This book's sole thesis is descriptive: ‘LIKE is systematic, layered and grammatically embedded’ (31); it is framed in large part as a corrective for sociolinguistically ignorant commentary—a function that it easily fulfils. I review the book here with a sociolinguistics readership in mind, though the book will also be of interest to historical linguists and those working on language ideologies.
There are seven chapters, which I discuss in turn below, after first outlining the structure of the book and making a few general observations. The first chapter is an introductory survey of the historical development of LIKE. This is followed by a chapter describing the corpora used in the book. Next are four chapters on the variable discourse-pragmatic use of LIKE viewed from various analytic perspectives (historical, grammatical, social, and ideological), and then a final chapter considering how these perspectives feed into three incipient research areas. These chapters are conceived as self-contained, which leaves the book a little uneven stylistically, particularly in terms of signposting within and between chapters. The book has twenty-eight pages of references, a short index, and an interesting thirty-page ‘anthology’ of all of the linguistic examples of LIKE in the volume (less the ambiguous ones) together with additional examples, all sorted by year of occurrence and pragmatic function, and including references to their locations in various corpora. It would be useful if an online version of this selection accompanied the book. The sixteen colour charts and graphs could generally do with some reformatting, for example, where irregularly scaled time axes give a misleading impression of the rate of grammatical change in apparent time.
In chapter 1, the introduction, D'Arcy mentions the thorny problem of functional equivalence and polysemy demonstrated by LIKE. Citing somewhat mixed phonological evidence, she strongly rejects the idea that LIKE is a single lexical item, while also noting that there is insufficient empirical data available at present to adequately explain its remarkably extensive homophony. D'Arcy then sets about introducing the ‘myriad’ uses of LIKE, invoking Romaine & Lange's (Reference Romaine and Lange1991:244) analytic distinction between discursive and nondiscursive functions and asserting that this ‘aligns with’ the distinction between, respectively, being remarkable and nonremarkable to speakers. Further elaboration of this alignment would be useful here, especially since exceptions quickly begin to accumulate (e.g. quotative be like, or LIKE as a conjunction or as an approximative adverb). A general issue in the classification system presented is that the reader is left unsure about whether its categories are exhaustive or mutually exclusive, and about the derivational principles on which they are based. For example, the subsection on approximative adverbial like (9–12), introduced as ‘remarked upon yet unremarkable’, includes propositional as well as metalinguistic functions. Approximative adverbial like alternates with functionally equivalent items such as about or approximately when operating at the propositional level, but D'Arcy claims that these alternants do not also share the metalinguistic function of like, that is, signalling a drop in speaker commitment. D'Arcy notes this issue but does not explain why these distinct functions are then lumped together, or why cases of sentence-final adverbial like sharing the same metalinguistic function are classed separately. It seems that the categories generally derive from an informal combination of remarkableness, present-day grammatical function and polysemy, and diachronic and etymological evidence, with historical provenance trumping functional equivalence.
The two forms of LIKE that this book focuses on in subsequent chapters are discourse particle like, which is supposed to serve an intersubjective function, and discourse marker like, which is supposed to contribute to the coherence of a turn-at-talk by signalling that subsequent material is related to what preceded it, for example, by clarifying or exemplifying it. The latter is an evidential function. Furthermore, by taking a more general statement and then supporting it with a more specific one, the marker repositions the speaker relative to the more general statement. This is an epistemic stance function. These evidential and epistemic functions are not addressed under the section on discourse marker like, but they are mentioned in the section on discourse particle like. D'Arcy notes that it is this form that attracts the most metalinguistic hand-wringing, citing its recent functional expansion as a cause of inter-generational variation in use. In addition to its epistemic function (i.e. hedging/mitigating authority/nonequivalence/loose literality), D'Arcy notes too that the discourse marker can also serve a focusing/highlighting function, as well as interpersonal functions (i.e. establishing common ground, building solidarity/intimacy). The internal coherence of these categories is again not obvious; a function-based typology would be more perspicuous.
D'Arcy returns to the issue of functional equivalence in her description of variationist work on LIKE to date (25–29), eschewing pragmatic function as analytically intractable: ‘I make no attempt to ascertain the pragmatic motivations of the speaker but rely instead on structural factors to determine the contexts of use’ (28). However, pragmatic function is invoked throughout the book, for example, in discussions of grammaticalization. It seems incongruous to bracket pragmatic function off for being inscrutable when defining a variable, only to then rely on it when interpreting diachronic change in that variable. It would be helpful to clarify the thinking behind this approach.
