O’ZBEKISTON RESPUBLIKASI

OLIY TA’LIM, FAN VA INNOVATSIYALAR VAZIRLIGI

URGANCH DAVLAT UNIVERSITETI
XORIJIY FILOLOGIYA FAKULTETI
TARJIM NAZARIYASI VA AMALIYOTI KAFEDRALARI
GUIDE HAMROHLIGI VA TARJIMONLIK FAOLIYATI
GURUHI
III BOSQICH TALABASI
______________________________________NING
TARJIMA NAZARIYASI VA AMALIYOTI FANIDAN

KURS ISHI
Theme: Translation as a process

Ilmiy rahbar: Saparboyeva Gulandom

URGENCH 2024
CONTENTS

1
INTRODUCTION………………………………...………..…………………......3
CHAPTER I. THE PHASES OF THE TRANSLATION PROCESS…………
6
1.1 Stages of
translation…………………………………………………………….6
1.2 The proof of the translation process is in the reading of the target
text………..11
CHAPTER II. TRANSLATION PROCESS AND PROBLEM OF
TRANSLATION IN WORLD
CLASSICS……………………………………..19
2.1 A Systems Theory Perspective on the Translation Process……………………
19
2.2 The translation process: an analysis of observational
methodology…………..26
CONCLUSION…………………………………………………………….…….31
THE LIST OF USED LITERATURE………………………….………………32

2
 INTRODUCTION

Translation is a process of converting messages, thoughts, feelings, orders or

any other verbal utterances expressed in one language into messages of the same
meaning and value in another language. Different languages use different names to
label this process, but all of these names indicate that this process consists of
‘conversion’ and ‘transmission’ of messages to people who do not use the source
language system and cannot understand the message coded without the processes
of ‘conversion’ and ‘transmission’ into the system that they use for
communication. It is highly important to be aware of the fact that what is translated
are not simply the elements of one system, being replaced with the corresponding
elements of another language system – what is, or should be, translated is the
expressed message. For the message to be translated into another language, or the
target language, it must be expressed in the source language, and it must remain
the same in both languages; the receivers of the translated message must receive
the same content as the receivers of the original message. Now, what I have
explained up to this point is that translation is a process.
However, another meaning implied in the term translation refers to the
product itself – the translated material or the concrete translation product which is
produced by the translator. In his book ‘Becoming a Translator’, Douglas
Robinson states the following: ‘Translation is different things for different groups
of people. For those who are not translators, it is primarily a text; for people who
are, it is primarily an activity.’ In order to fully understand this concept, we should
consider the following distinction. The distinction between these two is also drawn
out by the following definition found in Dictionary of Translation Studies:
Translation – An incredibly broad notion which can be understood in many
different ways. For example, one may talk of translation as a process or a product,
and identify such sub-types as literary translation, technical translation, subtitling
and machine translation; moreover, while more typically it just refers to the
transfer of written texts, the term sometimes also includes interpreting.

3
 It is very important to understand that in the history of translation, oral
translation appeared before written translation, primarily because of the fact that
many languages were spoken and not written. These two types of translation carry
the significance for the development of human culture and civilization in general.
Due to the never-ending development of modern technologies, it is highly
important to mention the machine-aided translation or simply machine translation,
although the results of such translations are not considered to be reliable or
completely accurate because of the complexity of language and especially the
study of word meaning – semantics.
However, although machine translation cannot be developed to such a great
extent due to many factors, the use of computer-aided tools or CAT which help
translators in translation activities should not be neglected. In short words, the
basic, but not the only concept for making a division of translation can be based
with respect to the type of the translator and can primarily be human or machine
translation, then, it can be written or oral (also referred to as interpreting, which
can be further divided into consecutive and simultaneous), and at last, but not least,
according to the type of the translation material, it is possible to consider
translation to be literary or non-literary.
Actuality of theme. At the present time it is important to learn Translation
as a process and its features.
The object of the course work is our scientific research about Translation
as a process that used to generate an in-depth, multi-faceted understanding of a
complex issue in its real-life context.
The subject of the work of the analyzed text implies the following scenes:
It is an established research design that is used extensively in a wide variety of
disciplines, particularly in the social sciences.
Aim of research. The aim of our scientific research is telling about What
Translation as a process is, it is the method of conveying a thought or message to a
reader using the correct way.
The objectives of this paper are:

4
 1. Identify types of Translation as a process;
2. Identify elements of it;
3. Discuss the qualities of Translation as a process; logical, language, and
style;
4. Analyze and write it.
In the course paper following methods were used to the first steps when
Translation as a process methodology are the same as for other types of research.
The scientific basis for this work was taken from some kinds of articles.
The practical value of the research work. Writing down a Translation as a
process is important as it helps you better understand the topic, better organizes
your thoughts, prevents any mistakes in your method, helping you become more
familiar with it, and boosts its overall quality.
Structurally the research work consists of
• Introduction
• Chapter 1.
• Chapter 2.
• Conclusion
• List of used literature
The introduction discovers information about the basics of writing various
types of speeches. It discusses language, style, and logic as well as presents
selected speeches by various speakers locally and internationally.
The first chapter presents information about Translation as a process. It is a
crucial skill to have in the modern information age.
The second chapter describes information about learning more about
Translation as a process in English language and how you can help your students
become experts in FLT.
Conclusion summarizes all information of the research.
The list of used literature presents information about the books which were
used during investigating.

5
 CHAPTER I. THE PHASES OF THE TRANSLATION PROCESS
1.1 Stages of translation

The phases of the translation process are a main topic in translation process
research (TPR) and translation scholars (e.g. Mossop, 2000; Jakobsen, 2002;
Englund Dimitrova, 2005) generally concur that the process is divided into three
clear-cut phases. However, different authors employ different terminology for the
same three phases, albeit defining the three phases very similarly.
For example, Jakobsen (2002) applies
i) orientation,
ii) drafting and
iii) revision, while Mossop (2000) employs
i) pre-drafting,
ii) drafting and
iii) postdrafting.
The current study utilises Mossop’s terminology and delineates the three
phases as follows: in the first phase, the translator familiarises himself/herself with
the source text (ST); a full version of the translation is produced in the second
phase; and the third phase begins “after sentence-by-sentence drafting is complete”
(Mossop, 2000 p.40). Translation Studies (TS) distinguishes between revising
one’s own work and revising other translators’ work, termed by Mossop “self-
revision” (2001/2010) and “other-revision” (2007) respectively. In self-revision
translators correct their own work, while in other-revision, the translation is revised
by a third party. In this study, the translation process is considered as ceasing when
the role of third parties begins, hence at the end of the post-drafting phase (Figure
1).

