Bilinguals’ Brand Perceptions Reported on Different-Language Questionnaires

ABSTRACT - This research examines the way Chinese-English bilinguals report their perceptions of a brand on Chinese-only, English-only, and bilingual questionnaires. Research results indicate that language has significant main effects on the bilinguals’ reported brand perceptions. This research also provides evidence that bilinguals shift from their dominant language (Chinese) to their secondary language (English) when the brand perception variable receives strong international influences. The implications of these findings are discussed.

Citation:

Kineta Hung and Roger Heeler (1998) ,"Bilinguals’ Brand Perceptions Reported on Different-Language Questionnaires", in NA - Advances in Consumer Research Volume 25, eds. Joseph W. Alba & J. Wesley Hutchinson, Provo, UT : Association for Consumer Research, Pages: 246-251.

Advances in Consumer Research Volume 25, 1998      Pages 246-251

BILINGUALS’ BRAND PERCEPTIONS REPORTED ON DIFFERENT-LANGUAGE QUESTIONNAIRES

Kineta Hung, Chinese University of Hong Kong

Roger Heeler, York University

ABSTRACT -

This research examines the way Chinese-English bilinguals report their perceptions of a brand on Chinese-only, English-only, and bilingual questionnaires. Research results indicate that language has significant main effects on the bilinguals’ reported brand perceptions. This research also provides evidence that bilinguals shift from their dominant language (Chinese) to their secondary language (English) when the brand perception variable receives strong international influences. The implications of these findings are discussed.

Among the many issues that cross-cultural researchers have to consider, language presents an ever important issue. This is because consumers in the target culture may speak or read a language other than the language of the source culture. Under these circumstances, the need to translate existing questionnaires from the source language (usually in English) to the target language would be obvious. However, if consumers in the target culture are bilingual, speaking both the source language and a native language, would it be necessary to translate existing questionnaires into the native language? Alternatively, should the investigator administer the research in the source language instead?

In a seminal study, Ervin (1964) compared two sets of Thematic Apperception Test (TAT) administered to two matched samples of French-American bilinguals. The results indicated that the content of the stories told by the subjects shifted wih language, with more English stories illustrating a need for achievement and more French stories illustrating the use of verbal aggression (against peers) and withdrawal (in family situations) to resolve interpersonal conflicts. More recently, Ralston, Cunniff, and Gustafson (1995) administered Chinese and English versions of the Schwartz Values Survey to bilingual Hong Kong managers. Although language did not have significant overall effects, significant effects were identified for four of the ten values under investigation. These values are Achievement, Hedonism, Tradition, and Security.

These findings suggest that, when bilingual subjects respond to questions relating to their personalities, they adjust their responses to conform to sociocultural expectations of different linguistic subgroups. This proposition presents an interesting question to consumer researchers: when products are extensions of consumers’ self-image (Belk 1988) and brands are characterized by different personalities (e.g. the Marlboro Man, the Green Giant), would bilinguals adjust their perceptions of a brand when research questions are posed in different languages?

MIX-LANGUAGE PROCESSING IN BILINGUALS

Bilingualism is a very common phenomenon on a worldwide basis (Harris and Nelson 1992). Many countries are officially bilingual (e.g. Canada) or multilingual (e.g. Singapore). Even in countries that are officially monolingual (e.g. the United States), many residents have acquired a second language for cultural or business reasons.

Bilinguals have often been treated in research as two monolinguals in one. For example, language skills in bilinguals have almost always been appraised in terms of monolingual standards. This "fractional" approach to bilingualism tends to treat language shifts as an abnormality or a sloppy use of language. However, researchers that take a more "wholelistic" approach recognize that language shift is a common and legitimate form of communications among bilinguals (Bantahila and Davies 1992; Grosjean 1992; ValdTs and Figueroa 1994). For example, a shopkeeper in Kenya who was exchanging greetings with his sister in their local Luyia diaclect had suddenly switched to Swahili when he asked her what she had wanted (cited by Scotton and Ury 1977):

Dada, sasa leo unahitaji nini? (code-switching)

(Sister, now today what do you need?)

