Research ethics in a multilingual world: A guide to reflecting on language decisions in all disciplines

1. Introduction

Academia is increasingly globalized, which is one of the reasons why academics increasingly discuss and write about ethical concerns in multilingual research settings.² In our daily life as researchers, we have also noticed a number of ethical dilemmas that are connected to language decisions we make while conducting research in multilingual contexts. As part of the European Network for Junior Researchers in the Field of Plurilingualism and Education (ENROPE), an Erasmus+ project in which all three authors participated, we discussed general challenges that novice researchers meet while working in the field of plurilingualism and education.³ Many of these challenges turned out to be ethical challenges, which resulted in interesting discussions and ultimately in this paper. Based on our own research experience, we have observed that Holmes et al.’s finding from 2013 still holds true, namely that “the complexities and possibilities of researching multilingually are not extensively covered in research training nor widely discussed in the research methods literature.”⁴

The literature⁵ suggests that we are not alone in thinking about language decisions at different stages in the research process. However, the ethical implications of such language decisions are not often discussed outside the field of language and cultural studies, and, according to Eaton’s work from 2020, they need to be addressed in more systematic ways.⁶ Based on regulatory documents from the UK, Australia, Canada, and the USA, Eaton found that “the guidance provided in official or regulatory documents from all four countries offers little explicit direction for research undertaken in additional languages.”⁷ She concludes that

Conducting research in additional languages is an aspect of research ethics that remains underdeveloped and merits more robust consideration among scholars, more explicit guidance from regulatory and research bodies, and more attention from those who serve on research ethics boards.⁸

Institutionalized codes or regulatory bodies are typically concerned with creating a safe space for the researchers and participants at all stages when conducting research, by doing no-harm⁹ or “minimizing harm.”¹⁰ In order to minimize harm, Von Köppen et al. and Kubanyiova emphasize researcher reflection as an important principle and invite researchers to reflect on ethics at all research stages.¹¹

Many authors believe that there are rarely straightforward ethical decisions to make in research¹² because ethical codes do not always follow the complexity and specificity of research in social sciences for instance.¹³ Instead, researchers often need to make informed decisions on a “case-by-case basis.”¹⁴

The contribution we make in this article is based on the premise that ethical codes and regulations alone cannot provide sufficient guidance for all possible research situations¹⁵ and that researcher reflection is an over-arching principle of conducting research ethically.¹⁶ Thus, the aim of this article is to work towards a framework for reflection that, on the one hand, complements ethical research regulations, and, on the other hand, facilitates systematic engagement with the nature and consequences associated with language decisions¹⁷ in multilingual research contexts.

For the purpose of this article, we take “languages” to mean standardized languages and other language varieties such as dialects or sociolects. In situations where languages, thus defined, come into contact with one another, language decisions and negotiations are required as they can either enable or hinder interactions between language groups. Such negotiations, which also occur in “multilingual research contexts,” are influenced by language status, norms and attitudes and are not value free as illustrated in this article.

Our work was guided by two questions. Firstly, we were interested in how multilingualism and research ethics intersect at different stages of the research process, i.e., what ethical problems arise at the different stages of doing research. Secondly, we wanted to develop an understanding of how language decisions can result in potential harm and how ethical concerns related to language decisions can be mitigated.

We will start this paper by exploring at what research stages ethical concerns occur more generally. This allows us to think through the need for language decisions at each stage. We then explain how we identified and analyzed relevant literature in section 3 that specifically considers language decisions, potential harm, and mitigation and led to the framework for reflection we present in section 4. This is followed by a discussion in section 5 that raises awareness of how ethical dilemmas related to language decisions can take on an organizational, social or ideological role. We close with recommendations for researchers and research development in all disciplines.

