03 · What You Need to Know
How to Select or Adapt a Research Measure for Another Language
Why Is Language a Measurement Issue Rather Than Just a Translation Problem?
Questionnaires depend on respondents interpreting words, instructions, examples, and response options in ways consistent with the intended construct.
When an instrument is translated, apparently equivalent words may carry different meanings, levels of formality, emotional associations, or cultural assumptions.
Consider a questionnaire assessing academic self-efficacy. An item referring to speaking confidently during classroom discussions may not represent the same educational experience in a context where participation is primarily asynchronous, written, or individually assessed.
A grammatically accurate translation may therefore remain conceptually inappropriate.
Language proficiency also matters. Participants who can communicate in the original language may nevertheless find it difficult to interpret abstract psychological constructs, nuanced response categories, or technical terminology.
The relevant question is whether respondents can understand the instrument as intended, not merely whether they can recognize its vocabulary.
Should You First Look for an Existing Translation?
Yes. Before beginning a new translation, investigate whether an appropriate language version has already been developed.
Search the instrument developer's website, original publication, supplementary materials, official instrument repository, and subsequent measurement studies. Where necessary, contact the developer or authorized distributor.
Determine whether the translation is an official version, an independently developed research adaptation, or an informal translation reproduced in another publication.
These distinctions matter because the existence of a translated questionnaire does not establish that its development was methodologically adequate or that its use is authorized.
Review the documentation supporting the translation. Ideally, it should explain how language decisions were made, whether members of the target population participated in reviewing the items, and what measurement evidence is available.
Also verify whether the translated version is appropriate for your particular language variety. A questionnaire translated for one national or regional population may not be equally comprehensible elsewhere, even when the language name is shared.
What Options Are Available When No Suitable Translation Exists?
Several approaches may be defensible, depending on the research objectives and available resources.
| Option |
When It May Be Appropriate |
What Requires Attention |
| Locate another established language version |
A documented version exists for the intended population or a closely related context |
Permissions, language variety, adaptation quality, and relevant measurement evidence |
| Select another suitable instrument |
A different measure adequately represents the intended construct in the study language |
Construct coverage, psychometric evidence, and comparability with earlier research |
| Translate and culturally adapt the preferred instrument |
The original measure is appropriate and sufficient resources exist for a defensible adaptation |
Translation procedures, cultural relevance, participant comprehension, and further evaluation |
| Use the original language version |
Participants have sufficient proficiency to understand the instrument as intended |
Language proficiency, accessibility, selection effects, and response interpretation |
| Reconsider the measurement strategy |
No available or feasible instrument adequately supports the intended inference |
Consequences for the research question, design, and scope of conclusions |
These alternatives are not necessarily equivalent. A different instrument may be linguistically appropriate but assess a narrower construct. A translated version may preserve more of the original content but require additional evaluation before its scores can be interpreted confidently.
The decision should follow the same principles used to determine whether a measure is appropriate for the study population and context.
What Is the Difference Between Translation and Cultural Adaptation?
Translation concerns expressing the instrument's content in another language while preserving its intended meaning.
Cultural adaptation examines whether the concepts, examples, references, situations, and response categories remain relevant and understandable in the target context.
Linguistic translation
Producing wording in the target language that communicates the intended meaning of the original instructions, items, and response options.
Cultural adaptation
Evaluating and, where justified, modifying culturally dependent content so that the instrument remains meaningful and suitable for the target population.
These processes overlap. A translation decision may require cultural knowledge, while a cultural adaptation may require changes to wording.
The objective is not necessarily literal correspondence between every word. It is to preserve the intended construct and support an appropriate interpretation of responses.
Nevertheless, cultural adaptation must be approached cautiously. Changing examples or activities may improve relevance while also altering the construct represented by an item.
What Does a Defensible Translation and Adaptation Process Involve?
The ISPOR Task Force's principles of good practice describe a structured approach to translating and culturally adapting patient-reported outcome measures. Although developed for that domain, the framework offers useful methodological considerations for questionnaire research more broadly.
A suitable process may include the following activities:
-
Preparation: Clarify the construct, obtain necessary permissions, and identify the target language and population.
-
Forward translation: Produce one or more translations from the source language into the target language, preferably with appropriately qualified translators.
-
Reconciliation: Compare translation alternatives and resolve differences through documented discussion.
-
Back-translation when appropriate: Translate the proposed target-language version back into the source language to help identify possible discrepancies.
-
Expert review: Examine linguistic meaning, construct alignment, and cultural relevance with suitable expertise.
-
Target-population review: Investigate how intended respondents understand the items and response options.
