Beware of Bilingual Experts

In spite of their assumptions, most “bilingual experts” are not equal to the task of translating documents for medical research. Translation by “bilingual experts” are often not comparable to translations that were done by real translators.

Maria Cornelio, director of the Hispanic Research and Recruitment Center at Columbia University Medical Center in New York City, comes to the same conclusion in an article Published by the American Translators Association (ATA) Chronicle as I did.

Maria Cornelio is director of the Hispanic Research and Recruitment Center at Columbia University Medical Center in New York City, where she oversees recruitment and language-support programs for Spanish-speaking participants in clinical studies. She also serves as Spanish-language consultant to the New York State Psychiatric Institute and teaches English-to-Spanish “Translation in Healthcare” at New York University’s Center for Foreign Languages and Translation. She is a member of the Board of Advisors of the University of Chicago Translation Studies Program and frequently conducts continuing-education seminars for professional translators. She has a master’s degree in international studies from the University of Denver and a Dipl6me d ‘Etudes Francaises from the University of Poitiers, France. She studied at the University of Seville, Spain, and has a B.A. in Spanish and French from Hunter College of the City University of New York.

Abstract:
In spite of their assumptions, most “bilingual experts” are not equal to the task of translating documents for medical research.

(This article was originally published in the ATA Chronicle (the journal of the American Translators Association), August 2003, Vol. 32, No. 8.)

Beware of the ” Bilingual Expert”

By Maria Cornelio

It is generally recognized that the translation of medical documents is a very specialized activity. But it has been my experience that people in the medical field who are not translators believe the difficulty in translating such documents is due only to the medical and scientific terminology and the rigorous standards demanded of scientific research. As a result, a large number of physicians, healthcare workers, and scientists assume that any bilingual person among them (usually called a “bilingual expert”) can do a better job than a professional translator who does not also have an advanced health-science or technical degree.

An article in the April 2003 issue of the ATA Chronicle brought the point home in a humorous and compelling way. As I read Steve Vitek’s “Is Technical Translation Really a Collaborative Activity?” [1], I couldn’t help applying his observations to my own professional situation. Although we labor in different areas of specialization, Vitek and I seem to have come to the same conclusions about the dangers of relying on so-called “bilingual experts” to produce quality specialized translations. Generally, such “experts” are people who have trained in their respective professions and also happen to speak another language. However, unlike professional translators, these individuals did not acquire their second language by studying it in a systematic way. Even when they are native speakers and have received their technical training in the language of the proposed translation, these people may be “field experts,” but they are not necessarily “language experts.” And this makes all the difference in the world with respect to the quality of the translations they produce. To quote Vitek: “Most bilingual experts are not translator material if they lack education emphasizing linguistic skills and translating experience [2]. Unfortunately, many don’t seem to realize this and attempt to translate documents, all too often with disastrous results.

After reading Vitek’s account of his frustrating experiences with such “experts” translating patents, I thought of the many medical translations I’ve had to rescue after they were botched by physicians, microbiologists, nurses, and other “bilingual experts.” In fact, for several years now I’ve kept a (constantly growing) file with the most egregious examples.

Two colleagues and I have described in print the many problems I found with one of these documents, a questionnaire that was translated by Spanish-speaking health-care
professionals for a major research study [3]. The investigator who brought me the translation for review assumed that it would need very few changes, since she had confidence in the language abilities of the translators. However, the document contained so many errors that it had to be almost completely re-translated.

The vocabulary and syntax gave me the impression that I was reading English with Spanish words. Many statements would have made sense only to someone familiar enough with the English language to decipher the meaning behind what was said. For example, the phrase “did you attend support groups” was translated as “atendio grupos de soporte.” Many bilinguals often use “false cognates,” words that look alike but have different meanings in the two languages. Atendio is one such word. The Spanish meaning is “to be attentive, to heed, to pay attention. ” It does not have the English meaning “to go to” or “to be present at.”Soporte” is another false cognate. It comes from soportar, which means “to carry a load, suffer, or tolerate. ” Support in the positive sense in which it was being used in the questionnaire should have been translated as apoyo. Success was translated as “succeso” [4], a word that does not exist in Spanish. The respondent was sometimes addressed with the formal usted and other times with the informal tu. Frequently, there was no agreement between adjective and noun, or between verb and subject. Often, the questions were simply mistranslated. For example, “how many other pregnancies have you had?” was turned into “have you had other pregnancy losses?” “I get cross at my friends ” became ‘I have confrontations with my friends [5].

