|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 EXPLORING ADVERTISING CAMPAIGNS AND BILINGUAL CONSUMERS: STUDYING THE EFFECTS OF CODESWITCHING AND THE ACCOMMODATION THEORY By KEITRA E. HARRIS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ADVERTISING UNIVERSITY OF FLORIDA 2009
2 2009 Keitra E. Harris
3 To future bilingual advertising campaigns
4 ACKNOWLEDGMENTS I would like to express my gratitude to all th e people who helped me in the completion of my Masters degree. First, I am eternally grateful for my family. To my wonderful parents, thank you for encouraging me to pursue my Mast ers degree and for your unconditional love and support. To my brother Devin, thank you for be ing the best friend a sister could ever have. I would also like to thank everyone who participated in my research. To Mari Luz, thank you for answering all my questions, for providing stress relief, and for being such a great friend. To all the participants in the experiment, thanks for taking the time to answer a few questions without you, I would not have any data. To a nyone I may have forgotten, thank you for being there when I needed you. Lastly, and certainly not least, I would also like to thank my thesis committee. To Dr. Michael Weigold and Dr. Cynthia Morton, thank you for agreeing to be part of my committee, for your time and support, and for your kind encouragement of my ideas. I would also like to give a special thanks to my chair Dr. Jorge Villegas, thank you for your kindness, your patience and your guidance.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...........................................................................................................4 LIST OF TABLES................................................................................................................. .....7 ABSTRACT....................................................................................................................... ........8 CHAPTER 1 INTRODUCTION ...............................................................................................................9 2 LITERATURE REVIEW...................................................................................................14 Some Background on the Res earch.....................................................................................14 Linguistic Theories of Communi cation ...............................................................................15 Psycholinguistics and Sociolingui stics ................................................................................23 The Accommodation Theory..............................................................................................31 Limitations of Previous Studies & Wh at This Thesis Will Contri bute.................................37 Hypothe ses..................................................................................................................... ....39 3 METHOD....................................................................................................................... ...42 Experimental Design..........................................................................................................43 Questionna ires ............................................................................................................. 43 Experimental Stimu li...................................................................................................43 Subjects....................................................................................................................... 46 The Experiment...........................................................................................................46 Control for Confoundi ng Variab les ............................................................................. 47 Operational Definitions......................................................................................................48 Independent Variab le: Bilingua lism ............................................................................. 48 Covariates...................................................................................................................50 Dependent Variables...................................................................................................51 Reliability a nd Valid ity ...............................................................................................52 Analysis......................................................................................................................5 5 4 RESULTS...................................................................................................................... ....57 Sample Profile................................................................................................................. ...57 Hypothesis Testin g............................................................................................................. 57 Hypothesis On e Resu lts ...............................................................................................57 Hypothesis Tw o Resu lts ..............................................................................................59 Hypothesis Thre e Resu lts ............................................................................................59 Hypothesis Four Resu lts .............................................................................................. 60
6 5 DISCUSSION AND CONCLUSIONS ...............................................................................67 Discussion..................................................................................................................... .....67 Implications................................................................................................................... ....69 Limitati ons.................................................................................................................... .....70 Future Research................................................................................................................ .72 APPENDIX A EXPERIMENTAL QUESTIONNAIRES ...........................................................................74 Initial Stimulus A dvertisem ent...........................................................................................74 Experimental Conditio n Advertis ements............................................................................75 English-only/Control...................................................................................................75 Spanish-only...............................................................................................................76 Spanish to English Codeswitch....................................................................................77 English to Spanish Codeswitch....................................................................................78 B QUESTIONNAIRES .........................................................................................................79 C IRB PROTOCOL AND CONSENT FORM .......................................................................98 LIST OF REFERENCES........................................................................................................100 BIOGRAPHICAL SKETCH...................................................................................................104
7 LIST OF TABLES Table page 4-1 Demographic makeup of subject s by experimental conditio n.........................................61 4-2 Multiple Regression of Familiarity Dependent on Language Proficie ncy .......................62 4-3 Multiple Regression of Attitude Toward th e Ad Dependent on Langua ge Proficiency...63 4-4 Multiple Regression of AAd Language Dependent on Language Proficie ncy .................64 4-5 Analysis of Variance for Familiarity De pendent on Ethnic ID and Accultu ration...........65 4-6 Analysis of Variance for AAd Dependent on Ethnic ID and Acculturation.....................65 4-7 Analysis of Variance for AAd Language Dependent on Ethic ID and Accultu ration ......66
8 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Advertising EXPLORING ADVERTISING CAMPAIGNS AND BILINGUAL CONSUMERS: STUDYING THE EFFECTS OF CODESWITCHING AND THE ACCOMMODATION THEORY By Keitra E. Harris August 2009 Chair: Jorge Villegas Major: Advertising Codeswitching occurs when a bilingual person, namely Spanish-English bilinguals switch between English and Spanish during a conversatio n, even within a single sentence. Although there are many recent studies about advertising to Spanish speakers, there is limited research on the specific topic of codeswitching and how it mig ht be used in advertising. Advertisers are already engaging in codeswitching in their ads in the hopes that their reflection of colloquial speech might improve attitudes towards advertiser s and their brands (e.g. McDonald's, Ford). This thesis tests four different hypotheses that examine the relationship between advertisers efforts to accommodate to Hispanics through c odeswitching and how these consumers perceive these ads. The methodology involves an experimental design in which subjects attitudes toward four types of advertising slogans (English only, Spanis h only, Spanish to English switch, and English to Spanish switch) are measured based on their level of bilingualism, strength of ethnic identification and level of acculturation. The results indicate a weak relationship between the accommodation effort of a codesw itched ad and the positive attitude of bilingual consumers. The results suggest that bilingual consumers are indeed complex requiring more than linguistic cues for advertising to have a more meaningful effect.
9 CHAPTER 1 INTRODUCTION Scrubs is a popular television show about a group of young doctors at a learning hospital. The cast is somewhat diverse featuring character s from a variety of ethnic backgrounds. Actress Judy Reyes plays the role of Dominican nurse Carla Espinosa. Although she is very fluent in English, Carla often peppers her language with Spanish terminology first to assert her ethnicity and second to illustrate a particular mood. She acknowledges her use of Spanglish as a means of power amongst her friends and colleagues. In the episode Her Story II (Schwartz, 2006) in which Carla narrates the story line, she and a group of friends go to a college club. Carla wants to drink away her feelings of being over the hill but the other women feel out of place and want to leave. The audience hears Ca rlas thoughts: Time to dial-up the Spanish. It always makes them feel guilty that they havent spent more time with minorities. She pleads to her friends, Cmon mis amigas [I really want to stay here], and she orders a round of drinks to solidify her request. Nurse Carlas internal dialogue and interacti ons with the other cast members is not the fantasy of an imaginative writer, it is a true refl ection of real life for bilingual individuals. This phenomenon is often referred to as Spanglish in po pular culture. The word itself is a combination of the English terms Spanish and English. Ther e is a Spanish equivalent, inglaol but this term has not gained as much popularity. However Sp anglish is a term that has come to mean a variety of bilingual linguistic c onsequences including the lexical use of false cognates (e.g. rentar means to rent, but the appropriate Spanish word is alquilar); calquing or loan translation (e.g. the use of librera to mean a library, when in standard Spanish librera is a bookstore. The appropriate word would be bibliot eca.); and the use of new vocabulary that has no linguistic match (e.g.correo electrnico is a calque for the English email). The term also
10 refers to what Silva-Corvalan (2003) calls a s implification or reduction of the Spanish language (e.g. the limited use of the subjunctive verb tense). Most importantly to this paper, Spanglish also refers to the sociolinguistic phenomenon of c odeswitching. David Luna and Laura Peracchio have conducted extensive research on bilingual individual behavior and codeswitching. Luna and Peracchio (2005a) define the word as the insertion of a foreign word or expression into a sentence (e.g. into an advertising slogan), re sulting in a mixed-language message (p. 760). Grosjean (1982), as cited in de Jongh (1990) extend s this definition: switching is different from borrowing a word from the other language and interpreting it phonologically and morphologically into the base language. In c ode-switching, the switched element is not integrated; instead there is a total shift to the other language (p. 276). How might interest in bilingual speech patterns be of interest to social researchers and marketers? U.S. Census data (2000) reveals that approximately 35 million or 13% of the population is composed of self-reported Hispanics or Latinos. A more recent American Community Survey (2006) noted an increase in the population of almost 10 million people in just six years. The rise of the Hispanic/La tino population implies that new immigrants, and next generation individuals must assimilate into the culture of the United States, which includes learning the language and adopting the customs. Hoyer and MacInnis (2007), citing Koslow, et al. (1994), describe this assimilation as acculturati on or how members of a subculture [learn] to adapt to the host culture [by acquiring] knowledge, skills and behavior through social interaction, modeling the behavior of others, and reinforcement (p. 317). Targeting Hispanics/Latinos is as difficult as it is to target many of the consumer segments within the United States without taking race an d ethnic identity into account. In fact, merely finding an appropriate label for the ethnic gr oup sparks debate among themselves and those
11 outside the ethnicity. Some have described either term to be identifiers of people from a specific geographic region of the Spanish-language countri es in Central and South America, Africa and Europe. Several studies (Deshpande et al., 1986; Koslow et al., 1994; Silva-Corvalan, 2003; Hoyer and MacInnis, 2007) divide Hispanics/Lati nos into three main categories based on their level of acculturation or assimilati on and the intensity of ethnic identification (Deshpande et al., 1986): (1) highly acculturated/assimilated individuals who were most likely born in the U.S. and speak mostly in English; (2) bilingual or bicult ural individuals who straddle the divide between the cultures and the language; and (3) traditiona l individuals who are more likely to speak Spanish and identify strongly with Hispanic/Latino cultures or values. Multicultural studies have revealed, however, that individuals may belong to any of these groups no matter where they were born, or what generation they belong to because of other socioeconomic factors such as family ties, education, occupation and income. A rapidly increasing and diversified bilingual population fosters a need for better communication towards and among these indivi duals. Current research and strategy by advertising agencies and advertisers have rev ealed that speaking ones language means more than simply targeting Latinos in Spanish. Jeff Valdez, cofounder of S TV, a Los Angeles based television cable network is quoted in the Washi ngton Post (Valdez, 2005) in an article about Spanish language advertising; When a lot of peopl e talk about the Hispanic market, they really mean the Spanish-speaking market, which is only part of the market youre only reaching a segment of a very big population ( 6-7). Jeff Valdez applauds recent real world experiments by marketers like Vehix.com who are inserting Span ish-language ads in Engl ish-language television spots. This particular advertis ing tactic recognizes that bilingua l highly acculturated Latinos are
12 not always watching Spanish-langua ge television. For Vehix.com, it has been a success they have noticed a 300% increase of Latino users visiting their website (Valdez, 2005). The Ford Motor Company is following suit. Last year, the company hired Miami-based Zubi Advertising agency to work on mixed language spots for the 2007 Ford Edge SUV. According to an article on Adweek.com (Lema nn, 2006), the spots were aired on major cable networks with a majority Latino audience, such as CNN en Espaol, Discovery en Espaol, and Telemundo and featured codeswitching in action. Perhaps the most obvious example of codeswitching in the media is on the radio. Latin urban radio, a relatively new format, serves as a major platform for English, Spanish, and Spanglish songs to coexist (Ben-Yehuda, 2007, 9). Not only are popular songs showing how codeswitching can communicate solidarity with the young bilingual community, radio personalities and local commercials constantly switch from English to Spanish to communicate with a diverse audience. Television is borrowing the radio format and hopes to have the same success. Long-time cable network Telemundo aired their first bilingual talk show, Mas Vale Tarde (Better late than never.) in November 2007 and hopes to target bilingual bicultural viewers (Wentz, 2007). Many in the profession see the codeswitching movement as a way to tell Hispanics/Latinos, and even other ethnicities, that its cool to speak Spanish (Ben-Yehuda, 2007, 16). Making it cool to speak Spanish might be a way for advertisers to reach Hispanic/Latino consumer segments. It is also a way for Latino/Hispanic culture to become part of the mainstream. Integrated marketing agency CMO has coined the term interacculturation to infer that when a new culture is introduced into a host culture, each will inevitably affect and influence the other (MacFarlane and Semple, 2007, para. 3). McFarlane and Semple (2007)
13 write that the assumption that Latinos will eventually assimilate and leave their cultures behind is an understatement of the affect of the melting pot that is the Un ited States. Codeswitching lies in the center of this phenomenon.
