With her considerable research experience on ‘innovative’ English quotatives from the perspective of variationist sociolinguistics, Isabelle Buchstaller is extremely well placed to offer the kind of state-of-the-art overview which only a book-length treatment can provide. The first two chapters of Quotatives in particular have benefited from the broader typological and constructional perspectives embraced in the equally important eponymous volume (though with a different subtitle), co-edited with Ingrid Van Alphen in 2012, while the remaining chapters build on and expand earlier studies on the global spread and localized variants of be like and go quotatives, their diachronic development, and the attitudes and ideologies attaching to them. The chapters are all firmly grounded in the relevant and extensive literature (somewhat redundantly listed at the end of each individual chapter) and accessibly written, although perhaps numbered subsections and systematic chapter summaries might have provided explicit support for the text's implicit structure and forward dynamics.

The introductory chapter briefly lays out the scene, revisiting the published literature on innovative quotatives such as be like and go as well as other variants such as be all and this is + NP. Existing evidence suggests that be like in its quotative use emerged in the early 1980s in California to subsequently spread geographically and broaden its use from introducing thoughts and stances to speech as well, giving the variant what Buchstaller calls a ‘wild card’ status (p. 9), with speakers opting for the form in part because of its indeterminate status. As for go, it has a long pedigree as an introducer of onomatopoeic quotes, but presumably it only came to be used with ‘non-mimetic’ quotes in the latter half of the twentieth century (p. 10). Factors involved in these relatively recent changes are argued to include the pragmatic function of marking key points in narrative and cross-linguistically converging tendencies in recruiting certain types of lexical source material for quotation (comparative, demonstrative, quantifying and motion/action sources). In addition, the conventionalization of a productive construction of the form ‘Noun Phrase + Copula + (Discourse Marker) + Quote’ (p. 17), which can also cover go in its copular function (as in to go mad), facilitates analogical extensions, particularly in the variable discourse marker slot, typically showing hedging (like, kinda, sorta) or intensifying (all, pure, totally, so) function.

Against the background of these preliminaries, chapter 2 tackles the fundamental question of what is meant by quotation. In a wide-ranging discussion, Buchstaller draws on notions of evidentiality, epistemic stance and quotation as performance, and aligns herself with proposals which view reported clauses not as direct objects of transitive verbs but as elaborating a conceptually dependent reporting clause (see, e.g., Vandelanotte Reference Vandelanotte2009: ch. 2). While recognizing the oft-noted point that quotations need not have a corresponding actual prior utterance, the discussion here focuses on the inevitable imprecision in depicting prior behaviour, and on the difficulty of establishing whether a quote was outwardly realized or only thought or felt. Arguably, the more fundamental point seems to be that quotes can be inherently and deliberately without ‘original’, whether spoken or thought, as in negated and hypothetical quotes (didn't say, might have said; see, e.g., von Roncador Reference Roncador1980), in objects, such as carrier bags declaring ‘I’m not a plastic bag’ (see, e.g., Pascual Reference Pascual2014), or in memes of the type ‘Can I wear your Crocs?’ said no one ever (Dancygier & Vandelanotte Reference Dancygier, Vandelanotte, Dancygier, Lu and Verhagen2016: 28–34). In this sense, Buchstaller's definition, ‘a performance whereby speakers re-enact previous behaviour (speech/thought/sound/voice effect and gesture) while assuming the dramatic role of the original source of this reported behaviour’ (p. 54), is both appealingly rich and coherent with the aims of sociolinguistic research on quotatives, but it does not sit entirely comfortably in the context of a theoretical discussion which explicitly aims for comprehensiveness (p. 34): clearly quotation does not, as Buchstaller claims, ‘always [involve] repetition’ (p. 65).

The limitation of the book's scope to direct speech, hitherto implicit, is explicitly addressed towards the end of chapter 2, after a brief summary and discussion of differences between direct and indirect speech, in which the existence of borderline or unclear individual cases is rightly stressed. What is less explicitly addressed is the restriction to spoken data, perhaps obvious given the specific research tradition that the case studies presented in later chapters are situated in, but less obvious in the context of this chapter, which reaches out to the many other branches of linguistics that have occupied themselves with questions of speech and thought representation, often based on written sources, literary or otherwise. Insights from these lines of research have also been applied to spoken data, as in debating the vexed question of spoken forms of free indirect speech or thought, or in distinguishing other ‘mixed’ types of reporting not to be confused with direct speech (see, e.g., Vandelanotte Reference Vandelanotte2009: 238 for an example).

The final part of chapter 2 further specifies the scope of data considered in later chapters. It excludes direct repetition and reporting of inner states and mental activity; as was the case with the restriction to direct quotes, the rationale is ‘to assure comparability – regarding the definition of the variable – with the majority of previous research’ (p. 75).

