The impact of task complexity and translating self-efficacy belief on students’ translation performance: Evidence from process and product data

Abstract

Previous studies that explored the impact of task-related variables on translation performance focused on task complexity but reported inconsistent findings. This study shows that, to understand the effect of task complexity on translation process and its end product, performance in translation tasks of various complexity levels needs to be compared in a specific setting, in which more factors are considered besides task complexity—especially students’ translating self-efficacy belief (TSEB). Data obtained from screen recording, subjective rating, semi-structured interview, and quality evaluation were triangulated to measure how task complexity influenced the translation performance of Chinese students with high and low TSEB. We found that the complex task led to significantly longer task duration, greater self-reported cognitive effort, lower accuracy, and poorer fluency than the simple one among students, irrespective of their TSEB level. Besides, the high-TSEB group outperformed the low-TSEB group in translation accuracy and fluency in both tasks. However, the interaction effect of task complexity and TSEB was not significant, due possibly to weak problem awareness among students. Our study has implications for effectively designing task complexity, getting the benefits of TSEB, and improving research on translation performance.

Full text

Frontiers in Psychology 01 frontiersin.org
The impact of task complexity
and translating self-ecacy
belief on students’ translation
performance: Evidence from
process and product data
Xiangyan Zhou
1,2, Xiangling Wang
1
* and Xiaodong Liu
2
1 College of Foreign Languages, Hunan University, Changsha, China, 2 School of Foreign Studies,
Hunan University of Humanities, Science and Technology, Loudi, China
Previous studies that explored the impact of task-related variables on
translation performance focused on task complexity but reported inconsistent
findings. This study shows that, to understand the eect of task complexity
on translation process and its end product, performance in translation tasks
of various complexity levels needs to be compared in a specific setting, in
which more factors are considered besides task complexity—especially
students’ translating self-ecacy belief (TSEB). Data obtained from screen
recording, subjective rating, semi-structured interview, and quality evaluation
were triangulated to measure how task complexity influenced the translation
performance of Chinese students with high and low TSEB. Wefound that the
complex task led to significantly longer task duration, greater self-reported
cognitive eort, lower accuracy, and poorer fluency than the simple one
among students, irrespective of their TSEB level. Besides, the high-TSEB group
outperformed the low-TSEB group in translation accuracy and fluency in both
tasks. However, the interaction eect of task complexity and TSEB was not
significant, due possibly to weak problem awareness among students. Our
study has implications for eectively designing task complexity, getting the
benefits of TSEB, and improving research on translation performance.
KEYWORDS
task complexity, translating self-ecacy belief, interaction, translation process,
product quality
Introduction
Translation process research, which has been ongoing for about 40 years, focuses on
human cognition during translation (Alves and Hurtado Albir, 2010; Jääskeläinen, 2016).
In translation process research, simultaneous analysis of the translation process and its end
product is vitally important since “looking only at the process or the product…is looking
at only one side of a coin” (O’Brien, 2013: 6). Translators, related both to the process and
product of translation, have thus attracted great attention in translation process research
TYPE Original Research
PUBLISHED 03 November 2022
DOI 10.3389/fpsyg.2022.911850
OPEN ACCESS
EDITED BY
Simone Aparecida Capellini,
São Paulo State University, Brazil
REVIEWED BY
Pierre Gander,
University of Gothenburg,
Sweden
Ghayth Kamel Shaker AlShaibani,
UCSI University, Malaysia
Md Adnanul Islam,
Monash University, Australia
Juan Antonio Prieto-Velasco,
Universidad Pablo de Olavide, Spain
Lieve Macken,
Ghent University,
Belgium
*CORRESPONDENCE
Xiangling Wang
xl_wang@hnu.edu.cn
SPECIALTY SECTION
This article was submitted to
Educational Psychology,
a section of the journal
Frontiers in Psychology
RECEIVED 03 April 2022
ACCEPTED 17 October 2022
PUBLISHED 03 November 2022
CITATION
Zhou X, Wang X and Liu X (2022) The
impact of task complexity and translating
self-ecacy belief on students’ translation
performance: Evidence from process and
product data.
Front. Psychol. 13:911850.
doi: 10.3389/fpsyg.2022.911850
COPYRIGHT
© 2022 Zhou, Wang and Liu. This is an
open-access article distributed under the
terms of the Creative Commons Attribution
License (CC BY). The use, distribution or
reproduction in other forums is permitted,
provided the original author(s) and the
copyright owner(s) are credited and that
the original publication in this journal is
cited, in accordance with accepted
academic practice. No use, distribution or
reproduction is permitted which does not
comply with these terms.

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 02 frontiersin.org
(e.g., Muñoz Martín, 2014; Lehka-Paul and Whyatt, 2016;
Araghian et al., 2018). Although several factors can aect
translators’ mental process and product quality, previous studies
have identied the task itself as a key factor (e.g., Kelly, 2005;
O’Brien, 2013). First, there is a need for translation teachers to
design appropriate tasks as per pedagogical objectives. However,
traditional translator training is oen criticized for the lack of
appropriate criteria for source text selection and task design in
general (Kelly, 2005). In addition, task design is also an important
element in research design (O’Brien, 2013) as tasks adopted shall
beappropriate for the research project. e past decades have seen
a growing interest in investigating how task properties may
inuence translators’ performance. Researchers have explored task
type (e.g., Jia etal., 2019), task modality (e.g., Chmiel etal., 2020),
task condition (e.g., Weng etal., 2022), and task complexity (e.g.,
Feng, 2017; Sun etal., 2020). Among them, studies dealing with
task complexity have remained a major focus and reported
interesting, albeit inconsistent, ndings.
Although a major line of task-based research investigated the
impact of task complexity, it focused primarily on the translation
process and largely did not relate its ndings to specic translator
factors (e.g., Feng, 2017; Liu etal., 2019). Translation has oen
been depicted as a cognitive task driven by problem-solving
(Angelone, 2018). As translation problems are individual and arise
from the interplay between task properties and characteristics of
task performers (Muñoz Martín and Olalla-Soler, 2022),
translators may deliver varied performance even on the same task.
Studies that investigated the interplay between task properties and
translator characteristics worked on such individual dierences as
L2 prociency (Pokorn etal., 2019), working memory capacity
(Wang, 2022), and emotional intelligence (Ghobadi etal., 2021),
among others. However, to the best of our knowledge, no study
has so far investigated how task complexity inuences the
translation performance of students with dierent levels of self-
ecacy belief, an important aective variable inuencing students’
motivation and learning (Bandura, 1997) and a construct recently
introduced into translation studies (Yang etal., 2021a). is study
attempts to contribute further empirical data to translation
performance research by analyzing both the process and the
product of written translation and by considering the interaction
of task complexity and translating self-ecacy belief (TSEB).
Literature review
Tapping into the process and product
data of translation performance
Literature regarding the nature of translation performance
reveals two assessment approaches: product-and behavior-based
assessment. Most studies perceived translation performance as the
result of translation activities and assessed it by evaluating the
quality of the products only (Jääskeläinen, 2016). While exploring
the relationship between personality type and performance in
translating expressive, appellative and informative texts, Shaki and
Khoshsaligheh (2017) concluded that the sensing-type students
delivered less successful performance than the intuitive-type
students in all three tasks. eir performance was evaluated
against Waddington’s (2001) rubric, which was also adopted by
Ghobadi et al. (2021) to assess participants’ translation
performance. Meanwhile, some studies considered translation
performance as the sum of behaviors that participants controlled
in the process of translation (Muñoz Martín, 2014). erefore,
participants’ translation performance was understood by
evaluating their behaviors in the translation process. For example,
to better understand the mental processes involved in translation,
Lörscher (1991) investigated the strategic translation performance
of 56 secondary school and university students by utilizing think-
aloud protocols. Besides, Rothe-Neves (2003) analyzed the process
features of translation performance with processing time and
writing eort measures.
To solve the riddles of translation as a process and a product,
there is a need to combine quality assessment and process ndings
for performance evaluation. is is in line with the suggestion of
Jääskeläinen (2016) and Angelone (2018) that translation
performance research should begrounded in both process and
product data. Combining the two data sources opens up new
research avenues and increases the possibilities of nding
explanations and generalizing results to real-life circumstances. It
may help explain, for instance, why longer task duration
sometimes leads to better performance, but sometimes to poorer
performance among the same group of students (e.g., Seufert
etal., 2017). Despite potential benets of using the integrated
approach, product quality assessment has been integrated with
only a few process-oriented studies (Saldanha and O’Brien, 2014).
is study aims to demonstrate the utility of this approach by
collecting empirical evidence on both the translation process and
its end product.
Task complexity and translation
performance
Task complexity, dened as “the result of the attentional,
memory, reasoning, and other information-processing demands
imposed by the structure of the task” on task performers
(Robinson, 2011: 106), can aect cognitive processing (Plass etal.,
2010). Translation is a high-order cognitive task that imposes
cognitive load on and engages cognitive eort of task performers
(Liu etal., 2019). erefore, wemust rst dierentiate cognitive
load and cognitive eort before discussing the relationship
between task complexity and translation performance. In the
present study, cognitive load is associated with the complexity of
a task as it refers to the demand for cognitive resources imposed
on students by the task, and cognitive eort associated with the
actual response by a student as it is the amount of cognitive
resources that the student expends to accomplish the task. is is
consistent with the constructs of cognitive load and cognitive

