SUPPLEMENTARY ONLINE APPENDIX
Sound and sound symbolism
Sound symbolism refers to the psycholinguistic process that explains the relationship between sounds and their meanings. Itimplies that the mere sound of a word provides cues about a word’s meaning while the word itself is meaningless (Yorkston Menon 2004). Sound symbolism applies to both vowels and consonants. Among others, one common classification of vowelsis byfront versus back distinction, taking into accounttheposition of the tongue during articulation. Front vowels are produced when the highest point of the tongue is positioned toward the front of the mouth as when pronouncing the sounds [i] (such as in the last sound of the word happy)or [e](such as in bed). In contrast, back vowels are produced when the highest point of the tongue is positioned toward the back of the mouth as for [o] (such as in for)or [u] (such as in actual). Among others, linguists classify consonants asplosives (i.e., a certain kind of stops; Klink 2000; Lowrey Shrum 2007; Yorkston Menon 2004) orfricatives. For the articulation of plosives, the mouth is closed and therefore the airflow is impairedfrom escaping the mouth. A sound like an explosion is then produced like for [k] (such as in key) or [t] (such as in ten). By contrast, fricatives are pronounced by not fully closing the mouth and causing a partial friction of the airstream as for [f] (such as in fat) or [s] (such as in soon).
Hypotheses
Sound-preference effects.In line with previous studies, we suggest that “preference for front versus back vowel sound words changes as a function of product category” (Shrum et al. 2012, p. 277; Lowrey Shrum 2007). We propose that brand name preferences derived from sound symbolism hold across languages of the same language family. As basis for our hypotheses, we expect thatconsumers preferbrand names for a two-seater convertible (SUV) that contain front vowels (back vowels) compared to back vowels (front vowels; Lowrey Shrum 2007). Replicating the study by Shrum et al. (2012), we hypothesize that these sound symbolism effects of vowels on brand name preference hold for English, French, German, and Spanish (H1). Extending the study byShrum et al. (2012)to consonants, weadditionally apply the same line of argument by varying brand names with respect to the consonant sound that is, whether brand names contain fricatives versus plosives (Klink 2000). Specifically, we propose a preference for brand names for a two-seater convertible (SUV) that contain fricatives (plosives) compared to plosives (fricatives). Again, we argue thatthese sound symbolism effects of consonants onbrand name preference hold for English, French, German, and Spanish (H2).
H1:Preference forbrand names of a two-seater convertible (SUV) that contain front (back) vowels compared to brand names that contain back (front) vowelsholdsacross languages (English, French, German, and Spanish).
H2:Preference forbrand names of a two-seater convertible (SUV) that contain fricatives (plosives) compared to brand names that contain plosives (fricatives) holdsacross languages (English, French, German, and Spanish).
Sound-perception effects. Research suggests that “people prefer particular words as brand names when the attributes connoted by the vowel sound of the word are congruent with the attribute of the product” (Lowrey Shrum 2007, p. 412;Yorkston Menon 2004).In the following, we will refer to this fit as sound-attribute fit. Sound-attribute fitexplains the previously discussed sound-preference effects by providing an explanation for the underlying mechanism (Klink 2000; Lowrey Shrum 2007; Yorkston Menon 2004). We propose that brandname perceptions derived from sound symbolism hold across languages of the same language family. In particular, we hypothesize that brand names that contain front vowels are perceived as smaller, faster, and lighter compared to brand names that contain back vowels (Klink 2000; Lowrey Shrum 2007) regardless of the languages of the Indo-European language family used (H3). We expect the same pattern for brand names that contain fricatives compared to plosives (H4).
H3:Across languages (English, French, German, and Spanish), consumers perceivebrand names that contain front vowels as (a) smaller, (b) faster, and (c) lighter compared to brand names that contain back vowels,and vice versa.
H4:Across languages (English, French, German, and Spanish), consumers perceivebrand names that contain fricatives as (a) smaller, (b) faster, and (c) lighter compared to brand names that contain plosives,and vice versa.
Stimuli Development
Selection of languages. Some evidence from psycholinguistics suggests that sound symbolism may elicit similar effects across different languages that belong to the same language family while they might vary between language families (O’Boyle, Miller, Rahmani 1987; Roper et al. 1976; Smith 1998). We therefore have selected the languages English, French, German, and French, which all belongto the Indo-European language family(Fortson 2010).
Development of fictitious international brand names. To create fictitious international brand names, we first used sound inventories to determine which vowels and consonants exist across English, French, German, and Spanish. We found that the front vowels [i] and [e] and the back vowels [o] and [u] as well as the plosives [k] and [t] and the fricatives [s] and [f] are employed in all four languages (International Phonetic Alphabet, IPA 2005). In a second step, we created 56 pairs of fictitious brand names for which eachonly varies in the vowel sound (i.e., front versus back vowel) and 77 pairs of fictitious brand names for which each only varies in the consonant sound (i.e., plosive versus fricative). Brand names were then transcribed in the respective phonographic writing system of each language. As a next step, we used dictionaries and an online brand search machine (United Domains 2012) to investigate whether these brand names exist. Brand names that were already available or provoked associations with existing brand names were eliminated. A set of 50 brand names was then pretested with five native speakers of each language to further exclude brand names that elicit negative, foreign, or any other semantic association and to ensure correct spelling and pronunciation when subjects read the brand names aloud (Shrum et al. 2012). The final set includes sixteen brand names of which fourword pairs differ in vowel (front versus back vowel) and fourword pairs differ in consonant (plosive versus fricative) sound.
