Australian and New Zealand Journal of Psychiatry (1975) 9: 263
FOREIGN LANGUAGE INTERPRETING IN PSYCHIATRIC PRACTICE* by J. PRZCE**
SYNOPSIS This article examines the problems of foreign language interpreting in psychiatric practice. It is demonstrated how an Merpreter’s apparent competence may readily be mistaken for true competence and that it is well worthwhile formally assessing an interpreter‘s command of hL second language. In interpreting from patient to psychiatrist, the omission of important material was found to be a noteworthy feature. Here the meaninglessness of material not only favoured its omission but also often indicated important psychopathology. It was found that the more psychotic the patient the more likely it was that mistakes in interpreting from patienk to doctor would occur. INTRODUCTION Perhaps the commonest situation arising in psychiatry where an interpreter is required is when it becomes necessary to interview a migrant or somebody travelling through a country not his own who becomes mentally ill. For psychiatrists in Australia this is not an infrequent experience (Australian Department of Labour & Immigration Report, 1973). Less commonly a psychiatrist may be working in a developing country without knowing the language or languages spoken. There his need for the services of an interpreter is likely to be much greater. It was in the latter situation that the present study was conceived and carried out. Its main purposes were to investigate the frequency and types of error in interpreting during psychiatric interviews and to see whether any useful criteria could be established for the selection of those individuals most likely to prove good interpreters. In addition, it seemed possible that different amounts of error might be a function of the psychiatrists’ skill as well as the interpreters’ competence and the methods used were designed to test this possibility. Finally the study ~
* **
~~
Received 15 April 1975. Reader in Psychiatry, University of Queensland.
was planned in such a way that the frequency of error could be compared with the degree of clinical disturbance. METHODS Three doctors and three interpreters were used in this study. Of the doctors one was of consultant status, (A), the second had recently obtained a Diploma in Psychological Medicine, (B), while the third, (C), had no postgraduate qualification but had had 10 years experience in the specialty and was rated as thoroughly competent. Doctors B and C spoke the languages used throughout the project (English and Hindustani) fluently. Doctor A spoke only English. The three interpreters consisted of a charge orderly, ( l ) , who was considerably experienced in interpreting and regarded as a very responsible and reliable person; a second orderly, (2), who was regarded as rather careless and easy-going and who had a more limited experience in interpreting, and a paranoid schizophrenic in full remission, (3), who had a good educational background in both languages but who had virtually no experience as an interpreter. For all three Hindustani was their mother tongue and English a second language. Each doctor worked with each interpreter in turn, Thus nine different doctor/interpreter pairs were created. Each pair interviewed a number of psychiatric patients who understood only Hindustani. Many of these patients were actively psychotic at the time of interview. The interpreters were not given any special instructions beforehand. The interviews were tape recorded throughout and errors assessed at playback later. Doctors B and C, who spoke only English throughout the interviews, assessed their tapes themselves, while Doctor A was assisted by a fluently bilingual psychiatric nursing sister (since he spoke no Hindustani). The doctors were not permitted to discuss their results among themselves until all assessments had been completed. For each doctor/interpreter pair at least one hundred questions and answers were analysed for error, this being defined as an alteration in meaning rather than as a failure to translate word for word.
