Variations in the core phoneme inventory

Superimposed on this idealized paradigm is considerable variation. Firstly, since the majority of the lexicon is derived from English, and Tok Pisin speakers come into increasing contact with Standard English, the likelihood of influence from English phonology is great. As Laycock has pointed out (1985: 25), Tok Pisin speakers familiar with English have potentially the whole of the English phoneme inventory at their disposal. More will be said about the relationship between the two languages below.

Another source of variability among speakers of Tok Pisin as a second language is the influence of the phonologies of other languages spoken. A pioneering study by Bee (1971) showed that this influence can be considerable. She, however, was describing a situation where Tok Pisin had been fairly recently introduced and its use marginal. Influence from substrate phonology is likely to decrease as speakers become more familiar with mainstream patterns of use.

Nevertheless, Laycock (1985) has shown that there is considerable influence from phonological patterns of other Papua New Guinea languages. The picture is undoubtedly very complex, as the 800−900 languages spoken within the country vary enormously in their phonology, so a consistent pattern of influence is only likely to be felt where there are widespread regional patterns. For example, intervocalic pre-nasalization of voiced stops is widespread in many of the languages of the Sepik and Madang areas so far described, and this feature may appear in the Tok Pisin spoken of this region. Similarly, unvoiced stops may be realized medially as fricatives in many Highlands languages, and this feature may be preserved in the Tok Pisin of some speakers in the Highlands region.

While these variations can be observed among second language speakers, the situation among first language speakers has received less attention. Romaine (1990) looks at the variation between /p/ and /f/ in initial position by young people in Morobe and Madang provinces, including first language speakers. Greater consistency in the correspondence between initial /f/ in Tok Pisin and their English equivalents is shown to be related to such factors as urban or rural status. In rural areas, such as Indagen, there was greater variation, and some interesting observations are made, including instances of hypercorrection where initial /f/ was used on words derived from English words beginning with /p/. Smith (2002) also observes that the distinction between /p/ and /f/ is often inconsistently applied, especially in Highlands speech samples, but also in other areas such as Sepik and North Solomons, with such items as pik ‘pig’ rendered as fik.

In another study on a single phoneme pair, Romaine (1995) discusses discrimination of the phonemes /r/ and /l/ in the same corpus of speech from Morobe and Madang Provinces used for the study of /p/ and /f/. She again relates the use of this distinction to urban and rural status, but does report that first language urban speakers are more consistent in distinguishing /r/ from /l/, even though there is considerable variation. Among the rural speakers, those in Waritsian village were most likely to confuse the phonemes, which could be due to substrate influence, as the Adzera language does not distinguish these two sounds. Smith (2002) also found some first language speakers showing quite marked variability with regard to this contrast. In the following extract from Eastern Highlands, for example, the expected forms lized ‘lizard’, long ‘to’, stilim ‘steal’ and lapun ‘old woman’ all appear with /r/ substituted for /l/, while /l/ replaces /r/ in rere ‘ready’:

(1) em i kam araun ro disa, a kam araun ro disa haus na stirim disa kiau blo rized na ranawe pinis. Em kukim i stap na leli lo(n)gen. Em stirim na go pinis na disa rapun meri i kam bek.

‘he came around to this house and stole the lizard’s eggs and ran away. He cooked them and got them ready. He had stolen them and taken them away when this old woman came back.’

Smith (2002) also found that the contrast between voiced and unvoiced stops was often inconsistently made by first language speakers, especially from the Highlands region, but also in other areas, particularly Manus and West Sepik. This leads to words such as pik ‘pig’ being heard as fik, antap ‘on top’ and paitim ‘hit’ as andap and paidim and liklik ‘small’ as liglig. In the case of velar stops, such words as pik ‘pig’ and dok ‘dog’ may thus have a pronunciation closer to the English source as pig or dog. There is also occasionally a tendency to pre-nasalise medial voiced stops, giving forms such as gondaun in place of godaun ‘go down’. Also typical of some Highlands speakers is the tendency to voice /s/ to give forms such as dizla for disla ‘this’. In some areas, the voiceless alveolar stop /t/ and spirant /s/ may alternate, especially in certain words such as sapos/tapos ‘if’.

While these features serve to give particular accents to speakers of various first languages, Laycock (1985: 304) notes that there is such internal diversity in all provinces that distinct regional accents are not likely to emerge. Although such variation generally is not so marked as to make comprehensibility a problem, he notes that the lack of a distinction between /t/ and /s/, carried over into Tok Pisin from many languages in New Ireland, New Britain, the Highlands and South Bougainville is actually “disturbing to communication” Laycock (1985: 302). More generally, peculiarities of pronunciation provide the basis for many jokes at the expense of less fluent Tok Pisin speakers, for example the humorous stories featured in the “Kanage” column of Wantok Niuspepa. This ridiculing of strongly nonstandard features is cited by Laycock (1985: 304) as another reason why distinct regional accents are unlikely to emerge.