Comprising six chapters, each accompanied by a dedicated reference section, this book provides a comprehensive overview of research on quotative variation and change, widely acknowledged to be an ‘extraordinarily dynamic domain’ (1). Isabelle Buchstaller is eminently qualified to address the diversity of topics covered in this book, having established herself as a leading scholar of quotative constructions.
Particular attention is paid to the emergence of innovations such as quotative be like and go, and the consequences of their incursion into the quotative system. Although the most detailed analyses are based on North American and British English, a notable strength of this book is the effort invested in situating quotative variation and change in a broader typological and crosslinguistic perspective (chapter 1). Another commendable feature is the discussion of methodological issues, including consideration of how quotation should be defined, as well as the problematic issue of distinguishing direct and indirect speech (chapter 2). The analytical framework underpinning the examination of the quotative system is embedded in quantitative sociolinguistics (4). This framework is used to elucidate variety-specific and cross-varietal constraints on variant selection (chapter 3) and to explore the dramatic restructuring of the quotative system in Tyneside English, in northeast England, drawing on a number of real-time cross-sections in the DECTE corpus (chapter 4). A quantitative methodology is additionally used to probe speaker perceptions of recent innovations in the quotative system, focusing on be like and go in British and US varieties (chapter 5). Chapter 5 also offers an insightful excursus on normative responses to be like and go, revealing, for example, that the association of the former variant with young people's presumed ubiquitous use of like has elicited sustained critique in the US, as demonstrated by a campaign waged against ‘mallspeak’ on American college campuses (235). A distillation of the major research findings is included in the final chapter (chapter 6), which engages with a range of highly topical and controversial issues, such as whether the dramatic ascendancy of quotative be like represents the ‘same’ innovation across different varieties, or is better construed as an independent parallel development (248). Of particular interest in this chapter is the discussion of how ongoing change in the use of be like should be modelled, an enterprise that arguably necessitates a real-time perspective supplemented by insights from age-grading and lifespan change (264). A valuable ‘take-home message’ foregrounded in the last chapter is that an adequate understanding of innovative quotative use can only be achieved when emergent variants are situated in the broader linguistic system of which they are part (251).
Buchstaller's felicitous combination of quantitative and qualitative perspectives, deftly integrated into a multifaceted vision of quotative strategies, greatly enhances the appeal of the book, which will be of potential interest to a broad range of researchers with different theoretical and methodological orientations to the study of reported speech, as well as scholars of language variation and change. Space constraints do not allow me to do justice to the book's many valuable offerings, so I confine my remarks to what I take to be the highlights of the book, in addition to focusing on controversial aspects meriting critical evaluation.
As indicated above, one of the strengths of this book resides in its efforts to situate change in the English quotative system in relation to the existence of superficially parallel innovations in a number of unrelated languages. From a typological perspective, there are clear crosslinguistic trends in the recruitment of a range of functionally polyvalent lexemes that are used in routinized constructions to introduce reported speech, thought, sound, and gesture, challenging the orthodox equation of ‘prototypical’ quotative markers with verbs of speech (Güldemann Reference Güldemann2008). Recent interest has focused on similative quotatives, such as be like in English and its crosslinguistic analogues (see Güldemann Reference Güldemann2008:317–28), in an attempt to elucidate the functional repertoire of such constructions, as well as the processes by which they emerge and subsequently spread. A key question concerns how innovations such as be like are rapidly diffusing across disparate geographical locales, and whether they are transmitted ‘wholesale’, with their social and linguistic conditioning intact, or undergo modification as they are integrated into variety-specific quotative systems. Among the contributions that Buchstaller makes to these questions is her suggestion that a constructional template, that is, NOUN PHRASE + BE + DISCOURSE MARKER (all, like, totally, etc.), serves as a productive schematic for the emergence of new quotative constructions in English (17). With regard to the supralocal spread of new quotatives, Buchstaller concedes that the media, uncritically assumed by many to be the primary agent in the long-distance transmission of be like, may play a role in the dissemination of the form, but is considerably less likely to be an important vector for the transmission of the underlying set of constraints on variant use (97). This is because the high-context knowledge associated with underlying grammatical constraints is believed to require sustained face-to-face interaction in order to be successfully transmitted (Meyerhoff & Niedzielski Reference Meyerhoff and Niedzielski2003:537). As innovations such as be like spread transnationally to other varieties from a presumed American epicenter (6), Buchstaller argues that they enter into localized competition with older variants in the quotative system (112), culminating in ‘locally specific constraints’ on usage (114).