2. Research stages and ethical concerns

In the literature, a distinction is made between more general institutionalized codes of conduct, also referred to as procedural¹⁸ or professional research ethics¹⁹, on the one hand, and research ethics that concerns a particular project in a specific context, also referred to as particularistic²⁰ or practical research ethics²¹, on the other hand. As Johansen argues, particularistic ethical decisions made in the day-to-day research practice relate to individual research projects, while procedural research ethics refers to the institutionalized codes of conduct and ethical guidelines, which often pertain to research more generally.²²

Both institutionalized codes of conduct²³ and academic literature²⁴ emphasize that ethical questions arise at all research stages. The British Educational Research Association (BERA) for instance recommends that “at all stages of a project – from planning through conduct to reporting – educational researchers undertake wide consultation to identify relevant ethical issues, including listening to those in the research context/site(s), stakeholders and sponsors,”²⁵ while Cohen et al. consider research ethics to span the ethics of designing, planning, conducting, analyzing, and disseminating the research.²⁶ In addition, the Italian Consiglio Nazionale delle Ricerche (CNR), or National Research Council, offers a useful overview of all the actors who are involved in ethical decision making: researchers, research institutions, scientific community, funders, editors and publishers and policy makers.²⁷ These different actors also represent different stages of the research process.

Thus, we take as a starting point that ethical considerations are part of all research stages²⁸ and that many people are involved.²⁹ The institutional and scholarly literature frame relationships with participants as an ethical concern. The literature refers to the pre-data collection stage, at which informed consent is a recurring topic. This is about participants’ autonomy, right to privacy and self-determination.³⁰ CNR suggests that researchers should inform participants through an adequate communication process to make them understand the objectives, methods and potential risks, as well as guarantee the respect towards individual autonomy and information transparency.³¹ At the post-data collection stage, both BERA and Cohen et al. provide guidance on data management in terms of privacy and data security, anonymity, and confidentiality.³² Horner and Minifie indicate that the ownership of the data is an additional ethical concern at the post-data collection stage.³³ It is argued that an agreement should be reached between the researchers and the participants as to whether the data is co-owned or who has control over these.³⁴ In addition, many scholars also emphasize ethical considerations regarding the analysis and interpretation of data.³⁵ Regarding data dissemination, Lujić believes that researchers should carefully select in which journals and conferences results are presented or published, as findings will be shared with certain language communities but not others.³⁶ This means, as the European Commission points out, that research findings should share “the benefits with disadvantaged populations, especially if the research is being carried out in developing countries.”³⁷ Thus, there is a power dimension that we need to consider in our dissemination strategies.

In sum, institutionalized guidance and literature on research ethics more generally suggest that as researchers we need to deeply reflect on ethical concerns at all stages of the research process from conception to dissemination, on the many actors that are involved and the relationships that we develop. These guiding principles – which are not specific to language decisions – inform the next part of this article.

3. Critical transdisciplinary literature review

In this section we lay open how we identified and analyzed literature that helped us answer our research questions. It also makes transparent our positionalities and language biases.

We opted for a critical and transdisciplinary literature review, guided by Grant and Booth’s overview on review types, Jesson and Lacey’s suggestions on what a critical review should contain, and Montuori’s transdisciplinary approach.³⁸ Drawing on published and relevant literature with an interest at the cross-over between multilingualism and research ethics, our review is critical as it goes beyond description to include analysis and conceptual innovation that results in a model. As suggested by Montuori, we immersed ourselves in an “ecology of ideas,” took stock, deconstructed and reconstructed ideas from the field of language education, sociolinguistics and applied linguistics, as well as other disciplines, to identify and map significant literature that “brings information from separate disciplines together so that it can be useful knowledge that allows us to act wisely.”³⁹

The literature search started with all three authors searching for literature covering language decisions to be made at the different stages of the research process, language decisions related to research ethics, and research ethics at the different stages of the research process more generally. Our literature search has been multilingual: we included knowledge produced in several languages that we as a multilingual researcher team could access, which is also part of the ethical principles we put forward in this article. We are aware that we can only read a small number of languages, which inevitably produces a language-bias in a literature review.⁴⁰ Having said that, our search strategy enabled us to learn from diverse research communities, as we draw on publications in Chinese, Danish, English, French, German, Italian, Norwegian, Polish, Spanish, and Swedish (languages are alphabetically listed – not ranked). One way in which we have tried to reduce the language bias is by using a snowballing technique. By also looking at the references used in the literature that we have found, we could include relevant literature in other languages than the ones that we read and that we searched in (e.g., Danish and Polish). We have read these articles through our knowledge of similar languages based on the principle of intercomprehension, e.g., written Danish is very alike Norwegian and Swedish, or in the case of Polish with help of online translation software and peers who spoke the language in question.

