Selecting a topic for PhD research in interpreting

Daniel Gile

Université Lumière Lyon 2

Published in Gile, Daniel, Helle Dam, Bodil Martinsen and Anne Schjoldager (eds). 2001. Getting Started in Interpreting Research. Amsterdam/Philadelphia: John Benjamins. 1-22.

 

 

1.      Introduction

 

Most scientific disciplines are essentially defined on the basis of their research objects (physics studies the physical aspects of reality, zoology studies animals, astronomy studies celestial bodies, linguistics studies language), and at discipline level, find themselves tackling specific research questions and topics as a matter of course. At the level of institutional bodies - such as research laboratories or academic departments - many research questions and objectives are determined by external actors such as government agencies, industrial sponsors and other funding organisations, and by circumstances (obvious examples can be found in medical research). At individual level, researchers working within these bodies are often assigned to research topics on the basis of their institution's needs. In other cases, experienced individual researchers take up research topics and objectives in some sort of natural sequence, based on their previous research.

            Selecting a topic for research is therefore not necessarily problematic. However, when beginning researchers, and more specifically (prospective) PhD students, have to select their own topic (sometimes they are given a list of potential research topics by their supervisor - Cenková's paper in this volume refers to a similar system for MA theses), it turns out that it is a very difficult task for them, especially in the field of interpreting. In fact, a sizeable proportion of PhD students never manage to take off, precisely because they fail to overcome this problem, and in spite of the absence of precise statistics, it is probably fair to say that this accounts for a large proportion of aborted dissertations. This paper sets out to provide some general guidance to prospective PhD candidates in the selection of a topic for research, with some emphasis on and illustrations from the field of interpreting.

 

2.   Expectations from a research project

 

Research is basically a target-oriented activity, and it is important for beginning researchers to know what the expectations are in order to choose an appropriate target for their endeavour.

            Essentially, science is a systematic, rule-bound, collective exploration of reality. A prevailing norm in the scientific community suggests that individual research efforts should contribute to such exploration by adding and/or correcting and/or consolidating either knowledge or a potentially useful tool for knowledge acquisition.

            In concrete terms, such contributions can take three forms:

 

 

2.1       Empirical contributions

 

These can be the discovery of unknown physical, social and behavioural entities and/or the unveiling of facts describing their nature and behaviour. In the field of interpreting, to take just a few examples of what could be a very long list, such empirical contributions can take the form of new information about:

 

·        language-specific interpreting strategies (Dawrant 1996),

·        the personal history and linguistic behaviour of "truly bilingual" conference interpreters (Thiéry 1975),

·        progress in the interpreting students' ability to listen and speak at the same time (Pinter 1969),

·        inter-subject variability in the vocabulary used to interpret a source speech (Lamberger-Felber 1998),

·        the interpreter's working memory span (Padilla 1995),

 

            Note that generally in the scientific establishment as a social institution, while new empirical findings are given the highest priority, corrections of previous findings are also considered very important, and confirmation of previous findings by replication is indispensable. In routine scientific activity, uncovering new information is therefore not a standard requirement.  At PhD level, some degree of innovation is generally required, but this need not be a new theory, a new model, or the discovery of new laws. Prospective PhD students should keep this in mind in order to avoid setting their sights too high unnecessarily.

 

 

2.2     Conceptual ('theoretical') contributions

 

Another acceptable type of  contribution is conceptual, in the form of new theories, new questions, new hypotheses to test, new analyses of known facts or findings: why do so many errors occur in source-speech segments which do not present any measurable objective difficulty (Gile 1989)? Why is it more difficult to interpret read speeches than impromptu speeches (Déjean Le Féal 1978)? Why is quality perception by delegates highly variable (Gile 1999)?

            Note that generally, conceptual contributions generate empirical research, which leads to empirical findings, whereas the reverse is not true.

            It may be worthwhile pointing out that in early interpreting research, most of the conceptual contributions were introspective, and published in the form of essays, with few references to the literature from the relevant cognate disciplines, and few attempts to actually test ideas. Over the past ten to fifteen years, the paradigm has shifted towards a more sophisticated form of research. Introspection-based speculation has been replaced to a large extent by conceptual research more in line with the behavioural sciences, referring extensively to existing theories as reflected in the literature, with a strong interdisciplinary component involving mostly linguistics and the cognitive sciences (see for example Dillinger 1989, Setton 1997). Prospective PhD students should be aware that a theory-oriented project now involves much reading of theory in cognate disciplines, and assess the feasibility of their project accordingly (see Section 3).

 

 

2.3   Methodological contributions

 

Exploration of reality in all its complexity is a difficult endeavour, and one important component of scientific work is to develop tools that help explore reality in a more sensitive, accurate and reliable way. Therefore, the value of research projects can be assessed not only on the basis of new information they collect and new conceptual/theoretical analyses and entities they introduce, but also on the basis of new exploratory tools they develop. Such contributions can materialise as new experimental designs, statistical procedures, the development of metrics, of measurement tools, of analytical procedures, of classification and categorisation tools. For example, Pöchhacker (1994) designed a tool to measure parameters of source speeches relevant to interpreting difficulty, and Tommola and Hyöna  (1990) looked at the feasibility of the use of pupil dilation measurements as an indicator of mental load during interpreting.

