LSA 2025: Illustrating African Phonologies

Implosives are found in 13% of the world’s languages (Maddieson, 1984) and are particularly widespread throughout the languages of sub-Saharan Africa, but they remain extremely understudied in phonetics and phonology (Clements & Osu, 2002). Although canonically described as voiced with an ingressive glottalic airstream, studies show their production is highly variable; for example, implosives can be produced with no ingressive airstream and without rarefaction (negative oral air pressure; Ladefoged, 1968, 1971), and they can be produced with or without full glottal closure (Ladefoged & Maddieson, 1996). In this study, we report our preliminary findings on the acoustics and aerodynamics of implosives in Nghlwa [ISO:gwa], a Kwa language spoken by the M’Bato people in the La Mé Region of Côte d’Ivoire. In fieldwork elicitations, we collected acoustic, electroglottograph, nasal airflow, oral airflow, and oral pressure measurements. We discuss the overall phonetic profile of Nghlwa implosives and how it relates to their split phonological patterning with both obstruents and sonorants.

The Lobi (Gur) consonant inventory contains implosive and labiovelar stops as well as plain and glottalized sonorants. Using distributional facts, articulatory EGG data, and acoustic data, we offer several insights into the set of sounds /ɓ, b, j', l', w', j, l, w, kp, ɡb/, including: (i) /ɓ/ shows positional allophony and distinct phonological behavior from /b/; (ii) all glottalized sounds show show semi-periodic supra-glottal activity leading up to segment onset; (iii) bilabial voiced stops and labiovelars pattern alike in voicing onset; (iv) all included sounds tend to have more dramatic phonetic cues present word initially than word medially. Depending on which metrics one uses, different phonetic and phonological natural classes emerge, though on all metrics /ɓ j' l' w'/ pattern together phonetically. This work contributes to our typological understanding of whether the phonetic realization of glottalized sounds is correlated with their phonological patterning (see Sande and Oakley 2023).

In this talk, we highlight the existence of non-dorsal vowels with consonant-like place features, closely resembling syllabic labial or coronal fricatives or approximants but not phonetically identical to either. Data from west and central African languages provides key evidence for these phonological representations. Non-dorsal vowels’ place features are most often intrinsic (as in Kom, Limbum, Fang, and Lendu) but sometimes arise through spread from the preceding onset consonant to underspecified vowels (as in Lus, Kung, Isu, or Aghem). Articulatory phonetic data (tongue ultrasound, lip video) provides direct evidence of labial and coronal constrictions during the non-dorsal vowels. Further evidence comes from co-occurence restrictions and variable assimilatory patterns between onsets and non-dorsal vowels which are most easily captured as interactions of the major places of the two segments.

This study aims to experimentally establish whether or not pharyngealization, a phonemic property of vowels in Toussian languages (Gur/Mabia, Burkina Faso), is “less strongly pronounced” in Northern Toussian (NT) than in Southern Toussian (ST) as claimed by Prost (1964). Since acoustic expressions of pharyngealization include lower F2, higher F1, and lower F2-F1 (i.e. more compact) (Shahin 1997; Mohamed 2001; Al-Tamimi 2017), it is expected that F2 would be much lower, along with a much higher F1 in ST than in NT per Prost’s claim. Four male speakers of Toussian (2 per variety) were recorded producing a set of carrier phrases containing target words with and without pharyngealized vowels. Preliminary results show that the mean Bark-converted F2 is indeed overall much lower in ST than in NT, with even more pronounced difference in the second half of the vowel. F1 results were more nuanced. The full analysis includes the vowels [i,ɪ,e,œ,a,o,ɔ].

Implosives have variable cross-linguistic behavior (Sande and Oakley 2023), wherein they can pattern phonologically with obstruents, sonorants, or both in different contexts. This study compares the acoustic properties of implosives to sonorants and obstruents in Guébie (Kru, Côte d’Ivoire), in order to examine whether implosives share phonetic properties with the natural class of sonorants, as predicted by their phonological behavior.

Five Guébie speakers were recorded in Gnagbodougnoa, Côte d’Ivoire. 1,364 voiced obstruents, 1,993 sonorants, and 656 implosives were included in the present analysis.

Implosives, sonorants, and voiced obstruents all differ in duration and intensity. Implosives have a positive intensity slope, whereas sonorants and voiced obstruents have a negative intensity slope. Implosives also have a longer voicing duration ratio than voiced obstruents.

Implosives seem to have acoustic measurements in-between sonorants and obstruents, which, if true across languages, may explain their variable cross-linguistic phonological patterning.

While click consonants have been the object of more than a century of articulatory and acoustic study, there is a relative dearth of literature on their perception. Ladefoged & Traill (1994) present a small-scale experiment about perceptability of clicks over background noise, and find that clicks are considerably easier to identify than non-cliks – presumably due to their extremely high intensity. Pioneering studies conducted by Best et al. (1981, 2003, et seq) find that clicks are on the whole extremely reliably perceived – even by speakers unfamiliar with them. However, descriptive work on South African languages shows rampant variation, abundant doublets, and distributional patterns that suggest that click distinctions are perhaps not so easy and reliable distinguish outside of the lab. In this talk, we present results from several preliminary experiments on click perception with Zulu and Xhosa listeners.

Previous research on Gbagyi (Nupoid, Nigeria) posits three contrastive tones, namely H(igh), L(ow), and M(id) (Hyman and Mogaji 1970). In certain bisyllabic words, the second tones surface with falling and rising contours when the initial tone is higher or lower than the second tone. In this study, we examine the acoustics of tones in the language, which has not been done previously.

In addition to the three lexical tones, we present evidence for a fourth lexically contrastive level tone, as illustrated with the minimal set [ snì ] “to fetch”, [ snū ] “to roast”, [ sní ] “drink” and [ sni̋ ] “drank”, and the minimal pair [wje̋gje̋] "next year" and [wjégjé] "seeing". Acoustically, the pitch of the fourth tone is higher than that of the H tone and patterns differently from other tones in the language. Results of our acoustic study also confirm the previous impressionistic observation regarding tone transfer. However, we find no evidence to suggest that tone transfer is optional in the language.