Salience in Visual Context: Effects on Appreciation of Advertisements

Salience in Visual Context: Effects on Appreciation of Advertisements Luuk Lagerwerf Communication Science Vrije Universiteit 3972 HV Amsterdam l.lagerwerf@fsw.vu.nl Abstract The notion of salience in the lexicon is used to predict effects of deliberate ambiguity in print ads. An advertisement is more appreciated if the salient meaning of the headline is not the appropriate meaning. When non-salient, secondary meanings are more relevant in the combination of words or the visual context, appreciation for the ad will enhance. In an experiment, advertisements were manipulated with respect to ambiguity (headline) and target group orientation (image). Results show that the ambiguous headlines are appreciated more than the straightforward ones, independent of the respondents trait openness to experience. Although the manipulation check shows that the manipulation in the materials was successful, direct appreciation effects of the visual context in target groups were not found. Instead, inappropriateness of the visual context seemed to be effective. 1 Introduction Thamar Siebring Policy, Communication and Organization Vrije Universiteit siebringt@hotmail.com There are different definitions of the term salience in linguistics. A common denominator is that salience is used as a background for explanations of other phenomena, such as anaphora resolution or disambiguation of discourses. In the research presented in this chapter we assume that the Graded Salience hypothesis provides explanations for disambiguation of headlines in (visual) contexts (Giora 2002). We further adopt the Optimal Innovation hypothesis to account for the differences in appreciation of the advertisements in this research (Giora et al. 2004). Graded salience refers to activation of lexical meanings ordered by salience (in a mental lexicon), rather than an ordering by contextual relevance, in the process of interpreting language in context. The Graded Salience hypothesis is best understood in contrast with Relevance Theory. According to the latter theory, relevance in context and situation is supposed to control lexical choice (Sperber and Wilson 1995). Graded salience states that interpretation starts with lexical alternatives and ends with contextual relevance. In addition to graded salience, the Optimal Innovation hypothesis states that metaphors are appreciated most if their interpretations consist of available, more and less salient, lexical alternatives, and contain only some new meaning inferences (Giora et al. 2004). The purpose of this chapter is to apply the salience and innovation approach to possible effects of advertisements on readers. We were interested in the interplay of the visual design of print ads with deliberate ambiguity of the headlines (Lagerwerf 2002), and in appreciation differences between respondents characteristics. We had the following research questions: 1. Are ambiguous headlines in ads appreciated more than straightforward ones? 2. Are ambiguous headlines appreciated more in highbrow visual contexts than in common visual contexts? 3. Are ambiguous headlines appreciated more in appropriate contexts than in inappropriate ones? In the next section we explain and underpin these questions, and formulate our hypotheses. We first illustrate our way of using salience and innovation in the experiment. Then we specify the role of visual design in creating types of contexts for the headlines. Finally, the role of appropriateness as a factor in both interpretation and appreciation is elucidated. The experiment is reported in the subsequent sections.