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 03 frontiersin.org
eort developed in educational psychology (Sweller etal., 1998).
While the cognitive load of a task can theoretically beidentical for
dierent students (Liu etal., 2019; Ehrensberger-Dow etal., 2020),
the cognitive eort expended in a task is individual since students
have certain freedom regarding how much eort to expend and
how to expend it (Feng, 2017; Sun etal., 2020).
Previous research shows that the highest level of cognitive
eort and task performance occur when the task imposes
moderate cognitive load (e.g., Plass etal., 2010; Chen etal., 2016).
erefore, translation tasks can optimize students’ opportunities
for performance and development if they are of moderate
complexity. Such a claim aligns with the social constructivist
approach to translator education, where Kiraly emphasizes the use
of scaolded learning activities (Kiraly, 2000). By far, task
complexity has been an issue central to curriculum and test
development in translator education (Sun and Shreve, 2014).
Myriad factors contribute to the complexity of translation tasks,
such as source text complexity (Sun and Shreve, 2014), source text
quality (Ehrensberger-Dow et al., 2020), the number of
simultaneous tasks (Sun etal., 2020), task familiarity (Pokorn
etal., 2019), and directionality (Whyatt, 2019). Among them,
source text complexity has been a constant focus. To investigate
the level of text complexity, researchers have resorted to various
measures, including readability (Sun and Shreve, 2014; Whyatt,
2019), word frequency and non-literalness (Liu et al., 2019),
degree of polysemy (Mishra etal., 2013), dependency distance
(Liang etal., 2017), text structure (Yuan, 2022), and cohesion
(Wu, 2019).
Although numerous studies have analyzed how task
complexity aects translation performance, the research
ndings are conicting. For example, by operationalizing task
complexity as the number of simultaneous tasks, Sun et al.
(2020) concluded that compared with translating silently,
translating while thinking aloud resulted in a higher level of
cognitive eort as indicated by task duration, xation duration
and self-ratings; however, the dual-task condition had no
inuence on translation quality when the source text was
complex. Sun etal.’s (2020) ndings were only partly consistent
with the ndings of Wu (2019). In Wu’s (2019) study, task
complexity, operationalized as several text characteristics, was
positively correlated with self-reported cognitive eort but
negatively correlated with translation quality. Despite their
discrepancy, Sun et al. (2020) and Wu (2019) revealed that
students devoted a higher level of cognitive eort with an
increase in task complexity. When task complexity was
operationalized as directionality, according to Fonseca (2015),
the consensus among translators and translation scholars was
that translating into a non-native language (also known as L2
translation) was cognitively more demanding than translating
into the native language (also known as L1 translation).
However, empirical studies dealing with the eect of
directionality on translation performance also reported
inconsistent ndings (Whyatt, 2019). One possible reason
underlying such inconsistency is that existing research on the
impact of task complexity does not consider its interplay with
translator characteristics.
Self-ecacy belief in translation
How task performance is inuenced by the interaction
between task properties and learner characteristics has been
consistently studied in second language acquisition (Robinson,
2011) and educational psychology (Sweller etal., 1998). Given that
individual dierences may inuence how many cognitive
resources to devote and how to expend them in task
implementation (e.g., Homan and Schraw, 2009; Wang, 2022),
there is a need to study translation performance by giving due
consideration to translator characteristics. Among the various
translator factors, it is critical to examine whether self-ecacy
interacts with task complexity in inuencing task performance—
an observation made in prior research in other disciplines (e.g.,
Judge etal., 2007; Homan and Schraw, 2009; Rahimi and Zhang,
2019). Interestingly, while Judge etal. (2007) reported that the
benets of self-ecacy were dicult to realize in more complex
tasks, some studies concluded that the role of self-ecacy was
more manifest when task complexity was higher (e.g., Homan
and Schraw, 2009; Rahimi and Zhang, 2019).
As an aective factor inuencing cognitive and motivational
processes (Bandura, 1997), self-ecacy can motivate learners and
encourage them to put in more eort once an action has been
initiated (Bandura, 1995). However, the construct has only
recently begun to draw attention from researchers in the eld of
translation (e.g., Bolaños-Medina, 2014; Muñoz Martín, 2014;
Bolaños-Medina and Núñez, 2018). Muñoz Martín (2014)
attached importance to the correlation between self-ecacy and
translation expertise by specically including self-ecacy as one
of the minimal sub-dimensions of self-concept, which constitutes
translation expertise together with knowledge, adaptive
psychophysiological traits, problem-solving skills, and regulatory
skills. Bolaños-Medina (2014) also proposed that self-ecacy was
a construct of relevance for translation process research, related
particularly to procient source language reading comprehension,
tolerance of ambiguity, general text translation, and
documentation abilities. Besides, Moores etal. (2006) pointed out
that an understanding of self-ecacy was required if training
programs were designed to develop expert performance level in
complex tasks.
Notably, the one-measure-t-all approach usually has
constrained explanatory and predictive value because “most of the
items in an all-purpose test may have little or no relevance to the
domain of functioning” (Bandura, 2006: 307). is idea aligns
with Alves and Hurtado Albir’s (2010: 34) proposal that translation
process research should “design its own instruments for data
collection.” erefore, empirical research on self-ecacy belief in
translation shall utilize measurement scales tailored for translation
tasks. So far, several such scales have been developed (e.g.,
Bolaños-Medina and Núñez, 2018; Yang etal., 2021a). Since this

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 04 frontiersin.org
study focused on the Chinese-English language pair, the
Translating Self-Ecacy China scale developed by Yang et al.
(2021a) was adopted, which was specically designed for students
with Chinese as their mother tongue and English as a
foreign language.
The present study
Considering the limitations in earlier studies, the current
study attempts to provide further empirical evidence on
translation performance. It collects data on both the process
and product of two written translation tasks, which are of
different complexity levels and performed by homogeneous
groups of students with high and low TSEB. A framework is
proposed to delineate variables contributing to and measures
of translation performance in the current study (see
Figure 1). We promote the idea that task complexity and
TSEB influence both students’ mental process and product
quality; moreover, task complexity might interact with TSEB
in influencing their translation performance. For process
features of translation performance, weadopt two measures
of cognitive effort following previous research
(Ehrensberger-Dow etal., 2020; Sun etal., 2020): time-on-
task and self-reported cognitive effort. Regarding the quality
of translation products, accuracy and fluency are discussed
against assessment guidelines.
In brief, to investigate how task complexity and TSEB
inuence students’ cognitive eort and product quality, and
whether there is an interaction eect between the two
independent variables, the following research questions (RQs)
are raised:
RQ1: What is the impact of task complexity on cognitive
eort and product quality of students?
RQ2: How does TSEB inuence students’ cognitive eort and
product quality?
RQ3: Does task complexity interact with TSEB in inuencing
students’ translation performance? If yes, how?
Materials and methods
Participants
Brysbaert (2019) proposed that an experiment involving
interaction required a minimum sample size of 100in psychology
research. us, 136 second-year translation students were recruited
for the study on a voluntary basis. e students were from one
comprehensive university in mainland China. ey all enrolled in
the “Translation eory and Practice” course, which constituted
their rst experience of intensive translation training aer a
foundation year with modules in their two working languages
(Chinese as their mother tongue and English as a foreign language),
an introduction to linguistics for translation, and instrumental skills
such as documentary research and computer skills. When the
experiment was conducted, the participants had learned English for
about 10 years. us, they were generally equipped with basic
translation skills and language abilities that could guarantee their
completion of the translation tasks. e students were aged between
19 and 22, and their gender was primarily female (N = 116, 85.3%).
FIGURE1
Proposed research framework of translation performance.

