From text to talk : Answers and their uptake in standardised survey interviews

Abstract

The standardised survey interview is a highly structured form of interaction. Interviewers ask questions from a script, and respondents select one of the pre-formulated answers. However, interviewers and respondents do more than just their pre-allocated tasks. This study approaches survey interviews from an original perspective, entering the stage after the scripted question has been released into the interactional space. Interviewers and respondents switch roles, the interviewers becoming recipients of the respondents’ talk. Questionnaires cannot provide for all the contingencies that arise in the interaction following the scripted question. Using Conversation Analysis and phonetic analysis of the interaction, the ways in which participants deal with these contingencies as they arise is explored in this dissertation. Interviewers employ response tokens to perform the task of answer recipient. In Dutch survey interviews the response token 'ja' ('yes') is used in several sequential positions, performing different actions. Analyses show that the token can be employed as a continuer token, as a receipt token, or as a mere acknowledgement of the prior talk. Moreover, through phonetic analyses it was found that the same token in a similar sequential position can perform different actions when it is realised with a different prosody. Respondents perform meaningful actions when they do not simply select one of the answer categories. Interviewees perform a range of actions with these expanded and non-conforming response turns. Minimal expansions are used to indicate imprecision, or to indicate the connectedness of the present answer with a previous answer. More extensive expansios are employed to resist the dichotomy that is set up by a yes/no-question or they can be used to resist presumptions embedded in the question. Responses that do not contain any of the answer categories ('non-conforming' responses) can be employed to perform similar actions to expanded answers. An interesting finding of this study is that the design of the non-conforming response influences whether the interviewer accepts that response or decides to probe for an answer in terms of one of the answer categories (or a 'type-conforming' answer). Non-conforming responses that start with hesitations, followed by 'nou' ('well'), and that contain an explicit or implicit contrast are treated as unacceptable, while responses that do not contain these features of a dispreferred turn are treated as acceptable. Finally, interviewers need to deal with these responses, always balancing the rules of standardisation with the goal of obtaining recordable answers. Their uptake of non-conforming responses ranges from simple acceptance to repeating the entire question. Interviewers may also reformulate the respondent's turn into one of the answer categories, and provide prompts when confronted with non-response. While only the exact repeat conforms to the rules of standardisation, phonetic analyses show that interviewers adjust the delivery of the redoing to the interviewee's response. Thus, even the most standardised way of dealing with the respondent's non-conforming turns, requires the interviewer to produce an understanding of those turns, and discover how it relates to the question or answer categories to which it is responsive.