Cognitive effort in direct and inverse translation performance: Insight from eye-tracking technology

Abstract

This case study examined the translation performance of four
professional translators with the aim of exploring the cognitive effort
involved in direct and inverse translation. Four professional translators
translated two comparable texts from English into Spanish and from Spanish
into English. Eye-tracking technology was used to analyze the total
time spent in each task, fixation time, and average fixation time. Fixation
count in three areas of interest was measured including: source text, target
text, and browser, used as an external support. Results suggested that
although total time and fixation count were indicators of cognitive effort
during the tasks, fixation count in the areas of interest data showed that
more effort was directed toward the source text in both tasks. Overall, this
study demonstrates that while more traditional measures for translation
difficulty (e.g., total time) indicate more effort in the inverse translation
task, eye-tracking data indicate that differences in the effort applied in
both directions must be carefully analyzed, mostly regarding the areas of
interest.

Full text

Esta obra utiliza uma licença Creative Commons CC BY:
https://creativecommons.org/lice
http://dx.doi.org/10.5007/2175-7968.2016v36n3p60
COGNITIVE EFFORT IN DIRECT AND INVERSE
TRANSLATION PERFORMANCE:
INSIGHT FROM EYE-TRACKING TECHNOLOGY
Aline Ferreira*
University of California, Santa Barbara
John Wayne Schwieter**
Wilfrid Laurier University
Alexandra Gottardo***
Wilfrid Laurier University
Jefferey Jones****
Wilfrid Laurier University
* Possui graduação em Língua e Literatura Espanhola e mestrado em Linguística
Aplicada – Estudos da Tradução pela Universidade Federal de Minas Gerais,
Brasil. Doutorou-se em Linguística Aplicada – Estudos da Tradução pela mesma
universidade, em 2013. Realizou o pós-doutoramento em Psicolinguística na Wil-
frid Laurier University, no Canadá. Atualmente é professora no Departamento de
Espanhol e Português da Universidade da Califórnia Santa Barbara (UCSB). Santa
Barbara, California, USA. Email: alineafe@gmail.com
** John W. Schwieter is an Associate Professor of Spanish and Linguistics and
Faculty of Arts Teaching Scholar at Wilfrid Laurier University in Canada and
a Visiting Professor of Applied Linguistics in the Centre for Applied Research
and Outreach in Language Education at the University of Greenwich in England.
Waterloo, Ontario, Canada. Email: jschwieter@wlu.ca
*** Dr. Alexandra Gottardo is a Full Professor in Psychology at Wilfrid Laurier
University in Waterloo, Ontario, Canada. She has a background in educational
psychology and speech-language language pathology. Her research interests in-

61
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
Abstract: This case study examined the translation performance of four
professional translators with the aim of exploring the cognitive effort
involved in direct and inverse translation. Four professional translators
translated two comparable texts from English into Spanish and from Spa-
nish into English. Eye-tracking technology was used to analyze the total
time spent in each task, fixation time, and average fixation time. Fixation
count in three areas of interest was measured including: source text, target
text, and browser, used as an external support. Results suggested that
although total time and fixation count were indicators of cognitive effort
during the tasks, fixation count in the areas of interest data showed that
more effort was directed toward the source text in both tasks. Overall, this
study demonstrates that while more traditional measures for translation
difficulty (e.g., total time) indicate more effort in the inverse translation
task, eye-tracking data indicate that differences in the effort applied in
both directions must be carefully analyzed, mostly regarding the areas of
interest.
Keywords: Directionality in translation. Cognitive effort. Bilingual
dominance
clude examining factors related to the development of reading, both word reading
and reading comprehension, in second language learners. As a developmental
psychologist she studies development across a wide range of ages, from prescho-
olers to adolescents. Although her focus is on cognitive-linguistic variables with
and across languages, she has recently become more interested in socio-cultural
variables as additional explanatory variables. Waterloo, Ontario, Canada. Email:
agottardo@wlu.ca
**** Jeffery Jones is a professor at the Department of Psychology at Wilfrid Laurier
University, Canada. He is interested in human communication through speech,
music and other acoustic signals. Research topics include sensory-motor control
during speech, singing, and music production, as well as multisensory perception.
Applied work looks at communication disorders caused by Parkinson’s disease,
stuttering, as well as attention and cognitive loads during communication while
driving. Techniques used include behavioural measures, EEG, and fMRI. Water-
loo, Ontario, Canada. Email: jjones@wlu.c

