Ehrensberger-Dow... (Et Al) - Describing Cognitive Processes in Translation Acts and Events (Benjamins, 2015)

Describing Cognitive Processes in Translation
Benjamins Current Topics

issn 1874-0081
Special issues of established journals tend to circulate within the orbit of the
subscribers of those journals. For the Benjamins Current Topics series a number
of special issues of various journals have been selected containing salient topics of
research with the aim of finding new audiences for topically interesting material,
bringing such material to a wider readership in book format.
For an overview of all books published in this series, please see
http://benjamins.com/catalog/bct
Volume 77
Describing Cognitive Processes in Translation. Acts and events
Edited by Maureen Ehrensberger-Dow, Birgitta Englund Dimitrova,
Séverine Hubscher-Davidson and Ulf Norberg
These materials were previously published in Translation and Interpreting Studies
8:2 (2013)
Describing Cognitive Processes
in Translation
Acts and events
Edited by
Maureen Ehrensberger-Dow
Zurich University of Applied Sciences
Birgitta Englund Dimitrova

Stockholm University
Séverine Hubscher-Davidson
Aston University
Ulf Norberg
Stockholm University
John Benjamins Publishing Company

Amsterdam / Philadelphia
TM
The paper used in this publication meets the minimum requirements of
8
the American National Standard for Information Sciences – Permanence

of Paper for Printed Library Materials, ansi z39.48-1984.
doi 10.1075/bct.77
Cataloging-in-Publication Data available from Library of Congress:
lccn 2015019668 (print) / 2015026242 (e-book)
isbn 978 90 272 4265 5 (Hb)
isbn 978 90 272 6820 4 (e-book)
© 2015 – John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. · https://benjamins.com
Table of contents
Introduction 1
Models of what processes? 7
Andrew Chesterman
Shared representations and the translation process: A recursive model 21
Moritz Schaeffer and Michael Carl
ELF speakers’ restricted power of expression: Implications for
interpreters’ processing 43
Michaela Albl-Mikasa
The role of intuition in the translation process: A case study 63
The effect of interpreting experience on distance dynamics: Testing the
literal translation hypothesis 85
Nataša Pavlović and Goranka Antunović
The impact of process protocol self-analysis on errors in the translation
product 105
Erik Angelone
Opening eyes to opera: The process of translation for blind and partially-
sighted audiences 125
Sarah Eardley-Weaver
Notes on editors 147
Index 149
Introduction
In the translation process, translators and interpreters have a central position as

thinking and acting human beings. This volume1 addresses the topic of translation
as an act and event, having as its main focus the cognitive, mental processes of the
translating or interpreting individual in the act of translating. But it also opens up
wider perspectives by including the social situation in the cognitive explorations
of the translation process, i.e., by also looking at the event of translating.
The mental processes of the translator/interpreter have been the subject of
a growing field of study in Translation Studies, beginning in the 1980’s and ex-
periencing a very strong development in the last 15 years (for an overview, see
Englund Dimitrova 2010 or Jääskeläinen 2011). This field, often called process-
oriented translation studies or translation process research, has been character-
ized from the start by a strong empirical orientation, thus inscribing itself in the
paradigm of Descriptive Translation Studies. Another characteristic feature has
been interdisciplinarity, as seen in the application of theories and methods from
disciplines such as cognitive psychology, cognitive linguistics, and psycholinguis-
tics. The latest, but certainly not last, evidence of this interest in interdisciplinarity
is another recent volume in the BCT series (Ehrensberger-Dow, Göpferich, and
O’Brien 2015).
The empirical focus of process-oriented studies has entailed the application
and further development of a number of research and data collection methods,
ranging from introspection (i.e., concurrent during think-aloud or retrospective
commentary) to observation of behavior such as writing (e.g., with keystroke log-
ging; see Jakobsen and Schou 1999) or eye movements and pupillometry through
eyetracking (e.g., O’Brien 2006). The introduction of such tools and methods has
brought with it increasing attention to and awareness of fundamental aspects of
research design and experimental rigor, but also, somewhat paradoxically perhaps,
the importance of ecological validity. For a discussion of developments over the
1. The chapters were originally published in a special issue of the journal Translation and
Interpreting Studies (8:2, 2013), bringing together papers presented at two international con-
ferences in 2011, Text-Process-Text, at Stockholm University, Sweden, and Translation Process
Research: Breaking New Ground, at Aston University, Birmingham, U.K.
doi 10.1075/bct.77.01int
2015 © John Benjamins Publishing Company
2 Introduction
past decade of process-oriented translation studies and its most prominent topics
and tendencies, see Muñoz (2014).
In parallel with the field of process-oriented studies, another important field
has evolved in Translation Studies, also having as one of its focal points the transla-
tor/interpreter, but seen from the perspective of being embedded in his/her social
environment: translation sociology or the sociology of translation (e.g., Wolf and
Fukari 2007; Kinnunen and Koskinen 2010; Angelelli 2012). This field could be
said to have originated in the core of Descriptive Translation Studies and in the
conceptualization of translation norms (see Toury 1995) but has since been fur-
ther developed by inclusion of theories and concepts from sociologists such as
Bourdieu and Latour.
Chesterman (2009) proposed a new subfield within Translation Studies, which
he called Translator Studies. He further suggested a tripartite division of this sub-
field, with cultural, cognitive, and sociological branches, the last two of which are
the focus of the present volume. The sociological branch would deal with “transla-
tors’/interpreters’ observable behaviour as individuals or groups or institutions,
their social networks, status and working processes, their relations with other
groups and with relevant technology”, and the cognitive branch would deal with
“mental processes, decision-making, the impact of emotions, attitudes to norms,
personality” (Chesterman 2009: 19). It is clear that there is overlap in some of the
objects of study in the cognitive and the sociological branches and hence a poten-
tial for a rapprochement and perhaps cooperation. Examples of studies which may
be seen as located at the interface between the two branches are Buzelin (2007) and
Risku and Windhager (2013/2015). The relevance of such potential convergence
seems especially clear in view of the rapid changes in the working conditions of
translators and interpreters. Increased use of technology, specialization of transla-
tion work processes in large enterprises as well as the outsourcing of translation
tasks are just some factors which characterize the development of modern transla-
tion professions. How those changes in the sociological translation event impact
the cognitive translation act is a fruitful avenue for future study.
The six chapters in this volume address various aspects of translators’ and
interpreters’ observable and non-observable processes, thus enabling readers to
reflect further on the concept of translator studies and a possible merging of cog-
nitive and sociological approaches to understanding the phenomenon of interest.
In the first chapter, Andrew Chesterman provides a theoretical framework for
the volume with Toury’s (1995) proposed distinction between different translation
processes: on the one hand, cognitive translation acts, i.e., the mental processes,
and on the other, sociological translation events, into which translation acts are
embedded. Chesterman here proposes a third term, translation practices, to desig-
nate the translation process at the historical and cultural level. The author further
Introduction 3
claims that translation acts can be studied from three different perspectives: as
virtual (potential) processes, as reverse-engineered (reconstructed) processes, or
as actual (observed) processes.
In terms of those distinctions, the other chapters of the volume can be charac-
terized as studies of the actual cognitive translation acts, of other processes related
to the translation acts, or of processes that are related to the sociological transla-
tion event. They share an interest in actual, observed processes but differ with
respect to the phenomena under investigation as well as the methods used to gain
an increased appreciation and understanding of those phenomena.
Focusing on bilingual processing, clearly part of the act of translation, Moritz
Schaeffer and Michael Carl re-evaluate the psycholinguistics of the bilingual lexi-
con and evidence from bilingual priming studies in order to gain insights into
automated translation processes. They argue that translation involves the activa-
tion of shared cognitive representations and report experimental evidence for a
facilitative effect of translation over simple reading on the recall of a source text.
In the light of their findings, they propose a recursive model of translation which
re-defines the literal translation hypothesis and the monitor model in terms of
horizontal and vertical bilingual processing during the production of a target text.
In her chapter, Michaela Albl-Mikasa considers the special challenges that
ELF (English as Lingua Franca) input presents in the act of mediated bilingual
processing in conference interpreting. She draws on a small case study of a student
interpreter’s rendering into German of short speeches given by ELF speakers. Her
analysis of the data suggests that interpreting ELF speech may be compromised
on two levels: non-standard input may affect the interpreter’s comprehension
processes; and unconventional expressions and structures prevent the interpreter
from being able to rely on previously established links in rendering the target text.
Especially interesting with respect to the act-event distinction is that the study also
includes the ELF speakers’ perspectives on their own discourse, as deliverers of the
source speech. In her conclusion, Albl-Mikasa discusses the didactic implications
of trying to prepare students for unpredictable ELF input.
Probing how non-conscious processes might influence the act, Séverine
Hubscher-Davidson examines the psychological construct of intuition and ex-
plores the role that it might play in translation. She reviews the literature that has
considered the importance of intuition in translation and explains the construct
in light of recent psychological research. Self-report data from a case study of a
student translating a literary text is analyzed in order to highlight the possible in-
fluence of intuition in decision-making behavior during translation. On the basis
of her analyses, she claims that the potential to enhance intuitive decision-making
should be exploited, since it might be an important part of the translation process.
4 Introduction
Similar to Schaeffer and Carl, Nataša Pavlović and Goranka Antunović focus
on the literal translation hypothesis. Using Translog recordings of interpreters’ and
translators’ processes, they test whether more literal renderings during the initial
act of translation become freer during self-revision. Hypothesizing that interpret-
ing experience will have an effect on the types of self-revisions made at various
stages of the process, they compare revisions in terms of distance to the source text
structure. Their hypothesis is only partly confirmed by their results. An important
finding is that many self-revisions are ‘neutral’ in terms of distance to the source
text, and the authors conclude that this has to be taken into account in future stud-
ies on self-revision.
By encouraging students to reflect on their own processes afterwards, Erik
Angelone moves them a step away from the immediate cognitive act. In his ex-
ploratory study of self-revision processes, he compares the efficacy of three types
of translation process protocols as prompts for post-drafting translation problem
recognition and revision. Students created a process protocol while translating
and then self-reflected about their problem-solving performance on the basis of
that protocol. They were given the opportunity to revise their translations follow-
ing their analyses, and then their final versions were marked for various types of
errors. Of the three protocol types, screen recordings were consistently associated
with the fewest errors in the final translations, suggesting that this type of prompt-
ing would be a useful revision technique in process-oriented translator training.
In the last chapter of this volume, Sarah Eardley-Weaver considers a very dif-
ferent type of translation process, one which is more closely linked to the concept
of a sociological translation event. She discusses the growing field of audiovisual
translation and examines audio description and touch tours for partially-sighted
and blind patrons of opera performances. Proceeding from the framework of actor
network theory also used in studies within sociology and translation (Abdallah
2010), she explores the roles of various actors involved in such translation process-
es and argues that the complex networks would be best described as iterative, with
the audience’s reception of the translation product feeding back into subsequent
processes. Audience reception studies, she points out, should be an integral part
of the investigation of any translation process and not just audiovisual translation.
With the chapters of this volume, we hope to have provided impetus for further
work, combining the act and event perspectives.
Birgitta Englund Dimitrova
Maureen Ehrensberger-Dow
Ulf Norberg
Introduction 5
References
Abdallah, Kristiina. 2010. “Translators’ Agency in Production Networks.” In Translators’ Agency,

ed. by Tuija Kinnunen, and Kaisa Koskinen, 11–46. Tampere: Tampere University Press.
Angelelli, Claudia (ed). 2012. The Sociological Turn in Translation and Interpreting Studies.
Special Issue of Translation and Interpreting Studies 7 (2): 125–128. DOI: 10.1075/tis.7.2.01int
Buzelin, Hélène. 2007. “Translations in the Making.” In Constructing a Sociology of Translation,
ed. by Michaela Wolf, and Alexandra Fukari, 135–169. Amsterdam: Benjamins.
DOI: 10.1075/btl.74.11buz
Chesterman, Andrew. 2009. “The Name and Nature of Translator Studies.” Hermes — Journal of
Language and Communication Studies 42: 13–22.
Ehrensberger-Dow, Maureen, Susanne Göpferich, and Sharon O’Brien (eds). 2015.
Interdisciplinarity in Translation and Interpreting Process Research. Amsterdam: John
Benjamins. DOI: 10.1075/bct.72
Englund Dimitrova, Birgitta. 2010. “The Translation Process.” In Handbook of Translation
Studies, ed. by Yves Gambier, and Luc Van Doorslaer, 406–411. Amsterdam: John
Benjamins. http://www.benjamins.com/online/hts/ DOI: 10.1075/hts.1.tra6
Jääskeläinen, Riitta. 2011. “Studying the Translation Process.” In The Oxford Handbook of
Translation Studies, ed. by Kirsten Malmkjaer, and Kevin Windle, 123–134. Oxford: Oxford
University Press.
Jakobsen, Arnt Lykke, and Lasse Schou. 1999. “Translog Documentation”. In Probing the
Process in Translation: Methods and Results, ed. by Gyde Hansen, 151–186. Copenhagen:
Samfundslitteratur.
Kinnunen, Tuija, and Kaisa Koskinen (eds). 2010. Translators’ Agency. Tampere: Tampere
University Press.
Muñoz Martín, Ricardo. 2014. “A Blurred Snapshot of Advances in Translation Process
Research.” In Minding Translation — Con la traducción en mente, MonTI Special Issue 1,
ed. by Ricardo Muñoz Martín, 49–84.
O’Brien, Sharon. 2006. “Eye-Tracking and Translation Memory Matches.” Perspectives: Studies
in Translatology 14 (3):185–205.
Risku, Hanna, and Florian Windhager. 2013/2015. “Extended Translation. A Sociocognitive
Research Agenda.” In Interdisciplinarity in Translation and Interpreting Process Research,
ed. by Maureen Ehrensberger-Dow, Susanne Göpferich, and Sharon O’Brien, 35–47.
Amsterdam: Benjamins. DOI: 10.1075/target.25.1.04ris
Toury, Gideon. 1995. Descriptive Translation Studies and Beyond. Amsterdam: Benjamins.
DOI: 10.1075/btl.4
Wolf, Michaela, and Alexandra Fukari (eds). 2007. Constructing a Sociology of Translation.
Amsterdam: Benjamins. DOI: 10.1075/btl.74
Models of what processes?
Andrew Chesterman
Toury (1995; 2012) distinguishes between cognitive translation acts on the

one hand, and sociological translation events on the other; a translation act is
embedded in a translation event, and both acts and events are seen as processes.
He also explains three senses of ‘translation problem,’ which relate to different
notions of the processes involved in the translation act. The present chapter
analyzes and develops these ideas. It distinguishes between what are here labeled
virtual, reverse-engineered, and actual processes of translation acts or events,
which correlate with Toury’s three senses of ‘translation problem.’ A few exam-
ples are given of models of each kind of process, both classical and more recent
ones. Also discussed is the extent to which the various models are predictive and
hence testable. To designate the translation process at the historical and cultural
level, alongside the mental act and the situational event, the term ‘translation
practice’ is suggested.1
Keywords: translation process, translation model, translation act, translation

event, translation practice
1. Act and event
Most process research on translation has so far focused on the cognitive dimen-
sion; the investigation of sociological processes has not attracted as much atten-
tion, although the recent “sociological turn” in translation studies may influence
future tendencies (e.g., Wolf and Fukari 2007). The distinction between these two
basic dimensions was already implied by Gideon Toury (1995), in his discussion of
natural translation and sources of feedback. Feedback, he wrote, may come from
the translation recipients, and also from:
1. My sincere thanks are due to the two critical referees who made many helpful suggestions on
the initial version of this chapter. I should undoubtedly have taken more of them on board than
I have been able to.
doi 10.1075/bct.77.02che
8 Andrew Chesterman
those who have commissioned the act of translating, and sometimes from the orig-
inator of the utterance to be translated as well. When realized by actual persons,
these roles (in the sociological sense) — all parts of the interactional makeup of a
translating event — may, of course, partially overlap. (1995: 249, emphasis original)
Reading somewhat between the lines here, many scholars (including myself) have
taken “act of translating” — or translation act — to refer to the cognitive process,
whereas the translation event is the observable sociological framework in which
the cognitive translation act takes place. In the revised version of the book (Toury
2012: 67–68), this distinction is drawn more clearly and developed at some length.
The locus of the translation act is said to be “the human brain.” Yet Toury insists
there that the cognitive cannot be completely separated from the environmental.
The relation between the mental act and the situational event is one of “comple-
mentarity and containment.” This is a position that has also been taken by other
scholars such as Risku (2010), who have explored the relevance of the notion of
situated cognition to Translation Studies. Toury calls for research on the interac-
tion between these two levels, the mental and the situational. A recent example of
how this interaction might be studied is Jones (2011), on poetry translation; this
study combines think-aloud protocols (TAPs), interviews, and agent networks.
An act is thus embedded in an event: the event is the sociological or situ-
ational context of the act. The event is directly observable: one can follow a trans-
lator’s overt behavior, observe phone calls, emails, use of the internet, physical
movements, and so on. But the act is not directly observable, one can only make
inferences from the behavior one can see. Even the insertion of electrodes in the
brain does not provide direct access to cognition itself, only to the electric pulses
and neuron activities etc. which manifest it. A translation event normally involves
other actors too, of course, who also perform relevant cognitive acts.
In principle, the distinction between the cognitive translation act and the so-
ciological translation event seems clear, although both terms refer to processes
taking place in time; moreover, some models appear to incorporate aspects of both
dimensions. In translation, these time scales are very different: one is measured in
seconds or microseconds, the other in hours or days, or even months. There does
not yet seem to be agreement about how to define precise starting and ending
points of a translation act or event, however, or how to conceptualize the overlap
between them. Aspects of the event, for instance, such as the details of the brief
and the definition of the intended addressees, presumably influence the mindset
of the translator and hence the cognitive translation act, perhaps even before he/
she even receives the source text. Let us nevertheless suggest that a translation
act begins when the translator begins to read the source text, and ends when the
translator decides to take no further action in revising the translation (although
there might still be further thought on the subject). The act may of course be in-
terrupted. A translation event, on the other hand, could be said to begin when the
translator accepts the job (or perhaps when the client begins to look for a transla-
tor?), and ends with, say, payment of the bill (or perhaps when the first recipient
reads the translation?). Translation events have been investigated in different ways,
via workplace studies, revision procedures, the analysis of translator teams and
networks, translator agency, and so on (e.g., Kinnunen and Koskinen 2010). Here
too, interruptions are common.
This chapter mainly concerns translation, but it is worth noting that, in inter-
preting, the distinction between act and event often seems less clear-cut. But the
interpreting act and event can still be studied separately, as indeed they are, using
different kinds of methods and models.
2. Models and problems
The term “model” is significantly polysemous in the philosophy of science, where

debates range about what a model is ontologically, what kinds of models there
are, and how models relate to theories. For instance, some models aim to be rep-
resentations of a phenomenon, in some way or other, while others have the form
of an explanation (like a law). Some models are more explicit than others; some
are formulated in mathematical terms, others are based on analogy (such as the
computer model of the mind). Within Translation Studies, too, there are many
different views. I take a model here to be a preliminary kind of theory, one which
claims some relation of similarity with the object that is modeled; a model, in this
definition, purports to be isomorphic with its object, in some way. In some kinds of
models this isomorphism is obvious. Think of those models of the solar system, be-
fore and after Copernicus, that you saw at school. In others, such as mathematical
or computational models, the isomorphism is more abstract. Models in this sense
show what are thought to be the main components or elements of a phenomenon,
what are thought to be the main relations between them and their main functions.
I shall assume here that models are basically systematic descriptions, descriptive
hypotheses, which claim to represent something. True, such a model may also im-
ply an explanation, for instance if the relations included are causal. Consider for
instance a simple model of a primitive steam engine, showing how increasing the
temperature of water in a container eventually causes an increase of pressure when
the water boils, and this pressure then causes something to move somewhere else
in the modeled system; the resulting movement is thus explained by the model.
Explanations of various kinds are also implied by the predictiveness of models,
insofar as they are indeed predictive. In a weak sense, a general descriptive model
10 Andrew Chesterman
implicitly predicts that it will also apply to yet-to-be-studied instances of the phe-
nomenon in question, indeed to all possible instances belonging to the same set.
In this weak sense, a descriptive model could be said to be explanatory in that it
generalizes, by predicting applicability to unknown instances (on explaining via
generalization, see Chesterman 2008). However, explanations and predictions do
not inevitably go hand in hand, as is well known.
In a stronger sense, a model can be predictive of consequences that can them-
selves be tested empirically. A classic example from chemistry is Mendeleev’s
nineteenth century model of the elements, the periodic table, which arranged the
elements by group and by atomic weight. The first versions of this model had gaps
at certain points; the model predicted that these gaps would eventually be filled, as
new elements were discovered. And they were.
As we shall see, not many models of the translation process appear to be par-
ticularly predictive.
In what follows I will distinguish between different kinds of models of the
cognitive translation process according to the ontological status of what is being
modeled. These different kinds of models actually represent different senses of
‘translation process,’ although all are concerned with the cognitive translation act,
not the sociological event. My presentation of model types is based on Toury’s
discussion (2002; 2012: Chapter 2) of three different senses of the term ‘translation
problem.’ Toury does not give labels to these different senses, but refers to them
as problem1, problem2, and problem3. In brief, Toury’s distinction is as follows.
The first sense is the potential problem of the translatability of a given source-text
item into a given target language, under given conditions: how might this ST item
be translated here? A problem in the second sense is identified by starting with a
given target-text item that functions as a factual translation solution, and then at-
tempting in retrospect to reconstruct the translation problem for which it has been
selected as a solution, and also to reconstruct the thinking that led the translator
to this solution. And the third sense is the notion of a problem as it is observed to
be experienced by the translator, during the process of a given act of translation,
via traces left e.g., by interim solutions, by verbal reflection, or by pauses in the
process. In Toury’s revised version of his book (2012), these three senses of ‘trans-
lation problem’ are argued to correspond to different senses of the translation act.
With these senses in mind, consider what they imply for an understanding
of different models of the process of the translation act. I propose to distinguish
three kinds of models: models of virtual processes, corresponding to Toury’s first
sense of ‘problem’; those of reverse-engineered processes, corresponding to Toury’s
second sense; and those of actual processes, corresponding to his third sense. I now
look at these types in more detail, and give some examples.
3. Models of virtual processes
First, consider a model of what I will call a virtual process: this would be the po-
tential path from one sense of ‘translation problem’ to a potential solution, show-
ing for instance the possible strategies for the translation of an allusion or a pun
(taken as translation problems, in Toury’s sense 1). Such a model is pedagogical or
advisory/prescriptive in nature, and starts with something in the source text that is
treated as a translation problem, or as we saw Toury put it, a problem of translat-
ability. It thus takes a prospective approach. Based on intuition, or experience, or
on the analysis of many translations, the model then outlines possible courses of
action leading to possible solutions, in theory. It is a simplified, idealized model, of
possible decision processes leading to acceptable solutions (although, clearly, real
translators might not behave optimally). Such a model might also purport to rep-
resent the translation process in general, not just the solution of a given problem.
Models of the virtual process are predictive in the weak sense that if translators
follow the advice illustrated by the model, it is presumably assumed that the results
will tend to be more acceptable than if the advice is not followed. In other words,
the model implies a prediction that its use will lead to beneficial effects: hence its
usefulness in the classroom. In principle, therefore, these models are testable: we
can test whether their use really does lead to better translations than cases where
they are not used. (But do we have any such tests?)
An early example of such a virtual model is Nida’s (e.g., 1969) well-known
three-stage model, comprising analysis, transfer, and restructuring, with obvious
pedagogical aims. Nord’s “looping model” (1991) takes Nida’s pedagogical ap-
proach a step further. She starts with the analysis of the skopos, then the analysis
of the source text, followed by the production of the translation. The model loops
back and forth between these three. Feedback comes from the emerging target text
itself, too, as later decisions affect earlier ones. Hönig (1995) proposes an “ideal”
flow-chart model, including the translator’s macro- and micro-strategies, moni-
toring, etc. Interestingly, he explicitly compares his virtual model with Krings’ ac-
tual one (see below).
Levý’s game-theoretical model (1967) takes a teleological point of view of the
virtual translation process, in which the translation act is represented as a series of
decisions, which are like moves in a game. Alternative solutions to a given ST seg-
ment are generated, assessed according to specified criteria (such as stylistic natu-
ralness vs. semantic closeness, or type of implied audience). Most of the examples
he analyzes are modeled in terms of binary decisions. Levý takes the process of
translating to be “a game with complete information”; this means a game where
every decision and move is influenced by the knowledge of previous decisions and
their consequences.
Along similar lines, but focusing on a particular translation problem (in

Toury’s sense 1) rather than translation in general, Leppihalme’s study (1997) of-
fers a model of a decision-making tree, incorporating an implicit suggestion that
this is a helpful way to arrive at a possible, or optimal, solution for the translation
of different kinds of allusions.
4. Models of reverse-engineered processes
Corresponding to Toury’s second sense of ‘problem,’ there are models of what I

will call a reverse-engineered process. These aim to reconstruct the possible or even
probable route taken to a given factual solution (or set of solutions), as in reverse
engineering. For instance, as outlined above, from a target-text segment we infer a
possible translation problem in the corresponding source segment, and then spec-
ulate on the most plausible sequence of actions or decisions leading to the given
solution, considering linguistic factors, likely motivations, available resources and
constraints, and so on. In other words, we ask “how could the translator have ar-
rived at this solution?”
Such models are often implied in studies focusing on translation or interpret-
ing errors (however these are defined): given the error, the possible or probable
antecedent decision-making stages are inferred, with the aim of determining likely
causes of the error. They are thus potentially predictive, in that they can in prin-
ciple predict that the presence of certain conditions or aspects of the decision-
making process will increase the probability of such errors. In pedagogical use,
such models are also used in the retrospective reconstruction of (probable) deci-
sions leading to successful translations (e.g., many examples in Kussmaul 2007,
and similar translation manuals).
Gile’s Effort Model for simultaneous interpreting (revised in Gile 2009) can
be interpreted as one example of this kind of model, because it looks like a pos-
sible representation of conditions for a reverse-engineered process, rather than a
model of the actual process itself. It infers the existence of three basic ‘Efforts’: the
Listening and Analysis Effort, the Production Effort (including self-monitoring),
and the short-term Memory Effort, plus a Coordination Effort for processing ca-
pacity management. Errors are predicted when one or another of these Efforts is
overburdened.
Another example of this kind of model is illustrated in a study by Shreve,
Lacruz, and Angelone (2011). Looking at speech disfluencies in sight translation,
the authors infer that these may be caused by what they call visual interference. This
study does not set out an explicit model, but appears to assume that such a model
would comprise components such as comprehension, transfer, production, and in
this case also reading. Like other TAP researchers, Shreve et al. assume in general
that (longer) pauses indicate problems or points where special efforts are needed.
5. Models of actual processes
The third kind of model aims to represent a process in real time. We could call it
a concurrent model, of an actual process. Such models make use of data derived
from observation of the process as it takes place, using such methods as think-
aloud protocols, keyboard logs, eye-tracking, and the like. They are thus different
from reverse-engineered models in that they observe the process as it unfolds,
rather than retrospectively inferring from the end-result how this result might
have arisen. (True, a model of an actual process also involves inferences concern-
ing the nature of the unobservable cognitive activities presumably taking place.)
Models of an actual process are different from virtual models in that they are de-
scriptive, not prescriptive (e.g., Kussmaul 2007: 91).
Krings’s model of the actual process (1986), based on TAP data, is a group-
ing of types of strategies (of comprehension, retrieval, decision-making, monitor-
ing, and reduction) to solve problems of different kinds. Lörscher’s model (1991)
aimed to represent the abstract kinds of “strategies” translators use when they
come to a segment of source text that they cannot translate routinely: they have
to stop and think. Lörscher’s strategies are thus problem-solving ones, and do not
represent the whole of the translation act. Examples of such strategies are: the
realization of a translation problem; its verbalization; formulation of a tentative
solution; assessing this solution. The classification is based on TAP data, with a few
non-professional translators. What is modeled is part of an observed process, at a
very general level. He concludes that the strategies he discusses are unlikely to be
specific to translation itself, but pertain to text-processing generally.
Jakobsen’s recent work with Translog and eye-tracking (e.g., 2011) is of con-
siderable interest as an approach to the observed cognitive process, i.e., what I
have called the actual process. Among his results he cites data (2011: 40) suggest-
ing that at least on some occasions, comprehension, formulation of a translation,
and actual typing could not possibly always be in sequence, linearly, but overlap;
this is because some recorded chunks are so long they could not possibly have
been held in short-term memory. In other words, there must have been some
parallel processing going on. An alternative interpretation of these results might
argue that they go against the assumption that there are separate “modules” for
memory, formulation, and physical execution (typing) in the first place.
On the basis of his eye-tracking and keystroke data, Jakobsen summarizes the
translation “micro-cycle” as a series of six steps, which he presents as a “small al-
gorithm” (2011: 48).
1. Moving the gaze to read the next chunk of new source text (and constructing
a translation of it)
2. Shifting the gaze to the target text to locate the input area and read the current
target-text anchor word(s)
3. Typing the translation of the source-text chunk
4. Monitoring the typing process and the screen outcome
5. Shifting the gaze to the source text to locate the relevant reading area
6. Reading the current source-text anchor word(s)
As thus formulated, these steps partly refer to the translation act and partly to the
event. Typing is an observable feature, and hence pertains to the translation event;
constructing a translation and monitoring, however, are cognitive acts. Jakobsen
adds that specifically steps 3 and 4 are not necessarily in linear sequence; that there
may be a good deal of recursion; and that steps can be skipped. Steps move some-
times from source to target and sometimes from target to source.
When models of actual processes are sufficiently formalized, computational
models of the translation act can be developed, with the goal of improving in-
teractive machine translation systems. Carl (2012), for example, reports research
of this kind, based on the actual behavior of expert translators. This behavior is
studied via keystroke logging and eye-tracking, but the derived “production rules”
include inferred mental operations such as shifting attention and actually translat-
ing, as well as the observable actions of shifting gaze and typing. Carl also outlines
a statistical model of the process studied, which appears to match actual human
behavior more closely. This model so far only deals with “unchallenged” (i.e.,
problem-free, routine) translation.
6. Relations between types of models
The relations between models of these three kinds of processes prompt a number
of questions. In the first place, whose processes are being studied? Much of the
research on actual processes has compared novice translators to professionals or
experts, on the assumption that something can thus be learned about the develop-
ment of translation competence, or about potential application in translator train-
ing (e.g., Jääskeläinen 1999; Englund Dimitrova 2005). There has thus been an
implicit or explicit quality variable involved: it is often the cognitive processes of
good translators that we are really interested in. Kussmaul (2007), for instance, is
particularly interested in successful, creative translation solutions.
There is a potential problem here, concerning the relation between virtual
processes and actual ones: our simplified models of virtual processes may be peda-
gogically useful, but they may not in fact correspond to what professionals actually
do. (For an early discussion of this issue, see Lörscher 1989.) Is there evidence, for
instance, that translators proceed through a logical series of binary decisions, as
suggested by some virtual models? (All translators? Or just some types?) Models
of virtual processes can thus generate predictions that can be tested.
Similarly, if reverse-engineered processes can be shown to lead, in theory, to a
given solution, to what extent might such a process correspond to an actual pro-
cess which did lead to this solution? What might be the significance of the real-
ization that two or more different processes might arrive at the same solution?
Might reverse-engineering even suggest more efficient ways of proceeding? Or is
the point of reverse-engineering models only to generate questions and then per-
haps hypotheses: could the process have been like this? And how might reverse-
engineered models relate to virtual ones?
Models of actual processes are likely to prove a good deal more complex, and
involve much more variation, than models of the first two process types, which
naturally tend to be idealized and simplified. Of course, any model may turn out
to be inaccurate, incomplete, or simply wrong, if it makes false predictions. Most
of the examples mentioned do not score highly on predictiveness, except in the
pedagogical sense of predicting their own usefulness in training, and in the weak
descriptive sense of purporting to be general models. None seem to be strongly
predictive, in the sense of a model exposing itself to the risk of being falsified by
predictions that are not realized.
In terms of testability, we could say that models of reverse-engineered process-
es could be checked to some extent by retrospective interviews, or tested against
other kinds of triangulated data elicited during the process itself, as indeed is done
(e.g., Englund Dimitrova 2005). However, there is no guarantee that models of a
reconstructed process necessarily represent the actual process which terminated in
a given translation.
7. Some models of the translation event
Models of the sociological event have been rarer than those of the translation act;
in many cases they are only implicit. My first, simple example is from the profes-
sional translator Robert Bly (1984), who wrote a classic essay about his own way of
translating poetry (specifically, a Rilke poem). He summarized this as eight steps.
1. Make a literal version of the poem.

2. Check the meaning in depth.
3. Polish the English.
4. Naturalize the language to spoken English.
5. Check that the translation is still true to the mood of the original.
6. Pay attention to the sound. Read the translation aloud.
7. Get a native speaker to react to the translation.
8. Re-read all previous drafts; check other people’s translations; do the final draft.
As a representation of what Bly says he himself does, the stages represent a gen-
eralized report of actual empirical processes, but as prescriptive “advice” they de-
scribe an ideal, virtual process.
One of the earliest and simplest event models proposed by a TS scholar was
Sager’s (1994) four-stage one: specification (understanding the client’s instruc-
tions, checking that the brief is appropriate and feasible); preparation (finding the
necessary resources, terminology, and so on); translation; and evaluation (revi-
sion). This model of the translation working procedure interestingly places trans-
lation itself (presumably that phase of the event that represents the translation
act) as a separate stage, distinct from other activities. Sager’s stages are in an obvi-
ous linear sequence, and also seem to represent an ideal type of process: a virtual
process. In real life, for instance, a professional translator may start translating at
once, before even reading the whole of the source text and before any preparation:
it would depend on the text in question and how routine the task was. Admittedly,
Sager’s stages are formulated at a very general level of delicacy or granularity, and
thus hide much potential variation.
Skopos Theory (Reiss and Vermeer 1984) and its close relative, the Theory of
Translational Action (Holz-Mänttäri 1984) are also implicitly event models. They
incorporate a number of actors, not just the translator. In Holz-Mänttäri’s model,
the stage even begins to look crowded: we find the Initiator, the Commissioner,
the ST Producer, the TT Producer (i.e., the translator), the TT User, and the TT
Receiver. And they view translation as an act of communication in a sociological
or situational context: hence the importance of the skopos.
Studies of interim solutions and the revision process map sequences of chang-
es made at different stages, as one could do with Bly’s evolving versions of his Rilke
translation. Englund Dimitrova’s (2005) model of the translation event has three
phases: initial planning, text generation, and revising. Self-revision is seen to oc-
cur both during the writing phase and after it. Englund Dimitrova notes that the
advent of computers has probably altered revision routines, as it allows easy and
instant revision, with more scope for non-linear working procedures (2005: 136).
Revision by others obviously occurs at the post-writing phase. Bly’s model

includes “another native speaker.” If there are several “others,” as in Nordman’s
(2009) study of legal translation from Swedish to Finnish, the whole translation
event can be plotted as a chain involving different actors in sequence: the initial
translator (or translator team), then various legal and language revisers in turn.
The various “others” involved in the translation event can be naturally mapped
as nodes in a network, resulting in a nexus model of the sociological process (e.g.,
Koskinen 2008; Kinnunen and Koskinen 2010). Actor Network Theory has been
used by a number of scholars, as a way of formalizing some aspects of such net-
works (e.g., Buzelin 2005; Jones 2011). Such models show multiple connections be-
tween nodes, with some nodes becoming more dominant than others. And nodes
do not have to be animate (in Actor Network Theory at least), which allows the
possibility of incorporating all kinds of electronic resources into the model.
Some research using an implicit event model highlights the difficulty of de-
termining when a translation event ends. For instance, Künzli and Ehrensberger-
Dow (2011) study the effects of a particular kind of subtitling on audiences.
Earlier, I suggested that an end-point might be specified when the translator
receives payment, or perhaps when the first recipient reads the translation. But
suppose the translation is never read? Studies of reader response and reception
nevertheless seem highly relevant to the translation event, if only because of
the way the translator’s knowledge or beliefs about implied readers can affect
translation decisions. This is in fact the point of the article in question: it appears
that readers, or at least some readers, may be able to cope with longer and more
complex subtitles than the current norms suppose, which would then allow the
subtitler to translate more fully.
In terms of testability, none of the event models mentioned is strongly predic-
tive.
8. The translation practice?
In addition to cognitive and sociological processes, we can also see Translation

Studies in the context of a larger process of historical and cultural evolution,
marked by changing traditions, norms and fashions, the careers of major influ-
ential translators, and so on. Translation Studies does not seem to have an ac-
cepted term to describe translation as a phenomenon on this level, a term that
would match ‘act’ and ‘event.’ Perhaps ‘practice’ might do. Long-term trends in this
historical process seem to include deprofessionalization and dehumanization: see
the increasing use of crowd-sourcing and of machine translation. Shorter-term
“historical” processes have been studied in terms of the acquisition of translator
competence or expertise, in longitudinal developmental studies (e.g., Schmidt

2005; Göpferich 2009). In the longer time-scale of human evolution, the practice
of translation can be seen as one means developed by homo sapiens for managing
a kind of heterogeneity, coping with communication across linguistic difference.
In terms of process modeling, it is worth noting that historical and cultural
studies tend not to aim at, or stop with, temporal models of processes: history is
more than just one damn thing after another… Such studies may set out to build
descriptive networks, like those used to represent the translation event (e.g., Pym
1998). But the aim is usually to find cause and effect relations (Pym 1998: 143).
9. Concluding remarks
If models incorporate causality in some way, they can make strong predictions and
thus be tested. Some of the cognitive models we glanced at above do include causal
factors: Gile’s Effort Model, for instance; and to some extent Levý’s decision-mak-
ing model, in that decisions are seen to be partly based on (i.e., caused by) cost-
benefit calculations on the pros and cons of various alternatives. But most of the
ones we have looked at are non-causal, so far; at least, they are not explicitly causal.
Retrospective methods such as interviews or error analysis may suggest causal
factors, but these are inferred via the logic of reverse engineering. The challenge
is then to test a reverse-engineered model against real-life translation, against an
actual process.
A similar challenge faces scholars who draw pedagogical inferences from their
descriptive models, as these inferences also imply a cause-and-effect dimension.
But we need more empirical tests of the validity of such inferences: whether, for
instance, translation quality would be improved by the adoption of such-and-such
a model as a basis for translator training. We do not know enough about how pre-
dictive our models might be.
Indeed, some scholars may feel that we are not at all ready yet to build any-
thing more than very simple models, and that research should proceed by gener-
ating and testing specific hypotheses: if supported, these hypotheses may then be
synthesized into a complex (and causal) model. One recurring observation is that
many studies indicate a great deal of variation across subjects, both in translation
and in interpreting. This makes generalization risky; but without generalizations
we cannot build models.
If one does set out to construct a model, I conclude, it is important to be clear
about what kind of process one is seeking to model; i.e., about what kind of model
one is building, based on what facts and/or hypotheses, and what assumptions.
What testable hypotheses might the model generate? What kind of predictions is
the model capable of making, if any? If it can make explicit predictions, it can be
tested, and maybe progress can be made.
References
Bly, Robert. 1984. “The Eight Stages of Translation.” In Translation. Literary, Linguistic and
Philosophical Perspectives, ed. by William Frawley, 67–89. Newark, NJ: University of
Delaware Press.
Buzelin, Hélène. 2005. “Unexpected Allies: How Latour’s Network Theory could Complement
Bourdieusian Analyses in Translation Studies.” The Translator 11 (2): 193–218.
DOI: 10.1080/13556509.2005.10799198
Carl, Michael. 2012. “A Computational Cognitive Model of Human Translation Processes.” In
Emerging Applications of Natural Language Processing: Concepts and New Research, ed. by
Sivaji Bandyopadhyay, Sudip K. Naskar, and Asif Ekbal, 110–128. Hershey, PA: IGI Global.
Chesterman, Andrew. 2008. “On Explanation.” In Beyond Descriptive Studies. Investigations in
Homage to Gideon Toury, ed. by Anthony Pym, Miriam Schlesinger, and Daniel Simeoni,
363–379. Amsterdam: John Benjamins. DOI: 10.1075/btl.75.27che
Englund Dimitrova, Birgitta. 2005. Expertise and Explicitation in the Translation Process.
Amsterdam: John Benjamins. DOI: 10.1075/btl.64
Gile, Daniel. 2009. Basic Concepts and Models for Interpreter and Translator Training. Revised
edition. Amsterdam: John Benjamins. DOI: 10.1075/btl.8
Göpferich, Susanne. 2009. “Towards a Model of Translation Competence and its Acquisition:
The Longitudinal Study TransComp.” In Behind the Mind: Methods, Models and Results in
Translation Process Research, ed. by Susanne Göpferich, Arnt Lykke Jakobsen, and Inger M.
Mees, 11–37. Copenhagen: Samfundslitteratur.
Holz-Mänttäri, Justa. 1984. Translatorisches Handeln. Helsinki: Suomalainen Tiedeakatemia.
Hönig, Hans. 1995. Konstruktives Übersetzen. Tübingen: Stauffenberg.
Jääskeläinen, Riitta. 1999. Tapping the Process: An Explorative Study of the Cognitive and Affective
Factors Involved in Translating. Joensuu: University of Joensuu.
Jakobsen, Arnt Lykke. 2011. “Tracking Translators’ Keystrokes and Eye Movements with
Translog.” In Methods and Strategies of Process Research, ed. by Cecilia Alvstad, Adelina
Hild, and Elisabet Tiselius, 37–55. Amsterdam: John Benjamins. DOI: 10.1075/btl.94.06jak
Jones, Francis R. 2011. Poetry Translating as Expert Action: Processes, Priorities and Networks.
Kinnunen, Tuija, and Kaisa Koskinen (eds). 2010. Translators’ Agency. Tampere: Tampere
University Press. http://urn.fi/urn:isbn:978-951-44-8082-9.
Koskinen, Kaisa. 2008. Translating Institutions. Manchester: St. Jerome.
Krings, Hans P. 1986. Was in den Köpfen von Übersetzern vorgeht: eine empirische Untersuchung
zur Struktur des Übersetzungsprozesses an fortgeschrittenen Französischlernern. Tübingen:
Gunter Narr.
Künzli, Alexander, and Maureen Ehrensberger-Dow. 2011. “Innovative Subtitling. A Reception
Study.” In Methods and Strategies of Process Research, ed. by Cecilia Alvstad, Adelina Hild,
and Elisabet Tiselius, 187–200. Amsterdam: John Benjamins. DOI: 10.1075/btl.94.14kun
Kussmaul, Paul. 2007. Verstehen und Übersetzen. Tübingen: Gunter Narr.

