- Tongan Accent1
This paper examines previous accounts of Tongan accent-especially those that have attempted to predict accent placement by using either syllable count or morpheme boundaries. Both means are shown to be inadequate. An alternative is suggested-that of positing the measure as a unit of accent, a prosodic unit between the syllable and the phonological phrase.
C. Maxwell Churchward (CMC), an astute observer of several Oceanic languages and a prolific writer of grammars and dictionaries, had an unusual method of organizing his grammars. For example, he began his description of Rotuman (1940) with a "general survey," then moved to sounds and inflections in detail, and finally treated "supplementary details." For Fijian (1941), he refined the system to just two parts: "general survey" and "additional details," and for Tongan (1953), he approached the problem of an overall verb classification by dividing the topic into a chapter on "verbs, subjects, and objects," "more about verbs," and "still more about verbs." Therefore, it comes as no surprise to find that in his Tongan grammar he divided the topic of accent into "the fundamental facts" and "some irregularities."2 In this paper, I shall try to unite these two categories into one description.
CMC's "fundamental facts" about Tongan word accent can be summarized as follows: The penultimate vowel or a final long vowel is accented. In a word consisting [End Page 307] of a single long syllable, its vowel is accented. Certain grammatical markers that consist of only one short syllable are accented if followed by another short marker.3
In short, according to CMC's description, Tongan has a perfectly regular and predictable accent system, so long as one deals only with words in isolation that are no longer than three short syllables, or one short syllable plus a long one. Moreover, because this part of the description avoids the problem of how to treat diphthongs, it is easier for the analyst if the words do not have two vowels in succession. Thus, the "facts," as given by CMC, fit such words as those in the two categories shown in table 1. To simplify this treatment further, some analysts (e.g., Feldman 1978) view a long vowel as a sequence of two short vowels, hence giving, for example, fále and fáa (fā) the same syllable structure and merging the two lists in table 1 into the first type: (CV)CVCV.4
2. The Problem.
A list of words of the shapes shown in table 1 would be of considerable length, but would certainly not cover the total vocabulary of Tongan.5 What of the other types-those that constitute the "irregularities" of accent? Not unexpectedly, they cannot be treated so simply. The following categories of exceptions or irregularities are based on those listed by CMC (pp. 4-10):
1. Certain sequences of vowels act as units in the assignment of accent. Accent is said to occur on the whole sequence, or on the first vowel, when the second vowel is in penultimate position.
2. Some words have more than one accent: compounds, fully reduplicated words, words with an affix longer than one short syllable, words with more than one long vowel, and three-syllable words in which the first or last syllable is long.6 Finally, although CMC did not mention this type specifically, one should add to the list any word that has four or more short syllables, whatever its morphological structure.
Click for larger view
View full resolution
[End Page 308]
3. Accent is affected when a word is followed by certain one-syllable grammatical markers, termed enclitics by CMC.
4. Accent can occur on the final (short) vowel to mark a type of definiteness.
In the next section, I discuss CMC's first three irregularities, along with related matters in four other descriptions of Tongan accent: Morton 1962, Clark 1974, Feldman 1978, and Krupa 1982. Because the phenomenon listed above as irregularity (4)-definitive accent-operates somewhat differently, I shall treat that later, in section 6.
The first irregularity is fairly straightforward. Certain sequences of vowels...