Embed Size (px)
Citation preview
3
English Teaching, Vol. 70, No. 2, Summer 2015
DOI: 10.15858/engtea.70.2.201506.3
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output
Ji Hyun Kim
(Keimyung University)
Kim, Ji Hyun. (2015). The role of models and error correction in L2 young
learners’ noticing and output. English Teaching, 70(2), 3-26.
Despite the increasing interest in noticing by second language (L2) learners during the
output process, little attention has been given to child L2 learners’ noticing and output.
Thus, the present study investigated what child L2 learners of different English
proficiency levels noticed as they composed a text and received written corrective
feedback (WCF) (error correction vs. models), and how they incorporated the noticed
features into their revised texts. Data were collected from twenty-six child pairs
throughout three stages (composition, comparison, and revision stages) and note-taking
was employed as a means of measuring learner noticing. It was found that learners
were able to initiate noticing on their own when composing texts, and that high-
proficiency learners tended to attend to grammatical problems more frequently than
medium/low-proficiency learners. WCF played a facilitative role in leading learners to
notice, and their attention to language was mediated differently by different types of
WCF. Learners incorporated the noticed features into their revisions, and textual
revisions were mostly lexical. Error correction (EC) triggered more grammatical
revisions, while the model text helped learners to notice something beyond what they
could produce on their own. These findings suggest a facilitative role for output and
WCF as a means of learner noticing, and the different roles of EC and model texts.
Key words: written corrective feedback, direct error correction, reformulation, learner
output, noticing
1. INTRODUCTION
The role of written corrective feedback (WCF) has been assessed from two different
perspectives, “writing to learn” and “learning to write” (Manchón, 2009). The former
approach is framed within second language acquisition (SLA), which explores the role of
WCF in second language (L2) learners’ linguistic development (Bitchener, 2012). In
4 Ji Hyun Kim
contrast, the “learning to write” approach puts an emphasis on the role of WCF in
facilitating L2 learners’ overall writing process and skills (Ferris, 2010). Although these
two approaches are certainly different with respect to the primary purpose for providing
WCF, they share some common interests – the value of written output process and learners’
use of WCF. The present research, within the theoretical frame of SLA, attempts to explore
these issues.
Theoretically, the potential role of written output and WCF in L2 learning is closely
related to the importance of noticing in SLA. According to Schmidt’s (2001) noticing
hypothesis, “people learn the things that they attend to and do not learn much about the
things they do not attend to” (p. 30) (Robinson, 1995; Schmidt, 1990, 2001). In other
words, attention at the level of noticing is a necessary condition for SLA to take place since
only noticed input can be converted into intake and lead to L2 learning. Schmidt (1990,
2001) especially points out the importance of noticing the gap (i.e., noticing the difference
between L2 learners’ interlanguage and the target forms that appear in the input).
One of the conditions that could enhance L2 learners’ noticing is being positioned to
produce output. Swain (1985) asserts that output may serve as “the trigger that forces the
learner to pay attention to the means of expression needed in order to successfully convey
his or her own intended meaning” (p. 249). Output constitutes not just the means by which
to improve learners’ language fluency through practice but also a significant factor in the
SLA process. More specifically, through the output process, learners may have unique
opportunities for active deployment of their cognitive resources (i.e., hypothesis testing,
feedback, noticing, and metalinguistic reflection) which may not be provided in the process
of meaning-oriented comprehension (Swain, 1985, 1995).
While learners produce output, learners may notice what they can and cannot express
(i.e., noticing the hole). Then they attempt to solve their linguistic deficiency in ways that
are appropriate in a given condition. If learners are left on their own to solve their problems,
they may consolidate existing knowledge or generate alternatives based on their current
interlanguage knowledge. On the other hand, when relevant information is immediately
provided through some form of WCF, learners may actively seek solutions to the problems,
referring to the WCF (Izumi, 2002, 2003). In this process, WCF may lead learners to notice
the difference between their IL and the target form and fill in the gaps in their IL
knowledge.
The aforementioned theoretical explanations of the relationships among noticing, output,
and WCF seem plausible. However, not many empirical studies have explored what
learners indeed notice in the process of written output and how such noticing affects their
process and use of WCF. In particular, child L2 learners have rarely been involved in such
studies. For this reason, the present research attempts to investigate child L2 learners’
noticing during the process of writing, and their noticing and use of WCF delivered in two
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 5
forms (i.e., direct error correction vs. models).
