51.07.02 · linguistics / sociolinguistics

Variationist sociolinguistics: the Labovian method and the apparent-time construct

shipped3 tiersLean: none

Anchor (Master): Labov 1963 The social motivation of a sound change (Word 19); Labov 1966 The Social Stratification of English in New York City (CUPT); Labov 1972 Sociolinguistic Patterns (Penn); Weinreich-Labov-Herzog 1968 Empirical foundations for a theory of language change

Intuition Beginner

Listen to two New Yorkers say the word "car". One says "cah"; the other says "ca-r", pronouncing the r. In 1962, William Labov walked into three Manhattan department stores — Saks, Macy's, and S. Klein — and asked each clerk for directions to a department on the fourth floor, then asked them to repeat the answer. The "fourth floor" answer told him whether the clerk pronounced the r. He found a clean pattern: Saks clerks (upper-middle-class clientele) pronounced the r most often, Macy's clerks (middle-class) less often, and S. Klein clerks (working-class) least often. The variable was not random. It followed social class, with a separate gradient by style (careful repetition versus casual answer).

This is the variationist programme in one study. Dialect variation is not free noise. The way people speak varies along axes that can be measured: social class, gender, age, ethnicity, and the attention the speaker pays to speech. The variationist method is the apparatus that measures those gradients and reads off whether a sound change is in progress, where it is in its trajectory, and what social meaning the variant carries.

Why this matters: before Labov, dialectology meant drawing lines on a map where one word gave way to another. The lines ignored cities, ignored class, and ignored change in progress. Labov's method turned dialectology into a quantitative social science. Sound change was no longer a thing of the past, to be reconstructed by the comparative method; it could be observed as it happened, in real time, in the streets of a living city.

Visual Beginner

The picture shows two panels. The left panel plots the frequency of rhotic r (pronounced r in "car", "fourth", "floor") by department store, with three bars: Saks highest, Macy's middle, S. Klein lowest. Each bar is split into casual speech (lower) and careful speech (higher). The right panel is the apparent-time plot: the frequency of a sound change variant (say, the Northern Cities Shift of "cat" toward "ket") on the vertical axis, plotted against the speaker's age on the horizontal. The points slope downward from younger to older speakers, indicating that the newer variant is used more by younger speakers and the change is in progress.

The two panels summarise the method: variation is structured by social factor and by style (left), and the age gradient reveals a change in progress (right). The structure is what turns casual speech into evidence.

Worked example Beginner

Run a miniature Labovian study of the rhotic r in three New York department stores, using counts and percentages.

Step 1. Sample. Visit three stores targeting three social strata (Saks upper-middle, Macy's middle, S. Klein working-class). Ask 68 clerks at each store for directions to a department on the fourth floor, noting whether each "r" in "fourth" and "floor" was pronounced. The principle of accountability requires counting every opportunity, so a clerk who said "fourth floor" twice gave four potential rhotic r's.

Step 2. Count. At Saks, 62 percent of possible r's were pronounced; at Macy's, 51 percent; at S. Klein, 20 percent. The class gradient is steep and monotone.

Step 3. Stratify by style. At each store, the casual answer (the first reply) had fewer pronounced r's than the careful repetition (the second reply, when the researcher feigned not having heard). The style-shift is in the same direction at all three stores.

Step 4. Interpret. Two gradients — by class and by style — point to the same conclusion. The rhotic r is an incoming prestige variant: it is more frequent in higher social strata and in more self-monitored speech. The pattern is not random; it is a change in progress, with the higher strata leading the adoption of the prestige norm.

What this tells us: a clean social gradient, replicated across three settings and two styles, is enough to identify a sound change in progress. The method reads the trajectory of change from a single afternoon's fieldwork.

Check your understanding Beginner

Exercise (easy, multiple choice).

What is the principle of accountability in variationist method?

A. Researchers must be accountable to their informants B. Every instance where the linguistic variable could have occurred must be counted, whether or not the variant of interest was used C. Researchers must publish their data D. Each informant must give consent

Hint

The principle is methodological. It governs how the denominator of a variation frequency is computed — by counting all opportunities, not just the occurrences of the variant.