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 05 frontiersin.org
Based on the measures of TSEB, students were divided into two
groups as per the guideline of median split (Malik et al., 2021).
Namely, the top50% students (N = 68) with their TSEB value above
the median value were assigned to the high-TSEB group, while the
remaining 50% (N = 68) assigned to the low-TSEB group.
Participants’ CET4 score, a national test designed to measure English
prociency of undergraduates in China, was used as a measure of
their L2 prociency. e results of the independent samples t-test
showed that the two groups were signicantly dierent in TSEB
(t = −13.704, p < 0.001) and in L2 prociency (t = −2.297, p < 0.05).
Students were involved for pedagogical considerations. e
research outcome is expected to help translation teachers make
informed decisions in translation task selection and help students
benet from high TSEB for performance improvement. Task
selection is crucial for translator training, particularly at the initial
training stage, as unrealistically complex tasks prove to bea source
of frustration for students (Kelly, 2005; Wu, 2019).
Translation tasks
Translation direction
Demand for translation from Chinese into English has remained
strong in China. According to Kelly (2000), translation into a
non-mother tongue is a professional necessity in many local
translation markets and a useful training exercise that contributes to
students’ understanding of translation problems. However, despite
the presence of L2 translation on the market, the performance of L2
translation remains under-researched (Pokorn etal., 2019). In the
hope of contributing to research on L2 translation, wedecided to
implement Chinese-to-English as the translation direction.
Text selection
e current study operationalized task complexity as
quantiable measures of text characteristics following Wu’s
(2019) suggestion. e two source texts are both about mobile
phones and between 160 and 170 Chinese characters (see
Appendix A for details). is precludes the possibility that
unfamiliarity with the subject domain would skew task
performance. e selection of quantiable measures was guided
by literature review. First, lexical polysemy indicates translation
ambiguity and task complexity (Mishra etal., 2013). As a word
with more senses may be ambiguous and thus slow down
processing for learners with a low level of skill and knowledge
(McNamara et al., 2014), word polysemy value is positively
correlated with task processing demands. Second, low cohesion
may increase reading time and disrupt comprehension
(McNamara etal., 2014). Connectives are very important in
establishing cohesion (Graesser et al., 2011). In Chinese-to-
English interpreting, connectives were added to enhance
cohesion by professional translators, so as to make implicit
information in the Chinese text explicit in the English text (Tang
and Li, 2017). erefore, the incidence score of connectives in the
Chinese text is reversely linked to task processing demands.
We utilized the Coh-Metrix Web Tool (Traditional Chinese
version) to analyze properties of the two source texts. Coh-Metrix,
a linguistic workbench that uses indices to scale texts on
characteristics related to words, sentences, and connections
between sentences, has been adopted to analyze text characteristics
in academic research (Graesser etal., 2011). Coh-Metrix reports
the average polysemy for content words in a text, and provides an
incidence score for all connectives (occurrence per 1,000 words).
Although the specic Coh-Metrix measures vary across versions
and tools, the measures are quite similar (McNamara etal., 2014).
According to the analysis results of the Coh-Metrix Web Tool
(Traditional Chinese version), Text II has a larger polysemy value
and a lower incidence score of connectives, which indicates a
higher level of text complexity. erefore, Task 2, which
corresponds to Text II, is more complex than Task 1. e details are
illustrated in Table1. Moreover, as experts’ intuition is reasonably
reliable when it comes to text complexity evaluation (Sun and
Shreve, 2014), an expert panel was recruited to assess task
complexity. e expert panel consisted of two translation teachers
with over 5 years’ teaching experience and three professional
translators with over 10 years’ translation experience. eir
conclusion also indicates that Task 2 is more complex than Task 1.
Quality assessment metrics
e produced translation texts were evaluated by two Chinese
translation teachers with over 5 years’ experience in Chinese–
English translation teaching and assessment. eir assessment
guidelines were adapted from Waddington’s (2001) rubric. e
original rubric consists of three measures—accuracy of transfer of
source text content (i.e., accuracy), quality of expression in the
target language (i.e., uency), and task completion degree. As
translation quality was discussed in terms of accuracy and uency
in our study, we only considered the rst two measures when
developing the assessment guidelines (see Appendix B for details).
Besides, for each measure, there are ve levels, and each level
corresponds to two possible marks; this is to comply with the
marking system of 0–10, and to give raters freedom to award the
mark according to whether the candidate fully meets the
requirements of a particular level or falls between two levels but is
closer to the upper one (Waddington, 2001).
e two raters were rst invited to get familiar with the
assessment guidelines and then worked together to negotiate
quality assessment in the marking process. Previous studies
TABLE1 Task complexity and quantifiable measures of source texts.
Task Task
complexity
Text Polysemy
value of
content
words
Incidence
score of all
connectives
(occurrence
per 1,000
words)
1 Simple Text I 4.692 19.231
2 Complex Text II 5.955 12.195

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 06 frontiersin.org
revealed that even precise guidelines were given, cognitive bias
and disagreement might still occur during the assessment process
(Eickho, 2018; Islam etal., 2022). e negotiation approach has
been widely adopted in writing assessment research and proved
as an eective way to reduce raters’ bias (Trace etal., 2015). During
the assessment process, the two raters analyzed product quality
and reached a consensus through discussions and consultations
(Yang et al., 2021b). Namely, when there was a discrepancy
between their scores, the raters negotiated by providing an
explanation and justication for their score assignment, in the
hope of reaching a consensus score. If there was still disagreement
in marking, they solved it by consulting a third person, a
professional translator with over 10 years’ working experience.
Procedure
e experiment took about two and a half hours. First,
participants signed an Informed Consent Form approved by the
university’s Ethics Committee, and completed a language
background questionnaire and the Translating Self-Ecacy China
scale. en, all participants performed two translation tasks, with
their task duration and translation behaviors observable on the
screen recorded by screen capture soware. Each of the translation
tasks was followed by a subjective rating with a questionnaire
adapted from NASA-Task Load Index (NASA-TLX) (Hart and
Staveland, 1988), which was aimed to measure their cognitive eort
invested in the preceding task. e revised NASA-TLX questionnaire,
which comprises mental demand, eort, frustration, and
performance subscales, is of good reliability and has been applied in
some previous research to measure the amount of cognitive
resources devoted to task implementation (e.g., Sun and Shreve,
2014; Sun etal., 2020; Yuan, 2022). For details, see Appendix C. To
avoid sequencing eects, tasks were pseudo-randomly ordered so
that participants would alternate between the two tasks. Participants
were told that their translations would be assessed for external
dissemination; therefore, they could review and revise their
translations. is was intended to encourage participants to try their
best in the experiment. is session had no time limits, and
participants were allowed to access the Internet.
Upon completing the translation tasks, about 30% of the
participants (N= 20) were randomly selected from each group to
participate in a semi-structured interview, which was designed to
understand students’ perceptions of task complexity and their
translation performance, solution to uncertainties and ambiguities
during translation, and willingness to invest cognitive eort in the
process. Please refer to Figure 2 for the ow chart of the
experiment procedure.
Data quality and statistical analysis
To measure how task complexity, TSEB and their interaction
inuence students’ cognitive eort and translation quality, a
mixed-methods approach was adopted to collect and analyze data.
Specically, subjective rating and quality evaluation were used to
collect quantitative data, semi-structured interview was adopted
to collect qualitative data, while screen recording was employed
to collect both quantitative data (participants’ task duration) and
qualitative data (participants’ translation behaviors observable on
the screen). Data quality was ensured with two measures: First, EV
Screen Recorder soware was installed on each computer to
monitor the translation process. Besides, aer gathering process
data, wefound some outliers in the task duration and translation
quality dataset. e recordings of EV Screen Recorder showed
that, seven students failed to record the translation process in a
complete manner or to submit their translation(s) due to technical
issues with the computer. eir data were therefore excluded from
the dataset. Consequently, there were 65 students in the low-TSEB
group and 64in the high-TSEB group in our data analysis.
In quantitative analysis, linear mixed-eects models
(LMEMs), which can compensate for weak control of variables in
naturalistic translation tasks (Saldanha and O’Brien, 2014), were
employed as one of the analytical techniques to account for high
variability among participants and increase the power of tests
(Mellinger and Hanson, 2018). Webuilt four LMEMs altogether.
e dependent variable of the four models was (1) time-on-task,
(2) self-reported cognitive eort, (3) accuracy score, and (4)
uency score, respectively. Burnham and Anderson (2004)
provided rules of thumb when assessing plausible models. ey
believed that the best model was considered as the one with the
lowest Bayesian information criterion (BIC) value. Obtained
results in this study suggested that the models with interaction
were better than the null ones.
For all four models, the random eects were always the
participants, while the xed eects were task complexity (simple
and complex) and TSEB (low and high). As previous studies
revealed a strong correlation between L2 prociency and translation
performance (e.g., Jiménez Ivars etal., 2014; Pokorn etal., 2019), the
inuence of L2 prociency was controlled by adding it to the four
LMEMs as a covariate. During data analysis, we rst veried
whether there was a signicant main eect and then checked the
interaction eect of task complexity and TSEB. All statistical
analyses were run on IBM SPSS Statistics 26. e signicance level
was set at p = 0.05. Cohen’s f
2
was used to measure the eect size. e
results of the four LMEMs are discussed in the following section.
Results
Process feature: Time-on-task
e rst dependent variable in our LMEMs is time-on-task.
Measured by the time from task onset to task completion, time-
on-task has oen been used as a measure of cognitive eort
(Sweller etal., 2011). Overall, the rst LMEM showed a signicant
main eect of task complexity (b = −215.625, SE = 38.239,
t= −5.639, p< 0.001, 95% CI −291.323 ~ −139.927, f
2
= 0.112); but

