The segmental and prosodic system of Bumthang is a work still in progress.
The analysis that is summarised in the following tables is an out-of-date account, which has since been superseded, but which does illustrate some of the phonetic distinctions that are important in the Ura dialect of Bumthang. We hope to make a more up-to-date description, and analysis, available in the near future.
There are 20, possibly 26, consonants, and 7 vowels in the language. At a superficial level the following broad phonetic distinctions are made:
|'aspirated'||pʰ||t̪ʰ||tsʰ||tʰ ~ tʴʰ ~ ʈʴʰ ~ ʈʰ||tʃʰʲ||kʰ|
|'voiceless'||p||t̪||ts||t ~ tʴ ~ ʈʴ ~ ʈ||tʃʲ||k|
|'voiced'||b||d̪||d ~ dʴ ~ ɖʴ ~ ɖ||dʒʲ||g|
The categories 'aspirated', 'voiceless' and 'voiced' are not contrastive in the same environment. These labels are only approximations to the fact that there is a contrast in Voice Onset Timing between the plosive series, a contrast that is mediated by the prosody (high or low) that applies to the word. The contrast is not consistently one of voicing; this is explained below. The status of ɸ is unclear, showing prescriptive variation with pʰ, but several lexemes are produced preferentially with a fricative. The 'retroflex' plosives are often purely alveolar, and frequently have a voiceless component to their release, making the contrast with aspiration difficult to perceive (t ~ tʴ ~ tʴ̥ ~ ʈʴ ~ ~ ʈ). In general, the degree of aspiration is not high, and the different VOT contrasts posited in the table above are not always easily contrasted.
It is entirely possible that the retroflex series is in fact a series of underlying clusters, such that "/ʈʴ/" < /t̪/ + /ɹ/, for instance. Similarly the palato-alveolar series might be analysed as clusters consisting of the alveolar consonants + /j/, so that "/dʒʲ/" < /dz/ + /j/. These two analyses are behind the indeterminacy in the number of consonants. Under the cluster-rich analysis the language would have only four plosive series, matching the nasals, and only two 'fricatives', /s/ and /h/.
The vowel height contrasts are well established by pairs such as [go̤] 'wheat' vs. [gɔ̤] 'start, beginning', or [bɻẓi̤] 'smell, odour', [bɻẓe̤] 'female yak', [bɻẓɛ̤] 'measuring container'. What are rounded vowels in other Bhutanese languages are realised as diphthongs in Bumthang. As an example of this, *dŋul 'silver' (in Written Tibetan) is /ŋyy/ in the Kurtoep (a sister to Bumthang), through a regular process of *l > j and subsequent vowel fronting (Michailovsky and Mazaudon 1994). The same lexeme emerges as /ŋoj/ in Bumthang, showing *l > j and and a semi-regular process of vowel-lowering (also described in Michailovsky and Mazaudon). (We should note that van Driem (2001: 909), working with a different dialect of Bumthang, reports "rü" for the bamboo species that we have recorded as [ɹuj], implying that the lack of front rounded vowels might not be universal across the language.
Diphthongs can be formed with one of four offglides: i~j, e, o, or u~w. The height differences in off-glides is often neutralised, but the effects on the nuclear vowel make the correct analysis clear. The differences are important; below we see the 1SG.AGT form of the pronoun; the 1SG genitive is /ŋae/, pronounced [ŋaj] ~ [ŋae̯], with a consistently lower nuclear vowel.
Vowel quality is affected by the register of the word (see below) and the nature of the rhyme. In particular vowels tend to be more lax in closed syllables, and to be raised preceding a final -j or -w. Coronal codas front a vowel, and velar codas or onsets back the low vowel. Examples of these processes are:
|/ae/||[aj] ~ [ae̯]||'who'|
|/ao/||[aw] ~ [ao̯]||'where'|
|/uj/||[ɻẓʊj]||'small bamboo (sp.)'|
|high||high, falling||longer VOT lag||clear ("modal")||long|
|low||low, level/falling||short or negative VOT lag||breathy||short|
This implies that there is a four-way phonetic contrast in initial plosives at each place of articulation, and this is generally true (the exception being that /ts/ and possibly /tsʰ/ do not occur with low register). The realisation of phrase-initial plosives is shown in the following table, which preempts an analysis with a two-way contrast in manner for the plosive segments. In the low prosody there is considerable overlap in the realisation of some of the segments, though given the rest of the places in the word that the prosody can be realised there is no real ambiguity involved.
|Phoneme||high prosody||low prosody|
|'voiceless'||[pʰV]||[pʰV] ~ [pV] ~ [p̬V] ~ [pb] ~ [b]|
|negative VOT lag||coincident VOT||slight VOT lag||high VOT lag|
|[bV̤]||[bV] ~ [p̬V] ~ [p]||[pV]||[pʰV]|
|Low prosody /b/||Low prosody /p/||High prosody /b/||High prosody /p/|
The two prosodies interact to create numerous differences in the segments, as well as differences in pitch height, vowel quality, and phonation. Words in isolation are all pronounced with a falling pitch contour, but the onset of this contour is depressed when the low prosody is found. Vowels are consistently centralised in words with the low prosody, and consonants show the realisations summarised in the table below.
|Phoneme||high prosody||low prosody||low prosody, intervocalic|
|/pV/||[pʰV]||[pV] ~ [p̬V]||[bV]|
One strong argument that register, rather than tone, is the prosody involved in Bumthang is the absence of lexical pitch contours in running speech. Outside elicitation contexts phrasal or clausal intonation dominates any lexical specification for pitch, though phonation and vowel quality remains.
The analysis that is most elegant, and accounts for more of the data than others, involves the vowels and prosodies described above, and posits the twenty underlying consonants shown in the table below.