2 Optimal Innovation in Advertising In order to illustrate salience and innovation in our experiment we refer to the ad variants in Appendix A (we manipulated 5 different ads this way). In two of the four conditions (the ambiguous heading conditions) the heading reads Speechless. In Dutch as well as in English, the most salient meaning of this word is something like silent in admiration. A less salient interpretation would be without speech. In the context of a camera cell phone, the less salient lexical alternative is more relevant within its context. Still, the salient alternative is contributing to the interpretation as well. According to the Optimal Innovation hypothesis (extended to deliberate ambiguity instead of solely metaphor), this ambiguous interpretation will be appreciated more than an interpretation in which only one, salient, meaning is possible (Giora et al., 2004; the hypothesis was two-sided: Metaphors conveying only non-salient and new meanings will also be less appreciated). In the other two conditions Love and kisses has only one salient meaning, supported by the visual context. The interpretation is straightforward. Optimal innovation predicts lower appreciation. The Optimal Innovation hypothesis corresponds to findings in the fields of literary and advertising research. Metaphors have been investigated in literary research for their effects on arousal and pleasure (Steen 1994; Giora 2002; Goatly 1997; Giora et al. 2004). Recognition of incongruity in metaphorical stimuli have a measurable psychophysiological effect (Hoorn 1997). For advertisements, it has been established that ambiguous headlines are appreciated more, unless they are not understood, irrelevant, or too obvious (Van Mulken, Enschot-Van Dijk, and Hoeken 2005; Lagerwerf 2002). Appreciation is an important feature for advertising, since it enhances the Attitude towards the ad (MacKenzie, Lutz, and Belch 1986; Mitchell and Olson 1981; Mitchell 1986), which in turn may affect purchase behaviour. A first hypothesis can thus be formulated: H1: Ambiguous headlines are appreciated more in print ads than are straightforward headlines. The example Speechless already indicated that salience is not the same as literal interpretation. In fact, salience is an attempt to replace the notion of literal meaning as the most immediate lexical meaning of terms in the lexicon, or compositional meaning as the most immediate meaning of a phrase. Conventionalized metaphors and idioms dominate our language use rather than decorate it with secondary meanings (Lakoff and Johnson 1980). In the mental lexicon, salient meanings are those meanings that pop up instantaneously, while less salient meanings do not need to come to mind immediately. With Speechless, the literal interpretation is less salient, and comes about only after suppression of more salient alternatives (Giora 2002). Different factors may cause lexical meanings to be more or less salient: (1) Lexical items that are more frequent are more likely to be salient (Landauer and Dumais 1997); (2) From individual experiences, some meanings may be more familiar, irrespective of relative frequencies (Gernsbacher 1984); (3) Lexical choices in conversation and writing are often determined by conventions, favouring one expression over an equivalent other (Gibbs 1983); (4) Prototypical members of categories are more likely to be referred to with the category name than are non-prototypical members of the same category (Armstrong, Gleitman, and Gleitman 1983; see also Lagerwerf 1998). How these factors interact in establishing saliency, is described in Giora (2002). The hypothesis that ambiguous headlines are appreciated more than straightforward headlines is in most cases tested on samples from students populations (Lagerwerf 2002; Van Mulken, Enschot-Van Dijk, and Hoeken 2005; Phillips 2000). However, respondents may differ in their appreciation when they have different educational backgrounds (Lagerwerf and Meijers 2005). In order to find out whether there are differences between respondents for liking rhetorical effects such as ambiguity, we tried to establish a respondents trait called Openness to experience (McCrae and Costa Jr. 1999). Openness to experience is one of the so called big five of personality traits (Piedmont 1998). Since resolving ambiguity enhances appreciation through optimal innovation, we expected that H1 would especially be confirmed for respondents high on openness to experience: One has to open his or her mind in order to appreciate ambiguity. Straightforward headlines would then be appreciated more by people low on openness to experience.