2. NOTICING AND CORRECTIVE FEEDBACK
Ever since Schmidt (1990) proposed the Noticing Hypothesis, the role of noticing in
SLA has been hotly debated. Schmidt (2001) claims that noticing is the conscious
perception of “the surface structure of utterances in the input – instances of language,
rather than any abstract rules or principles of which such instances may be exemplars” (p.
5). In addition, what is noticed is “not just the raw data of the input” but “input as
interpreted by existing schemata” (p. 30). Even though the level of noticing necessary for
SLA has long been a controversial issue, a consensus has been reached that a certain
degree of noticing is crucial to promote L2 learning process (Robinson, 1995). In fact, it
has been claimed that one of the advantages of CF over positive evidence is that CF would
lead to learner noticing of linguistic problems (Long, 1996).
The beneficial role of CF in SLA has been discussed on the premise that it would trigger
learner noticing. However, CF research has been outcome-oriented and focused on how CF
affects learners’ actual development in their output (Jang, 2012). As a result, how learners
actually perceive and use CF has been overlooked (Coyle & Roca de Larios, 2014). With
the acknowledgement of the importance of process-oriented research, researchers have
started looking into learner perception of CF, questioning whether or not CF actually
promotes learner noticing and further understanding of the linguistic problems it targets.
Many oral CF studies reported a considerable mismatch between learner perception and
teacher intent (Egi, 2007; Kim & Han, 2007; Mackey, Gass, & McDonough, 2000). Such
findings have given rise to follow-up research that explores the conditions which enhance
learners’ accurate understanding of the teacher’s CF (Egi, 2007; Kim, 2011).
While learner noticing of oral CF has been widely examined, little WCF research has
looked into how learners actually notice and process it. This lack of interest in learners’
processing might be due to the assumption that they may easily notice WCF since it is
provided in written form. For direct error correction, in particular, noticing seems to be
considered guaranteed, since it explicitly points out errors by crossing out an unnecessary
word or phrase, inserting a missing word, or providing the correct form. Noticing of CF,
however, does not guarantee learners’ accurate interpretation of teacher CF. For instance, in
her longitudinal case study, Han (2001) found that a learner’s persistent error was
attributed to her misinterpretation of WCF, notwithstanding a teacher’s consistent error
correction (mostly written direct CF). Through interviews, Han found that the learner
noticed CF, but her L1 misguided her interpretation of the teacher’s intended meaning and
the nature of errors. As Han’s study shows, it may be naive to assume that learners will be
6 Ji Hyun Kim
able to notice written CF, and that this will lead to the correct understanding of the errors.
One of very few studies which explored how learners perceive written direct error
correction is Kim’s (2013) investigation of the extent to which 32 Korean university
students of English understood written direct error correction they received from the
teacher in intact writing classes. In addition, the study investigated whether or not the
presence and quality of understanding of CF was related to learners’ immediate uptake and
retention of CF. Kim reported that the students did not understand WCF all the time even
though the CF was explicit (i.e., nearly half of CF was not understood). This empirically
proves that assuming that WCF will be easily understood is no more than a benign
supposition. When the study considered the quality of understanding, only a third of CF
was correctly understood and two-thirds of CF were not understood or were misinterpreted.
The result indicates that only a third of CF was processed in a form the learners could
utilize for L2 learning (Schmidt, 2001).
Learner processes of other types of written CF have also received attention.
Reformulation (i.e., a teacher rewriting a learner’s text while preserving all its ideas but
removing lexical and grammar errors) is one of them. It has been claimed that
reformulation has advantages over traditional feedback methods such as direct error
correction because it provides more extensive feedback on improving L2 learners’ essays,
especially in the area of cohesion (Cohen, 1982). Of particular interest to the present
research is the claim that reformulation pushes learners to actively engage in cognitive
comparison between their original texts and the reformulated ones (Adams, 2003), while
traditional feedback does not necessarily require learners to engage in active cognitive
processing (Kim, 2013). Qi and Lapkin (2001) provided more in-depth explanations for
noticing triggered by reformulation and its effects on L2 development by examining two
different levels of noticing: perfunctory (e.g., noticing only and without giving reason of
being given CF) or substantive (i.e., noticing and providing reasons of being given CF)
noticing. They explored what two Chinese learners of English noticed in the composing
stage and subsequent reformulation stage. In the reformulation stages, the learners were
given the opportunity to compare their original drafts with the reformulated ones and were
asked to report what they noticed. Later, the learners were asked to revise their initial drafts.