Downloaded from anp.sagepub.com at UNIVERSITE DE MONTREAL on July 6, 2015
264
FQREIGN LANGUAGE INTERPRETING
TABLE 1 Percentage error encountered in mterpreting from psychiatrist to patient. (Type A process) Interpreter 1 Interpreter 2 Doctor A Doctor B Doctor C
6 10 2
8 9 3
Interpreter 3 1 0 0
TABLE 2 Percentage error encountered in interpreting from patient to psycbitist. (Type B process) Interpreter 1 Interpreter 2
Interpreter 3 ~ _ _ _ _ _
Doctor A Doctor B Doctor C
12 7 13
17 4 10
TABLE 4 Assessor‘s scoring of proficiency in English compared with the mean emor for each interpreter
4 4 5
When a psychiatrist puts a question to a patient through an interpreter and receives an answer, two processes occur - first the translation and relaying of the question (designated here a Type A process), and secondly, the translation and relaying of the patient’s answer ( a Type B process). In this study these processes have been examined separately. RESULTS (a) The frequency of error The percentage of errors. encountered is shown in Tables 1 and 2, Table 1 referring to the Type A process, Table 2 to Type B. The data were subjected to Logit Analysis using the methods of Cox (1970). For this purpose the absolute numbers of errors, questions and answers were used rather than the percentages given above. The results of this analysis are shown in Table 3: there was a highly significant difference in the frequency of error between the interpreters and between the two processes (Types A and B) but not between the participating doctors. Errors in the Type B process (patient to doctor) were about twice as common as those for Type A (doctor to patient).
Interpreter 1 Interpreter 2 Interpreter 3 Assessor’s score Mean 70 error
4.5 8.3
3.5 8.5
8.5 2.7
One reason for presenting these data in full is to demonstrate that the original impression that interpreter 1 would be better than interpreter 2 with interpreter 3 possibly somewhere in between proved to be quite erroneous. Not only was interpreter 3 by far the best but there was very little to choose between the other two. (b) Further assessment of the interpreters Since interpreter 3 had a better educational background than the other two interpreters it seemed logical to investigate formally the command of oral English of all three interpreters. This was carried out in the English Department at the University of the South Pacific. The assessor had no knowledge of the psychiatrists’ assessments of the interpreters. He made his assessment on a 10 point scale ( a high score indicating greater proficiency than a low score). He first noted the interpreters’ responses to general questioning in English relating to job, home circumstances and school background. To each interpreter was then played the same piece of taped discussion about which the assessor subsequently asked a standard series of questions. The assessor’s estimation of Interpreter 1 was that “he has achieved some proficiency within a rather limited vocabulary area”; of Interpreter 2 that “he . . . is adequate within a limited range. His fluency masks, to some extent, his relatively limited capacity to handle English beyond a communication level”; and of Interpreter 3, “He has an accurate command of English structures in various situations and understands English even when spoken at reasonably rapid speeds”. Table 4 shows that assessor’s ratings for each interpreter. Included also is the mean percentage error for each interpreter (all doctors, Types A/B).
TABLE 7 Percentage error related to clinical disturbance Type A Process involved A: Acute Psychotics B: Chronic Psychotics C: Personality disorders, Neurotics, Psychotics in remission
Total Questions
% Errors
Total Answers
TvDe B < r
~
% Errors
256 124
19 4
7.4 3.2
252 123
39 10
15.5 8.1
330
14
4.2
329
17
5.2
Downloaded from anp.sagepub.com at UNIVERSITE DE MONTREAL on July 6, 2015
J. PRICE TABLE 5 Types of error encountered in interpreting from psychiatrist to patienk
No. of errors
Type of error Question turned into a leading question Distortion of question Addition by the interpreter Mistranslation due to carelessness Partial omission Mistranslation due to ignorance of English
12 10 8 5
3 1
Although these data are too limited for statistical analysis to be appropriate, they are consistent with the existence of an inverse relationship between the formal rating of verbal English skills and the mean percentage of error in interpreting. (c) Types of error encountered (i) Type A (Psychiatrist to Patient) These errors are tabulated in Table 5 . Interpreter 1, who was considerably experienced, was responsible for all the additions made. Frequently these reflected his own knowledge of questioning procedure: when he was asked to translate the question (put to a hallucinated schizophrenic) “Where d o these voices seem to come from?” he added “from inside or outside the head?”, a phrase obviously derived from his previous experience but an error in translation since clearly it is up to the psychiatrist not the interpreter to decide when the patient needs to be helped to explain what he has been experiencing. Interpreter 2, rated as rather lazy, was responsible for most of the leading questions: he was probably using this as a method of extracting information from the patient with the least amount of trouble. (ii) Type B (Patient to Psychiatrist) These errors are tabulated in Table 6. Omissions, which were almost always partial, were responsible for over half of the errors made. Frequently a patient would give a long involved answer and in translation part of this would be omitted. On one occasion the translator’s response to a long rambling answer was simply “He isn’t answering the question”. However, this answer had contained sufficient English words to make one suspect strongly the presence of delusions since references were made in English to “Queen Elizabeth”, “the Prime Minister” and “the Chief Justice”. Playback subsequently confirmed this suspicion fully. Sometimes, however, a highly significant phrase of a rather meaningless nature would be omitted in the translation of quite a short answer. For example, a woman was asked why she had stopped drinking her own urine. She answered, “I don’t know why! Since rlie well dried u p I’ve just stopped doing it.” In the interpreting, the italicized words were omitted and in consequence the presence of a number of primary delusions was completely overlooked.
