Within the great variety of qualitative methods, some emphasise language and language use. Any professional–client interaction relies on language as a major communication device. Conversation(al) analysis (CA) is a type of discourse analysis (DA) that examines the use of language and asks the question of how everyday conversation works; in its basic form it is the study of talk in everyday interaction (Hutchby and Wooffitt, 2008). This type of inquiry focuses on ordinary conversations and on the way in which talk is organised and ordered in speech exchanges. While researchers primarily examine speech patterns, they also analyse non-verbal behaviour in interaction such as mime, gesture and other body language. As Nofsinger (1991:2) explains: ‘If we are to understand interpersonal communication, we need to learn how this is accomplished so successfully’.

The origins of conversation analysis

Harold Garfinkel, Harvey Sacks, Emmanuel Schegloff and others initially developed CA in the 1960s and 1970s in the United States within the ethnomethodological movement. While other types of DA have their roots in the field of linguistics, CA originates in ethnomethodology, a specialist direction of sociology and phenomenology. Ethnomethodology focuses in particular on the world of social practices, interactions and rules (see Turner, 1974). Garfinkel attempted to uncover the ways in which members of society construct social reality. Ethnomethodologists focus on the ‘practical accomplishments’ of members of society, seeking to demonstrate that these make sense of their actions on the basis of ‘tacit knowledge’, their shared understanding of the rules of interaction.

The use of conversation analysis

CA focuses on what individuals say in their everyday talk, but also on what they do (Nofsinger, 1991). Through conversation, movement and gesture, we learn of people’s intentions and ideas. The sequencing and turn taking in conversations demonstrate the meaning individuals give to situations and show how they inhabit a shared world. Body movements too, are the focus of analysis. Conversation analysts do not use interviewing to collect data but analyse ordinary talk, ‘naturally occurring’ conversations. Most sections of talk analysed are relatively small, and the analysis is detailed. According to Heritage (1988: 130), CA makes the assumptions that talk is structurally organised, and each turn of talk is influenced by the context of what has gone on before and establishes a context towards which the next turn will be oriented. There are two other fundamental tenets of CA according to Heritage: sequential organisation and empirical grounding of analysis. Talk happens in organised patterns; the action of the member who takes part in the conversation is dependent on and makes reference to the context, and researchers should avoid generalities and premature theory building (Silverman, 2006).

CA is more often used in sociological or education studies than in nursing or other disciplines within the healthcare arena. Indeed, Jones (2003) found few in nursing journals. We think, however, that it can contribute a valuable research approach in nursing and healthcare and lead to changes in the interaction between health professionals and patients. Researchers generally audio- or videotape these interactions and transcribe the conversations in a particular way (see transcription techniques in detail, in Button and Lee (1987)) largely developed by Gail Jefferson.

There are examples in health research of doctor–patient or nurse–patient interaction, particularly in consultations which show how talk is generated and organised by the participants and follows an orderly process in which a turn-taking system exists (Sharrock and Anderson, 1987; Bergstrom et al., 1992). All the recent examples we discovered were in the area of medical consultations (see for instance Campion and Langdon, 2004). These interactions are usually taped and the tapes show what actually takes place in a setting.

Jones (2003) investigated the communication and interaction between nurses and patients in healthcare consultations and found that CA was a useful way to undertake this. He regrets that this way of studying talk is not often used in the field. Sequences of talk can be studied, and he suggests that they illuminate processes such as treatment, advice and assessment. The disadvantages of CA he sees as the length of time that is needed for this research approach, and the potential lack of context as the focus on direct interaction and communication might become isolated from the social context.

The analysis of CA includes the discovery of regularities in speech or body movement, the search for deviant cases and the integration with other findings without over-generalisation (Heritage, 1988). One of the disadvantages is the way in which conversation analysts emphasise the formal characteristics of interaction at the expense of content, however, much can be discerned from the way the communication and interaction proceeds. ten Have (2007) describes ways of analysing CA data, and researchers might find his book useful.

CA is difficult, highly complex and very detailed. Researchers may not find it easy, and we do not recommend it to novice researchers.

The critical incident technique (CIT) is a procedure designed to solve problems in clinical practice or educational settings. In the past, it was not often used in the health arena, but it has been applied in this field, and although neglected for a number of years, it has become more often used in the recent decade. It is not a specific method or methodology but a means of developing questions, whilst focusing on people’s behaviour in critical situations in order to solve problems in task performance. A critical incident is an observed event that is perceived as particularly important or critical. Researchers examine those events that are significant for a particular process. They collect examples of critical incidents in the situation under study by observation of behaviour and by asking participants to give an account of the way in which they deal in critical situations or times of crisis.