Leppihalme, Ritva. 1997. Culture Bumps. Clevedon: Multilingual Matters.
Levý, Jiří. 1967. “Translation as a Decision Process.” In To Honor Roman Jakobson. Janua
Linguarum Series Major 33, 1171–1182. Berlin: Mouton de Gruyter.
Lörscher, Wolfgang. 1989. “Models of the Translation Process: Claim and Reality.” Target 1 (1):
43–68. DOI: 10.1075/target.1.1.05lor
Lörscher, Wolfgang. 1991. Translation Performance, Translation Process and Translation
Strategies. Tübingen: Gunter Narr.
Nida, Eugene A. 1969. “Science of Translating.” Language 45 (3): 483–498. DOI: 10.2307/411434
Nord, Christiane. 1991. Text Analysis in Translation. Amsterdam: Rodopi.
Nordman, Lieselott. 2009. Lagöversättning som process och produkt. Helsingfors: Nordica,
Helsingfors universitet.
Pym, Anthony. 1998. Method in Translation History. Manchester: St. Jerome.
Reiss, Katharina, and Hans J. Vermeer. 1984. Grundlegung einer allgemeinen Translationstheorie.
Tübingen: Niemeyer. DOI: 10.1515/9783111351919
Risku, Hanna. 2010. “A Cognitive Scientific View of Technical Communication and Translation.
Do Embodiment and Situatedness Really Make a Difference?” Target 22 (1): 94–111.
DOI: 10.1075/target.22.1.06ris
Sager, Juan C. 1994. Language Engineering and Translation — Consequences of Automation.
Schmidt, Morena Azbel. 2005. “How Do You Do it Anyway?” A Longitudinal Study of Three
Translator Students Translating from Russian into Swedish. Stockholm: Stockholm
University. http://su.diva-portal.org/smash/record.jsf?pid=diva2:197703.
Shreve, Gregory M., Isabel Lacruz, and Erik Angelone. 2011. “Sight Translation and Speech
Disfluency: Performance Analysis as a Window to Cognitive Translation Processes.”
In Methods and Strategies of Process Research, ed. by Cecilia Alvstad, Adelina Hild, and
Elisabet Tiselius, 93–120. Amsterdam: John Benjamins. DOI: 10.1075/btl.94.09shr
Toury, Gideon. 2002. “What’s the Problem with ‘Translation Problem’?” In Translation
and Meaning 6, ed. by Barbara Lewandowska-Tomaszczyk, and Marcel Thelen, 57–71.
Maastricht: Hogeschool Zuyd, Maastricht School of Translation and Interpreting.
Toury, Gideon. 1995. Descriptive Translation Studies and Beyond. Amsterdam: John Benjamins.
DOI: 10.1075/btl.4
Toury, Gideon. 2012. Descriptive Translation Studies — and Beyond. Revised edition. Amsterdam:
John Benjamins. DOI: 10.1075/btl.100
Wolf, Michaela, and Alexandra Fukari (eds). 2007. Constructing a Sociology of Translation.
Shared representations
and the translation process
A recursive model
Moritz Schaeffer and Michael Carl1
The purpose of the present chapter is to investigate automated processing during

translation. We provide evidence from a translation priming study which suggests
that translation involves activation of shared lexico-semantic and syntactical repre-
sentations, i.e., the activation of features of both source and target language items
which share one single cognitive representation. We argue that activation of shared
representations facilitates automated processing. The chapter revises the literal
translation hypothesis and the monitor model (Ivir 1981; Toury 1995; Tirkkonen-
Condit 2005), and re-defines it in terms of findings from translation process
research. On the basis of the evidence, we propose a recursive model of translation.
Keywords: literal translation, priming, monitor model
1. Introduction
Guided by the question of what is a translation, Toury (1995: 144) put forward the
notion of “assumed translation” and postulated that a translation is a text for which
“there [is assumed to be] another text, in another culture/language, which has both
chronological and logical priority over it.” Toury (1995: 145) further suggested that
the latter postulate “entails the assumption that the process whereby [the translation]
came into being involved the transference from the assumed source text of certain
features that the two now share.” Toury (1995: 145) further argued that the latter two
postulates imply “that there are accountable relationships which tie [the translation]
to its assumed original, an obvious function of that which the two texts allegedly
1. The authors would like to thank the editors and the reviewers for their valuable and insightful
comments and suggestions.
doi 10.1075/bct.77.03sch
22 Moritz Schaeffer and Michael Carl
share…” In other words, the relationship between source and target text can be de-
fined in terms of what both texts share — during the production of the target text.
In terms of the investigation of the process of translation, a central question
is therefore when and how the features which are shared between source and tar-
get text are identified, because identification of interlinguistic equivalence is a
distinctly translational process and one which could clearly distinguish reading
and writing for translation from reading and writing for comprehension or other
forms of rewriting.
Two models of the translation process, which have been adapted from the
models representing the bilingual lexicon have been referred to as “vertical ver-
sus horizontal translation” (De Groot 1997: 30). These two translation models dif-
fer in terms of the assumed cognitive processes. The vertical model is essentially
composed of two monolingual systems: one for decoding the source language and
one for encoding the captured meaning in the target language. The horizontal
translation model implies that items in the lexicons of the translator are linked via
shared memory representations and that cognitive processes during translation
are specific to the combination of both languages involved (see horizontal arrows
in Figure 1). In both translation models a source text is read and a target text is
produced, but they differ in terms of the processes which are specific to transla-
tion. The X in equation (1) represents these processes:
(1) Translation = Source Text Reading + Target Text Writing + X
On the one hand, vertical translation assumes that the source text is parsed and
abstracted into more or less language specific concepts or even non-linguistic
concepts and then re-expressed in the target language. Only after the transforma-
tion of input tokens into concepts does the translator proceed to translate these
concepts into lemmas which belong to the target language and finally into the
output. In vertical translation, there is no clear distinction between monolingual
and bilingual processing, because higher level conceptual representations or pos-
sibly non-linguistic conceptual representations are not necessarily specific to any
given language and may be involved in monolingual and bilingual processing
equally: whether a conceptual meaning, which is not specific to a given language,
is re-expressed in a different or the same language does not necessarily distinguish
translation from re-writing or a general problem solving activity. In other words,
the vertical translation method involves monolingual source language decoding
processes, conceptual, non-linguistic representations, monolingual target lan-
guage encoding and comparison between the propositional content of source and
target language utterances, i.e., problem solving (this process is represented by the
vertical arrows in Figure 1).
Shared representations and the translation process 23
Problem solving
Contextualised Contextualised
propositional propositional
content content
Shared representations
(limited context)
Monolingual
Monolingual
decoding
encoding
NP VP NP
They pushed the boat Sie schoben das Boot
Figure 1. Horizontal and vertical translation
In horizontal translation, items in the two languages are linked via shared rep-
resentations. In this case, a source text noun phrase — verb phrase — noun phrase
construction activates a cognitive representation which it shares with the target
text. In vertical translation, source text items are decoded once enough context
is available and the resulting propositional content is encoded into the target lan-
guage once translation problems are solved.
The data elicited by the use of Think Aloud Protocols (TAPs) in research on
translation processes has focused on the description of translation as the employ-
ment of translation strategies (e.g., Krings 1986; Lörscher 1991). While these strat-
egies may deal with the relationship between texts in different languages, this does
not necessarily imply that the cognitive processing during this kind of problem
solving is any different from the one employed when solving problems in rela-
tion to a (number of) text(s) in one language. Diamond and Shreve (2010) review
studies which search for neural and physiological correlates of translation and find
that the evidence does not seem to suggest that neural activity during translation
is different from neural activity during non-translational language use. The evi-
dence does however seem to suggest that translation is a complex problem solving
activity, but not a “primarily linguistic one” (Diamond and Shreve 2010: 309). The
variable X in equation 2 represents translation specific processes, if translation is
viewed as primarily a problem solving activity.
(2) X = problem solving
Horizontal translation, on the other hand, assumes that translation does not nec-
essarily occur via conceptual, but via shared semantic and syntactic representa-
tions: via “replacement of SL linguistic structures of various types (words, phrases,
clauses) by the corresponding TL” (De Groot 1997: 30). Paradis (1994: 329) sug-
gests that in the horizontal method, equivalence is established on different levels
by the application of automatic rules which have been learned during training. In
most cases, morphology, syntax, and semantics are the most likely locus of “trans-
ference” (Toury 1995: 145) in translation.
Horizontal translation does not necessarily require higher level represen-
tations, i.e., it is not necessary to fully comprehend either source or target text.
Paradis (1994: 332) argues that the horizontal translation method is subserved by
explicit metalinguistic knowledge, whereas the vertical translation method is sub-
served by implicit linguistic competence. The horizontal translation model implies
that the cognitive processes during translation are specific to the two languages in-
volved, because items in the lexicons of the translator are linked via shared memo-
ry representations. Source text reading activates, thus, shared representations and
the same shared representations serve as a basis for target language writing. This
process is represented in equation 3.
(3) X = activation of shared memory representations
In the next sections we reformulate the literal translation hypothesis and the
monitor model (Ivir 1981; Toury 1995; Tirkkonen-Condit 2005), and we provide
evidence for the horizontal translation model as a default mode in translation
processing. We look at a number of experiments from psycholinguistics which
indicate, on the one hand, that recall does not occur on the basis of verbatim
representations, i.e., that the target text is regenerated on the basis of heightened
activation of different aspects of the items to be recalled. On the other hand, prim-
ing experiments show that much of the translation activities can be explained by
shared syntactical and semantic representations. We further provide evidence that
shared representations play a central role during automatic translation. We will
then discuss a recursive model of translation which integrates strategic problem
solving processes with automated translation processes.
2. The literal translation hypothesis and the monitor model
In equivalence theories such as that by Catford (1965), sameness is defined on the

basis of input and output, i.e., source and target text or language. The study by Al-
Kufaishi (2011) may serve as an example: Al-Kufaishi applies the notion of equiv-
alence put forward by Catford (1965) by analyzing obligatory departures from
formal correspondence, dictated by structural and other typological differences

between English and Arabic, in a variety of source and target texts as a measure
of quality. The assumption is that a literal translation, according to “formal cor-
respondence” (Catford 1965), is more similar to the source text than a non-literal
translation. Tirkkonen-Condit (2005), on the basis of what Ivir (1981) and Toury
(1995) proposed, assigns both the monitor and formal correspondence prominent
roles in the translation process. The data presented by Tirkkonen-Condit seeks
to assess whether the “monitor model” (Toury 1995: 191) is an accurate repre-
sentation of the translational process. This model assigns the translator the role
of reader of their own translation while producing the target text and simultane-
ously assumes that the translator’s default procedure is to translate according to
formal correspondence and that departures from formal correspondence only oc-
cur when the translator, in their role of reader, finds that contextual aspects of the
source text are not adequately referred to in the resulting target text.
Tirkkonen-Condit (2005: 408) argues that literal translations are triggered
through automatic processes. Literal translations may be interpreted as evidence
for the associated horizontal translation processes and activation of shared repre-
sentations, but it is equally possible to find literal translations which are the result
of a non-automatic process, as suggested by Englund Dimitrova (2005: 54). We
therefore suggest disentangling the literal and the automatic translation processes:
while literal translation refers to the translation product, automatic translation de-
scribes how the translation was produced, as described in equation 4:
(4) literal translation product ⇏ automatic translation processes
In other words, it is not possible to conclude whether the processes by which the
target text was produced were either horizontal or vertical if only the product is
taken into consideration. Whether a highly literal target text has been produced in
a very automatic manner or not cannot be determined on the basis of a compari-
son between source and target text alone.
3. Automatic translation
We understand the term automatic in Tirkkonen-Condit (2005) as referring to

procedural, implicit mechanisms and memory which are not accessible to con-
sciousness (Paradis 2004: 38). When shared representations form the basis for
translation, identification of shared aspects is automatic and there is no conscious
control over how source and target are aligned cognitively. Implicit procedural
processes cannot be verbalized and are not consciously controlled, once initiated.
Paradis (2004: 42) maintains that automatic processes can be stopped and replaced
by controlled, conscious processing, but only on the basis of the output of the
automatic process, not on the basis of the aspects of the automatic process itself.
Only after the output has been received and evaluated by some kind of monitor
can these processes become consciously controlled. Therefore, we assume that the
monitor plays a prominent role in the translation process and that shared rep-
resentations are accessed very early during the process, i.e., that the horizontal
method is a very early process while the vertical method is a later process.
This understanding of the horizontal method differs from how Paradis (1994)
understands the horizontal method, because Paradis (1994: 332) assumes that ex-
plicit metalinguistic knowledge regarding translation equivalents forms the basis
for the horizontal method. Our view is that priming forms the basis for the hori-
zontal method: the influence of a previously processed item or structure on a sub-
sequently processed item or structure forms the basis for horizontal translation.
This is in line with Pickering and Ferreira (2008: 447, italics in the original) who
argue that priming “reflects the operation of an implicit learning mechanism,” i.e.,
that repeated exposure to primes creates long-lasting memories. So rather than
learning about translational equivalents, implicit mechanisms during repeated ex-
posure to source and target texts establishes shared representations in the transla-
tor’s long term memory.
4. The regeneration hypothesis (Potter and Lombardi 1990; Lombardi and

Potter 1992)
When copying a text by re-typing it in the same language, “no lexical or structural
transfer is required” (Carl and Dragsted 2012: 129). Apart from corrections of ty-
pos and given adequate blind typing skills, reading and writing can thus occur
concurrently. In theory, it is possible to assume that during the process of copying
a text, the same representations which have been activated during reading are also
activated during writing. In other words, it is possible to assume that for copying,
lexical access is sufficient in order to allow for a reproduction of the text and that
higher levels of e.g., discourse integration are not necessary. But the data (eye-
tracking and keyboard logging) Carl and Dragsted (2012) present shows that par-
ticipants do reread the text to be copied even when there was no typo. In particular,
the data shows that regressions (re-reading of already fixated words) in the source
text are triggered when a particular item is literally easy to comprehend, but dif-
ficult to integrate contextually. Interestingly, re-reading is triggered by production
activity. Carl and Dragsted find re-reading occurred immediately before typing
corresponding items — both during copying and during translation. Regressions
during reading are normally associated with processes which integrate lexical
representations into larger units of meaning (e.g., Reichle, Warren, and McConnel
2009). This suggests that even when source and target are literally the same (as
is the case when copying a text), non-literal or non-verbatim representations are
activated and serve as a basis for regeneration in the output.
Carl and Dragsted (2012) also confirmed the finding from Dragsted (2010)
that translators only look a few words ahead when reading and writing at the same
time. This means that production of a target text item normally depends on recall
of the source text item. When typing the translation of word N, the translator’s
eyes normally do not fixate word N, but a different word a few positions ahead in
the text. These findings from Carl and Dragsted (2012) and Dragsted (2010) are
in line with the Potter-Lombardi Regeneration hypothesis (Potter and Lombardi
1990; Lombardi and Potter 1992). Potter and Lombardi (1998: 265) argue that re-
call does not occur on the basis of a verbatim representation in memory, but on
heightened activation, i.e., primed representations, but maybe more importantly,
that rather than reproducing a verbatim representation of the sentence, partici-
pants regenerate the sentence on the basis of the primed representations. In sum,
even within the same language and when re-writing results in a verbatim copy of
the source text, literality is not a concept which adequately describes the processes
associated with this reading and writing behavior. Instead, we suggest describing
translation processes in terms of automaticity and shared representations.
5. Recall as a measure of monolingual priming
In priming studies, participants are not aware of the effects of primes. Priming is
normally interpreted as evidence of implicit mechanisms and not explicit memory
retrieval. Potter and Lombardi (1998) presented native English participants first
with the target sentence and then with the prime sentence. The recall of the target
sentence was evaluated in terms of the influence from the prime sentence. For
example, when the target was a double-objective dative sentence, such as “The
prompt secretary wrote her boss a message every week,” the prime sentence was
a prepositional dative sentence, such as “The tycoon willed that mansion to his
young nephew grudgingly” (Potter and Lombardi 1998: 268). They found that par-
ticipants were significantly more likely to recall a sentence with the same syntactic
structure as that of the prime sentence than with an alternative syntactic structure.
6. Bilingual priming
The bulk of structural priming studies investigates language effects within one lan-
guage. Fewer studies have presented primes and targets in different languages. They
are, though, consistent in their findings: the results from a number of studies using
two languages (Heydel and Murray 2000; Meijer and Fox Tree 2003; Loebell and
Bock 2003; Desmet and Declercq 2006; Salamoura and Williams 2007; Schoonbaert,
Hartsuiker, and Pickering 2007; Kantola and van Gompel 2011) seem to suggest
that if particular features of syntax exist in both of the bilingual’s languages, then
they are likely to prime across languages. These studies used genetically and ty-
pologically close languages: English, French, German, Dutch, Spanish, and Greek.
These findings are in line with Hartsuiker, Pickering, and Veltkamp’s (2004) shared
syntax account. The authors propose that, where the syntax of the two languages of
the bilingual are sufficiently similar, they share a representation. In addition, there
is ample evidence for semantic representations which are shared across languages.
7. Semantic bilingual priming
In semantic cross-linguistic priming studies, participants are presented with primes

in either L1 or L2 and targets in the other language of the bilingual and are then
asked to carry out, for example, a lexical decision task, in which they are asked to
decide whether the target word exists or whether it is a non-word. The time it takes
to recognize e.g., the word in L2 correlates in most studies with its relationship to the
previously presented L1 word: if the L1 prime word is semantically related (e.g., boy–
niña [girl]) to or a translation equivalent of the L2 target word (boy–niño [boy]),
participants require significantly less time to recognize it than when it is semanti-
cally unrelated (boy–mesa [table]). Crucially, this effect is observed when the prime
is masked, i.e., when it is presented for a very short time (40–100ms) (Duñabeitia,
Perea, and Carreiras 2010). This suggests that the effect of the prime represents early
automatic processes which supports shared semantic representations.
In earlier studies, this effect is only observed from L1 to L2, but not from L2 to
L1 (e.g., Keatley, Spinks, and De Gelder 1994; Gollan, Forster, and Frost 1997; Jiang
1999; Jiang and Forster 2001). This is important in the context of translation, because
translators often translate in either both directions or from L2 into L1. But a number
of more recent studies addressed methodological inconsistencies within and across
studies and found that the effect in primes from L2 to L1 was comparable to the
reverse direction (Basnight-Brown and Altarriba 2007; Duyck and Warlop 2009;
Duñabeitia et al. 2010; Dimitropoulou, Duñabeitia, and Carreiras 2011; Schoonbaert
et al. 2011). These studies used a variety of language combinations: Spanish-English,

Dutch-French, Greek-Spanish, Basque-Spanish and English-French.
8. The shared syntax account (Hartsuiker et al. 2004)
Hartsuiker et al.’s shared syntax account (2004) extends Pickering and Branigan’s
(1998) monolingual model of how lemmas are represented cognitively. Hartsuiker
et al. (2004: 481) assume that entries in the bilingual mental lexicon consist of
“combinatorial nodes” which connect lemmas, concepts, word forms, and syntac-
tic information. This means that these nodes are fairly abstract and general and
far from verbatim representations and this means also that many different words
from both languages share nodes at the respective levels of lemma, concept, word
form, and syntactic information: “…such combinatorial nodes are connected to
all words with the relevant properties, irrespective of language” (Hartsuiker et al.
2004: 481). However, aspects which are shared between the two languages in the
bilingual mind are specific to the language combination, because not all languages
always share the same aspects. This makes shared representations very different
from non-linguistic conceptual representations which can be shared by all lan-
guages. It is very likely that transference during horizontal translation occurs via
these shared representations on different levels (semantics, syntax, etc.): accessing
one relatively abstract and general shared representation or combinatorial node,
rather than two language specific verbatim representations is faster and more effi-
cient. Regeneration (Potter and Lombardi 1990; Lombardi and Potter 1992) on the
basis of combinatorial nodes with a maximum amount of overlap between source
and target text resembles copying rather than problem solving: the source text is
read, shared representations are primed and the same shared representations serve
as a basis for the regeneration in the target language.
9. Syntactic bilingual priming
Meijer and Fox Tree (2003) report from a study which used recall as a measure of
priming, but participants in this study were Spanish-English bilinguals and prime
and target sentences were presented in two different languages. The English target
sentences were constructed with verbs which allowed both double-object and prep-
ositional object constructions, such as “the waitress will bring the noisy customers
a tray of drinks” which can also be expressed as “the waitress will bring a tray of
drinks to the noisy customers” (Meijer and Fox Tree 2003: 187). The Spanish prime
sentences always had a prepositional object. And after the presentation of Spanish
prime sentences with a prepositional object, participants were significantly more

likely to recall the English target sentence with a prepositional object rather than
a double-object construction. In Experiment 3, Meijer and Fox Tree reversed the
languages of prime and target sentences and still found that participants were sig-
nificantly more likely to recall the Spanish target sentence with the same structure
as the English prime sentence than in the control condition.
Prime and target sentences in these studies were not translations of each other,
but the syntactic form of the prime sentences still influenced the recall of the target
sentence. These priming studies suggest that bilinguals have access to shared rep-
resentations and that these representations are accessed automatically.
10. Priming in translation
Participants in priming studies are normally assumed to be unaware of the effects

of primes on targets. Structural priming studies try to trick participants into using
a particular syntactic structure when e.g., describing a picture after having heard
and repeated an unrelated sentence with that particular syntactic structure (e.g.,
Bock 1986). In these studies, prime and target sentences are normally unrelated in
terms of the semantic features. But during translation, using both the same syntac-
tic structure and the same semantic features, apart from other aspects, is not nec-
essarily something translators need to be tricked into: it is often the implicit aim to
produce that very semantic and syntactic sameness in the target text — when the
result is acceptable to target norms, the skopos or any other consideration regard-
ing acceptability.
The study by Vandepitte and Hartsuiker (2011) was designed with the aim
of testing whether cross-linguistic differences are cognitively more effortful than
cross-linguistic similarities in order to make translation students aware of these
differences and thus improve training. Vandepitte and Hartsuiker investigated
cross-linguistic priming and used stimuli which could not be translated using the
same agent and predicate as in the source sentence. The study by Vandepitte and
Hartsuiker raises a fundamental question in this regard: “Is it still possible to call
this type of priming, in which there is a combination of semantic features with a
specific structure, structural priming?” (2011: 81). This question is also raised by
the studies of Ruiz et al. (2008) and Macizo and Bajo (2006), who suggest that the
reading purpose (reading for translation versus reading for repetition) has an ef-
fect on the activation of target language representations during reading. In other
words, it is possible that translators encode the text during reading for transla-
tion in a way that favors the activation of shared representations. The study by
Vandepitte and Hartsuiker (2011) suggests that shared representations facilitate
the process of translation as long as the emerging target text is acceptable but also
that shared representations inhibit the process of translation when the emerging
target text does not conform to target norms. The stimuli in the study by Vandepitte
and Hartsuiker were tightly controlled in terms of the acceptability of a target
text with the same agent and predicate, i.e., using the same agent and predicate
in the translation of critical items was not acceptable, while this was acceptable
in the translation of control items. Vandepitte and Hartsuiker found significantly
longer translation (writing) times for critical items and interpreted these results
as evidence for cognitive effort during translation of items with cross-linguis-
tic differences as compared to items with cross-linguistic sameness. Vandepitte
and Hartsuiker instructed participants “to type the first translation they found”
(2011: 76, italics in the original). It is therefore possible that not only the nature of
the stimuli, but also the translation brief, i.e., the reading or translation purpose,
has an effect on how the stimuli are encoded. That how the source text is encoded
is a matter of choice, strategy or habit was already suggested by Isham (1994) on
the basis of a study which investigated recall in simultaneous interpreters.
Our study aimed to test whether translation primes semantic and syntacti-
cal representations, as described by Vandepitte and Hartsuiker (2011: 81), because
this could suggest that shared representations facilitate the process of translation.
We wanted to test whether reading a sentence in one language and translating it
into another language primes representations shared between the two languages
more than reading only for comprehension does. We used verbatim recall as a
measure of priming (Potter and Lombardi 1998; Meijer and Fox Tree 2003). The
source text was the target and the translation was the prime.
10.1 Participants
Six translators who habitually (at least 5 years’ experience) translated into German
from English and five translators who habitually translated into French from
English were selected from registers available at professional bodies for translators
(IOL, ITI, and AIIC). All participants had English as their L2. No other require-
ment was taken into account for the selection of participants. The results from one
French participant had to be excluded, because she recalled the target text instead
of the source text in the translation condition.
10.2 Procedure
Two excerpts (844 and 794 words) from The Old Man and the Sea (Hemingway
1975) were presented one sentence at a time using ePrime. The instruction for the
translation condition was to produce a rough written draft of each sentence, not
a polished translation. During the translation condition, the source text sentence
remained visible while translating. To indicate that they were happy with their
translation, the participants hit the escape key. A new sentence or a recall prompt
would appear and the previous sentence would no longer be visible. The procedure
during the reading for comprehension condition was exactly the same apart from
the fact that participants did not translate. Recall prompts appeared every 1–5 sen-
tences at irregular intervals during both conditions. The recall prompt instructed
participants to recall as much as possible of the sentences they had either just read
for comprehension or which they just translated, and to type as many words from
the last recall prompt up to and including the last word before the current recall
prompt. It was stressed that recall should be as accurate as possible. There were 24
recall prompts in each condition. The order of presentation of the two excerpts
and tasks were counterbalanced.
10.3 Scoring
Scoring was carried out in two different ways. The lexical score represents lexical
aspects of the recall and disregards syntax. The total number of possible correct
scores is the number of words in each of the two texts (844 and 794). For the lexi-
cal score, only words which were recalled exactly in the same form were counted
as correct, with the exception of singular/plural. Figure 2 is an example of how
the lexical score was calculated: for this sentence, the score was 6/7 (86%), be-
cause [the trousers] was recalled as [his trousers]. For the lexical score, words were
counted as correct irrespective of word order (see Figure 3). For the sentence in
Figure 3, the lexical score was 10/14 (72%).
Source text He put the newspaper inside the trousers.
Recall He put the newspapers inside his trousers.
Figure 2. Lexical score example 1
Source text To wake the girl from her sleep he went over to their little house.
Recall He went over to wake the girl in their little house.
Source text To wake the girl from her sleep he went over to their little house.
Correct to wake the girl He went over their little house.
Figure 3. Lexical score example 2

For the syntactical score, the recall and the source text were parsed using the
Berkeley Parser (Klein and Petrov 2007) and each parsed sentence was aligned
starting with the last word. Only words which were recalled in the same order
(starting from the end) and with the same part of speech tag were counted as cor-
rect. This is in line with Jarvella (1971) who also employed two recall measures:
recalled sentences aligned from the end and recall of words irrespective of order.
The total number of possible correct syntactical matches for the two texts was 1553
and 1598, which correspond to the total number of part of speech and phrasal tags
for each text. Only recalled words in exactly the same position (from the end of the
sentence) and with exactly the same part of speech and phrasal tags were counted
as correct disregarding lexical differences — the fact that [slid] was recalled as
[pushed] did not affect the recall score (see Figure 4). The syntactical score for this
sentence was 19/19 (100%).
original original translation recall recall

they ((S(NP(PRP they)) Sie they ((S (NP (PRP they))
lifted (VP(VP(VBD lifted) hoben lifted (VP (VP (VBD lifted)
the (NP(DT the) das the (NP(DT the)
boat (NN boat))) Boot boat (NN boat)))
and (CC and) an and (CC and)
slid (VP(VBD slid) und pushed (VP(VBD pushed)
her (NP(PRP her)) schoben her (NP(PRP her))
into (PP(IN into) es into (PP(IN into)
the (NP(DT the) ins the (NP(DT the)
water. (NN water))))))) Wasser. water. (NN water)))))))
Figure 4. Syntactical score example
10.4 Results
The lexical and syntactical scores for each participant in each condition in Table 1
are reported as mean correct recall as proportions of 1.
Within participants comparison showed that participants recalled on aver-
age significantly more after translation than after reading for comprehension on
both scores: Lexical recall (t(9) = 3.38, p < .008) and syntactical recall (t(9) = 3.34,
p < .009). Figure 6 shows the mean recall score for both measures and tasks; error
bars indicate the confidence interval (95%). The magnitude of the effect accord-
ing to Pearson’s correlation coefficient for the syntax score was large (r(8) = .584,
p > .076), though not significant and the magnitude of the effect for the lexical
score was also large (r(8) = .599, p > .067), but not significant.
Table 1. Mean recall scores (lexical and syntactical) for each participant
Lexical Syntactical
Participant Translation Reading Translation Reading
German 1 0.81 0.49 0.57 0.33
German 2 0.61 0.65 0.25 0.29
German 3 0.74 0.64 0.31 0.32
German 4 0.71 0.75 0.46 0.45
German 5 0.76 0.51 0.42 0.25
German 6 0.80 0.69 0.42 0.37
French 1 0.84 0.78 0.55 0.45
French 2 0.77 0.65 0.38 0.21
French 3 0.48 0.26 0.26 0.07
French 4 0.80 0.56 0.49 0.27
1.00
0.80
0.60
Mean
Translation
Reading
0.40
0.20
0.00
Lexis Syntax
RecallMeasure
Error Bars: 95% CI
Figure 5. Lexical and syntax mean recall scores for reading for comprehension versus
translation
10.5 Discussion
The significantly higher recall during translation from English into French and
into German than during English monolingual comprehension may be explained
through increased activation of shared representations during translation. The re-
sults suggest that during reading for comprehension, a single presentation of the
text activates monolingual English representations. During translation, monolin-
gual English representations are activated, but, where these representations coin-
cide with French or German target language representations, these shared aspects
are activated twice and hence facilitate recall and increase the overall recall scores.
Contextualised
propositional content
Monitoring
Decoding Decoding
SL Semantics TL Semantics
SL Syntax NP VP NP TL Syntax
SL Morphology TL Morphology
SL Phonology TL Phonology
SL Phonetics TL Phonetics
SL Orthography TL Orthography
Encoding Encoding
They pushed the boat Sie schoben das Boot
Figure 6. Recursive model of translation
Pickering and Branigan’s (1998) model of the lemma stratum predicts a lexi-
cal boost when primes and targets have identical verbs. Hartsuiker et al.’s model
(2004) also predicts a translation equivalent boost via shared semantic represen-
tation. This lexical boost is normally found from L1 to L2. During translation,
many words in a sentence are lexical translation equivalents and in this study, the
target text was the prime and was in the L1. In this sense, the higher recall is best
explained by reference to repeated access of the same representations.
Our experiment was not controlled in terms of the stimuli and is ecologically
valid to a certain extent: both tasks were relatively continuous in that the indi-
vidually-presented sentences in both tasks were part of a longer text. Traditional
priming studies require control stimuli in order to make sure that the use of a
particular structure in the target sentence is an effect of the prime structure. The
control stimuli in, for example, Potter and Lombardi (1998) were sentences with a
structure which could not prime the target, i.e., the authors used intransitive sen-
tences as controls for transitive target sentences. In translation, it is more compli-
cated to find control stimuli because only syntactic structures which cannot have
a shared representation can be said not to prime. In other words, only sentences
which cannot be translated using the same cognitive representation can be said
not to prime and therefore act as control stimuli. But to decide whether something
is untranslatable in this sense requires considerations of acceptability and this is

something which is not always easy to define: the distinction between acceptability
and unacceptability is not always clear-cut and may vary across contexts.
In terms of shared representations, translation is at least partially a naturally
controlled experimental environment in that the instruction to produce a rough
translation acted as an instruction to exclude target sentences which are (se-
mantically and syntactically) very dissimilar to the source sentence. This means
that for the translator, every sentence acts as a prime — when cross-linguistic
similarity allows this. In our study, we interpreted the recall score as a measure of
whether a particular sentence can prime between the two languages involved, or
not, as compared to reading for comprehension. A clear weakness of this study
is that we did not control the number of words and sentences to be recalled,
although this clearly has an effect on the recall score. We favored instead an eco-
logically more valid authentic text rather than a text which controlled for such
factors.
11. The revised literal translation hypothesis
Rather than describing the cognitive processes during translation in terms of the
literality of the target text in relation to the source text as in Tirkkonen-Condit,
Mäkisalo, and Immonen (2008), we propose to describe the literality of the pro-
cess in terms of shared representations and automaticity. Englund Dimitrova
(2005: 53) argues that “in the translation process there is an alternation between
automated and strategic problem-solving processing.” Automaticity is here in-
dexed by pauses in the typing activity and/or by problem indicators in the TAPs
(Englund Dimitrova 2005: 156). We propose that future studies could measure
automaticity in terms of the degree to which source text reading and target text
production occur at the same time. Hartsuiker et al. (2004: 487) predict that trans-
lation and simultaneous interpretation “will be facilitated when the two languages
have parallel syntactic structures; this is because processing the source structure
will act as a syntactic prime that leaves the target structure with residual activa-
tion.” In other words, the closer the cognitive alignment between source and target
text representations, the more efficient is the process. We propose that instances
of concurrent reading and writing during translation are indicative of automatic
processes and shared representations. The resulting target text may or may not be
considered a formally corresponding translation according to equivalence theo-
ries such as that proposed by Catford (1965).
In the following section, we will explain how and why both the horizontal and
the vertical model can be integrated into one model of translation which accounts
for both automatic processes and problem-solving activity. We understand hori-

zontal processes as priming processes as described by Vandepitte and Hartsuiker
(2011: 81) and as evidenced by the priming study reported above.
12. Recursive model of translation: the revised monitor model
The model we propose is very similar to the one proposed by Christoffels and de
Groot (2005: 474, Figure 22.2 b). Christoffels and de Groot (2005: 471, our brack-
ets) argue that “…there is some [indirect] evidence to suggest that, in addition to
meaning-based [vertical] interpreting, transcoding [horizontal] also takes place.”
De Groot (2011: 324) proposes separate input and output lexicons for source and
target language systems to allow for target text production monitoring at the same
time as source text comprehension. In our model, vertical processes act as the
monitor for target text production by the horizontal processes. In other words, we
do not assume separate input and output lexicons for source and target language
(i.e., 4 lexicons). During decoding, both horizontal and vertical processes are al-
ways active at the same time. We assume that the horizontal process is an early
process while the vertical processes depend on context which becomes available
later, as processing advances in the chunk or text. Early during source text reading,
shared representations are activated which then serve as a basis for regeneration
in the target language (circles in Figure 6). As long as the target text being pro-
duced conforms to the target norms and contextual considerations of the vertical
processes, regeneration on the basis of shared representations is not interrupted.
But when the target text is not acceptable, the interim translation, either kept in
working memory or already partially produced as target text, is adapted to target
norms by vertical encoding processes (vertical arrows in Figure 6). Adaptation
during encoding to target norms is an essentially monolingual process. In other
words, vertical processes, i.e., monolingual encoding and decoding processes act
as monitor. In this sense, De Groot (2011: 322) argues that for “…such monitoring
to take place no special mechanism needs to be postulated…” As represented in
Figure 6, horizontal processes access shared representations (NP/VP/NP in this
case) automatically and early during the process. Vertical processes monitor out-
put from the horizontal automatic processes.
The shared syntax account (Hartsuiker et al. 2004) predicts that syntax is not
separate for each language. The neurofunctional model proposed by Paradis also
predicts:
…each language is represented as a subsystem of the language system. When one
language is selected for expression, the activation threshold of the other language
is raised so as to avoid interference… However, it is not raised so high that it could

not be activated by incoming verbal stimulus… (Paradis 2004: 115)
While entries in each of the different aspects of the language system may belong
to a different language which can be inhibited selectively, there is only one lan-
guage system — with one subsystem per language of the bilingual. In line with
Paradis (1994: 321) our model predicts that the monolingual monitoring activity
and the automatic default procedure use one language system and balance the
activation threshold for source and target language since during concurrent read-
ing and writing the activation threshold for both source and target language is
similarly high. When the translator is not reading and writing at the same time,
the activation threshold for the non-active language is higher than for the active
language. In both these modes, both languages are never completely inhibited and
always active to a certain extent as described by the “monolingual-bilingual mode
continuum” (Grosjean 1997: 168).
The vertical processes control the acceptability of the target text, but also need
to monitor equivalence. These processes need to assess whether source and target
are the same. In order to do this, the monitoring processes need to have a tertium
comparationis: the monitoring processes assess whether source and target texts
are compatible in terms of propositional content and shared conceptual represen-
tations — when the activation threshold for each language is relatively balanced.
The model consists of a recursive cycle which integrates horizontal and verti-
cal source and target language processes: the monitor needs to compare whether
the source is the same as the target, but it is equally important to make sure that
the target is the same as the source. Vertical processes access the output from the
automatic default procedure recursively in both the source and the target language
and monitor consistency as the context during translation production increases.
13. Conclusion
In this chapter, on the basis of a translation priming study, we revised the lit-
eral translation hypothesis and the monitor model (Tirkkonen-Condit 2005) in
terms of translation processes: an automatic translation procedure produces de-
fault translations on the basis of shared representations (horizontal translation).
Vertical processes interrupt this automatic process if the resulting target text is not
acceptable. Malmkjær (2005) argues that the evidence put forward by Tirkkonen-
Condit (2005), Eskola (2004) and Gellerstam (1996) suggests that the literal trans-
lation hypothesis might be cognitively determined and that it is not restricted to
one language combination of closely related languages. The latter combination of
factors makes it therefore a good candidate for a phenomenon which can be ob-
served across languages and contexts. We explained the literal translation hypoth-
esis in terms of the translator’s bilingual lexicon. This increases the likelihood that
automatic translation on the basis of shared representations is a process which
occurs across language combinations.
In sum, our model describes how the relationship between source and target
text can be defined in terms of what both texts share — during the production of
the target text.
References
Al-Kufaishi, Adil. 2011. “Obligatory Translation Shift as a Sub-component of a Model of Quality

Assurance Specifications and Performance Translator Assessment.” Babel 57: 144–167.
DOI: 10.1075/babel.57.2.02kuf
Basnight-Brown, Dana M., and Jeanette Altarriba. 2007. “Differences in Semantic and
Translation Priming across Languages: The Role of Language Direction and Language
Dominance.” Memory and Cognition 35: 953–965. DOI: 10.3758/BF03193468
Bock, Kathryn. 1986. “Syntactic Persistence in Language Production.” Cognitive Psychology 18:
355–387. DOI: 10.1016/0010-0285(86)90004-6
Carl, Michael, and Barbara Dragsted. 2012. “Inside the Monitor Model: Processes of Default
and Challenged Translation Production.” Translation: Computation, Corpora, Cognition 2:
127–145. http://t-c3.org/index.php/t-c3/article/view/18/23.
Catford, John C. 1965. A Linguistic Theory of Translation: An Essay in Applied Linguistics. Oxford:
Oxford University Press.
Christoffels, Ingrid K., and Annette M.B. De Groot. 2005. “Simultaneous Interpreting: A
Cognitive Perspective.” In Handbook of Bilingualism Psycholinguistic Approaches, ed. by
Judith F. Kroll, and Annette M.B. De Groot, 454–479. New York, NY: Oxford University
Press.
De Groot, Annette M.B. 2011. Language and Cognition in Bilinguals and Multilinguals: An
Introduction. New York, NY: Psychology Press.
De Groot, Annette M.B. 1997. “The Cognitive Study of Translation and Interpretation: Three
Approaches.” In Cognitive Processes in Translation and Interpreting, ed. by Joseph H. Danks,
Gregory M. Shreve, Stephen B. Fountain, and Michael McBeath, 25–56. Thousand Oaks,
CA: Sage Publications.
Desmet, Timothy, and Mieke Declercq. 2006. “Cross-linguistic Priming of Syntactic Hierarchical
Configuration Information.” Journal of Memory and Language 54: 610–632.
DOI: 10.1016/j.jml.2005.12.007
Diamond, Bruce J., and Gregory M. Shreve. 2010. “Neural and Physiological Correlates: Recent
Perspectives.” In Translation and Cognition, ed. by Gregory M. Shreve, and Erik Angelone,
289–323. Amsterdam: John Benjamins. DOI: 10.1075/ata.xv.16dia
Dimitropoulou, Maria, Jon A. Duñabeitia, and Manuel Carreiras. 2011. “Masked Translation
Priming Effects with Low Proficient Bilinguals.” Memory and Cognition 39: 260–275.
DOI: 10.3758/s13421-010-0004-9
Dragsted, Barbara. 2010. “Coordination of Reading and Writing Processes in Translation: An