Chapter 2 is a description of the corpora used in the book, organized under two main headings—diachronic and synchronic. D'Arcy says that ‘almost all these corpora represent casual vernacular use in some way, be it genuine, scribal record, or constructed’ (36). Some exemplification of the vernacular content of, for example, the legal proceedings, literary fiction, and personal correspondence that constitute a large proportion of the chosen corpora would have been even more reassuring. UK, New Zealand and Australian, and Canadian and US Englishes are represented. A discussion of how and why these corpora were selected (and/or constructed) would have been helpful here, as too would a quantitative summary of their composition—especially their vernacular content—along with demographic and contextual data. Also absent from the book are details of corpus search procedure that could facilitate reproduction and comparison by others—despite reference to ‘using corpus methods to systematically target and extract instances of LIKE’ at the start of the next chapter (47).
Chapter 3 provides further diachronic analysis, noting at the outset that the only recent innovations in LIKE are quotative be like and ‘epistemic parenthetical’ like (presented in Chapter 1 as a function of sentence adverbial like). The chapter focuses on discourse marker like and discourse particle like. Although functional/semantic ambiguity is acknowledged as a feature of early grammaticalization (D'Arcy (49) cites Nevalainen & Rissanen Reference Nevalainen and Rissanen2002 and Hopper & Traugott Reference Hopper and Traugott2003), D'Arcy generally seems to regard ambiguous cases as analytically marginal. If, instead, semantic ambiguity were regarded as a core feature of discourse-pragmatic variables, then an additional analytic step would need to be explicated. In generalizing variationist analysis from sociophonetic variables, where it is assumed that variants are semantically equivalent, to what are effectively sociosemantic variables, where it should be assumed that variants are (or at least can be) semantically different, we need to make it clear exactly how we are loosening the criterion of semantic equivalence. In particular, if one variant is semantically or functionally ambiguous while another is not, then we need to spell out an argument for considering them to be variants of the same variable—taking care to disambiguate between analytic ambiguity and ambiguity as a sociolinguistic resource. For example, is it possible to epistemically hedge (with like) without simultaneously doing interpersonal work (see e.g. Du Bois Reference Du Bois and Englebretson2007)?
In chapter 4, the discussion of internal constraints on marker and particle like (73–80) is lucid and convincing, especially when providing and discussing constructed examples of infelicitous use. Summary tables and a clearly signposted and systematic structure make this chapter easy to read. We learn that marker and particle functions segregate with syntactic position—an insight that could have usefully informed the classification system developed in chapter 2.
Chapter 5 is a very brief survey of external factors conditioning marker and particle like. D'Arcy mentions that LIKE is often ideologically linked to white, suburban, and countercultural identities, before focusing on sex differences. Women make more frequent use of marker like, while men favour particle like. This is something of a puzzle given that both forms are subject to strong negative evaluations, and one would therefore expect men to lead both changes. D'Arcy cites work suggesting that the marker may index cooperativeness and conscientiousness, mitigating the negative evaluation by pointing to characteristics of hegemonic femininities. D'Arcy is unconvinced about the explanatory value here of a putative general trend towards greater discursive ambiguity as suggested by Dinkin (Reference Dinkin2016), saying that there is no systematic analysis so far. This book is possibly not the place to have provided one, but the need for such a piece of research was certainly made apparent. Perhaps I feel like as an epistemic hedge could be a good place to look next.
Chapter 6, an expanded version of D'Arcy (Reference D'Arcy2007), is devoted to evaluating popular ideologies about like. No doubt this will disabuse some readers of erroneous beliefs, but the sociolinguistically informed reader will probably find this chapter less illuminating.
Chapter 7 explores overlaps between the foregoing chapters by considering three research areas. First, children are noted to make more frequent use of particle than marker like, recapitulating the diachronic development of LIKE in their own acquisition. Second, D'Arcy returns to the principle of accountability. Noting that discourse-pragmatic variables can be defined either in terms of pragmatic function or in terms of grammatical or collocational criteria—and notwithstanding the fact that these approaches are not mutually exclusive—D'Arcy reiterates that ‘The solution I propose is to appeal entirely to linguistic structure, and focus on linguistic context rather than form or meaning’ (155). The third research area in which D'Arcy considers how the foregoing chapters interact is methodological, arguing for a complementary application of corpus and variationist methods, whose absolute and relative frequencies respectively sometimes provide conflicting pictures.
As a resource for sociolinguists and historical linguists interested in LIKE, this book is particularly valuable for cataloguing the current literature comprehensively, and for carefully documenting the diachrony of LIKE. Missing, perhaps, is a discussion or chapter drawing together and developing the full range of methodological difficulties and implications of the eight-hundred years of LIKE, including the role of prosody, which is mentioned several times in the book, but also a more explicit and critical approach to expanding the notion of the sociolinguistic variable.
