74 The Language of Fashion
recommended, by the stated aim of the research, which is to establish
a formal system and not to describe a concrete Fashion, to provide the
general ‘grammar’ of Fashion and not its sociology. First, those selected
in relation to time; if we are looking for a structure, it is a good idea to
work on only one state of Fashion, that is a synchronic one. now, as we
have said, the synchrony of Fashion is decided by fashion itself: it is the
Fashion for one year.^19 For this we decided to work on the magazines
from the year 1958–59 (June to June), but this date obviously has no
methodological importance; we could have chosen any other year, for
what we were looking to describe was not such and such a Fashion, but
Fashion; once collected, isolated from its year, the object (or utterance
[l’énoncé]) found its place in a purely formal system of functions.^20 So
there will be no indication here of any contingent Fashion, even less of a
history of Fashion: we did not want to look at the substance of Fashion,
but only at the structure of its signs.
Similarly (and here is the second limit placed on the corpus), there
would have been no interest in going through all the magazines for
one year unless we had wanted to isolate the substantial differences
(ideological, aesthetic or social) between them; from a sociological
point of view this would be a crucial issue, because each magazine
would refer both to a socially defined audience and to a particular body
of representations (and we have suggested continually how semantic
analysis could help with this problem), but the differential sociology of
magazines, audiences and ideologies was not the stated aim of this
research: we have never aimed here at anything but a pre-sociology
of Fashion.^21 So we went through in an exhaustive way only two
magazines (Elle and Jardin des Modes), without denying the possibility
of finding some things in other publications (mainly Vogue and Echo de
la Mode^22 ), as well as the weekly pages which certain daily newspapers
provide on Fashion. What was important, given the semiological project,
was to constitute a corpus which was reasonably saturated with all the
possible differences in vestimentary signs; conversely, it did not matter if
there was an element of repetition in these differences, for what makes
meaning is not repetition but difference; structurally, a rare Fashion trait
is as important as a frequent one, a gardenia as important as a long
skirt; the aim here was to distinguish the units, not to count them.^23
having then established these principles, we needed to decide what
we were looking for. The rule was obviously to work on the pure and