14 CHAPTER 2 LITERATURE REVIEW Some Background on the Research Early research concerning the growing minority population (Kassarjian, 1969; Wilkes and Valencia, 1989) notes that despite growing numbers, Hispanics were largely underrepresented in the traditional media. A content analysis of prime time television on the major networks by Wilkes and Valencia (1989) revealed that the re presentation of Hispanics in advertising did not mirror their proportion in the general population. The results of this study still apply, however advertising and media planners ar e making great strides to create a more realistic reflection of Hispanics in the media. Many advertisers of national products are reducing media expenditures and allocating budgets towards Spanish-language advertising campaigns, however special attention to bilingual phenomena will result in more effective multic ultural advertising. In his doctoral dissertation about how adve rtising in bilinguals involves controlling thoughts, Jaime Noriega (2006) discusses how langua ge can become an important part of an experience that its usage can create and influe nce thought. This basic principle has also been cited in Koslow, Shamdasani, and Touchstone s (1994) study of the Whorfian hypothesis of linguistic relativity in that language serves as a kind of schema through which the world is understood (p. 575) Noriega maintains the same idea in that context can cue the language we use to communicate (p. 6). The use of a particular language and the form of this usage exists in the linguistic use of codeswitching. Luna et al. (2005a.), MendietaLombardo and Cintron (1995) cite Carol MyersScottons Markedness Model (2002) as a mean s to explore the use of codeswitching in persuasive messages and poetic messages, respec tively. The model states that a codeswitched word becomes marked in that this is a consci ous choice to signify an individuals social or
15 ethnic identity and the meaning behind this identity. The codeswitched element sparks a contrast within the language. Mendieta-Lombardo et al. (1 995) cite particular codeswitched lines from poems that either (1) function as indices of specific Latino realities or cultural traditions, e.g. She was just a young woman. Another Puertorr iquea among many in Sandra Maria Esteves So Your Name Isnt Maria Christina (p. 567); (2) help to evoke identification with everyday scenes from any Hispanic community in the U.S. e.g. Mama looks real pretty today even doa olga looks pretty Mama says/ <
16 that are the most different from the first language (p. 22). Although the contrast analysis hypothesis applies to English-prof icient students in formal classroom settings, the basic idea is applicable to any situation of acquired bilingua lism. Hispanics/Latinos who learn Spanish or English as a second language ofte n look for matches between their first language and the second. When one cannot be found, it is difficult to learn the new lexicon, syntax, or part of speech, and individuals look for shortcuts to help them lear n or to help them avoid the use of misunderstood language transference. The contrast analysis also relates to the Ma rkedness Model for monol ingual individuals in that those linguistic elements that occur the most often are least marked compared to the elements that rarely occur in conversation and writing. Acquisition is more difficult for these highly marked elements because students have nothi ng to relate them to in their first language and these elements appear to be the most obvious hurdles for bilinguals (Koike and Klee, 2003). Silva-Corvaln (2003) describes ma rkedness and contrast analysis as examples of bilinguals who are exposed to a reduced input of the Spanish language. In her extensiv e conversations with multiple generations of bilingua l individuals, Silva-Corvaln (2003) has noted that late generation bilinguals find it difficult to learn certain Spanish elements because they were not exposed to them as children, when they are they most impressionable towards learning language. As a result these individuals resort to examples of Spanglish use, an action the author notes as a weakening of the language. The use of Englis h terminology or English pragmatics directly relates to a bilinguals tende ncy to relate their second language to their first. David Luna and Laura Peracchio have extende d basic linguistic theories to hypothesize about particular moderators that increase understanding of mixed la nguage advertising, advertising messages that can be made more persuasive through th e use of linguistic models, and
17 the effectiveness of codeswitching through structural constraints. In their study of moderators of understanding of mixed language adve rtising, researchers investigated the application of the Revised Hierarchical Model (RHM) which explains how bilinguals process words corresponding to two languages (Luna and Per acchio, 2001, p. 284). The RHM model implies that processing a message in an individuals s econd language is always in relation to prior knowledge of the first language. According to th e RHM, individuals have difficulty processing second-language stimuli and as a re sult have a weakened memory of the message, compared to the processing and memory retention of the first language (Luna and Peracchio, 2001). The researchers conducted two studies that suggest that conceptual processing and enhanced memory for second language messages can be enc ouraged by designing ads in which the picture and text are highly congruent (p. 284-285). The second study added a second moderator to processing and memory motivation. After a pa rticularly complex experiment, the discussion noted that motivation to process an advertised message would increase memory, but only if there also existed congruency between the picture and the text. The conditions of Luna and Peracchios resear ch used ads in which the one language was consistent throughout the experime ntal materials. They list other limitations of their study and emphasize that the experiments also featured rela tively simple print advertisements. Traditional print media is perhaps the most useful forum for observing linguistic phenomena because the words are read and often can be referred to on multiple occasions to increase understanding and motivation. OGuinn and Meyer, as cited in L una and Peracchio (2001), Wilkes and Valencia (1989) and Hernandez and Newman (1992), infe r that Hispanics are heavy consumers of television and radio (Luna and Peracchio, 2001 p. 294). Picture and text congruity and motivation might have different effects if the message were more complex, more interactive, and
18 more visual as in television vehicles. Finally the study opens the arena for future research related to the idea of how bilinguals acquire their second language and the context of the advertisement. For example, the effect of classr oom learning, or the context of buying high-tech items, in which most words are consistent in mu ltiple languages or in which there are no Spanish equivalents (e.g. Disco compacto in Spanish is the equivalent of compact disc in English. In many situations, even Spanish monolinguals will simply use the term CD.) In their second study of linguistic affects on advertising, Luna and Perrachio (2005a) attempt to use the Myers-Scotton Markedness Model (2002) to describe the effect of codeswitching on persuasion. R ecall the relationship between the linguistic practice of codeswitching and its relation to markedness, or the salience of the linguistic element. This second language effects study aims to measur e existing schemas and attitudes toward the majority or minority language in terms of the persuasiveness of the message. According to Luna et al. (2005a) the term major ity language (vs. minority langua ge) denotes the language spoken by the group that holds the political, cultural, and economic power within a country (p. 761). In the U.S. the majority language is English, and highly acculturated, weak Hispanic identifiers often associate more positive feelings toward Englis h than they do Spanish. In fact, interaction with the host culture and assimilation tends to lead to inferior feelings about ones own ethnic identity (Koslow et al., 1994). Armed with this knowledge, Luna and Peracchio (2005a) hypothesize that codeswitched messages are more pe rsuasive when they involve a particular direction from minority to majority versus fr om majority to minority languages if attitudes toward a language are consistent with their research. They have termed this phenomenon the code-switching direction effect (p. 761). The researchers also hypothesize that more positive associations toward a particular directio n will lead to higher product evaluations.
19 Understanding this hypothesis may be clearer if we use some examples. Consider this advertising slogan taken from Luna and Peracchio s (2005d) experiment w ith fictional products: In my kitchen, I would never make coffee with any other coffeemaker (p. 49). An example of a message that is a switch from minority to majority would be: En mi kitchen nunca hara caf con ninguna otra cafetera. On the other hand, a majority-to-minority switch might read: In my cocina, I would never make coffee with any other coffeemaker. In the experiment, participants were exposed to a number of code switched slogans, similar to the example. The researchers manipulated attitudes toward codeswitching, or ACS, by giving participants newspaper articles that revealed fi ctitious research touting the importance of one language over another (Luna et al 2005a, p. 763). The results show that in general, respondents maintained a positive attitude to ward the majority language, th an the minority language. This would appear to be an appropriate response fo r highly acculturated/assimilated individuals. The participant pool consisted of Hisp anics who were either foreign-born first generation Americans, which implies that the participants were not very highly acculturated. Hoyer and MacInnis (2007) insist that acculturation is quite labored the level of acculturation increases to a high level in four generations (p 319). Wallendorf and Reilly (1983) also maintain that full acculturation, or assimilation, has multiple le vels and takes place ove r multiple generations. There are a number of explanations for this result. De Jongh (1990) notes that observing codeswitching is a difficult task in social research, when identified observers make speakers conscious of what they are being exposed to and th eir response to it. In addition, high levels of exposure to various media, such as English-la nguage television programs and participating in online chat rooms and blogs might increase the process of assimilation. Perhaps codeswitching in
20 and of itself is a cause of high acculturation, however there is little research to support this explanation. Finally Luna and Peracchio (2005c) attempt to study the structural constraints of codeswitched messages that is, despite its seemingly ungra mmatical context, codeswitched messages must follow the grammatical rules of syntax in order to be processed and have some effect on the receiver. The research is based on the Matrix Language Frame, or the MLF, another model by Carol Myers-Scotton,(Luna et al, 2005c). Previous research by Grosjean (1982) insist that codeswitching is a relativ ely unconscious process, however Luna and Peracchio insist that bilinguals choose to codeswit ch to communicate a partic ular attitude toward the matrix or embedded language. In this case, ra ther than experimenting with the direction of majority to minority, the matr ix language is English and the embedded language is Spanish. According to the MLF model, the rules of English syntax will apply to a codeswitched message. The MLF is based on two underlying rules, the morpheme order principle and the embedded language isla nd hypothesis. To understand either rule requires some basic understanding of linguistics. In basic linguistics, phonemes are the smallest unit of sounds that create meaning. These phonemes are combined to create morphemes, or the basic units of language. For example a word like incoming cont ains three morphemes: (1) in; (2) come; (3) ing. The MLF model specifies two types of morpheme s, system and content, that help define the parts of a message. System morphemes are the qua ntifiers or determiners of speech that link various elements of a sentence together and help to form their meaning. Content morphemes are the nouns and verbs that convey the central meaning of the utterance (Luna and Peracchio, 2005c, p. 417). For example in the sentence: A student writes a paper on a computer, the system morphemes are -s, on, and a. The content morphemes are student, paper, and
21 computer. In this division of morphemes, content morphemes may be replaced with codeswitched elements as long as the system mor phemes of the matrix la nguage remain in place. The morpheme order principle continues by stating: the sequential order of the morphemes must follow the order of the matr ix language, not the order of the embedded language (Luna et al., 2005c, p. 417). This is especially essential when codeswitching from English to Spanish because the syntax of each language is different. For example, in the phrase intelligent student, a codeswitched element mu st read intelligent estudiante although in Spanish, this phrase would read student intelligent (Estudiante intelligente is the correct Spanish syntax). The embedded language island hypothe sis is an extension of the morpheme order principle in that a speaker ignores the rules of syntax for a more pragmatic meaning. A codeswitched element becomes an entire phrase in the embedde d language that follows the embedded language grammatical rules. It is said to be an isla nd because one languages grammatical context is surrounded by another language. For example, in the sentence An intelligent student writes her paper on a computer, the embe dded language island might read: An intelligent student writes her paper en una computadora, rather than An intelligent students writes her paper en a computer. In this sentence, en is a system morpheme and cannot be altered. Finally, Luna et al. (2005c) posit that although codeswitching must follow a strict structural context, bilingual individuals might skip the codeswitched term/phrase and its underlying meaning if: (1) the context helps de fine the embedded language meaning and (2) individuals are processing conceptually vs. pro cessing in a data-driven mode. In two betweensubjects, mixed design experiments, the researchers concluded that grammaticality is a factor in persuasiveness, but only when individuals are act ively paying attention to the language of the
22 slogans, and understand the advertisers cu ltural meaning by using codeswitching. The researchers discuss the importance of ensuring that efforts designed to target linguistic minorities are well conceived, including the seemingly smallest details (Luna, Lerman, and Peracchio, 2005c, p.422). All three of these studies by David Luna, et al. discuss multiple linguistic rules and theories that may be extended to the use of codeswitching in advertising. There are probably many more applications of linguistic theory to codeswitching including Koike and Klees (2003) hypotheses based on the formal acquisition of a second language. These hypotheses include factors such as the elements an individual affixe s his/her attention, inte raction between English and Spanish monolinguals and bilinguals, the variab ility of language use in particular contexts, and sociocultural influences. The implications of these social research studies on linguistics and advertising infer that Spanish-langua ge advertising is more than s imply advertising in Spanish in Spanish media, a thought once supported by early bilingual research (Wilkes and Humberto, 1989; Hernandez and Newman, 1992). They also confirm Mendieta-Lombardo et al.s (1995) three connotative meanings of the use of code switched or mixed language advertising: (1) a reflection of cultural tradition; (2) exhibition of common Hispanic/Latino events; and (3) a reproduction of speech patterns. These studies also raise questions about how advertisers approach bilingual messages are they going through these various processes; or are they merely mirroring the language of their intended audi ence without consciously reviewing linguistic theories? Luna et al.s (2005c) study of the MLF model reveals that copyw riters and advertising planners may not be considering the implications of the role of grammar in advertised messages. Finally every study implicated measures related to the accommodation theory, an aspect of multicultural marketing that will be discussed further.
23 Psycholinguistics and Sociolinguistics The best method to relate codesw itching to the various studies of linguistics is to define the basic principles of each. The American Heritage Dictionary (2006) defi nes psycholinguistics as the study of the influence of psychologi cal factors on the development, use, and interpretation of language. The same dictionary defines socio linguistics as the study of language and linguistic behavior as influenced by social a nd cultural factors. The studies that have been discussed so far have fulfilled both of these definitions. Language and its denotative and connotative meanings involve the rules of lin guistics, psychological factors of interpretation and social influences. This implies that codeswitching creates a deeper meaning than what is written on the surface. Koslow, Shamdasani, and T ouchstone (1994) indicate an important factor that relates psycholinguistics and sociolinguistics: Hispanics value the use of Spanish less for what the advertisements communi cate about products than for what Spanish usage signals about the importance of Hispanics as consumers (p. 582). The level of acculturation can be seen as a measure of the importance of Hispanic consumers. Much of the research concerning th e affects of advertising to bilingual consumers has used a number of factors in determining Hi spanic acculturation in relation to language. The question of whether or not to advertise to Hispan ics in English or in Spanish requires a complex answer, not only related to la nguage fluency and level of acculturation, but also the product category, the type of media vehicles to be us ed, and geographical location. Researchers must remember the danger in targeting Hispanic s/Latinos as a homogenous group because they descend from a common culture, when clearly ther e exists differences among them that extends beyond kinship, political solidarity common cust oms, language, religion, values, morality, and etiquette (Deshpande, et al., 1986, p. 214). De shpande, Hoyer, and D onthu introduce another
24 important factor in the study of bilingualism and th e sociological aspects of Hispanic advertising the intensity of ethnic affiliation and what it means in terms of consumption patterns. Deshpande and colleagues (1986) maintain that ethnicity is defined by a number of indicators that are both objective, or defined by the people about themselves, and subjective, or a classification of Hispanic consumers based on sociocultural categories such as education, income, media consumption, and origin. In another related study, Stayman and Deshpande (1989) posit that ethnicity is not just who one is but how one feels in and about a particular situation (p.361). Traditionally, individuals w ill identify to which ethnic group they belong, which the researchers call self-designated ethnicity ; and individuals will indicate how strongly they feel about their identity in that group, termed felt ethnicity. However the former study does not create a measure of ethni c affiliation and only suggests that some other research should create an index of those indicators of ethnic af filiation and the intensity based on self-reported data and assigned categorization of participants. Because they do not create a universal measure of intensity of ethnic affiliation, the associated conclusions of their survey study, including political affiliation, brand loyalty, and ethnic prid e, can only be confirmed in relation to the specific participant sample used in the study. In the latter study, ethnicity and the strength of ethnic identification were considered intermediary variables that affect how an individual behaves in certain social situations. The researcher s found that subjects that were in a heightened state of ethnic awareness, often made purchase decisions that did not seem ethnically relevant when they were in an ethnically diverse group, such as during a business lunch. However the same subjects would be more likely to make et hnically relevant decisions when in homogenous groups of people of the same ethnicity as th e subject, such as dining with parents.
