Kabiye is a Gur (Voltaic) language that belongs to the Eastern Grusi (also Gurunsi) sub-branch of Central Gur, which in turn sub-branches from Gur that is part of the greater Niger-Congo language family (Naden Reference Naden and Bendor-Samuel1989: 147). The number of native Kabiye speakers is estimated at 730,000, with approximately 700,000 speakers living in Togo, 30,000 in Benin, and a very small number in Ghana (Grimes Reference Grimes2000: 242). The original homeland of the Kabiye people is in Northeastern Togo, with the region of Kozah serving as the community's cultural center. From the early 20th century, however, the community began to spread southward and now encompasses locations in the central and southern parts of Togo, as well as in the neighboring countries (Ali-Napo Reference Ali-Napo1997: 20).
Regional varieties of Kabiye have been grouped into four types according to their geographic centers: Kɛˋwɛ, Laˊˌmaˊɖɩˊsɩ, Lɩgbaˊ, and Laˊmba (Delord Reference Delord1976: 3). As a language with a limited literature and only a recent history of writing, no one variety of Kabiye has been clearly established as the standard. However, considered the most prestigious variety of the language, Kɛˋwɛ, which is also the variety with by far the largest number of speakers, functions as the de facto provisional standard, representing Kabiye as a whole to the outsider. Therefore, it is also often learned and adopted by the speakers of the other three varieties. The examples in this description were recorded from a female speaker of the Kɛˋwɛ variety from the town of Somdina.
Consonants
The chart above illustrates the most significant sounds of Kabiye. Phonologically, the consonant system comprises 21 sounds. The labial-velar plosives /
/ are found in Kabiye, just as in other larger and more extensively studied West African languages such as Igbo and Yoruba. Phonetically, voiceless plosives are moderately aspirated and voiced sounds tend to have a degree of breathiness in the pronunciation of some speakers. The current orthography, which uses several graphs from the IPA, is generally very close to the phonological form on which it is based. However, the orthography departs from the IPA in several cases: IPA [tʃ ʃ ɲ j ʈ] are represented by the graphs 〈c j ñ y ɖ〉, respectively. The graph 〈ɖ〉 is used in the orthography to represent both the phonologically underlying /ɖ/ and its phonologically conditioned variants [ʈ], [r] and [ɽ]. The allophone [ʈ] is restricted to initial positions, e.g. /ɖoˊnˋ/ [ʈoˊŋˋ] ɖoŋ ‘strength’, whereas [ɽ] appears between vowels as in /h
ɖ
/ [h
ɽ
] haɖaa ‘farmers’, and [r] is found in the type of words called ideophones such as [t
r
r
r
] ‘very’ in [leˋleˊŋˋ t
r
r
r
] ‘very sweet’ or in [n
m
t
r
r
r
] ‘very straight path’. The graph 〈ɣ〉 is also found in the orthographic inventory of consonants. Phonologically however, we analyze it not as a separate sound but as a secondary feature of velarization [ɣ] (also referred to as ‘pharyngealization’ in Paaluki 1995) that affects the quality of front vowels (as well as /
/) in Kabiye. Velarization can be contrastive, for instance, only velarization distinguishes between some forms of words (cf. [s
] saa ‘drive!’ vs. [s
ɣ
ɣ] saɣ ‘sculpture!’, [tìì] tii ‘descend!’ vs. [tìɣìɣ] tiɣ ‘rub!’, and [teˋeˋ] tee ‘sing!’ vs. [teˋɣeˋɣ] teɣ ‘(keep) singing’).
