Methodological Aspects of Interpretation and Translation Research

Methodological Aspects of Interpretation and Translation Research Daniel Gile INALCO and CEEI (ISIT) 1. Introduction Because of their importance in the development of culture and because they involve language, interpretation and translation have been the focus of much reflection and have generated a large body of literature, from St. Jerome and Calvin to Catford, Nida, Mounin, Steiner, not to mention countless other less well-known authors. However, most of the writings on I/T are normative or philosophical, and scientific research in the field is very recent, only several decades old. One of the driving forces behind the development of I/T research were efforts in the field of machine translation, which started in the fifties, dwindled temporarily and later regained momentum. Other vectors of This is an offprint from: translation research include the increasing use of mathematical methods and statistics in behavioural and linguistic sciences and the interest psychologists and Sylvie Lambert and Barbara Moser-Mercer (eds) psycholinguists have taken in I/T. Bridging the Gap: Empirical research in simultaneous interpretation Many practitioners reject the idea of scientific research on translation and interpretation: some fear these disciplines may lose the aura of mystery John Benjamins Publishing Company surrounding their profession if dissected under a microscope; others do not Amsterdam/Philadelphia believe science can shed much light on the processes and interactions involved; 1994 whereas others appear to fear that their intuitive theories and position in (Published as Vol. 3 of the series academia, or even that their social status may be threatened by scientific studies BENJAMINS TRANSLATION LIBRARY) of I/T. As put very lucidly by Shlesinger (1989): ISBN 90 272 2144 8 (Eur.)/1-55619-481-1 (US) ...those describing interpretation from the practitioner's standpoint are © Copyright 1994 - John Benjamins B.V. sometimes prone to a certain lack of detachment which surfaces in these writings in the form of a sense of awe at an impossible job incredibly done. No part of this book may be reproduced in any form, by print, photoprint, microfilm or any other means, without written permission from the publisher. 40 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 41 and by Stenzl (1983): 2.1. In I/T, Input and Output can be Observed and Recorded Under Particularly Favourable Conditions We are quite pleased when psychologists confirm that ours is a complex job which requires a number of highly developed skills, but we are perhaps less In translation, both input and output are entirely contained in written inclined to document the limits of our skills and to face the occasions when we did not properly understand a speaker or were unable to adequately render a documents, providing an extraordinarily favourable observation and recording message even if we had understood it. condition from the research angle. In simultaneous interpretation, input comes essentially through the interpreters' earphones and can therefore be observed However, other practitioners and many I/T teachers have been fostering 'real and recorded. However, some interpreters stress that visual input is important research* so as to be able to test intuitions and go beyond them through to them, but as yet, this claim is still unsupported experimentally (Anderson observation and experimentation (in the field of interpretation, see for example 1979). The need for such visual input could be observed and recorded by way Stenzl 1983, Arjona-Tseng 1989, as well as Gran and Taylor 1990). of video cameras placed in the interpretation booth, for example. With regard Interpretation and translation can be studied under a wide variety of angles: to the interpretation output as it reaches the audience, it is entirely contained in the economics of I/T, market development of I/T, the psychology of translators the message sent to the listeners' earphones, and is therefore easy to observe and interpreters, their sociology, their health, etc. This paper deals with and to record. In consecutive interpretation, the visual components of both research on and around I/T processes, which raises more I/T specific input and output are more important and more difficult to record because the methodological issues. The subject has been discussed by Toury (1991) in an relevant visual field is larger, but it could nonetheless still be captured to a article with a rather theoretical view of the issues and focusing on translation. certain extent. The present paper tackles more practical issues in research methodology and leans essentially towards interpreting, though most comments also apply to 2.2. Processes are NOT Easy to Observe and Record translation. In written translation, it is possible to observe some parts of the processes involved, such as the reading and writing as well as the documentary and 2. Basic I/T Characteristics and Their Implications on Research terminological 'research'. In some cases, intermediary target language texts may be observed as well. But the actual mental process of translation remains If translation and interpretation are represented as a process P acting on an invisible. Even the Thinking Aloud Protocols used in research on translation input I and producing an output O (Figure 1), the following comments become are only a reflection of the mental processes involved (Krings 1986). Similarly, evident: in consecutive interpretation, it