Page 385 - The ISA Handbook in Contemporary Sociology

9781412934633-Chap-24 1/10/09 8:55 AM Page 356

356 THE ISA HANDBOOK IN CONTEMPORARY SOCIOLOGY

Our approach aims at overcoming some of This method was applied to the five ques-
these limitations (Duchastel and Armony, tions in the two countries included in the first
1996). By asking simple and general phase of our research (El Salvador and
(non-context-specific) questions and empiri- Honduras). The tables show the main distinc-
cally observing consistent patterns in the tive terms elicited for each question in each
interviewees’ discourse, we obtain robust country, that is, the words that the activists in
data from a comparative and a cumulative each country tend to prioritize when they
perspective, while preserving to some extent enunciate their response to a particular ques-
the complexity, heterogeneity, and internal tion. The rationale behind this procedure is
logic of activists’ representations. quite simple and it is based on the same
We have applied to the interview transcripts assumptions underlying the commonly used
a computer-assisted procedure based on the word-association technique. We aim at
principle of ‘distinctive notions’. 3 This observing recurrent, non-random patterns in
principle is theoretically linked to the con- the subjects’choice of words when they react
cept of paradigmatic preferences, which to a representation (usually conveyed by a
refers to the recurrent linguistic choices concept or a sentence). The columns show
made by speakers when several similar the total frequency of each term (the number
words are available to them. The paradig- of times it appears in the country database),
matic selection can be seen as a menu of pos- the partial frequency (the number of times it
sible choices between similar but not fully appears in the answers to a specific ques-
equivalent words that can be used to fit a slot tion), and the level of significance regarding
in a given sentence. If a given individual con- the difference between expected and
sistently uses the term ‘immigrant’ when observed frequencies. If we observe that the
referring to alien residents, instead of other word x shows a significantly higher than
semantically equivalent terms such as ‘new- expected frequency in the answers to ques-
comer’ or ‘foreigner’, it is possible to infer a tion y, we consider that the word x is statis-
paradigmatic preference (which can be, and tically associated with the question y. All
often is, involuntary and unconscious). terms that are semantically empty were elim-
Interestingly, this phenomenon can be inated from the tables, as well as those whose
observed through statistical means. Words overrepresentation is mostly due to their fre-
that are significantly overused or underused quent use by particular respondents. We set
by a group of speakers when responding to a the confidence limit (the chance that the
given question may reveal a meaningful observed differences are due to an underly-
pattern of preferences within the stream of ing reason) at 99% (a z-value of 3). The terms
discourse. A common criticism of this kind in the tables were sorted by statistical signif-
of word-based statistical approach is that icance (the higher the z-value, the lower the
words mean different things to different probability of a random difference), and
people and in different contexts. While this translated from Spanish into English. All
remark is in itself obviously true (language instances of the selected words were
is, by definition, fluid), it misses a crucial observed in context in order to make sure that
point: this analysis focuses on the signifiers. they convey a relatively stable meaning.
The fact that most respondents, with different Table 24.1 shows the results obtained from
personal and cultural backgrounds, chose the the distinctive-notion analysis of the answers
same words to respond to certain ideas tells to the first question: ‘What are the main
us about a shared disposition to frame social injustices in this country?’ We observe a
reality in a particular way. The analyst does strikingly clear pattern in terms of cross-
not presuppose that this meaning exists; she country and thematic coherence. The respon-
observes it as a social fact (see Laclau and dents dwell on the notion of ‘injustice’,
Mouffe, 1985). and while they link it to all-encompassing