            As is the case of conceptual/theoretical contributions, methodological contributions are generally associated with empirical findings which support them. Without them, it is difficult to assess the value of a tool.

 

 

2.4    What is the minimum required contribution for a research study?

 

One frequent misunderstanding among beginners is the idea that a research project has to "find an answer" or "solve a problem". Associated with this misunderstanding is their somewhat unrealistic ambition, often expressed in the introduction of dissertations, to "bridge the gap between theory and practice", to "determine/develop the best training methods", to "show the mental operation of interpreting", to "see how the interpreter's personality is expressed in his/her interpretation style", to "show that performance when working into a B language is better/worse than working into an A language", to "show that a particular way of using a computer in the booth is most efficient", etc. Reality is generally far too complex to allow a single study to find complete and final answers to such questions. Most studies 'only' contribute one or a few bricks to the building which is being erected by many other members of the community, and more comprehensive (albeit generally provisional) answers generally result from the efforts of dozens, if not hundreds of researchers over long periods. This is often a disappointment to beginners, who may be surprised when their supervisor tells them after they have worked on their thesis or dissertation for some time that they can stop, because they have "enough" for the degree.

            This disappointment is somewhat puzzling at first sight: while reviewing existing literature in the course of their preparatory work, budding scholars have ample opportunity to realise that previous theses and dissertations do not provide comprehensive answers either. Expectation bias introduced by the strength of their own motivation in starting a doctoral project may be the reason for their reaction. Be it as it may, calling the attention of beginners to the fact that individual research projects generally make small contributions to the community's collective exploratory endeavour may help them accept more modest objectives than those they would have contemplated otherwise.

 

 

3.   Topic selection and feasibility

 

One further point to remember when selecting a topic for research is that science is 'opportunistic', in the sense that its objectives and methods adapt to what is feasible, gradually extending the territory and pushing the limits further as more tools, concepts, financial and human resources become available. Generally, doctoral students who do not take on board existing constraints when planning their study either do not manage to complete their dissertation, or find themselves with a flawed end-product. Frequent feasibility problems include the following:

 

 

3.1     Lack of time

 

MA and PhD regulations often impose limitations on time-to-completion. Besides these formal constraints, each researcher also takes part in life outside the academic microcosm, and experiences other motivations and constraints which may compete with the demands of doctoral work. Some PhD candidates need the degree for their career, in which case this requirement sustains their motivation, but imposes time constraints of its own (for instance if their scholarship or grant has a set duration, or if they need to complete the degree within a certain period in order to have an employment contract offered or renewed). Many PhD candidates also have an important professional or personal activity outside university. In the latter case, especially in the field of interpreting, where successful professionals may be very busy and earn their living handsomely, once the coursework is over, it may be difficult for PhD candidates to sustain their motivation, as other activities become increasingly important. As time goes by, more interpreting work, and possibly an enlarged family and more active social life make the benefit associated with completing an advanced degree less attractive at the required cost in time and efforts, and many PhD candidates drop out.

                    Two distinct but interdependent time constraints therefore need to be addressed when selecting a topic (actually, when considering the feasibility of doctoral studies for a given prospective doctoral student): total duration to completion, and total availability of time on a regular basis (daily, weekly or monthly). A project involving much field work and/or laboratory work is not feasible if the candidate is too busy to spare the necessary time slots. And yet, this consideration is often neglected. Similarly, a project designed around events lasting several months or longer (for instance, when studying progress made by interpreting students over two years of studies) or occurring periodically with long intervals in-between (such as graduation examinations) may require several years for data collection even if it is well planned and implemented, let alone if there are problems with the data collected the first time.

                    Beginners also tend to underestimate the time required to draft the dissertation. Many believe that once the data have been analysed and results are available, 'writing it up' is fast and effortless. And yet, writing a document of up to several hundred pages presenting a large volume of information and thoughts in a clear, precise style and along a logical and convincing path which takes the reader from a general introduction to the conclusion is a lengthy process, which may take several months to the better part of a year even if the student's effort is sustained. Actually, many a PhD student who has completed his/her design, data collection and data analysis has given up when the 'only' remaining task was to "write up" the dissertation.

 

 

3.2       Lack of baseline expertise

 

Another issue which arises in any study which involves transdisciplinary components is baseline expertise, both declarative ('knowing about things', in particular knowing about relevant existing theories and facts) and procedural ('know-how', for instance as regards experimental design or data processing), insofar as the project may require expertise which the researcher does not yet have. Lack of awareness of the gap may lead to aborted projects (generally when the problem arises in the early stages of actual work on the concept, and the researcher finds him/herself unable to overcome it), or to seriously flawed studies (generally, when awareness of the problem arises at a later stage, and the researcher decides to go ahead with the study and complete it nevertheless). In the field of interpreting, gaps in declarative expertise are often found in psychological and linguistic concepts, theories and tools, and there are frequent weaknesses in procedural expertise as regards inferencing (see Gile's second paper in this volume), let alone experimental design, statistical data processing, linguistic analysis of the corpus, etc.