H2: People with high Openness to experience appreciate ambiguous headlines in print ads more than do people with low Openness to experience. Besides this personality trait, we controlled for gender, age and education. Although graded salience is initially contextindependent in offering meaning alternatives, visual context is important for both interpretation and appreciation of the ads. We investigated two aspects of the visual context: Effects of visual design regarding target group differences, and the interplay of headline and visual context. Advertising research of visual rhetoric increased in the past ten years (Mothersbaugh, Huhmann, and Franke 2002; Phillips and McQuarrie 2002; Phillips 2000; McQuarrie and Mick 2003, 1999, 1996; Scott 1994; Forceville 1996). In fact, the analysis of visual rhetoric strongly resembles the way in which verbal metaphors are analysed (McQuarrie and Mick 1996). Although cultural differences in understanding visual metaphors have been tested (for different national backgrounds, McQuarrie and Mick 1999), there is little attention for effects of visual design in different target groups. Since visual rhetoric research has been oriented on students populations, there is a risk of a sample bias. Does everyone like the same kind of visuals? In order to find a tentative answer, we conducted a corpus analysis as preliminary research. We selected 808 print advertisements from 40 different magazines. These magazines were divided in magazines with target groups in different socio-economic classes (high: AB; low: CD). There were significant differences in the following features: few or no people (AB) versus many people (CD); sad or neutral looking people (AB) versus happy looking people (CD); shaded colors (AB) versus bright colors (CD) (differences were significant using Chi-square tests). We decided that it would be possible to manipulate these features to form highbrow visualizations versus common visualizations (see the examples in the appendix). In the same line of thought, we expected deliberate ambiguity in the headlines to be more highbrow, and straightforward interpretations to be more common (but this was not tested in preliminary research). Combining text variants with visual variants, we developed the materials in four conditions, exemplified in Appendix A. We expected an interaction effect of text and visual context on the basis of this manipulation: H3: Ambiguous headlines are appreciated more than straightforward headlines in highbrow visual contexts. H4: Straightforward headlines are appreciated more than ambiguous headlines in common visual contexts. Although graded salience implies that lexical meaning comes first, the visual context is important for the interpretation of the headlines. In Love and kisses the salient meaning is the goodbye greeting, in Dutch expressed with little kiss. Without context (but with the quotation marks) it would be unclear in which circumstances the greeting took place, or whether something else had to be interpreted. Within the visual context it becomes clear that the camera cell phone is expressing the goodbye visually. The salient meaning is confirmed. In the common visual context, some feelings of happiness seem to be expressed by the man in the picture, but this is not inconsistent. In the Speechless variant, the salient meaning is not consistent with the visual context. This meaning is suppressed and the less salient, more literal meaning comes about. The headline is used inappropriately with respect to its context, and consequently, less salient meanings are probed to get a more appropriate interpretation. Recognizing the ambiguity is recognizing inappropriateness (see Attardo 2000). An interpretation is successful when both salient and less salient meanings intertwine (optimal innovation). This process of interpretation is not restricted to lexical ambiguity. Attardo (2000) defines relevant inappropriateness to analyse irony. If an interpretation that comes forth from salient meanings is not appropriate in its context, we may look for an interpretation with another intention than purely informational. This way, we are able to recognize utterances as ambiguous, humorous, ironic, sarcastic, or metaphorical. The theory of relevant inappropriateness may shed another light on the hypotheses 3 and 4. The inappropriateness may relate to other properties than lexical meaning, such as the weirdness of the boy in the common visual context with the headline Speechless. Is he visualizing speechlessness? Also, the straightforward headlines are quite expressive, and this expressiveness does not match very well with the shaded colors of the high brow

