The Kindige variety of Bondu So (Dogon, Mali), displays [+/–ATR] vowel harmony patterns that interact with root-final consonants. The language displays agglutinating, suffixing morphology. This study focuses on the behavior of verb stems in two paradigms: the perfective aspect and past tense. It is posited that these two verb stems share the same suffix, but that the quality, height, and length of the suffix is determined by a combination of factors, including [ATR], sonorancy, and metrification. Examples from the 3rd person singular and plural in the perfective and past stems are shown in Table 1.
Table 1: Obstruent versus nasal-final roots following [+/–ATR] vowels in Perfective and Past stems
Table 1 demonstrates that verb roots spread their [+/–ATR] value to a suffix specified for backness but not [ATR]. Front suffixal vowels are associated with the third singular person and back vowels with plural.
The vowel is long in the Perfective and short in the Past, or, following nasal-final roots with [+ATR] vowels as in (1c), absent. In the Perfective, the suffixal vowels are mid unless the root-final consonant is a nasal and the root vowel is [+ATR] such as in (1c). In the Past, a similar pattern is found in that obstruent-final roots with [+ATR] vowels are followed by high vowels, but those with [–ATR] take mid-vowel suffixes.
Green and Hantgan (2019)1 explore the possibility that the difference between the two stems is due to the past stem vowel being epenthetic. However, here, we propose an alternative account that relies upon structural requirements inherent in Bondu So. More specifically, the stem final vowel (the suffix) of Perfectives, because of its phrase-final position, is realized as long, yielding an ideal unbalanced iambic foot. The suffix in the Past, when followed by a clitic, however, may appear as a short vowel, or may be absent, but only when followed by nasals licensed as moraic by [ATR] spreading - a postulation put forth in Green and Hantgan (2019)1.
To further substantiate this new analysis, we present a phonetic study illustrating that, on average, root-final sonorants following [+ATR] vowels are longer than those following [–ATR] vowels. These outcomes contribute to a small but growing literature (Archangeli et al. 20102; Carnie 20023; Ní Chosáin 19914) on the role of [ATR] in sonorant mora licensing. While these earlier studies illustrate cases of underlying sonorant moraicity, motivated (for example) by their ability to trigger compensatory vowel lengthening, Bondu So verb stem’s behavior instead presents a mirror image case of coerced moraicity, as evidenced by sonorant lengthening in the [+ATR] context.
Archangeli, Diana, Jeff Berry, Andrew Carnie, Nicole Hunt, Sunjing Ji & Keisha Josephs. 2011. ATR in Scottish Gaelic tense sonorants: A preliminary report. Formal Approaches to Celtic Linguistics, 283-306. ↩︎
Carnie, Andrew. 2002. A note on diphthongization before tense sonorants in Irish: An articulatory explanation. Journal of Celtic Linguistics 7, 129-148. ↩︎
Ní Chiosáin, Maire. 1991. Topics in the phonology of Irish. PhD dissertation: UMass-Amherst. ↩︎