Approaches of Language Attitudes

There are three assessment techniques used in language attitude studies: content analysis of societal treatment, direct measurement and indirect measurement (Garrett, 2010: 37). Content analysis includes autobiographical, observational and case studies.

Research not involving explicit requests to respondents for their views or reactions is also classified under this category (Ryan, Giles, and Sebastian, 1982: 7).

Direct assessment includes questionnaires and interviews, and this is an approach relying on overt elicitation of attitudes. Questions involved in this measurement include language evaluation, language preference, motivation for learning a particular language, judges of social groups who use a certain linguistic variety, self-reports about language use, views regarding language policies, etc. (Ryan, Giles, and Sebastian, 1982: 7). Fasold (1984: 152) reports on two further types of questions:

open and closed. Open questions provide the respondents maximum freedom to express their opinions toward the questions in research. However, the respondents might also give an answer that is not related to the question, while it also problematic to score the responses to open questions. As for closed questions, a particular format is established for the respondents such as semantic differential scales, yes/no questions, multiple choice, or ranking allocation. Closed questions provide respondents with an easy way to answer, with the drawback that the answers are restricted to the researchers’ terms. The advantage of closed questions is the relatively easy way in scoring. Since both types of questions have their own advantages and weaknesses, Fasold (1984: 152) suggests that a pilot research with open questions is recommended in order to construct a closed-question questionnaire.

The indirect approach is best known for the “matched-guise technique.” As the name suggests, instead of asking straight questions about people’s attitudes, indirect approaches adopt a subtle or even deceptive way to elicit respondents’ judgments

18

(Edwards, 1985; Garrett, 2010). For instance, some studies aim to elicit respondents’

evaluation reactions toward a speech presented and some researchers investigate whether informants make behavioral adjustments, like accommodation in speech style or non-behavioral reaction, such as a change in attitudes (Ryan, Giles, and Sebastian, 1982: 8). This approach looks at the assessment of listeners’ affective responses and their beliefs about the speaker that is presented to them. This approach furthermore deals with other issues like the discrepancies between evaluations of a variety as representative of a particular group, as well as attitudes toward the speech characteristics (e.g., beautiful, inarticulate or improper).

Cooper and Fishman (1974:16-17) conducted an investigation by applying the matched-guise technique with the purpose of looking at the effectiveness of languages on persuasion. In their study, respondents were asked to listen to passages recorded by a fluent speaker of Arabic and Hebrew. The two scholars presumed that Arabic was more persuasive in uttering traditional point of view, and that Hebrew was more suitable in expressing a scientific argument. Regarding the contents of the passages, one was concerned with the disadvantage of having tobacco, supported with scientific evidence. The other described the effects of liquor by citing arguments from a traditional point of view. After listening to the passages, the respondents were asked direct questions and indirect questions. The former was mainly designed to explore if the respondents agreed with the position provided in each argument, and whether the latter asked if an increased tax was appropriate to decrease relevant consumption.

Cooper and Fishman found that respondents held the same attitudes in answering direct question, whereas dramatic differences were shown when responding to indirect questions. When the passage was delivered in Hebrew with scientific arguments for increasing tax, the number of respondents who agreed with the message was twice as high compared to those delivered in Arabic. On the other hand, when the

19

respondents heard the message in Arabic from traditional point of view, the numbers of agreements were twice as high as when it was heard in Hebrew. Hence, Cooper and Fishman (1974) concluded that, in indirectly comparing the effectiveness of Hebrew and Arabic, the former was more persuasive in terms of scientific arguments, whereas the latter was a better choice when considering traditional arguments.

Another approach often used with the matched-guise technique is semantic differential scales (Fasold, 1984: 150). The scales are designed with two opposite extremes in terms of a trait at either end. A typical example regarding semantic differential scales is shown in figure 2.1:

friendly __ __ __ __ __ __ __ unfriendly Figure 2.1 Seven-Point Semantic Differential Scale

The listener who participates in the matched-guise technique is asked to judge if the speaker on tape sounds friendly or unfriendly by placing a mark on one of the blank spaces. The closer the mark is placed to the left-most end on the scale, the friendlier the speaker’s sounds are judged; the reverse way represents more unfriendly sounds.

The procedure used in scoring semantic-differential scales is to firstly assign numbers to each of the spaces on the scale. Take figure 2.1 for example, 7 points may be assigned to represent ‘friendliest’, which is followed by 6 on the next space, and 1 point indicates the least friendly. After the responses have been collected, the number of participants who place their evaluation for the speaker on the scale are multiplied by the value of the spaces. Finally, the scores will be divided by the total number of listeners. The final result then represents the mean evaluation for this speaker on the scale. Take figure 2.1 for instance again, the number represents the mean evaluation of the listeners for the speaker on the friendly guise. If the result is 5.10, it is interpreted as friendly since the number locates more than the middle space on the

20

seven-point scale (Fasold, 1984: 151).