Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

Embed Size (px)

Citation preview

  • 7/29/2019 Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

    1/16

    69Journal of Marketing

    Vol. 72 (September 2008), 6983

    2008, American Marketing Association

    ISSN: 0022-2429 (print), 1547-7185 (electronic)

    Jaime Noriega & Edward Blair

    Advertising to Bilinguals: Does theLanguage of Advertising Influence

    the Nature of Thoughts?When targeting bilingual consumers, advertisers have a choice to advertise in a bilinguals native language or inthe countrys dominant language. Within the Hispanic community in the United States, for example, Kellogg has achoice to advertise Frosted Flakes in English or in Spanish. But which is the better choice and why? This researchconsiders whether the choice of language in advertising to bilinguals influences the types of thoughts they have inresponse to an advertisement. In other words, for a bilingual, can the exact same selling message cue differentassociations depending on the language in which it is presented? The underlying issue is whether advertiserscan use language of execution as a strategic variable with which to generate certain types of associations thatmay facilitate persuasion. The authors consider this issue from a social cognition perspective. They hypothesizethat a native-language advertisement is more likely to elicit self-referent thoughts about family, friends, home, orhomeland, which in turn may lead to more positive attitude measures and behavioral intentions. Furthermore, theauthors show that these effects are moderated by the consumption context presented in the advertisement.

    Keywords: information processing, bilingualism, advertising, psycholinguistics, consumer research

    Jaime Noriega is Assistant Professor of Marketing, Marketing Depart-ment, College of Commerce, DePaul University (e-mail: [email protected]). Edward Blair is Michael J. Cemo Professor and Chair of Marketing &Entrepreneurship, C.T. Bauer College of Business, University of Houston(e-mail: [email protected]).

    Aprimary topic of interest in the U.S. literature oncross-cultural communication is whether advertisingdirected at bilingual minorities is more effective if

    presented in the countrys dominant language or in thebilinguals native language. According to the U.S. CensusBureau (2005), the United States has more than 50 millionpeople who speak a language other than English at home,and the trend toward acculturation rather than assimilationmeans that many minorities are choosing to preserve ele-ments of their ethnic identity, perhaps the most distinctiveof which is language.

    Previous research on language choice and ad effective-ness has used three perspectives to explain why languagechoice can make a difference in advertising effectivenessfor bilingual markets. First, the earliest research consideredhypotheses related to identity and accommodation. Thisstream of research is conditional on in-group versus out-group identities and presumes that the target group is madeup of minorities for whom an advertisement is more likelyto be persuasive if it features a character similar to them oris written in their native language (Deshpand, Hoyer, andDonthu 1986; Deshpand and Stayman 1994; Koslow,Shamdasani, and Touchstone 1994; Whittler 1991). This

    effect should occur as long as group members believe thatthe act of translating or otherwise making the advertisementculturally accessible is a sign that the advertiser acknowl-

    edges, values, and respects them. The moderators suggestedby this mechanism are anything that would enhance ordiminish these effects. For example, Deshpand and col-leagues (Deshpand, Hoyer, and Donthu 1986; Forehandand Deshpand 2001) find that ethnic self-identity affectswhether a native-language execution aids ad persuasiveness,such that a native-language execution is more persuasive forpeople whose ethnic self-identity is high rather than low.

    Second, and more recently, cross-cultural communica-tion research has suggested that language choice can berelated to ad effectiveness through ease of processing. Inthis regard, Luna and Peracchio (1999, 2001) find that it ispreferable to advertise to bilinguals in their first/native lan-guage not because of any social or cultural considerationsbut simply because second-language words are more diffi-cult to process for bilinguals. Because conceptual links aremore difficult to come by for second-language words thanfor first-language words, less of a message will be recalledwhen it is presented in the persons second language. Themoderators suggested by this stream of research are any ele-ments that affect the level of verbal processing required byan advertisement or the ease of doing this processing. Anobvious choice is the audience members language fluency,

    and Luna and Peracchio (2001; see also Luna, Peracchio,and DeJuan 2003) also find that high levels of picturetextcongruity make the process easier and allow for better recallof a second-language message.