            Note that such lack of expertise may also be found in the supervisor in charge, whose own background may be of little relevance to the dissertation's topic and paradigm, and who was assigned or chosen as a supervisor because no one else was available locally (also see Section 4.2.3).

 

 

3.3       Lack of access to the literature

 

A pandemic problem in interpreting research is difficult access to the relevant literature. Few university libraries, even in well-known translation and interpretation schools in West-European countries, let alone academic departments not specialised in translation and interpretation, offer adequate coverage of interpretation literature. Ironically but understandably, it is much easier for interpretation students to find specialised texts in cognitive psychology and linguistics. Many research projects by serious scholars from the adjacent disciplines have gone astray because of a lack of insight into the mechanisms of interpreting, whereas more available literature might have prevented this from occurring. The lack of availability of interpretation literature has also generated much repetition and loss of opportunities within the field.

 

 

3.4       Lack of access to subjects and/or materials

 

Difficult access to subjects is a well-known problem to all researchers in the field and is mentioned time and again in the literature (though personal accounts in this volume do not confirm this - see the conclusion at the end of this book). The difficulty arises not only because of  the small size of the total population of conference interpreters (court interpreters and community interpreters are much more numerous in many countries) and because it is spread out geographically, but also due to the interpreters' reluctance to serve as subjects for research. Authentic source speeches can be found, but those having precise pre-determined linguistic, informational and other characteristics are sometimes hard to come by, and so are recordings of their authentic target-speech renderings, with the exception of those that have been broadcast on radio, TV or the Internet.

 

Some supervisors can help students plan their work adequately (see in particular Cenková's paper in this volume on the guidance provided institutionally at MA level at the Institute of Translation Studies at Charles University, Prague). Others consider that it is up to the students to take responsibility for their work. Still others are too overloaded to be able to devote to each of their students as much time as would be required to check feasibility thoroughly. Finally, as mentioned above, supervisors may find themselves in charge of a student working on a topic with which they are not familiar, and do not have the necessary knowledge to guide him/her efficiently through feasibility analysis. In all these cases, students have to take their fate in their own hands.

 

 

4.   Dealing with feasibility issues: project planning

 

The best way to avoid the trap of feasibility problems arising unexpectedly after the project has started and jeopardising its success is to select one's topic only after careful consideration of the requirements and potential problems versus available resources.

 

 

4.1       Areas of interest, topics and objectives

 

The first step would be to identify one or several areas of interest, such as interpretation quality, interpretation training, the interpreter's personality, the interpreter's status in society, etc. In each such area of interest, one or several possible topics can be considered. For instance, in the area of interpretation quality assessment, Collados Aís (1996) chose to focus on the effect of monotonous intonation on quality perception, which is just one out of the many dimensions of quality. In the area of cognitive abilities of interpreters, Padilla (1995) checked their memory span for one type of data, again a very specific location in a wide space of possibilities.

            Within each topic, specific objectives should be considered. These can be exploratory, as was the case Thiéry's doctoral dissertation (1975) devoted to an initial description of the personal history and behaviour of 'true bilingual' interpreters. They can also be focused around specific questions, as is the case of Collados Aís's dissertation mentioned above. They can also consist of tests of a hypothesis, a series of hypotheses or a theory.

 

 

4.2       Method and resources

 

No final decision about a research objective should be taken before a method is selected and its associated requirements are assessed against available resources. Such an assessment involves concrete planning, far beyond the choice of a general approach, so that precise needs in terms of resources are known: prospective PhD candidates should, inter alia, decide what background literature to explore, see where it can be found, what subjects and/or materials are relevant and where to find them, what precise method(s) is/are going to be used, calculate the approximate time required to go through the various steps of the study, and see whether all other required resources are available. In particular, the following important areas deserve to be highlighted:

 

4.2.1    Baseline expertise

If the method contemplated is theoretical discussion or development of a particular theoretical entity, researchers should make sure they have access to the relevant background texts, understand the languages in which these are published, and have enough expertise in the relevant field. For instance, the theoretical model of simultaneous interpreting adapted by Moser in her 1976 dissertation from another model by Massaro required theoretical knowledge in cognitive psychology. It is important to realise that minimum baseline expertise may call for much preparatory learning, up to several years of coursework in linguistics, cognitive psychology, neurophysiology, etc. Browsing through a few textbooks or papers is not enough to turn a novice into a genuine linguist, cognitive psychologist or neurophysiologist. Guidance through discussions and comments with and by experts is generally necessary to acquire a coherent, well balanced and correct perception of the theories, thinking modes and research methods that make up the relevant fields. The same applies to experimental design. Reading descriptions of various classical designs in textbooks is not enough, no matter how intelligent the students are. Implementing the reasoning process while avoiding the numerous traps requires hands-on guidance and correction until the skills are truly mastered.

            If prospective PhD candidates do not have the required baseline expertise, they have three options: acquire it, enlist the help of an expert, or choose another topic. If their supervisor happens to be an expert in the field, s/he may help decide whether acquiring a minimum level of expertise is feasible. Without such help, the decision to go ahead with the study is risky.

