Authors
Payame Noor University, Tehran, Iran
Abstract
Keywords
Main Subjects
Introduction
For many years, researchers have been eager
to find out whether different levels of task
structure and cognitive complexity have any
noticeable impact on learners’ oral
production. This subject has always been
controversial since the degree to adjust the
task, according to learners’ ability, has
always attracted the researchers’ attention.
There is general consensus on the claim that
planning and other factors such as cognitive
complexity of the task might have crucial
impact on oral production (Ahmadian, 2012;
Ahmadian&Tavakoli, 2011; Ellis, 2000;
Skehan & Foster, 1996; Wendel, 1997; Ellis
2003).
Jeon and Hahn (2006) express that task-based language teaching has a substantial
implication for the area of language
learning. They maintain that learning is a
developmental process with the aim of
promoting communication and social
interaction rather than acquiring a product
by practicing language items. Besides, they
believe that learners learn the target
language more effectively when they are
naturally exposed to meaningful task-based
activities. Although early empirical studies
strongly support the use of task as a
beneficial way to conceptualize language
teaching, the amount of research in this area
is still not sufficient. Preparing suitable tasks
demand a great deal of exploration in related
studies and a deep insight through the
influence of the task type on learners’ oral
production in terms of accuracy and
linguistic complexity. Therefore, the use of
task-based programs will be open to more
research (Skehan, 1998).
According to Revesz (2011), task
complexity can affect attentional allocation
and the focus on second language (L2)
constructions during task performance, and
this can influence the quality of learning.
However, there were some inconsistencies
among the findings of applied linguistics
regarding the impact of task type on
accuracy and linguistic complexity. To
bridge the gap and to understand the
importance of selecting appropriate tasks, it
is necessary to conduct more studies in the
field.
Literature review
Since 1980, second language acquisition
(SLA) researchers have been interested to
explore the impact of task cognitive
complexity on oral production. A clear
understanding of the load of cognitive
demands on participants can help material
producers to design appropriate tasks for
learners. Hence, tasks’ management is a
crucial basis for communicative language
syllabus.
Skehen’s model of task complexity
To Skehan (1998), three factors are
associated with task difficulty: code
complexity (the syntactic and lexical
difficulty of input), cognitive complexity
(the processing demands of the tasks), and
communicative stress (time pressure and the
modality demand). These factors can
produce different demands, and therefore
can influence the quality of learners’
performance (Taguchi, 2007).
Another factor included in Skehan’s model
is planning time. Previous research has
explored the effect of planning time on L2
output. Planning time would help learners to
produce more accurate as well as greater
level of lexical complexity. However, in
case of accuracy, results could be different.
Some studies showed that certain task types
may lead to more accurate speech while
others proved that task condition is an
influential factor which determines the
degree of accuracy. Skehan (1998), as
reiterated in Iwashita, McNamara, and Elder
(2001), believes that different numbers of
factors have impact on task difficulty. To
him, task dimensions such as abstractness
and familiarity of task information can
influence the difficulty of the task. He
maintains that performance conditions (e.g.,
concrete vs. abstract information) play a
crucial role in determining the level of task
difficulty.
Iwashita, McNamara, and Elder’s task
dimensions
According to Iwashita et al. (2001),
Skehan’s framework was encouraging;
however, some aspects of the framework
were questionable such as: (1) the notion of
difficulty, (2) assessing the candidate’s
performance, (3) inconsistent results, and (4)
the complexity of task performance.
Iwashita et al. (2001) consider four
dimensions for narrative tasks, with two
different performance conditions (+ or -).
These task dimensions and performance
conditions are defined as follows:
Perspective: When a story is told as if it
happened to the participant, more accurate
but less complex response is produced.
However, telling the story from others’
viewpoint will make a task more difficult
and the result would be different.
Immediacy:If learners have access to the
pictures while telling a story, their speech
would be more accurate, but less complex.
On the other hand, telling a story without
pictures in view would be less accurate;
hence, narrative tasks considering there-and-
then condition are cognitively more complex
than those referring to here-and- now
condition. Iwashita et al. (2001) associate
this complexity with more complex syntax
and multi-propositional utterances.
Adequacy: Using a complete set of pictures
while narrating (- condition) would make a
task less difficult; in this case learners would
produce more accurate, but less complex
sentences. On the other hand, if some
pictures are missing, the result would be
opposite.