6
 Some translators jot down words/phrases and/or carry out research, some
read the whole ST, others browse it quickly, whereas certain translators start
translating immediately, skipping this phase altogether. In Phase 2, the translation
is drafted but this is not the only action taken. Translators read the ST (again) and
engage with it, resort to external resources such as dictionaries and the Internet,
evaluate the emerging text and self-revise it. Most studies have found the second
phase to be the longest (e.g. Jakobsen, 2002; Englund Dimitrova, 2005; Dragsted
and Carl, 2013) but individual differences and exceptions were reported here too in
terms of duration, approach and activities performed.1
In the third phase, translators review and/or hone the translation. Self-
revision, however, is not restricted to this phase but often takes place both in the
drafting and the post-drafting phases. Moreover, “the post-drafting phase can
include non-checking work such as term research” (Mossop, 2001/2010, p.168)
and some decisions could be postponed until this phase. Still, “the main activity is
monitoring of existing text” (Jakobsen, 2002, p.193). Some translators choose to
subdivide this phase in various subphases (Englund Dimitrova, 2005), which
begins when the translator finishes a first draft of the translation and ends when
1
The Phases of the Translation Process: Are They Always Three? Claudine Borg claudine.borg@um.edu.mt

7
s/he considers the translation complete (Jakobsen, 2002; Englund Dimitrova,
2005). It will be interesting to find out what the translation process of this study’s
participant is composed of, and whether it is also divided into three phases.
This case-study forms part of a larger research project (Borg, 2016) which
investigated in-depth the coming-into-being of a Maltese literary translation: Is-Sur
Ibrahim u l-Fjuri fil-Koran (henceforth Is-Sur Ibrahim) rendered from French by
Toni Aquilina (Schmitt, 2014). It collected data through draft translations, the ST
and the final TT, think-aloud, observation, and interviews. The data elicited were
triangulated and analysed qualitatively and quantitatively.
The present study focuses on the phases of the translation process and
analyses pertinent data extracted from Drafts 1-4 (D1-4), the initial interview,
think-aloud and observations (for details on the methodology see Borg, 2016).
Since D1 of Is-Sur Ibrahim was completed in 2008, prior to the start of the
research project, data about Phases 1 and 2 were obtained through an initial
interview, in which the translator described all the phases of his translation
process; the results are presented in section 4.1. The researcher then observed the
translator as he self-revised D1 and D2 while thinking aloud, corresponding to
Phases 3 and 41 respectively; these data were recorded in Translation Process
Protocols (TPPs). Data about Phase 5 were gathered through a telephone call and
analysis of D4. These results are found in section 4.2.2
The translator immediately asserted that he has his own translation method:
“I always follow my method” (ISSI/TA047). He perceives his translation process
in terms of phases, each one of which is analysed below:
Phase 1
For Aquilina, the first step in translating a literary work consists of reading
and understanding the text. He first reads the whole text, as many times as required
to get a full grip of the work. A thorough understanding of the text to be translated
and a complete grasp of the meaning of words in context are considered crucial.
During this phase, he also carries out research which includes background reading

2
The Phases of the Translation Process: Are They Always Three? Claudine Borg claudine.borg@um.edu.mt

8
about the author, reading other works by the same author and searching words in
dictionaries. Only once he is confident of having achieved a good grasp of the text,
will he move to the next phase.
Phase 2
The next step consists of drafting the translation. First, he divides the text
into what he calls units of work. This involves segmenting the ST in chunks
comprising between five and eight pages, based on the amount of time available
for translation. A unit of work is tackled in one go. During this phase he works
regularly on the literary translation at hand until D1 is completed, in order to
maintain the impetus gained as well as to capitalise on the research done in the
previous phase, while this is still fresh in his mind.
No recourse to dictionaries is made during drafting, in order to produce the
draft swiftly, in an uninterrupted flow. During the production of the draft, he does
not necessarily look at the research undertaken because he gets carried away. It
could be argued that in the phase preceding drafting he internalises the research
done, stocking up on his internal resources so that the draft TT is produced in a
natural burst. Once a paragraph is translated, he rereads it and checks whether it
can be improved. However, he clarified that changes done at this phase are minor.
D1 is not intended to be perfect. In D1 he is not concerned with details such as
spelling; optimisation of the TT is reserved for later.
Phase 3
Aquilina asserts that he likes to leave an interval between this phase and the
previous one as it allows him to approach the draft with fresh eyes. Here, the
translation is typed and fine-tuned and he sets himself no targets and time limits. In
fact, he explained that this phase normally takes longer than the preceding one.
Fine-tuning the TT involves carrying out minor and not so minor changes. At this
stage, the focus is on improving the text. Use of dictionaries and other external
resources resurfaces here. Interestingly, he envisages this phase not as the revision
stage, but as the stage where he improves the writing.
Phase 4

9
 Now, he focuses on the TT, enhancing and polishing it, making it sound as
natural as possible, ensuring that it works on its own. All this is done with the
potential reader in mind and thereby for acceptability reasons. Reference to the ST
is made at this stage too, in order to quickly eliminate any remaining doubts.
Phases 5
Phase 5 is only discussed fleetingly; it is nonetheless referred to as a phase.
At this point the translator takes on the role of a proofreader and endeavours to
spot as many mistakes as possible before the TT is passed on to the actual
proofreader. Translating Is-Sur Ibrahim (Phases 1-2) D1 of Is-Sur Ibrahim was
created while on holiday in France. He had already read the book in Malta and
reread it during this vacation. He estimates having worked on Phases 1 and 2 for
circa ten days. As part of the groundwork, he read four or five other books by
Schmitt to acquaint himself with the author’s writing style. He affirms having
adopted his normal translation process for the production of D1.3

3
The Phases of the Translation Process: Are They Always Three? Claudine Borg claudine.borg@um.edu.mt

10
 1.2 The proof of the translation process is in the reading of the target text