Chapter 3 makes a convincing case for what Buchstaller (Reference Buchstaller2008: 15) referred to as the ‘localization of global linguistic variants’: be like in particular, while having spread globally, has adapted in its different locales to the local quotative system, displaying sensitivity to social and stylistic parameters. This finding argues against simplistic models of be like being adopted wholesale from a US model spread via mass media influence: instead, while the media can create awareness of a form, they presumably play only a facilitative role as to its adoption, all the more so given the existence of cross-linguistic tendencies supporting the emergence of innovative quotatives independently of media influence.

A detailed literature review considers past findings on the social factors age, sex and socioeconomic status, and on the intralinguistic factors temporal reference, grammatical person, content of the quote (‘speech’ vs ‘thought’) and mimetic re-enactment (traces of voice effects, sounds, gestures, postures), focusing on the UK and the US but including data for other varieties too (Canada, Ireland, Australia, New Zealand). While this review reveals a lot of diverging or even conflicting results, some generalizations do emerge concerning linguistic factors: go and (with few exceptions) be like are favoured with mimesis reporting and in conversational historic present tense contexts. Be like has also started out with a preference for thought reporting, an effect which ‘might be weakening across time’ (p. 110). Social factors, on the other hand, turn out to be localized, suggesting quite strongly that quotative be like and go are not, in Eckert's (Reference Eckert2004) terminology, ‘off-the-shelf’ changes, but instead involve ‘weak transfer’ (Meyerhoff Reference Meyerhoff2009) with some constraints remaining intact in the transfer, and others undergoing substantial transformation (pp. 116–17). In all this, Buchstaller rightly stresses the possibility that cross-study comparability may not be fully guaranteed, due to ‘methodological and conceptual differences across investigations’ (p. 110), as with the (partial or total) exclusion of it from the analysis in some studies.

In this connection, it is worth noting that, from a qualitative perspective, some discussion of the functions of it's like might have been welcome. Contexts of use for it says are easily understood, with it referring to some written source or institution. However, in some of the relatively few examples quoted by Buchstaller, the meaning of it's like seems to be ‘the situation is such that one might say’, whereas in other cases it seems almost interchangeable with I’m like. Rather than treating it as just an option for the grammatical person parameter, it merits fuller discussion in its own right.

Returning to Buchstaller's third chapter, its second half is devoted to a comparison of the findings reported in the literature with two recent datasets (2004–8), one collected in Northern California and the other in the North East of England (Newcastle), both involving adolescents and young adults. The results for the different conditioning factors investigated are rather complex, often ‘support[ing] some studies and contrast[ing] with others’ (p. 135), but one conclusion which stands out is that ‘the innovative quotative form continues to be governed by idiosyncratic patterns on both sides of the Atlantic’ (p. 136), underscoring the localization factor discussed above.

While chapter 3 focuses in the main on the UK and the US, some other inner- and outer-circle varieties of English are also reported from existing studies. To sketch in more of the global context of quotative be like in particular, the corpus of Global Web-based English (GloWbE; Davies Reference Davies2013) should prove useful, as it represents mainly informal language across twenty countries (with nearly 60 per cent of texts taken from blogs; see Davies Reference Davies2014). Usage levels of be like expressed in occurrences per million words can be seen to range from the low end (e.g. Sri Lanka, 0.28; Pakistan and Bangladesh, 0.68) all the way to the top scorers Canada (2.29) and the US (2.50). Above and beyond crude statistics, a wealth of data from this source awaits qualitative exploration.

The fourth chapter provides a longitudinal study of change in the quotative system based on three sets of spoken English data collected in Tyneside (Newcastle, Gateshead and surrounding urban areas), stratified by age. The repeated reference throughout the chapter to the ‘60 years of data from Tyneside’ (p. 184) – with a single exception where ‘50’ is mentioned (p. 162) – is rather puzzling and hard to substantiate, since the earliest data included stem from the Tyneside Linguistic Survey (TLS), collected from 1969 onwards and into the early 1970s, and the most recent of the three data sets ‘was assembled during a corpus-building project that started in 2007’ (p. 150) and, apparently, concluded in 2011.Footnote ¹ This surely suggests something closer to forty years being covered by the data.

Putting this minor puzzle aside, the interesting thing about the data sets is that, thanks to the TLS, they include ‘a snapshot of speech and thought reporting before be like and fully quotative go had gained currency in the UK’ (p. 151), when say and think together made up around 60 per cent of the variable, independently of age. By the first half of the 1990s, be like and go took up 6 and 14 per cent respectively; by the late 2000s, the former increased to 21 per cent while the latter dropped off to 4 per cent. An age gap had also developed by the 1990s and widened by the 2000s, with the younger speakers’ system relying on be like in 40 per cent of cases, and relegating say to a mere 17 per cent, as against 3 and 58 per cent respectively for older speakers. In terms of genre, the data for younger speakers from the 1990s and the 2000s show an increase in use of other forms than say in narrative contexts (showing only 14 per cent say use among young speakers in the 2000s). It might be noted in passing that the distinction between ‘narrative’ and ‘other’ (including, e.g., ‘chit chat’ and ‘reminiscing about the past’, p. 176) used here seems rather sweepingly broad, and not very clearly drawn. In general, the diachronic findings show that the repertoire of quotative forms has expanded and become decidedly more diverse in terms of functional niches.

