This book provides an exhaustive accounting of English like in its many functions and, as the book’s organizational motif, its many contexts. Through like, Alexandra D’Arcy offers a useful case study in discourse-pragmatic variation and (though it is not in the title) change, contributing to theories of variation and change, grammaticalization and syntax. The book illustrates the importance of triangulating multiple sources of data, such as diverse corpora, and complementary methodologies, including corpus-based frequencies and variationist-style proportions. Thorough in its documentation and analysis, the book is also full of exciting examples, and should find interest from across linguistic subfields.
Discourse-pragmatic variation in context consists of an introductory chapter, five main chapters, a concluding chapter, and an appendix compiling instances of like. As per the title, the five main chapters each address one of like’s ‘contexts’: empirical, historical, developmental, social and ideological. The book’s organization allows one to find information specific to a particular research question in a single chapter, which makes it highly functional as a reference for future projects. More than once while reading the book, I was struck by the idea that one could organize an entire introductory English linguistics course around like – or, at least, one could usefully incorporate like into each unit. This is a testament to D’Arcy’s ample and multifaceted documentation of related phenomena in morphology, syntax, semantics, pragmatics, sociolinguistics, historical linguistics, language acquisition, and even small bits of phonology and psycholinguistics. The volume compellingly illustrates how a seemingly simple word can open a big window onto the depth of language’s systematicity.
When one is talking about like, though, one is really talking about a collection of likes with different functions, patterns and social evaluations. This multiplicity is the starting point for D’Arcy’s introductory chapter, which is one of the most valuable in the volume. The chapter begins with a brief but detailed tour of the multiple likes, overviewing and exemplifying their syntactic functions and histories in the language. One of the main tasks here is to distinguish the discourse-pragmatic functions of like from its referential and syntactic functions; the rest of the book focuses on the two functions that are discourse-pragmatic: the discourse marker and discourse particle.
As she lays out each like’s function and history, D’Arcy situates them relative to their status in public discourse and academic research, using a framework of ‘un/remarkability’. In the first section on the ‘unremarkable’ likes, D’Arcy overviews like as verb, adjective, noun, preposition, conjunction, complementizer and suffix. The verbal use of like, for instance, is considered ‘unremarkable’ since neither public commenters, grammarians, nor linguists have found anything particularly special about its use. On the other hand, the other functions have been ‘remarked on’ by these constituencies to varying extents: the adverb (both approximative and sentence) and the quotative (with be), in addition to the discourse marker and particle.
This framework of ‘remarkability’ establishes that the book’s aims are not only to document the linguistic facts about like, but also to use those facts to counter popular myths about like. While the myths are discussed in greater detail in later chapters (specifically chapter 6), they are invoked throughout the book as the background against which to present linguistic data. D’Arcy’s ‘remarkability’ framework suggests that words that have indexical, interpersonal functions are more readily villainized than words with lexical meaning. This observation echoes those made about (Western) language ideology more generally by, for instance, Silverstein (Reference Silverstein1979, Reference Silverstein1981). Namely, speakers believe that referring is language’s primary (or even only) function, and speakers have more difficulty accurately describing non-referential features (see also Preston Reference Preston1996; note that linguists have also not been immune to this ideology). With like, we see that the two discourse-pragmatic functions are both heavily ‘remarked upon’ and are central to the language ideologies around it – they are clearly available to folk analysis – yet those espousing critiques of like do not recognize these functions as being distinct from any others, and are typically not accurate in their descriptions.
What are the linguistic properties of the discourse-pragmatic likes? According to D’Arcy, the discourse marker ‘encodes textual relations by relating the current utterance to prior discourse’ (p. 14); syntactically, it occurs clause-initially. The discourse particle ‘signals subjective information’ (p. 15), including epistemic stance and focus; syntactically, it occurs in multiple clause-internal positions. Examples of the marker (1) and particle (2) are given below.
(1) They never went out in a small canoe. Like, we went from here to Cape Beale. They had great large war canoes. [DCVE/87f/1875] (p. 14)
(2) His father had like a restaurant cafe in Regent Street. [DAR/51m/1955] (p. 15)
Chapter 1’s final task is to introduce D’Arcy’s methodology. This section includes a brief discussion of central issues in variationist research on discourse-pragmatic features (e.g. Lavandera Reference Lavandera1978). To obviate concerns over the ‘subjective’ domain of pragmatic judgments, D’Arcy uses structural criteria – syntactic criteria – to delimit the occurrences and potential occurrences of like to form an accountable analysis (Labov Reference Labov1972). This method assumes that discourse-pragmatic features ‘are systematically constrained by a probabilistic choice mechanism’ (p. 25) as are other elements of the grammar. The following four chapters of the book establish historical and social patterns in the marker and particle, outlining the nature of those probabilistic constraints and tracing the features’ intergenerational transmission.