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 07 frontiersin.org
neither TSEB nor the interaction of the two independent variables
proved signicant (p > 0.05; p > 0.05). Table2 shows the descriptive
statistics for time-on-task in the simple and complex tasks for
both groups, and the interaction between task complexity and
TSEB. As indicated by task duration, the complex task engaged
more cognitive eort than the simple one.
Process feature: Self-reported cognitive
eort
e second variable in our LMEMs is self-reported cognitive
eort, which was measured with the revised NASA-TLX
questionnaire mentioned above. Regarding the measurement of
cognitive eort invested in task implementation, self-rating scales
were more sensitive and far less intrusive (Sweller etal., 2011). e
overall results showed statistically signicant eect of task
complexity (b= −0.383, SE= 0.130, t= −2.938, p< 0.01, 95% CI
−0.641 ~ −0.125, f
2
= 0.024); however, the main eect of TSEB and
the interaction eect of task complexity and TSEB did not reach
statistical signicance (p> 0.05; p> 0.05). Table3 summarizes the
descriptive statistics for self-reported cognitive eort. According
to self-ratings of cognitive eort, students put in more cognitive
eort in the complex task than in the simple one.
Table4 provides details of an independent samples t-test for
mental demand rating, a subscale of the revised NASA-TLX
questionnaire. e table shows that the mean perceived mental
demand of Task 1 and Task 2 were 5.51 and 6.03in the low-TSEB
group, and were 5.75 and 6.00 in the high-TSEB group,
respectively. e two groups did not vary signicantly in the
perceived mental demand of Task 1 and Task 2, respectively
(t = −0.822, p > 0.05; t = 0.108, p > 0.05). In addition, to assess
whether the self-rated mental demand of Task 1 and Task 2
diered signicantly in each group, a paired samples t-test was
employed for both groups. e results showed that the perceived
mental demand increased signicantly from Task 1 to Task 2 for
the low-TSEB group (t = −2.790, p < 0.01), but not for the high-
TSEB group (t = −1.183, p > 0.05). In a word, the two groups had
similar perceptions of task demands in Task 1 and Task 2,
respectively; besides, only the low-TSEB group realized a
signicant increase in task demands when task complexity
changed from simple to complex.
Table5 summarizes the results of an independent samples
t-test for performance rating, which is also a subscale of the
revised NASA-TLX questionnaire and ranges from good (coded
as one point) to poor (coded as 10 points). It is shown that the
mean perceived quality of Task 1 and Task 2 were 5.72 and 5.63in
the low-TSEB group, and were 5.13 and 5.75in the high-TSEB
group, respectively. e two groups diered signicantly in the
perceived quality of Task 1 (t = 2.091, p < 0.05), but not in that of
Task 2 (t = −0.447, p > 0.05). In short, the high-TSEB group was
signicantly more condent in their translation quality than the
low-TSEB group in the simple task. But this was not true for the
complex task.
Product feature: Accuracy
e quality of translation products is analyzed in terms of
accuracy and uency. In this paragraph, the dependent variable
discussed is the accuracy score. e overall results revealed that
both xed factors (task complexity and TSEB) signicantly
inuenced translation accuracy (b= 1.625, SE= 0.121, t= 13.457,
p< 0.001, 95% CI 1.386 ~ 1.864, f
2
= 0.456; b= −0.405, SE = 0.198,
t= −2.044, p< 0.05, 95% CI −0.796 ~ −0.014, f2= 0.017). However,
FIGURE2
Flow chart of experiment procedure.
TABLE2 Time-on-task—significant eect of task complexity.
Eect Descriptive statistics for time-on-task in seconds
Factor level Factor
level
NMean SE
Tas k
complexity
(TC)
Simple / 129 1259.217 26.618
Complex / 129 1465.338 26.618
TSEB Low / 129 1376.701 31.562
High / 129 1347.854 33.355
TC*TSEB Simple Low 65 1278.393 36.825
High 64 1240.042 38.446
TC*TSEB Complex Low 65 1475.008 36.825
High 64 1455.667 38.446

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 08 frontiersin.org
their interaction eect was not signicant (p > 0.05). In other words,
students produced a signicantly less accurate translation in the
complex task than in the simple one, regardless of their TSEB level.
Besides, students with high TSEB signicantly outperformed their
counterparts with low TSEB in terms of accuracy in both tasks. e
descriptive statistics for accuracy are provided in Table6.
Product feature: Fluency
e fourth LMEM was built with the uency score as the
dependent variable. Statistically signicant eects of task
complexity and TSEB were found on translation uency (b= 1.750,
SE= 0.099, t= 17.599, p< 0.001, 95% CI 1.553 ~ 1.947, f2 = 0.391;
b = −0.666, SE = 0.222, t = −3.003, p < 0.01, 95% CI
−1.104 ~ −0.228, f
2
= 0.042). However, the interaction eect of task
complexity and TSEB did not reach statistical signicance
(p > 0.05). is means that both groups produced signicantly
poorer uency in the complex task than in the simple one. In
addition, high-TSEB students achieved signicantly greater
uency than low-TSEB students. e descriptive statistics for
uency are provided in Table7.
Discussion
e Results section shows complex eects of task complexity
and TSEB on students’ translation process and product quality, and
prove the importance of TSEB in investigating the impact of task
complexity on translation performance. Wefound that the complex
task led to signicantly longer time-on-task, greater self-reported
cognitive eort, lower accuracy, and poorer uency than the simple
one in both groups. Moreover, the high-TSEB group achieved
signicantly higher accuracy and greater uency when compared
with the low-TSEB group in both tasks. However, the interaction
eect of task complexity and TSEB was not statistically signicant.
e ndings are further discussed in the following paragraphs.
Eects of task complexity on translation
performance
Eect of task complexity on cognitive eort
Irrespective of their TSEB level, students put in a higher level
of cognitive eort in the complex task as measured by the time-
on-task and self-reported cognitive eort. Our nding
corresponds to some previous ndings that complex tasks engage
greater cognitive eort. For example, Feng (2017) reported that L2
translation, which was cognitively more demanding than L1
translation, involved greater cognitive eort as indicated by longer
TABLE3 Self-reported cognitive eort—significant eect of task
complexity.
Eect Descriptive statistics for self-reported cognitive
eort
Factor level Factor level NMean SE
Tas k
complexity
(TC)
Simple / 129 5.493 0.089
Complex / 129 5.826 0.089
TSEB Low / 129 5.703 0.105
High / 129 5.615 0.111
TC*TSEB Simple Low 65 5.561 0.123
High 64 5.424 0.129
TC*TSEB Complex Low 65 5.846 0.123
High 64 5.807 0.129
TABLE4 Independent samples t-test for mental demand subscale.
Task Group NMean SD Independent
samples t-test
tSig. (two-
tailed)
1 Low-TSEB 65 5.51 1.501 −0.822 0.413
High-TSEB 64 5.75 1.834 / /
2 Low-TSEB 65 6.03 1.714 0.108 0.914
High-TSEB 64 6.00 1.512 / /
TABLE5 Independent samples t-test for performance subscale.
Task Group NMean SD Independent
samples t-test
tSig. (two-
tailed)
1 Low-TSEB 65 5.72 1.452 2.091 0.039
High-TSEB 64 5.13 1.777 / /
2 Low-TSEB 65 5.63 1.409 −0.447 0.655
High-TSEB 64 5.75 1.613 / /
TABLE6 Accuracy—significant eects of task complexity and TSEB.
Eect Descriptive statistics for accuracy
Factor
level
Factor
level
NMean SE
Tas k
complexity
(TC)
Simple / 129 6.692 0.099
Complex / 129 5.134 0.099
TSEB Low / 129 5.677 0.122
High / 129 6.148 0.131
TC*TSEB Simple Low 65 6.423 0.136
High 64 6.961 0.144
TC*TSEB Complex Low 65 4.931 0.136
High 64 5.336 0.144

