The intermediaries’ task is usually essential in this type of research. In this case, the strategy of access to informants provided very good results, as it is not at all easy to gain the trust of helpers with an intrinsic intellectual curiosity, as secondary school students tend to have. It is also important to note that the role of teacher played by the son/daughter, grandchild or brother/sister establishes a relationship between researcher and informant that encourages receptivity and eliminates suspicion to an extraordinary extent. This appraisal stands in stark contrast to the initially suspicious attitude of some of the informants that did not come from the forms at the Institute. Nevertheless, it is necessary to take into consideration possible linguistic effects that may be caused by the display of superior status by the researcher. If we wish to avoid the formality involved in an interpersonal hierarchical relationship, it will be necessary to use all our powers of persuasion in order to attain a good understanding and a friendly atmosphere.

2.4. The participants

Moreno (1989:71-77) makes an interesting distinction between researcher and explorer (or in our case, interviewer). The sample size may make the participation of several explorers necessary. It should be said, however, that most studies in the sociolinguistic field have been carried out by the same researchers. We feel that if it is sure that the linguistic data can be gathered in a realistic time span, it is advisable for the interviewer to be just one person (6), the researcher him/herself if possible. In this way, consistency and homogeneity in data collection would be ensured. This homogeneity is a methodological imperative of supreme importance. The manner data in which is obtained from each informant should be as similar as possible.

The interviewer’s role is crucial to the registration’s success. The rule is that an osmotic attitude should be adopted as far as the informant is concerned. We have already stated that the possible perception of superior status should be avoided. At the same time, age, gender, educational level and other differences that may inhibit the speaker should also be given careful consideration. The interviewer must adapt him/herself – as far as possible – to the informant’s linguistic and extra-linguistic behaviour. And finally, a disproportionate involvement in the conversation should be avoided. If the interviewer belongs to the community being studied, it will easier for him/her to have access to casual speech. The other side of the coin is that it will probably be more difficult for him/her to increase the formality of the informant’s discourse.

If he/she is not a member of the community, there are two alternatives (apart from osmotic adaptation terms). One is to become a member (although full integration into the group is never guaranteed). We are in the field of ethnography of communication, where we can place, at least partially, the renowned work by Milroy (1987). This research introduces the concept of the social network to sociolinguistics. This is a more realistic approach to linguistic facts, which is essentially qualitative, and more suited to studies of linguistic interaction or evaluation of communicational norms than studies of sociolinguistic correlation. The other alternative is to train a member of the community to do the fieldwork.

The objective of the research will determine whether the interviews take the form of individual conversation or with multiple informants. While this is especially advisable in studies of linguistic interaction, it is usually used in research in studies of sociolinguistic correlation, especially if phonetic-phonological variables are being considered. Despite group conversations providing greater encouragement for a relaxed atmosphere between acquaintances, they lead to many technical problems while recording (different microphones must be used, different contributions must not be superimposed on each other).

Finally, the informant’s home is a frequent choice for the venue of the interview. This choice is a result of the need to encourage the maximum familiarity and comfort and at the same time, the quiet necessary for recording. In the study by Pradilla (1993a), the place where recording took place was shown to be a variable needing to be controlled, as a highly formal linguistic and excessively tense linguistic behaviour was seen in the interviews carried out in a public place. These interviews subsequently had to be eliminated from the sample because they invalidated the homogeneity of the methodological design.

3. Data gathering. Complementary techniques to the sociolinguistic interview

In the first section of this methodology description (The sociolinguistics of variation: a methodological approximation (I)) we concerned ourselves at length with one of the basic tools in the gathering of data – the Labovian sociolinguistic interview. It is also a good idea to consider some methodological tools that complement this – anonymous observation and the subjective evaluation trial.

3.1. Anonymous observation

This technique has its origins in Labov’s study (1966a) of the implosive /r/ in New York. It consists of asking one or several questions, the reply to which should supply instances of the variable under study.

In quantitative studies, anonymous observation should be taken to mean a complementary technique with no statistical value, but which supplies qualitative data of great interest concerning casual discourse. If we attain our objective of obtaining linguistic data in its natural context, we have overcome the methodological obstacle of the observer’s paradox.

In research carried out by Pradilla (1993a), the strategy designed consisted of placing the researcher in a street next to the Creu Roja Hospital, with the air of one who is lost. The next step was to ask passers-by for information, at wide time intervals, using the following question: "Do you live here?" In the case of an affirmative answer, the researcher continued "Well, perhaps you can tell me where there is a casualty ward around here." The answer was obvious, as the Creu Roja Hospital was nearby. The researcher finished off with the question "And do you think they’ll be able to help me there?" This question generally led to an answer including the word metge(s) - (doctor(s)).