Comparative Conversation Analysis
- Introduced by
- Harvey Sacks, Emanuel Schegloff, Gail Jefferson (CA foundation); comparative extension developed across the field from the 1980s onward
- Year
- 1974 (CA foundation); comparative applications from 1980s–1990s
- Type
- Qualitative micro-analytic research design
- Family
- Qualitative design / analysis
What does it do?
Comparative Conversation Analysis (comparative CA) applies the rigorous micro-analytic methods of Conversation Analysis across two or more contrasting interactional settings, languages, cultures, or participant groups.
It examines how the sequential organisation of talk — turn-taking, repair, adjacency pairs, and action formation — varies or remains stable across contexts, producing cross-contextual evidence about the architecture of human interaction.
The intuition behind the method
Ordinary CA studies one setting in depth to reveal how participants coordinate talk moment by moment. Comparative CA asks a further question: do the same interactional practices appear when we examine doctor–patient consultations in Japan versus Sweden, or service encounters in formal versus informal settings? By building collections of structurally similar instances drawn from different contexts and examining what changes and what stays constant, researchers can make claims about interaction that reach beyond a single cultural or institutional site. The comparison is always grounded in detailed sequential analysis of real recorded talk — not in questionnaire responses or observer ratings.
History
Conversation Analysis was founded by Harvey Sacks, Emanuel Schegloff, and Gail Jefferson in the 1960s–70s, with their landmark 1974 Language paper establishing the turn-taking system. From the 1980s onward, CA researchers working across multiple languages (notably in the cross-linguistic CA programmes led by Margret Selting, Elizabeth Couper-Kuhlen, and later the EMCA tradition) developed comparative methods to test whether CA findings from English held in other languages and settings. The publication of the Handbook of Conversation Analysis (Sidnell and Stivers, 2012) consolidated comparative and cross-linguistic CA as a recognised strand of the field, with dedicated chapters on cross-linguistic and cross-cultural comparison.
When to use it
Use comparative CA when you want to know whether a conversational practice operates the same way across different languages, cultures, or institutional settings, and when you have access to recordings of naturally occurring talk from at least two contrasting sites. It is appropriate for cross-cultural pragmatics, cross-institutional interaction research, and typological questions about the universality of interactional structures. Do not use it when you lack high-quality recordings and Jefferson transcripts from multiple sites — thin or paraphrased data cannot support the sequential claims comparison requires. It is also inappropriate when the research question concerns attitudes, beliefs, or experiences rather than the moment-by-moment organisation of talk.
Limitations and considerations
- Requires high-quality audio or video recordings and detailed Jefferson transcripts from multiple sites, which is resource-intensive.
- Comparative claims are limited to the specific corpora analysed; statistical generalization to populations is not the aim or the product.
- Building collections large enough to support cross-contextual comparison demands substantial time and often a team of transcribers.
- Analysts need command of CA conventions in each language/variety studied; even minor transcription differences can affect cross-site comparisons.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Grounds cross-contextual claims in actual recorded behaviour rather than in participant reports or analyst inference.
- Can distinguish universal interactional structures from context-specific or language-specific variations.
- The collection-based method makes the analytic evidence transparent and open to scrutiny by other analysts.
- Produces findings with genuine cross-cultural or cross-institutional reach without sacrificing micro-analytic rigour.
- Applicable to any form of talk-in-interaction — institutional, conversational, digital, or multilingual.
Applications
- Comparing turn-taking and repair practices in medical consultations across different national health systems or languages.
- Examining how service requests are formulated and granted in formal versus informal institutional encounters across cultures.
- Investigating whether sequence-closing procedures (e.g., topic termination, pre-closing moves) operate similarly in phone calls across different linguistic communities.
- Analysing how participants accomplish disagreement or agreement in multilingual or cross-cultural business meetings.
- Studying child–caregiver interaction sequences across cultures to identify universal versus language-specific acquisition environments.
- Comparing instruction sequences in classroom interaction across educational systems or pedagogical traditions.