We collected all the relevant literature in a table in a shared Excel document, where we added language information and English summaries and quotes relevant to the different topics that we identified. Based on the literature thus collected, we carried out our critical review by identifying the different stages of the research process, the language decisions to be made at each stage, and their ethical implications. This review, combined with our own experiences as multilingual researchers, has been the basis for the development of our framework.

Literature identified and analyzed

Our multilingual search strategy resulted in a number of relevant publications associated with the field of linguistics, applied linguistics and sociolinguistics⁴¹, or in related interdisciplinary studies.⁴² These have been published in Chinese, Danish, English, French, Norwegian and Spanish. Some studies can (also) be associated with the study of culture,⁴³ which were published in English and Spanish; as well as in psychology⁴⁴ and medicine,⁴⁵ published in English. It is important to note that ours is not a systematic review, and our study like others is limited by the languages we can read. However, we feel we have included relevant voices internationally that can help us develop a systematic and critical understanding.

It is crucial to include regulatory frameworks, as our article is designed to complement such guidance. Thus, we build on Eaton’s work, who provides a useful overview of regulatory ethical guidance from UK, Canada, USA, Australia.⁴⁶ To understand ethical concerns raised in such documents more widely, we build on this by including relevant documents from Italy, the European Union, and Germany. ⁴⁷ These documents regulate ethical behavior in education, social sciences and humanities, psychology as well as bioethics.

Potential biases

Our linguistically diverse backgrounds as Swiss, Dutch and Chinese nationals, working in UK and in Swedish universities at different stages in our careers, and our experiences of doing research multilingually with majority and minoritized language groups are visible in this paper, and may mitigate a European bias. We utilized our diverse linguistic repertoires to access literature in all languages that we can read and even beyond. Nonetheless, the only language that we all three have in common is English, so from a practical point of view English has been the working language between the three of us and this article is written in English which means this knowledge is not available in other languages. To reduce this bias, we follow Canagarajah’s example⁴⁸ and make available an abstract in several languages through IRIS.⁴⁹ Lastly, there is a visible disciplinary bias in our work, as publications from language-related disciplines dominate the literature included. The reason for this is not only that we are working in the field of applied linguistics, but also that researchers interested in multilingualism may be particularly sensitive to ethical implications of language decisions and mention these in their work.

4. Language decisions with ethical implications

We thematically organized the ethical concerns we identified in the literature, and added concerns based on our own experience as researchers and researcher developers. This analysis resulted in over-arching language-related ethical concerns that can occur at four stages of the research process. These overarching themes include: conception of research projects, working with participants, data analysis and interpretation, as well as dissemination and project management. The latter frames the entire research process. Each of these five overarching themes contain a number of sub-themes. In order to raise awareness of the concerns and potential risks of each of these sub-themes, we offer a definition and insights regarding potential consequences and mitigation for each one.

Conception of research projects

Some publications emphasize the ethical concerns related to the understanding of concepts across languages and cultures, and what languages we read when compiling a literature review.

Conceptual understanding:

Concepts operationalized in research can be hard to translate from a lexical and a cultural perspective, as outlined by translation specialists Okolie and Okoedion⁵⁰. While these authors do not explicitly refer to ethics, Nasri and Talib highlight “conceptual equivalence” across languages as an ethical issue.⁵¹ Potential harm: Words have meaning related to their cultural settings, concepts and knowledge. Such words, formulated by communities working in another language, may be misrepresented, distorted or lost in another language. Proposed mitigation: Any translational issues should be made transparent in the research project. Okolie and Okoedian list a number of strategies that can be used to do this, for instance by borrowing a word from the source text, and use it in its original form with an explanation. Here, intercultural awareness of the terminology and concepts in the different languages is required, making a case for working in multilingual and multicultural teams.