25 Desphande et al. (1986) do discuss a number of limitations to the study. One important limitation and grounds for future research include the idea of varying levels of intensity in ethnic affiliation. The researchers acknowledge that a s cale measure is probably best suited to this Hispanic/Latino identifier, however all the findings were positioned in relation to weak identifiers and strong without an in-between measure. Although Stayman and Deshpandes (1989) research does attempt to eradicate some of the limitations of the previous study, the researchers suggest that future research should fu rther assess the situational effect of social surroundings, such as the presence or absence of others, and physical surroundings, such as the geographic location of the subject (p. 362). Another dimension of acculturation, assimilation, is a concept that defines how a minority or migrating culture becomes a part of the ma jority or host culture (Wallendorf and Reilly, 1983). Wallendorf and Reillys (1983) assimilation st udy developed a model of assimilation that was used to determine the consumption patterns of Hispanic Americans at various stages of assimilation. Their assimilation model offers a compelling argument for the relationship between acculturation and language, and the types of Hispanic consumers that advertisers might target in particular campaigns. The researchers propose th at the traditional ass imilation model has seven levels and that full assimilation indicates that all levels have been achieved by the minority culture. The seven dimensions of assimilation ar e progressive and may overlap. The first four dimensions are of interest to this particular thesis and their basic definitions are outlined as follows (p. 293):
26 1. Cultural assimilation or acculturation: This dimension involves changes in behavior patterns by the minority culture. Indicators of these changes are the adoption of the host cultures language, dress, food and consumption patterns. It is interesting to note that language changes occur at the first level of assimilation. This indicator is a reason why advertisers simply decide to advertise in Spanish when targeting Hispanic/Latino consumers rather than considering other cultural concepts. 2. Structural assimilation: This dimension is defined by the desire of individuals in the minority culture to enter into certain occupa tional categories and majo rity-exclusive groups. 3. Marital assimilation: When intermarriage occurs between the minority culture and majority culture, researchers maintain that the minority culture has achieved marital assimilation. 4. Identificational assimilation: The immigrating culture will develop a sense of identity based on the host society. It is interesting to note here that although it appears to be closel y related to language, ethnic identification occurs much later in the process of assimilation than do changes in language. This progression of assimila tion infers that advertising me ssages with varying levels of language use, whether in Spanish, English or a combination of both communicates to varying levels of Hispanic/Latino consumers in the United States. Although Koslow et al. (1994) admit to not deeply investigating this particular aspect in their research, their final hypothesis predicts that of those Hispanic consumers who more positively identify with the mainstream culture, it is more likely that they would have negative associations with Spanish-language advertisements (p. 478). Their results showed that the more English-dominant the subject was, the more likely they were to perceive the advertiser who chose to advertise in Spanish as culturally sens itive (p. 581). This finding may be related to the more recent surge in ethnic pride among Hispanics who would like to maintain the Spanish language and culture related to its use. This trend is also evident in the New American Dimensions (2008) Hispanic segmentation study in which some of the more acculturated segments identified in the study look for Spanish cultural cues in the media.
27 Ethnic identification introduces a number of vari ous issues related to type and strength of an ethnic affiliation as reported by consumers themselves. The U.S. Census is a reliable source of information about how consumers self-identif y based on ethnic and racial terminology. The latest Census in 2000 noted a special effort to identify Hispanics as an ethnicity and to ask contingency questions of respondents that i nvolved identifying themselves ethnically as Hispanic, and racially as any one of the si x categories including white, black or African American, American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific islander, or some other race as defined by the Office of Management and Budget (OMB), the federal department that approves census questionnaire formation. According to the OMB, Hispanic or Latino is defined as a person of Cuban, Mexican Puerto Rican, South or Central American, or other Spanish culture or origin regardless of race (Grieco and Cassidy, 2001). The questions on the Census reflect both the fe deral governments official view of race and Hispanic origin as separate entities and the Census Bureaus effort to change with developing lifestyles and ethnic makeup (Grieco and Cassidy, 2001). The question about Hispanic origin was asked directly before questions about racial makeup, rather than after other demographic questions and the Census Bureau expected the close proximity of the questions to affect how people living in the U.S. might respond. Grieco and Cassidys (2001) report of race and Hispanic origin as reported by the census offers some engaging figures about the Hispanic population. Approximately 13 percent of census respondents decl ared themselves as Hispanics. Of those 35 million Hispanic ethnic identifiers, 48 percent declared themselves racially as white and two percent as black (p. 10). The census does not measure strength of ethnic identification, however some inferences can be made about the intensity through the pr esentation of the data. For example, the high
28 coincidence of Hispanic identifiers that claim some other race (42 percent) might reflect the Hispanic consensus that Hispanic is a racial identifier and not just an ethnic one. Golash-Boza and Darity (2007) also note the changing nature of racial and ethnic self-ide ntifiers as presented in the 2000 U.S. Census (p. 3). Golash-Boza and Darity (2007) take an interest in the U.S. Census report of predicted projections of population change in the year 2050. The researchers infer that the Census predictions were based on [then-current] patterns of ethnic and racial identification without taking into account the pos sibility that Hispanics racial and ethnic identifications can and do change (p. 900). Thes e identification changes are the result of a number of factors including level of acculturati on, language proficiency, personal experience, and geographic location. There is an overwhelming amount of data that support research motivations to study and report ethnic identification or affiliation based on a number of possible categories for the diverse Hispanic population. Neither the Census (2001) nor Golash-Boza and Daritys (2007) ethnic and racial categories take into account the strength of these identifications on some scale. Early research by Deshpande, Hoyer and Donthu (1986) e xplores the belief that there may exist some fundamental consumption-related differences between Hispanics who identify strongly or weakly with their ethnic group (p. 215). They explore of a number of hypotheses related to attitude toward business and government and use of Spanish media to determine the consumption-related differences between strong and weak identifiers. These hypotheses include the relationship between strong Hispanic identifiers and (1) more frequent use of Spanish language media, (2) brand loyalty, (3) purcha se of prestige products, and (4) likelihood of purchasing products advertised to their ethnic group. Note that the fourth hypothesis is closely related to accommodation efforts toward strong Hispanic identifiers. The New American
29 Dimensions (2008) study provides support for these hypotheses. The six segments identified by the study are characterized by their attitudes towa rd business and government (e.g. Latin Flavor Cubans are more likely to be Republican), media usage, and attitudes toward advertisements (e.g. Accidental Explorers penchant for trustin g Spanish-language advertisements but lack of brand loyalty.). One of the theoretical perspectives of Greens (1999) research on the interaction of strength of ethnic identification, media placement, and ad racial compos ition involves a review of the literature involving strength of ethnic id entification. Green (1999) makes an important link between ethnic identity and acculturation and introdu ces a new theory that might be useful in research concerning the Hispanic/Latino community. He defines ethnic identity as a combination of acculturation related to how members of an ethnic group relate to the group as a subset of the larger society (p. 51). Green exte nds strength of ethnic identification to the theory of in-group bias in which bias towards members of ones own group represents favoritism toward the in-group, but does not imply disfavor toward members of the out-group (p. 51). In this theory social distance is a dimension of strength of ethnic affiliation and the perception of accommodation efforts made by advertisers is affected by the advertisers social distance from the minority target market. Perhaps social distance can by minimized by the choice to use Spanish, English or a combination of both by advertisers practicing accommodation. Language choice is the most engrossing indicator of accommodation theory towards Hispanics. Advertising messages that target Hispanic Americans indicate catering to a culture that is an important consumer group. When advertiser s first began allocati ng media budgets towards Hispanic/Latino bilingual advertising, there arose the question of whether to advertise in English or Spanish. Hernandez and Newman (1992) begin their literature review of those studies that
30 attempted to answer this question by referring to the basic models of communication and language. They state that a fundamental link must occur if the receiver is to interpret a message in the way an advertiser intended. This fundame ntal link is language. At the most basic level, then, advertisers have assumed that advertising to Hispanics in Spanish is the best policy. But issues such as the dialect to be used in the message, and whether or not to create separate campaigns or to merely translate English-only campaigns complicates this approach. Hernandez and Newmans review does not simplify the method in which advertisers target Hispanics, but brings to attention other critical factors that are key in target marketing. The study examined five issues inherent to the choice of language ques tion and created a number of conclusions based on existing literature. Finally, Hern andez and Newman (1992) discuss a number of propositions that might lead to further research or real-impleme ntation and testing. The five issues and their conclusions are discussed: 1. Media Language Preference and Level of Acculturation: The researchers cite early works by Wilkes and Valencia (1983) and (Deshpa nde et al., 1986) and their reliance on the variable intensity of ethnic affiliation as a measure of acculturation outside of language fluency. They conclude that high in acculturation Hispanics are more likely to use English language media than those who are low in acculturation (p. 36). 2. Empirical Research on Advertising Effectiveness: English vs. Spanish: Although later studies by Luna and Peracchio (2001; 2005a,b,c,d ) have confirmed that recall of English for Spanish-language bilinguals can be enhanced by picture and text congruency, the results of the literature review insists that there is no major difference in recall between English-only advertisements and Spanish-only advertisemen ts. This finding is probably due to other advertising elements, such as the use of an ethnic spokesperson, the product being advertised, and the context of the advertisement (e.g. An ad for a South American travel agency during The George Lopez Show .) 3. Language Used by Type of Media: Most of the Hispanic media consumption is devoted to the television, however the audiences am ong Spanish-language programs and Englishlanguage programs are divided in relation to acculturation. Traditional individuals are more likely to watch Spanish-language sta tions than the highly acculturated.
