In this book, Aijmer suggests that, as the study of pragmatic markers develops into a ‘major area in pragmatics’ (p. 1), a fruitful way into this rich topic is through the prism of variational pragmatics (Barron & Schneider Reference Barron and Schneider2008). Research into this topic, she claims, needs to go beyond a ‘general analysis’ and look more closely at use across ‘different varieties (of English), text types and activity types’ (p. 2). This is exemplified by the case studies in this book, which examine the use of well and in fact/actually in different contexts in British English (chapters 2 and 3 respectively), and general extenders such as and things like that in different varieties of English (chapter 4). Although variational pragmatics might be more typically associated with the study of pragmatic variation across regional lines, it does also concern variation in social space (Barron & Schneider Reference Barron and Schneider2008:1), so the author fulfils the promise suggested by the book's subtitle.

Chapter 1 provides some important background information. Aijmer clarifies her definition of pragmatic markers, describes the formal and functional properties she will consider, and introduces the methodology and linguistic theories adopted for the studies. Pragmatic markers are introduced first of all as indices of the speaker's metacognitive awareness, that is to say, the speaker's awareness of how the interaction is proceeding, their role within it, and whether any real-time corrections are required. They are also described as having a structural role, being used to ‘mark off segments in the discourse thus helping the hearer to understand how the stream of talk is organised’ (p. 6). The analyses in the later chapters amply bear out these two perspectives. Of particular relevance throughout the book is the notion that ‘pragmatic markers do not have a fixed meaning but a meaning potential’ (p. 12), that is to say, their meaning develops in different ways according to the context in which they are used, with different parts of the meaning foregrounded or backgrounded according to the circumstances (see also Norén & Linell Reference Norén and Linell2007). I will return to this point below.

I now turn to the two main case studies, on well and in fact/actually, each of which takes up over fifty pages. The analysis of these markers is very rich in both quantitative and qualitative detail, and this review cannot adequately summarise all the data presented. I will instead provide an overview of the case studies, drawing attention to points of particular interest or novelty. As will become clear, I am full of admiration for the content and detail of this work and at times frustrated that, in a few places, it is presented in such a way that it becomes difficult for the reader to appreciate its depth and richness.

In both cases, the data were taken from the 1-million word British component of the International Corpus of English (ICE-GB), as described in chapter 1. Among the corpus's strengths are the inclusion of a wide range of text types, allowing for the kind of inter-textual analysis envisaged, and the availability of audio recordings, which allow for prosodic analysis of the pragmatic markers. On the other hand, the reader needs to be made aware of two potential weaknesses of this corpus, namely its age – the data were collected between 1990 and 1993 – and its lack of speaker diversity, as most of the speakers are from southern England. While this does not in any way invalidate the study's findings, a brief reflection on these limitations and the possible developments of the markers studied over time might have been appropriate.

Well, the subject of chapter 2, is an interesting topic for analysis because, despite having been the object of numerous studies, these do not agree on its meaning and functions, nor do they consider how it may vary across different text types. Aijmer provides us with a description of the word's formal properties, such as position and prosodic features, and emphasises that ‘for a more fine-grained analysis of the extended meaning of well the importance of collocations cannot be underestimated [sic]’ (p. 29) since they can constrain the item's meaning. Accordingly, some of the most frequent collocates are listed (okay, now, at least, anyway) and mentioned throughout the subsequent analysis. However, the explicit link between them and meaning constraints is not discussed as fully as one might perhaps wish, given the emphasis on their importance earlier in the chapter.

Functionally, following the principle of meaning potential, well is analysed as having not a core meaning but rather ‘several different core aspects’ which carry out a variety of communicative functions (p. 30). These relate to coherence, involvement and politeness, and are exemplified with a wealth of examples in the remainder of the chapter. The three main functions are further subdivided in turn into ‘micro-functions’ or ‘sub-senses’. For example, the coherence function, which maintains connections between the different parts of discourse, includes four micro-functions: word search and self-repair, new turn projection, transition according to an agenda and transition to a quotation (p. 32). The latter, as the name suggests, occurs when the speaker is reporting another's direct speech, possibly to keep the discourse varied and entertaining (pp. 36–7), and so is easily recognised. The difference between turn-taking device and transition according to an agenda could at first appear trickier to identify as both occur utterance-initially and usually turn-initially as well (on p. 34 turn-taking well is described as a ‘device pointing forwards to the next turn or discourse unit’). The transitional function, however, is typical of contexts other than informal conversation, such as doctor–patient exchanges or classroom discourse – that is, situations where there is a structure to the exchange, and conventions that the participants are expected to follow (p. 35). In these contexts, well is a marker of the speaker's awareness of the role they inhabit, in addition to its function as marking the transition to a new turn. Activity-based distinctions such as these are not often encountered in more general discussions of pragmatic markers; Aijmer's emphasis on this is one of the great achievements of this book.