Planning time:Considering an appropriate
planning time in narrative task would make
learners’ oral production more accurate.
However, narrating without considering the
time limitation may lead to some mistakes in
learners’ oral production.
Robinson’s task complexity dimension
Robinson’s (2005) cognition hypothesis
could be the most prominent model which
was devised to examine the impact of
increase in task cognitive complexity on oral
production. According to Robinson (2005),
the cognition hypothesis reveals that more
difficult tasks may produce L2 production
with more accuracy and more linguistic
complexity. Based on the hypothesis,
complex tasks produce interactional
processes such as corrective feedback and
noticing of input. He believes that
researchers need to determine what
differences L2 tasks, with different levels of
complexity, make to learners’ performance,
so that they can sequence and grade the
tasks on a proper basis.
When learners perform more than one task
at the same time, they actually experience a
real world situation; in this case, task
complexity is increased along resource-dispersing dimensions. On the other hand,
increasing task complexity along resource-directing dimension (e.g., asking for
justification) can motivate learners to use
specific L2 constructions. Resource-directing variables of task complexity
demand a great deal of attention and
working memory, and make learners focus
on linguistic forms. Some examples of
resource-directing factors are: [± few
elements], [±here and now], and
[±reasoning]. The low complexity
conditions include [+ few elements], [+ here
and now], and [- reasoning] and the high
complexity conditions are associated with [-
few element], [-here and now], and
[+reasoning] (Robinson, 2001; 2005).
According to Robinson, learners’ attention
can be diverted over many L2 elements
when the task complexity is increased along
recourse-dispersing dimensions. Some
examples of resource-dispersing factors
include: [± planning], [±single task], and [±
prior knowledge]. The low complexity
conditions are [+planning], [+single task],
and [+prior knowledge] while the high
complexity conditions include [-planning],
[-single task], and [-prior knowledge]
(Robinson, 2005). The cognition hypothesis
proves that when we increase the cognitive
complexity of the task, learners show more
accurate, but less fluent language.
Furthermore, in more complex tasks,
interactional processes such as noticing and
corrective feedback are noticeable (Kim,
2009).
This study was an attempt to explore the
extent to which cognitive complexity in
tasks could have an effect on Iranian EFL
learners’ oral production. To identify the
relations, contradictions, and gaps in the
literature, the following research questions
were formulated to check the aim of the
study:
1. Does increase in task cognitive
complexity affect the accuracy of
Iranian EFL learners’ oral
production?
2. Does increase in task cognitive
complexity affect the linguistic
complexity of Iranian EFL learners’
oral production?
Methodology
This quasi-experimental research drew
preliminary on Robinson’s cognition
hypothesis which was a foundation for
investigating the impact of cognitive
complexity on the aspects of oral
performance such as accuracy and linguistic
complexity.
Participants
The participants in this study were 30
students at the intermediate level between
the ages of 21 to 34. They were all female
students studying general English at Kish
Institute in Tehran, Iran. The participants
were all native speakers of Persian, and on
average they had been studying English for
three years. Based on non-random sampling,
the participants were selected from two
intact classes, with 15 participants in each.
A version of Test of English as a Foreign
Language (TOEFL) was administrated to
assure the homogeneity of the participants in
the study.
Materials
Testing materials
In order to evaluate learners’ oral
production, two tests were used by the
researchers: speaking tests from TOEFL
IBT book by McGraw (2006) for checking
the homogeneity of the participants, and a
post-test with the aim of measuring and
comparing oral skill of the two groups.
Teaching materials
For teaching materials and treatment, some
narrative tasks were selected from English
Result by Hancock and McDonald (2012)
with tasks and the exercises designed for
intermediate level. The teacher and the
researchers agreed that the tasks in this book
were appropriate for the aim of the treatment
because the content of the book covered
both simple as well as complex tasks which
could be suitable for assessing learners’ oral
skills according to resource-directing
elements.
Data collection and procedure
The data in this study drew mostly from the
participants’ oral production which was in
the form of picture narration tasks. All the
necessary data were collected during one of
the students’ regular term. The classes lasted
two hours, and two experimental groups
participated in this study. For the aim of the
study, three elements of Robinson’s
resource-directing model, +/- few elements,
+/- reasoning demand, and +/- here and now
condition, were checked along accuracy and
linguistic complexity.
