Karin Aijmer and Christoph Rühlemann's Corpus pragmatics: A handbook represents a coming of age for the growing field of corpus pragmatics, ‘filling a gap in the literature’, as the cover notes tell us. A general introduction is followed by sixteen original research articles by leading scholars in their fields that overview existing work, providing new perspectives on the use of corpora to deal with problems of pragmatics. The contributions are amply illustrated with figures and tables; a final author and subject index provides an excellent reference tool.

Christoph Rühlemann and Karin Aijmer's introductory chapter, ‘Corpus pragmatics: laying the foundations’, begins with a brief definition of pragmatics and corpus linguistics in turn. On the one hand, the ‘intricate contextual embeddedness of communication’ that interests researchers in pragmatics requires close qualitative analysis of texts, a ‘horizontal reading methodology’ (p. 3), as the authors put it. On the other hand, corpus linguistics, which Aijmer and Rühlemann characterise in terms of the type and size of corpus, the annotation conventions, etc., tends to operate along a ‘vertical reading methodology’ (p. 8), epitomised by the KWIC window of automated corpus analysis tools, providing quantitative information about frequencies within texts. These two approaches are brought together in corpus pragmatics in what the authors present as an integrated approach. One may, for example, submit the results of corpus queries to the close readings of pragmatics. Another possibility is to use purpose-made corpus annotations for functions which do not correspond neatly to forms, in the case of speech act analysis, for instance. The contributions to Corpus pragmatics: A handbook are organised into six broad sections: speech acts, pragmatic principles, pragmatic markers, evaluation, reference and turn-taking. Each section consists of two or three articles by leading researchers, organised in two main parts: an overview of the issues involved and of the main currents of research in the particular field, and an original case study. These editorial constraints are intended ‘both to overview and expand the field of corpus pragmatics’ (p. 13). This review will present each contribution individually before making an overall assessment of the volume.

Chapter 1 of part I, ‘Corpora and speech acts’, ‘Speech acts: a synchronic perspective’, by Paula Garcia McAllister considers the problem posed by the automatic detection in corpora of indirect speech acts, which have no easily recognised linguistic expression. After a brief review of some contributions to corpus-based research on speech acts, Garcia moves on to a case study on directives in academic contexts, including student–lecturer office hour exchanges, study groups and service encounters. A recorded corpus was manually tagged and coded for ten speech act types, defined with examples. Corpus tools were used to observe modals, hedges, amplifiers and syntactic structure (in particular as this refers to mood). Results are summarised in tables, showing that requests and suggestions are by far the most frequent speech acts in the chosen context. Garcia McAllister provides analysis for each type of situation in turn. The study shows a predictable correlation between situation type and the types of speech act. Requests for information were more frequent in service encounters while suggestions/commands predominated in office hours situations, for example. A final discussion and conclusion highlight the findings – and some shortcomings – of the study and indicate directions for further research.

Chapter 2 by Thomas Kohnen is entitled ‘Speech acts: a diachronic perspective’. As in chapter 1, Kohnen divides his study into an overview of research and then an original case study. He points out that diachronic analysis of speech acts poses the question of whether we know enough about communicative situations of the past to interpret historical data meaningfully. Kohnen divides previous research into three approaches. The first of these, dubbed ‘illustrative eclecticism’, involves simply using corpus examples for illustration. He quotes in this respect studies on the decreasing power of the legally binding spoken promise over the medieval period, for example. The second approach, ‘structural eclecticism’ might select manifestations of a speech act for study over a period. In this approach, he quotes a previous paper of his, showing that directive performatives in Old English in the Helsinki Corpus are some seven times more frequent that in the LOB Corpus. A third section, ‘Beyond eclecticism’, looks at extending time-consuming ‘manual’ analysis of speech acts in small corpora to developing extrapolative search algorithms for automatic speech act retrieval on larger bodies of texts. The final section provides a corpus-based study of the development of English performatives from the fifteenth to the late nineteenth century. Frequencies of commissives and directives are shown to have dropped significantly; representatives have undergone a rise–fall movement, while frequencies for expressives have remained relatively stable. These developments might be explained in terms of changes in face and literacy over the period, Kohnen hypothesises, before suggesting several directions for future research.

