upon the clinician’s fluency in the patient’s language, the patient’s fluency in English for example, the accuracy of interpretation services and the availability of appropriate vocabulary in the patient’s language for signs and symptoms of mental illness described in western diagnostic classifications.
Consider the case of many ethnic minority elders who do not speak English [Barker 1984; Manthorpe and Hettiarachy 1993; Lindesay et al. 1997a, 1997b]. It is plausible to think that, ideally, such patients should be assessed by a clinician who is able to speak the patient’s language and belongs to the patient’s culture. But this is often not possible. The use of bilingual health workers to facilitate such assessments has been advocated, but they are uncommon [Phelan & Parkman 1995]. Either of these approaches might seem to promise the transparent communication of signs and symptoms of mental illness. But this may nevertheless fail because transparent transmission by testimony is blocked.
It may be, for example, that there is a lack of matching vocabulary for signs and symptoms of psychiatric illness in the patient’s language. Interpretation will thus fail to be fluid and even a bilingual clinician will struggle to ask questions on symptoms based on western diagnostic classifications. There is, for example, no matching vocabulary for depression in some languages, like Gujarati [Shah 1999].
Disorders which of their nature undermine interpretation can block transmission by testimony. Formal thought disorder in languages with significantly different grammar than English, such as Gujarati, are difficult to elicit [Shah 1999]. Similarly, testing for cognitive impairment in the context of dementia can also be difficult in patients without fluency in English because many of the existing screening and diagnostic tests depend upon fluency in English.
In cases where direct communication in a shared language is not possible, relatives, non-clinical staff, clinical staff and professional translators (with and without special training in mental health) have been used [Phelan and Parkman 1995; Shah 1997a, 1997b].
To take another example: A 65 year old woman who had been deaf and dumb all her life was referred by her GP for a psychiatric assessment of depression. The patient could only communicate by sign language. She was, therefore, seen with a sign language interpreter. Other than non-verbal communication all the communication was through the sign language interpreter. The psychiatrist had no alternative but to rely on an interpreter. It would not have been practical to find a psychiatrist with such sign language skills and the psychiatrist could not learn them.
In addition to the problems already mentioned, there may be everyday difficulties in this process that render it less than transparent and thus undermine the direct transmission of knowledge about signs and symptoms. Interpreters may inaccurately translate the content of the questions or answers. They may translate their opinions rather than facts, and they may be emotionally biased. Patients may be too cautious to reveal signs and symptoms of mental illness because of doubts about confidentiality.
In the context of the discussion of testimony as a general method of inheriting another’s knowledge, one kind of bias or breakdown is particularly relevant. An interpreter may not accurately translate questions to the patient or accurately translate the patient’s response back to the clinician because of the interpreter’s beliefs about the patient’s illness [Shah 1997a, 1997b].
This has been observed in clinical settings by one of the author of this paper (AS) who understands Hindi, Punjabi and Urdu, but not sufficiently fluently to conduct a psychiatric interview without an interpreter. The clinician’s questions were frequently badly translated and leading questions were introduced by the interpreter, a practice frowned upon in psychiatric interview technique. In some cases, patients’ answers were translated only after interpreters had added their own diagnostic interpretations of the patients’ response. In other cases, there was an absence of matching vocabulary in the patients’ languages for the