Differences in attitudes towards RP and GA

(1)

Department of English Bachelor Degree Project English Linguistics

Autumn 2018

Supervisor: Raphaël Domange

Differences in attitudes

towards RP and GA

A study of Swedish university students with

English as their L2

(2)

Differences in attitudes

towards RP and GA

A study of Swedish university students with English as their L2

Robert Vidén

Abstract

The purpose of this study is to broaden our understanding of the attitudes Swedish university students hold towards RP and GA. Although several studies have been conducted on the topic, none have sought to describe the three components traditionally assumed in the mentalist approach to language attitudes. The present study aims to bridge this gap. This was done by investigating students at Stockholm university using the verbal guise method. The present study found that students of Stockholm university with English as their L2 think that RP sounds more arrogant and boring, but also more responsible, gentle and intelligent than GA. The investigated population also felt more irritation while listening to RP than GA. They also felt more trust and liking for speakers of GA than of RP. Lastly, there were also differences regarding the accent the investigated population perceives it uses. GA was heavily favored in this respect. The findings of the present study could help to solve educational issues such as finding which English accents are suitable or most suitable for teaching. Keywords

(3)

1.0 Introduction

Attitudes have been the focus of a great deal of research in the social sciences. Research on this concept has been conducted by social psychologists since the 1920s (McKenzie, 2010, p. 19). Even though attitudes have been studied for a long time, the very foundation of attitude research such as what constitutes an attitude is still highly debated. In the early years of attitude studies, the difficulty in defining attitudes led some to argue that the concept of attitudes was unsuitable for scientific research (Bain, 1928, p. 940). In 1928, Bain listed reasons for sociology being ridiculed by the scientific community, and among these were “the inaccuracy, indefiniteness, and anarchistic confusion of our concepts. […] We cannot do valid scientific research until we know what we are talking about” (Bain, 1928, p. 942). According to Bain, attitudes are a good example of a concept that is ill-defined or even undefined, used in a pseudo-scientific manner (Bain, 1928, p. 942). In the contemporary debate, the question discussed is not whether we can study attitudes, but rather how we should do it (Garett, 2010, p. 20). According to Agheyisi and Fishman (1970), the definitions used in this domain of research often reflect certain research interests (p. 137). Therefore, certain definitions are more prone to be adopted in certain fields of attitude research. In language attitude research, which will be the focus of this study, attitudes are often defined as pro/con evaluations of certain language varieties and speakers of these (Carrie, 2016, p. 429). By using this type of pro/con definition, researchers often end up only measuring the sum of participants feelings towards the attitude object, even though attitudes are generally said to consist of beliefs and behavioral tendencies too (Agheyisi and Fishman, 1970, p. 137). The reason for using this pro/con definition of the term ‘attitude’ is most likely because of the many theoretical challenges that stem from its complexity (Garett, 2010, p. 20).

(6)

2

Considering that research in this domain has targeted only one of the three components of attitude, the purpose of the present study is to broaden the scientific community’s

understanding of the attitudes that Swedish university students with English as their L2 hold towards RP and GA. To accomplish this, 50 Swedish university students were asked to rate recordings of a GA speaker and an RP speaker. The findings suggest that the investigated population found RP more arrogant and boring, but also more responsible, gentle and

intelligent than GA. Also, the respondents expressed more irritation while listening to RP than GA, and felt more trust and liking for speakers of GA than of RP. Lastly, they perceived themselves as using GA to a larger extent than RP. The findings of the present study could help to solve educational issues such as which English accents are suitable or most suitable for teaching.

2.0 Background

2.1 What are attitudes?

According to Agheyisi and Fishman (1970), “[t]he concept ‘attitude’ has been variously defined and characterized by almost every theorist or researcher who has concerned himself with attitude studies” (p. 137). The definition used often reflects particular research interests, or the theoretical approach adopted (McKenzie, 2010, p. 19). In general, the term attitude refers to a social and socio-psychological concept, and historically, researchers of attitudes have adopted either a so-called mentalist approach or a behaviorist approach (McKenzie, 2010, pp. 19-21).

(7)

3

instance, some studies show certain individuals having a positive attitude towards an object while also displaying a negative behavior towards it (Garett, 2010, p. 6).

