Cheshire stresses that language attitude surveys are important to language planning, teaching and the status of a language in public life generally (Cheshire 1991: 8).
Baker has listed some possible difficulties connected with measuring an individual's attitudes. The most relevant for my study are that people may, consciously or unconsciously, give socially desirable answers, or, for instance, the purpose of the research may affect them (Baker 1988: 117). To eliminate this, I have used a questionnaire which consists of three parts, all of which are independent, though closely interconnected. With this I wanted to ensure that it would be more difficult to try to give, for instance, "politically correct" answers. Also, to eliminate the effect of knowing the purpose of the research, it was not specified to the informants that their attitudes towards English were being studied.
Although attitude scores are imperfect representations of individuals' attitudes, if the attitude test has been constructed well, its' results can be relatively reliable. Baker believes in attitude tests, claiming that "in attitude change lies one hope for language life and resurrection". Attitudes, once recognized, can be turned into action (ibid, 141).
In Fasold's view, even the question of one's mother tongue may present the first problem in attitude surveys in multilingual countries: an Indian person's answer to the question can be based on the desire to be associated with a particular language, to appear patriotic, or to show belonging to a local ethnic group (Fasold 1984: 23).
The study was, indeed not wholly unproblematic: some people clearly wanted to give socially desirable answers, the kind of answers they thought would be expected from them, or maybe the kind of answers they ideally would like the language use to be like. In my opinion, however, domain analysis helped to minimize this problem, because ambivalencies in people's language use in different domains and their claimed attitudes could be compared with each other. Some informants, argued, for instance, that English was not at all important to them when, e.g. looking for a job: in the domain analysis, however, it turned out that the same people claimed very high frequencies for the use of English in that particular situation.