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 09 frontiersin.org
task duration. Besides, Wang (2022) also concluded that
translation students invested more cognitive eort in the complex
task than in the simple one, which was indicated by their longer
production time and longer pausing time.
e time-consuming eect resulting from task complexity
may beexplained by dierences in participants’ strategic behaviors
since cognitive load can impact mental processes (Muñoz Martín,
2014). Analysis of the screen recordings gave us some insights into
students’ translation process. First, regarding cognitive resources
allocated to the three phases of translation (Jakobsen, 2002),
students’ time on initial orientation increased with task
complexity, although they generally spent short time on initial
orientation in both tasks. Participant 62 spent about 20 s on initial
orientation in the simple task, compared to 105 s in the complex
task. For Participant 111, her orientation time on the simple and
the complex tasks were 25 and 75 s, respectively. Second, when
faced with higher task complexity, students tended to improve
their output by monitoring the translation process during both the
draing and the revision stages. For example, Participants 42 and
80 exhibited a higher level of product monitoring (evaluation) in
the complex task than in the simple one. Finally, students had a
higher frequency of pauses in the complex condition than in the
simple condition, such as Participants 16 and 88. Angelone (2010)
proposed that pauses or hesitations were a diagnostic sign of
uncertainties in the problem-solving process, which could occur
at any translation phase. Such uncertainties might cause students
to doubt their comprehension of the source text, ability to work
out a solution, or solution evaluation capacity.
However, it is interesting to nd out that no statistically
signicant correlation was observed between time-on-task and self-
reported cognitive eort (Pearson’s r = 0.098, p> 0.05). is indicates
that task complexity inuenced the two measures of cognitive eort
in a separate manner. Ogawa (2021) also revealed that task
complexity may aect translators’ task duration and subjective
rating in a dierent way. A possible explanation is that students
recruited in the current study were undergraduates and had weak
problem awareness, which led to underrating of cognitive eort and
in turn to insignicant correlation between the two measures. Such
an idea was corroborated by data from the semi-structured
interview. e interview data demonstrated that students on the
whole had low problem awareness. To bespecic, when asked how
to assess task complexity in the semi-structured interview, 18 of the
20 high-TSEB interviewees stated that task complexity depended
on the frequency of new words, and only one interviewee mentioned
the number of connectives. In the low-TSEB group, 16 of the 20
interviewees responded that they evaluated task complexity based
primarily on the number of new words; besides, topic familiarity,
lexical polysemy, and use of connectives were, respectively,
mentioned by two interviewees. Such a nding highlights the
importance of recruiting students with diverse education
backgrounds in future studies so as to compare their performance.
Eect of task complexity on product quality
A signicant main eect of task complexity was observed on the
accuracy and uency scores. Higher task complexity led to poorer
translation quality. Our nding lends support to Michael etal. (2011),
who claimed that ambiguous words, indicative of high complexity,
were translated less accurately as compared to unambiguous words.
Whyatt (2019) reported that participants made more grammar
mistakes in L2 translation than in the less demanding L1 translation,
indicating that higher task demands led to reduced uency. Wu
(2019) also reported that higher text complexity led to greater
inaccuracy and dysuency in students’ performance.
However, the nding conicts with Sun etal. (2020), who
arrived at their conclusion when using a complex text as the
source material and operationalizing task complexity as the
number of simultaneous tasks. According to Cognitive Load
eory, students generally “increase cognitive eort to match
increasing task demands up until they reach the limit of their
mental capacities” (Chen etal., 2016: 3). As a result, with an
increase in task complexity, students can adjust their level of
cognitive eort to maintain the quality level achieved in the less
complex task. at explains why students’ translation quality had
no signicant change when the condition changed from single-
task to dual-task in Sun etal. (2020).
Previous research showed that the relationship between cognitive
eort and performance quality was not linear: Increased eort may
lead to enhanced, unchanged, or reduced quality depending on
whether task complexity is low, moderate, or high (Charlton, 2002;
Seufert etal., 2017). In the current study, students produced poorer
translation quality in the complex task despite investment of more
cognitive eort, as they, with weak problem awareness, failed to
adequately increase their cognitive eort to match increasing task
demands and properly tackle the translation problems.
Eects of TSEB on translation
performance
Eect of TSEB on cognitive eort
The low-TSEB and high-TSEB groups were similar in
cognitive effort as per time-on-task and self-reported
TABLE7 Fluency—significant eects of task complexity and TSEB.
Eect Descriptive statistics for uency
Factor
level
Factor
level
NMean SE
Tas k
complexity
(TC)
Simple / 129 5.801 0.111
Complex / 129 4.142 0.111
TSEB Low / 129 4.593 0.143
High / 129 5.350 0.155
TC*TSEB Simple Low 65 5.378 0.151
High 64 6.225 0.162
TC*TSEB Complex Low 65 3.809 0.151
High 64 4.475 0.162