The mentalist approach views attitudes as a mental and neural state of readiness, thus implying “that they are not directly observable but have to be inferred from the subject´s introspection” (Agheyisi and Fishman, 1970, p. 138). Today, it is the principal theoretical approach to the investigation of attitudes (McKenzie, 2010, p. 21). The mental and neural state of readiness, now more readily defined as “an internal state of readiness” will affect the responses of the individual when aroused by stimulation of some sort” (McKenzie, 2010, p. 21). Also, mentalists view attitude as having a tripartite structure, with cognitive, affective and conative components (Carrie, 2016, p. 430). The cognitive component refers to an individual’s beliefs about X. The affective component refers to an individual’s feelings towards X and the conative component refers to an individual’s perception of their behavioral tendencies towards X (McKenzie (2010, p. 22). However, it should be noted that all three components are not necessarily represented in an attitude towards a certain entity (McKenzie (2010, p. 22). One advantage to using the mentalist approach instead of the behaviorist approach is that attitude “still remains an independent variable in the form of a latent

psychological constant which is not tied to the specific external stimulus situations in which the responses are made” (Agheyisi and Fishman, 1970, p. 138). Therefore, it allows

researchers to use their findings to explain how the investigated population might think, feel and behave in other similar situations (Alexander and Norman, 1966, pp. 280-281). This is arguably a major advantage, however, using the mentalist approach does come with several challenges that will be discussed further in the context of language attitude studies.

(8)

4

Attempts to standardize the tools for evaluation of language attitudes have been made by Zahn and Hopper (1985). Statements from several different studies were tested in regard to how meaningful they were for undergraduate university students in the southwestern United States. All the statements were designed to measure the cognitive aspect of attitudes in a certain language variety, so even though attitudes were never defined explicitly, it is evident that attitudes were thought of as consisting solely of beliefs in this study. The 30 most

meaningful statements were then chosen to constitute the new instrument for the evaluation of language (Zahn and Hopper, 1985, pp. 118-119). The attempt was arguably unsuccessful for two reasons. Firstly, the 30 statements were by no means exhaustive regarding all the

different beliefs one can have about language. Therefore, it is likely that some researchers would want to know about beliefs that are not reflected in these statements. Secondly, Garett (2010, p. 56) argues that there is no universal set of statements that is applicable in the study of every population since different sets of statements are meaningful to different populations. Therefore, even if researchers who equate attitudes with belief or simply want to study beliefs in language use these statements, some of those statements will most likely not be meaningful to their investigated population. Because of this, it is unlikely that a standardized instrument for language attitude research will ever be accepted by the scientific community.

The challenge of defining the object of study in language attitude research appears to have a long history. Agheyisi and Fishman (1970) argue that even though attitudes are said to include three components, researchers usually only measure one of them, this being the affective component (p. 139). Not including all three components could be a way of avoiding a significant problem in the mentalist approach. According to Agheyisi and Fishman (1970), a major criticism of the multi-component view of attitudes is the impossibility of determining the organization and interrelation of the various attitude components an individual hold to a certain attitude object (p. 139). By reducing attitudes to one component, the problem is avoided.

Defining attitude as a mono-component concept has been the subject of some contemporary critique. Carrie argues that:

(9)

5

Instead of using a monolithic definition of language attitude, Carrie suggests the following definition: “[O]ne’s evaluation of and disposition towards a speech variety and its speakers, consisting of thoughts, feelings and behavioural tendencies” (2016, p. 430). This definition is for speech variety and speakers, but these can be replaced with any other attitude objects. It should be noted that Carrie´s definition brings nothing new into the theoretical framework of language attitudes but derives ultimately from the mentalist approach to attitudes. However, by using it herself and presenting it, she hopes to offer a more theoretically informed

definition of language attitude by including attitude’s tripartite structure into the definition (2016, p. 443).

Despite differences in how researchers define their object of study, the large amount of research still allows for some generalizations to be made in the field of language attitude research. For instance, if one study has measured the affective part of Y’s attitude towards X, this result can be compared with another study that has researched the cognitive parts of Y’s attitude towards X. Therefore, the ability to do this is largely dependent on the amount of research done about the attitude object. As an example, evidence suggests that depending on which English accent is perceived as the standard in a certain population, the attitudes of this population towards different English accents differ (Carrie, 2016, p. 430). Also, in an EFL context, which will be the focus of this study, there seems to be a general preference for RP in Europe and a preference for GA outside of Europe (Carrie, 2016, p. 433). In the following section, the practical value of such language attitude generalizations will be explored. Also, some previous language attitude research of students with English as their L2 will be discussed.

2.3 Previous studies on language attitudes in second language research

Carrie (2016) investigated the attitudes Spanish university students with English as their L2 hold towards RP and GA. Most notably, she found that speakers of RP were thought to be more competent than speakers of GA. However, GA speakers were thought to be more

(10)

6

while listening to them (Carrie, 2016, p. 440). Even so, this study is important for the present study in several ways. Firstly, some of the research instruments she used will be reused in this study. Secondly, the present investigation relies on the same working definition of language attitudes. Moreover, lastly, by comparing the findings of the present study with studies such as Carrie´s, generalizations can be made which constitute the practical value of such research.

