Consonant voicing is relevant to the tonology of many languages, where in nearly every case, voiced obstruents condition a lower tone. While this is the most common pattern, questions remain as to other possible consonant-tone interactions. This paper describes verbal tone in the Chadic language Zina Kotoko, spoken in northern Cameroun. Kotoko is relevant to the study of consonant-tone interaction because the entire system of tone assignment in verbs is based on interaction between tense-aspect and the consonants in the stem: there are no lexical tone properties in verb roots.

Tone in verbs is determined primarily by the nature of the root-initial consonant, and in some tenses also by the final consonant. In each tense, the initial (and possibly final) syllable of the stem selects between one of two tones, the surface tone being selected on the basis of the nature of the initial consonant and the tense in question. Thus in the recent past the first syllable either has L or M tone, the choice determined by the consonant in the first syllable such that M is selected after voiceless consonants (sə̄k-ə́m "sent") and L appears after voiced consonants (gàh;-ə́m "poured"). In the habitual, on the other hand, H is selected after voiceless consonants (sə́kə́ "sends") and M appears after voiced consonants (gàhə̀ "pours"). Other tenses exhibit different tone patterns in terms of choices of tones and conditioning classes of consonants. The consonants of Kotoko can be classified hierarchically in terms of the strengh of a consonant's tone-lowering properties. Voiced obstruents always manifest tone-lowering, sonorants do in virtually all contexts, glottal stop does in very many contexts, implosives do in half of the contexts, and voiceless obstruents and h never do.

It will be shown how the tense-and-consonant induced alternations can be explained in terms of specific floating tones which mark tense-aspect (M in the past, H in the habitual, MH in the imperative, etc.), plus tone lowering rules that are sensitive to the neighboring consonant. One rule, H-Lowering, lowers H to M after voiced obstruents, sonorants, and glottal stop. A second rule, M-lowering, lowers M to L after voiced obstruents, sonorants, glottal stop and implosives: implosives thus behave differently for the two rules. The third rule, Pre-Depressor Lowering, lowers M to L before a voiced obstruent (but not before a sonorant). The tonal separation of glottal stop and sonorants is found in the CV-reduplicant progressive tense, where voiced obstruents and sonorants, but not glottal stop, trigger lowering of the underlying H tone.

Consonant-tone interaction in Kotoko presents a number of theoretically interesting points. First, most languages with such effects are 2-tone-level languages: Kotoko lowers M to L and H to M, indicating at least that "lowering by one step" is a possible tone-depressor strategy for three-level languages. Sonorants also (usually) fall into the class of depressors, which is unusual since sonorants are almost always tonally neutral in other languages. Kotoko also demonstrates that glottal stop can be a depressor, something not attested (or testable) in any other language. Contrary to the otherwise universal trend that only a preceding consonant is relevant for tone-lowering effects, following voiced obstruents lower tone in Kotoko. Finally and quite surprisingly, implosives function as depressors of M -- but not H -- even though depressors are the phonetically least likely to cause tone lowering, and have never been seen to function as depressors in any other language.