Answer

B. To compute the frequency of a variant fairly, the analyst must count every site where the variable could have occurred (the denominator) and every site where the variant did occur (the numerator). Counting only the occurrences biases the estimate. The principle of accountability is the methodological core that makes variation frequencies comparable across speakers and across studies.

Exercise (easy, multiple choice).

The apparent-time construct infers a sound change in progress by:

A. Comparing the speech of older and younger speakers at a single moment B. Following the same speakers across decades C. Reading historical records D. Using acoustic equipment to measure tiny differences

Hint

Apparent time uses age stratification in a cross-section. Real time tracks the same community across actual years.

Answer

A. If younger speakers use an innovative variant more often than older speakers do, and if we assume that individual speakers' habits are relatively stable across their lifetimes, then the age gradient in a cross-section is evidence of change in progress: the younger cohort represents the future state of the community. The construct is called "apparent" because it uses age as a proxy for time, distinguished from real-time studies that resample the same community decades later.

Formal definition Intermediate+

Variationist sociolinguistics treats linguistic variation as the systematic conditioning of a categorical or quantitative choice by both internal (linguistic) and external (social) factors. Its core notions are the linguistic variable, the principle of accountability, the Labovian interview, and the variable rule.

Definition (linguistic variable). A linguistic variable is a point in the grammar where two or more variants can occur in the same syntactic and phonological environment with the same truth-conditional meaning. The variants are denoted $(v_{1}, v_{2}, \dots)$ and the variable's distribution is described by the frequency of each variant conditioned on the relevant factors. The classic example is the English (ing) variable with variants /iŋ/ and /ɪn/ ("running" versus "runnin"), which carries no semantic difference but varies systematically by class, style, and following sound.

Definition (principle of accountability). To estimate the frequency of a variant, the analyst counts every site in the corpus where the variable could have occurred (the envelope of variation) and every site where each variant was used. The denominator is the count of all such sites, not the count of any subset. The principle excludes sites where the variants are not in free alternation (for instance, where one variant is blocked by a grammatical constraint), so that the variation analysed is genuine.

Definition (Labovian sociolinguistic interview). The Labovian interview is a structured but conversational data-elicitation protocol that moves the speaker through a gradient of attention to speech, from casual (emotionally involving personal narratives) to careful (formal interview questions) to word-list (maximal self-monitoring). The gradient yields style-shifts that reveal whether speakers monitor the variable.

Definition (variable rule). A variable rule is a logistic model expressing the probability that a variant $v_{i}$ occurs as a function of linguistic factors (preceding and following phonological context, grammatical category, stress) and social factors (age, gender, social class, style). Formally $Pr (v_{i} ∣ context) = logit^{- 1} (β_{0} + \sum_{j} β_{j} x_{j})$ , where the $x_{j}$ encode the factor groups and the $β_{j}$ are the factor weights. The model lets the analyst separate the linguistic constraint (which variants are favoured by which phonological environments) from the social constraint (which speakers use which variants at which rates).

Counterexamples to common slips Intermediate+

Treating the variant frequency as a speaker property. The frequency is a corpus-level statistic computed from many opportunities, not a stable trait. Two speakers with the same underlying grammar can produce different frequencies in a short sample because of which opportunities happened to arise.
Conflating real time and apparent time. Apparent time uses the age gradient in a single cross-section as a proxy for change in progress. The construct is valid only if individual speakers' habits are stable across their lifetimes, which has been tested and largely confirmed for phonological variables but is less reliable for lexical ones.
Ignoring the envelope of variation. Computing a frequency over sites where only one variant is grammatical yields a meaningless ratio. The envelope must exclude sites where the variants are not in free alternation, such as those where a higher constraint (phonotactic, morphological) blocks all but one variant.

Linguistic theory Intermediate+

Theorem (Labov's department-store stratification, 1966). The frequency of the rhotic r in New York City department stores stratifies monotonically by the social class of the store's clientele and by the style of the elicitation. Across 264 informants in three stores, the percentage of possible r's that were pronounced was 62 percent at Saks (upper-middle-class clientele), 51 percent at Macy's (middle), and 20 percent at S. Klein (working), with a parallel stratification by style (casual versus careful) within each store.