31 4. Growth Projections of Hispanic Audiences by Language Use: The rapidly increasing number of Hispanics/Latinos in the U.S. has already been discussed. This augmentation is not only due to increased immigration but a hi gh birth rate among Hispanic Americans. Media proliferation has increased the assimilation of first generation and immigrating Hispanics/Latinos and has increased their pref erence for English-language vehicles. Spanishlanguage media may only be used to reach ol der, traditional Hispanics and new immigrants and not highly acculturated bilingual consumers. Fu rthermore, ethnic pride is an explanation for why Hispanic individuals do not completely assimilate into the host culture and why advertisers need to create special targeting for them. 5. Marketers Perceptions of the Role of Spanish vs. English Advertising: Despite all the research to the contrary, advertisers are still simply advertising to Hispanics in Spanish in Spanish-language media. Perhaps there is not enough research, or perhaps the population of Hispanic consumers has not reached a level in which their purchase decisions seem very profitable to advertisers, or perhaps the effect of different types of bilingual advertising has not yet been revealed. These five areas of research are important in understanding the psychol ogical and sociological implications of bilingual consumer ism and have led to a number of propositions about how to effectively target a bilingual/high in acculturation audience. Hernandez and Newman (1992) suggest the S panish-only argument says a Hispanic is defined by language alone a Spanish-only advertising policy is based on an oversimplification of the market (35, 43). One of the most interesting propositions that the researchers list in their discussion is the idea that English language ads can be used to target both high in acculturation/English dominant and low in acculturation/Spanish-dominant Hispanics if there is an ethnicized message one that reflects the culture and customs of the audience. Perhaps codeswitching theory might fulfill the requirements of this proposition in that in majority-to-minority messages, the language is English dominant and insertion of the codeswitched element creates an ethnicized message. The Accommodation Theory All of the research cited in relation to codeswitching and adve rtising to bilingual consumers possesses an important link between how thoughts are linguistically communicated
32 and the accommodation theory. Hoyer and MacInnis (2007) define the accommodation theory as one that predicts that the more effort a s ource puts into communicating with a group by, for example, using role models and the native langua ge, the greater the response by this group and the more positive their feelings (p. 320). Koslow Shamdasani, and Touchstone (1994), have solidified the link between the application of the accommodation theory and attribution of positive consumer attitudes to a culturally sensitive advertiser. The use of codeswitching indicates connotativ e meanings to the receiver including the reference to cultural traditions, the identificatio n of scenes from Hispanic communities, and the realistic reproductions of everyday speech. Koslow et al. (1994) notes that the framework for testing the accommodation theory in consumer behavior requires the borrowing of sociolinguistic theory. At this point, researcher s can observe how the accommodation is directly related to linguistic theories of communicati on and psychoand sociolinguistic theories. In their proposition of a research question in terms of accommodation theory, Koslow et al. (1994) writes that: the use of Spanish language in advertising to Hispanic consumers triggers the dynamics of accommodation theory. For accommodation to occur, however, Hispanics must perceive the choice and use of Spanish in the advertisement as an indicator of the advertisers respect for the Hispanic culture and desire to break down cultural barriers through reduction of linguistic dissimilarities. (p. 576). The researchers do predict that the effect of accommodation is directly related to the use of Spanish. The more Spanish used in an advertisement, the higher the level of accommodation. This direct correlation does not imply that the use of Spanish language means the exclusive use of Spanish. In fact, Koslow and his colleagues observe a limit to the use of advertising solely in Spanish. Multiple hypotheses relate to the idea of a la nguage-related inferiority complex of Spanishlanguage speakers. Advertising cannot always a ssume that Hispanic consumers have positive
33 feelings toward their culture and language. Th e process of assimilation includes obstacles in which the host culture may discriminate against the Hispanic culture and disparage the exclusive use of Spanish in formal settings. In relation to the intensity of ethnic affiliation, these linguistic insecurities are most present among highly acculturated, weak identifiers. The stimulus material used in the personal interviews conducted by Koslow et al. (1994) included a scale of advertisements for the sa me product ranging from an all-Spanish to two bilingual to an all-English language. The mixed language ads included one in Spanish with English subtitles, and the other in English with Spanish subtitles; and the bilingual advertisements were needed to investigate the degree of accommodation in response to increasing levels of language usage (p.679). The results revealed that the direct relationship between the increased the use of Spanish and a higher perception of advertiser sensitivity proved the null hypothesis there was no large differen ce between an all-Spanish ad and the mixed language ads. However, when the researchers cont rolled for the effect of perceived advertiser sensitivity, the negative relationship between increased use of Spanish and negative attitudes toward the advertisement held true. The results showed that advertisements with some Spanish would meet the criteria for advertiser cultural sensitivity. Once this has been achieved, advertisers must look for ways to effectively r each these consumers and persuade them towards some action. The conclusions of Koslow et al.s (1994) res earch and others like it have reached the same idea, that in order to target Hispanics, adver tisers simply cannot design advertisements in only Spanish or English. But the idea of accommodation is not limited to language. In fact simply concentrating on the language aspect of accommo dation might only help to address a few segments of Hispanic population. Current resear ch and strategy by advertising agencies and
34 advertisers have revealed that speaking ones language means more than simply targeting Latinos in Spanish. Consider the definition of accommodation as defined by Holland and Gentry (1999) in their extensive research concerning ethnic consumers reactions to target marketing related to intercultural accommodation: When a marketer borrows cultural symbols from ethnic consumers in an attempt to enhance communication with them, and when the consumer recognizes the symbols, makes attributions about their use, has an a ffective response to them, and changes his or her behavior as a consequence, interpreta tion of meaning through a mediated social interaction has occurred (p.68). Note that in both Koslow et als (1994) and Holland and Gentrys (1999) definition, accommodation can only take place if there is some reciprocal action by the consumer: an acknowledgement of the marketer solidarity a nd/or some affective response (e.g. improved attitude toward the advertis er, recalling message content, or purchase intention). In applying accommodation theory beyond la nguage choice, Holland and Gentry (1999) combine a number of research studies related to minority cultures to create an intercultural accommodation model based on those factors that influence the strength and type of reaction of the marketers accommodation attempt, as well as the consequences of that reaction (p. 69). Their model of intercultural accommodation is represented both in graphical form and in a number of important propositions that will be discu ssed further because of their applications in accommodation theory research. The researchers begin with the center of the model: the accommodation response. For a response to occur, the consumer must first notice the message and the related accommodation attempt. Holland and Gentry (1999 ) suggest that advertising clutter can play a major role in preventing accommodation from taking place. It is assumed then that advertising must also be compelling and creative to the target market. Given that the message is noticed by the consumer,
35 the researchers propose that the reaction will consist of two parts: a cognitive response and an affective response in which the cognitive re sponse involves consciously recognizing the accommodation attempt and the affective respons e involves the consumers questioning the motivations for the accommodation attempt. Ho lland and Gentry (1999) do acknowledge that it is possible to have an affective response or to make attributions about the reasons for a message without recognizing the accommodation attempt, a lthough they give little empirical evidence of this effect. We can look at a similar study on the effect of target marketing on targets and nontargets by Grier and Brumbaugh (1999), co incidentally in the same issue of Journal of Advertising as Holland and Gentrys article. According to Grier and Brumbaugh, non-target consumers may evaluate an ad without fully recognizing the accommodation attempt because they are not part of the target market and th erefore cannot decode those cultural cues. Although Holland and Gentry (1999) assume the model will be applied to target market reactions, Grier and Brumbaugh (1999) may adequately explain the formation of an affective response without a cognitive response. Also, Vakratsas and Amblers (1999) explanation of the various models of advertising effectiveness shows that affectiv e-only models depend on emotional and visual elements to enhance preference (p.30). We can assume then that the accommodation may not be very obvious in those instances where a cons umer response is not preceded by accommodation recognition, such as an ad without any copy (in which no copy means there is no question of language choice) or culturally id entifiable models/spokespersons. The next consideration of the intercultura l accommodation model involves the antecedents, or the pools of knowledge and experience that consumers use to evaluate accommodating advertising. Holland and Gentry (1999) recognize six different antecedents, four of which are borrowed from the persuasion knowledge mode l (PKM) proposed by Friestad and Wright
36 (1994). The link between successful accommodati on and persuasion helps to solidify the importance of accommodation in advertising since the various definitions of advertising often involve persuasion as a motivation of mark eters (Richards and Curran, 2002). The six antecedents and a brief definition of each are as follows: 1. Accommodation experience: Previous experience with advertisers attempting successful accommodation and the length of that experience might affect the attributions and affective response of consumers. In terms of length of experience, Holland and Gentry (1999) assume that some ethnic groups have been targeted mo re heavily for a longer period of time than others and therefore may have increasesd experience in handling accommodation efforts. (p. 71) 2. Attitude toward accommodation: Beyond simply collecting accommodation experiences, consumers also rely on stored memories about their attitudes toward a history of exposed accommodation attempts. (p. 71) 3. Agent knowledge: What consumers know about the company or individuals outside of the current accommodation attempt is closely linked to attitudes towards previous attempts (if there are any). Knowledge about the agent incl udes the perception of the communicator to be of the dominant culture or a member of their own ethnic group and whether or not the accommodation attempt is genuine. (p. 72) 4. Newness: This factor refers to perceived novelty of the accommodation attempt. If an advertiser has never before attempted some cultural accommodation, it may be acknowledged more readily than an adver tiser who predictively makes accommodation attempts. (p. 72) 5. Strength of ethnic identification: According to Holland and Gentry (1999), Ethnic groups or individual members who identify strongly w ith their heritage are likely to have an emotional response to the use of cultural sy mbols in marketing communications (p. 72). 6. Attitude toward the mainstream culture: Also referred to the level of acculturation, this factor refers to how a consumer feels about the mainstream culture in which they share some cultural similarities and from which they may be culturally differentiated and how these feelings may affect the attributions c onsumers make about the communicators motives for the accommodation and their response to the attempt. Finally, accommodation attempts are only made successful through measurable consequences or evaluations of the attempt. Holland and Gentry (1999) operationalize the consequences as either (1) development or ch anging of attitudes towa rd the brand, company
37 and/or ad; (2) effectiveness measures such as comprehension and recall; and/or (3) reciprocal accommodative behavior such as purchase intenti on or positive word of mouth effects (p. 73). Holland and Gentrys (1999) intercultural acco mmodation model is very much applicable to accommodation theory. As a final note, the model will enable researchers to identify differences among ethnic groups that may particularly useful for companies that target more than one group (Holland and Gentry, 1999, p. 74). This proposition helps support the hope that codeswitching, which was introduced earlier in the literature review can be an advertising application of multicultural marketing and help promote understanding between cultures and among various segments of the same culture. Limitations of Previous Studies & What This Thesis Will Contribute The literature review noted a number of studies pertaining to Hispanic consumers in the American host culture. There are a few of these studies that are particularly important to the premise of this thesis. The limitations enumerated in these studies are the inspiration for the experimental design. In Koslow, et al. (1994), the researchers noted that future research should consider the intensity of ethnic identification, level of ethni c pride and the degree of assimilation as important factors that interact with language usage a nd that might affect how Hispanics perceive accommodation in advertising messages (p. 583). In the following experime ntal design, language will still be considered the the most important cultural factor, however other ethnic indicators such as the intensity of ethnic identification and the level of assimilation will be controlled by measuring the moderating effects of these factors. Koslow et al. (1994) also suggests that their research did not truly address the effect of t he stigmatization of Spanish usage in the United States (p. 583) or the idea of a language-related in feriority complex. These covariates will also be manipulated to reveal some information a bout language-related inferiority. Subjects will be
38 asked to participate in the experiment in two different external conditions based on the geographic location of the subject: one in which they would consider themselves part of the majority population because they attend a pre dominantly Spanish-language university; and one in which they would consider themselves part of the minority population because they attend a predominantly English-language university. Recall that Stayman and Deshpande (1989) suggested future research in the particular area of the relationship between ethnicity and consumer behavior. The manipulation of the subj ects felt ethnicity through different geographic situations in this thesis is also an expans ion of Stayman and Deshpandes (1989) research on situational ethnicity. This thesis will also expand on Deshpande et. als (1986) suggestion that future research should test the levels of ethnic affiliation as indi cator of how strongly a subject feels about his or her minority culture in relation to other elements. For the research presented in this study, the relative strength of ethnic identification will be used in comparison with the subjects attitude toward an accommodating ad either in Spanish or a Spanish-English codeswitched ad. Codeswitching can serve of number of purposes simultaneously: (1) that it encourages acculturation while maintaining ethnic pride; (2) it answers the language question by incorporating the best of both worlds; and (3) it can create more favorable attitudes towards advertisers not only by the targeted bilingual consumers but by other consumers who might see and interpret the message. Researchers understa nd that there is no singular Hispanic/Latino identifier in the U.S. they must be segmented according to their multiple identities. Codeswitching incorporates linguistic fluency, level of acculturation and intensity of ethnic affiliation; and it can be used a variety of ways that will indicate successful use of the accommodation theory. However there is a need for more research that incorporates these ideas
39 and the use the mixed language advertising (Koslo w et al., 1994) and this study aims to break ground on this lack of research. Hypotheses Language serves both as a communication tool and a means of understanding a particular culture. Sociolinguistic theorist Benjamin Lee Whorf has expanded on the original relationship between language and thought origina lly posited by Edward Sapir. Often referred to as the SapirWhorfian hypothesis of linguistic relativity (K oslow et al., 1994), Whorf quotes Sapir: It is quite an illusion to image that one adju sts to reality essentially without the use of language and that language is merely an incidental means of solving specific problems of communication or reflection. The fact of the matter is that the real world is to a large extent unconsciously built up on the language habits of the group . We see and hear and otherwise experience very largely as we do b ecause the language habits of our community predispose certain choices of interpretation. (Caroll, 1998, p. 134) For Hispanics living in the United States, this passage encompasses their ability to use Spanish, English, and oftentimes a combination to both to communicate to people contained in various circles. For example, a Spanish-English bilingual might use Spanish at home with her parents, English during school hours, and some mix, usua lly involving codeswitching, with friends. Bilingualism is often used a cultural identifier in social research (Deshpande et al., 1986; Grosjean, 1982; Hernandez and Newman, 1992; Kosl ow et al., 1994; Luna and Peracchio, 2001) because it illustrates solidarity among Hispanics beyond a means of communication. That same bilingual student might codeswitc h with friends because the language use communicates her social or ethnic identity, an identity that straddles the divide between the host/majority culture and minority culture and between the majority la nguage and the minority language. In this case, the majority culture and language is American English and the minority is any Spanish dialect. Advertisers that engage in codeswitched messages are communi cating an ethnicized message that reflects Hispanic values and culture (H ernandez and Newman, 1992). The intent is that
40 Spanish-English bilinguals will understand th e deeper cultural message embedded in the bilingual codeswitche d communication. Recall that codeswitched, or mixed langua ge, advertising have three important connotations for bilingual audiences: it is (1) a refl ection of cultural tradition; (2) an exhibition of common Hispanic/Latino events; and (3) a reproduction of natural bilingual speech patterns (Luna and Peracchio, 2001; 2005 a, b, and c; Mendieta-Lombardo, et al., 1995). This is the underlying assumption of the accommodation theory and its use by culturally sensitive advertisers. Considering that bilingual individua ls are cognizant of the codeswitching occurring in an advertising message, they wi ll generate positive feelings towa rd that ad and the advertiser. The link between bilingualism, c odeswitching and the accommodation theory leads us to the first two hypotheses to be tested in this experimental design: H1 : Primarily bilingual individuals of Spanish origin will be more likely than primarily monolingual individuals of Spanish origin to understand the code switched advertising slogan. H2 : Primarily bilingual individuals of Spanish origin will be more likely than primarily monolingual individuals of Spanish origin to have a more positive attitude toward the ad and the advertiser of a code switched advertising slogan. The first two hypotheses emphasize highly bilingual individuals in comparison to lowly or nonbilingual individuals. This research will only be testing bilingual indivi duals, however language fluency differs among Spanish-English bilinguals for a number of reasons including how an individual acquired either language and when an individual uses either language. Therefore the experimental materials will m easure the level of language fluency of English and Spanish to determine the level of bilingualism or language proficiency. This c onstruct will be explained in further detail in the next section. Dual language fluency, or bi lingualism, might affect the participants comprehension of the ad and the participants positive attributions to the ad, therefore this experiment will consider
41 two intervening variables that can be responsible for the various levels of bilingualism. These variables are (1) the level of acculturation and (2) the strength of ethnic identification. Because these constructs are compelling a lternative independent variables that might be responsible for the dependent variable, they will be measured as covariates. H3 : Level of acculturation and strength of ethnic identification are covariates affecting a bilingual individuals understanding of a codeswitched adver tising slogan. H4 : Level of acculturation and strength of ethnic identification are covariates affecting a bilingual individuals attitude toward a codeswitched a dvertising slogan. Hypotheses three and four are fashioned like the first two because of these covariates importance to bilingualism. Although they will be operationalized in the next section, acculturation refers to the level of assimilation in to the majority culture based on Wallendorf and Kellys (2003) assimilation model. Strength of et hnic identification is defined as both an indicator of who one is culturally and how they fe el in particular situations (Desphande et. al, 1986); Stayman and Deshpande, 1989 p. 361).