According to Chesterman (1998), translation is done to bring effects on

individuals and society. The primary effects are cognitive effects referring to the
change in the mental or emotional state of the reader who reads a translated text.
The secondary effects are behavioural in nature. As a consequence of reading a
translation the reader might implement some changes in the behaviour, knowledge,
actions, and aesthetic experiences. The tertiary effects reach further and expand to
the entire target culture, the intercultural relationship between the source and target
culture and affect society at large. These theoretical assumptions, although very
appealing, have rarely been empirically validated.
Walker (2021a) in his article, “Investigating How We Read Translations: A
Call to Action for Experimental Studies of Translation Reception”, points out the
imbalance between a robust body of research into how translations are produced
using experimental study designs and data collection tools including keylogging
and eyetracking, and the lack of empirical investigations into how translated texts
are received and experienced by readers.
Over the last thirty years, TPR has provided empirically grounded findings
to demonstrate the complexity of the cognitive processes in the translator's mind
(Xiao and Muñoz, 2021). Keylogging programmes (Translog, Inputlog) as data
collection tools allowed researchers to see how the target text is produced in real
time without engaging the translator in an additional task in a way Think Aloud
Protocols did (Jakobsen, 2003). Keylogging studies provided evidence for the
phases of the translation process – orientation, drafting and revision (Jakobsen,
2003), different translation styles (Dragsted and Carl, 2013), and the impact of
translation expertise on the decision-making processes, among others. The deeper
understanding of the complexity of the translation process was further expanded
thanks to the use of eyetracking and studies looking at how translators interact with
the source and target texts as well as with information sources, and how they use
the available technology to increase productivity.
11
 Much less effort has been devoted to testing the relationship between the
translation process and the quality of the end product or the way it is read and
comprehended by the target reader. Kruger and Kruger (2017) review the few
studies which explored the effect the translator's choices have on the target
audience (Kenesei, 2010; Moran and Perego, 2012; Puurtinen et al., 1994; Rojo et
al., 2014; Zasyekin, 2010) and conclude that, “there is clear evidence that the way
in which translators choose to translate a text affects the kinds of cognitive
representations that readers subsequently construct from the linguistic features …”.
For example, Zeven and Dorst (2021) argue that the linguistic choices in two
Dutch translations of The Great Gatsby influence the way the fictional character –
Daisy Buchanan is perceived by the reader. With the loss of the author's ambiguity
which encourages the readers to form their own opinion, the translators of the two
versions which are being compared offer a biased characterisation and encourage
potential gender stereotyping by readers. This interesting comparative analysis of
three narrative elements (theme, setting, and point of view) and theory-driven
conclusions have not been validated empirically but they could be. As pointed out
by Kruger and Kruger (2017), “studies of the cognitive dimensions of producing
and receiving translated materials have developed largely separately, with few
attempts to connect the two processes.”4
This chapter is an attempt to bring together the translation process and the
reception of the end product – the translated text. The first hurdle in these
empirical explorations of the production-reception crossover is that the reader
interacts with the product. What we know about the translation process was very
often driven by our curiosity about how and why particular decisions are made,
and which are cognitively more demanding than others. Much less attention has
been paid to the relationship between the translator's cognitive effort and the
quality of the final translation version, that is the text which is read by the target
readers. Very few studies have included any evaluation of the end product or

4
The proof of the translation process is in the reading of the target text: An eyetracking reception study
Author links open overlay panelBogusława Whyatt, Olga Witczak, Ewa Tomczak-Łukaszewska, Olha Lehka-Paul

12
triangulation of the process and product data (Carl and Buch-Kromann, 2010;
Hubscher-Davidson, 2009; Lehka-Paul, 2020).
Translation quality is a complex concept. As argued by Kotze et al. (2021),
the individual and social expectations of what constitutes a good translation are
shaped by norms defined as “cognitive-evaluative templates”. These normative
constructs have conceptual, evaluative, and affective dimensions which shape the
translator's choices and the reader's expectations when interacting with translated
texts. Still, translation quality is relative to the purpose of the target text and
remains difficult to measure. One of the ways to operationalise it is to measure the
equivalent effect a translated text has on the reader.
To date, research into eye movements in Translation Studies (TS) and in
Cognitive Translation and Interpreting Studies (CTIS) has mainly been reported in
translation process research where eyetracking measures (fixation count and
duration) were used as correlates of the increased cognitive effort when reading for
translation (Alves et al., 2011; Jakobsen and Jensen, 2008), evidence for parallel
processing of the source and target language (Schaeffer et al., 2016), coordination
of reading and writing (Dragsted, 2010; Hvelplund, 2017), or the use of on-line
resources (Whyatt et al., 2021).
Very few studies have used eyetracking to investigate the effect the
translator's choices have on the readers outside the research in AVT into how
subtitles are read by viewers (Gambier, 2018; Gerber-Morón and Szarkowska,
2018; Hu et al., 2020; Orrego-Carmona, 2016). Kruger (2013) and Walker (2019,
2021b) are known examples of studies which applied eyetracking to explore the
effect of the translation strategy on the readers of translated literary texts. Both
studies measured the reader's cognitive effort when processing selected phrases in
the translations.5
Kruger (2013) used eye movements to investigate how Afrikaans children
and adult readers processed and responded to foreignised and domesticated textual
elements in children's picture books translated from English into Afrikaans. The
5
The proof of the translation process is in the reading of the target text: An eyetracking reception study
Author links open overlay panelBogusława Whyatt, Olga Witczak, Ewa Tomczak-Łukaszewska, Olha Lehka-Paul

13
specific research questions asked whether foreignised elements affect the cognitive
effort and comprehension scores of both reader groups. The eyetracking measures
included first fixation duration, dwell time, fixation count, and glances count for
areas of interest (AOIs) corresponding to domesticated and foreignised textual
elements. The effect on comprehension was checked by short questionnaires (adult
readers) and interviews (child readers). The author concluded that, “the findings of
the experiment demonstrate that while there are perceptible effects on processing
and comprehension associated with the use of foreignising strategies, these effects
are not straightforward or uniform, with notable differences not only for different
AOIs, but also for child and adult readers” (Kruger, 2013).
Walker (2019, 2021b) used eye movements to compare the reading
experience of source language (SL) readers who read extracts from a novel Zazie
dans le métro originally written in French by Raymond Queneau in 1959 with the
reading experience of target language (TL) readers who read the same extracts
from the English translation – Zazie in the Metro translated by Barbara Wright in
1960. Using eye movements as correlates of cognitive effort, Walker focused on
the cognitive equivalent effect on the TL readers. This expectation is in line with
key concepts from literary reception studies of narrative engagement and the
sensations on the reader activated by the reading process (Scott, 2012) –
Chesterman's (Chesterman, 1998) primary effects.
Similar to Kruger (2013), the effect on the TL reader is measured with
respect to specific stylistic devices which are quasi-phonetic forms (‘concertina
words’). Processing such stylistic elements required substantial effort from the SL
readers. Walker assumed that if the TL readers experienced comparable effort
when reading the English translation of these concertina-words then the translator
managed to create an equivalent effect. Native French speakers (N17) read four
French extracts ranging from 700 to 1500 words and English speakers (N14) read
the same extracts translated into English while their eye movements were recorded.
Three parameters were analysed – first fixation duration, gaze duration and total
fixation duration in AOIs which included altogether 7 concertina-words in the SL