Reconstruction. Labov's interpretive claim was that the rhotic r is an incoming prestige variant: it is more frequent in higher social strata and in more self-monitored speech, two gradients pointing in the same direction. The stratification by class identifies the social meaning of the variant (upper-middle prestige); the stratification by style identifies speakers' awareness (they monitor the r and produce it more in careful speech). The two gradients together diagnose a change in progress in the direction of the prestige norm. The apparent-time gradient — younger speakers using the rhotic r more — would confirm the trajectory in real time, and later studies (Labov 1984 resampling) did so.

Theorem (the apparent-time construct). If individual speakers' phonological habits are stable across their adult lifetimes, then the age gradient in a community cross-section is a valid proxy for the change in progress: the frequency of an incoming variant is higher in younger cohorts because they acquired it as the community norm. The construct was validated by real-time replications of the 1966 New York study, which confirmed that the rhotic r continued to rise across the decades, matching the apparent-time prediction.

Bridge. The apparent-time construct builds toward 51.06.01 historical-comparative linguistics by providing the real-time window that the comparative method lacked — change can be observed as it happens, not only reconstructed from its end state, and appears again in 51.07.01 sociolinguistics foundations as the methodological engine that turned variation from a descriptive observation into a quantitative science. The foundational reason the construct works is the stability assumption, which has been tested for phonology and largely confirmed (speakers do not shift their phonological variants substantially past adolescence). This is exactly the assumption that identifies a community's age gradient with a slice of its historical trajectory, and the bridge is from synchronic variation to diachronic change within a single analytical framework. The pattern generalises across the major variationist studies — Martha's Vineyard, the Northern Cities Shift, the (ing) variable, t-glottalisation — each of which diagnoses an ongoing change by reading the social and stylistic gradients as the footprint of its trajectory.

Exercises Intermediate+

Exercise 1 (easy, multiple choice).

The envelope of variation refers to:

A. The geographic region in which a variant occurs B. The set of grammatical sites where the variants of a linguistic variable could in principle alternate C. The historical period during which a variant existed D. The range of acoustic realisations of a single phoneme

Hint

The envelope defines the denominator of the variationist frequency. It is the set of sites where the choice between variants is genuinely open.

Answer

B. The envelope of variation is the set of grammatical and phonological contexts in which the variants of a linguistic variable could in principle alternate. The principle of accountability requires counting every site within the envelope, and excluding sites where one variant is blocked (for example, where a word-final consonant cluster must simplify by a higher regular sound rule). Computing frequencies over the envelope rather than over the whole corpus is what makes them comparable.

Exercise 3 (medium, symbolic).

Write the variable rule (logistic regression) for the probability that a speaker uses the [ɪn] variant of the (ing) variable, conditioned on style (casual versus careful), following sound (vowel versus consonant), and speaker sex. Identify the factor weights whose sign and magnitude the variationist interprets.

Hint

Use dummy coding for the categorical predictors. The intercept is the baseline log-odds; each factor weight is the change in log-odds when that factor is present.

Answer

Let $v_{i} = 1$ if the [ɪn] variant is used, $0$ if [iŋ]. Let $S_{i} = 1$ for casual style (0 for careful), $F_{i} = 1$ if the following sound is a consonant (0 for vowel), $M_{i} = 1$ if the speaker is male (0 for female). The variable rule is

Pr (v_{i} = 1) = logit^{- 1} (β_{0} + β_{S} S_{i} + β_{F} F_{i} + β_{M} M_{i}) .

The $β$ coefficients are the factor weights. A positive $β_{S}$ means casual style favours [ɪn]; a positive $β_{F}$ means a following consonant favours [ɪn]; a positive $β_{M}$ means male speakers favour [ɪn]. The magnitudes are interpreted in log-odds (or converted to odds ratios); the variationist convention reports factor weights in the $[0, 1]$ range centred at 0.5, with weights above 0.5 favouring the variant.

Exercise 4 (medium, multiple choice).

In the Northern Cities Shift (Chicago, Detroit, Buffalo, Rochester), the most characteristic change is:

A. The loss of rhotic r B. The raising and fronting of the vowel in "cat" toward the vowel in "ket" C. The merger of "pin" and "pen" D. The loss of the vowel in unstressed syllables

Hint

The Northern Cities Shift is a chain shift affecting the short vowels. The triggering event is the raising and fronting of /æ/ (the vowel in "cat").