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 10 frontiersin.org
cognitive effort. Such a finding contradicts Araghian etal.
(2018), who found that high self-efficacy led participants to
spend less time on the translation task as highly efficacious
students had greater confidence in dealing with larger
translation units and reported fewer lexical and sentential
problems. However, the comparison should bemade with
caution since the task involved in their study was translating
an English text into Persian, a low-resource language (Fadaei
and Faili, 2020). Translating from a high-resource language to
a low-resource language poses challenges related to word
ordering (Fadaei and Faili, 2020), semantic and sentence
representations (Gu etal., 2018), and so on. It is different from
Chinese-to-English translation implemented in the current
study, as both Chinese and English are high-resource
languages (Aysa etal., 2022).
According to Cognitive Load Theory, devoting greater
cognitive effort is on the condition that task performers have
consciously realized increased task demands and/or feel
motivated to do so (Chen etal., 2016). However, first, the two
groups had similar mental demand ratings (i.e., perceived task
demands) in Task 1 and Task 2, respectively (see section
Process feature: Self-reported cognitive effort for more
details). This was corroborated by the interview data. As is
mentioned in section Effect of task complexity on cognitive
effort, although the low-TSEB group performed slightly better
in assessing task complexity than their counterparts, students
overall had weak problem awareness as they mainly referred
to new words for complexity assessment. In this study, task
complexity was operationalized as word polysemy value and
incidence score of connectives. Ignorance of translation
problems resulted in their failure to accurately assess
processing demands of the complex task and in turn
adequately increase cognitive effort to match increased task
demands. Our finding corresponds to the finding of
Jääskeläinen (1996: 67) that students “translate quickly and
effortlessly” because they problematized less than
semi-professionals.
Besides, according to the semi-structured interview, high-
TSEB students were more willing to put in greater cognitive eort
than their counterparts with low TSEB, but largely on the
condition that “the task becomes more demanding.” However, as
previously mentioned, the high-TSEB group failed to realize that
Task 2 was signicantly more demanding than Task 1 (see section
Process feature: Self-reported cognitive eort). In other words,
the two groups did not vary signicantly in cognitive eort due
possibly to similar perceptions of task demands in each task and
lack of strong motivation to invest more cognitive eort in the
complex task.
Eect of TSEB on product quality
TSEB had a signicant eect on students’ translation accuracy
and uency. is nding lends support to Jiménez Ivars et al.
(2014) who concluded that self-ecacy could boost translation
quality. Given that TSEB was a strong predictor of translation
quality but not of cognitive eort, it was possible that self-ecacy
enhanced product quality through resourceful use of strategies
rather than changing task duration, which echoes the ndings of
Homan and Schraw (2009). Araghian etal. (2018) also concluded
that self-ecacy might inuence students’ strategy use. According
to Bandura (1993), it required a strong sense of ecacy to remain
task oriented in the face of pressing demands and to eectively
process information that contained many ambiguities and
uncertainties. erefore, when faced with a translation problem,
students with high TSEB might be more resourceful in the
allocation and adaptation of alternative strategies than the
low-TSEB students, which in turn led to higher translation quality.
An analysis of data collected via the semi-structured interview
underpins such an explanation. For example, a low-TSEB
participant mentioned in the interview that she mainly resorted
to external resources to reach a denitive solution to translation
problems, while a high-TSEB participant stated that depending on
the nature of the translation problem, she alternated between
relying on her own knowledge and using external resources to
reach a solution. Both internal and external support can help
address unfamiliar terms whose equivalent expression in the
target language is available on the Internet. However, it might
befutile to resort solely to external resources when it comes to
translation uncertainties and ambiguities arising from polysemous
words or text cohesion. Resourceful strategy use by the high-TSEB
group is indicative of their better allocation of cognitive resources
during the translation process.
Interaction eect of task complexity and
TSEB on translation performance
No statistically signicant interaction eect of task complexity
and TSEB was found on students’ cognitive eort and product
quality. Our nding is inconsistent with that of Rahimi and Zhang
(2019), who identied an interaction eect between task
complexity and self-ecacy. First, writing tasks were used in their
study, which are dierent from translation, a complex cognitive
task that comprises source text reading and target text production
(Feng, 2017). Second, increased task complexity did not result in
evident dierences in the cognitive eort of the two groups
because neither group put in signicantly more eort in the
complex task than in the simple one. ese reasons could
potentially explain the contradiction in the ndings.
However, despite insignicant interaction eect on cognitive
eort and product quality, the two groups displayed obvious
dierences in other aspects with increased task complexity. First,
although the two groups did not expend signicantly more
cognitive eort in the complex task, the main reason behind their
decision was dierent: e high-TSEB group did not devote more
eort due to their failure in realizing a signicant increase in task
demands, whereas the low-TSEB group had low willingness to
devote more eort. Second, the two groups had observable
dierences in quality perception in the simple task, but such

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 11 frontiersin.org
dierences diminished in the complex task (refer to section Process
feature: Self-reported cognitive eort for details). is shows that
high task complexity may reduce the eect of TSEB, which lends
support to Judge etal. (2007), who believed that the role of self-
ecacy was more evident in less complex tasks.
Conclusion
is study examined students with high and low TSEB when
they performed written translation tasks across two complexity
levels. To the best of our knowledge, this study is the rst to examine
the eects of task complexity and TSEB on both the process and the
product of written translation. e research questions raised at the
beginning of the paper are addressed based on qualitative and
quantitative analysis of data from screen recording, subjective rating,
semi-structured interview, and quality evaluation.
First, the impact of task complexity was found on both the
translation process and the end product of students. Irrespective
of their TSEB level, students had longer task duration, higher
self-ratings of cognitive eort, lower accuracy, and poorer
uency in the complex task than in the simple one. e evidence
seems to reveal that, when faced with a higher level of cognitive
load, students would put in more cognitive eort. However,
unrealistically high cognitive load would reduce their
translation quality. Second, high TSEB was associated with
higher accuracy and greater uency, but did not cause
signicant dierences in time-on-task and self-reported
cognitive eort. e evidence seems to indicate that highly
ecacious students produced higher translation quality through
more exible allocation of cognitive eort rather than
expending more cognitive eort in the translation process. at
may also explain why the interaction eect of task complexity
and TSEB was not signicant on cognitive eort.
Examining the ndings in this study together with those in
previous studies, it becomes evident that the relationship
between cognitive eort and task performance is not linear,
depending on the level of task complexity. e nding proves
the importance of quantiable measures for categorizing task
complexity. Otherwise, a task considered simple in one study
might not bedened as such in another. Quantiable measures
were adopted in the present study to categorize task complexity,
which can provide a reference for future translation studies to
compare research results. Besides, the study also highlights the
necessity of problem awareness cultivation among students
since awareness of cognitive load increase is one prerequisite for
students to put in more cognitive eort. With problem
awareness in hand, students are in a better position to know
what to look for in their performance so that their performance
can beself-assessed, not just from the perspective of the end
product, but also from the perspective of the translation process
that contributes to its production.
e research ndings can, rstly, inform translation teachers
to gear task complexity to students’ developmental levels of
translation competence and to pedagogical objectives. For
instance, simple tasks can beassigned to help students build self-
ecacy, and moderately complex tasks beassigned to facilitate
their development (Graesser etal., 2011). If challenging tasks are
assigned for a particular objective, scaolds can be used to
reduce the impact of task complexity. For example, Jia et al.
(2019) found that neural machine translation can help students
address terminology issues and reduce their cognitive eort
when specialized texts, indicative of high complexity, were
assigned to develop their background knowledge. Secondly, our
ndings also highlight potential benets of TSEB. To help
students benet from high TSEB, teachers can draw on existing
research ndings on self-ecacy development, which relies on
enactive mastery experience, vicarious experience, verbal
persuasion, and physiological and emotional states (Bandura,
1997). Lastly, from a methodological perspective, webelieve that
the integrated approach adopted in this study, namely combining
process and product data for translation performance research,
allows us to bring to light results that might have been more
dicult to identify using the onefold approach. By observing
participants’ translation process, and not only their products,
future studies may develop a better understanding of
translation performance.
Although a mixed-methods design was adopted to collect
data from several sources for triangulation purposes, this study
still has some limitations. First, key-logging and eye-tracking
data could beutilized to better observe students’ translation
behaviors, so as to illustrate and explain their translation
process more vividly. Second, there are only a limited number
of source texts, single text type and language pair, and students
with similar education background involved in the experiment.
ird, the current study focuses on human translation.
Considering the recent success of neural machine translation
(Almansor and Al-Ani, 2018; Islam et al., 2021), it will
contribute further to translation performance research if
dierent task types (i.e., human translation, and post-editing of
neural machine translation) are taken into account. Future
studies could diversify the design of task features (e.g., task
type) and select participants with dierent language pairs and
diverse education backgrounds, so as to explore further the
relationships between variables in task complexity, learner
factors, and translation performance with larger samples.
Data availability statement
e raw data supporting the conclusions of this article will
bemade available by the authors, without undue reservation.
Ethics statement
e studies involving human participants were reviewed and
approved by Ethics Committee of Hunan University. e