62
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
ESFORÇO COGNITIVO NO DESEMPENHO EM
TRADUÇÃO DIRETA E INVERSA: UMA PERCEPÇÃO
A PARTIR DA TECNOLOGIA DE RASTREAMENTO
OCULAR
Resumo: O presente estudo de caso analisou o desempenho tradução de
quatro tradutores profissionais com o objetivo de explorar o esforço cog-
nitivo envolvido durante a tradução direta e a inversa. Quatro tradutores
profissionais traduziram dois textos comparáveis do inglês para o espanhol
e do espanhol para o inglês. A tecnologia de rastreamento ocular foi usada
para analisar o tempo total gasto em cada tarefa, o tempo de fixação, e o
tempo médio de fixação. A contagem de fixação em três áreas de interesse
foi medida e inclui: a) o texto de partida, b) o texto de chegada e c) o nave-
gador de internet, usado como apoio externo. Os resultados sugerem que,
embora o tempo total e contagem de fixação sejam indicadores de esforço
cognitivo durante as tarefas, os dados da contagem de fixação nas áreas
de interesse mostram que mais esforço foi alocado ao texto de partida em
ambas as tarefas. De maneira geral, este estudo demonstra que, enquan-
to as variáveis tradicionalmente usadas para aferir níveis de dificuldades
durante uma tradução (e.g., o tempo total) indicam um maior esforço na
tarefa de tradução inversa, os dados de rastreamento ocular sugerem que
as diferenças no esforço aplicado em ambas as direções devem ser cuida-
dosamente analisadas, sobretudo em relação às áreas de interesse.
Palavras-chave: Direcionalidade em tradução. Esforço cognitivo. Domí-
nio bilíngue.
1. Introduction
Translation Studies (TS) is a relatively new area of research that
builds on theories and methodologies from other fields, especially
psycholinguistics, cognitive science, and expertise studies. At the
same time, TS has been successful at adapting experimental designs
that construct a unique tradition (Ferreira, Schwieter, and Gile
2015). Recently there has been a significant increase in the empirical
work drawing on what exemplified in special issues of international
journals (e.g., Ehrensberger-Dow, Englund Dimitrova, Hubscher-

63
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
Davidson, and Norberg 2013) and book volumes (e.g., Ferreira
and Schwieter 2015; Schwieter and Ferreira 2014a; Shreve and
Angelone 2010). Directionality in translation has offered important
contributions in understanding the translation process and in
studying how translation performance may be dependent on a
variety of variables involved in the process (Buchweitz and Alves
2006; Ferreira and Schwieter 2017; Pavlović and Jensen 2009).
Different levels of cognitive effort are applied depending on the
task at hand (Ferreira 2014) and the amount of attention that is
directed towards the source text (ST) and target text (TT) depends
on how the translator understands and processes the task. We hope
this analysis can contribute to the still incipient body of work on
TPR that uses eye-tracking data, a rather complex field, not only
because of the number of participants in this sort of experiment but
also because eye fixations tend to vary among subjects.
2. Background
2.1. Directionality in translation
Despite the taboo against IT practice (Feltrin-Morris 2008), IT
translation is performed on a regular basis by professionals and
novices alike. Globalization in our modern world demands translators
and interpreters (Ordoñez López, 2010), leading to more studies on
translation processing and a better understanding of the intervention of
translation and interpreting services, in both directions of translation.
In this sense, directionality in translation has recently resulted in an
increase in the number of studies that contribute to understanding
the cognitive mechanisms that are involved in the translation process
(e.g., Alves and Gonçalves 2013; Ferreira 2012, 2014; PACTE
2011; Pavlović and Jensen 2009). There remain several gaps in what
we know about the practice of inverse translation (IT) in contrast to
direct translation (DT) and as such, additional studies are necessary
to make further advancements in TPR.