42 CHAPTER 3 METHOD In order to test the relationship between an accommodating advertiser and the Hispanic consumers attitude toward the ad in terms of these three cultural indicators: bilingualism, or language fluency, le vel of acculturation, and strength of ethnic identification, the methodology for this thesis incorporated an experimental design that manipulated the language of four different language advertisements, Spanish-only, English only, a Spanish to English codeswitch and an English to Spanish codeswitch (Leedy a nd Ormrod, 2005). As mentioned in the previous chapter, there are a number of hypotheses pertaini ng to this relationship and an experimental design was used to test these hypotheses. Every experiment consists of at least one i ndependent variable that is manipulated to determine the effect of the dependent variable (Davis, 1997). This exper imental design contained two independent variables, two de pendent variables and three covariates. Covariates are possible independent variables that might also be used to predict the out come of the dependent variable. In this experiment, the covariates were of interest, but their inclusion served to eliminate some possible interacting or confounding variables. The research question of this study asked how consumers of Spanish/Hispanic or Latino origin would respond to the advertisers who a ttempted to accommodate them through the use of the Spanish language, particularly, in the cond ition of a mixed language advertisement. The independent variable be ing manipulated was the language of the advertisement. Subjects were asked to look at an ad and answer questions a bout their understanding of the ad and their attitude toward the ad and its language. Therefore, language of the ad was used as a predictor for the dependent variable, attitude toward the ad. The following sections describe the constructs that
43 make up the independent and depe ndent variables, and explain the experimental questionnaire items that will be used to measure these variables. Experimental Design This thesis used a nonrandomized pretest-pos ttest experimental de sign with a control group. Its administration is explained in the following sections. Questionnaires The questionnaires began with the persona l inventory section that included basic demographics, information about citizenship, r ace and ethnicity, and la nguage proficiency and usage. The next section of the questionnaire contained the initial stimulus advertisement. The questions following exposure to the initial stimulus ad and following exposure to the experimental ads were very similar. The first set of questions involved a recall and recognition test based on the experimental design by Luna and Peracchio (2001). These two questions served as a validity check to ensure that participants recognized the language of the ad the ad and understood the advertisement and its meaning. Th e second set of questions was a series of questions used to determine internal validity and served as a manipulation check on the state of the respondents. The final set of questions tested the effect of the dependent variable through a series of scaled questions (Koslow et al., 1994; Jung and Kellaris, 2006). Experimental Stimuli Section 2 of the experimental packet i nvolved the manipulation of bilingualism through one initial stimulus and four varying experimental stimuli. The initial stimulus was administered before exposure to the experimental stimuli. Th e language of the advertisement was manipulated to determine the effect of codeswitching and language proficiency on understanding and attitude toward the ad.
44 Participants were shown an E nglish language ad for a fictitious brand in order to create an antecedent state in which they develop an attitude toward the ad and the advertiser based on their primed linguistic and cultural state and the ad presented to them. Priming individuals to recognize their antecedent state, particularly based on culture an d language, is based on Stayman and Deshpandes (1989) research on situational ethnicity. The 1989 study allowed researchers to ensure that participants were answering ques tions based on their cultural background and not on some other criteria. The order of the questionnaire for this thesis helped to motivate participants to think about their language a nd culture while evaluating differe nt language advertisements. For this thesis, the questionnaire that follows the ad was used to test the internalization of the effect of language on attitude toward the ad. After establishing their antecedent state, pa rticipants were exposed to one of four experimental conditions: 1) anot her English language ad for the same product; 2) a Spanish language ad for the same product; 3) an Engl ish-to-Spanish codeswitched ad for the same product; and 4) a Spanish-to-Englis h codeswitched ad for the same product. Those participants exposed to a second English ad were part of th e control group because the English language ad was not considered an accommodating ad. The other three conditions were part of the experimental group. All of the experimental stimuli were part of a campaign for a fictitious brand of chocolate called Chocolate. (see Appendix A). The brand name was selected because the word chocolate is considered an easily understanda ble Spanish-English cognate. Photographs of food products that have chocolate as an ingredie nt were selected from a stock image website and manipulated by the researcher using digital imaging and desktop publishing software. In each of the print ads, the photograph covered the entire pa ge. The brand name was located at the bottom
45 of the page, along with the brand tagline Making delicious moments. Brief body copy describing the use of chocolate in the photograph accompanied each ad. The initial stimulus ad showed a photograph of a chocolate chip cookie being dipped into a glass of milk. The body copy was written in English and read: chocolate chips./ eggs, sugar, flour./ bake at 350./ warm chocolate chip cookies./ a perfect afternoon snack. The four experimental stimuli ads depicted a photo graph of a slice of chocolate cake and a side of raspberries. The body copy for the English-language control ad read: Moist chocolate cake/ a side or raspberries/ a sweet combination/ for an after dinner delight. The tagline of the Spanish-only ad was translated into Spanish, para momentos deliciosos and the copy was a direct translation of the English copy: Un pedazo de pastel chocolate/ acompaado de frambuesas frescas/ una combinacin dulce/ para un deleite despus de la cena. Finally the two codeswitched ads were similar in structure to the English-only and Spanish-only ads with changes in the third line a sweet combination, and at least one other word. The English tagline was also in English for the English-to-Spanish switch and the Spanish tagline was used for the Spanish-to-English switch. The English-to-Spanish and Spanish-to-English copy read respectively: Moist chocolate cake/ a side of raspberri es/ una combinacin dulce/ for an afternoon deleite. Un pedazo de chocolate cake/acompaado de frambuesas frescas/ a sweet combination/ para un deleite despus de la cena. The design of the ad is largely based on Luna and Peracchios (2001) research on picture and text congruity as a facilitato r for bilingual communications. In order to avoid any effect on the reliability and validity of the experiment, the experimental stimulus contained the same
46 product as the first English language ad. Although the experimental ad message was different than the initial stimulus, the copy and tagline wa s the same for all four experimental stimuli. Spokespersons or models were not included in the advertisements to avoid any interaction effects; therefore the ads only f eatured a picture of the product. A panel of bilingual experts reviewed the language of the advertisements usi ng backward translation. The questionnaire that follows the second ad will be used to measure the dependent variables. Subjects Potential subjects that were part of the experimental sample were taken from two different populations. The first population consisted of st udents of Spanish origin that attended a predominantly English-language university in Florida. The second population consisted of the same types of students that a ttended a predominantly Spanish-la nguage university in Texas. The majority of the sample comprised graduate students of varying age and gender. Students voluntarily agreed to participate in the study and did not receive any compensation for their participation. The Experiment The research was administered online, using the online software program, SurveyMonkey, that designs experimental ques tionnaires and collects the responses based on the researchers established coding procedure. The experimental questionnaire was accessible via a URL web address from February 8, 2009 until March 1, 2009. Each participant was asked to read a consent form that outlines the purpose of the research, what will be asked of each participant, the time required to complete the study, the risks and bene fits associated with the study, confidentiality, voluntary participation, right to withdraw and c ontact information for the researcher. This was a primarily undisguised questionnaire. Subjects shoul d understand what the research is about and
47 what will be expected of them before signing the consent form. Clicking the I agree button will be considered an electronic signature and allow th e participant to maintain his or her anonymity. A software program randomly assigned the particip ant to either the control group or one of the three experimental groups. The subjects were first asked the personal inventory questions in the first section of the questionnaire. They were required to answer a particular question before moving on to the next question. The software progra m will prompt the participant with directions for answering each question. Any matrix-type questions were combined on one page to prevent misunderstanding. Next, the participant was informed that they would be shown the initial stimulus ad. They were asked to pay attention to product and the copy in the ad for as long as they pleased. Participants were advised to pay close attention to the ad because they would not able access the ad in order to answer the next wet o questions When the participant clicked next the second set of questions was administered to the participant in the same manner as the personal inventory questions. Finally, the participan t was exposed to the second expe rimental stimulus ad and the process repeated itself. Control for Confounding Variables In an attempt to help strengthen the results of this study, the researcher must control for confounding variables. The possibility of inte ractional effects of acculturation and ethnic identification and how they are being mani pulated in this study through both explicit questionnaire items and the geogra phic location of the experimental subjects have already been mentioned. The layout of the advertisements used in th e experimental design has also already been explained. The product used in the ad was relativ ely simple in that purchase decisions should be made with a low level of motivation, meaning that subjects did not require a large amount of
48 information in order to answer questions about their attitudes toward the ad and the product. For example, using a product such as underwear can il licit emotional feelings about body image that might interact with a subjects attitude toward the ad. The product was fictitious and should not have elicited any previously held attitudes toward the brand. However asking participants if they are familiar with the product and the ad did control fo r this effect. The simplicity of the ad also helped to avoid the effect other advertising elements, such as the gender or race of models used within the ad, the resemblance of the product to another existing product, and extraneous copy, illustrations or photographs. Operational Definitions Independent Variable: Bilingualism The primary independent variable that was m easured in this experiment was bilingualism or language proficiency. The decision to m easure bilingualism through a combination of indicators is rooted in the linguistic theories di scussed in the literature including the contrast analysis described by Koike and Klee (2003) and Silva-Corvaln (2003) and the Meyers Scotton markedness model discussed in Silva-Corvaln (2003) and in Luna and Peracchio (2005a). Bilingualism was defined as a variable that combin es dimensions such as the level of language fluency of Spanish and English respectively, wh ether Spanish or English is a first or second language, how the individual acquire d or learned Spanish and English, and whether or not the individual regularly engages in c odeswitching. All of these indicat ors were revealed using selfreported data. The experiment involved a questionnaire p acket (see Appendix B) that began with questions relating to bilingualism in Part 3, Language Proficiency. Measurement was based solely on bilingual subjects. Non-bilingual partic ipants were excluded using the contingency question that introduces Part 3. Question number 10 asks participants if they are indeed bilingual
49 speakers of both English and Spanish. If subject s responded no to this question, they were debriefed about the nature of th e experiment and excused from furt her participation. If subjects responded yes to this question, they were given directions for responding to the following questions about bilingualism. The next question of the experimental packet asked participants to indicate which language they learned or acquired first. It is possible that participants first lear ned a language outside of Spanish or English and the question allowed for thes e participants to write in this third language. A follow-up question, question number 12, asked par ticipants to clarify their acquisition of either language by indicating how they learned English and Spanish. Luna and Peracchio (2001) suggest that la nguage proficiency or fluency is a better construct for operationalizing bilingualism because it is possible that a person could have learned Spanish (English) first and yet be more proficient in other languages (p. 285). Therefore question number 13 asks participants to report th eir level of proficienc y or language knowledge on a scale of 1 to 7, where 1 indicates a low le vel of knowledge and 7 indi cates a high level of knowledge. Using this data, par ticipants that chose a higher number on the scale for both languages were considered primar ily bilingual; participants that chose a higher number for one language over another we re considered primarily monolingual. In their study of the effects of accommodation in advertising, Koslow et al. (1994) created a language dominance variable measured by which language participants used at work and at home. Therefore, for the current study, question nu mber 14 asked participants to indicate on a 7point scale which language they used in particular situations, either (1) at home, (2) at work, (3) with friends, and (4) watching TV and (5) readin g magazines. On the 7-poi nt scale, 1 indicated English-only usage and 5 indicates Spanish-only usag e. For this question, a selection closer to
50 the middle indicated that participants engage in some mixed-language communication, probably involving codeswitching. As me ntioned earlier, these questions were combined to create a measure for the independent variable, bilingualism. Covariates The questions in Part 3 served multiple purposes. Not only did they help to create a multivariable index of bilingualism, they also helped to prime the participants to think about language and its meaning as they progress to the next section. Recall that the two covariates, level of acculturation and strength of ethnic identification, are closely related to bilingualism and have an effect on participants attitude toward a dvertisements. Wallendorf and Reillys (1983) assimilation model shows that the basic level of the concept of assimilation includes changes in behavior patterns that are evident in language usage. Because neither variable is determined to be the most important cultural indicator, level of acculturation was not directly measured by any particular question. The language acquisition questions were used as a proxy for acculturation. Question number 14 was used to measure acculturation. Based on the Wallendorf-Reilly assimilation model, the assumption is that if participants and use English in most social situations, they are highly acculturated individuals and the same can be assumed for Spanish and a low or developing level of acculturation. Not until the fourth tier of the Wallendor f-Reilly (1983) assimila tion model does the minority culture begin to exhibit what they term identificational assimilation in which immigrants develop an identity based on the host society (p. 293). Strength of ethnic identification will be explicitly measured during th e experiment in the last two questions of Part 2. In question 8, participants were asked to indicate to which race or ethnicity they are most likely to identify. The extensive list of ethnic origins was taken directly from the U.S. Census Bureaus American Community Survey (2006) because it identifies the majority of races
