The real group is all the bar of the same colour. For example, in table 9, the exclusive Spanish speaking group is delimited correctly in a 79% of the cases, only a 2% of pupils "more Spanish speaking" are included (the little yellow band of the base).

Otherwise a 21% is incorrectly situated in the bar of "more Spanish speakers" (the blue band).

4.3.2. The evolution forecasts of linguistic groups since 1993

The reference group (or willed linguistic group) plays an essential predictive paper upon the fulfilled predictions of 1993. We proposed this variable as a fact to preview the evolution of the composition from the established linguistic group. We remarked that the willed group in 1993 could be the real use after a generation (33 years). Let us see it in the following table:

Table 11. The confirmation of the forecasts

With the details of 1993 we knew the composition of four linguistic groups, (the first horizontal line of the table: 12.90%, 26,60%, 44.10% and 16.40%) and the groups were the pupils wanted to be (the third horizontal line of the table: 23.90%, 41.10%, 27.80% and 7.20%). With the details that we have just obtained (second horizontal line) we noticed that the three groups follow the forecasts that we made in the research of 1993: Both groups of Catalan speakers increase and the ones that speak more Spanish, decrease. Therefore the Spanish speakers group takes the contrary way of the predictions, and increases in 0.30%. Fact that is not very significant because the other evolutions are of 4%, 7.50% and 11.80%, all of them in the direction that we indicated.

The details obtained in year 2000, even allows us to shorten the period of 33 years, because what we affirm has already been completed in a 36.36%, in a 51.72% and in a 71.08%. Except in the exclusive Spanish speaker group that has taken this slight tendency opposed to the forecasts.

5. Conclusions

The objectives that we concrete in point 2, have been completed in excess. The first, was to know the evolution of the uses of Catalan in the seven years that go form 1993 to 2000; in the point 4.1 we have seen them clearly. In second place, we wanted to know if the variables that favourite the use of Catalan changed or not, and we noticed that they have not done in a substantial way (point 4.2). In third place, we have checked if the forecasts made eight years ago have been completed in a very high way. And, finally, we have corroborated our theoretical proposal for third time (1193, 1998 and 2000).

Therefore the dominant theories did not arrive to explain even the 10% of the characteristics from the language, with our own theoretical and methodical point of view, we have obtained a 75% of justification of these linguistic characteristics. But not only this. We’ve been further on to the level of co-variation, we’ve used instruments of multivariable analysis, like the multiply regression and the induction techniques by graphs (SIPINA), also the discriminate analysis; that has allowed us, between other things, to prove the predictive capacity of our model, that reaches 83.8% of the classified cases.

We can ask ourselves why have we obtained these results. A very easy answer would be that we have passed from a mono-casual diagram (or attitude or motivation or identity…) to a conception that articulates three principal variables, and also includes the facts that pretended to explain the three of them separately; like the attitude ones, that form part of our theory of social representations, that we purposed as a frame of our conception.

By other way, we have improved in the classifications and we have divided the contexts of linguistic replacement in four groups of speakers: two that are nearer from the mono-linguistic and two bilinguals, with a priority language in each case. Said in a different way, we have justified the dependent variable (linguistic use) with the interaction of the independent variables: the representation of the reference group, the representation of languages and the representation of social nets.

Finally, the recourse to the catastrophe theory has allowed us to realise our discursive reflection of agreement, not only with the empirical rules, but as well to a minimal formal system, thing that means that the proposed model has a great coherence: As theoretical as formal, and we corroborate it, again, empirically with the investigation done in year 2000.