-
Revision and finalization: Resolve identified problems, document decisions, and prepare the version for further measurement evaluation.
This is a general workflow rather than a universally mandatory sequence. The appropriate procedures depend on the instrument, disciplinary conventions, intended use, and resources.
Importantly, the quality of the adaptation depends on the reasoning and evidence generated during the process, not merely on whether each procedural step was completed.
Is Back-Translation Necessary?
Back-translation is commonly used as a quality-control procedure. It can reveal differences between the original instrument and a translated version that deserve further investigation.
However, back-translation is not proof of equivalence.
A translation can return to wording similar to the original while still sounding unnatural or carrying an unintended meaning for target-language respondents. Conversely, an appropriate culturally sensitive translation may produce a different literal back-translation because the target language expresses the concept differently.
Back-translation should therefore be interpreted alongside expert judgment, conceptual review, and evidence from the intended respondents.
The goal is to identify and resolve meaningful differences, not to achieve word-for-word symmetry at the expense of comprehensibility.
Why Should Members of the Target Population Review the Translation?
Language experts can evaluate grammar and meaning, but they may not anticipate how every intended respondent will interpret an item.
Cognitive interviewing or cognitive debriefing can help identify misunderstandings, ambiguous expressions, unfamiliar terminology, and problems with response categories.
For example, a respondent may understand an item grammatically but interpret "academic engagement" as attendance rather than active participation in learning activities.
Asking respondents to explain what they believe a question means, how they selected an answer, and whether the response options reflect their experiences can reveal problems that ordinary proofreading may miss.
Participants involved in this review should reasonably represent the language proficiency, educational background, and relevant characteristics of the intended study population.
A small qualitative review can identify important comprehension problems, but it does not establish the instrument's complete psychometric performance.
Does Translation Require New Reliability and Validity Evidence?
Translation may change the response processes and measurement properties of an instrument. Consequently, evidence established for the original language version should not automatically be assumed to apply unchanged to the translated version.
The extent of additional evaluation depends on the intended interpretation and the existing evidence.
Relevant investigations may include content validity, response processes, structural validity, reliability, and relationships with other variables.
If the study involves comparisons between language groups, evidence concerning cross-cultural validity or measurement invariance may also be necessary.
For example, comparing latent mean scores across language groups generally raises stronger equivalence requirements than using a translated questionnaire descriptively within a single population.
Researchers should distinguish translation quality from psychometric evaluation. A carefully translated instrument may still lack sufficient evidence to support particular statistical comparisons.
Can You Administer the Original Instrument to Bilingual Participants?
Possibly. If the intended participants can understand the original instrument sufficiently well, administering it without translation may be defensible.
However, conversational fluency does not necessarily establish comprehension of technical or abstract questionnaire language.
Researchers should consider whether differences in language proficiency could affect responses or systematically exclude participants with lower proficiency.
In multilingual settings, respondents may also differ in the language they ordinarily use for academic, professional, or personal experiences. These differences can influence how naturally particular concepts are understood.
Using the original language version should therefore be justified through participant characteristics and the instrument's linguistic demands, not simply through the assumption that bilingual respondents can answer any questionnaire accurately.
What If Translation and Adaptation Are Beyond the Research Budget?
Formal adaptation can require translators, subject specialists, target-population participants, and additional data collection. These demands may be substantial.
When resources are insufficient, investigate whether an alternative measure is already available in the target language and has suitable measurement evidence.
If no appropriate alternative exists, consider whether the study can be redesigned, the outcome narrowed, or additional resources obtained.
The same reasoning applies when measurement cost or participant burden makes the preferred instrument impractical.
Unexamined machine translation or informal translation should not be treated as an adequate substitute for evidence that respondents understand the items and that the resulting scores support the intended interpretation.
What Should You Document When Using a Translated Instrument?
Your methodology should identify the original instrument, its source language, the target language, the translation's origin, and the version administered.
When you conduct the adaptation, describe the translators' relevant qualifications, the reconciliation process, expert and participant review, substantive modifications, and the available measurement evidence.
If you use an existing translation, identify its source and explain why it is suitable for your study population.
Also report any limitations that could affect interpretation, particularly when evidence of linguistic or measurement equivalence remains incomplete.
This documentation allows readers to evaluate the measurement process rather than accepting the label "translated and validated" without knowing what was actually done.
Watch Out
A translated questionnaire should not be described as validated solely because it was reviewed by bilingual experts, back-translated successfully, or produced a satisfactory Cronbach's alpha. These procedures provide different kinds of evidence, none of which independently establishes validity for every intended score interpretation.