Insofar as there are defects in the quality of Buchstaller's argumentation, these lie primarily in the quantitative reasoning associated with the analyses of the synchronic and diachronic datasets presented in chapters 3 and 4. Somewhat disconcerting are the competing distributional models used by Buchstaller to depict the influence of independent variables on the choice of quotative variant. In addition to showing the distribution of variants within each independent factor conditioning variant selection, as is conventional in variationist analysis (Tagliamonte Reference Tagliamonte2006:192–93; Walker Reference Walker2010:26), Buchstaller chooses to display distributional rates using the individual variant as the denominator (121), calculating the proportion of tokens across factors as well as within factors. The motivation for this decision resides in Buchstaller's questionable conviction that these alternate distributional analyses are fully justified by ‘multiple skews’ in the data (120), and offer a ‘more encompassing picture of the patterning in linguistic space’ (129). Among the problems associated with this decision are the difficulties that inevitably arise when alternate distributional models lead to conflicting interpretations of the data. To take one example, the results for the effect of mimesis on younger US speakers' choice of quotative variant, calculated according to the proportion of tokens across factors in Table 3.17a and within factors in Table 3.17b (133), indicate that be like is more likely to occur with mimetic quotes in the former, but is less likely to do so in the latter model. This conflict is not satisfactorily resolved by Buchstaller, who, ‘orienting to majority trends and tendencies’ (135), simply opts to base her synthesis of the results in chapter 3 on the proportion of tokens within factors, in line with standard variationist practice.
These problems are further compounded in cases where Buchstaller's interpretation of the results is not fully accountable to the data. Thus, in chapter 4, in Table 4.30 depicting the proportion of quotative variants used by young Tyneside speakers in narrative and nonnarrative contexts respectively (178), Buchstaller notes that the overall proportion (40%) of be like in narrative contexts indicates that it is the lead variant in this environment (179). While this is true, at least according to Buchstaller's method of calculation based on the distribution of variants across factors, there is no concomitant acknowledgement that be like occurs at a proportionally higher rate (60%) in nonnarrative contexts. Inspection of the corresponding distribution of tokens within factors in Table 4.30 (relegated to Appendix 2, p. 281) reveals that rates of be like in narrative and non-narrative contexts pattern in the opposite direction to the data tabulated in Table 4.30 in chapter 4. To be sure, expert readers can navigate their way around these shortcomings, but those who are less experienced are liable to find themselves disorientated by conflicting distributional information, especially in cases where there is no explicit guidance on how to adjudicate between competing analyses. The real issue here is the failure to confront the fact that distributional imbalances in linguistic data are an intrinsic property of naturally occurring discourse (Bayley Reference Bayley, Chambers and Schilling2013:90), rather than a defect of any specific quantificational model. More important, the conventional approach to conducting a distributional analysis by calculating the proportion of variants within each factor is not simply dictated by superficial methodological considerations, it is rooted in the primordial Labovian principle of accountability. This foundational principle is based on a particular theoretical conceptualization of the linguistic variable, as well as an empirically accountable framework for modelling speaker choices (Labov Reference Labov1969).
In sum, any regrettable blemishes in the exposition of the quantitative analysis should not detract from the overall calibre of the book, its rich and variegated findings, and its irrefutable capacity to stimulate further research.
