3.1 Hypotheses

Types of language behaviour (6) are the result of the interaction of a set of social representations that the subject makes. Interaction between the following three is particularly important:

- representation of each of the languages at issue
- representation of the interpersonal communication network
- representation of the social reference group

3.1.2 Subhypotheses

Representation of languages and the social network

First Subhypothesis: There is a positive correlation between the representation of languages in contact and the individual social network: the more compact the social network in Catalan (i.e. the more possibilities of contact with Catalan-speakers), (7) the higher the representation in this language.

Second Subhypothesis: The representation of the languages (and social network) are correlated with the use of these languages in contact.

Third Subhypothesis: The representation of the Catalan language, as seen by Catalan-speakers, will be greater overall than that of the Catalan language as seen by Spanish-speakers.

Fourth Subhypothesis: Interaction between the social network and the language representations will be conditioned by the social reference group. In other words, this latter will mark the perception of linguistic groups. Hence, this will produce a desire to converge with or diverge from one’s own group. Spanish-speakers’ desire for distinction from their group results from the need to form part of the social network of Catalan-speakers.

Our research tool was the survey, mainly because we wanted to compare the results obtained with the concept of language attitude and these latter were always quantitative.

4.1 Delimitation of the sample

The basis of the sample are fourth-year ESO pupils from public and private schools across the Balearic Islands.

4.1.1.1 Selection of the level

The starting point for our empirical research already focused the field of analysis on secondary education, which was the scope of our work as teachers, with the aim of possibly drawing up different learning proposals for other research. We decided that the population to analyse would be fourth-year ESO pupils. The reasons for this were numerous. The most influential factor, however, was that this is the point at which compulsory secondary education is completed and, therefore, all pupils reach this level. This compulsory point also adds certain special features because it represents a type of crossroads for pupils, since their decisions on which options to take as regards education must begin here: this could be continuing their education with the Baccalaureate, studying on professional training schemes, or entering the world of work. In this last case, there may also be the option of taking further studies courses. Otherwise, at the very least, it could be used to extract sociolinguistic data for analysing future tendencies and uses of the different languages. The complexity of the questionnaire was also a reason for choosing the final year, since a more mature age group was preferred.

4.1.1.2 Sample Confidence Interval

The confidence interval we chose for the sample was of two sigmas, i.e. 95.5%; which is habitually used for working with. It indicates that the values of the universe must fall within this quantitative space formed by the average, the percentage obtained ±2 (plus or minus 2%). This sampling error is merely the maximum statistical error of the sample. It is also generic, because it is valid for the set of all the diverse samples of the same size that could be taken from the same population.

4.1.1.3 Sample Unit

As we said earlier, this is made up of fourth-year ESO pupils from public and private schools across the Balearic Islands. On the basis of the population of pupils across the Balearic Islands in the fourth year of ESO during the 2001-2002 academic year, we established what the population had to be to ensure a representative sample.