Abstract This article reports on a set of language attitude experiments undertaken in Belize in 2013–2014. The experiments tested attitudes toward Belizean Kriol in two different situational settings among 96 Belizean university students, exploring the extent to which experimental ‘setting’ had an effect on survey results. We administered verbal guise tests to 48 university students in classroom settings at the University of Belize campuses in Belize City and Punta Gorda. We then administered the same tests to 48 students in non-classroom settings in these same two cities. We found significant differences in both sets of results, with classroom participants registering stronger preferences for Belize City Kriol in status and solidarity traits. Our results add to a small but important body of literature in which setting is shown to influence language attitudes, and they provide support for an understanding of the quantitative survey as a meaning-making activity. 1. INTRODUCTION Language attitude research plays a central role in illuminating many of the fundamental questions in applied linguistics. It has had important influences on second language learning, bi/multilingualism, assessment and pedagogy, planning and policy, endangerment, and revitalization, to name a few areas. Over the years, attitudes have been gathered and assessed using a range of different quantitative and qualitative types of methods. One aspect that has been largely overlooked, however, is the role of situational setting in the gathering of the attitudes themselves. That is, to what extent does experimental setting play a role in the kinds of responses given by test participants, and to what extent are language attitudes context dependent? Questions such as these have important implications for our understanding of what language attitudes are and for how they should be studied, and they provide the motivation for the present study, which explores experimental setting effects in the Creole context of coastal Belize. In an important paper published in 1983, Creber and Giles argued that situational settings could have an effect on the outcome of language attitude experiments, claiming essentially that language attitudes could be shown to vary depending on the situation in which they were tested. Specifically, C&G found that attitudes toward RP English among English school children varied depending on whether their experiment was administered in a formal setting (i.e. a school) or an informal setting, such as an evening youth club, the effect being that language attitudes were seen to be context-dependent. Surprisingly, this problem has received very little attention in the literature since that time, and 30 years later, in 2013, Prikhodkine and Preston make a similar claim introducing a formal workshop that they organized on some of these same issues, writing that ‘sociolinguists have granted relatively little attention to the social context of language attitude production’. The question of setting effects has been underexplored for every language and setting, but this is especially true for lesser-studied, minority languages such as the Creole language of the Central American/Caribbean country of Belize. The present article adds to a vanishingly small body of literature on the subject of attitudes and setting, and it does so with attention to a part of the linguistic world that is itself quite understudied. Our interest in the area of applied linguistics, then, is to shine light on these problems of language attitudes, both in terms of collecting data and, importantly, in interpreting previous studies which do not take into account potential setting effects. We attempt to replicate Creber and Giles (1983), but in the Creole context of a non-majority language. As mentioned above, language attitudes are at the heart of many of the core questions addressed by applied linguists; this study seeks to sharpen our understanding of how attitudes are gathered, produced, and interpreted. The outline of the article is as follows. In the next section, we provide a brief overview of the extent to which experimental setting has been explored in the attitude literature generally, including a close look at the results and methods of Creber and Giles (1983). We then discuss the linguistic situation in Belize, as reported in Salmon (2015) and Salmon and Gómez Menjívar (2016), followed by a discussion of the experimental methods utilized in the present work. A presentation of our results follows, and we then conclude with a discussion of implications for our findings and for attitude research in general. 2. SOCIAL CONTEXT AND LANGUAGE ATTITUDE SURVEYS Giles and Ryan (1982: 219) write that ‘[t]he extent to which language variety A is or is not preferred over language variety B depends upon the situation in which the assessment is made’. The implications of this statement are immense; surprisingly, though, experimentation with situational context has played almost no role in language attitude studies to the present time. To our knowledge, the best-known study that does manipulate situational context is Creber and Giles (1983), which specifically investigates the effects of situational variation on attitudes toward two dialects of English. C&G write of their results: [W]e have at last an empirical demonstration that the social setting of evaluation can affect language attitudes […]. Obviously, further research with children and adults in this genre is urgently required so as to place previous so-called established findings empirically in their situational frames of reference. (Creber and Giles 1983: 159) This clarion call for more research is as urgent and unheeded now as it was three decades ago for research in language attitudes. Accordingly, we have modeled much of this present study on Creber and Giles (1983), and so it is worth considering that study in more detail here. C&G investigated attitudes toward RP English and Welsh-accented English among ‘middle-class, standard-accented English (Buckinghamshire) subjects’. They constructed a 2×2 design, with two levels of content and two accents, and they tested 36 participants: 18 participants in the formal school setting and 18 in an informal youth club setting. They relied upon a matched-guise test (Lambert et al. 1960), with a single male speaker who was