the base form, since a double vowel occurs only in a marked environment. More specifically, a long vowel can occur anywhere in the phonological phrase, but a double vowel can occur only at the phrase peak.
I was pleased to see that the authors took note of my measure approach to accent in some Polynesian languages, and supported my position that accent in longer words cannot be predicted by counting syllables. It was also interesting to read that “Hawaiian long vowels and diphthongs act as single units that...