Digital Humanities Workbench


Home page > Data analysis > Conversation

Conversation analysis

Conversation or interaction analysis is used to analyse face-to-face conversations in formal or informal settings. This means that all types of conversations can be examined, from those in organizations (companies, courtrooms, police stations, hospitals) to those around the coffee table. What all these conversations have in common is that they are all 'natural' conversations: they would also have taken place if they hadn't been examined. Conversation analysis is almost never used for conversations recorded in experimental settings, but is often used for the analysis of CMC (computer-mediated communication), such as in research on interaction on a forum and in online therapy, or on chat conversations. These are all conversations/texts where there is a lot of interaction between people.

Before the conversations can be analysed, they are first recorded (audio, video) and transcribed (verbatim, including stress, intonation, pauses).

Research using conversation analysis generally aims to detect patterns in the way interlocutors structure their conversations themselves. This includes patterns in how people start or end (phone) conversations, how interviewers ask questions or interviewees try to avoid answers; how judges question suspects or defendants try to exonerate themselves; how managers lead meetings and introduce or close topics; or how job interviews are held.

Conversation analysis is usually qualitative research: it is usually not about how often something occurs, but how it occurs, and what that means for the people involved in the conversation. You usually start with a case study (a single comprehensive and detailed analysis of a particular conversation, fragment or phenomenon), before continuing with a corpus of multiple cases. Here, you compile a corpus of related cases in which you try to detect a broader pattern in the phenomenon you are researching.

You can look at many different aspects of interaction in your research: how people alternate turns (turn-taking organization), for instance, how people coordinate their actions with each other (for example, question-response), or how people choose to formulate what they want to say. Whatever the object of your research is, you are always trying to gain more insight into how people construct, influence and change their relationship to the other interlocutors during the conversation, and how they use language to do so.

Tasks

Collecting data The quality of your research stands or falls by the quality of your data: it is important to use good recording equipment when recording conversations. For example, it is a good idea to use multiple microphones or cameras for larger groups, depending on whether you need to observe eye contact or gestures. For information about borrowing or using recording equipment, please contact the faculty's ICT Support Unit.
It is also very important that you ask the people you are recording for permission to record and to use the recordings for research or teaching purposes. For this purpose you can use existing 'consent forms' (consult your teacher).
Transcribing data
After recording, you make general or detailed transcriptions of the conversations, depending on your research questions. At the VU, we use the program Transana for transcription. There are conventions on how to transcribe conversations properly. There are also other transcription programs. You need an account in order to use Transana (consult your teacher).
Analysing data
After recording and transcription, you can start analysing your data. This can involve one or more case studies or a corpus. In most cases, you will use qualitative analysis on various aspects of interaction: turn-taking, sequence analysis and formulation, for example, in order to determine how people construct their relationship to their interlocutors in conversations.

Tools

In our faculty, Transana is the most important tool for conversation analysis.

Other topics in this section: Introduction   Textual data   Visual data   Structured data