Conducting a literature review:

There is evidence that scientific research is indeed published in several languages,⁵² but publications in languages other than English are not always considered in literature reviews. In addition, languages considered in the search process are rarely made transparent. Potential harm: Using search terms in one language only is problematic, as it can lead to a language bias in our understanding.⁵³ If literature published in other languages is disregarded,⁵⁴ it can lead to “duplicated efforts and lost knowledge.”⁵⁵ It can also negatively affect the status of those authors who publish in languages other than English, as their work gets cited less⁵⁶. Proposed mitigation: In a literature review, researchers should make transparent the languages they use in their search processes and mention the languages of the literature that guided their study – or which languages were not considered. This ought to be routinely mentioned as a limitation of the study, as illustrated by our approach in section 3.

Assumptions held by research team

Professional socialization is an important factor in forming a person’s worldview and how they approach understanding.⁵⁷ Depending on their own language socialization, researchers develop different research interests and expertise. With this comes a particular worldview, potentially including a language bias, that guides the understanding of what is important, what research counts and what needs to be a research priority.⁵⁸ Potential harm: If researchers do not make transparent their language bias and positionalities, findings may be misrepresented as the only possible interpretation, whereas researchers with different linguistic backgrounds may ask different questions, invite different participants, and may thus develop different understandings. Proposed mitigation: Reflecting on, and making transparent, linguistic positionalities and backgrounds may indicate how the findings can be understood, and lead to more humble statements. In this article we make the linguistic bias and our positionalities transparent, which by implication means we can offer a partial understanding only.

Working with participants, creating a safe space

Research publications and regulatory frameworks tend to agree that there are ethical implications to language decisions around informed consent and communication with participants, but literature shows that there is more to consider when it comes to research projects in particular settings, namely contextual sensitivity.

Participant recruitment and data collection

This concerns communication with participants in preparation of and during data collection. The literature we consulted agrees that participants need to understand what their participation in the project implies,⁵⁹ including those participants with limited language proficiency or literacy levels.⁶⁰ Authors and regulatory guidance emphasize the need for accessible language (written or spoken) especially in terms of informed consent.⁶¹ According to the literature, informed consent is a crucial element in research ethics as this is the key approach to ensure the research process is transparent to the participants.⁶² Potential harm: Some participants – and their voice – may be excluded from research because they cannot understand the main language used at a university or by researchers. If participants are invited, but find it hard to understand the information provided, they may not be fully informed about the research process, which may make them vulnerable and exploitable. During data collection, participants may not be able to express themselves fully, unless they can use their strongest or preferred language, for instance in an interview.⁶³ Proposed mitigation: The literature tends to agree that communication should take place in an accessible language and format⁶⁴ that respects individuals’ autonomy⁶⁵ and self-determination.⁶⁶ From our experience, we know that some ethics committees require researchers to submit consent forms and project information in two languages, namely in the main university language for the benefit of the ethics committee and in the language and a format (oral and/or visual) the participants can comfortably access. Translation can be used to develop accessible consent forms and research instruments. This may require translators⁶⁷ and researchers who can use the participants’ preferred language.⁶⁸ It also requires the ethics committee to trust the researchers’ ability to adequately translate their forms. Piloting of instruments – which forms part of existing research routines in our experience – is important to maximize accessibility and understandability.⁶⁹

Contextual sensitivity:

We identified cultural sensitivity as an important ethical aspect that has a language angle. This includes respecting desired or imagined linguistic identities and different worldviews and realities that are associated with certain languages.⁷⁰ Potential Harm: Researchers can interfere with participants’ self-concepts and worldviews/realities that are associated with certain (indigenous) populations, and impose their understanding of individuals, communities and the world. Participants could become identifiable if their language background or accent is revealed in a publication, especially if theirs is unusual in the participant group. Proposed mitigation: Researchers ought to consider whether the languages used, for instance in an interview, reflect the proficiency of the participant or a desired linguistic identity, or both.⁷¹ We may need to be aware that several languages can be used during data collection, without keeping languages separately if this reflects the participants’ practice. In addition, our findings suggest that different languages can be related to different realities and philosophies that may be unknown outside certain communities, e.g., in indigenous populations in South America.⁷²

Data analysis and interpretation

The analysis of data in a bi- or multilingual research contexts is recognized as a challenge in the literature, and here we identified three areas of concern, related to translation, software and knowledge construction.