is possible to observe and record the note-taking part of the process and to monitor interpreters listening and providing their oral rendition, but the mental interpretation process itself is invisible. I O 2.3. Although Many Rules on Output, Including Relations between Output and Figure 1. Translation/Interpretation process Input Have Been Described, Much of the Process LeadingfromInput to Output Remains Unknown This third principle is a corollary of the second. Given the invisible nature of the processes, they can be explored only by inference. There is a large consensus on the general directions, if not on the precise quantitative assessment, of I/T quality parameters, such as fidelity, linguistic correctness, 42 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 43 stylistic and terminological acceptability (see discussion below). There are also (except when a TV monitor is used, as is often the case in satellite conferences certain widely-accepted operational rules regarding I/T preparation strategies, and for TV interpretation). documentary and terminological 'research*, problem-solving tactics, etc. However, most of the I/T processing which does not involve conscious decision-making, i.e. the reading and speech recognition operations, the unconscious inferencing during comprehension of source language speech or With respect to scientific research on I/T, these basic characteristics suggest text, the meaning-to-language or language-to-language retrieval of target the following methodological principles: language terms and structures, remains largely unknown. Research in speech a) The most logical way to undertake any study of I/T mental processes production and comprehension in psycholinguistics, studies on reading and would be through the study of correlations between input and output or input experiments and models on attention in cognitive psychology as well as some and .process product. As Hoffstaedter puts it, "A critical point in the artificial intelligence research, particularly as applied to machine translation, investigation of mental processes is...that the processes themselves can never certainly do contribute their share of ideas. However, at this time, no single be observed directly...their investigation is always based on the PRODUCTS paradigm is fully accepted nor accounts for all phenomena observed in I/T of such processes" (Hoffstaedter 1987, quoted in Toury 1991). processes. This comes as no surprise, given that the study of language b) Because of variability, general conclusions regarding I/T mental comprehension and production, both in the oral and the written mode, is still processes can only be drawn on the basis of observations covering a wide range in a developmental phase. Furthermore, I/T processes also involve language- of variables and parametric values. This means that empirical studies, both switching, processing capacity constraints and linguistic interference between observational and experimental, need to be plentiful before data can be the source and target languages. considered as being representative of more than a limited population of practitioners and of more than a limited range of environmental I/T conditions 2.4. I/T Situations and Operations are Highly Variable and tasks. Until now, not only have there been few scientific studies of I/T, but, as Variability in I/T is striking, even to the casual observer. Variability is a noted by Toury (1991) regarding translation, and by Gile (1990a) and others, salient feature of I/T in communication situations with respect to subject matter, with respect to interpretation, most studies were limited to very small samples source language material and speeches, content density and logic, linguistic of subjects, languages and text or speech types. Generalizations are therefore quality, the range of working conditions, differences between individual still premature. translators and interpreters in terms of talent, training, experience, working c) Observational studies are the logical starting point. They are methods and motivation. This variability can also be observed in the same methodologically simple, as input and output can be easily observed and individual who does the same translation or interpretation twice, during training recorded without influencing translators' and interpreters' behaviour, whereas in I/T schools for example, as well as in the rare case when professionals are experimental studies may produce data unrepresentative of I/T if not conducted required to interpret or translate the same source language speech or text twice. with proper methodological care and sufficient know-how. Moreover, observational studies are inexpensive, because material requirements are 2.5. I/T Operational Environments are Fairly Simple generally limited to photocopies of documents and cassette tapes. In some cases, video cameras and video-players may also be necessary, but no complex, Technically speaking, in the case of translation, 'physical' I/T environments are expensive and sensitive equipment and no carefully prepared settings are easily described as consisting of a chair, a desk, paper and pencil, a typewriter required. or a computer, and sometimes a dictating machine. For consecutive d) Experimental studies are technically easier to carry out in I/T than in interpretation, they include any room or space where two persons meet and hold most other fields of human behaviour. In translation, the fact is obvious, since a discussion. For simultaneous interpretation, a