The volume Prosody in interaction provides a multi-faceted collection of papers addressing a variety of prosodic aspects and the roles they play in spoken natural interaction. It impressively reflects the establishment of the conversation analytic and interactional linguistic framework over the last decades and confirms the need for a socially-oriented functional analysis of linguistic concepts in spontaneous data.

The book is divided into an introduction plus three major sections, dealing with (I) prosody and other levels of linguistic organisation in interaction (e.g. syntax and pragmatics), (II) prosodic units as a structuring device in interaction (such as intonation phrases), and (III) prosody and other semiotic resources in interaction (i.e. multimodality in various contexts such as doctor-patient interaction or in the communicative means employed by people suffering from aphasia). The concept of the book is very appealing in that nine of the twelve papers are followed by a short commenting paper, adding a different perspective from a second expert in the field. Another great advantage is the open access to video clips and audio files accompanying the papers (at http://dx.doi.org/10.1075/sidag.23.media). In this review, I will briefly discuss a selection of papers taken as examples of the book's main topics, looking at them from a non-conversation-analytic point of view.

In her introductory paper on the state of the art in the field, Margret Selting outlines the development and improvement of the methodology as proposed in Couper-Kuhlen & Selting's (Reference Couper-Kuhlen and Selting1996) seminal book Prosody in conversation setting out to use only spontaneous data, conduct data-driven analyses of functions, analyse the interrelation between phonetic realisations and the sequential organisation of dialogues, and validate results by looking at the orientation of participants to the investigated functions. Moreover, Selting gives an overview of interactional approaches to the study of prosody and the current as well as yet unsolved research questions. She explains the relevant concepts and units of conversation analysis which are encoded by prosodic means (e.g. turns and actions), including methodological aspects and principles. In particular, Selting presents the transcription systems used, such as GAT 2 (Selting et al. Reference Selting, Auer, Barth-Weingarten, Bergmann, Bergmann, Birkner, Couper-Kuhlen, Deppermann, Gilles, Günthner, Hartung, Kern, Mertzlufft, Meyer, Morek, Oberzaucher, Peters, Quasthoff, Schütte, Stukenbrock and Uhmann2009), an improved version of the original system, which has recently been adapted for American and British English (http://agd.ids-mannheim.de/html/gat_en.shtml). As a whole, the introduction is a very helpful reference text for the terminology used and basic questions asked in the field of prosody in interaction which the other papers in the book can be related to.

Part I starts with a paper by Gareth Walker providing a phonetic account of a sample of ‘rush-throughs’ used as a turn-holding device in English. Walker's analysis confirms the function of the prosodic parameter duration (higher articulation rate at the end of the final foot in a turn-constructional unit (TCU) in contrast to final lengthening at the end of a turn, plus lack of a pause) and the non-prosodic parameter phonation (continuous vocal fold vibration between the TCUs) in making it impossible for the other interlocutor to begin his/her turn. In addition, the author discovers that anticipatory assimilation of articulatory gestures projects more talk by the current speaker. In contrast, and rather surprisingly, it was found that low pitch – usually signalling the end of a complete TCU – does not prevent a rush-through. Although the investigation actually separates different parameters in a fine-grained acoustic analysis of rush-throughs, it is at the same time claimed to be holistic in the sense that all perceivable phonetic parameters which may have interactional relevance are looked at. Such an account is problematic if it lacks a previous selection of potentially relevant parameters, which may blur the results and consequently lead to difficulties in systematising the relation of form and function (see also Selting, this volume, pp. 25–26). In particular, it is impossible to weigh the relevance of specific prosodic/phonetic parameters in constructing a certain function in a strictly holistic approach. Still, many conversation analytic (CA) studies are based on a small number of qualitative case studies from a holistic perspective rather than a quantitative perspective (see ten Have Reference ten Have2007, Hutchby & Wooffitt Reference Hutchby and Wooffitt2008). Accordingly, Walker's analysis is based on a ‘representative sample of a larger collection of 20 instances of rush-throughs’ (p. 53). Apart from the fact that the small number of cases is problematic, one can also ask what makes the sample ‘representative’.

