The research skills required for scholarly investigation into translation/interpreting issues when empirical methods are involved.


Delia Chiaro



Choosing topics

Over and above the research paradigm in which a student intends to work, in my personal experience, choosing a topic often appears to be easier said than done. In fact, at the outset of their research projects, most students tend to have very vague ideas regarding what they actually want to do, let alone how they are going to do it. In other words, at the first meeting with me as a potential supervisor, a typical postgraduate student will tell me that s/he wants to work on anything from bilingual children to translating Monty Python or cross-cultural irony. Although they will have already chosen a topic in which they are often passionately interested, over the years, I have developed the firm impression that their idea of research can be summed up in “I want to find out everything there is to know about x ”. This is all well and good, and indeed part of a correct research process within (presumably) either paradigm, but I would argue that it is, nevertheless, one step away from real research. While being an important part of the research process, library research is not the whole picture. Given that true research involves originality, background reading should be step one in the entire process, a stepping stone, a basis for new ideas. At university, and especially at postgraduate level, more than this is required.

            Directly connected with a certain lack of focus in choosing a project, is what I would like to label the “is it doable?” factor. In the same way that, initially, the average student in translation studies sees research as a potted compilation of everything written on a certain subject, or at best a comparative study of some kind of source and target text,  once guided towards a concrete research question, the next step is to direct him/her towards a realistically achievable objective. Students embarking on empirically based projects may often know very little about the subject in question and the complexity of working within this paradigm. Their goals are often somewhat unrealistic. Furthermore, if, as is usually the case, the student lacks the necessary research tools, then actually being able to develop the project of his/her choice may not be feasible. Thus the task of the mentor is to mediate with the young researcher without enforcing his/her own ideas. And this task is not easy either. An established researcher may well have very clear ideas of how a piece of research is to be carried out yet s/he must seek to strike a balance by guiding, rather than coercing, the student towards a task which is reasonably doable within constraints such as time and money available. The task should be simple enough for the student to see it through, but at the same time the venture should stretch his/her capabilities to recognize and become skilled at what are usually totally new ways of doing research.


Planning one’s study

A formal research design requires clear specification of the what, why, when, who, where and way (the six Ws) of research.[1] After deciding what s/he is going to research, what information is being sought, the reason why the student is conducting the project should follow naturally. Empirical research should in some way be relevant to the wider community beyond academia. Of course anything can be investigated for the sake of intellectual exercise and woe betide a world in which enquiry were limited to issues of direct and immediate relevance yet, by the same score, in the particular case of translation and interpreting studies, there is indeed a great need for new information regarding numerous translational circumstances occurring world wide which are of direct bearing in the world in which we live. In which way is this information to be obtained? The student can choose from qualitative methods such as interviews, focus groups and observation or quantitative methods such as experimentation and surveys. Whether the student opts for one of these single methods or a combination of more, s/he then has to decide upon where to gather data, when to do so and who from.

First and foremost students need to focus on a question or set of questions which they must set out to answer through their research. In order to get them thinking, I ask them to reflect upon the meaning of the word ‘thesis’. According to Merriam-Webster’s a thesis is “a position or proposition that a person advances and offers to maintain by argument ” or a “proposition to be proved or one advanced without proof”. At this point I ask them to consider these definitions and to try and express exactly what they are proposing; what they want to find out and whether they have any hunches regarding what they will discover. After a short brainstorming session, I get them to formulate their ideas into a ‘proposition’ or a ‘thesis’. By doing so, the typical student who arrives in my office wanting to say, compare the original version of Monty Python’s The meaning of life with the dubbed Italian version Il senso della vita  will have transformed a vague notion of a dissertation into a thesis that will explore, for example, the impact of dubbing on the response of Italian viewers to the film.[2]

            Once the student has focused on the right research question, s/he needs to plan how to find the right answers.[3] In this particular case, the student decided to set up an experiment basing himself on Paul Eckman’s Facial Action Coding System (FACs) which codifies facial expressions according to emotions, to explore how far the expressions of Italian viewers differed from those of English speaking viewers when watching the Python’s film. In order to do this we decided to record the facial expressions of a group of Italian speakers and a group of English speakers while watching the film to be able to see to what extent their expressions converged, the working hypothesis being that the Italian viewers would laugh and smile less than English speaking viewers. This would raise the issue of translational impact on the humour response. At this point, a whole string of both practical and theoretical questions arose. As far as practicalities were concerned, problems ranged from whether to show the whole film or just some clips, to where to actually record the people who would volunteer to take part in the experiment. Talking of which, where would these volunteers come from? And how many would we need? Where would we perform the experiment? And should we tell our ‘guinea pigs’ what we are looking for? Furthermore, surely people will tend to accommodate to one another, thus if a group are watching the film together they may well smile and laugh more than they would if they were watching it alone. So, as we can see, empirical research requires much foresight and planning as, I would argue, it requires more of what I would like to label physical input on behalf of the researcher compared to the very different kind of planning required in the Liberal Arts Paradigm. And the task of the mentor of such a project is to use his/her experience to tailor it to the abilities and above all, to the true possibility of realization of the task by the student – in other words, how can I help him/her make it be ‘doable’ over and above the many constraints s/he will have to face?

