This book is about doing variation analysis. My goal is to give you a manual which will take you through a variationist analysis from beginning to end. Although I will cover the major issues, I will not attempt a full treatment of the theoretical issues nor of the statistical underpinnings. Instead, you will be directed to references where the relevant points are treated fully and in detail. In later chapters, explicit discussion will be made as to how different types of analysis either challenge, contribute to, or advance theoretical issues. This is important for demonstrating (and encouraging) evolution in the field and for capturing a sense of its ongoing development. Such a synthetic perspective is also critical for evolving our research in the most fruitful direction(s). This book is meant to be a learning resource which can stimulate methodological progression, curriculum development as well as advancements in teaching and transmission of knowledge in variation analysis. With any luck new discoveries will be made.
What Is Variation Analysis?
Variation analysis goes by different names, sometimes it is called ‘Labovian sociolinguistics’ after its founder William Labov; another term is ‘variationist sociolinguistics’, yet another is ‘language variation and change’, as in the name of this subdiscipline’s major journal. In this book, I will use these terms somewhat interchangeably; however, the emerging term encompassing worldwide developments is variation linguistics and variation analysis.
Variation analysis combines techniques from linguistics, anthropology, and statistics to investigate language use and structure (Poplack, Reference Poplack and Preston1993:251). For example, a seven-year-old boy answers a teacher’s question by saying, ‘I don’t know nothing about that.’ A middle-aged woman asks another, ‘You got a big family?’ An octogenarian might say, ‘I did see it.’ Are these utterances instances of dialect, slang, or simply performance errors, mistakes? Where on the planet were they spoken, why, by people of what background and character, in which sociocultural setting, under what conditions? How might such utterances be contextualised in the history of the language and with respect to its use in society? This book provides an explicit account of a method that can answer these questions, a step-by-step ‘user’s guide’ for the investigation of language use and structure as it is manifested in situ.
At the outset, however, I would like to put variationist sociolinguistics in perspective. First, how does the variationist tradition fit in with the field of sociolinguistics as a whole? What is its relationship to linguistics?
Linguistics
The enterprise of linguistics is to determine the properties of natural language. Here, the aim is to examine individual languages with the intention of explaining why the whole set of languages are the way they are. This is the search for a unified theory of grammar which can specify the permissible rules of one language, say English or Japanese, but which is also relevant for the grammar of any natural language. In this way, linguistics puts its focus on determining what the component parts and inner mechanism of languages are. The goal is to work out ‘the rules of language X’ – whether that language is English, Welsh, Igbo, Inuktitut, Niuean, or any other human language on the planet.
The type of question a linguist might ask is, ‘How do you say X?’ For example, if a linguist was studying Welsh, they would try to find a fluent speaker of the language, and then they would ask that person, ‘How do you say “dog” in Welsh?’ ‘How do you say “The child calls the dog”, “The dog plays with the children”?’ and so on. This type of research has been highly successful in discovering, explaining, and accounting for the complex and subtle aspects of linguistic structure. However, in accomplishing this, modern theoretical models of language have had to set aside certain aspects, consigning them to the lexical, semantic, or pragmatic components of languages, or even outside of language altogether in the socio-stylistic components of its use. For example, in a syntactic account of grammatical change, Roberts and Rousseau (Reference Roberts and Roussou2003:11) state:
Of course, many social, historical, and cultural factors influence speech communities, and hence the transmission of changes (see Labov Reference Labov and Labov1972c, Reference Labov1994). From the perspective of linguistic theory, though, we abstract away from these factors and attempt, as far [sic] the historical record permits, to focus on change purely as a relation between grammatical systems.
Linguistic theory focuses on the structure of the language. It does not concern itself with the context in which the language is learned, and more importantly, it is not interested in the way the language is used. However, see an early review of attempts at rapprochement in phonological theory in Coetzee and Pater (Reference Coetzee, Pater, Goldsmith, Riggle and Yu2011). Only in the late 1990s and into the 2000s have researchers begun to make the link between variation theory and syntactic theory (e.g. Beals et al., Reference Beals, Denton, Knippen, Melnar, Suzuki and Zeinfeld1994; Meechan & Foley, Reference Meechan and Foley1994; Cornips & Corrigan, Reference Cornips, Corrigan, Auer, Hinskens and Kerswill2002; Adger & Smith, Reference Adger, Smith, Cornips and Corrigan2005).