Meanwhile, a French emigrant to the United States was found to have shifted from one language to another in the course of a sentence when she asked her husband to wake up their son (cited by Grosjean 1982):

Va chercher Marc (go fetch Marc) and bride him avec un chocolat chaud (with a hot chocolate) with cream on top

(code-switching).

In addition to code-switching from one language to another, bilinguals also"borrow" words from one language and adapt it to another language. An example can be found in the following (cited by Grosjean 1982):

¦a a poppT. (It popped.)

(borrowing-popper (to pop) is not a proper French word)

These examples show that, bilinguals tend to interact with the world around them using both languages. It has been postulated that bilinguals’ choice of language in a conversation is affected by the situation (e.g. presence of monolinguals), the function of interaction (e.g. to create/reduce social distance), the content of the discourse (e.g. topic and vocabulary) as well as the participants’ language preference and proficiency (Grosjean 1982).

Despite these evidence for mix-language processing, research involving bilinguals are typically conducted in a monolingual mode (e.g. using a questionnaire translated into the local language). This procedure allows the participants to respond in a language that they may be more proficient in. However, when bilinguals are forced to respond in a monolingual mode, the language they use in responding to the research instrument may not be appropriate for the specific research topic or environment.

The current study examines the way bilinguals respond to brand perception questions that are posed in different languages. Because this is an exploratory study, we will use a grounded research approach (Lincoln and Guba 1985) to identify issues relevant to bilinguals’ perceptions.

METHOD

Research Design

The current study was conducted as part of a larger research project that examines the effects of advertising in a cross-cultural setting. However, because the questionnaire (written in English) has been developed to test commercials in the source culture, we had to translate and adapt the original questionnaire to test Chinese-English bilinguals in Hong Kong.

We have developed three versions of the questionnaire in this research: a Chinese-only version, an English-only version, and a bilingual version. We believe that the bilingual questionnaire could provide a benchmark measure for differential language use and thereby illustrate the strategies that bilinguals use to process different variables.

Technically, this study follows a 3 (language) x 4 (commercials) factorial design. We have included four commercials in this study to provide a replication of four and to rule out potential interactive effects. Brislin, Lonner, and Throndike (1973) have suggested that examining potential interactive effects can enhance the determination of the main effects in cross-cultural research. However, we recognize that language is the subject of interest in this research and we will refrain from elaborating on the main effects of the commercials in this study. Finally, this study follows a between-subject design. Thus, each subject would complete only one of the questionnaires and be exposed to only one of the commercials.

Subjects

The subjects were 139 undergraduate business students enrolled in marketing courses taught by one of the authors at a university in Hong Kong. All the subjects were of Chinese origin. Linguistically, the subjects are ChineseEnglish bilinguals and all of them speak Chinese at home. English is taught early on at school. Most students received their first lessons in English either in kindergarten or in Garde 1. At the university level, lectures are often delivered in a mix-language mode. However, text books were typically written in English and the students complete their assignments and examinations for most courses in English. Thus, eventhough the subjects are Chinese-dominant, they have many opportunities to read and write in English on a day-to-day basis.

Stimuli

The stimuli were television commercials provided by an industry source that promote a national brand of coffee. These commercials emphasize different lifestyle images and have been used in previous research (Hung and Rice 1995). The first commercial was comprised of scenes of nature and people working in a jungle environment and drinking coffee. The music in this commercial was broad and slow. The second commercial contained scenes of people drinking coffee and socializing in a modern cafT. The music was upbeat. The third and fourth commercials were produced by miss-matching the scenes and music in the first two commercials.

The original commercials were produced in Australia and have not been aired in Hong Kong before. These commercials were 60-seconds in length. During editing, the verbal elements (written and/or spoken) were taken out and the commercials were reduced to 30-seconds each. This was possible because the commercials utilized a collage format without following specific storylines. Thus, the commercials were able to retain a continuous flow even with the editing. To reduce familiarity effects, the original brand names in the commercials were replaced by a fictitious name in English. The same fictitious name was used to replace the original brand names in each of the four commercials.