The analysis also stands out for its rich level of granularity. The various sub-senses are, in the main, explained well, but I was on occasion unconvinced by the detailed differentiation proposed. For example, among the sub-senses of the involvement function (signalling ‘the speaker's involvement with the hearer, co-operation and interest in what is going on in the discourse’, p. 37) are – among others – agreement and disagreement as separate senses. Aijmer herself acknowledges that these different functions can sometimes be ‘slippery’ to identify (p. 37) and dependent on factors such as prosody and body language. According to her analysis, agreement well ‘avoids overt disagreement . . . [it] can be used for agreement with some reservation’ (p. 38) as well as resignation and other negative emotions – ‘i.e. it is associated with disagreement rather than agreement’ (p. 39). I wonder therefore why this discussion is not subsumed in the very short section on disagreement instead, or why the two sections are not combined. Furthermore, mitigating disagreement and offering agreement are, in my view, strongly linked to politeness. Therefore I am unsure about the creation of a separate function of politeness for well (with no sub-senses), which is (briefly) defined as ‘mitigating illocutionary force in contexts which can be experienced as face-threatening’ (p. 42). My understanding is that this function includes uses where the marker is used in combination with other phrases (I don't know, perhaps, you know, I think) and is mitigating assertive and authoritative statements rather than disagreement. Despite this difference, I feel that the politeness aspect of the (dis)agreement function is sufficiently prominent to warrant grouping these functions more closely.

The second part of the chapter analyses well from the perspective of text type, in keeping with the variationist outlook of the book. It looks at its use in private dialogue (distinguishing between face-to-face and telephone conversation), public dialogue (broadcast discussion, legal cross-examination) and sports commentaries. The analysis is formal and functional. We therefore learn, for example, that the main function of well in face-to-face conversation is involvement; Aijmer describes it as a ‘friendly marker’ (p. 48). The wealth of information in these pages, including careful quantification of several aspects of the marker's behaviour, is distributed across twenty-one different tables. It might have been more helpful for the reader to have a smaller number of tables, containing more comparative data, to allow a more immediate identification of differences and similarities in the data – for example, that even in the two closely related fields of the two types of private conversation, the coherence function is more prominent in telephone conversations. Similarly, having the public discourse functions in raw frequencies rather than percentages (e.g. tables 2.17, 2.20, 2.21) makes it difficult to derive direct comparisons to the data for the private discourse functions. This is an unfortunate oversight because the private/public comparison is particularly valuable, as a clear example of how context, register and activity type give rise to and modify the pragmatic meanings of particular words, reminding us that we need more than one type of text to discover the full extent of a word's behaviour.

Chapter 3 looks at in fact and actually, defined ‘adversative’ markers (p. 74). These are dealt with together because they have some semantic similarities and have often been discussed as a pair in previous work (p. 75). The chapter follows a similar structure to chapter 2, beginning with several tables summarising quantitative information. Salient findings include the fact that both markers are more frequent in speech than in writing, that actually is more frequent than in fact, and that there is a clear difference in their distribution among text types: in fact occurs more often in monologues (parliamentary debates, legal cross-examinations, commentaries, speeches), actually more often in dialogues (conversations, telephone calls, interviews, broadcast news and talks, unscripted speeches). It would appear therefore that in fact is associated with more formal registers than actually.

A detailed formal and quantitative description follows, including information on position, prosody, collocation with pauses and – very briefly – with other lexical terms. Unlike well, the analysis also looks at written language, though it's worth noting that, as Aijmer herself acknowledges, the frequencies in written text are quite low overall (p. 78), so the findings need to be treated with some caution. An interesting point is the comparison with the markers’ frequency in the London–Lund Corpus (LLC), which covers an earlier period of time (1960s – early 1980s). It suggests that use of actually is on the rise, though Aijmer notes that this could also be due to ‘the greater formality of the conversations recorded in the LLC’ (p. 79), and, as noted above, actually appears more typical of less formal language. A similar comparison is made for the use of well in the LLC and the Bergen Corpus of Teenage Language, leading her to conclude that ‘it is possible that it is losing its popularity’ (p. 25). These tantalising observations are not pursued further, leaving the reader to expand the diachronic comparison to other data sources from both earlier and later time periods on their own.

Functional analysis reveals that in informal conversation in fact has three main functions, each with further sub-senses: adversative, elaborative and politeness. The elaborative function is found to be the most prominent, though this is on the basis of a very small sample (68 instances). The discussion covers many specific types of spoken language, but many of these share the same main functions: for example, in unscripted speeches it is used to express adversativity, counter-expectation, and to elaborate, much as in the other domains. In short, in fact is found, like well, to have a ‘meaning potential organised around several core functions’ (p. 103), activated by different contexts. The variational approach taken by this study is particularly valuable in relation to the less-examined types of spoken language, where the functions of the marker ‘depend on the social identity of the speaker, the speaker-hearer relationship and the goals of the activity’ (p. 103), as it points to new perspectives on the study of specialised language.