    Third, in their most recent research, Luna and Peracchio(2002, 2005) also consider language effects from the per-spective of affective response. Luna and Peracchios argu-ment is that some words have more of an emotional attach-ment when presented in the native language than in thesecond language. This may occur as a result of sociolinguis-

  • 7/29/2019 Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

    2/16

    70 / Journal of Marketing, September 2008

    tic differences. A bilinguals native culture may value cer-tain concepts (e.g., family, relationships, religion) morehighly than other cultures, and the language in which themeaning of the concept is first learned then becomes theprototypical representation of that concept. A possible mod-erator suggested by this stream of research is the nature ofthe appeal of any given advertisement. The impact of affect-laden words might be more important when using an emo-tional appeal instead of a rational or functional appeal basedon the products attributes. Note that the effects of this

    stream of research are not directly conditional on in-groupversus out-group identities.

    In this article, we consider a fourth perspective toexplain how the choice of language might affect ad persua-siveness. Our approach is based on social cognition. Weargue that each of a bilinguals two languages may cue dif-ferent associations for the same message, and thus each lan-guage execution has the potential to lead to different levelsof persuasiveness. In particular, we argue that the two lan-guages are likely to be differentially associated with a bilin-guals experiences among family and friends and thus aredifferentially likely to cue self-referent associations withthese experiences, with possible implications for persua-

    sion. The moderators suggested by this line of research areany aspects of the stimuli that are somehow related to thiscueing. We examine how the consumption context pre-sented in an advertisement may moderate the relationshipbetween choice of language and the resultant thoughts andpersuasion.

    All these research streams indicate that it may be moreeffective to advertise to members of a minority group intheir native language, depending on their level of accultura-tion. What differentiates these theories and makes each onesingularly valuable is that each suggests its own set of mod-erators. Multiple moderators drawing from different per-spectives lead to a better understanding of the overall phe-

    nomenon and enable us to suggest several different practicalapplications.We organize the remainder of this article as follows:

    First, we examine different perspectives on language andmemory and how language and context intersect. Second,we develop the hypotheses and describe the studies andresults. Third, we discuss the theoretical and managerialimplications of our findings. Finally, we provide the limita-tions of our research and their implication for furtherresearch.

    Conceptual Background andHypotheses

    Perspectives on Language and Memory

    The literature on language and cognition maintains that lan-guage can serve as an attribute of an experience (Lambert,Ignatow, and Krauthamer 1968; McCormack 1976; Saegert,Hamayan, and Ahmar 1975; Winograd, Cohen, and Barresi1976). Experiences can be characterized by attributes thathave a distinct language identity. The language spoken dur-ing an encounter is one such attribute, as is the presence of

    people who speak a particular language. The number ofattributes within an experience that share a distinct lan-guage tag and the salience of those attributes may determinethe extent to which the memory for that experience as awhole is linked to a given language. More broadly, anygiven attribute with a distinct language tag has the potentialto serve as a pathway through which language cues theexperience as a whole.

    Language can also serve as a medium of experience,which means that it is intrinsic to capturing the event; every

    aspect of that incident is infused with a language tag suchthat language becomes the gateway to that memory, and itcan be reexperienced or shared fully only in the same lan-guage in which it was experienced. Under this view, lan-guage becomes a sort of superattribute. The idea that lan-guage can serve as a medium of experience began as aphilosophical argument, which in its strong version sug-gests that language is so intrinsic to thought that it com-pletely guides comprehension and representation (Whorf1956). Although the strong version of this argument hasbeen discarded, there is common agreement that languagedoes influence thought (Larsen et al. 2002; Marian andNeisser 2000), and this might be particularly true for social

    experiences (Fivush 1998; Schrauf 2003). Language is atool of communication, intimately tied to how peopleexperience life as social creatures. As Fivush (1998, p. 486)suggests, language is a critical tool of human cognition,one which allows us to move beyond individual cognitionand engage in culturally mediated cognition.