Sociolinguistics
Sociolinguistics argues that language exists in context, influenced by the individual who is using it, and dependent on where it is being used and why. Individuals mark their personal history and identity in their speech as well as their sociocultural, economic, and geographical coordinates in time and space. Indeed, some researchers would argue that, since language is obviously social, to study it without reference to society would be like studying courtship behaviour without relating the behaviour of one partner to that of the other. As Joseph (Reference Joseph, Boas and Pierce2020) argued, the ‘evaluation problem’ of Weinreich et al. (Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968:183–187) entails that ‘Someone has to do the evaluating, and someone has to produce a word or an utterance or a piece thereof that can be evaluated, and this means that change is not just something in an isolated individual but involves at least two people. It is inherently social in nature, as a result, and requires contact between speakers.’
Two important arguments support the integral social role of language. First, you cannot take the notion of language X for granted since this is a social notion insofar as it is defined in terms of a group of people who speak X. Therefore, if you want to describe the English language you must define it based on the group of people who speak it. Second, speech has a social function, both as a means of communication and as a way of identifying social groups.
Standard definitions of sociolinguistics read something like this: the study of language in its social contexts and the study of social life through linguistics (Coupland & Jaworski, Reference Coupland, Jaworski, Coupland and Jaworski1997:1); the relationship between language and society (Trudgill, Reference Trudgill2000:21); the correlation of dependent linguistic variables with independent social variables (Chambers, Reference Chambers2003: ix). However, the many ways that society can impinge on language make the field of reference extremely broad. Studies of the various ways in which social structure and linguistic structure come together include personal, stylistic, social, sociocultural, and sociological aspects. Depending on the purposes of the research, the different orientations of sociolinguistic research in the 1960s and 1970s was subsumed by one of two umbrella terms: ‘sociolinguistics’ and ‘the sociology of language’ (Fasold, Reference Fasold1984; Reference Fasold1990). A further division could also be made between qualitative (ethnography of communication, discourse analysis, and so on) and quantitative (language variation and change) approaches. Sociolinguistics tends to put emphasis on language in social context, whereas sociology of language puts emphasis on society and the social interpretation of language. Variation analysis is embedded in sociolinguistics, the area of linguistics which takes as a starting point the rules of grammar and then studies their links with society. But then the question becomes, how and to what extent? Methods of analyses, and focus on linguistics or sociology, are what differentiate the different subdisciplines of sociolinguistics. From this perspective, variation analysis is inherently linguistic, analytic, and quantitative.
Variationist Sociolinguistics
Variationist sociolinguistics has evolved since the 1960s as a discipline that integrates social and linguistic aspects of language. Perhaps the foremost motivation for the development of this approach was to present a model of language which could accommodate the paradoxes of language change. Formal theories of language were attempting to determine the structure of language as a fixed set of rules or principles, but because language changes perpetually, the structure must be fluid. How does this happen? The idea that language is structurally sound is difficult to reconcile with the fact that languages change over time. Structural theories of language, so fruitful in synchronic investigation, have saddled historical linguistics with a cluster of paradoxes which have not been fully overcome (Weinreich et al., Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968:98).
Change in language does not happen in a linguistic vacuum. Because it is used by human beings in a social world, there is also a need to consider the social world. This interface between language as a system and language as a social phenomenon makes sociolinguistics an unusually expansive field of research, with researchers having a myriad of unique foci. Sociolinguistics often comes across as either too restricting to social categories, using such categorisations as class, gender, style, geography (external, social, factors), or too restricting to linguistic categories, using concepts such as systems or complexity (internal, linguistic, factors). When variationist methods have focused on the linguistic system, as opposed to the social aspects of the individual and context, it has garnered considerable critique (e.g. Cameron, Reference Cameron, Joseph and Taylor1990; Rickford, Reference Rickford1999; Eckert, Reference Eckert2000), restating the bipartite underpinnings of the field (Milroy & Gordon, Reference Milroy and Gordon2003:8). When attempting to synthesise both internal and external aspects of language, the challenge will always be how to explore both without compromising one or the other. While this will likely always be tempered by researchers’ own predilections, it is also the case that the research questions, data, and findings may naturally lead to a focus on one domain over the other. Having said all this, the variationist enterprise is essentially the complex study of the interplay between variation, social meaning, and the evolution and development of the linguistic system itself.
