Replicating in Naxi (Tibeto-Burman) an Experiment Designed for Yorùbá: An Approach To ‘Prominence-Sensitive Prosody’ vs. ‘Calculated Prosody’

The starting-point of this study is the hypothesis (suggested by an overview of typologically varied languages) that it may be useful to characterise prosodic systems in terms of the degree to which they rely on the calculation of tone sequences. Each language could be placed at a certain position along a typological continuum between two types of prosodic organisation: (i) ‘calculated prosody’, in languages such as Yorùbá (in which tone serves complex morphophonological functions), whose prosodic structure hinges on the calculation of a tone sequence, by categorical processes such as the association of lexical tones and/or boundary tones, reassociation/‘tone floating’, and downstep; and (ii) ‘prominence-sensitive prosody’, typologically more common, found in languages such as Chinese, which have fewer elements of categorical tonal calculation, and in which intonation appears to reflect phrasing and informational structure in a largely noncategorical way. In an effort to test (and refine) this hypothesis, an experiment used for Yorùbá [8, 9] is adapted to Naxi, a Sino-Tibetan language which, like Yorùbá, has three lexical tones (High, Mid and Low), but which is hypothesised to be closer to the ‘prominencesensitive prosody’ type, whereas Yorùbá would be closer to the ‘calculated prosody’ type. This pilot study on sentences in which all syllables bear the same tone does bring out differences between the two languages in terms of phenomena of phrasing and of prominence. 1. A typological proposal: ‘prominencesensitive prosody’ vs. ‘calculated prosody’ 1.1. Some definitions As a preliminary to the proposed typological overview, some indications on the terms used may be useful. Prosody is here understood in its broadest sense: it includes lexically distinctive suprasegmentals, intonation, the expression of attitudes and emotions, and performance factors: rhythm and speech rate [15]. Lexically distinctive suprasegmentals may be, depending on the language, lexical stress, lexical tone, lexical pitch accent... Intonation is an abstract structure, not to be confused with the parameters whereby it manifests itself (in particular, it is a different notion from F0, which is a physical parameter, and pitch, which is a percept). Intonation can be conveniently divided into two components: (i) intonational phrasing, sometimes called syntactic intonation because of its strong (though not bijective) link to syntax, (ii) pragmatic intonation, determined by the information structure of the utterance. The term intonational group is used here loosely as a cover term for the various levels of intonational phrasing. 1.2. An overview of typologically varied tone languages Typologically, there appears to be considerably more diversity among prosodic systems than labels such as ‘tone languages’, ‘pitch accent languages’ and ‘stress languages’ suggest [7, page 1]. ‘Tone languages’ do not make up a unified typological pole. An overview suggests that some prosodic systems, more than others, go by a calculation of tone sequences. The prosody of Mambila (Niger-Congo, Bantoid) seems to be highly constrained by its four level lexical tones. Ngamambo (Bantoid) [1] also appears highly constrained, as a five-term paradigm is open after a High-tone syllable: High, Mid, Low, downstepped Low (!L), or flat Low (L°). On the other hand, in Hausa, an Afro-Asiatic language which only has two tones, speakers apparently play to a larger extent on intonation to convey degrees of informational prominence (A. Rialland, p.c.). It thus seems that the degree of constraint on the margin of freedom left to the speaker depends on the number of level tones and of categorical processes of tone modification. 1.3. Formulation proposed The proposal put forward here is that prosodic systems could be placed along a typological continuum between (i) prosodic systems that hinge on the calculation of a tone sequence (by categorical processes of association of lexical tones and/or boundary tones, reassociation/‘tone floating’, and downstep), and (ii) prosodic systems in which phonetic realisation is sensitive to the prominence of individual items within the utterance, i.e. where intonation reflects informational structure (with such typical phenomena as contrastive emphasis). ‘Prominence-sensitive prosody’ can be illustrated by languages such as Mandarin Chinese. Despite having four lexical tones, Mandarin allows local intonational modifications which carry a communicative load comparable to the one they have in a language such as English (as has been known since [3]): the phonetic realisation of tones depends on the degree of prominence of individual syllables, and on intonational phrasing. Rather than unambiguously categorical processes such as downstep and tonal reassociation, Mandarin Chinese shows many noncategorical modifications of lexical tones, and few categorical modifications. In the study of Mandarin tones, much attention has been directed to the sandhi process whereby a sequence of two tone-3 syllables yields a tone 2+tone 3 sequence; on the other hand, the many cases of “minor sandhi” that have been proposed for Mandarin (e.g. by Chao Yuen-ren) are not clearly categorical, so much so that “It is not clear whether it is desirable or even possible to segregate tonal coarticulation [i.e. a noncategorical modification] from tone sandhi proper [i.e. a categorical modification]” [4, page 25]. Speech Prosody 2006