64
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
In a study investigating directionality in translation processes by
means of eye tracking, Pavlović and Jensen (2009) looked at the
effects of directionality on gaze time, average fixation duration,
total task length, and pupil dilation, all of which were interpreted as
indicators of cognitive effort. Students and professionals translated
two comparable texts from Danish into English (DT) and from
English into Danish (IT). The results showed that for both groups
inverse translation tasks on average lasted longer than DT tasks.
Furthermore, pupil dilation values, previously shown to be positively
related to cognitive effort, were higher in the inverse tasks. In
terms of average fixation duration, results showed that professional
translators had higher average fixation durations in IT, but students
had lower average fixation durations in the IT in comparison to DT.
Interestingly, when comparing the data of students with professionals
on both tasks, it was found that professionals do not always present
higher cognitive effort than students. Results of average fixation
duration, for instance, showed higher cognitive effort-related values
for professionals, which would “challenge traditional assumptions”
(p.108) about DT and IT.
By analyzing keystrokes and retrospective protocols,
Ferreira (2014) carried out a study with professional translators
that investigated recursiveness patterns and the production
of retrospective protocols. Participants translated two sets of
comparable STs, with each set comprising a text to be translated
from English into Portuguese (DT) and another text to be translated
from Portuguese into English (IT). The first set was comprised of
texts on the same topic, whereas the second set was composed of
STs on different topics. Analyses on time spent, recursiveness, and
retrospective protocols, more specifically related to lexical problems
and spontaneous solutions versus solutions from external resources,
suggested that in the context of related texts, DT translation was
more demanding than IT. However, in the context of unrelated
texts, results showed that IT required more effort than DT, which
indicates that in the absence of a facilitating effect, IT translation
required more effort than DT. From keystroke activities, Ferreira

65
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
(2012) investigated the effects of directionality on the segmentation
product of the same group of professional translators, and the
results indicated that translators produced smaller segments in the
IT, resulting in more segments in the IT task. As a consequence of
a higher number of segments in the IT task, they produced more
segments at the word level—the lower level—in the IT task in
comparison to the DT task. Thus, smaller segments produced by
the translators in the IT can be interpreted as an indicative of more
cognitive effort in the inverse direction of translation.
Ferreira’s (2013) carried out a study with eight professional
translators who translated two comparable texts on the same topic
and two comparable texts on different topics, from English into
Portuguese and from Portuguese into English. By analyzing time,
pause, recursiveness and segmentation patterns as indicators of
cognitive effort, the study showed that a translator’s performance
is strongly related to the task at hand, when translators have to
apply more effort to a specific task depending on the ST type
(texts on the same topic vs. texts on different topics), as well as
the task order. The results also showed that lexical solutions are
more effortful during IT compared to DT. However, questions
remained regarding the extent to which bilingual linguistic
competence affects the task at hand, mostly in terms of lexical
decisions. As studies on directionality increases, more questions
arise regarding how translation is conducted in both directions
and whether translation processing is affected depending on the
direction of the translation.
2.2. Attention and cognitive effort during translation tasks
Attention distribution during translation tasks can be measured
by recording eye fixations (Dragsted 2010), assuming that a specific
item which is fixated at a particular moment might indicate that that
item is being processed at that time (Just and Carpenter 1980).
In a study which compared eye-tracking data from reading-for-

66
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
translation task with silent reading and reading-for comprehension
tasks, Dragsted showed that “professional translators load their
cognitive capacity with something that is not only merely reading
for comprehension indicating that a pre-translation takes place”
(p. 47). The researcher explained that different fixation data (e.g.,
gaze data) can be combined with other variables (e.g., total reading-
for-translation task time) to indicate different levels of cognitive
“load” during a task at hand. Some variables are commonly used
in TPR studies, such as total gaze time, average fixation duration,
total task length, pupil dilation, among others (for a review see
Hvelplund 2014).
The present study examined cognitive effort in DT and IT
tasks. We investigated the coordination of attention in three areas
of interest (AoI): total dwell time of fixation on the source text, on
the target text, and on the external support area (available for the
translators via browser). In addition, we examined time spent in
each task, total fixation count and average fixation count.
3. Present study
3.1. Participants
Four professional translators were recruited from an English-
speaking region of Ontario, Canada. Two of the participants
were highly-proficient English-Spanish bilinguals and two were
highly-proficient Spanish-English bilinguals. Information from the
questionnaires (see Table 1) and Verbal Fluency Measure (VFM)
validated their high proficiency in both languages. In terms of their
educational background, all of the participants possessed at least a
bachelor’s degree and had a minimum of six years of experience
with DT and IT.