265
One patient was asked where he was born: his answer, “In Madras” was interpreted as “In Bombay”. Errors of this type have been classified as ‘mistranslation due to carelessness’ in Table 6. Occasionally, probably in an attempt to achieve a more meaningful reply the interpreter added a word or two of his own in his translation: One patient, discussing hallucinatory voices said, “They just talk”. This was translated incorrectly as, “They just talk with me”. In fact, the voices were talking (and arguing) among themselves and talking about the patient in third person. Hence the translator, by his two word addition, led the psychiatrist to overlook symptoms designated by Schneider (1957) as of first rank in the diagnosis of schizophrenia. Generally omissions led to more serious oversights than mistranslation, since frequently the latter could be recognised as inadequate and the psychiatrist was then able to question the patient further. One patient was asked, “How can you see the hand when your eyes are shut?” She replied, “As soon as I close my eyes I can feel this hand on my chest, just like a dream”. This was interpreted, extremely crudely, as “It was a hand and also a dream”. This answer was so unsatisfactory that the psychiatrist immediately went on to ask further questions which successfully elucidated exactly what the patient had been experiencing. One patient asked her age replied, “56”. This was interpreted as “about 56” and exemplifies what appears in Table 6 as an “Exact answer interpreted as inexact”. This is clearly an error since an impression of inexactitude was conveyed where none was contained in the patient’s reply. (d) The relationship between transla4or error and clinical dkturbance For this purpose only interview material relating to patients in the following three categories was used: A - grossly disturbed psychotics; B - patients who were chronically psychotic; C - patients with personality disorders and neurosis and patients pre-
TABLE 6 Types of error encountered in interpreting from patient to psychiatrist Types of error Omission of part or whole of the answer Mistranslation due to carelessness Mistranslation due to ignorance of English Distortion of answer Additions made by interpreter Faulty summary by interpreter Exact answer interpreted as if inexact Inexact answer interpreted as if exact Mistranslation due to ignorance of Hindustani
Downloaded from anp.sagepub.com at UNIVERSITE DE MONTREAL on July 6, 2015
No. of errors 40
9 6 6 6 6 2 1
1
FOREIGN LANGUAGE INTERPRETING
266
TABLE 3 Logit analysis of data from Tables 1 and 2 Effect Interpreters Doctors, after allowing for differences between interpreters Types ( A & B), after allowing for above effects Interaction (Doctors x Interpreters), after allowing for main effects) Residual Total
d.f.
X2
27.1965
2
1.9632 1 1.6443
2
3.4484 22.6587
4 8
66.9111
17
1
P
< 0.001 > 0.1 < 0.001 > 0.1
Wilcoxon’s two sample tests for linear trend in proportion gives, for the Type A process, x2 = 2.595 ( 1 d.f.) with p > 0.1 and for the Type B process, xz = 17.253 ( 1 d.f.) with p 0.001.