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 12 frontiersin.org
participants provided their written informed consent to participate
in this study.
Author contributions
XZ and XW contributed to the conception of the study. XZ
conducted the experiment and draed the manuscript. XW and
XL contributed to the revision of the manuscript. All authors
contributed to the article and approved the submitted version.
Funding
is research was funded by National Social S cience Foundation
of China (grant no. 22BYY015), Education Department of Hunan
Province (grant no. 21C0791) and Social Science Evaluation
Committee of Hunan Province (grant no. XSP22YBZ035).
Acknowledgments
We would like to thank all the participants in this study.
Conflict of interest
The authors declare that the research was conducted
in the absence of any commercial or financial relation -
ships that could be construed as a potential conflict of
interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their aliated
organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or
claim that may be made by its manufacturer, is not guaranteed or
endorsed by the publisher.
Supplementary material
e Supplementary material for this article can be found
online at: https://www.frontiersin.org/articles/10.3389/fpsyg.
2022.911850/full#supplementary-material
References
Almansor, E. H., and Al-Ani, A. (2018). “A hybrid neural machine translation
technique for translating low resource languages,” in Proceedings of International
Conference on Machine Learning and Data Mining in Pattern Recognition, ed.
MLDM. Cham: Springer. 347–356.
Alves, F., and Hurtado Albir, A. (2010). “Cognitive approaches,” in Handbook of
translation studies, eds. Y. Gambier and DoorslaerL. van (Amsterdam/Philadelphia:
John Benjamins Publishing Company), 28–35.
Angelone, E. (2010). “Uncertainty, uncertainty management, and metacognitive
problem solving in the translation task,” in Translation and cognition. eds. E.
Angelone and G. M. Shreve (Amsterdam/Philadelphia: John Benjamins Publishing
Company), 17–40.
Angelone, E. (2018). “Reconceptualizing problems in translation using
triangulated process and product data,” in Innovation and expansion in translation
process research. eds. I. Lacruz and R. Jääskeläinen (Amsterdam/Philadelphia: John
Benjamins Publishing Company), 17–36.
Araghian, R., Ghonsooly, B., and Ghanizadeh, A. (2018). Investigating problem-
solving strategies of translation trainees with high and low levels of self-ecacy.
Transl. Cogn. Behav. 1, 74–97. doi: 10.1075/tcb.00004.ara
Aysa, A., Ablimit, M., Yilahun, H., and Hamdulla, A. (2022). Chinese-Uyghur
bilingual lexicon extraction based on weak supervision. Information 13, 1–18. doi:
10.3390/info13040175
Bandura, A. (1993). Perceived self-ecacy in cognitive development and
functioning. Educ. Psychol. 28, 117–148. doi: 10.1207/s15326985ep2802_3
Bandura, A. (1995). Self-ecacy in changing societies. Cambridge: Cambridge
University Press.
Bandura, A. (1997). Self-ecacy: e exercise of control. New York: W. H. Freeman
and Company.
Bandura, A. (2006). “Guide for constructing self-ecacy scales,” in Self-ecacy
beliefs of adolescents. eds. F. Pajares and T. Urdan (Greenwich, CT: Information Age
Publishing), 307–337.
Bolaños-Medina, A. (2014). Self-ecacy in translation. Transl. Interpreting Stud.
9, 197–218. doi: 10.1075/tis.9.2.03bol
Bolaños-Medina, A., and Núñez, J. L. (2018). A preliminary scale for assessing
translators’ self-ecacy. Across Lang. Cult. 19, 53–78. doi: 10.1556/084.2018.19.1.3
Brysbaert, M. (2019). How many participants do wehave to include in properly
powered experiments? A tutorial of power analysis with reference tables. J. Cogn. 2,
1–38. doi: 10.5334/joc.72
Burnham, K. P., and Anderson, D. R. (2004). Multimodel inference: understanding
AIC and BIC in model selection. Sociol. Methods Res. 33, 261–304. doi:
10.1177/0049124104268644
Charlton, S. G. (2002). “Measurement of cognitive states in test and evaluation”
in Handbook of human factors testing and evaluation. eds. S. G. Charlton and T. G.
O’Brien. 2nd ed (Mahwah: Lawrence Erlbaum), 97–126.
Chen, F., Zhou, J. L., Wang, Y., Yu, K., Arshad, S. Z., Khawaji, A., et al. (2016).
Robust multimodal cognitive load measurement. Cham: Springer Nature
Switzerland AG.
Chmiel, A., Janikowski, P., and Cieślewicz, A. (2020). e eye or the ear? Source
language interference in sight translation and simultaneous interpreting. Interpreting
22, 187–210. doi: 10.1075/intp.00043.chm
Ehrensberger-Dow, M., Albl-Mikasa, M., Andermatt, K., Hunziker Heeb, A., and
Lehr, C. (2020). Cognitive load in processing ELF: translators, interpreters, and
other multilinguals. J. Engl. Lingua Franca 9, 217–238. doi: 10.1515/jelf-2020-2039
Eickho, C. (2018). “Cognitive biases in crowdsourcing,” in Proceedings of WSDM
2018: e Eleventh ACM International Conference on Web Search and Data Mining.
ed. ACM. NewYork: Association for Computing Machinery. 1–9.
Fadaei, H., and Faili, H. (2020). Using syntax for improving phrase-based SMT in
low-resource languages. Digit. Scholarsh. Human. 35, 507–528. doi: 10.1093/llc/
fqz033
Feng, J. (2017). Comparing cognitive load in L1 and L2 translation: evidence from
eye-tracking. Foreign Lang. China 14, 79–91. doi: 10.13564/j.cnki.
issn.1672-9382.2017.04.012
Fonseca, N. B. (2015). Directionality in translation: investigating prototypical
patterns in editing procedures. Transl. Interpret. 7, 111–125. doi: ti.106201.2015.a08
Ghobadi, M., Khosroshahi, S., and Giveh, F. (2021). Exploring predictors of
translation performance. Transl. Interpret. 13, 65–78. doi: 10.12807/ti.113202.2021.
a04
Graesser, A. C., McNamara, D. S., and Kulikowich, J. M. (2011). Coh-Metrix:
providing multilevel analyses of text characteristics. Educ. Res. 40, 223–234. doi:
10.3102/0013189X11413260