 

4.2.2    Sample representativeness

In any research endeavour which scrutinises a selected part of the entity to be explored, that is a sample of the population (the name traditionally given to the target entity), it is essential to know whether there is good reason to believe it is representative of the whole population, in which case it said to be a "representative sample", or whether it represents a sub-population that may differ in some relevant characteristic from the overall population. It is important to understand that representativeness is only loosely dependent upon sample size. For instance, women interpreters probably make up well over 50% of the population of interpreters, but may differ from the rest of the population (male interpreters) as regards social behaviour, self-image, personal ambitions, etc. In a study focusing on such parameters, a sample composed of female interpreters, be it 50% of the total population of interpreters, could only be taken as representative of  the female sub-population of interpreters, and any inference on the whole population would be justifiably criticised. On the other hand, a well-chosen sample of a few thousand people yields fairly accurate estimates of behaviour in a population of millions of people.

            In interpreting research, the lack of access to subjects and the limited volume of empirical research conducted so far make this issue particularly sensitive, because there is little data to help determine the limits of generalisability. To what extent and in what areas is it legitimate to extrapolate from one type of speech to another, from one language to another, from students to professionals, from a particular working environment to another?

            Two fundamental rules can be recommended in this respect:

            Firstly, use all available information to decide whether a sample of subjects to whom you have access can be representative of your population in respect of the phenomenon you are studying; if not, consider changing the focus of your study. For instance, suppose you are interested in the study of automated cognitive processes of a complex nature in interpreters, and it turns out that you only have access to student interpreters. Reading the literature and consulting with experienced researchers may make you decide that students cannot be taken as representative of the population of interpreters with respect to the focus of your study. In such a case, it does not make sense to go ahead with the project. A somewhat different situation occurs when you want to analyse the personality of interpreters, and find out you only have access to staff interpreters in an international organisation. An analysis and some reading of the relevant literature in sociology and psychology may make you decide that the personality of staff interpreters may differ from that of free-lancers. In such a situation, you can go ahead with your project, but with a narrowed-down objective, namely to study the personality of staff interpreters only.

                    Secondly, be cautious when generalising from your results. Clearly, in the project about the personality of staff interpreters, you can only generalise to staff interpreters, and possibly only to staff interpreters in certain types of organisations. The issue of generalisability is tricky: after all, when implementing a policy of scientific scepticism and caution in strict terms, one could argue that unless a strict random sampling procedure is used, something which is not feasible in the interpreting research environment (and in most behavioural disciplines), there is always the possibility that some hidden sampling bias exists, and that only claims about individuals in the actual sample studied are legitimate. The second best (and the only feasible) strategy in most cases is to consider all available evidence and reflections by fellow researchers, to decide how far one wishes to generalise, and to explain in one's text whether, why and with what reservations one could consider that the findings also apply beyond the 'generalisation perimeter'.

           At this point, and as long as no comparative data is available to help determine the limits of generalisability, such decisions will be largely subjective. Making them carefully on the basis of all available data and making the limitations clear to the reader (see Schjoldager in this volume) are the acceptable way to proceed.

 

4.2.3   Sample size

One other matter which deserves particular attention in interpreting research is sample size. In an empirical study, the number of subjects that can be observed and/or submitted to experimental procedures is an essential parameter. In a hypothesis-testing endeavour, random variation makes it difficult to contemplate any generalisation from very small samples, unless intra-sample variability is small. This is a matter of common sense, and mathematics only add precision to the principle: generalising on the basis of a 4-unit sample with measured values 2, 3, 3 and 2 is more reasonable than generalising on the basis of a 4-unit sample with measured values of 2, 3, 75 and 60. This is particularly critical if the idea is to compare two samples and decide whether they come from the same population, which is the basic idea in most experimental procedures (for instance, when testing student performance against professional performance, or performance when interpreting from an A-language into a B-language vs. work in the opposite direction).

            On the other hand, if the question is whether some mean value (such as a metric for quality, a physiological value, a proportion of a population displaying a particular feature, etc.) is higher or lower than some threshold, or even whether some mean value is 'rather high' or 'rather low', some generalisation is possible even in a highly variable sample. For instance, suppose the research question is whether names are truly problematic in interpreting, as is often reported in the literature, and correct name-rendition rates are measured in a random sample of subjects performing an interpretation task. If the measured values are 30%, 85%, 50%, 60%, 25%, in spite of the small size of the sample and its variability, it is clear that all subjects experience difficulties with names, and performing statistical tests will not contribute anything towards finding the answer to the precise question as defined above.

            The mathematics of probability theory and inferential statistics can give more generalisation power to findings if more is known about variable distributions, especially if these are close to the Gaussian curve (also called "bell-shaped curve" or "normal curve"). For further technical details and precise suggestions on sample size, read one of the many available books on practical statistics for the behavioural sciences, and/or consult statisticians.

            However, as long as no unreasonable generalisations are made, there is nothing wrong with studies on very small samples. Even case studies with single subjects are legitimate scientific endeavours, especially when they serve as exploratory studies. Many important scientific discoveries originated in case studies. In fact, Bromley (1986, quoted in Coolican 1999:126), has called case studies "the bedrock of scientific investigation", while Robson (1993) argues that experiments should also be considered case studies, which makes sense insofar as each is based on a single sample or set of samples.

