There is substantial variation in language: both within and across language varieties. We’ll see some examples of both of these kinds of variation and I’ll introduce one of the central concepts used in variationist sociolinguistics: the linguistic variable.
All languages exhibit variation
Many linguistic approaches to the study of language are concerned with language variation. As you’ve read about in other chapters, theories about how language works rest on evidence that comes about by contrasting the way something is said or signed in two or more different languages, dialects, or varieties.
Language, dialect, variety. Colloquially, the term dialect is used to refer to ways of speaking that people perceive to be substandard, low status, associated with working class, non-prestigious, geographically-isolated, or some derivation or aberration from a ‘standard’ version of the language. The linguistic fact though is that everyone has a dialect. Rather than think about languages and dialects in a hierarchical way, linguists think about dialects as subdivisions of a language. Sometimes, linguists might talk about the “standard dialect” but it’s important to emphasize that no dialect, not even what we might call the “standard dialect” is objectively (linguistically) superior to any other dialect of the language. Another term, ‘variety’ doesn’t have the same negative connotations that ‘dialect’ has, and so we’ll usually use that to refer to subdivisions of a language in this chapter.
For example, consider the Icelandic and Danish sentences in (1) and (2). Both sentence express the same meaning.
Obviously some of the lexical items and morphemes differ between (1) and (2) – as expected given that they come from two different languages. At the same time, you can see similarities between them – also as expected since these two languages are quite closely related (they are both North Germanic languages). One syntactic difference between the two examples is the order of the negative marker and the main verb in the embedded clause (hafði ekki ‘have not’ in Icelandic and ikke havde ‘not have’ in Danish). If we were looking at this data like a syntactician, we might look at (1) and (2) and use the two different word orders (i.e., VERB-NEG vs. NEG-VERB) as evidence in an analysis of verb raising. We’ll call this cross-linguistic variation: different ways of doing the same thing in different languages or varieties.
But within a single language or variety – or even a single person – the specific realization of abstract structures (like word order) can vary. Consider the rhyming couplet in (3) from Shakespeare’s Romeo and Juliet (386, 670).
Example (3) shows the same word order variation that we see in (1) and (2) but here, the two different ways of doing the same thing appear within the same language! And while technically spoken by two different characters, the two sentences were written by the same person! This isn’t that surprising though because Early Modern English allowed for both options: an Icelandic-like/Romeo-like VERB-NEG order and a Danish-like/Juliet-like NEG-VERB order (together with ‘do-support’, see Chapter 6 and Chapter 14). Within this one rhyming couplet, we see sociolinguistic variation: two or more ways of doing the same thing within a language, variety, and individual.
What’s a linguistic variable?
When we approach language from a variationist sociolingistic perspective, we call the choices between a set of options that mean the same thing a linguistic variable. The individual options that people choose between in the course of language use, we call variants. Linguistic variables exist in all languages and varieties, in all modalities, and at all domains of language from the phonetic to the pragmatic. A linguistic variable is an abstract set; there’s nothing out there in the world that we can point to and be like “hey, that’s a linguistic variable!”. We only ever see or hear the abstract variable as one of its concrete variants. Let’s have a look at some examples of linguistic variables from different languages and different domains of language.
In Auslan, NZSL, and ASL, some signs that are prescriptively located at the forehead, like KNOW, THINK, and NAME, have variable location: at the forehead or lowered to near the cheek (Bayley, Schembri, and Lucas 2015). You can see these two variants in Auslan/NZSL in Figure 12.1 and watch a video at this link in ASL.
This is an example of phonetic-phonological variation with variants differing with respect to an articulatory parameter (in this case, location). Another kind of phonetic-phonological variable involves variation between the presence and absence of a sound segment. In Beijing Mandarin, open syllables (i.e., syllables without a coda) can variably be rhotacized (i.e., produced with a rhotic coda). For example the word meaning ‘bag’ can be said with the open syllable variant (包 bao [paw]) or the rhotic coda variant (包儿 baor [pawr]) (Zhang 2008). We also find linguistic variables in the morphophonological domain of languages. Standardized English contains a categorical alternation with the indefinite article between an and a with an occurring prior to a vowel and a occurring elsewhere (cf. an apple vs. a pineapple). However, in contemporary London English, especially among immigrant youth of colour, the pre-vocalic context exhibits variation between an and a; both an apple [ənæpl] and a apple [əʔæpl] are possible (Gabrielatos, Torgersen, Hoffmann, and Fox 2010).
