Introduction
Research has shown that the use of a foreign language can have an effect on our moral judgement and decisions (e.g., Cipolletti et al., Reference Cipolletti, McFarlane and Weissglass2016; Costa et al., Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014; Geipel et al., Reference Geipel, Hadjichristidis and Surian2015a, Reference Geipel, Hadjichristidis and Surian2015b, Reference Geipel, Hadjichristidis and Surian2016). In a seminal study, Costa et al. (Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014) administered the trolley dilemma task, where saving the life of five people actively involves sacrificing the life of one person. The authors showed that presenting the dilemma in a language that participants had learnt as a second language (L2), greatly increased their stated willingness to sacrifice that one person. In the domains of morals, judgements and decisions, it has been confirmed that the foreign language elicits less intense emotional reactions if compared to a native language and that using a foreign language seems to make people more rational and it increases utilitarian choices (Corey & Costa, Reference Corey and Costa2015). The foreign language has been found to reduce risk aversion, making people more prone to opt for a sure outcome instead of taking the risk of a gamble with unsure outcomes (Hadjichristidis et al., Reference Hadjichristidis, Geipel and Savadori2015; Keysar et al., Reference Keysar, Hayakawa and An2012; Winskel & Bhatt, Reference Winskel and Bhatt2020; Xing, Reference Xing2021). It has also been shown to reduce causality bias, the illusion of a relation of causality between two events when they are not causally related (Díaz-Lago & Matute, Reference Díaz-Lago and Matute2019) and to suppress superstition (Hadjichristidis et al., Reference Hadjichristidis, Geipel and Surian2019). Research has also shown that bilinguals perceive dishonesty as less inappropriate in their foreign language (Alempaki et al., Reference Alempaki, Dogan and Yang2017), and perceive crimes described in a foreign language as less severe (Woumans et al., Reference Woumans, Van der Cruyssen and Duyck2020).
Tolerance of ambiguity is the ability of an individual to perceive, process, and respond to ambiguous or uncertain situations without experiencing undue stress or anxiety (Budner, Reference Budner1962). Furnham and Ribchester (Reference Furnham and Ribchester1995) found that individuals who score high on measures of tolerance of ambiguity are more likely to engage in creative problem-solving and to embrace novel and unconventional ideas. On the contrary, individuals with low tolerance of ambiguity may feel more anxious or stressed in ambiguous situations and may be more prone to rigid thinking, avoidance, or decision-making based on incomplete information (Furnham & Ribchester, Reference Furnham and Ribchester1995).
Measurements of tolerance of ambiguity have been used in various contexts and fields. For example, the ability to tolerate ambiguity has been shown to have an impact on medical doctors’ level of perceiving work-related stress (Ianello et al., Reference Ianello, Mottini, Tirelli, Riva and Antonietti2017). Low tolerance for ambiguity has been seen to be associated with the tendency to order more diagnostic tests (Ianello et al., Reference Ianello, Mottini, Tirelli, Riva and Antonietti2017) and it has been shown to interfere with medical decision making (Kruglanski et al., Reference Kruglanski, Webster and Klem1993; Roets et al., Reference Roets, Raman, Heytens and Avonts2013). It has also been seen to be related to lower tolerance of uncertainty (Gärtner et al., Reference Gärtner, Bußenius, Prediger, Vogel and Harendza2020). It has been assumed that measuring the ToA in medical school applicants could be interesting in the selection process in order to choose the best applicants for undergraduate medical schools (Gärtner et al., Reference Gärtner, Bußenius, Prediger, Vogel and Harendza2020).
Previous studies have pointed out how bilingual and multilingual individuals are more tolerant to ambiguity if compared to monolingual speakers (e.g., Dewaele & Li, Reference Dewaele and Li2013). In a subsequent study, Dewaele and Botes aimed to investigate the effect of multilingualism on five high-order personality traits and extended previous findings showing that knowing more languages provides positive traits to individuals (Dewaele & Botes, Reference Dewaele and Botes2020). They found that knowing more languages was positively related to openness to experience and extraversion. The study suggests that the earlier someone learns additional languages and the more they use them, the greater the effect on personality traits. The findings contribute to the growing body of literature on the relationship between multilingualism and personality.
