Methodological Issues in Sociolinguistics: A Comparison of Data Elicitation Methods

Ronald Geluykens

University of Oldenburg, Germany


This paper highlights the enormous methodological difficulties in pragmatics research, where the bulge of the data is still elicited exlusively through questionnaires and role plays. While there is no doubt that these methods can yield interesting and valid results, it seems important to complement them with naturally occurring speech (see also Beebe, 1996; Golato, 2003; Hartford & Bardovi-Harlig, 1992; Kasper, 2000; Yuan, 2001).

The paper deals with the realization of complaints in native British English and compares a variety of data collection methods. The corpus under scrutiny here consists of elicited complaints (production questionnaires; role play interactions) and naturally occurring complaints (telephone conversations; material from a fly-on-the-wall documentary).

The main research questions include:
- To what extent do elicited complaints differ from natural complaining behaviour?
- Is it possible to draw conclusions about speech behaviour from elicited data?
- Do role plays resemble naturally occurring speech in terms of turn-taking mechanisms and emotional involvement?
- Quantitative and qualitative analysis – how can pragmatics research benefit from a combination of different methodologies?

A comparison of the four different data gathering methods used here served to highlight the complexity of the speech act complaint. Its realization is dependent on a multitude of variables, such as the situational context, the severity of the offence, the role relationship of speaker and hearer, and the interlocutors’ temperaments. The elicitation of data allows the researcher to control these variables, which is especially useful for cross-cultural comparisons on a large scale.

Complaining usually involves a high degree of emotional involvement for the interlocutors. We found that this aspect was not represented realistically enough in the elicited material. Whereas test subjects seemed to be aware of the emotional component, their anger seemed artificial (or was absent) and the reactions of people who the complains were addressed to lacked the kind of involvement that was found in the naturally occurring complaints. However, the role plays were very similar in reference to turn-taking, with only one significant difference, a significantly lower occurrence of overlaps.

From our analysis we conclude that a combination of role plays (where variables can be controlled throughout) for quantitative analysis and naturally occurring data for qualitative analysis can yield the best results.

Session: Paper session
Various Topics
Saturday, April 5, 2008, 13:45-15:15
room: 05