Conversation Analysis

Conversation analysis (CA) is a qualitative research method rooted in ethnomethodology and sociology that studies the structure, organization, and sequential patterns of naturally occurring talk in interaction. Rather than analyzing what people say about their experiences (as in interviews), CA examines how people actually communicate in real time, focusing on features such as turn taking, repair sequences, adjacency pairs (like question and answer), overlap, pauses, and preference organization. CA treats conversation as a social activity governed by systematic practices that participants orient to, produce, and hold each other accountable for.

Conversation analysis reveals the organized practices through which social actions are accomplished in talk. It has transformed our understanding of how medical consultations work, how news interviewers hold politicians accountable, how emergency calls are managed, and how everyday conversations maintain social relationships. Because CA examines real interactions rather than hypothetical or recalled ones, it provides ecologically valid evidence about how communication actually functions. This evidence has practical applications in training clinicians to communicate more effectively with patients, designing more intuitive voice interfaces, and improving institutional interactions in legal, educational, and therapeutic settings.

CA begins with the collection of naturally occurring interactions, typically audio or video recorded with participants' consent. The data are transcribed using specialized conventions (the Jefferson transcription system is most widely used) that capture not only words but also pauses (timed to tenths of a second), overlapping speech, intonation, volume, and other prosodic features. Analysts then examine the data repeatedly, identifying recurrent interactional phenomena. Analysis proceeds from unmotivated looking (approaching the data without a predetermined hypothesis) to detailed examination of specific sequences, such as how patients present concerns in medical consultations or how conversationalists manage topic transitions. Collections of similar phenomena across multiple interactions are assembled to identify systematic practices. CA findings are validated by demonstrating that participants themselves orient to the identified patterns, as shown by their responses and repairs.

Use conversation analysis when your research question concerns the organization of interaction: how actions are performed through talk, how sequences unfold, or how participants coordinate their contributions. CA is especially appropriate for studying institutional interactions (medical encounters, courtroom proceedings, helpline calls, classroom discussions), ordinary conversation, and technology mediated communication. It is also valuable for applied purposes such as evaluating communication training programs, identifying sources of miscommunication, or informing the design of conversational agents.

A common mistake is using interview data, which are researcher elicited rather than naturally occurring, as a stand in for recordings of the interaction under study. While CA techniques can illuminate how interviews work as interactions, they cannot tell you about participants' everyday conversational practices through interview data. Another error is producing transcripts that lack the level of detail CA requires; a verbatim transcript that omits pauses, overlaps, and intonation is insufficient for CA. Researchers sometimes treat CA as a method for theme identification, which conflates it with thematic analysis; CA focuses on the mechanics and sequential organization of interaction, not on content themes. Finally, making analytic claims that are not demonstrably oriented to by the participants themselves undermines CA's empirical grounding.

A team of health communication researchers video records 80 primary care consultations to study how physicians deliver bad news. Using conversation analysis, they identify a recurrent sequence: the physician produces a pre announcement ("I want to talk to you about your test results"), waits for the patient's acknowledgment, and then delivers the news in a stepwise manner that allows the patient to display anticipatory understanding. When patients do not provide the expected acknowledgments, physicians recycle the pre announcement or provide additional framing. This sequential pattern reveals a systematic practice for managing the emotional and informational aspects of bad news delivery, with implications for communication training.