To ensure that the tasks’ design
manipulation was appropriate for the
purpose of the study, two raters, experienced
English teachers, cooperated with the
researchers in estimating the level of
cognitive complexity of the tasks. For each
element of this model, the participants were
asked to perform two versions of the same
task (+ condition and – condition). Hence,
six sources of data were analyzed to answer
the two research questions.
Checking homogeneity
Prior to the administration of the tasks, to
assure homogeneity of participants, some
speaking tests from TOEFL IBT book by
McGraw (2006) were chosen and the
participants’ voices were recorded; based on
Speaking Rubric the responses were scored
from 4 to 0. The results of the independent t-test [t (28) = .843, p = .406] indicated that
the two groups were homogeneous in terms
of their general language proficiency. The
mean scores for the two groups were 17.66
and 17.31 respectively which point to
homogeneity of the two groups.
Treatment. The treatment plan for the first
experimental group included working on
more difficult tasks (see Appendix A).
These tasks demanded more causal
reasoning as well as justification for the
replies. The practice plan for these
participants included all the aspects which
were aimed at the post-test such as +/-
reasoning demand, +/- few elements, and
here and now vs. there and then condition.
For the reasoning demand aspect of the task,
two sets of pictures were selected, one with
correct order and the other with scrambled
pictures. For checking the impact of number
of elements on learners’ oral production, the
researchers asked the participants to narrate
the story once with 9 pictures and the other
time with 6 pictures (the three last pictures
were omitted).
For the last aspect of Robinson’s (2001)
resource-directing model, here and now vs.
there and then condition, the participants
were first required to tell the story with
pictures in front of them. Next, the
participants were asked to turn the picture
strips over before beginning their narration.
For both narrations, the participants were
given prompts and instruction. For each
dimension, two conditions (+ condition and
– condition) were needed to be tested, so all
the participants in both experimental groups
were supposed to Perform six tasks. Table 2
illustrates a brief description of test tasks
based on Robinson’s model. The
participants in the second experimental
group were exposed to simpler tasks in the
form of picture narration (see Appendix B).
For the simpler task, a set of four pictures
were selected with a topic familiar to the
participants, and the task did not require
causal reasoning, justification of beliefs, or
any kind of interpretation.
Post Test. The post-test was administered
one week after the treatment. The
researchers preferred a monologic picture
description task to elicit participants’ oral
performance (see Appendix C).
As mentioned earlier, each group performed
two versions of the same narrative task: a
simple and a complex version. The
participants had three minutes to read the
instructions, take note, and prepare their
answers.
Therefore, the three mentioned dimensions
in Robinson’s resource-directing model, [±
few elements], [±here and now], and
[±reasoning], were examined in the study.
As Table 1 shows, the low complexity
conditions include [+ few elements], [+ here
and now], and [- reasoning] while the high
complexity conditions are [- few element], [-here and now], and [+reasoning].
It is also worth mentioning that in order to
avoid practice effect, counterbalancing was
suggested by the researchers.
Rating Scale. To measure participants’ oral
production in terms of accuracy and
linguistic complexity in the post-test, an
analytic rating scale proposed by Iwashita et
al. (2001) was chosen. In this rating scale,
linguistic control as well as managing forms
and grammar are among the most prominent
factors while assessing accuracy. Based on
Iwashita et al. (2001), attempting a variety
of verb forms (e.g., passive, modals, and
tense), taking grammatical risks in order to
express complex meaning, and using
coordination and subordination to transfer
ideas were the basis for checking linguistic
complexity.
Results
The results of MANOVA are presented in
two subcategories: The impact of the
increase in task cognitive complexity on
accuracy and linguistic complexity.
The impact of the increase in task cognitive
complexity on accuracy
This part depicts the analysis and findings
for the first research question: Does increase
in task cognitive complexity affect the
accuracy of learners’ oral production?
Based on the outcome of MANOVA, the two
experimental groups’ means were compared
on the three accuracy tests (the elements of
Robinson’s model). According to Table 3,
the following results were obtained:
F (3, 36) = 11.24, p = .00, and Partial η2 =
.56.
As the Table reveals, the F-observed value
calculated for the effect of the difficulty
level on the students’ overall accuracy in
oral production was statistically significant.
The statistics on the mean score displayed in
Table 4 and Table 5 revealed that on average
the experimental group 1 (M = 1.93, SD =
.65) was more successful than the
experimental group 2 (M = 1.33, SD = .57)
on the accuracy in tasks checking reasoning
demands , F (1, 28) = 12.06, p = .002, and
Partial η2 = .30.
