The Routledge Handbook of Translation and Cognition

Fabio Alves, Arnt Jakobsen, Fabio Alves, Arnt Lykke Jakobsen

• Publisher: Routledge
• Year: 2020
• ISBN:
  9781351712453
• Edition: 1
• Topic: Lingue e linguistica
• Subtopic: Linguistica

The Routledge Handbook of Translation and Cognition provides a comprehensive, state-of-the-art overview of how translation and cognition relate to each other, discussing the most important issues in the fledgling sub-discipline of Cognitive Translation Studies (CTS), from foundational to applied aspects.

With a strong focus on interdisciplinarity, the handbook surveys concepts and methods in neighbouring disciplines that are concerned with cognition and how they relate to translational activity from a cognitive perspective. Looking at different types of cognitive processes, this volume also ventures into emergent areas such as neuroscience, artificial intelligence, cognitive ergonomics and human–computer interaction.

With an editors' introduction and 30 chapters authored by leading scholars in the field of Cognitive Translation Studies, this handbook is the essential reference and resource for students and researchers of translation and cognition and will also be of interest to those working in bilingualism, second-language acquisition and related areas.

Part I

Foundational aspects of translation and cognition

1

Translation, epistemology and cognition

Andrew Chesterman

1.1 Introduction

The human brain, writes Nobel Prize winner Gerald Edelman (1992, p. 23), is “the most complicated material object in the known universe”. Note the careful “known”. After all, it seems that currently we can only perceive about 5% of the universe, the rest being mysterious dark matter and equally mysterious dark energy (so we lay folk are told, at least). Edelman’s qualification “known” is thus eminently justified, and a useful reminder of the virtue of modesty in making claims about the world. But in what sense, and to what extent, do we even really “know” that 5%?

How much do we know about the human brain? Less than 5%? Do we even know how much we don’t know?

And further: what do we think we know about “knowing”? From Plato on, philosophers have usually started from the assumption that “knowledge” is justified true belief; but such a definition carries a number of problems and has given rise to a good many debates. (For a relatively reader-friendly survey, see Steup, 2018.) The debated issues include the questions of how we can know—for certain—anything at all, what counts as justification (and for whom), and of course, what is truth. These are problems of epistemology, the philosophical study of knowledge. In the present context, my rather sceptical focus will be on the limitations of knowledge and of our methods of acquiring it. My position is thus a fallibilistic one: I assume that when we think we know something, we may be mistaken.

So, when Krings (1986) asked “What goes on in translators’ heads?”—in one of the most influential early publications in what was to become the “cognitive turn” in Translation Studies (TS)—we can at least answer: well, it is unlikely that we will ever know completely. But we might get to know something, and this something might eventually “evolve” into greater knowledge, better descriptions, better explanations and better understanding.

Suppose I introspect for a minute: what happened in my head when I just wrote what goes on in translators’ heads? For this is not what Krings actually wrote, which was: Was in den Köpfen von Übersetzern vorgeht. Actually, I first translated this as what happens …. But then I paused, and changed what happens to what goes on. Why? I’m not sure: perhaps it sounds more informal? Perhaps I was unconsciously influenced by the morphology of the German vor-geht? Would this interference hypothesis be strengthened by other examples of me being influenced by German morphology when translating into English? What about my translations from (or indeed into) other languages? Tests could be done …. Or was the change just due to intuition? The same kind of intuition that prompted me to write in translators’ heads rather than in the heads of translators, which would have followed the German form more literally? I just don’t know, however hard I try to introspect and to retrospect on what I just did. I do know that in the first line of this paragraph I wrote happened in my head rather than went on precisely because in the previous paragraph I had just shifted happens to goes on and did not want to repeat the same lexical item so close to the previous mention. But if I had not made this shift, would I have written goes on in my head? Again, I don’t know.