Eye on Unchartered Territory.” In Translation and Cognition, ed. by Gregory M. Shreve, and
Erik Angelone, 41–62. Amsterdam: John Benjamins. DOI: 10.1075/ata.xv.04dra
Duñabeitia, Jon A., Manuel Perea, and Manuel Carreiras. 2010. “Masked Translation Priming
Effects with Highly Proficient Simultaneous Bilinguals.” Experimental Psychology 57: 98–
107. DOI: 10.1027/1618-3169/a000013
Duyck, Wouter, and Nele Warlop. 2009. “Translation Priming between the Native Language
and a Second Language: New Evidence from Dutch-French Bilinguals.” Experimental
Psychology 56: 173–179. DOI: 10.1027/1618-3169.56.3.173
Eskola, Sari. 2004. “Untypical frequencies in Translated Language: A Corpus-based Study on
a Literary Corpus of Translated and Non-translated Finnish.” In Translation Universals:
Do They Exist? ed. by Anna Mauranen, and Pekka Kujamäki, 83–99. Amsterdam: John
Benjamins. DOI: 10.1075/btl.48.08esk
Gellerstam, Martin. 1996. “Translation as a Source for Cross-linguistic Studies.” In Languages in
Contrast, ed. by Karin Aijmer, Bengt Altenberg, and Mats Johansson, 53–62. Lund: Lund
University Press.
Gollan, Tamar H., Kenneth I. Forster, and Ram Frost. 1997. “Translation Priming with Different
Scripts: Masked Priming with Cognates and Noncognates in Hebrew-English Bilinguals.”
Journal of Experimental Psychology: Learning Memory and Cognition 23: 1122–1139.
DOI: 10.1037/0278-7393.23.5.1122
Grosjean, François. 1997. “The Bilingual Individual.” Interpreting 2: 163–187.
DOI: 10.1075/intp.2.1-2.07gro
Hartsuiker, Robert J., Martin J. Pickering, and Eline Veltkamp. 2004. “Is Syntax Separate
or Shared between Languages? Cross-linguistic Syntactic Priming in Spanish-English
Bilinguals.” Psychological Science 15: 409–414. DOI: 10.1111/j.0956-7976.2004.00693.x
Hemingway, Ernest. 1975. The Old Man and the Sea. London: Jonathan Cape.
Heydel, Maren, and Wayne S. Murray. 2000. “Conceptual Effects in Sentence Priming: A Cross-
linguistic Perspective.” In Cross-linguistic Perspectives on Language Processing, ed. by Marica
De Vincenzi, and Vincenzo Lombardo, 227–254. Dordrecht: Kluwer Academic Publishers.
DOI: 10.1007/978-94-011-3949-6_9
Isham, William P. 1994. “Memory for Sentence Form after Simultaneous Interpretation:
Evidence both for and against Deverbalization.” In Bridging the Gap. Empirical Research in
Simultaneous Interpretation, ed. by Sylvie Lambert, and Barbara Moser-Mercer, 191–211.
Amsterdam: John Benjamins. DOI: 10.1075/btl.3.15ish
Ivir, Vladimir. 1981. “Formal Correspondence vs. Translation Equivalence Revisited.” Poetics
Today 2: 51–59. DOI: 10.2307/1772485
Jarvella, Robert J. 1971. “Syntactic Processing of Connected Speech.” Journal of Verbal Learning
and Verbal Behavior 10: 409–416. DOI: 10.1016/S0022-5371(71)80040-3
Jiang, Nan. 1999. “Testing Processing Explanations for the Asymmetry in Masked Cross-
language Priming.” Bilingualism: Language and Cognition 2: 59–75.
DOI: 10.1017/S1366728999000152
Jiang, Nan, and Kenneth I. Forster. 2001. “Cross-Language Priming Asymmetries in Lexical
Decision and Episodic Recognition.” Journal of Memory and Language 44: 32–51.
DOI: 10.1006/jmla.2000.2737
Kantola, Leila, and Roger P.G. van Gompel. 2011. “Between- and Within-Language Priming is
the Same: Evidence for Shared Bilingual Syntactic Representations.” Memory and Cognition
39: 276–290. DOI: 10.3758/s13421-010-0016-5
Keatley, Catharine W., John A. Spinks, and Beatrice De Gelder. 1994. “Asymmetrical Cross-
language Priming Effects.” Memory and Cognition 22: 70–84. DOI: 10.3758/BF03202763
Klein, Dan, and Slav Petrov. 2007. “Improved Inference for Unlexicalized Parsing.” In Proceedings
of Human Language Technology Conference of the North American Chapter of the Association
of Computational Linguistics (HLT-NAACL), Rochester, New York, 22–27 April 2007, ed. by
Candace L. Sidner, Tanja Schultz, Matthew Stone, and Cheng Xiang Zhai. Rochester, NY:
The Association of Computational Linguistics.
Krings, Hans P., 1986. Was in den Köpfen von Übersetzern vorgeht:eine empirische Untersuchung
zur Struktur des Übersetzungsprozesses an fortgeschrittenen Französischlernern [What
Happens in the Heads of Translators: An Empirical Investigation of the Structure of the
Translation Process in Advanced Students of French]. Tübingen: Gunter Narr.
Loebell, Helga, and Kathryn Bock. 2003. “Structural Priming across Languages.” Linguistics 41:
791–824. DOI: 10.1515/ling.2003.026
Lombardi, Linda, and Mary C. Potter. 1992. “The Regeneration of Syntax in Short Term Memory.”
Journal of Memory and Language 6: 713–733. DOI: 10.1016/0749-596X(92)90036-W
Lörscher, Wolfgang. 1991. Translation Performance, Translation Process and Translation
Strategies. A Psycholinguistic Investigation. Tübingen: Gunter Narr.
Macizo, Pedro, and Maria T. Bajo. 2006. “Reading for Repetition and Reading for Translation:
Do They Involve the Same Processes?” Cognition 99: 1–34.
DOI: 10.1016/j.cognition.2004.09.012
Malmkjær, Kirsten. 2005. “Norms and Nature in Translation Studies.” Synaps 16: 13–19.
Meijer, Paul J.A., and Jean E. Fox Tree. 2003. “Building Syntactic Structures in Speaking: A
Bilingual Exploration.” Experimental Psychology 50: 184–195.
DOI: 10.1026//1617-3169.50.3.184
Paradis, Michel. 2004. A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins.
DOI: 10.1075/sibil.18
Paradis, Michel. 1994. “Toward a Neurolinguistic Theory of Simultaneous Translation: The
Framework.” International Journal of Psycholinguistics, 10 (3): 319–335.
Pickering, Martin J., and Holly P. Branigan. 1998. “The Representation of Verbs: Evidence from
Syntactic Priming in Language Production.” Journal of Memory and Language 39: 633–651.
DOI: 10.1006/jmla.1998.2592
Pickering, Martin J., and Victor S. Ferreira. 2008. “Structural Priming: a Critical Review.”
Psychological Bulletin 134: 427–459. DOI: 10.1037/0033-2909.134.3.427
Potter, Mary C., and Linda Lombardi. 1990. “Regeneration in the Short-Term Recall of Sentences.”
Journal of Memory and Language 29: 633–654. DOI: 10.1016/0749-596X(90)90042-X
Potter, Mary C., and Linda Lombardi. 1998. “Syntactic Priming in Immediate Recall of
Sentences.” Journal of Memory and Language 38: 265–282. DOI: 10.1006/jmla.1997.2546
Reichle, Erik D., Tessa Warren, and Kerry McConnell. 2009. “Using E-Z Reader to Model
the Effects of Higher Level Language Processing on Eye Movements during Reading.”
Psychonomic Bulletin and Review 16: 1–21. DOI: 10.3758/PBR.16.1.1
Ruiz, Carmen, Natalia Paredes, Pedro Macizo, and Maria T. Bajo. 2008. “Activation of Lexical
and Syntactic Target Language Properties in Translation.” Acta Psychologica 128: 490–500.
DOI: 10.1016/j.actpsy.2007.08.004
Salamoura, Angeliki, and John N. Williams. 2007. “Processing Verb Argument Structure
across Languages: Evidence for Shared Representations in the Bilingual Lexicon.” Applied
Psycholinguistics 28: 627–660. DOI: 10.1017/S0142716407070348
Schoonbaert, Sofie, Robert J. Hartsuiker, and Martin J. Pickering. 2007. “The Representation of
Lexical and Syntactic Information in Bilinguals: Evidence from Syntactic Priming.” Journal
of Memory and Language 56: 153–171. DOI: 10.1016/j.jml.2006.10.002
Schoonbaert, Sofie, Phillip J. Holcomb, Jonathan Grainger, and Robert J. Hartsuiker. 2011.
“Testing Asymmetries in Noncognate Translation Priming: Evidence from RTs and ERPs.”
Psychophysiology 48: 74–81. DOI: 10.1111/j.1469-8986.2010.01048.x
Tirkkonen-Condit, Sonja, Jukka Mäkisalo, and Sini Immonen. 2008. “The Translation Process
— Interplay between Literal Rendering and a Search for Sense.” Across Languages and
Cultures 9: 1–15. DOI: 10.1556/Acr.9.2008.1.1
Tirkkonen-Condit, Sonja. 2005. “The Monitor Model Revisited: Evidence from Process
Research.” Meta 50: 405–414. DOI: 10.7202/010990ar
DOI: 10.1075/btl.4
Vandepitte, Sonia, and Robert J. Hartsuiker. 2011. “Metonymic Language Use as a Student
Translation Problem: Towards a Controlled Psycholinguistic Investigation.” In Methods and
Strategies of Process Research: Integrative Approaches in Translation Studies, ed. by Cecilia
Alvstad, Adelina Hild, and Elisabet Tiselius, 67–92. Amsterdam: John Benjamins.
DOI: 10.1075/btl.94.08van
ELF speakers’ restricted power of expression
Implications for interpreters’ processing
Michaela Albl-Mikasa
At the crossroads of English as a lingua franca (ELF) research and interpreting

studies, it is paramount to examine why interpreters are at odds with ELF com-
munication in general and with the effects of the output of the growing number
of non-native English conference speakers on their work in particular. On the
basis of a small-scale case study, the stumbling blocks resulting from non-native
English input are examined. The findings point toward what may be a major
impediment: activation and retrieval constraints can result from ELF speakers’
restricted power of expression and have an adverse effect on the interpreter’s
inferential processing and target text rendering. In the discussion, it is argued
that a key problem for devising compensatory (strategic or didactic) measures
may lie in the unpredictable and open-ended nature of the means of expression
creatively constructed by ELF speakers, which makes it extremely difficult for
interpreters to build up a stock of resources that will match the ongoing input
items, allowing them to function as activating cues.
Keywords: Conference interpreting, English as a lingua franca, inferential

processing, activating cues, routine transfer links, mental translation memory
1. Introduction
The study of English as a lingua franca (ELF) is an even younger discipline than
that of interpreting studies. Although both belong under the wider umbrella of ap-
plied linguistics, it is only very recently that the two fields have begun to consider
the mutual relevance of each other (see Cook 2012). With studies so far having
concentrated on the effect of non-native accents on the interpreting task (espe-
cially McAllister 2000; Sabatini 2000; Cheung 2003; Kurz 2008) and the advantage
of having the non-native speaker’s first language as one of the interpreter’s work-
ing languages (see Taylor 1989; Basel 2002; Kurz and Basel 2009), consideration of
doi 10.1075/bct.77.04alb
44 Michaela Albl-Mikasa
the wider impact of ELF on interpreting has only just started to take off (see Albl-
Mikasa 2010; 2012a; 2013b; Reithofer 2010; 2013). It has become clear from this
research that interpreters are disconcerted by the unprecedented spread of English
as a global lingua franca. However, it is not simply a potential existential threat to
the profession — ELF developments threaten the very multilingualism on which
it thrives, (see Cook 2012: 256) — that gives rise to major concerns on their part.
In fact, changing market conditions — such as more conferences in English only,
more ‘bilingual ELF conferences’ (namely, interpreting only between ELF and one
local language instead of multilingual conferences with a greater number of dif-
ferent language booths), less appreciation of interpreter services by customers, a
greater pressure to deliver higher quality to demonstrate the added value of those
services, and so forth (see Albl-Mikasa 2010) — are not the only disturbing factors.
It is also, and quite fundamentally, on the micro-level of working conditions or ac-
tual processing that interpreters face new challenges. In my questionnaire survey
among 32 professional interpreters on the implications of ELF on the interpret-
ing profession (Albl-Mikasa 2010), the respondents complained that their work
was increasingly strenuous and tiring due to additional cognitive load; faced with
ELF speakers, they had to expend extra capacity in source text comprehension (to
grasp foreign accents and recover unfamiliar expressions, to resolve unorthodox
syntactic structures and compensate for the lack of pragmatic fluency on the part
of the non-native speaker); faced with non-native English listeners, they had to do
so in the target text production (for accommodation to differing proficiency levels
of non-native addressees). As a result, in a number of qualitative statements to
open questions, the interpreters expressed that job satisfaction was on the decline
(2010: 141). Furthermore, the restricted power of expression on the part of ELF
speakers was identified as a major trouble spot (see Albl-Mikasa 2013a).
It is arguable that the perceived burden is simply a reaction to new circum-
stances, which deviate from well-trodden paths, internalized ways, and proven
experience, and that such perceptions are exacerbated by the (as yet) unavailability
of special training measures to help would-be interpreters build up the required
compensatory strategies geared to ELF conditions. It needs to be noted, however,
that the complaints voiced in the questionnaire survey (see Albl-Mikasa 2010)
and in the in-depth interview corpus of 10 professional conference interpreters
(see Albl-Mikasa 2012b) were made by very experienced professionals who have
been in the profession for decades rather than years; this means that they have had
the chance to become accustomed to a constantly growing number of non-native
English speakers and audiences in conference settings by gradually working their
way through the emerging challenges. The complaints about taxed resources, ex-
haustion, and declining job satisfaction due to the ‘ELF factor’ ran across the whole
spectrum of interpreters with a working experience of 10 to 40 years. Whichever
ELF speakers’ restricted power of expression 45
way one may look at the situation, it is a difficult one for interpreters and begs the
question as to what exactly it is that makes their work in settings with ELF speak-
ers and listeners especially challenging. The focus of this chapter is on exploring
the nature of processing difficulties arising from ELF.
2. Theoretical background — a cognitive discourse model of interpreting
Today’s understanding of the process dimension of the interpreter’s work with

(source and target) texts takes as its theoretical background the paradigmatic shift
from text linguistics to discourse analysis. This, in turn, proceeds from the seminal
work of Gillian Brown and George Yule (1983). By making reference to Henry
Widdowson’s highly relevant (1979) distinction of text (the product) and discourse
(a pragmatic process of meaning negotiation), Brown and Yule took the step from
looking at text as an object (“text-as-product view”) to its discourse dimension
(“discourse-as-process view”), that is, “to describe linguistic form, not as a static
object, but as a dynamic means of expressing intended meaning” (1983: 24). It is
against this paradigmatic background that a cognitive discourse model of inter-
preting was developed by what Franz Pöchhacker calls the “Heidelberger Schule”
(2012: 19) with its psycholinguistic orientation and tradition. Different from the
intuitive théorie du sens of the Paris School or the neurophysiological approach of
the Trieste School, the Heidelberg School approach brings into focus (interpret-
ers’) text understanding as mental model building. This is described in some detail
by Sylvia Kalina (2009: 159–163) as well as by Sabine Braun and Kurt Kohn (2012):
According to a cognitive and pragmatic discourse model of interpreting (Kohn
/ Kalina 1996; Kalina 1998; Kohn / Albl-Mikasa 2002; Braun 2004; Albl-Mikasa
2007), the interpreter’s understanding of a source text can be depicted as the cre-
ation of a ‘mental model’ of what the text is about (Brown / Yule 1983; Johnson-
Laird 1983); or as the pragmatic engagement in a meaning negotiating ‘discourse’
event, of which the text is a mere product (Widdowson 2004: 8). The cognitive-
pragmatic unfolding of discourse from text involves strategic processes of con-
tinuous and cyclic utterance meaning formation (including monitoring and
revision) based on linguistic as well as world-related knowledge.
(Braun and Kohn 2012: 190)
It is within this theoretical and methodological framework that I wish to explore

the new challenges interpreters have to face in light of the global spread of ELF,
that is, the growing number of non-native English speakers at conferences. This
perspective has provided the justification for why communication is not and need
not be based on complete, well-formed, and well-pronounced speech output and
why, therefore, (unmediated) non-native communication can work. Proceeding

from the non-native speaker example “Nowadays, in our country, Sweden, there is
so well of all sort of eating that man light come to big overweight. What to doing?”
(1983: 235), Brown and Yule go on to explain:
It is the predictive power of top-down processing that enables the human reader to
encounter, via his bottom-up processing, ungrammatical or mis-spelt elements in
the text and to determine what was the most likely intended message.(1983: 235)
More precisely, it is the intricate bottom-up/top-down interaction of source speech

input, general background, contextual, domain-specific, discourse-related, and
other knowledge, and the extensive drawing of inferences that also explain why
translation works despite adverse processing conditions (see Kohn 1990). This is,
finally, the basis that underlies Daniel Gile’s (2009) Efforts Model and Kalina’s
(1998) strategic approach, which illustrate why interpreting is possible, even un-
der extreme time constraints, and how resources are variably allocated in source
text comprehension or target text production.
One might, at first glance, concur that the ‘ELF factor’ represented by the
growing number of non-native English speaking conference participants with
varying degrees of proficiency may be no more than one variant of the numerous
and constantly changing input conditions with which interpreters have to cope,
and for which they have to find compensatory measures. These input conditions
include more and less strong accents, also in the case of native speakers; more and
less complex, technical and dense source texts; more and less able speakers/source
text producers; or more and less adequate working conditions. However, while
interpreters like to go on about “poor speakers,” “fast presentation rates,” “below-
standard technical installations” or “read-out rather than spontaneously delivered
talks,” their ELF-related complaints (as voiced in the questionnaire survey and
interviews) are of a more fundamental and poignant nature (see also Reithofer
2010). What then is the foundation of these complaints? This is, of course, a crucial
question in providing a basis for future ELF-oriented interpreter training.
3. Changing contexts — the ELF dimension of bilingual processing
To understand why interpreters keep complaining about ELF and the exact par-
ticularities typical of non-native English speakers that impinge upon the interpret-
ers’ processing, and thus to understand the consequences for the transfer process
and the resulting difficulties and possible new strategic requirements, one has to
bear in mind that interpreters work under conditions that are different from those
of monolingual communication settings in which ELF is usually used (and which
have been the basis of the specialist ELF literature so far). As Braun and Kohn
point out, under bilingual conditions people’s “routinely practiced comprehen-
sion and production skills are on unfamiliar territory” (2012: 190). The particular
and aggravating processing conditions in mediated multilingual communication
(as opposed to unmediated monolingual communication) are described in some
detail by various authors:
1. Kohn (1990: 109–112; 2004: 220–223) outlines the three sources of conflict
inherent to translation: (a) the continuing presence of the source text, (b) the
lack of semantic autonomy on the part of the translator, and (c) the need for
deeper reception and analysis as a basis for choosing adequate expressions in
target text production.
2. According to Anna Kucharska (2009: 14–18), interpreting typically takes place
in “deficient situations” due to a great number of mode-, time-, interpreting-
and situation-specific restrictions (see also Braun 2004).
3. Furthermore, what makes interpreting an especially complex task is the cogni-
tive need of interpreters to focus their attention — contrary to monolingual
processing (see Brown and Yule’s Sweden example in Section 2 above) — on
all levels of mental representation and processing at the same time. Focus on
the level of linguistic surface structures is necessary for the re-textualization of
linguistic expressions and equivalence effects in the target language; focus on
the level of the propositional text base for complete rendering and better mem-
orization of details; and focus on the level of the mental model for a grasp of the
overall meaning or overarching coherence and control of the communicative
situation (see Albl-Mikasa 2011: 56–57).
These particular bilingual conditions of mediated communication are the backdrop
against which we need to look at ELF as an additional aggravating factor. It is in this
context that it becomes clear why perceptions of the efficiency of ELF communica-
tion differ. While ELF researchers have collected ample evidence emphasizing the
effectiveness of ELF communication (see Seidlhofer 2011), findings in interpreting
studies are much less positive. On one hand, the factors facilitating and support-
ing monolingual ELF communication in largely interactional settings do not apply
to mediated communication when the presentation of monologic speeches leaves
no room for meaning negotiation and co-construction and when the requirement
of their full and detailed rendering does not allow for let-it-pass strategies (see
Reithofer 2010: 149). On the other hand, as Karin Reithofer (2013) finds in a larger-
scale study, understanding of source speech propositional content in conference
settings can be significantly higher among conference participants listening to the
interpreter than those listening to the non-native English original. As far as the
interpreters are concerned, respondents to my 2010 questionnaire survey seemed
to agree that non-native English speakers can be easier to interpret for than native
speakers, especially when they speak slowly and use simple structures. At the same
time, a clear majority (69%) of them favored native-speaker source text producers
who were felt to be not necessarily easier but better or more pleasant to interpret
for, as their speech flow, speech structure, and precision in expression facilitated
the interpreters’ task. In contrast, ELF speaker-induced pronunciation, structure,
and expression-based comprehension problems were perceived to impede the un-
derlying processes of anticipation and inferencing (see Albl-Mikasa 2010: 136).
To gain further insight into what might be behind interpreter complaints and
perceptions regarding input by ELF speakers, I should like to draw on a small-
scale case study. The study is too restricted in scope to be representative or to allow
for broader generalization of the results. Moreover, it involves a student interpreter
rather than a professional. The reason is that this study, carried out for a B.A. thesis
under my supervision (see Dschulnigg and Hansen 2011), was conceived as a kind
of pilot study for a larger study involving professionals.1 As such, it studied a stu-
dent in order to amplify a problem that the professional interpreters interviewed
had previously identified as restrictions in the power of expression on the part of
ELF speakers (see Albl-Mikasa 2013a). The idea was to investigate the basic dif-
ficulties posed by these restrictions and to identify different types of restriction
before testing their validity with professional interpreters.
4. Evidence from a small-scale case study — ELF-related stumbling blocks

for interpreter trainees
In the small-scale B.A. study, three German native-speaker students in the B.A. in
Translation program who did not have English as one of their study languages but
had taken an ELP (English for Language Professionals) course were asked to give
a five-minute speech in English on one of the following randomly assigned topics:
the nuclear disaster in Japan, the Libyan crisis, and WikiLeaks. They were given a
sheet with topic-related content notes in German and had 20 minutes to prepare
and look up some of the necessary terminology on the Internet or in a diction-
ary. All three speeches were simultaneously translated into German by the same
1. I did not ask Nadine Dschulnigg and Leslie Hansen to co-author this chapter (which I nor-
mally do in the case of publication of M.A. thesis results), because the setup, study design, and
guiding ideas behind the thesis resulted from a joint discussion, in which my input was salient,
and because the classification of interpreter stumbling blocks I propose in this chapter does not
figure in the thesis. Full credit goes, of course, to the two students for collecting and discussing
highly interesting data and for rendering a clearly relevant B.A. account.
interpreter: a student of conference interpreting in the second semester of a three-

semester M.A. course with German as her strongest (A) and English as her second
strongest (B) language. In retrospective interviews held immediately following the
interpretation, the speakers were asked whether they had been capable of express-
ing what they wanted to convey and (on the basis of the transcribed source and
target texts a week after the interpretation) whether they felt that their speech had
been adequately rendered by the interpreter. The interpreter, in turn, was shown
those parts of the source and target texts where the speakers felt they had been
misrepresented and was asked to explain possible reasons for the discrepancy in
speaker intention versus interpreted rendering. This combination of product-
based analysis with retrospective interviews was chosen in order to look beyond
surface evidence and to gain insights that could not be obtained by either product-
or process-based methods alone. The idea was to see whether ‘typical interpreter
mistakes,’ such as omissions in the target text rendering, may have resulted from
problems in ELF speaker source speech production (which had been identified as
a major source of problems by professional interpreters in Albl-Mikasa 2013a).
As it turned out, in 22% of source speech “propositions” (which corresponded, in
actual fact, to full sentences), the speakers felt that renderings were not adequate
(Dschulnigg and Hansen 2011: 18). According to Dschulnigg and Hansen, in the
majority of those cases (63%), inadequate rendering was attributable to the speak-
ers’ (self-confessed) difficulties in expressing their intended meaning while 33%
were related to interpreter-specific problems.
Upon further analysis of the data, I identified three main areas of difficulty in
self-expression on the part of the speakers which presented stumbling blocks or
sources of difficulty for the interpreters. They were categorized as shown in Table 1.
Table 1. ELF-related sources of interpreting difficulties
Table 1. ELF-related sources of interpreting difficulties

In the following, I will discuss one or two examples for each category.
4.1 Incoherent input--investment of extra capacity to clear up the input
In the example in table 2, the speaker points out in the retrospective interview that the interpreter
In the following, I will discuss one or two examples for each category.
4.1 Incoherent input — investment of extra capacity to clear up the input
In the example in Table 2, on the nuclear disaster in Japan, the speaker points out in
the retrospective interview that the interpreter did not render the idea of a benevo-
lent gesture on the part of the emperor, which she had wanted to express by saying
“he was really kind with (eh) with the folk there, cause he has like a building.” The
interpreter, by contrast, reports that she was rather irritated by the phrase “the
emperior imperior like the king [clearing throat] the king of Japan.” Upon hearing
“king” as the last piece of information, she started off with “der König” (king), only
to rephrase it as “Kaiser” (emperor) when she realized it was the emperor that was
being referred to. All this, however, distracted her to such an extent that she did not
hear the rest of the proposition and omitted it. It should be added that the omitted
part of the proposition was not well expressed either and would hardly have given
away the meaning of the benevolent gesture intended by the speaker.
In the second part of the example in Table 2, the speaker points out that
“Badezimmer” (bathroom) as used by the interpreter does not convey what she
meant. At the same time, she admits to having been uncertain about what kind of
building or Badeanlage (bathhouse) it actually was that she had found described in
a newspaper some time ago. This is not the only concept, however, that the speaker
found difficult to convey. As seen from her use of “I don’t know” and “like,” she
Table 2. Example of incoherent input and resulting interpretation (Dschulnigg and

Hansen 2011: 87)
Table 2. Example of incoherent input and resulting interpretation (Dschulnigg and Hansen
2011:87)
In the second part of the example in table 2, the speaker points out that “Badezimmer”
(bathroom) as used by the interpreter does not convey what she meant. At the same time, she
had general difficulty in expressing things. In fact, the interpreter emphasizes that
it was the very presentation of the content matter by the speaker that caused a
comprehension problem on her part, as she was unable to imagine what exactly
was meant and therefore summarized the source text, which may have led to the
omission of some details.
Looking at the example in Table 2, it becomes clear how difficult it must be for
the interpreter to figure out the exact meaning of an utterance on the basis of poor
source text input. It is almost impossible to infer the actual circumstances from the
very general term “bath” unless one has the corresponding background knowledge;
this is even more so if such a term is embedded in an overwhelmingly unclear pas-
sage (laughter, “take a half half bad,” etc.). In other words, the example highlights
the extent to which explicit input supports the interpreter’s task.
The example in Table 2 illustrates a case where the ELF speaker becomes con-
fused and is unable to complete her output production in an efficient way (evidence
that this is not a rare occurrence in ELF speech can be found in the TELF database).2
This can lead to serious comprehension and processing problems on the part of the
interpreter, who has to invest extra energy to clear up and reorganize the confused
and incoherent speaker output in order to reconstruct what the speaker meant or
would have said had she been able to express herself more adequately.
4.2 Imprecise input — need for additional resources and/or background

knowledge to pin down the intended meaning/message
In the example in Table 3, concerning the crisis in Libya, the speaker criticizes in
the retrospective interview the interpreter’s use of “Versammlung” (assembly) in
the target text, which to her mind does not relate to the governmental body consti-
tuted by the opposition she intended to refer to. However, had the speaker known
that in English-speaking media this body is called ‘National Assembly’ and had
she used the phrase accordingly, the interpreter could have more easily come up
with the German equivalent Nationalrat (National Council). “General Assembly,”
by contrast, conjures up Generalversammlung (general meeting) in people’s minds.
From the fact that the interpreter hesitated about which word to use (saying “ein
eine Versammlung”), it seems likely that she must have figured out that the speaker
could not have been referring to a Generalversammlung; as a result, her strategy
was to use the more general term Versammlung.
2. Evidence can be found in TELF, the Tübingen English as a Lingua Franca corpus and data-
base (http://www. telf.uni-tuebingen.de).
up with the German equivalent Nationalrat (national council). “General Assembly,” by contrast,
conjures up Generalversammlung (general meeting) in people’s minds. From the fact that the
interpreter hesitated about which word to use (saying “ein eine Versammlung”), it seems likely
52 that
Michaela
she mustAlbl-Mikasa
have figured out that the speaker could not have been referring to a
Generalversammlung; as a result, her strategy was to use the more general term “Versammlung.”
Table 3. Example of imprecise input and resulting interpretation (Dschulnigg and
Hansen 2011: 89)
Table 3. Example of imprecise input and resulting interpretation (Dschulnigg and Hansen
What is also striking in the example in2011:89)
Table 3, but goes unnoticed by the speaker,
is her use of “foreign minister of justice.” The person in question is actually the
‘former’ minister of justice, which suggests that the words ‘former’ and ‘foreign’
53
were confused by the ELF speaker. For unknown reasons, this part is omitted by
the interpreter. It is possible, however, that the omission was caused by irritation
about the contradictory collocation and the fact that too much energy would have
been needed to draw on background knowledge and decide whether the former
minister of justice or of foreign affairs was meant by the speaker. By contrast, the
step from ‘former minister of justice’ to früherer Justizminister (former justice min-
ister) would have been straightforward, not taking up much capacity. Thus, while
omissions are often seen as ‘classical interpreter mistakes,’ some of them may not
be attributable to the interpreter, but to the (ELF) speaker.
This example demonstrates that the restricted ability of the speaker to express
precisely what she intends to convey can lead to general comprehension problems
or to the interpreter being puzzled and needing to invest additional resources to
draw on background knowledge and make decisions that are normally pre-empt-
ed by the speaker upon output production.
4.3 Comprehensible but unconventional input — erosion of direct SL-TL

links3
In the example in Table 4, concerning the Japanese nuclear disaster, the speaker
expresses her feeling in the retrospective interview that the target text expression
“schwimmen” (swim) is too positive a term for what she wanted to convey, namely,
“dass die Häuser ‘weggespült’ wurden” (that houses were washed away). The inter-
preter, however, upon hearing “flow away” (which conjures up the picture of cars
floating around), would have had to invest some effort to pin down the intended
meaning and come up with wegspülen. Had the speaker provided the standard
term ‘to wash away,’ this might have — at least in the given context — narrowed
3. SL stands for source language; TL for target language.

down the meaning immediately and helped the interpreter retrieve the direct
equivalent wegspülen from her multilingual lexicon, without requiring any extra
inferential effort.
Table 4. Example of comprehensible but unconventional input and resulting interpreta-

tion (Dschulnigg and Hansen 2011: 85)
Table 4. Example of comprehensible but unconventional input and resulting interpretation

(Dschulnigg and Hansen 2011:85)

A similar
A similar point
point cancanbebemade
made by thethe
with
(Dschulnigg example in table
example
and Hansen in 5. The speaker
Table 5,
2011:85) points out
concerning thein Libyan
the
crisis. Theinterview
retrospective speaker that
points
she out in the
did not know retrospective interview
the exact expression that she didbut
for Sondergipfel not know
expected
thethe exact expression
interpreter withSondergipfel
to come up for it on the basisbut expected
of the context.theFrominterpreter to come
the interpreter’s up with
perspective,
it on Athe
similar
basis point
of cancontext.
the be made by the example
However, had inthe
table 5. The used
speaker speaker points out in the
‘extraordinary sum-
however, one could say that had the speaker used ‘extraordinary summit,’ the interpreter might
retrospective interview that
mit,’ the interpreter she did
might notmore
have know easily
the exact expression
retrieved thefor Sondergipfel
respective but expected
equivalent.
more easily have retrieved the respective equivalent.
the interpreter to come up with it on the basis of the context. From the interpreter’s perspective,
Table 5. Example of comprehensible but unconventional input and resulting interpreta-
however, one could say that had the speaker used ‘extraordinary summit,’ the interpreter might
tion (Dschulnigg and Hansen 2011: 89)
more easily have retrieved the respective equivalent.

The examples in Tables 4 and 5 illustrate
(Dschulnigg that there
and Hansen are cases where the interpreter
2011:89)
has no difficulty in understanding what is expressed. While the ELF performance
isTable
clear5.and easy of
Example to comprehensible
understand, itbut differs from the input
unconventional expected conventionalized
and resulting ex-
interpretation
The examples
pression that wouldin tables 4(Dschulnigg
trigger and 5 illustrate
established that
and directthere
Hansen are cases
As awhere
2011:89)
links. the interpreter
consequence, the has no
inter-
preter in
difficulty hasunderstanding
difficulty inwhat
retrieving the corresponding
is expressed. The ELF performance translation equivalents.
is clear and easy to The
same goes
understand, butfor irritation
differs from thecaused
expectedbyconventionalized
differences in expression
style of expression. These cases
that would trigger
The examples in tables 4 and 5 illustrate that there are cases where the interpreter has no
demonstrate the difference it makes for the interpreter whether she
established direct links. As a consequence, the interpreter has difficulty in retrieving simply has to
difficulty in understanding what is expressed. The ELF performance is clear
link up two related terms (translation equivalents) in her multilingual lexicon or and easy to
corresponding translation equivalents. The same would go for irritation caused by differences in
understand, but differs from the expected conventionalized expression that would trigger
established direct links. As a consequence, the interpreter
55 has difficulty in retrieving
corresponding translation equivalents. The same would go for irritation caused by differences in
55
has to infer exactly what the speaker meant when using an imprecise term in order
to then activate the corresponding expression.
5. Discussion — effects of ELF speakers’ appropriation of resources on

interpreters’ comprehension and transfer processes
The examples given in the previous section point to two major processing con-
straints arising from ELF speaker source text production. First, incoherent or im-
precise input may hamper the interpreter’s comprehension processes and make it
difficult for her to grasp exactly what the speaker is trying to get at, thus under-
mining the interpreter’s decision-making directed at finding adequate target lan-
guage solutions. Second, the use of conventional (lexical or structural) expressions
by the speaker might make it easier for the interpreter immediately to retrieve
corresponding translation equivalents, while ELF speaker output can impede the
interpreter’s falling back on established transfer links. Even though they are from a
small study, these findings may be relevant when considered against the backdrop
of the cognitive discourse processing framework given in Section 2 above as well
as results from ELF research.
The two constraints are interesting in that they are closely related to the two
basic strategies of interpreting, namely, meaning-based and word-based interpret-
ing. The former is assumed to take place on the basis of the (mental) representa-
tion of a (more or less) fully comprehended source text meaning, whereas the
latter involves “the literal transposition of words or multiword units” (Christoffels
and de Groot 2005: 459). During their task, interpreters “inevitably shift from
one pattern to the other according to internal and external circumstances” (Gran
1989: 98). This is explained by the Heidelberg School cognitive discourse model
of interpreting (see Section 2 above) as follows: processing takes place in terms of
more global or more local coherence building, in other words, it is geared toward
higher (the mental model) or lower (the surface structures) levels of mental repre-
sentation (see Albl-Mikasa 2011). The first constraint refers to (global coherence-
oriented) meaning-based processing, which is dealt with in detail in the specialist
literature on interpreting where researchers describe how comprehension and
analysis of the source text are fundamental for target speech production (see Albl-
Mikasa 2007: 11–132) and serve as the basis for translational decisions (see the
third source of translational conflict specified by Kohn in Section 3 above). From
Gile’s Effort Models (2009), it can easily be inferred that successful interpreting is
undermined when additional resources have to be expended to grasp unfamiliar
source language accents and structures produced by ELF speakers. I should, there-
fore, like to concentrate on the second constraint in the following discussion.
The second constraint is situated at the level of surface structures because it fo-
cusses on (matching) linguistic items. Interpreters, not being experts in each and
every field, will, at times, have to forego deeper understanding (meaning-based in-
terpreting) and process at the level of the lexical and terminological input (word-
based interpreting). What is often called “transcoding” (see Kalina 1998: 118) is
not an emergency measure, but a full-fledged strategy in circumstances that do
not allow for meaning-based interpreting. Psycholinguistic research has produced
evidence that transcoding occurs and that it “implies the existence of direct links
between representations of the linguistic elements of one language and the cor-
responding representations in the other language” (Christoffels and de Groot
2005: 471). This is in line with reports by an experienced interpreter (I-1) saying
that, in many cases, interpreters choose to rely on activation of their “autopilot”
(see Albl-Mikasa 2012b: 75), that is, on surface-level transcoding and on transfer
trails carved out by high-frequency use rather than in-depth analysis. A similar
remark was made by another interpreter in the same interview corpus: “I do a lot
of patterns, technical collocate patterns. I don’t even try to understand them; I still
believe that I can do a good job” (I-2). Given time constraints and capacity limi-
tations, important strategies to relieve pressure include activation of translation
routines (see Göpferich 2008: 156) as well as the routine assembly of translation
equivalents, or “reflex-like equivalence relations” (see Kutz 2002: 190; my transla-
tion and emphasis). Such direct and easy links and paths between source text items
and target text solutions are either established through rehearsal and practice or
ad hoc. An example of how such routines emerge is provided in my interview
corpus (see Albl-Mikasa 2012b: 79). One of the interpreters (I-6) reported render-
ing the source language term “first pass rate” by “Erstläuferrate” even though the
English term rate is typically translated as Quote in German. The reasons given
for this decision were that this is the term the German experts in the audience
were familiar with and that it made interpreting easier because it took less effort
to cross over from rate to Rate rather than from rate to Quote. It greatly facilitates
the interpreter’s task to store such expressions, phrases, formulaic sequences, and
their corresponding target language expressions in a routine way, and to build up
a ‘mental translation memory’ to fall back on during interpreting. Based on the
following explanations, I hypothesize that this represents a major impediment for
interpreters when it comes to ELF speaker source text production.
Retrieval, in general, and from the ‘mental translation memory,’ in particular,
depends on conducive, activating input items, as explained by recourse to the “prin-
ciple of encoding specificity” (formulated by van Dijk and Kintsch 1983, who view
memory as consisting of the traces that result as a function of depth of processing):
In order to retrieve an item from memory, the retrieval cue must match, at least
partially, the encoded item. The cue must be appropriate for the encoding. The
efficiency of memory is determined not by either encoding or retrieval alone, but
by the encoding-retrieval interaction. This is the principle of encoding specificity
[…]. (van Dijk and Kintsch 1983: 334)
According to this principle, ease of retrieval, transfer processing, and source text
rendering are to some degree determined by the extent to which an input cue (as
provided by the source speech producer) matches the item encoded and stored by
the interpreter (see Albl-Mikasa 2007: 82–84). If ELF speaker output is not made
up of the idiomatic expressions the interpreter has learned and internalized, there
will be a mismatch between the activating (ELF speaker) input item and the item
encoded (by the interpreter).
Such a mismatch is likely to occur, if one considers the differences between na-
tive and non-native English speakers identified in the literature on ELF. Of native
speakers, it can typically be said that “a language user has available to him or her a
large number of semi-preconstructed phrases that constitute single choices, even
though they might appear to be analyzable into segments” (Sinclair 1991: 110). As
Barbara Seidlhofer (2009) points out, such prefabricated chunks serve to relieve
memory, ease processing, support memorization, and make for efficient retrieval.
If all interlocutors have access to conventionally prefabricated phrases, they do not
have to make sense of each linguistic component bit by bit. In the case of prover-
bial sayings, the very beginning of such formulaic patterns (e.g., When in Rome…)
will suffice for the listener to complete the phrase (… do as the Romans do). The
idiom principle, therefore, relates to the least effort principle, since idiomatic us-
age of language “reduces the language user’s online processing load and so facili-
tates communication” (Seidlhofer 2009: 197) in that “[o]ne linguistic item primes
expectation for the occurrence of others, with different degrees of predictability”
(2011: 130). Contrary to native or highly proficient speakers of a language (such as
interpreters), however:
ELF users will tend to construct what they have to say more analytically, in a
bottom-up fashion, drawing on what is semantically encoded in the grammar
and lexis of the language — in other words, by recourse to what Sinclair calls the
‘open-choice’ principle. (Seidlhofer 2009: 202)
According to the ‘open-choice’ principle, ELF speakers, with little knowledge of

