Article of the Day:
Is personality modulated by language?
by Veltkamp, G. M., Recio, G., Jacobs, A. M., & Conrad, M. (2012)
(International Journal of Bilingualism, 17(4). 496-504. DOI: 10.1177/1367006912438894)
Background & Research
The complex relationships between language and culture, as well as thought and personality, have been the subject of great interest and research since at least the 19th century, for psychologists, linguists, philosophers and even romantic/nationalist thinkers. Linguistic relativity, for instance, was a particularly hot topic in the first half of the 20th century. As regards more recent psychological research, there are indications that different languages carry different emotional tones, in that people fluent in several languages behave and feel in different ways when speaking one language as compared to another. There is also evidence that language learning on the whole is closely tied to the internalization of cultural norms and values that appear to be inherently present in the language. It has been said that culture provides different formulaic ways for expressing thoughts and feelings, and if this is the case, language is the most likely process for this to occur through.
For multilingual individuals, researchers have suggested that by learning new languages, individuals simultaneously gain access to new cultural meaning systems, and can then switch between different culturally appropriate behaviors. If that is the case, might expressed dimensions of personality also change when using another language?
For the most part, standard cross-cultural Big Five personality tests (e.g., the BFI as well as the NEO-FFI used here) show good cross-linguistic reliability with strong item correlations for bilingual populations, seemingly contradicting the idea that personality might change with the language used. However, language differences have rarely been explicitly addressed, and between-item correlations are probably not adequate to discover them. One previous study examining this specific issue (Ramirez-Esparza et al., 2006) concluded that Spanish–English bilinguals in the United States and Mexico were more extraverted, agreeable, and conscientious in English than in Spanish, as assessed by the BFI personality test.
The study this article reports on extends such research by studying late (post-adolescence) learners of a second language, taking into account possible differences between L1 (“native”) versus L2 (second, nonnative) language use. The intention was to investigate whether an “automatic cultural frame shift” that has been observed with regard to personality in bilingual individuals who have learned both languages in their childhood would also be seen in late L2 learners. As personality is thought to be relatively stable by this stage, it is conceivable that these late learners would no longer be able to achieve a new “culturally framed and language-modulated persona” by learning another language.
To examine these issues, 68 German-Spanish “late bilingual” individuals (40 native German speakers and 28 native Spanish speakers) were asked to complete both the German and Spanish language versions of the NEO-Five-Factor Personality Inventory, and the mean levels of scores in each domain (Neuroticism, Extraversion, Openness, Conscientiousness and Agreeability) were then statistically compared. The authors considered it conceptually legitimate to employ mean levels (aggregate domain scores) for their comparisons, based on the fact that correlations between symmetrically translated items in the German and Spanish language versions were high (Cronbach’s α = .791 and α = .760 for the German and Spanish native groups, respectively)
Results & Discussion
An ANOVA with test language as the within-subject factor showed statistically significant effects for Neuroticism (F(1,49) = 5.15; p = .028), Extraversion (F(1,43) = 30.65; p < .0001) and Agreeability ( F(1,51) = 16.98; p < .0001.) but not for Conscientiousness or Openness. In other words, when participants took the test in Spanish, mean scores were significantly higher for Neuroticism (difference in means 1.74 points) and Extraversion (difference in means 3.11 points, and significantly lower for Agreeability (difference in means -2.98 points), as compared to the German version of the test.
The rule out possible effects of the overrepresentation of German native speakers, equal-sized subsets of the participants, matched on a number of control variables were used for further analyses.
A second ANOVA with test language as a within-subjects factor and L1 group as a between-subjects factor replicated the findings of the first one. Statistically significant differences across test language were again observed in the same NEO-FFI domains: Scores were higher in the Spanish version for Neuroticism (F(1,38) = 8.74, p = .006) and Extraversion (F(1,40) = 31.01, p < .0001), and lower for Agreeability (F(1,40) = 14.86, p = .0004), as compared to the German version. No significant effects were found for L1 group or test language X L1 group interactions for any of the domains. The authors thus conclude that the observed differences in personality dimension scores appear to be due to test language only.
The findings that differences in the three personality dimensions were modulated by test language are in line with previous research by Ramirez-Esparza et al. (2006) who demonstrated for the first time that personality test scores of bilingual individuals were affected by the language used for the test. The direction of the differences as regards the Spanish language was similar for Agreeability (lower in Spanish than in German or English) and Neuroticism (higher in Spanish than in German or English), but different in Extraversion, where this study found higher scores for Spanish compared to German but Ramizer-Esparza et al. reported lower scores for Spanish as opposed to English. The samples, of course were quite different, with the participants hailing from Northern America in the Ramirez-Esparza et al. study, and these partly differing results underline the complexity of the interactions between language, culture and personality. However, these results do on the whole agree with normative results reported by McCrae and others for the specific culture groups under study.
In any case, put together, these two studies point to a cultural frame shift in the way bilingual individuals present their personality depending on the language used. Crucially, the direction of the observed effects of test language was the same for the two groups with different native language studied. Hence, these effects do not appear to be limited to, e.g., individuals growing up bilingual as children and having to adapt to different cultural frames, which might otherwise explain differential personality patterns when using different languages. Contrariwise, consistent cultural frame shifts appear to occur in late second-language learners too, regardless of native language or cultural background. In the authors’ view, this suggests that “learning a second language always implies the automatic representation of new cultural frames associated with this language.” In other words, it seems an individual learning a new language is presented with new ways of perceiving and displaying his/her own personality, probably considered a blessing by most second-language learners.
The simple study reported on here provides us with very interesting results regarding the differences in (the expression of) personality within the same individual when using different languages. Its findings certainly merit reproduction with larger samples and a wider array of languages and measures.
Some potentially interesting information was left out of this article for some reason. For one, the authors do not report effect sizes for the examined effects, which should be standard by now. Second, mean scores for the different domains of the NEO-FFI previously observed in the different language versions with general or monolingual populations would have been of great interest for comparison.
Though the authors consider their use of the mean levels of domain scores of one self-rated personality test legitimate for comparing personality features when using a particular language, some questions remain in my mind about this measure. First, the observation that scores in some domain in one language version of a test tend to be overall higher than in another language version can be related to a wide array of issues, and differentiating between them seems difficult. We may be dealing with “cultural” differences in personality expressed through language, but the differences may also be due to the translation process (how the questions are formulated) or dissimilarities in test-taking attitudes (e.g., how important it is to present oneself in a positive light, how careful people are in reporting shortcomings). Similarly, high item correlations between two languages may speak of many different issues. In any case, employing, at a minimum, two separate measures would greatly have improved the credibility of the evidence. Even more optimally, an observational/task-type element could have been included, e.g., to see if bilingual individuals would act differently in a task related to personality features when using a different language.
On a somewhat far-fetched, but still related note, it is interesting to ponder whether “forced” changes in the meanings of words and language use more generally could affect the (displaying of) personalities of an entire language group. It seems to me something akin to this has occurred on a relatively large scale in some countries (e.g., historical fascist nations, authoritarian China, and to a more extreme degree, North Korea). The cultural values of a nation can clearly be fundamentally affected by the draconian policies of totalitarian/authoritarian states. To what extent might this take place through the manipulation of the very language used?