visual contexts (the boy is more meaningful if we imagine that he is cheering for the kiss he receives). These kinds of inappropriateness may initiate a process of looking for speaker intentions that fit the context. Relevant inappropriateness predicts that if this fit is not existent, the interpreter will be frustrated, resulting in low appreciation. In other words, we expect matching pairs (or: headlines that have appropriate visual contexts) to be more appreciated than the other pairs. H5: Headlines with appropriate visual contexts are appreciated more than are headlines without. 3 Method 3.1 Respondents A purposive sample was taken from the Dutch population of people more than twenty years of age (M age =39.7 years). The sample consisted of 86 men (48%) and 94 women (52%). Respondents highest education was university (8%), higher vocational education (39%), intermediate vocational education (26%), secondary education (27%). 3.2 Design and materials In a 2x2 design, advertisements in Dutch were created in 4 conditions: highbrow image with ambiguous headline, highbrow image with straightforward headline, common image with ambiguous headline, and common image with straightforward headline. In Appendix A, an example of these four conditions is given for one product. The difference between the kinds of image was created by changing the colors (bright/shaded), the number of people on the images, and the gravity of their looks. Within each condition, five ads were created, for different products: photo camera, food product, mobile phone, winter sports, summer holiday. For each product, a variant was made to fit in one of the cells in the design. In total, twenty ad variants were used for the materials. These variants were divided into five portfolios. No variants of the same product occurred in one portfolio. 3.3 Measures Besides the demographic variables gender, education, and age, a psychographic variable was measured in order to verify between-subjects effects. This was the trait Openness to experience. This trait is one of the Big Five, operationalized in an extensive question list instrument (Piedmont 1998). Our own earlier research revealed no or modest effects of Need for Cognition (Cacioppo, Petty, and Kao 1984) and education. Need for Cognition disturbs experiencing humor (Zhang 1996). Education, but not Need for Cognition, moderates the liking of visual metaphors (Lagerwerf and Meijers 2005). We wanted to establish a personality trait that would explain liking of rhetorical effects. Openness to experience consists of six dimensions (fantasy, esthetics, feelings, actions, ideas, and values). We used two items for each dimension. Cronbach s alpha =.67; a factor analysis revealed that fantasy and actions are the most prominent dimensions, while esthetics and ideas did not come about. We decided to maintain the instrument. Two items per dimension was not enough to measure separate dimensions, but reliability was sufficient for the scale as a whole. The manipulation was checked by asking for judgements on each ad: The people look happy and/or have fun; the ad is colored brightly; text and image fit together well; the headline makes me think. The dependent variable was the Attitude towards the ad (Bhat, Leigh, and Wardlow 1998). This variable consists of a cognitive (the ad is good, amusing, informative, credible), affective (is irritating, makes me cheerful, interests me, bores me), and conative (I want to use this product in future, I m inclined to use this product) dimension. Cronbach s alphas across all ads were.85,.82, and.76 respectively. 3.4 Procedure All respondents received one question list, and one of the five portfolios. Having completed the questions concerning demographic and psychographic information, the respondent looked at the first ad, and completed the Attitude towards the Ad items. Then, the next ads and attitude questions were presented. After reviewing the whole portfolio, respondents were asked to look at the ads again and complete, per ad, the manipulation check questions.

3.5 Data analysis Manipulation check (except for number of people), Openness to experience and Attitude towards the ad were measured on five point Likert scales or five point semantic differential scales. Manipulation check and hypotheses were tested by performing a variance analysis in a general linear model with the four conditions as repeated measures. A post hoc regression analysis was conducted as well. 4 Results A manipulation check was performed to check whether the manipulations were recognized by the respondents. The judgement in the ad one or more persons are depicted was answered with yes more often for the common visual context ads (χ 2 =122,883, df=3, p<.001). In table 1, the means of the other manipulation check questions are presented. Visual context Common Highbrow Headline Headline straight ambiguous straight ambiguous Depicted persons have fun* 4.16 (0.68) 4.27 (0.58) 2.79 (0.91) 2.82 (0.93) Bright colors* 3.71 (0.91) 3.84 (0.89) 2.96 (0.92) 2.94 (0.85) Text and image fit well* 3.59 (0.80) 3.41 (0.96) 3.33 (1.01) 3.66 (0.83) Headline makes me think* 2.63 (0.96) 2.89 (0.96) 2.68 (0.99) 3.19 (1.00) *: p <.001 (F-tests, main effects indicated in bold face) Table 1. Manipulation check means and standard deviations for each condition (N=180) The manipulated conditions differ from each other significantly, in the expected direction. For depicted persons and bright colors there was a main effect of common visual context (F(1,175)=269.094, and F(1,175)=160.694, respectively). For fit of text and image, there was an interaction effect (F(1,175)=14.697), where common visual contexts with straightforward headlines and highbrow visual contexts with ambiguous headlines score higher than the other two conditions. The ambiguous headlines were making the respondent think more than the straightforward ones (main effect of headline, F(1,175)=35.694). Also, there is an interaction effect of headline and visual context. Apparently, respondents recognize the appropriateness of the visual context. In table 2, the means for the Attitude towards the ad are given. Visual context Common Highbrow Headline Headline straight ambiguous straight ambiguous Cognitive component* 3.00 (0.68) 3.33 (0.67) 3.01 (0.76) 3.19 (0.73) Affective component* 3.02 (0.68) 3.25 (0.65) 3.01 (0.72) 3.13 (0.66) Conative component* 2.35 (0.84) 2.66 (0.84) 2.53 (0.86) 2.62 (0.88) Aad (comprehensive)* 2.90 (0.63) 3.19 (0.60) 2.93 (0.69) 3.07 (0.64) *: p <.001 (F-tests, main effects indicated in bold face) Table 2. Means and standard deviations of Attitude towards the ad for each condition (N=180) There are only main effects of the headline: on all components, ambiguous headlines score higher than do straightforward headlines. Hypothesis 1 is confirmed. However, there is no interaction of appreciation values for headline and context. Hypotheses 3 and 4 are therefore rejected. Openness to experience did not moderate the Attitude towards the ad. Hypothesis 2 is rejected. Education does moderate the Attitude towards the

ad, but in the opposite direction: Higher educated respondents appreciate common visual contexts more than do lower educated respondents, with only a small difference in appreciation for the high brow visual contexts (F(1,178)=13.650, p<.001). On the basis of the results in table 2, hypothesis 5 should be rejected as well. However, the manipulation check showed that fit of the headline in the visual context was recognized by the respondents. Therefore, a post hoc analysis is performed: a regression analysis of attitude towards the ad on the manipulation check questions, dichotomous dummy variables representing the conditions, and their multiplications as interaction variables. The results of this analysis are shown in table 3. Attitude towards the ad on: Beta R²=.386 Persons have fun.030 p=.386 Bright colors.083 p<.05 Headline and image fit well.206 p<.001 Headline makes me think.456 p<.001 Visual context condition -.002 p=.966 Headline condition.079 p<.05 Bright colors * visual context condition.006 p=.849 Bright colors * headline condition.063 p<.05 Bright colors * visual context cond. * headline cond..024 p=.478 Headline and image fit well * visual context condition -.019 p=.590 Headline and image fit well * headline condition -.102 p<.01 Headline and image fit well * visual context cond. * headline cond. -.026 p=.474 Headline makes me think * visual context condition.018 p=.600 Headline makes me think * headline condition -.004 p=.900 Headline makes me think * visual context cond. * headline cond. -.030 p=.395 Table 3: Regression of attitude towards the ad on manipulation check and condition In table 3, it appears that headline fit (β=.206, p<.001) and the headline perception (β=.456, p<.001) explain most of the variance for Attitude towards the ad (in total, the model itself explains 38,6 % of the variance). Bright colors in the visual context positively affect the Attitude towards the ad (β=.083, p<.05). The main effect of the ambiguity of the headline is reflected in the positive affect on appreciation of the headline condition (straightforward or ambiguous; β=.079, p<.05). There is an interaction between the headline condition and headline fit (β=-.102, p<.01), as well as between headline condition and bright colors (β=.063, p<.05). These interactions need further interpretation. This is done in the Figures 1 and 2, by calculating the appreciation values in the regression equation for the maximum and minimum values of the interaction factors. Attitude towards the ad 3.4 3.2 3 2.8 2.6 2.4 Straightforward Ambiguous -1 1 Fit of text and image Figure 1. Interaction of Straightforward and Ambiguous headlines and Fit of text and image for Attitude towards the ad. From Figure 1, we may infer that straightforward headlines are less appreciated when they do not fit their visual context, whereas contextual fit does not affect the appreciation for ambiguous headlines to the same extent.