Answer

B. The Northern Cities Shift, documented in Labov, Ash, and Boberg's Atlas of North American English (2006), is a chain shift of the short vowels. Its triggering event is the raising and fronting of /æ/ (as in "cat") toward [eə] or [ɪə]. The shift then pulls /ɑ/ ("hot") forward to fill the gap left by /æ/, and so on through the short-vowel subsystem. The shift is strongest in the Inland North (Chicago, Detroit, Buffalo, Rochester) and is a leading case of change in progress in apparent time.

Exercise 5 (medium, numeric).

A variable rule fit yields a factor weight of 0.65 for casual style on the (ing) variable (baseline 0.5). Convert this weight to an odds ratio (relative to the baseline). The conversion is $OR = w / (1 - w)$ , with the baseline odds ratio 1.0.

Hint

The factor weight $w$ in $[0, 1]$ is converted to a log-odds by $ln (w / (1 - w))$ , and the odds ratio relative to the 0.5 baseline is $w (1 - 0.5) / ((1 - w) \cdot 0.5) = w / (1 - w)$ .

Answer

1.86. $OR = 0.65/0.35 \approx 1.86$ . A weight above 0.5 corresponds to an odds ratio above 1, meaning the factor favours the variant. Casual style thus nearly doubles the odds of the [ɪn] variant relative to careful style. The interpretation: the variationist reads a factor weight of 0.65 as "casual style moderately favours the non-standard variant", consistent with the (ing) variable being a Labovian marker subject to style-shifting.

Exercise 6 (medium, multiple choice).

The community of practice (Eckert) refinement of the variationist programme argues that:

A. Social class is irrelevant to variation B. Variation correlates more strongly with locally meaningful social categories (the "jocks" and "burnouts" in a Detroit high school) than with macro-sociological class alone C. Acoustic analysis should replace impressionistic coding D. Apparent time should replace real time

Hint

Eckert's ethnographic study of a Detroit-area high school showed that the speakers driving the Northern Cities Shift were the "burnouts" (working-class-oriented students) more than the "jocks" (middle-class-oriented students), within a single macro-class.

Answer

B. Penelope Eckert's Linguistic Variation as Social Practice (2000) showed that macro-sociological class categories are too coarse: the same macro-class can contain locally meaningful social groups (the "burnouts" and "jocks" of a Detroit high school) whose members align with or against a sound change in distinct ways. The community of practice — a group that comes together around mutual engagement in a shared enterprise — is a finer social unit than class and predicts variation more strongly. The refinement does not reject class but situates it within the locally meaningful categories that speakers actually use.

Exercise 7 (hard, symbolic).

Weinreich, Labov, and Herzog (1968) formulated five problems for a theory of language change: the constraints problem, the transition problem, the embedding problem, the evaluation problem, and the actuation problem. State each problem in one sentence and identify which the variationist method addresses directly.

Hint

The constraints problem: what changes are possible? Transition: how does the change move through the grammar? Embedding: how is it situated in social and linguistic structure? Evaluation: how do speakers evaluate the variants socially? Actuation: why did this change happen at this moment in this community?

Answer

Constraints problem: What are the possible types of language change, and what general principles rule out impossible changes? (Addressed by phonological typology and the comparative method.)
Transition problem: How does a language move from one grammar to another across the span of the change? (Addressed by the variationist variable rule, which models the trajectory of frequency change.)
Embedding problem: How is the change situated within the linguistic system and within the social structure? (Addressed by the variationist method's conditioning of variant frequency on linguistic and social factors.)
Evaluation problem: How do speakers evaluate the variants socially (positive, negative, neutral), and how does this evaluation drive or retard the change? (Addressed by matched-guise experiments and style-shifting analysis.)
Actuation problem: Why did this particular change happen in this particular community at this particular moment? (Largely unsolved; the variationist method can document the trajectory but not explain the initial trigger, which is the famous "actuation riddle".)