[The] language variety employed by a given speaker can have wide-ranging social implications, including influencing job hiring and career progression, teachers’ perceptions of their students’ educational abilities and the perceived persuasiveness and credibility of the message itself. (McKenzie and Carrie, 2018, pp. 3-4)

This passage illustrates several fields in which language attitude research could be considered valuable. If we can discover more about attitudes towards language varieties or accents, we can also use this knowledge to influence the many areas that it affects. Furthermore, in an EFL context, Carrie argues that language attitude studies provide invaluable insight regarding the status of English, but also the preferences of EFL learners (2016, p. 429). The value of such insight is arguably important in countries where English is used in the educational system.

(11)

7

Modiano (1993) asked 125 Swedish university students which English accent sounded most pleasant to the ear. 64 of them answered RP, and 45 answered GA (pp. 44-45). He returned to the topic in 2002, this time asking Swedish upper secondary school students which English accent they preferred. 61.3% answered that they preferred GA and 33.3% of the student’s preferred RP (Modiano and Söderlund, 2002, p. 152). The shift in students´ preference from RP to GA is seen as “striking evidence of the ongoing Americanization” (Modiano and Söderlund, 2002, p. 152). This shift that these two studies supposedly show is somewhat questionable. The questions used in the questionnaires do not match, and since “which accent do you find most pleasant to the ear” is not equivalent to “which accent do you prefer”, the comparison between them may be found objectionable. Also, the findings from these studies arguably say little about the participants´ attitude towards the two accents. The details of the participant’s attitude are lost when not looking at a specific aspect of the attitude, i.e., the cognitive, affective or conative. Even though the studies carried out by Modiano has some validity issues, the findings still provide some evidence of a significant change in Swedish university students preference from RP towards GA.

Axelsson (2002) also investigated Swedish university students’ attitudes towards RP and GA by asking 149 participants about the accent they preferred and why. 54% answered that they preferred RP and 25% answered that they preferred GA. The answer to which accent they preferred and why was to be specified on a single row on the questionnaire, thus leaving only a short space to answer why they preferred the chosen accent (Axelsson, 2002, p. 146). The short answers are evident in the results of the study where she lists some of the reasons advanced by the participants (Axelsson, 2002, p. 143). Apart from a possible problem in not giving the participants enough space to provide an elaborated answer, Axelsson does not specify how many participants agree with each observed attitude. Also, she does not specify if she excluded any answers. Furthermore, she claims that she had investigated the attitudes towards these accents in 1992 and that “no student had anything positive to say about AmE” (Axelsson, 2002, p. 142), referring to GA accent. This statement is interesting since Modiano found that 45% of Swedish university students preferred GA in 1993. As to the results of her study from 2000, the data also conflicts with what Modiano found in 2002.

(12)

8

the fairytale in RP, and 71% read the word list in RP (Mobärg, 2002, p. 121). From this result, Mobärg concluded that the students associated GA with informal speech styles and RP with formal speech styles (Mobärg, 2002, p. 128). This corresponds to the traditional Labovian approach to style variation as a function of attention paid to speech. Although this method has proved effective in sociolinguistics, this approach has definite behaviorist underpinnings (Gal, 2016).Although, as mentioned above, behaviorist methods have been largely rejected in attitude studies, those results should not be dismissed fully. According to Carrie, the

psychological process of similar attraction prompts individuals to express solidarity towards an accent through speech (2016, p. 441). Since Mobärg finds most speakers to pronounce the words in RP, this should indicate that they feel more positively towards RP.

A more recent study investigated Swedish upper secondary school students´ attitudes towards four different accents, including RP and GA. Attitudes were inferred by having the

participants listen to recordings of the different accents and asking them to rate the speakers on 22 different traits (Sahlström, 2005). Since she had the participants rate the speakers on these many different traits, the detail in her findings was relatively large. Unfortunately, only a few of these cognitive traits are the same as the ones I will be testing in this study. It may nevertheless prove a useful source for comparison for the results of the present study.

3.0 Research questions

(13)

9

Do Swedish university students with English as their L2 have different attitudes towards RP and GA? If yes, what are the differences?

The working definition of ‘language attitude’ in this study will be the same as the one that Carrie uses in her study. “[O]ne’s evaluation of and disposition towards a speech variety and its speakers, consisting of thoughts, feelings and behavioural tendencies” (2016, p. 430). A definition such as this one acknowledges the complexity of language attitudes and using it should, therefore, produce results that reflect language attitude in more detail than the monolithic definitions do. The method used to investigate this is described in the following section.

