Critique of Bilingualism and Its Effects on the Lexical Interpretation of Emotions

Introduction:

Acevedo-Molina (2021) from the Association for Psychological Science states that culture and language significantly impact one's interpretation of emotions. One’s bilingualism is often composed of how early the two languages were acquired and their level of proficiency. People learn languages at varied ages, typically at home, and/or at an institution. Learning two or more languages at the same time can be challenging. Psychology Today’s Grosjean (2011) explains that the expression of emotions through language can vary. Depending on someone’s language proficiency, they can express a variety of emotions in their language. 

On the topic of variance, it is essential to understand that the term bilingual is nuanced. Even though someone may know two languages, their abilities can differ between reading, writing, and speaking. Furthermore, it is difficult not to have a dominant language. Language context, immigrant generation, and cultural factors are some of the key factors in the variation of bilingualism. Both studies discussed in this paper have varied levels of language proficiency in their participants. 

Dekeyser et al.’s (2024) study, “Do Bilinguals Have an Emotional Advantage?” explores how bilingual adolescents who are ethnic minorities have varying levels of emotional competency (EC).  Two methodological areas in need of improvement are the number of participants and the study design. There was a limited pool of participants, and the researchers solely relied on questionnaires. Another study discussed in this paper is Gu and Chen’s (2024), “Two languages, two emotional minds in one brain: processing emotion-label and emotion-laden words by Chinese-English bilinguals,” in their methodical approaches and implications. 

While both studies had good measures, only Dekeseyer et al.’s (2024) study noted its limitations. 

Methods:

In Dekeyser et al.’s (2024) study, they administered an in-class, online survey given to 924 children (ages 10-12) at 19 Dutch-speaking schools in Antwerp. The participants’ demographics were selected from low socioeconomic statuses. Home language (HL) vs institutional language (IL) media exposure was from reading, music, or TV. They were placed into 3 clusters based on their proficiency in Dutch: high Dutch and high HL, intermediate Dutch and high HL, and high Dutch and low HL. A dummy variable was created to control for gender. This was used as a control along with the other variables. 

Their statistical analyses and measures were as follows. The Differentiating Emotions subscale of the Emotion Awareness Questionnaire (Rieffe et al., 2007) results were used with R coding to determine the degree of EC. Additionally, they used the Kruskal-Wallis test (comparing 3+ significant independent groups) and linear regression models. Researchers collected data about the participants’ demographics, language proficiency, and their associated EC, and how mediums of exposure affect that. 

A comparable study would be Gu and Chen’s (2024) study. They were interested in how emotional effects from reading related to bilinguals (Chinese and English-speaking students from Ocean University, China). The study was composed of two cohorts. The first cohort had 25 people, ages 19-22, who were normal (eyesight and no neurological impairments). Both cohorts took a survey to indicate their proficiency with the two languages and signed a consent form to participate. But, they also had emotional categorization tasks as opposed to Dekeyser et al.’s (2024) primary survey. The participants were given 145 English nouns. 35 English majors acted as judges to evaluate the stimuli based on the level of valence they had through event-related potentials (ERPs) and response time recordings. ERPs illustrate how timing and events are related to brain activity. In cohort 2, there were 27 bilinguals, ages 19-23, who were normal to the researchers’ standards. 12 more judges were added. 12 English majors measured the translation equivalency, 15 evaluated grammar, and 20 ranked the emotion states. Response times and brain activation regions were recorded.
 

Key Findings:

Table 1. Note that Model 2 proved to have a higher R². The context and proficiency for language are related to EC.

Coefficient of Determination (R²) was measured for different levels of proficiency in Dutch and home language (HL) in association with emotional competency (EC) levels. 

Dekeyser et al. (2024) found that the three clusters had distinct levels of EC. In their methods, 18.55% were monolingual, 41.43% of the participants were highly proficient in both Dutch and HL, 23.21% in the middle, and 16.81% for high levels of proficiency in Dutch and low HL. Ordinary least squares (OLS) linear regression values between language proficiency and EC were: High Dutch and High HL, Intermediate Dutch and High HL, and High Dutch and Low HL. The first cluster had the highest levels of EC, the 2nd cluster was lower than monolinguals, and the last cluster was marginally lower, yet similar to monolinguals in terms of EC. While the R²  (model fit) values aren’t the closest to 1, model 2 performed better than model 1. The key difference is that model 2 included only the bilingual population, rather than the monolingual population. Thus, language proficiency does affect EC. Language learning mediums, such as exposure to Dutch media, yielded positive correlations with EC. But interestingly, the Belgian socialization efforts (family members teaching Belgian culture directly to the child) led to slight negative correlations. 

The authors also discussed how cultural differences in socialization and emotions led to differences in the development of positive and negative emotions. Western cultures tend to be more subjective to the individual, whereas other cultures can be more objective and contextual regarding social situations. Two instances of this would be American and Spanish culture. American culture stresses an emphasis on individualism, and this can be expressed through emotions and feelings of anger and discontent. This can lead to more direct and confrontational arguments and can clash with people from other cultures. English-speaking cultures can also be more rigid when expressing emotions through language. For example, there is only one way to say I love you in English. On the other hand, Spanish culture focuses on familialism, respect, and collectivism (Acevedo-Molina, 2021). There are multiple ways to express emotions for one another. Beyond the English, I love you. With this, there can be more feelings of shame with shortcomings in social and performance contexts. 