This language exhibits some noteworthy phonetic intricacies. Two rules constrain the surfacing of the underlying voicing specifications of obstruents. One rule categorically disallows surface voiced obstruents in initial positions after a pause of any kind, namely the absolute utterance-initial position, the beginning of a word in isolation, and after a pause in a sentence. Another rule has the opposite effect after morpheme boundaries in an utterance, where obstruents become voiced. The application of these rules can be seen in the following examples: /s
w/ [s
w] ‘to greet’ vs. [píy z
w] ‘to greet children’ and [-z bíy] ‘he greeted children’, /zt
/ [st] ‘to cut’ and [stnìmìyeˊ] ‘cut the rope!’ vs. [píy ztnìmìyeˊ] ‘the children cut the rope’. The paradigmatic consequence of the application of these rules is often a complete overlap of the voicing specifications of obstruents on the surface that can cause confusion. Adding to this confusion is the independent fact that some grammatical tenses in this language such as the Simple Present, the Simple Past, the Immediative, the Past Progressive, the Future, etc. appear to carry an inherent floating morpheme for voicing that also affects initial obstruents. The underlying voicing specifications, however, surface in a limited number of contexts, specifically, with those grammatical tenses that do not have an inherent floating morpheme for voicing such as the Consecutive, the Past Perfect, the Past Perfect Continuous, the Optative, etc. (viz. /kɖ/ [kɽ] ‘to lock’: [-ɡɽɣɣ] ‘he locks’ vs. [-kɽ] ‘then he locked’ and [-ɡɽ] ‘he locked’ vs. [
-kɽ] ‘he had locked’, and /ʃt/ [tʃt] ‘to lie’: [-ʃtɣɣ] ‘he lies’ vs. [-ʃt] ‘then he lied’ or [-ʃt] ‘he lied’ and [-ʃt] ‘he had lied’), and after long vowels and negation morphemes (viz. /kɖ/ [kɽ] ‘to lock’: [-ɡɽ] ‘he locked’ vs. [-d-kɽ] ‘he did not lock’ and [-kɽ] ‘he had locked’; and /ʃt/ [tʃt] ‘to lie’: [-ʃt] ‘he lied’ vs. [-d-ʃt] ‘he did not lie’ and [-ʃt] ‘he had lied’). In word medial position where there is no morpheme boundary, the word structure constraint is not operative and therefore, minimal contrasts between voiced and voiceless obstruents are found. Peculiarly, the voiced /b/ and /ɖ/ are found in all environments other than initial positions after pauses, where their respective allophonic variants [p] and [ʈ] occur exclusively, thus resulting in a defect in the voiceless obstruents’ phonological inventory. Distributional restrictions also affect all consonants in general in syllable coda or word-final. Only the nasals [m ŋ] are found in these positions where they are syllabic as well as in syllable peak. [ŋ] is found in syllable coda as a significant variant of either of the phonemes /n/ or /ɲ/.
Previous inventories showed different sets of phonemic obstruents. In Delord (Reference Delord1976), for example, all the obstruents but /b/ and /ɖ/ are voiceless. In contrast, the inventories posited by Lébikaza (Reference Lébikaza1999) and Roberts (Reference Roberts2002) recognize the voiced alveolar sounds /d/ and /z/ but they do not include Delord's voiced obstruents. Moreover, Lébikaza on the one hand agrees with Delord with regard to the presence of a velar phoneme /ɣ/ while Roberts on the other hand agrees with Delord in including a velar nasal /ŋ/ in his inventory.
Vowels
The vowel formant plot presented above is based on three tokens of each word used. Kabiye has nine contrastive vowel qualities. Long vowels also occur in the language, most often as a result of inflectional processes as seen in the Imperative forms that contrast with the short vowels in the words illustrated on the vowel chart. Long vowels show no noticeable difference in quality from the short vowels and are analyzed as sequences of short vowels. This is represented in the writing system by doubling the graph. The vowel system also shows two contrastive sets that have traditionally been described in terms of a difference in tongue root position (‘Advanced Tongue Root’ or ATR). Recent research (Edmondson & Esling Reference Edmondson and Esling2006, and Edmondson, Padayodi, Hassan & Esling Reference Edmondson, Padayodi, Hassan and Esling2007) has shown that the mechanism involved in their articulation has to do not only with tongue root movement but also with aryepiglottic fold constriction as had been previously predicted by Esling (Reference Esling1996: 81; see also Esling Reference Esling2005).
Traditionally, the difference in tongue root position has been expressed by means of a diacritic: either
signifying [+ATR] or
for [–ATR] and V signifying [–ATR] or [+ATR], respectively (cf. IPA 1999). The critical point is that the V quality is usually represented by the same symbol shape, with or without the diacritic. This paper, following Edmondson et al. (Reference Edmondson, Padayodi, Hassan and Esling2007: 2067f.), argues that the two vowel sets of Kabiye differ in their articulatory postures not just in their resonance features: specifically, ‘Kabiye constricted [i.e. [+RTR]] vowels are systematically narrowed forwards and upwards across the aryepiglottic sphincter, tongue-retracted, and larynx-raised’. Therefore, we have chosen a transcription that recognizes two vowel sets [i
, u
,
, e
, o
] in which the [+RTR] set is represented both by symbol and diacritic to express the pervasive differences of retraction of the tongue, forward and upward narrowing of the aryepiglottic sphincter, and the raising of the larynx. The [+RTR] set in this description corresponds to the traditional [–ATR] vowels, and the [–RTR] set corresponds to what has been called [+ATR].
This division of the vowels into two sets is known to impose a constraint on the combination of vowels occurring in a word in a pattern of ‘Vowel Harmony’. The Vowel Harmony in Kabiye allows only vowels of one set to appear together in a phonological word. The only exception we found is the case of compounds in which each constituent keeps its original vowel harmony, and borrowings such as [oˋtl] ‘hotel’ (from the French hôtel, Lébikaza Reference Lébikaza1999: 177). Although the vowel // has a special status that allows it to combine quite freely with either set, there is sufficient evidence to support that it belongs to set 2. For instance, when // is the first vowel in a verbal stem, any co-occurring vowel is from set 2 (cf. [tmˊs] ‘to bind’ but never *[tmˊsìuˋ]). Further confirmation comes from the articulatory phonetic perspective of the laryngoscopic studies by Edmondson & Esling (Reference Edmondson and Esling2006), which clearly show that the articulation of the vowel // involves aryepiglottic constriction just as in the articulation of set 2 vowels in this language.