4.0 Method

According to McKenzie and Carrie (2018), the verbal guise method has been widely used in language attitude research and has shown to provide a high degree of consistency concerning the data collected (p. 3). This method constitutes a methodological framework for collecting data in language attitude research. Respondents in this method are first asked to listen to a recording at least one time. In research about accents, each recording consists of a different individual talking in the investigated accent. After that, the respondent is asked to fill in attitude rating scales concerning certain statements about the speaker in the recording (Garett, 2010, p. 41). The result produced from having the respondent answer the attitude rating scales are thought to represent the attitude towards the accent and its the speaker (McKenzie, 2010, p. 46). The following section is a detailed account for what the data is, how it was collected, what the tools for collection are and how they were designed.

4.1 Data

(14)

10

of attitude as this study. This similarity arguably lowers the risk of the definition or theoretical approach conflicting with the research instrument. It is important to note that the statements included in this questionnaire should not be thought of as exhaustive. For instance, a

participant might have strong feelings towards the tested accent, involving sentiments that have not been tested in this study.

Furthermore, a section asking for background information was added to the questionnaire. This section was added to ensure that the sample size would not consist of a hidden overrepresentation of a certain type of student. Such an overrepresentation could take the form of a bias towards a certain gender or age. However, also, in qualities that could affect responses to a great extent, such as having lived in English-speaking countries or attended English speaking schools. The questions regarding whether the participants have Swedish as their first language and whether they are a Swedish citizen were added as a formality, the participants were asked these questions before partaking in the test to avoid unnecessary testing.

The recordings that were used in collecting the data were not adopted from Carrie (2016). However, in order to have increased comparability with the results, an attempt was made to adopt the recording she used. Unfortunately, since I was not able to obtain speaker consent, this was unsuccessful. Therefore, the recordings were selected from other studies. Two recordings were used, one with spoken language in RP and one with GA. The criteria for choosing these recordings were that they should be in in the appropriate accent and that the speakers should be of the same sex. Also, attention was paid to the fact that they should be comparable in terms of quality, speech rate, and volume. The RP speech was selected from Brulard, Carr, and Durand (2015) and the GA from Hughes, Trudgill and Watt (2012). Since the topics of conversation in these recordings could potentially influence the responses of the participant, shorter excerpts were selected to increase the neutrality of the topic. The RP recording chosen is 29 seconds long, and the GA recording is 30 seconds. Both recordings have female speakers.

(15)

11

research. The data was collected in the campus library because it attracts students from different faculties.

Before the data was collected, the research instruments were tested. Dörnyei argues that one should always pilot the research instruments before launching a project. This is done to ensure a high-quality outcome within a specific context (2016, p. 75). This was done on 5 individuals that match the target population of this study. The data collected from this pilot study showed two alarming concerns. Firstly, two participants felt that it was difficult to agree or disagree with the statement about whether they felt that they could identify with the speaker. This was because the statement was difficult to understand. The statement about identification was not removed before collecting the data since an option for “I am not sure” is available for the participant. Secondly, a slight humming was reported to be disturbing in the background of the GA recording. Therefore, the GA recording did not match the criteria of being comparable to the RP recording in terms of sound quality. The humming reported in the GA recording was reduced to a great extent by the supervisor of this project, thus eliminating the issue. The collection of the data was conducted in the following manner: students in Stockholm university main library were arbitrarily asked to participate in the study and selected if they matched the criteria for this study and agreed to participate. They were then led to one of the group rooms in the library where the test was conducted. Firstly, the participants were asked to answer the background information sheet. Secondly, in accordance with the verbal guise method, students were first exposed to either the RP or GA recording twice in a row. Whether the GA or RP recording was played first alternated between each participant. Carrie argues that listening to a recording of this length twice is sufficient exposure for participants to be able to evaluate the speaker (2016, p. 435). After being exposed to the first recording twice, students were asked to respond to the various statements in the questionnaire. After

responding to these statements, this procedure was repeated, but now with the unheard recording. Lastly, participants were asked to read and sign a consent form.

Data was collected from 50 participants. Dörnyei argues that sample sizes should be thought of as the larger, the better (2016, p. 99). However, the rule of thumb seems to be a sample size of at least 30 in quantitative studies in order to achieve normal distribution (Dörnyei, 2016, pp. 99-100). A sample size of 50 was chosen since it matched this criterion and was

(16)

12

distribution of 20 males and 30 females. This choice was made to better reflect the gender distribution of students as reported by Stockholm University (Sifferfakta, 2017)

The data mostly consists of responses to the various statement about the speaker in the recordings. These responses are either “I am not sure” or to what degree participants agree with a certain statement on a 7-point scale. Also, some data consists of responses to the background questions posed in the questionnaire. In the next section, I will describe how this data was analyzed.