booth with electronic material requirements for experiments are limited to 'source texts' which can equipment as well as an electronically equipped meeting room are required be controlled at will. In consecutive interpretation, material requirements 44 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 45 involve space and at least one 'speaker' and one 'client'. In the case of motivated, represent an especially valuable human resource. Moreover, they simultaneous interpretation, a meeting room, an electronically-equipped booth, can gain a lot from the exercise and the careful scrutiny of their work, which at least one 'speaker' and an audience are required. But in all these cases, it is part of the process, provided ethical rules and respect for them are observed. is technically easy to create an experimental environment that is quasi-identical However, caution is called for when using them in experiments or observational to actual field conditions. As pointed out by cognitive psychologist Flores studies of professional translation or interpretation. They can be used when d'Arcais (1978): "...after all, there are probably very few "real life" situations sufficient evidence is available to indicate that they are comparable to which are more similar to a laboratory psychological experimentation than the professionals in the type of behaviour under study. When no such evidence situation of an interpreter in a conference booth...". exists, one possible suggestion might be to combine samples of professionals In spite of these rather favourable conditions for its scientific study, I/T was and students, enabling comparisons between the two, and possibly accumulating not scientifically researched until approximately three decades ago. Since then, data on comparability or non-comparability. This has indeed been done, though several theses and dissertations, as well as a number of studies and papers have not with this aim in mind, in Dillinger's (1989) and Viezzi's (1989) recent been devoted to the subject. A significant proportion of these suffer from studies, not to mention studies having the specific objective of comparing methodological weaknesses. The following is an attempt to present those which students' and experienced professionals' performance (see for example Pinter, occur most frequently. 1969). 3.2. Materials 3. Major Methodological Issues and Problems in I/T Research Similarly, investigators have conducted experiments using input materials which 3.1. Subjects in I/T Experimentation were very different from materials found in the field. This problem is more tangible in interpretation than in translation, since written translation is Access to translators and interpreters for research purposes may be difficult for performed on a wide sample of texts, whereas interpretation is performed on geographic reasons, because of a lack of physical availability of practitioners, speech types occurring in a smaller number of communication situations. for professional reasons (confidentiality of translated material), and for personal Therefore, although interpreting sentences read out of context or written reasons: many translators and interpreters do not appreciate having their work material not intended for oral communication may deepen insight on aspects of carefully scrutinized by anyone but their clients. Researchers therefore tend to language comprehension and production or on other aspects of language-related take as subjects either I/T students or 'bilinguals', that is, non-practitioners skills, both processes and results may be quite different from those that occur having knowledge of languages. in actual interpretation. For example, in a recent paper (1990), Dillinger This strategy is widely challenged by practitioners, especially in the field of defended the texts he used for an earlier experimental study of comprehension conference interpretation (see for example Gile 1990a). A professional's in simultaneous interpretation (1989) against a critic who challenged the validity approach and appropriate working methods in I/T are the result of training, of these texts, written especially for the experiment rather than taken from an experience and selection: not only do they have to be acquired and fine-tuned, authentic conference. Dillinger's reaction was: but some students may never become professional translators or interpreters because they prove incapable of mastering and/or applying them. There may One might ask...why this would make a difference. No one knows what the be very significant differences in the way professionals on the one hand, and differences are (if there are any) between the texts used here and the sorts of students or amateur translators/interpreters on the other, perform I/T tasks, and texts found in actual conference situations. The hypothesis that the texts are experiments conducted with the latter subjects may not be representative of the different and the processes involved in comprehending and interpreting them are activity to be examined. qualitatively different because one text was presented to an audience and the other was not is not only unsupported by any evidence, but also seems entirely This does not mean that no I/T studies using students or amateurs as subjects implausible (p.43). should be undertaken. Students, who are readily available, numerous and 46 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 47 No scientific evidence indicated the existence of such a difference, but Error counts have been used