FOREIGN LANGUAGE INTERPRETING IN PSYCHIATRIC PRACTICE* by J. PRZCE**
SYNOPSIS This article examines the problems of foreign language interpreting in psychiatric practice. It is demonstrated how an Merpreter’s apparent competence may readily be mistaken for true competence and that it is well worthwhile formally assessing an interpreter‘s command of hL second language. In interpreting from patient to psychiatrist, the omission of important material was found to be a noteworthy feature. Here the meaninglessness of material not only favoured its omission but also often indicated important psychopathology. It was found that the more psychotic the patient the more likely it was that mistakes in interpreting from patienk to doctor would occur. INTRODUCTION Perhaps the commonest situation arising in psychiatry where an interpreter is required is when it becomes necessary to interview a migrant or somebody travelling through a country not his own who becomes mentally ill. For psychiatrists in Australia this is not an infrequent experience (Australian Department of Labour & Immigration Report, 1973). Less commonly a psychiatrist may be working in a developing country without knowing the language or languages spoken. There his need for the services of an interpreter is likely to be much greater. It was in the latter situation that the present study was conceived and carried out. Its main purposes were to investigate the frequency and types of error in interpreting during psychiatric interviews and to see whether any useful criteria could be established for the selection of those individuals most likely to prove good interpreters. In addition, it seemed possible that different amounts of error might be a function of the psychiatrists’ skill as well as the interpreters’ competence and the methods used were designed to test this possibility. Finally the study ~
* **
~~
Received 15 April 1975. Reader in Psychiatry, University of Queensland.
was planned in such a way that the frequency of error could be compared with the degree of clinical disturbance. METHODS Three doctors and three interpreters were used in this study. Of the doctors one was of consultant status, (A), the second had recently obtained a Diploma in Psychological Medicine, (B), while the third, (C), had no postgraduate qualification but had had 10 years experience in the specialty and was rated as thoroughly competent. Doctors B and C spoke the languages used throughout the project (English and Hindustani) fluently. Doctor A spoke only English. The three interpreters consisted of a charge orderly, ( l ) , who was considerably experienced in interpreting and regarded as a very responsible and reliable person; a second orderly, (2), who was regarded as rather careless and easy-going and who had a more limited experience in interpreting, and a paranoid schizophrenic in full remission, (3), who had a good educational background in both languages but who had virtually no experience as an interpreter. For all three Hindustani was their mother tongue and English a second language. Each doctor worked with each interpreter in turn, Thus nine different doctor/interpreter pairs were created. Each pair interviewed a number of psychiatric patients who understood only Hindustani. Many of these patients were actively psychotic at the time of interview. The interpreters were not given any special instructions beforehand. The interviews were tape recorded throughout and errors assessed at playback later. Doctors B and C, who spoke only English throughout the interviews, assessed their tapes themselves, while Doctor A was assisted by a fluently bilingual psychiatric nursing sister (since he spoke no Hindustani). The doctors were not permitted to discuss their results among themselves until all assessments had been completed. For each doctor/interpreter pair at least one hundred questions and answers were analysed for error, this being defined as an alteration in meaning rather than as a failure to translate word for word.