14
text – 4 of them were rendered “in a stylistically ‘equivalent’ manner in the TT,
and three of which have not”. The results showed that the effect on the TL reader
was similar to that on the SL reader in terms of comparable eye movement
measures only if the translator had applied orthographic manipulation to challenge
the reader's imagination.
Both studies show that eye movement data are a valid measure of the effect
translator's decisions have on the readers of translated texts. In both studies the
reader's reactions are measured in response to the final well-groomed product – a
published literary translation.
In the study reported here, we go beyond measuring reactions to selected
phrases in the translated texts. We want to show the process in which the
translator's decisions are made, the end product of these decisions and the process
of receiving the target text by the reader. In particular, we want to explore whether
the effort invested by the reader when reading the translated text in any way relates
to the cognitive effort invested by the translator.
At first glance, the cognitive effort of the translator and the reader seems
disproportional, but we really would like to know if the translator's effort pays off
when the text is being read – well translated coherent text should read with ease.
Taking into consideration the measures of the cognitive effort of the translators and
readers, we take the target text as a point of departure – cognitive effort is needed
to produce it and to read it.6
The first methodological challenge we face is finding a measure of cognitive
effort that could reliably capture the effort invested by the translator and the reader
of the translation. The concept of cognitive effort defined as the amount of mental
resources invested in the task of producing a translated text has been central to
TPR (Krings, 2001; Kruger, 2016; O'Brien, 2011; Lacruz, 2017; Vieira, 2014). The
cognitive effort of translators has been operationalised depending on the method
chosen to collect the data. In keylogging studies, following Jakobsen (2014)
assumption that there is a correlation between problem-processing and effort
6
The proof of the translation process is in the reading of the target text: An eyetracking reception study
Author links open overlay panelBogusława Whyatt, Olga Witczak, Ewa Tomczak-Łukaszewska, Olha Lehka-Paul

15
duration, time on task was taken as a proxy for cognitive effort with other
measures including number of long pauses, typing speed and the use of online
resources, among others (Hvelplund, 2019; Koponen et al., 2012; O'Brien, 2006).
In eyetracking studies, the translator's effort was measured mostly by total fixation
duration, fixation count, and duration (Jakobsen and Jensen, 2008; Schaeffer et al.,
2016; Hvelplund, 2019). There is a substantial body of evidence that the increase
in cognitive effort can be related to the features of the source text (lexico-semantic,
stylistic, and syntactic complexity) and the level of translation expertise – well-
seasoned translators capitalise on their experience. Another factor which is known
to contribute to the effort invested in producing a translation is time pressure (Carl
and Kay, 2011; Rojo López et al., 2021).
The reader's effort can be indexed by the reading rate – number of words per
minute (wpm). Brysbaert (2019) conducted a meta-analysis of 190 studies based
on 18,573 participants and estimated that the average silent reading rate for
educated adults in English is 238 wpm for non-fiction (the range is 175–300 wpm)
and 260 wpm for fiction (range: 200–320 wpm). Fiction is read faster because the
words used are usually high in frequency, shorter, and the sentence structures are
less complex than in non-fiction texts which are likely to include fewer known
words, specialist terminology and more sophisticated syntax. The use of the
eyetracking technology has brought more fine-grained measures of the reader's
cognitive effort when reading for comprehension. The early reading measures
(first-run viewing) reflecting the word recognition processes are indexed by first
fixation duration and gaze duration (time spent viewing a word until a saccadic
movement is made to fixate on another word) and are related to lexical access
(mapping the written word with its meaning stored in the long term memory). Late
reading measures refer to comprehension-monitoring processes which kick in
when more time is needed for meaning integration and are indexed by regressions
to the previously read words and second- and third-run dwell time – re-reading or
re-viewing for re-processing (Inhoff et al., 2019; Rayner, 1998). A global measure

16
of cognitive effort can be also indexed by total fixation duration (dwell time) and
fixation count.7
Numerous studies have shown that more frequent words as well as shorter
words are processed faster than less frequent and longer words, and words which
are highly predictable from the context need very short fixations or are skipped by
the eyes (Ehrlich and Rayner, 1981; Juhasz et al., 2008; Rayner and Duffy, 1986;
Staub, 2015). Short words, if not skipped, are fixated once and at the preferred-
viewing locations (PVL) usually to the left of the word's centre (Rayner, 1978).
Longer words, if fixated twice, will receive the first fixation near their beginning
and the second near the end. Rayner and Well (1996) found that highly predictable
words (86% cloze probability) were read faster than words with a medium level of
contextual probability (41%), whereas low-predictable words (4%) were read more
slowly. Additionally, factors which slow down the reading process include
semantic and syntactic ambiguity (Rayner and Duffy, 1986; Frazier and Rayner,
1982; Wiley and Rayner, 2000) and inconsistency with what has been read or with
what the readers know.
The fine-grained knowledge of the reading process comes from studies
which focused on single word recognition and sentence reading. In whole-text
reading, the reader's effort is modulated by the linguistic features of the text –
lexico-semantic, stylistic, and syntactic complexity, information density, and the
reader's language skills, cognitive abilities including executive control, working
memory, and knowledge of the theme.
It seems that the cognitive effort of both translators and readers of the
translated text is dependent on the features of the text and on their own
contribution in terms of knowledge and skills derived from their experience with
texts. It is tacitly assumed that good translators are aware of their readers’
expectations and communicative needs (Apfelthaler, 2014; Shreve, 2009), and
therefore they should be able to predict the readability of the translation they
consider good enough to be read by the target reader. Translators take much longer
7
The proof of the translation process is in the reading of the target text: An eyetracking reception study
Author links open overlay panelBogusława Whyatt, Olga Witczak, Ewa Tomczak-Łukaszewska, Olha Lehka-Paul

17
to produce a translation which a reader will take a minute to read. Therefore, to
find a common denominator for both, we calculate the cognitive effort of the
translator as the total time needed to produce the translation divided by the number
of characters with spaces in the target text. This has pragmatic relevance –
translators are paid for the text they have produced. This simple formula gives us a
character-adjusted measure of effort for the translation process to overcome the
differences in the length of words and sentences.
In a parallel fashion, we derive a character-adjusted measure of the reader's
effort – dwell time per character with spaces to mirror the translator's effort.
Although we know that when reading, we do not process every letter, we also
know that the length of words is a strong predictor of the cognitive effort needed to
process them (Brysbaert, 2019). Character-adjusted measures of effort have been
used by Kruger (2013) and Walker (2019) in translation reception studies and by
Hyönä and Niemi (1990) in whole text reading research. O’Brien (2010) analysed
14 data sets in an eyetracking study investigating the effect of controlled language
rules on the readability of texts and noted that looking at “fixation count as a
function of characters per text” was a more accurate measure because the
experimental texts differed in the number of words and characters.8