    The idea that language might serve as an attribute or amedium of experience creates a theoretical connectionbetween language and experience. At a minimum, languageis one of the many attributes of an experience that peopleassociate with it, and therefore language and experiencesshould have the capacity to cue each other.

    Context and LanguageThere are many examples of context (the people, places,things, and symbols in a persons environment) cuing thelanguage people use to communicate. For a monolingual,surroundings help determine the proper vocabulary and theformality with which it should be used. For example, a doc-tor might speak of a hematoma at the hospital but refer tothe same condition as a bruise at home. A lawyer is notlikely to use slang in the courtroom but may do so whenvisiting with friends over the weekend.

    Bilinguals also use context as a guide for the vocabularythey should use and the formality with which to use it, butin addition, the context within which bilinguals find them-

    selves can cue which language seems most appropriate. Forexample, a ChineseEnglish bilingual making a reservationat a Shanghai hotel might handle the transaction in Chinese,but when in Boston, he or she will handle the transaction inEnglish.

    These examples are a clear indication that context cancue language; however, from an advertising perspective, weare most interested in knowing whether language can cuecontextthat is, whether different languages can cue differ-ent associations.

  • 7/29/2019 Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

    3/16

    Advertising to Bilinguals / 71

    This question has received some attention in the psy-cholinguistics literature. Research on language-dependentmemory has shown that for bilinguals, the recall of autobio-graphical memories may depend on the language withwhich the memory is cued. In a study of RussianEnglishbilinguals, Marian and Neisser (2000) find that participantsshared more experiences in Russian when interviewed inRussian and more experiences in English when interviewedin English, in support of the hypothesis that language ofinquiry can cue the language of recall for autobiographical

    memories. Larsen and colleagues (2002) go a step furtherby proposing that bilinguals semantic and conceptualstores can be both language and culture specific. They findthat when Polish immigrants to Denmark were asked inPolish to recall a life experience, the reply was more likelyto be given in Polish and to correspond to an event experi-enced in Poland before immigration, whereas if the requestwas made in Danish, the reply was more likely to be givenin Danish and to correspond to an event experienced inDenmark after immigration.

    Although these studies provide some indication that dif-ferent languages can cue different associations, they haveimportant limitations. The social conditions of their

    methodology (i.e., an interviewer making an explicitrequest for autobiographical memories in a specific lan-guage) make the presence of demand effects a distinct pos-sibility. It could be that bilinguals assume out of reciprocitythat questions should be answered in the language in whichthey are stated. If this is so, it could be argued that it was therespondents presumption of language expectations that ledto the reporting of linguistically matched memories. Like-wise, when immigrants were asked to recall a life experi-ence, the language of inquiry may have cued certain experi-ences in memory but may also have been taken as anindication of which experiences were of interest. Further-more, even if these studies suggest that language has the

    potential to cue context, they do not provide a clear indica-tion that it can do so spontaneously, because both studiesexplicitly requested that participants recall an episodicmemory.

    Another stream of research that suggests that languagecan cue associations is related to cultural frame switching.This recent literature in psychology has considered howlanguage and symbols can cue different personality and cul-tural characteristics in biculturalbilingual people. Forexample, Hong and colleagues (2000) consider how differ-ent cultural icons primed collectivist versus individualistresponses in ChineseEnglish bilinguals. Ross, Xun, andWilson (2002) examine the possibility that language servesas a cue to distinct cultural mind-sets. In a study of

    EnglishSpanish biculturalbilinguals, Ramrez-Esparzaand colleagues (2006) find that language is capable of cuingeither U.S. or Mexican personality characteristics (as identi-fied in monolinguals from each country).

    Hypothesis Development

    If language can cue associations, the general literature oninformation processing suggests that the linguistic diagnos-ticity of any given association will determine whether lan-

    guage cues it. For example, if some experiences systemati-cally occur in one language context, it is more likely thatthis language context will cue those experiences. There mayalso be experiences that carry a unique or special culturalsignificance that can only be experienced, elaborated on,and shared in one specific language.