The translators were obviously unfamiliar with the proper wording for demographic questionnaires in Spanish, including not being aware that “marital status” is translated as “estado civil.” This last phrase is something they could have found in any good bilingual dictionary. But being native speakers, they assumed no equivalent term existed in Spanish or else they would have known it. Clearly, they did not feel the need to consult the dictionary. As a result, while the English had a section with the heading “marital status” the Spanish heading read, “Por favor de marcar. Estaba casado?” (“Please to mark. Were you married?”).

In short, the translation as it was would have been completely useless as a cross-cultural research instrument on psychosocial variables, even though it had been done by “bilingual experts”. As a professional translator, I was able to fix this particular translation so the study could be carried out successfully. I am happy to say that not only did the research make an important contribution to the field, but the investigator also obtained a Ph.D. for her work.

Even peer-reviewed medical journals sometimes unknowingly publish translation disasters that would have been simple enough to prevent if the translation had been entrusted to a professional translator. A case in point is an article published a few years ago in the Journal of Clinical Epidemiology [6], one of the most prestigious medical journals in the U.S.

This article was brought to my attention during the course of my work reviewing translations to be used in clinical trials. An investigator submitted for review the translation of a pain questionnaire that he wanted to use in his study. After looking over the questionnaire, I informed the study coordinator that I could not approve it because there were serious problems with the translation. She did not believe me and said there had to be something wrong with my method of evaluation, since this translation had been done and statistically validated by bilingual health-care professionals at the University of Texas. In addition, an article had been published describing the methodology that was used in the translation and validation of the questionnaire. She gave me the complete citation and I consulted the journal.

I found the article, which I proceeded to read with great care. The article looked impressive enough. According to the authors, their methods “were designed to assure cross-cultural equivalence …and to conform to the guidelines …proposed …for the cross- cultural adaptation of [health-related quality-of-life] instruments. Content equivalence was assured by having an expert panel of health care providers fluent in Spanish and knowledgeable about Mexican-American culture evaluate the relevance of each item…to the culture of Spanish-speaking Mexican­ Americans. Semantic equivalence was ensured by using a rigorous forward- and back-translation process …An integral part of the translation process was evaluation by an expert panel to assure comparability of item meanings across research instrument on psychosocial variables, even though it had been done by “bilingual experts” in the two 1anguage versions [7].

The article then describes the translation and validation methodology in great detail:

• One translation committee formed by “nine bilingual health researchers with extensive
experience studying Mexican Americans.” Five of these translators had Spanish as their first language. The other four knew Spanish and ”were…familiar with local Spanish usage [8].

• Two evaluation committees, one made up of “eight health-care providers,” six physicians, one nurse, and one social gerontologist. “Spanish was the first language for four of these individuals, but all were fluent and able to read and write in both languages, and had many years of clinical experience with Mexican-American patients. [9] The second evaluation committee was formed by 10 bilingual health-care consumers [10], all of whom were Hispanic (three had Spanish as their first language).

• “A total of five iterations of the translation process were accomplished, including a formal back-translation.” According to the article, great care was taken to preserve “the original structure of the [questionnaire]” and “for each of the [pain] descriptors in English, a Spanish equivalent was found that was considered to convey similar qualitative and quantitative dimensions of pain. [11]

• Once the translation and back­ translation were done, the “evaluation committee formed by health-care providers examined the appropriateness of the semantic content and
grammatical form of each Spanish descriptor in the translation [12].