the recurrent idiomatic features and shared conventions, use unconventional
wordings to meet their communicative needs and so present the interpreter with
an open and ad hoc devised inventory of idiosyncratic phrases and constructions.
Interpreters (with English as a B language) who invest a great deal of effort in ac-
quiring the Standard English repertoire of idioms, collocations, and conventional
phrases will find that many ELF speakers’ unidiomatic expressions do not match
the choices in their repertoire and, therefore, do not help to activate encoded
items, including the corresponding translation equivalents. This undermines re-
trieval and may put the interpreter off track. Such derailment may not only apply
to trainees. For, while one could argue that, in the example in Table 5, a profes-
sional would have had enough background knowledge to immediately come up
with Sondergipfel (extraordinary summit) upon hearing “special meeting,” the
same professional may not manage to do so in the less obvious cases that often
arise in interpreting situations.
Obviously, such retrieval constraints can have an adverse effect on funda-
mental strategic interpreting processes, such as anticipation and segmentation
(see Chernov’s 2004 probability-prediction model of inference and anticipation).
Conscious knowledge of preconstructed, conventional phrases as part of the in-
terpreters’ competence not only supports fast retrieval of established direct links,
but, in addition, facilitates anticipatory processes (see the example above, When
in Rome… do as the Romans do). Similarly, prefabricated chunks point the way
toward meaningful segmentation supportive of processing. Whereas the boundar-
ies of indecomposable holistic items are easily recognized (e.g., take a gap year), it
takes more time to reconstruct decomposable semantic chunks (e.g., work abroad
as a volunteer), which rely on sense creation processes. As a consequence, the ELF
speakers’ creative appropriation of resources to serve their communicative pur-
poses may be a major asset for unmediated interactional ELF communication, as
pointed out by ELF researchers (see Seidlhofer 2011: 97–111), but may become an
impediment to the interpreters’ processing.
6. Conclusion
Complementing the traditional emphasis of ELF research on the strong points

of ELF communication (e.g., Seidlhofer 2011), this chapter takes a look at inter-
preters’ complaints about ELF. They arise, to a great extent, from the particular
bilingual mediated processing conditions under which the interpreter’s task is
carried out. In an attempt to encourage research into exactly what causes the dif-
ficulties reported by professional interpreters, I discussed the data of a small-scale
(B.A. thesis) case study against the backdrop of the Heidelberg School’s cognitive
discourse model of interpreting and findings from research into ELF. The (non-
representative) study data point out that stumbling blocks for interpreters may
result from incoherent, imprecise, or unconventional ELF speaker output, which
prevents the interpreter from making sufficiently reliable inferences about the in-
tended meaning or from activating relevant translation equivalents. Upon closer
analysis, it is possible to identify two major ELF-induced problem areas. One is

the possible adverse effect on interpreter-specific comprehension, since in order
for the interpreter to make appropiate translational decisions, input analysis and
source text comprehension need to be at a level of processing that is deeper than
that required in non-mediated ELF communication. The second is the possible
adverse effect on activation of direct SL-TL links, since in order for the interpreter
to have access to her ‘mental translation memory’, source speech input must act
as an activating cue (i.e., matching the stored and encoded expressions and struc-
tures). These two problem areas may help explain interpreter complaints about the
extra effort they have to invest when interpreting non-native English speeches. It is
worth noting in this context that the features interpreters mentioned as preferable
in native-speaker speech correspond with predictable input that matches with pre-
viously encoded items; they include: more predictable structures; fewer mistakes
and false friends; more conscious, deliberate, and purposeful use of concepts and
terms; higher accuracy in expression; and more logical and differentiated argu-
mentation (see Albl-Mikasa 2010: 135).
Some of the difficulties may be caused by the novelty effect, as discussed
above (see Section 1), and may call for different and special ELF-related strategies
with consequences for the modeling of interpreter competence (see Albl-Mikasa
2012b) and the design of interpreter training (see Albl-Mikasa 2013b). However,
this may still leave a major problem unsolved. If, against the backdrop of the en-
coding specificity principle (i.e., that retrieval rests on matching input items, as
dicussed in Section 5 above), a major handicap to interpreters’ processing were
found to be obstruction of activation of direct links between SL and TL, then this
would make it extremely difficult for interpreters to prepare themselves and take
preventive measures under the particular conditions of ELF speech in mediated
bilingual settings. While it may be true that “ELF is not a chaos of idiosyncra-
sies which arise at random,” as Anna Mauranen (2009: 232) finds from a corpus
perspective, interpreters will still be challenged by ELF “phraseology […] at the
interface of linguistic convention and creativity” (Mauranen 2009: 231). ELF us-
ers’ output is (cross-linguistically) influenced by a myriad of first languages and
marked by unexpected and unpredictable formations of a more idiosyncratic and
more unconventional nature than the output of native speakers. This leaves little
space for automated routine processes or the establishment of direct links, which
greatly facilitate the interpreting task. Moreover, it may undermine established
automatisms and stretch capacity limitations as higher order (and thus more ca-
pacity-consuming) inferences become necessary. Against this backdrop, the ELF
speaker’s creative appropriation of unconventional expressions may pose a greater
challenge than the English native speaker’s predictable idiomatic repertoire.
The above analysis of ELF-induced interpreter difficulties needs to be followed

up by much more comprehensive and rigorous empirical research. The hypoth-
esis that the ELF speaker’s idiosyncratic or unconventional linguistic creations
undermine the establishment or activation of transfer links and routines needs to
be tested. Furthermore, ELF studies aimed at filtering out the commonalities or
prototypical features of ELF, which are undertaken on the basis of ELF corpora
such as ELFA, TELF, or VOICE (see Mauranen 2012; Kohn 2011; Seidlhofer 2011,
respectively), should be more directly focused on interpreter needs in order to be of
benefit for them.4 In addition, more fine-grained analysis of ELF-related stumbling
blocks in source speeches should be conducted, possibly on the basis of the source
text surface triggers that Braun and Kohn (2012) identify as indicative of the chal-
lenges interpreters face in (a) source text understanding, (b) source text retention,
and (c) target text production. Such an analysis also has the potential of isolating
ELF-induced communication difficulties in unmediated settings. Cross-fertilization
of ELF research and interpreting studies is undoubtedly a promising prospect.
References
Albl-Mikasa, Michaela. 2007. Notationssprache und Notizentext. Ein kognitiv-linguistisches

Modell für das Konsekutivdolmetschen. Tübingen: Gunter Narr.
Albl-Mikasa, Michaela. 2010. “Global English and English as a Lingua Franca (ELF):
Implications for the Interpreting Profession.” trans-kom 3 (2): 126–148. http://www.trans-
kom.eu/ihv_03_02_2010.html.
Albl-Mikasa, Michaela. 2011. “Das integrative Potenzial des kognitiv-konstruktivistischen
Forschungsparadigmas für die Translationswissenschaft.” In Translation — Sprachvariation
— Mehrsprachigkeit. Festschrift für Lew Zybatow, ed. by Wolfgang Pöckl, Ingeborg
Ohnheiser, and Peter Sandrini, 41–67. Frankfurt: Peter Lang.
Albl-Mikasa, Michaela. 2012a. “Interpreting Quality in Times of English as a Lingua Franca
(ELF): New Variables and Requirements.” In Translation Studies: Old and New Types
of Translation in Theory and Practice. Proceedings of the 1st International Conference
TRANSLATA. Translation & Interpreting Research: Yesterday? Today? Tomorrow?, May
12–14, 2011, Innsbruck, ed. by Lew N. Zybatow, Alena Petrova, and Michael Ustaszewski,
267–273. Frankfurt: Peter Lang.
Albl-Mikasa, Michaela. 2012b. “The Importance of Being not too Earnest: A Process- and
Experience-based model of Interpreter Competence.” In Dolmetschqualität in Praxis, Lehre
und Forschung. Festschrift für Sylvia Kalina, ed. by Barbara Ahrens, Michaela Albl-Mikasa,
and Claudia Sasse, 59–92. Tübingen: Narr Francke Attempto.
4. VOICE (Vienna Oxford International Corpus of English): <http://univie.ac.at/voice/>. TELF

(Tübingen English as a Lingua Franca): <http://www.telf.uni-tuebingen.de/>. ELFA (English as
a Lingua Franca in Academic Settings): http://www.helsinki.fi/englanti/elfa/.
Albl-Mikasa, Michaela. 2013a. “Express-ability in ELF Communication.” Journal of English as a

Lingua Franca 2 (1): 101–122. DOI: 10.1515/jelf-2013-0005
Albl-Mikasa, Michaela. 2013b. “Teaching Globish? The Need for an ELF Pedagogy in Interpreter
Training.” International Journal of Interpreter Education 5 (1): 3–16.
Basel, Elvira. 2002. English as Lingua Franca. Non-Native Elocution in International
Communication. A Case Study of Information Transfer in Simultaneous Interpretation.
Ph.D. dissertation, University of Vienna.
Braun, Sabine. 2004. Kommunikation unter widrigen Umständen? Einsprachige und gedol-
metschte Kommunikation in der Videokonferenz. Tübingen: Gunter Narr.
Braun, Sabine, and Kurt Kohn. 2012. “Towards a Pedagogic Corpus Approach to Business and
Community Interpreter Training.” In Dolmetschqualität in Praxis, Lehre und Forschung.
Festschrift für Sylvia Kalina, ed. by Barbara Ahrens, Michaela Albl-Mikasa, and Claudia
Sasse, 185–204. Tübingen: Narr Francke Attempto.
Brown, Gillian, and George Yule. 1983. Discourse Analysis. Cambridge: Cambridge University
Press. DOI: 10.1017/CBO9780511805226
Chernov, Ghelly. V. 2004. Inference and Anticipation in Simultaneous Interpreting: A Probability-
prediction Model. Amsterdam: John Benjamins. DOI: 10.1075/btl.57
Cheung, Andrew. 2003. “Does Accent Matter? The Impact of Accent in Simultaneous
Interpretation into Mandarin and Cantonese on Perceived Performance Quality and
Listener Satisfaction Level.” In Evaluación de la calidad en interpretación de conferencias:
investigación, ed. by Ángela Collados Aís, Manuela M. Fernández Sánchez, and Daniel Gile,
85–96. Granada: Editorial Comares.
Christoffels, Ingrid K., and Annette M.B. de Groot. 2005. “Simultaneous Interpreting: A
Cognitive Perspective.” In Handbook of Bilingualism: Psycholinguistic Approaches, ed. by
Judith F. Kroll, and Annette M.B. de Groot, 454–479. New York, NY: Oxford University
Press.
Cook, Guy. 2012. “ELF and Translation and Interpreting: Common Ground, Common Interest,
Common Cause.” Journal of English as a Lingua Franca 1 (2): 241–262.
DOI: 10.1515/jelf-2012-0018
Dschulnigg, Nadine, and Leslie Hansen. 2011. Englisch als Lingua Franca (ELF) in
Dolmetschsituationen. Eine empirische Untersuchung zur Ausdrucksfähigkeit von
ELF-Sprechenden in einer Dolmetschsituation. Unpublished B.A. thesis, ZHAW Zurich
University of Applied Sciences.
Gile, Daniel. 2009. Basic Concepts and Models for Interpreter and Translator Training. Amsterdam:
John Benjamins. DOI: 10.1075/btl.8
Göpferich, Susanne. 2008. Translationsprozessforschung: Stand — Methoden — Perspektiven.
Tübingen: Narr Francke Attempto.
Gran, Laura. 1989. “Interdisciplinary Research on Cerebral Asymmetries: Significance and
Prospects for the Teaching of Interpretation.” In The Theoretical and Practical Aspects of
Teaching Interpretation. Proceedings of the Trieste Symposium, 1986, ed. by Laura Gran,
and John Dodds, 93–100. Udine: Campanotto.
Johnson-Laird, Philip. 1983. Mental Models. Cambridge, MA: Harvard University Press.
Kalina, Sylvia. 1998. Strategische Prozesse beim Dolmetschen. Theoretische Grundlagen, em-
pirische Fallstudien, didaktische Konsequenzen. Tübingen: Gunter Narr.
Kalina, Sylvia. 2009. “Das Modell der gemittelten Kommunikation in interkulturell geprägten
Settings.” In Dimensionen der Zweitsprachenforschung/Dimensions of Second Language
Research. Festschrift für Kurt Kohn, ed. by Michaela Albl-Mikasa, Sabine Braun, and Sylvia
Kalina, 159–181. Tübingen: Narr Francke Attempto.
Kohn, Kurt. 1990. “Translation as Conflict.” In Confli(c)t. Proceedings of the International
Symposium‚ Contact + Confli(c)t, Brussels, 2–4 June 1988, ed. by Peter H. Nelde, 105–113.
Brussels: Association Belge de Linguistique Appliquée (ABLA papers 14).
Kohn, Kurt. 2004. “Zur Rolle des Übersetzens beim Sprachenlernen.” In Neue Perspektiven in
der Übersetzungs- und Dolmetschwissenschaft. Festschrift für Heidrun Gerzymisch-Arbogast
zum 60 Geburtstag, ed. by Juliane House, Werner Koller, and Klaus Schubert, 217–233.
Bochum: AKS-Verlag.
Kohn, Kurt. 2011. “English as a Lingua Franca and the Standard English Misunderstanding.”
In English in Europe Today. Sociocultural and Educational Perspectives, ed. by Annick De
Houwer, and Antje Wilton, 71–94. Amsterdam: John Benjamins. DOI: 10.1075/aals.8.06koh
Kohn, Kurt, and Michaela Albl-Mikasa. 2002. “Note-taking in Consecutive Interpreting.
On the Reconstruction of an Individualised Language.” In Linguistics and Translation
Studies. Translation Studies and Linguistics. Linguistica Antverpiensia1, ed. by Leona Van
Vaerenbergh, 257–272. Antwerpen: Hoger Instituut voor Vertalers & Tolken.
Kohn, Kurt, and Sylvia Kalina. 1996. “The Strategic Dimension of Interpreting.” Meta 41 (1):
118–138. DOI: 10.7202/003333ar
Kucharska, Anna. 2009. Simultandolmetschen in defizitären Situationen. Strategien der translato-
rischen Optimierung. Berlin: Frank & Timme.
Kurz, Ingrid. 2009. “The Impact of Non-Native English on Students’ Interpreting Performance.”
In Efforts and Models in Interpreting and Translation Research, ed. by Gyde Hansen, Andrew
Chesterman, and Heidrun Gerzymisch-Arbogast, 179–192. Amsterdam: John Benjamins.
DOI: 10.1075/btl.80.15kur
Kurz, Ingrid, and Elvira Basel. 2009. “The Impact of Non-native English on Information Transfer
in SI.” Forum 7 (2): 187–212.
Kutz, Wladimir. 2002. “Dolmetschkompetenz und ihre Vermittlung.” In Übersetzen und
Dolmetschen. Eine Orientierungshilfe, ed. by Johanna Best, and Sylvia Kalina, 184–195.
Tübingen: Francke.
Mauranen, Anna. 2009. “Chunking in ELF: Expressions for Managing Interaction.” Intercultural
Pragmatics 6 (2): 217–233. DOI: 10.1515/IPRG.2009.012
Mauranen, Anna. 2012. Exploring ELF. Academic English Shaped by Non-native Speakers.
Cambridge: Cambridge University Press.
McAllister, Robert. 2000. “Perceptual Foreign Accent and its Relevance for Simultaneous
Interpreting.” In Language Processing and Simultaneous Interpreting, ed. by Kenneth
Hyltenstam, and Birgitta Englund Dimitrova, 45–63. Amsterdam: John Benjamins.
DOI: 10.1075/btl.40.05mca
Pöchhacker, Franz. 2012. “Qualität, die man versteht: ein funktional-kognitiver Ansatz.” In
Dolmetschqualität in Praxis, Lehre und Forschung. Festschrift für Sylvia Kalina, ed. by
Barbara Ahrens, Michaela Albl-Mikasa, and Claudia Sasse, 19–31. Tübingen: Narr Francke
Attempto.
Reithofer, Karin. 2010. “English as a Lingua Franca vs. Interpreting — Battleground or Peaceful
Co-existence.” The Interpreters’ Newsletter 15: 143–157.
Reithofer, Karin. 2013. “Comparing Modes of Communication. The Effect of English as a Lingua
Franca vs. Interpreting.” Interpreting 15 (1): 48–73. DOI: 10.1075/intp.15.1.03rei
Sabatini, Elisabetta. 2000. “Listening Comprehension, Shadowing and Simultaneous Interpreting

of two ‘Non-Standard’ English Speeches.” Interpreting 5 (1): 25–48.
DOI: 10.1075/intp.5.1.03sab
Seidlhofer, Barbara. 2009. “Accommodation and the Idiom Principle in English as a Lingua
Franca.” Intercultural Pragmatics 6 (2): 195–215. DOI: 10.1515/IPRG.2009.011
Seidlhofer, Barbara. 2011. Understanding English as a Lingua Franca. Oxford: Oxford University
Press.
Sinclair, John M. 1991. Corpus, Concordance, Collocation. Oxford: Oxford University Press.
Taylor, Christopher. 1989. “Towards a Typology of Speakers.” In Coming of Age. Proceedings
of the 30th Annual Conference of the American Translators Association, ed. by Deanna
Lindberg Hammond, 183–188. Medford, NJ: Learned Information.
Van Dijk, Teun A., and Walter Kintsch. 1983. Strategies of Discourse Comprehension. New York,
NY: Academic Press.
Widdowson, Henry. 1979. “Rules and Procedures in Discourse Analysis” In The Development
of Conversation and Discourse, ed. by T. Myers, 61–71. Edinburgh: Edinburgh University
Press.
Widdowson, Henry. 2004. Text, Context, Pretext: Critical Issues in Discourse Analysis. Oxford:
Blackwell. DOI: 10.1002/9780470758427
The role of intuition in the translation process
A case study
This chapter explores the psychological construct of intuition and its influence
in decision-making behavior. Intuition was defined by Jung (1971) as a primary
mode of perception operating subconsciously. As opposed to sensing personal-
ity types who prefer concrete details, intuitive personality types prefer to acquire
information by imagining possibilities (Myers and Myers 1995). In this chapter,
an analysis of verbalization data from a translation process study is discussed
in order to demonstrate the influence of intuition on decision-making during
the translation process and to explore the implications of this influence. Recent
studies have found that intuition plays a role in learning and decision-making
tasks involving affect (Laborde, Dosseville, and Scelles 2010: 786). Intuition is
therefore here viewed as a potentially vital component of translator behavior
which could predict individuals’ translating effectiveness.
Keywords: intuition, individual differences, translation process research,

decision-making, think aloud protocols
You are translating along, and stumble briefly on a word. “What

was that in the target language?” All of a sudden it comes to
you, out of nowhere, it seems, and your fingers type it. How do
you know it’s right? Well, you just know. It feels right. It feels
intuitively right. (Robinson 1997: 114–115)
1. Introduction
Translation Studies (TS) scholars would probably like to think that the field has
moved on from a time when it was thought that translation was mostly an un-
conscious process that translators would have difficulty explaining (Weaver 1989).
Certainly, studies in the area of Translation Process Research have multiplied over
the last few years and provided useful insights into the workings of the black box.
doi 10.1075/bct.77.05hub
64 Séverine Hubscher-Davidson
Nonetheless, much remains to be explained, and experts in the field still question
what exactly constitutes decision-making. In 1989, Weaver commented that, when
faced with a choice between two words, the translator hears the words in some
corner of his mind and likes the sound of one better than the other (1989: 117); in
1998, Newmark argued that intuition and imagination are as valuable in transla-
tion as awareness of what one is doing (1998: 77); in 2008, Munday postulated that
translation is not scientific and that the translation process always involves an ele-
ment of poetic taste (2008: 227). Over the years, it would seem that the idea of an
objective, conscious, straightforward decision-making process in translation has
not been universally accepted. Could there be a mysterious je ne sais quoi within
each translator that informs their choice of translation solutions? Could this inex-
plicable element be intuition? When weighing up different potential solutions to the
translation of a particular segment, can a first instinctive or unconscious thought
produce a more successful solution than one that is the result of careful analytical
deliberation? This chapter will attempt to provide an answer to these questions.
The concept of intuition can be said to have been rather neglected in academic
circles. It is often perceived to be an irrational and unreliable concept pertaining to
popular psychology rather than a scientific phenomenon worthy of investigation.
Despite this, the last two decades have witnessed a surge of interest in the topic. In
TS, process research has not yet fully explored the depths of nonconscious infor-
mation processing despite the will to go beyond traditional approaches to under-
stand the mechanisms underlying the translation act. Translators have often been
thought to deviate from rational patterns of behavior (Wilss 1996: 37), but I would
argue that what may appear irrational in the course of a translation act can actually
be the result of a highly effective nonconscious phenomenon.
In this chapter, an attempt is made to discuss the concept of intuition and
to demonstrate its influence on decision-making during the translation process.
First, relevant TS studies that have addressed intuition in a psychological context
are reviewed. Second, the construct is clarified and recent research in the field of
intuition is discussed. Third, an analysis of verbalization data from a translation
process study (henceforth referred to as ‘the case study’) is presented. The final
section of this chapter addresses some key issues in relation to the translator’s intu-
itive behavior. Intuition is viewed as a potentially vital component of translator be-
havior, and one which could be predictive of individuals’ translating effectiveness.
2. Intuition in Translation Studies
While it has been acknowledged for some time that cognitive and attitudinal fac-
tors can be very important during the translation process (e.g., Laukkanen 1996;
The role of intuition in the translation process 65
Jääskeläinen 1999; Dancette 2010), the concept of intuition has received little at-
tention in TS and, to my knowledge, only a handful of scholars have employed the
term. In order to understand what role intuition plays in the translator’s decision-
making process, this section presents a brief chronological review of studies in
translation which have explored the concept.
As early as 1981, Straight argued that translation decisions are a result of the
translator’s own reading of the original, and personal responses to the source text
(ST) and to the translation-in-progress (1981: 45). If that reading process matches
that of ST readers, then the translation is more likely to be successful, he claimed.
Although the idea that a personal reading of a ST might be helpful when translat-
ing is conceivable, Straight made other, less convincing, observations. He argued
that the translator’s decisions may be “influenced by an essentially ‘intuitive,’ right-
hemispheric mental apparatus” (1981: 49). In his view, the influence of the right
hemisphere is the reason why it is not always possible to provide logical explana-
tions for translation decisions. However, there is no evidence supporting the idea
that different thinking styles lie within each hemisphere and, in fact, the two brain
hemispheres are said to work together during cognitive effort (OECD 2002; Wager
et al. 2003; Pinel and Dahaene 2010). As a result, Straight’s theory does little to
enlighten us.
In 1988, Wilss made a distinction between objective and subjective transla-
tor behaviors, arguing that subjective intuition is used when a translator is faced
with a problem requiring a new approach in order to find a solution.1 In his view,
“the sphere where intuition is needed depends on the translator’s knowledge and
experience” (qtd. in Komissarov 1995: 351). Both Straight and Wilss conceived in-
tuition as a cognitive process influencing judgment and decision-making in trans-
lation, though they remained less clear as regards what activates and characterizes
intuitive processes.
Robinson (1991) argued that more work needed to be done on the ways in
which our bodies send us signals regarding what we know and how we should re-
act. Convinced that “we are guided much more powerfully […] by those autonomic
responses called “intuition” or “gut reactions”” (1991: x), Robinson calls transla-
tion a largely intuitive process, and postulates that we have somatic responses to
words on an unconscious level. The idea that translators’ body conditioning will
provoke a reaction, a feeling that a word may be ‘right’ before they are consciously
aware of what is happening, hints at the powerful influence of intuition.
Komissarov (1995) picks up Wilss’s distinction between objective and subjec-
tive behaviors during the translation process, and suggests:
1. The author’s comments in this section are based on Komissarov’s review of Wilss’s work, due
to the original only being available in German.
Decisions can be taken either after an objective analysis of facts and situations or
on the strength of subjective intuition. […] In many cases it is the intuitive type of
decision-making that plays the decisive role in the translating process. (1995: 349)
Kiraly (1995: 106) makes similar observations to Komissarov and highlights a use-

ful distinction between the conscious (controlled) and subconscious (intuitive)
areas of the brain, suggesting that translation problems emerge from the intuitive
workspace and are then considered in the controlled processing centre. In his dis-
cussion on how to become an effective professional, Robinson (1997: 2–3) empha-
sizes the need to balance conscious analysis with subliminal discovery, and for the
former to serve as a check on the latter.
While intuition scholars would broadly agree with TS scholars that an individ-
ual makes use of both intuition and conscious analysis, (e.g., Laborde et al. 2010),
it has also been argued that knowledge initially acquired by conscious reasoning
processes becomes automatic, or “experientialized,” through practice, thus becom-
ing tacit and intuitive (Epstein 2010: 307). In this case, intuition is based on highly
specific domain knowledge which makes it a valid resource for resolving problems
requiring objective consideration (Sadler-Smith and Shefy 2004; Pretz and Totz
2007; Hodgkinson et al. 2009; Evans 2010).
These early attempts to investigate the translation process marked a step for-
ward in our understanding of the role of intuition in decision-making. Around the
same time, an increasing number of TS scholars (e.g., Lörscher 1991; Jääskeläinen
1993; Hansen 1999) turned to cognitive psychology in order to investigate the
translation process, translation competence, and the translators’ use of strategies.
This heightened interest in cognitive processes led to the design of new studies
exploring translators’ decision-making. One example is Bǎlǎcescu’s (2006) study
on the strategies of student translators, whose conversations were recorded while
they worked on a translation task, and subsequently analyzed. Like Kussmaul
(1995: 244), Bǎlǎcescu observes that students who ‘intuitively’ come up with appro-
priate and creative solutions do not always make use of these in their translations.
She believes this behavior to be linked to a lack of confidence in their intuitions
and suggests that, during their training, translators should be made aware of the
validity of their intuitions (2006: 213). Bǎlǎcescu is quick to accept intuitive judg-
ments as valid, but intuition scholars are less convinced of their trustworthiness, a
point which is addressed in the following section.
This brief review of TS studies concerned with intuition has demonstrated
that the field has been interested in nonconscious and complex translation phe-
nomena for several decades. Intuition has been acknowledged as having an im-
portant role to play in the translation process, and researchers in translation have
offered informative glimpses of the workings of intuition. Recently, work in the
area of cognitive psychology has produced new information on what intuition is

and how it works. This calls for a review of the construct in order to clarify the role
of intuition in decision-making and problem solving during the translation act.2
3. Intuition in Psychology
3.1 The dual process model
Defining intuition is not an easy task. Epstein (2008: 23) highlights that intuition
has been given so many meanings that one might wonder whether the term has
any meaning at all. Nevertheless, recent work in cognitive psychology has placed
intuition firmly within the context of dual process models:
In the dual-process framework, intuition is […] automatic, holistic, affective, fast,
and associative, as contrasted with rational thought which is deliberate, analytical,
non-affective, slow, and rule-based. (Pretz and Totz 2007: 1249)
Dual process models make a clear distinction between intuitive and deliberate
thinking, qualifying the former as effortless and perception-like — the “low road”
— and the latter as analytic and controlled — the “high road” (Myers 2010: 371–
372). Our brains therefore are believed to process information using both conscious
analysis and nonconscious intuition. Whereas the former is said to lead to reasoned
decision-making, the latter occurs simultaneously and automatically, thus result-
ing in gut-feelings and implicit, spontaneous thoughts. Interestingly, as intuition
is extremely quick and works in parallel with conscious analysis, it is believed to
exceed the capacity of working memory, thus only making the outcome of intuition
available to us (Evans 2010: 314). This outcome is known as an intuitive judgment
(Dane and Pratt 2009: 3). Therefore, intuition processing is perceived to be mostly
nonconscious, and intuition outcomes are said to be experienced consciously.
Dual process models have been adopted in many different areas of psychology,
including the study of learning, memory, thinking, reasoning, and decision-making
(Evans 2010: 313). Though the categorical distinction between conscious analysis
and intuitive behavior is useful in terms of structuring our thinking about intuition,
Hammond (2010: 330) argues that judgments are not necessarily made at either
pole of the cognitive continuum, but could come from somewhere near the middle.
Betsch and Glöckner suggest that intuition and analysis share the cognitive work,
with intuitive processes working in the background while analytic processes are
2. In this chapter, it is understood that problem solving may consist of more than one instance
of decision-making.
Conscious and analytical
The mind
Automatic Deliberate
judgments resulting and
from acquired self-aware
expertise (inferential decision-
Nonconscious intuition) making
and intuitive
Complex,
non-sequential
judgments
(holistic intuition)
Figure 1. Dual processing activity
responsible for controlling and changing information (2010: 280). A balance there-

fore needs to be struck between the two processes, as translators could be process-
ing information in a manner that is neither purely analytical nor purely intuitive.
Figure 1 is my illustration of the cognitive activity continuum and the process
of intuition within the dual process framework.
3.2 Intuitive judgments: Characteristics, process and reliability
3.2.1 Different types of intuition

When exploring the unconscious part of dual process models, a further useful
distinction can be made. Different types of intuitions have been posited over the
years (e.g., associative, matching, accumulative) and Dane and Pratt (2009: 4) offer
a clear review, categorizing intuitions into three broad types: a vehicle for prob-
lem-solving, an input for making moral decisions, and an instrument facilitating
creativity. These types are said to differ in terms of their holistic associations, af-
fect, and speed, though they are all nearly instantaneous and produced through
nonconscious processes (Dane and Pratt 2009: 4). Due to the relatively low rel-
evance of moral intuition for the case study presented in this chapter, I will focus
on problem-solving and creative intuition.
According to Dane and Pratt (2009: 5), problem-solving intuition involves
“pattern matching, which is often honed through repeated training [and is] con-
nected to domain knowledge, or expertise.” The assumption is that a person will
view a situation in terms of whether it is similar to a past experience, and will
“match patterns between their environment and deeply held knowledge struc-
tures” (Dane and Pratt 2009: 6).
Creative intuition, on the other hand, is believed to be an input into the cre-
ative process and is defined as being “linked to a creative act of synthesis in which
disparate elements are fused together in novel combinations [to generate] some-
thing ‘truly new’ ” (Dane and Pratt 2009: 9–10). Creative intuitions are also said
to be preceded by an incubation period, i.e., “an antecedent to the rapid, holistic,
and associative operations that produce the intuition” (Dane and Pratt 2009: 11).
Table 1 was first published in Dane and Pratt (2009: 5), and I have adapted it so as
to illustrate the distinction between the two types of intuitions.
Despite employing different terminology, other scholars studying the concept
(e.g., Pretz and Totz 2007; Hogarth 2010; Sinclair 2011) also refer to two types
of intuitions. Sinclair (2011: 5–6) calls these intuitions ‘inferential’ and ‘holistic.’
While inferential processing, like problem-solving intuition, relies on matching
information with existing knowledge and experience, holistic processing, like cre-
ative intuition, relies less on existing knowledge.3
Pretz (2011: 24) argues that holistic intuition is likely to be most appropriate
when problems are perceived as highly complex, and inferential intuition is best
used when problems are perceived as relatively simple, either due to prior experi-
ence or the straightforward nature of the task. It could be argued that the use of
holistic intuition would be particularly well-suited to a literary translation task
where problems are likely to be complex in nature and to require the integration
of knowledge from different domains, whereas inferential intuition would be best
used in a technical or domain-specific translation task where pattern matching
may be more easily done.
Table 1. Two types of intuition (adapted from Dane and Pratt 2009: 5)
Intuition Type Description Nature of Affect Level of
Associations Incubation
Problem-Solving Automatic acts of Largely convergent/tight; Relatively Low to none
(inferential) recognition due to Based on highly specific low intensity
pattern matching domain knowledge
Creative Feelings that arise Largely divergent/broad; Relatively Often high

(holistic) when knowledge is Based on integration of high intensity
combined in novel knowledge across different
ways domains
3. From this point onwards, the terms ‘inferential’ and ‘holistic’ will be consistently used to refer
to these two types of intuition.
3.2.2 Accuracy and reliability of intuitive responses

Despite the appeal that processing knowledge without conscious effort might hold,
intuition experts are divided in their opinion of what Hammond terms ‘unjustified
cognitive activity’ (2010: 327). According to Myers, intuition generally functions
adaptively and efficiently, but can lead to ill-fated decisions if gut reactions are not
checked against reality: “Intuition is often wise but sometimes perilous, especially
when it leads us to overfeel and underthink” (2010: 374).
Intuitive responses can therefore inspire feelings of confidence in the wrong
answers. This observation seems particularly pertinent in light of Bǎlǎcescu’s
previously discussed argument regarding the validity of intuitions in translation.
Hammond (2010: 329) argues that intuition should not be encouraged solely on
the grounds that it supposedly outperforms the dull hard work of justification,
and Hogarth (2010: 344) highlights that errors in intuitive thought can be due to
bias induced from momentary affective states.4 However, Strick and Dijksterhuis
(2011: 28) do not see intuition as either useful or misleading, arguing that its qual-
ity is determined by the knowledge it was based on and the psychological process
that led to it.
Having considered various conceptualizations of intuition, it now seems per-
tinent to address how it can be measured.
3.3 Measurement
As it is difficult to access nonconscious processes, there is little agreement among

researchers as to how best to measure intuition (Hodgkinson et al. 2009: 342).
Dane and Pratt (2009: 16) admit that, although there is broad agreement on the
usefulness of combining analysis and intuition when making decisions, scholars
have not fully explored the best methods to integrate these approaches. The most
popular method currently used to measure the construct is a self-report personal-
ity test.
Personality tests generally include attitudinal statements aiming to capture
individual differences in preferences for intuitive or analytical processing. A test
measuring holistic intuition will contain items referring to a preference for abstract
processing and reliance on incubation in decision-making, while a test measuring
inferential intuition will contain items assessing one’s trust in snap judgments and
analytical short-cuts (Pretz and Totz 2007: 1256). It has been suggested that the
4. Many authors have discussed the role of affect in intuitive judgments, and the impact of
emotional states on intuition and decision-making. Although an important avenue for further
research, dealing with the concept of ‘affectively charged’ intuitions (Dane and Pratt 2009) is
beyond the scope of this chapter.
kind of rational or intuitive intervention that takes place in decision-making can

be determined by personality (Evans 2010: 314). In this sense, a translator with a
preference for intuition would rely more on initial feelings and be less disposed to
call upon rational thinking when making decisions.5
Although easy to administer, personality tests are subject to limitations (e.g.,
susceptibility to deliberate impression management) and they remain an indirect
measure of intuition. Evidence can be produced for one’s tendency to use intu-
ition, but scales do not capture intuition when it actually occurs.
Pretz and Totz (2007) found that the MBTI, or Myers Briggs Type Indicator
(Myers and Myers 1995), purports to measure holistic intuition as a preference
for abstract, conceptual thought. According to Myers and Myers (1995: 57–58),
intuitive types are attuned to the intuitions that spring from their unconsciousness
with great speed and “pop into the conscious mind with an effect of inspiration.”
In the case study described in the next section, I explore how a potentially strong
preference for using intuition manifests itself during the translation act and how a
translator can use it to his advantage.
4. The case study
In this section, I analyze three extracts from the Think Aloud Protocols (TAPs)
of a former student on the Master’s program in Translating and Interpreting at
the University of Bath. This data was collected as part of the doctoral work that I
conducted on the translation process, the aim of which was to better understand
the influence of students’ personality traits on their translations, as expressed
through their decision-making behaviors. In brief, the study took place in 2004;
the students were all English native speakers, with an average age of 27.4; and they
were asked to produce a translation of a literary text from French into English.
The text was 221 words in length and was originally published in a 1960s volume,
presented like a tourist guide of Paris and describing its mysterious aspects, with
stories of witches, plots and haunted places (Hubscher-Davidson 2009: 179). This
text was chosen due to its appealing content, which would, it was hoped, help keep
students interested and motivated. There was no specific brief, access to dictionar-
ies or time constraints.
The participants were required to think aloud and to complete a version of the
MBTI as well as other questionnaires. Once the participants finished, the target
5. Over 15 years ago, Robinson (1997: 73) hinted at the idea that certain types of learners pro-
cess information in a more intuitive manner than others. His early observations are consistent
with personality theory.
texts were collected and assessed anonymously. Interestingly, out of the twenty stu-
dents who participated in the experiment, the five most successful were all classed
as intuitive types by the personality measure. Thus, results indicated a strong cor-
relation between success in the task and the presence of an intuitive trait among
participants. The following examples are extracts from the TAPs of one of these
intuitive students, T2, who performed particularly well in the task and scored very
highly on intuition. His protocols were selected as they seemed to present traces
of both intuitive and rational behaviors, thus demonstrating the interplay of both
systems during translation.
The intention here is not to reassess the correlation between the personality
trait and translation competence, nor to judge the quality of translation solutions
adopted. Rather, the aim of reviewing the protocols is to highlight the mechanisms
of intuitive behavior during the translation process and to gain a better under-
standing of its influence during decision-making in light of the literature reviewed
above.
Please refer to the appendix for the complete ST, the ST and TT extracts, and
the three excerpts from T2’s TAPs, which are analyzed below.
Excerpt 1
In the first excerpt (see Appendix), the student is wondering how to translate the
first part of the seventh sentence of the ST, i.e., la lune est pompeuse ou fugitive. The
student starts by verbalizing his understanding of the moon’s trajectory through
Paris and ‘behavior’ during this time. If intuition is, as Robinson (1991: 15) puts it
“a feeling that a word is or is not ‘right,’ ” it could be argued that T2 has an initial
intuition that using the word pompous to describe the moon is ‘not right’: ‘I don’t
know, I don’t really like pompous here.’ Having verbalized this subjective feeling,
T2 then proceeds to verbalize what can only be described as an attempt to reason,
to rationalize his initial reaction: ‘I don’t think it lends itself very well to the, well
I suppose it is metaphorical language […] there’s no reason to stay quite literal.’
Despite this effort to bring back decision-making to a conscious and analytical
process which is based on knowledge and experience (i.e., recognition of meta-
phorical language and the preferred norm for non-literal renderings), T2 cannot
seem to shake off his initial intuition that pompous is not the best translation solu-
tion: ‘but my instinct is that it’s not…I prefer imperious.’ It might seem surprising
that T2 does not opt for imperious in his final TT, but this decision is consistent
with findings from other process studies (see Kussmaul 1995). Although it could
be said that T2 has come full circle, i.e., that he did not act on his intuition but
opted for a predictable literal translation, two important features can be extracted
from this specific process.
Firstly, it would seem that the student experiences some awareness that his de-
cision-making process could be influenced by intuition or instinct: ‘I’m not really
sure whether [my] hesitation with the word pompous is to do with a well-informed
instinct or a lack of appreciation of [2 sec. pause] the meaning of the passage.’
This is consistent with outcome-oriented intuition research, which highlights that
it is possible to be aware that intuition processing is happening on a conscious
level (Sinclair 2010: 380). T2’s observation is evidence of a tension between the
underlying process and the point at which it is consciously experienced. This ‘in
between’ state is called fringe consciousness (Topolinski and Strack 2009). In this
case, the realization that there may be some processing mechanism that escapes
his immediate awareness leads T2 to postpone making a decision: ‘I’ll just put a
note over that one.’
Secondly, T2 subsequently makes the decision to personify the moon as being
coy. He expresses dissatisfaction with his first translation of fugitive as fleeting, and
settles on a solution which transfers the elusive and playful nature of the moon.
In this instance, T2 opts for a creative solution without much debate or, it seems,
without really knowing why: ‘coy I suppose, I can’t really think of another mean-
ing, word, right now.’ Clearly this is not true, since he previously came up with an
alternative, fleeting. Although the reasons underlying his choice may have seemed
unfathomable to the student, it could be argued that his previous intuition and
reflection on the personification of the moon as imperious might have somehow
triggered an appropriate subsequent creative solution. I would therefore venture
that T2’s prior experience with a potentially well-informed instinct may have sub-
consciously encouraged him to rely on inferential intuition and prevented him
from analyzing alternative solutions.
Excerpt 2
In the second excerpt, T2 exhibits continued doubt in relation to the first part
of the seventh sentence. He is aware that continuing to play around with la lune
est pompeuse will not help him find a solution to au hasard des arrondissements
(‘it still doesn’t resolve the issue of depending on’). Nonetheless, having chosen
a translation solution that went counter to his initial intuition, the translation he
settled on could still feel problematic to him and he may be attempting to find
an alternative that might better match his initial somatic response. The verbaliza-
tions also suggest that T2 is working on at least two ST segments simultaneously,
as he repeats the moon is pompous several times while trying to solve au hasards
des arrondissements, thus perhaps still processing one problem subconsciously
while verbalizing in relation to another. Betsch and Glöckner (2010: 281) claim
that extensive information integration can occur even if cognitive capacities are
constrained by another task. Interestingly, in one of her TAP studies, Dancette