to evaluate quality in general, and fidelity in interpreters report reacting to factors that have not yet been quantified or even particular. However, researchers have had difficulties in such endeavours. specifically identified. Those who engage in research themselves have stressed One specific example is that of Henri Barik (1969), whose definition of errors time and time again the need to use experimental materials and conditions as includes linguistic transformations not necessarily considered as errors in an close as possible to field conditions because of this. In Dillinger's case, since interpreter's mind (see Gile 1989). It would seem that Barik did not consult the whole validity of the experiment relies on the speeches, we feel that starting interpreters when defining errors. The question is not so much whether his out with the assumption that there is no difference constitutes a methodological definition is better than that of the practitioners' ; the point is that the latter aim error. to reduce the number of errors as they see them and may do nothing to try to reduce the number of errors as defined by Barik, in which case his error count 3.3. Experimental Conditions and Tasks may be uncorrelated with the difficulty of the task. Bank's definition of errors has also been challenged by Gerver (1976) as Along the same lines, one might note that experimental studies have been being "purely subjective" (Gerver applied this qualifier to his own definition as conducted under conditions or with tasks somewhat far removed from field well), and by Stenzl (1983) as too vague. Stenzl reports that when she tried to conditions, including translation into and from languages which are not the replicate Barik's results and to identify errors in his transcripts according to his practitioners1 usual active or passive languages or 'translation' of random own definitions, she found that her data did not agree with his. sequences of words (Treisman 1965). Another quality assessment problem resides in language acceptability Again, experimenters may argue that no evidence points to a significant evaluations. Not only are both the sensitivity and standards of native speakers difference between such tasks and authentic translation or interpretation. Again, highly variable, as illustrated in Gile (1985), but no evidence is available when no evidence or rationale in favour or against is available, the best policy regarding the sensitivity and standards of delegates in international conferences seems to take as a starting point the hypothesis that the general feeling of I/T or the quality standards aimed for by practitioners while interpreting. practitioners may be right. One should not forget that many research decisions The very basic question of quality definition in interpretation has only been are partly made on the basis of intuitive, scientifically unsubstantiated dealt with intuitively and prescriptively until very recently. The International knowledge of the subject or phenomenon under study. Association of Conference Interpreters (AIIC) tried to organize meetings with interpretation users in order to elicit some kind of feedback, but these attempts 3.4. Quantification and Quality Assessment were never very successful. The most straightforward scientific way of collecting data on actual quality perception delegates would clearly be by way Scientific researchers in I/T still hear practitioners say that translation and of questionnaires or interviews (Gile, 1983). The first study along such lines interpretation cannot be measured because they are "akin to the arts" or very was conducted in Trieste by Meak (1983). Another was carried out by Buehler complex and "cannot be reduced to formulas". A thorough discussion of this (1986), a third by Kurz (1989), and a fourth by Gile (1990c). But in view of issue is beyond the scope of this paper. However, it may be appropriate to the variability mentioned earlier, these studies constitute only a beginning and stress that quantification in the behavioural sciences is not tantamount to much replication and further studies are definitely required before quality of attempting to equate a behaviour with a set of equations or figures. The idea interpretation can be said to be understood. Methodologically, quality is to find indicators that can be 'measured', if only approximately, and yield assessment of translation seems less daunting, at least as regards non-literary data that will contribute to a better knowledge of the phenomenon. In this translation, because language standards relating to formal written texts are respect, many indicators seem to be relevant to I/T, depending on the aspect of easier to identify and because variables and parameters of reception conditions interpretation or translation under study: translation speed, translation costs, are less complex in translation than in interpretation. However, as pointed out source text length to target text length ratio, etc. In quality assessment, one by Stenzl (1983), Gerver's method of having interpretation assessed by 2 judges common indicator is the number of language errors in a target language text or who "...had experience of marking translation from the French, up to a first speech; another is the number of 'translation errors'. year University standard..." and who then corrected interpretation transcripts 48 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 49 (1976), is likely to be a poor reflection of the reactions of delegates sitting in rejection