Changing Perspectives on Data: Interviews as Situated Speech

Much of the work done today in sociolinguistics and dialectology is based on tape recordings of spoken language. Other branches of linguistics which rely on linguists' intuitions about how they themselves use language, on written documents, or on responses to questionnaires about language use. Although some researchers have come up with innovative ways of collecting recordings of "natural" language use, much of the data in the field come from interviews. Appropriately, the social identity of the interviewer, relative to the social group of the research participant, has often been incorporated into the analysis of the interview data. However, the speech and discourse patterns of interviewers have only recently been included in the analysis of interview data. It is the perspective of interviews as situated speech that interests me and that I see as a growing factor in research in the field.

We often hear people say, "So-and-so has some really good data!" We all know what they mean: the data illustrate interesting features of the variety being studied and/or include colorful characters who have interesting insights about their own language and community. Similarly, many of us have had the experience of coming away from an interview feeling that we have just collected some really "good"--or really "bad"--data; our data are "good" when our research participants speak to us as we imagine they speak when we're not around.

Clearly, many researchers in the field are very good at collecting casual, unselfconscious speech; tapes and transcripts from data presented as examples often indicate that the interviewer has developed great rapport with the research participants. In my own research, people have told me personal stories, used their native dialects with me, and invited me to come back next week to talk some more. I used to take this to mean that I was getting data that was the "real stuff," the in-group speech. But one day, while transcribing an interview and listening to myself backchannel politely and never once interrupt, it occurred to me that since I don't talk the way I do with my friends when I'm doing an interview, why should I assume that the research participants do? That doesn't mean that the data aren't "good," but it does mean that the interviews were just that, interviews, with each participant playing an assigned role. For example, when I was collecting data to study Pennsylvania German, I considered an interview successful if the participant did not switch back to English frequently but answered my questions and told stories in Pennsylvania German. However, in my observations of interactions among community members, it became obvious that switching to English was often the norm for in-group conversations. Thus, while my data suit my purposes, they do not necessarily reflect the everyday speech of the research participants.

Over the years, what started out in my mind as two different research questions to ask when looking at data (what are the features of this variety and what is interesting about this discourse?) have become one question: How does the nature of the discourse relate to the features I find in this variety? For me, the interview has changed from method to research topic. As a speech event, the sociolinguistic interview has been underanalyzed, and one of the most exciting developments in the field of late has been the greater attention paid to interview data as situated speech.

What do we have to gain from looking at data in this way? Researchers are already conscious of the impact of their social identity (e.g., their ethnicity, gender, native dialect, and familiarity with the speakers) on their research participants; expanding this awareness to include an assessment of the nature of, and norms for, the interaction can only deepen our understanding of how the research participants speak. A more in-depth examination of interviewer behavior in the data can...