Since the F-value of 11.24 indicated a
significant difference between the
experimental group 1 (with difficult tasks)
and the experimental group 2 (with easy
tasks), the two groups’ performances on the
three tasks including reasoning demands,
number of elements, and here-and-now vs.
there-and-then condition were compared.
The results displayed that the experimental
group 1 (M=1.93, SD=.50) had higher means
than the experimental group 2 (M = 1.46, SD
= .69) on the accuracy tests while checking
number of elements. As shown in Table 5,
there was a significant difference between
the two groups’ means on the accuracy tests
as far as the number of elements was
concerned, [F (1, 28) = 6.86, p= .00, and
Partial η2 = .19].
Accordingly, another MANOVA test was run
to check the results of groups’ performance
in tasks considering accuracy with two
conditions (here-and-now vs. there-and-then). A simple comparison, based on Table
4, demonstrated that the experimental group
1 (M = 1.93, SD = .47) showed greater
efficacy than the experimental group 2 (M =
1.33, SD = .55) on the accuracy in tasks
checking the mentioned conditions.
Furthermore, according to Table 5, there
was a large effect size [F (1, 28) = 12.06, p
= .002, and Partial η2 = .30]. In other words,
there was a significant difference between
the two groups’ performance on the
accuracy of here-and-now vs. there-and-then
condition.
As Figure 1 and the obtained information
reveal, the first null hypothesis could be
rejected.
The impact of the increase in task cognitive
complexity on linguistic complexity
In this section, we investigate the findings
for the second research question: Does
increase in task cognitive complexity affect
the linguistic complexity of learners’ oral
production?
The results of performing another MANOVA
test indicated that the F-observed value
obtained from the students’ performances on
overall linguistic complexity was
statistically significant [F (3, 36) = 24.68,
p= .00, and partial η
2
= .74]. The analysis
also specified a large effect size (see Table
6); hence, the second null-hypothesis could
be rejected
According to Table 6, the F-value of 24.68
revealed a significant difference between the
means on linguistic complexity of oral
production; however, the two groups’
performance on the three elements of
Robinson’s model was also compared.
Based on the descriptive statistics displayed
in Table 7 and Table 8, it could be realized
that on average the experimental group 1 (M
= 2.06, SD = .60) had better results than the
experimental group 2 (M = 1.13, SD = .60)
on the linguistic complexity in reasoning
demands tasks. Comparing the results in
Table 8 proved a large effect size revealing a
significant difference between the two
groups’ means on linguistic complexity [F
(1, 28) = 27.44, P = .00, and Partial η2 = .49].
The data in Table 7 depicted that the
experimental group 1 (M = 2.20, SD = .69)
outperformed the experimental group 2 (M =
1.06, SD= .68) on linguistic complexity
while performing tasks with different
number of elements. According to Table 8,
there was a large effect size [F (1, 28) =
36.78, p = .00, and Partial η2 = .56]. the
data proved a significant difference between
the two groups’ means on linguistic
complexity in tasks with different number of
elements.
As displayed in Table 7, on average, the
experimental group 1 (M = 2.06, SD = .74)
got better scores than the experimental
group 2 (M = 1.53, SD = .49) on linguistic
complexity on here- and-now vs. there-and-then condition. The obtained data from
Table 8, [F (1, 28) = 6.89, p = .01], reveals a
significant difference between the two
groups’ means on linguistic complexity
while performing the tasks in two
conditions. Figure 2 supports the statistical
information.
Discussion
In the case of oral production, adjusting the
complexity of the task appears to be one of
the prominent aims of syllabus designers
because based on different studies such as
the ones conducted by Revesz (2011),
Iwashita, et al. (2001), and Robinson (2005),
human beings possess a limited processing
capacity and are not capable of attending
fully to all aspects of a task. When different
aspects of oral production are concerned, an
appropriate level of cognitive complexity in
tasks would help learners to promote their
speaking. This study sought to explore the
effects of increase in task cognitive
complexity on accuracy and linguistic
complexity of Iranian EFL learners’ oral
production in narrative tasks.
In general, the results of the study indicated
that the group that had exposure to more
complex tasks presented more accurate
speech with more complex structure. This
difference in the outcome was noticeable in
all the three aspects of Robinson’s resource-directing model.
