The study presented here explores the question: what is an adequate phonological model for global intonation in tone languages? This question is warranted because studies of the intonation of African tone languages often approach intonation either as emergent from the cumulative effects of local interactions between sub-tonal features (Welmers, 1959; Inkelas & al, 1986; Connell & Ladd, 1990; Clements, 1979) or as the manifestations of prosodic domain boundary tones (see Rialland 2007, 2009 for a review). Global effects that fail to be accounted for via toneto-tone interactions are considered to be phonetical in nature. One instance of this is found in Inkelas & al’s (1986) analysis of Hausa, where the characteristic Global Raising in Yes/No Questions is analyzed as a “low-level phonetic implementation” (1986:328). This conclusion agrees with the predictions of the Frequency Code that questions universally tend to be realized with a rising or overall high f0, unlike statements (Ohala, 1984; Gussenhoven, 2002; Cahill, 2013). While a phonetic analysis of global raising makes sense in questions, such an analysis extends poorly to cases where global shifts go in the opposite direction, i.e. global lowering in questions or global raising in statements. The results of two studies in Ede Chaabe, production and perception establish that statements are globally lower than questions. The Ede Chaabe data does not submit to a tone-to-tone interaction account ; instead, I argue that the global f0 shift is best analyzed as an independent feature in the phonological representation of intonation. The feature in question, which in Ede Chaabe is -1 in the last Phonological Phrase of the utterance, defines the baseline or reference line on which lexical tones are realized. This approach also accounts for changes in the onset of the global shift, as a function of utterance length.
The study involved 24 native speakers of Ede Chaabe, where participants were instructed to produce twelve (12) different utterances in pairs of question/statement, long/short utterances grouped by the three lexical tones in the language (i.e all H, all M and all L tones), andrepeated three times in three blocks of subject-randomized orders. The results show that yes/no questions in Ede Chaabe not only (1) have the boundary tone (i.e L%) known to characterize the putative ‘lax’ prosody languages (Rialland 2007, 2009), but also (2) have a significantly lower global f0than statements. The apparent graduality in the lowering of the f0 in questions can be partially attributed to phonetic coarticulation between the boundary L% and the immediately preceding lexical tones as confirmed by the perceptual study presented in Mamadou & D’Imperio (2020).
Selected References
Gussenhoven, C. (2002). Intonation and interpretation: phonetics and phonology. In Speech Prosody 2002, International Conference.
Rialland, A., 2009. The African lax question prosody: Its realization and geographical distribution. Lingua, 119(6), pp. 928-949.
The ACLC seminar series is a two weekly lecture series organized by the ACLC, the Amsterdam Center for Language and Communication.