The variationist method addresses transition, embedding, and evaluation directly; constraints is addressed in conjunction with phonological theory; actuation remains the hardest problem and is not directly addressed by the method.

Exercise 8 (hard, multiple choice).

The curvilinear principle (Labov) states that, for sound changes in progress below the level of social awareness:

A. The change is led by the lowest social class B. The change is led by the highest social class C. The change is led by the interior social groups (upper-working and lower-middle), not the extremes D. The change is led equally by all social classes

Hint

Labov's empirical generalisation across many studies: the highest and lowest extremes of the social hierarchy do not lead changes; the interior groups do.

Answer

C. Labov's curvilinear principle: sound changes in progress (especially the "change from below" that operates below conscious awareness) are typically led by the interior social groups — the upper-working and lower-middle classes — rather than by the highest or lowest extremes. The finding held across many studies of vowel shifts and consonantal changes. The interpretation is that the interior groups are sufficiently central to the community's network to spread an innovation but not so anchored to either prestige or stigmatised norms as to block it. The principle is one of the empirical generalisations that distinguish variationist findings from armchair theory.

Advanced results Master

Result 1 (Labov 1963, Martha's Vineyard centralisation). The centralisation of the vowels /ay/ and /aw/ on Martha's Vineyard correlates positively with orientation toward the island (resistance to summer tourists) and with age cohort, indicating a change in progress whose social meaning is the assertion of local identity. The study is the founding document of variationist sociolinguistics and the first demonstration that a sound change could be shown to be in progress by quantitative field methods.

Result 2 (Labov 1966, department-store stratification). The rhotic r in New York City department stores stratifies monotonically by the social class of the clientele and by the style of the elicitation, demonstrating that variation in a single phonetic feature carries information about both social structure and speaker self-monitoring. The study established the Labovian interview as the standard data-collection protocol and the apparent-time construct as the standard diagnostic of change in progress.

Result 3 (Weinreich-Labov-Herzog 1968, the five problems). A theory of language change must address five problems: constraints (which changes are possible), transition (how the change moves through the grammar), embedding (how the change is situated in social and linguistic structure), evaluation (how speakers evaluate the variants), and actuation (why this change happened here and now). The framework organised the field and identified the actuation problem as the standing frontier.

Result 4 (Eckert 2000, community of practice). Variation correlates more strongly with locally meaningful social categories (communities of practice) than with macro-sociological class alone. The "burnouts" and "jocks" of a Detroit-area high school aligned differently with the Northern Cities Shift despite belonging to the same macro-class, showing that the social meaning of a variant is constructed within local networks rather than imposed from above.

Result 5 (Labov-Ash-Boberg 2006, the Atlas of North American English). Telephone-survey data from across the urban United States revealed systematic regional dialect regions defined by ongoing vowel shifts: the Northern Cities Shift (Inland North), the Southern Shift (the U.S. South), the Canadian Shift, the Pittsburgh and Mid-Atlantic chains, and the merger of /ɑ/ and /ɔ/ (the "cot-caught" merger) spreading westward from western Pennsylvania. The Atlas established that the major regional variation in North American English is driven by chain shifts in progress, each diagnosable in apparent time.

Result 6 (the variable rule, Sankoff 1978 onward). The probability of a variant can be modelled as a logistic regression on linguistic and social factors, with factor weights interpretable as the linguistic and social constraints on variation. The variable rule succeeded earlier impressionistic coding by giving quantitative estimates of effect size and significance, and modern mixed-effects versions (Johnson, Drager) add random effects for speaker and word to handle nested sampling.

Synthesis. The Labovian method builds toward 51.06.01 historical-comparative linguistics by giving the comparative method a real-time window — change can now be observed as it happens, not only reconstructed from its end state — and appears again in 51.07.01 sociolinguistics foundations as the methodological engine that turned dialect geography into a quantitative social science. The foundational reason the method works is that variation is structured along measurable axes (class, age, gender, style), and reading those axes' gradients diagnoses the trajectory and social meaning of a change in progress. This is exactly the structure that identifies a linguistic variable with a quantitative choice conditioned on internal and external factors, and putting these together with the apparent-time construct and the variable rule, the bridge is from observed variation to the underlying change, with the social gradient as the footprint and the age gradient as the clock. The pattern generalises across the major variationist findings — Martha's Vineyard, the Northern Cities Shift, the (ing) variable, t-glottalisation, AAVE copula absence — each of which reads the social and stylistic gradients as the signature of a change with a specific trajectory and social meaning, and the central insight is that the structured heterogeneity of speech is the empirical face of change in progress, not noise around a homogeneous norm.