Based on the analyses of the learners’ verbal reports, Qi and Lapkin reported that while
composing and reformulation promote noticing, the quality of noticing was closely related
to L2 writing improvement (Kim, 2013).
When the extent of noticing triggered by reformulation is compared with other types of
feedback, however, researchers have found reformulation relatively less effective than
others (Sachs & Polio, 2007). For instance, Sachs and Polio (2007) reported, based on their
analysis of verbal reports through think-aloud, that learners who received error correction
(i.e., direct error correction) showed better noticing of their problems, and this led to more
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 7
improvement in L2 linguistic accuracy compared to those who received reformulation as
feedback. Storch and Wigglesworth (2010) also investigated the efficacy of reformulation
and indirect error correction by examining collaborative dialogues between learners. It was
reported that indirect error correction elicited more language-related episodes and a higher
level of engagement with feedback than did reformulation. However, considering that there
has not been sufficient research regarding relative effects of reformulation, it would be
premature to reach any conclusion.
Written CF with respect to learner processing has also been examined in relation to
providing models. While reformulation seems practically difficult to be used in intact
writing classrooms (Coyle & Roca de Larios, 2014), models seem more feasible since it
does not require teachers to provide individual feedback. Instead, learners receive model
texts which are “tailored to the learners’ age and proficiency level as well as to the content
and the genre of the writing task at hand” (Coyle & Roca de Larios, 2014, p. 3). The
purpose of using models is to provide learners with a means to compare their own
productions with models, but not to imitate them. Namely, models are used as resources
which can help learners solve the problems in their writing. Producing the TL can
encourage learners to recognize what they can and cannot express, and this eventually
leads learners to actively look for relevant information in the input. In this regard, models
provided after learners have produced writing offer an adequate and useful resource which
can solve the recognized problems. In spite of the potential usefulness of models as
feedback in writing classes, using models as feedback has been overlooked thus far.
Hanaoka (2007b) is one of the few studies exploring how learners process models of
writing. Thirty-seven Japanese EFL learners participated in a multiple-stage writing task
consisting of composition, comparison, and revision. The learners autonomously noticed
their linguistic problems while writing a composition, and actively sought to solve the
problems through a model text provided to them at the comparison stage, incorporating
them in subsequent revisions as well. Hanaoka’s research shows that models lead to learner
noticing of linguistic problems. However, Martínez Esteban and Roca de Larios (2010)
reported incompatible findings: in their study, learners attended more to the content of the
models than language forms. The researchers explained that the different results might be
due to their learners’ low level of L2 proficiency. More recently, Hanaoka and Izumi
(2012) investigated the role of models and reformulation by looking at how they
differentially played a role in disparate types of problems–overt and covert. Overt
problems referred to instances where participants tried out their hypothesis in their writing,
while covert problems referred to the cases where participants did not address their
hypothesized forms in their output. The study found that both overt and covert problems
were both equally solved in the feedback texts; however, the model text provided solutions
to covert problems more frequently than did the reformulation.
8 Ji Hyun Kim
The aforementioned studies account for how learners actually process written CF and
provide some evidence that partly supports the role of CF in triggering learner noticing.
However, all these studies were interested in adult learners, and child L2 learners have
remained outside of researchers’ scope of interest. One of the few studies in child L2
learners is Coyle and Roca de Larios (2014). They examined how 11- and 12-year-old L2
learners noticed two forms of feedback (i.e., error correction and model texts) provided in
response to their collaborative writing task. It was found that error correction resulted in
more noticing errors and more incorporation of the noticed errors into revisions, which
coincided with the findings from studies carried out with adult learners. However, it is too
premature to conclude that children process CF as adults do with this very limited finding.
