Translation protocol

The translation of the Balance Recovery Confidence Scale (BRC) or the Multi-dimensional Falls Efficacy Scale (MdFES) will follow internationally recognised guidance for cross-cultural adaptation of patient-reported outcome measures (PROMs), including Beaton et al. (2000) and the WHOQOL Translation Methodology. The process will ensure semantic, conceptual, and cultural equivalence through forward and backward translation, expert review, and cognitive debriefing with target users.

Source version: The original English version is the starting point. Translate all items from the full version before deriving any short form (if applicable), to preserve content validity.

Example Workflow for Translating the BRC

  1. Forward translation
    • The English version of the BRC will be translated from English into the local language by at least two translators independently. These translators need to be (a) native speakers of the local language, and be (b) familiar with the concepts of falls efficacy and balance-related confidence measures.
  2. Reconciliation meeting
    • A first consensus meeting of the translators is held which has to result in a provisional local version.
  3. Cognitive debriefing
    • Each translator recruits two individuals (from the target population) for a try-out of the BRC in written form. The questionnaire should be completed by the individuals independently. Afterwards, the 19 BRC items are discussed between the translator and the older person (Were all items clear? Is it necessary to reformulate items?).
  4. Refinement
    • Each of the translators may adjust the wording of items.
  5. Second reconciliation meeting
    • A second consensus meeting is held to create consensus about a next preliminary local version of the BRC.
  6. Back translation
    • A back translation from the local language into English is done by a professional translator whose native language is English.
  7. Third reconciliation meeting
    • A third consensus meeting of the translators is held to review the back translation. Important for the reviewing is the intentional meaning of the back translation, not the literal meaning. The objective is a valid translation of the local version of the BRC, not a new English one. If necessary, the professional back translator will be consulted for additional information.

Please email shawn.soh@singaporetech.edu.sg with the final local version and brief documentation so the community of practice can remain informed and updated.

Example Workflow for Use of AI Translating the BRC

There is emerging use of artificial intelligence (AI) to assist translation of PROMs as discussed in recent literature (e.g., Kunst et al., 2023). AI models can accelerate the translation process and enhance consistency across items, but human oversight remains essential to ensure conceptual, semantic, and cultural equivalence.

Source version: The original English version is the starting point. Translate all items from the full version before deriving any short form (if applicable), to preserve content validity.

An adapted approach using Forward-Back Machine Translation + Human Quality Check

  1. Preparation
    • Select at least two suitable AI translation model (e.g., ChatGPT, Copilot, or domain-trained neural machine translation systems). Provide the model with clear instructions, including: The instrument’s conceptual background (e.g., falls efficacy and balance-related confidence measures) using the latest references
    • Provide prompt guidance on maintaining rating-scale consistency (e.g., phrasing of response anchors). Contextual notes for technical terms or culturally sensitive expressions.
  2. AI-Assisted Forward translation
    • The English version of the BRC will be translated from English into the local language by at least two genAI model.
    • Input each item, instruction, and response option separately into the AI system.
    • Request natural, culturally appropriate phrasing rather than literal translation.
    • Generate two or more AI-assisted translations using different models or prompts to compare nuances.
    • Document all versions and note any model-generated alternatives or ambiguous outputs.
  3. Human Reconciliation meeting
    • A first consensus meeting of at least two translators is held to compare versions which has to result in a provisional local version.
    • Experts ensure that:
      • Each item preserves the intended construct meaning.
      • Linguistic choices are appropriate for the target older-adult population.
      • Terminology aligns with local clinical or cultural norms.
  4. Cognitive debriefing
    • Each translator recruits two individuals (from the target population) for a try-out of the translated scale in written form. The questionnaire should be completed by the individuals independently. Afterwards, the translated scale’s items are discussed between the translator and the older person (Were all items clear? Is it necessary to reformulate items?).
  5. Refinement
    • Each of the translators may adjust the wording of items.
  6. Second reconciliation meeting
    • A second consensus meeting is held to create consensus about a next preliminary local version of the scale.
  7. Back translation
    • A back translation from the local language into English is done by a different AI translation model (e.g., ChatGPT, Copilot, or domain-trained neural machine translation systems), which was not used in the forward translation process.
  8. Third reconciliation meeting
    • A third consensus meeting of the translators is held to review the back translation. Important for the reviewing is the intentional meaning of the back translation, not the literal meaning. The objective is a valid translation of the local version of the scale, not a new English one. If necessary, a professional translator will be consulted for additional information.

Note that AI-generated text must always undergo human verification to prevent semantic drift or cultural misalignment.

Please email shawn.soh@singaporetech.edu.sg with the final local version and brief documentation so the community of practice can remain informed and updated.

A appended documentation report will be helpful for future work:

  • AI model(s) used and the prompting instructions.
  • Any issues highlighted during human reviewer meetings.
  • Summary of cognitive debriefing findings.
  • Any known limitations or recommended local adaptations.