The first ‘context’ is empirical: chapter 2 is a short summary of the extensive source material from which D’Arcy draws her data, offering a valuable listing of available data for projects on grammatical variation and change in English. Some of these are established, widely used corpora (e.g. Penn Parsed Corpora of Historical English; International Corpus of English), others less so (e.g. Corpus of Southwest Tyrone English). There are 20 overviewed corpora, 10 diachronic and 10 synchronic; they comprise mostly vernacular language and reach internationally from North America to Great Britain and the Pacific. D’Arcy notes in the book that like’s developmental trajectory is similar across Inner Circle varieties, but its status in the wider English-speaking world is only hinted at. These corpora also provide the sourcing for the book’s impressive appendix, which contains 364 examples of the discourse marker, 416 of the discourse particle, and 30 of the sentence adverb.
Three data and analysis chapters follow. Each is central to combating the myth of like that D’Arcy’s book positions itself against: like is not new, as the history shows (chapter 3); it is not random, as its syntactic distribution shows (chapter 4); and its use is not limited to young or female speakers, as the sociolinguistic data show (chapter 5).
In the historical investigation in chapter 3, D’Arcy traces the marker and particle to the sentence adverb usage of like (contra arguments that trace them to the conjunction, e.g. Romaine & Lange Reference Romaine and Lange1991). D’Arcy finds the discourse marker already in use in the mid-1800s; importantly, the fact that it occurred in multiple Inner Circle varieties suggests that the marker is likely even older than what is documented. She argues that the marker and particle reflect the natural course of transmission via generations. The particle’s earliest documentation is newer, as it unambiguously occurs only from the early 1900s. In its development from adverb to marker to particle, like shows key features of grammaticalization, such as decategorialization and semantic bleaching (p. 65).
This analysis is developed further in chapter 4, which investigates like’s syntactic trajectory (‘developmental context’). The discourse-pragmatic uses spread over time as like went from appearing clause-initially (as a discourse marker) to appearing in five separate positions in the clause (as a discourse particle). The data come from the Toronto English Archive, a corpus of speakers from 11 to 87 years old – allowing the investigation of change in apparent time – and approximately balanced across men and women. The analysis combines variationist methodology with basic Minimalist syntactic architecture in order to delineate like’s contexts of occurrence. D’Arcy treats the marker and particle as syntactic adjuncts (‘optional’ elements) that adjoin to phrase- and clause-level projections. She looks at seven projections as possible ‘adjunction sites’: CP and TP (clausal domain); DP and nP (nominal domain); DegP and AP (adjectival domain); and vP (verbal domain). These are the contexts in which variation in the use of like is found; an example of each is below in (3)–(8). Note that the marker occurs in the clausal domain (3)–(4), while the particle occurs in the clause-internal domains (5)–(8).
(3) (a) CP-matrix: Like my first experience with death was this Italian family. [TEA/82f/1921] (p. 83)
(b) CP-subordinate: So I get it all done [like when I get home]. [TEA/17f/1986] (p. 87)
(4) TP: I think [that like there’s been a desire instilled in me]. [TEA/21f/1982] (p. 87)
(5) (a) DP: I haven’t seen [like a huge difference]. [TEA/45m/1958] (p. 94)
(b) nP: They have [this like energy], you know? [TEA/21f/1982] (p. 98)
(6) (a) DegP-without adverb: They remained like aloof. [TEA/52m/1951] (p. 100)
(b) DegP-with adverb: Everything is like so complicated. [TEA/50m/1953] (p. 100)
(7) AP: I get really like flabbergasted. [TEA/24f/1979] (p. 100)
(8) vP: I was like playing in bands like all the way through high school. [TEA/22m/1981] (p. 102)
D’Arcy extracts each instance in the corpus of these projections, excluding syntactic contexts in which the marker and particle are known to be very rare or nonexistent. This procedure produced just under 20,000 total tokens to analyze for presence/absence of like.
D’Arcy finds that the discourse marker like occurs in 14.2 percent of both matrix and subordinate CPs; however, the development of the two is not identical. For matrix CPs, the proportion is similar to the proportion of all other discourse markers combined – and only among younger speakers is like the clear discourse-marking preference. At the same time, the oldest speakers in the sample do not evidence like in subordinate CPs, which clearly shows a generalization of the form’s use from matrix to subordinate CPs over time. The TP context shows an even later development. This kind of continued generalization, D’Arcy argues, is in line with the development of other discourse markers, and shows variation data to be critical to the analysis of syntactic change.