applicable - see for instance Hansen, Hurwitz and Madow 1953 and the conference room. Snedecor and Cochran 1967); neither are we referring to cases in which non- professionals are selected for experiments on professional practice. The 3.5. Drawing Conclusions problem is that even when professionals are given tasks that can be considered valid as I/T tasks, even in observational studies, which deal by definition with Experimental design and conclusion-drawing issues are not much different in real I/T tasks, researchers tend to extrapolate somewhat uncautiously. I/T from what they are in other disciplines of the behavioural sciences. Most One such example can be found in Dillinger's (1989) general conclusions on specific methodological difficulties and flaws seem to occur because of the interpretation based only on a single set of experiments with a total of 24 researchers1 background. The examples discussed below illustrate this fact, but subjects, working on two speeches in only two languages. Another example do not point at difficulties inherent to I/T research as such. involves Seleskovitch (1975) making general statements on interpretation on the The most basic difficulties in I/T conclusion drawing are related to the basis of a single experiment with a total of 12 subjects working in the validity of experiments with respect to professional I/T, as explained above consecutive mode on two English speeches. Inter alia, she considers that regarding subjects, materials, environmental conditions, tasks and evaluations. deverbalization is an established fact because in her experiment, with the Methodological weaknesses associated with this type of validity are found exception of some specific types of words, notes taken by interpreters do not mostly in research undertaken by non-practitioners of I/T, because they may not list "word-equivalents" in the target language. Other possible explanations are be aware of specific I/T rules and effects in speech or text production standards, not considered. This weakness is not linked to sophisticated methodology or strategies and tactics. highly specialized knowledge and can be spotted by analysis of the logic of her For instance, some non-interpreting researchers engaged in interpretation conclusions. research may be unaware that interpreters may deliberately opt to change the Toury (1991) mentions other weaknesses in the rationale of I/T researchers. surface structure of a sentence or select words which are morphologically The most striking example he refers to is a study by Sandrock (1982) on the use remote from the source language so as to avoid linguistic interference. In other of the Thinking Aloud Process (TAP) in translation research: a comparison was words, a deviation from the source language structure may mean the interpreter made between two translations of the same text, one traditional and the other is controlling the situation, whereas the selection of target language structures resulting from a verbalization and TAP process. The problem, as pointed out similar to source language structures indicates that the interpreter may be short by Toury, is that one translation was produced by a learner of English, and the of processing capacity. Similarly, the use of one word as opposed to another other was given as a "Model translation" in a manual. It is therefore possible, which resembles the corresponding source language word, could be an and even plausible, that differences resulted not only from differences in the indication that the interpreter has sufficient processing capacity left to processes, but also from differences in the translation expertise of the authors consciously resist linguistic interference, a fact which should not be mistaken of the two target language texts. for an indication of failure to find the 'right' word. One can point out in this context that the interpreters' view of fidelity may be quite different from that of the translators', since they deal with externally-paced message reception, as 4. Strategies for I/T Research opposed to internally-paced visual reception. This may lead to errors both in quality assessments and in evaluations of difficulty. Barik's approach to errors I/T research has been making some headway over the past decade or so, mostly is a good case in point, as mentioned above. because of importation of experimental methods from other disciplines (Toury A second major weakness in I/T research involves extrapolation. The flaws 1991), As Toury also points out, the implementation of such methods as well referred to here are not the technical problems caused by 'convenience ai observational methods in I/T is still suffering from weaknesses. We believe sampling' in statistical inference ('convenience sampling1 is the selection of that proper scientific research is required, because too many widely accepted subjects because they are available, as opposed to random sampling-based ideas are based solely on intuitive personal speculation and have resulted in methods which make inferential mathematics and hypothesis rejection or non- 50 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 51 some stagnation in I/T, as also noted by Toury (1991 - see also a discussion of but a formal agreement between institutions offering I/T research training at the the problem in Gile 1990a). graduate or doctoral level, which would allow their students to choose board However, we do not share Toury's view that empirical methods require members as M. A. or Ph.D. supervisors outside