Skip to content

Compare methods

Compare source explanations, strengths, limitations and references for up to four methods.

7 results
Criterion Matched-Guise Technique Variationist Sociolinguistics Discourse Completion Task
Overview
  • The matched-guise technique is an indirect experimental method for measuring attitudes toward languages, dialects, and accents. Developed by Wallace Lambert and colleagues in 1960, it has the same bilingual or bidialectal speaker record the same passage in two or more language varieties ('guises'); listeners, believing they are hearing different speakers, rate each recording on personality and status traits. Because the voice, content, and delivery are held constant, any differences in the ratings can be attributed to listeners' attitudes toward the variety itself.
  • Variationist sociolinguistics is the quantitative study of how linguistic variation is structured by social and linguistic factors. Pioneered by William Labov in the 1960s and 1970s, it treats alternative ways of saying the same thing — the 'linguistic variable' — as systematically conditioned by speaker characteristics (class, age, sex, ethnicity), stylistic context, and the surrounding linguistic environment, and it uses statistical modeling of natural speech to reveal the orderly heterogeneity beneath apparent randomness.
  • The discourse completion task (DCT) is an elicitation instrument widely used in pragmatics to gather data on how people perform speech acts such as requests, apologies, refusals, and compliments. Respondents read short descriptions of situations and write (or say) what they would utter in each, allowing researchers to collect comparable speech-act data across many speakers, languages, and cultures under controlled conditions. It was popularized by the Cross-Cultural Speech Act Realization Project (CCSARP) of Blum-Kulka, House, and Kasper in 1989.
Use and interpretation
  • Use the matched-guise technique when you want to measure private or socially sensitive attitudes toward language varieties that respondents would not report honestly if asked directly. It is the classic tool for studying accent prejudice, the prestige of standard versus non-standard varieties, and reactions to bilingual or minority languages. It is less appropriate when the research question concerns conscious, reportable beliefs (where direct questionnaires suffice), when suitable bidialectal speakers cannot be found, or when the deception or the artificiality of decontextualized speech threatens validity or ethics.
  • Use variationist sociolinguistics when you want to know how the choice between equivalent linguistic forms is conditioned by social and linguistic factors, and especially to detect and explain language change in progress. It suits phonological, morphosyntactic, and discourse variables that recur often enough in speech to count. It is less appropriate for rare constructions that yield too few tokens, for meaning differences that violate the 'same thing' assumption, or for purely qualitative questions about identity where ethnographic or perceptual methods (such as matched guise) are better suited.
  • Use a DCT when you need comparable speech-act data from many speakers under controlled conditions — for cross-cultural pragmatics, interlanguage and second-language pragmatics, or studies that systematically vary power, distance, and imposition. It excels when natural occurrences would be too rare or uncontrolled to compare. It is less appropriate when the research question concerns authentic interactional dynamics, turn-taking, or negotiation, where role plays, conversation analysis, or naturalistic recordings are preferable, since DCTs capture offline, idealized production rather than real-time interaction.
Strengths
  • Accesses covert, often socially undesirable attitudes that direct questioning fails to elicit.
  • Holds voice, content, and delivery constant, so rating differences can be attributed cleanly to the language variety.
  • Produces quantitative, comparable attitude scores amenable to statistical analysis and replication.
  • Established the durable competence-versus-solidarity framework for understanding language attitudes.
  • Reveals the systematic social and linguistic conditioning behind variation that earlier approaches dismissed as free or random.
  • Provides a powerful method for detecting and tracking language change in progress through apparent-time and real-time comparison.
  • Grounds claims in quantified analysis of natural speech, combining rigorous statistics with authentic data.
  • Has a mature methodological apparatus — the sociolinguistic interview, factor coding, and variable-rule/regression analysis — that supports replication and comparison across communities.
  • Yields large amounts of comparable data quickly by holding situations constant across many respondents.
  • Allows systematic control and manipulation of social variables (power, distance, imposition) that shape speech acts.
  • Enables clean cross-cultural and cross-linguistic comparison of pragmatic strategies.
  • Produces data in a form well suited to established coding schemes and statistical comparison, supporting replication.
Limitations
  • Relies on deception, raising ethical concerns and requiring careful debriefing.
  • Finding speakers who are genuinely authentic and equivalent across all guises is difficult; imperfect guises confound the design.
  • Decontextualized reading of a neutral passage is artificial and may not reflect attitudes in real interaction.
  • Listeners may still detect the matched design or react to the reading style rather than the variety, threatening validity.
  • The 'same meaning' criterion that defines a variable is straightforward for phonology but contested for syntax and discourse, where variants may carry subtle differences.
  • Collecting natural vernacular speech is labor-intensive and confronts the observer's paradox, limiting sample sizes.
  • Predefined social categories (class, sex) can essentialize speakers and obscure how identity is actively constructed, a critique from third-wave sociolinguistics.
  • Rare variables generate too few tokens for stable estimates, and coding decisions can materially affect results.
  • Elicits what respondents think they would say, not what they actually say, so responses may diverge from authentic use.
  • Written DCTs lose interactional features such as turn-taking, repair, and negotiation present in real conversation.
  • Responses are sensitive to scenario wording and to how much context is supplied, which can shape the output.
  • The offline, reflective nature of the task may favor more idealized, normative, or lengthy responses than spontaneous speech.
Common pitfalls
  • Using a speaker whose two guises differ in pitch, pace, or warmth, so that ratings reflect those features rather than the variety.
  • Placing matched guises too close together in the sequence, allowing listeners to recognize the same voice and defeat the deception.
  • Generalizing from a single speaker's guises as if they represented the variety, when speaker idiosyncrasies confound the result.
  • Neglecting the artificiality of read-aloud neutral text, which may evoke different attitudes than spontaneous, contextualized speech.
  • Defining the envelope of variation loosely, so that tokens that could not have varied are counted, distorting the rates.
  • Treating apparent-time (age) differences as evidence of change without considering age-grading, where speakers shift across the lifespan.
  • Ignoring speaker as a random effect, which inflates significance when some individuals contribute many tokens.
  • Imposing social categories that do not reflect the community's own meaningful distinctions, missing the locally relevant pattern.
  • Treating DCT responses as equivalent to naturally occurring speech rather than as elicited, metapragmatic production.
  • Writing scenarios that are culturally unequal across the groups compared, so the prompt itself biases the comparison.
  • Failing to balance or document the social variables (power, distance, imposition), making strategy differences uninterpretable.
  • Using inconsistent or unreliable coding of head acts and modifications, undermining cross-group comparability.
Introduced by
  • Wallace Lambert and colleagues
  • William Labov
  • Shoshana Blum-Kulka, Juliane House & Gabriele Kasper (CCSARP project)
Year
  • 1960
  • 1972
  • 1989