Q-age group: 35-44
I-father born in
yod segregation geographical area (Peníscola).
born in Benicarló
/- no information,
as this is defined by occupation.
level: secondary education
0- no knowledge of
Valencian language norms.
The inclusion of mathematical procedures is essential in obtaining a description
of languages as heterogeneously ordered structures. In this respect, applying the
quantitative model requires a degree of training in statistics and probability theory.
Since the first
probabilistic additive model, used in 1969 by Labov in his study of copula contraction and
elision in English in New York, up to the latest generations of the logistic-
multiplicative model represented by the VARBRUL programme, there has been permanent
progress. The group of researchers in Montreal has without any doubt played a leading role
in this field. (11)
Indeed, taking the step from the analysis of frequencies to probabilities based
on Cedergren and Sankoff was the most important qualitative leap. The new multiplicative
model meant that the linguistic competence of the speakers was described, based on the
idea that behaviour data are a statistical reflection of competence. The calculation of
probabilities is therefore carried out according to the frequencies observed in a body
representative of the linguistic behaviour of the speakers in a community. Silva-Corvalán
(1989:56) describes the first version of the VARBRUL program as follows:
this mathematical procedure converts the real frequencies of occurrence of a variant X
according to a range of factors (the independent variables) into theoretical probabilities
and assigns a probabilistic index to each factor which reflects the effect that this
factor has on the probability of occurrence of the variant X. The VARBRUL 2 program shown
here introduces a logistic model with the following mathematical formulation:
p Po Pi Pj Pk
1-p 1-po 1-pi 1-pj
represents the probability that a variant occurs in a given context, Po is a
probability common to all the factors which is assumed to have a given effect on the
variant and Pi, Pj, Pk, etc. are the probabilities of contribution of each factor
(which in total make up the variants context) to the production of a variant.
Using this logistic-multiplicative model, VARBRUL estimates the value of each of these
probabilities based on the frequencies observed by the researcher. These values are in the
range 0-1. If the value of a pi factor is 0.5, this will have no effect on the
variants occurrence; values over 0.5 will have a positive effect and those under
0.5, a negative effect. The sum of the values in each group always gives a multiple of
A new version,
named VARBRUL 2S, includes a calculation of the logarithmic probability every time a group
of factors is analysed, which enables different analyses of the same group of data to be
compared. This is done by a multiple regression analysis with a staggered factor
selection. The programme selects the factors that have a significant correlation with the
variable being studied and only calculates the indices of probability for these. Finally,
the VARBRUL III version includes a substantial modification while the application
of previous models only accepted binary variables with two variants analyses
can now be carried out without having to make several computations of data.
BASTARDAS, A. i
SOLER, J. (eds.) (1988). Sociolingüística i llengua catalana, Barcelona:
BELL, A. (1984).
"Language style as audience design", Language in Society, 13,
(1988). "Les actituds lingüístiques", dins A. BASTARDAS i SOLER, J. (eds.)
BOURDIEW, P. i
BOLTANSKI, L. (1975). "Le fetichisme de la langue", dins Actes de la
recherche en Sciences Sociales, 4, pàgs. 2-35.
(1973). The interplay of social and factors in Panama, Tesi Doctoral: Cornell
(1987). "Consideraciones sociolingüísticas sobre la micro-evolución
lingüística", dins LÓPEZ MORALES, H. i VAQUERO, M. (eds.) (1987). ). Actas del
I Congreso Internacional sobre el Español de América, San Juan: Academia
Puertorriqueña de la Lengua Española.