determined to be proficient in both guises: that is RP and Welsh-accented English. They tested 19 seven-point bipolar scales (Intelligent–Unintelligent, Friendly–Unfriendly, etc.), which were grouped as either status traits or solidarity traits. They then performed 2×2 analysis of variances with factors of accent and context for each of the speaker evaluation traits. Among other results, C&G found that the RP speaker was rated significantly higher than the Welsh speaker in the school setting for status traits such as successful, self-confident, etc. They took this to be confirmation of their hypothesis that the formal school setting would affect the outcome of their research. To our knowledge, there is little other work that explicitly tests situational setting as an experimental variable in this manner. We do find studies, however, in which other aspects of context have been shown to influence speaker evaluations. For example, Carranza and Ryan (1975) tested status and solidarity traits among bilingual Anglo- and Mexican-Americans, and they found that Mexican-American Spanish speakers rated Spanish more favorably when the speaker represented a home domain than a school domain. This group also rated English more favorably when the speaker represented a school domain. Studies have also shown that manners of speaking in the test materials are rated differently depending on context. Thus, Brown et al. (1985) have shown that rate of speech can be evaluated differently in different contexts, noting that ‘attitudes towards a speaker’s slow rate of speech would be different in the context of a nuclear physics lecture than during introductions at a cocktail party’. Internal relations of test conditions have also been shown to affect results. Thus, Abrams and Hogg (1987) found with respect to Scots that participants from Dundee rated Dundee speakers more favorably than they did Glaswegian speakers on status and solidarity. However, when asked to rate Glaswegian speakers compared to standard English speakers, the same participants rated the Glaswegian speakers ‘much higher even than they had rated their own ingroup (Dundee) speakers in the earlier comparison’. In addition, as reported in Cargile et al. (1994), Street (1985) has shown that in an interview situation, participants in the interview rate interviewees different than non-participant observers do. Changing course a bit, Preston and Prikhodkine (2013) shows that lexis choice within test conditions has an effect on results, while Giles et al. (1983) found that results changed when participants discussed their evaluations with each other before making their ratings. Thus, test conditions have been manipulated in various ways over the years, and some of these manipulations have had significant effects on the survey outcomes. As stated above, however, there is very little previous work that specifically takes into account the situational setting of the experiment. 3. THE BELIZEAN KRIOL LANGUAGE The official language of Belize is English; yet, Kriol is the native language of many Belizeans, regardless of their ethnic group, and it is the language adopted by newly arrived Spanish-speaking immigrants from neighboring Honduras, Guatemala, and El Salvador. Similarly, we have observed firsthand Kriol being acquired as a first language among current generations of Mopan and Kekchi Mayan teenagers in the Toledo district of the country. There is currently a strong movement underway by the Belize Kriol Project and National Kriol Council of Belize to promote Kriol for use in education and government and to raise the status of the language in the country in general. To this end, there is an English/Kriol dictionary (Herrera et al. 2007), which defines many Kriol words and includes a chapter of grammatical description, and there is also a Kriol translation of The New Testament, which was completed in January 2013. These projects complement a literary tradition of short fiction and poetry in Kriol, as well as a weekly newspaper column in The Reporter and a weekly radio show on Wave Radio FM 105.9. Finally, there have been three linguistic grammars written of the language—that is Young (1973), Greene (1999), and Decker (2005). To some extent, Kriol is discussed in public primary and secondary school in Belize. The country’s official educational policy as of 2008 says that students are taught to see the differences between English and Kriol, and that there should be discussion of when it is appropriate to use Kriol as opposed to English; however, it is not taught as a subject, and it is not the language of instruction for other subjects (Belize National Standards 2008). Further, according to many of our participants, Kriol is not discussed in the public schools at all, and they do not consider it to be a legitimate language. Many referred to Kriol as ‘broken English’, and this exact phrase was one that that we heard almost everywhere by Belizeans with respect to Kriol. The linguistic focus of this present article is on attitudes toward regional varieties of Kriol within Belize. Apart from the aforementioned works by Salmon and Gómez Menjívar, there is no previous research at all on this topic. 4. HYPOTHESES, METHODOLOGY, AND FIELDWORK SITES In the following sections, we describe the two sets of experiments undertaken in the present study. As both sets of experiments utilized the same test recordings and survey methods (differing only in situational setting), we will discuss these shared aspects first, with the individual details of each set of experiments following in turn. 4.1 Hypotheses In Salmon and Gómez Menjívar (2016), which is the only work we are aware of that test attitudes toward varieties of Kriol, BC Kriol is seen to be more traditional than PG Kriol in general, and it is rated as more prestigious than PG Kriol along several dimensions. Further, based on perception tasks, Salmon and Gómez Menjívar found that all participants were much more adept at identifying BC Kriol than they were PG Kriol—even among residents of Punta Gorda. This depicts a situation in which BC Kriol is held in higher regard and in which PG Kriol, while not stigmatized, was an unidentifiable ‘other’ Belizean Kriol. Given these previous findings, we hypothesized: H1: There would be no significant difference between the classroom and casual groups’ ratings of BC Kriol in status, or between the groups’ ratings of BC Kriol in solidarity. H2: There would be no significant difference between the classroom and casual groups’ ratings of PG Kriol in status, or between the groups’ ratings of PG Kriol in solidarity. H3: The means of (H2) would in general be lower than the means of (H1). Our use of the status and solidarity groupings follows a large body of work in the social psychology literature, which has shown that comparisons between individuals are frequently made along the broad axes of status and solidarity. See, for example, Locke (2003) and the many sources cited therein. 