The paper by Beatrice Szczepek Reed in Part II of the volume starts with the assumption that the phonological category intonation phrase (IP) could be a relevant category for the analysis of talk-in-interaction, since an IP is often coextensive with a chunk of speech used in conversation. However, since this is not always the case, Szczepek Reed suggests another, functionally defined, category to replace the notion of IP which she calls Turn Constructional Phrase (TCP). A TCP is defined as an interactional unit which is treated by participants in a conversation as a multi-layered event. The boundaries of these units are said to be marked by pauses and breathing rather than pitch movements, i.e. boundary tones. Nevertheless, the exact (formal as well as functional) properties of a TCP remain vague. In fact, the rejection of the IP as a relevant chunk in the analysis of spontaneous conversation seems to stem from the somewhat idealised and thus misleading assumption of the unit as being ‘delivered as one recognisable overall pitch movement’ (p. 191). However, also hesitant and distorted speech in spontaneous conversation can just as well be analysed in terms of IPs. The only prerequisite for an IP in most non-CA approaches is the existence of at least one pitch accent in the phrase (see for example work on spontaneous data by Shari Speer and colleagues on American English or by Klaus Kohler and colleagues on German).

Another paper on prosodic units as a structuring device in interaction, by Bill Wells, deals with the acquisition of the system of tones and their meanings in young children. It is claimed that previous, primarily phonological studies on the topic, which prioritised formal aspects, did not succeed in showing progressive acquisition of adult tones in children (but see e.g. de Ruiter Reference de Ruiter2010). In contrast, Wells proposes a strictly interactional approach claiming that the child's choice of tone (in most cases the nuclear pitch movement, i.e. nuclear accent plus boundary tone) depends on the previous speaker's turn. More specifically, Wells demonstrates in a case study of a one-and-a-half-year-old boy and his mother that the interpretation of a tone as the same (‘tonal repetition’) or different (‘tonal contrast’) determines its communicative function. That is, the meaning of a tone is not derived from its form as such, e.g., a rise or a fall, but from its form in relation to a tone used by the previous speaker. Wells concludes that there is no predefined set of tones with intrinsic meanings in an intonational lexicon, but that the meanings of the tones develop in the context of their usage. This aspect, which is also present in Wittgenstein's philosophy of language, is further elaborated in Traci Walker's comment on Wells' paper. She strongly supports the idea that the basic meanings of tones are learned and used in social interaction, and ranks this interactional function higher than the human conditioning by ‘biological codes’ (see Gussenhoven Reference Gussenhoven2004).

As an example of the multimodality of human interaction (Part III), Charles Goodwin shows in a case study of a person suffering from aphasia that the same type of speech act (here: disagreement) can be expressed in a wide variety of nuances of meaning, even if lexical information is missing. The aphasic mainly uses prosody (comprising intonation and timing), in combination with other semiotic resources such as hand and body gestures. The paper clearly reveals that prosody is part of the gestural system of language, and that its intrinsically iconic nature can be crucial in order to enable a recipient of a complex speech sign to reach the interpretation intended by the speaker. Goodwin convincingly shows that the aphasic is able to build meaningful and varied conversational chunks by making use of a rich prosodic system, which even allows him to encode rather complex topic–comment structures.

Having looked at a variety of papers in this excellent collection, it is striking to see that most papers try to avoid the accusation of being too impressionistic and thus provide detailed phonetic analyses of the prosodic phenomena investigated. This may be taken as an important step towards a fruitful combination of the advantages of different types of approaches. Nevertheless, the strictly functional view of spontaneous data in social interaction in CA approaches will have to be combined with quantitative experimental techniques, including statistical analyses, to enable a systematic investigation of the relation between prosodic forms and functions. To conclude, the book is not only highly relevant for specialists in the field – in fact for all linguists adhering to approaches that deal with prosody or spoken language in general – but also for graduate and undergraduate students of linguistics.