Questionnaire

As part of a larger research project, the original questionnaire was developed using a pilot study conducted in the source (English-speaking) country. During the pilot study, the informants wrote in an open-ended manner the attributes and images that they associate with the coffee advertised in each of the four commercials. The protocols were analyzed by two coders using a hermeneutics approach (Arnold and Fischer 1994) and the more "trustworthy" (Lincoln and Guba 1985) categories of attributes and images were retained from the protocols. These categories of attributes and images were subsequently developed into 23 five-point agreement scales to test the viewers’ perceptions of the coffee advertised in the four commercials. Because the original questionnaire was intended for use in the source country, the questions were stated in English.

We recognize that people in Hong Kong do not consume as much coffee and the varieties of coffee available in Hong Kong are more limited. Thus, before translating the questionnaire into Chinese, we reviewed the items on the original questionnaire and deleted nine items that could be meaningless to subjects in Hong Kong. Meanwhile, we retained 14 items that described the viewers’ perceptions of the coffee (10) as well as the viewers’ attitudes towards the ad (2) and the coffee (2). Questions relating to the subjects’ demographic variables were also included on the questionnaire.

To develop a Chinese equivalent version of the revised (shorter) questionnaire, two graduate students worked independently to translate the English questionnaire into Chinese and then back-translate the Chinese version into English. The translate/back-translate procedure has been used in many studies to develop comparable questionnaires in different languages (Brislin et al. 1973; Parameswaran and Yaprak 1987; Price and Mombour 1967). After the translation and back-translation had been completed, we examined the revised (shorter) English version, the Chinese version, and the back-translation (in English) and were satisfied with their inguistic equivalence.

Based on the above procedures, we developed three different-language questionnaires, including an English-only version, a Chinese-only version, and a bilingual version. On the bilingual questionnaire, each question was stated initially in English and then repeated immediately in Chinese.

Dependent Measures

Three dependent measures were used in this study: trendy, Aad, ABrand. The measure "trendy" evaluates the image viewers associate with the advertised brand. "Trendy" was identified from a principal component factor analysis (with VARIMAX rotation) involving 10 image items describing the brand promoted in the four commercials. As an image measure, "trendy" evaluates a form of brand "attribute" (or belief) that is useful to consumers in assessing their attitudes towards a brand (Fishbein and Ajzen 1975). Further, "trendy" is a lifestyle concept that is heavily influenced by consumption images in countries in the Western World (Craik 1994; Ewen 1988). There were three items in the "trendy" measure: stylish, fashionable, trendy. Cronbach’s a for this measure ranges from .74 to .84 across the three language groups.

In addition to the image measure, two attitudinal measures were included. Attitudes are more abstract forms of consumer perceptions (Gutman 1982; Mitchell and Olson 1981). There were two items in each of the "Aad" (ad good, like ad) and "ABrand" (coffee good, like coffee) measures. Pearson r ranges from .65 to .68 for the "Aad" measure and from .59 to .79 for the "ABrand" measure (Table 1).

Research Procedures

The current study was conducted during the coffee break in three marketing classes as an optional class project. Before the break, students were informed that they could participate in a marketing research project. Volunteers were divided into two groups according to their student numbers: odd numbers were assigned to Group 1 and even numbers were assigned to Group 2. Students in Group 1 were asked to remain behind while students in Group 2 were asked to return to the classroom in ten minutes.

TABLE 1

RELIABILITY OF MEASURES ACROSS LANGUAGE GROUPS

After the students in Group 2 have left the classroom, students in Group 1 were randomly handed an English-only, Chinese-only, or bilingual questionnaire. Because there was a limited number of available subjects, we have handed out more English-only and Chinese-only questionnaires (test groups) than bilingual questionnaires (control group). Afterwards, the students were instructed to watch a commercial of coffee (Commercial #1) before they filled out the questionnaires. Because the classes from which the students were recruited were taught in English, the instructions for the experiment were delivered in English. The subjects were allowed to watch the commercial one time. Before the students in Group 1 left the classroom to take their break, they were thanked and were instructed not to discuss the research with other students. The above procedures were repeated for students in Group 2 when they returned to the classroom. However, students in Group 2 were shown a different commercial (Commercial #2). After the students in Group 1 have returned from their coffee break and rejoined the class, both groups of students were debriefed and thanked.