4.2 Shared aspects of research design We chose as fieldwork sites a northern urban area, Belize City, and the southernmost sizeable town, Punta Gorda. Both are coastal cities, in which fishing plays an important economic and cultural role and in which various kinds of maritime tourism—fishing, scuba diving, etc.—are important to the economy. Belize City is the country’s cultural center for ethnic Creoles, with a population of 61.2 per cent ethnic Creole (Belize Census 2010). It is also the country’s largest city, with an overall population of approximately 68,000 residents. Punta Gorda, on the other hand, has a smaller percentage of ethnic Creoles, at 14.7 per cent, and a much smaller overall population of approximately 6,000 (Belize Census 2010). Importantly, there is little contact between the two cities due to travel conditions, as well as to the perception among many Belizeans that Belize City is an exceptionally dangerous place. We chose these two sites as they are both urban and both have universities drawing students from their respective local communities. The main university campus in Belmopan, on the other hand, is much more likely to draw students from around the country—including the western Spanish-speaking districts—and would thus inevitably have provided a much more linguistically diverse participant sample with more variables to take into account. 4.3 Instruments For all experiments, we used a verbal guise test (Cooper 1975; Huygens and Vaughn 1983; Kristiansen 2009, 2011) to gauge attitudes toward the Kriol spoken in the two sites. The basic setup included recordings we made of local speakers in BC and PG, as well as a semantic differential survey (Osgood et al. 1957) which was filled out by test participants in their ratings of the recordings, and which is reproduced below in Figure 1. The same voice recordings and surveys were used in the classroom and casual contexts. This experimental setup, according to Soukup (2013: 252), ‘combining some form of the matched-guise technique […] with a questionnaire on the basis of Osgood scales seems to have become the standard in the (quantitative) measurement of attitudes towards variation in language use’. Figure 1: View largeDownload slide Rating sheet given to all participants Figure 1: View largeDownload slide Rating sheet given to all participants 4.4 Test recordings Unlike the matched-guise test (Lambert et al. 1960), which uses recordings of one speaker performing multiple guises, the verbal guise test uses different speakers for each language variety. Like the matched-guise test, however, the verbal guise conceals the identities and distinguishing information of the speakers from the participants. Kristiansen (2011) notes that the verbal guise test can result in a relaxed control of voice and content effects, as compared to the matched-guise test, since there are two different speakers in the former. We controlled for this as much as possible by recording speakers who were the same age, sex, and ethnicity, and who shared the same occupation. We deemed it unlikely that we would be able to find individual speakers who controlled the different varieties of Kriol spoken in Belize City and Punta Gorda. Thus, our test recordings were made by an individual from Belize City and an individual from Punta Gorda. Both of these individuals were ethnic Creole men in their early 40s whom we have gotten to know quite well over our two years of doing research in Belize.1 Each speaker has lived his entire life in Belize City or Punta Gorda, as the case may be, and each speaker is a taxi driver by occupation. This ensures that the Kriol varieties we recorded are representative of our sites of interest and that there are no issues with authenticity as there can be with the matched-guise tests. (See Garrett 2010 on authenticity questions in the match-guised test.) Similarly, we chose speakers who shared the same occupation as a means of controlling against potential differences in social class. Each speaker was told that we were interested in making recordings of ‘broad Kriol’ to compare the ‘sounds of Belize City’ with ‘the sounds of Punta Gorda’, and each was paid $40BZ for his time. In both recordings, we asked speakers to tell a Belizean folk tale, involving Brer Anansi, Brer Scorpion, Brer Frog, Brer Tiger, etc. The contents are thus comparable; there are, however, clear phonological and morphological distinctions evident in the two recordings which we know to be indicative of the two varieties of Kriol. Salmon and Gómez Menjívar (2016) provide a detailed analysis of the phonological and morphological differences in the two recordings; we include a few examples here. Thus, in the PG Kriol, third-person singular pronouns are pronounced [hi], while in the BC Kriol recording, they are pronounced [i]. In addition, in BC Kriol, the mid-front vowel [e] is frequently raised to [i]. Thus, a word like pain is pronounced closer to ‘peen’ in BC Kriol, but not in PG Kriol. Another difference in vowel quality can be seen in words with the low diphthong [au], such as how. In the BC Kriol sample such words are pronounced with back, mid, rounded vowels, such that BC Kriol how and out are pronounced closer to ‘hoe’ and ‘oat’. In PG Kriol, on the other hand, these vowels are also raised, but are pronounced with a central vowel closer to schwa, similar to the GOAT-fronting process described in the sociolinguistics literature. Thus, in PG, how and now are pronounced closer to [həw] and [nəw]. The recordings were made on a Fostex FR-2LE digital recorder, using a Beyerdynamic M58N(C) microphone, which was fitted with a felt windscreen. Both recordings were made outside in garden environments, and the sound quality is very clear, with very little noise from the recording equipment or surrounding environment. 