How it works
How it works
The researcher defines two or more contrasting sites, languages, or participant groups whose interactional practices are to be compared. Sites may differ by institution (e.g., medical vs. legal settings), language (e.g., English vs. Finnish), culture, or participant role. The selection is theoretically motivated: the contrast should be capable of revealing whether a practice is universal, language-specific, or institution-specific.
Naturally occurring interaction at each site is audio- or video-recorded with participant consent. Recordings are transcribed using Jefferson notation, which captures turn boundaries, overlap, latency, pitch, and other prosodic features essential for sequential analysis. Transcription is itself an analytic act requiring repeated listening.
Before comparison begins, analysts examine instances within each corpus independently, applying standard CA procedures: identifying action sequences, mapping turn-design choices, and tracing how participants display their understanding of prior turns through their responses. This step establishes what the practices look like in each site on their own terms.
Instances of the target practice (e.g., repair initiations, requests, topic closings) are identified across all corpora and assembled into a collection. Each instance is documented with transcript and contextual metadata so that structural similarities and differences can be examined systematically.
The assembled collections are compared across sites. Analysts ask: Is the sequential slot the same? Are the turn designs structurally parallel? Do participants orient to the practice in the same way? Differences and invariances are noted precisely, with evidence drawn from the sequential unfolding of each instance rather than from analyst inference alone.
Comparative patterns — whether a practice is universal across all sites, modified at one site, or absent in some contexts — are synthesised into findings. Claims are qualified according to the scope of the corpora and are expressed in the conditional language appropriate to collection-based CA: the analysis describes what participants demonstrably do rather than asserting cognitive or cultural causes without sequential evidence.
Common pitfalls
- Comparing thin or paraphrased transcripts — if only one site has Jefferson-level transcription and the other has summary notes, the comparison is methodologically incoherent.
- Selecting sites for convenience rather than theoretical contrast — the comparison gains analytical leverage only if the sites differ in ways that are relevant to the target practice.
- Importing cultural or cognitive explanations prematurely — comparative CA describes sequential patterns; causal claims about why differences exist require additional evidence.
- Treating small collections as definitive — a handful of instances per site may illustrate a pattern but cannot establish its distribution; authors should report collection sizes and acknowledge scope.
Frequently asked questions
How does comparative CA differ from cross-cultural pragmatics?
Cross-cultural pragmatics often relies on elicited data (discourse completion tasks, questionnaires) and focuses on norms or politeness strategies as reported or rated by participants. Comparative CA works exclusively with recordings of naturally occurring interaction and analyses the sequential organisation of actual talk, turn by turn. It describes what participants demonstrably do rather than what they say they would do.
Do I need the same number of instances from each site?
Strict numerical balance is not required, but analysts should be transparent about collection sizes at each site. Marked asymmetries — for example, 40 instances from Site A and 3 from Site B — undermine the comparative claims. The goal is to have sufficient instances at each site to identify the structural pattern within that corpus before comparing.
Can I use comparative CA with online or text-based interaction?
Yes. CA has been extended to computer-mediated communication, and comparative CA can compare interaction across platforms (e.g., video calls vs. in-person meetings) or across online communities in different languages. The analytic logic — building collections of structurally comparable instances and comparing their sequential organisation — remains the same, though the notation may need to be adapted for text-based turn-taking.
Is comparative CA compatible with ethnographic fieldwork?
It is compatible and often enhanced by ethnographic observation. Field notes and contextual knowledge help the researcher understand what activities are underway in each setting and how participants categorise their own actions. However, the primary analytic object remains the recordings and transcripts; ethnographic context informs interpretation but does not substitute for sequential evidence.
How large does my corpus need to be?
There is no fixed minimum, but the corpus at each site should be large enough to find multiple instances of the target practice and to judge whether variants are systematic or idiosyncratic. In practice, comparative CA studies often draw on corpora of several hours of recording per site. Small pilot corpora are useful for refining the research question before larger data collection.
Related methods
Sources
Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of turn-taking for conversation. Language, 50(4), 696–735.
Open DOI referenceCopied.Sidnell, J., & Stivers, T. (Eds.). (2012). The Handbook of Conversation Analysis. Wiley-Blackwell.
ISBN: 978-1444330564Copied.