The procedures carried out in the first class were repeated in the other two classes in which the experiment was conducted. However, instead of watching Commercials #1 and #2, students in the second class were shown Commercials #3 and #4. The students in the third class were shown Commercials #4 and #1.

During the debriefing session, some students indicated that they were aware of the different-language questionnaires. However, they thought that the different-language questionnaires were included in the study to clarify the questions. None of them felt that the purpose of the research was to examine the way they respond to questionsstated in different languages. Further, the students did not feel that they would adjust their responses to accommodate the different languages even after they have been debriefed of the objective of the study. Thus, eventhough some of the subjects were aware of the different questionnaires, there should be little demand artifacts in their responses.

The students responding to bilingual questionnaires indicated during the debriefing session that they were aware of the two languages on the questionnaire. However, when they were asked which language they had focused on when they were reading the questionnaires, they suggested that they had "taken in" both the English and Chinese questions "as a whole" before responding to the questions.

RESULTS

Profile of Subjects

The profile of the students who have participated in this study is reported in Table 2. As indicated, there were more female students than male students in each language condition. The break-down of male/female students generally agrees with the gender ratio of students enrolled in the business faculty at the participating university. As the students were full-time undergraduate students, all of them were 24 years of age or under. Meanwhile, the table also indicated that the majority of students did not drink coffee regularly. The household income of the students were roughly evenly distributed.

To examine the potential differences among the three language groups, chi-square tests were calculated for each demographic variable (except age which has one value). Results indicated that there were no significant differences among the three groups in terms of sex (c²=1.29, p>.50), coffee consumption rate (c²=2.77, p>.25), or household income (c²=5.17, p>.50). Thus, the results confirmed that there were no significant differences among the students in the three language groups.

Overall Effects

As discussed in the Method section, two types of manipulations were used in this study. Results of two-way MANOVA indicated that both "language" (F(6,250)=7.39, p<.0001) and "commercials" (F(9,304)=9.74, p<.0001) have significant main effects on the dependent variables. However, while "language" has significant effects on all three dependent variables, "commercials" only have significant effects on "trendy" image, with the commercials incorporating cafT scenes ( = 9.90, 10.59) scoring higher mean values than the ones incorporating nature scenes ( =7.14, 7.31). Meanwhile, even though both "language" and "commercials" have significant main effects, there were no significant interactive effects between these variables (F(18,354)=1.00, p>.45). Thus, research results suggest that, despite reasonable care in ensuring linguistic and conceptual equivalence in the dependent variables, the language in which research questions were stated has significant main effects on the brand perceptions reported by bilingual subjects.

TABLE 2

PROFILE OF SUBJECTS

As an additional check to verify the main effects of "language", a series of MANOVA was conducted to examine the effects of demographic variables on the subjects’ responses. Results indicated that the respondents’ "gender" (F(2,136)=3.43, p<.05) has significant main effects. However, results of two-way MANOVA indicated that the interaction between "gender" and "language" did not have significant effects (F(4,264)=.72, p>.58). Meanwhile, the respondents’ rate of "coffee consumption" (F(6,268)=.65, p>.69) and "income" (F(10,260)=1.59, p>.10) did not have significant main effects on the subjects’ attitudes or image perceptions.

Pairwise Comparisons-MANOVA

Because the language in which research questions were stated has significant main effects on the way bilinguals respond, a series of pairwise MANOVA (with effect size) was conducted to further investigate the issue. Results indicated that there were significant differences between the Chinese/English pair (F(3,101)=8.95, p<.0001, h=.21) and the English/bilingual pair (F(3,82)=5.05, p<.01, h=.16). However, there were only marginal differences between the Chinese/bilingual pair (F(3,83)=2.49, p<.10, h=.08). These findings show that responses to the bilingual questionnaire were more similar to the Chinese questionnaire than the English questionnaire. Indeed, this pattern is even more prominent when we compare the correlation coefficients among the dependent variables across the three language groups.

