Lets see other interesting aspects about
the comparatives between both researches:
-The representation
of Spanish (the image that the speakers make) is one of the principal facts correlation
the use of Catalan. It increases in all linguistic groups, but it is still very far from
Catalan.
-The Catalan
representation had in 1993 a very high average inside the four linguistic groups. (Between
645 and 413), that decreased slightly in year 2000, except in the exclusive Spanish
speakers, that increase 13 points.
-The social net
in Catalan, this means the whole group of persons with who the speaker relations with
habitually) decreases slightly in all linguistic groups.
-The social net in
Spanish (that went from 88 to 162) descends in all groups (more than the Catalan net did)
except in the exclusive Spanish speakers, that increases one point and it established
between 63 and 163.
-In all the groups
decreases the identification with the Catalan speakers except in the exclusive
Spanish speakers, that increases 0.26.
-The average of
identification with Spanish speakers increased all.
- The only desired
group by the pupils (the group were they would like to be) that has increased is
the one that speaks more Catalan. (1.5%) The exclusive Catalan speakers decrease in the
same proportion and the Spanish-speaking groups keep exactly the same.
-The results among
the variables that influence on the use of Catalan are very similar, independently
from the year and analysis used. Let us see in the following table:
Table 6. Variables from the
discriminate and regression attitudes from 1993 and year 2000
1993 |
2000 |
A. DISCRIMINATE |
A. OF REGRESSION |
A. DISCRIMINATE |
A. OF REGRESSION |
Representation of Spanish |
Representation of Spanish |
Representation of Spanish |
Representation of Spanish |
Social net in Catalan |
Social net in Spanish |
Social net in Spanish |
Social net in Catalan |
Identity |
Identification with spanish speakers |
Social net in Catalan |
Social net in Spanish |
Social net in Spanish |
Language of the parents |
Identity |
Identity |
Representation of Catalan |
|
|
|
-The correlation,
these means, the co-variations between the variables of the hypothesis all increase in
year 2000, and were already high in year 1993. This fact means that all the procedures
that we have seen in the previous epigraph (whether the analysis of regression whether it
was discriminate, or the correlation) corroborate the hypothesis in a very clarifying way.
4.2.2. The techniques of induction by graphs (SIPINA)
We wanted to add
an analysis method of the details named explanative. (Like the correlation, the
discriminate analysis and of multiply regression that we have just seen) so it will be
very helpful because it allows us to explain a qualitative variable using the explicative
variables, by themselves discredited qualities. We are speaking of the acknowledgement by
induction graphs. This method consists in building a forecast function under the shape of
a graph and decision tree. This allows us to describe o foresee the value taken from a
particular variable named endogenous, in function with a name of variables named
exogenous.
The program that
we have used has been elaborated by D. A. Zighed and Ricco Rakotomalala, members of the
ERIC (Equipe de Recherche en Ingénierie des Connaissances) from the University of
Lumière Lyon 2. This program is called SIPINA that, as we already advanced, means
Interactive System for the Processes of Investigation Non-Ramificated.
It extracts acknowledgement from details, it produces decision graphs, that are a
generalisation of the decision trees and allows us to treat the explanation problems.
These type of
problems can be found frequently in the human science (sociology, biology, psychology,
medicine
). The SIPINA is used for example, as help in diagnostics, for the
evaluation of a risk, for the establishment of profile characteristics from certain towns,
etc (8)
What we obtain
with this method is similar to the establishment of profile characteristics in certain
towns. We characterise the four linguistic groups that we have established:
-Exclusive Catalan
speakers (ECAT),
-More Catalan (+CAT),
-More Spanish (+CAST),
-Exclusive Spanish speakers (ECAS),
With some of the
variables that we have introduced in the research. The proceeding is similar to the
discriminate analysis (we will comment at point 4.3.1). In first place, lets see in table
nr. 7 the global results that this new technique has gave us. |