 

4.2.4    The availability of a competent supervisor

For historical, sociological and institutional reasons (see Gile 1995), there are still few supervisors for research into interpreting who have both research expertise, especially as regards empirical research, and knowledge of the world of interpreting, and such supervisors with sufficient competence for the particular topic the candidate has chosen may not be available locally (the subject is discussed further in the context of interdisciplinarity in Section 6.1 and illustrated in several contributions in this volume). In such a case, candidates may decide to choose a topic for which there is a local supervisor, or to enlist the help of experts from other institutions as co-supervisors. Thanks to modern technology, it is even possible to work with people from afar (see Mead, in this volume). However, there may be institutional and personal obstacles: some universities may not allow supervision by scholars from other institutions, or the main supervisor may be reluctant to accept the very principle of co-supervision. It is also possible to enlist informally the help of experts from other institutions, but this is also associated with some risk if the official supervisor's views are at odds with the informal supervisor's.

 

 

5.    Strategies

 

Somewhat ironically, while doctoral training programs are said to teach graduate students what research is all about and how to go about it, in many cases, the PhD project, which is the culmination of the learning process, may differ significantly from research projects as they are conducted in the field. The essential reason for the difference is that while in research as practised by qualified scholars, projects generally aim at advancing towards solutions to research problems, PhD projects aim at fulfilling academic requirements, and are driven by them and largely constrained by the local academic environment and its norms. The two are not always compatible, and a choice may have to be made. Doctoral projects are a necessary gateway in academia and do not restrict their authors' freedom to engage in any type of research they choose after obtaining their degree. It may therefore be best, when students have long-term academic objectives, to give priority for a time to the specific requirements of the doctoral project rather than to norms of research as it is practised more generally in the field. This implies in particular specific strategies relevant to the selection of topics:

 

 

5.1       Feasibility as a priority

 

While research proper in a PhD project must be in line with the prevailing standards of research, as mentioned above, the fundamental aim underlying a PhD project is the acquisition of  the degree. Consequently, the candidate's priorities are not quite the same as those of non-degree research. More specifically, adapting to environmental resources and constraints may take precedence over the quest for a solution to a specific research problem, and candidates would be well-advised to consider the possibility of choosing a  topic most likely to facilitate their doctoral work rather than a topic closer to their research interests but more difficult to tackle at the PhD stage in their local environment. In particular, as already mentioned, choosing a topic in line with the expertise and interests of a locally available supervisor may be a wise strategy, although before doing so, candidates should attempt to predict how they will react to several years of work on a subject which may not be their favourite.

 

 

5.2       Alternative routes

 

Planning alternative routes is another strategy specific to research-for-a-degree as opposed to ordinary research. Although some research projects are straightforward, when innovative or original elements come in, problems and outcomes are somewhat unpredictable in spite of careful planning. This is why it is wise to plan not only the implementation of the initial action plan, but also alternative routes, which become increasingly important as the project advances. If, during the early stages of a study, PhD candidates discover that access to background literature is more difficult than anticipated, or that their project requires baseline expertise that they do not have and do not have the time to acquire, they can still choose a different direction for their study without incurring dramatic loss of time and other resources. If, however, they find after much investment that their material is not suitable for their project, or that there is too much variability in their data to find the answer that they had sought, a radical change may be much more painful and/or costly.

            Alternative routes should be planned rather than improvised, the idea being to use to the largest possible extent the work already done at the time the change in direction is made. This means in particular thinking of several scenarios that will be using to a large extent the same background literature, the same material, and the same sample of subjects. Without such planning, there is a risk of having to discard much of what was achieved before the insurmountable obstacle was found.

 

 

5.3       Determining the scale of the study

 

The issue of alternative routes is linked to another question, namely how and when to determine the scale of the study. Theoretically, careful planning at all levels is highly recommended, and in the traditional experimental paradigm, the scope of the study, including the number of experimental groups and their respective sizes, as well as the corpus to be used (most often source speeches), is supposed to be determined in advance. However, in the field of interpreting studies, especially in the case of time-constrained doctoral dissertations, this is not always possible, or indeed wise, simply because of the difficulties involved in finding authentic speeches and enlisting the co-operation of professional interpreters for research purposes. Determining in advance in a non-modifiable way the scope of the study in such terms may entail the risk of failure, if the sights are set too high, or of missing opportunities for collecting valuable information, if they are set too low. An open-ended design with minimum sample sizes and materials which can also accommodate further materials and/or information may lead to better results.