    The focus of the current research is U.S. minority bilin-guals. These people often live linguistically bifurcated lives(Linton 2004). Life at home and in their ethnic communityis experienced in their native language, whereas life at

    school, work, and in the community at large is experiencedin English. This pattern is so common that the U.S. CensusBureau (2005) recognizes it as a categorypeople whospeak a language other than English at homethat appliesto more than 50 million Americans. The resultant duality ofexperiences may lead to distinct thoughts when faced withadvertisements written in either of the languages a bilingualcomprehends.

    Given the foregoing discussion, our basic argument isthat advertisements presented in bilinguals native languageare more likely to evoke associations that match that lan-guagespecifically, the bilinguals associations with fam-ily, friends, home, or homeland (hereinafter FFHH)than

    advertisements presented in English. Furthermore, thiseffect is likely to be stronger if the consumption contextrepresented in the advertisement is consistent with such use.We define native-language versus second-language contextsas follows: Native-language contexts are more likely toevoke settings in which native language is typically or tradi-tionally spoken and/or people are present with whom abilingual either typically communicates in the native lan-guage or shares aspects of the native culture, whereassecond-language contexts are less likely to evoke these set-tings. Thus, our first two hypotheses are as follows:

    H1: Native-language advertisements elicit a higher propor-tion of FFHH-related thoughts than second-language

    advertisements.H2: The phenomenon predicted in H1 is moderated by context

    such that the effect is stronger for advertisements innative-language contexts than for those in second-language contexts.

    From an advertising point of view, it might be askedwhether these types of thoughts will influence ad effective-ness in terms of better attitudes toward the ad and productand higher purchase intentions. We argue that they will, atleast in some circumstances. The key issue is that FFHH-related thoughts are a form of self-referent thoughts.

    Self-referent processing consists of autobiographicalthoughts (thoughts about life experiences) and thoughts

    about targets associate with the self. Other people, espe-cially those who are close to someone, form part of the self-concept because people define themselves in part by theirinteractions with others and by the groups to which theybelong (Ogilvie and Ashmore 1991). Salient interpersonalrelationships are incorporated into the self-concept (Aron etal. 1991). In this regard, Brewer and Gardner (1996) arguethat people have different versions of self-concept: the per-sonal self, which is characterized by personal attributes andcomparisons with others; the relational self, which consists

  • 7/29/2019 Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

    4/16

    72 / Journal of Marketing, September 2008

    of traits related to status as spouses, parents, siblings, andso forth; and the collective self, which consists of group-related traits. Furthermore, Johnson and colleagues (2002)find that processing a message that refers to groups towhich a person feels close has similar results to the moreindividualistic self-referent thoughts.

    The results of self-referent processing are well docu-mented. Prior research has shown that information relatedto the self has an advantage over other types of processingin terms of the strength and accessibility of the resultant

    memory (Rogers, Kuiper, and Kirker 1977). Self-referentprocessing results in increased elaboration of a message(Andersen, Glassman, and Gold 1998; Craik and Lockhart1972; Keenan, Golding, and Brown 1992), and more elabo-ration makes strong arguments more persuasive (Burnkrantand Unnava 1995). The self can also serve as an efficientorganizing framework (Klein and Kihlstrom 1986), andorganized or categorized elements are easier to recall. Self-referent processing also facilitates positive thoughts andmemories, the affect for which can get transferred to theadvertisement or brand (Stayman and Unnava 1997). Fur-thermore, Rogers (1981) notes that the self-referencingprocess may have an affective quality, regardless of its con-

    tent. Thus:H3: A rise in the proportion of FFHH-related thoughts results

    in more positive attitudes toward the ad and the brand andhigher purchase intentions.

    We report two studies to test these hypotheses. The ideathat each of a bilinguals languages can evoke differentthoughts (H1) is a fundamental thesis of our research; there-fore, as a first step, we test this point in Study 1. Study 2attempts to confirm the language effect (H1) and tests ourremaining hypothesesthe premise that context can mod-erate the effects of language on thoughts (H2) and whetherthese differences in thoughts translate into differences in adeffectiveness (H3).