Full proof set Master

Proposition (The apparent-time construct under the stability assumption). Suppose that individual speakers' phonological habits are stable across their adult lifetimes (the stability assumption). Then a cross-sectional age gradient in the frequency of a variant at time $t_{0}$ is a valid proxy for the change in progress: the cohort born in year $y$ approximates the community's state at year $y + A$ (where $A$ is the age of acquisition), and the gradient $\partial f / \partial age$ estimates the rate of change.

Proof. Let $f (c, t)$ denote the community-level frequency of a variant in cohort $c$ at time $t$ . Under the stability assumption, $f (c, t) = f (c, t_{0})$ for all $t$ in the speakers' adult lifetimes; speakers carry the frequency they acquired at age $A$ . So a cross-section at time $t_{0}$ samples $f (c, t_{0}) = f (c, t_{0} - age + A)$ across cohorts of varying age. The age gradient $\partial f / \partial age$ in the cross-section therefore equals the time gradient $\partial f / \partial t$ in the community's history. Real-time replications test the stability assumption by resampling the same community at $t_{1} > t_{0}$ ; if the real-time change matches the apparent-time gradient, the assumption is validated for that variable. $□$

Proposition (Curvilinear principle as an emergent property of network structure). If the spread of an innovation through a community follows a contact-network model in which central nodes (interior social groups) have higher degree than peripheral nodes (extreme groups), then the interior groups reach the critical threshold for adoption before the extremes, and lead the change in apparent time.

Proof sketch. Model the community as a graph $G$ with vertices as speakers and edges as social contact. An innovation spreads by a threshold contagion process: a vertex adopts when the fraction of its neighbours that have adopted exceeds $θ$ . Interior vertices (upper-working, lower-middle) have higher degree $d$ and so reach threshold sooner for a given spreading rate; extreme vertices (highest and lowest) have lower degree and reach threshold later. The age gradient in apparent time therefore shows the interior groups leading, recovering the curvilinear principle. The model is qualitative but the prediction matches Labov's empirical generalisation across many studies. The interpretation is that the curvilinear principle is not a sociological primitive but an emergent property of network structure in communities where interior groups are the most connected. $□$

Connections Master

Sociolinguistics — variation, identity, and language in society 51.07.01. The variationist method is the quantitative engine that turned the broader concerns of sociolinguistics (language and identity, language and power) into testable empirical claims. The variable rule and the apparent-time construct are the methodological foundations of the field's modern form.
Historical-comparative linguistics foundations 51.06.01. The variationist method gives the comparative method a real-time counterpart. Where the comparative method reconstructs a finished change from its end state, the variationist method observes change in progress and tests the regularity hypothesis against data the comparative method can only assume. The two methods are complementary wings of historical linguistics.
Social stratification: class, race, gender, intersectionality 30.04.01. The Labovian stratification by class, style, and gender is the linguistic analogue of the stratification analyses in sociology. The community-of-practice refinement (Eckert) parallels the intersectional move in sociology, situating macro-categories within locally meaningful identities.

Historical & philosophical context Master

William Labov's 1963 study of Martha's Vineyard ^[Labov1963] in Word is the founding document of variationist sociolinguistics. Labov showed that the centralisation of the diphthongs /ay/ and /aw/ on the island correlated positively with orientation toward the island (resistance to the summer-tourist influx) and with age cohort, demonstrating that a sound change could be observed in progress and that its social meaning could be quantified. The 1966 Social Stratification of English in New York City ^[Labov1966] consolidated the methodology: the department-store stratification, the Labovian interview, the apparent-time construct, and the indicator-marker-stereotype distinction. Weinreich, Labov, and Herzog's 1968 paper ^{[WeinreichLabovHerzog1968]} articulated the five problems of language change and set the research programme.

