Phonogenres without borders? A cross-linguistic study of speaking style variation
(2014) 3rd Swiss Workshop on Prosody (SWIP) — Location: Geneva, Switzerland (10.September.2014)
Files
No attached file found for this publication.
Details
- Authors
Christodoulides, George 
- Abstract
- Social situations condition linguistic choices, giving rise to differences in register (Biber 1995), and leading to the emergence of genres (Beacco 2004). The study of this situational variation in spoken language (Llisteri 1992; Léon 1993) has gained new insights by recent work based on large corpora (e.g. Goldman et al. 2014) that established wide-ranging differences in the prosodic features between the speaking styles adopted under different communicative situations, at the global and local level. We attempt to incorporate a cross-linguistic perspective in the investigation of speaking style variation. We have compiled a trilingual corpus of 6 different speaking styles. The corpus consists of samples in English, French and Greek, while the genres were selected based on the following considerations: (a) comparability with previous monolingual studies; (b) broad coverage of different speaking situations and contexts and (c) choosing speaking styles that are defined consistently across the three languages studied. The corpus consists of 15-20 minutes of speech, per language, for each of the following situations: reading; scientific lecture; political speech during parliamentary debates; short radio news bulletin; stand-up comedy sketches; and sports commentaries. Speaking styles are characterised along multiple dimensions (Hirschberg 2000; Koch & Oesterreicher 2001) and five situational features are identified: degree of interaction; speaker preparation; external time pressure; presence of an audience; and medium of communication. These situational features are correlated with a set of measures on each corpus sample, in an effort to find the characteristics that are typical of each genre and distinguish it from the others, within and across languages. The cross-linguistic nature of the study creates an additional challenge of finding a set of features that is valid across languages. Such features include the distribution of silent pauses (Goldman et al. 2010); speech rate (De Looze 2010); lexicometric measures such as density (Ure 1971); pitch range patterns and prosodic phrasing (Frazier et al. 2006); the distribution and prevalence of disfluencies (Christodoulides & Avanzi 2014); and discourse marker use and distribution. In this study, we present preliminary findings on automatically extracted prosodic features. Our aim is to answer the following research questions: Are phonogenre differences perserved accross languages? Are phonogenres separated equally well using a combination of features, accross languages? And which set of features/descriptors performs best in the task of distinguishing between phonogenres accross languages?
- Affiliations
Citations
Christodoulides, G. (2014). Phonogenres without borders? A cross-linguistic study of speaking style variation. 3rd Swiss Workshop on Prosody (SWIP), Geneva, Switzerland. https://hdl.handle.net/2078.5/65854