            One example is a study by Gile (1999) - not a doctoral project - on variability in the perception of fidelity, which involved several steps and an incrementally growing number of participants. The idea was to check variability in the perception of interpreting quality by various groups, including interpreters and non-interpreters, and depending on the mode of presentation of the target speech (audio recording or transcript). Data on reactions in the written mode (the transcript) was obtained incrementally by questionnaire, as interpreters and other subjects became available. Data on reactions in the oral mode (the audio tape) had to be collected on a large enough sample using a method and environment ensuring identical conditions for all subjects. The opportunity presented itself when I was invited to give a one day workshop on research into conference interpreting at an interpreting school. I thus had enough time for the experiment and a large sample of participants - it goes without saying that deciding to wait for such opportunities is only an appropriate strategy if the researcher knows that such opportunities will present themselves within an acceptable timeframe. One reviewer of the paper reporting the study, who had a traditional experimental background, commented that the study seemed to be improvised, rather than planned. In fact, precise planning and the preparation of materials was necessary in order to seize the opportunity when it arose. Admittedly, there was some uncertainty as to when exactly it would arise and what the precise size and composition of the sample of subjects would be, but a close-ended study could probably have been conducted only at a much smaller scale.

            Sharing corpora and using the same speech and some findings from completed research for further studies is one way of increasing the amount of available, comparable data. It is also, in a rather loose sense, a way of increasing 'sample size' (but not in the strict sense, since the data may be available from a set of distinct samples, with differences in the task and/or environmental conditions under which it was obtained). The open-ended design is legitimate and efficient, though one must be careful to identify the differences between conditions in the various incremental stages so as to avoid making unjustified inferences.

 

 

5.4       Canonical and non-canonical steps in the research process

 

In general accounts of science, the point is made that research is based on a theoretical foundation, insofar as one or several theoretical account(s) of a phenomenon or set of phenomena are studied and then either developed (in theoretical research) or tested (in empirical research). In actual life, research often focuses on a pragmatic problem, without an extensive theoretical basis: Is there a difference between the expectations of various user groups as to interpreting quality? Do interpreters work better from an A-language into a B-language or vice-versa? How useful are specific conference preparation methods for interpreters?

            While some form of "theory" in a very general sense of the word is always present behind the very questions, in empirical studies, it need not be as comprehensive, cohesive and well-articulated as a theory presented as such in a scientific text. In the humanities, and in particular in a humanities-oriented view of Translation Studies (as opposed to a social-sciences-oriented view of Translation Studies), the prevailing norms seem to require from doctoral dissertations a substantial introduction, at least one chapter long, simulating the researcher's route from theoretical considerations to the precise research question(s). While it is necessary for PhD candidates to explore the theoretical background to a research topic, as well as previous empirical studies and their findings, making a final selection of the topic on this sole basis may not be the most efficient strategy, because feasibility issues may then force the student to backtrack. It is often better to start with one or several tentative research questions, do some planning and check feasibility and previous work on the subject, and defer the drafting of the detailed literature-based introduction to a later stage, so that it can introduce the reader to the actual research question tackled. In other words, while the way the dissertation is structured seems to suggest that its author started out with theoretical exploration of the issue, developed a set of hypotheses and went on to test them, the actual order in which the research project was carried out may be quite different.

 

 

6.   Further issues

 

6.1       Interdisciplinary work

 

The idea of interdisciplinary work (meaning, in this context, work on interpreting involving a substantial input from an adjacent discipline such as psychology, linguistics, sociology, etc.) has become very popular in the last decade or so, and many prospective PhD candidates are attracted to the idea of working on cognitive or neurophysiological topics. While a number of literature reviews and simple empirical studies at MA level have been completed successfully, at PhD level, feasibility issues are more problematic and should not be underestimated. In particular, in the cognitive paradigm, experimental design at PhD level needs to be very strict (see for example the sophisticated design of Dillinger 1989 or Liu 2001), and it takes specific training or much experience to acquire the required expertise.

      In this connection, the following points deserve special attention:

-     In principle, doing interdisciplinary research at PhD level requires the input of supervisors or co-supervisors from the relevant adjacent discipline. Such supervisors may assume that the students have much better knowledge of that discipline than is actually the case, be reluctant to coach PhD candidates into the acquisition of basic knowledge or skills which they should normally have at their fingertips at that stage, and/or require them to attend many courses to bring them up-to-date; some of these courses may seem irrelevant or too remote from interpreting-related matters. Candidates for interdisciplinary projects should be aware of these possibilities and decide whether such requirements are acceptable to them.

-     In some cases, especially when students are experienced interpreters themselves, supervisors from adjacent disciplines may trust them to 'do the right thing' without actually supervising them, and be too lenient when the completed work is presented to the assessment committee, because of these students' special status as 'outsiders' and as experienced interpreters. This may result in weaknesses both as regards interpreting proper (due to the supervisor's insufficient knowledge of the world of interpreting and of the associated literature) and as regards the adjacent discipline.

-     There may also be differences between the student and the supervisor when choosing metrics and when interpreting the data. For instance, in the interest of rigorous measurement, non-interpreting researchers may insist on using formal textual criteria such as propositional matching between source speech and target speech to assess fidelity, without taking into account interpreting strategies. In experiments having performance as the dependent variable, this can be very problematic.