4.2 Analysis

The analysis of the data is fairly simple. Responses to each statement designed to measure the affective and conative component of attitude generate a score according to the form described by Dörnyei, i.e. “’strongly agree’ = 7 … ‘strongly disagree’ = 1” (2016, p. 105). Data on the cognitive component generate a score according to the same form as the Likert scales, even though it is a semantic differential scale, i.e. ‘responsible’ = 7… ‘not responsible’ = 1. The response “I am not sure” is not counted. The mean score calculated from the responses of a statement either reflect a part of how the investigated population thinks, feels or perceive as behaving in regard to the investigated accent. These scores are then compared to find

differences in the students’ attitudes to RP and GA.This type of score system is used in Carrie (2016) and using it in the present study will therefore increase the comparability of the

findings.

5.0 Results and Discussion

The results and discussion presented in this section are subdivided into four different sections. These sections are Cognitive evaluation, Affective evaluation, Conative responses, and

(17)

13 5.1 Cognitive evaluations

This section presents the results for the cognitive part of the questionnaire. As stated earlier, the cognitive component measures beliefs. The beliefs investigated in this study are

perceptions or prejudices of and information about RP and GA. As previously mentioned, the traits selected to represent this component of attitude should not be regarded as exhaustive. Instead, each trait should be thought of as measuring a certain belief.

Figure 1.

Figure 1 illustrates the mean of each cognitive trait tested in the questionnaire. The blue bars represent the evaluation of RP, and the grey bars represent the evaluation of GA. Even though the questionnaire from Carrie (2016) was used in this study, the results are not presented in the same manner. Carrie calculated the mean score out of all tested cognitive traits, whether they are perceived as negative or not (2016, pp. 436-437). Therefore, it is difficult to

(18)

14

form (not arrogant and not boring) as in figure 1, the mean score of all the tested cognitive traits better reflect which accent is most favored overall in terms of students’ cognitive part of attitude. Also, Figure 1 provides the mean and standard deviation for all cognitive traits tested combined.

The mean of all tested cognitive traits is 5.578 for the GA accent and 5.288 for the RP accent. This result suggests that Swedish university students with English as their L2 think of GA in slightly more positive terms than RP. Furthermore, the standard deviation is 0.978 RP and 0.618 for GA. The standard deviation reflects a difference in spread between the evaluated traits related to each accent. As shown in Figure 1, the mean score of the traits responsible (6.1), calm (6.1) and gentle (6.0) was more than any mean score generated by the cognitive evaluation of GA. However, some mean scores that reflect the cognitive evaluation of RP were also far below that of any mean score that reflects the cognitive evaluation of GA, these being the traits ‘not arrogant’ (3.5) and ‘not boring’ (4.1).

As Figure one illustrates, there is a difference to some degree in every investigated trait. Since little research has been done on Swedes´ attitude to RP and GA that translates well into the findings of the present study, it is difficult to say whether the differences found in the present study are expected or surprising. Even so, there is some previous research done in Sweden that is comparable. Sahlström (2005) tested several cognitive traits that were to a large degree different from what has been tested in this study. However, she did test arrogance,

(19)

15

secondary students. Despite this distance in population, the findings of Sahlström do support a prejudice towards the RP accent as sounding more arrogant than GA.

Some other notable differences in beliefs about RP and GA are that RP speakers are considered to be more responsible (0.7-point difference) and gentler (0.8-point difference) than GA. Also, GA speakers were thought of as less boring than RP speakers with a 1.1-point difference. However, there is no previous research done on Swedish students that can support these findings.

5.2 Affective evaluations

This section presents the results for the affective part of the questionnaire. As stated in the background section, the affective component measure feelings. In short, the recordings have a different effect on different people. The findings in this section reflect which feelings are experienced after listening to either RP or GA speech and to what degree this feeling is felt. However, just like the cognitive evaluation of this study, the sentiments tested are not to be thought of as exhaustive for every feeling that can be felt.

(20)

16

Figure 2 illustrates the mean of each affective sentiment that was tested in the questionnaire. As in Figure 1, the blue bars represent the evaluation of RP, and the grey bars represent the evaluation of GA. Also, Figure 1 provides the mean and standard deviation for all affective sentiments tested combined. The mean score for RP is 3.886 and for GA 3.943. Furthermore, the standard deviation is 1.634 for RP and 2.088 for GA.

As shown in figure 2, the result from the question about whether students felt that they could identify with the speaker was removed. As discussed in the background section, different sentiments are meaningful to different populations. The sentiment of identity seems not to be meaningful in a Swedish context. Seventeen participants chose the option of responding “I am not sure” to this question. This finding could suggest that they have no formed attitude

regarding this specific statement. However, since some of the participants of the pilot study stated that they found the sentiment difficult to understand, it is arguably more likely that this is the case.