Indeed, as Weinreich et al. (Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968:188) stated:
Explanations of language which are confined to one or the other aspect – linguistic or social – no matter how well constructed, will fail to account for the rich body of regularities that can be observed in empirical studies of language behaviour …
This ‘duality of focus’ has been fondly described by Guy (Reference Guy and Preston1993:223) as follows:
One of the attractions – and one of the challenges – of dialect research is the Janus-like point-of-view it takes on the problems of human language, looking one way at the organisation of linguistic forms, while simultaneously gazing the other way at their social significance.
In my view, variationist sociolinguistics is most aptly described as the branch of linguistics which studies the foremost characteristics of language in balance with each other – linguistic structure and social structure; grammatical meaning and social meaning – those properties of language which require reference to both external (social) and internal (systemic) factors in their explanation.
Therefore, instead of asking the question ‘How do you say X?’, as a linguist might, a sociolinguist is more likely not to ask a question at all. The sociolinguist will just let you talk about whatever you want to talk about and listen for all the ways you say X.
There is a distinct ‘occupational hazard’ to being a sociolinguist. You will be in the middle of a conversation with someone, and you will notice something interesting about the way they are saying it. You will make note of the form. You will wonder about the context. You may notice a pattern. Suddenly, you will hear that person saying to you, ‘Are you listening to me?’ and you will have to say, ‘I was listening so intently to how you were saying it that I didn’t hear what you said!’
The essence of variationist sociolinguistics rests on three facts about language that are often ignored in the field of linguistics. First, the notion of ‘orderly heterogeneity’ (Weinreich et al., Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968:100), or what Labov (Reference Labov, Lehmann and Malkiel1982:17) refers to as ‘normal’ heterogeneity; second, the fact that language is always changing; and third, that language conveys more than simply the meaning of its words. It also communicates abundant non-linguistic information. Let us consider each of these in turn.
Orderly Heterogeneity
Heterogeneity is the observation that there is variability in language. Individuals have ‘more than one way to say more or less the same thing’, that is, accomplish the same function. Variation can be viewed across whole languages, in the choice of one language or the other by bilingual or multilingual individuals, for example French, Tamil, Inuktitut. However, linguistic variation also be observed across an entire continuum of choice types ranging between different word orders, morphological affixes, constructions, right down to the minute microlinguistic level where there are subtle differences in the pronunciation of individual vowels, consonants, intonation contours, and tone. Importantly, this is the normal state of affairs: ‘The key to a rational conception of language change – indeed, of language itself – is the possibility of describing orderly differentiation in a language serving a community … It is absence of structural heterogeneity that would be dysfunctional’ (Weinreich et al., Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968:100–101). Furthermore, heterogeneity is crucially not random, but patterned. It reflects order and structure within the grammar. Variation analysis aims to characterise this complex system.
Language Change
Language is always in flux. The English language today is not the same as it was 100 years ago, or 400 years ago. For example, ain’t used to be the normal way of doing negation in English, but now it is stigmatised. Another good example is not. It used to follow the verb (e.g. I know not). Now it precedes the verb, along with a supporting word, do (e.g. I do not know). Double negation (e.g. I don’t know nothing) is ill-regarded in contemporary English. Not so in earlier times. Similarly, use of the ending -th for simple present was once the favoured form (e.g. doth, not do), and pre-verbal periphrastic do (e.g. I do know) and use of the comparative ending -er (e.g. honester, not more honest) used to be much more frequent; see studies of historical corpora such as the Corpus of Early English Correspondence (Nevalainen & Raumolin-Brunberg, Reference Nevalainen and Raumolin-Brunberg2003).
Variation analysis aims to put linguistic features such as these in the context of where each one has come from and where it is going – with a focus on how and why.
Social Identity
Language serves a critical purpose for its users that is just as important as the obvious one. Language is used for transmitting information from one person to another, but at the same time it is used by individuals to make statements about who they are, what their group loyalties are, how they perceive their relationship to their hearers, and what sort of speech event they consider themselves to be engaged in. The only way all these things can be carried out at the same time is precisely because language varies. The choices individuals make among alternative linguistic means to communicate the same information often conveys important extralinguistic information. While you can sometimes identify a person’s gender from a fragment of their speech, it is often nearly as easy to identify their age and even their socioeconomic class, but these judgements can be misleading. Further, depending on one’s familiarity with the variety, it can be relatively straightforward to identify nationality, locality, community, etc. For example, are the following excerpts from the late twentieth century from a young person or an old person?