Translation as part of the data analysis

When working across languages, researchers regularly translate qualitative data either for analysis or to cite participants’ words in publications. However, all translation is a type of interpretation of what was said in the original language.⁷³ The topic of “translation of qualitative research data, such as interview transcripts, is largely absent from regulatory and guiding documents.”⁷⁴ Potential harm: Concerns occur with relation to translating data for analysis because translation is neither value-free not straightforward and subtleties or ambiguities may be lost in translation.⁷⁵ There is a concern of distorting or misinterpreting the participants voices and intended meaning across languages and cultures. This risk gains salience as machine translation is becoming readily available, which may open new possibilities but also harbors challenges of mistranslations.⁷⁶ Proposed mitigation: According to our analysis, researchers ought to develop greater awareness of challenges related to translating research data⁷⁷ and “use a variety of techniques to ensure accurate translation.”⁷⁸ It may be beneficial to work with bi-/multilingual researchers who are familiar with the research context to mitigate misrepresentation and distortion of participants’ voices.⁷⁹ This requires trust in the competences of bi- and multilingual research colleagues and brokers. Ideally, any participant citations ought to be presented in their original form and in the language of publication,⁸⁰ which may require journals to be more flexible with word counts, as this requires more space. In research reports any translation processes should be made transparent.⁸¹

Software availability

Based on our professional experience, we found that software that supports qualitative data analysis typically supports big European languages (NVivo), plus Arabic and Chinese (MaxQDA). However, for smaller languages with different scripts, there may be no adequate software support available, forcing researchers to work by hand. Potential harm: Some researchers may not consider research with certain language communities for lack of software support, and thus exclude communities from participation and knowledge production. Alternatively, it may disadvantage those researchers who do choose to work with minority language communities, as their work is not equally supported by technology. Proposed mitigation: This topic, which was not mentioned in the literature but emerged from our own experience and discussions, suggests that researchers need to collaborate with software providers and language communities to support a greater range of languages. Alternatively, costing for extra time and staff needs building into funding bids when working with languages that are not supported by traditional software. This is a systemic or organizational problem.

Knowledge construction

Translation and interpretation is not just an issue of respect towards participants, but researchers also construct knowledge based on their understanding of the world⁸². Translation of meaning “can extend to broader concerns related to basic beliefs and general worldview.”⁸³ Potential harm: There is a risk of misrepresentation, deculturization⁸⁴ and “epistemicide,” as knowledge can be lost when translated into dominant discourses.⁸⁵ Furthermore, translation of participant data may make communities and their languages invisible, which may further disempower for instance minority language communities.⁸⁶ Proposed mitigation: Based on our own experiences of working with minority language groups, researchers making transparent their own worldviews, assumptions and beliefs they bring to the interpretation of results may help indicate that there is more than one way of making sense of the world. Critical understandings of language issues, based on articles such as this one, may help create awareness of the language factor as a wider ethical concern that also has an ideological dimension.

Dissemination

Dissemination in languages accessible to participants seems to be a widely accepted ethical concern. However, there is an additional ethical conundrum related to academic publications.

Impact and stakeholder information

Eaton found that regulatory guidance tends to require researchers to communicate their research in a certain way.⁸⁷ As described by BERA, researchers should “communicate their findings, and the practical significance of their research, in a clear, straightforward fashion, and in language judged appropriate to the intended audience(s).”⁸⁸ Potential harm: If findings are not shared with participants in an understandable format, they might feel exploited, while they and communities concerned may not benefit from the results, which limits impact opportunities.⁸⁹ Proposed mitigation: It is good practice to share findings with participants and community stakeholder groups in accessible language and formats⁹⁰ and make them “relatable and usable for participants.”⁹¹ When disseminating research, presentations can be multilingual, e.g., the slides can be in another language than the verbal presentation. We have come across academic conferences in the field of language education where this was required or encouraged.