    Study 1The purpose of the first study was to test our central thesis;namely, engaging a bilinguals native language versus his orher second language may gain access to different thoughts/associations. Specifically, in this study, we propose thatbilingual people who are asked to translate an English-language advertisement into their native language will listmore FFHH-related thoughts than those who are not askedto translate the advertisement.

    Method

    Participants and design. Eighty-two adults enrolled inadvanced ESL (English as a second language) classes inHouston and Los Angeles participated in Study 1 for thechance to win a $25 gift certificate to a store of their choice.We enlisted ESL instructors to help identify and contactpotential respondents who were sufficiently proficient inEnglish (given the specific demands of our study). Wechose this sampling frame because these people are bilin-guals for whom English is a second language but whoshould have the ability to understand English well enough

    to follow the study instructions. The sample consisted of 41women and 41 men, and the average age was 27 years.

    The study was a one-way between-subjects design inwhich the language condition had two levels: translate anddo not translate. Respondents were randomly assigned toone of the two language conditions. Of the respondents, 39completed the translate condition, and 43 completed theno-translate condition.

    To engage their native language effectively, we askedparticipants in the translate condition to translate an

    English-language advertisement into their native language.Presenting this approach rather than a pretranslated adver-tisement served a dual purpose. First, asking respondents toprovide their own translation enabled us to circumvent thenecessity of ensuring semantic equivalency of the ad copyacross any two languages because each respondent pro-vided his or her own translation according to his or her ownunderstanding of the advertisement. Second, this approachenabled us to accommodate a wide range of languages,which in this study included Cambodian, Catalan, Czech,Dutch, Farsi, French, Indonesian, Italian, Japanese, Korean,Mongolian, Russian, Spanish, Turkish, and Vietnamese.

    Stimuli and procedure. The stimuli consisted of twoadvertisements: a practice advertisement and a focal adver-tisement. The study was available online, and all partici-pants logged on and participated at their leisure and at theirown pace.

    Before viewing the first advertisement, respondents inthe translate condition saw the following instructions: Weare interested in knowing how people translate ads intoother languages. Please take some time to look at the nextadvertisement. Look at the ad as you would normally lookat ads when you are reading a magazine. In the box to theright of the ad, please type how the ad would read in yournative language. When you are finished, click the next but-ton. Participants in the no-translate condition saw the fol-

    lowing instructions: We are interested in peoples reactionsto advertisements. Please take some time to look at the nextadvertisement. Look at the ad as you would normally lookat ads when you are reading a magazine. When you are fin-ished, click the next button. The first advertisement was aprint advertisement for a tire-cleaning product and wasincluded so that participants would be comfortable with thetask of viewing advertisements and, in the translate condi-tion, translating copy.

    Exposure to the first advertisement was followed by asecond set of instructions that were identical to the firstexcept that all respondents were also instructed as follows:This time, please keep track of all your thoughts as you are

    looking at the advertisement. The second stimulus was aprint advertisement for a fictitious restaurant named TheNeighborhood Kitchen. We chose the restaurant categorybecause it is neutral with regard to consumption context;people can eat at a restaurant with family or friends, bythemselves, or with coworkers. The advertisement read,Imagine ... Organically grown chicken, flame cooked in asizzling cast iron skillet, with a picture to match. We chosethis ad copy to encourage ideation, but the object ofideation was the product itself. The advertisement did not

  • 7/29/2019 Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

    5/16

    Advertising to Bilinguals / 73

    ask for self-referencing of any kind. For the actual stimuliused in this study, see Figure 1.

    Measures. Respondents were asked to write down everythought they remembered having as they looked at the sec-ond advertisement. Thoughts about friends or family mem-bers, in general or specifically, and thoughts about the homeor the respondents native country were coded as 1; all otherthoughts were coded as 0. The thoughts were coded by twoindependent coders who were blind to the condition fromwhich they were elicited, and conflicts were resolvedthrough discussion.