67
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
Table 1. Participant characteristics.
L1 L2 Translation
experience
(in years)
University
degree
Self-reported
L2 Proficiency
Frequency
of DT
Verbal
Fluency
L1
Verbal
Fluency
L2
S1 English Spanish 6-10 Yes Very Proficient Up to 70% 121 124
S2 English Spanish 6-10 Yes Very Proficient Up to 70% 118 110
S3 Spanish English >10 Yes Very Proficient Up to 70% 123 125
S4 Spanish English 6-10 Yes Very Proficient Up to 40% 127 118
Because there are specific aspects related to each direction
(e.g., editing by means of recursiveness; time; pause patterns;
segmentation) that can be measured, quantitative analyses of total
time spent in each task, as well as fixation time and fixation count
were carried out in order to investigate whether translators applied
more cognitive effort in DT or IT.
3.2. Procedure and design
The participants described above carried out two translation
tasks, one from English into Spanish and another from Spanish
into English. These tasks were recorded as a movie file and carried
out separately. Only the translator and two researchers were
presented during the data collection. Participants first filled in a
pre-questionnaire and carried out the translation from English into
Spanish. At the conclusion of the translation, the video was shown to
them and retrospective protocols were recorded. Next, they translated
the text from Spanish into English and retrospective protocols were
recorded. The task order was the same for all participants. The STs
were two popular science texts on different topics, yet similar in
terms of length (Spanish ST = 180 words; English ST = 187 words)
and structure according to Rhetorical Structure Theory (Taboada
and Mann 2006). STs are also similar in terms of coherence (i.e.,

68
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
how one text span has a specific role relative to another text span)
(Taboada and Mann 2006).1 Translation process data was recorded
with eye-tracking technology (EyeLink II) and Translog. The eye
tracking data provides insight on attention directed to each of the
tasks–fixation count and fixation duration. The amount of attention
directed to three different parts of the screen in the same task was
also analyzed. Therefore, the following variables were explored:
1. Total task length (i.e., the total time a translator spent in
each task);
2. Fixation count (i.e., the total number of fixations in each task);
3. Average fixation duration (i.e., the average time that the
fovea is directed at some location); and
4. Gaze time (i.e., the dwell time spent in each area of interest:
ST, TT, and browser).
3.2.1. Eye-tracking technology and Translog
The translation tasks were recorded with EyeLink II, a head-
mounted eye-tracking device which consists of three miniature
cameras situated on a comfortable padded headband2. Translation
tasks were also recorded with Translog (Jakobsen and Schou 1999),
a computational tool that logs keystrokes and mouse clicks during
translation. Due to the space constraints, in the present paper,
we will only report the results from the eye-tracking technology.
EyeLink II allowed for us to calculate the total time spent on
1 For comparable reasons, original texts had to be adapted. The text in English
was originally published at “A New Report Explains the Physics of Crumpled
Paper,” Scientific America, accessed on February 5th, 2002: http://www.
scientificamerican.com/article/a-new-report-explains-the. The original text
in Spanish was entitled, “Desarrollan una lengua electrónica para producir
mejor vino,” accessed on September 3rd, 2014: http://www.bbc.co.uk/mundo/
noticias/2014/01/140108_tecnologia_lengua_electronica_fabricacion_vino.
2 More information about the system can be found in http://www.sr-research.
com/EL_II.html.