The mean of all these sentiment ratings tells us little about any preference for either GA or RP. The tested sentiments that together generate this score are a mix of positive and negative ones, such as ‘Liking’ and ‘Feeling bored’. For instance, if most of the investigated

population express liking for an accent, it is unlikely that they would express that they feel bored by it. Therefore, this mix is also the cause of the relatively high standard deviation, since data become more spread out.

On looking further, the results show some interesting differences. The participants felt that they liked the GA speaker more than the RP speaker, there was an average 1,2-point

difference in evaluation regarding this sentiment. The GA speaker was also evaluated as more trustworthy than the RP speaker; there was a 1.4-point difference in the evaluation. Another significant difference was that the RP speaker was considered to be more irritating than the GA speaker, this is illustrated by the 1.3-point difference in evaluation regarding this

sentiment. These differences found do show some preference towards the GA accent in terms of trust, liking and being less irritating. This result conflict with the findings of Mobärg

(21)

17 5.3 Conative responses

This section presents the results for the conative part of the questionnaire. As stated earlier, the conative component measures self-perceived behavioral tendencies. In this case, it

measures, how the investigated population perceives themselves to behave towards RP or GA. There are many types of possible self-perceived behavioral tendencies one can have to RP and GA. The investigation of this component is only done on two specific ones and is not thought to be representable for all of them.

Table 1.

When speaking English, I have a pronunciation

similar to speaker X. Mean SD

RP 3.38 1.778

GA 5.37 1.916

Table 2.

When speaking English, I make a conscious effort to have a pronunciation

similar to speaker X. Mean SD

RP 2.98 1.827

GA 3.48 1.735

The mean and standard deviation of the conative responses are reported in Table 1 and 2. Table 1 represents the results from the first question posed in the conative section of the questionnaire. The mean score for the responses to this question is 3.38 for RP and 5.37 for GA. The standard deviation of these responses is 1.778 for RP and 1.916 for GA. Table 2 represents the second and last question. The mean score for the responses to this question is 2.98 for RP and 3.48 for GA. The standard deviation of these responses is 1.827 for RP and 1.735 for GA.

(22)

18

shows that a majority of Swedish students report that they speak GA and not RP (Pettersson, 2008, pp. 11-17).

At first glance, it might appear as if the investigated population expressed a relatively neutral stance to whether they made a conscious effort to have a pronunciation similar to the RP and GA speaker. The mean scores generated from the responses to this question were 2.98 for RP and 3.48 for GA, with 3.5 being a neutral position. However, most of the participants did not express a response close to 3.5 regarding this question. In the responses to this question, we see a high amount of standard deviation. As reported in table 2, the standard deviation for the responses to this question was 1.827 for RP and 1.735 for GA. Because of this high spread in data, the findings indicate that that the majority of the investigated population do not hold a neutral stance regarding these statements about perceived behavioral tendencies. Instead, the findings suggest that the investigated population holds conflicting perceptions regarding these statements. Whether a majority hold a negative or positive stance is inconclusive from this result. Since no previous research in Sweden has investigated this question of conscious effort, more research is needed to support the findings of the present study.

5.4 Further discussion

Some notable findings in this study suggest that students of Stockholm university with English as their L2 think RP sounds more arrogant and boring, but also more responsible, gentle and intelligent than GA. The investigated population also felt more irritation while listening to RP than GA. They also felt more trust and liking for speakers of GA than of RP. Lastly, there was also a difference in which accent the investigated population perceived that they used themselves. GA was heavily favored in this respect.

As previously mentioned, Carrie (2016) investigated the same traits, sentiments, and self-perceived behavioral dispositions as the present study. Carrie used both male and female speakers to represent each accent. Since the present study only had recordings representing female speakers, comparisons will only be made with the evaluation of the female speakers in Carrie (2016).

(23)

19

(Carrie, 2016, p. 337). Her findings suggest that both RP and GA is thought to be arrogant and boring but did not find evidence that one accent is thought of as more boring or arrogant than another (Carrie, 2016, p. 337). In contrast to this, the present study found that the investigated population did not think of these accents as boring or arrogant, but still thought RP sounded more arrogant than GA. Also, the present study found that RP was believed to sound more responsible, gentle and intelligent. Although Carrie (2016) did not find any difference in belief that RP sounded more responsible, she did find that RP was thought to sound gentler and more intelligent than GA (p. 337).

As for the affective component of attitude, the present study found that not much irritation was felt when listening to RP or GA, but that more irritation was felt when listening to RP than GA. This finding was not supported by Carrie (2016) who found that a high degree of irritation was reported when listening to both accents, more precisely, a mean score of 6 out of 7 was generated by the responses to this sentiment (p. 440). With 7 reflecting a mean of ‘Strongly agree’, this is arguably a strong reaction.