I don’t know, it’s jus’ stuff that really annoys me. And I jus’ like stare at him and jus’ go … like, ‘huh’.
How about the following? Man or woman? Old or young?
It was sort of just grass steps down and where I dare say it had been flower beds and goodness knows what.
It was just a fun experience in general, like, the experience of like, you know, debating random problems and stuff.
To a certain point sweeping decisions on social judgements may be accurate. The first is a young woman, aged 30 in 2018 (YRK 2018, spickering, woman, 30). The second is a woman, aged 79 in 1997 (York, UK, rbaker, woman, 79). The third is a man, aged 19 in 2021 (TOR 2021, rbarman, man, 19).
Key Characteristics of Variationist Sociolinguistics
Given these three aspects of language – inherent variation, constant change, and pervasive social meaning – variationist sociolinguistics rests its method and analysis on a number of key concepts.
The Vernacular
A specific goal of variationist methodology is to gain access to what is referred to as the ‘vernacular’. The vernacular has had many definitions in the field. It was first defined in sociolinguistics as ‘the style in which the minimum attention is given to the monitoring of speech’ (Labov, Reference Labov1972d:208). Later characterisations of the vernacular reaffirmed that the ideal target of investigation for variation analysis is ‘everyday speech’ (Sankoff, G., Reference Sankoff, Bauman and Sherzer1974:54), ‘real language in use’ (Milroy, Reference Milroy1992:66), and ‘spontaneous speech reserved for intimate or casual situations’ (Poplack, Reference Poplack and Preston1993:252) – what can be described as informal speech.
Access to the vernacular is critical because it is thought to be the most systematic form of speech. Why? First, because it is assumed to be the variety that was acquired first. Second, because it is the variety of speech most free from hypercorrection or style-shifting, both of which are considered to be later overlays on the original linguistic system. Third, the vernacular is the style from which every other style must be calibrated (Labov, Reference Labov, Baugh and Sherzer1984:29).
The position of the vernacular is pivotal, positioned maximally distant from the idealised norm (Milroy, Reference Milroy1992:66; Poplack, Reference Poplack and Preston1993:252; Poplack et al., Reference Poplack, Jarmasz, Dion and Rosen2015). Once the vernacular baseline is established, the multi-dimensional nature of speech behaviour can be revealed. Bell (Reference Bell1999:526) argues that performance styles are defined by normative use, making unmonitored speech the focus for taping the dimensions of the speech community. Moreover, as Labov (Reference Labov1972d:208) argued, the vernacular provides the ‘fundamental relations which determine the course of linguistic evolution’. The vernacular is the foundation from which every other speech behaviour can be understood, and in which change in progress must be situated.
Many of my students report that their roommates switch into their vernacular when talking to their family on the phone. You will also notice it shining through whenever a person is emotionally involved (e.g. excited, scared, angry, moderately drunk). Listen out for it!
The Speech Community
To ‘tap the vernacular’ (Sankoff, D., Reference 347Sankoff and Newmeyer1988b:157), a vital component of variation analysis, analysts are required to immerse themselves in the speech community, entering it both as an observer and as a participant. In this way, analysts may record language use in its sociocultural setting (e.g. Labov et al., Reference Labov, Cohen, Robins and Lewis1968; Trudgill, Reference Trudgill1974; Milroy, Reference Milroy1987; Poplack, Reference Poplack and Preston1993:252). This methodology’s focus on unmonitored speech behaviour has allowed it to overcome many of the analytical difficulties associated with intuitive judgements and anecdotal reporting use in other paradigms (Poplack, Reference Poplack1980; Sankoff, D., Reference 347Sankoff and Newmeyer1988b). This is crucial in the study of non-standard varieties, as well as ethnic, rural, informal, and other less highly regarded forms of language, where normative pressure inhibits the use of vernacular forms.