their school, could have distinct "translational hypotheses" to be tested. Scientific investigation in most fields advantages. starts with exploration of reality in the field, that is with observational studies. Systematic observation of reality is a valuable scientific act per se (see 4.2. Research Programs Fourastié 1966), which should not be snubbed by more fashionable experimentation, which practically "haunts" the scientific scene (Toury 1991). Besides training, actual research programs would be helpful in improving the Another type of experimenting, known as open experimenting, consists in overall situation, and I/T schools seem once again to be in the best position to generating controlled conditions not in order to test a specific hypothesis, but foster such programs. One obvious possibility is to involve research trainees for the purpose of examining the resulting situation as a whole and trying to in programs as described above in actual research. Another approach is the make inferences on that basis. Open experimenting is also very valuable (see "thesis" required at the end of basic I/T training in some schools, such as the Gile 1990b). In our view, the most fundamental reasons for methodological Scuola Superiore di Lingue Moderne per Interpreti e Traduttori in Trieste. weaknesses, as discussed above, lie in what can be summarized as insufficient Such "theses" are a good opportunity for students to carry out actual research, expertise in scientific research on the part of I/T practitioners engaged in as demonstrated by several theses published in Gran and Taylor (1990) and in research (were they better trained, I/T researchers outside I/T could avoid most The Interpreter's Newsletter. of their own errors by reading their work and cooperating with them). In research programs of this type, schools have an opportunity to replicate We believe that improvement of the situation can be accelerated by observations and experiments, which is a good way to corroborate results and appropriate strategies. Some suggestions are formulated below with this goal to enlarge samples. in mind. "Institutional strategies" aim at creating a favorable I/T research environment by the action of I/T schools, associations and governmental 4J. Interdisciplinary Contacts agencies entrusted with I/T responsibilities. "Personal strategies" are proposed to individual researchers who wish to engage in I/T research. Another role of institutions with respect to I/T research is to establish and improve interdisciplinary contacts. This can be done by inviting scientists from 4.1. Institutional Strategies Other disciplines such as linguistics, philosophy of science, and psychology, to lecture to students or to participate in I/T research projects. This can also be Highest priority should clearly be given to research training. The logical forum done by actively fostering information exchanges, for instance by sending such for such training would be the postgraduate I/T schools attached to universities. scientists information about I/T studies and projects which could be of interest Clearly, interdisciplinary cooperation is required at least in the early stages to them* of methodological training of I/T students and practitioners for research. We believe the programme should include some concepts from the philosophy of 4,4» Dissemination of Information in I/T Research Circles science, training in statistics for the social sciences and the presentation of research methods used in behavioural sciences. At a later stage, the I/T The situation of translation and interpretation research differs greatly with teaching staff could adapt the concepts and methods to I/T research. respect to dissemination of information. The translation research community Since there are I/T instructors with a scientific background who teach at il much Older, larger, wealthier, and better organized. There are numerous various locations, pooling together I/T research expertise through "international ttmslation journals and an increasing number of conferences dealing with I/T research advisory boards", possibly specialized in translation or trtittSlation studies. Furthermore, translator researchers have traditionally kept interpretation, may provide institutions and individuals with better consultative in touch with the academic communities of linguists, philosophers, literature services and research guidance. The activity of such boards could be informal, 52 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 53 scholars, and more recently, have been drawing the attention of the artificial - Research design and methods: Unless they are sure to be able to rely on intelligence community. their advisors for close monitoring and methodological help, budding I/T Interpretation researchers are still to a large extent isolated and unaware of researchers should aim at very simple design and methods. Observational each other's work, though much headway has been made in the past few years Studies and "open experimenting" (in which a situation is generated and with the Trieste school's initiatives, in particular its seminar on interpretation observed but no statistical inferencing for hypothesis-testing purposes is training (Gran and Dodds 1989) and the launching of an interpretation journal, done) are methodologically simpler that the more traditional hypothesis- The Interpreter's Newsletter. Another effort is underway through the creation, testing. Inferential statistics are often challenging, even for experienced in 