• Finally, in order to establish its validity and reliability, the results subjected to a series of statistical analyses. These tests included Pearson correlation coefficients, chi-square, and paired t-tests. The results of these analyses are presented in a total of six scatter plots, two bar charts, two tables, and two graphs. [13]

In their conclusion, the authors assert that they have “provided evidence that the translation fulfills published criteria for cross-cultural equivalence” and that it “is suitable for studying Spanish-speaking Mexican Americans in South Texas, and probably in other locations in the Southwestern U.S. [14]

Seeing all this, how could I, a mere translator without a medical or scientific degree pass judgment on work produced to such exacting standards by “bilingual experts” who were M.D.s, Ph.D.s, and R.N.s? As a translator, I am interested in words and their meaning, the context in which those words are used, and whether they follow the logic of the language. Looking over the list of pain descriptors these researchers had produced, it was clear to me that in the real world very few Spanish-speaking patients, Mexican­ American or not, would understand such a questionnaire – let alone be able to give meaningful answers.
In addition to my language skills, I also have to be familiar with research methodology so that I can understand and evaluate the translations of the protocols that come across my desk for review. Despite its complicated statistics, tables and charts, the study violated one of the basic principles of research design. This principle states that in order to assure the validity and reliability of the research instrument, categories must be mutually exclusive. That is, the categories that are being studied must be defined in such a manner that each piece of information obtained during the research can fit into only one category and no other. The English questionnaire had 78 distinct word categories, each one a different adjective describing a unique type of pain. The translation purported to have a distinctive, perfectly matched Spanish word for each English one. However, two of the categories of the translation consisted of exactly translated questionnaire of the the same word, punzante. [15]

Several other were tested with both bilingual and monolingual patients and the categories were too close in meaning to serve as unique descriptors. For example, three categories were all modifications of the word dolor (Spanish for pain): doliente, doloroso and adolorido. [16] Two more categories were simply variations on the word forcer. Neither of those categories would work because no Spanish­ speaking person would characterize his pain as “un dolor torciendo” or “un dolor torciente,” [17] since those phrases would make no sense semantically. A pain described as fearful in the English questionnaire becomes horrificante (a word that does not exist in Spanish), a radiating pain is translated as radiante (which means radiant in Spanish), and a wretched pain becomes afligido [18], which can mean upset, grieving, sad, troubled, or tormented. A patient who is in pain can certainly feel any of those emotions, but it would be nonsensical to describe the pain itself in that way.

The entire translation was riddled with such problematic words. How did these terms manage to pass the evaluation committee’s aforementioned test of “appropriateness of the semantic content and grammatical form”?

Evidently, the members of this committee did not have the necessary language skills to carry out the task. After hearing my comments, it was obvious to the study coordinator that in spite of its having been published in a prestigious journal, no amount of “bilingual expert” input or statistical analysis would make this translation achieve its objective.

This and many other such incidents have made it clear to me that no one can ever take the place of a translator who has the education and practical experience to render meaning faithfully from one language to another while at the same time respecting the conventions of the specialized field in which he or she labors. In working with medical researchers, I let them know I believe in the concept of division of labor: they are medical professionals and I am a language professional.

We each have our own area of expertise – theirs is medical and mine is linguistic. It’s simple. If you’re sick, see a doctor. If you need a translation, see a translator.

To quote Vitek once again:

“A bilingual expert is not necessarily a good translator, and a good translator is much more than a bilingual expert…if I have a choice between a doctor and a professional translator, I will always choose the latter.”

I could not have said it better.

Notes

[1] Steve Vitek. April 2003. “Is Technical Translation Really a Collaborative Activity?” The ATA Chronicle, pp.24-27.
[2] Ibid., p.25.
[3] Capitulo, Kathleen L., Maria A. Cornelio, and Elizabeth R. Lenz. August 2001. “Translating the Short Version of the Perinatal Grief Scale: Process and Challenges,” Applied Nursing Research. Vol. 14 No. 3, pp. 165-170.
[4] Ibid., p. 168.
[5] Ibid., p. 168.
[6] Escalante, Agustin, et.al. 1996. “Measuring Chronic Rheumatic Pain in Mexican Americans: Cross-Cultural Adaptation of the McGill Pain Questionnaire,” Journal of Clinical Epidemiology, Vol. 49 No. 12, pp. 1389-1399.
[7] Ibid., p.1390.
[8] Ibid., p. 1390.

[9] Ibid., p. 1390.
[10] Ibid., p. 1390.
[11] Ibid., p. 1390.
[12] Ibid., p. 1390.
[13] Ibid., pp. 1392-1396.
[14] Ibid., p. 1398.
[15] Ibid., p. 1396.
[16] Ibid., p. 1396.
[17] Ibid., p. 1396.
[18] Ibid., p. 1396.