3.1 PILOT STUDY
3.1.3 Key findings and revised methods
In this section I will briefly sketch some key findings from the pilot study and also how I addressed methodological limitations in preparation for the main study. This section is not an exhaustive report of all results from the pilot study, which are reported alongside the results from the main study in chapters 4, 5, 6, and 7.
In brief, the multidimensional approach proposed in the pilot study was appropriate, though not exhaustive. As stated in the Literature Review, measuring EV brings up two important problems: reliability and the appropriateness of the methods to capture the multiple dimensions of EV. Regarding reliability, while the pilot study was informative, no major predictions can be made beyond this particular case because of the impact of linguistic attitudes
on language use and because EV is relative to the context. Hence, I implemented a standardized measure of EV in the main study. The use of standardized methods produces reliable measures of EV and contributes to a unitized worldwide methodological approach. Regarding the methodological appropriateness to address the complexity of EV, the pilot study revealed that a more refined methodology was needed. The revised methods needed to take into account the subjective component and that would give more relevance to the participants’ emic viewpoint and also incorporated more qualitative and integrative approaches.
3.1.3.1 Limitations on Objective EV. The analysis of demographic trends from census data (Dane, 2005, 2014) provided objective evidence of unfavorable demographic factors for language maintenance in San Andrés and stable favorable factors for language maintenance in Providencia. This analysis disclosed a significant rate of language shift (23.13%) in San Andrés as compared to a lower rate (5.76%) in Providencia. These percentages correspond to Raizal people (especially young people) who declared they did not to speak Creole and instead have acquired Spanish as their native language. However, these rates were stated from the demographic analysis alone, while none of those who might have shifted to Spanish were recruited given the selection criteria of the pilot study, which targeted fluent Creole speakers. In order to document the predicted shift trend with attitudinal and linguistic data, the main study included 16 new participants who were deemed to be undergoing a process of language shift to Spanish (see section 3.2.1.3).
3.1.3.2 Limitations on Subjective EV. The assessment of the subjective EV showed a mixed set
of patterns. Firstly, it produced higher EV scores for all participants from Providencia, which is consistent with the demographic factors favoring language maintenance in this island. On the other hand, and contrary to the expectations, the young group from San Andrés scored higher than any other group, which contrasts with the language shift trend in San Andrés.
In general, the youngsters appeared to be very optimistic and positive about their culture and language, while the older adults seemed more critical of the community practices and more aware of the social problems of the islands (e.g. discrimination). It is possible that speakers simply reproduced stereotypical responses intended as appropriate for a non-native researcher. Also, the private setting of the interview might have facilitated the provision of such
stereotypical responses. Therefore, the main study addressed people’s opinions and ideological positions in a way that they were publically displayed, discussed, and negotiated. This was done using group and peer interview techniques. The group interview would encourage the confrontation of multiple viewpoints, their negotiation, and perhaps the display of privileged information otherwise undeclared, whereas the peer interview technique would encourage a more natural display of participants’ perceptions than what is displayed for a researcher.
Similarly, the analysis of perception data disclosed differential attitudes of the participants toward the stimuli-languages. Using the matched guise questionnaire, the English stimuli received the highest rates in different categories such as intelligence, friendliness, and linguistic accuracy, suggesting that the lexifier language is perceived as a prestigious model language for Creole speakers. The Creole stimuli also received high rates especially from the young adults and the females, given that Creole appeared to have some covert prestige as an identity marker among these groups. On the other hand, both the youngsters and the females gave the lowest rates to the Spanish stimuli, suggesting some negative attitudes toward this language, probably because it might be perceived as a threat for the local language and culture.