Zhou et al. 10.3389/fpsyg.2022.911850
Frontiers in Psychology 13 frontiersin.org
Gu, J. T., Hassan, H., Devlin, J., and Li, V. O. K. (2018). “Universal neural machine
translation for extremely low resource languages,” in Proceedings of NAACL-HLT
2018, ed. ACL. New Orleans: Association for Computational Linguistics.
344–354.
Hart, S. G., and Staveland, L. E. (1988). “Development of NASA-TLX (Task Load
Index): results of empirical and theoretical research” in Human Mental Workload.
eds. P. A. Hancock and N. Meshkati (Amsterdam: North-Holland), 139–183.
Homan, B., and Schraw, G. (2009). e inuence of self-ecacy and working
memory capacity on problem-solving eciency. Learn. Individ. Dier. 19, 91–100.
doi: 10.1016/j.lindif.2008.08.001
Islam, M. A., Anik, M. S. H., and Islam, A. B. M. A. A. (2021). Towards achieving
a delicate blending between rule-based translator and neural machine translator.
Neural Comput. Applic. 33, 12141–12167. doi: 10.1007/s00521-021-05895-x
Islam, M. A., Mukta, M. S. H., Olivier, P., and Rahman, M. M. (2022).
“Comprehensive guidelines for emotion annotation,” in Proceedings of the 22nd
ACM International Conference on Intelligent Virtual Agents, ed. ACM. NewYork:
Association for Computing Machinery. 1–8.
Jääskeläinen, R. (1996). Hard work will bear beautiful fruit. A comparison of two
think-aloud protocol studies. Meta 41, 60–74. doi: 10.7202/003235ar
Jääskeläinen, R. (2016). “Quality and translation process research,” in Reembedding
translation process research. ed. R. Muñoz Martín (Amsterdam/Philadelphia: John
Benjamins Publishing Company), 89–106.
Jakobsen, A. L. (2002). “Translation draing by professional translators and by
translation students,” in Empirical translation studies: Process and product. ed. G.
Hansen (Copenhagen: Samfunds 1itteratur), 191–204.
Jia, Y. F., Carl, M., and Wang, X. L. (2019). How does the post-editing of neural
machine translation compare with from-scratch translation? A product and process
study. J. Spec. Transl. 31, 60–86.
Jiménez Ivars, A., Pinazo Catalayud, D., and Ruizi Forés, M. (2014). Self-ecacy
and language prociency in interpreter trainees. Interpret. Transl. Trainer 8,
167–182. doi: 10.1080/1750399X.2014.908552
Judge, T. A., Jackson, C. L., Shaw, J. C., Scott, B. A., and Rich, B. L. (2007). Self-
ecacy and work-related performance: the integral role of individual dierences. J.
Appl. Psychol. 92, 107–127. doi: 10.1037/0021-9010.92.1.107
Kelly, D. (2000). “Text selection for developing translator competence,” in
Developing translation competence. eds. C. Schäner and B. J. Adab (Amsterdam/
Philadelphia: John Benjamins Publishing Company), 157–167.
Kelly, D. (2005). A handbook for translator trainers. Manchester: St Jerome Publishing.
Kiraly, D. (2000). A social constructivist approach to translator education. London
and NewYork: Routledge.
Lehka-Paul, O., and Whyatt, B. (2016). Does personality matter in translation?
Interdisciplinary research into the translation process and product. Poznań Stud.
Contempor. Linguist. 52, 317–349. doi: 10.1515/psicl-2016-0012
Liang, J., Fang, Y., Lv, Q., and Liu, H. (2017). Dependency distance dierences
across interpreting types: implications for cognitive demand. Front. Psychol . 8:2132.
doi: 10.3389/fpsyg.2017.02132
Liu, Y. M., Zheng, B. H., and Zhou, H. (2019). Measuring the diculty of text
translation: the combination of text-focused and translator-oriented approaches.
Targets 31, 125–149. doi: 10.1075/target.18036.zhe
Lörscher, W. (1991). Translation performance, translation process and translation
strategies: A psycholinguistic investigation. Tübingen: Narr.
Malik, A. A., Williams, C. A., Weston, K. L., and Barker, A. R. (2021). Inuence
of personality and self-ecacy on perceptual responses during high-intensity
interval exercise in adolescents. J. Appl. Sport Psychol. 33, 590–608. doi:
10.1080/10413200.2020.1718798
McNamara, D. S., Graesser, A. C., McCarthy, P. M., and Cai, Z. Q. (2014).
Automated evaluation of text and discourse with Coh-Metrix. New York: Cambridge
University Press.
Mellinger, C. D., and Hanson, T. A. (2018). Order eects in the translation
process. Transl. Cogn. Behav. 1, 1–20. doi: 10.1075/tcb.00001.mel
Michael, E. B., Tokowicz, N., Degani, T., and Smith, C. J. (2011). Individual
dierences in the ability to resolve translation ambiguity across languages. Vigo Int.
J. Appl. Linguist. 8, 79–97.
Mishra, A., Bhattacharyya, P., and Carl, M. (2013). “Automatically predicting
sentence translation diculty,” in Proceedings of the 51st Annual Meeting of the
Association for Computational Linguistics (Volume 2: Short Papers ), eds. H. Schuetze, P.
Fung, and M. Poesio. Soa: Association for Computational Linguistics. 346–351.
Moores, T. T., Chang, J. C. J., and Smith, D. K. (2006). Clarifying the role of self-
ecacy and metacognition as predictors of performance: construct development
and test. SIGMIS Database 37, 125–132. doi: 10.1145/1161345.1161360
Muñoz Martín, R. (2014). “Situating translation expertise: a review with a sketch
of a construct,” in e development of translation competence: eories and
methodologies from psycholinguistics and cognitive science. eds. J. W. Schwieter and
A. Ferreira (Newcastle upon Tyne: Cambridge Scholar Publishing), 2–56.
Muñoz Martín, R., and Olalla-Soler, C. (2022). Translating is not (only) problem
solving. J. Spec. Transl. 38, 1–26.
O’Brien, S. (2013). e borrowers: researching the cognitive aspects of translation.
Targets 25, 5–17. doi: 10.1075/target.25.1.02obr
Ogawa, H. (2021). Diculty in English-Japanese translation: Cognitive eort and
text/translator characteristics. Dissertation. Ohio: Kent State University.
Plass, J. L., Moreno, R., and Brünken, R. (2010). Cognitive load theory. Cambridge:
Cambridge University Press.
Pokorn, N. K., Blake, J., Reindl, D., and Peterlin, A. P. (2019). e inuence of
directionality on the quality of translation output in educational settings. Interpreter
Transl. Trainer 14, 58–78. doi: 10.1080/1750399X.2019.1594563
Rahimi, M., and Zhang, L. J. (2019). Writing task complexity, students’
motivational beliefs, anxiety and their writing production in English as a second
language. Read. Writ. 32, 761–786. doi: 10.1007/s11145-018-9887-9
Robinson, P. (2011). “Second language task complexity, the cognition hypothesis,
language learning, and performance,” in Second language task complexity. ed. P.
Robinson (Amsterdam/Philadelphia: John Benjamins Publishing Company), 3–38.
Rothe-Neves, R. (2003). “e inuence of working memory features on some
formal aspects of translation performance,” in Triangulating translation: Perspectives
in process oriented research. ed. F. Alves (Amsterdam/Philadelphia: John Benjamins
Publishing Company), 97–119.
Saldanha, G., and O’Brien, S. (2014). Research methodologies in translation studies.
New York: Routledge.
Seufert, T., Wagner, F., and Westphal, J. (2017). e eects of dierent levels of
disuency on learning outcomes and cognitive load. Instr. Sci. 45, 221–238. doi:
10.1007/s11251-016-9387-8
Shaki, R., and Khoshsaligheh, M. (2017). Personality type and translation
performance of Persian translator trainees. Indonesian J. Appl. Linguist. 7, 122–132.
doi: 10.17509/ijal.v7i2.8348
Sun, S., Li, T., and Zhou, X. (2020). Eects of thinking aloud on cognitive eort
in translation. LANS-TTS 19, 132–151. doi: 10.52034/lanstts.v19i0.556
Sun, S., and Shreve, G. M. (2014). Measuring translation diculty: an empirical
study. Targets 26, 98–127. doi: 10.1075/target.26.1.04sun
Sweller, J., Ayres, P., and Kalyuga, S. (2011). Cognitive load theory. New York:
Springer Science + Business Media.
Sweller, J., van Merrienboer, J. J. G., and Paas, F. G. W. C. (1998). Cognitive
architecture and instructional design. Educ. Psychol. Rev. 10, 251–296. doi:
10.1023/a:1022193728205
Tang, F., and Li, D. C. (2017). A corpus-based investigation of explicitation
patterns between professional and student interpreters in Chinese-English
consecutive interpreting. Interpreter Transl. Trainer 11, 373–395. doi:
10.1080/1750399X.2017.1379647
Trace, J., Janssen, G., and Meier, V. (2015). Measuring the impact of rater
negotiation in writing performance assessment. Lang. Test. 34, 3–22. doi:
10.1177/0265532215594830
Waddington, C. (2001). Dierent methods of evaluating student translations: the
question of validity. Meta 46, 311–325. doi: 10.7202/004583ar
Wang, F. X. (2022). Impact of translation diculty and working memory capacity
on processing of translation units: evidence from Chinese-to-English translation.
Perspectives 30, 306–322. doi: 10.1080/0907676X.2021.1920989
Weng, Y., Zheng, B. H., and Dong, Y. P. (2022). Time pressure in translation:
psychological and physiological measures. Targets 34, 601–626. doi: 10.1075/
target.20148.wen
Whyatt, B. (2019). In search of directionality eects in the translation process and
in the end product. TCB 2, 79–100. doi: 10.1075/tcb.00020.why
Wu, Z. W. (2019). Text characteristics, perceived diculty and task performance
in sight translation: an exploratory study of university-level students. Interpreting
21, 196–219. doi: 10.1075/intp.00027.wu
Yang, Y. X., Cao, X., and Huo, X. (2021a). e psychometric properties of
translating self-ecacy belief: perspectives from Chinese learners of translation.
Front. Psychol. 12:642566. doi: 10.3389/fpsyg.2021.642566
Yang, Y. X., Wang, X. L., and Yuan, Q. Q. (2021b). Measuring the usability of
machine translation in the classroom context. Transl. Interpreting Stud. 16, 101–123.
doi: 10.1075/tis.18047.yan
Yuan, R. J. (2022). Material development for beginner student interpreters: how
does text structure contribute to the diculty of consecutive interpreting?
Interpreter Transl. Trainer 16, 58–77. doi: 10.1080/1750399X.2021.1950979