(2010: 74) also notes that a student spontaneously comes up with two translations
naturally while working on another segment, suggesting that leaving a problem
momentarily unresolved is an efficient strategy that enables the continuation of a
translation.
Another interesting feature of this excerpt is T2’s difficulty with syntax: ‘just
trying to think of a way of getting that phrase into the middle of the […] previous
part of the sentence.’ T2 is clearly trying to find a translation that will be structur-
ally sound and somehow combine the two ST clauses harmoniously in English.
This attempt seems to require so much effort that T2 considers leaving this com-
plex problem aside temporarily. Making an analytical decision was perhaps not
possible here, as it required the integration and weighing of many pieces of in-
formation (e.g., target text syntax, accuracy of meaning) rendering it particularly
difficult for T2 to ‘try and crack this phrase.’ This verbalization indicates that T2
found this segment challenging and akin to a mystery that needed to be deci-
phered. According to Pretz (2011: 21) some problems involve so many factors that
deliberate analysis becomes impossible and intuitive judgments are more likely to
be given consideration.
Excerpt 3
In the third excerpt, T2’s initial verbalization, frosty, immediately triggers a feel-
ing of uncertainty and an alternative TL word is offered (‘um […] don’t know
about frosty, icy I think’). This observed behavior recalls Epstein’s (2010) definition
of intuition as an almost instant sense of knowing seemingly resulting from un-
conscious information processing. Interestingly, T2 then approves of his solution
by expressing a positive feeling: ‘I think I’m quite happy with that.’ According to
Sinclair (2010: 381), a confirmatory feeling can accompany the outcome of intu-
ition processing as it emerges into consciousness, and it could be argued that T2
experiences this feeling. It could also be argued that T2 is experiencing the kind of
somatic response discussed by Robinson (1991) whereby his reasoning seems to
be emotionally directed (e.g., ‘I think I’m quite happy with that’) and his gut feel-
ing is intellectually monitored (e.g., ‘there might be a better way of saying that’).
T2 then tackles l’Esplanade des Invalides. The first (literal) translation that T2
verbalizes is Esplanade of the Invalides, a solution he initially seems content with
(‘think I’m just gonna put that’). However, he then expresses the need to ‘check
that,’ and verbalizes a ‘temptation’ to shorten the translation to the Invalides.
This could be an indication that T2 is experiencing a vague sense of knowing
that is emerging into consciousness, i.e., the outcome of an intuition. Not having
heard part of the verbalization related to this decision, I made the exception of
interrupting the verbalization in order to clarify what led to this second solution.
The student provided two justifications for his ‘temptation.’ The first, ‘[be]cause it’s
simpler,’ seemed to be an attempt at humor. The second justification seemed to be
an attempt to reason his choice: ‘because I think it’s something that is quite easily
recognizable to a lot of people.’ It could be argued that a conscious and systematic
consideration of relevant factors may not yet have taken place and that T2 is there-
fore struggling to explain the outcome of his intuition in terms that would not
amount to admitting the use of embarrassing guessing behavior.
T2’s second justification is guided by his assumption that target readers, like
him, will be educated, Francophile, and thus familiar with the green open space
that extends north of the Parisian Hôtel des Invalides. However, T2 quickly retracts
his assumption and decides that perhaps not all of his target readers would under-
stand what the Invalides are, berating himself for his subjectivity and insufficient
reflection: ‘that’s quite a subjective point of view […] when I actually think more
about it […] on second thought.’ T2 is clearly emphasizing the fact that his initial
intuitive solution had not been the result of careful deliberation, and that now he
‘knows better.’ It could be argued that the student is making a focused attempt to
bring back the decision-making process to a rational analysis of facts and to steer
clear of what Straight would call his personal, intuitive responses to the ST and to
the translation-in-progress (1981: 45).
Another salient feature of T2’s justification is the reference to how target read-
ers would behave upon reading the translation and to what knowledge they might
have as regards the Invalides. This verbalized concern for target readers is relevant
for at least two reasons. First, it could be argued that verbalized sensitivity to tar-
get reader behaviors is a feature of T2’s personality type and, more specifically, a
feature of his intuition. Second, this problem-solving behavior is consistent with
holistic intuition. Indeed, if innovation draws on experience from everyday life in
order to create something new (Sinclair 2010: 382), it is clear that T2 is drawing
on his life experience in terms of reader expectations to create an adequate TL so-
lution.6 This combination of experience and current stimulus results in disparate
patterns being associated in new and creative ways (Sinclair 2010: 383).
6. This is in conformance with Kussmaul and Tirkkonen-Condit’s observation that good trans-
lators are more prepared to use their world knowledge and their own inferences about text and
text type as a basis for their decisions (1995: 190).
5. Discussion
The analysis produced a number of interesting findings in relation to the interplay

between T2’s intuitive and rational systems, but also in relation to the expression
and management of different types of intuitive judgments. The fact that this ex-
plorative study analyzed only one student’s verbalizations clearly does not enable
generalizations to be made about typical intuitive translator behavior. The case
study, however, illustrated a number of interesting processing behaviors which can
be summarized as follows:
1. Some intuitive translation judgments are immediate and are followed by a
conscious effort to reason and rationalize;
2. Intuitions are not necessarily used in final translation decisions, even when
they might lead to appropriate solutions;
3. Regardless of their accuracy, intuitive translation judgments can be consid-
ered suspiciously and can be difficult to shake off;
4. The translator may have conscious awareness of the presence of some aspects
of intuition, which can lead to tension;
5. Both inferential and holistic intuitions seem to be experienced when translat-
ing, with evidence of attempts to match information with stored patterns but
also to combine different types of information to create something new;
6. It would appear that translation problems can be processed on conscious and
subliminal levels simultaneously;
7. When faced with translation problems perceived as particularly complex, the
translator may favor intuition over deliberate analysis.
In addition to these points, a further observation can be made. It could be hypoth-
esized from the case study that a personality preference for using intuition may
indicate increased sensitivity to target readership expectations. This corroborates
arguments by both Straight (1981) and Robinson (1991) that self-projection and
sensitivity to the feel of the SL text are necessary to produce a successful TL text. A
translator’s personality preference for using intuition during decision-making in
translation would therefore seem to be a key avenue for further research.
In relation to the above, three key areas will now be discussed in further detail:
the intuitive personality trait; the relationship between intuitions and expertise;
and the potential for developing intuition.
5.1 An intuitive personality trait
The analysis demonstrated that T2 experienced the outcome of rapid intuitive

processes, often verbalized as feelings or instincts. More often than not, these
intuitions were then followed by reflective processes, thus verifying the initial feel-
ings with explicit reasoning. This is what Evans (2010: 314) calls the default-inter-
ventionist structure, suggesting that dispositional thinking style determines the
likelihood of this intervention. Interestingly, the personality instrument identified
T2 as an INFJ or Introverted Intuitive Feeling Judging type, which suggests that
he is exceptionally intuitive, and that this intuition is responsible for his imagina-
tion, originality, and visions (Myers and Myers 1995: 110). However, INFJs are also
judging types, which implies that they do not usually exploit unplanned events,
and that they are rational and depend on reasoned judgments (Myers and Myers
1995: 75). T2’s preference for intuition is therefore tempered by his propensity to
judge. According to Myers and Myers (1995: 110) introverted intuitive types need
to cultivate this judging process to balance their intuition and to combine vision
and practicality successfully.
T2’s rational thinking could be perceived on several occasions (e.g., ‘when I
actually think more about it’), and it is interesting to note this link between per-
sonality traits and the handling of intuitive judgments, which is in line with the
observation made by Robinson (1997: 73) that learning styles influence informa-
tion processing. As alluded to in Section 3, successfully integrating intuition and
analysis can improve performance by drawing on both conscious and noncon-
scious resources (Kuhnle 2011: 228). As a result, people who may not have well-
balanced traits may be inclined to rely on their impulsive and effortless intuitions,
thus running into trouble. Promoting a reliance on ‘gut feelings’ would therefore
be ill-advised, as they can be misleading in the absence of an active reasoning
process.
5.2 The relationship between holistic/inferential intuitions and expertise
One of the key findings from the analysis is that both holistic and inferential in-
tuitive judgments were present alongside conscious analysis in the TAPs, which
suggests that the translator had to deliberate on the appropriateness of both types
of intuitions.
Interestingly, the literature suggests that participants in reasoning experiments
tend to rely on intuition as it requires less effort than reasoning, particularly when
tasks become more complex (Frederick 2005; see also Stanovich 2009; Evans 2010;
Betsch and Glöckner 2010). The issue here is that an experienced translator will
probably have useful inferential intuition to rely on, but novices could instead rely
on biased judgments. According to Hogarth (2010: 343), inferential intuition can
be trusted when a person’s past experience is representative of the current situa-
tion but, if this is not the case, one should be careful.
The development of experience is a continuum along which novices become

semi-competent before becoming fully-fledged professionals. Sinclair argues that
this is why semi-experts tend to deliberate very carefully: “they know enough to
be cautious and not enough to be intuitive” (2010: 382). As a semi-expert, some
of T2’s inferential intuitions may be successful due to previous experience with a
similar task, but less reliable holistic intuitions could be used for perceived com-
plexity where analysis is unlikely to bring about a successful decision. As such, it
is important to exercise caution when drawing conclusions from studies on the
behavior of a semi-expert.
5.3 Developing intuition
Despite the fact that intuition processing is not a conscious process, recent re-
search suggests that it is possible to train ourselves to pay attention to cues in
our environment that may help uncover solutions to problems. Interestingly, this
would suggest that translators may be trained to become more in tune to the out-
comes of their intuition. Sinclair (2010: 380) suggests that this requires a goal-
directed thought process aimed at a specific task, and that operating on a primed
nonconscious level might enable intuition processing to be activated. This ties in
with Robinson’s (1991) argument that translators should pay particular atten-
tion to their somatic responses, and with Bǎlǎcescu’s (2006) point that translators
should be made aware of valid intuitions during training. This study provides ad-
ditional evidence that thorough training in the handling of intuitive judgments
when translating would be beneficial.7
Dane and Pratt (2009: 31) suggest that different types of intuition may be use-
fully considered during different stages of the decision-making process, and that
individuals should return to their initial intuitions once they have undertaken an
analytical assessment of a problem in order to re-evaluate these. This would also
enable individuals to stay attuned to the potential emergence of holistic intuitions
which, we have noted, require a period of incubation and can become useful in the
final stages of complex decision-making.
Although consensus is yet to be reached on the best way to develop intuition,
the potential to enhance translation decision-making processes should be exploit-
ed. Dane (2011: 221) suggests that even attending to inaccurate intuitions gives
one the opportunity to compensate for them, and decide what role they should
play in decision-making.
7. For those wishing to apply formative techniques in the translation classroom, Robinson’s
1997 volume provides exercises designed to develop subliminal skills alongside conscious pro-
cessing.
6. Conclusion
The present study aimed to shed light on the role that intuition played in one
translator’s decision-making in order to enhance our understanding of the cir-
cumstances when trusting intuition is an appropriate strategy in translation. It
is clear that validation studies are needed to build on this subjective analysis, but
the study demonstrated the worthwhile nature of further investigating the mecha-
nisms underlying intuition.
The method used to collect information on the translation process in this
study remains the topic of a vigorous debate, since the validity of verbal reports
as a way to tap into the translator’s mental processes has been hotly contested.
Jääskeläinen (2011: 16) argued that the method is based on the idea that partici-
pants can report accurately on information that is consciously processed but not
on automated processes. This implies that intuitive outcomes may feature in the
protocols, but not intuitive processes, which are not consciously experienced. In
addition, among other issues, Göpferich and Jääskeläinen (2009: 179) report that
think-aloud data do not necessarily reflect the order in which mental processes
occur. It must therefore be borne in mind that the instances of intuitive behavior
gauged from T2’s protocols are just a partial representation of the complex and
(mostly) unconscious processes actually taking place.
However, I would argue that TAPs are still useful for exploring intuition. One
area where the design of intuition studies can be improved is by experimenting
in multimodal situations (Betsch and Glöckner 2010: 290), and thinking aloud
could be one of the multimodal methods with which intuitive thinking processes
are investigated. Wilss (1998: 59) highlights that understanding translators’ pre-
choice behaviors must be a priority in future studies, particularly as the develop-
ment of internalized processes is not well understood. The impact of intuition in
translation processing may be far more important than we imagine, and the topic
should no longer be a marginal area of enquiry. For example, it has been found
that speeded tasks encourage a greater reliance on intuition, to the detriment of
logical reasoning (Evans 2010: 318). This is clearly relevant for those studying the
translation process of professionals working under real-world time constraints.
This study demonstrated that it is time to raise awareness among the transla-
tion community not only of the benefits of taking ‘the high road’ but also the im-
portance of intuitive processing. Long gone is the time when translators’ intuitions
were considered to be imponderable and inaccessible to rational explanation.
References
Bǎlǎcescu, Ioana. 2006. “Intuition et Contraintes Sémantiques: le “Joker Sémantique.” Atelier de

Traduction — Ştefan cel Mare de Suceava University 5–6: 209–220.
Betsch, Tilmann, and Andreas Glöckner. 2010. “Intuition in Judgment and Decision Making:
Extensive Thinking Without Effort.” Psychological Inquiry: An International Journal for the
Advancement of Psychological Theory 21: 279–294.
Dancette, Jeanne. 2010. “Understanding Translators’ Understanding.” In Meaning in Translation,
ed. by Barbara Lewandowska-Tomaszczyk, and Marcel Thelen, 65–77. Frankfurt: Peter
Lang Publishing.
Dane, Erik. 2011. “Capturing Intuitions ‘In Flight’: Observations from Research on Attention
and Mindfulness.” In Handbook of Intuition Research, ed. by Marta Sinclair, 217–226.
Cheltenham: Edward Elgar Publishing.
Dane, Erik, and Michael Pratt. 2009. “Conceptualizing and Measuring Intuition: A Review of
Recent Trends.” International Review of Industrial and Organizational Psychology 24: 1–40.
Epstein, Seymour. 2008. “Intuition from the Perspective of Cognitive Experiential Self-Theory.”
In Intuition in Judgment and Decision Making, ed. by Henning Plessner, Cornelia Betsch,
and Tilmann Betsch, 23–37. New York, NY: Erlbaum.
Epstein, Seymour. 2010. “Demystifying Intuition: What It Is, What It Does, and How It Does It.”
Psychological Inquiry: An International Journal for the Advancement of Psychological Theory
21 (4): 295–312.
Evans, Jonathan St. B.T. 2010. “Intuition and Reasoning: A Dual-Process Perspective.”
Psychological Inquiry: An International Journal for the Advancement of Psychological Theory
21: 313–326.
Frederick, Shane. 2005. “Cognitive Reflection and Decision Making.” Journal of Economic
Perspectives 19 (4): 25–42. DOI: 10.1257/089533005775196732
Göpferich, Susanne, and Riitta Jääskeläinen. 2009. “Process Research into the Development of
Translation Competence: Where are We, and Where do We Need to Go?” Across Languages
and Cultures 10 (2): 169–191. DOI: 10.1556/Acr.10.2009.2.1
Hammond, Kenneth R. 2010. “Intuition, No!… Quasirationality, Yes!” Psychological Inquiry: An
International Journal for the Advancement of Psychological Theory 21: 327–337.
Hansen, Gyde (ed). 1999. Probing the Process in Translation: Methods and Results. Copenhagen:
Samfundslitteratur.
Hodgkinson, Gerard P., Eugene Sadler-Smith, Marta Sinclair, and Neal M. Ashkanasy. 2009.
“More than Meets the Eye? Intuition and Analysis Revisited.” Personality and Individual
Differences 47: 342–346. DOI: 10.1016/j.paid.2009.03.025
Hogarth, Robin M. 2010. “Intuition: A Challenge for Psychological Research on Decision
Making.” Psychological Inquiry: An International Journal for the Advancement of
Psychological Theory 21: 338–353.
Hubscher-Davidson, Séverine. 2009. “Personal Diversity and Diverse Personalities in
Translation: A Study of Individual Differences.” Perspectives: Studies in Translatology 17 (3):
175–192. DOI: 10.1080/09076760903249380
Jääskeläinen, Riitta. 1993. “Investigating translation strategies.” In Recent Trends in Empirical
Translation Research, ed. by Sonja Tirkkonen-Condit, 99–120. Joensuu: University of
Joensuu.
Jääskeläinen, Riitta. 1999. Tapping the Process: An Explorative Study of the Cognitive and Affective
Factors Involved in Translating. Joensuu: University of Joensuu.
Jääskeläinen, Riitta. 2011. “Back to Basics: Designing a Study to Determine the Validity and
Reliability of Verbal Report Data on Translation Processes.” In Cognitive Explorations of
Translation, ed. by Sharon O’Brien, 15–29. London: Continuum Studies in Translation.
Jung, Carl Gustav. 1971. Psychological Types. Princeton, NJ: Princeton University Press.
Kiraly, Don. 1995. Pathways to Translation: Pedagogy and Process. Kent, OH: Kent State
University Press.
Komissarov, Vilen Naumovich. 1995. “Intuition in Translation.” Target 7 (2): 347–354.
DOI: 10.1075/target.7.2.10kom
Kuhnle, Claudia. 2011. “The Benefit of Intuition in Learning Situations.” In Handbook of
Intuition Research, ed. by Marta Sinclair, 227–236. Cheltenham: Edward Elgar Publishing.
Kussmaul, Paul. 1995. Training the Translator. Amsterdam: John Benjamins.
DOI: 10.1075/btl.10
Kussmaul, Paul, and Sonja Tirkkonen-Condit. 1995. “Think-Aloud Protocol Analysis in
Translation Studies.” TTR: Traduction, Terminologie, Rédaction 8 (1): 177–199.
DOI: 10.7202/037201ar
Laborde, Sylvain, Fabrice Dosseville, and Nicolas Scelles. 2010. “Trait Emotional Intelligence and
Preference for Intuition and Deliberation: Respective Influence on Academic Performance.”
Personality and Individual Differences 49: 784–788. DOI: 10.1016/j.paid.2010.06.031
Laukkanen, Johanna. 1996. “Affective and Attitudinal Factors in Translation Processes.” Target 8
(2): 257–274. DOI: 10.1075/target.8.2.04lau
Lörscher, Wolfgang. 1991. Translation Performance, Translation Process, and Translation
Strategies. Tübingen: Gunter Narr.
Munday, Jeremy. 2008. Style and Ideology in Translation. London: Routledge.
Myers, David G. 2010. “Intuition’s Powers and Perils.” Psychological Inquiry: An International
Journal for the Advancement of Psychological Theory 21: 371–377.
Myers, Isabel Briggs, and Peter B. Myers. 1995. Gifts Differing: Understanding Personality Type.
Mountain View, CA: Davies-Black Publishing.
Newmark, Peter. 1998. “The Customer as King: A Response to Hans G. Honig.” In Translation
and Quality, ed. by Christina Schäffner, 75–78. Clevedon: Multilingual Matters.
OECD. 2002. Understanding the Brain — Towards a New Learning Science. Paris: OECD
Publishing.
Pinel, Philippe, and Stanislas Dehaene. 2010. “Beyond Hemispheric Dominance: Brain Regions
Underlying the Joint Lateralization of Language and Arithmetic to the Left Hemisphere.”
Journal of Cognitive Neuroscience 22 (1): 48–66. DOI: 10.1162/jocn.2009.21184
Pretz, Jean E. 2011. “Types of Intuition: Inferential and Holistic.” In Handbook of Intuition
Research, ed. by Marta Sinclair, 17–27. Cheltenham: Edward Elgar Publishing.
Pretz, Jean E., and Kathryn Sentman Totz. 2007. “Measuring Individual Differences in Affective,
Heuristic, and Holistic Intuition.” Personality and Individual Differences 43: 1247–1257.
DOI: 10.1016/j.paid.2007.03.015
Robinson, Douglas. 1991. The Translator’s Turn. Baltimore: John Hopkins University Press.
Robinson, Douglas. 1997. Becoming a Translator. London: Routledge.
Sadler-Smith, Eugene, and Erella Shefy. 2004. “The Intuitive Executive: Understanding and
Applying ‘Gut Feel’ in Decision Making.” Academy of Management Executive 18 (4): 76–91.
DOI: 10.5465/AME.2004.15268692
Sinclair, Marta. 2010. “Misconceptions About Intuition.” Psychological Inquiry: An International

Journal for the Advancement of Psychological Theory 21: 378–386.
Sinclair, Marta. 2011. “An Integrated Framework of Intuition.” In Handbook of Intuition Research,
ed. by Marta Sinclair, 3–16. Cheltenham: Edward Elgar Publishing.
DOI: 10.4337/9780857936370
Stanovich, Keith E. 2009. What Intelligence Tests Miss. The Psychology of Rational Thought. New
Haven, CT: Yale University Press.
Straight, Stephen. 1981. “Knowledge, Purpose, and Intuition: Three Dimensions in the
Evaluation of Translation.” In Translation Spectrum. Essays in Theory and Practice, ed. by
Marilyn Gaddis-Rose, 41–51. Albany, NY: State University of New York.
Strick, Madelijn, and Ap Dijksterhuis. 2011. “Intuition and Unconscious Thought.” In Handbook
of Intuition Research, ed. by Marta Sinclair, 28–36. Cheltenham: Edward Elgar Publishing.
Topolinski, Sascha, and Fritz Strack. 2009. “Scanning the ‘Fringe’ of Consciousness: What is
Felt and What is Not Felt in Intuitions About Semantic Coherence.” Consciousness and
Cognition 18: 608–618. DOI: 10.1016/j.concog.2008.06.002
Wager Tor D., Phan K. Luan, Liberzon, Israel, and Taylor Stephan F. 2003. “Valence, Gender, and
Lateralization of Functional Brain Anatomy in Emotion: a Meta-analysis of Findings from
Neuroimaging.” NeuroImage 19: 513–531. DOI: 10.1016/S1053-8119(03)00078-8
Weaver, William. 1989. “The Process of Translation.” In The Craft of Translation, ed. by John
Biguenet, and Rainer Schulte, 117–125. Chicago, IL: Chicago University Press.
Wilss, Wolfram. 1988. Kognition und Übersetzen: Zu Theorie und Praxis der menschlichen und
der maschinellen Überzetzung. Tübingen: Niemeyer.
Wilss, Wolfram. 1996. Knowledge and Skills in Translator Behavior. Amsterdam: John Benjamins.
DOI: 10.1075/btl.15
Wilss, Wolfram. 1998. “Decision Making in Translation.” In Routledge Encyclopedia of Translation
Studies, ed. by Mona Baker, and Kirsten Malmkjær, 57–60. London: Routledge.
Appendices
T2 Experiment Data: Text, Translation, and Think Aloud Protocols

Guide de Paris mystérieux.
Paris est une ville mystérieuse. Rien n’est plus mystérieux que Paris. Il n’y a qu’à voir la Tour
Eiffel promener ses gros yeux sur la ville pour sentir qu’il se passe des choses. Lesquelles ? On
ne sait pas trop, mais c’est très inquiétant. La Seine est noire et roule une eau sale. La lune est
pompeuse ou fugitive, au hasard des arrondissements. Elle étale sa lueur glacée sur l’Esplanade
des Invalides ; ailleurs elle passe en 15 secondes tant le ciel est étroit. Elle éclaire d’un rayon
oblique le tombeau des poètes, Baudelaire, Henri Heine, ceux d’Abélard et d’Héloïse, qui furent
si malheureux et si intelligents (je tiens la chose de ma femme de ménage). Autant de fantômes,
autant de mystères. Encore faut-il vouloir les voir. Wilde assurait qu’ « un gentleman ne regarde
jamais par la fenêtre ». Il habitait alors Quai Voltaire. Moins de préjugé aristocratique lui aurait
permis de s’étonner. Il était surpris de voir les arrondissements se succéder en escargot, et la
Seine couler d’est en ouest, ce qui la met à deux pas de l’Océan, et fait de Paris l’un des plus
grands de nos ports de mer. Caprices de la nature et hasards de l’Histoire semblent s’être ainsi
donné le mot pour faire éclore et conserver mystérieusement l’originalité de Paris.
(Caradec, François and Jean-Robert Masson (eds). 1985. Guide de Paris Mystérieux. Collection
Les Guides Noirs. Nouvelle Édition. Paris: Éditions Tchou)
Source Text Extract

La lune est pompeuse ou fugitive, au hasard des arrondissements. Elle étale sa lueur glacée sur
l’Esplanade des Invalides ; ailleurs elle passe en 15 secondes tant le ciel est étroit.
Target Text Extract

The moon is pompous one minute, coy the next, as she moves from arrondissement to ar-
rondissement. She spreads her icy glow over the Esplanade of the Invalides; elsewhere she is
obscured in 15 seconds, so narrowed is the sky.
TAPs excerpt 1: can the moon be pompous?

ailleurs elle passe en 15 secondes (4s) the sky is so straight, so narrow (4s) ok so over the Invalides
it’s being pompous (2s) it’s showing itself, it’s got time (2s) elsewhere (2s) it’s less visible, it’s cut
out by the, by the buildings (2s) I assume. (2s) ok, so the moon is (2s) pompous I don’t know, I
don’t really like pompous here, umm I don’t think it lends itself very well to the, well I suppose
it is metaphorical language actually so (2s) there’s no reason to stay quite literal but my instinct
is that it’s not (2s) I prefer imperious. Cos I’m not quite sure what, what the French is getting at
by saying it’s pompous (2s) but I suppose no, I mean I don’t know, the moon is pompous, it’s a
point of view (10s) see I’m not really sure whether that’s my (2s) hesitation with the word pomp-
ous is to do with uh a well informed instinct or a lack of appreciation of (2s) the meaning of the
passage as a whole, so I’ll just put a note over that one. so the moon (2s) is pompous or (2s) I’m
tempted to say fleeting but that doesn’t really (2s) convey the same meaning as fugitive or (2s)
the moon is pompous (4s) or (2s) coy I suppose, I can’t really think of another meaning, word,
right now to, for fugitive (4s)
TAPs excerpt 2: the one minute option

the moon is pompous (5s) what is posing me the problem is all I can think up for the minute is
for au hasard des is depending on and that just doesn’t sound right, the moon is pompous (2s)
I suppose if I say the moon is at times pompous, at times coy (6s) although this kinda does, if
I say at times, this does kind of suggest that it’s quite retracted, that it’s not necessarily (5s) hm
cos I could say at one minute, the moon is at one minute, the moon is pompous, one minute it’s
coy (3s) it still doesn’t resolve the issue of depending on (3s) I could say either or (3s) or (3s) one
minute (2s) one minute (5s) au hasards des arrondissements uh (5s) just trying to think of a way
of getting that phrase into the middle of the, of the previous part of the phrase, of the previous
part of the sentence (6s) I’m tempted to come back to it but I kind of want to actually try and
crack this phrase cos, I don’t want to leave myself too much to do afterwards, I’d like to get the
(1s) the essential bits done on (2s) my first reading (2s) my first working (20s) uh I might just
settle for something like as it moves between (2s) the arrondissements (2s) although I’m not sure
whether that’s going to be accurate enough (4s) if I take the one minute option then perhaps that
might be accurate enough (1s)
TAPs excerpt 3: icy glow on the (Esplanade des) Invalides

she spreads her frosty glow (3s) um (2s) don’t know about frosty, icy I think (1s) ok so I think
I’m quite happy with that, elle étale (2s) there might be a better way of saying that, elle étale (3s)
but I think I’ll go with that for the minute (3s) so she spreads (1s) her icy glow (1s) over the
(1s) Esplanade des Invalides um (5s) Esplanade of the Invalides (3s) think I’m just gonna put
that, Esplanade of the Invalides (3s) I might have to check that (2s) see (…?…) (5s) I’m almost
tempted just to say the Invalides but (3s) elsewhere {researcher asks: ‘why just the Invalides?’} cos
it’s simpler uh *laughs*(2s) uh because I think it’s something that is quite easily recognisable to
a lot of people but then that’s quite a subjective point of view I think so (2s) I think well when I
actually think more about it then there are probably quite a lot of people that haven’t got a clue
what the Esplanade des Invalides is (3s) no actually on second thoughts so I think that (2s) to
keep it as the Esplanade of the Invalides is probably more fitting to the text anyway because it
ties in with, with the pompe (2s)
The effect of interpreting experience
on distance dynamics
Testing the literal translation hypothesis
Nataša Pavlović and Goranka Antunović
The aim of this process-oriented study is to test the literal translation hypothesis,
i.e., the frequently held view that, in their search for the most suitable translation
solution, translators move from more literal towards freer renderings. The study
uses keystroke-logging to investigate translation processes involving self-revision
in terms of ‘distance dynamics.’ The latter term refers to the fact that each new
solution to a particular translation problem can be closer or more distant from
the source item than the previous one. A more specific hypothesis is also tested,
namely that experience in interpreting has an effect on distance dynamics. To
this end, processes of two groups of subjects — translators without substantial
interpreting experience, and interpreters — are compared. The findings chal-
lenge the literal translation hypothesis, and highlight some methodological
issues that should be tackled in future studies.
Keywords: literal translation hypothesis, deliteralization, translation process,

self-revision, keystroke logging
1. Introduction
In this chapter we present the findings of a study in which we tested the literal trans-
lation hypothesis, more aptly named ‘deliteralization hypothesis’ (see Chesterman
2011). This hypothesis reflects the frequently held view (see Ivir 1981; Toury 1995;
Tirkkonen-Condit 2005; Englund Dimitrova 2005; Shlesinger and Malkiel 2005;
Jakobsen, Jensen, and Mees 2007) that translators first consider cognate solutions
and that their self-revision involves a move from such literal solutions towards
freer ones. For example, Tirkkonen-Condit (2005) posits that “literal translation is
a default rendering procedure, which goes on until it is interrupted by a monitor
doi 10.1075/bct.77.06pav
86 Nataša Pavlović and Goranka Antunović
that alerts about a problem in the outcome,” building on Ivir (1981; 1988), who
claimed:
[t]he translator begins his search for translation equivalence from formal corre-
spondence, and it is only when the identical-meaning formal correspondent is
either not available or not able to ensure equivalence that he resorts to formal cor-
respondents with not-quite-identical meaning or to structural and semantic shifts
which destroy formal correspondence altogether. (Ivir 1981: 58)
It is not quite clear whether Ivir’s statement is a recommendation for translators or

a description of what presumably happens during the translation process, and if it
is the latter, whether it refers to cognitive processing or to actual text production.
This is pointed out by Toury (1995: 192), who brings up this issue in the context of
‘interim solutions.’ However, it can be assumed that at least some of the authors are
referring to actual text production, therefore, to something that can be accessed,
studied and measured. In particular, Englund Dimitrova (2005: 234) suggests that
translators “use literal translation as part of their translation process, as a step in
a process.” In her think-aloud study involving nine subjects she identified three
prototypical patterns in which literal translations appear in the process: a. literal
translations are formulated orally, then revised orally before a version is written; b.
literal translations are initially written, then evaluated and revised during drafting;
and c. literal translations are written as provisional solutions, and revised dur-
ing post-drafting revision. The interpretation of the literal translation hypothesis
as applying to text production is also assumed in Chesterman’s discussion of the
merits of the hypothesis (2011: 28).
Another moot point is whether the literal translation hypothesis refers to the
whole translation process or only to the first revision of each problematic element.
If the former is the case, as we would be inclined to believe, it might be more
appropriate to use Chesterman’s term and speak about “a deliteralization hypoth-
esis, i.e., a move from more literal to less literal” (Chesterman 2011: 27, italics in
the original). Doing that would also solve the rather unfortunate possibility for
the traditional name of the hypothesis to be interpreted as implying that the first
translation solution is necessarily a literal translation (see Chesterman 2011: 26).
Furthermore, the claim says nothing about how strong a tendency we may
expect: if we are assuming that literal solutions are the default, and that subsequent
solutions present a move away from the source, do we expect this to be true in all
cases, most of the time or simply more often than not? Is the other type of ‘move’
— that in which an initial freer solution is followed by a more literal one — to be
expected at all? If both situations are possible, what is the ratio between them, and
which variables may be expected to affect these dynamics, which we have decided
to term ‘distance dynamics’ (reflecting the fact that a solution can be closer or
The effect of interpreting experience on distance dynamics 87
more distant from the source item than the previous solution)? This brings us to
the final question that needs to be addressed: does the claim apply to all translation
processes, regardless of who translates, in which direction, what type of source text
is involved, whether there is time pressure or not, etc.?
2. Aims and hypotheses
In this research we set out to test the literal translation hypothesis, restricting its
scope to actual text production rather than applying it to cognitive processing in
general. We assumed, as does Chesterman (2011: 28), that keystroke logging is
a suitable method to study such text production. The general hypothesis of our
research was therefore formulated as follows: in their search for the most suitable
translation solution, translators move from more literal towards freer renderings.
In an earlier study (Antunović and Pavlović 2012), which involved students
translating from their second and third languages (L2 and L3), we had already
tested this hypothesis but the evidence supporting it was not convincing enough.
In the present study we decided to test professional subjects and, more specifically,
to compare translation processes of translators (without substantial interpreting
experience) with interpreters working on the same translation task. Interpreters
were chosen because of their experience in immediately producing acceptable TL
solutions, enhanced by deverbalization / conceptualization habit (the relevance
of the concept in interpreter training is confirmed in e.g., Gile 1990: 33 or, in par-
ticular, Dejean Le Féal 1993). We assumed that the ability to immediately produce
acceptable TL solutions would be activated also in the written translation tasks,
and that the first solution suggested by interpreters would be less literal than that
proposed by translators. Instances of deliteralization were therefore expected to be
less frequent in the interpreter group.
The above assumptions led to our more specific hypothesis, namely that expe-
rience in interpreting has an effect on ‘distance dynamics.’ The effect is expected to
be evident in that translation processes of interpreters, when compared to those of
translators, involve: a. fewer self-revisions on the whole; b. fewer self-revisions that
involve deliteralization; c. a relatively smaller share of deliteralizing self-revisions
in the overall number.
Subhypotheses (b) and (c) build on the findings of Jakobsen et al. (2007). In
their study inspired by Shlesinger and Malkiel (2005), they compared the process-
ing of idioms by interpreters and translators. They had expected both groups of
subjects to start off by looking for solutions based on idiom-to-idiom translation,
which, however, proved to be true only of the translator group. Contrary to the au-
thors’ expectations, occurrences of ‘direct transfer’ (literal translation) were found
to be more numerous in the translators’ processes (2007: 235). Translators also had

a larger number of ‘falsely coined cognate idioms,’ that is, idioms constructed by
means of literal translation.
With regard to the Shlesinger and Malkiel (2005) study, there is a major dif-
ference in the focus between their research and ours. While they compared the
incidence of cognates found in the final products of interpreting and (later) trans-
lation tasks by the same subjects, we examined the move from more literal to freer
renderings (or vice versa) within the processes of two groups of subjects with dif-
ferent (training and) habits.
3. Method
In this study we tested six professional translators and six interpreters (four of
whom were ESIT graduates) with Croatian as their first language (L1 or A) and
English as L2 or B-language. At the time of the experiments, all of the subjects
had five or more years of professional experience in translation and interpreting
respectively, and were working for EU institutions.
The two groups — the translators and the interpreters — were asked to trans-
late the same text, 168 words long, from the EU portal at www.europa.eu, en-
titled Documents you will need. The subjects were informed about the origin of
the source text (ST) and were given a realistic translation brief. The translation
sessions were recorded with the help of the keystroke-logging program Translog
2006,1 based on Jakobsen and Schou (1999) and further developed by Jakobsen
(2006). Having been introduced to the program, the subjects were asked to do
a warm-up task in order to familiarize themselves with the application. No ex-
ternal resources were available, and the time was limited in order for the logging
program to capture a maximum number of solutions. We believed that with time
constraints the subjects would be less likely to formulate solutions in their mind
before writing them down and thereby make them inaccessible for analysis.
Following the experiments, all problems and self-revisions were identified
and listed. For the purposes of this research, ‘problem’ was defined as every seg-
ment of the ST that involved at least one self-revision. Self-revisions encompassed
all the changes introduced into the target text (TT) by the translator during the
process of translation. As in our previous research on self-revision (Antunović
and Pavlović 2011; 2012), we excluded typos and followed the method used by
Englund Dimitrova (2005: 116) and Brenda Malkiel (2009: 156) in deciding what
counted as an instance of self-revision. They both observe that several distinct,
1. http://translog.dk/
unrelated changes can be made in a single operation and deem it appropriate in

such instances to count each change as a separate revision. On the other hand, as
Malkiel (2009: 156) points out, “when one change demands another change of the
same type, this was counted as a single self-correction.” For instance, if changing a
noun from singular to plural involves a change in the verb ending to ensure agree-
ment, this would be counted as a single change.
Next, the self-revisions were labeled in terms of ‘distance dynamics,’ as be-
longing to one of the following three types (see Englund Dimitrova 2005: 117):
1. a move towards target (further ‘T-move’); 2. a move towards source (further
‘S-move’); or 3. neutral change. The following examples illustrate the three types:
1. T-move is a revision involving a change from a more literal towards a freer
rendering, and thus reflecting target circumstances rather than those at the
source end of the process. For instance, if the ST segment (Schengen) member
was initially translated literally as ‘članica’ [member] and then changed into
‘potpisnica sporazuma’ [agreement signatory], this was counted as a T-move.
2. S-move, on the other hand, was considered to have occurred if a more literal
solution followed a freer one, as in the following example: the ST segment
opened its borders was first translated as ‘otvorila granice’ [opened borders]
and later revised to read ‘otvorila svoje granice’ [opened its borders].
3. Neutral revision. In addition to the two ‘moves’ — away from and towards
source — we also encountered a considerable number of revisions which did
not involve a change in ‘distance,’ that is, which were neither closer nor more
removed from the source. A neutral revision was, for example, the change
from ‘politika o vizama’ [policy about visas] to ‘politika za vize’ [policy for
visas] for the ST segment visa policy.
Finally, a quantitative analysis was performed in which total values, mean values
and percentages were calculated. It needs to be pointed out that some revisions
were difficult to classify as a particular type. In such cases we ascribed 0.5 to both
types the revision could be said to belong to. We compared the two groups of sub-
jects — ‘the translators’ and ‘the interpreters’ — in terms of the number and type
of self-revisions made a. in the whole process, b. as a first revision of an initial solu-
tion, and c. self-revisions in the post-drafting phase of the process. The latter phase
is defined as beginning after the last sentence of the draft version of the translation
is finished and lasting until the end of the process (see Mossop 2001/2007: 167).
Since the original literal translation hypothesis could be interpreted in different
ways, we analyzed the first revisions and the revisions in the post-drafting sepa-
rately in order to see if there are striking differences in distance dynamics between
those values and the values for the whole process. The following examples from
our transcripts may illustrate the different categories:
Example 1:
The ST segment a common visa policy was initially translated as ‘zajednička poli-
tika za vize’ [common policy for visas]. At a later point in the drafting process, this
initial solution was changed into ‘zajednička politika u području viza’ [common
policy in the area of visas]. While still in the process of drafting the translation
(i.e., before the whole ST was translated), the translator was presumably not happy
with the solution resulting from the first revision and changed it into ‘zajednička
politika viznog režima’ [common visa regime policy]. In our analysis this solu-
tion counted as the second revision. It produced the solution that was kept in the
final version of the TT, which means that this particular example involved no self-
revisions in the post-drafting phase.
Example 2:
The ST contained the segment the Schengen rules, which produced ‘pravilima o
Schengenu’ [rules about Schengen] in one of the transcripts. The translator appar-
ently felt this initial solution should be revised, and substituted it with ‘pravilima
o schengenskom sustavu’ [rules about the Schengen system]. Deciding to apply
a different orthographic rule, the translator then changed the spelling and used
a capital initial S- in the adjective ‘schengenskom.’ The second revision thus pro-
duced ‘pravilima o Schengenskom sustavu’ [rules about the Schengen system],
which was kept until the end of the drafting process. Once the whole ST was trans-
lated, the subject started reading it again, introducing some new revisions. Among
those self-revisions in the post-drafting phase were two further changes of the
segment taken up here, i.e., the third and the fourth revision of the initial solu-
tion. While the third revision involved a return to the initial solution (‘pravilima o
Schengenu’), the fourth read ‘schengenskim pravilima’ [the Schengen rules] (note
that this phrasing is closer to the ST segment than any earlier solution). As can be
gathered from this description, Example 2 provided solutions relevant for all three
stages of our analysis reported here: (a) four self-revisions with regard to the whole
process, (b) a first revision of an initial solution, and (c) two self-revisions in the
post-drafting phase.
4. Findings
4.1 Number of self-revisions
The specific hypothesis formulated in this study states that substantial experi-
ence in interpreting affects interpreters’ self-revision when they undertake writ-
ten translation. The first subhypothesis focuses on the number of self-revisions:
since interpreters are normally required to voice nothing but an acceptable TL

solution and to refrain from revising it unless it is absolutely necessary, they could
be expected to produce a satisfying written TL solution sooner than translators.
In other words, there should be fewer self-revisions in their translation processes
than in those of translators. As the figures in Table 1 indicate, this was confirmed
in our study: the interpreter group produced a total of 107 self-revisions, 34 fewer
than the translator group. The average number of interpreter self-revisions was
thus 17.83, compared to the average of 23.5 in each translator’s process. Since there
is only a very slight difference in the number of problems that motivated self-
revision in interpreters’ and in translators’ processes (14.17 vs. 14.67), the figures
suggest that the number of revisions per problem was lower in the interpreter than
in the translator group. Interestingly, this had no impact on the time that the two
groups took to produce the final version of the translation: the difference in the
average duration of the task between the interpreter and the translator group was
only 3 seconds (13:37 vs. 13:40).
Table 1. Number of self-revisions, interpreters vs. translators