The discourse particle’s clause-internal distribution shows some fascinating evidence for just how nuanced the patterning of like is. For example, within the nominal domain, D’Arcy shows that LIKE is more frequent with DPs that are arguments versus complements, with indefinite versus definite articles, and with quantified versus unquantified nouns (p. 97). Yet each of these uses rise in frequency across apparent time, being more frequent for younger speakers. That is, speakers share the same internal constraints on like – its variable grammar is not changing in kind, only frequency of application. The same consistency of constraints is also in place within the verbal domain: use of particle like in the vP increases over time, but across the generations, the frequency of particle like is sensitive to verb type (unergative > transitive > unaccusative) and subject thematicity (agentive > non-agentive). For the relative lack of attention ‘optional’ elements like discourse markers have received in theory, D’Arcy makes a strong case for their inclusion in any comprehensive accounting of syntax.
In the next two chapters, D’Arcy confronts head-on the popular myths around like. Chapter 5 focuses on like’s use according to gender of speaker and, incidentally, age (as shown in the prior chapter, the marker and particle are more frequent among younger speakers, though their internal constraints are the same across generations). The primary, and surprising, finding is that the marker and particle have different gender associations: the marker is favored by women, while the particle is favored by men. The latter runs counter to popular ideologies holding that like is used primarily by females; it also runs counter to the ‘classic’ sociolinguistic finding showing that women lead linguistic changes, and that the changes led by men tend to be ‘isolated’ ones rather than systemic (e.g. Labov Reference Labov1990). Importantly, the gender differences for both features are larger among the younger generations.
For the syntactic distributions and gender findings, one naturally wonders about pragmatic motivations potentially underlying them. Discourse analysis – the ‘conversational context’ as it were – is not the terrain of this book, but D’Arcy offers some ideas on the gender differences. She suggests that differing orientation to conversation may motivate differing use of the marker and particle; the discourse marker signals conversational cooperation, which is related to politeness, and politeness moves are typically feminized. However, elsewhere, D’Arcy suggests that the particle too aids ‘cooperative aspects of communication’ (p. 129), so it is unclear that orientation to personal dynamics such as politeness is enough to explain these patterns. There is also other work remaining to be done on the social distribution of like, including investigating racial/ethnic differences, gender-related categories that may not be captured through the binary of ‘men’ and ‘women’, and class- or status-based stratification. While D’Arcy has covered a massive amount of ground here, it is only lucky for sociolinguists that there is still more to do.
In chapter 6, D’Arcy presents six aspects of the ‘myth’ of like, and summarizes the evidence to the contrary of each (for instance, that it is only used by young females; that it is meaningless; that it is random). The chapter would make a fine reading on its own as a snapshot of public discourse about like from the last half-century or so, and an example of how linguistic science can attempt to correct unfounded, and often discriminatory, beliefs about language (for recent discussion of ‘error correction’, see Lewis Reference Lewis2018 and responses, e.g. Rickford Reference Rickford2018). The ‘myth’ of like, just like like itself, is actually a multipronged thing: its precise values differ across language varieties/cultures. The chapter also reveals starkly the incongruity between linguistic facts and ideologies about like, echoing work on the enregisterment of linguistic features and misalignments between practice and ideology (e.g. Johnstone & Kiesling Reference Johnstone and Kiesling2008).
The final chapter (‘Contextual interfaces’) addresses some remaining questions about like, including its acquisition, and then turns to a sort of meta-analysis. D’Arcy summarizes the timeline of like’s development, then she discusses the implications of her methodological choices. This includes a very useful section comparing the approaches of corpus linguistics (whose data are summarized as frequencies) and variationist sociolinguistics (whose data are summarized as proportions). The two approaches test for different things – with corpus methods being especially useful for questions of diachrony, and variationist methods more so for questions of synchrony. Both are needed, D’Arcy suggests, to fully understand a feature’s incrementation and diffusion across the speech community. This section will prove helpful especially for early-career researchers, or anyone just getting into discourse-pragmatic work.
This book sets out to refute the widespread beliefs that like is random, meaningless, a marker of inarticulateness, and the purview of Valley Girls. D’Arcy shows thoroughly that like is none of these, through an impressive amount of data, clear analysis and compelling argumentation. Of course, to a linguist, it is not surprising that like is systematic. It is nonetheless thrilling to see all the ways in which the systematicity emerges, through time in the language and throughout its grammar. D’Arcy’s book joins a handful of other recent volumes dedicated to discourse-pragmatic variation and change (e.g. Pichler Reference Pichler2016; Aijmer & Lewis Reference Aijmer and Lewis2017). This research shows the importance of accounting for linguistic features often considered ‘little’, ‘meaningless’, ‘random’, or otherwise set aside by matters of theory. Such features are structured, meaningful (even if not referentially so) and deeply embedded in the grammatical layers of the linguistic system. As such, their use and development can shed light on questions interfacing with every other linguistic field. This book will stand as an exemplar of both how to do this kind of work, and why it should be done.