8
The proof of the translation process is in the reading of the target text: An eyetracking reception study
Author links open overlay panelBogusława Whyatt, Olga Witczak, Ewa Tomczak-Łukaszewska, Olha Lehka-Paul

18
 CHAPTER II. TRANSLATION PROCESS AND PROBLEM OF
TRANSLATION IN WORLD CLASSICS
2.1 A Systems Theory Perspective on the Translation Process

Within translation and bilingualism studies, a number of models have been

proposed aiming to explain how the bilingual mind learns, stores, and produces
translations (for an overview, see Schwieter 2017). While some scholars in
translation studies promote a stratificational model of comprehension-transfer-
generation, there seems to be a consensus in the bilingualism community that
lexical access is non-selective, where source and target items are activated at the
same time.
As Halverson (2003, 210) points out, the dispute is whether “the conceptual
layer contains one common store or separate stores for each language, whether the
conceptual level is conceived of as “local” or as “globally distributed”, whether or
not there are connections between the lexical elements of the two languages, and
whether or not there are direct connections between the lexical elements of L2 and
the conceptual level”.
Some of these models assume that translation knowledge is distributed over
several nodes which are shared across the source language (SL) and the target
language (TL) and in which the meanings of words (De Groot 1992)—and/or their
grammatical properties (Hartsuiker, Pickering & Veltkamp 2004)—are represented
in form of distributed nodes. These models aim at explaining why some words and
expressions are easier to translate than others, assuming that a larger overlap and
stronger connections of representations between the SL and the TL items also
facilitate their translation.9
Other models such as BIA+ (Dijkstra & van Heuven 2002) and Multilink
(Dijkstra et al 2018) attribute lexical decision and retrieval times to the non-
selectivity of lexical access: “word recognition processes involve simultaneous
activation of words in both lexicons, followed by competition between words with
9
A Systems Theory Perspective on the Translation Process Michael Carl, Andrew Tonge, Isabel Lacruz Kent State
University, Ohio, USA

19
similar orthographies or semantics, and lastly, inhibitory processing in order to
assist with selection of the appropriate choice”. (Basnight-Brown 2014, 15) While
these models are mainly based on translation experiments of single words
presented out of context, the basic findings can also be confirmed for sentence
translations in context. Schaeffer et al (2016) and Carl & Schaeffer (2017), for
instance, find that the first fixation duration on source text (ST) words significantly
increases if they allow for many different translations, as compared to ST words
that have only few possible translations. This is in line with the non-selective
hypothesis of lexical activation, which predicts longer reaction times for words that
activate a large set of competitors.
However, there is no clear understanding how “competition between words
with similar orthographies or semantics” could be modelled in sentence translation.
In this paper we address how translators cope with several simultaneously
activated word translation systems as a sequence of words (e.g., a sentence) is
translated, and discuss the underlying mechanism that integrates competing
translation solutions. We will take a systems theory and information philosophy
approach to address the questions of how translators arrive from a set of non-
selective activated word translation systems to a final coherent translation.
We argue that word translation systems are open, dissipative systems, which
possess different amounts of internal structure (entropy), and which are
surprisingly constant across different languages. We introduce the notion of word
translation entropy which indicates the amount of organization in activated word
translation systems. Dissipative systems are integrated into dissipative structures,
which follow certain dynamic patterns, and which emerge on various levels of
linguistic and other description, which then lead to the translation of a coherent
piece of text. In the process of sentence translation, some components of the
activated word translation systems are bound together (or selected) through a
process of entropic gravity, just as physical systems are bound together by the
force of gravity acting on the matter that comprises them. Interestingly, physicists

20
(Verlinde 2011) have proposed to explain how gravity arises through quantum
information theory.
In this view, gravity is an entropic force that derives from "information
associated with the positions of material bodies" and that tends to lower (statistical)
entropy. We show that this process is compatible with models of bilingual
activation, in particular the BIA+ and the Multilink model (Dijkstra & van Heuven
2002; Dijkstra et al 2018). We undertake a comparative study using a variety of
language pairs and different measures of entropy based on existing data from
previous studies by several researchers. Results across very different languages are
strikingly similar. This suggests that the framework we introduce in this paper is
sufficiently promising to merit further study and refinement. We conclude with an
information philosophy outlook on the translation process.
Numerous models of translation processes have been proposed, and many (if
not most) of them imply some sort of stratificational processes in cycles of
comprehension-transfer-production. Gile (2005), for instance, suggests a
stratificational translation process model, in which first the translator creates a
“Meaning Hypothesis” for a ST chunk (i.e., a Translation Unit) which is consistent
with the “context and the linguistic and extra linguistic knowledge of the
translator” (p. 107), from which a translation is produced in a second step.
Angelone (2010) supports that translators process in cycles of comprehension-
transfer-production and maintains that it should be possible to attribute all
“uncertainties” of translators to any of the comprehension, transfer, or production
phases, but regrets that “non-articulated indicators, such as pauses and eye-
fixations, give us no real clue as to how and where to allocate the uncertainty” (p.
23).
Also Shreve & Lacruz (2017) consider reading, writing, and transfer the
“constituent activities” of the translation task. According to them, a translator first
constructs a “situation model” from a source text which is then a “template for the
production of the target text” (p. 133). Their “construction–integration model”
aims to shed light on “the transfer processes that occur between source text reading

21
and target text writing” (p. 134). As pointed out by Carl & Dragsted (2011), such
stratificational models have been challenged. Mossop (2003) assumes the existence
of “direct linkages in the mind between SL and TL lexicogrammatical material,
independent of ‘meaning’”, and that a translator “automatically produces TL
lexical and syntactic material based on the incoming SL forms”. Similarly,
Jakobsen & Jensen (2008) showed that reading purpose has a “clear effect on eye
movements and gaze behaviour” and conclude “that a fair amount of pre-
translation probably enters into the reading of a text as soon as it is taken to be the
source text for translation” (p. 116). Ruiz et al (2008) provide evidence that direct
links exist between the SL and TL, at least at the lexical and syntactic level of
processing, and Eddington & Tokowicz (2013, 442) found that translation
ambiguity slows down translation recognition regardless of the source of
ambiguity (synonym translation-ambiguous or meaning translation-ambiguous).
That is, participants were slower and less accurate to respond to words that
had more than one translation compared to unambiguous words. Van Heuven et al
(1998) conducted experiments with Dutch-English bilinguals to measure reaction
times for recognizing English words. He found that the lexical decision time for
English words correlated with the size of a competitor set of orthographic similar
words (orthographic neighbors) in English and Dutch. For the English word hood
the competitor set would include English words like food, hold, and hoot, and
Dutch words like lood, hond, and hoos (Dijkstra et al 2018), and possible
translations into Dutch kap or hoed. Finally, Wu & Thierry (2010) observed non-
selective processing of brain activity using ERP, which provides additional support
at the neurological level for the co-activation of both languages during lexical
access. All these findings indicate that “the translator [automatically] engages in
partial reformulation while reading for the purpose of translating the source text”
(Ruiz et al 2008: 490), making a clear distinction between comprehension, transfer,
or production phases impossible.10