            So, let’s take a look at a few of these practical problems. Following experimental-style theses on cross-cultural perception of filmic texts (films, sitcoms etc.), the first issue which arises is usually “Do I need to show my guinea-pigs/volunteers the whole film/programme or just a few clips?”. In an ideal world, we would of course show the whole film so that viewers are able to contextualize what they are watching. But one of the major constraints of empirical research is the time factor. In today’s busy world not many people will be willing to sit down for an hour or two to watch a film and then be tested on it in some way. Thus, students in perception based empirical work in screen translation will tend to opt for showing a series of clips rather than the complete programme. But then, how do they choose which clips to show? And this is where the issue of bias and subjectivity raises its ugly head. If we are testing the humour response (i.e. laughter and smiling) cross-culturally via translated products, we are immediately tempted to include examples which make us laugh. And of course this will not do. We need to see what makes the non-expert in humour, the non-linguist, the non-translator laugh. Not only, but the researcher really should try his/her best to keep his/her feelings at bay. Now, in a full blown research project with funding behind it, at this point, we would most likely carry out pretty extensive qualitative research in the form of interviews and/or focus groups to try and identify the scenes in a certain product where people laugh most. The scenes which would make the largest number of respondents react positively would then be utilized in the experimentation proper. And this raises question number two. How do we decide what “laughing most” and “react positively” mean? Is a smile the same thing as a burst of laughter? Not only, but people have different personalities and unpredictable moods, how do we know they are not laughing at a stimulus because of a bad mood rather than because they don’t find it funny. Vice-versa, they may be laughing more because they have just won the lottery or fallen in love. And lots of people can be highly amused but show little expression. So, how is the researcher to deal with all this? Well, firstly s/he needs to establish operational definitions. For the sake of this research project s/he will need to decide what qualifies at laughter and smiling. If the definition is too loose, then s/he will see laughter and smiling occurring everywhere leaving him/her with a huge amount of unmanageable data. If, on the other hand, an operational definition is too limiting, s/he may end up with few or no instances of what s/he is looking for. As for the state-trait variable (i.e. respondents’ personalities and mood at the time of the experiment) inherent to the respondents, this is an important factor to be borne in mind before drawing far-reaching conclusions from the results of the study.

            The next big problem regards our guinea-pigs, our volunteers – our sample. Where do we find them? How many people do we need? Who will give me the input for which clips to use? The easiest thing to do would be to encourage the student to use family and friends, which, at a stretch could be feasible in an undergraduate dissertation as long as s/he is aware of the fact that such a sample is extremely convenient and points this out in his/her writing up. Again, ideally, respondents in such research are usually paid or compensated for their inconvenience, money at student level is a significant constraint. And let us not forget that translation is, by default, cross-cultural. We need respondents from two cultures, a control group from the source language culture. Thus, either this involves the researcher traveling, or hunting out speakers of language x from within the home environment. Neither option is simple, cheap or time efficient. And let’s not forget that we needed a sample to help us identify the clips to use in the first place and a fresh sample for the testing proper. Usually students choose volunteers from their own social network, and I always advise them to deliberately exclude students of translation and interpreting in order to try to reduce the bias. Finding a control group of English speakers can be more difficult, but with some imagination, this is never impossible. As for sample size, true research requires time, energy and money, the last of which is typically lacking in students. In my experience, when possible, I have attempted to fund students to, for example, travel abroad to collect ample samples of respondents. And this can be worthwhile if the research design is well structured. In other words, it is often a real pity to throw away a well thought out project on a small, local sample. Having said that, however, let’s not forget we are working with students, thus exploratory, pilot studies should be perfectly acceptable – as long as we are aware that they are simply pilot studies and will thus not result in gospel truth. In survey work, for example, I normally get students to work with a minimum of 30 respondents, and preferably with multiples of this number.

