03 · What You Need to Know
How Translation Can Remove Cultural Distinctions From Research Evidence
Cultural Concepts Do Not Always Have Direct Equivalents
Languages provide different ways of expressing social relationships, experiences, values, and institutional practices.
Some concepts have established translations that communicate their intended meanings effectively. Others overlap with familiar expressions without matching them completely.
For example, the Filipino expression "utang na loob" may involve gratitude, reciprocity, interpersonal responsibility, or perceived obligation, depending on how it is used.
Translating it simply as "gratitude" may omit relevant relational expectations. Translating it as "debt" may introduce an overly transactional interpretation.
Neither expression should be assumed to capture every use of the concept.
Researchers should examine what a participant means in the particular context rather than treating a cultural term as though it has one fixed interpretation shared by every speaker.
Translation Can Collapse Different Concepts Into One Category
One form of meaning loss occurs when several distinct source-language expressions are translated using the same target-language word.
For example, expressions concerning social harmony, personal obligation, respect, and interpersonal accommodation might all be represented as "social responsibility."
This may simplify analysis, but it can also remove differences in how participants understand their relationships with others.
Conceptual compression
Different source-language meanings are represented through a narrower or more general target-language category.
Conceptual preservation
The translation or accompanying explanation retains distinctions necessary to understand the original meanings.
Conceptual compression is not always inappropriate. Researchers routinely use broader analytical categories. The problem arises when the broader category obscures differences that are relevant to the research question or is presented as a direct equivalent without justification.
Fluent Translation Can Conceal Cultural Meaning Loss
A translated sentence may read naturally while omitting important contextual information.
For example, an expression describing hesitation to challenge a teacher may be translated as "lack of confidence."
Although confidence could be relevant, the original statement might concern respect for authority, classroom norms, fear of public mistakes, or a combination of factors.
Replacing the expression with a generalized psychological category can change the explanation implied by the participant's account.
Generative AI may produce such substitutions when asked to make translations clear, concise, or academically polished.
Researchers should therefore distinguish readability from fidelity to culturally situated meaning.
Social Relationships May Be Encoded Differently Across Languages
Some languages use honorifics, pronouns, particles, or other linguistic features to communicate social relationships.
These features may express respect, familiarity, distance, or expectations about interaction.
When translated into a language without directly corresponding forms, some relational information may disappear.
For example, a participant's choice of address when discussing a teacher or supervisor might contribute to an interpretation of authority relationships.
However, researchers should avoid assigning cultural significance to every polite expression. The analytical importance of a linguistic feature depends on the participant's context and the study's objectives.
Where relational meaning matters, the original expression may need to be retained or explained.
Idioms and Metaphors Can Carry Locally Specific Knowledge
Participants sometimes communicate experiences through idioms, metaphors, and expressions grounded in local practices.
A translator may replace an unfamiliar metaphor with a conventional target-language expression that communicates the general message.
For ordinary communication, this may be appropriate.
In qualitative research, however, the original metaphor may reveal how participants conceptualize an experience.
For example, a participant's description of academic pressure through a locally familiar image may contain distinctions that disappear when translated simply as "stress."
Researchers should consider whether the figurative expression itself contributes to the analysis before replacing it with an abstract category.
Translation Can Change the Emotional Intensity of a Response
Expressions describing discomfort, fear, embarrassment, frustration, or obligation may not correspond neatly across languages.
A translation may make a statement sound more intense or less intense than the original.
For example, a participant who reports occasional hesitation may be represented as experiencing persistent anxiety.
Such changes can influence qualitative codes and the conclusions researchers draw about participants' experiences.
Researchers should examine whether the translation preserves qualifications, frequency expressions, and the scope of the participant's claim.
Emotional interpretations should remain grounded in what participants actually communicated.
AI May Favor Dominant English-Language Categories
Generative AI systems may interpret unfamiliar expressions through concepts that are prominent in widely available English-language material.
In educational research, this may involve categories such as engagement, motivation, autonomy, or self-efficacy.
These concepts can be analytically useful when justified by the study's framework.
However, they should not automatically replace local concepts whose meanings differ.