10
A Systems Theory Perspective on the Translation Process Michael Carl, Andrew Tonge, Isabel Lacruz Kent State
University, Ohio, USA

22
 Several models of the bilingual mind have been proposed to explain these
observations. The Revised Hierarchical Model (RHM; Kroll & Stewart 1994) is a
model of lexical representation in which L1 and L2 have two distinct but
connected mental stores on a lexical level and a common, shared store for
conceptual representations. The model assumes that new vocabulary words in L2
are mapped onto concepts that already exist within the L1 framework. The RHM
explains a number of observed phenomena, e.g., faster (and better) L2→L1
translation, as compared to L1→L2 translation but could not explain other
observations, e.g., that bilinguals can encode semantic representation of the new
L2 word directly, rather than via an activation of the L1 representation. In her
Distributed Feature Model (DFM), De Groot (1992) accounts for the fact that
semantic features of words and their translations do not always have the same
degree of overlap.
The DFM explains, for instance, that words with more than one translation
(there are many ways to translate advice into Spanish) are semantically less
similar, as compared to words that only had a single translation (e.g., father) across
languages (Tokowicz et al 2002). The DFM was mainly inspired to predict that
concrete translation equivalents show greater semantic similarity than abstract
translation equivalents. A word like table is typically rated higher in concreteness
and translated faster than a comparable abstract word. Words with multiple
translations in another language trigger a “Fan Effect” (Degani & Tokowicz 2010),
which explains that retrieval processes are slowed down when multiple concepts
are activated, and thus translation production takes longer (see also Eddington &
Tokowicz, above).
The DFM thus attributes an effect of “concreteness” on translation speed as
being due to the number of overlapping features. The Bilingual Interactive
Activation Model (BIA and BIA+, see Dijkstra & van Heuven 1998, 2002) takes
into account a large number of research findings, which show that lexical access is
nonselective (bottom-up) where two (or more) languages are simultaneously active
during word recognition.

23
 The integrated lexicon assumes an interactivity between orthography,
phonology, semantics, and the co-activation of words across different languages,
which acknowledges the findings of the hood example above. While the earlier
BIA model also allows for language specific (top-down) processes in word
recognition, this was removed in the BIA+ model, as it was found that higher
task/decision systems have no impact on word identification and retrieval. A
task/decision system controls, however, many executive processes, such as
monitoring, associated with the prefrontal cortex. The Multilink model (Dijkstra et
al 2018) builds on RHM and BIA+. It is a computational model to simulate
bilingual word recognition and word translation times. Simulations of word
recognition and translation times obtain moderate to strong correlation with the
empirical data. Multilink is implemented as a localist-connectionist model (i.e., a
network of connected nodes with analytically derived meanings and activation
functions) in which word nodes are activated depending on their orthographic
similarity to the input word and their frequency of usage. Similarly to BIA+,
Multilink assumes one integrated bilingual lexicon—i.e., one store for words from
different languages, and in contrast to the DFM “holistic meaning representations
that are fully shared or fully separate across languages" (Dijkstra et al 2018, 5).
Semantically active representations spread their activation to other items,
and trigger what we will refer to below as word-translation systems, i.e “co-
activation of many word candidates from different languages that are similar to the
input". (p2) Dijkstra et al (2018) explain that “[i]n sentence translation, the specific
meaning of translation that is appropriate to the context must be retrieved. To make
the correct selection, the meaning of the sentence must be related to the specific
shades of meaning of the lexical items under consideration” (p. 18). Because
activated word-translation systems consist of several co activated lexical-
phonological representations of both the input and output language, there must be a
task-dependent decision system that makes sure the correct translation is produced
in the correct language.

24
 This task/decision component is independent of the activation component
and makes sure that “the link between phonology and semantics is checked before
the word’s translation is selected and produced” (p. 18). The models described
above are mainly concerned with lexical and conceptual representations of
individual words and their impact on L1 and L2 retrieval. Most of the experiments
reported are based on reaction time tests involving single words, presented out of
context. Translation Studies, however, is concerned with the translation of
sequences of words in context, and less so with translation of individual words.11
Entropy is a fundamental concept in physical and information systems.
Entropy describes the amount of disorder in a system. In the equilibrium situation,
where all possible states of a system are equally likely, the entropy is proportional
to the logarithm of the number of possible states. In information systems, such as
humans or machines engaged in the creation of translations, entropy is measured in
terms of bits, where one bit is the quantity of information needed to decide
between two equally likely possibilities. A system of N states and an entropy of
log2(N) bits is in its equilibrium, which indicates maximum unpredictability
(anything is possible). A system with an entropy of 0 bits indicates complete
determinism. For physical systems, the second law of thermodynamics states that
the total entropy of a closed, isolated system never decreases over time, and thus
tends to develop towards its equilibrium.