Interpreters Translators
Task duration (mean) 13:37 13:40
Number of problems (mean) 14.17 14.67
Number of self-revisions (mean) 17.83 23.5
Number of self-revisions (total) 107 141
4.2 Distance dynamics
As it was pointed out earlier (see Section 1), the various formulations of literal
translation hypothesis in the literature do not state clearly to what part of the
self-revision process the hypothesis applies. With no specific phases mentioned,
it seems justified to take it as applying to the process as a whole. We therefore
counted the overall number of the three different types of self-revisions in the
interpreters’ and in the translators’ process. In addition, allowing for different in-
terpretations, we also analyzed first self-revisions in their own right, as well as
self-revisions done in the post-drafting phase.
4.2.1 Distance dynamics over the whole self-revision process

All three types of self-revisions occurred in the processes of all our subjects. Before
contrasting the two groups, we looked at all their processes together, in order to
check our general hypothesis (the one applying to translation as such, formulated
in line with the generally held view).
As the general hypothesis would suggest, T-moves (deliteralizing ones) were

the most numerous type. However, the prevalence was not very pronounced. They
accounted for 39.47% of all revisions, along with 33.77% distance neutral ones and
26.75% revisions involving S-moves (Figure 1).
45%
39.47%
40%
35% 33.77%
30%
26.75%
25% T-moves
S-moves
20% Neutral revisions
15%
10%
5%
0%
Figure 1. Three types of self-revision, whole process, all subjects
The ordering of the three types of self-revisions changed when the two groups
of subjects were studied separately. In the translator group, T-moves (40.87%)
were still predominant, but S-moves proved more common than neutral revisions
(34.13% vs. 25%). In the interpreter group, however, T-moves were less common
than distance neutral revisions (37.74% T-moves, 44.61% neutral changes, while
S-moves accounted for only 17.65% of all revisions) (Figure 2).
50%
44.61%
45%
40.87%
40% 37.74%
35% 34.13%
30%
25% T-moves
25% S-moves
17.65% Neutral revisions
20%
15%
10%
5%
0%
Figure 2. Three types of self-revisions, whole process, interpreters vs. translators
The fairly high values for neutral self-revisions indicate that this type must not
be neglected in studies of revision processes. However, since such revisions result
in TL solutions which, compared to the solutions they have replaced, are neither
closer nor more distant from their respective source elements, they provide no
information on the direction of distance change. What is more, they can make the
comparison between the two groups of distance changing revisions less clear. We
therefore decided to also look at the number of T- and S-moves alone, excluding
neutral revisions (Figure 3).
38.5(68%) 18(32%)
Interpreters
T-moves
S-moves
51.5(54.50%) 43(45.50%)
Translators
0 20 40 60 80 100
Figure 3. Number of T- and S- moves only, whole process, interpreters vs. translators
With regard to our general hypothesis, it can be noted that T-moves were indeed
more numerous than S-moves with both groups of subjects. However, with the
45.5% share of S- moves in the translator group, the difference between the two
types is not very convincing. The interpreter group is more in line with the general
hypothesis, even if it also has a fair share of S-moves (almost one third of all dis-
tance changing revisions).
Figures 2 and 3 point at another interesting finding at this stage of our study.
In contrast to our specific subhypothesis (c) (stating that interpreters’ translation
processes, when compared to those of translators, involve a relatively smaller share
of changes from a more literal to a freer solution), the proportion of T-moves
turned out to be larger in the interpreter group (68.14% interpreters vs. 54.5%
translators).
4.2.2 Change of distance in first self-revisions only

Our study of the whole self-revision process thus did not provide conclusive evi-
dence for the literal translation hypothesis. We wanted to test whether that could
be due to the interpretation of the hypothesis with respect to the part of the pro-
cess to which it applies. A likely alternative interpretation seemed to be that the
hypothesis should apply to first self-revisions only, rather than to the whole pro-
cess. One might claim that the self-revision process becomes more complicated as
it develops, while first revisions are most directly influenced by the nature of the
initial solutions that they replace. If literal translations are default solutions, the
first self-revision will then have to involve a T-move, which means that the share
of such moves among first revisions would have to be very high.
As can be seen in Figures 4–6, the T-move values are indeed somewhat higher
for first revisions than they were for the whole process: for the combined group
of all the 12 subjects (Figure 4), the share has risen from 39.47% to 42.45%; in
the interpreter group from 37.74% to 43.21% (Figure 5) and from 68% to 71%
with neutral revisions excluded (Figure 6); in the translator group from 40.87%
to 41.67% (Figure 5) and from 54.5% to 59% with neutral revisions excluded
(Figure 6). However, it is just as obvious that the increase is not particularly high
in any of these comparisons.
45% 42.45%
40%
35% 33.96%
30%
25% 23.58% T-moves
S-moves
15%
10%
5%
0%
Figure 4. Three types of self-revisions, first revisions only, all subjects
50%
45% 43.21%
41.67%
40% 38.89%
35%
30% 29.49% 28.85%
T-moves
25% S-moves
20% 17.90% Neutral revisions
15%
10%
5%
0%
Figure 5. Three types of self-revisions, first revisions only, interpreters vs. translators
The picture for first revisions differs from that for the whole process in two re-
spects. One is the order of the three types of changes in the interpreter group:
neutral revisions are here less numerous than T-moves (although they still account
for 38.89%, i.e., more than a third of all revisions). The other interesting difference
has to do with the comparison of the interpreter and the translator group with re-
gard to the number and the share of T-moves among first revisions: contrary to our
specific subhypotheses, interpreters made a larger number of deliteralizing first
revisions than translators (both in absolute figures and in terms of their share).
35(71%) 14.5(29%)
Interpreters
T-moves
S-moves
32.5(59%) 23(41%)
Translators
0 10 20 30 40 50 60
Figure 6. Number of T- and S- moves only, first revisions, interpreters vs. translators
With regard to our general hypothesis, the predominance of T-moves over S-moves
is manifest in all three groups (interpreters and translators combined, and the two
groups taken separately). However, the question still remains how predominant
they would have to be to consider the description of the self-revision process as
moving from more literal towards freer renderings unequivocally substantiated.
For, even among the first revisions, T-moves never account for more than half of
the revisions done if all three types of changes are considered (the values range
from 41.67% to 43.21%). The predominance seems more convincing if only dis-
tance changing moves are considered, but even then the relatively high share of
S-moves would have to be acknowledged (approximately one third of all distance
changing moves: 29% and 41% for the interpreter and the translator group respec-
tively (Figure 6), or 35.71% for the combined group of all twelve subjects).
4.2.3 Distance dynamics in the post-drafting phase

The post-drafting phase involves self-revisions introduced after the whole text has
been translated. With a complete draft translation on hand, translators may be
expected to concentrate more on the target circumstances and pay less attention to
the ST, which was the primary influence during the drafting phase. Such translator
behavior during the post-drafting phase would have to be reflected in a very pro-
nounced dominance of T-moves over S-moves. We therefore decided to establish
the values for each self-revision type realized by our subjects in that phase of the
translation process, and thereby check whether the post-drafting phase would cor-
roborate our general hypothesis more conclusively than the former two situations
(the whole process and first revisions only).
As it turned out, just the opposite was true in our study: the values for T-moves
in the post-drafting phase were lower than in the two situations described ear-
lier (4.2.1 and 4.2.2). For example, for the combined translator-and-interpreter
group, the T-move value was 24.32% in the post-drafting phase, which compares
to 39.47% for the whole process and 42.45% for first self-revisions. As can be seen
in Figures 7 and 8, T-moves accounted for less than one third of all revisions in the
combined and the interpreter group (24.32% and 15.91%) and only slightly more
in the translator group (36.67%).
60% 56.76%
50%
40%
T-moves
30% S-moves
20% 18.92%
10%
0%
Figure 7. Three types of self-revision, post-drafting phase, all subjects
80%
72.73%
70%
60%
50%
T-moves
40% 36.67% S-moves
33.33%
30%
20% 15.91%
11.36%
10%
0%
Figure 8. Three types of self-revision, post-drafting phase, interpreters vs. translators
The revision type whose share rose most noticeably in the post-drafting phase,
in parallel to the pronounced drop in the share of T-moves, were neutral revi-
sions. The rise was quite dramatic in the interpreter group (72.73%, or 16 out of
the post-drafting total of 22) but there was a certain rise in the translator group
as well (33.33% vs. 25% for the whole process and 28.85% for first revisions). This
has once again confirmed the significance of distance neutral moves for any self-
revision typology.
As regards the ratio between the two distance changing types of revisions in
the post-drafting phase, T-moves were once again more numerous than S-moves,
but their dominance could hardly be considered substantial: 58% : 42% in the
interpreter group and 55% : 45% in the translator group (Figure 9). It is worth not-
ing, though, that the absolute figures in this case are very low (6 and 10 distance
changing revisions in the interpreter and the translator group respectively), which
challenges the relevance of the percentages quoted here.
3.5(58%) 2.5(42%)
Interpreters
T-moves
S-moves
5.5(55%) 4.5(45%)
Translators
0 2 4 6 8 10 12
Figure 9. T- and S- moves only, post-drafting phase, interpreters vs. translators
5. Conclusions with regard to the hypotheses
Inspired by various formulations in literature of the so-called literal translation hy-

pothesis, this study proposed a general hypothesis stating that “in their search for
the most suitable translation solution, translators move from more literal towards
freer renderings.” In addition to testing the general hypothesis, the study set out to
check whether substantial interpreting experience influences distance dynamics
patterns. The specific hypothesis governing that part of the study stated that sub-
stantial interpreting experience does have an effect, and the three sub-hypotheses
expressed three parameters manifesting that effect: (a) a smaller overall number of
self-revisions in interpreters’ than in translators’ processes; (b) a smaller number
of manifestations of the literal translation hypothesis in interpreters’ processes; (c)
a smaller share of such deliteralizing changes in the total number of self-revisions.
Based on our findings, we can conclude that the hypotheses have been partially
confirmed.
5.1 Conclusions with regard to the general hypothesis
Our general hypothesis applies to the translation process as such, and it emphasiz-
es the significance of deliteralizing changes (‘T-moves’ in our study) in the process
of self-revision, claiming that they are essential to the process. As we pointed out
in the introductory section of this chapter, there are a number of unclear points in
various formulations of the literal translation hypothesis in literature. The fact that
it can be interpreted in different ways has led us to refrain from making only one
conclusion with regard to its truthfulness.
We base our conclusions on the values established in our study for three differ-
ent types of self-revisions undertaken by our subjects. Since we analyzed different
parts of the process, we obtained several values for each revision type. To reflect
that finding, the share of each of the three types in the overall number of self-
revisions is expressed as a range, with the lowest and the highest share established
as its endpoints (Table 2). The values included here refer to the largest group in our
analysis (the combined group, interpreters and translators taken together), which
we deem most appropriate in the context of the general hypothesis. As has been
done throughout this chapter, two sets of values are presented: the values in the
first row show the shares with all three self-revision types taken into account; the
values in the second row indicate the shares between T-moves and S-moves only,
with distance neutral revisions excluded from the calculation:
Table 2. Share of each self-revision type, all subjects

T-moves S-moves Neutral
self-revisions
Including neutral self-revisions 24–42% 19–27% 34–57%
Excluding neutral self-revisions 56–64% 36–44%
As Table 2 indicates, moves from a more literal towards a less literal translation
solution (i.e., T-moves) always account for more than half of all self-revisions
(56–64%) when only two revision types are studied, and for between 24% and
42% when neutral changes are also included (the arithmetic mean of the range
being exactly 33%). Given that fact, we feel that the figures substantiate the view
that cognitive processing in translation involves a fair number of literal solutions,
as well as the expectation that a considerable portion of self-revisions involve del-
iteralization.
On the other hand, the other two revision types also appear regularly in our
study, in the processes of all subjects and regardless of the part of the process ana-
lyzed. The values for S-moves (36–44% when neutral revisions are excluded, and
19–27% when they are included) and for neutral changes (34–57%) are too high to
warrant the interpretation of our general hypothesis as implying that literal trans-
lations are default solutions in actual text production. The high share of neutral
changes (higher than that of T-moves!) prompts us to draw yet another conclu-
sion, namely that descriptions of distance dynamics cannot be limited to T- and
S-moves only but they have to take into account changes that are neutral with
respect to the distance between the original segment and a particular translation
solution.
It was also pointed out in the Introduction that formulations of the literal
translation hypothesis in the literature do not normally limit the group of transla-
tors to which it applies. The hypothesis is therefore taken to apply ‘in general,’ and
an implication for empirical studies is that it should be tested with different groups
of subjects, and that it is not essential to keep subjects with different characteristics
strictly apart. We therefore deem it appropriate to refer at this point to the results
of an earlier study (Antunović and Pavlović 2012). At that stage we tested the same
general hypothesis by analyzing twenty logs produced by ten language and transla-
tion students, each of them working from two different source languages.
As can be seen in Figure 10, the shares of T- and S-moves in the students’ self-
revisions (excluding neutral changes) correspond fully to the values established
in this study, and corroborate the above conclusions regarding the two different
interpretations of the literal translation hypothesis just mentioned.
80%
70% 68.14%
65.49%
60% 54.50%
50% 45.50%
40% T-moves
34.51% S-moves
31.86%
30%
20%
10%
0%
Interpreters Translators Students
Figure 10. Shares of T- and S- moves over the whole process (combined data from two
studies)
Students’ self-revisions also involved a large number of distance neutral changes;

indeed, such changes were the most frequent revision type in the students’ logs,
with a share of 38.23% in the 633 self-revision total (Antunović and Pavlović 2012).
This corroborates our conclusion that the significance of distance neutral changes
should be acknowledged in the context of self-revision process, in particular when

studying distance dynamics.
If we combine the results of our two studies, we get values describing the share
of each of the three self-revision types in the processes of 22 subjects producing
32 translations of two language versions of the ‘same’ text (22 translations from
English and 10 translations from Swedish). These values (Figure 11) bear out all
the conclusions reached in the present study with regard to the general hypothesis.
45%
40.80%
40%
34.33%
35%
30%
24.88%
25% T-moves
S-moves
15%
10%
5%
0%
Figure 11. Shares of the three self-revision types, all subjects (combined data from two
studies)
5.2 Conclusions with regard to the specific hypothesis
The study discussed in this chapter compares self-revision processes of profes-

sional translators with those of professional interpreters, in an attempt to test the
specific hypothesis holding that substantial interpreting experience affects dis-
tance dynamics. Three subhypotheses (a)–(c) were tested, each of them describing
one expected effect.
Subhypothesis (a), stating that the translation process of interpreters, when
compared to those of translators, involve fewer self-revisions on the whole, has
been confirmed in this study (see Table 1). This conclusion about the process as a
whole is also true of first revisions alone and of self-revisions in the post-drafting
phase.
Another expected effect of interpreting experience was that there would be
fewer T-moves among interpreters’ self-revisions (subhypothesis b.). The subhy-
pothesis has been confirmed if interpreted as we do, i.e., as applying to the process
as a whole (38.5 T-moves by interpreters vs. 51.5 T-moves by translators). It could
not be considered confirmed if applied to first revisions only since the interpreters’
figure is in that case higher (35:32.5).
According to our subhypothesis (c), there is a less pronounced predominance

of T-moves over S-moves in interpreters’ processes (that is, the share of T-moves
in the total number of distance changing self-revisions is smaller with interpreters
than with translators). This subhypothesis has not been confirmed. As a matter of
fact, just the opposite was true in all the three situations analyzed: the share of inter-
preters’ T-moves vs. translators’ T-moves was 68% vs. 55.5 % for the whole process,
71% vs. 59% among first revisions, and 58% vs. 55% in the post-drafting phase.
6. Methodological concerns and future research
6.1 Lack of routine in the interpreter group
As we saw in 4.2.1, the interpreter group proved to have made a strikingly large
number of distance neutral revisions (44.6% vs. 25% in the translator group),
which may be interpreted as suggesting a lack of routine in solving some problems
typical of written translation (e.g., spelling or punctuation). Hence our first meth-
odological concern: to what extent is the assumed activation of the interpreters’
ability to immediately produce acceptable TL solutions hampered by their lack
of routine in the written translation mode? The challenges of comparing modali-
ties are well known (e.g., Shlesinger and Malkiel 2005); in this case a solution for
future studies might be to use a group of subjects who are equally skilled in both
modalities, and to compare their written tasks with those of the control group
(translators without any interpreting experience).
The large number of neutral moves in our study could also be interpreted as
being TL specific, considering the existence of several orthography manuals in
Croatia, with some mutually exclusive spelling and punctuation rules that often
make native speakers feel insecure about their writing. It seems plausible that in-
terpreters, who are not likely to write as much on a daily basis as translators, are
more insecure in this respect.
6.2 Small samples
Another limitation of this study is the small sample (12 subjects), a characteristic
it shares with most process oriented translation research. One or two subjects who
get stuck on a translation problem and make an exceptionally large number of
revisions can skew the results, as can be seen from Table 3. The first row shows
the average distribution of T-moves and S-moves for the whole translator group
(6 subjects), and the second the average figures without the two subjects who were
markedly different from the rest. It can be seen that the figures for the latter group
are much closer to those obtained for the interpreters (and, incidentally, to those
obtained for the students, see 5.1). Statistical measures such as the trimmed mean
are not worth performing on small samples, but that situation would change if the
experiments were replicated to involve more subjects. Future studies should thus
present a more convincing picture of distance dynamics, in particular if the find-
ings corroborate those obtained so far.
Table 3. Limitation of small samples

T-moves S-moves
Translators (6 subjects) 54.50% 45.50%
Translators (6–2 = 4 subjects) 66.35% 33.65%
Interpreters (6 subjects) 68 % 32%
6.3 Terminological inconsistency
In the existing literature dealing with what we have referred to as the literal trans-
lation hypothesis, a plethora of terms are used (literal solution, formal corre-
spondent, contrastive correspondent, default solution, cognate, reflex translation,
unthinking translation, direct transfer, etc.), and the implications of differences
among those terms — and, more importantly, among the concepts behind them
— are insufficiently discussed. In our research, we adopted a very broad definition
and considered as ‘literal solutions’ those solutions that are obviously source-in-
spired. Future studies should pay closer attention to this issue and perhaps differ-
entiate among the concepts more clearly.
6.4 Other possible variables
In the present study all subjects translated from their L2 (B-language) into their L1
(A-language). The students from our previous research, mentioned above in 5.1,
worked from their L2 and L3 into L1, and yielded similar results regarding dis-
tance dynamics. Future studies could further test directionality as a variable, e.g.,
by introducing translation from L1 into L2, which is a regular practice in many
settings. Language pairs might prove a relevant variable, as formal closeness of the
two languages could play an important role in the deliteralization process (see also
the comment in 6.1 regarding the Croatian language, indicating how specificities
of the TL may influence self-revision and distance dynamics). Another variable
that deserves further attention is the relationship between distance dynamics and
individual translating style. Finally, future studies could use different text types or
translation tasks, as the requirement of more literal vs. freer translation can also be
expected to influence distance dynamics.
References
Antunović, Goranka, and Nataša Pavlović. 2011. “Moving on, Moving back or Changing it here
and now: Self-Revision in Student Translation Processes from L2 and L3.” Across Languages
Antunović, Goranka, and Nataša Pavlović. 2012. “Dinamika otklona od izvornika tijekom iz-
rade prijevoda [Distance dynamics in the translation process].” In Aktualna istraživanja u
primijenjenoj lingvistici. Zbornik radova s 25. međunarodnog skupa HDPL-a održanog 12.-
14. svibnja 2011. u Osijeku [Current research in applied linguistics. Proceedings of the 25th
International Conference of the Croatian Applied Linguistics Society], ed. by Leonard Pon,
Vladimir Karabalić, and Sanja Cimer, 199–214. Osijek: HDPL.
Chesterman, Andrew. 2011. “Reflections on the Literal Translation Hypothesis.” In Methods and
Strategies of Process Research: Integrative Approaches in Translation Studies, ed. by Cecilia
Alvstad, Adelina Hild, and Elisabet Tiselius, 23–35. Amsterdam: John Benjamins.
DOI: 10.1075/btl.94.05che
Dejean Le Féal, Karla. 1993. “Pédagogie raisonnée de la traduction.” Meta 38 (2): 155–197.
DOI: 10.7202/003451ar
Gile, Daniel. 1990. “Scientific Research vs Personal Theories in the Investigation of Interpretation.”
In Aspects of Applied and Experimental Research on Conference Interpretation, ed. by Laura
Gran, and Christopher Taylor, 28–41. Udine: Campanotto Editore.
Ivir, Vladimir. 1981. “Formal Correspondence vs. Translation Equivalence Revisited.” Poetics
Today 2 (4): 51–59. DOI: 10.2307/1772485
Ivir, Vladimir. 1988. “Translation and Backtranslation.” In Yugoslav General Linguistics
(Linguistic and Literary Studies in Eastern Europe), ed. by Milorad Radovanović, 131–143.
Amsterdam: John Benjamins. DOI: 10.1075/llsee.26.09ivi
Jakobsen, Arnt Lykke. 2006. “Research methods in translation: Translog.” In Computer
Keystroke Logging and Writing: Methods and Applications [Studies in Writing, 18], ed. by
Eva Lindgren, and Kirk P.H. Sullivan, 95–105. Oxford : Pergamon Press, 2006.
Jakobsen, Arnt Lykke, and Lasse Schou. 1999. “Translog Documentation.” In Probing the Process
in Translation: Methods and Results [Copenhagen Studies in Language, 24], ed. by Gyde
Hansen, 151–186. Copenhagen: Samfundslitteratur.
Jakobsen, Arnt Lykke, Kristian T.H. Jensen, and Inger M. Mees. 2007. “Comparing Modalities:
Idioms as a Case in Point.” In Interpreting Studies and Beyond, ed. by Franz Pöchhacker,
Arnt L. Jakobsen, and Inger M. Mees, 217–249. Copenhagen: Samfundslitteratur.
Malkiel, Brenda. 2009. “From Ántonia to My Ántonia: Tracking Self-Corrections with
Translog.” In Behind the Mind. Methods, Models and Results in Translation Process Research.
Copenhagen Studies in Language, ed. by Susanne Göpferich, Arnt L. Jakobsen, and Inger M.
Mees, vol. 37, 149–166. Copenhagen: Samfundslitteratur.
Mossop, Brian. 2001/2007. Revising and Editing for Translators. Manchester: St Jerome.
Shlesinger, Miriam, and Brenda Malkiel. 2005. “Comparing Modalities: Cognates as a Case in
point.” Across Languages and Cultures 6 (2): 173–193. DOI: 10.1556/Acr.6.2005.2.2
Tirkkonen-Condit, Sonja. 2005. “Evidence from Process Research.” Meta 50 (2): 405–414.
DOI: 10.7202/010990ar
DOI: 10.1075/btl.4
The impact of process protocol self-analysis
on errors in the translation product
Erik Angelone
This chapter presents the initial results of an exploratory study examining the
efficacy of Integrated Problem and Decision Reporting (IPDR) logs, recorded
verbalizations, and screen recordings when used by translation students as revi-
sion tools for purposes of recognizing problems and mitigating errors in their
translations. Students were given the task of creating one of these three process
protocol types in conjunction with nine translations. They were then asked to
self-reflect on their problem-solving performance while paying particularly close
attention to a series of concrete problem indicator types found in the respective
protocols during a post-task retrospective session. The students had the chance
to make any desired revisions to their translations as a result of analyses before
submitting a final version. The revised versions were marked up for punctuation,
spelling, lexical, syntactic, stylistic, and mistranslation errors. The frequency of
errors for each of these types was calculated for each student in conjunction with
the process protocol utilized. The total number of errors in conjunction with
each process protocol type used, regardless of textual level, was also calculated.
Screen recording consistently proved to be the most efficacious process protocol
type in mitigating errors. The chapter concludes by positing several possible ex-
planations for the greater success rate of screen recording as a revision tool and
calls for screen recording to be implemented as a core component of a process-
oriented translator training curriculum.
Keywords: process-oriented translator training, problem-solving, IPDR logs,

recorded verbalizations, screen recording
1. Introduction
Recent technological and methodological advancements in the arena of transla-

tion process research has paralleled a rekindled interest in process-oriented ap-
proaches to translator training. Learning activities and objectives are becoming
doi 10.1075/bct.77.07ang
106 Erik Angelone
increasingly oriented toward fostering the trainee’s cognizance of how one trans-
lates, as shaped by the behaviors and strategies that tend to occur during the stages
of comprehension, transfer, and production. This cognizance, in large part, im-
plies that learners have a conscious awareness of the problems they encounter
while translating as well as of viable problem-solving strategies that can be ex-
ecuted in their presence. Large-scale longitudinal research projects, such as the re-
cently-completed Capturing Translation Processes project at the Zurich University
of Applied Sciences (2009–2012), the TransComp project at the University of
Graz (2008–2011), and the Copenhagen Retrospection Project at the Copenhagen
Business School (2004), have not only shed light on problems and problem-solv-
ing, but have also established some preliminary frameworks as to how learner
awareness of both can be trained through a variety of analyzable protocols, such as
keystroke logs (Jakobsen 1999; Hansen 2006) and screen recordings (Pym 2009;
Massey and Ehrensberger-Dow 2011; Angelone 2013). These protocols have con-
tinued to make their way from the research lab to the translator-training environ-
ment, and students now have the opportunity to engage in critical self-analysis of
their own problem-solving performance to an unprecedented extent.
Although translator training still tends to be biased in the direction of the
translation end product in and of itself (Dam-Jensen and Heine 2009: 1), particu-
larly when it comes to addressing errors, process-oriented translator training holds
great potential for both students and trainers when attempting to discern what po-
tentially triggered these errors in the first place. Process protocols can be utilized
as diagnostic tools, with problem indicators contained therein marking areas of
interest for purposes of problem-solving analysis. While a plethora of research has
shed light on how such protocols document various cognitive processes, such as
the utilization of internal and external support (PACTE 2005; Alves and Liparini
Campos 2009) and the allocation of cognitive effort during planning, drafting,
and revision (Pym 2009), relatively little research has been done on how translator
trainees can utilize protocols documenting their own performance for purposes of
problem awareness training and as a potential means for mitigating errors.
This chapter reports on the initial findings of an exploratory study aimed at
examining potential correlations between students’ utilization of different pro-
cess protocol types for purposes of problem recognition and error detection,
namely Integrated Problem and Decision Reporting (IPDR) logs (Gile 1994;
2004), recorded verbalizations of their thought processes, and screen recordings
documenting all on-screen activity, and the types and frequencies of errors ap-
pearing in corresponding translation products post retrospection and revision.
On a broader scale, this study explores efficacy in problem recognition and error
mitigation in a comparative manner when students use these three protocol types
as revision tools. It is hoped that the results will ultimately have an impact on
The impact of process protocol self-analysis 107
shaping the process-oriented translator training curriculum, particularly in the

domains of process knowledge and metacognition (Shreve 2002; 2006), problem
recognition, and strategy execution/evaluation. Before presenting and discussing
obtained data, this chapter will provide a more detailed overview of what process-
oriented translator training sets out to accomplish insofar as problem awareness is
concerned and discuss some of the methods of choice in reaching these objectives.
2. Primary objectives of process-oriented translator training
Process-oriented translator training concerns itself with helping students become

more cognizant of how they translate through (primarily) retrospective self-reflec-
tion on the problems they encounter, or what Hansen calls sources of disturbance
(2006: 1), as well as the strategies they execute (or fail to execute) in their presence.
Two primary objectives are (1) fostering the student’s process awareness, or abil-
ity to define and describe problems and problem-solving in a relatively granular
manner, and (2) honing the student’s capacities for monitoring and assessing their
own performance. The first of these two objectives concerns itself with helping
students learn to pinpoint the nature of the problems they encounter in a relatively
delineated fashion, for example in terms of a concrete textual level (lexis, syntax,
style, etc.) or a given locus (comprehension, transfer, or production). Training in
this area is critical, particularly in light of the fact that student problem reporting
has been shown to be relatively weak in comparison with that of professionals
(Angelone 2010: 32). Even when students do in fact recognize that a problem has
occurred, they often struggle with determining the exact nature of the problem
and identifying which resources would be optimal in addressing it. Interestingly,
the preliminary results of this study, as described in greater detail below, seem
to suggest that students’ retrospective access to concrete protocols documenting
their performance facilitates error detection.
The second primary objective of process-oriented translator training involves
getting students to successfully apply declarative and procedural knowledge in a
targeted manner based on the task at hand. Declarative knowledge involves aware-
ness of one’s cognitive resources in relation to the task. For example, in the con-
text of problem-solving, the student might indicate knowing three potential target
language (TL) equivalents for a source language (SL) term, or not knowing if a
produced target text (TT) segment adheres to established TL genre conventions.
Procedural knowledge, on the other hand, involves knowledge of strategies and
ways to solve problems, for example, knowing how to scan TL parallel texts for
contextualizing and ultimately narrowing down three potential term equivalents
to one. Problem indicators, such as extended pauses, episodes of information
108 Erik Angelone
retrieval, and revision behavior, as encoded in various process protocol types,

serve as an ideal point of departure in raising student awareness of their declara-
tive and procedural knowledge capacities through retrospective self-reflection.
Problem indicators will be discussed in greater detail below as stimuli associated
with various process-oriented training methods.
A process-oriented approach also sets out to enhance the student’s metacogni-
tion, or strategic, conscious, and volitional control over complex cognitive tasks
(Shreve 2009). In translation, these complex cognitive tasks include recogniz-
ing problems, proposing solutions, and then evaluating solutions at the stages of
comprehension, transfer, and production (Angelone 2010; Angelone and Shreve
2011). As already mentioned, extended pauses (in on-screen activity, strings
of recorded verbalizations, etc.) during task progression may potentially signal
problems, and having students focus on them when analyzing protocols of their
own performance reinforces problem recognition and potentially error detection.
Interestingly, having students do so may result in problems being recognized that
may have otherwise gone unnoticed without an opportunity for such retrospec-
tion. Problem-solving strategies generally follow the occurrence of problem indi-
cators and can take the form of such observable indicators as information retrieval
and revision behavior. In a training scenario, students could be asked to reflect on
how problems, as represented through indicators, were addressed, as represented
through subsequent solution proposal and solution evaluation strategies. In the
event that problems resulted in errors, students could be asked to take a reverse
engineering approach in which they trace possible causes of the errors from the
perspective of faulty problem-solving strategies. Angelone and Shreve refer to the
translator’s relatively linear progression through problem recognition, solution
proposal, and solution evaluation as metacognitive bundling, and have found that
professionals often engage in such bundling as a deliberate procedural strategy
(2011: 111).
Ultimately, in light of the fact that student problem reporting has been shown
to be relatively weak, and in light of the fact that metacognitive bundling and
problem-solving in general depend on problems being discerned in the first place,
training problem recognition should be regarded as a top priority in process-ori-
ented translator training. This chapter will now describe a series of self-reflection
methods geared towards fostering problem recognition.
3. Methods for training problem recognition
3.1 IPDR logs
One of the longest-standing methods for fostering problem awareness through

self-reflection involves having students create IPDR logs, as conceptualized by
Gile (2004). Over the course of the translation task, students are asked to keep a
running log of (1) the problems they encounter and the nature of these problems
(problem description), (2) the strategies they utilize in addressing these problems
(problem-solving), (3) the rationales for doing so. Table 1 depicts a sample IPDR
log excerpt (translation direction German–English).
Table 1. An Integrated Problem and Decision Reporting log excerpt

Source text Problem description Problem-solving strategy Solution
Leberkäs Lexical and conceptual Since a one-to-one equivalent Leberkäs, a kind
gap. doesn’t exist, a description is of lunchmeat,
An equivalent product needed.
doesn’t exist in the TL/TC. Borrowing in conjunction with
explicitation.
Mit Google auf German noun construc- Transposition to a verb construc- Chasing votes
Stimmenjagd tion sounds awkward if tion, more specifically denominal- with google.
kept as a noun in English. ization, was used after consulting
parallel texts.
IPDR logs require the student to temporarily break away from the task at hand
for documentation purposes. My students reported occasionally losing their train
of thought as a result of this necessary disruption in natural translation flow, yet
were happy to have a concrete, personalized point of departure when reflecting on
their problems and problem-solving post task completion. Ultimately, the prom-
ise IPDR logs hold as a method for training problem awareness is dependent on
the student’s efficacy in recognizing and documenting problems in the first place.
Research has shown that problems in translation often go unnoticed by students
(Angelone 2010), implying that the logs they construct potentially run the risk of
being incomplete (Göpferich and Jääskeläinen 2009: 172) in terms of both overall
scope and the way in which problems and corresponding problem-solving strate-
gies are described. Interestingly, while incompleteness was in fact occasionally an
issue in the logs produced by my students, many of them expressed uncertainty
as to how much was too much when it came to documenting content in a log.
They mentioned feeling bogged down by the need to document “each and every
little problem.” Nevertheless, when it came time for them to self-reflect on the
110 Erik Angelone
problems they encountered, the data found on their IPDR logs were personalized,
learner-centered, and relatively straightforward.
3.2 Audio recordings of verbalized thought processes
A second process-oriented training method involves having students self-reflect on

the thought processes in which they engaged while translating, as captured and doc-
umented in the form of an audio recording. The “data” that the students ultimately
work with can be regarded as a type of think-aloud protocol, albeit likely much less
complex than a fully marked-up TAP as typically found in the translation process
research arena. In generating such verbal reports, students are asked to articulate
everything that goes through their minds while translating, including the problems
they encounter and subsequent problem-solving decisions. Problem awareness
ensues through strategy observation (here, listening to verbalizations of thoughts)
during a retrospective session after the translation task has been completed.
In recorded verbalizations, problem indicators take the form of such observ-
able phenomena as direct articulation (“I don’t know what this term is in the
target language”), indirect articulation (“hm… [student articulates term posing
problems]…hm”), or a series of speech disfluencies (Shreve et al. 2011: 94–95)
including repeated articulation of the problematic text passage, stuttering, cut-
offs mid-utterance, use of filler words, and extended periods of silence. Insofar
as training is concerned, students would be encouraged to look for instances of
such problem indicators in the process protocols documenting their performance.
After recognizing problems through such indicators, students would then focus
on subsequent articulations in analyzing corresponding problem-solving behav-
ior. For example, was internal or external support (PACTE 2005) utilized? In the
case of external support, which resources were utilized for information retrieval
and were look-ups optimized in relation to the nature of the problem? Finally, the
most important questions of all: Did errors occur and are revisions needed?
The verdict is still out on whether or not such articulated thoughts are in fact
an accurate reflection of what was going through the translator’s mind while trans-
lating (Toury 1995: 235). This sense of doubt, in turn, stems from the question of
whether or not the two processes (articulating and translating) can be performed
simultaneously without an interference effect (Jääskeläinen 2000). When asked to
create process protocols of recorded verbalizations, my students, on the one hand,
felt that having to articulate thoughts “forced” them to focus on the task at hand
to a greater extent than they perhaps otherwise would have. On the other hand,
they were quick to point out that thinking aloud over an extended period of time
proved to be both mentally and physically exhausting. Upon closer examination of
their recorded verbalizations, a fatigue effect quickly becomes apparent in that the
degree of articulation slowly, but gradually drops off over the course of the transla-
tion task. In anticipation of this trend, the text lengths for all translation activities
in this study were deliberately kept relatively short (not exceeding 250 words) in
an attempt to mitigate fatigue.
3.3 Screen recording
Screen recording, one of the most recent methods used in translation process re-
search, is rapidly attracting interest within the field of process-oriented transla-
tor training as a user-friendly, ecologically valid tool for fostering the learner’s
self-awareness. For example, Pym (2009) had his students self-reflect on screen
recordings of their performance to foster greater awareness of planning, drafting,
and revision tendencies. Kujamäki (2010) had his students create unguided ret-
rospective commentaries to document their own performance as well as that of a
fellow student while watching respective screen recordings. The popularity of this
method as a revision tool is particularly evident at the Zurich Institute of Applied
Sciences, where all undergraduate and graduate students of translation have been
required to record some of their translation assignments using screen recording
(Massey and Ehrensberger-Dow 2011).
During a screen recording session, a software application unobtrusively re-
cords all on-screen activity that transpires over the course of the translation task.
Practically everything the translator does on the computer that is rendered on-
screen, from the CAT tools and online resources utilized for information retriev-
al, to the textual levels at which target text generation and revision occur, to the
pauses the occur while translating, is captured in video format and made readily
available for retrospective analysis in a highly-visual, naturalistic manner. Some of
the paramount problem indicators in screen recordings include extended pauses
in screen activity, information retrieval (using software or online), and revision
behavior.
My students, almost unanimously, stated that screen recording was their
preferred method (over IPDR logs and recorded verbalizations) when asked to
self-reflect on translation processes. They appreciated the naturalness of workflow
rendition and the fact that they did not have to deviate from the way they usually
translate for purposes of protocol compilation. Furthermore, my students felt it
was easiest to track their performance when watching screen recordings (as op-
posed to when analyzing IPDR logs or listening to recorded verbalizations) thanks
to the verisimilitude that the video playback offered, mirroring each step of the
process from beginning to end in a granular fashion.
112 Erik Angelone
3.4 Eye-tracking: A promising method on the horizon?
By no means do the methods described above represent an exhaustive list. Several

additional methods have the potential to eventually find their place in the process-
oriented training curriculum. Eye-tracking, for example, can provide the learner
with valuable visual attention data, that is, data outlining where one looks on the
screen while translating, in which sequence, and for how long. Process awareness
could hypothetically stem from self-analysis of saccade and fixation patterns, as
distributed within and across areas of the screen containing the source text and
the emerging target text. Although eye-tracking has not been extensively explored
as a method for direct application in translator training to date, several recent re-
search studies point towards a potential tie-in for fostering awareness of such phe-
nomena as segmenting tendencies and the textual level of translation (Dragsted
and Hansen 2008), distributions of comprehension (fixations on the source text),
transfer (saccades running from the source text to the target text and vice versa),
and production (fixations on the target text) (Sharmin et al. 2008), and CAT tool
utilization (O’Brien 2008). Perhaps the greatest obstacle impeding the integration
of eye-tracking as a process-oriented translator training tool is cost. At the time
of writing, an eye-tracker still costs tens of thousands of U.S. dollars, making this
tool much less feasible than some of the others discussed in this chapter. However,
Tobii, one of the leading manufacturers of eye-tracking technology, has released a
portable model, the X2–60, that can snap onto practically any computer monitor.1
Time will tell if this new, more affordable model can catalyze eye tracking-based
process-oriented translator training.
The rest of this chapter will discuss a recently-completed pilot study exam-
ining potential correlations between the students’ utilization of different process
protocol types for purposes of problem recognition and error detection during
a draft stage of translation and the frequencies and types of errors appearing in
corresponding translation products post reflection and revision. A brief overview
of the study’s research design and methods will be followed by a more granular
presentation of some of the pertinent results, along with a discussion of possible
explanations for the trends suggested by the data.
4. Research design and methods
The primary objective of this study is to document the relative efficacy of IPDR
logs, recorded verbalizations, and screen recordings as revision tools used by
1. For more information see http://www.tobii.com/en/eye-tracking-research/global/products/

students of translation for purposes of problem recognition and error mitigation.

These three methods were selected for inclusion based on the fact that students:
(1) would not incur any costs to use the respective technologies, (2) could use
their own computers in their own natural translation environments, and (3) could
make sense of the obtained data in a relatively straightforward manner requiring
minimal, if any, reliance on the instructor in the process.
4.1 Research questions
This study is based on the following interrelated research questions.