69
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
each task, total fixation count, and average fixation duration.
Furthermore, fixation count in each area of interest (ST, TT, and
Browser) is also analyzed to verify whether these measures can
be used as indicators of different cognitive effort applied based on
the direction of translation. The mean fixation duration threshold
for the analysis was based on Pavlović and Jensen’s (2009) study
which specifies that the lower fixation threshold to discriminate
fixation from non-fixation is set to a temporal resolution of 100 ms
and a spatial resolution of 40 pixels, to include maximum gaze data
directly related to the translation task. The EyeLink Data Viewer
tool was used to display, filter, and report output of the data files.
We focused on defining three AoIs: ST, TT, and Browser. The
inclusion of these AoIs was based on the idea that the potential gaze
area can be the area related to ST, TT, or browser.
In order to verify fixation in each AoI, the computer screen
was divided into two parts: On the left half, the Translog screen
was displayed with TT and ST and, on the right half, browser was
displayed so that translators could use it for external support at their
convenience, such as online dictionaries, search tools, etc. Total
time spent on each task was calculated with Translog. Total gaze
time, fixation count, and average fixation duration were calculated
for each AoI using the EyeLink II’s Data Viewer tool.
3.2.2. Verbal fluency measure
The ability to access words from the mental lexicon is a key
component of human language processing and production, and is of
great importance to psycholinguists and clinicians alike (Isacoff and
Stromswold 2014). The VFM (Gollan, Montoya, and Werner 2002)
is one of the most commonly used assessments of vocabulary size
in bilingual studies. In the present study, participants generated as
many words as possible according to specified semantic categories
(clothing, country, animals, academic majors, colors, fruits,
vegetables, things with wheels, musical instruments, and sports).
They were given 30 seconds to do so for each category, first in

70
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
Spanish, then in English.3 When adding together the total number of
exemplars verbalized for each language, we are left with an arbitrary
verbal fluency score. In this case study, we will investigate whether
verbal fluency in either language can be used as an indicative of
fluency in both languages, which might endorse information
found in the prospective questionnaire on their proficiency level
in both languages. As described in Table 1, overall, it seems that
participants have almost the same vocabulary size in each language.
Indeed, a t-test confirmed no significant difference between English
and Spanish verbal fluency (t = .941, p = 0.416).
3.3. Hypotheses
The following predictions guided the research objectives of the
present study:
H1. Translators will spend more time and present higher
fixation count during the IT task compared to the DT task;
H2. Translators will present higher average fixation duration in
the IT task than in the DT task;
H3. Translators will invest more attention in processing the ST
than the TT in the DT task. Conversely, in the IT task,
they will invest more attention processing the TT than the
ST; and
H4. Translators will direct more attention to the browser during
the IT task in comparison to the DT task.
4. Results
4.1. General patterns in translation performance
The sample size of this case study suggests that we must first
caution against making sweeping generalizations based on the

71
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
results of the present study simply because small samples may not
fully provide an accurate representation of a general population.
Thus, we cannot assume that the data are normally distributed.
We also consider the general context of translation directionality:
all variables analyzed are considered to be indicative of cognitive
effort, depending on the translation direction (DT or IT).
4.1.1. Total time
The total time spent in each task appears in Table 2. The results
show that all translators spent more time in the IT task. A t-test
showed that the difference was significant (t = -2.799, p < .05).
Table 2. Total time spent (in seconds) in each task.
Direct Translation Inverse Translation
S1 1,014 1,397
S2 1,034 1,268
S3 1,020 1,036
S4 1,356 1,600
Mean 1,106 1,325
4.1.2. Total fixation count and time spent in each task
The results from total fixation showed that fixation count varied
considerably among participants, in both tasks (refer to Table 3). A
t-test showed that the difference was significant for all participants
but S3 (t = -1.194, p = .233). Total time presented in Table
3 confirmed our second hypothesis, that subjects would present
higher fixation count in IT. In this sense, in terms of total fixation
count, our results do not indicate that effort was be more intense
in IT if we consider both total time and total fixation as measures
of cognitive effort.