4.5 Test surveys The rating sheet with survey items that we gave to participants is reproduced below in Figure 1. The traits were pilot-tested with 10 Kriol speakers to insure consistency in the terms’ extensions. Participants listened to the first recording (BC Kriol) and filled out the survey form. Once this task was completed, participants were asked to listen to the recording of the second speaker (PG Kriol), and then follow the same survey procedure. Participants in the casual setting and the classroom setting listened to recordings and filled out the rating surveys in the same order. These aspects of our experiments were shared across both contexts. In the next sections, we discuss the aspects of our methods that differed across the two experiments. 4.6 Participants We surveyed equal numbers of participants in each experimental setting. In the casual setting, this included 23 college-aged Belizeans in Belize City, and 25 college-aged Belizeans in Punta Gorda. The average age of the participants was 20.14 years old, and the sample included equal numbers of women and men. The surveys were anonymous, but we kept track of demographic information such as race, age, sex, native language, and occupation. The number and demographic information of our participants in the classroom setting are almost identical to those of the casual setting: that is we had 23 participants in Belize City and 25 in Punta Gorda. The average age of the participants in the classroom setting was 21.09 years old, and the sample included equal numbers of women and men. 4.7 Procedures in the casual setting Potential test participants were approached on the grounds of the University of Belize campuses in Belize City and Punta Gorda. The grounds around the universities are open, grassy environments where students sit and chat while waiting for class. There are picnic tables, outdoor basketball courts, and food venders, and it is common to see students talking and laughing with each other, texting on cell phones, kicking the soccer ball, etc., in the casual environment. The atmosphere was much more informal than the classroom environment of the other half of the study. We initiated contact with potential participants by introducing ourselves as linguists from the University of X and stating that we were conducting a study on Kriol, without informing the participants what the actual interest in the language was. If they agreed to take part in the study, we then handed them the headphones and clipboard with the survey for the first speaker and began the process. Participants first listened to the recording of the BC speaker, which was simply called ‘Speaker 1’ and then completed the five-level semantic differential survey seen above in Figure 1. Recordings were approximately 30 s each.2 After completing the first survey, participants then listened to the second recording (PG Kriol) and filled out the second survey. Each survey was printed on its own sheet of paper and attached to a clipboard, which the participants held as they filled it out. This first survey of Speaker 1 also included one open-ended question, which asked participants to indicate the origin of the speaker. Once this task was completed for Speaker 1, participants were asked to listen to the recording of the PG speaker and then follow the same survey procedure as described above. When test participants were finished with the survey for the second speaker, they were asked to respond to a few additional qualitative questions. All were asked to indicate where they believed it was appropriate to use Kriol, the first language or dialect they had learned, and all of the languages or dialects they spoke with any proficiency. Importantly, these qualitative questions were not seen by the respondents until after they had filled out the primary surveys. Each interview took 5–30 min, depending on how much the participants wished to add regarding the qualitative aspects on the survey. When the participant was finished with both surveys and had returned them to us, we immediately wrote any further observations or comments that arose during the process on the back of the survey form in question. 4.8 Procedures in the classroom setting As in the casual interviews described above, the classroom attitude surveys were also administered in both Belize City and Punta Gorda. However, these rating tasks were undertaken in university classrooms at the respective University of Belize campuses. The classroom setting of the experiment was chosen for a variety of reasons. Classroom environments are exceedingly common across language attitude studies, with frequently little or no comment as to what impact this might have on researchers’ results. We wanted to see for our own purposes—to gain as clear a picture of the Kriol attitude system as possible—if this environment would affect our results. The fact that it did so was useful for us in understanding the character of our larger project; it also opens up important questions for the many classroom attitude studies across the general literature that make no comment on their experimental environments. Another reason for choosing the classroom environment comes from the inspiration for our study, which is Creber and Giles (1983). In this article C&G find context effects by varying tests between classroom and social settings. To our knowledge, C&G (1983) is the only experiment that manipulates setting in this way. For our purposes, we wanted to replicate this manipulation of setting as much as possible in a non-majority language context, as a means of better understanding the larger issues in play for gathering attitude data. We arranged beforehand with our colleagues at the two universities to visit their classrooms, and we visited two classes per university. When we arrived at each classroom, we were introduced by the Belizean professors as linguists from the University of X, and it was explained that we wished to survey the students as part of a research project on the Kriol language, and that the results would not be seen by their professors. The exact nature of the project was not revealed nor the identity or regional information of the test recordings. The Belizean professors then turned their classes over to us, and we explained the process of listening to the sound clips and filling out the surveys. We used the same 30-s sound clips described above in §4.3.1. This time, however, the clips were played via iTunes on a MacBook and projected through a Bose SoundLink Mini portable Bluetooth speaker. The classrooms were small, and the sound quality was very clear. As in the earlier study, test participants listened to the individual recordings and then filled out the corresponding surveys. The whole process took approximately 10 min of class time. The Belizean professors were present during the entire procedure, and they participated in handing out surveys, clarifying instructions, and so forth. We viewed this presence and participation of the Belizean professors as an important part of maintaining the integrity of the formal classroom setting. When we were finished, we thanked the students and the professors and then left the classroom. 