Research with Spanish-Speaking Populations in the United States: Lost in the Translation A Commentary and a Plea

Abstract

Verbal material used to assess the cognitive abilities of Spanish-speakers in the the United States is frequently of linguistically unacceptable quality. The use of these materials in research settings is thought to pose a serious threat to test validity and hence to the validity of claimed results or conclusions. The authors explain how and why incorrect language finds its way into cognitive tests used in research and other settings and suggest solutions to this serious problem.

Notes

This is not to say that these practices generalize to all individuals who work with Spanish language instruments in this country. Indeed, the use of multiple consultants from a variety of Spanish speaking countries has become standard within the commercial testing industry during all phases of test translation, adaptation, and construction as well as phases of pilot work with new tests. There is in many instances evidence of healthy feedback from test users that doubtless has led to improvement not only in the correct use of the Spanish language but also in heightened psychometric sophistication.

Incidentally, both translated and adapted instruments can contain threats to validity with the introduction of bias. This bias has been described in detail by Van de Vijver and Hambleton (1996) and may occur at three levels: item, method, and construct. For details on these levels of bias the reader is referred to Artiola i Fortuny & Mullaney, 1997.

“Condominiums Saint Window”. Even if Canon Law permitted the canonization of a window, there remains an error of gender concordance.

“Panther Street” error in gender concordance.

“Green Eye Way” error in number concordance.

“Old Mission” errors in gender concordance, spelling, and accentuation.

This approach has been applied already. for example, in the Catie Project, an ongoing investigation of the neurocognitive aspects of schizophrenia, Keefe and his colleagues (2003) decided that, when possible, it would be important to test patients who use Spanish as their primary language. A manual for Spanish-speaking subjects in the United States was developed with assistance from an experienced bilingual neuropsychologist educated both in English and Spanish. Spanish language translations were executed by professional translators and back translated by different professional translators. The final version of the manual was reviewed by the neuropsychologist for language accuracy and equivalence to the source language, and proofed by at least two masters or doctoral level Spanish speakers. for those tests not available in Spanish, test instructions, materials, and procedures were translated into Spanish using the same procedure. Sites with testers who planned to test patients in Spanish were required to demonstrate testing competency in Spanish and English through background check, oral interview, and submission of a sample of written Spanish.

Santiestevan, S. (1991) Use of the Spanish Language in the United States: Trends, Challenges, and Opportunities. U.S. Department of Education, Office of Educational Research and Improvement. (ERIC Document Reproduction Service No. ED ED335176)