    Thoughts also were coded according to Sauer, Dickson,and Lords (1992) more general scheme. This scheme codeseach thought across four dimensions: (1) target of thethought (product, brand, advertisement, or other, includingself), (2) type of thought (expression of intentions, expres-sion of feelings toward product or advertisement, expres-sion of usage consequences, expression of beliefs, or othertypes of thoughts), (3) personal (self-) relevance of thought(personalized to the self, personalized to others, or a deper-sonalized third person), and (4) polarity of thought (posi-tive, neutral, or negative).

    The purpose of coding thoughts according to Sauer,Dickson, and Lords (1992) recommended method was todetermine whether the expected change in specific thoughtpatterns (with respect to FFHH-related thoughts) wouldextend to a more general level. For example, FFHH-relatedthoughts should mostly be positive and self-relevant, buteven if so, these thoughts may simply displace other posi-tive and self-relevant thoughts that people could have inresponse to an advertisement, such that there is no netchange in the general profile of thoughts as characterized by

    Sauer, Dickson, and Lord. If this occurs, the choice ofnative versus second language for an advertisement mightinfluence the specific thoughts of bilinguals in response tothe advertisement without differentially influencing theirattitudes and purchase intentions.

    Hypothesis Test

    Table 1 shows the results for thought listings in the translateand no-translate language conditions. The results indicatethat the mean number of thoughts the respondents listed didnot differ significantly between language conditions (trans-late = 2.69, no-translate = 2.97; t = 1.16, not significant

    FIGURE 1Study 1 Stimuli

    A: Practice Advertisement B: Focal Advertisement

  • 7/29/2019 Advertising to Bilinguals: Does the Language of Advertising Influence the Nature of Thoughts?

    6/16

    74 / Journal of Marketing, September 2008

    TABLE 1

    Thought-Listing Results: Study 1

    Translate Condition (n = 39) No-Translate Condition (n = 43)

    Total number of thoughts 105 128

    Mean number of thoughts per participant 2.69 2.97

    FFHH-related thoughtsa 10.5% 1.6%

    Target of Thoughts

    Product 19.1% 11.7%Brand 20.0% 22.6%

    Advertisement 53.2% 53.2%

    Other 7.5% 12.6%

    Type of Thoughts

    Intention 8.6% 4.7%

    Feeling 30.4% 24.2%

    Consequences .9% 2.4%

    Belief 28.6% 27.3%

    Other 31.5% 41.3%

    Personal (Self-) Relevance of Thoughtsa

    Personalized self 76.1% 70.3%

    Personalized other 10.5% 1.6%Depersonalized 13.3% 28.1%

    Polarity of Thoughts

    Positive 29.5% 22.6%

    Neutral 43.7% 57.0%

    Negative 26.6% 20.4%

    Positive personalized thoughts (self or other) 25.8% 19.6%

    Positive thoughts about the product 10.5% 3.9%

    Positive thoughts about the ad 5.7% 9.4%

    Positive thoughts about the brand 11.4% 6.3%

    Positive intentions 4.8% 2.3%

    aDifferences between conditions are significant at p< .05.

    [n.s.]). This suggests that the request to translate did notaffect participants overall level of cognitive elaboration inresponse to the advertisement. However, whereas only 1.6%of thoughts listed in the no-translate condition related toFFHH, 10.5% of thoughts listed in the translate conditiondid. This difference is significant (t = 2.80,p < .05). Thus,the findings support H1.

    By way of example, the following are some of thethoughts coded as FFHH-related thoughts:

    I was reminded of the delicious food my mother makes;

    ... about my aunts fried steak fingers;

    ... chicken cooking in the skillet and about neighborhoods inMexico;

    ... chicken looked tasty, like something my mom could makeat home; and

    I remember thinking how people back home really growchickens.

    Additional Results

    At the broader level of thought coding (Sauer, Dickson, andLord 1992), language condition had a significant effect onthe personal relevance of thoughts (2 = 14.4, d.f. = 2,p