Flanagan (1954) who first developed this technique, suggests that in this approach, an incident has to occur in a situation with definite consequences and effects. CIT was initially developed as a result of the Aviation Psychology Program in the United States to collect information from pilots about their behaviour when flying a mission. In particular, psychologists asked for reports about critical incidents that helped or hindered the successful outcome of the mission. Through analysis of these reports, a list of components for successful performance was generated from the data. Flanagan refined the procedure for industrial psychology to assess the outcomes of task performance, and it was used also in other fields such as personnel selection. Although the method was neglected after the 1950s, it can be a useful, effective and qualitative approach to studying critical events in order to improve task performance and is thus very useful for the health professions. Flanagan (1954: 335) states that the technique is ‘a procedure for gathering certain important facts concerning behaviour in defined situations’.

Critical incidents might occur in clinical practice, make visible problems in care and are noticed by health professionals who then decide to examine these, as well as the reasons for their occurrence. They then develop the specific plan and aim of their research and start collecting and analysing data. It can be seen that this type of inquiry follows the traditional path of data collection and analysis in qualitative research.

Researchers examine those events that are significant for a particular process. They collect examples of critical incidents in the situation under study, and participants give an account of the way in which they act in critical situations or times of crisis. Direct observation is the most important part of the data collection. Generally the researchers ask about the critical event and gain a perspective about effective and ineffective behaviour in specific decisive and important situations.

The process of critical incident technique

Schluter et al. (2007) in a short guide to CIT (p.108) describe the five steps which Flanagan suggests:

The aim of the technique is to obtain information about each specific incident. This will include choosing the type of events on which researchers wish to focus, generally critical events or incidents in care or educational settings. The second stage involves selecting a purposive sample of incidents and people from which to collect data. The sample size depends on the number of critical incidents, not on the number of people interviewed or observed (Kemppainen, 2000). Initially, health researchers find out about critical incidents through incidental and casual observation, but when they decide to do the research, they observe and ask questions more purposefully. The data are analysed in a similar way as in other approaches. There is, however, a slight difference: Researchers choose a more defined frame of reference in this type of research as they wish to focus on particular events. It is also important that the terminology used is clear and appropriate for CIT; indeed, Butterfield et al. (2005) deplore the lack of consistency in studies which use this technique.

The goal of the researcher is to investigate a recurring problem and to find a solution to it. To examine the critical incidents, the health researcher has to be familiar with the setting and the nursing or midwifery tasks that are performed. Kemppainen (2000) advises that the responses of the participants to the researcher’s questions should be specific and accurate, and not vague or unclear. We feel that this way of researching is under-explored and could be specifically useful in the healthcare field.

Cheek (2004) suggests that CA and DA are complex concepts and that confusion exists about this area of research. DA cannot be easily identified, as people use it in different ways; it is more a framework or holistic theoretical stance. Discourse in general is applied to talk and text, such as conversations, interviews or documents. Traynor (2006) re-emphasises that it is an analysis of naturally occurring talk, one of the most important sources of data, although talk is not the only discourse that might be analysed.

DA in psychology is an analysis of text and language drawing on ‘accounts’ of experiences and thoughts that participants present. This type of DA has been carried out mainly by psychologists. Accounts consist of forms of ordinary talk and reasoning of people, as well as other sources of text, such as historical documents, diaries, letters or reports and even images such as photographs, drawings or paintings. DA is not a method but a specific approach to the social world and research (Potter, 1996; Cheek, 2004). It focuses on the construction of talk and text in social action and interaction. In common with other types of qualitative inquiry, discourse analysts initially use an inductionist approach by collecting and reviewing data before arriving at theories and general principles as do other qualitative researchers. The way people use language and text is taken for granted within a culture (Gill, 1996) and this shows that discourse is context-bound. DA as the structural analysis of discourse, is often used in media and communication research to analyse data. An example would be an analysis of the speech messages of politicians.

Language itself and reality are socially constructed. The vocabularies which individuals and groups use are located in interpretive ‘repertoires’ that are coherent and related sets of terms. Crowe (2005: 55) adds to this that discourses show how ‘social relations, identities, knowledge and power are constructed in spoken and written text.’

It is important to read the documents and transcripts carefully before interpreting them. The first step in the analysis is a close look at and detailed description of other, less language-based sources. The relevant documents are read and re-read until researchers have become familiar with the data, be they textual or visual. Immersion in the data, after all, is a trait of all qualitative research. Important issues and themes can then be highlighted. The analysis proceeds like other qualitative research: analysts code the data, look for relationships and search for patterns and regularities that generate tentative propositions. Through the process, they always take the context into account and generate analytical notes as in other forms of qualitative inquiry.

Also, like for other qualitative research, the findings from DA are not instantly generalisable; indeed researchers are not overly concerned with generalisability, because the analysis is based on language and text in a specific social context. There are a number of similarities between CA and DA: both CA and DA focus on language and text. While DA generally considers the broader context, CA emphasises turn-taking and explains the deeper sense of interaction in which people are engaged, particularly ‘naturally occurring’ talk, while discourse analysts look at the material more holistically, and they can also use records, newspaper articles or reports of meetings, etc.

Discourse analysts are interested in the ways through which social reality is constructed in interaction and action. DA is based on the belief that language (and presentation of images) does not just mirror the world of social members and cultures but also helps to construct it.