For example, when you hear people use utterances such as: ‘I ain’t gotta tell you anything’, certain social judgements may arise. Whatever judgements come to mind are based on hypotheses that arise from interpreting the various linguistic features within these utterances. What are those features? Most people, when asked why someone sounds different, will appeal to their ‘accent’, their ‘tone of voice’, or their ‘way of emphasising words’. However, innumerable linguistic features of language provoke social judgements.
One way to explore this is to contemplate the various ways the utterance cited above could have been said, (1).
(1)
a. I ain’t gotta tell you nothing/anything. b. I haven’t got to tell you nothing/anything. c. I don’t have to tell you nothing/anything. d. I don’t need to tell you nothing/anything.
Each possible utterance has its own social value, ranging from the highly vernacular to standard. Notice, too, how each feature of language varies in its own particular ways. Ain’t appears to vary with haven’t and possibly don’t. Gotta appears to vary with have to as well as got to and need to. Nothing varies with anything. In this way, each item alternates with a specific set – different ways of saying the same thing.
The linguistic items which vary amongst themselves with the same referential meaning constitute the set of linguistic items, the linguistic variable, which are the substance of variation analysis. But the next question becomes, How do you determine what truly varies with what?
Form–Function Asymmetry
The identification of ‘variables’ in language use rests on a fundamental view in variation analysis – the possibility of multiple forms to achieve the same function. Do all the sentences in (1) mean the same thing? Some linguists might assume that different forms can never have identical meaning. In variation analysis, however, it is argued that different options such as these can indeed be used interchangeably for the same function, particularly in the case of ongoing linguistic change. There is a basic recognition of instability in linguistic form–function relationships (Poplack, Reference Poplack and Preston1993:252; Reference Poplack, Shin and Erker2018) and, further, that differences amongst competing forms may be neutralised in discourse (Sankoff, D., Reference 347Sankoff and Newmeyer1988b:153). Where functional differences are neutralised is always an empirical question. It must first be established what varies with what and how. Notice that you can’t say I ain’t haven’t to tell you nothing. Why? The goal of variation analysis is to pinpoint the form–function overlap and explain how this overlap exists and why.
Linguistic Variables
Different ways of saying more or less the same thing may occur at every level of grammar in a language, in every variety of a language, in every style, dialect, and register, in every individual, and often even in the same interaction, discourse, and sentence. In fact, variation is everywhere, all the time. Consider the examples in (2) to (10), all of which are taken from the York English Corpus (YRK),Footnote 1 which documents the variety spoken in the city of York in the north of England (Tagliamonte, Reference Tagliamonte1998).
(2) Phonology/morphology, variable (t,d):Footnote 2
I did a college course when I lefØ school actually, but I left it because it was business studies. (YRK, kdilks, woman, 26)
(3) Phonology/morphology, variable (-ing):
We were having a good time out in what we were doin’. (YRK, nheath, woman, 20)
(4) Morphology, variable (-ly):
You go to Leeds and Castleford, they take it so much more seriously … They really are, they take it so seriousØ. (YRK, sdonaldson, woman, 41)
(5) Tense/aspect, variable future temporal reference forms:
… I think she’s gonna be pretty cheeky. I think she’ll be cheeky. (YRK, kyoung, woman, 31)
(6) Modal auxiliary system, deontic modality:
‘I’ve got to cycle all the way back and then this afternoon I’ll be cycling back up again!’ … You have to keep those thoughts err thoughts to yourself. (YRK, rslater, man, 59)
(7) Intensifiers:
I gave him a right dirty look … and I gave him a really dirty look. (YRK, kyoung, woman, 31)
(8) Syntax/semantics, variable stative possessive meaning:
He’s got bad-breath; he has smelly feet. (YRK, cbiggs, woman, 33)
(9) Syntax, agreement:
She were a good worker. She was a hell of a good worker. (YRK, rfielding, 70 in 1986)
(10)
Discourse/pragmatics, quotative use: a. She was like, ‘What are they saying?’ I was like, ‘We need to leave now.’ And she goes, ‘Why?’ (YRK 2018, vevans, woman, age 20) b. I thought, ‘That’s – that’s my dog.’ (YRK, woman, tlaxton, age 48) c. And we said, ‘No.’ And then Ned said, ‘Would you like me to go to a cash point?’ (YRK 2013, jjubb, man, age 19)
How can such alternation become interpretable? It is necessary to refer to more than just social meaning. Such variation might be explained by external pragmatic factors; however, more often this variation is the reflex of social, linguistic, and historical implications. In the case of variable (-ly), adverb morphology, have got, stative possessive meaning, the modal auxiliary system, intensifying adverbs, and others, variation amongst forms can be traced back to longitudinal change in the history of the English language. In the case of adverb placement and variable agreement, synchronic patterns may address issues pertaining to the configuration of phrase structure, feature checking, and other matters of theoretical importance. Indeed, much of the work on historical syntax has highlighted the complexity of how linguistic structures evolve in the process of grammatical change (e.g. Kroch, Reference Kroch1989; Warner, Reference Warner1993; Taylor, Reference Taylor1994; Pintzuk, Reference Pintzuk1995).