For example, a participant's account of family responsibility should not necessarily be reduced to individual motivation simply because the latter is a familiar educational psychology construct.
The possibility of English-language conceptual assumptions influencing non-English research deserves attention when AI translates or summarizes culturally situated material.
Meaning Loss Can Influence Qualitative Coding
When researchers code translated transcripts, they may assign labels based on the translated wording rather than the original expression.
If several distinct concepts have been translated into one term, the resulting codes may conceal differences among participants.
For example, responses involving social obligation, gratitude, and interpersonal accommodation might all receive the code "family support."
The code may capture a broad relationship but overlook how participants understand that relationship.
Researchers should consider whether coding should occur before or after translation when culturally specific meanings are central to the research question.
Source-language coding may help preserve distinctions, although it also requires appropriate linguistic and methodological expertise.
Translation Can Produce Artificial Similarities Across Groups
Cross-cultural researchers often compare patterns identified in different language groups.
If translation replaces distinct expressions with identical labels, the analysis may suggest greater similarity than the original material supports.
For example, several language groups might appear to share a theme called "respect for teachers."
However, the source responses may refer to different forms of interaction, institutional expectations, or interpersonal relationships.
Researchers should examine whether the common theme represents a defensible conceptual relationship or merely a shared translated label.
The challenge of comparing qualitative themes across languages requires attention to original-language evidence and context.
Translation Can Also Create Artificial Differences
Meaning loss does not always make experiences appear more similar.
Different translations of conceptually related expressions may create distinctions that are not supported by the source material.
For example, one language group's responses may be translated using terminology associated with anxiety, while another group's similar responses are described as hesitation.
Researchers may then infer psychological differences between groups that partly reflect translation choices.
Both artificial similarity and artificial difference can undermine cross-language interpretation.
Researchers should evaluate the conceptual relationships among source expressions rather than relying exclusively on translated terminology.
Cultural Distinctions Matter in Quantitative Instruments Too
The problem is not confined to qualitative research.
Questionnaires and assessment instruments may contain terms whose meanings depend on cultural or institutional context.
For example, an item about students independently challenging a teacher's explanation may be interpreted differently across educational settings.
Translation can alter the meaning or intensity of the item, while cultural differences may influence response processes even when the wording is linguistically accurate.
Researchers should distinguish translation quality from measurement equivalence.
The possibility of measurement nonequivalence in AI-translated instruments may require cognitive interviewing, expert review, and appropriate psychometric evaluation.
Retaining Original Terms Can Preserve Important Meaning
When no concise translation adequately represents a culturally significant concept, researchers may retain the source-language expression.
The term can be accompanied by a contextual definition explaining how it is used in the study.
This approach may be preferable to repeatedly substituting a familiar but incomplete English equivalent.
However, retaining original terms should not become a decorative practice that makes local concepts appear exotic or mysterious.
The decision should serve analytical clarity and accurately represent participants' meanings.
Researchers should also explain important variations in usage rather than presenting the term as culturally uniform.
Human Review Should Include Contextual Expertise
Bilingual proficiency is valuable, but culturally consequential translation may require additional contextual knowledge.
A reviewer who understands both languages may still be unfamiliar with the particular community, institution, or social practice being described.
Researchers should therefore consider reviewers with appropriate linguistic, disciplinary, and contextual expertise.
Where ethically appropriate, consultation with community collaborators may help evaluate interpretations.
However, no reviewer should automatically be treated as the sole authority on an entire culture.
The objective is to examine alternative meanings and preserve relevant distinctions rather than impose one supposedly definitive cultural interpretation.
Translation Decisions Should Be Documented
When culturally specific expressions influence findings, researchers should record how they were translated and interpreted.
Useful records include source-language expressions, proposed equivalents, contextual notes, reviewer comments, and reasons for selecting particular translations.
Researchers should also preserve relevant original quotations where ethically permissible.
If AI contributed to the translation, the methodology should distinguish AI-generated wording from human-approved interpretations.
Transparent documentation of translation decisions helps readers understand how culturally situated evidence was represented in the final research report.