As I continue my retrospective introspection, the thought occurs that there is actually a semantic difference between happens and goes on (when neither is used in a progressive form), such that happen co-occurs more with events and go on with states or processes. (What happens when I press this switch? What goes on when the light is out?) I find this thought supported by a quick Internet search. But I was certainly not consciously aware of it when I made my change. Was this just my native language intuition, working below the level of consciousness? Perhaps. But why, then, did I write happens in the first place? Because I have lived for several decades outside an English-speaking culture and have lost touch with some of the niceties of my native English? And further: how do you know that I am introspecting and retrospecting, and reporting, truthfully? On what grounds do you trust me (or not)?

This trivial example illustrates just some of the problems faced by cognitive research into the translation process: problems of causality and explanation; the reliability of different data sources; the plausibility of conclusions about decision making and choices, and about the possible influence of earlier choices on later decisions; accessing the possible roles of the unconscious and of intuition; and even perhaps the influence of personal life history.

1.2 Historical perspective

A sense of the historical development of research on translation and cognition can be framed as a series of epistemological assumptions. Initial assumptions are always necessary, of course; but they may turn out to be misleading or mistaken, or only partially justified. We can start with the idea of the translator’s mind as a black box, with an input arrow on one side and an output arrow on the other, and a question mark in the box itself. This box model itself reveals an initial assumption: that we can learn something about the mind by considering it in isolation from its environment and from its own history, both phylogenetic and ontogenetic. Such an assumption can of course be defended as a necessary heuristic preliminary at the early stages of an investigation, a useful simplification. True, if it is borne in mind that it is just such an assumption; otherwise, there is the risk that the simplistic map is actually taken to be the complex territory it represents.

Looking at this box, first, we just wonder and speculate—surely where all science starts. We can observe what is going in and coming out, and attempt to draw inferences from what we see. The “interpretive school” of interpreter training (based at L’École supérieure d’interprètes et de traducteurs (ESIT) in Paris) arose from the initial inference that good interpreting (and translating) requires the deverbalization of the incoming message, so that its “sense” can be separated from its form and then reformulated in the target language (Seleskovitch & Lederer, 1984). This view of the cognitive process involved was based mainly on experience and intuition, and led to successful training methods that produced excellent professional interpreters. But it was not supported by empirical research on cognition, and the status of the deverbalization phase was challenged, e.g. by research supporting the “literal translation hypothesis”. This claims that interpreters and translators are highly influenced by the form of the source text and tend first to select target forms that match those of the source when possible. If this is the case, it would imply that deverbalization does not take place—or at least not always. Interesting evidence supporting the literal translation hypothesis comes from studies of interim drafts and the revision process (see e.g. Englund Dimitrova, 2005). The difference between these two positions may be partly explained by the ESIT focus on what should happen in ideal interpreting, and therefore on what would be pedagogically relevant. But inferences may also be mistaken.

Continuing to ponder our black box, we might assume that our translator’s mind contains a number of smaller, interconnected boxes (modules, components) with different functions: one deals with comprehension, for instance, another with target-language formulation, yet another is perhaps engaged in some kind of quality checking, etc. So we assume that the mind is modular, or at least that it is useful to regard it as such at the initial stages of an investigation. The image is of the mind as some kind of machine, with various components all doing their own jobs, converting input into output.

Then we try manipulating the input and see how the output changes: i.e. we adopt an experimental method. We also begin scratching the surface of the box, with the idea of getting below the surface, at least to some small degree. We thus assume that by entering the outer layer of the brain we can begin to glimpse the workings of the mind: we can, for instance, use EEG (electroencephalography) or PET (positron emission topography) and examine which bits of the brain appear to be activated under different conditions (e.g. Kurz, 1994; Tommola, 1999). Some later research, however (e.g. García et al., 2016), has cast doubt on the idea that the brain has areas that are dedicated specifically to translation or interpreting.

But do these methods give us any knowledge about the mind? Is it reasonable to assume that the brain is a window on the mind? Not a transparent one, at least, it would seem. The huge mind–brain issue (or more generally, the mind–body issue) has not been resolved to general satisfaction, although a strict dualist position seems now to be out of fashion. The jury is still out and may well remain out for some time to come. Indeed, perhaps this issue is actually not solvable at all. (But see later.) There is, however, the risk that we too easily assume that new information about the brain tells us something about the mind. It may, but it may not: reservations are needed.