In chapter 3, ‘Speech act annotation’, Martin Weisser addresses issues involving the manual or automatic annotation of pragmatic phenomena in corpora. Existing practices are reviewed and directions for improvement explored, in the form of Weisser's own research tool DART. To illustrate the problems annotators face, Weisser focuses on backchannels which, if they appear as separate turns, become structurally indistinguishable from interruptions, whereas their function is on the contrary to provide uptake and encouragement to continue. He goes on to compare three main schemes for speech act annotation of texts, known acronymically as DAMSL, SWBD DAMSL and his own DART. He first considers the SGML conventions of each scheme before moving on to the design. The implementation of the schemes is compared on a number of corpora, with Weisser pointing out inconsistencies or incongruities as the discussion progresses. A brief final section provides an illustration of the use of the DART annotation scheme applied to a corpus of call centre exchanges involving a British agent. The schema appears to identify with precision the pragmatic features that contribute to the service efficiency of one particular employee.

Part II, ‘Corpora and pragmatic principles’, begins with Gunther Kaltenböck's ‘Processibility’. The chapter focuses on the development of what are termed ‘pleonastic matrix clauses’ such as I think that. . . it's just that. . . etc. One explanation for the use of such clauses is that they are considered to render information processing easier. Kaltenböck's chapter is organised into three main parts. Firstly, he provides an overview of the concept of ‘processibility’, looking at the given/new and topic/focus distinctions and the related principles of end-focus and end-weight. Secondly, he cites work by Boye and Harder which hypothesises a link between ‘primary’ and ‘secondary’, or ‘foreground’ and ‘background’, uses of linguistic items and their progressive grammaticalisation, i.e. the notion that linguistic structure is moulded by usage patterns. Thirdly, he investigates the development of a number of such structures, including I think that. . ., it turns out that. . ., The thing is. . . and It's just that / like. . . and related structures. Corpus observation demonstrates a common pattern towards grammaticalisation, involving the omission of the dummy subject it, or the complementiser that and also, interestingly, the emergence of new forms of determination (my thing is. . .) or the replacement of that by like (it's just like. . .). The diachronic development of each phenomenon is effectively illustrated by graphs and histograms. Kaltenböck's conclusion confirms Boye and Harder's suggestion, in that he sees processibility affecting usage, which in turn affects structure. This appears also to demonstrate the usefulness of corpus study of such phenomena, although the author does point out the problems of reliable or sufficient spoken language data in diachronic studies of this type.

The following contribution by Gisle Andersen is entitled simply ‘Relevance’. Andersen first provides a useful and complete overview of Relevance Theory and its relation to corpus studies. He adopts a critical position with regard to the frequent neglect of diachronic phenomena within relevance-oriented studies. He also notes that the theory has been reproached with focusing on invented examples and contexts, to the detriment of naturally occurring language, but considers nonetheless that more recent relevance-theoretical work pays more attention to empirical corpus evidence. Andersen's case study focuses on the emerging discourse markers as if! and duh! the uses of which he has explored in a range of English and Norwegian corpora. He is interested both in characterising these markers from a relevance-theoretical perspective and in observing the conditions under which they have been borrowed into Norwegian. As if! involves a backward-oriented emphatic rejection, while duh! is used patronisingly to indicate that a proposition is obviously true. Andersen does not aim at quantitative analysis in his use of corpora, which here provide authentic examples from English and Norwegian to illustrate the processes of emergence and borrowing of two innovatory interjective pragmatic markers.

Giuliana Diani's chapter entitled ‘Politeness’ begins by charting the recent rise of research in politeness and the relationships between this field and work on corpora. The main body of the chapter is a comparative study of how politeness – and the ‘mitigation strategies’ thereof – manifests itself in academic book reviews written in English and Italian. In previous studies of academic discourse, there is some consensus around a general praise–criticism–praise format, but the use of oppositional strategies appears to be subject to cultural variation. To verify this, the author works on a corpus of forty or so English and Italian academic book reviews, using Wordsmith Tools. Diani differentiates direct from indirect criticism, which is expressed via four mitigation strategies: praise–criticism, hedges, criticism–suggestion, praise–suggestion. The first two strategies represent almost 90 per cent of the instances in English texts, which also feature a far greater proportion of direct criticism than the Italian corpus (by a ratio of over 5:1). The texts in Italian favour the strategy of praise–suggestion for indirect criticism, display a higher frequency of criticism–suggestion and employ fewer hedges. The differences between English and Italian owe something, Diani suggests, to research cultures based on high competitiveness, in one case, and tighter, collaborative networks, in the other.