Although, as previously stated, actually has a different distribution from in fact, it is still a little distracting, in my view, to have its analysis entirely separate from that of in fact, especially when one of the aims of this chapter is to compare their behaviour; comparisons are made in the course of the discussion, but not consistently. Aijmer concludes that the functions of actually are the same as those of in fact: adversative, elaborative and conversation-specific, and that actually is, however, more adversative in nature. Findings such as this can have wide resonance, for example in cautioning speakers against its overuse, to avoid projecting too much hostility in their speech or writing. Aijmer also reminds us that ‘a pragmatic marker can index features such as authority and power’ (p. 125), as demonstrated by her detailed analysis of its use by authority figures such as lawyers and classroom instructors. These observations can have important implications for cross-disciplinary research into language and power, reminding us that it is all the words in an utterance which contribute to projection of identity, not just the more obvious ‘contentful’ ones.

Given the adversative nature of actually and the discussion of well as a marker of disagreement (and their occasional co-occurrence, as noted for example on p. 106), a curious (or greedy!) reader is left wishing that Aijmer had also applied her considerable powers of description to a comparison between these two markers. I can easily imagine contexts where they can be used with similar functions, as in the following (invented) dialogue:

1. A: Are you still OK to drive me to the airport?

2. B: Well/Actually, it turns out the car is misbehaving . . .

Indeed, Aijmer notes a further (implicit) functional similarity between the two: ‘actually in conversation is associated with hedging and apologising’ (p. 107).

Chapter 4, on general extenders in six varieties of English, is much shorter than the others (twenty pages). It addresses a topic which, though often focusing on a specific variety of language, is rarely viewed through a comparison of several regional varieties (though see e.g. Overstreet Reference Overstreet2012). To ensure wide coverage of all the possible variations of these phrases, Aijmer uses a ‘restricted number of collocational frames’ which combine different connectives, modifiers, nouns and comparatives to give rise to a wide range of expressions (p. 130) such as and that/this kind/sort of thing/stuff, and all those/these kind(s)/sort of things. The comparative nature of the analysis is given by the use of ICE corpora from the UK, Australia, Canada, New Zealand, Singapore and the US (in the form of the Santa Barbara Corpus of Spoken American English). As is typical of this book, the data are presented in thorough quantitative detail, highlighting numerous trends, such as the higher frequency of stuff in the US, Australia and New Zealand versus things in the UK and Singapore, with Canada – appropriately enough – not showing a preference for either. There is so much fascinating information that the reader regrets not having the chance to learn more about these differences, and the contexts in which they occur. Different functions are described, but not clearly quantified or associated with particular varieties. The strengths of this cross-varietal comparison should inspire further similar studies, especially as the ICE family of corpora continues to grow.

In the conclusions, Aijmer returns to the concept of meaning potential, which ‘consists of a specifiable set of formal and functional features’, from which, in a given social and linguistic situation, ‘a particular function is selected along with formal features such as position, prosody or collocations’ (p. 149). I believe Aijmer has been convincing in showing how this variability of meaning takes shape across so many different communicative situations, and how formal and functional features need to be considered together to fully appreciate the subtle pragmatic differences intended by the speaker. I am not sure how different this is from simply saying that we are dealing with words with very vague or unspecified meaning, which take different senses depending on context. Can there be a core meaning for at least some of these markers after all? When, in everyday conversation, someone starts a sentence with well or actually, we tend to immediately brace ourselves for some sort of negative utterance. This suggests to me that, beyond the detailed differences identified by Aijmer, there rests a commonly – if perhaps unconsciously – agreed usage for these terms.

I have dwelt at greater length on chapters 2 and 3 to illustrate both the strengths and (relative) weaknesses of this book. The latter relate mainly to occasional frustrations regarding the ‘packaging’ of the material, not its quality – because there is a richness of information, the finer details of which could sometimes elude the reader in need of more guidance. For example, the figures pertaining to the functional profiles of each marker do not appear until the end of the relevant sections, whereas I feel they would have proved more useful at the start of the discussion. The discussion of regional differences in chapter 4 is tantalisingly brief, and I was left wondering about how the data of the earlier chapters would have been enriched by analysis based on regional variation. The focus on different text types is certainly one of the key strengths of this book, as it provides us with a more comprehensive and representative view of the behaviour of these pragmatic markers.

Overall, this is a very valuable contribution to pragmatics and corpus linguistics. Although the subtitle of the book emphasises its relation to variational pragmatics – to which it clearly belongs – its strong ties to corpus linguistics are no less significant. It shows us how investigations in pragmatics can acquire a broad and comprehensive range when conducted on richly annotated corpora, and how even something as ‘slippery’ as a pragmatic marker can receive a methodologically rigorous treatment and description. As well as scholars of pragmatics, discourse and related areas of linguistics, I can also see this work being of relevance to researchers in English as a foreign language and as a lingua franca, and in the field of Human–Computer Interaction (Aijmer briefly acknowledges this in her discussion of Green Reference Green, Birner and Ward2006, on pp. 22–3). If one of the goals of the latter discipline is to create artificial systems which approximate natural human conversation, a thorough understanding of the formal and functional behaviour of these particles is necessary: as Aijmer has demonstrated in this book, they have a key pragmatic role to play in communication.