Skip to content
Language, power and voice in monitoring, evaluation, accountability and learning: a checklist for practitioners

Insufficient attention to language barriers systematically excludes many marginalised groups[1] from decision-making, essential services and monitoring, evaluation, accountability and learning (MEAL) frameworks. Displaced people who do not speak or understand the majority languages used in their host communities are less likely to be able to communicate their own needs and priorities effectively. More generally, they are less likely to obtain the information they need to access services and make decisions, or to report abuse. Unless humanitarian practitioners are sensitive to the impact of language on power and voice when designing and implementing MEAL systems and analysing the resulting data, these problems will persist.

CLEAR Global’s work in forced displacement contexts in Asia, Africa and Europe provides insights into potential pitfalls and how to avoid them. Below we summarise these as a checklist that MEAL practitioners can use to minimise the risks of language-related distortion and exclusion in their efforts to listen to displaced individuals.

Survey design

We can better understand people’s needs if we design appropriate, accessible surveys for them.

  • Is the language clear and simple? Do the questions avoid jargon and abbreviations?

By using plain language, designers of MEAL tools can increase the likelihood that both enumerators carrying out surveys and respondents understand the questions in the way they are intended to be understood. Our comprehension testing with enumerators in northeast Nigeria found that commonly used abbreviations, technical terms and certain other terms were not widely understood without an explanation.[2]

  • Does the survey focus on the needs and interests of the affected population?

A short, clear, contextualised survey that allows respondents to express their needs and views is more likely to lead to programming that is responsive to the affected population. It is also likely to produce better quality data, as data quality depends on the active participation of both enumerators and respondents.

  • Do you know what languages affected people speak?

With limited understanding of what languages affected populations speak and their preferred means of communication, agencies may find it difficult to plan adequately for effective data collection. This essential background information can be gathered as part of initial programme design. General language and communication data on certain contexts of forced displacement are available from Multi-Sector Needs Assessments (MSNAs) and from census results mapped by CLEAR Global.[3]

  • Did you include language preference questions?

Including language questions as standard in MEAL tools can provide valuable data to improve future data collection and programming. If a school collects data on which language pupils speak at home, for example, the school can then provide support for those being educated in a second language. Language questions can also be used to identify groups that data collection may have missed and to adapt tools to enable such groups to express their views.[4]

  • Are the tools translated into the right languages?

Enumerators in multilingual contexts face significant challenges in managing translation in their work. Translating questions into the relevant languages beforehand reduces pressure to ‘sight translate’ – where the enumerator has to translate questions on the spot – during data collection. As such, it can increase consistency and free up enumerators to focus on accurately recording the answers. If that is unfeasible or if the enumerators prefer an English text, a glossary of terminology specific to the sector or organisation can be helpful.

  • Have you field-tested comprehension?

Testing comprehension of MEAL tools with a sample of community members helps correct for information distortion or loss during translation. For example, words like ‘stigmatisation’ and ‘trauma’ may not have direct equivalents in other languages and can be difficult to explain. Moreover, conservative communities may use euphemisms to refer to sensitive concepts such as sexual violence, using words like ‘dishonour’ or ‘stain’ instead.[5] Failing to use culturally appropriate and easily understood terms increases the risk that data about people’s views and experiences is not recorded.

Role of enumerators

MEAL data is better if the enumerators are trusted and use the languages that respondents are most comfortable speaking.

  • Do the enumerators speak those languages? Did you ask?

High linguistic diversity among displaced populations could mean local enumerators may not be able to meet the language needs of all respondents. Similarly, host communities may speak different languages from those of the displaced population. Enumerators who only speak majority languages and lack adequate support and resources to manage multilingual data collection may be inclined to avoid interviewing people who speak minority languages. This results in data that is unrepresentative of marginalised sections of the community.

  • Are you accounting for power dynamics in your selection of enumerators?

Involving affected people in data collection and service provision provides a range of benefits. First, they are more familiar with the cultural aspects of the languages being used, and more likely to understand nuances and euphemisms. Second, respondents may be more likely to disclose opinions (including those that may be seen as socially undesirable, such as being dissatisfied with aid, when they know and trust the enumerator. Organisations working in the Rohingya response in Bangladesh have shown that involving affected populations in data collection “can help build trust and strengthen comprehension, resulting in more nuanced data that most accurately represents the needs and experiences of affected communities”.[6] It is worth bearing in mind that an external enumerator may be preferred for heavily stigmatised topics.

  • Is your group of enumerators sufficiently diverse, including in gender and language skills?

This is particularly important in communities where it would be inappropriate for male enumerators to speak with women in private, for example. An enumerator with a disability may also be better placed to engage with and understand the perspectives of other persons with disabilities in the community. Failing to account for this could lead to the exclusion of certain perspectives from your data.

Language support for enumerators

  • Have you given enumerators access to vetted, trained interpreters for any community languages that they do not speak?