Part III, ‘Corpora and pragmatic markers’, begins with ‘Pragmatic markers’, the contribution of co-editor Karin Aijmer. Her chapter is divided into a first section, which looks at the problems of research into pragmatic markers, as well as a number of different theoretical approaches, and a second section, which investigates the marker I think, across text types. Aijmer suggests that we need to go beyond merely Gricean maxims to explain the use of pragmatic markers. She looks in turn at the role played by questions of processibility (cf. Kaltenböck), principles of politeness, indexicality/reflexivity, position, function and context, before considering the more recent use of electronic corpora – whether synchronic, comparative, parallel or diachronic – in the study of pragmatic markers. Her case study presents statistics for I think in different text types (drawing on the ICE-GB corpus), and focuses on a limited number of examples in conversation as compared with broadcast discussion. In conversation I think was used initially most of the time where its functions include the expression of subjectivity, evaluation, hedging and floor holding. There is for Aijmer little overlap between these and uses in the more rule-bound activity of broadcast discussion, where the marker plays a more prominent argumentative role and where speaker authority is important for interpretation.

‘Stance markers’, jointly written by Bethany Gray and Douglas Biber, opens with an overview of the linguistic means of marking stance and the main approaches in research on these questions. Markers of stance include lexicon, complement clauses controlled by stance markers, adverbials, modals and certain nouns which are classified according to explicitness. Corpus studies typically adopt a lexico-grammatical approach, targeting lexical items that express stance in certain grammatical environments. In terms of register, stance is marked more overtly in spoken texts, and, among written texts, more in fiction or news. Gray and Biber ask whether academic genres might not also encode stance, but in more implicit ways. First they investigate the relatively common stance adjective- and noun-based constructions. Tables show that extraposition is particularly productive in academic writing compared to other registers, and that, within this category, markers of epistemic certainty or likelihood are more common than in the other categories. The authors then consider the use of the same adjectives and nouns outside contexts governing complement clauses, to see whether they might also carry indications of stance in other constructions. They identify seven new stance-bearing lexico-grammatical patterns that have passed under the radar of traditional searches, suggesting that stance markers in academic register are more productive than has often been presumed. The study concludes with directions for further research and a useful appendix of adjectives and nouns used in the analysis.

Neal Norrick's ‘Interjections’ first presents the difficulties in conducting corpus research on interjections, including the delimitation of the set of interjections and inconsistent mark-up, which often implies a need for close qualitative analysis. An overview of the literature reveals that an initially dismissive view of the category has given way to a recognition of the place and diversity of interjections within the linguistic system. One can distinguish primary (e.g. uh) and secondary (e.g. boy) interjections, the first of which can additionally be divided into expressive, conative, phatic interjections and routine formulas, though Norrick recognises that research still has a long way to go before firm categories can be established. The corpus studies pinpoint a number of key areas of enquiry. He first shows that primary interjections tend to figure clause-initially. Secondary interjections are much harder to search for, since they have other functions in the language. Certain interjections show affinities with exclamatives, while others appear to flag sequences as inserted dialogue. An interesting category of rarely tagged phrasal interjections (e.g. I tell you what and variants) also poses challenges for corpus research, as do combinations of interjections: Norrick shows how the interpretation of an oh yeah sequence depends very much on its prosodic contour. A short summary concludes the contribution.