From the mid-1980s on, Krings and others experimented with a new way of collecting data on what were assumed to be the cognitive processes involved in translation. This was introspection in the form of spoken think-aloud protocols (TAPs). The method had previously been used in psychology (see Ericsson & Simon, 1993). Part of my brief introduction earlier was a kind of thinking aloud, not spoken in this case, but slightly delayed and to some extent monitored before being expressed in writing. In my case, I was retrospecting; but in the protocols used by the early TAP scholars the idea was to have the subjects talk aloud at the same time as they translated (usually working on a computer, with a separate recording machine).

It was widely recognized that the TAP method had limitations (it was obviously not appropriate for research on interpreting) and flaws. Assumptions were made that now look hard to justify. Critics pointed out that the very act of talking aloud would surely interfere with the normal cognitive process that was allegedly being studied. Indeed, the very act of being observed may have interfered (the observer effect). How reliable would the protocols be? Wouldn’t they be rather selective, focusing only on what can be verbalized, and within that, on translation problems at the expense of routine flows? How much of what goes through the subject’s mind would actually be verbalized? And wouldn’t the TAPs actually concern the results of cognitive processes rather than the processes themselves? Such criticisms were often discussed, and often acknowledged, but TAP research nevertheless went ahead. Attempts were made to rectify some of the problems, for instance by setting up TAPs with pairs or small groups of subjects, on the assumption that this would allow a more natural dialogue rather than a possibly unnatural monologue. TAPs were also combined with retrospective interviews. Yet despite these developments, the numbers of subjects studied in each project remained very small—for the obvious reason that this kind of research is extremely time-consuming. (See Jääskeläinen, 2002, for a survey.)

Further refinements added new kinds of data gathered from new kinds of technical tools, the most noteworthy being Jakobsen’s Translog keystroke logging program, and later the application of eye-tracking technology. (See e.g. Jakobsen, 2017.) Here too there are interesting assumptions. Now, for instance, it is the eye that is assumed to serve as a window on the mind. There is also the assumption that when the subject pauses, i.e. stops typing for a few seconds or minutes, this absence of observable activity is a significant indication of non-observed cognitive activity, such as grappling with an awkward translation problem (rather than, say, day-dreaming). One big advantage of these new methods was their use in triangulating results, e.g. cross-checking the results of a TAP with those of keystroke logging and/or eye-tracking data and/or a retrospective interview. These innovations allowed more specific hypothesis formulation and more robust hypothesis testing (for instance of the above-mentioned literal translation hypothesis). TAPs have suggested significant differences in the way professionals or experts translate in comparison with students or untrained translators. They have also generated promising new lines of research, such as on the relevance of attitudinal and emotional factors, which have further complicated our original image of the black box. Much of this research has been methodology driven (we do it because we now can) rather than theory driven (e.g. to test theoretically significant hypotheses) or problem driven (e.g. to solve translation quality problems). It has often been exploratory: let’s do this, and see what we can find. This is not to say that such research does not bring new knowledge, but such knowledge may appear fragmentary if its theoretical framework is unclear. Observation is used to test theory, yes, such as when hypotheses are tested; but theory is also needed to make sense of what is observed.

Research into the cognitive processes involved in translation has been increasingly influenced by aspects of the related fields of psycholinguistics, cognitive linguistics and cognitive science. Translation is a complex activity, true, but on what grounds might we assume that the nature of translators’ cognitive processes would be significantly different from those of electricians, or dentists? Apart from the factor of the two languages, of course. More recently, “Cognitive Translation Studies” has been launched as a cover term for a broader framework of study that would open up to other relevant fields such as bilingual research and language acquisition, and more broadly still, neuroscience. (See e.g. Muñoz Martín, 2014, 2016a.)

We started with the assumption that the translator’s mind could be initially conceptualized as a black box, and it has seemed that the study of translation cognition also started with a similar image of itself, i.e. as a sel...