5. RESULTS We arranged our data into two super groups including all of the participants of this study. The first group—that is the classroom-setting group—contains all participants from the classroom settings in Belize City and Punta Gorda. Likewise, the casual-setting group contains all participants from the casual settings tested in the two cities. In addition, we combined several of the individual survey items from Figure 1 into the summative groups of status and solidarity, with status composed of: attractive, educated, eloquent, intelligent, and modern. Solidarity, on the other hand, combines friendly, hardworking, sense of humor, polite, and trustworthy.3 In Table 1, below, we provide the means (Ms) and standard deviations (SDs) to show general tendencies in the different pairings. To indicate where the differences between each pair of means are significant, we relied on independent samples t tests for each classroom/casual pairing.4 As can be seen, in the classroom setting, both Kriols are ranked higher almost without exception. Classroom group participants ranked BC and PG Kriol higher in status, and BC Kriol higher in solidarity. The ranking for solidarity with respect to PG Kriol was equal between the two groups. Table 1: Ms and SDs of casual/classroom pairs, rating status, and solidarity Casual (M, SD) Classroom (M, SD) Status: BC Kriol (3.2, 1.11) (3.5, 1.11) Status: PG Kriol (3.0, 1.07) (3.4, 1.25) Solid: BC Kriol (3.7, 1.18) (4.0, 1.06) Solid: PG Kriol (3.4, 1.10) (3.4, 1.23) Casual (M, SD) Classroom (M, SD) Status: BC Kriol (3.2, 1.11) (3.5, 1.11) Status: PG Kriol (3.0, 1.07) (3.4, 1.25) Solid: BC Kriol (3.7, 1.18) (4.0, 1.06) Solid: PG Kriol (3.4, 1.10) (3.4, 1.23) Note: N = 96: 48 casual and 48 classroom. Table 1: Ms and SDs of casual/classroom pairs, rating status, and solidarity Casual (M, SD) Classroom (M, SD) Status: BC Kriol (3.2, 1.11) (3.5, 1.11) Status: PG Kriol (3.0, 1.07) (3.4, 1.25) Solid: BC Kriol (3.7, 1.18) (4.0, 1.06) Solid: PG Kriol (3.4, 1.10) (3.4, 1.23) Casual (M, SD) Classroom (M, SD) Status: BC Kriol (3.2, 1.11) (3.5, 1.11) Status: PG Kriol (3.0, 1.07) (3.4, 1.25) Solid: BC Kriol (3.7, 1.18) (4.0, 1.06) Solid: PG Kriol (3.4, 1.10) (3.4, 1.23) Note: N = 96: 48 casual and 48 classroom. We predicted that there would be no significant differences found between language ratings in the classroom and casual settings for both BC and PG Kriols. This is not what was found, however. Differences between three of the comparison pairs given above in Table 1 were found to be significant; these are highlighted in blue. There was a significant difference in rankings for the status of BC Kriol in the casual environment (M = 3.2, SD = 1.11) and the classroom environment (M = 3.5, SD = 1.11); t(478) = −3.02, p = 0.002, d = 0.30. Similarly, there was also a significant difference in rankings in status between PG Kriol in the casual environment (M = 3.0, SD = 1.07) and the classroom (M = 3.4, SD = 1.25); t(478) = −3.85, p = .0001, d = 0.36. For the solidarity pairings, there was a significant difference in the ranking of BC Kriol in the casual environment (M = 3.7, SD = 1.18) and classroom (M = 4.0, SD = 1.06); t(478) = −2.43, p = .015, d = 0.29. However, there was no significant difference found in the solidarity pairing for PG Kriol, with the casual environment (M = 3.4, SD = 1.10) and the classroom environment (M = 3.4, SD = 1.23); t(478) = −0.23, p = 0.81, d = 0.02. 6. DISCUSSION Our hypothesis H1, which held that there would be no significant difference between casual and classroom ratings of BC Kriol with respect to status and solidarity, was rejected, with BC Kriol showing higher ratings in the classroom setting than in the casual setting in both status and solidarity. Similarly, our hypothesis H2, which held that there would be no significant difference between casual and classroom ratings of PG Kriol with respect to status and solidarity, was rejected, as the classroom setting rated PG Kriol higher than the casual setting in status, but not solidarity. Finally, our hypothesis H3, which held that the mean results for PG Kriol would be lower than the mean results of BC Kriol in general, was correct. This is not surprising, as it follows from the study reported on in Salmon and Gómez Menjívar (2016), in which BC Kriol in general is regarded as higher prestige than PG Kriol. Clearly the interesting aspect of our results is found in H1 and H2, in which there are significant differences in rating among the classroom/casual groups. As far as we can tell, there are no demographic differences among the two groups which can account for this variation. All participants were 18–25 years of age, with equal numbers of men and women, and all were university students. Similarly, the experimental materials—recordings, surveys, etc.