Attitude towards the ad 3.2 3.1 3.0 2.9 2.8 Straightforward -1 1 Bright colours in image Ambiguous Figure 2: Interaction of Straightforward and Ambiguous headlines and perception of Bright colors in image for Attitude towards the ad. From Figure 2, we may infer that ambiguous slogans are appreciated more in a brightly colored visual context than without bright colors, whereas the appreciation for straightforward headlines is not affected by bright colors. Hypothesis 5 is rejected. From Figure 1 we infer that although the Attitude towards the ad is enhanced for straightforward headlines with contextual fit, it is not affected by ambiguous headlines. In Figure 2 the Attitude towards the ad is positively affected by ambiguous headlines in inappropriate (brightly colored) visual contexts. 5 Discussion This research has re-established the effectiveness of ambiguity in headlines (hypothesis 1). However, the effects of and the interaction with the visual context was not very impressive (rejection of the other hypotheses). A first explanation could be sought in the manipulation of the images. However, both the preliminary research and the manipulation check were quite convincing in the choice of the image features and their manipulation. It has not been too subtle, and it imitates what copy writers design for their target groups (but maybe wrongly so). A more subtle explanation is that playing with salience was not just done in the headline, but also between headline and visual context. That is, the inappropriateness of the wordplay was perhaps quite obvious for everyone, and the less salient meaning easy to grasp. The support of the images for the headlines was not really necessary. Instead, people might have experienced the common visual context as inappropriate and appreciate the inappropriateness as more tickling than the supportive, highbrow contexts (as the post hoc analysis pointed out). Perhaps the use of bright colors was over the top, and appreciated as deliberately inappropriate (irony, by overstating the advertisement properties of the ad). An indication for this line of reasoning might be that the difference in Attitude towards the ad between higher and lower educated respondents was opposite to our expectations: The common visual contexts were appreciated more by the higher educated. In case of recognizing visual properties of images, the resolution of inappropriateness cannot always be spelled out into (lexical) meanings. In sum, the results of the experiment support the graded salience and the optimal innovation hypotheses (Giora 2002; Giora et al. 2004), but only for the appreciation of the headlines. The interaction with the visual context might be explained by relevant inappropriateness (Attardo 2000), but that would require a more elaborate account of the extralinguistic aspects of this theory. McQuarrie and Mick s (1996) definition of rhetorical devices as artful deviations from audience expectations is a fluid one: when people are exposed to a certain rhetorical device too often, they do not respond to it in the desired way, but appreciate it as just the ordinary way of advertising. This process of conventionalization affects the saliency of both verbal and visual information: what was inappropriate at first will become salient later on. Inappropriateness seems to be the factor that enhances appreciation, be it linguistic or visual. The extent to which inappropriateness should be relevant as well in order to get the appreciation is a topic for further research. References Sharon L. Armstrong, Lila R. Gleitman, and Henry Gleitman. 1983. What Some Concepts Might Not Be. Cognition 13:263-308. Salvatore Attardo. 2000. Irony as Relevant Inappropriateness. Journal of Pragmatics 32:793-826. Subodh Bhat, Thomas W. Leigh, and Daniel L. Wardlow. 1998. The Effect of Consumer Prejudices on Ad Processing: Heterosexual Consumers' Responses to Homosexual Imagery in Ads. Journal of Advertising 27 (4):9-28.