Table 1: Classification of international brand names tested in this study
Participants and Procedure
In total, 456subjects participated in our online survey,N =213 native speakers of German (126 women, Mage = 30.3), N =74 native speakers of English (46 women, Mage = 30.8), N =74 native speakers of Spanish (42 women, Mage= 28.2), and N =95native speakers of French (58 women, Mage = 27.1). Participants were told that for each accomplished questionnaire, the authors donated10 cents to Engineers without Borders, a non-governmental organization supporting international development work. Each participant received the survey in his/her mother tongue that is, in English, French, German, or Spanish. Before, we had pre-tested our survey for clarity in task instruction and consistency across languages.
In line with Shrum et al. (2012), participants self-selected their mother tongue, while they were randomly assigned to a car type (SUV, N = 229; two-seater convertible, N = 227). Apart from that, all participants received the same set of stimuli that differed only in the language of the survey. Accordingly, phonetic symbolism for vowels and consonants was a within-subjects factor, and both language and car type a between-subjects factor.
Participants were told that the study’s objective wasto identify a brand name for the respective car type (Shrum et al. 2012). Throughout the experiment, we continuously asked them to read the brand names aloud, or, in case the subjects were not able to speak loudly, to mentally read the brand names aloud (Baddeley 2003).
In the first part of the questionnaire, participants were presented a series of eight word pairs. Four word pairs differed by one vowel (Nooran-Neran, Nonra-Neenra, Nooma-Neema, and Nolam-Nelam) representing our manipulation of phonetic symbolism of front ([e] and [i]) versus back vowel ([o] and [u]) sounds.The remaining four word pairs differed by one consonant (Calrum-Falrum, Tumral-Sumral, Tanum-Fanum, and Kemrum-Semrum)representing our manipulation of phonetic symbolism of consonant sounds for plosives ([k] and [t])versus fricatives([s] and [f]). The order of word pair presentation was randomized. Participants were asked to state their preference for one brand name of each word pair.
The second part of the questionnaire dealt with brand perceptions to investigate sound-attribute fit. Participants evaluated brand names with respect to size (big versus small), speed (fast versus slow), and weight (heavy versus light) on seven-point semantic differential scales for the above mentioned word pairs. Order of word pair presentation was again randomized. Following this exercise,participants were asked to evaluate a typical two-seater convertible (SUV) in terms of size, speed, and weight (Lowrey Shrum 2007). Finally, subjects were asked to indicate their age, gender, and nationality.
Resultsfor sound-preference effects
Table 2 provides an overview on the results of the original studies (Lowrey Shrum 2007; Shrum et al. 2012) and our study with respect to sound-preference effects across languages. Specifically, we indicate which cross-language and language-specific effects we were able to replicate for different vowelsandto extend to consonants.
Replication: Sound-preference effectsof vowels in international brand names.We first created continuous dependent variables that represented the proportion of front and back vowel words chosen for each car type (Lowrey Shrum 2007; Shrum et al. 2012).We then ran a 2 (vowel sound) x 2 (car type) x 4 (language) mixed-model analysis of variance (ANOVA) to test H1.As expected the sound by car type interaction issignificant (F(1, 448) = 10.67; p = .001) with post hoc paired t-tests indicating a preference for brand names of a two-seater convertible (SUV) that contain front vowels (back vowels) compared to back vowels (front vowels) (Mconvertible = 57.5%, t(226) = -4.10, p .00; MSUV =53.9%, t(228) = 2.05, p < .05). Furthermore, a non-significant 3-way interaction by language, sound, and car type confirms the assumption of a language-independent preference for brand names of a two-seater convertible (SUV) that contain front (back) vowels compared to brand names with back (front) vowels (Shrum et al. 2012) (F(3, 448) = 1.33; p = .26).
However, a significant language by vowelinteraction (F(3, 448) = 3.22; p = .02) indicates a language-dependent preference for front versus back vowels regardless of car type.While post-hoc performed paired t-testsfor sound and language indicate that there is no general preference for front versus back vowels for German, French, and Spanish native speakers, English native speakers generally prefer the tested front vowels over tested back vowels (MEnglish = 60.8%, t(73) = -3.14, p < .01). In addition, post-hoc performed paired t-tests within each language and within each car type reveal three non-significant differences between the meansof front versus back vowels (MFrench_convertible = 42.9%, t(48) = -1.59, p > .05; MFrench_SUV= 53.3%,t(45) = .85, p > .10; MSpanish_convertible = 54.9%,t(35) = -1.23, p .10). It is important to note that sample sizes may account for these unsuccessfully replicated effects.
