Building Bridges Between Computational Methods and Human Translation: An English to Brazilian Portuguese Application of Machine Translation in the Cross-Cultural Adaptation of Psychological and Health-Related Assessments.
The present study evaluated the effectiveness of machine translation (MT) in both forward (English to Brazilian Portuguese) and backward translation (Brazilian Portuguese to English) of psychological and health-related assessments. The quality of the translations was assessed using the COMET (Crossl...
| Published in: | Journal of Cross-Cultural Psychology Vol. 57; no. 5; pp. 829 - 846 |
|---|---|
| Main Authors: | , , , |
| Format: | Article |
| Published: |
Sage Publications Inc.
Jul2026
|
| Subjects: | |
| Online Access: | View this record in EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ssf&AN=194489190&site=ehost-live header: @attributes: shortDbName: ssf uiTerm: 194489190 longDbName: Social Sciences Full Text (H.W. Wilson) uiTag: AN controlInfo: bkinfo: jinfo: jid: 00220221 CCP jtl: Journal of Cross-Cultural Psychology issn: 00220221 maglogo: Y pubinfo: dt: Jul2026 vid: 57 iid: 5 pid: 344 pub: Sage Publications Inc. artinfo: ui: 194489190 10.1177/00220221261418605 ppf: 829 ppct: 17 formats: tig: atl: Building Bridges Between Computational Methods and Human Translation: An English to Brazilian Portuguese Application of Machine Translation in the Cross-Cultural Adaptation of Psychological and Health-Related Assessments. aug: au: Albuquerque, Maicon Rodrigues Souza, Renan Pedra de Miranda, Débora Marques de Romano-Silva, Marco Aurélio affil: Neurosciences of Physical Activity and Sports Research Group, Universidade Federal de Minas Gerais, UFMG, Brazil Universidade Federal de Minas Gerais, Laboratório de Biologia Integrativa, Grupo de Pesquisa em Bioestatística e Epidemiologia Molecular, Brazil Instituto René Rachou, Fundação Oswaldo Cruz, Minas Gerais, Brazil Departamento de Pediatria, Universidade Federal de Minas Gerais, Minas Gerais, Brazil Departamento de Psiquiatria, Universidade Federal de Medicina de Minas Gerais, Brazil su: Brazil Self-evaluation Health status indicators Compulsive behavior Artificial intelligence Parenting Quantitative research Impulsive personality Psychometrics Psychological tests Authority Activities of daily living Diagnosis of eating disorders Secondary analysis Research funding Smartphones High performance computing Questionnaires Research methodology evaluation Pilot projects Descriptive statistics Structural equation modeling Research methodology Research Deep learning Data analysis software Comparative studies sug: subj: Self-evaluation Health status indicators Compulsive behavior Artificial intelligence Parenting Quantitative research Impulsive personality Psychometrics Psychological tests Authority Activities of daily living Brazil Diagnosis of eating disorders Secondary analysis Research funding Smartphones High performance computing Questionnaires Research methodology evaluation Pilot projects Descriptive statistics Structural equation modeling Research methodology Research Deep learning Data analysis software Comparative studies keyword: computational approach cross-cultural validation quantitative assessment translation computational approach cross-cultural validation quantitative assessment translation ab: The present study evaluated the effectiveness of machine translation (MT) in both forward (English to Brazilian Portuguese) and backward translation (Brazilian Portuguese to English) of psychological and health-related assessments. The quality of the translations was assessed using the COMET (Crosslingual Optimized Metric for Evaluation of Translation) metric, and statistical modeling was performed using Generalized Estimating Equations (GEE). In forward translation, COMET scores from DeepL (β = 0.0020, p = 0.667), OpenAI (β = 0.0041, p = 0.256), and Widn.AI (β = 0.0027, p = 0.505) showed no statistically significant differences from human outputs, whereas, Azure (β = −0.0143, p = 0.024) showed statistically significant underperformance. W-ADL and SCOFF showed lower scores, often below the 0.940 threshold, suggesting greater cultural adaptation demands. In back-translation, DeepL (β = −0.000075, p = 0.965), OpenAI (β = −0.0002, p = 0.883), and Widn.AI (β = −0.0047, p = 0.227) matched human performance, but Azure again underperformed (β = −0.0103, p = 0.013). Lower COMET scores were observed for SPAI, PSDQ, BIS-11, W-ADL, and SCOFF compared to DII (all p < 0.05). Despite this, the overall quality of back-translation remained high. Overall, COMET appears to be a robust metric for evaluating semantic fidelity in both directions, particularly in back-translation when the original version serves as a reference. These results support the integration of MT into cross-cultural adaptation approaches, suggesting that this approach is not a replacement but a supportive tool that enhances translation efficiency while maintaining the indispensable role of human expert judgment. pubtype: Academic Journal doctype: Article src: R language: English refInfo: copyright: @attributes: flag: N holdings: @attributes: islocal: N |
|---|