The Quantitative Method
Perhaps the most important aspect of variation analysis that sets it apart from most other areas of linguistics, and even sociolinguistics, is its quantitative approach (Labov, Reference Labov2008). The combination of techniques employed in variation analysis forms part of the ‘descriptive-interpretative’ strand of modern linguistic research (Sankoff, D., Reference 347Sankoff and Newmeyer1988b:142–143). Studies employing this methodology are based on the observation that individuals make choices when they use language and that these choices are discrete alternatives with the same referential value or grammatical function. Furthermore, these choices vary in a systematic way and as such they can be quantitatively modelled (Labov, Reference Labov1969a; Cedergren & Sankoff, D., Reference Cedergren and Sankoff1974); see also restatements in Young and Bayley (Reference Young, Bayley, Bayley and Preston1996:254), Poplack and Tagliamonte (Reference Poplack and Tagliamonte2001:88), which are affirmed in later textbooks (e.g. Van Herk, Reference Van Herk2012; Meyerhoff, Reference Meyerhoff2013). This is perhaps most candidly put by Sankoff, D. (Reference 347Sankoff and Newmeyer1988b:151):
whenever a choice can be perceived as having been made in the course of linguistic performance, and where this choice may have been influenced by factors such as the nature of the grammatical context, discursive function of the utterance, topic, style, interactional context or personal or sociodemographic characteristics of the individual or other participants, then it is difficult to avoid invoking notions and methods of statistical inference, if only as a heuristic tool to attempt to grasp the interaction of the various components in a complex situation.
The advantage of the quantitative approach lies in its ability to model the simultaneous, multi-dimensional factors impacting on individual choices, to identify even subtle grammatical tendencies and regularities in the data, and to assess their relative strength and significance. These measures provide the basis for comparative linguistic research. However, such sophisticated techniques are only as good as the analytic procedures upon which they are based: ‘The ultimate goal of any quantitative study … is not to produce numbers (i.e. summary statistics), but to identify and explain linguistic phenomena’ (Guy, Reference Guy and Preston1993:235).
The Principal of Accountability
According to Labov (Reference Labov1972d:72), ‘the most important step in sociolinguistic investigation is the correct analysis of the linguistic variable’. ‘Correct’ in this case means ‘accountable’ to the data. In variation analysis, accountability is defined by the ‘principle of accountability’, which holds that every variant that is part of the variable context, whether the variants are realised or unrealised elements in the system, must be taken into account. You cannot simply study the variant forms that are new, interesting, unusual, or non-standard – ain’t, for example, or got. You must also study the forms with which such features vary in all the contexts in which either of them would have been possible. In the case of ain’t, this would mean all the cases where ain’t is used as well as all other negation variants with the same referential value as ain’t – for example, I haven’t got nothing or perhaps even I don’t got nothing – whatever occurs in the same context.
Sometimes students struggle to understand what an unrealised variant might be. ‘Unrealised’ means that that structure or form is not discernible on the surface, not spoken; not written, but its function and meaning are present. There are many unrealised variants in language. Here is a good example (Tagliamonte, Reference Tagliamonte2012b:5). Can you spot the realised and unrealised variants?
i. To prove I could do it, I had to prove that I could do it.
Hint: It is a ubiquitous structural feature that signals clause structure. This example illustrates what I call a ‘super token (Tagliamonte, Reference Tagliamonte2012b:111), ‘two variants by the same individual in the same stretch of discourse’. Look for them!
