The Full Wiki

More info on Tsou language

Tsou language: Map


Wikipedia article:

Map showing all locations mentioned on Wikipedia article:

The Tsou language is the Austronesian language of the aboriginal Taiwanese Tsou people.


The dialectal variation of Tsou is not great. There are four recorded dialects, Tapangu, Tfuea, Duhtu, and Iimcu, of which Tapangu and Tfuea are still spoken. Iimcu is not well described. The grammar of the other three dialects is nearly identical, and phonological variation is marginal: In certain environments, Tapangu /i/ corresponds to Tfuea and Duhtu /z/ or /iz/, and Duhtu had /r/ for Tfuea and Tapangu /j/. (Actually, older speakers were recorded to vary between [r] and [j], but at that point the dialect was moribund.)



Tsou has six vowels, . Vowel sequences occur, including sequences of like vowels ( etc.), but these are separate moras rather than long vowels or diphthongs. Vowels, especially back vowels, are centralized when flanked by voiceless alveolar consonants ( ). This may involve a central offglide, so that is pronounced as a diphthong or in this environment.


labial alveolar velar glottal
nasal m n ŋ
plosive p t k ʔ
implosive ɓ ɗ
affricate ts
fricative f v s z h
approximant w j
The approximants and may surface as non-syllabic mid vowels and , even (for ) in initial position ( "fishes"; does not occur in initial position), explaining the spelling Tfuea ( ) for the name of the dialect. However, stress assignment ( ) and restrictions on consonant clusters (see stress and phonotactics below) demonstrate that they behave as consonants.

The plosives are not aspirated. Phonetically aspirated stops are actually sequences of stop plus , as can be seen by the fact that they cannot cluster with a third consonant (see phonotactics below), and by morphological alternations such as ~ "to trade".

According to spectrum analysis, appears to be a glottal fricative in most environments, but approaches a velar next to the central vowel , as in 'palm, sole'. However, the fact that the sequences and occur, when no other homorganic sequence is allowed, suggests that and may not both be glottal. (Additional evidence that might best be analyzed as velar is the fact that is not found, and that is only found medially, in the single known word "fox".)

The voiceless sibilants, and , are palatalized to and before the front vowels and . However, the voiced sibilant is not affected by this environment.

The implosives and are uncommon. Both may be glottalized ( or maybe ) in intervocalic position. In addition, alveolar has some unusual allophony: About a third of speakers pronounce it with a lateral release, or before as a lateral approximant , as in "maple". Indeed, Tsuchida (1976) transcribed it as a preglottalized lateral, .


With a few exceptions, stress is not only predictable, but shifts when suffixes are added to a word. It falls on the penultimate vowel, or on the penultimate mora if a moraic analysis is adopted. That is, a final heavy syllable (double vowel) receives stress ( "house"); otherwise, stress falls on the penultimate syllable ( "his child"). Additional stress falls in a trocheicmarker pattern: Every other light syllable (single vowel) also receives stress. Unstressed vowels are deleted, except at word boundaries (initial or final vowel) and unless doing so would create a forbidden consonant cluster (see below).

For example, the verb "to cut with a bolo" takes stress on the syllables and , and is realized as . However, this does not explain all consonant clusters, many of which are lexically determined.


The most complex syllable in Tsou is CVV. Tsou is unusual in the number of consonant clusters that it allows. Homorganic clusters are not allowed, unless one is a nasal consonant, and a maximum of two consonants may occur together, but otherwise about half of possible sequences are known to occur. For example, all non-homorganic sequences starting with /t/ and /ts/ are found. Missing clusters may not be allowed, or simply accidental gaps due to limited knowledge of the lexicon.

In clusters of oral stops, both have a release burst. This is true even between vowels, an environment where the first stop is unreleasedin most languages, supporting an analysis of these clusters as part of the syllable onset, with not syllable codasoccurring in the language.

Stops, oral or nasal, may or may not have a release burst before a nasal stop, depending on the speaker. The initial clusters are unusual cross-linguistically. The spectrum shows that the tongue moves towards an alveolar articulation during the of , demonstrating that it is not articulated as a velar. The initial clusters and are sometimes realized as two released stops, but sometimes with a single release, resembling ejective consonantsin other languages. ( is again notably missing, except intervocalically, despite the fact that is the most common ejective cross-linguistically.)



  • Richard Wright & Peter Ladefoged (1994). "A phonetic study of Tsou". In UCLA Working Papers in Phonetics 87: Fieldwork Studies of Targeted Languages II.

Further reading

  • Tsuchida, K. (1976). Reconstruction of Proto-Tsouic phonology. [Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa, Tokyo Gaikokugo Daigaku.

External links

Initial or medial Medial only

Embed code:

Got something to say? Make a comment.
Your name
Your email address