The methodological refinement to mixed-effects variable rules is due to Sankoff (the original VARBRUL programme, 1978 onward) and to modern Bayesian and frequentist implementations (Johnson, Drager, Tagliamonte ^{[Tagliamonte2006]}). Penelope Eckert's community-of-practice refinement ^[Eckert2000] brought ethnographic depth to the quantitative framework. The deeper lineage runs through dialect geography (Gilliéron's Atlas linguistique de la France 1902-1910), whose limitations Labov's method addressed: dialect geography drew lines on rural maps, ignored cities, and could not detect change in progress. The variationist method reoriented the field toward the urban speech community and toward quantitative measurement of structured heterogeneity.

Bibliography Master

@article{Labov1963,
  author = {Labov, W.},
  title = {The social motivation of a sound change},
  journal = {Word},
  volume = {19},
  pages = {273--309},
  year = {1963},
}

@book{Labov1966,
  author = {Labov, W.},
  title = {The Social Stratification of English in New York City},
  publisher = {Center for Applied Linguistics},
  address = {Washington, DC},
  year = {1966},
}

@book{Labov1972,
  author = {Labov, W.},
  title = {Sociolinguistic Patterns},
  publisher = {University of Pennsylvania Press},
  address = {Philadelphia},
  year = {1972},
}

@incollection{WeinreichLabovHerzog1968,
  author = {Weinreich, U. and Labov, W. and Herzog, M. I.},
  title = {Empirical foundations for a theory of language change},
  booktitle = {Directions for Historical Linguistics},
  editor = {Lehmann, W. P. and Malkiel, Y.},
  publisher = {University of Texas Press},
  address = {Austin},
  year = {1968},
}

@book{Tagliamonte2006,
  author = {Tagliamonte, S. A.},
  title = {Analysing Sociolinguistic Variation},
  publisher = {Cambridge University Press},
  address = {Cambridge},
  year = {2006},
}

@book{Eckert2000,
  author = {Eckert, P.},
  title = {Linguistic Variation as Social Practice: The Linguistic Construction of Identity in Belten High},
  publisher = {Blackwell},
  address = {Oxford},
  year = {2000},
}

@book{LabovAshBoberg2006,
  author = {Labov, W. and Ash, S. and Boberg, C.},
  title = {The Atlas of North American English: Phonetics, Phonology and Sound Change},
  publisher = {Mouton de Gruyter},
  address = {Berlin},
  year = {2006},
}

Prerequisites

51.07.01
51.06.01

Tier anchors

beginner: Labov — Sociolinguistic Patterns (Penn 1972) Ch. 1-4; Wardhaugh-Fuller — An Introduction to Sociolinguistics (Wiley 2014) Ch. 6-7
intermediate: Tagliamonte — Analysing Sociolinguistic Variation (Cambridge 2006); Sankoff — Sociolinguistics: An International Handbook (de Gruyter 2017) Ch. on variationist method
master: Labov 1963 The social motivation of a sound change (Word 19); Labov 1966 The Social Stratification of English in New York City (CUPT); Labov 1972 Sociolinguistic Patterns (Penn); Weinreich-Labov-Herzog 1968 Empirical foundations for a theory of language change

References

Labov — The social motivation of a sound change · Word 19, 273-309 (1963); originator of the variationist method via the Martha's Vineyard centralisation study
Labov — The Social Stratification of English in New York City · Center for Applied Linguistics 1966; originator of the department-store methodology and the apparent-time construct
Labov — Sociolinguistic Patterns · University of Pennsylvania Press 1972; the consolidated statement of the Labovian paradigm
Weinreich, Labov, and Herzog — Empirical foundations for a theory of language change · in Lehmann and Malkiel (eds) Directions for Historical Linguistics, Texas 1968; originators of the five problems of language change (constraints, transition, embedding, evaluation, actuation)
Tagliamonte — Analysing Sociolinguistic Variation · Cambridge University Press 2006; the canonical methodological manual for variationist analysis including VARBRUL and mixed-effects modelling
Eckert — Linguistic Variation as Social Practice · Blackwell 2000; the Detroit-area study reframing variation in terms of Communities of Practice

Estimated time

beginner: 15m
intermediate: 40m
master: 80m