The effect of increase in task cognitive
complexity on accuracy
The findings revealed that scrambled
pictures, designed for checking reasoning
demands, resulted in more accurate speech.
The major explanation for this difference
could be that while telling a story with
scrambled pictures, participants tried to rely
on their logical concepts in order to put the
pictures in the correct order on their own
way, and for doing that they needed to get
help from their knowledge repertoire.
The results concerning accuracy are in line
with Skehan’s (1998, as cited in Taguchi,
2007) study. The findings also support the
claim of Iwashita et al. (2001) about the
effect of complex tasks on directing
learners’ attentional resources to forms
which can lead them to induce risk avoiding
behavior. Robinson (2001; 2005, as cited in
Revesz, 2011) also emphasizes the
beneficial role of complex tasks on making
learners focus on linguistic forms. Besides,
Tavakoli’s (2009) cognitive approach
towards language learning supports the
findings in this study.
In the case of number of elements, the group
that performed the narration task with more
pictures outperformed the group that did the
task with fewer pictures. This might be
because using more pictures while narrating
produces more sentences, so participants
were concerned with the correct connection
in order to make their sentences
comprehensive. This made the participants
process their speech in their minds before
revealing it. This reformulation and
consistent monitoring resulted in more
accurate speech. The outcome bears out the
previous studies conducted by Kim (2009),
Iwashita et al. (2001), and Robinson (2001,
2005, as cited in Revesz, 2011).
The study also highlights the findings
regarding accuracy while checking here-and-now vs. there-and-then condition. In the
case of telling a story with pictures in view
(here-and-now), the participants produced
less accurate sentences than when they had
performed the same task without pictures in
front of them (there-and-then). The point is
that there-and-then condition (telling the
story without pictures) put more burdens on
participants’ working memory and the
participants paid more attention to the forms
before producing their speech, therefore
their speech were more accurate. This
viewpoint is in accordance with Revesz’s
(2011) statement.
The effect of increase in task cognitive
complexity on linguistic complexity
The results of the study revealed that
increasing the level of task difficulty led to
more complex speech in all the three
elements of Robinson’s model. In fact, in
telling a story with scrambled pictures,
participants produced more complex
sentences than the time they were exposed
to pictures in the correct order. Scrambled
pictures made participants process the
information in their minds while seeking for
some new and more complex structures.
Besides, participants were more willing to
provide reasons to prove why they put the
pictures in that order. This led them to use
new and more complex structures; this
stance is supported by Robinson’s (2001,
2005, as cited in Revesz, 2011) model.
The findings demonstrated that narrating the
story with more pictures led to more
complex language. This could be due to the
use of more complex syntax which was
associated with the tendency of using more
transitions and coordination conjunctions.
Other scholars such as Talmy (2000, as cited
in Revesz, 2011) pointed to this impact as
well.
As far as here-and-now vs. there-and-then is
concerned, it could be realized that in
second condition participants tended to
remember the events and they relied on their
memory to connect the sentences in a
coherent way and that led the participants to
produce not only more accurate speech but
also more complex syntax. This attitude
supports Iwashita et al.’s (2001) approach in
that more attention to forms and planning of
production is due to the greater demands on
memory and the attempt to make transition
between the events.
Conclusion
Generally, Iranian EFL learners have
difficulty producing an accurate speech. To
solve this problem, we need a
comprehensive insight into methodology in
general and designing and adjusting the
level of tasks in particular. Based on the
findings of the present study, the group that
performed more cognitively difficult tasks
was more successful in terms of accuracy
and linguistic complexity; therefore, the first
and second null hypotheses were rejected.
The results confirm the beneficial impact of
increasing the task cognitive complexity on
speaking, especially in accuracy and
linguistic complexity. This in turn can be an
acceptance for Robinson’s cognition
hypothesis with his emphasis on the
promotion of learners’ oral skills along with
opting for more challenging tasks. This
notion and the previous studies as well as
the results of the current research indicate
that increasing the difficulty of the task to a
reasonable level can be highly effective for
learners’ improvement in speaking.
However, there were some limitations in the
current study. First, gender differences were
not considered in this research; the
participants were all females, hence the
results may not be generalized to
coeducational systems. Second, choosing the
non-random sampling method and selecting
the participants from intact classes as well as
small sample size could have influence on
external validity.
Acknowledgment
The present study has been supported by
Payame Noor University, Iran.
)