Academic Publications

Researchers need to decide in which language(s) they publish their work for an academic audience.⁹² Peer-reviewed academic articles can typically be published in one journal only, thus limiting the author to one language. Potential harm: If researchers decide to publish in English, their work is more likely to be seen, considered and cited by international colleagues, which is an important factor in career development, but the communities concerned may not be able to access the information. Vice versa, if researchers publish in a language accessible to stakeholder communities, it may affect their status and career as a researcher negatively, as many universities expect research to be published in international (English medium) journals.⁹³ In addition, the fact of researchers not writing in their strongest language may be disadvantageous, and their work may be more often rejected by publishers.⁹⁴ Proposed mitigation: Include an abstract in more than one language where possible. Abstracts in additional languages could be deposited in external directories, such as www.iris-database.org, as demonstrated by Canagarajah⁹⁵ and our own multilingual abstract⁹⁶. Articles in languages other than English can be made findable through multilingual directories, such as the ENROPE annotated bibliography.⁹⁷ Furthermore, stakeholder reports can be published in additional languages, alongside an academic article in English, thus creating multiple access options.

Project management

Through our discussions and review of the literature, we found that language decisions can present ideological challenges when obtaining ethical approval, in relation to funding and in terms of power relations between actors.

Ethics application process

Our research experience informs us that ethics committees typically work in one language, e.g., Swedish in Sweden, and English in the UK. This requires all researchers, including international colleagues, to submit their ethics documents in the main language used at the university, even if it is an international research project. Potential harm: International researchers may struggle writing and/or reading ethics forms in the locally used language, e.g., Swedish, and may be excluded from conducting research or depend on colleagues. In addition, any forms would need translating for international colleagues. Proposed mitigation: There may be practical reasons why ethics forms can be submitted in one language only. However, given that academics are required to use English in many contexts, ethics committees could consider accepting documentation in English and the main university language to make the process more inclusive and allow for international collaboration. Moreover, a linguistically diverse ethics committee may be able to process ethics documentation in more than one language, where this may be required. While this may not cover all researcher languages, it may help smooth the process for some international researchers, and value the language repertoire of international ethics committee members. While there may not be a straightforward answer, universities should be aware that international staff and students may need linguistic support when developing their ethics applications.

Cost and funding

Doing research multilingually means additional cost and time is required, as language brokers may be needed at different stages in the research process.⁹⁸ Potential Harm: Researchers who work multilingually may incur additional costs and thus may be disadvantaged in a competitive funding system, or they may need longer to develop research findings, thus slowing down career prospects. Proposed mitigation: It is recognized that costs of language brokering need to be considered in grant applications.⁹⁹ For example, some PhD scholarships provide extra funds for students to “acquire or develop a working ability with a difficult language in order to carry out fieldwork (including UK fieldwork) or other parts of their research; this is over and above the funded length of the studentship,”¹⁰⁰ which we deem to be good practice.

Power distribution between actors

There is a body of literature that is concerned with decolonizing research processes,¹⁰¹ for instance in situations of linguistic power differentials, such as may be found in international research teams, or in contexts of forced migration or indigenous language communities, as well as in relation to local and global crises that would require heightened ethical sensitivity. Language-based asymmetries can reflect power hierarchies, where one language may have greater status than another.¹⁰² In addition, researchers with high-level competences in a dominant research language, typically English, may have greater rhetorical strategies for effective communication¹⁰³ compared to those with other language backgrounds.¹⁰⁴ Potential harm: Actors in the research process, whose language has a lesser status, or whose competences in the dominant research language may be less developed, may feel intimidated and inferior to other actors, e.g., between members of a research team, between researcher and participant, between authors and publishers, etc. This may result in a feeling of inferiority, oppression or discrimination, especially in contexts of additional power asymmetries, such as (post-)colonial contexts. In addition, it may result in unequal access to publication channels for authors with less developed competence in the dominant research language.¹⁰⁵ Mitigation: Colleagues who have high-level competences in high-status languages, such as English, may need to become aware of their respective power. Intentional conversations related to power balance and awareness of privileges may be necessary in research teams¹⁰⁶. To reduce the power imbalance, members in a team could acknowledge and value the language competences and expertise of participants and colleagues and support them in their identity constructions and career development endeavors, e.g., by making them co-authors.¹⁰⁷ Participatory research can be a way to make the relationship between the research participants and the researcher(s) more equal.