1. Do any relationships exist between self-reflection activity type used and certain
types of errors/ error frequencies in the corresponding translation products?
2. Do any relationships exist between self-reflection activity type used and over-
all error frequency in the corresponding translation products?
4.2 Participants
This study tracked the performance of six first-year M.A. students (German–
English Translation) at Kent State University over the course of a semester (Fall
2011) in a translation practice course. Five of the students were English L1 and one
was German L1. All of the students had had formal coursework in translation at
the undergraduate level. Prior to taking the course in which data were collected,
none of the six participants had ever taken part in IPDR, think-aloud, or screen re-
cording activities. Although this study was not set up to explore the potential of an
expertise effect, it is worth noting that two of the students had some professional
experience (< five years with 100% of income earned through translation activity
on a regular basis), one working as a project manager and the other as a freelance
translator. The other four students had no professional experience.
4.3 Materials and procedures
Over the course of the semester, the students were asked to create and then self-
reflect on one of the following three protocol types in conjunction with a cor-
responding translation task: (1) an IPDR translation log, (2) a recording of all
articulated thought processes while translating, and (3) a screen recording docu-
menting all screen activity while translating. All protocols were created on the stu-
dents’ own computers in their own natural working environments. IPDR logs were
created using a word processing application. Think-aloud protocols were created
using a pre-installed voice recording application in conjunction with a built-in or
114 Erik Angelone
external microphone. Screen recordings were created using Blueberry FlashBack

Express, a free software application.2
Each protocol type was explained in detail and its creation modeled in class
prior to the students creating their own. Students were then given the task of close-
ly analyzing the protocols they created for the presence of problem indicators, as
described above for each of the respective methods. After identifying potential
problem indicators, they were instructed to check on corresponding passages in
the translation products to see if errors occurred. Students had the opportunity to
make any desired revisions in light of their protocol self-analysis prior to submit-
ting a final version of the translation.
Each student created a total of nine process protocols, in conjunction with
nine translation tasks as described in Table 2.
Table 2. Translation tasks3

Week Text Topic Reflection Activity Text type classification3
1 Internet café ad IPDR translation log Primarily operative
2 Daycare center ad Think-aloud recording Primarily operative
3 Computer store flyer Screen recording Primarily operative
4 Dentist website IPDR translation log Balanced operative/informative
5 Veterinarian website Think-aloud recording Balanced operative/informative
6 Naturopathic website Screen recording Balanced operative/informative
7 Tennis club by-laws IPDR translation log Primarily informative
8 Youth hostel rules Think-aloud recording Primarily informative
9 Cell phone plans Screen recording Primarily informative
Each of the nine source texts was a self-contained, full-length text, approximately
250 words in length and of comparable difficulty. The translation tasks were not
timed, but students were asked to complete the translation task, corresponding
self-reflection activity, and any desired revisions in one sitting. Each text was care-
fully selected based on its exhibiting a balance of lexical, syntactic, and stylistic
Rich Points (PACTE 2011: 38) that were intuitively expected to pose challenges
in translation. Ultimately, for each self-reflection activity type, students translated
(1) a primarily operative text, (2) a balanced operative/informative text, and (3) a
primarily informative text. The self-reflection activity types were deliberately stag-
gered in such a manner so that students were using all three types at the beginning,
middle, and end of the semester. After making any desired revisions, students sub-
mitted their translations along with their respective process protocols.
2. http://www.bbsoftware.co.uk/BBFlashBack_FreePlayer.aspx
3. Classification based on the text typology model proposed by Reiss (1976).
4.4 Error classification
After the students submitted their materials, errors in the translation products
were codified in terms of type (punctuation, spelling, lexis, syntax, style, mis-
translation) and frequency of each type. In this study, lexical errors included such
phenomena as using incorrect terminology, false cognates, and incorrect or weak
collocations. Syntactic errors involved errors in word order, verb tense, incom-
plete or run-on sentences, and subject-verb agreement. Stylistic errors stemmed
from inappropriate register (too formal or too colloquial), lexical or grammatical
inconsistency (when the text would call for consistency), or lexical and grammati-
cal redundancy (when the text would call for variation). Finally, mistranslation
errors resulted from inappropriate additions or omissions and errors resulting in
incorrect meaning transfer. Furthermore, errors were tracked in the following two
manners: (1) by individual student, according to error type/frequency and activity
type used, and (2) collectively, according to error type/frequency and activity type
used.
5. Results and discussion
Figures 1–6 provide an overview of obtained results, followed by a brief discussion

and a series of potential explanations for these findings.
14
12
10
Frequency
8
IPDR Log
6 Think-aloud
Screen recording
4
0
Student 1 Student 2 Student 3 Student 4 Student 5 Student 6
Participants
Figure 1. Punctuation and spelling error frequency by self-reflection activity type
In Figure 1, the data show that four of the six students made the most punctu-
ation and spelling errors when using IPDR logs for purposes of self-reflection.
116 Erik Angelone
Upon closer examination of the submitted protocols, it quickly became evident

that punctuation and spelling problems were never documented in this process
protocol type. In other words, if students were relying on IPDR logs as a revision
tool for keeping errors of these two types at bay, it is plausible that the lack of docu-
mentation would further contribute to the errors going unnoticed. Instead, stu-
dent attention was likely drawn to problems encountered at larger textual levels, as
documented in their logs. Insofar as post-task problem recognition is concerned,
changes were likely made while translating, that is, concurrently with task comple-
tion, as opposed to retrospectively. Five of the six students made the fewest punc-
tuation and spelling errors when using screen recording for reflection purposes.
Perhaps the highly visual, relatively granular, “real-time” rendition of translation
as captured through screen recording is conducive in catching even the smallest
(from a textual unit sense) errors during retrospection.
14
12
10
Frequency
8 IPDR Log
Think-aloud
6 Screen recording
0
Participants
Figure 2. Lexical error frequency by self-reflection activity type
In Figure 2, as well, the data reveal that four of the six students made the most
lexical errors when using IPDR logs, perhaps further suggesting that this particu-
lar method is less helpful in catching errors at the level of smaller textual units.
As it turns out, the vast majority of problems documented in IPDR logs by most
of the students had to do with syntactic and stylistic issues. Comments pertinent
to terminology issues appeared early in their logs, but then tapered off. On many
occasions, my students would ask exactly how much content should be included
in an IPDR log (“Do I really need to document each and every word that caused
problems”?). It almost seems as if students preferred documenting “bigger issues”
when entering content rather than entering each and every problematic lexical
item. Since there is no other point of departure besides the self-generated con-
tent found in a log, it is quite likely that undocumented lexical problems went
unnoticed in the product during retrospection and revision and, hence, remained
as errors.
The majority of students made the fewest number of lexical errors when us-
ing screen recording. In light of this method’s efficacy in mitigating punctuation,
spelling, and lexical errors, it seems as if screen recording engages a visual process-
ing system for relatively granular problem recognition, successfully guiding atten-
tion to detail at the level of such smaller textual units.
14
12
10
Frequency
8 IPDR Log
Think-aloud
6 Screen recording
0
Participants
Figure 3. Syntactic error frequency by self-reflection activity type
Noteworthy in Figure 3 is the relative consistency in syntactic error frequencies

across all three self-reflection activity types. Overall, student errors of this type
were low in frequency. Syntax is the only textual level in this study where the uti-
lization of screen recording as a revision tool did not consistently result in fewer
errors in the translation product. Perhaps this finding points towards the tendency
for syntactic problems to be somehow more salient than problems at other textual
levels, to the extent that there seems to be a natural inclination to successfully de-
tect problems of this type regardless of process protocol type used. The idea that
syntax error recognition routines in translation might be well-established neural
networks as a result of reading tendencies certainly warrants further empirical
investigation. It is interesting to note that screen recording did prove to be more
efficacious in syntax error mitigation for student 5, as compared with the oth-
er students. Student 5 was the only student in this study who was translating L1
(German) into L2 (English). Perhaps the well-established syntactic error recogni-
tion routines mentioned above are weaker when working in one’s L2. With this
in mind, screen recording might ultimately prove to be particularly beneficial for
students translating into their non-native language when it comes to recognizing
problems and mitigating errors at this particular textual level.
118 Erik Angelone
14
12
10
Frequency
8 IPDR Log
Think-aloud
6 Screen recording
0
Participants
Figure 4. Stylistic error frequency by self-reflection activity type
At the level of style (see Figure 4), the students’ utilization of both IPDR logs
and recorded articulations tended to yield relatively high frequencies of errors.
Interestingly, stylistic problems tended to be well-documented in the IPDR logs, yet
were seldom articulated in the recorded verbalizations. In the case of the latter, it
is plausible that errors occurred in the product because stylistic problems were not
directly articulated in the context of the process protocol. When it came time for
self-reflection, perhaps the students were looking for such direct (or indirect) artic-
ulations pertaining to stylistic problems, and were less inclined to look for problem
indicators as found in the form of speech disfluencies or extended periods of silence.
The fact that stylistic errors occurred when IPDR logs were used in spite of
the fact that problems of this kind were well-documented is intriguing. A higher
frequency of documentation did not result in fewer errors. Perhaps the external,
self-documentation format inherent to problem reporting when creating IPDR
logs is less conducive to the mitigation of stylistic errors, regardless of degree of
documentation.
As far as the mitigation of stylistic errors is concerned, once again, the utiliza-
tion of screen recording during retrospective analysis proved to be most effica-
cious for most of the students. At this stage of data analysis, it starts to become
evident that screen recording was the most conducive of the three self-reflection
activity types at mitigating errors not only at the sub-sentential level, but also at
the supra-sentential level.
The Figure 5 data indicate that the translations of four of the six students con-
tained the most mistranslation errors when recorded verbalizations were used for
purposes of self-reflection. While listening to the recordings in conjunction with
errors documented in the corresponding products, it quickly became apparent that
14
12
10
Frequency
8 IPDR Log
Think-aloud
6 Screen recording
0
Participants
Figure 5. Mistranslation errors by self-reflection type
students articulated at a steadily declining rate as the task progressed, while mis-
translation errors appeared at an increasing rate. In other words, during retrospec-
tive analysis, students generally had nothing concrete to go on besides extended
periods of silence in trying to locate potential mistranslation problems/errors,
which may explain why errors of this type went largely unnoticed. The utilization of
screen recording yielded considerably fewer mistranslation errors for the majority
of the students, further warranting a closer examination of what potentially makes
this method particularly conducive vis-à-vis IPDR logs and recorded verbalizations.
An examination of error frequencies in aggregate for each of the five error
types documented in this study (see Figure 6) indicates that the utilization of
screen recording during retrospective self-reflection resulted in the fewest number
70
60
50
Spelling/Punc.
Frequency
40 Lexis
Syntax
30 Style
Mistrans.
20
10
0
IPDR Log Think-aloud Screen recording
Self-reflection activity type
Figure 6. Overall error frequency by self-reflection activity type
120 Erik Angelone
of errors across the board. At this stage, it is important to posit some further pos-
sible explanations for the efficacy of screen recording insofar as problem recog-
nition and error mitigation are concerned. This starts with a close examination
of what transpires during retrospection involving screen recording that does not
transpire in a similar fashion when using IPDR logs and recorded verbalizations.
First and foremost, it is possible that the guided (cued) eye movement that is
activated when analyzing a screen recording makes problem indicators (and er-
rors) embedded in the process protocol more salient. When working with screen
recordings, the students’ visual attention is inherently drawn to problems as they
unfold in real-time. As a result, even in the absence of concrete problem indicators
(in the case of screen recording, look-ups, extended pauses in screen activity, and
revision behavior, among others), this guided visual attention catalyzes a height-
ened state of cognitive awareness, to the extent that a highly effective problem
recognition monitor remains activated throughout the retrospection session. The
results of this study suggest that guided visual attention is a more optimal channel
for error mitigation than the the audio processing system activated when analyz-
ing recorded verbalizations and the more offline processing system activated when
analyzing IPDR log content.
There are additional possible explanations for the greater overall success of
screen recording as a revision tool above and beyond this inherent guided visual
cuing system. For example, screen recording is the only method of the three ex-
amined in this study that did not require some degree of parallel processing on
the student’s part during translation and concurrent process protocol generation.
The protocol was basically generated automatically by the software application
and students were not required to perform a simultaneous task while translating
(documenting problems and problem solving in the case of IPDR logs and articu-
lating thought processes in the case of recorded verbalizations). This may have
resulted in a lower cognitive load, and, in turn, less translation errors in the first
place. Finally, another possible explanation for screen recording’s success might be
that the problem indicators encoded in this particular process protocol type (look-
ups, extended pauses in screen activity, and revision behavior) are somehow more
overt in the sense that they draw greater attention. A more granular follow-up
analysis of problem indicator frequencies in the various protocol types in relation
to revisions made (or not made) to corresponding translation passages in their
presence will hopefully shed more light in this regard.
6. Conclusion and future directions
The data obtained in this study point towards screen recording as being the most
efficacious self-reflection activity type for purposes of error mitigation. Larger-
scale follow-up studies involving a higher number of participants and additional
language pairs are needed at this stage to further assess the efficacy of all three
self-reflection activity types (as well as others not included in this pilot study).
Although this study is primarily geared towards translation pedagogy, it might
also be interesting to explore the impact of various translator and text variables on
error patterns in conjunction with the three process protocol types. For example,
is screen recording equally effective as a revision tool for problem recognition and
error mitigation when used by professionals as opposed to students? What im-
pact does text length have on process protocol efficacy? How does directionality
(translating into L1 or L2) correlate with process protocol efficacy? In summary, it
is hoped that the findings obtained here will encourage translator trainers who are
dedicated to a process-oriented approach to consider integrating screen recording
as a fundamental tool to foster the learner’s problem recognition capacities, pro-
cess awareness, and metacognition in general.
References
Alves, Fabio, and Tania Liparini Campos. 2009. “Translation Technology in Time: Investigating
the Impact of Translation Memory Systems and Time Pressure on Types of Internal and
External Support.” In Behind the Mind: Methods, Models and Results in Translation Process
Research: Copenhagen Studies in Language 37, ed. by Susanne Göpferich, Arnt L. Jakobsen,
and Inger M. Mees, 191–218. Copenhagen: Samfundslitteratur.
Angelone, Erik. 2010. “Uncertainty, Uncertainty Management and Metacognitive Problem
Solving in the Translation Task.” In Translation and Cognition, ed. by Gregory M. Shreve,
and Erik Angelone, 17–40. Amsterdam: John Benjamins. DOI: 10.1075/ata.xv.03ang
Angelone, Erik. 2013. “Watching and Learning from ‘Virtual Professionals’: Utilising Screen
Recording in Process-oriented Translator Training.” In New Prospects and Perspectives
for Educating Language Mediators, ed. by Don Kiraly, Silvia Hansen-Schirra, and Karin
Maksymski, 139–155. Tübingen: Gunter Narr.
Angelone, Erik, and Gregory Shreve. 2011. “Uncertainty Management, Metacognitive Bundling
in Problem-Solving, and Translation Quality.” In Cognitive Explorations of Translation, ed.
by Sharon O’Brien, 108–130. London: Continuum.
Dam-Jensen, Helle, and Carmen Heine. 2009. “Process Research Methods and their Application
in the Didactics of Text Production and Translation: Shedding Light on the Use of Research
Methods in the University Classroom.” trans-kom 2 (1): 1–25. http://www.trans-kom.eu/
ihv_02_01_2009.html
Dragsted, Barbara, and Inge G. Hansen. 2008. “Comprehension and Production in Translation:
A Pilot Study on Segmentation and the Coordination of Reading and Writing Processes.”
122 Erik Angelone
In Looking at Eyes: Eye Tracking Studies of Reading and Translation Processes, Copenhagen
Studies of Language 36, ed. by Susanne Göpferich, Arnt L. Jakobsen, and Inger M. Mees,
9–29. Copenhagen: Samfundslitteratur.
Gile, Daniel. 1994. “The Process-Oriented Approach in Translation Training.” In Teaching
Translation and Interpreting 2, ed. by Cay Dollerup, and Annette Lindegaard, 107–112.
Amsterdam: John Benjamins. DOI: 10.1075/btl.5.17gil
Gile, Daniel. 2004. “Integrated Problem and Decision Reporting as a Translator Training Tool.”
JoSTrans 2: 2–20. http://www.jostrans.org/issue02/art_gile.php.
Göpferich, Susanne, and Riitta Jääskeläinen. 2009. “Process Research into the Development of
Translation Competence: Where are We, and Where Do We Need to Go?” Across Languages
Hansen, Gyde. 2006. “Retrospection Methods in Translator Training and Translation Research.”
JoSTrans 5: 2–41. http://www.jostrans.org/issue05/art_hansen.php.
Jakobsen, Arnt Lykke. 1999. “Logging Target Text Production with Translog.” In Probing the
Process in Translation, ed. by Gyde Hansen, 9–20. Copenhagen: Sammfundslitteratur.
Jääskeläinen, Riitta. 2000. “Focus on Methodology in Think-aloud Studies on Translating.” In
Tapping and Mapping the Processes of Translation and Interpreting, ed. by Sonja Tirkkonen-
Condit, and Riitta Jääskeläinen, 71–82. Amsterdam: John Benjamins.
DOI: 10.1075/btl.37.08jaa
Kujamäki, Pekka. 2010. “Auf der Suche nach treffenden Worten. Bildschirmvideos als Mittel zur
Analyse von studentischen Übersetzungsleistungen.” In Infinite Kontrastive Hypothesen.
Beiträge des Festsymposiums zum 60. Geburtstag von Irma Hyvärinen, ed. by Leena
Kolehmainen, Hartmut Lenk, and Annikki Liimatainen, 141–164. Frankfurt: Peter Lang.
Massey, Gary, and Maureen Ehrensberger-Dow. 2011. “Commenting on Translation:
Implications for Translator Training.” JoSTrans 16: 26–41. http://www.jostrans.org/issue16/
art_massey_ehrensberger_dow.php.
O’Brien, Sharon. 2008. “Processing Fuzzy Matches in Translation Memory Tools: An Eye-
Tracking Analysis.” In Looking at Eyes: Eye Tracking Studies of Reading and Translation
Processes, Copenhagen Studies of Language 36, ed. by Susanne Göpferich, Arnt L. Jakobsen,
and Inger M. Mees, 79–102. Copenhagen: Samfundslitteratur.
PACTE. 2005. “Investigating Translation Competence: Conceptual and Methodological Issues.”
Meta 50(2): 609–619. DOI: 10.7202/011004ar
PACTE. 2011. “PACTE Translation Competence Model: Translation Project and Dynamic
Translation Index.” In Cognitive Explorations of Translation, ed. by Sharon O’Brien, 30–56.
London: Continuum.
Pym, Anthony. 2009. “Using Process Studies in Translator Training: Self-discovery through
Lousy Experiments.” In Methodology, Technology and Innovation in Translation Process
Research: Copenhagen Studies in Language 38, ed. by Susanne Göpferich, Fabio Alves, and
Inger M. Mees, 135–156. Copenhagen: Samfundslitteratur.
Reiss, Katharina. 1976. Texttyp und Übersetzungsmethode: Der operative Text. Kronberg:
Skriptorverlag.
Sharmin, Selina, Oleg Spakov, Kari-Jouko Räihä, and Arnt L. Jakobsen. 2008. “Where on the
Screen do Translation Students Look While Translating, and for How Long.” In Looking
at Eyes: Eye Tracking Studies of Reading and Translation Processes, Copenhagen Studies
of Language 36, ed. by Susanne Göpferich, Arnt L. Jakobsen, and Inger M. Mees, 31–51.
Copenhagen: Samfundslitteratur.
Shreve, Gregory M. 2002. “Knowing Translation: Cognitive and Experiential Aspects of

Translation Expertise from the Perspective of Expertise Studies.” In Translation Studies:
Perspectives on an Emerging Discipline, ed. by Alessandra Riccardi, 150–171. Cambridge:
Cambridge University Press.
Shreve, Gregory M. 2006. “The Deliberate Practice: Translation and Expertise.” Journal of
Translation Studies 9 (1): 27–42.
Shreve, Gregory M. 2009. “Recipient-orientation and Metacognition in the Translation Process.”
In Translators and their Readers: In Homage to Eugene A. Nida, ed. by Radica Dimitriu, and
Miriam Shlesinger, 255–270. Brussels: Les Editions du Hazard.
Shreve, Gregory, Isabel Lacruz, and Erik Angelone. 2011. “Sight Translation and Speech
Disfluency: Performance Analysis as a Window to Cognition.” In Investigating the Process-
What it Tells Us About Translating and Interpreting, ed. by Cecilia Alvstad, Adelina Hild,
and Elisabet Tiselius, 93–120. Amsterdam: John Benjamins. DOI: 10.1075/btl.94.09shr
DOI: 10.1075/btl.4
Opening eyes to opera
The process of translation for blind
and partially-sighted audiences
Sarah Eardley-Weaver
In today’s rapidly developing digital age and increasingly socially-aware society,

the notion of media accessibility is evolving in response to shifting audience
expectations. Performing arts and media, such as opera, are called upon to
include all audiences, and related audiovisual translation methods are progress-
ing in this direction. These comprise audio description and touch tours for the
blind and partially-sighted, two relatively new translation modalities which
are consumer-oriented and require an original research design for the analysis
of the translation processes involved. This research design follows two funda-
mental principles: (1) audience reception studies should be an integral part of
the investigation into the translation process; and (2) the translation process is
regarded as a network. Therefore, this chapter explores the unique translation
processes of audio description and touch tours within the context of live opera
from the perspective of actor-network theory and by providing an overview
of a reception project. Through discussion of the methodology and findings,
this chapter addresses the question of the impact of audience reception on the
translation process.
Keywords: opera accessibility, translation process, audience reception, audio

description, touch tour, actor-network theory
1. Introduction
Opera translation is evolving to embrace all audiences, including the blind and
partially-sighted, and the process of translating this multisemiotic art form is ad-
vancing towards widened accessibility. As numbers of people with sight loss are
rising in aging populations, innovative translation techniques to provide for this
growing minority group are becoming more widespread in various audiovisual
doi 10.1075/bct.77.08ear
126 Sarah Eardley-Weaver
(AV) media, such as television and opera (Díaz-Cintas and Anderman 2009).1
These developments reflect an increasing general acknowledgement of accessibil-
ity issues in today’s society, also demonstrated by legislation such as the European
‘Audiovisual Media Services Directive’ (European Parliament 2010) and the UK
‘Disability Discrimination Act 2005’ (Great Britain Parliament 2005). Similarly,
in the field of translation studies, the concept of translation and the scope of AV
translation research are expanding in line with social and politico-economic ad-
vancements to encompass media accessibility (Remael 2010). However, the pro-
cess of raising awareness about access facilities, such as opera translation for blind
and partially-sighted patrons (BPSPs), is still in progress (Greening and Rolph
2007: 127; Díaz-Cintas and Anderman 2009: 6). It is within this sociocultural con-
text that this chapter investigates the process of translation of live opera for BPSPs
in the UK, employing an original research design. This research design is based
on the notion of the translation process as a network (see Section 2) and takes
into account that analyzing audience reception is an integral part of translation
process research (see Künzli and Ehrensberger-Dow 2011: 188; Tiselius and Jenset
2011: 273; Mateo 1997: 99–102). The central focus in this chapter is the examina-
tion of audience feedback on opera translation facilities for BPSPs and of the im-
pact of audience reception on the translation process.
Firstly, the notion of translation in this study is problematized in relation to
the two main opera translation modalities targeted at BPSPs in the UK: audio de-
scription (AD) and touch tour (ToTo). The theoretical framework is established,
and the translation processes of ToTos and of the different types of AD currently
available in UK opera houses are examined. Secondly, the audience’s response to
the translations is investigated with reference to the reception project carried out
in collaboration with the UK opera company Opera North (ON) at performances
of Bizet’s Carmen in May 2011. Finally, the study findings and methodological ap-
proach are discussed in relation to the wider research framework, and conclusions
are drawn regarding their impact on the translation process.
2. Problematizing translation in opera audio description and touch tours
The live performance of opera is an intrinsically multisemiotic experience which

appeals to multiple senses. Hence, for the purposes of this study, translation is
regarded as a form of communication between sign systems: a transfer process (or
1. Results show “more than a doubling (115% increase over 2010) in the numbers of people with
partial sight and blindness in the UK, to nearly 4 million people by 2050” (Access Economics
2009: 44).
Opening eyes to opera 127
product thereof) between one sensory communication channel and another. As

such it can be labeled “intersensorial” (De Koster and Mühleis 2007: 189), “mul-
tidimensional” (Gerzymisch-Arbogast 2005: 5), or “intersemiotic” translation (a
nuanced redefinition of Jakobson’s original 1959/2000 term by Gottlieb 2005: 3).
Translation is also viewed as “a form of accessibility” and vice versa (Díaz-
Cintas et al. 2007: 13–14). Therefore, the role of the translator is to facilitate com-
munication and access by overcoming “not only linguistic but also sensorial
barriers” (Orero and Matamala 2007: 262). In AD and ToTos, the translator’s role
is to convey the visual and other semiotic features primarily via the audio and
tactile sensory channels. The translator can be described as a narrator, interpreter,
and performer, but above all as a facilitator of social inclusion. Therefore, the defi-
nition of translator is a multifaceted and dynamic notion. This multifunctional
role of the translator highlights the complexity of the translation processes in op-
era translation for BPSPs, and a parallel can be drawn here with sign language
interpreters, who frequently “switch between confidant, co-worker, interpreter,
assistant and advocate” (Dickinson and Turner 2008: 231; see also Swabey and
Gajewski-Mickelson 2008; Roy 2000; 2002). In AD and ToTos, the translator’s role
is also collaborative. For instance, in addition to the audio describer(s), audience
members may contribute to the AD translation process, whether intentionally or
not, through comments and reactions. In ToTos, the collaborative aspect of the
translation process is more complex, as discussed in Section 2.2.
In this study, which investigates the translation process and the translator’s role
in ToTos and AD by examining audience reception, the methodology employed
is a combination of a top-down design and a bottom-up, data-driven approach.
The top-down design is based on a multisemiotic model of translation (Eardley-
Weaver 2010: 2), which facilitates the identification of the intersemiotic opera
translation processes for BPSPs. The bottom-up approach refers to the methods
for analyzing these processes. This analysis, which is conducted from a sociologi-
cal perspective using ethnography, draws on actor-network theory (Latour 2005)
and grounded theory (Glaser and Strauss 1967).2
In actor-network theory, a process is viewed as a network consisting of agents
and the interactions between them. Agents can be human or non-human and are
characterized as “anything that can induce […] an action” (Buzelin 2005: 197).
This theory can be applied in translation process research, for example, in es-
tablishing which networks exist and in exploring the links between the agents
(Chesterman 2006: 22). Furthermore, as Buzelin argues, the actor-network
2. Ethnographic research is characterized by the collection of data in natural settings using a
variety of techniques to explore and understand human social worlds from the perspective of
the research participants (Hammersley and Atkinson 2007: 2; Hubscher-Davidson 2011: 4).
concept “enables us to grasp both the complexity — and nonlinear character —

of the translation process, and the hybridity of the translating agent” (Buzelin
2005: 212). In this study, the opera translation process for BPSPs is regarded as a
network and the agents include, among others, the audience, set, audio describ-
ers, and service providers. Thus, actor-network theory provides a theoretical
framework for the investigation in this chapter, which focuses on establishing a
link in the network between the audience and audio describer for both AD and
ToTos through audience feedback.
Grounded theory can be defined as “the discovery of theory from data systemat-
ically obtained from social research” (Glaser and Strauss 1967: 2). This ethnographic
style of research is particularly suited to the investigation of the opera translation
process for BPSPs because it can accommodate the complex and contradictory evi-
dence that qualitative data may present. As Koskinen (2008: 36–39) argues, it not
only enables flexibility and complexity in data collection and analysis within a real-
life environment, it also supports “theory-building by providing methods for elicit-
ing new kinds of qualitative data on the social aspects of translation.” Furthermore,
it promotes reflection (Hammersley and Atkinson 2007: 4), the benefits of which for
translation process research have been highlighted by Hubscher-Davidson (2011).
Reflection within the context of opera translation for BPSPs has the added advan-
tage of raising awareness of the processes involved in the access facilities.
This discussion relates to the sociocultural aspect of the translator’s role which
determines the social impact of the research. This research engages with audi-
ence responses to translation modalities which contribute to social inclusion by
granting further opera accessibility to all and by allowing a shared experience.
Consequently, collective awareness is raised about these modalities and dialogue
is promoted among the various agents involved in the translation process. The
importance of considering the translation process within its sociocultural envi-
ronment is emphasized by Holz-Mänttäri (1984). Actor-network theory facilitates
this, “looking not only at the translators, but also at the other parties involved in
the translation process (clients, subject-matter experts, colleagues) and the many
tools that are part of modern translation work” (Risku and Windhager 2013: 36).
The study presented in this chapter is only one component of a larger research
project which uses an extension of the research methodology outlined above, most
notably to include the viewpoint of the translator as well as the audience. Thus, the
entire translation process, regarded as a network, is analyzed from the production
of the translation to its reception (see Eardley-Weaver 2014). This chapter focuses
on examining the audience responses to the translation and the role of audience
reception in the translation process.
In addition to AD and ToTos, resources for BPSPs include large-print or braille
program notes and audio-subtitling described by Orero (2007: 141) as “the media
accessible mode of reading aloud, or voicing, subtitles.” Since these methods are
rarely used in the UK, the focus in this chapter is restricted to AD and ToTos.
2.1 Definition and types of AD
AD is defined by Orero (2005: 7) as:

the descriptive technique of inserting audio explanations and descriptions of the
settings, characters and action taking place in a variety of audiovisual media,
where information about these visual elements is not offered in the regular audio
presentation.
In essence, AD “transforms visual information to words, translates the visual into

spoken language” (López Vera 2006: 1). In the opera AD process, the translator
verbalizes the opera’s visual aspects, with mode-shifts firstly from the visual to
the written form (the audio describer writes a printed script, having seen the per-
formance) and then to the audio format (the script is spoken aloud; see Eardley-
Weaver 2010: 3; Braun 2008: 15–17).
In this chapter, AD is used as an umbrella term which encompasses the dif-
ferent types of this translation modality, including audio introduction (AI), au-
dio through-description, and various other subcategories. Figure 1 illustrates the
classifications used here to distinguish between the varying forms of opera AD for
comparison purposes.
AUDIO
DESCRIPTION
(AD)
Audio
Audio ‘through’
introduction
description
(AI)
Live Pre-
recorded
Standalone Preparatory Standalone Preparatory, extended version

version version version (hard copy &/or online)
Condensed Extended
version (to be version (hard
played in the copy &/or
theatre) online)
Figure 1. Subdivisions of opera AD used in the UK.
AI refers to audio notes provided prior to the performance, whereas through-

description signifies live intermittent commentary throughout the performance.
Distinctions are also made between pre-recorded and live AI, both of which have
subcategories: a standalone version designed to be listened to on its own, and a
preparatory version designed to be listened to before a through-described perfor-
mance. These different types of AD are not mutually exclusive. The UK AD com-
pany VocalEyes provides both a preparatory version of live AI to be played in the
auditorium via wireless headphones immediately prior to the performance, and a
pre-recorded extended version available online or by request via post in advance
of the performance.
2.2 Touch tours
Live AD in some UK performances is often accompanied by a ToTo which oc-

curs shortly before curtain-up and adds an interactive element to the opera ex-
perience. As Figure 2 shows, this involves a guided tour of the stage by the audio
describer(s) with opportunities to touch items of the set, props, and sometimes a
cast member in costume (Orero and Matamala 2007: 273). In this chapter, ToTos
are studied in conjunction with AD, because at present opera ToTos are only of-
fered at audio-described performances and, although patrons can use AD or ToTo
alone, these two facilities are envisaged in tandem. Moreover, ToTos involve oral
explanation by the audio describer(s).
In ToTos, a combination of intersemiotic translation processes occurs between
multiple sensory communication channels, predominantly tactile and audio. The
audio channel is engaged primarily by the audio describer who orally describes
various visual features such as the set, props, hairstyles, and costumes. There may
be additional agents in this translation process; sometimes assistant directors, stage
designers, backstage staff, and cast members are willing to offer extra comments
or explanations during the ToTo (Udo and Fels 2010: 232). Also, companions of
BPSPs may provide physical guidance, directing which items to focus on, and
orally describing visual aspects. The amount of involvement of the various agents
depends on the audio describers’ translation strategy in the ToTo, which varies ac-
cording to the nature of the production as well as other factors such as the numbers
and ages of participants. Sometimes audio describers guide all participants around
the stage throughout the ToTo, or they might begin the ToTo with a brief introduc-
tion and then allow BPSPs to freely explore the stage with their companions.
The translation process engaging the tactile channel is conducted by the
BPSPs themselves, adopting the role of translator “as they translate the informa-
tion gathered by means of touch, such as texture and weight, in order to create
their own mental image without any external agent” (Eardley-Weaver 2010: 3–4).
Figure 2. Touch tour by VocalEyes. Photo courtesy of ON.
Thus, the roles of translator and target receiver merge. This merging of roles is rare
in other translation contexts, although a parallel can be drawn with fansubbing,
“the most important form of crowdsourcing” translation in which members of a
community of fans provide subtitles of a film or program and make them avail-
able online for this community (European Commission 2012: 45). By recogniz-
ing that fans are not passive consumers and are confident about the contributions
they can make to the translation process, the term ‘prosumer’ has been used to
describe fansubbers, as the boundary between producer and consumer is blurred
(European Commission 2012: 31). There is a similarity with ToTo participants,
as they are given opportunities to ask questions and are active in the translation
process. Another parallel can be drawn with crowdsourcing, as a ToTo is a joint
translation process performed by the multifarious network of the aforementioned
agents. The decisions of the various translating parties will be mutually influ-
ential, and this is similar to crowdsourcing translation in which work is carried
out collaboratively to help other members of the community to access products
(European Commission 2012: 23).
Where used, ToTos have certainly become part of the multisemiotic opera
translation process for BPSPs. Like any translation process, social and contextual
factors have a significant impact on the format and route of the ToTo. For instance,
the choice of elements to touch may be determined by safety concerns (e.g., set
fragility or dangerous objects) or by time constraints. Similarly to any other trans-
lation strategy — such as the impossibility of using computer-aided translation
(CAT) tools for certain language combinations — practical issues determine the
feasibility of consistently providing ToTos because some opera houses do not have
an accessible stage. In these situations, guiding BPSPs may prove to be dangerous.
3. The impact of audience reception on the translation process
Research into the audience’s perspective is fundamental to gain further insights

into the multisemiotic opera translation process. As Di Giovanni (2011: 10) high-
lights, “media audiences are indeed the driving force behind the success or decline
of the media themselves; they are a dynamic, increasingly diverse and demanding
force, whose specific identity and requirements are also extremely relevant for the
translation process.”
Audience expectations are changing as people become increasingly accus-
tomed to multimedia environments (Gambier and Gottlieb 2001). Therefore, au-
dience reception studies are of paramount significance in order to respond to this
sociological shift. Although, as Denton claims, “audience reception is fast becom-
ing a leading feature of audiovisual translation research” (2007: 26), translation
studies focusing on audiences remain limited (Di Giovanni 2011: 10). Gambier
emphasizes the urgent demand “to know the viewers’ needs and reception capacity,
whatever the modality of AVT being offered” (2006: 5; see also Chiaro 2006, 2007).
Within the network of opera translation for BPSPs, audience feedback pro-
vides a link between audience and translator. This may be viewed as a dialogue,
referring to Hansen’s definition of this term: “a process of negotiation with the
purpose of reaching subjective and inter-subjective identification and clarification
of a phenomenon of interest” (Hansen 2008: 394). Thus, audience feedback can
promote collective understanding of translation quality among the agents in the
network (Chesterman 2006: 23).
With this in mind, in 2011 a pioneering pilot reception project was carried out
at two performances of ON’s production of Carmen. This dramatic, sensuous op-
era tells the tragic story of the love triangle between a soldier, a bullfighter, and the
passionate seductress Carmen, concluding with her murder. Audio description,
touch tour, sign interpreting, and surtitles were provided at each of the perfor-
mances on 8 May at The Grand, Leeds, and 24 May at Nottingham Theatre Royal
(UK). The reception project aimed to evaluate the effectiveness of these opera ac-
cess facilities from the audience’s perspective. In this chapter, the objective is to
analyze the translation processes of AD and ToTos by studying BPSPs’ responses to
these modalities. The hypothesis is that while existing translation modalities meet
audience needs to a certain degree, there are individual elements, especially relat-
ing to the notion of a shared and inclusive experience, which could be improved.
3.1 Reception project research methodology
The overall research design for this reception project drew on Oppenheim’s model
for data collection and survey (1992: 6–8). A combined methodology using a two-
fold design (Creswell 2003: 114) which involved questionnaires and feedback via
discussion was employed. These data collection methods were chosen as they were
considered the most ecologically valid and reliable way to investigate reception of
the actual translation techniques in their natural environment, given the live nature
of AD and ToTos and the exploratory approach of this study. Feedback was gath-
ered from users of the AD and/or ToTo, including BPSPs and their sighted com-
panions. Respondents were asked to complete two questionnaires: a preliminary
questionnaire prior to the performance to gather demographic details and other
personal information, and a post-performance “group-administered” question-
naire (Oppenheim 1992: 103) to gain feedback on the access facilities provided.3
In all questionnaires there were both closed-ended and open-ended items.
In the post-performance survey, questions required graded responses using
a variation on the Likert scale technique which replaced the standard set of re-
sponses (strongly agree to strongly disagree) with the numerical scale 1=not at all,
2=not very much, 3=so-so, 4=quite a lot, and 5=very much. Each question on the
hard copy post-performance questionnaires also included room for comments,
and further qualitative data was gathered from informal group discussion.
In this chapter, the initial analysis of the questionnaire results is presented,
combining quantitative and qualitative data.4 Firstly, the profile of the whole group,
including BPSPs and their sighted companions, is briefly discussed. Secondly, data
collected regarding the BPSPs’ reception of the AD and ToTo is analyzed. The fo-
cus of this data analysis is three-fold, concerning the audience reception of: (1)
different types of AD; (2) the translation of features provoking an emotional reac-
tion, such as humor and shock; and (3) the ToTo. Humor and shock are focused on
because this production of Carmen included dark comedy and explicit references
3. See following link to questionnaires: www.dur.ac.uk/cim/members/seweaver.
4. The AD and ToTo at the survey performances were provided by VocalEyes (VocalEyes 2011).
to sexuality and violence. Although previous literature mentions the different

types of AD and the translation of emotion in opera AD (see Matamala and Orero
2007; Puigdomènech, Matamala, and Orero 2008; Holland 2009; York 2007), there
are very few audience reception studies into opera AD (Matamala 2005; Cabeza
2010), and little or no research on opera ToTos. Furthermore, UK opera access
providers have expressed interest in the survey results in relation to the develop-
ment of their accessibility provisions regarding the types of facilities offered and
translation strategies employed.
3.2 The participants
The profile of the whole group is discussed here because the opera experience is
a social activity shared among BPSPs and their companions, who may, moreover,
be agents in the translation process (see Section 2.2). In total, 32 people partici-
pated in the questionnaire on AD and ToTo: 11 sighted patrons and 21 BPSPs,
of whom 4 were totally blind, 16 were legally blind and 1 was partially-sighted.5
Among the BPSPs, 8 were male and 13 were female. Among the sighted patrons,
there were 5 males and 6 females. Figure 3 shows the distribution of age across the
whole group: half of the participants were aged 66 and above. These results could
arguably suggest a lack of awareness among younger people of the availability of
Age groups
10 – 17 18 – 25
26 – 35
36 – 45
10 − 17
18 − 25
26 − 35
36 – 45
66 + 46 – 55
46 – 55 56 – 65
66 +
56 – 65
Figure 3. Distribution of age in the participants using AD and touch tour.
5. Categories are based on definitions of visual acuity by the RNIB (Access Economics
2009: 11–12).
these access facilities, although many other factors could explain the low numbers
of younger AD and ToTo users, such as the opera genre itself, or this particular
production of Carmen, which was given a 12+ rating by ON. Interestingly, com-
ments made by respondents highlighted a general lack of awareness and publicity
regarding access facilities. One participant commented that “access is not always
advertised, you have to ask” and another remarked “I’d forgotten about that” with
reference to the pre-recorded preparatory AI.
The results analyzed in subsequent sections refer to responses from the 21
BPSPs participating in the survey. The analysis involves examination of quanti-
tative and qualitative data collected from the post-performance questionnaire in
the form of graded responses and comments. This accumulated quantitative and
qualitative data of 21 BPSPs is significant, because on average 27 BPSPs register
to use AD and/or ToTo annually at ON productions.6 Due to the nature of the
ethnographic research approach adopted, results are necessarily situated and cau-
tion is needed in making inferences (Hubscher-Davidson 2011: 13). However, as
Blommaert points out, referring to Pierre Bourdieu’s views, “there is a clear sug-
gestion that single cases, even if they don’t speak to the totality of the population
or the system, can speak to theory” (2005: 228).
3.3 Reception of different types of AD
Figure 4 displays the average scores given by the BPSPs to questions regarding the
amount and helpfulness of the different types of AD listened to. Standalone AI was
not provided at the reception project performances, so this translation modality
cannot be investigated. However, live AI and pre-recorded preparatory AI were
Amount of types of Helpfulness of

AD listened to different types of AD
5 5
4 4
Through description
3 3 Live AI
Preparatory AI
2 2
1 1
Figure 4. Average perceived amount and helpfulness of different types of AD listened to.
6. Personal communication with ON, May 2011.