We can also find morphosyntactic variables in languages. In North Baffin Inuktitut, transitive constructions can variably occur with ergative alignment or antipassive alignment. These alignment types differ in terms of the morphological case that arguments have and the kind of agreement that appears on verbs. With ergative alignment, as in (4a), the object is marked with absolutive case (which appears as a null morpheme -ø) and the verb agrees with both subject and object (which appears as the morpheme -jara). With antipassive alignment, as in (4b), the object is marked with an oblique case (it occurs with the morpheme -mit) and the verb agrees only with the subject (which appears as the morpheme -vunga) (Carrier 2020).
Languages can also have linguistic variables with variants that differ in multiple ways, across different domains. For example, in Tagalog, the meaning of adjectives can be intensified with several variants that differ lexically, morphologically, and morphosyntactically from each other variant, as in (5) (Umbal 2019).
One variant, in (5a), uses a free morpheme sobra (similar to English very, really etc.). Reduplication of the adjective in (5b) and affixation of the morpheme napaka (5c) are two morphosyntactic variants. Finally, the exclamative construction in (5d) is a fourth, syntactic, variant.
These are just a small handful of examples of linguistic variables in different languages and in different domains of language. All languages have variation like this in all the different parts of a language’s grammar.
What isn’t a linguistic variable?
Whatever domain of language a linguistic variable exists in, the variants ‘do the same thing’ in some way. This should give you a good idea about what a linguistic variable is, but before moving on, it’s important to point out a couple of linguistic concepts that are similar to but are not linguistic variables: 1) synonyms and 2) categorical alternations.
Synonyms are a concept that is often confused with linguistic variables – and for good reason: some synonyms can be linguistic variables, but not all of them! Synonyms are pairs or sets of words that share the same or similar meaning like car, automobile, ride, horseless carriage, jalopy, hooptie and paddock basher, which all denote those four-wheeled, motor-powered vehicles that many people drive. These options certainly seem like two or more ways of doing the same thing but critically, different synonyms are generally not interchangeable in the same way that variants of linguistic variables are. As we saw in Chapter 7, languages have only very few absolute synonyms. The different options may have different connotations or social meanings that make one option much more suitable than another option. For example, jalopy, hooptie, and paddock basher connote that the vehicle is old or run-down; ride might be used in informal contexts and automobile in formal contexts. Some options may also only appear in particular regional or social varieties. For example, jalopy is an older North American English term (you’ll find it used several times in Jack Kerouac’s On The Road, written in the late 1940s), hooptie is typically associated with Black English (having been the topic of the 1989 hip-hop song, ‘My Hooptie’ by Sir Mix-a-Lot), and paddock basher is a term mostly only found in Australia, referring to a car only suitable to drive around on a farmer’s field (which, non-coincidentally, is referred to as paddock there). Some of these options have become obsolete: you might only hear horseless carriage today if you’re watching something like Downton Abbey. Because of the differing connotations or limited regional and social usages, synonyms like these are not generally interchangeable in the same way as linguistic variables. However, sometimes they can be! Critically, if the choice between options systematically co-varies with social and/or linguistic constraints, synonyms can be analyzed as linguistic variables. For example, in English adjectives of positive evaluation like cool, awesome, sick, neat, and great have been found to correlate with linguistic and social constraints (Tagliamonte and Pabst 2020).
As you’ve already seen in previous chapters, languages are full of categorical alternations (e.g., the phonological rules discussed in Chapter 4). Categorical alternations are a second concept that can be easily confused with linguistic variables. Now it’s true: categorical alternations represent variation within a language and the options are indeed two or more ways of doing the same thing. However, they depend strictly on the linguistic context that they appear in. In other words, the choice between the options is deterministic. A linguistic rule like Canadian Raising (the nucleus of the /aɪ/ and /aʊ/ diphthongs is raised to [ʌ] before voiceless consonants) in Canadian English is an example of a categorical alternation: if we know what phoneme comes after the vowel then we know if the nucleus will be [a] or [ʌ]… it’s predictable! This differs from linguistic variables because a variable can be realized as its different variants even within identical linguistic contexts!