5. Discussion and Conclusion

Our analysis revealed how language use and ethical challenges intersect in five major ways. Such ethical challenges can affect not only participants and stakeholder communities, but also individual authors, research team members and communities as there are power hierarchies organized along language dimensions. Thus, responsibilities to minimize harm lie with all actors mentioned in the CNR guidance: researchers, research institutions, scientific community, funders, editors and publishers, policy makers.¹⁰⁸ The framework we propose in section 4 offers ideas of how ethical concerns related to potential misrepresentation and deculturalization of knowledge and power asymmetries between actors in the research process could be mitigated.

We expect our framework to be relevant for the wider academic research community across disciplinary boundaries, and we invite colleagues working in market or industrial research, as well as in other initiatives to critique, expand and develop the framework we propose.

Our take on the literature – combined with our own research experiences – suggests that there are ethical dilemmas in which languages play an organizational (e.g., understanding, communication), social (e.g., developing trust, relationships, collaboration) or ideological role (e.g., power, justice, worldviews). We consider these dilemmas to be connected to both particularistic and procedural research ethics,¹⁰⁹ as the differences between these two are rather blurred and dynamic, which is also visible in our review. In the following we discuss the three dimensions in which languages and research ethics intersect.

Organizational dimension

Our results indicate that there are organizational factors that influence the language decisions made at the different stages of the research process. This is about what research design is selected, how access is gained to participants, how findings are generated and disseminated. Regulatory frameworks often require researchers to reflect and make transparent their language choices related to certain aspects.¹¹⁰ Literature and our professional experiences corroborate that regulatory frameworks and ethical committees often seek transparency over the languages that are used to communicate with participants, when gaining access to participant groups, when collecting data and when disseminating findings to stakeholder groups. Language decisions made at all other stages, however, are rarely mentioned in respective guidance.¹¹¹

Further language organizational aspects we identified consider the languages used in communication with ethics committees, as well as additional time and cost related to translation or brokering. This dimension is more obvious, whereas social and ideological dimensions tend to be less visible.

Social dimension

Based on our findings and experiences there are linguistic implications related to the ethics of collaboration and relationships with the multitude of actors involved in research, including researchers, research institutions, scientific community, funders, editors, brokers, publishers, and policy makers.¹¹² Participants as well as communities and other stakeholder groups are another important group of actors. Thus, the social dimension is about who the actors are in the research process, what roles they play and any underlying power relations or asymmetries that might exist. It is about how individuals and groups can be instrumental in supporting, informing, developing, expanding, interpreting, and cascading knowledge.

Our findings suggest that researchers need to reflect beyond the organizational dimension about how to create a safe space based on trust, as suggested by von Köppen et al. in which participants as well as members of the project team of different language backgrounds and expertise are included and respected.¹¹³ The generation of such a safe space may support the feeling of belonging and generate joint ownership of the project. Recognizing the social nature of language decisions as an ethical concern in multilingual research contexts might help mitigate the feeling of being exploited or disrespected. In projects where actors with multiple languages come together, it may be worth recognizing – and celebrating – linguistic expertise, to make diverse voices heard and use the linguistic resources wisely and constructively, which may contribute to creating a safe space where all research actors can belong regardless – or because – of their language background.