Regarding the conative component of attitude, the present study found that the investigated population perceived themselves to use GA to a greater extent than RP. Carrie (2016) found no such difference (p. 441).

In conclusion, the most notable findings of the present study contrast with those found by Carrie (2016). The only notable findings of the present study that are supported by Carrie (2016) is that RP was thought to sound gentler and more intelligent than GA. Furthermore, the overall results of the present study arguably conflict with the previously mentioned

generalization that Europeans prefer RP to GA. However, as we have seen, language attitude research can tell us more than which accent is preferred or not preferred. Although RP was not preferred overall, the findings of the present study suggest that RP was preferred

(24)

20

6.0 Conclusion

The present study has been an investigation of the differences in attitude that Swedish university students hold towards RP and GA. The purpose was to broaden the scientific community´s understanding of this population regarding their attitudes to the accents in question. This was successful. By using the 3-component model in the investigation of these attitudes, the findings of the present study are in greater detail compared to the findings of previous studies on the topic. The investigation conducted was on students at Stockholm University. The position of this university as one of the largest in Sweden and that it is in the capital city increase the representativeness of the findings. However, it is questionable whether these results are representative for all of Sweden. The lack of nationwide representativeness is partly due to the extent of this study, but partly since most of the

previous research that exists on this topic does not translate well to the findings of the present study. Therefore, any strong nationwide generalizations are not possible to make from these findings. However, some findings could be supported by previous research, such as the findings that RP is thought to sound gentler, more intelligent but also more arrogant.

Furthermore, the finding that more students perceive themselves to speak GA than RP was in accordance with previous research in a Swedish context.

Some notable findings could not be supported by previous research, but are of value to resolve educational issues, so should be further investigated. As mentioned in the background section of the present study, negative comments from students of Stockholm University regarding the English accents of their teachers prevail across different faculties and

(25)

21

References

Agheyisi, R., & Fishman, J. A. (1970). Language Attitude Studies: A Brief Survey of Methodological Approaches. Anthropological Linguistics, 12(5), 137-157.

Alexander, C. Norman Jr. (1966). Attitude as a Scientific Concept. Social Forces, 45(2). 278-81. Axelsson, M.W. (2002). “Refined” or “Relaxed” English Pronunciation: Usage and Attitudes among

Swedish University Students. In M. Modiano (Ed), Studies in Mid-Atlantic English (pp. 132-146). Gävle : Högskolan i Gävle.

Bain, R. (1928). An Attitude on Attitude Research. American Journal of Sociology, 33(6), 940-57. Brulard, I, Carr, P. & Durand, J. (Eds.). (2015). La Prononciation de l’anglais contemporain dans le

monde: Variation et structure. Toulouse: Presses Universitaires du Midi.

Carrie, E. (2016). ‘British is professional, American is urban’: Attitudes towards English reference accents in Spain. International Journal of Applied Linguistics, 27(2), 427-447. doi:10.1111/ijal.12139

Dörnyei, Z. (2016). Research methods in applied linguistics: Quantitative, qualitative, and mixed

methodologies. Oxford: Oxford University Press.

Gal, S. (2016). Labov in anthropology. Journal of sociolinguistics, 20(4), 453-463. Garett, P. (2010). Attitudes to Language. Cambridge: Cambridge University Press.

Hughes, A., Trudgill, P. & Watt, D.J.L. (2012). English accents & dialects: an introduction to social

and regional varieties of English in the British Isles. (5th ed.) London: Hodder Education.

Kingsley Bolton & Maria Kuteeva. (2012) English as an academic language at a Swedish university: parallel language use and the ‘threat’ of English. Journal of Multilingual and Multicultural

Development, 33:5, 429-447.

McKenzie, R & Carrie, E. (2018), Implicit–explicit attitudinal discrepancy and the investigation of language attitude change in progress. Journal of Multilingual and Multicultural Development, 39(9), 830-844.

McKenzie, R. (2015). UK university students’ folk perceptions of spoken variation in English: the role of explicit and implicit attitudes. International Journal of the Sociology of Language, 2015(236), 31-53.

McKenzie, R. M. (2010) The social psychology of English as a global language: attitudes, awareness

and identity in the Japanese context. London: Springer.

Mobärg, M. (2002). RP or GA? On Swedish School Students´ Choice of English Pronunciation. In M. Modiano (Ed), Studies in Mid-Atlantic English (pp. 119-131). Gävle : Högskolan i Gävle.

Modiano, M. (1993). American English and higher education in Sweden. American studies in

Scandinavia, 25(1), 37-47.