—were the same for each group. The primary difference we see here is the experimental setting: one group was surveyed in their university classrooms, and the other group was surveyed on university grounds but in non-classroom environments. We can suggest several possible reasons for why the classroom environments should rate both Kriols more favorably—though these will only be suggestions at this point.5 A factor which cannot be ignored is the presence of the Belizean professors. As mentioned above, they were present in the classroom at all times while the surveys were being conducted, and they had sanctioned our presence in the classrooms by introducing us and asking their students to cooperate with us. It is not inconceivable that this could have had an effect on student ratings. As we see, the ratings of the classroom groups are higher than those of the casual groups almost without exception. The surveys in the classroom group were obviously anonymous—much more so even than the individualized surveys of the casual settings—and students were assured that their professors would not see the results. Still, it is possible that students felt constrained by the classroom environment to give ratings which they believed would please their professors. While this reasoning is possible, we deem it unlikely for a few reasons. First, as mentioned, the surveys were anonymous; there is no way to know to whom a particular survey belonged, and this was obvious to the students. As such, there is no risk to the students in filling out the survey as honestly as possible. Secondly, most students were unable to identify the origin of the PG Kriol. Given this, how would they know which way to answer the surveys to best please their professors? Thus, we do not believe that the presence of the professors themselves influenced the results. More likely to have affected the results, we believe, is the high regard in which the Belizean university is held. Many of the Belizean students we spoke to in our ethnographic interviews made very positive statements about the university and about its importance in the country of Belize. We found over and over again that students respect the university as a ‘Belizean’ institution, and so viewed it as participating in something that was uniquely Belizean. In this context, it is perhaps unsurprising that attitudes toward Kriol, the burgeoning national language, should be elevated. This idea is supported by the internal relations between BC and PG Kriol, which are also evident from the results given in Table 1, though they are not explicitly mentioned in that context. While the classroom setting generally rates Kriol higher than does the casual setting, BC Kriol is frequently rated higher than PG Kriol, regardless of experimental setting. Salmon and Gómez Menjívar (2016) and Gómez Menjívar and Salmon (forthcoming) provide lengthy discussions rooted in history and contemporary social dynamics for why this should be so. Essentially, as presented in those works, the variety of Kriol as spoken in BC has become the prestige variety, following a long history of slavery, colonialism, border disputes with Guatemala, and finally Independence in 1981. Although English is the official language of Belize, since Independence, there has been a rise in prestige of the Creole language and cultural identity, as the young country continues to try and distinguish itself from its former colonial ties. Similarly, the border tension between Belize and its much larger Spanish-speaking neighbor has contributed to the growing sense of Belizean nationalism, which is largely represented in terms of Creole identity. Thus, the Kriol language continues to rise in prestige as a result of such internal and external pressures. As the largest percentage of ethnic Creoles in the country is found in the Belize district and Belize City, it follows naturally that that variety of Kriol would be taken as most representative. This brief discussion of history and social dynamics provides a natural starting point to answering our final question: that is why the different groups should give equal ratings in the solidarity category for PG Kriol. Given the one-sided ratings of the three other categories, one might expect this rating to follow suit. We attribute this equality in PG solidarity to the overall growing prestige of Kriol in Belize. Neither group was able to pinpoint reliably the location of PG Kriol, though there was at least agreement that it was Belizean, and this can be seen in the fact that while PG Kriol did rate lower than BC Kriol, it is not dramatically lower. If, then, as noted above, Belizean Kriol has come to be associated as a marker of Belizean identity, perhaps it was enough for both groups to recognize that the PG Kriol was Belizean, even if they could not locate its origin precisely within the borders of their country. Since neither group recognized the origin, strong attitudes toward it did not develop one way or the other, other than that it was some variety of their national language. 7. CONCLUSION What is most interesting of all in this article—above and beyond the predictions discussed in the previous section—is that there are significant differences in the results of the classroom and the casual situation ratings. In both of the breakdowns we report here, three out of four categories show significant differences in the comparisons. This is interesting, given that the only obvious variances in the surveys conducted are the situational settings. Recall the quote from Creber and Giles from above: ‘we have at last an empirical demonstration that the social setting of evaluation can affect language attitudes […]’. This present study provides strong additional evidence for this claim, but in a cross-cultural Creole context. This fact is an important one. Creber and Giles’ results were got in England (Buckinghamshire) among speakers of a majority language. The present results, however, which strongly resemble those of Creber and Giles, were got in the Creole contexts of Belize City, as well as in Punta Gorda, which is located in a sparsely populated district of southern Belize. As we mentioned in the introduction to this article, echoing Creber and Giles (1983) and Preston and Prikhodkine (2013), such evidence is sorely needed in the study of language attitudes, and even more so in minority language situations such as that in coastal Belize. The situational dependence of our results suggests that setting effects should be taken into consideration in design of future experimental work. That is, if surveys are carried out solely in classrooms or solely in one type of environment, acknowledgement should be made that results might differ in other contexts, and efforts should be made to complement results with a variety of other settings and methodologies. This is in line with Soukup’s (2012: 218) recasting of the quantitative language attitude experiment