51 represented within the United States. In question 9, participants were asked to indicate their level of ethnic identification toward their self-identified racial heritage on a scale of 1 to 7, where 1 indicates a weak identificati on and 7 indicates a strong identification. This questions construction is directly linked to Deshpande et al. (1986) and their suggestion that future language studies create a scale meas ure for ethnicity. The responses to this question were useful when comparing ethnic identification to the racial makeup of respondents. Ethnic identification was also measured as if it were an independent variable and this effect will be important to participants attitude toward the ad. Potential participant samples were taken from two different student populations: 1) a prima rily Spanish-language uni versity in Texas and 2) a primarily English-language university in Flor ida. In these populations, Hispanics/Latinos are most likely to feel a stronger ethnic identifica tion at the primarily Spanish-language university than they would at the primarily English-la nguage university. This ma nipulation is strongly linked to Greens (1999) research on social dist ance as an effect on ethnic identification and Stayman and Deshpandes (1989) situation ethnicity dimension of social surroundings. This manipulation can help to further analyze the e ffect of ethnic identification on language and subsequently attitude toward the ad. Dependent Variables The questionnaire following the advertisem ents was used to operationalize the two dependent variables, understandi ng and attitude toward the ad. The first set of questions within the questi onnaire was the recall and recognition test. This was a relatively simple test ba sed on Luna and Peracchios (2001) research on advertisements with slogans that may or may no t be consistent with the product advertised in the ad. The recall test was an open-ended question asking respondents to remember the text of the advertisements slogan or tagline. Because the codeswitch occurred within the text of the tagline, this question
52 can be used to determine whether responde nts acknowledged the code switch by rewriting the slogan verbatim or translated the slogan into its underlying meaning. The recognition test asked respondents to choose what product was featured in the ad from a list of possible product categories. The matrix question number 17/25 also served as a continuously measured indicator of understanding or familiarity with the ad. Re spondents should not be familiar with the ad because it is a fictitious brand. Higher scores on these questions indicated an understanding (or non-familiarity) of the ad. Questions 18 through 21 and questions 26 to 29 were used to determine respondents attitude toward the ad itself with 7-point, 4-item semantic differential scales (Jung and Kellaris, 2006) labeled pleasant/unpleasant, likeable/unlikable interesting/boring, good/bad. Finally, the questionnaire asked respondents to respond to a set of statements that will determine the effect of language usage on their attitudes toward the ad. Th ese items were adopted from Koslow, et al.s (1994) study measuring attitude toward English and Spanish based on the subjects perceived friendliness of the language, the convincing ability of the ad copy, and influence of the language use. Reliability and Validity Reliability refers to the quality of a measurem ent technique that suggests that the measure would produce the same results over multiple observations or administrations (Davis, 1997; Babbie, 2007). A measure is said to be reliable if it yields consistent results. There are three types of methods for testing relia bility: test-retest reliability, inte rnal consistency and inter-coder reliability. The test-retest method issues the same measure to the same group of people to ensure that the results are stable and that responses will not change. This experimental design did require a pretest to ensure the validity of the questions, or that the questionnaire actually
53 measured what it intend to measure. However th e questionnaire was not a ltered in any way after the pretest and these results were considered part of the experiment as a whole. Internal consistency is an indicator of the degree of correlation of a measure. Cronbachs alpha was used to determine this method of reliability for the semantic differential questions and the Likert scale questions used to determine attitude toward the ad. Reliability is acceptable at the .70 level. The four items that make up the attitude toward the ad index do have internal consistency at .93. The internal consistency of the five Likert scale questions making up attitude toward the ad language was also acceptable at 93. This means that these questionnaire items consistently measure attitude toward the ad. Reliability tests were used for other variables used throughout the hypothesis testing. These alpha scor es will be discussed in the next chapter. Validity refers to the quality of a measure to correspond to what the researcher wants to measure, based on the operational definition. Many of the measures used in the questionnaire were taken or adopted from previously used expe riments that have proven the validity of these measures. The experimental questionnaire itself al so contains a manipula tion check that can be used to test validity issues. However, for the purpose of this thesis, there are some internal and external validity issues that should be addressed in the analysis of the findings. Internal validity is especially important to ensure that no extraneous variables can be used to determine the effect of the dependent variable (Babbie, 2007). There are multiple sources that can affect the experiments internal validity in cluding premeasurement and interaction, history, maturation, selection bias, experimental mortality, and instrumentation. Premeasurement and interaction are threats that occur before expos ure to the experimental stimulus in which the respondents become sensitized to the independent variable (Davis, 1997). The purpose of the pre-experimental questions was to sensitize the respondents to the underlying accommodating
54 effect of the ad. Therefore intentionally manipulating premeasurement and interaction can help to control for these effects. History is the effect of events that occur outside the researchers ability to control. Considering that many advertisers are already engaging in codeswitched and bilingual campaigns and that the population for the experimental sample c onsists of the same target market for these campaigns, there may be some historical interaction effect. Maturation refers to the changes in respondents attit ude and behavior throughout the experiment. Maturation was not be a major effect on internal validity because the experiment was administered in one sitting and was relatively short. I have already mentioned that the convenience sample of students will affect the inte rnal validity of the experimental data because randomization was not be used to select potential s ubjects. This effect is referred to as selection bias. Experimental mortality is the effect of subjects who decide not to complete the experiment, which can have an effect on statistical comp arisons (Babbie, 2007). Removing any incomplete questionnaires and ensuring that there are enough questionnaires for each experimental stimulus can help to avoid the effect of mortality. Instrumentation affects occur when changes in the questionnaire occur during the experiment. Ag ain, there were no changes made to the questionnaire. The experimental questionnaire also provides some checks on validity. In Section 2, the second set of questions following exposure to th e experimental stimulus (questions 17 and 25) was a series of statements that represent a ma nipulation check to determine if respondents are paying attention throughout the e xperiment, and if they are answering the questions truthfully. These questions were on an interval level, using a 7-point Likert scale, in which 1 means that the respondent strongly agrees with the statement and 7 indicates that they strongly disagree with the statement. Because a fictitious product was used for the experiment, respondents should answer
55 that they strongly disagree with the manipulati on statements. If they responded otherwise, the assumption is that there is some lack of internal validity, perhaps subject maturation, in which subjects willingness to answer the questions changed throughout the experiment (Davis,1997 ; Babbie, 2007). External validity relates to the ability of the results to be generalized to the larger population. If the study lacks external validity, the findings cannot reflect what might actually occur in a real world situation. Researchers usually ensure external validity through random sampling, using a sample that is representativ e of the real world population, and selecting a heterogeneous sample of subject s. A quasi-experimental design lacks these three elements and therefore lacks validity. Therefore it will be difficult to create strong links between the results of the experiment and the gene ral population (Davis, 1997). Analysis The data was organized and analyzed using the statistical analysis software SPSS 16.0. A variety of research methods were employed incl uding crosstabulations, multiple regression and analysis of variance. These methods provided support for the research hypotheses at a .05 confidence level. Multiple regression is an analysis of dependence that is used to test the linear association between multiple continuously measured independ ent variables and one continuously measured dependent variable. The statistic s generated from a regression analysis can reveal the importance of each independent variable, relative to each other, in predicting the dependent variable. Analysis of variance is another method that tests mean differences between two or more groups or variables. In this analysis, the two c ovariates were considered continuously measured independent variables that are used to deter mine the relationship between the categorically measured independent variable and the one or mo re continuously measured dependent variables.
56 ANCOVA allows the researcher to hold the effect of the covariates constant to determine their effect on the relationship between the independent and depe ndent variables. Each research hypothesis is accompanied by a null hypothesis, which asserts that there is no real relationship between variable or that the relationship is due to chance or error. The main purpose of statistical analysis is to conduct hypothe sis testing. In order to accept either the null or the research hypothesis, the reported statistics ar e described in relation to a confidence interval and confidence level. A confidence interval is a statistical range of numbers within which the population mean should lie. The confidence level is the probability that the population mean actually lies within the confidence interval. A 95% confidence level was used in this thesis. This means that at an alpha of .05, there exists 95% confidence that the population mean will lay within a given range. For the purpose of this study, support for the research hypotheses would occur at the .05 level. The methods explained were used to analyze the four research hypotheses. Crosstabulations were used to determine frequencies and proportions of the demographic makeup of subjects based on each experimental co ndition. Multiple regression was used to test the first two hypotheses and analysis of variance was used to test the final two hypotheses.
57 CHAPTER 4 RESULTS Sample Profile Ninety students volunteered to participate in this experiment, however there were only 66 complete experimental que stionnaires. The majority of participants were female (n= 43, 65%), were in their junior year or above (n = 59, 90%) and the average age was 24 years. The Spanishonly control experimental condition had 22 pa rticipants assigned to it. The English-only experimental condition ha d 11 participants, the Spanish to English (S E) codeswitched condition had 22 participants and the English to Spanish (E S) condition had 11 participants. The majority of respondents has been in the United States since birth (n = 29, 44%), is of Mexican origin (n = 45, 68%) and consider themselves to be White or Caucasian (n = 27, 41%). Note the number of respondents that reported a Hisp anic racial identity. Table 4-1 presents three cross tabulations of the four experimental cond itions along with the participants gender, their Latino origin, and their racial makeup. Hypothesis Testing The main goal of the study is explore the relationship between language, ethnic identification, and acculturation and understandi ng and attitude toward the advertisement. Hypothesis One Results H1 predicts that more bilingual responde nts would have a bette r understanding of the codeswitched ad than primarily monolingual respondents. Multiple regression was used to test this hypothesis with familiarity of the ad as the dependent variable and language prof iciency and the language of the experimental stimulus as the independent variab les. The independent variable bilingualism, or language profic iency was determined by res pondents answers to question number 13, in which they were asked to indicate their level of English and Spanish proficiency
58 on a scale of one to seven. The interval level data was combined into a language proficiency index. The manipulation of the language of the experimental stimulus was also considered an independent variable called level of codeswitching (LOC). There were four versions of the experimental stimulus: a Spanish-only ad, an English-only ad, a Spanish to English ad (S E) and an English to Spanish ad (E S). The first two items in question 25 was used to determine the dependent variable understa nding. The responses to these Likert scale questions were combined into an index called familiarity. Reliability measures suggested high internal consistency for familiarity ( = .88). In order to test the possible interaction effect of the two independent variables in a regr ession model, language proficienc y and LOC were multiplied in order to create a dummy variable, interaction effect (Jaccard and Turrisi, 2003). Table 4-2 represents the results of the regression analysis. Neither of the independent variables had significant t-ratios. The correlation coefficient (R = .21) shows there is a weak correlation between the observed y-valu e and the predicted y-value. The R2 value is .04 which means that 4% of the variance in the dependent variable familiarity is explained by the independent variables. The in teraction effect of the dummy variable produced an R2 value that indicated that the independent variables did not explain any of the variance in the dependent variable. According to the regression equati on for familiarity (4-1), the more bilingual the individual, the more likely they are to understa nd the codeswitched adver tisement. However, the dramatic decrease in the R2 value and the lack of statistical significance reveals that the regression model cannot be projected onto the sample population and that there is no linear association between language prof iciency and familiarity of th e codeswitched ad. Based on the results as reported, H1 is not supported.
59 Hypothesis Two Results The second hypothesis predicts that more bilingual individuals have a more positive attitude toward the code switched ad. The independent variab le language profic iency was coded in a similar fashion to H1. The dependent variable attitude toward the ad was coded into two variables called attitude toward the ad (AAd ) and attitude toward the ad language (AAd Language). The semantic differential questions (Q 26-Q29) were combined into an AAd index. Recall that this index was also internally consistent ( = .93). The Likert scale questions (Q30) were combined into an AAd Language i ndex with a high internal consistency ( = .93). Multiple regression was run on each dependent variable and the results are represented by Tables 4-3 and 4-4 respectively. The results of H2 were similar to H1. Although the attitude toward the ad regression equation reveals a positive correlation between bilingualism and attitude toward the ad and attitude toward the ad language, the correlation is relatively weak (R = .12, .18) and not statistically significant. The decrease in the R2 value also indicates that neither regression equation can be projected onto the sample populati on. Non-significant F-values also reveals that there is no linear relationship between the inde pendent and dependent va riables. Based on the results as reported, H2 was not supported. Hypothesis Three Results The third hypothesis posits that strength of ethnic identification and acculturation are covariates affecting a bilingual individuals unde rstanding of the ad. Analysis of covariance (ANCOVA) was used to test H3. Familiarity was used as the continuous dependent variable for understanding. The categorical independent variable was LOC, wh ich was the language of the four experimental stimuli. The covariate ethnic identification was coded using the responses to question nine.
60 Acculturation was coded using an index of the re sponses to question 14, which asks subjects to indicate their level of language usag e in particular situations. This index was also considered to have a high internal consistency ( = .80). Table 4-5 illustrates the results of the ANCOVA for familiarity. The effect of the language of the ad (LOC) was not statistically sign ificant (p > .05) therefore there is no main effect the ad language on familiarity. The effect of the covariates ethnic identification and acculturation on familiarity was also not statistically significant (p >.05). The covariates do not have any effect on familiarity. The results as reported indicate that H3 cannot be supported. Hypothesis Four Results The fourth hypothesis predicts that strength of ethnic identification and acculturation are covariates that affect a bilingual respondents attitude toward the ad. Analysis of covariance (ANCOVA) was also used to test the final hypothesis. To ensure that the same concepts were represented by the same variables, the independent variable level of codeswitching (LOC) was coded in a similar fashion to H3, as were the two covariates. The dependent variables, attitude toward the ad (AAd) and attitude toward the ad language (AAd Language) were also coded as they were for H2. Table 4-6 illustrates the results of the ANCOVA for attitude toward the ad (AAd). The effect of the language of the ad (LOC) was not statistically significant (p > .05) therefore there is no main effect the ad language on attitude toward the ad. The effect of the covariates ethnic identification and acculturation on attitude toward the ad was also not statistically significant (p >.05). The covariates do not have any effect on familiarity. Table 4-7 shows the results of the ANCOVA for attitude toward the ad language (AAd Language), however the results were also not statistically significant and reveals no effect between the covariates and attitude toward the ad language. The results as reported i ndicate that H4 cannot be supported.