 

It is therefore particularly important, when contemplating an interdisciplinary doctoral project, not only to make sure that a supervisor from the relevant discipline is available locally, but also that s/he is mentally available and willing to help over the whole duration of the PhD work, and to ascertain what his/her attitude and requirements are. When the dissertation is completely within the paradigm of the adjacent discipline (as was the case in Padilla's 1995 dissertation on the interpreters' memory span), things are relatively straightforward, but as soon as genuine interpreting work comes in, problems arise easily. On the whole, interdisciplinarity seems less problematic with linguistics and sociology, with their strong descriptive and naturalistic component, than with cognitive psychology, which tends to favour 'hard' experimental procedures (see the comments in Shlesinger 2000).

 

 

6.2       Scientific norms

 

Another issue which is far from trivial has to do with the selection of scientific norms for interpreting research, since these are diverse and evolve over time. In the mid-seventies to mid-eighties, the prevailing interpreter-researchers' norms (see Gile 2000) were associated with essay-type explorations of the applicability of the prevailing paradigm (in the form of "why the théorie du sens also applies to Korean" - see Choi 1986), as opposed to strictly experimental studies or to theoretical development in the strict sense such as done in Moser (1976), in Mizuno's work on a 'dynamic model of simultaneous interpretation' (1993) and in Setton's pragmatics-based theory (1997). Towards the end of the eighties, a shift occurred, with a strongly interdisciplinary trend, with respect not only to the topics and methods used, but also to scientific standards. As mentioned above, the now central position of cognitive psychology among the proponents of interdisciplinary research is associated with a strong position of the experimental paradigm in the strict sense, and within the ranks of interpreter-researchers, there seems to be a strong feeling that experimental research is the only one which truly meets the requirements of Science.

                    The experimental paradigm certainly provides a relatively powerful rationale to deal with confounding variables (variables which exert influence over the phenomena under investigation but which are not controlled by the experimenter). However, it also entails a number of drawbacks other than the well-known issue of ecological validity (to what extent do laboratory conditions reflect the genuine phenomena they are supposed to measure?), and is being challenged from within the ranks of cognitive psychology - for an interesting discussion of the issue, and the presentation of an alternative qualitative approach, see Coolican 1999.

                    Besides this fundamental rationale, practical reasons such as the lack of access to a sufficiently large number of subjects, the lack of knowledge about variable distribution, the lack of knowledge about inter- and intra-subject variability, the lack of validating evidence for performance metrics, etc., may erect severe methodological and feasibility obstacles in the way of the conventional hypothesis-testing type of experimental paradigm in the world of conference interpreting, and reduce considerably its potency.

                    It is important to remember that the experimental paradigm is by no means the only legitimate one in science, and that naturalistic paradigms are standard in many disciplines acknowledged as scientific (including astronomy, geology, botany,  zoology, epidemiology, etc.). As regards behavioural disciplines, and in particular cognitive psychology, quoting from Coolican (1999) again, "it is perfectly possible to test hypotheses without an experiment. Much psychological testing is conducted by observing what children do, asking what people think and so on." (p.12). He explains that "scientists are pragmatic in their progress of knowledge, rather than following the rigid code which is often presented as the ideal to work towards in psychology methods courses" (p.197). Allport, a promoter of psychology as a conventional science, explains that "We should adapt our methods so far as we can to the object, and not define the object in terms of our faulty methods" (quoted in Coolican 1999:198). Going further into the critique of the conventional experimental paradigm is a feminist perspective which sees it as

 

characteristic as a male approach to research…: preoccupation with quantifying variables; an emphasis on control, mastery and manipulation; a tendency to remain distant rather than be involved with the subjects of research; a preference for gadget-oriented research over naturalistic enquiry; competition and ego building. (Coolican 1999:207).

 

                    Without necessarily going as far in the critique of the experimental paradigm, PhD candidates should realise that some margin of freedom in the choice of their scientific norms is legitimate. Unless forced to do otherwise for practical reasons, for instance when enrolled in doctoral programs which impose the conventional hypothesis-testing experimental approach, PhD candidates should ideally be free to choose one out of several scientific paradigms, to be determined according to their own tastes and as a function of feasibility - with the proviso that their underlying rationale is solid. In that respect, the paradigms of sociology, with their emphasis on naturalistic studies, which are not less rigorous than experimental studies by nature, may provide a good reference to present to reluctant prospective supervisors.

 

 

7.   Conclusion: preparing students for research topic selection

 

As can be seen, selecting a topic for a PhD project is not to be seen as a short, straightforward preliminary to the study proper. Rather, it is a planning process, sometimes complex, often lengthy, which may last well into the actual literature review, data collection and even data analysis stages (as illustrated by Dubslaff's and Schjoldager's accounts in this volume), sometimes forcing a change in direction somewhere along the line (see Dubslaff, in this volume). Optimising the process requires much reflection, as well as a flexible and realistic attitude. In interpreting research, PhD students are often experienced interpreters who do not need the doctoral degree for their career and who undertake their PhD project with the purpose of solving an interpreting problem or finding an answer to a question. This may make it more difficult for them to come to terms with the reality of research, and with the idea that their project will probably not bring forth the ultimate solution or answer.