11
A Systems Theory Perspective on the Translation Process Michael Carl, Andrew Tonge, Isabel Lacruz Kent State
University, Ohio, USA

25
 2.2 The translation process: an analysis of observational methodology

Over the past 15 years, the field of translation studies has seen considerable
theoretical and experimental activity in the area of cognitive translation processes.
Many observational studies have been conducted with the attempts of
understanding what translators actually do compared to what they are assumed to
be doing. Accessing the black box and gaining a better understanding of what goes
on during translation will advance the field of study, open new areas of research,
and improve the way translation is viewed and taught.
Researchers have focused on various aspects of the process including
revision (Mossop, 2001), creativity (Kussmaul, 1997), professional and student
approaches (Séguinot, 1989; Tirkkonnen-Condit, 1989; Jääskeläinen and
Tirkkonnen-Condit, 1991), time pressure (Jensen, 1999) and many more. While
observing translators, researchers make use of data collecting tools such video
cameras, think-aloud verbalizations, retrospective interviews and computer
software programs. This article aims to critically evaluate the methods used to
observe the translation process. Within the context of a study conducted at Toyota
Canada, my objective is to test the methods and to make recommendations for
more advanced practices of observation.12
The data used to evaluate observational research methods for this paper was
collected while observing a junior and a senior translator at Toyota Canada, and a
third translator who has worked with the Ontario Government’s Translation
Service. I would like to thank all three translators as well as Toyota’s linguistic
services in Scarborough, Ontario for their enthusiasm and willingness to
participate in the study. The background and experience of the three translators
varied slightly.
At first I thought this might be a problem because I had been trying to
assemble a homogeneous group, but the differences brought forth the strengths and
weaknesses of each observational approach in relation to experience, and how they
12
THE TRANSLATION PROCESS: AN ANALYSIS OF OBSERVATIONAL METHODOLOGY Sabine Lauffer
Glendon College, York University

26
can be used together to get a better understanding of the translation process. The
source text was chosen by the manager of Toyota linguistic services because it was
an example of a typical translation that the department would complete. It was of
average difficulty and did not contain any confidential information. The translation
was carried out in the natural work setting of all three translators.
In the case of Toyota, I went to their Scarborough office and observed the
translators in their work environment with the tools they were accustomed to
using. They had their computers, databases, and electronic and traditional reference
materials. It was an open concept office and the translators were free to consult
with their colleagues. I observed the third translator in her natural work
environment with a computer, dictionaries and the electronic resources she is used
to consulting. I prepared each computer in advance by installing two software
programs, Translog and Camtasia. Translog is a translation recording program
used to obtain quantitative results. Camtasia is a recording software that creates a
video file of the computer screen and records all moves on the computer. The
video camera was set up to capture facial expressions and work done away from
the computer. It also recorded think-aloud verbalizations.13
The data collection was a three step process: informing and preparing the
subjects; the actual translation process combined with think-aloud, computer
recordings, direct observation, and audio and video recording; and retrospective
interviews with and without computer playback of the translation process. During
the pretranslation discussion, the translators were introduced to the study and the
methods of observation. They were encouraged to ask questions and use Translog
in order to become familiar with the program and its features.
All three translators were asked to verbalize whatever came to their mind
and to speak freely during the translation process. I emphasized the need to feel
comfortable and to work as if it were an average day on the job. Once the
translators were ready to begin, I acted as a strict observer and had no contact with
them. After the translation was complete, I conducted two retrospective interviews.
13
THE TRANSLATION PROCESS: AN ANALYSIS OF OBSERVATIONAL METHODOLOGY Sabine Lauffer
Glendon College, York University

27
The first included general questions on how the translator felt about the translation
and the methods of observation. The second interview included a playback of the
Camtasia recording allowing the translator to see the entire process and make
further comments.
A great number of studies on cognitive processes in translation have used
think-aloud protocols (TAP) to extract information about the underlying mental
processes required to complete a given task. TAPs originated in the field of
psychology and have been used in many disciplines. Although each study may
vary slightly, the expectation is that subjects faced with a specific task, such as
translating, will verbalize whatever comes to mind while performing the assigned
task.
The verbalizations are recorded, transcribed and analyzed. The anticipated
outcome is that these verbalizations will give a better understanding of “the levels,
steps, units of processing, the role of the interaction of the source and target
language, the amount of proceduralization, the origin and course of search
processes, and the times used for these processes.” (Dechert and Sandrock,
1986:115) Since thought processes are not directly observable, researchers use
TAPs as indications of what might be going on in the black box.
There has been much debate surrounding the use of think-aloud during
information processing, and whether the subjects can even verbalize their thought
processes while they are thinking. In their often-cited article “Verbal Reports as
Data”, Ericsson and Simon support the use of TAPs, but acknowledge that only
information stored in short-term memory is accessible to verbalization. Therefore,
as processes become more procedural, they may no longer be as available for
description. There are also concerns about forced speaking during a normally
unvoiced procedure altering results.14
Direct observation is usually conducted in two ways. Firstly, the researcher
is present to take notes and observe the overall process. Secondly, the event is
recorded with a video camera so it can be analyzed in closer detail at a later time.
14
THE TRANSLATION PROCESS: AN ANALYSIS OF OBSERVATIONAL METHODOLOGY Sabine Lauffer
Glendon College, York University

28
In my study, I used the camera to record facial expressions and body language
because they can be indicators of mental processes. It also recorded the think-aloud
verbalizations and the non-computerized tasks such as consulting with colleagues
and using paper reference materials. Before computer programs such as Translog
existed, cameras were also used to record computer, mouse and keyboard activity
which were later transcribed into detailed logs.
Translog is a relatively recent method of observation. It is a computer
program developed in Denmark at the Copenhagen School of Business. Professor
Arnt Jakobsen kindly allowed me to use this software for research purposes and I
would like to thank him for the opportunity. Translog was developed to obtain
quantitative reinforcement of assumptions about translation. It is ideal for
gathering data such as the number of keystrokes, total number of characters, ratios,
time delays, number of electronic dictionary look-ups, and more.
Translators are required to translate directly into Translog on a split screen
that displays the source text on the top and the active target text on the bottom. The
program then records the translation process into a video format that can be played
back at a later time. Translog also produces a detailed analysis so the researcher
can examine what went on during the process with accurate time indications and
statistics. Camtasia is a software program that can be set to record the entire
computer screen. I used it to document what the translators were doing on the
computer outside of Translog, for example how they consulted electronic
dictionaries, databases, websites and other tools.
The retrospective interview was conducted in two parts. The first part
consisted of prepared questions that examined how the translators felt about the
translation and the observation process. During the second part of the interview,
the translators were asked to comment on their translation process with the
Camtasia playback. This was done to trigger information that was not verbalized
during the think-aloud process.
Traditionally, reviewing a translation after the fact has not yielded much
information because once the work is on paper, it is difficult to resurface the

29
thought processes. However, when the process is played back in real time with
actual mouse cursors and keyboard input, the translator is more likely to remember
the thoughts and actions that surrounded the decisions that were made
Combining the results of observational methodologies yielded very
interesting results about the translation process. They complimented each other and
gave a detailed report that could then be evaluated. I was quite surprised by the
results of the think-aloud verbalizations. As part of the pre-translation preparation,
the participants were asked to verbalize whatever came to mind. I told them that
once they began translating I would no longer be there, and that they should
behave as if it were a normal work day.
Once the preparation was finished and I was confident that the translators
had understood what was expected of them, I started the camera and stepped aside.
The first translator started by reading the entire text very carefully. I assumed he
would begin talking once he started translating. Several minutes later, he had still
not said a word. Even though I knew not to interfere with the process, I quietly said
to him “remember to think-aloud.” He looked at me with great surprise and said “I
am not thinking about anything” and went back to typing. Several minutes later, he
was still not saying anything.
Once again, I quietly asked him to think-aloud. He said he was just looking
for a word in the dictionary, but did not give any details. Luckily, I had the camera
to zoom in and could see that he was looking up the word écarter. He then
continued to translate and did not say another word until he finished two hours
later. In the retrospective interview, I asked why he had not verbalized and he
replied that he had found it very difficult and that having to think about talking
while translating took him away from the actual work.15

