Clean language interviewing


Clean language interviewing, sometimes shortened to clean interviewing, aims to maximise the reliability that information collected during an interview derives from the interviewee. CLI seeks to address some of the "threats to validity and reliability" that can occur during an interview and to increase the "trustworthiness" of the data collected. It does this by employing a technique that minimises the unintended introduction of interviewer content, assumption, leading question structure, presupposition, framing, priming, tacit metaphor and nonverbal aspects such as paralanguage and gesture that may compromise the authenticity of the data collected.
At the same time clean language interviewing seeks to minimise common interviewee biases, such as the consistency effect, acquiescence bias and the friendliness effect which may mean an interviewee looks for cues from the interviewer about how to answer.
Furthermore, a systematic application of a 'cleanness rating' protocol provides a quantitive measure of adherence to interview guidelines and by extension the trustworthiness of the data collected.
CLI can be considered a phenomenologically-based interview method, similar in intent to neuro- and micro-phenomenology, psycho-phenomenology, phenomenography, and Interpersonal Process Recall. Clean interviewing can be seen as a method of operationalising the phenomenological aim of bracketing. CLI is also an integral part of a new qualitative research methodology, Modelling Shared Reality which suggests that by paying careful attention to the language they use, qualitative researchers can reduce undesired influence and unintended bias during all stages of research—design, data gathering, analysis and reporting.

History

Clean interviewing derived from clean language, the language model of therapeutic processes devised by David Grove which has since found application in education, business, organisational change, health and academic research.
Grove devised the principles of clean language in the 1980s and continued to develop a specific question set until his death in 2008. Although Owen first suggested applying clean language to phenomenological research in 1996, it was not until 2010 that the first systematic research into the veracity of the method was conducted and published in the British Journal of Management.

Features

The features of clean interviewing include: the specificity of the technique; minimising unintended influence; data collection from the perspective of the interviewee; its applicability to in-depth interviews; elicitation of naturally occurring autogenic metaphor; investigating tacit knowledge; evaluation research; modelling mental models; and the verifiability of the adherence to the method.
CLI seeks to address the issues raised by research that demonstrates responses can materially be affected by: question construction; framing; changing a single word; introduced metaphors; presupposition; and the nonverbal behaviour of the interviewer such as paralanguage and gesture. Furthermore, in these studies, not only did the subjects show little or no awareness of being subtly and systematically influenced, many reported their answers with a high degree of confidence.
The amount of training required to become a proficient interviewer is commonly underestimated and because of the rigour of the CLI technique, training and practice are required to become a competent clean interviewer.

Cleanness rating

Roulston maintains "researchers must demonstrate the quality of their work in ways that are commensurate with their assumptions about their use of interviews" yet systematic methods for assessing the reliability and validity of the data collected during an interview are noticeably absent from published research.
A "cleanness rating" can be used to measure the proportion of 'clean' to 'leading' interviewer interventions and hence the authenticity or trustworthiness of the data collected. An independent examination of interview transcripts allocates one of four categories to every question or statement made by the interviewer. The tabulated results are used to arrive at a summary assessment for each interview. Interviewers experienced in CLI can consistently achieve over 85% of 'clean' interventions. Whereas as many as two-thirds of the questions asked by an untrained interviewer may be 'leading'.
An inter-rater reliability analysis of 19 interviews from five research studies demonstrated substantial agreement among raters with an average intraclass correlation coefficient of 0.72.

Critique

The assumption that interviewers can minimise their influence on the collection of data has been questioned for several decades and the social constructivist school claims all knowledge is co-constructed. Practitioners of CLI do not claim it does not influence nor that it is "non-directive". They acknowledge that the interviewer's role is to set the overall frames and direct the general flow of the interview and they emphasise that all questions, including clean questions, set frames, contain presuppositions and invite the interviewee to attend to certain aspects of his or her experience. Clean questions aim to stay within the lexicon and inherent logic expressed by the interviewee and leave them maximally free to answer from within his or her frames of reference. Tosey maintains that, "accounts are co-constructed through the process of selecting and asking questions. At the same time, Clean Language aims to minimise the co-construction of the content."
As yet there has been no comprehensive test of the premise that interviews with a high cleanness rating result in higher quality data. Rather, an interview with a low cleaness rating may raise questions about compliance with research protocols and doubts about the authorship of the data collected.
Other forms of interviewing would be more appropriate where the researcher intentionally aims for the 'co-construction of meaning' or to challenge and change the understandings of participants, as in 'transformative interviews'.
Some critics have maintained that the metaphor 'clean' is pejorative to other methods. Grove settled on the metaphor after discovering how others 'contaminate' another person's perceptions with introduced content and presupposition – including Carl Rogers the originator of "non-directive therapy".

Applications

While clean interviewing is ideally suited to research, it has also be used by police interviewers, critical incident investigators and in commercial projects such as requirements gathering, market research and organisational change. Examples of published research include:
In addition clean interviewing has been used in a range of PhD research topics: