Phonetics of intonation in South African Bantu languages

Much is already known about the prosodic systems of the indigenous South African languages from descriptions and analyses in the existing literature. All of the existing work has been carried out in the field of African studies or formal linguistics. In order to be able to implement the generalisations obtained into computational models in speech processing, the existing sources and results must be made accessible to researchers in these areas. In the opposite direction, the modelling of Bantu intonation in speech processing shows the need for more quantitative data. The present article presents a review of durational and pitch-related tonal aspects relating to intonation in South African Bantu languages. Its aim is, on the one hand, to make the results of Africanist and linguistic studies accessible to research in speech processing, laying a foundation on which work in the speech-processing branch of Human Language Technologies can be based. On the other hand, by pointing out gaps in our knowledge, it wants to draw attention to the fact that more quantitative research is needed in order to advance our knowledge also in the theoretical linguistic field. Introduction The term intonation is loosely defined in Crystal (2003), who refers to ‘the distinctive use of patterns of pitch’, whereas Hirst and Di Cristo (1998) refer to ‘the melody of speech’ (thus encompassing not only pitch but also patterns of rhythm and loudness). Traditionally, the term intonation is used in stress languages in order to refer to meaningful pitch changes at the sentence level. However, also in tone languages, we find meaningful alternations in pitch across the sentence (for Chichewa see Downing et al., forthcoming; for question intonation in equatorial African languages see Rialland, forthcoming; and Xu, 1999, for focus in Chinese). Many different aspects feed into intonation. These aspects include the use of intonation and tone to express emotions, to encode information structure and to signal sentence types; the sensitivity of tone and intonation to lexical specification, word classes and syntactic structure; the phonetic aspects of timing, alignment and segmental influences. Acoustically, intonation is the modulation of fundamental frequency, intensity and duration across an utterance. Formal linguistics and speech processing share a common interest in the investigation of intonation in South African Bantu languages. Both want to provide an account of Bantu tone systems that predicts the realisation of tones so that tonal contours and intonation over utterances can be modelled, either theoretically or practically. The current article presents a state-of-the-art report on tonal research into South African Bantu languages in formal linguistics. South African Bantu languages are among the better-documented languages of the African continent. With respect to their intonation systems, a number of case studies exist for all nine official languages. Existing work is cited in Table 1 and serves as the basis for the exposition in the rest of the article. The article is structured as follows: the section entitled ‘Duration’ presents durational aspects of intonation in southern Bantu languages. It discusses how phonology, syntax and pragmatics influence vowel length. ‘Tone and intonation’ reviews the tonal aspects of intonation in southern Bantu languages, discussing the tone inventory, acoustics of low and high tones, as well as tones Zerbian and Barnard 236 within different tonal and segmental contexts. ‘F0 versus intensity’ addresses the role of intensity in intonation. These three sections taken together thus lay out how duration, fundamental frequency and intensity interact in the intonation of southern Bantu languages. The summary and critical assessment of the available literature on this topic is meant to facilitate access to basic information for use in speech processing. Though many aspects will need to be oversimplified owing to space restrictions, the list of references is fairly extensive and should allow following up specific aspects for specific varieties or languages. The section entitled ‘Implications for speech technology’ discusses the importance of the above matters for speech-processing systems. (Text processing and speech processing are the two main branches of human language technologies; the former is not directly impacted by considerations related to intonation.) The last section of the article provides a conclusion and summary. Duration Duration is one parameter that feeds into intonation. As a parameter of intonation, the duration of segments is modified owing to changes in phonological boundaries and/or alternations in discourse prominence. However, not all durational alternations in a language relate to intonation. The present section discusses segment duration in southern African Bantu languages. It starts out with durational alternations which are unrelated to intonation, but which are grounded in the sound inventory of the language (subsection ‘Phonemic vowel length’) or in phonetic universals (subsection ‘Phonetic influences on vowel duration’). It then discusses the durational alternations due to syntactic (subsection ‘Syntactic factors of vowel length’) and pragmatic factors (subsection ‘Pragmatic factors’). The last two aspects fall under intonation. Phonemic vowel length Depending on the sound inventory of a specific language, length in vowels can be phonemic or not. If length is phonemic, the language has both long and short vowels and the durational difference of the vowel leads to a difference in meaning. In Kinyarwanda, a Bantu language spoken in Rwanda, vowel length is contrastive. An example of a minimal pair differing only in vowel length is given in example 1. Kinyarwanda: contrastive vowel length (from Myers, 2003) 1(a) [gutaka] ‘to scream’ 1(b) [guta:ka] ‘to decorate’ Table 1: Literature sources for tone systems of southern African languages Family Language Source Nguni isiZulu Beuchat (1966), Cassimjee and Kisseberth (2001), Clark (1988), Cope (1959; 1970), Downing (2001), Khumalo (1981; 1982; 1987), Laughren (1984), Peterson (1989), Roux (1995b), Rycroft (1963; 1980a) isiXhosa Cassimjee (1998), Cassimjee and Kisseberth (1998), Claughton (1983), Goldsmith et al. (1989), Jokweni (1995; 1998), Louw (1968), Roux (1995a;