Constraints on the onset: /j/ and /w/

The composition of the onset may be constrained by the nature of the vowel(s) occupying the nucleus. We saw previously on JamC vowels that vocalic sequences /ui/ and /iu/ do not occur. Put another way, however, phonetic [iu] and [ui] sequences are only possible provided the initial vowel in the sequence occupies a C-slot, i.e. functions as the semi-vowels /w/ and /j/ respectively. These produce the phoneme sequences /wi/ and /ju/. This is demonstrated in the following examples.

There is an uneasy relationship between /j/ and /w/ on one hand and their vocalic equivalents, /i/ and /u/, on the other. The occurrence of semi-vowels in the onset is subject to a constraint which follows from their relationship with vocalic segments. Underlyingly, syllable onsets tend not to consist of a semi-vowel as the sole consonant, followed immediately by the vowel which is its vocalic equivalent. This blocks underlying sequences such as */ji/ and */wu/. Where [ji] and [wu] sequences do occur phonetically, it is in variation with a form without the initial semi-vowel, e.g. [unu] ~ [wunu] ‘you (plural)’, [jimba] ~ [imba] ‘a yam variety’ (Cassidy and Le Page 1980: 225, 457).

In JamC, the onset may have a maximum of two consonants. In such combinations, the first item is always an obstruent and the second an approximant. Combinations with /w/ as the second consonant are /pw, bw, tw, dw, kw, w, sw/. Those involving /j/ include /pj, bj, tj, dj, kj, j, fj, vj, sj, mj, nj/. Of these, /tj/, /dj/ and /sj/ do not have transparent realizations at the surface level. The matching phonetic forms, *[tj], *[dj] and *[sj] are blocked, in spite of a contrary suggestion by Wells (1973: 21). They may be blocked because the underlying phoneme sequences /tj/ /dj/ and /sj/ have their surface phonetic manifestations merged by speakers with those of the affricate and fricative consonant phonemes, /ʧ/ , /ʤ/ and /ʃ/ . Both sets of sequences become realized phonetically as [ʧ] , [ʤ] and [ʃ] respectively. The fact is, however, that the consonants /ʧ/ , /ʤ/ and /ʃ/ also occur in the coda, e.g. /matʃ/ ‘match’, /ʤoʤ/ ‘judge’, /jjaʃ/  ‘cash’. This establishes that [ʧ] , [ʤ] and [ʃ] can and do represent the consonant phonemes /ʧ/ , /ʤ/ and /ʃ/ rather than just underlying /tj/, /dj/ and /sj/. We suggest nevertheless that in the onset, speakers do treat [ʧ] , [ʤ] and [ʃ] as representing a merger at the phonetic level between [ʧ] , [ʤ] and [ʃ] , on one hand, and /tj/, /dj/ and /sj/ on the other.

The only consonants occurring in the JamC onset which are blocked from occurring before /j/ are /l/, /r/, /z/, /ʃ/ , /ʧ/ and /ʤ/. Given the position of /l/ and /r/ in the sonority hierarchy, we may regard them as sonorant consonants which, like /j/ and /w/, only occur in second position in the onset. An onset */zj/ cluster fails to occur because it cannot be phonetically reinterpreted. The expected form, * /Ʒ/, does not exist as a phoneme in JamC. The blocking of */ʃj/ , */tʃj/ * /ʤj/and are, we would suggest, the result of the unacceptability of the alternative /sjj/, /tjj/ and /djj/ underlying representation. These would require a */jj/ sequence. The analysis is presented below.

The apparent occurrence of /dj/ on the surface as in /djam/ ‘damn’ really involves a disyllabic sequence /dijam/, with prominence on the second syllable.