Downloaded from anp.sagepub.com at UNIVERSITE DE MONTREAL on July 6, 2015
264
FQREIGN LANGUAGE INTERPRETING
TABLE 1 Percentage error encountered in mterpreting from psychiatrist to patient. (Type A process) Interpreter 1 Interpreter 2 Doctor A Doctor B Doctor C
6 10 2
8 9 3
Interpreter 3 1 0 0
TABLE 2 Percentage error encountered in interpreting from patient to psycbitist. (Type B process) Interpreter 1 Interpreter 2
Interpreter 3 ~ _ _ _ _ _
Doctor A Doctor B Doctor C
12 7 13
17 4 10
TABLE 4 Assessor‘s scoring of proficiency in English compared with the mean emor for each interpreter
4 4 5
When a psychiatrist puts a question to a patient through an interpreter and receives an answer, two processes occur - first the translation and relaying of the question (designated here a Type A process), and secondly, the translation and relaying of the patient’s answer ( a Type B process). In this study these processes have been examined separately. RESULTS (a) The frequency of error The percentage of errors. encountered is shown in Tables 1 and 2, Table 1 referring to the Type A process, Table 2 to Type B. The data were subjected to Logit Analysis using the methods of Cox (1970). For this purpose the absolute numbers of errors, questions and answers were used rather than the percentages given above. The results of this analysis are shown in Table 3: there was a highly significant difference in the frequency of error between the interpreters and between the two processes (Types A and B) but not between the participating doctors. Errors in the Type B process (patient to doctor) were about twice as common as those for Type A (doctor to patient).
Interpreter 1 Interpreter 2 Interpreter 3 Assessor’s score Mean 70 error
4.5 8.3
3.5 8.5
8.5 2.7
One reason for presenting these data in full is to demonstrate that the original impression that interpreter 1 would be better than interpreter 2 with interpreter 3 possibly somewhere in between proved to be quite erroneous. Not only was interpreter 3 by far the best but there was very little to choose between the other two. (b) Further assessment of the interpreters Since interpreter 3 had a better educational background than the other two interpreters it seemed logical to investigate formally the command of oral English of all three interpreters. This was carried out in the English Department at the University of the South Pacific. The assessor had no knowledge of the psychiatrists’ assessments of the interpreters. He made his assessment on a 10 point scale ( a high score indicating greater proficiency than a low score). He first noted the interpreters’ responses to general questioning in English relating to job, home circumstances and school background. To each interpreter was then played the same piece of taped discussion about which the assessor subsequently asked a standard series of questions. The assessor’s estimation of Interpreter 1 was that “he has achieved some proficiency within a rather limited vocabulary area”; of Interpreter 2 that “he . . . is adequate within a limited range. His fluency masks, to some extent, his relatively limited capacity to handle English beyond a communication level”; and of Interpreter 3, “He has an accurate command of English structures in various situations and understands English even when spoken at reasonably rapid speeds”. Table 4 shows that assessor’s ratings for each interpreter. Included also is the mean percentage error for each interpreter (all doctors, Types A/B).
TABLE 7 Percentage error related to clinical disturbance Type A Process involved A: Acute Psychotics B: Chronic Psychotics C: Personality disorders, Neurotics, Psychotics in remission
Total Questions
% Errors
Total Answers
TvDe B < r
~
% Errors
256 124
19 4
7.4 3.2
252 123
39 10
15.5 8.1
330
14
4.2
329
17
5.2
Downloaded from anp.sagepub.com at UNIVERSITE DE MONTREAL on July 6, 2015
J. PRICE TABLE 5 Types of error encountered in interpreting from psychiatrist to patienk
No. of errors
Type of error Question turned into a leading question Distortion of question Addition by the interpreter Mistranslation due to carelessness Partial omission Mistranslation due to ignorance of English
12 10 8 5
3 1
Although these data are too limited for statistical analysis to be appropriate, they are consistent with the existence of an inverse relationship between the formal rating of verbal English skills and the mean percentage of error in interpreting. (c) Types of error encountered (i) Type A (Psychiatrist to Patient) These errors are tabulated in Table 5 . Interpreter 1, who was considerably experienced, was responsible for all the additions made. Frequently these reflected his own knowledge of questioning procedure: when he was asked to translate the question (put to a hallucinated schizophrenic) “Where d o these voices seem to come from?” he added “from inside or outside the head?”, a phrase obviously derived from his previous experience but an error in translation since clearly it is up to the psychiatrist not the interpreter to decide when the patient needs to be helped to explain what he has been experiencing. Interpreter 2, rated as rather lazy, was responsible for most of the leading questions: he was probably using this as a method of extracting information from the patient with the least amount of trouble. (ii) Type B (Patient to Psychiatrist) These errors are tabulated in Table 6. Omissions, which were almost always partial, were responsible for over half of the errors made. Frequently a patient would give a long involved answer and in translation part of this would be omitted. On one occasion the translator’s response to a long rambling answer was simply “He isn’t answering the question”. However, this answer had contained sufficient English words to make one suspect strongly the presence of delusions since references were made in English to “Queen Elizabeth”, “the Prime Minister” and “the Chief Justice”. Playback subsequently confirmed this suspicion fully. Sometimes, however, a highly significant phrase of a rather meaningless nature would be omitted in the translation of quite a short answer. For example, a woman was asked why she had stopped drinking her own urine. She answered, “I don’t know why! Since rlie well dried u p I’ve just stopped doing it.” In the interpreting, the italicized words were omitted and in consequence the presence of a number of primary delusions was completely overlooked.