John T. Cacioppo, Richard E. Petty, and Chuan F. Kao. 1984. The Efficient Assessment of Need for Cognition. Journal of Personality Assessment 48:306-307. Charles Forceville. 1996. Pictorial Metaphor in Advertising. London: Routledge. Morton A. Gernsbacher. 1984. Resolving Twenty Years of Inconsistent Interactions between Lexical Familiarity and Orthograhy, Concreteness, and Polysemy. Journal of Experimental Psychology: General 113:256-281. Ray W. Gibbs Jr. 1983. Do People Always Process the Literal Meaning of Indirect Requests? Journal of Experimental Psychology: Learning, Memory, and Cognition 9:524-533. Rachel Giora. 2002. On Our Mind : Salience, Context, and Figurative Language. Oxford: Oxford University Press. Rachel Giora, Ofer Fein, Ann Kronrod, Idit Elnatan, Noa Shuval, and Adi Zur. 2004. Weapons of Mass Distraction: Optimal Innovation and Pleasure Readings. Metaphor and Symbol 19 (2):115-141. Andrew Goatly. 1997. The Language of Metaphors. London: Routledge. Johan F. Hoorn. 1997. Metaphor and the Brain: Behavioral and Psychophysiological Research into Literary Metaphor Processing. PhD, Faculty of Arts, Vrije Universiteit, Amsterdam. Luuk Lagerwerf. 1998. Causal Connectives Have Presuppositions. Effects on Discourse Structure and Coherence. Utrecht: LOT.. 2002. Deliberate Ambiguity in Slogans. Recognition and Appreciation. Document Design 3 (3):245-260. Luuk Lagerwerf, and Anoe Meijers. 2005. Inferred Metaphors or Inappropriateness? Effects of Trope and Openness. Paper read at the International Communication Association, 55th Annual Conference, May 26-30, New York. George Lakoff, and Mark Johnson. 1980. Metaphors We Live By. Chicago: The University of Chicago Press. Thomas K. Landauer, and Susan T. Dumais. 1997. A Solution to Plato's Problem: The Latent Semantic Analysis Theory of Acquisition, Induction and Representation of Knowledge. Psychological Review 104:211-240. Scott B. MacKenzie, Richard J. Lutz, and George E. Belch. 1986. The Role of Attitude Towards the Ad as a Mediator of Advertising Effectiveness: A Test of Competing Explanations. Journal of Marketing Research 23:130-143. Robert R. McCrae, and Paul T. Costa Jr. 1999. Age Differences in Personality across the Adult Life Span: Parallels in Five Cultures. Developmental Psychology 35 (2):466-478. Edward F. McQuarrie, and David G. Mick. 1996. Figures of Rhetoric in Advertising Language. Journal of Consumer Research 22:424-438.. 1999. Visual Rhetoric in Advertising: Text- Interpretive, Experimental, and Reader- Response Analyses. Journal of Consumer Research 26:37-54.. 2003. The Contribution of Semiotic and Rhetorical Perspectives to the Explanation of Visual Persuasion in Advertising. In Persuasive Imagery. A Consumer Response Perspective, edited by L. M. Scott and R. Batra. Mahwah, NJ: Erlbaum. Andrew A. Mitchell. 1986. The Effect of Verbal and Visual Components of Advertisements on Brand Attitudes and Attitude Towards the Advertisement. Journal of Consumer Research 13 (1):12-24. Andrew A. Mitchell, and Jerry C. Olson. 1981. Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude? Journal of Marketing Research 18:318-332. David L. Mothersbaugh, Bruce A. Huhmann, and George R. Franke. 2002. Combinatory and Separative Effects of Rhetorical Figures on Consumers'Effort and Focus in Ad Processing. Journal of Consumer Research 28:589-602. Margot van Mulken, Renske Enschot-Van Dijk, and Hans Hoeken. 2005. Puns, Relevance and Appreciation in Advertisements. Journal of Pragmatics 37 (5):707-721. Barbara J. Phillips. 2000. The Impact of Verbal Anchoring on Consumer Response to Image Ads. Journal of Advertising 29 (1):15-25. Barbara J. Phillips, and Edward F. McQuarrie. 2002. The Development, Change and Transformation of Rhetorical Style in Magazine Advertisements. Journal of Advertising 34 (4):1-13. Ralph L. Piedmont. 1998. The Revised Neo Personality Inventory: Clinical and Research Applications. New York: Plenum Press.

Linda M. Scott. 1994. Images in Advertising: The Need for a Theory of Visual Rhetoric. Journal of Consumer Research 21:252-273. Dan Sperber, and Deirdre Wilson. 1995. Relevance: Communication and Cognition. 2nd ed. Cambridge, MA: Harvard University Press. Gerard Steen. 1994. Understanding Metaphor in Literature: An Empirical Approach. Edited by G. Leech and M. Short, Studies in Language and Linguistics. London: Longman. Yong Zhang. 1996. Responses to Humorous Advertising: The Moderating Effect of Need for Cognition. Journal of Advertising 25 (1):15-32. Appendix A. Example of one advertisement in four conditions Common visual context straightforward heading Love and kisses Common visual context ambiguous heading Speechless

Highbrow visual context straightforward heading Love and kisses Highbrow visual context ambiguous heading Speechless