as a ‘communicative event’, similar to the ‘interviews attitudinal discourse analysts prefer’. That is, both techniques of information gathering—the quantitative and the qualitative interview—can be seen as meaning-making interactions, in which language attitudes are emergent and context-dependent. As Soukup writes: [A]sking informants in a speaker evaluation experiment to listen to and rate samples of different linguistic varieties can be quite straightforwardly recast as engaging them in a meaning-making activity during which they construct contextually situated evaluated responses. Of course, such a view puts even more emphasis on the fact that findings from experiments (or any other type of data elicitation) are always and necessarily relative to the contextual frame in which they were generated. Any neglect of this relativity may lead to undue generalizations and implausible conclusions. (Soukup 2012: 218) For the purposes of this present work, the last sentence here is the clincher, and drawing conclusions based on just one survey setting—be it the classroom or the schoolyard—would lead to an undue and incomplete picture of attitudes toward Belizean Kriol. This fact is especially important, given the current multilingual situation in Belize and the postcolonial relationship of Kriol to English. As mentioned above, there have been ongoing efforts to raise the status of Kriol to that of a recognized national language in post-Independence Belize. Our results provide a clear picture of the venues and manner in which those efforts have seen success. They also provide a foundation in the educational arena for further language planning by the Belizean government and the Kriol recognition movement. Additionally, that there are different attitudes toward regional varieties suggests it might be useful to incorporate discussion of regional variation into the curriculum; as Kriol begins to be accepted as the national language, there is the inherent danger that regional variation could become stigmatized. Footnotes 1 In verbal guise tests, researchers often rely upon more than one speaker per guise. However, due to fact that the informal half of our testing was done in impromptu settings outside on university grounds, it would have been awkward for participants to listen to and rate four or more speakers. We controlled carefully, however, for potential interference from register, diction, affect, and rate of speech on the two recordings, as well as insuring that demographic features such as sex, age, and social class of the speakers were held consistent. We believe that factors such as these had minimal impact on participants’ ratings of the two varieties (see also Kristiansen 1997; Ladegaard 1998; McKenzie 2008 for research using similar procedures.) 2 In total, 30-s samples are shorter than the 2-min samples employed in the classic matched-guise survey of Lambert et al. (1960); however, due to time constraints of the survey technique, it was necessary to shorten the recordings to a more workable length. Other influential studies have also relied upon 30-s samples in their verbal tests. For example, Patrick (1999: 48) employed 30-s samples in his study of urban and rural dialects of Jamaican Creole, and Kristiansen (2009) employed 30-s samples in his investigation of Danish accents in Copenhagen. Creber and Giles (1983), upon which much of the present article is modeled, employed 60-s samples, though Creber and Giles did not have the same temporal pressures as the current study. Thus, while our 30-s samples are on the short end of test samples across a survey of the literature, they are not without precedent in the literature. 3 The individual items were not combined on a statistical basis such as factor analysis. Due to the smaller overall number of items in the survey, we deemed this to be unnecessary. Rather, the individual items were selected for the study, in consultation with Belizean colleagues, with an eye toward composing the summative groupings from the outset of the experiment. This presents a potential limitation of the study, though it does not affect the larger point, which is the significantly different results obtained in the casual and classroom environments. 4 A question that often arises with respect to the use of t tests with Likert-style surveys is the question of interval vs. ordinal data. It is often argued that t tests require interval data and that Likert-style surveys produce ordinal data. We believe that the question of interval vs. ordinal data in such survey methods is attenuated in our work by the fact that we use the Osgood semantic differential survey, which is built on a numeric rating system rather than words like agree/strongly agree, etc., which are at greater risk to subjective interpretation on the part of test participants. Further, the sample sizes that we use are identical, so the parametrical vs. non-parametrical question is of less significance, given the assumptions of the test. Finally, we report only summative scales, in which there is general agreement that parametric methods such as the t test can be more safely used. 6 An anonymous reviewer reminds us that attitude ratings can be affected by the topic of the discussion in the recorded guises. The Anansi stories we used are relatively informal and are associated with traditional Belizean culture. A different topic in the recorded guise—politics, current events, love and marriage, etc.—might very well have aroused different attitudes among test participants. (See also above discussion of Carranza and Ryan 1975 in Section 2.) A follow-up study in these same environments with different guises and topics would be extremely worthwhile; it will have to wait for future research, however. For our present purposes, it is enough to see that the setting effects are there at all. Acknowledgements The authors have had fruitful discussions with Patricia Cukor-Avila, Geneviève Escure, Carol Klee, Mike Linn, Dennis Preston, and Haj Ross, and portions of this article have been presented at the 2014 Annual Meeting of the Linguistic Society of America in Minneapolis, Berkeley Linguistics Society 40, Southeastern Conference on Linguistics 81 at Myrtle Beach, NWAV 43 in Chicago, Chicago Linguistic Society 52, and Departmental Colloquia at Bucknell University, the University of Minnesota Duluth, the University of Minnesota Twin Cities, and the University of North Texas. The authors are extremely grateful to our Belizean participants, who number far too many to list here. The authors would like to single out the following people for special thanks, however: Leela Vernon, Yvette Herrera, Anthony Brown, Ubaldimir Guerra, Gillian Flowers, Tracey Sangster, and Silvaana Udz, and to Nigel Encalada at the National Institute of Culture and History in Belize City and James Garber in the Department of Anthropology at Texas State University in San Marcos. References Abrams D. , Hogg M. . 