72
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
Table 3. Total fixations in DT and IT.
Direct Translation Inverse Translation
S1 1,645 2,793
S2 2,634 1,504
S3 2,002 1,830
S4 4,179 3,342
Total 10,460 9,469
Mean 2,615.00 2,367.25
4.1.3. Average fixation duration
In terms of average fixation duration, results in Table 4 show
that all participants produced longer average fixation duration in
IT. Therefore, our second hypothesis was confirmed.
Table 4. Average fixation duration (in milliseconds) in each task.
Direct Translation Inverse Translation
S1 283 338
S2 272 258
S3 338 343
S4 223 271
Mean 279.00 302.50
We could assume that these participants needed to allocate more
effort in the IT if we also assume that average fixation duration can
be used as a measure of effort in translation.

73
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
4.1.4. Areas of interest
The third hypothesis was that translators would invest more
attention processing the ST than the TT in DT, which was partially
confirmed by the data. Only S3 presented longer dwell time in
the target text than in the source text area during DT. We had
also hypothesized that they would invest more effort in the TT
processing than the ST processing in IT, which was not confirmed.
Total dwell time spent in Table 5 shows that participants tended to
present longer dwell time in the ST area in both tasks.
Table 5. Dwell time spent (in milliseconds) for source text, target
text, and browser.
Direct Translation Inverse Translation
ST TT Browser ST TT Browser
S1 327060 102432 36436 442712 235220 199120
S2 253328 201172 246572 217500 115180 55688
S3 257660 329252 115004 103276 27836 11376
S4 556748 188296 205196 511296 306408 89100
Total 1170012 746556 603208 1274784 684644 355284
Mean 292503 186639 150802 318696 171161 88821
4.1.4.1 Source text and target text
In this study, we assume that longer dwell time would indicate
more attention in one specific AoI in comparison to others, which
might indicate ST comprehension and TT production. Results show
that during IT, all participants presented higher dwell time in the
ST than in the TT, which did not confirm our third hypothesis. It
appears that this group needed to allocate more cognitive resources

74
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
to ST comprehension than to the TT reformulation in the IT.
Regardless the source language, results indicate that translators
made more effort to decode the source language than to encode
the meaning in the target language, assuming that the time spent in
each AoI is an indicative of effort in translation.
We had also hypothesized that translators would invest more
attention in processing the ST than the TT in the DT task. This
hypothesis was partially confirmed because only S3 presented
higher dwell time in the source text area in comparison to the target
text area. If translators present similar patterns in both directions
(e.g., focusing in the ST area more than in the TT area), it might
indicate that translators attempt to make sense of the sentences
and text as a whole and understanding the meaning of that ST
in the original culture demands more thinking than conveying the
information in the target language, as the target text, regardless the
direction of the translation.
4.1.4.2 Browser
Our fourth hypothesis, that translators would direct more
attention to the external resources (browser) during IT, could
not be statistically confirmed, as each participant present a very
different pattern for in each task. Comparing the AoI’s, results
show that during DT participants spent longer dwell time in the ST
area, followed by the browser area (except for S3). In the IT, all
translators spent longer dwell time in the ST.

75
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
Table 6. Total dwell time (in seconds) in the browser for direct
and inverse translation.
Direct
Translation
Inverse
Translation
S1 36.436 199.120
S2 246.572 55.688
Total 283.008 254.808
S3 115.004 11.376
S4 205.196 89.100
Total 320.200 100.476
Grand Total 603.208 355.284
Results showed that most translators spent higher dwell time in
the browser during their DT task, which was the opposite of what
we had expected. It might be the case that vocabulary search in the
DT was more demanding because translators are more critical of
lexical decisions in their first language
We had hypothesized that during DT, processing the ST would
be more demanding because the ST is in the foreign language,
which would demand more cognitive effort to comprehend than to
render the TT in the first language. Comparing only the results for
ST area in DT and IT, we can confirmed that participants indeed
spend longer time in the ST area in the DT. During the IT, on the
other hand, we hypothesized that conveying the target text in the
foreign language would demand more effort. Therefore, translators
would present higher dwell time in the TT area than in the ST
area during the IT, and that dwell time would also be higher in the
browser area during the IT, assuming that translators would need
to rely more in external support during the IT than DT.