Coyle and Roca de Larios’ (2014) research makes the research into learner noticing of
CF move forward by dealing with child L2 learner’s noticing and learner proficiency levels,
which have rarely been explored. However, as Coyle and Roca de Larios admitted, its
effect on noticing was not reasonably examined in their study due to the imbalanced
number of learners across feedback conditions, which consequently hindered them from
claiming any relationship between learner proficiency levels and noticing. In addition, the
absence of control group in their study cast doubt on whether learner noticing was
triggered by feedback or by other variables.
In order to explore the on-going issues surrounding learner noticing of CF and fill in the
gaps of previous research, the present study aims to investigate how child L2 learners of
different proficiency levels process and use two different types of written CF (i.e., error
correction and a model text) provided to their written tasks in intact classes. More detailed
research questions are as follows:
1. What features of languages do child EFL learners notice when they compose a text?
Is this related to different proficiency levels?
2. What do child EFL learners notice in the process of written feedback (i.e., error
correction or model texts) provided for their written texts? Is this associated with
their proficiency levels?
3. How is the child learners’ reported noticing related to their revised texts? Is this
related to their proficiency levels?
3. METHODS
3.1. Participants
Twenty-six pairs of children, 28 girls and 24 boys aged 10-11, participated in the study.
They were enrolled in a private English academy in Seoul and had been learning English
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 9
for 2 or 3 years. Their classes were based on communicative language teaching covering
four skills, and the classes met three times a week for 80 minutes. The children were
accustomed to various writing tasks including collaborative writing. This kind of writing
was chosen for the present study for the following reasons. First, it has been argued that
writing in pairs can lead to gains in accuracy (Storch, 2007; Storch & Wigglesworth, 2007;
Wigglesworth & Storch, 2012). In addition, collaborative writing can reduce children’s
cognitive demands while composing texts and can, in turn, create a safe learning
environment (Yarrow & Topping, 2001).
The children were placed in proficiency-matched pairs based on performance tests
administered in class. The tests consisted of listening comprehension and reading and
writing. Six high, six medium, and six low-proficiency pairs (total 18 pairs) were formed
as an experimental grouping. Three high, two medium, and three low-proficiency pairs
(total 8 pairs) were formed as a control group.
3.2. Data Collection
Data were collected for a period of three weeks. For the purpose of collecting natural
data as much as possible, instructors teaching children’s regular classes were in charge of
collecting data. A simple four-frame picture story taken from the Cambridge Young
Learner English Language Movers speaking test (Cambridge ESOL, 2012) was used to
prompt students’ writing since it could control the propositional content (see Appendix B).
Note-taking was employed as a means of measuring student noticing for the following
reasons. First, the present study aimed to create natural classroom settings as much as
possible, considering that contextual factors may impact the outcome of learning (Bruton,
2009). Thus, all participants worked on their writing tasks together in the same classroom
as they would in their regular classes. Since the think-aloud protocol was not optimal for a
noisy classroom setting, note-taking was used. Along with this practical constraint, it was
believed that written data could provide “a good indication of the main focus of the
children’s attention and highlight differences between proficiency levels with regard to the
quantity and quality of the notes produced” (Coyle & Roca de Larios, 2014, p. 10).
The data collection procedures consisted of three stages (see Figure 1). At Stage 1
(Composition Stage), the students, following the teachers’ guidelines, engaged in the joint
story-writing task. In this stage, a simple four-frame picture story and two paper sheets
were given to the students. On the sheet entitled “Your Story,” they were asked to compose
a story based on the picture, and on the sheet with the title “Problems” they were asked to
note down any linguistic difficulties they experienced while doing so. In each pair, one
student was in charge of writing a story and the other was asked to note down the problems
they encountered while writing a story. The children voluntarily chose whether they would
10 Ji Hyun Kim
work as a writer or a note-taker. This idea was borrowed from the concept of writers and
helpers in Yarrow and Topping (2001). The hope was that this could reduce cognitive
demands caused by simultaneous note-taking and writing when students performed
individually (Hanaoka, 2007b).
FIGURE 1
Data Collection Procedures
Three days after Stage 1 (Composition Stage), the students received their composition.
