The morphology and phonology of the nominal domain in Tagbana. Yranahan Traoré
variations. First, stops can be phonemically voiced and voiceless. In some cases, however, the voicing difference is an allophonic realization; see Section 2.5.1. The second allophonic variation in Section 2.5.2 concerns [g]; and its nasal variant [ŋ]. Section 2.5.3 compares [g] and [ɣ] in Tagbana dialects and Mending, a neighbouring Gur language. Section 2.5.4 shows that [j] and [ɲ] stand in an allophonic relationship with each other, and Section 2.5.5 deals with the lengthening of vowels before liquids.
The usual test for the phonemic status of a speech sound is conducted with minimal pairs: two words contrast by virtue of having two different phonemes, everything else being equal. In English, cat and bat are different words with different meanings, because both [k]; and [b] are phonemes. Allophones, by contrast, do not distinguish words: they realize the same phoneme in two particular phonological environments. Whether [k] in cat is realized with aspiration or not ←47 | 48→does not change the meaning of the word. And the same is true when one or both of the stops in this word is glottalized. In these cases, there is allophonic variation. Allophonic sounds are phonetically similar. They are often mutually exclusive or in complementary distribution. Two sounds are said to be in complementary distribution if the distribution of sounds in their respective phonetic environments is such that one never appears in the same phonetic context as the other in the same grammatical category.1
This section deals with the distribution of the voiced and voiceless stops of Fròʔò. In nouns, the initial stop is either voiced or voiceless, and there is no minimal pair between voiced and voiceless stops. It is illustrated in (7) that if a word starts with a specific voiced or voiceless stop, there can be no identical word starting with the inverse value for [voice].
The [±voice] stops are mutually exclusive in the onset position. The same effect is also observed word-medially. Voiceless and voiced stops word-internally cannot create minimal pairs in the same grammatical category. This means that in a structure CV.XV where X is a voiceless stop [p, t, k…] a structure CV.XV cannot be found where X = a voiced stop [b, d, g…] and is in a minimal pair relation with the former structure. Since both voiceless and voiced stops may occur word-initially, no prediction can be made as to which appears in which word.
In verbs, stop voicing has a grammatical role. When in citation form, verb stop onsets are always voiceless. Their substitution with their voiced counterpart ←48 | 49→triggers a change of aspect, from infinitive to past or progressive. Some examples appear in (8), one of them in a sentence.
2.5.2 Allophonic relation between [g]; and [ŋ]
The dorsal stop [g]; can appear between a nasal and an oral vowel in either order; see (9)a–b. It can be found between two oral vowels as in (9)c. The voiced dorsal stop also appears word-initially before any vowel, nasal or oral; see (9)d–e. And it can also occasionally be part of a complex onset (9)f.
The segment [ŋ] can appear word-initially followed by a liquid [l]; or [r] and also by the velar oral [g]; see (10)a–d. It only appears in a nasal environment; and when it is followed by a vowel, the vowel is always nasal (e–f), except for in cases where the vowels do not have nasal equivalents, that is, the vowels [e] and [o] as in the examples in (10)g–i.
←49 | 50→
Both intervocalic distributions are summarized in Fig. 13, based on (9) and (10). Between the two mid vowels [e]; and [o], both the velar and the nasal velar are found, as shown above. The examples in (10)a–c show that nasality is subject to harmony and that it can spread across a segment.
Tab. 13 sums up the environments of the two segments [g]; and [ŋ].
Tab. 13: Environments of [g]; and [ŋ]
| Environments | [g]; | [ŋ] |
| V[+nas] _ V[-nas] | + | - |
| V[-nas] _ V[+nas] | + | - |
| V[+nas] _ V[+nas] | - | + |
| V[-nas] _ V[-nas] | + | - |
A possible analysis could take the phoneme [g]; as underlying. [g] is replaced by [ŋ] between identical nasal vowels.
2.5.3 Free variation of [g];, [ɣ], [ʔ], or [Ø] in some Senufo languages
This section gives a comparison of the use of [g]; and [ɣ] in some Gur languages, particularly Mending and Tafiré, as compared with Fròʔò. In Mending, which is widely spoken in Côte dʼ Ivoire, more especially in the Northern part of the country, where Fròʔò is spoken, [g] alternates with the voiced dorsal fricative [ɣ]. The segment [ɣ] is analysed as a lenized pronunciation of the dorsal velar [g] by Herault and Mlanhoro (1973).
←50 | 51→
| (11) | mɔ̀ɣɔ́ | ‘human man’ |
| sàɣá | ‘sheep’ | |
| nòɣó | ‘intestine’ | |
| tɔ́ɣɔ́ | ‘noun, name’ |
In Tafiré, another neighbouring dialect close to Fròʔò, the dorsal velar stop [g]; is used instead of the glottal stop [ʔ] as shown in (12). The glottal stop in Fròʔò raises a number of questions. In some languages and more especially in the Fròʔò dialect spoken in Katiola, there is a process of fusing and lengthening of identical vowels which are separated by a glottal stop, as in tīʔī ‘tree’ realized as tīː ‘tree’ in Katiola. In Fròʔò, the two morphemes, lexical root and class marker, are separated by a glottal stop that is the onset of the CM.
Where Fròʔò uses [Ɂ], Tafiré prefers [g];, Katiola deletes the segment, and Mending prefers [ɣ]. This variation may be the result of free variation.
2.5.4 Allophonic relation between the palatal glide [j]; and palatal nasal [ɲ]
Consider