As indicated in Table 3, the correlations between "trendy" and "Aad", and between "trendy" and "ABrand" were not significant in either the Chinese questionnaire or the bilingual questionnaire. However, these correlations were significant in the English questionnaire. These findings suggest that, when bilingual questionnaires are used, respondents will rely more heavily on the Chinese portion of the questionnaire than the English portion of the questionnaire. These findings are not surprising since Chinese is the native and dominant language of the respondents. However, it is interesting to note that, according to their comments, the subjects exposed to bilingual questionnaires were not aware that they have relied more heavily on one language than another when they filled out the questionnaires.

Pairwise Comparisons-ANOVA

The above analysis indicated that, in general, respondents relied more heavily on the dominant language, Chinese, when they have a choice of languages. This section will examine whether the reliance on the dominant language could be extended to individual variables. Results of the pairwise ANOVA are reported in Table 4.

TABLE 3

PEARSON r ACROSS THE THREE LANGUAGE GROUPS

TABLE 4

RESULTS OF THE PAIRWISE COMPARISONS

As indicated on the table, the subjects’ perceptions of "trendy" image were significantly different between the Chinese/bilingual pair (F(1,85)=6.60, p<.05). However, their perceptions of either "Aad" or "ABrand" were not significant in the same language pair. Meanwhile, the subjects’ perceptions of "Aad" (F(1,84)=15.27, p<.001) and "ABrand" (F(1,84)=5.81, p<.05) were significantly different between the English/bilingual pair but their perceptions of "trendy" image were not significantly different (F(1,84)=.08, p>.70) in the same language pair. These results suggest that, when the subjects reported their perceptions of "trendy" image on bilingual questionnaires, they have relied more heavily on the English portion of the questionnaire. However, when the subjects reported their "attitudes" (Aad or ABrand) on bilingual questionnaires, they have relied more heavily on the Chinese portion of the questionnaire. The perceptions reported on the Chinese/English pair were mixed, showing significant differences in "trendy" image (F(1,103)=6.52, p<.05) and "Aad" (F(1,103)=15.32, p<.001), but no significant differences in "ABrand" (F(1,103)=2.05, p>.10).

While the shift in language reliance yielded contradictory results at the univariate level, we could discern meaningful patterns from the results if we take into considerations the characteristics of the variables under investigation. It has been suggested earlier that attitudinal measures, "Aad" and "ABrand", represent perceptions that are more abstract. Thus, it is conceivable that, because "attitude" requires more inferences from the subjects, the subjects have relied more heavily on their dominant language when they process and respond to attitudinal questions. Conversely, products that are considered "trendy" in Hong Kong (e.g. fashion) have often received much international influences. Thus, the subjects may have associated "trendy" images with international sources and thereby process iage-related questions in English instead of in Chinese.

DISCUSSION

This research has provided an exploratory investigation into the approaches bilinguals use to respond to questions relating to their perceptions of ads and brands. Previous research on bilingualism has shown that bilingual subjects may communicate in a mix-language mode, ranging from code-switching morphemes to using different languages for different functions (Bentahila and Davies 1992; Harris and Nelson 1992; ValdTs and Figueroa 1994). The results of this study lend support to these research by showing that bilinguals process some marketing stimuli in their dominant language and other marketing stimuli in their secondary language. However, even though bilinguals are more likely to process and respond to questions written in their dominant language, the reliance on the dominant language may not apply in every situation. When bilingual subjects associate a variable more strongly with their secondary language, they may process this variable in their secondary language instead of their dominant language. Thus, unless there is a strong theoretical basis determining which monolingual response would be more valid, it would be preferable to test bilinguals with bilingual questionnaires, allowing the subjects to choose the language they prefer for a given topic, in a given situation. Indeed, Drasgow and Hulin (1987) have suggested that using bilingual questionnaires could minimize interpretation errors in research involving bilingual subjects.

The findings of this research may also affect the way research findings are interpreted in other studies when bilinguals are involved as participants. If language has significant main effects on bilinguals’ responses, research results could be confounded if the research design has not taken into consideration potential language effects.

While this study focuses on examining language issues relevant to marketing research, results of this study could be extended to marketing communications as well. Rather than restricting all aspects of marketing communications (e.g. brand names, advertisements) to the target consumers’ dominant language, marketers working in bilingual/multilingual markets (e.g. Hong Kong, India, Singapore, etc.) should consider the contents of the communications before deciding on the language of communications.