            Is the difficulty of the topic selection process universal? It is obviously a critical process, and my own experience as a supervisor and the experience of fellow researchers in several other disciplines, including translation studies, literary studies and linguistics, with whom I have discussed the issue, seem to suggest that it may well be difficult for many students. In established empirical disciplines, the process should be easier. While studying for a second degree, and even for an undergraduate degree, students are introduced to research methods, and their study of the evolution of concepts and theories in their discipline makes them analyse closely and critically the designs and methods used throughout its history. Often, they also have to design and conduct their own empirical projects in the course of their studies. When their turn comes to select a topic for a PhD, they already have a basis for their planning and decision process. Moreover, in such disciplines, a number of research paths are well charted out and validated, and there are many qualified supervisors to guide the students in the appropriate direction.

                    In interpreting research, where students often have to chart their own course and do not have the benefit of a syllabus which prepares them naturally to research, besides training in critical reading (see Gile's other paper in this volume), specific training in topic selection/research planning is probably a useful component of coursework for beginning researchers. Students are presented with a research question or topic and asked to plan a course of action for a PhD project that would seek to explore it and find a specific answer or solution, listing all expected difficulties, available resources, possible strategies, alternative routes, etc. Their analysis and design are then discussed in the classroom, and compared to actual research done on the same topic or question. In the absence of such courses, students should be encouraged to seek as much help as possible from their supervisor, as well as from other experienced researchers.

 

 


References

 

Bromley, D.B. 1986. The Case Study Method in Psychology and Related Disciplines. Chichester: Wiley.

Choi, Jungwha. 1986. L'interprétation consécutive coréen-français du point de vue de son enseignement. Unpublished doctoral thesis, ESIT, Université Paris III.

Collados Aís, Á. 1996. La entonación monótona como parametro de calidad en la interpretación simultánea: la evaluación de los receptores. Doctoral dissertation, Universidad de Granada.

Coolican, H. 1999. Research Methods and Statistics in Psychology. London: Hodder & Stoughton. 3rd Edition.

Dawrant, A. 1996. Word Order in Chinese-English Simultaneous Interpretation: An Initial Exploration. Unpublished M.A. thesis, The Graduate Institute of Translation and Interpretation Studies, Fu Jen Catholic University, Taipei.

Déjean Le Féal, K. 1978. Lectures et improvisations - Incidences de la forme de l'énonciation sur la traduction simultanée. Unpublished doctoral dissertation, ESIT, Université Paris III.

Dillinger, M. 1989. Component Processes of Simultaneous Interpretation. Unpublished doctoral dissertation, McGill University, Montreal.

Gile, D. 1989. La communication en réunion multilingue. Les difficultés de la transmission informationnelle en interprétation simultanée. Unpublished doctoral dissertation, Université Paris III.

Gile, D. 1995. Regards sur la recherche en interprétation de conférence. Lille: Presses Universitaires de Lille.

Gile, D. 1999. "Variability in the perception of fidelity in simultaneous interpretation". Hermes 22. 51-79.

Gile, D. 2000. "The history of research into conference interpreting: a scientometric approach". Target 12:2. 299-323.

Lamberger-Felber, H. 1998. Der Einfluss kontextueller Faktoren auf das Simultandolmetschen. Eine Fallstudie am Beispiel  gelesener Reden. Unpublished doctoral dissertation, Karl-Franzens Universität Graz.

Liu, Minhua. 2001. Expertise in simultaneous interpreting: a working memory analysis. Unpublished doctoral dissertation, The University of Texas at Austin.

Mizuno, A. 1993. "Doujitsuuyaku no doutaimoderu ni mukete". ("Towards a dynamic model of simultaneous interpreting"). In Conference Proceedings of the 4th International  Japanese/English Translation Conference. University of Auckland, Brisbane, Australia. 421-443.

Moser, B. 1976.  Simultaneous translation: Linguistic, Psycholinguistic, and Human Information Processing Aspects. Unpublished doctoral dissertation, University of Innsbruck, Austria.

Padilla, P. 1995. Procesos de memoria y atención en la interpretación de lenguas. Doctoral dissertation, Universidad de Granada.

Pinter, I. 1969. Der Einfluss der Übung und Konzentration auf Simultanes Sprechen und Hören. Unpublished doctoral dissertation, Universität Wien.

Pöchhacker, F. 1994. Simultandolmetschen als komplexes Handeln. Tübingen: Gunter Narr.

Robson, C. 1993. Real World Research. Oxford: Blackwell.

Setton, R. 1997. A Pragmatic Theory of Simultaneous Interpretation. Doctoral dissertation, The Chinese University of Hong Kong.

Shlesinger, M. 2000. Strategic Allocation of Working Memory and Other Attentional Resources in Simultaneous Interpreting. Unpublished doctoral dissertation, Bar-Ilan University.

Thiéry, C. 1975. Le bilinguisme vrai chez les interprètes de conférence professionnels. Unpublished doctoral dissertation, ESIT, Université Paris III.

Tommola. J and Hyöna, J. 1990. "Mental Load in Listening, Speech Shadowing and Simultaneous Interpreting: a Pupillometric Study". In Foreign Language Comprehension and Production. AFinLA Yearbook 1990, Tommola, J.(ed). 179-188. Turku: Publications de l'Association finlandaise de linguistique appliquée AfinLA.