Phonemic variation

Although many speakers operate with the consonant inventory just presented, there is considerable individual (and regional) variation in the maintenance of this set of contrasts with particular words. No comprehensive regional study of phonological diversity has ever been carried out on Bislama, nor has there been any empirically-based quantitative study of phonological variation. Phonological variation is also often related in informal comment locally to an individual’s language of education – whether one is considered to be ‘anglophone’ or ‘francophone’ – though such comments have once again not been subjected to detailed empirical scrutiny.

It is difficult to present statements which cover all possibilities regarding variation from this basic pattern of consonantal contrasts given that there is a fairly extensive range of possibilities. The following general observations can be made about the loss of phonemic contrasts vis-à-vis the basic consonant inventory, though it should be recognized that some additional phonemic mergers may be encountered among small groups of speakers, or in particular lexical sets with some speakers:

(i) The contrast between voiced and voiceless stops is not consistently made. For some speakers, there appears to be little contrast at all, with only voiceless unaspirated stops found in all environments. It is far more common, however, for a contrast to be made, but for the contrast to be lost with some words. That is, while some speakers may contrast /dok/ ‘dog’ and /tok/ ‘talk’ on the one hand and /draim/ ‘dry (something)’ and /traim/ ‘try’ on the other, other speakers may merge /dok/ and /tok/ as /tok/ while maintaining a contrast between /draim/ and /traim/, and yet other speakers may merge /draim/ and /traim/ as /traim/ while maintaining a contrast between /dok/ and /tok/. If any merger takes place, it is most likely to be in the direction of the voiceless stops rather than the voiced stops.

(ii) The contrast between /v/ and /f/ is also not very stable. The /v/ segment is not nearly as widely distributed as /f/ in any case, and some speakers lose the contrast entirely, having only /f/. This results in alternations such as /vanuatu/ and /fanuatu/ ‘Vanuatu’ within the speech community.

(iii) For many, perhaps even most, speakers, the contrast between voiced and voiceless stops is lost in homorganic nasal-stop clusters, this time in the direction of phonetically voiced segments. Thus, while for some speakers there may be a voicing difference in pairs such as /stampa/ ‘base (< from English stump)’ and /namba/ ‘number’, most people pronounce /stamba/ and /namba/ respectively.

(iv) A small minority of speakers may go further than this in tending to lose the contrast between voiced and voiceless stops and homorganic nasal-stop clusters, pronouncing all as voiced prenasalized stops, particularly in word-initial position. Thus, a word that will be pronounced by many as /pik/ ‘pig’ may occasionally be encountered as /mbik/.

(v) There also is a substantial amount of unpredictable alternation between voiceless stops and the corresponding voiceless fricatives, with /pik/ ‘pig’ and /faia/ ‘fire’ occasionally being heard as /fik/ and /paia/ respectively. This kind of alternation is strongly stigmatized with some words, but quite widespread with others.

(vi) There is a tendency for the distinction between /c/ and /s/ to be lost among some speakers, or with some words, resulting in alternations such as /calus ~ salus/ ‘jealous’ and /cenis ~ senis/ ‘change’.

(vii) The glottal fricative /h/ is often lost. This is especially frequent intervocalically with pronunciations such as /biain/ ‘behind’ being far more common than /bihain/, though it can also be lost word-initially, resulting in not-in-frequent alternations such as /harem ~ arem/ ‘hear’. (Note that /h/ is never found word-finally in Bislama.)

Given that for the vast majority of speakers, Bislama is acquired after the acquisition of one of 80 or so local vernaculars in childhood, these kinds of phonological mergers, as might be expected, correspond to some extent to the distribution of particular features in the substrate languages. It has been noted, for example, that in a number of languages from the island of Malakula, while there is a prenasalized /^mb/ phoneme, there is no correponding plain voiceless /p/, and it is precisely with speakers of such languages that more widely distributed pronunciations such as /pik/ ‘pig’ are encountered as /mbik/. The stigmatized retroflex flap articulation of /r/ that was mentioned earlier also appears to correspond closely to the distribution of retroflex rather than alveolar flap realizations of /r/ in local vernaculars, particularly those of northern Efate and parts of Pentecost island.

However, having pointed to a correlation between such variations from the basic phonological pattern described above and differences between local vernacular phonologies, we should exercise some caution in assuming that all regional phonological variation shares the same explanation. Not only do we have an inadequate knowledge of the distribution of variants to this basic phonological system of Bislama, but we have a detailed knowledge of the phonologies of only a small number of vernaculars (Lynch and Crowley 2001: 14–19). Even with the limited knowledge that we do have, it is not difficult to point to features of vernacular phonologies which are not carried over into Bislama. In the Paamese language, for instance, there is word-final neutralization of the contrast between /p/ and /v/ with phonetic free variation between stop and fricative realizations, though this does not seem to correspond to any tendency among speakers of Paamese to loose their contrast between the stop and fricative word-finally when they are speaking Bislama.

In addition to the kinds of phonological mergers just described, there are speakers who operate with somewhat expanded consonant and vowel inventories, at least for some words. This seems to correspond to a considerable extent to a higher command of English or French. With such speakers, we tend to find that not only is the contrast between /s/ and /c/ maintained, but there is also a tendency to distinguish between /s/ and /ʃ/ in words of English or French origin. Thus, in contrast to the majority pronunciations of /sup/ ‘soup’ and /sus/ ‘shoe’ we may encounter /sup/ and /ʃus/ respectively.

There also appears to be a tendency among better-educated speakers for the contrast between long (or diphthongized) and short (monophthongal) vowels in English – which is ordinarily completely lost in Bislama – to be maintained in the form of a tense-lax distinction. Thus, while /set/ for many speakers is the pronunciation for ‘shirt’ and ‘agreed’ (< set), some speakers may make a contrast between /sεt/ ‘agreed’ and /set/ ‘shirt’. It should be pointed out, however, that as far as I am aware, such an observation has not been offered in any previously published account of the language and study needs to be carried out by a well-trained phonetician to verify (or disconfirm) this.

Another area of phonemic uncertainty involves the relationship between vowel quality and phonemically contrastive voicing with stops in word-final position in words of English origin. It was indicated above that there is no contrast in Bislama word-finally between /p, t, k/ on the one hand and /b, d, g/ on the other, with minimally contrasting pairs in English ending up as homophones in Bislama. Although I am fairly confident that there is indeed no final voicing contrast in Bislama, it may be worth investigating the possibility that there may be some kind of surviving contrast in nature of the preceding vowel. My suspicion is that there may be some kind of acoustically detectable laxness in the vowel of forms such as /pik/ ‘pig’ in contrast to a more tense vowel in /pik/ ‘plectrum (< pick)’. Such a test would need to be carefully constructed so that it is based on natural pronunciations without any possibility of contamination from spelling pronunciations.