By definition, an accountable analysis demands of the analyst an exhaustive report for every case in which a variable element occurs out of the total number of environments where the variable element could have occurred. In Labov’s (Reference Labov1972d:72) words, ‘report values for every case where the variable element occurs in the relevant environments as we have defined them’. ‘As we have defined them’ is the important point here. What does this mean?
Circumscribing the Variable Context
How does the analyst determine the variants of a variable and the contexts in which they vary? This procedure is most accurately characterised as a ‘long series of exploratory maneuvers’ (Labov, Reference Labov1969a:728–279):
1. Identify the total population of utterances in which feature varies. Exclude contexts where there is only one variant.
2. Decide on how many variants can be reliably identified. Set aside contexts that are indeterminate, neutralised, and so on.
These manoeuvres accentuate that variation analysis is not interested in individual occurrences of linguistic features but requires systematic study of the recurrent choices an individual makes (Poplack & Tagliamonte, Reference Poplack and Tagliamonte2001:89). Analysis of these recurrent choices enables the analyst to ‘tap in’ to an individual’s use of the targeted forms. In this case, a ‘pattern’ refers to ‘a series of parallel occurrences (established according to structural and/or functional criteria) occurring at a non-negligible rate in a corpus of language use’ (Poplack & Meechan, Reference Poplack and Meechan1998:129). What does non-negligible mean? Guy (Reference Guy, Ferrara, Brown, Walters and Baugh1988) suggests between 5 per cent and 95 per cent; however, the lower and upper boundaries may be unrealistic to model quantitatively. The decision to study a linguistic variable with very low or very high rates of occurrence should be motivated, sometimes because it is the last opportunity to study a moribund phenomenon (e.g. Jones & Tagliamonte, Reference Jones and Tagliamonte2004; Rupp & Tagliamonte, Reference Rupp and Tagliamonte2017:93–94; Needle & Tagliamonte, 2022) or an opportunity to describe an incipient form (e.g. Tagliamonte, Reference Tagliamonte2021); sometimes because it reveals reanalysis of frequently occurring collocations into discourse pragmatic devices (Thompson & Mulac, Reference Thompson and Mulac1991a; Reference Thompson, Mulac, Traugott and Heineb; Franco & Tagliamonte, Reference Franco and Tagliamonte2020). So now the question is: How do you find the patterns?
Testing Hypotheses
Labov’s (Reference Labov1969a:729) third exploratory manoeuvre is to ‘identify all the sub-categories which would reasonably be relevant in determining the frequency with which the rule in question applies’. These categories allow the analysis to detect the underlying patterns, the internal linguistic contexts that are hypothesised to influence the choice of one variant over another. How does one find them? Sometimes these are discovered by scouring the literature, both synchronic and diachronic. Sometimes they ‘emerge from the ongoing analysis because of various suspicions, inspections, and analogies’ (Labov, Reference Labov1969a:729). Sometimes they are stumbled upon by chance in the midst of analysis, and a ‘Eureka!’ experience unfolds. More often, the very worst days of variation analysis come when you are floundering in reams of statistical analyses and data and numbers, and you just can’t see the forest for the trees! As long as one’s practice has been ‘carried out with a degree of accuracy and linguistic insight’, Labov promises that ‘the end result is a set of regular constraints which operate upon every group and almost every individual’ (Reference Labov1969a:729). Indeed, it never ceases to amaze me what patterns underlie linguistic variables that one has no inkling of. One prominent example in the field is the French inflected future, which is nearly categorical in negative contexts but which native speakers seem to be entirely unaware of (Emirkanian & Sankoff, D., Reference Emirkanian, Sankoff, Lemieux and Cedergren1985; Poplack & Dion, Reference Poplack and Dion2009). Another example comes from my study of verbal -s on verbs outside of third person singular. A paper from the late 1980s (Tagliamonte & Poplack, Reference Tagliamonte and Poplack1988) did not notice the contrast between pronouns and noun phrases. A later paper revised the methodology leading to an important discovery that helped explain the variation and its ancestry (Poplack & Tagliamonte, Reference Poplack and Tagliamonte1989). Later, cross-variety comparison (Tagliamonte, Reference Tagliamonte2015) revealed how the variable has evolved from Scotland to Southern England to North America and how verbal -s continues to advance knowledge on variable processes across time and space (e.g. Wolfram, Reference Wolfram2000; Poplack & Tagliamonte, Reference Poplack, Tagliamonte and Hickey2005; José, Reference José2007; Britain & Rupp, Reference Britain and Rupp2024).