61 Table 4-1. Demographic makeup of subjects by experimental condition Spanish Only S E E S English Only Total Group N Percent N Percent N Percent N Percent N Percent Gender Female 12 27.91 17 39.53 6 13.95 8 18.60 43 65.15% Male 10 43.48 5 21.74 5 21.74 3 13.04 23 34.85% Latino Origin Not Latino 2 40.00 3 60.00 5 7.58% Mexican 17 37.78 16 35.56 6 13.33 6 13.33 45 68.18% Puerto Rican 1 12.50 3 37.50 2 25.00 1 12.50 7 10.61% Cuban 0 0 1 33.33 1 33.33 2 3.03% Other Latino 2 28.57 3 42.86 2 28.57 0 7 10.61% Racial Identity White 8 29.63 10 37.04 4 14.81 5 18.52 27 40.91% Asian 1 100.00 1 1.52% Hispanica 7 28.00 10 40.00 5 20.00 3 12.00 25 37.88% Bi/Multiracial 5 83.33 1 16.67 6 9.09% Other 2 28.57 2 28.57 2 28.57 1 14.29 7 10.61% Note: All percentages are rounded to the nearest one hundredth. aAlthough Hispanic is not an official race noted by the U.S. Census, the number of self-reported data was so overwhelming as to necessitate a Hispanic category.
62 Table 4-2. Multiple Regression of Famil iarity Dependent on Language Proficiency Unstandardized Coefficient (B) ANOVA Model IVs B SE t-ratio p-value SS Df MS F P 1 Constant 4.44 1.74 2.55 .01* Language Proficiency .27 .26 .13 1.03 .31 Level of Codeswitch -.23 .20 -.15 -1.17 .25 8.37 2 4.18 1.46 .24 2 Constant 5.59 4.26 1.31 .19 Language Proficiency .08 .69 .04 .12 .90 Level of Codeswitch -.62 1.34 -.39 -.47 .64 Interaction Effect .06 .22 .25 .30 .77 8.62 3 2.87 .99 .41 Descriptive statistics: R =.21, .21; R2 = .04, .05; R2 Change = .04, .00; F Change = 1.46, .09 ; *p .05 Familiarity Regression Equation, Model 1 (4-1) y = 4.44 +[.27X1 .23X2] Familiarity Regression Equation, Model 2 (4-2) y = 5.59 + [.08X1 .62X2 + .06X3]
63 Table 4-3. Multiple Regression of Attitude Towa rd the Ad Dependent on Language Proficiency Unstandardized Coefficient (B) ANOVA Model IVs B SE t-ratio p-value SS Df MS F P 1 Constant 2.45 1.60 1.54 .13 Language Proficiency .06 .24 .03 .26 .79 Level of Codeswitch -.21 .18 -.15 -1.16 .25 3.77 2 1.89 .78 .46 2 Constant 1.38 3.91 .35 .73 Language Proficiency .24 .63 .13 .38 .71 Level of Codeswitch .15 1.23 .11 .13 .90 Interaction Effect -.06 .20 -.26 .30 .76 3.99 3 1.33 .54 .66 Descriptive statistics: R =.16, .16; R2 = .02, .03; R2 Change = .02, 00; F Change = .78, 09 Attitude Toward the Ad Regression Equation, Model 1 (4-3) y = 2.45 + [.06X1 .21X2] Attitude Toward the Ad Regression Equation, Model 2 (4-4) y = 1.38 + [.24X1 + .15X2 .06X3]
64 Table 4-4. Multiple Regression of AAd Language Dependent on Language Proficiency Unstandardized Coefficient (B) ANOVA Model IVs B SE t-ratio p-value SS Df MS F P 1 Constant 1.42 1.57 .90 .37 Language Proficiency .22 .24 .12 .92 .36 Level of Codeswitch .06 .18 .04 .33 .75 2.05 2 1.02 .44 .65 2 Constant 5.05 3.83 1.32 .19 Language Proficiency -.37 .61 -.83 -.61 .33 Level of Codeswitch -1.17 1.20 -.20 -.98 .55 Interaction Effect .20 .20 .89 1.04 .30 4.59 3 1.53 .65 .59 Descriptive statistics: R =.12, .18; R2 = .01, .03; R2 Change = .01, 02; F Change = .44, 1.08 AAd Language Regression Equation, Model 1 (4-5) y = 1.42 + [.22X1 + .06X2] AAd Language Regression Equation, Model 2 (4-6) y = 5.05 + [.37X1 1.17X2 + .20X3]
65 Table 4-5. Analysis of Variance for Familia rity Dependent on Ethnic ID and Acculturation Descriptive Statistics LOC N Mean ( ) Std. Dev. (s) Spanish only 22 5.52 1.73 Spanish to English 22 6.27 1.12 English to Spanish 11 5.09 1.92 English only/control 11 4.86 2.11 Total 66 5.59 1.71 Between Subjects Effects (df =1, 3) IVs Type III SS MS F-value P-value Ethnic ID .88 .88 .32 .58 Acculturation 4.39 4.39 1.59 .21 Survey Version 17.89 5.96 2.16 .10 Error (df = 60) 165.78 2.76 R2 = .13 Table 4-6. Analysis of Variance for AAd Dependent on Ethnic ID and Acculturation Descriptive Statistics LOC N Mean ( ) Std. Dev. (s) Spanish only 22 2.41 1.60 Spanish to English 22 2.78 1.85 English to Spanish 11 2.11 1.24 English only/control 11 1.80 .87 Total 66 2.38 1.55 Between Subjects Effects (df =1, 3) IVs Type III SS MS F-value P-value Ethnic ID .29 .29 .12 .73 Acculturation .55 .55 .22 .64 Survey Version 8.78 2.93 1.19 .32 Error (df = 60) 147.40 2.46 R2 = .06
66 Table 4-7. Analysis of Variance for AAd Language Dependent on Ethic ID and Acculturation Descriptive Statistics LOC N Mean ( ) Std. Dev. (s) Spanish only 22 2.60 1.31 Spanish to English 22 3.20 1.77 English to Spanish 11 3.02 1.58 English only/control 11 2.64 1.36 Total 66 2.88 1.52 Between Subjects Effects (df =1, 3) IVs Type III SS MS F-value P-value Ethnic ID 1.11 1.21 .46 .50 Acculturation .02 .02 .01 .94 Survey Version 5.46 1.82 .76 .52 Error (df = 60) 144.11 2.40 R2 = .04
67 CHAPTER 5 DISCUSSION AND CONCLUSIONS Discussion The purpose of this study was to explore the relationship between an accommodating advertiser that engages in code switched, or mixed language, adve rtisements and the attitude of Spanish-English bilinguals. An experimental de sign was devised to determine the strength of this relationship based on a number of psycholinguis tic and cultural factors. Overall the results of the study did not the support the assumption that bilingual consumers, who perceive the accommodating effect, will, first, understand the cultu ral connotations of such an advertisement, and secondly, have a more positive attitude toward the ad. In the first hypothesis, this study proposed th at primarily bilingual consumers would have a better understanding of the code switched ad than primarily m onolingual consumers. The basis for this hypothesis was provided by research by Hernandez and Newman (1992) that suggests that bilinguals are more linguistically equippe d to understand the implied cultural message of a codeswitched advertisement. The convenience sample of students was comprised almost completely of primarily bilingual consumers. A multiple regression model was used to test the research hypothesis by determining the linear asso ciation between the leve l of codeswitching of the ad, or the language of the experimental stimulus advertisement, and familiarity, or understanding, of the fictitious ad. The results of the regression analysis showed a weak positive correlation between the level of codeswitching and familiarity, which means that the more codeswitching in the body copy and tagline of the ad, the more likely bilingual consumers would understand that ad. However the results were not statistically significant and could not support the first hypothesis.
68 A number of limitations within the study suggest possible reasons why the first hypothesis was not supported. As mentioned earlier, the sa mple was composed of nearly all bilingual students. There were no primarily monolingual subjects with which to compare understanding of the ad. In addition, similar and s imple copy was used for each of the experimental stimulus ads. It appears that a stronger relationship might have followed if the language of the ad was more complex and more narrowly targeted to a Latino/Hispanic consumer. These limitations also explain some reasons why the second hypothesis was not supported by the results of the study. The second hypothesis predicts that primarily bilingual consumers would have a more positive attitude toward the code switched ad than primarily monolingual individuals. This assumption is based on research by Luna and Peracchios (2001, 2005) research that codeswitched advertisemen ts have important cultu ral connotations for bilingual individuals that they can use to generate a more positive attitude toward advertisers. Multiple regression was used to determine the linear relationship between the level of codeswitching in the experimental stimulus ads and attitude toward the ad. There were two separate measures for attitude toward the ad: (1) a traditional semantic differential attitude measurement based on Jung and Kellaris (2006) attitude research; and (2) a Likert-scale m easurement of attitude toward the language of the ad based on Koslow et. al (1994) research on bilingual consumers attitude toward mixed language advertisements Despite there being separate m easures of attitude within the experiment, neither revealed a very strong relationship between bili ngualism and attitude. The results were also not significant and could not support the second hypothesis. The third and fourth hypotheses suggest th at understanding and attitude are not simply affected by linguistic cues. Wallendorf and Reilly (1983)s assimilation model offers evidence that bilingual individuals perception of codesw itched ads depends on their strength of ethnic
69 identification and their level of acculturation within the American culture, two closely related concepts. Strength of ethnic identification was di rectly measured within the experiment by asking subjects to indicate their ethnic identity and how strongly they feel about this identity. Acculturation was measured through a proxy of language-related questions. For the purpose of testing the third and fourth hypotheses, acculturation was measured through subjects selfreported level of language usage in particular social situations, in which the more English they used, the more acculturated they were presumed to be. ANCOVA analysis was used to test both hypotheses. The covariates had no significant e ffect on the dependent variables understanding and attitude. The final two hypot heses were not supported. The limitations of the study also provided so me explanation why the results did not support the third and fourth hypotheses. Sin ce ethnic identification and acculturation were measured as covariates and not as independent variables, their effect was somewhat limited by the weakness of the independent variable used to test the first and second hypotheses. In addition, subjects were not asked more specific questions about their experiences in the United States, which has some effect on their strength of ethnic identification and level of acculturation. Despite these and other limitations, the results of the research are still important to the advertising research community and offers opportunities for future research. Implications Although the results were not very significan t, this study provides some understanding of how bilingual individuals percei ve codeswitched adve rtisements. There is some correlation between the amount of codeswitching in an adve rtisement and a bilingual individuals attitude toward the ad. This research suggests that adve rtisers that decide to engage in mixed language advertising must first, understand that they ar e evoking a cultural message to their target
70 audience, and second, ensure that bili ngual consumers are actively acknowledging the codeswitch and its meaning. Furthermore simply translating a word or two into another langua ge may not create any meaningful change in attitude or purchase ha bits of Spanish-English bilingual consumers. Finally, language proficie ncy is not the only indicator of bilingual consumers acknowledgement of and attitude towards codeswitc hed ads. The effect of codesw itching for bilingual consumers is somewhat sensitive to consumers level of accultu ration within the culture of the United States and their self-reported level of ethnicity. Limitations Some limitations to this study have already been mentioned, including the lack of primarily monolingual subjects. This limitation is a di rect result of a small sample size. In the original design of the experiment, the researcher selected a sample of students from two different populations, a Spanish-majority and an English-ma jority, with the expectation that there would be a comparable number of monolingual and bilingual respondents for each experimental condition (language of the ad). There was no Spanish-language questionnaire, therefore the expectation was for monolingual subjects to be primarily English speakers. During data collection, the majority of respondents were from the Spanish-majority population and selfidentified as primarily bilingual. A sample size of this nature prevented any comparison between primarily monolingual and primarily bilingual indi viduals. The small sample size also did not allow for any deep exploration of some of the theories suggested by Luna et al.s extensive codeswitching research, including the code-switc hing direction effect (2005a), which suggests bilinguals have a preference for the majority (English) language, a nd including codeswitching persuasion techniques (2001), e.g. rhyming, using cognates, or using English when there is no official Spanish vocabulary equivalent.
71 A second previously mentioned limitation to this study was the simplicity of the print ads. This was a relatively simple study that required the elimination of intervening factors by simplifying the image and copy of the experimental stimuli. Unfortunately, in the real world, consumers consider advertisements on a number of complex factors including their comparison to other ads in other media, the image used in the ad, the spokesperson, etc. These other factors have an impact on acknowledgement of the accomm odation attempt and attit ude toward the ad. For example, if an individual encountered this studys ad in a mixed language magazine full of other codeswitched ads, they may not notice the ad itself, much less register the accommodation attempt. Finally, consumers might be more receptive to codeswitched ads in other media. This study could be duplicated using more interactive media, such as television and the Internet. The close relationship between bilingualism, ethnic identification and acculturation might also explain the lack of significant evidence in support of this studys hypotheses. These measures are similar, measure overlapping c oncepts, and are somewhat dependent upon each other. These subtle differences might explain the weak relationship between the independent variables, the covariates and the dependent variables. For example, there was no direct measure of acculturation because of the complexity of the concept. It is possible that acculturation or assimilation could be a more discerning variable than language profic iency but without any direct measure, this study cannot confirm this possibility. Although all of the measures used in this study were considered highly reliable, many of the series of questions were brief. Perhaps a longe r, more in-depth series of questions might have revealed the subtle differences between bilingua lism, ethnic identification, and acculturation and their effect on attitude toward the ad. Furthe rmore a longer study that allowed respondents to study ads in more realistic situations might have revealed more meaningful results.