Ideological dimension

Our results indicate that linguistic decisions carry ideological underpinnings. Specifically, when it comes to decisions about what is worth knowing, which language groups and actors have relevant knowledge, what there is to understand, how this can be understood, and who needs to know. Languages are highly politically charged and viewed differently by different groups.¹¹⁴ This means there are power differentials between linguistic groups, between researchers and the researched, between more or less experienced researchers in a team, or between institutions, such as ethics committees and funding bodies and research teams. The ideological dimension is to a large part about the researcher stance, cultural sensitivity and reflexivity, as emphasized above all in research with indigenous language populations, for instance in South America.¹¹⁵ Being aware of the ideological nature of language decisions and how they intersect with ethical concerns, is about recognizing that the world can be understood in different ways, which in its turn will be influenced by a person’s language socialization and positionality. Some concepts cannot, therefore, be easily translated from one language to another, as certain languages are associated with certain worldviews or understanding the world, so mere translation might not suffice to make sense of data. Depending on what languages researchers read, what language groups they belong to, and who they invite as participants, they may come with different understandings of concepts and what we need to find out. Language socialization and exposure may also influence decisions taken in funding bodies, where some research may be prioritized over others.

Recommendations

Based on our findings and interpretation, we make some recommendations that complement the regulatory frameworks that form part of procedural research ethics. These recommendations concern what is called particularistic or context-sensitive research ethics and concerns individual research projects.

For researchers of all disciplines

Even though the present article was conceived by authors with an interest in multilingualism, sociolinguistics and applied linguistics, it invites researchers of all disciplines, and beyond, to reflect on language decisions and their ethical implications. It has been found that, for example, 36% of research papers on biodiversity are published in a language other than English,¹¹⁶ and it is argued that “a bias toward English-language science can result in preventable crises, duplicated efforts and lost knowledge.”¹¹⁷

We invite researchers to reflect on and be transparent about how languages are used at each of the stages in the research process and the ethical implications that come with it. When we look at our overview of all the language decisions that can be made at the different stages of the research process and their ethical implications, reflexivity and being open about positionality are recurring themes, as well as the choice of language of publication. The framework we presented in section 4 is designed to guide such reflection. When it comes to reflexive research ethics, it could also be worthful to have a look at indigenous research ethics, in which reflecting on researcher positionality and being open about it is central.¹¹⁸

For researcher development

We also make recommendations for researcher development and researcher education in- and outside of academia. We designed the framework (see section 4) with research training in various disciplines in mind. Thus, it can complement regulatory research ethics and requirements so that language decisions and their ethical implications can be considered. See also Holmes, Fay, Andrews & Attia¹¹⁹ for valuable guidance on researcher development. Moreover, we invite researcher developers and mentors to encourage novice, and more established researchers to enable a reflexive and transparent space to reflect on and discuss ethical implications related to language decisions in their own projects. Our article may serve as a basis for such discussions. Furthermore, we suggest that participatory research involving the research participants as co-researchers and co-authors in the research process is a way to reduce the power imbalance between the research actors.

Concluding statement

In this article, we have shown that the language decisions we all make as researchers are of significance beyond data collection and representation, as is often stated in regulatory frameworks. Thus, we contribute to an increasing body of literature that aims to understand the complex ways in which language decisions intersect with ethical implications, and how these occur in research processes. In particular, we provide an easy-to-understand framework that complements regulatory ethical guidance and can be used by research teams and in researcher development.

The framework we present in this article reflects our own positionality as researchers, insofar as it is based on a literature review conducted in the languages that we can read, our life and research experiences in the field of language education, as well as our joint reflections across language and cultural boundaries. The fact that we draw on literature from a limited number of languages could be seen as a limitation. At the same time, we feel it is also a strength that we considered several languages. In any case, we offer a partial understanding. Cognizant of this limitation, we now invite colleagues from various disciplines, and beyond, to critique, use, expand and adapt our framework, which we envisage as an accessible starting point to develop a deeper and more systematic understanding of ethical implications carried by language decisions required in research situations and projects.

Acknowledgements: Special thanks go to Adrianna Surmiak for inviting us to contribute to this special issue, to colleagues who participated in the ENROPE project and triggered our interest in this topic, as well as to colleagues and peer-reviewers who offered valuable feedback on previous versions of this article. Special thanks go to our multilingual research community who translated the abstracts into several languages (see Meier, van der Voet, Yan, 2023).