Research has seen the use of a second language implying a reduction in the emotional response in that language, leading to a certain degree of emotional distance (Costa et al., Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014; Iacozza et al., Reference Iacozza, Costa and Duñabeitia2017). The less emotion is experienced in a situation, the more control it is possible to have in that situation and the use of a foreign language alters the emotion caused by the message (Gross, Reference Gross2012; Keysar et al., Reference Keysar, Hayakawa and An2012). For this reason, people might be more rational and logical in a situation involving their second language (e.g., Costa et al., Reference Costa, Foucart, Hayakawa, Aparici, Apesteguia, Heafner and Keysar2014). They might be more prone to invest in situations they are not used to and that are distant and unfamiliar from them. They might invest more easily in challenging or unfamiliar situations and might see change as an opportunity. Previous research also suggests that moral FLE appears to be stronger among bilinguals with lower self-reported foreign language proficiency (Stankovic et al., Reference Stankovic, Biedermann and Hamamura2022).
The present study investigated whether there is a L2 effect on tolerance of ambiguity and whether presenting information in a foreign language (as opposed to the native language) makes people more tolerant to ambiguity. It consisted of a questionnaire that participants completed in either their first or second language. We predicted that participants who answered the questionnaire in their second language (L2 = English) were more tolerant to ambiguity compared to participants who answered the same questionnaire in their first language (L1 = Italian).
Methods
Participants
Sample size was determined a-priori using G*power with: between participants design, d = .4 (small/moderate size; cf. Circi et al., Reference Circi, Gatti, Russo and Vecchi2021), alpha = .05, and power = .8. Results indicated a minimum sample size of 156 (see pre-registration: https://osf.io/7rn9z)
Five hundred and twenty six participants volunteered to take part in the study and completed an online questionnaire measuring ToA. They also provided biographical and linguistic information. After eliminating incomplete answers, three hundred and eighty seven participants were included in the analysis (251 female, 118 male, 10 non binary, 8 did not provide this information); mean age = 27.15 years (41 participants did not provide this information, age range: 18-76 years). 207 participants were randomly assigned to the Italian questionnaire (L1_Italian condition), and 180 to the questionnaire in English (L2_English condition).
All participants had studied English as a second language as part of their previous scholastic education. On average, participants in the L2_English condition have had English education since the age of 9.33, with speaking in L2: 9.42 (39 answers not given) – reading in L2: 9.13 (41 answers not given) – writing in L2: 9.44 (43 answers not given) and have studied English on average for 8 14.98 years (41 answers not given).
Participants were asked to self-assess their foreign language proficiency in terms of speaking, reading, writing and understanding levels on a 7-point Likert scale (1 = no competence through 7 = high/native competence). Across the 7 measures, participants in their foreign language condition rated their foreign language skills at 5.34 (between 5 and 6 of the Likert scale (L2_English: M = 5.34), with speaking at 5.02, reading at 5.68, writing at 4.97 and understanding at 5.71. Participants were also asked to self-rate their level of English according to the Common European Framework. They could choose between B1 (independent user), B2 (advanced independent user), C1 (proficient user) and C2 (advanced proficient user). Participants who answered the questionnaire in the foreign language condition reported, on average, a level between B2 and C1 (mean_CFR: 2.59 (between b2 and c1; see Table 1). Only 235 participants completed this particular section of the questionnaire. 152 participants did not report this information.
Table 1. L2 descriptives table. Only participants in the foreign language condition (L2) are considered.
Most of the participants were born in Italy, with the largest group (n = 141) being born in Trentino Alto Adige, followed by Veneto (n = 71), Piedmont (n = 9), Friuli Venezia Giulia (n = 18), Lombardy (n = 41), Liguria (n = 2), Emilia Romagna (n = 15), Tuscany (n = 11), Marche (n = 2), Lazio (n = 4), Campania (n = 6), Apulia (n = 13), Sicily (n = 19), Abruzzo (n = 2), Basilicata (n = 3), Calabria (n = 1), Umbria (n = 1) and Sardinia (n = 3). A map of Italy with the participants’ regions is provided in Appendix B. Fifteen participants were born outside Italy (n = 2 in India, n = 4 in the United Kingdom, n = 2 in Romania, n = 1 in Sri Lanka, n = 1 in Germany, n = 1 in Austria, n = 1 in Mexico, n = 1 in Albania, n = 1 in Brazil and n = 1 in Bangladesh) and ten participants did not report the place of birth. All our participants currently live in Italy.