72 Future Research A relatively simple product category was used to gather data in this study. Chocolate is a low-involvement product and it appeared th at many respondents simply duplicated their responses between the initial stimulus ad and the e xperimental stimulus ads, indicating a lack of interest in the accommodation attempt. However the weak results indicate that perhaps different product categories might produce more meaningf ul results for accommodating advertisers. Understanding and duplicating colloquial speech in social situations might provide more positive attitudes for certain product categories. Fo r example conducting the study using a product category that is more salient to the Latino population or even to the convenience sample (e.g. Latino students and cell phones) may bring out more positive attitudes toward the accommodation attempt. Not only does this study lay the groundwork for future research that explores the effect of codeswitching in different produc t categories, the study can also be duplicated using more sophisticated print media. Luna and Peracchios (2001, 2005) research suggest that using print ads is the best-suited method for measuring code switching recognition. In their research (2005a), they primed subjects attitudes toward Spanish or English by having them read an article about the importance of either language. Researchers might consider priming subjects attitudes based on language by duplicating Luna and Peracchios pe rsuasion study or perhaps inserting the print ads within a mixed language periodical. Encourag ing subjects to be more involved about their attitudes provides an opportunity for more meaningful results. Ethnic and social identities have a grand impact on how bilingual consumers perceive accommodation attempts. Recall that many of this studys participants identified their racial identity as Hispanic although it was not listed as a possible racial choice. This identification with Hispanic as a racialized ethnic label is a testament to Golash-Boza and Daritys (2007)
73 research on Latino racial choices. In their st udy, the researchers analyze two demographic studies on Latinos and discover that different ge nerations of Latinos at varying levels of acculturation self-identify differently on the scale of possible racial choices depending on a number of cultural reasons including how others pe rceive them racially, their skin color and subsequent ability to identify as white, and whether or not they have experienced any type of racial or ethnic discrimination. Their conclusions suggest that Hispanic/L atino racial and ethnic self-identities are constantly changing. How one self-identifies in one situation, or even in one survey or experiment, may not be how they identify in another. Future research will be required to determine a more static Hispanic/Latino identify that advertisers can target, even if temporarily, for successful codeswitching attempts. Research has discovered that targeting bilingual consumers requires more than simply advertising in Spanish; however codeswitching research is still a relatively new topic in communication and advertising studies. Advertisers require further research in order to better understand their bilingual target markets. Langua ge proficiency is only one measure of how bilingual individuals perceive codeswitched accommodation a ttempts. Understanding other important factors, such strength and directio n of cultural indicators, purchasing habits, involvement, history, etc. are required in future re search in order to better target Hispanic/Latino bilingual consumers.
74 APPENDIX A EXPERIMENTAL QUESTIONNAIRES Initial Stimulus Advertisement
75 Experimental Condition Advertisements English-only/Control
77 Spanish to English Codeswitch
78 English to Spanish Codeswitch
79 APPENDIX B QUESTIONNAIRES
97 Author Note: The copy of the questionnaire s hown is of the Spanish-only experimental condition. All of the questionnaires were the same, however the ads were different for each experimental condition. See Appendix A for a lis ting of all of the experimental condition ads.
98 APPENDIX C IRB PROTOCOL AND CONSENT FORM Informed Consent Protocol Title Exploring advertising campaigns and bilingual consumer: Studying the effects of codeswitching and the accommodation theory Please read this consent document carefully before you decide to participate in this study. Purpose of the research study: The purpose of this study is to determine consum er attitudes towards bilingualism and its use in the media. What you will be asked to do in the study: You will be asked to participate in an experiment involving a series of different advertisements. First, you will be asked to answer a series of personal inventory questions about some basic demographics, your ethnic identity and your languag e proficiency in both English and Spanish. In the next section of the experiment, you will be given an English language advertisement for a fictitious brand, and you will be asked to answer a series of questions about your attitude toward that brand. In the final section of the expe riment, you will be given randomly one of four different advertisements for the same fictitious br and and you will be asked to answer a series of questions similar to that of the second section. Age required: You must be at least 18 years of age to participate in this study. Minor participants that are below 18 years of age will be asked not to participate in this study. Time required: 10 minutes Risks and Benefits: Minimal risk Compensation: There will be no compensation for participating in this study.
99 Confidentiality: This experiment will require you to answer personal questions. Your responses will be confidential and will not be disclo sed to any party outside the scope of the research. You will not be asked to give any identifying informa tion. When you complete the experiment, your information will be assigned a code number. The data will be analyzed using this code number only. Voluntary participation: Your participation in this study is completely voluntary. There will be no penalty if you decide not to participate. Right to withdraw from the study: You have the right to withdraw from this study at any time without consequence. Whom to contact if you have questions about the study: Keitra E. Harris, Graduate Student, Univers ity of Florida College of Journalism and Communications, Department of Advertising, Weimer Hall PO Box 118400, Gainesville, FL 32611; phone (352) 392-5059; e-mail: email@example.com Whom to contact about your rights as a research participant in the study: IRB02 Office, Box 112250, University of Florida, Gainesville, FL 32611-2250; phone (352) 392-0433. Agreement: I agree and I wish to participate. By clicking here, I agree that I have read the procedure described above. I am at least 18 years of age, and I voluntarily agree to participate in the procedure and I have received a copy of this description. I agree and do not wish to participate. By clicking here, I agree that I have read the procedure described above. I voluntarily agree not to participate in the procedure and I have received a copy of this description. Please print a copy of this page for your records.
100 LIST OF REFERENCES American Psychological Association (2005), Publication Manual of the APA (5th ed.), Washington, DC: APA. Ben-Yehuda, Ayala (2007, February 3), Spanglish Strategy, Billboard.com ., Retrieved December 4, 2007 from Lexis-Nexis database. Campbell, Donald T. and Julian C. Stanley (1970), Experimental and quasi-experimental designs for research, Chicago: Rand McNally. Caroll, John B. (ed.) (1998), Language, thought and reality: selected writings of Benjamin Lee Whorf. (24th ed.), Cambridge: Massachusetts Institute of Technology Press. Davis, Joel J. (1997), Advertising research: theory and practice New Jersey: Prentice Hall. de Jongh, Elena M. (1990), Interpreting in Miamis federal courts : code-switching and Spanglish, Hispania [Publicaciones peridicas], 73 (March), 274-278. Deshpande, Rohit, Wayne D. H oyer, and Naveen Donthu (1986), The Intensity of Ethnic Affiliation: A Study of the Sociology of Hispanic Consumption, Journal of Consumer Research 13 (September), 214-220. Friestad, Marian and Peter Wright (1994), The Persuasion Know ledge Model: How People Cope With Persuasion Attempts, Journal of Consumer Research, 21 (June), 1-31. Golash-Boza, Tanya and William Darity, Jr. (2007), Latino Racial Choices: the Effects of Skin Colour and Discrimination on Latinos and Latinas Racial Self-identifications, Ethnic and Racial Studies, 31 (September), 899-934. Grieco, Elizabeth M. and Rachel C. Cassidy (2000 ), Overview of Race and Hispanic Origin, (Census 2000 Brief, March), Washington D.C.: U.S. Census Bureau, Retrieved October 11, 2007, from http://www.census.gov/prod/2001pubs/c2kbr01-1.pdf Green, Corliss L. (1999), Ethnic Affiliations of Advertising: Interaction Effects of Strength of Ethnic Identification, Media Placement, and Degree of Racial Composition, Journal of Advertising, 28 (1), 49-64. Grier, Sonya A. and Anne M. Brumbaugh (1999), Noticing Cultural Differences: Ad Meanings Created by Target and Non-target Markets, Journal of Advertising, 28 (1), 79-93. Grosjean, Franois (1982), Life with two languages Cambridge, Massachusetts: Harvard University. Hernandez, Sigfredo A. and Larry M. Newman (1992), Choice of English vs. Spanish Language in Advertising to Hispanics, Journal of Current Issues and Research in Advertising 14 (2), 25-45.
101 Holland, Jonna and James W. Gentry (1999), Ethnic Consumer Reaction to Targeted Marketing: A Theory of Interacultural Accommodation, Journal of Advertising, 28 (1), 65-77. Hoyer, Wayne D. and Deborah J. MacInnis (2007), Consumer Behavior (4th ed.). Boston: Houghton Mifflin. Jaccard, James and Robert Turrisi (2003), Interaction Effects in multiple regression, (2nd ed), In Sage University Papers Series on Quantitative Applications in the Social Sciences 07072, ed. Michael S. Lewis-Beck, Thousand Oaks, CA: Sage University, 32-36. Jung, Jae Min and James Kellaris (2006), Responsiveness to Authority Appeals Among Young French and American consumers, Journal of Business Research, 59 (6), 735-744. Kassarjian, Harold H. (1969), The Negr o and American Advertising, 1946-1965, Journal of Marketing Research, 6 (1), 29-39. Koike, Dale and Carol Klee (2003). Chapter 2: El campo de investigacin de la ASL, Lingistica aplicada: Adquisicion del espaol como segunda lengua Somerset, NJ: Wiley, John & Sons, Inc., 21-44. Koslow, Scott, Prem N. Shamdasani, and E llen E. Touchstone (1994), Exploring Language Effects in Ethnic Advertising: A Sociolinguistic Perspective, Journal of Consumer Research, 20 575-585. Leedy, P. D., and J.E. Ormrod (2005), Practical research: Planning and design, (8th Ed.) Upper Saddle River, NJ: Pearson Prentice Hall. Lemann, Mariana (2006, December 19). Ford speaks Spanglish for Edge SUV, Adweek.com Retrieved December 4, 2007 fro m Lexis-Nexis database. Luna, David, Dawn Lerman, and Laura A. Peracchio (2005b), Structural Constraints in Mixed Language Ads: A Pscholinguistic Analysis of the Persuasiveness of Codeswitching [Abstract], Advances in Consumer Research 32 647-648. Dawn Lerman, and Laura A. Peracchio (2006c), Structural Constraints in Codeswitched Advertising, Journal of Consumer Research 32(3), 416-423. and Laura A. Peracchio (2001), Moderators of Language Effects in Advertising to Bilinguals: A Psycholinguistic Approach, Journal of Consumer Research, 28 (September), 284-295. and Laura A. Peracchio (2004), Language Processing, Affect and Cognition: Word and Sentence Structure Effects Across La nguages, [Special Session Summary]. Advances in Consumer Research, 31 148-151. and Laura A. Peracchio (2005a,), Advertising to Bilingual consumers: the Impact of Code-switching on Persuasion, Journal of Consumer Research 31(March), 760-765.
102 and Laura A. Peracchio (2005d), Socio linguistic Effects on Code-switched Ads Targeting Bilingual Consumers, Journal of Advertising 34(2), 43-56. McFarlane, Cynthia and Laura Semple (2007, May 14), Get Connected with Latinos in Nuevo American: Interacculturation: Blended Cultures Fuel Mainstream, Advertising Age Retrieved December 4, 2007 fro m Lexis-Nexis database. Mendieta-Lombardo, Eva and Zaida Cintron (1995). Marked and Unmarked Choices of Codeswitching in Bilingual Poetry, Hispania [Publicaciones peridicas], 78 (September), 565-573. Retrieved November 18, 2007 from http://www.cervantesvirtual.com/servlet/SirveObras/02473844544571839644424/p000000 4.htm 19 Myers-Scotton, Carol (2002), Contact linguistics: Bilingual encounters and Grammatical Outcomes Oxford: Oxford University Press. New American Dimensions (2008), Acculturation and Beyond (Presented by David Morse). Los Angeles, Califonina: Author. Noriega, Jaime (2006), Advertising to Bilingua ls: Does the Language of Advertising Influence the Nature of Thoughts? (Doctoral dissertation, University of Houston, 2006), Dissertation Abstracts International, 67(7). psycholinguistics (n.d.), The American Heritage Dictionary of the English Language, Fourth Edition Retrieved November 18, 2007, from Dictionary.com website: http://dictionary.reference.com/browse/psycholinguistics Richards, Jef I. and Catharine M. Curran (2002), Oracles on Advertising: Searching for a Definition, Journal of Advertising, 31 (2), 63-77. Schwartz, Mike (Writer) and Chris Koch (Direct or) (2006), Her Story II [Television series episode], In Bill Lawrence (producer), Scrubs Burbank, CA: ABC Studios. Silva-Corvaln, Carmen (2003), El espaol en Los ngeles: aspectos morfosintcticos, nsula, 678-680 (July-August), 19-38. sociolinguistics (n.d.), The American Heritage Dictionary of the English Language, Fourth Edition Retrieved November 18, 2007, from Dictionary.com website: http://dictionary.reference.com/browse/sociolinguistics University of Alabama at Birmingham IRB ( 2007, November 9), Sample informed consent document, Retrieved November 16, 2007, from http://www.uab.edu/irb/forms/sampleconsent-form.doc U.S. Census Bureau (2006), American Community Survey, Washington D.C.: Author, Retrieved October 12, 2007, from http://factfinder.census.gov/servlet/DTTable?_bm=y&geo_id=01000US&-ds_name=ACS _2006_EST_G00_&-redoLog=false&mt_name=ACS_2006_EST_G2000_B03001
103 Vakratsas, Demetrios and Tim Ambler (1999), H ow Advertising Works: What do We Really Know? Journal of Marketing, 63 26-43. Vargas, Jose Antonio (2005, May 31), Spanish ads on English TV? An experiment, The Washington Post p. C01. Wallendorf, Melanie and Michael D. Reilly ( 1983), Ethnic Migration, Assimilation, and Consumption, Journal of Consumer Research, 10 292-302. Wentz, Laurel (2007, November 29). Telemundo Launches Bilingual Late-night Talk Show, Advertising Age, Retrieved November 30, 2007, from http://adage.com/hispani c/article?article_id=122324 Wilkes, Robert E. and Humberto Valencia( 1989), Hispanics and Blacks in Television Commercials, Journal of Advertising 18 (1), 19-25.
BIOGRAPHICAL SKETCH Keitra Harris is from Atlanta, Georgia. She earned her Bachelor of Science in advertising from the University of Miami in 2007. The inspiration for her thesis came from her experience with learning Spanish as a second language. She received her Master of Advertising from the University of Florida in the summer of 2009. Keitra plans to work professionally in the advertising field. She hopes that her research w ill be beneficial to both her professional career and the bilingual advertising community.