Large blue keyboard button, inscribed: 'translate'

Thinking about how to present multilingual data

So far in this series of posts on doing multilingual research, I have probed the intricacies of multilingual research settings, presented some dilemmas about obtaining informed consent, and talked about language choice and data generation. This post moves the discussion forward to the most visible outcome of a multilingual research project: How might one present multilingual data on a research report or article?

The short answer is: it depends. Rather than suggest a representational strategy that must always be used, I would argue that the representational choices one makes should be informed by reflection about the specifics of one’s research project. In the paragraphs that follow, I will illustrate such a process of refection, by drawing on my own PhD study.

The challenge

My research was a case study of a ELT language school in Greece, so in the process of doing fieldwork, I collected data in both Greek (the mother language of most participants) and English (the working language of the school). As I was proficient in both languages, I was able to analyse the data without translating them. However, in the interim reports that I sent to the University, and in my thesis, these data had to be translated to English, for obvious pragmatic reasons. From this stipulation, a number of questions emerged, such as:

  • How literal should my translation be? Literal translations sounded stilted, but taking too many liberties in translation would add a layer of opacity to the data.
  • Should the original be presented alongside the translation? What advantages would such a representational choice offer? Might there be any hidden complications involved?
  • If I decided to present both translation and the originals, how would the texts relate to each other? Would the original versions of the extract be consigned to an appendix? Or should the appendix house the translations? Should the texts be presented side by side, and if so, which language should be presented first?

Developing a representational strategy

I first thought that it would be possible to solve these problems by consulting the University regulations, or the APA style manual, or maybe by referring to the literature in order to find out what standard practice was. These strategies proved less than helpful. Although my study was by no means unique, and perhaps not uncommon, there seemed to be little guidance available. More importantly, the proposed strategies that seemed to work well for some parts of my dataset, seemed less suitable for others.

Soon enough, I decided to develop a tailor-made representational strategy, by going back to first principles. This involved finding a way to reconcile four sets of considerations: political, theoretical, practical and ethical.

Political considerations: countering linguistic hegemony

There is a vibrant discourse in the literature regarding the ways in which the global spread of English is threatening the viability of other languages, and this discourse seems to have extended to the role of English as an academic lingua franca. The need to promote the visibility of languages other than English in academic discourse, as a means of counteracting this trend, is -I think- uncontroversial. For me, this meant that it was important to preserve the Greek data in some form in my thesis.

But this led to another question, namely, which variety of Greek should I make? Sometimes, my research participants would use pronunciations and lexis that were particular to North-Western Greece, as a way of indexing our shared heritage, and in order to show that I was accepted as a member of their in-group (perhaps also as a subtle reminder that I was expected to behave as ‘one of them’?). While I would have liked to faithfully transcribe such linguistic cues, I was also conscious that Modern Greek is highly standardised, and that regional varieties tend to be associated with lack of education. I did not want to risk stigmatising participants by making them ‘sound like villagers’, but I also had to reflect on whose standards were being enforced by my representational choices.

Theoretical considerations: preserving the participants’ voices

A second set of considerations related to the way I understood language. From the theoretical perspective that informed my study, language was not a ‘thing’ that exists independent of context. Rather, it emerges from local situations, it is influenced by dynamics of interaction that are particular to those settings, and it is intended to cause specific effects within them .

When a text is picked out of this context, and the ‘voice’ of the original is replaced by that of the translator, all the above changes. A successful translation, my reasoning was, should perform a similar function to the original in a different ecology, so (paradoxically?) it should be different from the original. In a sense, this is similar to the way that a prosthetic limb is functionally similar to what it replaces, but it need not be an exact replica down to the level of blood vessels, sinews or hair follicles.

It followed that it would be epistemologically problematic to deny my readers’ access to the participants’ voices. But it also meant that I could not expect to create ‘perfectly accurate’ translations. Moreover, I began to realise that the goal should be functional equivalence.

Practical considerations: the challenges of translation

Added to the lofty considerations of politics and epistemology, I had to consider was the nitty-gritty of translation methods. Far from being accurate renditions of the originals, translations are selective and contingent texts, and I would like to illustrate this by reference to an example.

Slide11
Figure 1. Literal and functional translations

Figure 1 shows a description, by a student, of a typical lesson (top). On the left-hand column, I have presented a literal translation, and the right hand column presents my rendition, as it appears in one of my interim reports. Stylistics aside, in the space of these few lines I had to make several interpretative decisions. For instance, the phrase ‘we say the lesson’ («λέμε το μάθημα») is semantically ambiguous: taken in isolation it could refer to an oral examination (i.e., ‘we are asked to recount what we were taught in the [previous] lesson’) or to a presentation of new material (as in: “we go over the next few pages in the coursebook”).

In this instance, I was confident in choosing the latter interpretation because of the co-text. This interpretation was consistent with my knowledge of the research setting, and my experience as a person who was educated and has taught in Greece. Still, this was a personal interpretation, and it was important to give readers the opportunity to make alternative interpretations, or at least to help them understand where my interpretation was derived from. To that end, I felt it would be helpful to present original forms alongside the translations.

Ethical considerations, or ‘do no harm’

The last set of considerations that weighed on my mind pertained the principle of non-malfeasance (‘first, do no harm’). In the previous post, I already hinted at the the implications of how the English language teachers were portrayed, when they used non-standard forms. The same line of reasoning could be extended to students, as seen below:

Slide18
Figure 2. Unedited vs. standardised data

Figure 2 shows a response given, in Greek, by a learner who seemed unenthusiastic about language learning. The student used a phonetic spelling of the word “βαριέμαι” (: I’m bored), which violates the orthographic expectations of Modern Greek in a somewhat amusing way (top). It would be possible to render this answer accurately, and attempt to recreate the spelling mistake in the translation (middle). However, the analytical advantages of such a choice seem to be outweighed by the risk of stigmatising the student. In such a case perhaps a standardised rendition of his answer (bottom) would be a better option.

So, what is one to do?

Taken together, all the considerations above posed a number of challenges for representation. Some considerations seemed to privilege maximum transparency, whereas others seemed to suggest a need for measured opacity. In the the next blog post in this series, I will focus on some possible solutions, and describe a ‘cline of representational positions’ which can be used as a framework for developing a representational strategy.


Featured Image Credit: Pixabay (Public Domain). Note: this post draws on a paper I presented in the Researching Multilingually seminar in Manchester in April 2012.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s