If we are filming respondents, we need to have their permission before doing so. But how much can we tell them of exactly what we are looking for? Surely if they know we want to see whether or not they smile and laugh they may deliberately control themselves and skew the experiment. In the case of Monty Python, the student told his respondents a little white lie. He said that he was recording their hand gestures to see whether it was true that Italians gesticulate more than Anglo-Saxons. All the respondents believed him.

Finally, students working with a quantitative framework need to know how they will test the data which they have gathered for significance. At the outset they should have a clear idea of which statistical test they intend to administer.


Reading the literature and/or collecting information from other sources

Reading the literature and collecting secondary data should follow the same rules within either paradigm. However, as translation students tend to have little or no know-how of empirical research methods, included in the “reading up” stage, at least one research manual will be included..


Reasoning on the information available

If a student has followed the six W process, s/he should have a clear idea of what s/he is looking for in the data. But how is s/he to reason with the new information that has been obtained? This information will either be qualitative or, in the case of quantitative research, in the form of numbers. In the latter case, the student should already know how he/she will test his/her initial hypothesis and how to pass from the sample to the whole of the population. Do we want to examine difference or correlations in our data? If we are analyzing in descriptive terms, are data distributed normally or are they deviant? Is our sample big or small? Are our samples dependant or independent? These questions guide the student towards the best statistical test needed to answer the initial research questions. The test adopted will depend on a series of factors such as the nature of  variables (ordinal, nominal, continuous etc.) the type of sample (independent or dependant) and the distribution of data (parametric vs non-parametric) although the latter can only be known after having gathered our data. In other words, the researcher has to examine data to decide whether to use para-inferential statistics or non parametric. For example, if we need to verify differences between the mean of two independent samples, we can use a t-test for differences. However, the important thing is, that the statistical test should be decided upon before data collection.


Making inferences

In quantitative research, our results will tell us whether to accept or reject our initial hypothesis. Usually we want our initial research hypothesis to be true, so we want to reject the null hypothesis. Depending on our statistical test, we make a decision. Each test has its own characteristics which the researcher should be familiar with a priori.


Writing one’s paper/thesis/dissertation

The rules of writing up a piece of work based upon empirical methods are pretty strict. The empirical genre requires that a dissertation or essay begins with the presentation of aims and objectives of the research, followed by a detailed and description of tools and methods. Next follow the  results and a critical discussion of these results followed by a conclusion. As in the liberal arts paradigm, a review of the literature is also necessary, plus references to all previous similar surveys and/or experiments.

            However, students who have often carried out a survey or experiment extremely well, will have problems writing it up. Normally, they tend to be unable to disentangle a set of simultaneous processes that they themselves have carried out, into what should be a streamlined narrative which is reader-friendly and perfectly clear. In my experience, on first draft, they have problems with moving from the general to the particular and zig-zag around in bottom-up confusion. Writing up an empirical piece of research should leave nothing implicit. One of the academic loops to jump we must jump through is that of crossing every ‘t’ and dotting every single ‘i’. To this purpose, kKeeping a research log can be extremely useful. At the outset of their research, I advise students to buy an exercise book and number the pages in ink (this discourages “cheating” by ripping out pages). I ask them to write down everything they do in the order in which they do it. They are told to include the smallest detail – even if it might seem insignificant. By recording every single step they take, not only will they have fewer problems in remembering all the steps of the methodological process but they also have a ready made order from which to write up their methodology chapter. They also leave nothing out. I also encourage students to jot down any ideas relative to the project, including things such as, what findings could mean. Examination of a log can reveal the evolution of the entire research process, and, in the case of error, can even reveal its source.



[1] See Naresh  K. Malhotra, Marketing Research : an applied orientation, 2nd ed. (New Jersey: Prentice Hall, 1993), p.91; Harper W. Boyd, Jr.; Ralph Westfall and Stanley F. Stasch, Marketing Research :text and cases, 7th ed. (Homewood IL: Richard D. Irwin, 1989), p.286 and Gilbert A. Churchill, Jr., Marketing Research, 5th ed. (Chicago: Dryden Press, 1991), p.145.

[2] I would like to thank Matteo Camporesi for allowing me to use his research project he undertook for his undergraduate dissertation in order to exemplify the various problems faced in carrying out empirical research in translation studies.

[3] Space does not allow me to explore the issue of what the ‘right’ research questions are, however, the minimum requirement would be one of relevance i.e. a good research question will be able to respond positively to the “So what?” factor.