Eight pairs in the control group had their stories returned to them without any feedback,
while 16 pairs in the experimental groups received their stories with either explicit
corrections (3 high, 3 medium, and 3 low-proficiency pairs) or the model text of the story
(3 high, 3 medium, and 3 low-proficiency pairs) (Stage 2: Comparison Stage). At Stage 2,
the students in the model group were asked to compare their versions of the story and the
model text (see Appendix A), and the students in the explicit correction group were also
asked to check the teacher’s corrections. Both groups were encouraged to use the
“Problems” sheets as a checklist to help pay attention to the differences between their
original texts and the model text or the teacher’s corrections. During this process, the
students were asked to note down anything they noticed on the sheet entitled “What I
found.” In contrast, the students in the control group were asked to read their compositions
along with the “Problems” sheet and to write any problems they recognized which were
not stated at Stage 1 (Composition Stage). After the learners finished the comparison stage,
the “Problems” and the “What I found” sheets were collected. Then, the students were
given the same picture story they received at Stage 1 (Composition Stage) and were asked
to rewrite the story on the sheet entitled “Revised Story” (Stage 3).
3.3. Data Analysis
The data consisted of 26 original stories composed by the students, 26 sets of notes
made during the writing process, the notes made during the comparison stage, and 26
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 11
revised stories. In order to investigate what features of languages the child learners noticed
during the composition and comparison stages, language-related episodes (LREs) were
identified in their notes (“Problems” and “What I found”). Following Hanaoka (2007b),
LREs were defined as any aspects of the language the learners wrote about in their notes.
Following Coyle and Roca de Larios (2014), the LREs were classified into lexis,
grammar, spelling, or ideas and they were coded as either (a) problematic features noticed
(PFNs) by the learners at Stage 1 or (b) features noticed (FNs) by the learners at Stage 2. In
the present study, noticing was operationalized as a nontechnical term which referred to
“the children’s perception of surface features in their output or in the feedback” (Coyle &
Roca de Larios, 2014, p. 11). The detailed descriptions and examples of the four categories
of the LRE are presented in Table 1.
TABLE 1
Coding Scheme for LREs as Problematic Features Noticed and Features Noticed
LRE Descriptions Examples
Lexis Learners refer to their incapability to express a lexical item at Stage 1 or recognize lexical items at Stage 2.
I don’t know how to say gapjagi in English. I found suddenly in the teacher’s writing.
Grammar Learners have problems with verb tense/formation, word order, etc. at Stage 1 or identify their writing and the feedback at Stage 2.
I don’t know whether I need to use present tense or past tense. The teacher put -ing, -crying.
Spelling Learners cast doubts regarding accurate word spelling at Stage 1 or notice how a word is spelled at Stage 2.
I don’t know how to spell appear. I need two m’s, swimming.
Ideas Learners express their doubts concerning how to describe the given picture at Stage 1, or get ideas of how to express the picture at Stage 2.
I don’t know what children said to their parents. Oh, now, I see. That’s a toy.
Note. LRE = Language-Related Episodes
To answer the third research question, the relationship between the learners’ initial
noticing and their noticing in response to feedback (or during the self-reflection for the
control group) was analyzed. The total number of PFNs and FNs were counted for each of
the four linguistic features, and the features noticed at both stages were also recorded.
Kruskal-Wallis tests (a nonparametric equivalent of one-way ANOVAs) were conducted
in order to examine group differences in terms of the amount of noticing and textual
revisions among the treatment groups across proficiency levels. In addition, Mann-
Whitney tests (an alternative to the parametric t-test) were employed as a means of post-
hoc analyses (p < .05).
12 Ji Hyun Kim
4. RESULTS
To investigate the features of language child EFL learners noticed while they composed
a text (research question 1), the data from Stage 1 were analyzed. Table 2 shows the
amount of noticing across language features and proficiency levels in three groups.
As Table 2 shows, all of the learners, regardless of proficiency levels and treatment
conditions, reported lexical problems most frequently (65% for the EC group, 68% for the
model group, and 59% for the control group). Difficulties in grammar were reported 21%
of the time (22% for the EC group, 18% for the model group, and 23% for the control
group). When three groups were compared in the total amount of noticing, a group
difference did not exist (χ2 = .056, df = 2, p = .972). This indicates the different learner
pairs assigned to either of EC group, model group, or control group were similar insofar as
the focus of their attention.