Part IV, ‘Corpora and evaluation’, opens with Alan Partington's presentation of ‘Evaluative prosody’. The simplest form of evaluative prosody is found in collocations, which he illustrates with the opposition between KWIC examples of the positive brimming with and the negative fraught with taken from an original newspaper corpus. Partington uses a short extract to illustrate how evaluative prosody employed over stretches of text results in what he terms evaluative consistency, or harmony. He then goes on to consider nine commonly held properties of evaluative prosody. Evaluative prosody is said to be collocational. It has been claimed that speakers are unconscious of this aspect of language and that it does not attach to individual items, but is essentially distributive, affecting groups of words and phrases. It has also been argued that evaluative prosody is related to semantic preference, and that it is distributed synchronically across co-text. At the same time, new prosodies are thought to develop diachronically. It is, from one researcher to the next, considered either a pervasive or a marginal feature of language use. Lastly, evaluative prosodies may vary across different text types such that what is evaluative in one genre, is not in another. The main part of Partington's contribution is devoted to a brief presentation and critical discussion of each of these purported properties.

Ivor Timmis’ ‘Tails’ looks at right-dislocation (e.g. People said they'd never, never catch on, teabags) in terms of the syntax, pragmatics and variational properties of the construction, using three comparative corpora of spoken English (BNC, ICE-Ireland, and the 1930s Bolton corpus). Four tails are distinguished by type: full noun phrase, pronoun, inverted operated and simple operator. The frequency of tails is disputed and hard to assess but the author considers it at least as significant statistically as ought to or the get-passive. Timmis does not subscribe to the view of tails as mere afterthoughts; he considers the views that they express emphasis, clarification or focus, concluding that tails are essentially evaluative. Frequencies of tails in the three corpora studied vary considerably. Timmis explains this, in particular for the Bolton corpus, by the conditions in which the corpus was compiled. He then looks at sociopragmatic variation with tails by type and by corpus, presenting results in table form. His analysis shows that pronoun tails are typically associated with evaluative comment and operator tails with negative evaluation, while inverted operator tails are associated with named individuals. Specific usage and frequency of tails in the Bolton and Irish English corpora prompt Timmis to conclude that the use of tails appears to correlate with the affirmation of community identities.

The first chapter of part V, ‘Corpora and reference’, entitled simply ‘Deixis’, is written by co-editor Christoph Rühlemann and Matthew Brook O'Donnell. Various facets of deixis ‘an interface linking language and situational context’ (p. 331) are characterised by the authors in the first part of the chapter, who look in turn as its relationship with reference, anaphora and gesture. They consider the translation of deictic origin, or ‘origo’, and distinguish categories of deixis, including time, space, person, discourse and social deixis. The case study investigates the use of introductory this in oral narratives from an original corpus derived from part of the BNC with enriched annotation specific to the oral narrative genre. This includes, for example, a division within narrative into initial, medial and final sequences. Using computer-assisted statistical analysis the authors test the hypothesis that introductory this might be found more frequently in initial sequences of narrative than elsewhere, exploring the marker's syntactic and semantic environment in terms of colligations and collocations. This does indeed appear from the results to avoid narrative-final sequences and to occur most frequently narrative-initially, where it most often introduces new referents, functioning as a discourse-deictic theme marker and contributing to the processibility of the narrative as a whole. The conclusion summarises the preceding text and points to new directions for related research.

‘Vagueness’, by Winnie Cheng and Anne O'Keeffe, opens with a consideration of how certain expressions can, despite their vagueness, construct successful reference in appealing to common grounds between speaker and hearer. The authors present different subcategories of vague language and then move on to a short overview of research on the theme, which has focused on a variety of different genres and registers to show how vague language performs ‘important social, interpersonal and interactional functions’ (p. 363), in face management for example. Cheng and O'Keeffe's case study investigates vague approximators in number-related expressions, working on two corpora, the Hong Kong Corpus of Spoken English and the LCIE of spoken Irish English. This approach is hoped to favour intercultural comparison of the use of vague language. However, the results appear on the whole fairly homogeneous between varieties, although the authors do claim to detect cultural and contextually specific differences in the use of about + number. The study finishes with consideration of the problems created by the fact that spoken corpora are not generally annotated for vagueness, and so close qualitative analysis is necessary to apprehend the phenomenon.