76
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
5. Discussion
This case study has provided some preliminary insight on
the effects of directionality by looking at patterns in translation
performance. The data painted a complicated picture which
presented idiosyncratic patterns related to each of the participant’s
performance. Returning to our hypotheses, H1 posited that
translators would spend more time and present higher fixation
count during IT. This prediction could not be confirmed because
participants tended to present higher fixation count in the DT, even
though they spent more time in the IT. Our second hypothesis,
that subjects would present higher average fixations in IT, could
not be confirmed because S2 presented longer average fixation
during DT. However, we to find this to be true for the other three
participants. Our third hypothesis, that participants would invest
more effort, or attention, processing the ST than the TT during
DT, was also partially confirmed. While the overall mean across
participants was not significantly distinct, we do see that three out
of four translators invested more attention in processing the ST
than TT in DT. We had also hypothesized that participants would
direct more attention towards the TT than the ST, either English
or Spanish during IT. This hypothesis was rejected. During IT,
more attention was directed towards the ST area compared to the
TT area. All translators with the exception of S3 directed more
attention towards ST in both directions contrary to what we had
expected. Our results do not confirm that higher fixation in the
TT area could indicate that a translator is more concerned with
conveying the most accurate information to the reader than with
the comprehension of the ST. It seems that translators were more
concerned with understanding the source text, regardless of the
language, in order to be able then to convey the information in
the target text. While reading the target text, they could have been
thinking about how to produce the segment in the target language,
in which case the analysis of the areas of interest in an isolated
fashion does not allow us to confirm whether they were indeed

77
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
thinking of the target text, or simply understanding the source text.
The fourth hypothesis which predicted that translators would direct
more attention to the browser during IT was not confirmed. Only
S1 presented higher fixation in the browser during the IT translation
task in comparison to the DT. It is important to highlight that the
translators did not dedicate similar time to the external support in
the tasks and that S3’s fixation during DT was 10 times higher in
the browser area of interest than during IT.
In this study, professional translators, regardless of the direction
of the task, adapted their behavior to the task at hand in order to
render the TT, directing more or less attention depending on the
challenge that they were facing, and potentially focusing on the ST
area while pre-translating the TT. Assuming that longer fixations on
the ST or TT reflect text production problems (Carl and Dragsted
2012), it might be fruitful to investigate how two different groups of
translators (English-dominant and Spanish-dominant) face similar
problems during translation in both directions, and how they solve
these complications. It is still not clear whether DT would require
more external support research (e.g., searching by means of an
internet browser) because a translation into the more dominant
language would require more research than into the less dominant
language simply because translators would be more critical in their
first language, mostly in terms of lexical selection in L1 (Ferreira
2014). Future studies could also focus on the analysis of the final
TT in order to assess whether more or less effort applied to the
task can actually be observed in terms of the quality of the product
output. But more importantly, studies with English-dominant and
Spanish-dominant bilingual professionals in the same experiment
are necessary in order to investigate effects of language dominance
in the tasks and whether there is a relationship between the direction
of translation and dominance.

78
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
References
Alves, Fabio, and José Luiz Gonçalves. “Investigating the Conceptual-Procedural
Distinction in the Translation Process: A Relevance-Theoretic Analysis of Micro
and Macro Translation Units.” Target 25.1 (2013): 107-124.
Buchweitz, Augusto, and Fabio Alves. “Cognitive Adaptation in Translation: An
Interface between Language Direction, Time, and Recursiveness in Target Text
Production.” Letras de Hoje 41 (2006): 241– 272.
Carl, Michael, and Barbara Dragsted. Inside the Monitor Model: Processes of
Default and Challenged Translation Production. TC3, Translation: Computation,
Corpora, Cognition, 2.1 (2012): 127-145.
Dragsted, Barbara. “Coordination of reading and writing processes in translation:
An eye on uncharted territory.” Translation and Cognition. Ed. Gregory M.
Shreve and Erik Angelone. Amsterdam/Philadephia: John Benjamins Publishing.
2010. 41-62.
Ehrensberger-Dow, Maureen, Birgitta Englund Dimitrova, Séverine Hubscher-
-Davidson, and Norberg, Ulf. Describing Cognitive Processes in Translation:
Acts and events. Special Issue of Translation and Interpreting Studies, 8. 2
(2012): 151-153.
Feltrin-Morris, Marella. Into forbidden territory: The audacity to translate into
a second language. Diss. State University of New York, 2008. Ferreira, Aline.
“Investigando o Processamento Cognitivo de Tradutores Profissionais em Tradu-
ção Direta e Inversa no Par Linguístico Inglês-Português.” Cadernos de Tradução
1 (2012): 73–92.
Ferreira, Aline. Analyzing recursiveness patterns and retrospective protocols of
professional translators in L1 and L2 translation tasks. Translation and Inter-
preting Studies, 9.1 (2014) 109-127. Amsterdam/Philadelphia: John Benjamins
Publishing.
Ferreira, Aline, John. W. Schwieter, and Daniel Gile. “The position of psycho-
linguistic and cognitive science in translation and interpreting: An introduction.”