Sound-preference effects of consonants in international brand names. To test whether a similar pattern as shown by Shrum et al. (2012) also exists for consonants, we created continuous dependent variables that represented the proportion of fricative and plosive words chosen for each car type, and ran a 2 (consonant sound) x 2 (car type) x 4 (language) mixed-model ANOVA. As expected the sound by car type interaction is significant (F(1, 448) = 10.43; p < .01). In contrast to the overall sound-preference effect of vowels in international brand names, we found a significant interaction by consonant sound, car type, and language (F(3, 448) = 2.80; p = .04) and thus, cannot confirm H2. In addition, the interaction between sound and language is significant (F(3, 448) = 4.83; p < .01). Post-hoc performed paired t-tests for sound and language indicate a general preference for plosives over fricatives for German (MGerman = 60.3%, t(212) = -5.33, p < .00) and English (MEnglish = 58. 5%, t(73) = -2.99, p < .01) native speakers. Individual paired comparisons within each car type and each language demonstrates a tendency toward the preference for brand names that contain plosives for SUVs across all languages (MSpanish_SUV = 61.2%, t(37) = -2.90, p < .01) besides French (MFrench_SUV = 51.1%, t(45) = -.28, p > .10).Due to sample size constraints, comparisons for English are significant at p < .1.
1
Table 2: Results of replication and extension of sound-preference effects in international brand names
1
Results for sound-perception effects
Table 3 provides an overview on the results of our study with respect to sound-perception effects across languages. Specifically, we indicate which cross-language and language-specific effects we were able to extend for vowelsand consonants. These effects were only pretested in the study by Lowrey & Shrum (2007).
Car type–attribute fit. Toensurethatcar types are associated with typical attributes in the same way across all of the four languageswhich were examined in the present study, we analyzed how participants rated an SUV compared to a two-seater convertible in terms of size (big versus small), speed (fast versus slow), and weight (heavy versus light; all items measured on seven-point semantic differential scales) using t-tests. In comparison to a two-seater convertible, an SUV is generally considered bigger in size (M = 1.96, t(454) = -31.6, p < .01), slower (M = -.39, t(454) = -11.6, p < .01), and heavier (M = 1.7, t(454) = -26.7, p < .01). These findings do not vary across the four languages examined in this study (all ps .01).
Sound–perception effectsof vowels in international brand names. We investigated whether differences in sound symbolism for front vowels versus back vowels exist for the four languages in terms of brand name perception. Accordingly, we ran three separate 2 (vowel sound) x 4 (language) mixed-model ANOVAs. Allanalysesshow a significant main effect of sound (speed: F(1, 452) = 29.46; p = .00). The results confirmthat front vowels compared to back vowels in brand names elicit associations of smaller size in English, French, German, and Spanish (H3a: F(3, 452) = 1.61; p = .19). We find similar patterns for fastness (H3b: F(3, 452) = 2.33; p = .07) and lightness (H3c: F(3, 452) = .17; p = .91). Individual paired t-tests confirm these findings exceptfor Spanish people’s perception of speed(M=-.79,t(73) =.54, p > .10).
Sound–perception effectsof consonants in international brand names. We ran the same analyses for the comparison between fricatives and plosives across the four languages. All ANOVAs show a significant main effect of sound (speed: F(1, 452) = 14.98; p = .00). We find support for H4a, which predicts that regardless of the language, consumers associate brand names that containfricatives compared to brand names that contain plosives with smaller size (F(3, 452) = .59; p = .62), higher speed (H4b:F(3, 452) = 1.71; p = .16),and less weight (H4c: F(3, 452) = .96; p = .41).Again, we conducted paired t-tests, which confirm the results of the ANOVAs except for the perception of speed by Germans(M=-.21, t(212) =.90, p > .10) and Spanish people (M=-.70, t(73) =1.52, p > .05).
1
Table 3: Results of sound–perceptioneffects in international brand names
1
References not cited in the manuscript
Baddeley, A. (2003). Working Memory: Looking Back and Looking Forward.Nature ReviewsNeuroscience, 4 (10), 829-839.
Fortson, B. W. (2010). Indo-European Language and Culture: An Introduction. (2nd ed.).Malden, Mass. et al.: Blackwell.
IPA (2005). URL:
Klink, R. R. (2000). Creating Brand Names With Meaning: The Use of Sound Symbolism. Marketing Letters, 11 (1), 5-20.
O’Boyle, M. W., Miller, D. A., & Rahmani, F. (1987). Sound-Meaning Relationships in Speakers of Urdu and English: Evidence for a Cross-Cultural Phonetic Symbolism.Journal of Psycholinguistic Research, 16 (3), 273-288.
Roper, C. W., Dixon, P. W., Ahern, E. H., & Gibson, V. L. (1976). Effect of Language and Sex on Universal Phonetic Symbolism.Language and Speech, 19 (4), 388-396.
United Domains (2012). URL:
Yorkston, E., & Menon, G. (2004). A Sound Idea: Phonetic Effects of Brand Names on Consumer Judgments.Journal of Consumer Research, 31 (1), 43-51.
1