Gunnel Tottie opens part VI, ‘Corpora and turn-taking’, with ‘Turn management and the fillers uh and um’, an analysis of uh and um, considered together as UHM, using the Santa Barbara Corpus of Spoken American English. She briefly presents conversation analysis and linguistic perspectives on turn-taking before moving on to corpus-based studies of UHM specifically as a turn-holding device. Contra previous studies, the author does not consider that preventing other speakers from taking the floor is necessarily the main function of UHM. The case study shows UHM to be more frequent in formal, institutional contexts, and less so in private contexts. Tottie first considers the practical (and consequently theoretical) difficulties in identifying and coding occurrences of UHM for position, in view of phenomena of overlap or ‘turn sharing’. The data show, in keeping with previous studies, that UHM is largely attested in medial position. Initial UHM is frequent in answers to questions, medial UHM is associated with forward-planning and final UHM is turn-yielding, although whether this is planned for by the speaker or simply a de facto change of turn is debatable. In conclusion, Tottie questions what she terms an ‘economy-based model’ of turn-taking and favours a more collaborative effort ‘rather than . . . a competitive fight for the floor’ (p. 403).

Pam Peters and Deanna Wong look in chapter 15 at ‘Turn management and backchannels’. Backchannels are presented in standalone or string usage (i.e. yeah vs mm yeah right), providing support in ‘turn maintenance’ (p. 410) for the current speaker. Working on telephone conversations from the Australian component of the International Corpus of English, Peters and Wong align written transcripts and oral sources, paying particular attention to the duration of the most common backchannels (mm and yeah) and the preceding and following pauses. Five research aims and the procedures used to address the questions raised are clearly presented. Results show that the longer the speaker's turn the more backchannels accompany it, suggesting a turn maintenance function. However, the authors note significant temporal differences between backchannels according to their position in the turn, suggesting a corresponding change in function. Differences also emerge between backchannels used as standalone items or in strings. Standalone uses are functionally similar to string-final uses, suggesting turn-changing functions. Consensual uses of yeah display phenomena of temporal alignment between speaker and backchanneller, while mm may by contrast provoke temporal disalignment as speakers take longer to resume their discourse. The authors conclude that backchannelling might play a greater role than previously recognised in conversation management.

The final chapter of the volume is Brian Clancy and Michael McCarthy's ‘Co-constructed turn-taking’. While, for the researcher, transcripts of conversation often appear ‘messy and incoherent’, speakers on the other hand have the impression that things are running smoothly. This, it is claimed, is because talk is co-constructed. The authors’ aim in this chapter is to look at collaborative completion or extension of utterances. Their overview of research on conversational co-construction points out in particular the difficulty in reconciling either traditional clausal or syntactic divisions of autonomous speech acts with the construction of meaning in authentic conversation. The case study uses material classed as intimate or socialising in the British English CANCODE and Irish English LCIE corpora, focusing on turns beginning if, when and which, appended to a main clause uttered by a previous speaker. A series of corpus examples is presented, showing how turn-initial if, when and which are used to complete the clause along similar lines, but also to qualify the clause, commenting, modifying or assessing its content in some way. The conclusion notes ‘regularities in the co-construction of utterances’, including ‘conventionally sanctioned opportunities for co-constructions’ (pp. 448–9). Such observations plead for a concept of conversational confluence in opposition with individually assessed fluency.

Corpus pragmatics: A handbook is a very welcome publication, providing in one volume an excellent introduction to a range of original research in this new and burgeoning field. The articles are of a consistently high quality and the careful editing has produced a very coherent final result. It would be possible to take issue with certain points in certain contributions, but such is the nature of academic research. The overarching editorial project of organising each article according to the same form – an overview followed by a case study – has its advantages and disadvantages. In its favour is the fact that the overview of each contribution provides an excellent starting point for the researcher who wishes to become familiar with a new area, including a full, commented bibliography of the most important research. The case studies that follow provide the reader with hands-on applications of research perspectives, but sometimes lack the necessary space to present things in a fully satisfactory manner. This is a minor quibble, however, and probably unavoidable in an ambitious volume which by and large more than succeeds in reconciling thorough and helpful presentations of the state-of-the-art with exciting original research.