For the MS-teams link, write to Kim.Groothuis@UGent.be.
In studies on African lexical tone languages, intonation is often approached either as emergent from the cumulative effects of local interactions between sub-tonal features like the register feature (Welmers, 1959; Inkelas & al, 1986; Connell & Ladd, 1990; Clements, 1979) or as limited to domain boundary manifestations (Rialland, 2007). When global effects are mentioned, they are often treated as phonetic in nature (Inkelas & al, 1986; a.o), supposedly because they match the predictions of the Frequency Code, which holds that questions are realized with a higher pitch than statements (Gussenhoven, 2002; Cahill, 2013). Here, we present the results of a case study (production and perception) of yes/no question intonation in Ede Chaabe (cbj, Benin). We found that questions’ register is lower and not higher, but also have a L% edge tone known to characterize ‘lax’ prosody languages (Rialland, 2009); hence contrary to the Frequency Code. Considering these findings, we argue that the observed global effects are represented in the grammar in the form of a Register feature, which is treated in the present account as an intonational feature than spans specific prosodic domains. We go a step further in proposing a new sub-tonal feature model that does not use a register feature (like previous models do), but rather treats lexical tones as pitch change instructions, where Polarity (+/-) indicates the direction of the change and Step (1/2) would indicate its magnitude. In this sense, any given lexical tone (with Polarity and Step features) is projected on the intonational Register plane, post-lexically.