The study protocol was approved by the Ethics Committee of the University of Trento according to the principles expressed in the Declaration of Helsinki. We obtained the participants’ consent at the very first stage of our online questionnaire. The participants had to choose if they agreed or not by clicking on the chosen box before moving on to completing the questionnaire.
Materials
The participants were asked to complete the adapted version of the Tolerance for Ambiguity questionnaire (Herman et al., Reference Herman, Stevens, Bird, Mendenhall and Oddou2010 – see Appendix A for the full texts of the questionnaire in English and in Italian) that was used to investigate if multilingualism was linked to a higher tolerance of ambiguity (Dewaele & Li, Reference Dewaele and Li2013). Herman et al. (Reference Herman, Stevens, Bird, Mendenhall and Oddou2010) had developed a psychometrically sound measure of ToA and presented the Tolerance for Ambiguity Scale (TAS), which they describe as a “conceptually clear, internally consistent assessment tool” (p. 60). Theirs was a 12–item questionnaire with five-point Likert scales. They found four distinct dimensions which were labelled as follows: 1) valuing diverse others; 2) change; 3) challenging perspectives; and 4) unfamiliarity. One question was removed by Dewaele and Li, who have worked with the 11-item questionnaire with five-point Likert scales. They also made some minor stylistic adaptations. We used the new version of the questionnaire (Dewaele & Li, Reference Dewaele and Li2013) and analysed the results following Herman et al.'s categorization into four dimensions. The four dimensions together with the corresponding items are presented in Table 2 above.
Table 2. List of items sorted by dimension.
In the current study, the original materials were in English and then translated into Italian by highly proficient bilingual speakers. One independent judge controlled the translated version for consistency with the English version. Some questions had already been reverse coded in Herman et al.'s (2010) original manuscript (see Appendix A). Their numerical scoring scale ran in the opposite direction in order to check if respondents were giving consistent answers. The language versions were also closely matched for word count (English questionnaire = 163 words, Italian questionnaire = 129 words).
Procedure
Participants had to judge to what extent they agreed with the items of the ToA questionnaire by selecting the appropriate box of a 5-point Likert scale (1 = strongly disagree, 2 = disagree, 3 = neither agree nor disagree, 4 = agree, 5 = strongly agree). The presentation order of the 11 items of the Tolerance of Ambiguity questionnaire has been randomised. In each condition, participants received the questionnaire entirely written in one language: English or Italian. After completing the questionnaire, participants had to spend a few more minutes completing our informative questionnaire on their linguistic and cultural background. This final part was in Italian for all participants, as only participants with Italian as their first language were recruited.
Results
ANOVAs and a MANOVA were carried out to test if there were any differences in the language in which the questions were presented. Analysis of covariance (ANCOVA) was also used to analyse whether age, the number of years of study of the second language, L2 proficiency and the distance from home had an effect on ToA scores.
In order to explore if there was an effect of the language of the questionnaire on ToA, an ANOVA was carried out, with ToA as a dependent variable and the language of the questionnaire as a fixed factor. This indicated a significant effect of the language of the questionnaire overall, F(1, 385) = 5.239, p = 0.023, ηp2 = .013, with participants who completed the questionnaire in English (mean = 37.67, sd = 4.26) scoring higher than participants who completed the questionnaire in Italian (mean = 36.72, sd = 3.88; see Figure 1).
Figure 1. Sum of ToA scores, sorted by Language of the questionnaire. Vertical bars represent standard errors, while black dots represent the aggregated means.
To evaluate the effect of language on the four dimensions – that is, 1) valuing diverse others; 2) change; 3) challenging perspectives; and 4) unfamiliarity – a MANOVA was performed with questionnaire language as the between-subject factor and the four ToA dimensions as dependent variables.