While I was writing my dissertation, I came to a particularly impassable and apparently dead end in my analysis. I could not see any patterns! In desperation, I wailed at one of my mentors, ‘There are just no patterns at all.’ The response was empathetic, but firm: ‘Take it from an old variationist like me – there will be patterns. Keep looking.’ And, of course, there were.
Once it can be established that a variable exists in a body of materials, the variationist sociolinguist will embark on the long process of studying the feature: circumscribing the variable context, extracting the relevant data from corpora, coding the material according to reasoned hypotheses gleaned from the diachronic and synchronic literature, and then analysing and interpreting the results.
The question inescapably arises, Why use variation analysis? My answer is this. It is an area of the discipline that involves language as it is being used, sometimes referred to as ‘real language’ (e.g. Milroy & Milroy, Reference Milroy and Milroy1993), so it is inherently hands-on and practical; it employs a methodology that is replicable and accountable; it provides you with the ‘tools’ to analyse language, not simply on an item-by-item basis, but at the level of the underlying system. Finally, variation analysis puts language in context, socially, linguistically, synchronically, and diachronically. In the end, by conducting a variation analysis you get closer to knowing what language is and what human beings are all about.
Organisation and Logic of This Book
The book is organised so that the chapters take the reader from the first stages of research right through to the last ones. The chapters build from simple observations to basic conceptual initiatives, to training procedures. I then move to general research problems and issues and gradually turn to explaining how to resolve complex data handling, computational, and linguistic problems. In Chapters 8–11, the focus turns to performing analyses and developing skills in R. Then I will return to interpretation, explanation, and communication skills. Examples from my own research demonstrate problem-solving at each stage in the research process, providing the techniques of variation analysis, but also the process through which it unfolds. Each chapter ends with an exercise highlighting the topic just covered.
I will use two linguistic variables as test cases, which will be shared with readers online at www.cambridge.org.tagliamonte. One variable is binary, variable (hwat), with the variants [w] and [ʍ]; the other variable has many variants, variable (adj_pos), all the adjectives in the semantic field of positive evaluation, including great, cool, and lovely and many others. The data files are adj_1-2-24.tsv and hwat_2-13-24.tsv. Also included are several RMD files for student practice and their output as PDF files.
In order to tap into the structured heterogeneity that is rich in living language, it is necessary to gain access to language in use, whether it is in written language, literature, in the media or, as is most typical of the variationist approach, ‘in the street’ (Labov, Reference Labov and Labov1972c:99), that is vernacular interaction. To study the living vernacular, it is necessary to go out of the office, beyond the anecdotal, and into the speech community. Fieldwork and data collection will be described in Chapter 2.
The purpose of this exercise is to develop your ability to observe linguistic variability. In the process you will begin to develop a sociolinguist’s ear and eye.
Find some language material. Any data will do, for example an audio-recording, a video, a podcast, a newspaper, a novel, an email message, an online conversation. Then examine it – carefully.
Consider different areas of grammar (e.g. phonology, morphology, syntax, discourse). Notice morphological, syntactic, or discursive alternations discourse markers, variation in quotative use, syntactic structure, and so on. Be attentive so that you spot unrealised forms (e.g. missing plurals, possessives, prepositions, articles). Provide an inventory of the variants that occur in the data. Illustrate intra-individual alternation of forms. Try to find a super token. For example:
I mean I was real small and everything you know really tiny built … (YRK, kyoung, woman, 31)
Summarise the nature of the features you have identified. Are there any features that are unfamiliar to you? Are there features that are typical of older rather than younger individuals? One gender more than another? Standard vs non-standard, etc.? Do they vary across your sample? Is one variant known to be predisposed to certain sectors of the population over the other? How? Are any forms geographically or regionally correlated? Also make note of the linguistic and social contexts in which each variant occurs. Can you spot any trends?
When you make a linguistic observation from a data set, always record it and provide an example from the data. Further, ensure that the example is referenced to the location in the original data sample (i.e. individual and coordinates), audio-recording time stamp, or whatever is suitable for the data you are examining. I will often take the front page of a newspaper from the day I present this topic to a class and use it to illustrate how normal variation in language really is. There are usually some good examples. Look for some of the variants in (1). Try it.
