Description
Micro-analyses naturally occurring talk, including talk with machines, for turn-taking, sequence structure and repair, revealing the implicit rules users bring to interaction.
Conversation Analysis studies the sequential organisation of real talk: how turns are allocated, how adjacency pairs (question and answer, offer and acceptance) structure exchanges, how trouble is repaired, and how dispreferred responses are delayed and mitigated. It works from recordings transcribed in Jefferson notation, capturing pauses to a tenth of a second, overlaps and intonation. Because CA treats interaction as orderly at the finest level, it is unusually powerful for conversational interfaces: it shows exactly where a voice assistant or chatbot breaks users' expectations of how the next turn should go. Rigour comes from using naturally occurring rather than elicited data, detailed transcription, and grounding every claim in participants' own displayed orientations in the next turn.
Attributes
- Discipline
- Linguistics, Semiotics & Discourse
- Loss type
- Siloed
- Stage
- Discovery, Evaluation
- Data
- Qualitative
- Effort
- Weeks
Steps
- Collect naturally occurring interaction data: CA requires authentic data — actual user-chatbot logs, voice assistant recordings, or customer service chat transcripts.
- Jefferson transcription notation: Capture overlapping speech ([ ]), latency ((0.5)), elongated sounds (::), cut-off words (-). The transcript is an analytic document.
- Identify sequential structure: What adjacency pairs exist (question-answer, greeting-greeting)? Where do sequences complete normally and where do they break? Breakdowns are your design data.
- Repair sequences: When non-understanding is signalled (explicit question, repetition, silence), how is repair initiated? Failure to achieve repair is the UX problem.
- Preferred vs. dispreferred responses: Dispreferred responses (disagreement, rejection) are typically delayed, prefaced, or mitigated. CA reveals the social logic of how users phrase requests.
Sources
- Handbook of Discourse Analysis
- Qualitative Research from Start to Finish (Yin)
- Moore, R. J. & Arar, R. (2019) Conversational UX Design
Related talks
Also known as
- CA
- Sequential analysis of talk
Categories
- Content
- Linguistics, Semiotics & Discourse