Nevertheless, the pilot perception experiment had some issues that made it difficult to reach any conclusive statements. First of all, the adjectives included in the questionnaire did not reflect the native categories used by the participants to perceive and judge the local languages and their speakers. Secondly, each subset of listeners listened to a different experimental speaker: the young adult males listened to a different speaker from the older adult males, and these were different from the experimental speakers listened to by the young and the older adult females. This increaed speaker variability and made the data less comparable across the listener subsets, given the different narrative styles of the speakers and other differences between them. Finally, the 4-point Likert scale and its alignment to a categorical scale of agreement and disagreement was problematic for capturing middle-point responses and sometimes was confusing for the participants, especially when expressing disagreement to items with negative connotations, such as the speaker sounds inaccurate. Some listeners attempted to choose 4 (maximum agreement with the statement) when they actually wanted to express that the speaker sounded accurate to them, in which case 1 or 2 (disagreement) were to be chosen.
Therefore, for the main study, the perception study and its MG questionnaire were redone. Rather than setting the researcher’s viewpoint, a new procedure was implemented to
capture the participants’ emic viewpoint, while a new questionnaire was created using emergent native categories from the participants (Campbell-Kibler, 2006, p. 72; Gaies & Beebe, 1991, p. 167). This is not to say that, once these adjustments were done, the MG technique would provide a neat reflection of the participants’ linguistic attitudes. The complexity of these attitudes and other limitations of the MG technique (Gaies & Beebe, 1991; Ihemere, 2006) are more extensively discussed in sections 6.2 and 3.2.2.2, respectively. Nevertheless, the triangulation of this instrument with other techniques enabled a comprehensive understanding of the underlying ideologies behind the different outcomes of EV.
3.1.3.3 Limitations on the linguistic evidence. The analysis of production data showed that the
participants from San Andrés used more conservative Creole features, such as the plural marker dem, the progressive marker deh, and the locative deh, than those from Providencia. These features were comparatively higher among young adults and females. The participants from Providencia preferred to use less conservative features, such as the suffixes –s/–es, –ing, auxiliary verbs (to be, to do), and the adverbs here and there, which appeared to be closer to the lexifier language. This result suggested that, in the absence of a significant perceived threat from Spanish, English was targeted as a model language for Creole speech in Providencia. The higher rates of conservative Creole features in San Andrés and among the young adults and females were consistent with their positive attitudes toward Creole as disclosed in the EV interview and the MG questionnaire. Thus, the higher use of conservative Creole features also appeared to be a response to the perceived threat from Spanish, especially from those who have been more exposed to the language shift trend: the young generations from San Andrés.
Although the analysis of this small set of linguistic features was productive, in the main study I collected a larger oral Creole corpus. The larger corpus increased the probability for some other features to surface. For example, Spanish loanwords were more frequent in the larger corpus of the main study, suggesting some possible effects from the threatening language, especially for the Creole-shifting participants. Furthermore, the larger corpus allowed the implementation of more appropriate statistical analyses –such as logistic regression and mixed effects models– based on token frequency and weighted means across speakers. Finally, increasing the size of the corpus was also an opportunity to explore some local genres, such as those described in section 1.2.5.3.
In all, the analysis of some linguistic features, the perceptual judgments of linguistic input, the subjective perspective of participants, and the demographic measures of EV seem to be related but not linearly aligned. The inclusion of both the demographic and the subjective components along with production data has been important for a multidimensional understanding of EV, as suggested by Giles et al (1977) and Abrams et al (2009). Moreover, the EV interview showed some participants’ concerns regarding social processes (e.g. land alienation, discrimination) and cultural processes (e.g. changes on traditional activities, beliefs) that were beyond the language itself. The youngsters not only declared a positive position toward the local values, but they also appeared to be actively resisting the shifting process by promoting conservative Creole features in their own speech.
Consequently, the refinements implemented in the main study targeted a more comprehensive understanding of EV in the specific ecology of the islands, as shown in the next section. For example, I analyzed the objective language shift trend that threatens the local language, but I also took into account the participants’ emotional attachment to the local language and culture, some possible motivations for language shift, and a complex set of language ideologies. Although these refinements do not exhaust all possible dimensions of EV, addressing them in the main study was instrumental to gain a deeper understanding of EV in the islands.