Critiques of Dekeyser’s findings:

 Dekeyser et al. (2024) mention that their limitations were due to a lack of variation in multilingual profiles, analytical models for correlation, and the context of learning language beyond just school and home settings. Their data tables reveal the relationship between R values and participants’ EC in relation to their cluster and/or demographics. Their study could be more comprehensive if they used better programming software to evaluate their R values, which would result in higher reliability and accuracy. The lack of variation makes sense as they based their study on a certain demographic - socioeconomically disadvantaged students at the 19 Dutch-speaking schools in Antwerp. 

As for the other study, Gu and Chen’s (2024), it is questionable whether there were any limitations listed in their article. This is concerning as this shows a lack of proper reflection in their research. Faults with their study design should also be noted. There is a discrepancy in the number of observers between experiments 1 and 2. Experiment 1 had 35 bilinguals rank the stimuli of words on a 9-point scale. Experiment 2 had 47 English majors (sets of 12, 15, and then 20) act as judges to evaluate the matching of translation definitions. The imbalance of judges led to questions about the accuracy of the stimuli. Despite having fewer participants, they did have varied mediums for evaluating EC. They had a diagnostic questionnaire to assess their level of proficiency, judges, ERPs, and brain scans. Thus, they didn’t have the most representative sample, but their data would be reliable and valid.  

Furthermore, there could have been a wider range of sociodemographic classes. Drekeyser et al. (2024) mentioned that having more variance of sociodemographic factors (i.e., gender, age, socioeconomic status) could reveal more accurate findings about the spectrum of language proficiency and EC. It could be possible to predict what the results would be, low EC. Regardless, their study had a lack of variation, and little information was collected about their language learning in home and school contexts. Whether the contexts were family dynamics, specific language classes, and/or teaching methods, analyzing these factors would give more insight into the traits that led to differences in language proficiency and EC. 

Dekeyser et al.’s (2024) method of collecting data via self-reporting and questionnaires is simple and cost- effective. However, it is susceptible to bias, limited to the respondent’s perspective, and lacks context. The participants are 10-12 years old, so they might not fully understand the questions. As such, questionnaires can be subjective. If the participants interpret the questions differently, then the researchers are not collecting accurate data. Perhaps having the participants take the questionnaire several times may help identify inconsistent responses, improving the reliability of data. mitigate false data. In general, it’s more thorough to have multiple tests rather than an overreliance on one.  The number of participants may have helped mitigate these limitations by revealing the overall trends in the collected data; individual deviations would not be accounted for. 

Consequently, there are many factors that affect one’s language proficiency and EC, which is a perspective that should be noted for all bilingual studies. This could be achieved by setting up the research methods to have at least 2+ cohorts to account for the variation of bilingualism in languages. Or else the data would not be representative of population differences in language proficiency. Having proper representation and multiple age groups can be more representative of today’s society.

Critiques of Implications and Today’s Relevance:

Dekeyser et al.’s (2024) findings reveal that being a strong bilingual has a positive correlation with EC. Their data and interpretations are reliable and accurate to a certain extent. However, the study is limited to socioeconomic class (low), cultural context (Flemish and Belgian culture), learning context (school, home, etc.), and age (only ages 10-12). More studies need to be conducted to observe how variance in these areas affects EC. It has been observed that the more proficient bilinguals are, the better they can understand and interact with others in different contexts. Unfortunately, imbalances in proficiency can result in EC disparities. Generally, positive emotional detection is developed during the end of childhood, and negative detection is developed until adolescence (Grosjean, 2011). Thus, the ability to detect emotion is correlated with one’s age. Considering how adolescents are an understudied population, it would be more representative to include multiple age groups. Using a cross-sectional or a longitudinal study design would be more effective than relying on one questionnaire, as this would allow researchers to observe multiple age groups at once or be able to track the same participants over a longer period of time. In today’s world, interracial relationships and cross-cultural communication with technology are becoming increasingly prevalent. Overall, there does seem to be a positive correlation between EC and bilingualism vs monolingualism. It is important to note that other factors, such as SES, language proficiency, and contexts for language learning, are important to keep in mind. EC is a key skill to have in everyday social interactions and situations.

References:

1. Acevedo-Molina, M. C. (2021, December 29). Student notebook: Emotion processing in Bilinguals. https://www.psychologicalscience.org/observer/student-notebook-acevedo-molina 

2. Dekeyser, G. N. M., Dewaele, J., & De Leersnyder, J. (2024). Do bilinguals have an emotional advantage? An exploratory study on the interplay of language, culture and emotion. Language, Culture and Curriculum, https://doi.org/10.1080/07908318.2024.2424966

3. Grosjean, F. (2011, August 18). Emotions in more than one language. https://www.psychologytoday.com/us/blog/life-bilingual/201108/emotions-in-more-one-language 

4. Gu, X., & Chen, S. (2024). Two languages, two emotional minds in one brain: processing emotion-label and emotion-laden words by Chinese-English bilinguals. International Journal of Bilingual Education and Bilingualism, 28(2), 216–236. https://doi.org/10.1080/13670050.2024.2417827