Pettersson, J. (2008). British, American or Mid-Atlantic English: What accent do Swedish learners use

and where do they get their influences from?. Karlstad: Karlstads universitet.

Sahlström, C. (2005). Upper Secondary Students’ Assessment of Four Women Speaking Four Different

Varieties of English. Karlstad: Karlstads universitet.

(26)

22

Söderlund, M. & Modiano, M. (2002). Swedish Upper Secondary School Students and their Attitudes Towards AmE, BrE and Mid-Atlantic English. In M. Modiano (Ed), Studies in Mid-Atlantic English (pp. 147-171). Gävle : Högskolan i Gävle.

Zahn. C. J & Hopper. R. (1985) Measuring Language Attitudes: The Speech Evaluation Instrument.

(27)

23

Appendix 1 Background information

Age __________

Gender ________________________

How long have you studied at the level of university? ___________________________

What have you studied/what do you study at university level?

___________________________________________________________________________

Is Swedish your first language? Yes

□

No

□

Are you a Swedish citizen? Yes

□

No

□

Have you lived in an English-speaking country? Yes

□

No

□

If yes, which one/ones? ________________________________________________________

(28)

24

Part one

responsible not responsible I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

not serious serious I'm not sure

I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

confident not confident I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

not arrogant arrogant I'm not sure

calm not calm I'm not sure

(29)

25

kind not kind I'm not sure

not intelligent intelligent I'm not sure

boring not boring I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

When I listen to Speaker One, I feel…

strongly agree strongly disagree I'm not sure that I like him/her ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

that I trust him/her ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

bored ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

(30)

26

relaxed ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

interested ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

overwhelmed ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

irritated ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

When speaking in English, I have a pronunciation similar to Speaker One.

certainly false certainly true

❍ ❍ ❍ ❍ ❍ ❍ ❍

When speaking in English, I make a conscious effort to have a pronunciation similar to Speaker One.

certainly true certainly false

(31)

27

Part two

responsible not responsible I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

not serious serious I'm not sure

confident not confident I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

not arrogant arrogant I'm not sure

calm not calm I'm not sure

(32)

28

kind not kind I'm not sure

not intelligent intelligent I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

boring not boring I'm not sure I think Speaker One sounds ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

strongly agree strongly disagree I'm not sure that I like him/her ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

that I trust him/her ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

bored ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

(33)

29

relaxed ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

interested ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

overwhelmed ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

irritated ❍ ❍ ❍ ❍ ❍ ❍ ❍ ❍

When speaking in English, I have a pronunciation similar to Speaker One.

certainly false certainly true

❍ ❍ ❍ ❍ ❍ ❍ ❍

When speaking in English, I make a conscious effort to have a pronunciation similar to Speaker One.

certainly true certainly false

❍ ❍ ❍ ❍ ❍ ❍ ❍

(34)

30

Jag heter Robert Vidén och är student vid Stockholms universitet. Just nu håller jag på med en studie om svenska universitetsstudenters attityder till engelska accenter. Detta brev innehåller information om min studie och vad det innebär att delta.

Studiens syfte och genomförande

Studien handlar attityder till engelska accenter. Du har nu fått lyssna på inspelade konversationer som är tänka att reflektera dessa dialekter. De svar du angivit i relation till dessa inspelningar kommer att analyseras för att vi ska kunna bilda en uppfattning om din attityd till dessa accenter.

Etik och sekretess

Studien följer noga de etiska föreskrifter som gäller för god forskningssed. Allt insamlat material behandlas med största aktsamhet och förvaras på ett säkert sätt. Materialet kommer bara att användas i forskningssyfte och samtliga medverkande elever och lärare, samt skola, kommer att vara anonyma i de sammanhang där studien presenteras och publiceras. Om du har frågor eller önskar mer information är du välkommen att kontakta mig på mejl.

Student Handledare

robban9406@hotmail.com raphael.domange@english.su.se

Medgivande

Genom att kryssa för ”Ja” och skriva under detta dokument du ditt medgivande till att informationen får användas till denna studie enligt ovanstående beskrivning. Ett ”Nej” innebär att informationen inte används.

Ja, jag tillåter att informationen jag lämnat får användas i denna studie. Nej, jag tillåter inte att informationen jag lämnat får användas i denna studie.

_____________________________________ namn

_____________________________________ underskrift

(35)

(36)

Differences in attitudes towards RP and GA

Differences in attitudes

towards RP and GA

A study of Swedish university students with

English as their L2

Differences in attitudes

towards RP and GA

Abstract

Contents

1.0 Introduction

2.0 Background

3.0 Research questions

4.0 Method

5.0 Results and Discussion

6.0 Conclusion

References

Appendix 1

Background information

□

□

□

□

□

□

Part one

Part two