79
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Aline Ferreira, John Wayne Schwieter, Alexandra Gottardo & Jefferey Jones
Psycholinguistic and cognitive inquiries into translation and interpreting. Ed.
Aline Ferreira and John W. Schwieter. Amsterdam: John Benjamins Publishing,
2015. 3.15.
Ferreira, Aline, and John W. Schwieter. Directionality in translation. The Hand-
book of Translation and Cognition. Ed. John W. Schwieter and Aline Ferreira.
Malden, MA/Oxford, England: Wiley-Blackwell.
Gollan, Tamar, Rosa I. Montoya, and Werner, Grace A. Semantic and letter
fluency in Spanish-English bilinguals. Neuropsychology, 16.4 (2002): 562–576.
Hvelplund, Kristian T. Allocation of cognitive resources in translation: An eyetra-
cking and key-logging study. Diss. Copenhagen Business School, 2011.
Hvelplund, Kristian T. Eye tracking and the translation process: Reflections on
the analysis and interpretation of eye-tracking data. MonTI - Special Issue on
Cognitivism and Translation. Ed. Ricardo Muñoz Martín. Alicante: Universidad
de Alicante, 2014. 201-223.
Isacoff, Nora M., and Stromswold, Karin. “Not All Verbal Fluency Tasks Are
Created Equal: Lexical Access development between three and five.” First Lan-
guage 34.1 (2014): 43-57.
Jakobsen, Arnt Lykke, and Lasse Schou. “Translog Documentation.” Probing
the Process in Translation. Methods and Results. Ed. Gyde Hansen. Copenhagen:
Samfundslitteratur. 1999. Appendix, 1-6.
Just, Marcel A., and Patricia A. Carpenter. “A theory of reading: From eye fixa-
tions to comprehension.” Psychological Review 87.4 (1980): 329-354.
Ordoñez López, Maria del Pilar. 2010. “De mayor quiero ser traductor. Estudio
de las preconcepciones de los alumnos de primer curso del Grado en Traducción
e Interpretación.” Intralinea, 2010. Web. 21 May 2016. <http://www.intrali-
nea.org/archive/article/1659>
PACTE Group. “Results of the Validation of the PACTE Translation Competence
Model: Translation Project and Dynamic Translation Index.” Cognitive Explora-
tions of Translation.Sharon O’Brien. London: Continuum, 2011. 30-56.

80
Cad. Trad., Florianópolis, v. 36, nº 3, p. 60-80, set.-dez./2016
Cognitive effort in direct and inverse translation performance:...
Pavlović, Nataša, and Kristian Tangsgaard Hvelplund Jensen. “Eye tracking trans-
lation directionality.” Translation Research Projects. Ed. Anthony Pym and Ale-
xander Perekrestenko. Tarragona: Universitat Rovira i Virgili. 2009. 101-119.
Schwieter, John W., and Aline Ferreira, Eds). The development of translation
competence: Theories and methodologies from psycholinguistics and cognitive
science. Newcastle, England: Cambridge Scholars, 2014.
Shreve, Gregory, and Erik Angelone. (eds). Translation and cognition. Amster-
dam: John Benjamins, 2010.
Taboada, Maite, and William C. Mann. “Rhetorical Structure Theory: Looking
Back and Moving Ahead.” Discourse Studies 8. 3 (2006): 423-459.
Recebido em: 02/05/2016
Aceito em: 18/06/2016
Publicado em setembro de 2016