TABLE 2
Problematic Features Noticed Across Proficiency Levels and Treatments
Treatment Level N Lexis Grammar Spelling Ideas
All pairs Low 36 25 33 34 92
(69%) (70%) (56%)(64%)
2 7
21 30
(6%) (15%) (34%) (21%)
5 43
12
(14%)(9%) (5%) (8%)
433
10
(11%) (6%) (5%) (7%)
Med 47
High 61
Total 144
Error correction (EC) (9 groups)
Low 13 8 13 12 33
(62%) (71%) (60%) (65%)
137
11
(8%) (17%) (35%) (22%)
2103
(15%) (6%) (6%)
2114
(15%) (6%) (5%) (7%)
Med 18
High 20
Total 51
Mean 3 .67 1.22 0.33 0.44
Models (9 groups)
Low 12 9 12 12 33
(75%)(70%)(60%)(68%)
1269
(8%) (12%) (30%) (18%)
2 2 1 5
(17%) (12%) (5%) (10%)
0112
(6%) (5%) (4%)
Med 17
High 20
Total 49
Mean 3 .67 1 0.55 0.22
Control (8 groups)
Low 11 8 8
10 26
(73%)(67%)(48%)(59%)
02 8
10
(17%) (37%) (23%)
1 12 4
(9%)(8%) (10%) (9%)
2114
(18%) (8%) (5%) (9%)
Med 12
High 21
Total 44
Mean 3 .25 1.25 0.5 0.5
As regards the learners’ noticing across their proficiency levels, there were no
differences in difficulties reported for lexis, spelling, and ideas across proficiency levels in
all three groups, as shown in Table 3. However, the results of the Kruskal-Wallis tests
proved that the learners of high proficiency levels reported the difficulties in grammar
more often than low- and medium-proficiency learners. This result indicated that the
learners of high proficiency levels paid more attention to accuracy than the other two
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 13
groups.
TABLE 3
Differences in Problematic Features Across Proficiency Groups
Treatment Difference Lexis Grammar Spelling Ideas
Error correction
χ2
4.829 6.889 2.667 .800
df 2 2 2 2
Asymp. Sig. .089 .032 * .264 .670
Model χ2
2.451 6.222 .444 1 .143
df 2 2 2 2
Asymp. Sig. .294 .045 * .801 .565
Control χ2
4.533 6.731 .583 .583
df 2 2 2 2
Asymp. Sig. .104 .035 * .747 .747
*p < .05
In order to answer the second research question, which addressed what child EFL
learners noticed in the process of written feedback (EC and a model text), the data
collected at Stage 2 were analyzed. The frequencies and percentages of the noticed features
at Stage 2 are presented in Table 4. The total number of noticed features at Stage 2 (n =
218) was higher than at Stage 1 (n = 144), and similar to Stage 1, the learners’ attention
was drawn to lexical items most frequently in all three groups (55% for the EC group, 66%
for the model group, and 56% for the control group). In addition, to investigate the
differences among three groups, Kruskal-Wallis tests were performed. The results proved
that group differences were reported in all language areas as Table 5 shows.
Mann-Whitney tests employed as post-hoc analyses revealed the following outcomes.
First, when the EC and the model groups were compared, there was no difference in
noticing of lexical items between the groups (z = -.331, p = .741). However, the differences
in noticing of grammatical features (31% for the EC group and 17% for the model group)
was statistically significant (z = -2.866, p = .004): the EC group reported more grammatical
features (M = 3.11) than the model group (M = 1.44). The same result was found for
spelling (z = -2.766, p = .006).
In contrast, the model text was found to bring about noticing about ideas more often than
EC (M = 0.66 for EC and M = 1.33 for the model text), and this difference was significant
(z = -2.062, p = .039). In sum, while EC was more effective in drawing learner attention to
grammar and spelling than the model text, the model text was more useful for triggering
the noticing of gaps in ideas. The type of feedback did not make a significant difference in
drawing learner attention to lexical items. When the EC and the control groups were
compared, the learners in the EC group reported more noticing than the learners in the
control group in all areas (z = -3.254, p = .001 for total number of noticing).