The final section in the fourth chapter suggests that in the late 1960s/early 1970s, when speakers rely heavily on say, they ‘compensate for the comparatively low formal diversity of their quotative system by creating variability via aspectual and temporal marking’ (p. 185), as well as via the position of the quotative clause and the use of so-called ‘non-veridical’ (interrogative, negated and conditional) reporting clauses. This idea does not seem to be convincingly demonstrated, given that the effects observed in the data are rather small and, in the case of non-veridical clauses, not consistent with the hypothesis for the later period. Buchstaller admits that ‘speakers might not aspire to be maximally variable’ (p. 189).

Chapter 5 opens with a number of amusing illustrations of prejudice against speakers using like – or indeed against researchers expending valuable time and energy writing seriously about the form. The bulk of the chapter is devoted to case studies investigating attitudes towards the new quotatives, their association with personality traits, their perceptual geographies (where do people think users of the quotatives come from?) and social perceptions attaching to them, based on a survey conducted in New York and Newcastle. The chapter frames this as a comparison between the US and the UK, which perhaps might have been hedged a little more. Indeed, at one point Buchstaller points out that results for attitudes towards go in a study with respondents from Edinburgh were very different from those found in Newcastle in this survey (p. 220). Unsurprisingly, older informants in the attitude survey are much less accepting of the quotatives (significantly so for be like) than younger ones in both locales. As for perceptual geographies, to give one example, users rate the use of be like as more British in Newcastle and more American in New York, whereas go is not felt to be typically British or American in Newcastle and New York respectively. One of the most striking results of the New York survey was that the question ‘Where do you think it [like] comes from’ was mostly left blank (86 per cent), suggesting that the easy equation between be like and the cultural stereotype of the (inarticulate, underachieving and be like using) Valley Girl has less reality than it gets credit for.

The sixth and final chapter considers the theoretical significance of the book's findings for language change and linguistic globalization. While the extension of be like and go is ‘motivated by cross-linguistic channels’ (p. 248), the process is not globally uniform, as local adaptations have given rise to ‘idiosyncratic routines, balances and trajectories’ (p. 249). Considering the question on which level of linguistic structure (lexical, discourse-structural or morpho-syntactic) quotative innovations are situated, Buchstaller's discussion concludes by stating inconclusiveness (p. 256). After reminding us that the ‘innovations’ are not truly ‘new’ – in that they ‘cannot have come into the language with the current generation of young speakers’ (p. 256) – the chapter turns to questions of the present and future of the forms. Based on aggregated data from a number of studies, Buchstaller argues that to paint a full picture of the sociolinguistic reality of these forms, and particularly of be like, we need to draw not only on age grading and change in progress (a change driven by younger speakers, but apparently arrested by the time speakers reach their thirties), but also on lifespan change, evidenced in the behaviour of older speakers who start to ‘embrac[e] an ongoing youth trend which middle-aged speakers shy away from’ (p. 264). As for go, it turns out to show ‘variability that is neither consistently age graded nor does it ever change to completion’ (p. 268).

There are, perhaps inevitably in a book-length study, the occasional minor editorial glitches, including misspelt names and a handful of imprecise references. The style is at times a little wordy and, particularly in the last chapter, sometimes repetitive, but always clear and easy to follow. In terms of methodology, some choices are perhaps debatable from a qualitative perspective; in particular, the decision to ‘consistently [code] combined forms as the respective first quotative item’ (p. 140), discussed in a lengthy endnote to chapter 3, effectively means that these forms disappear from view, since instances such as go like and all like are subsumed under go and all respectively. However, the important point is that time and again methodological decisions are explicitly recorded and addressed, providing excellent groundwork for possible future follow-ups and replications.

Summing up, this book engages with the wider research traditions in speech and thought representation and offers state-of-the-art analyses of variation and change in the English quotative system across time, place and attitudes, with a focus on be like and go. There is more work to be done in bridging the gap between variationist sociolinguistics and other kinds of constructional, diachronic and discourse-oriented studies of the forms and functions of speech and thought representation, but current and future students of the field will ignore this landmark study at their peril. For its exemplary methodological rigour and finesse, its comprehensive meta-analysis of previous studies coupled with original research on new data sets, and its inclusive embedding in existing literature on quotation and sociolinguistic innovation, it is to be highly commended.