This can help prevent people from being excluded or misunderstood because they do not speak the dominant language, and help reduce reliance on family members and neighbours who are not trained interpreters. When discussing topics like sexual exploitation and abuse, it may be better to have an enumerator and interpreter from outside the community in order to protect privacy.

  • Can the enumerators ask questions and get clarifications?

Ideally, enumerators should be able to speak with designers of MEAL tools to resolve any confusion regarding the questions before using the data collection tools. This is challenging when designers roll out pre-approved tools from headquarters-level and the same set of questions is used in multiple contexts for cost-effectiveness and to obtain comparable data across contexts. In such cases, organisations should ensure that an experienced staff member is available to answer questions and encourage enumerators to raise any issues they foresee.

  • Do the enumerators have terminology resources?

Enumerators are rarely professional translators. Relying on them to translate questions and answers can lead to mistranslations and inconsistency, resulting in inaccurate data. Glossaries and pre-recorded questions can help prevent misunderstandings.[7] Either way, testing the enumerators’ comprehension of both the questions and the answer options is essential for accurate data collection and takes only 5-10 minutes, depending on the number of words assessed. For example, if people understand ‘rape’ to apply only to women, or if the enumerator only translates it in that way when posing a question, then sexual violence against men and boys is even less likely to be reported.[8]

Language technology

  • Can you record, transcribe and translate at least a sample of the interviews?

Ideally, all survey interviews would be recorded, transcribed and translated. This would not only improve quality assurance but also complement survey data with rich qualitative narratives and quotes. Translating and transcribing recordings requires significant investment, however, especially for under-resourced languages. But organisations can take steps to increase the likelihood that the data they receive matches the respondents’ answers. Recording all interviews and transcribing a sample of them for spot checks is feasible, especially for languages for which automated transcription and translation tools exist and produce high quality results.

Follow-up and analysis

  • Have you planned validation meetings?

Results and analyses are seldom translated back into the languages spoken by affected populations. Affected populations therefore have no opportunity to correct any mistakes or contribute their perspectives on how to incorporate findings from MEAL activities into programming. Validation workshops with affected communities could help you identify and address misunderstandings and increase accountability to affected populations.

  • Do you disaggregate and analyse data by language?

While disaggregating data by age and gender has become common practice, the same is not true of language. Disaggregating data by language can enable organisations to identify and support marginalised groups. In a 2021 MSNA for Somalia conducted by REACH with analysis from CLEAR Global, for example, almost all respondents using Somali Sign Language said they do not feel that they can influence site-level decisions. Equipped with this information, organisations can now take steps to address communication barriers for site residents with hearing impairments.

Conclusion

Improvement is not only possible; it is happening. There is a growing awareness of the ways in which language and communication issues affect who is heard and who can access services.[9] As more practitioners take this on board and try new approaches, we continue to learn as a sector about how we can make language an enabler of inclusion. Checking practice against the simple questions above can be an important part of that process.

 

Daniel Davies dnledvs@gmail.com @Daniel_E_Davies

Former Senior Advocacy Officer, CLEAR Global

 

Emily Elderfield emily.elderfield@clearglobal.org

Advocacy Officer, CLEAR Global

 

[1] While marginalised language speakers are the most prominently affected by insufficient attention to language barriers, so are speakers of dominant languages with low or no access to education, people with disabilities that affect how they can communicate in any language, people who speak a dominant language but do not understand technical or unfamiliar vocabulary, and people who face communication barriers due to social discrimination.

[2] In 2018, Translators Without Borders (now CLEAR Global) found that just 1 in 24 enumerators in northeast Nigeria could explain what ‘extremism’ meant, and 78% could not explain ‘stigmatisation’. Translators without Borders (2018) The Words Between Us: How well do enumerators understand the terminology used in humanitarian surveys? A study from Northeast Nigeria  https://bit.ly/enumerator-comprehension

[3] See Translators Without Borders Language Data by Country https://translatorswithoutborders.org/language-data-by-country/

[4] See Translators without Borders Language Questions in Humanitarian Data Collection https://bit.ly/language-questions and Translators without Borders (July 2021) Five easy steps to integrate language data into humanitarian and development programs https://bit.ly/language-data-guide

[5] Translators without Borders (March 2019) Rohingya Language Guidance: Building a better dialogue around gender issues https://bit.ly/Rohingya-language-gender

[6] For example, see ACAPS, IOM (April 2021) Our Thoughts: Rohingya Share Their Experiences and Recommendations  https://bit.ly/Rohingya-experiences-recommendations. See also Ground Truth Solutions (May 2021) For Rohingya, trust begins with who is asking the questions https://bit.ly/ethnicity-interviewer-effects

[7] See Translators without Borders TWB Glossaries https://translatorswithoutborders.org/twb-glossaries/ 

[8] Resource & Support Hub (2021) How to consider language when researching Sexual Exploitation, Abuse and Sexual Harassment (SEAH)  https://bit.ly/language-SEAH

[9] Kemp, E. (2018) Language and the Guiding Principles, Forced Migration Review issue 59  https://bit.ly/language-guiding-principles

DONATESUBSCRIBE