MANOVA revealed a main effect of questionnaire language, F(1,385) = 4.934, p < 0.001. This indicates that language selectively influences the four different dimensions (see Figure 2). In order to examine whether the foreign language effect is present throughout all dimensions, four one-way ANOVAs were carried out, one for each dimension. Whereas the main effect of the dimensions valuing diverse others, and unfamiliarity were not significant (Fs < 1), the effect of the dimension challenging perspectives was significant, F(1,385) = 5.964, p = 0.015, ηp2 = .015, with participants who completed the English questionnaire scoring higher on ToA (mean = 4.25, sd = 0.62) than those who completed the questionnaire in Italian (mean = 4.1, sd = 0.61). Change was also significant, F = 6.040, p = 0.014, ηp2 = .015, with participants who completed the English questionnaire scoring higher on ToA (mean = 2.93, sd = 0.70) than those who completed the questionnaire in Italian (mean = 2.72, sd = 0.66). This shows that participants who responded to the item of the dimension in English were more tolerant to ambiguity than those who responded to the Italian version.
Figure 2. Mean of ToA scores for each one of the four dimensions, sorted by Language of questionnaires. Vertical bars represent standard errors, while black dots represent the aggregated means.
The role of age, gender, years of L2 acquisition/exposure, L2 proficiency and geographical distance
In two ANCOVAs we have examined whether the years of L2 acquisition/exposure and L2 proficiency had an effect on ToA. 19 participants were excluded from these analyses because they did not provide the relevant linguistic information in the questionnaire.
A one-way ANCOVA was conducted with the language of the questionnaire (English vs Italian) as an independent variable, the participants’ overall level of ToA as dependent variable and with L2 proficiency as a covariate. L2 Proficiency was rated based on the answers provided by the participants in the linguistic assessment section: the higher the score obtained in L2 Proficiency, the higher the participant's proficiency. This indicated a significant effect of the language of the questionnaire, F(1,365) = 5.132, p = 0.024, ηp2 = 0.014, with L2 proficiency being a significant covariate, F(1,365) = 5.018, p = 0.026, ηp2 = 0.014, showing that the more proficient participants were in the foreign language, the more tolerant to ambiguity they were (see Figure 3).
Figure 3. Sum of ToA scores sorted by L2 proficiency, in both the English (ENG) and Italian (ITA) condition. Each dot corresponds to the ToA score of at least one participant. Grey and brown regions indicate the confidence interval (95%).
However, from a different ANCOVA, the variable number years of L2 acquisition/ exposure did not reveal any significant effect (F = 1.1, p = 0.3).
In a third one-way ANCOVA, we have examined whether age had an effect on ToA. 41 participants were excluded because they did not report their age. We have set Language of questionnaire as an independent variable, ToA as dependent variable and Age as a covariate. There was a significant effect of the language of the questionnaire, F(1,343) = .5.471, p = 0.020, ηp2 = 0.016. Age also resulted in a significant covariate, F(1,343) = 5.577, p = 0.019, ηp2 = 0.016, showing that older participants were less tolerant to ambiguity if compared to younger participants. To further analyse this relation, linear regression analysis was conducted for each of the two questionnaire conditions (English and Italian). This revealed that Age was a reliable predictor of ToA (β = -0.18, p = 0.22), but only for the participants who completed the questionnaire in their second language (English). No relation was found for participants who completed the questionnaire in their first language, Italian (p = 0.26) (see Figure 4).
Figure 4. Sum of ToA scores sorted by Age, in both the English (ENG) and Italian (ITA) condition. Each dot corresponds to the ToA score of at least one participant. Grey and brown regions indicate the confidence interval (95%).
We then performed a two-way ANOVA in order to examine if there were any gender-related differences on ToA. 18 participants were excluded from this analysis (10 = non binary, 1 = other, 7 participants did not provide this information), as their number was too small to be consistent. We have set Gender and Language of the questionnaire as independent variables and ToA as dependent variable. The main effect of gender resulted to be significant, F(1,365) = 5.013, p = 0.026, ηp2 = 0.013, with female participants being more tolerant to ambiguity (mean = 37.422, sd = 4.088) if compared to male participants (mean = 36.50, sd = 3.92). Moreover, Gender significantly interacted with Language of the questionnaire, F(1,365) = 5.761, p = 0.017, ηp2 = 0.015. Bonferroni corrected post-hoc tests showed that female participants who answered the English questionnaire displayed higher ToA scores than both female participants with an Italian questionnaire, t = 3.059, p = 0.014, and male participants with an English questionnaire, t = 3.229, p = 0.008 (see Figure 5).
Figure 5. Sum of ToA scores, sorted by Language of the questionnaire and Gender. Vertical bars represent standard errors, while black dots represent the aggregated means.
We have also explored the role of geographical distance on tolerance of ambiguity, but this resulted to be non-significant (F = 2.5, p = 0.1).
Discussion
This study addressed the question whether the use of a second language (L2) has an effect on tolerance of ambiguity (ToA), a personality trait in which an individual perceives ambiguous situations or stimuli as desirable, challenging and interesting (Furnham & Ribchester, Reference Furnham and Ribchester1995). Specifically, it was hypothesised that processing information in L2 promotes higher levels of ToA in bilingual speakers. Three hundred and eighty seven speakers of Italian and English completed an online questionnaire measuring ToA. The questionnaire, originally available in English (Herman et al., Reference Herman, Stevens, Bird, Mendenhall and Oddou2010), was translated into Italian as well. Two hundred and seven of them were randomly assigned to the Italian questionnaire and one hundred and eighty to the questionnaire in English. Participants had to judge to what extent they agreed with the items of the ToA questionnaire by selecting the appropriate box of a 5-point Likert scale. We analysed the data following Herman et al.'s (2010) categorization into four dimensions: 1) Valuing diverse others, 2) Change, 3) Challenging perspectives and 4) Unfamiliarity.
The results provided evidence that the use of a foreign language influences tolerance of ambiguity. In line with prior research and with our predictions, this study demonstrated that the use of a foreign language increases ToA overall.
When first categorising the four different factors that we ourselves have explored in the present article, Herman et al. have provided conceptualization in order to understand how and why each of the four factors might or might not be linked to bilingualism and multilingualism. Keeping in mind his analysis as a starting point, we have tried to think about why there could be a foreign language effect on our four factors. As to the factor labelled as valuing diverse others, this is consistent with interpersonal interaction, choosing situations where people share or don't share the same values and being comfortable with all kinds of people. It might reflect a person's inclination to appreciate and respect individuals from various backgrounds, cultures, and with different values or perspectives. A foreign language effect might be possible, as when using a second language participants are more likely to feel at ease with people who are different to them. As to the dimension labelled as change, it “reflects the dynamic nature of intercultural situations such as cross-cultural transitions and global management.” (Herman et al., Reference Herman, Stevens, Bird, Mendenhall and Oddou2010). A FLE is likely to be present here, as people might be more prone to go out of their comfort zone and follow effective coping strategies in their foreign language, when following their rationality. As to unfamiliarity, as new challenges “unfreeze and change prior mental models” (Hollenbeck & McCall, Reference Hollenbeck and McCall2002; Oddou & Mendenhall, Reference Oddou and Mendenhall1984), this could easily be more true in the foreign language, as part of the foreign language effect. Participants in their second language might be more prone to choose a situation that is not that familiar to them. This can be linked and be applied to the factor challenging perspectives, where people will tend to be more prone to engage in something that is different to them to a certain extent, if that perspective is presented in the foreign language.
When analysing the four different factors separately, our results provided evidence of increased ToA in the dimension of challenging perspectives and change. However, the effect was not statistically reliable when unfamiliarity and valuing diverse others were analysed. The foreign language effect was robust in the factors challenging perspectives and change. This confirms our hypothesis that participants who had had to answer the questionnaire in their foreign language would have been more likely to be tolerant to ambiguity if compared to the participants who had completed the questionnaire in their first language in situations where it is necessary to undertake effective coping strategies and engage in decisions that might lead to an evolution or a shift from the status quo. In the factors valuing diverse others and unfamiliarity, no significant effect of the foreign language on tolerance of ambiguity was found. One possible explanation could be that these items, that were very clear in Italian (L1), might have been more challenging to understand in L2, as the sentences were a bit longer or contained difficult constructions or double negatives, that might have created some confusion to our participants. This possible result may be affected by subtle differences in translation, that we have included among the possible limitations of our study (page 20).
Some evidence suggests that variables, such as language proficiency, may affect this reduction in emotionality associated with a foreign-language use (Caldwell-Harris, Reference Caldwell-Harris2014). On this basis, we decided to set some variables as covariates and investigate any possible outcomes.