1990, of the Paris-based IRTIN, or International Interpretation Research and researchers: see for example Gore and Altman (1982) for a wide discussion Theory Information Network, which publishes a bulletin twice a year. The of the issues. It is interesting to note that in a study of 149 medical papers International Association of Conference Interpreter's Research Committee also published in several journals, only 28% were judged of acceptable statistical endeavours to foster dissemination of information on interpretation research and quality (Schor and Karten 1966). It is therefore easier and at least as valid is due to publish a rather extensive interpretation bibliography soon. to use descriptive statistics rather than inferential statistics, though analyses Multi-centre contacts have been a reality in I/T for many years, but multi- cannot be as fine as with inferential statistics (see a discussion of the issue centre research is still to come. This, along with replication, is a good way of in Gile 1989 and 1990c). overcoming sample size and sample diversity (in particular as regards - Drawing conclusions: Bearing in mind the weaknesses discussed earlier in languages) limitations. Because so little empirical research has been carried out this paper, I/T practitioners should be very careful in drawing conclusions to date, simple projects which cost little and which are easy to coordinate are and extrapolating. Unlike researchers in most other disciplines, when they still very valuable. engage in I/T research, they are required to deal objectively and without bias with an activity in which they are intimately involved in their daily life and regarding which they have developed their own intuitive ideas. Again, the S. Strategies for Individuals presence of non-I/T scientists as advisors can be very valuable, as their views are more objective and their thinking more in line with the Cartesian 5.1. Strategies for I/T Practitioners discipline of scientific reasoning. I/T practitioners with no research training or background, wishing to engage in I/T research can do so because much can still be done with very simple 6. Strategies for Non-Practitioners methods. The following guidelines may be useful in such an endeavour: - The selection of research projects: Research topics should be well defined Ntott*practitioner I/T researchers do not have to overcome personal involvement and not overly ambitious. A completed project can be extended or followed Ilid are already familiar with and experienced in scientific thinking. This does up, but many initial projects are never completed because sights are set too ll§t prevent them from having preconceived ideas about I/T, but scientific high. The methods, potential difficulties and ways around them as well as discipline probably makes it easier to examine them more objectively on the a timetable should be clear from the start. The advice of more experienced basis of scientific data. On the other hand, like the public at large, they often I/T researchers should be very useful in this respect. Consider tfaey are familiar with I/T because of their experience with school - Research advisors: Pending the arrival of a generation of well-trained I/T 'translation1 as part of foreign language learning or because they have had some researchers, it seems advisable to have two advisors, one from I/T and the occasional experience in translating and interpreting. Such experience is often other from science, preferably from psychology or linguistics, even if only fallacious, a§ school translation and amateur translating and interpreting may be one of them is the official advisor. Of course, each advisor must be open- p l t i different from professional practice as regards strategies, tactics, user minded enough to welcome the contribution of the other. Institutional action expectations etc. This deceptive sense of familiarity may explain partly the fact along the lines advocated above may be of help in this respect. that many non-I/T practitioners who engaged in I/T research did not consult  54 DANIEL GILE METHODOLOGICAL ASPECTS OF RESEARCH 55 practitioners before and during their study on methodological questions, which Qll©, D. 1990a. "Scientific Research vs. Personal Theories in the Investigation of resulted in errors as discussed in Toury (1991) and in this paper. Interpretation". L. Gran and C. Taylor, eds. Aspects ofApplied and Experimental For all these reasons, we believe it is essential for investigators coming into Research on Conference Interpretation. Udine: Campanotto Editore. 28-41. I/T to consult with practitioners. Special attention should be given to the first Oîli, D. 1990b. "Observational Studies and Experimental Studies in the Investigation three items mentioned in the section on methodological issues and problems in Of Interpretation". Paper presented at the Scuola Superiore di Lingue Moderne per an earlier part of this paper, namely subjects, materials, experimental conditions Interpreti e Traduttori, Universita degli Studi di Trieste, March 19, 1990. and tasks, as this is where critical comments from practitioners may prove to (Mimeo.). be most useful. Oil©, D. 1990c. "L'Evaluation de la Qualité de l'Interprétation par les Délégués: Une Etude de Cas". The Interpreter's Newsletter 3. 66-71. Qll©, D. 1990d. "Research Proposals for Interpreters". L. Gran and C. Taylor, eds. Aspects of%Applied and Experimental Research on Conference Interpretation. Udine: References Campanotto Editore. 