Another type of harmony exclusively affects prefixes where the vowel in the prefix agrees with the first vowel of the stem with respect to front and back features. In this process, there is either a complete assimilation of the prefix vowel so that it becomes an exact copy of the stem vowel (e.g. /bˋV-ʃtmˋ/ [pʃtmˋ] pɛcɛtɩm ‘their lie’, /bˋV-toˊkoˋ/ [poˋdoˊkoˋ] potoko ‘their shirt’, /bˋV-smɖ/ [pzmɽ] pasaamɩɖɛ ‘their corn’, /bˋV-tk/ [pˋdk] pɔtɔkɩ ‘they eat’, /ˋV-liídìjeˊ/ [ìlíídìjeˊ] eliidiye ‘his money’), or the prefix vowel is partially assimilated, taking on only the [-back] or the [+back] features of the stem vowel (e.g. /ˋV-líɖeˋ/ [eˋlíɽeˋ] or [ìlíɽeˋ] eliɖe ‘his root’, /ˋV-toˊkoˋ/ [eˋdoˊkoˋ] or [oˋdoˊkoˋ] etoko ‘his shirt’). The vowels // and // exceptionally do not partake in this harmony (viz. /bˋV-bn/ [pbn] pabʊnʊ ‘their goat', /bˋV-bj/ [pbj] papɩyɛ ‘their stone’).
Prosodic features
Two distinctive tone levels are recognized in Kabiye: high /ˊ/ and low /ˋ/. The language also exhibits a downstepped variant of the high tone [↓ ˊ]. Syllabic nasals [m] and [ŋ] and vowels are the Tone Bearing Units with long vowels appearing to allow for the presence of limited tonal contours. It is also important to mention that the tone phenomena variously termed ‘declination’, ‘downstep’, and ‘downdrift’ in the literature are very apparent. Likewise, the process of tone spread is very common. To date, there has not been any solid study of these phenomena, so the very complex tonal processes of the language are not well understood.
Transcription of recorded passage
In the current orthography, tone is still not represented, a lacuna that researchers are seeking to address. The practice of previous researchers is to show tone only in a privative way by leaving the low tone unmarked and only indicating the high tone and the downstepped high. In the phonetic transcriptions of the illustrations for this paper, however, we have chosen to indicate all the tones. The transcription style adopted in this illustration is relatively narrow in certain respects, reflecting the particular pronunciation in the recorded passage.
Orthographic version
Hayikiŋ Heelim nɛ Wɩsɩ
Hayikiŋ Heelim nɛ Wɩsɩ paa maɣzɩnɩ mbʊ se pana zɩ akɩlɩ ɖoŋ. Pɩkɛdaa ɛlɛ
nʊmɔwoɖu nɔɔyʊ ɖɛwaɣ, ɛhɔkɩ ɛtɩɩ toko kpiŋgizim ñɩŋgʊ nakʊyʊ taa.
Panɩɩna ɖama se weyi ɛɖa pɩzʊʊ ɛla nʊmɔwoɖu ɛnʊ ɛhɔzɩ etokoo lɛ ɛnʊ kɩlɩna ɖoŋ.
Kɛlɛ Hayikiŋ Heelim pazɩ pɩma ɖeŋɖee pɩpɩzɩɣ yɔ, kɔyɔ pɩmakɩ ɖoŋ lɛ
nʊmɔwoɖu ñi kpiikiɣ ɛtɩɩ etoko taa kikpiiku. Pɩnɩɩ Hayikiŋ Heelim nɛ pɩtɛzɩ yebu.
Kɛlɛ Wɩsɩ ña sɩŋgɩ sɩñaɣ ɖeŋɖee sɩpɩzɩɣ yɔ, kaʊʊ. Nʊmɔwoɖu tɩ hɔzʊ etoko ɛlɔ.
Peeɖee pɩɖɛ Hayikiŋ Heelim nɛ pitisi se pataa lɛ Wɩsɩ kɩlɩna ɖoŋ noo.
English version
The North Wind and the Sun
The North Wind and the Sun were disputing which was the stronger, when a traveler came along wrapped in a warm cloak.
They agreed that the one who first succeeded in making the traveler take his cloak off should be considered stronger than the other.
Then the North Wind blew as hard as he could, but the more he blew the more closely did the traveler fold his cloak around him; and at last the North Wind gave up the attempt.
Then the Sun shone out warmly, and immediately the traveler took off his cloak.
And so the North Wind was obliged to confess that the Sun was the stronger of the two.
Acknowledgments
I am grateful to Dr. Jerold A. Edmondson and Dr. David J. Silva for their useful suggestions and assistance in the preparation of this illustration.
