%A ZHU Lin, LIU Jinru, LI Jing, LIU Conghui %T Moral foreign language effect and its moderating variables: A systematic review and meta-analysis %0 Journal Article %D 2022 %J Advances in Psychological Science %R 10.3724/SP.J.1042.2022.00032 %P 32-50 %V 30 %N 1 %U {https://journal.psych.ac.cn/xlkxjz/CN/abstract/article_5711.shtml} %8 2022-01-15 %X

Individual decisions in the field of moral judgement are often related to "hurting or sacrificing the innocent" and "tolerating immoral behaviors." Previous studies have shown that when presented with the moral decision-making situation within a foreign language context, the individuals will show a stronger utilitarian and a more tolerant moral evaluation tendency compared with the same situation within the native language context. This phenomenon is defined as the moral foreign language effect. The influence of the language context on the moral judgement has been investigated by numerous studies. However, the results were far from consistent. To this end, we used meta-analysis to explore the effect of the language type (native language vs. foreign language) on the individuals’ utilitarian tendency in moral judgments, and we analyzed several moderating variables.
A total of 19 papers were retrieved from literature, with 46 independent samples, 97 effect sizes and 9672 participants that met the inclusion criteria of the meta-analysis. First, we analyzed the effect of the language type (native language vs. foreign language) on the utilitarian tendency in moral judgments using the ‘metafor’ R package. Next, the potential moderation effects of several factors were examined, including the moral dilemmas story type (personal moral dilemmas vs. impersonal moral dilemmas vs. daily moral evaluation situations), sex, scoring method (two-point scoring vs. multi-point scoring) and language family type (same vs. different). In addition, we used Bayesian factor estimation for secondary exploration of the results that had a nonsignificant moderating effect.
Our meta-analysis resulted in the following findings. First, the main effect test indicated that the language type has a significant effect on the utilitarian tendency in moral judgment, with a small but stable moral foreign language effect (g = 0.23). Second, the moderation analysis indicated that the moral foreign language effect was influenced by the story type; there was a small but stable effect of the language type for personal moral dilemmas (g = 0.32), but not for impersonal moral dilemmas (g = 0.11) or daily moral evaluation situations (g = 0.12). The foreign language effect under impersonal moral dilemmas was affected by the scoring method; a significant effect was found under multi-point scoring (g = 0.27), but not under two-point scoring (g = 0.05). On the other hand, there was no significant moderating effect for the sex or language family type. In addition, Bayesian analysis showed only moderate evidence for the absence of moderating effect regarding the factors of sex, scoring methods and language family type. The stability of these conclusions can be further verified in future research.
In summary, this study used meta-analysis to systematically explore the robustness and influencing factors of foreign language effects in moral judgment and answered the disputes about the stability of the moral foreign language effect. The results showed a small but relatively stable effect of the language type on the utilitarian orientation in moral judgment. We analyzed the moderating effects of multiple variables, including variables that have not been well-considered in previous studies, such as the scoring methods (two-point scoring vs. multi-point scoring). Our work did not only find the moderating effect of the type of moral dilemmas, but it also revealed the potential impact of the scoring method on the effect size. This provides certain enlightenment and guidance for future empirical studies when selecting the experimental materials and statistical methods. Finally, we used a variety of data processing methods to increase the robustness of the results. For example, robust variance estimation (RVE) was used to control the correlations between dependent effect sizes and compare our results with those of traditional meta-analysis, so as to understand how the results of the meta-analysis are influenced by the correlations between multiple dependent effect sizes.