226-236. §or©, S.M. and Altman, D.G. 1982. Statistics in Practice. London: British Medical Anderson, L. 1979. Simultaneous Interpretation: Contextual and Translation Aspects. Association. Concordia University, Montreal, Canada. [Unpublished Masters' Thesis.] Qritt, L, and Dodds, J. 1989. The Theoretical and Practical Aspects of Teaching Arjona-Tseng, E. 1989. "Preparing for the XXIth Century" (Keynote speech). Conference Interpretation. Proceedings of the First International Symposium on Proceedings of the Twentieth Anniversary Symposium on the Training of Teachers Conference Interpreting at the University of Trieste. Udine: Campanotto Editore. of Translation and Interpretation, Monterey Institute of International Studies, Qnn» L, and Taylor, C. 1990. Aspects of Applied and Experimental Research on Monterey, California. Conference Interpretation. Udine: Campanotto Editore. Barik, H.C. 1969. A Study in Simultaneous Interpretation. University of North Hlfttin, M., Hurwitz, W. and Madow, W. 1953. Sample Surveys Methods and Carolina, Chapel Hill. [Unpublished Doctoral Dissertation.] Theory. New York, London and Sydney: John Wiley & Sons, Inc. Buehler, H. 1986. "Linguistic (Semantic) and Extra-linguistic (Pragmatic) Criteria for Mofhtaadter, P. 1987. "Poetic Text Processing and its Empirical Investigation". the Evaluation of Conference Interpretation and Interpreters". Multilingua 5. 231- Poetics 16. 75-91. Hun, I. 1989. "Conference Interpreting: User Expectations". ATA - Proceedings of Dillinger, M. 1989. Component Processes of Simultaneous Interpreting. McGill the 30th Annual Conference. Medford, NJ: Learned Information Inc. 143-148. University, Montreal, Canada. [Unpublished Doctoral Dissertation.] IflflgS» H.P. 1986. Was in den Koplen der Ubersetzer vergeht. Tubingen: Narr. Dillinger, M. 1990. "Comprehension During Interpreting: What Do Interpreters MîfÉ, L. 1983. La Selezione dellTnformazioneper ITnterpretazione Simultanea della Know that Bilinguals Don't?". The Interpreter's Newsletter 3. 41-58. Llteratura Medica. Trieste, Universita degli Studi di Trieste, Scuola Superiore di Flores d'Arcais, G.B. 1978. "The Contribution of Cognitive Psychology to the Study Lingua Moderne per Interpreti e Traduttori. [Unpublished monograph]. of Interpretation". D. Gerver and W. H. Sinaiko, eds. Language Interpretation and Pltltir» I, 1969. Der Einfluss der Uebung und Konzentration auf Simultanés Sprechen Communication. New York and London: Plenum Press, Nato Conference Series und Hdren. University of Vienna. [Unpublished Doctoral Dissertation]. 385-402. llfldfOek, U. 1982. Thinking-Aloud Protocols (TAPs - Ein Instrument zur Fourastié, J. 1966. Les Conditions de l'Esprit Scientifique. Paris: Gallimard. DekomposltiondesKomplexenProzesses'Ubersetzen'). Kassel: Gesamthochschule Gerver, D. 1976. "Empirical Studies of Simultaneous Interpretation: A Review and Kiilêl. (Mimeo.). a Model". R. Brislin, ed. Translation. New York: Gardner Press. 165-207. iillStovitch, D. 1975. Langage, Langues et Mémoire. Paris: Minard. Gile, D. 1983. "Aspects Méthodologiques de l'Evaluation de la Qualité du Travail en SUningtr» Miriam. 1989. Simultaneous Interpretation as a Factor in Effecting Shifts Interprétation Simultanée". META 28:3. 236-243. in the Position of Texts on the Oral-Literate Continuum. Tel-Aviv University. Gile, D. 1985. "La Sensibilité aux Ecarts de Langue et la Sélection d'Informateurs ^Unpublished Masters' Thesis.] dans l'Analyse d'Erreurs". The Incorporated Linguist 24:1. 29-32. Stfhor» S. and Karten, I. 1966. "Statistical Evaluation of Medical Journal Gile, D. 1989. La Communication Linguistique en Réunion Multilingue - Les MmumW, JAMA 195. 1123-1128. Difficultés de la Transmission Informationelle en Interprétation Simultanée. Sntisaof, Q.W. and Cochran, W.G. 1967. Statistical Methods, 6th Edition. Ames, Université de Paris III. [Unpublished Doctoral Dissertation.] taWii The Iowa State University Press. 56 DANIEL GILE Ste ^ 1 , J C ; l 983 * Simultaneominterprétation - Groundwork Towards a Comprehensive Model. University of London. [Unpublished Masters' Thesis.] Toury G. 1991. "Experimentation in Translation Studies: Achievements, Prospects In die \\\ NJ AMINS TRANSLATION LIBRARY the following titles have been published and some Pitfalls". Sonja Tirkonnen-Condit. ed. Empirical Research in Translation fliUN fur or arc scheduled to appear in the course of 1994: %U"%Z?ltural *tudies: Selected P*P™ of the TRANSIF Seminar, Savonlinna !. SA( iEK, Juan C: Language Engineering and Translation: Consequences of automation, 1988. Tubingen: Gunter Narr. 45-66. IW. Treisman, A. 1965. "The Effects of Redundancy and Familiarity on Translation and L SNEI .1,-1 IORNBY, Mary, Franz POCHHACKER and Klaus KAINDL (eds): Trans- lation Studies: An interdiscipline. Selected papers from the Translation Congress, Vie- 3® 379* B a C k a N a d V e a n d F ° r e i g n L a n g u a g e "- BritishJourml of Psychology 56. nna, 0-12 September 1992. 1994. Viezzi, M. "Information Retention as a Parameter for the Comparison of Sight 3, LAMBERT, Sylvie and Barbara MOSER-MERCER (eds): Bridging the Gap: Empiri- Iranslation and Simultaneous Interpretation: An Experimental Study" The cal research in simultaneous interpretation. 1994. y Interpreters'Newsletter 2. ' 4, T( HIRY, ( iidcon: Descriptive Translation Studies — and beyond. 1994. 5, POLLHRUP, Cay and Annette LINDEGAARD (eds): Teaching Translation and hltwprvtiHfi 2: Insights, aims, visions. Selected papers from the Second Language Inter^ national Conference, Elsinore, 4-6 June 1993. 1994.