Finally, the current research suffers from a number of limitations. First, we have not adequately pretested the equivalency of the Chinese and English questionnaires. Thus, the findings of this research may be confounded with differences in the translation of the questionnaires. Further, we have not pretested the bilingual subjects’ language dominance, their level of proficiency in the two languages, and characteristics of the dependent variables that may be language-dependent. Without insights into these issues, we cannot accurately assess the generalizability of the findings of this study. Given these limitations, future research could explore these issues in more controlled experiments to improve our understanding of the strategies bilinguals use to process marketing information. Further, future research could replicate this study in a) other Chinese-speaking countries, and b) other bilingual countries to help resolve language issues relating to the testing of bilinguals.

REFERENCES

Arnold, Stephen J. and Eileen Fischer (1994), "Hermeneutics and Consumer Research," Journal of Consumer Research, 21 (June), 55-70.

Belk, Russell W. (1988), "Possessions and the Extended Self," Journal of Consumer Research, 15 (September), 139-168.

Bentahila, AbdelGli and Eirlys E. Davies (1992), "Code-switching and Language Dominance," in Richard J. Harris, ed., Cognitive Processing in Bilinguals, Amsterdam: North-Holland, 443-458.

Brislin, Richard W., Walter J. Lonner, and Robert M. Throndike (1973), Cross-Cultural Research Methods, New York: John Wiley & Sons.

Craik, Jennifer (1994), The Face of Fashion: Cultural Studies in Fashion, London: Routledge.

Drasgow, Fritz and Charles L. Hulin (1987), "Cross-cultural Measurement," Revista-Interamericana-de-Psicologia, 21 (1-2), 1-24.

Ervin, Susan M. (1964), "Language and TAT Content in Bilinguals," Journal of Abnormal and Social Psychology, 68 (5), 500-507.

Ewen, Stuart (1988), All Consuming Images: the Politics of Style in Contemporary Culture, New York: Basic Books.

Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley.

Grosjean, Frantois (1982), Life with Two Languages: An Introduction to Bilingualism, Cambridge, MA: Harvard University Press.

Grosjean, Francois (1992), "Another View of Bilingualism," in Richard J. Harris, ed., Cognitive Processing in Bilinguals, Amsterdam: North-Holland, 51-62.

Gutman, Jonathan (1982), "A Means-End Chain Model Based on Consumer Categorization Processes," Journal of Marketing, 46 (Spring), 60-72.

Harris, Richard J. and Elizabeth M. Nelson (1992), "Bilingualism: Not the Exception Any More," in Richard J. Harris, ed., Cognitive Processing in Bilinguals, Amsterdam: North-Holland, 3-14.

Hung, Kineta and Marshall Rice (1995), "A Comparative Examination of the Perception of Ad Meanings in Hong Kong and Canada," Fifth Symposium on Cross-Cultural Consumer and Business Studies, ed. Scott M. Smith, 262-266.

Lincoln, Yvonna S. and Egon G. Guba (1985), Naturalistic Inquiry, Beverly Hills, Sage Publications.

Mitchell, Andrew A. and Jerry C. Olson (1981), "Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude?" Journal of Marketing Research, 18 (August), 318-332.

Parameswaran, Ravi and Attila Yaprak (1987), "A Cross-National Comparison of Consumer Research Measures," Journal of International Business Studies, 18 (Spring), 35-49.

Prince, Raymond and Werner Mombour (1967), "A Technique for Improving Linguistic Equivalence in Cross-Cultural Surveys," International Journal of Social Psychiatry, 13, 229-237.

Ralston, David A., Mary K. Cunniff, and David J. Gustafson (1995), "Cultural Accommodation: the Effect of language on the Responses of Bilingual Hong Kong Chinese Managers," Journal of Cross-Cultural Psychology, 26 (November), 714-727.

Scotton, Carol M. and William Ury (1977), "Bilingual Strategies: the Social Functions of Code-Switching," Linguistics, 193, 5-20.

ValdTs, Guadalupe and Richard A. Figueroa (1994), Bilingualism and Testing: A Special Case of Bias, Norwood, NJ: Ablex Publishing Corporation.

----------------------------------------