Preparing The Study

52

The VGT was the chosen technique in this study for several reasons. Firstly, the time frame was limiting.

Ideally one could have tried to find a person who was able to mimic all English accents, as well as the Swedish language flawlessly. This was not considered doable during the time frame of the study.

Secondly, even if applying the MGT and thus finding a person who was fluent in all languages, the authenticity of the speaker would have been limited since the languages originate from different countries.

By choosing to apply VGT it was possible to select four different speakers with similarities in voice qualities. The advantages of VGT outweighed the disadvantages for the purpose of this study.

53

can be found in Appendix 8 and the reasoning behind the choice of each of the survey components will be discussed below.

4.2.1. QUESTIONS ON ATTENTION

In order to unveil any effects of language variety on attention given to the commercial and its message, the short-term memory was tested. Paying attention is a necessary process the consumer must undergo before (s)he can respond to the commercial in terms of brand liking and purchase decision. Even if this study used only standard accents, it is of interest to control for attention effects. Therefore a standard short-term memory test was included in the survey. Short-term memory lasts for up to 20 seconds and is achieved when attention is paid to a sensory stimuli. (Solomon, 2015). Therefore, the respondent could be assumed to have given more attention to the commercial if (s)he has a better short-term memory of it.

Because of the short-term memory’s small time-span, this element was placed first in the survey, and consumer attention was tested by asking questions about informational pieces of the commercial; what the name was of the advertised brand, and for how long it had existed. Attention was measured by providing fixed answers, out of which one was the correct one. The option “don’t know” was also included.

4.2.2. QUESTIONS ON BRAND PERCEPTION

Brand personality

The first part of the brand perception component of the empirical study was brand personality. In order to measure brand personality, Aaker’s (1997) Brand Personality Scale was employed. Jennifer Aaker is considered the founder of the brand personality theory and her framework is still, about two decades later, frequently applied (Ahmad & Thyagaraj, 2014). Aaker’s brand personality scale is built up by five brand personalities, each containing a number of sub-traits; sincerity (down-to-earth, honest, wholesome, cheerful¹), excitement (daring, spirited, imaginative, up-to-date), competence (reliable, intelligent, successful), sophisticated (upper class, charming) and ruggedness (outdoorsy, tough) (Aaker, 1997). In

1 In the sincerity personality, the sub-trait cheerful was exchanged to friendly. This because the direct translation of cheerful to Swedish was considered irrelevant and not being sincere. Since friendly, according to Aaker (1997), in turn is a sub-trait to cheerful, this was used in this study.

54

order to test the brand personality, the respondents were asked about all sub-traits in parentheses above, 15 in total. All traits were measured using a five-point Likert scale (see section 2.4.5.)

Source credibility

With the logic of brand personality, i.e. that human characteristics can be transferred to a brand (Aaker, 1997), it was of interest to investigate if the concept of source credibility could also transmit to a brand.

Hence, source credibility was measured as a part of brand perception in the empirical study. Ohanian (1990) suggest that source credibility can be tested through the credibility traits trustworthiness, expertise and physical attractiveness, and the two former have been applied in this study. Trustworthiness includes the sub-traits credible, honest, reliable, sincere and trustworthy. As for expertise, the traits are expertise, experience, knowledge, qualifications and skill. The last dimension physical attractiveness, however, cannot be mimicked since the spokesperson will not be visible in the commercial shown to respondents.

This means that Ohanian’s (1990) original attractiveness traits can then not be applied. Instead, like many other scholars (Tsalikis et al., 2013) social attractiveness is used by measuring the sub-traits friendly, communicate status, humble, cheerful, charming, nice and attractive. All source credibility traits were measured using a five-point Likert scale.

Likability

Because brand equity implies consumers’ favourable associations towards a brand (Kapferer, 2008) it was of interest to test if the different commercials lead to different levels of brand likability. Likability is a subjective view on the brand, and was therefore considered part of the brand perception concept and was tested in the survey by asking the respondent to rate the statement; “I like this brand”. The question was measured on a five-point Likert scale.

4.2.3. QUESTIONS ON PURCHASE INTENTION

As Aaker (1991) points out in his book on brand equity, the intangible value of a brand is favourable if it is converted into financial equity and becomes realised profit for the company. This means that brand equity is of little value if it does not affect the bottom-line. In order to see if the choice of language variety in commercials can have a realised financial value for a company, a section of purchase intention was

55

included in the survey. This element was an adapted form of Ohanian’s (1990; 1991) research on source credibility and purchase intention and included the questions; “Is it likely that you would look for more information about this brand?”, “Would you consider buying a product of this brand?” and “Would you purchase a product of this brand as a gift?”. In addition to these, a question on likelihood of recommendation was added (“Is it likely that you would recommend this brand?”). This because word-of-mouth is easily accessible information, especially considering the emergence online social networks, which can affect consumers purchase decision greatly (Jalilvand & Samiei, 2012). All four questions were measured on a five-point Likert scale, although with different measurements (likelihood or level of agreement) to match the wording of the question.

4.2.4. QUESTIONS ON PERSONAL FACTS

Relationship to accent

There were two reasons for testing respondents’ relationship to the English accent they listened to. The first, to which most question of the element belonged, tested if familiarity with an accent affected the brand perception. Whether there is a correlation or not has been discussed in previous research on accent and brand perception (Powesland & Giles, 1975; Tsalikis et al., 1991), and so far there is little support for the theory. For example Powesland and Giles (1975) found that standard-accent speakers and non-standard-accent speakers tend to evaluate a message from each respective accent similarly. Tsalikis et al.

(1991) did not find enough support for their hypothesis that there would be an interaction between a respondent’s rating of an accent and his/her exposure to it. The reason this was included in the questionnaire despite previous research finding little evidence for it, was that preceding papers have compared standard accents with non-standard accents. This study analysed differences between accents/languages that can all be considered as “standard”.

The second reason for testing relationships reads stereotypes, which is the theory of people having set and over-simplified pictures of something they do not fully know (Cardwell, 1996). The respondents’

stereotypical pictures of the different accents and their country origin were not tested per se, since it was beyond the scope of this study, but the respondents were asked to identify the accent they were exposed to in order to afterwards be able to discuss the result of the empirical study in the light of previous research on stereotypes. The questions in this element were not measured on scales, but the respondent could choose from a fixed set of answers which were mutually exclusive and collectively exhaustive (MECE).

56

The fact that MECE was applied to the measurement was important in order to obtain quality in the answers, since it minimized the risk that the respondents will be in doubt which answer applies to her (Bryman, 2016).

Control facts

The last part of the empirical study survey was an element of control factors including gender, age, occupation and level of education. Just like with the accent relationship part of the personal fact questions, the control facts were measured using fixed answer developed through the MECE lens.