15
THE TRANSLATION PROCESS: AN ANALYSIS OF OBSERVATIONAL METHODOLOGY Sabine Lauffer
Glendon College, York University

30
 CONCLUSION

To obtain the most accurate results, it is crucial that translators be observed

in an everyday environment that is natural to them. This being said, it is important
to use unobtrusive methods of observation that do not risk altering or interfering
with the process. Translation has changed considerably over the past few years,
especially with the wave of new technology. Therefore, the way we observe
translation must also evolve. Computer programs such as Translog are an excellent
start. With continued development, Translog could become embedded into a
computer’s operating system and give statistics on all computer actions, not just
those performed within the program. Furthermore, integrating audio and video into
the program would ensure that facial expressions and vocalizations go with the
computer actions. Think-aloud protocols, computer programs, audio and video
recordings and retrospective interviews with playback should all be used together.
They each yield information about different parts of the process and help us better
understand the rapidly changing field. Continued work with improved methods
will shape the way translation is viewed, taught and understood. Developing
programs that keep statistics and monitor the translation process could also be
useful in other areas of translation. Currently, many professional translation
associations do not allow candidates to use computers during testing out of fear
that they may cheat and consult with on-line translators outside the exam. Forcing
translators to use paper methods when they are used to translating directly onto the
computer is an unfair method of evaluation. If a monitoring program such as
Camtasia or a systemwide Translog were to be implemented, the judges would be
able to view the video file and base their decision on a more realistic translation
process. There are many opportunities for translation observational methodologies
and they need to be further explored.

31
 THE LIST OF USED LITERATURE

1. Alves, F., Pagano, A. and da Silva, I. (2009). A new window on translators’

cognitive activity: methodological issues in the combined use of eye tracking, key
logging and retrospective protocols. In I. M. Mees, F. Alves and S. Göpferich, eds.
Methodology, technology and innovation in translation process research: a tribute
to Arnt Lykke Jakobsen, Copenhagen: Samfundslitteratur, 267–291.
2. Angelone, E. (2010). Uncertainty, uncertainty management and metacognitive
problem solving in the translation task. In G. Shreve & E. Angelone, eds.
Translation and cognition, Amsterdam: John Benjamins, 17–40.
3. Antunović, G. & Pavlović, N. (2011). Moving on, moving back or changing it here
and now: self-revision in student translation processes from L2 and L3. Across
Languages and Cultures, 12(2), 213–234.
4. Arnold, A. (1968). Genius with a dictionary: reevaluating D. H. Lawrence’s
translations. Comparative Literature Studies [on-line]. 5(4), 389–401. Available
online at http://www. jstor.org/stable/40467785 [Accessed 28 November 2015].
5. Borg, C. (2016). A literary translation in the making: an in-depth investigation into
the process of a literary translation from French into Maltese. PhD Thesis, Aston
University.
6. Breedveld, H. (2002). Writing and revising processes in professional translation.
Across Languages and Cultures, 3(1), 91–100.
7. Bush, P. (2006). The writer of translations. In S. Bassnett & P. Bush, eds. The
translator as writer, London: Continuum, 23–32.
8. Carl, M., Dragsted, B. and Jakobsen, A. L. (2011). A taxonomy of human
translation styles. Translation Journal [on-line]. 16(2), 155–168. Available at
http://translationjournal.net/ journal/56taxonomy.htm [Accessed 21 January 2015].
9. Chandler, D. (1993). Writing strategies and writers’ tools. English Today, 9(2),
32–38.

32
10. Dam-Jensen, H. & Heine, C. (2013). Writing and translation process research:
bridging the gap. Journal of Writing Research [on-line]. 5(1), pp.89–101. Available
at http://www.jowr. org/ [Accessed 12 June 2016].
11. Dragsted, B. & Carl, M. (2013). Towards a classification of translation styles
based on eyetracking and keylogging data. Journal of Writing Research [on-line].
5(1), 133–158. Available at http://www.jowr.org/ [Accessed 10 April 2014].
12. Englund Dimitrova, B. (2005). Expertise and explicitation in the translation
process. Amsterdam: John Benjamins.
13. Göpferich, S. (2010). The translation of instructive texts from a cognitive
perspective: novices and professionals compared. In S. Göpferich, F. Alves and I.
M. Mees, eds. New approaches in translation process research, Copenhagen:
Samfundslitteratur, 5–55.
14. Hansen, G. (2013). Many tracks lead to the goal: a long-term study on individual
translation styles. In C. Way, S. Vandepitte, R. Meylaerts and M. Bartłomiejczyk,
eds. Tracks and treks in translation studies: selected papers from the EST
Congress, Leuven 2010, Amsterdam: John Benjamins, 49–62.
15. Hayes, J. R., Flower, L., Schriver, K. A., Stratman, J. F. and Carey, L. (1987).
Cognitive processes in revision. In S. Rosenberg, ed. Advances in
psycholinguistics, Vol. 2. Reading, writing and language learning, Cambridge:
Cambridge University Press, 176–240.
16. Jakobsen, A. L. (2002). Translation drafting by professional translators and by
translation students. In G. Hansen, ed. Empirical translation studies: process and
product. Copenhagen: Samfundslitteratur, 191–204.
17. Johnsen, Å. (2014). Revision and cohesion in translation. Translation and
Interpreting Studies, 9(1), 70–87.
18. Mossop, B. (2000). The workplace procedures of professional translators. In A.
Chesterman, N. G. San Salvador and Y. Gambier, eds. Translation in context:
selected papers from the EST Congress, Granada 1998, Amsterdam: John
Benjamins, pp.39–48.

33
19. Mossop, B. (2001/2010). Revising and editing for translators. St Jerome:
Manchester.
20. Mossop, B. (2007). Empirical studies of revision: what we know and need to
know. Journal of Specialised Translation [on-line]. 8, 5–20. Available at
http://www.jostrans.org/issue08/ art_mossop.php. [Accessed 28 March 2013]

34