1987 . ‘ Language attitudes, frames of reference, and social identity: A Scottish dimension ,’ Journal of Language and Social Psychology 6 : 201 – 13 . Google Scholar Crossref Search ADS Belize Census. 2010 . ‘Belize Population and Housing Census, 2010,’ available at http://www.statisticsbelize.org.bz/dms20uc/dynamicdata/docs/20110505004542_2.pdf. Accessed 15 May 2015. Belize National Standards. 2008 . ‘Belize National Standards and Curriculum Web for Language Arts,’ available at http://www.moe.gov.bz. Accessed 15 May 2015. Brown B. , Giles H. , Thakerar J. . 1985 . ‘ Speaker evaluations as a function of speech rate, accent, and context ,’ Language and Communication 5 : 207 – 20 . Google Scholar Crossref Search ADS Cargile A. , Giles H. , Ryan E. , Bradac J. . 1994 . ‘ Language attitudes as social process: A conceptual model and new directions ,’ Language and Communication 14 : 211 – 36 . Google Scholar Crossref Search ADS Carranza M. , Ryan E. . 1975 . ‘ Evaluative reactions of bilingual Anglo and Mexican American adolescents toward speakers of English and Spa ,’ Linguistics 13 : 83 – 104 . Google Scholar Crossref Search ADS Cooper R. 1975 . ‘ Introduction to language attitudes II ,’ International Journal of the Sociology of Language 6 : 5 – 9 . Creber C. , Giles H. . 1983 . ‘ Social context and language attitudes: The role of formality-informality of the setting ,’ Language Sciences 5 : 155 – 62 . Google Scholar Crossref Search ADS Decker K. 2005 . The Song of Kriol: A Grammar of the Kriol Language of Belize . The National Kriol Council . Garrett P. 2010 . Attitudes to Language . Cambridge University Press . Giles H. , Harrison H. , Creber C. , Smith P. , Freeman N. . 1983 . ‘ Developmental and contextual aspects of British children’s language attitudes ,’ Language and Communication 3 : 1 – 6 . Google Scholar Crossref Search ADS Giles H. , Ryan E. . 1982 . ‘Prolegomena for developing a social psychological theory of language attitudes’ in Ryan E. , Giles H. (eds): Attitudes Towards Language Variation . Edward Arnold , pp. 208 – 23 . Gómez Menjívar J. , Salmon W. . Forthcoming. Tropical Tongues: Attitudes, Endangerment, and Minority Languages in Belize . University of North Carolina Press . Greene L. 1999 . A Grammar of Belizean Creole . Peter Lang . Herrera Y. , Manzanares M. , Udz S. , Crosbie C. , Decker K. . 2007 . Kriol-Inglish Dikshineri/English-Kriol Dictionary . The National Kriol Council . Huygens I. , Vaughn G. . 1983 . ‘ Language attitudes, ethnicity, and social class in New Zealand ,’ Journal of Multilingual and Multicultural Development 4 : 207 – 23 . Google Scholar Crossref Search ADS Kristiansen T. 1997 . ‘Language attitudes in a Danish cinema’ in Coupland N. , Jaworski A. (eds): Sociolinguistics: A Reader and Coursebook . McMillan , pp. 291 – 305 . Kristiansen T. 2009 . ‘Investigating language in space: Experimental techniques’ in Auer P. , Schmidt J. (eds): Language and Space: Theories and Methods . Walter de Gruyter , pp. 528 – 49 . Kristiansen T , et al. 2011 . ‘Attitudes, ideology, and awareness’ in Wodak R. (ed.): The SAGE Handbook of Sociolinguistics . SAGE , pp. 265 – 78 . Ladegaard H. 1998 . ‘ National stereotypes and language attitudes ,’ Language and Communication 18 : 251 – 74 . Google Scholar Crossref Search ADS Lambert W , et al. 1960 . ‘ Evaluational reactions to spoken languages ,’ Journal of Abnormal and Social Psychology 60 : 44 – 51 . Google Scholar Crossref Search ADS PubMed Locke K. 2003 . ‘ Status and solidarity in social comparison: Agentic and communal values and vertical and horizontal directions ,’ Journal of Personality and Social Psychology 84 : 619 – 31 . Google Scholar Crossref Search ADS PubMed McKenzie R. 2008 . ‘ The role of variety recognition in Japanese university students’ attitudes towards English speech varieties ,’ Journal of Multilingual and Multicultural Development 29 : 139 – 53 . Google Scholar Crossref Search ADS Osgood C. , Suci G. , Tannenbaum P. . 1957 . The Measurement of Meaning . University of Illinois Press . Patrick P. 1999 . Urban Jamaican Creole: Variation in the Mesolect . John Benjamins . Preston D. , Prikhodkine A. 2013 . ‘Attitudes, variation, and language detail: Effects of specifying linguistic stimuli.’ Congrès International des Linguistes, available at: http://www.cil19.org/cc/en/abstract/contribution/1140/. Accessed 15 May 2015. Salmon W. 2015 . ‘ Language ideology, gender, and varieties of Belizean Kriol ,’ Journal of Black Studies 46 : 1 – 21 . Google Scholar Crossref Search ADS Salmon W. , Gómez Menjívar J. . 2016 . ‘ Language variation and dimensions of prestige in Belizean Kriol ,’ Journal of Pidgin and Creole Languages 31 : 316 – 60 . Google Scholar Crossref Search ADS Soukup B. 2012 . ‘ Current issues in the social psychological study of language attitudes: Constructionism, context, and the attitude-behavior link ,’ Language and Linguistics Compass 6 : 212 – 24 . Google Scholar Crossref Search ADS Soukup B. 2013 . ‘The measurement of language attitudes—A reappraisal from a constructionist perspective’ in Kristiansen T. , Grondelaers S. (eds): Language (De)standardization in Late Modern Europe: Experimental Studies . Novus , pp. 251 – 66 . Street R. 1985 . ‘ Participant-observer differences in speech evaluation ,’ Journal of Language and Social Psychology 4 : 125 – 30 . Google Scholar Crossref Search ADS Young C. 1973 . ‘Belize Creole: A study of the creolized English spoken in the city of Belize in its cultural and social setting,’ Ph.D. dissertation, York University. © Oxford University Press 2017 This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
Applied Linguistics – Oxford University Press
Published: Apr 1, 2019
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera