Use of a language intervention to reduce vaccine hesitancy

Vaccine hesitancy is a major global challenge facing COVID-19 immunization programs. Its main source is low public trust in the safety and effectiveness of the vaccine. In a preregistered experimental study, we investigated how using a foreign language when communicating COVID-19 vaccine information influences vaccine acceptance. Hong Kong Chinese residents (N = 611) received COVID-19 vaccine information either in their native Chinese or in English. English increased trust in the safety and effectiveness of the vaccine and, as a result, reduced vaccine hesitancy. This indicates that language can impact vaccine attitudes and demonstrate the potential of language interventions for a low cost, actionable strategy to curtail vaccine hesitancy amongst bilingual populations. Language interventions could contribute towards achieving the United Nations Sustainable Development Goal of health and well-being.


Scientific Reports
| (2022) 12:253 | https://doi.org/10.1038/s41598-021-04249-w www.nature.com/scientificreports/ recycled wastewater, when these were described in a foreign language rather than in their native tongue 17 . The foreign language promoted more positive feelings towards these products, which resulted in higher acceptance.
Other studies have also shown that how people feel about novel products is driven by differences in trust 18,19 . For example, enhanced social trust increased positive feelings and decreased negative feelings towards the novel avian flu vaccine, which, in turn, increased intentions to get the vaccine 18 . Given that trust and feelings are closely related concepts 19 , and that people particularly rely on trust when judging things that are novel to them 20 , these results raise the possibility that communicating COVID-19 vaccine information in a foreign language might reduce mistrust in the vaccine and, therefore, decrease vaccine hesitancy.
Hong Kong Chinese bilinguals provided an ideal opportunity to test this theory for a couple of reasons. First, COVID-19 vaccine hesitancy was relatively high in Hong Kong compared to other countries 10,12,21 . Second, Chinese and English are official languages in Hong Kong and, hence, many government and healthcare resources are readily available in both languages. Therefore, the language manipulation represents an actionable intervention to reduce COVID-19 vaccine hesitancy in Hong Kong.
To study the impact of foreign language use, we provided COVID-19 vaccine information to unvaccinated Hong Kong residents and randomly assigned them to receive the information either in their native Chinese or in English. We then asked them whether they intend to get the vaccine and how much they trust the vaccine, among other questions.

Results
The data, analysis script and materials are publicly available online on Open Science Framework. The study design and number of participants was preregistered on www. AsPre dicted. org.
Descriptive data of vaccine hesitancy. Dovetailing with existing research, we found a high degree of COVID-19 vaccine hesitancy. Out of the 611 participants, only 36.0% (220) said they plan to get vaccinated, 45.2% (276) indicated that they were unsure, and 18.8% (115) indicated that they would not get vaccinated. Most importantly, vaccine hesitancy depended on language as we predicted. The use of English reduced vaccine hesitancy, with more people saying they intend to get vaccinated in the English (39.9%) than in the Chinese (32.5%) condition, and fewer saying they are unsure in English (41.2%) than in Chinese (48.8%). Language did not impact the rate of outright refusal ("No": English: 18.9%, Chinese: 18.8%). In sum, the foreign language English helped turn hesitancy into acceptance (see Fig. 1).

Predicting vaccine hesitancy by language.
Our main interest was to examine whether language affects vaccine hesitancy. Hence, we grouped responses into ones that indicated no hesitancy (0 = Yes) and ones that indicated hesitancy or refusal (1 = Unsure, 1 = No). A binary logistic regression was conducted examining the dichotomous COVID-19 vaccine hesitancy variable as a function of language (0 = Chinese, 1 = English), gender (0 = female, 1 = male), age, education, and general health. The logistic regression model was statistically significant, χ 2 (5) = 27.98, P < 0.001. The model correctly classified 62.4% of the cases. As anticipated, language accounted for a significant proportion of variance in vaccine hesitancy (B = 0.41, Wald = 5.43, P = 0.020, odds ratio = 0.67). Participants reading the COVID-19 vaccine information in English were less hesitant about getting the vaccine (Mean = 0.60) than were those reading the same information in their native Chinese (Mean = 0.68). Gender also accounted for significant variance in vaccine hesitancy (B = 0.54, Wald = 9.19, P = 0.002, odds ratio = 1.

Discussion
Vaccine hesitancy presents a major barrier to improving health and well-being around the globe. We investigated how the language used to communicate COVID-19 vaccine information influences vaccine hesitancy. We provide evidence that the use of a foreign language increased trust in the safety and effectiveness of the vaccine compared to identical information communicated in the native language. In turn, the higher trust associated with the foreign language reduced COVID-19 vaccine hesitancy. These findings suggest that feelings of trust when making health decisions depend not only on the content of health information but also on the nature of the language used to communicate it.
Studies have shown that foreign language use can influence judgment and decision making in different domains including risk taking 22 and morality 23 . Here we demonstrate that language can also influence an extremely consequential health decision, namely whether to get vaccinated during a pandemic. Research has further suggested that language is a powerful social cue that can influence trust 24 . We showed that the language used in communications can influence trust and, as a result, the decision to vaccinate.
Because English was a foreign language for our participants, they might have experienced more disfluency when comprehending the information in English than in Chinese. This could have influenced perceived trust in the vaccine and consequently willingness to be vaccinated. However, this account would predict the opposite of what we found. Disfluency decreases trust rather than increases it 25 . For example, in the "trust game" players show lower trust in people with disfluent names than in those with fluent names 26 . Hence, if English communication is more disfluent, then it should have prompted lower trust in the vaccine and thus increased vaccine hesitancy. However, the opposite was true.
The impact of language on vaccine hesitancy should depend on how it affects trust. Here we showed that when the native language context is associated with relatively low trust in the vaccine, the use of a foreign language increases trust and, therefore, reduces vaccine hesitancy. But in situations where the native language is associated with higher trust than the foreign language, we would expect the opposite. For example, consider the case of first generation immigrant communities such as Arab immigrants in Europe. For such communities, trust in the information that is provided in their native tongue, Arabic, might be higher than in information provided in the local language. In such cases, communications through the foreign language would be predicted to lead to lower trust in the vaccine, thereby increasing vaccine hesitancy. In this sense, language interventions should consider local conditions by understanding how each language impacts trust.
Other determinants of trust associated with a given language could be further explored in the service of reducing vaccine hesitancy. For example, it is possible that when one language of bilinguals has a higher status, people will trust the information provided in it more. In Hong Kong, English does not have higher status than Cantonese 27 , which suggests that the effect we found is not a function of differential language status. Yet in situations where one language has a higher status than the other, the higher status language may increase trust in the information thereby reducing hesitancy.
Limitations. This study examined a particular population, and a specific native-foreign language combination. It would be important to further investigate the generalizability of the current findings in different populations and with different native-foreign language combinations. Furthermore, the intervention that we identified only applies to bilingual populations. However, estimates show that more than half of the global population uses two or more languages in everyday life 28 , suggesting that this language intervention could be widely actionable. www.nature.com/scientificreports/ In monolingual populations, other language interventions could be explored, such as using a dialect towards which people have positive attitudes that might lead to higher trust 29 and reduced vaccine hesitancy.

Conclusion
We provide evidence for a low cost and actionable language intervention to reduce vaccine hesitancy amongst Hong Kong Chinese residents. Such language interventions can influence other health decisions and extend to other cultures. However, the selection of language should consider the local conditions. In cases where the native language context is associated with low trust, the use of a foreign language can enhance trust and reduce vaccine hesitancy. In cases where the foreign language is associated with low trust, the native language should be preferred. Public health campaigns therefore could use such language interventions strategically to boost vaccination uptake and other beneficial preventative behaviors such as cancer screening. Such language strategies can promote the United Nations Sustainable Development Goal 3 of "good health and well-being" 30 .

Methods
The study design, sample size and materials were preregistered on www. AsPre dicted. org. The data and study materials are available online in the Supplementary Materials and on https:// osf. io/ bdhvx/? view_ only= 02c42 3439e ac40a f8a9a 57c58 0bb05 88. All participants provided written informed consent prior to participation. All procedures were approved by the Social and Behavioral Sciences Institutional Review Board at the University of Chicago. All methods were carried out in accordance with the Declaration of Helsinki.

Materials and measures.
Participants read about why they should get the vaccine, how the vaccine works, and possible side effects of the vaccine in either their native Chinese or English adapted from the Hong Kong Department of Health (for the full descriptions see Table S1 in the Supplementary Materials available online). At the time of the study, Hong Kong residents could select the type of vaccine with which to be inoculated, so we did not mention a specific vaccine in the information 31 . We measured intention to vaccinate by asking participants, "If a vaccine that protects you from COVID-19 disease was available free of charge, would you get it?" (Yes, Unsure, No). We also asked participants to evaluate their trust in the effectiveness and safety of the vaccine, "Overall, how much do you trust that the COVID-19 vaccine will be effective?" and "Overall, how much do you trust that the COVID-19 vaccine will be safe?" (1 = Do not trust at all, 2 = Hardly trust, 3 = Trust a little, 4 = Mostly trust, 5 = Completely trust). Furthermore, we collected a number of exploratory measures of secondary interest (see Supplementary Information for the full set). In order to ensure that all participants had a sufficient proficiency in English to understand the materials, at the end of the study all participants, regardless of language condition, were asked to translate three key sentences from English to Chinese. Finally, participants were asked about their general health, age, gender, and education level.