offered in conjunction with through-description and therefore observations can

be made of these facilities.
The left graph in Figure 4 shows that on average the BPSPs perceived that they
listened to more through-description than both live and preparatory AI. This re-
sult is not surprising in light of the effort required to gain access to each type of
AD. For example, listening to preparatory AI requires time and effort prior to go-
ing to the performance. The high scores in the right graph of Figure 4 show that
all types of AD were considered very helpful. These results are not intended to
assess which type of AD is the most helpful, as the participants were not asked to
rank them. Certain comments from the survey respondents help explain the high
scores regarding the helpfulness of the different types of AD. For example, one
BPSP remarked with reference to the preparatory, pre-recorded AI:
Because of the unorthodox interpretation and imaginative production with so
much happening, the basic storyline was a bit complicated to understand. Having
had the pre-show notes, it was easier to follow the plot — less confusing — just
like for sighted people it’s helpful to read the program to know what is happening,
who is who, which character is which and so on.
This observation highlights the benefit of increased familiarity with the opera and
production gained by listening to the preparatory AI. Similar comments empha-
sizing this element were made regarding the live AI, such as “it’s a good reminder
even if you’ve heard it before.” Several participants noted the value of both of these
types of AD in setting the scene. Regarding through-description, participants
commented on the helpfulness of the immediacy of the description and commu-
nication of details as they happen.
3.4 Reception of features provoking an emotional response
The average scores given by the BPSPs to questions concerning emotional response
to the opera are displayed in the first three columns in Figure 5.
The first column in Figure 5 shows the average response to a question asking
participants if they felt emotionally engaged with the opera on this occasion. The
score of 3.7 reveals a fairly positive response, although it also suggests that some
participants did not feel that they were able to fully experience the emotion of the
opera. The following remark by a survey participant referring to the second-hand
experience in AD for BPSPs, although emphasizing the social role of translation
in contributing to inclusion, might provide a partial explanation for this result.
With AD you are experiencing something second-hand. As a listener you don’t
really have an opinion, you don’t have your own reaction to it, you have the audio
describer’s reaction to it. But it really helps to be able to share visual aspects with
4.5
3.5
2.5
1.5
1
Emotion Humor Shock Enjoyment of Helpfulness of
touch tour touch tour
Figure 5. Average response to the translation of emotional aspects of the opera and to the
touch tour
other audience members, which is the most important dimension. AD provides

an independent source, so you can discuss it after with other audience members.
This comment highlights the paradoxical nature of the reception of AD, the in-
evitable subjectivity of the audio describer’s role and the interpretative element
in the AD process as a whole. The AD allows the BPSP to interpret the opera
independently from other audience members, but the BPSP’s interpretation is not
independent because it is influenced by the audio describer’s rendering; it is sec-
ond-hand. Another BPSP alluded to this distancing effect, but this time in relation
to the translation of humor in the AD. The participant stated:
It’s quite hard to convey humor in the third person because it’s reported speech.
It’s hard to portray something that’s funny because of the timing of the punchline
— there’s a delay in the delivery of the humor. […] The reaction of the audience is
not simultaneous and so it loses impact.
The average scores of 3.8 and 4.0 given to questions about the AD’s helpfulness
in conveying humorous and shocking aspects, respectively (see columns 2 and 3,
Figure 5), indicate a positive response overall, although participants’ comments
reveal mixed reviews. Acclamations included, for example, “AD has brought the-
ater to life for me. I received the humor of Carmen well because of the AD,” and “I
liked the comic moment in the AD where it said something like ‘Zuniga sits there
in all his massiveness.’ I don’t mind subjectivity — it’s inevitable so why not have
fun!” On the other hand, there were also mildly critical comments such as “I don’t
think that was as good. I imagine humor is difficult to convey in AD.” Similarly, re-
garding shocking elements, several BPSPs’ comments revealed a lack of awareness
of any shocking details, such as nudity and allusions to sexual violence, suggesting
that the translation in the AD was not entirely successful for some. In particular,
the lack of a shared reception of this feature was noted. For instance, one partici-
pant highlighted the difference in verbal and visual communication:
If I compare my feelings with my [sighted] wife’s feelings, my wife found it a vio-
lently disturbing production. I didn’t receive this idea of physical violence. If one
is told ‘Zuniga kicks Carmen after she has fallen over’ it has less impact than if
you see it. I didn’t wince but it did make me think. I did notice the use of shock-
ing language in the AD to convey the shock of the production, for example ‘she
punched him in the balls’ — that was a clue that the gloves were off. However,
although it gave a hint, still the language used didn’t help me receive the same
impact of visual violence.
This remark also raises questions concerning the notion of a shared experience
which is particularly significant in the context of live performance received in a
group context, as demonstrated by other participants’ comments. There were also
very positive remarks regarding the helpfulness of AD in conveying shocking as-
pects, such as:
The AD was very helpful for this, not just for the stabbing of Carmen but also the
sinister characters Escamillo and the leader of the guards Zuniga. I got a sense of
the sinister characters, and also the fact that Carmen is a bit of an outcast — the
idea of racism. That shocking aspect came across loud and clear. The use of lan-
guage and not using euphemistic language was also effective — they [the audio
describers] said it as it was ‘she kicked him in the balls.’
Further research is needed to explore this diversity in audience response, which

may have been affected by other factors such as the amount of AD listened to (see
Eardley-Weaver 2014). Nevertheless these preliminary results underline the chal-
lenge for the translator to provide a translation facility for patrons with differing
opinions and expectations. Moreover, they highlight the importance of dialogue
between translator and audience in discerning possible refinements needed in the
AD process according to consumer requirements.
3.5 Reception of touch tours
Columns 4 and 5 in Figure 5 show the average scores given by the BPSPs who at-
tended the ToTo (11 out of 21) to questions about enjoyment and helpfulness of
the ToTo. The average score given for enjoyment is marginally higher than that
for helpfulness. This result might be explained by the respondents’ emphasis on
the “fun” aspect of the ToTo, as mentioned by several participants, referring, for
example, to the enjoyment of speaking to the actress who attended the ToTo. Also,
while some explicitly remarked on the helpfulness of the ToTo in providing op-
portunities to ask questions and in visualizing the stage, the following participant
comment highlights a limitation of the ToTo: “it depends on the set as to whether
the ToTo is helpful.” Further audience reception research is needed regarding fac-
tors affecting the helpfulness and translation processes of ToTos, such as the set,
agents involved, and numbers and ages of patrons. Studies focusing on BPSPs’
preferences regarding interpretative description and explanation of elements of
the director’s and design team’s vision are also required. From this perspective, the
following comment is also noteworthy:
I have some reservations about ToTos but I can’t resist them. It’s so nice to be part
of the business. It adds to the experience of the performance. But it’s a bit like
seeing a film of a book you like. In some ways I prefer to create my own mental
image of the stage from the AD without interference of the tactile experience and
details on the ToTo. […]
The same BPSP continued as follows, emphasizing the social aspect of the transla-
tion process in providing an inclusive experience:
As a totally blind person who has never seen, all the details given about color, cos-
tumes, props, and so on are not particularly helpful for me. I realize though that
audio describers have to cater for all audiences with differing degrees and types
of visual ability. For me, the important part of the opera experience is what you
talk about with other members of the audience. That is what is so valuable about
ToTos and AD — that you can share the experience with the rest of the audience.
This remark also emphasizes the need for further research into the varying re-
quirements of a diverse audience.
3.6 Discussion of findings from the reception project
The results support the initial research hypothesis that, while current translation
processes in AD and ToTos meet BPSPs’ needs to a certain degree, improvements
could be made. The quantitative and qualitative data suggest a generally positive
response from BPSPs to AD and ToTos, but comments highlight some of their
perceived limitations. While it must be acknowledged that these findings are con-
text-bound due to the ethnographic research approach adopted (Hammersley and
Atkinson 2007: 159; Glaser and Strauss 1967), they are clearly valuable, especially
given the pioneering nature of the reception project.
In relation to the actor-network theoretical framework discussed in Section 2,
the findings corroborate the idea that AD and ToTos form part of a translation
network. For example, this notion is tacitly understood by BPSPs, as demonstrated
by comments referring to the involvement of various agents in the translation pro-

cesses of AD and ToTos, including companions, cast members, the set, and the
BPSPs themselves. The comment about the delay in the delivery of the humor in
AD implies that the BPSP realizes that the audience’s reaction has translated an
aspect of the performance.
The findings also provide critical appraisal of the translation of specific fea-
tures such as emotion. BPSPs’ comments reveal the importance of sharing the
opera experience with other audience members, including sighted companions,
and acknowledge the helpfulness of AD and ToTos in facilitating this. However,
the results also suggest that AD and ToTos in their current format do not allow
BPSPs to fully share the emotion of the opera, which relates to the BPSPs’ sense of
a second-hand experience. It is implied by BPSPs’ comments that factors affecting
their reception of emotion include: (1) delay in delivery and level of subjectivity in
the AD; and (2) the influential role of companions in the translation process, who
may affect the BPSPs’ perception of the quality of the AD by providing another
translation with which to compare the AD. This highlights the complexity of the
audio describer’s role within the collaborative network of agents and emphasizes
the delicate balance required between using subjectivity to provide a vivid AD and
allowing the BPSPs to interpret the performance themselves.
Many of the inherent difficulties in the translation processes of AD and
ToTos stem from the differences in visual and verbal communication. This results
in inevitable subjectivity, as the audio describer Holland states: “there is no di-
rect equivalent between a moment on stage and the words chosen to describe it”
(2009: 184). Consequently, the incorporation of audience feedback into the trans-
lation process is crucial in establishing BPSPs’ preferences, which may inform the
production of AD and ToTos. This relates to the idea in fansubbing of embracing
consumers as “active agents” in the translation process through reception studies
“in favour of a collaborative method based on a circular flow of different skills
and profiles and the interaction among them” (European Commission 2012: 31).
Similarly, this chapter proposes that the translation network is cyclical with audi-
ence feedback as both the final stage (Matamala and Orero 2007: 209) and the first
stage of the process. This builds on the idea of translation as a constantly evolving
notion and an “open concept” (Tymoczko 2005: 1091). Given that performances
are often repeated within a short timeframe, it is particularly relevant to regard
the translation of live opera for BPSPs in this way. This notion may apply to other
translation contexts, such as fansubbing, especially within consumer-oriented en-
tertainment media.
Following on from this reception project, avenues of future research may ad-
dress the following questions: what exactly is a shared experience according to
receivers? Do ToTos reduce the distancing effect of AD and enhance emotional
engagement? Are there core features that patrons consider fundamental to any
ToTo or other details that should not be revealed? Furthermore, the theoretical
framework drawing on actor-network theory and grounded theory provides a
foundation from which to develop larger-scale audience reception studies, also
examining opera AD and ToTos in conjunction with other facilities such as au-
dio subtitling and braille program notes. This theoretical framework may also be
transferable to other translation contexts in view of the aforementioned analogies
between AD or ToTos and other forms of translation such as crowdsourcing and
sign interpreting. In addition, research bringing together investigation into the
translator’s role and audience reception is needed in order to unite and promote
dialogue between these agents in the translation network and to further explore
the sociocultural, collaborative aspects of the translation process. As Gambier
states: “in this changing mediascape, the translators have and will have a major
role, if they fully realize their socio-cultural function” (2006: 7).
4. Concluding remarks
As audience expectations are constantly evolving in the increasingly multimedia

world of today, translation process research must become ever more versatile and
receptive to these changes. This need is particularly evident in the relatively unex-
plored field of opera accessibility, especially given that rapid changes are occuring
as experimentation with new technological translation techniques is already in
progress. For example, at the Gran Teatre del Liceu opera house in Barcelona, a
prototype developed at the Universitat Autònoma de Barcelona is to provide ac-
cess via hand-held screens such as smartphones (Oncins et al. 2013).
Within this dynamic arena, audience reception studies are of key importance
and actor-network theory provides a flexible theoretical framework for studying
the translation process in all its complexity (Chesterman 2006; Buzelin 2005; Risku
and Windhager 2013). This includes investigation of the fundamental link between
audience and translator, as discussed in this chapter, which reveals a merging of
roles of these two agents in the highly collaborative network of opera ToTos. The
advantages of considering the translation processes of AD and ToTos as collabora-
tive, cyclical networks which involve various agents and links are highlighted by
the reception project. Through dialogue, reception research promotes identifica-
tion and clarification of the various agents and links, which may help participants
to feel part of the network and contribute to BPSPs’ sense of a shared experience.
It can also foster recognition among BPSPs and their companions of the intrica-
cies of the translation process, which may in turn encourage reflection concerning
their expectations and perceptions of these facilities. Furthermore, it promotes
mutual understanding among the agents in the network, which may contribute to
changes in translation practice. In addition, reception research heightens aware-
ness of the access facilities in general and of any innovations in the field, thus
contributing to a universally inclusive, multisemiotic opera experience. In sum-
mary, audience reception studies may open eyes to the collaborative, complex, and
dynamic nature of the opera translation network.
References
Access Economics. 2009. “Future Sight Loss UK (1): The Economic Impact of Partial Sight and
Blindness in the UK Adult Population.” RNIB. http://www.rnib.org.uk/knowledge-and-
research-hub/research-reports/general-research/future-sight-loss-uk-1.
Blommaert, Jan. 2005. “Bourdieu the Ethnographer: the Ethnographic Grounding of Habitus
and Voice.” The Translator 11 (2): 219–236. DOI: 10.1080/13556509.2005.10799199
Braun, Sabine. 2008. “Audiodescription Research: State of the Art and Beyond.” Translation
Studies in the New Millennium 6: 14–30.
Buzelin, Hélène. 2005. “Unexpected Allies: How Latour’s Network Theory Could Complement
Bourdieusian Analyses in Translation Studies.” The Translator 11 (2): 193–218.
DOI: 10.1080/13556509.2005.10799198
Cabeza, Cristóbal. 2010. “Opera Audio Description at Barcelona’s Liceu Theatre.” In New
Insights into Audiovisual Translation and Media Accessibility, ed. by Jorge Díaz-Cintas, and
Josélia Neves, 227–238. Amsterdam: Rodopi.
Chesterman, Andrew. 2006. “Questions in the Sociology of Translation.” In Translation Studies
at the Interface of Disciplines, ed. by João Ferreira Duarte, Alexandra Assis Rosa, and Teresa
Seruya, 9–27. Amsterdam: John Benjamins. DOI: 10.1075/btl.68.03che
Chiaro, Delia. 2006. “Verbally Expressed Humour on Screen: Reflections on Translation and
Reception.” JoSTrans 6: 198–208.
Chiaro, Delia. 2007. “The Case of Dubbed Comedy in Italy.” In Doubts and Directions in
Translation Studies, ed. by Yves Gambier, Miriam Schlesinger, and Radegundis Stolze, 137–
152. Amsterdam: John Benjamins. DOI: 10.1075/btl.72.16chi
Creswell, John W. 2003. Research Design: Qualitative, Quantitative, and Mixed Methods
Approaches. Thousand Oaks, CA: Sage.
De Koster, Karin, and Volkmar Mühleis. 2007. “Intersensorial Translation: Visual Art Made up
by Words.” In Media for All: Subtitling for the Deaf, Audio Description and Sign Language,
ed. by Jorge Díaz-Cintas, Pilar Orero, and Aline Remael, 189–200. Amsterdam: Rodopi.
Denton, John. 2007. “Audience Reception of Translated Audiovisual Texts and the Problem of
Visual and Verbal Shared/Unshared Knowledge.” In La Lingua Invisibile. Aspetti Teorici e
Tecnici del Doppiaggio in Italia [Invisible Language. Theoretical and Technical Aspects of
Dubbing in Italy], ed. by Giuseppe Massara, 23–36. Roma: Nuova Editrice Universitaria.
Díaz-Cintas, Jorge, and Gunilla Anderman (eds). 2009. Audiovisual Translation. Language
Transfer on Screen. Basingstoke: Palgrave Macmillan.
Díaz-Cintas, Jorge, Pilar Orero, and Aline Remael (eds). 2007. Media for All: Subtitling for the
Deaf, Audio Description and Sign Language. Amsterdam: Rodopi.
Dickinson, Jules, and Graham H. Turner. 2008. “Sign Language Interpreters and Role Conflict in
the Workplace.” In Crossing Borders in Community Interpreting: Definitions and Dilemmas,
ed. by Carmen Valero Garcés, and Anne Martin, 231–244. Amsterdam: John Benjamins.
DOI: 10.1075/btl.76.12dic
Di Giovanni, Elena. 2011. “Connecting the Dots in Audiovisual Translation Research:
Translation, Reception, Accessibility and Children.” In Between Text and Receiver:
Translation and Accessibility, ed. by Elena Di Giovanni, 7–19. Frankfurt am Main: Peter
Lang.
Eardley-Weaver, Sarah. 2010. “Opening Doors to Opera: The Strategies, Challenges and General
Role of the Translator.” InTralinea 12: 1–8.
Eardley-Weaver, Sarah. 2014. Lifting the Curtain on Opera Translation and Accessibility:
Translating Opera for Audiences with Varying Sensory Ability. Durham: Durham theses,
Durham University. Available at Durham E-Theses Online: http://etheses.dur.ac.uk/10590/.
European Commission. 2012. “Crowdsourcing Translation.” Directorate General for Translation.
http://bookshop.europa.eu/en/crowdsourcing-translation-pbHC3112733/.
European Parliament, Council. 2010. “Audiovisual Media Services Directive 2010/13/EU.”
European Parliament. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32
010L0013:EN:NOT.
Gambier, Yves. 2006. “Multimodality and Audiovisual Translation.” MuTra 2006 — AV
Translation Scenarios: Conference Proceedings. 1–8. http://www.euroconferences.info/
proceedings/2006_Proceedings/2006_Gambier_Yves.pdf.
Gambier, Yves, and Henrik Gottlieb (eds). 2001. (Multi) Media Translation: Concepts, Practices,
and Research. Amsterdam: John Benjamins. DOI: 10.1075/btl.34
Gerzymisch-Arbogast, Heidrun. 2005. “Introducing Multidimensional Translation.” MuTra
2005 — Challenges of Multidimensional Translation: Conference Proceedings. 1–15. http://
www.euroconferences.info/proceedings/2005_Proceedings/2005_GerzymischArbogast_
Heidrun.pdf.
Glaser, Barney G., and Anselm L. Strauss. 1967. The Discovery of Grounded Theory: Strategies for
Qualitative Research. New York, NY: Aldine de Gruyter.
Gottlieb, Henrik. 2005. “Multidimensional Translation: Semantics turned Semiotics.” MuTra
2005 — Challenges of Multidimensional Translation: Conference Proceedings. 1–29. http://
www.euroconferences.info/proceedings/2005_Proceedings/2005_Gottlieb_Henrik.pdf.
Great Britain Parliament. 2005. “Disability Discrimination Act 2005: Chapter 13.” The Stationery
Office. http://www.legislation.gov.uk/ukpga/2005/13.
Greening, Joan, and Deborah Rolph. 2007. “Accessibility — Raising Awareness of Audio
Description in the UK.” In Media for All: Subtitling for the Deaf, Audio Description and Sign
Language, ed. by Jorge Díaz-Cintas, Pilar Orero, and Aline Remael, 127–138. Amsterdam:
Rodopi.
Hammersley, Martyn, and Paul Atkinson. 2007. Ethnography: Principles in Practice. New York,
NY: Routledge.
Hansen, Gyde. 2008. “The Dialogue in Translation Process Research.” In Translation and
Cultural Diversity: Proceedings of the Eighteenth FIT World Congress, Shanghai, 2–7
August 2008, 386–397. Shanghai: Foreign Languages Press.
Holland, Andrew. 2009. “Audio Description in the Theatre and Visual Arts: Images into Words.”
In Audiovisual Translation. Language Transfer on Screen, ed. by Jorge Díaz-Cintas, and
Gunilla Anderman, 170–185. Basingstoke: Palgrave Macmillan.
Holz-Mänttäri, Justa. 1984. Translatorisches Handeln: Theorie und Methode. Helsinki:

Suomalainen Tiedeakatemia.
Hubscher-Davidson, Séverine. 2011. “A Discussion of Ethnographic Research Methods and
their Relevance for the Translation Process.” Across Languages and Cultures 12 (1): 1–18.
DOI: 10.1556/Acr.12.2011.1.1
Jakobson, Roman. 1959/2000. “On Linguistic Aspects of Translation.” In The Translation Studies
Reader, ed. by Lawrence Venuti, 113–118. London: Routledge.
Koskinen, Kaisa. 2008. Translating Institutions: An Ethnographic Study of EU Translation.
Manchester: St Jerome Publishing.
Künzli, Alexander, and Maureen Ehrensberger-Dow. 2011. “Innovative Subtitling. A Reception
Study.” In Methods and Strategies of Translation Process Research: Integrative Approaches in
Translation Studies, ed. by Cecilia Alvstad, Adelina Hild, and Elisabet Tiselius, 187–200.
Amsterdam: John Benjamins. DOI: 10.1075/btl.94.14kun
Latour, Bruno. 2005. Reassembling the Social: An Introduction to Actor-Network Theory. Oxford:
Oxford University Press.
López Vera, Juan Francisco. 2006. “Translating Audio Description Scripts: The Way Forward?
— Tentative First Stage Project Results.” MuTra 2006 — AV Translation Scenarios:
Conference Proceedings, 1–10. http://www.euroconferences.info/proceedings/2006_
Proceedings/2006_Lopez_Vera_Juan_Francisco.pdf.
Matamala, Anna. 2005. “Live Audio Description in Catalonia.” Translating Today 4: 9–11.
Matamala, Anna, and Pilar Orero. 2007. “Accessible Opera in Catalan: Opera for All.” In Media
for All: Subtitling for the Deaf, Audio Description and Sign Language, ed. by Jorge Díaz-
Cintaz, Pilar Orero, and Aline Remael, 201–214. Amsterdam: Rodopi.
Mateo, Marta. 1997. “Translation Strategies and the Reception of Drama Performances: A
Mutual Influence.” In Translation as Intercultural Communication, ed. by Mary Snell-
Hornby, Zuzana Jettmarová, and Klaus Kaindl, 99–110. Amsterdam: John Benjamins.
DOI: 10.1075/btl.20.11mat
Oncins, Estella, Oscar Lopes, Pilar Orero, Javier Serrano, and Jordi Carrabina. 2013. “All Together
Now: A Multi-language and Multi-system Mobile Application to Make Live Performing
Arts Accessible.” JoSTrans 20: 147–161. http://www.jostrans.org/issue20/art_oncins.php.
Oppenheim, Abraham, N. 1992. Questionnaire Design, Interviewing and Attitude Measurement.
London: Pinter.
Orero, Pilar. 2005. “Audio Description: Professional Recognition, Practice and Standards in
Spain.” Translation Watch Quarterly 1: 7–18.
Orero, Pilar. 2007. “Audiosubtitling: A Possible Solution for Opera Accessibility in Catalonia.”
TradTerm 13: 135–149. DOI: 10.11606/issn.2317-9511.tradterm.2007.47470
Orero, Pilar, and Anna Matamala. 2007. “Accessible Opera: Overcoming Linguistic and Sensorial
Barriers.” Perspectives: Studies in Translatology 15 (4): 262–278.
DOI: 10.1080/13670050802326766
Puigdomènech, Laura, Anna Matamala, and Pilar Orero. 2008. “The Making of a Protocol for
Opera Audio Description.” In La Traduction del Futuro: Mediacíon lingüística y cultural
en el siglo XXI [Translation of the Future: Linguistic and Cultural Mediation in the Twenty
First Century]: Proceedings of the Third International Conference of the Asociación
Ibérica de Estudios de Traducción e Interpretación, Barcelona, 22–24 March 2007, ed.
by Luis Pegenaute, Janet DeCesaris, Mercè Tricás, and Elisenda Bernal, vol. I, 381–392.
Barcelona: Promociones y Publicaciones Universitarias.
Remael, Aline. 2010. “Audiovisual Translation.” In Handbook of Translation Studies. Volume I,

ed. by Yves Gambier, and Luc Van Doorslaer, 12–17. Amsterdam: John Benjamins.
DOI: 10.1075/hts.1.aud1
Risku, Hanna, and Florian Windhager. 2013. “Extended Translation. A Socio-cognitive Research
Agenda.” Target 25 (1): 33–45. DOI: 10.1075/target.25.1.04ris
Roy, Cynthia. 2000. Interpreting as a Discourse Process. New York, NY: Oxford University Press.
Roy, Cynthia. 2002. “The Problem with Definitions, Descriptions, and the Role Metaphors of
Interpreters.” In The Interpreting Studies Reader, ed. by Franz Pöchhacker and Miriam
Shlesinger, 344–353. New York, NY: Routledge.
Swabey, Laurie, and Paula Gajewski-Mickelson. 2008. “Role Definition: A Perspective on Forty
Years of Professionalism in Sign Language Interpreting.” In Crossing Borders in Community
Interpreting: Definitions and Dilemmas, ed. by Carmen Valero Garcés and Anne Martin,
51–80. Amsterdam: John Benjamins. DOI: 10.1075/btl.76.04swa
Tiselius, Elisabet, and Gard B. Jenset. 2011. “Process and product in simultaneous interpreting:
What they tell us about experience and expertise?” In Methods and Strategies of Translation
Process Research: Integrative Approaches in Translation Studies ed. by Cecilia Alvstad,
Adelina Hild, and Elisabet Tiselius, 269–300. Amsterdam: John Benjamins.
DOI: 10.1075/btl.94.20tis
Tymoczko, Maria. 2005. “Trajectories of Research in Translation Studies.” Meta 50 (4): 1082–
1097. DOI: 10.7202/012062ar
Udo, John Patrick, and Deborah Fels. 2010. “Enhancing the Entertainment Experience of Blind
and Low-vision Theatre-goers Through Touch Tours.” Disability and Society 25 (2): 231–
240. DOI: 10.1080/09687590903537497
VocalEyes. 2011. Audio description to Bizet’s Carmen. Audio recording produced for Opera
North, voiced by Margaret Spittles, and Anne Hornsby. VocalEyes.
York, Gregory. 2007. “Verdi Made Visible. Audio-introduction for Opera and Ballet.” In Media
for All: Subtitling for the Deaf, Audio Description and Sign Language, ed. by Jorge Díaz-
Cintaz, Pilar Orero, and Aline Remael, 215–229. Amsterdam: Rodopi.
Notes on editors
Maureen Ehrensberger-Dow is Professor of Translation Studies in the Zurich

University of Applied Sciences’ Institute of Translation and Interpreting and
principal investigator of both the Capturing Translation Processes (CTP) and the
Cognitive and Physical Ergonomics of Translation (ErgoTrans) research projects.
Her publications focus on the development of translation expertise and its interac-
tion with metalinguistic awareness, information literacy, and self-concept.
Birgitta Englund Dimitrova holds a Ph.D. in Slavic Linguistics and is Professor
Emerita of Translation Studies at Stockholm University, Sweden. Her main re-
search interest is in cognition, bilingualism and translation, as evidenced by her
monograph Expertise and Explicitation in the Translation Process, published by
Benjamins in 2005. She has also published on the interaction in interpreter-medi-
ated encounters and on the translation of dialect in fiction.
Séverine Hubscher-Davidson is Lecturer in Translation Studies at Aston
University, Birmingham. She teaches both translation theory and practice, and
her research interests are in the area of Translation Process Research, translators’
personalities and individual differences, as well as translator training. She has pub-
lished articles mainly on the psychological aspects of the translation process.
Ulf Norberg is Assistant Professor of Translation Studies at Stockholm University.
He received his Ph.D. in German from the University of Uppsala. His current re-
search interests include translation didactics, cognitive processes in translation,
translation of children’s literature and speech-to-text interpreting.
doi 10.1075/bct.77.10not
Index
A cognitive alignment 36 deliteralization 85, 87, 98, 102

acceptability 30, 31, 36, 38 cognitive awareness 120 deliteralization hypothesis 85,
activating cues 43 cognitive discourse model 45, 86
actor-network theory 4, 17, 125, 54, 57 deliteralization process 102
127, 128, 141 cognitive discourse processing deliteralizing changes 97, 98
AD see audio description framework 54 deprofessionalization 17
adaptation 37 cognitive effort 31, 65, 106 descriptive model 9, 10, 18
affect 3, 68, 69, 70 cognitive load 44, 120 direct articulation 110, 118
AI see audio introduction cognitive model 18 direct links 53, 55, 57, 58
anticipation 48, 57 cognitive processes 14, 22, 24, direct transfer 87, 102
anticipatory processes 57 36, 66, 106 directionality 102, 121
attitudinal factors 64 cognitive processing 23, 86, discourse 3, 26, 45, 46, 54, 57
audience feedback 126, 128, 87, 98 discourse integration 26
132, 140 cognitive representation 3, 21, distance dynamics 85, 86, 87,
audience reception 4, 125, 126, 23, 35 89, 91, 95, 97, 99, 100, 102
127, 128, 132, 133, 134, 139, cognitive translation act 2, 3, distancing effect 137, 140
141, 142 7, 8, 10 distributions of comprehension
audience responses 128 combinatorial nodes 29 112
audio description 4, 125, conceptual meaning 22 dual process framework 67, 68
126–129, 130, 132, 133, 134, 135, conceptual representations 22, dual process models 67, 68
136, 137, 138, 139, 140, 141 29, 38
audio introduction 129, 130, conference interpreting 3, E
135, 136 43, 49 Effort Model 12, 18, 54
audio processing system 120 control stimuli 35 ELF see English as a lingua
audio recording 110 creative intuition 68, 69 franca
audio subtitling 128, 141 cross-linguistic differences 30, encoding 22, 23, 35, 37, 55, 56, 58
audio through-description 129 31 English as a lingua franca 3, 43,
automated processing 21 cross-linguistic similarities 30, 44, 45, 46, 47, 48, 49, 51, 52,
automated translation processes 36 53, 54, 55, 56, 57, 58, 59
3, 24 crowdsourcing 131, 141 error detection 106, 107, 108,
automatic default procedure 38 112
automatic processes 25, 28, D error frequency 113, 115, 116, 117,
36, 37 decision-making 2, 3, 12, 13, 18, 118, 119
automatic translation 24, 25, 54, 63, 64, 65, 66, 67, , 70, 71, error mitigation 106, 113, 120,
38, 39 72, 73, 75, 76, 78, 79 121
automaticity 27, 36 decision-making model 18 expertise 18, 68, 76, 77, 113
decision-making process 12, extended version 129, 130
B 64, 65, 73, 75, 78 external support 106, 110
bilingual lexicon 3, 22, 39 declarative knowledge 107 eye-tracking 13, 14, 26, 112
bilingual processing 3, 22, 46 decoding 22, 23, 35, 37
default procedure 25, 38 F
C default-interventionist structure fansubbing 131, 140
CAT tools 111, 132 77 formal correspondence 25, 86
cognitive act 4, 8, 14 dehumanization 17 formative techniques 78
150 Index
fringe consciousness 73 isomorphism 9 personality measure 72

personality theory 71
G K post-drafting phase 89, 90, 91,
game-theoretical model 11 keyboard logging 26 95, 96, 97, 100, 101
grounded theory 127, 128, 141 keystroke logging 1, 14, 85, 87 preparatory AI 135, 136
preparatory version 129, 130
H L primed representations 27
Heidelberg School 45, 54, 57 lemma 22, 29, 35 primes 26, 27, 28, 30, 31, 35, 56
holistic intuition 68, 69, 70, 71, lexical errors 115, 116, 117 priming 3, 21, 24, 26, 27, 28, 29,
75, 76, 78 Likert scale 133 30, 31, 35, 37, 38
horizontal method 24, 26 linear progression 108 principle of encoding specificity
horizontal processes 37 linguistic surface structures 47 55, 56
horizontal translation 22, 23, literal translation 21, 25, 72, 74, problem awareness 106, 107,
24, 25, 26, 29, 38 85, 86, 87, 88, 94, 99 109, 110
literal translation hypothesis 3, problem description 109
I 4, 21, 24, 36, 38, 39, 85, 86, 87, problem indicators 36, 106, 107,
imagination 64, 77 89, 91, 93, 97, 98, 99, 102 108, 110, 111, 114, 118, 120
implicit procedural processes literality 27, 36 problem recognition 4, 106,
25 107, 108, 109, 112, 113, 116, 117,
incubation 69, 70, 78 M 120, 121
indicators 36, 106, 107, 108, 110, MBTI see Myers Briggs Type problem recognition monitor
111, 114, 118, 120 Indicator 120
indirect articulation 110, 118 meaning-based interpreting 55 problem reporting 107, 108, 118
individual differences 63, 70 mental lexicon 29 problem solving activity 22,
individual translating style 102 mental model 45, 47, 54 23, 37
inferential intuition 68, 69, 70, mental translation memory 43, problem solving processes 24
73, 77. 78 55, 58 problem-solving analysis 106
inferential processing 43, 69 metacognition 107, 108, 121 problem-solving behavior 75,
information processing 64, 74, metacognitive bundling 108 110
77, 78 metalinguistic knowledge 24, 26 problem-solving decisions 110
information retrieval 108, monitor model 3, 21, 24, 25, problem-solving intuition 68,
110, 111 37, 38 69
input tokens 22 monolingual process 37 problem-solving performance
Integrated Problem and monolingual processing 47 4, 105, 106
Decision Reporting 105, multisemiotic opera translation problem-solving strategies 13,
106, 109, 110, 111, 112, 113, 114, process 132 106, 108, 109
115, 116, 118, 119, 120 Myers Briggs Type Indicator 71 procedural knowledge 107, 108
interference 12, 38, 110, 139 procedural strategy 108
interim solutions 10, 16, 86 N process awareness 107, 112, 121
interpreter competence 58 neural activity 23 process knowledge 107
interpreter training 46, 58, 87 neural networks 117 process protocol 4, 105, 106,
intersemiotic translation 127, neurofunctional model 37 108, 110, 112, 114, 116, 117, 118,
130 nexus model 17 120, 121
intuition 3, 11, 63, 64, 65, 66, 67, non-automatic process 25 process-oriented translator
68, 69, 70, 71, 72, 73, 74, 75, training 4, 105, 106, 107, 108,
76, 77, 78, 79 O 111, 112
intuition processing 67, 73, omissions 49, 52, 115, 152 product-based analysis 49
74, 78 opera accessibility 125, 128, 141 propositional text base 47
intuition research 73 opera translation 125, 126, 127,
intuitive processes 65, 67, 76, 79 128, 132, 142 R
intuitive translator behavior 76 re-reading 26
IPDR see Integrated Problem P recall prompt 32
and Decision Reporting parallel processing 13, 120 recorded articulations 118
Index 151
recorded verbalizations 105, self-reflection 107, 108, 109, 113, transcoding 37, 55
106, 108, 110, 111, 112, 118, 114, 115, 116, 117, 118, 119, 121 transfer 11, 12, 26, 46, 54, 55, 87,
119, 120 self-report personality test 70 102, 106, 107, 108, 112, 115, 126
recursive model of translation self-revision 4, 16, 85, 87, 88, 89, transfer links 43, 54, 59
3, 21, 24, 35, 37 90, 91, 92, 93, 94, 95, 96, 97, transfer processing 56
reflex-like equivalence relations 98, 99, 100, 101, 102 transference 21, 24, 29
55 semantic representations 24, 28 translation act 2, 3, 7, 8, 10, 11,
regeneration 27, 29, 37 semi-experts 78 13, 14, 15, 16, 64, 67, 71
regeneration hypothesis 26, 27 sensitivity 75, 76 translation brief 31, 88
regressions 26 sensory (communication) translation competence 14,
retrieval 13, 27, 43, 55, 56, 57, 58, channel 127, 130 66, 72
108, 110, 111 shared representations 21, 23, translation event 2, 3, 4, 7, 8, 9,
retrospection 106, 108, 116, 24, 25, 26, 27, 29, 30, 31, 34, 14, 15, 16, 17, 18
117, 120 36, 37, 38, 39 translation model 7, 22, 24
retrospective analysis 111, 118, sight translation 12 translation practice 2, 7, 17,
119 sign interpreting 132, 141 113, 142
retrospective interview 15, 49, situated cognition 8 translation problem 4, 7, 10, 11,
50, 51, 52, 53 Skopos Theory 16 12, 13, 23, 66, 76, 85, 101
retrospective methods 18 sociological (translation) event translation purpose 31
retrospective self-reflection 2, 3, 4, 7, 8, 10, 15 translation routines 55
107, 108, 119 solution evaluation 108 translation strategies 23, 134
reverse engineering (models) somatic response 65, 73, 74, 78 translational conflict 54
12, 15, 18, 108 sources of disturbance 107 translator training 4, 14, 18, 105,
reverse-engineered processes speech disfluencies 12, 110, 118 106, 107, 108, 111, 112
10, 12, 15 standalone version 129, 130 Translog 4, 13, 88
revision 4, 9, 16, 17, 86, 88, 89, strategy execution 107
90, 91, 92, 93, 94, 95, 96, 97, strategy observation 110 U
98, 99, 100, 102, 105, 106, 110, stylistic errors 115, 118 unjustified cognitive activity 70
111, 112, 114, 117, 121 stylistic problems 118
revision behavior 108, 111, 120 subliminal skills 78 V
revision process 16, 91, 92, 93, surtitles 132 verbal reports 79, 110
95, 100 syntactic error 115, 117 verbalization 13, 63, 64, 73, 74,
revision tool 105, 106, 111, 112, syntactic representations 24 75, 76, 105, 106, 108, 110, 111,
116, 117, 120, 121 112, 118, 119, 120
Rich Points 114 T verbatim representation 24,
routine transfer links 43 T-move 89, 92, 93, 94, 95, 96, 27, 29
97, 98, 100, 101 vertical method 26
S TAP see think-aloud protocol vertical model 22, 36
S-move 89, 92, 93, 94, 95, 96, Theory of Translational Action vertical processes 37, 38
97, 98, 99, 100, 101, 102 16 vertical translation 22, 23, 24
screen recording 4, 105, 106, think-aloud protocol 8, 13, 23, virtual processes 10, 11, 15
111, 112, 113, 114, 116, 117, 118, 36, 63, 71, 72, 74, 79, 82, 83, visual attention 112, 120
119, 120, 121 110, 113 visual interference 12
segmentation 57 think-aloud recording 114 visual processing system 117
segmenting tendencies 112 through-description 129, 136
self-analysis 105, 106, 112, 114 ToTo see touch tour W
self-awareness 111 touch tour 125, 126, 127, 128, working memory 37, 67
self-correction 89 129, 130, 131, 132, 133, 134, 135,
self-projection 76 137, 138, 139, 140, 141

Ehrensberger-Dow... (Et Al) - Describing Cognitive Processes in Translation Acts and Events (Benjamins, 2015)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ehrensberger-Dow... (Et Al) - Describing Cognitive Processes in Translation Acts and Events (Benjamins, 2015)

Uploaded by

Copyright:

Available Formats

Describing Cognitive Processes in Translation

Benjamins Current Topics

Birgitta Englund Dimitrova

John Benjamins Publishing Company

the American National Standard for Information Sciences – Permanence

© 2015 – John Benjamins B.V.

Notes on editors 147

In the translation process, translators and interpreters have a central position as

Abdallah, Kristiina. 2010. “Translators’ Agency in Production Networks.” In Translators’ Agency,

Toury (1995; 2012) distinguishes between cognitive translation acts on the

Keywords: translation process, translation model, translation act, translation

1. Act and event

2. Models and problems

The term “model” is significantly polysemous in the philosophy of science, where

3. Models of virtual processes

Along similar lines, but focusing on a particular translation problem (in

4. Models of reverse-engineered processes

Corresponding to Toury’s second sense of ‘problem,’ there are models of what I

5. Models of actual processes

6. Relations between types of models

7. Some models of the translation event

1. Make a literal version of the poem.

Revision by others obviously occurs at the post-writing phase. Bly’s model

8. The translation practice?

In addition to cognitive and sociological processes, we can also see Translation

competence or expertise, in longitudinal developmental studies (e.g., Schmidt

Kussmaul, Paul. 2007. Verstehen und Übersetzen. Tübingen: Gunter Narr.

Moritz Schaeffer and Michael Carl1

The purpose of the present chapter is to investigate automated processing during

Keywords: literal translation, priming, monitor model

They pushed the boat Sie schoben das Boot

Figure 1. Horizontal and vertical translation

2. The literal translation hypothesis and the monitor model

In equivalence theories such as that by Catford (1965), sameness is defined on the

formal correspondence, dictated by structural and other typological differences

We understand the term automatic in Tirkkonen-Condit (2005) as referring to

4. The regeneration hypothesis (Potter and Lombardi 1990; Lombardi and

5. Recall as a measure of monolingual priming

7. Semantic bilingual priming

In semantic cross-linguistic priming studies, participants are presented with primes

et al. 2011). These studies used a variety of language combinations: Spanish-English,

8. The shared syntax account (Hartsuiker et al. 2004)

9. Syntactic bilingual priming

prime sentences with a prepositional object, participants were significantly more

10. Priming in translation

Participants in priming studies are normally assumed to be unaware of the effects

Source text He put the newspaper inside the trousers.

Recall He put the newspapers inside his trousers.

Figure 2. Lexical score example 1

Recall He went over to wake the girl in their little house.

Correct to wake the girl He went over their little house.

Figure 3. Lexical score example 2

original original translation recall recall

They pushed the boat Sie schoben das Boot

Figure 6. Recursive model of translation

is untranslatable in this sense requires considerations of acceptability and this is

11. The revised literal translation hypothesis

for both automatic processes and problem-solving activity. We understand hori-

12. Recursive model of translation: the revised monitor model

is raised so as to avoid interference… However, it is not raised so high that it could

Al-Kufaishi, Adil. 2011. “Obligatory Translation Shift as a Sub-component of a Model of Quality

Dragsted, Barbara. 2010. “Coordination of Reading and Writing Processes in Translation: An

At the crossroads of English as a lingua franca (ELF) research and interpreting

the American National Standard for Information Sciences – Permanence

Figure 1. Horizontal and vertical translation

Figure 2. Lexical score example 1

Figure 3. Lexical score example 2

Figure 6. Recursive model of translation

Table 1. ELF-related sources of interpreting difficulties

Table 2. Example of incoherent input and resulting interpretation (Dschulnigg and

Table 4. Example of comprehensible but unconventional input and resulting interpreta-

You are translating along, and stumble briefly on a word. “What

Figure 1. Dual processing activity

Table 1. Number of self-revisions, interpreters vs. translators