265
One patient was asked where he was born: his answer, “In Madras” was interpreted as “In Bombay”. Errors of this type have been classified as ‘mistranslation due to carelessness’ in Table 6. Occasionally, probably in an attempt to achieve a more meaningful reply the interpreter added a word or two of his own in his translation: One patient, discussing hallucinatory voices said, “They just talk”. This was translated incorrectly as, “They just talk with me”. In fact, the voices were talking (and arguing) among themselves and talking about the patient in third person. Hence the translator, by his two word addition, led the psychiatrist to overlook symptoms designated by Schneider (1957) as of first rank in the diagnosis of schizophrenia. Generally omissions led to more serious oversights than mistranslation, since frequently the latter could be recognised as inadequate and the psychiatrist was then able to question the patient further. One patient was asked, “How can you see the hand when your eyes are shut?” She replied, “As soon as I close my eyes I can feel this hand on my chest, just like a dream”. This was interpreted, extremely crudely, as “It was a hand and also a dream”. This answer was so unsatisfactory that the psychiatrist immediately went on to ask further questions which successfully elucidated exactly what the patient had been experiencing. One patient asked her age replied, “56”. This was interpreted as “about 56” and exemplifies what appears in Table 6 as an “Exact answer interpreted as inexact”. This is clearly an error since an impression of inexactitude was conveyed where none was contained in the patient’s reply. (d) The relationship between transla4or error and clinical dkturbance For this purpose only interview material relating to patients in the following three categories was used: A - grossly disturbed psychotics; B - patients who were chronically psychotic; C - patients with personality disorders and neurosis and patients pre-
TABLE 6 Types of error encountered in interpreting from patient to psychiatrist Types of error Omission of part or whole of the answer Mistranslation due to carelessness Mistranslation due to ignorance of English Distortion of answer Additions made by interpreter Faulty summary by interpreter Exact answer interpreted as if inexact Inexact answer interpreted as if exact Mistranslation due to ignorance of Hindustani
Downloaded from anp.sagepub.com at UNIVERSITE DE MONTREAL on July 6, 2015
No. of errors 40
9 6 6 6 6 2 1
1
FOREIGN LANGUAGE INTERPRETING
266
TABLE 3 Logit analysis of data from Tables 1 and 2 Effect Interpreters Doctors, after allowing for differences between interpreters Types ( A & B), after allowing for above effects Interaction (Doctors x Interpreters), after allowing for main effects) Residual Total
d.f.
X2
27.1965
2
1.9632 1 1.6443
2
3.4484 22.6587
4 8
66.9111
17
1
P
< 0.001 > 0.1 < 0.001 > 0.1
Wilcoxon’s two sample tests for linear trend in proportion gives, for the Type A process, x2 = 2.595 ( 1 d.f.) with p > 0.1 and for the Type B process, xz = 17.253 ( 1 d.f.) with p 0.001.