That said, even though linguistic variables are not deterministic they also aren’t random! Instead, linguistic variables are probabilistic in nature. To use the wording of one of the foundational studies in variationist sociolinguistics, there is order amid the heterogeneity (Weinreich, Labov, and Herzog 1968: 100). The choice between different variants of a linguistic variable is subject to probability given many different possible conditioning factors (also called constraints). Like with categorical alternations, these conditioning factors can include aspects of the linguistic context. So think about (6).
(6) I’m fishin’ this morning.
You’ve probably noticed that the end words like fishing and morning that end with -ing sometimes get pronounced as [ɪn] instead of [ɪŋ]. That’s another linguistic variable! Like all linguistic variables though, the choice between [ɪn] and [ɪŋ] isn’t random. In most varieties of English, the [ɪn] variant is more likely to occur in verbs (like fishin’) than in nouns (like morning). That’s a conditioning factor for this variable! However, where the socio- comes into variationist sociolinguistics – and why the analysis of linguistic variables is so important to the field – is that these conditioning factors also include social factors as well. In other words, whether someone uses [ɪn] or [ɪŋ] in a particular moment depends on social facts about the speaker/signer, their interlocutors (the other people in the conversation), and other aspects of the sociocultural context of the interaction. By counting and quantifying variants of a linguistic variable, social facts, in addition to linguistic facts, can be uncovered.
Source of linguistic constraints. You might be wondering… where do these linguistic constraints on variation come from? Why is [ɪn] more likely to occur in a verb than in a noun? These constraints have many different sources for different variables but some are rooted in historical structural patterns found in earlier versions of the language. Believe it or not, there’s good evidence that the reason English speakers are more likely to use [ɪn] in a verb than in a noun today goes all the way back to a pattern in Old English, spoken between the 5th and 11th centuries CE! The modern -ing morpheme came about through the coalescence (or merger) of two different grammatical morphemes found in Old English: -ende, which marked the present participle (I am teaching today > Old English tǣċende) and -ung the verbal-noun marker (Teaching is fun > Old English tǣċung). In the Middle English period, these two morphemes started to merge together as -ing but the alveolar nasal found in the Old English present participle marker stuck around as a variant! As the morphemes merged, people lost track of the older categorical ‘alveolar-in-verbs and velar-in-nouns’ pattern and both variants were used with verbs and nouns. However, traces of the old pattern are still apparent in the form of a conditioning factor!
There are many different types of linguistic variation. Some variation distinguishes varieties from each other (cross-linguistic variation), other variation exists within a single variety or person (sociolinguistic variation). Variation within a single variety between variants of what we call linguistic variables is subject to probability given social and linguistic conditioning factors that favour or disfavour certain options.
Bayley, R., Schembri, A., and Lucas, C. (2015). Variation and change in sign languages. In A. Schembri and C. Lucas (eds.) Sociolinguistics and Deaf Communities. Cambridge University Press. pp. 61–94.
Carrier, J. (2017). The ergative-antipassive alternation in Inuktitut: Analyzed in a case of new-dialect formation. The Canadian Journal of Linguistics / La revue canadienne de linguistique 62(4), 661-684. https://www.muse.jhu.edu/article/678231.
Gabrielatos, C., Torgersen, E. N., Hoffmann, S., & Fox, S. (2010). A Corpus-Based Sociolinguistic Study of Indefinite Article Forms in London English. Journal of English Linguistics, 38(4), 297–334. doi.org/10.1177/0075424209352729
Tagliamonte, S. A., & Pabst, K. (2020). A Cool Comparison: Adjectives of Positive Evaluation in Toronto, Canada and York, England. Journal of English Linguistics, 48(1), 3–30. doi.org/10.1177/0075424219881487
Umbal, P. (2019). Contact-induced change in Heritage Tagalog: Evidence from adjective intensification. PhD generals paper: University of Toronto.
Weinreich, U., Labov, W., & Herzog, M. (1968). Empirical foundations for a theory of language change. University of Texas Press.
Zhang, Q. (2005). A Chinese yuppie in Beijing: Phonological variation and the construction of a new professional identity. Language in Society, 34(3), 431-466. https://doi.org/10.1017/S0047404505050153