The current study found a significant relationship between age and ToA. Dewaele and Li (Reference Dewaele and Li2013) reported a strong positive relationship between participants’ age and their ToA scores (Dewaele & Li, Reference Dewaele and Li2013) and a similar correlation was found between multilinguals’ age and scores on Trait Emotional Intelligence (Dewaele et al., Reference Dewaele, Petrides and Furnham2008). From these studies, it seems that with age, people might become more tolerant to ambiguity. Contrary to this, the current study found an inverse correlation. It seemed that older participants were, the less tolerant to ambiguity they resulted to be, with younger participants being more tolerant to ambiguity. This was particularly true when the questionnaire was presented in the participants’ foreign language. It needs to be said, as a possible limitation for this study, that participants who volunteered to take part in our experiment were on average quite young. Also, we unfortunately do not have any information about age for 41 participants. It may be useful to replicate this study with a higher number of older participants.
The relationship between ToA and the participants’ self reported proficiency in the second language was also analysed and it showed that participants who were more proficient in their L2 were significantly more tolerant to ambiguity. This is not in line with previous research, where high self-rated language proficiency, a possible indicator of more frequent language use (as suggested by Luk & Bialystok, Reference Luk and Bialystok2013) has been seen to facilitate stronger emotional experience in the foreign language (Caldwell-Harris et al., Reference Caldwell-Harris, Tong, Lung and Poo2011; Dewaele et al., Reference Dewaele, Petrides and Furnham2008), leading to more deontological choices. This finding is worth investigating at a deeper level in future studies, possibly having the participants complete a placement test in the foreign language in order to objectively assess their actual level of proficiency in that language. As proficiency has been self reported, we cannot be sure of the participants’ response.
When analysing the relationship between ToA and the participants’ gender, we found that female participants were more tolerant to ambiguity if compared to male participants. Moreover, Gender significantly interacted with Language of the questionnaire, showing that female participants who answered the English questionnaire displayed higher ToA scores than both female participants with an Italian questionnaire and male participants with an English questionnaire. This is a very interesting result that should be explored in the future in more detail.
Although our study provides valuable insights into the relationship between FLE and ToA, it is important to acknowledge several limitations, that will need to be addressed in future studies. First, the ToA Questionnaire has not been validated in the sense of demonstrating that it can predict behavioural outcomes and not just correlate with other self-ratings of this trait or logically related other traits. Second, it would be useful to have more homogeneous groups of male and female participants. Third, as a recent study by Paap et al. (Reference Paap, Anders-Jefferson, Balakrishnan and Majoubi2023) has shown, a scale's predictive validity and factor structure can be influenced by subtle changes in wording (e.g., translation). In the current study, participants who have completed the questionnaire in English have been presented with the same questionnaire that had been used in the previous study (Herman et al., Reference Herman, Stevens, Bird, Mendenhall and Oddou2010). For this reason, we have decided, in the Italian version, to conduct an analysis maintaining the same structure of the original paper (Herman et al., Reference Herman, Stevens, Bird, Mendenhall and Oddou2010). For future studies, it might be worth investigating whether translations may change the original factor loading structure.
Conclusions
The present research investigated for the first time the foreign language effect on tolerance of ambiguity. It offered new evidence that processing information in a foreign language does influence a personality trait like tolerance of ambiguity. We had the most significant results in sentences that were related to challenging perspectives and change, suggesting that people are more likely to be tolerant to ambiguity in situations that are linked to those two dimensions. Older participants resulted to be less tolerant to ambiguity if compared to younger participants, especially in the second language. Female participants were more tolerant to ambiguity if compared to male participants and the more proficient participants were in their second language, the more tolerant to ambiguity they were.
This study provided novel data of how processing information in L2 may change the perception of the world and affect decisions. These novel data are, in our opinion, very relevant today more than ever, given the amount of people who make decisions in a foreign language every day, often dealing with delicate subjects, where tolerance of ambiguity is desirable, if not required. This issue could have serious socioeconomic implications, especially in our globalised world, where decisions are taken on a daily basis in international, multilingual settings.
Acknowledgements
Many thanks to all participants who took part in the study.
Appendix A
Appendix B
Map of Italy with regions of interest highlighted in grey.
Open access