14 Ji Hyun Kim
TABLE 4
Language Features Noticed Across Proficiency Levels and Treatments
Treatment Level N Lexis Grammar Spelling Ideas
All pairs Low 62 38 45 46
129
(61%)(62%)(55%)(59%)
11 16 27 54
(18%)(22%)(33%)(25%)
5 5 3
13
(8%)(7%) (4%) (6%)
8 7 7
22
(13%) (9%) (8%) (10%)
Med 73
High 83
Total 218
Error correction Low 26 15 18 17 50
(58%)(56%)(52%)(55%)
6 10 12 28
(23%)(32%)(36%)(31%)
3 3 1 7
(12%)(9%) (3%) (8%)
2 1 3 6
(7%) (3%) (9%) (6%)
Med 32
High 33
Total 91
Mean 5.55 3.11 0.77 0.66
Models Low 22 15 18 18 51
(68%)(67%)(64%)(66%)
2 4 7
13
(9%)(15%)(25%)(17%)
1 001
(5%) (1%)
4 5 3
12
(18%) (18%) (11%) (16%)
Med 27
High 28
Total 77
Mean 5.66 1.44 0.11 1.33
Control Low 14 89
11 28
(58%)(65%)(50%)(56%)
3 2 8
13
(21%)(14%)(36%)(26%)
1 2 2 5
(7%)(14%)(9%) (10%)
2 1 1 4
(4%) (7%) (5%) (8%)
Med 14
High 22
Total 50
Mean 3.50 1.63 0.63 0.50
As a further analysis, an additional Mann-Whitney test was conducted to examine the
difference between the model group and the control group. The test revealed that the
learners who received the model text attended to lexical items more often than the learners
in the control group (M = 5.66 for the model group and M = 3.50 for the control group; z =
-2.909, p = .004).
TABLE 5
Differences in Noticed Features
Lexis Grammar Spelling Ideas Total
Chi-square 12 .238 12.661 8.387 9.750 14.413
df 2 2 2 2 2
Asymp. Sig. .002 * .002* .015* .008* .001 *
*p < .05
The same result was found in the area of ideas (M = 1.33 for the model group and M =
0.50 for the control group; z = -2.776, p = .006). In contrast, the learners paid more
attention to spelling when they had the self-reflection time (the control group) than when
they received the model text (M = 0.11 for the model group and M = 0.63 for the control
group; z = -2.147, p = .032). With regard to grammatical features, there was no group
difference (z = -.160, p = .873). The results showed that while the model text afforded the
learners more opportunities to notice errors in the areas of lexis and ideas, for spelling
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 15
problems, the self-reflection time given to the control group generated more noticing
compared to the model text. The learners in both groups showed no difference in their
noticing of grammatical features.
As a subsequent step, the relationship between learners’ noticing and their proficiency
levels in each group was examined. In the EC group, the learners’ noticing was not related
to the areas of lexis, spelling, and ideas across proficiency levels. However, a difference
was observed for grammatical features (χ2 = 6.222, p = .045), and this difference originated
from the difference between the medium-proficiency pairs and the low-proficiency pairs (z
= -2.121, p = .034). No difference was found between the high-proficiency pairs and the
medium-proficiency pairs (z = -.943, p = .346). In the model group, no difference was
found in the number of learners’ noticing across their proficiency levels in the four
language features. For the control group, the high-proficiency pairs reported more
grammatical features compared to the medium-proficiency (z = -2.000, p = .046) and low-
proficiency pairs (z = -2.236, p = .025).
The third research question asked how the learners’ noticing was related to the
corrections they made in the revised texts. In order to answer this question, the study
classified learner changes to their revised texts as (a) those reported at Stage 1 only, (b)
those noticed at Stage 2 only, (c) those noticed at both Stage 1 and Stage 2, and (d) those
unreported at the previous stages but drawn from the feedback. Table 6 below presents the
frequencies and proportions of features incorporated into the revised texts across stages
and feedback types.
In general, more than half of the changes made by all pairs were lexical (62%). In the
EC group, 58% of incorporated features were lexis, 71% for the model group, and 52% for
the control group. Almost half of lexical items incorporated in the revision texts had been
reported at both stages (49%). For the EC group, among 55 lexical changes, 53% were
noticed at both stages, 31% were at Stage 2, 9% were not reported, and 7% were reported
at Stage 1. In the model group, out of 54 lexical changes 43% were noticed at both stages,
28% were not reported, 22% were at Stage 2, and 7% were at Stage 1. In the control group,
of 16 changes, 56% were reported at both stages, 31% were at Stage 2, and 13% were at
Stage 1. When the group differences were examined through Kruskal-Wallis tests, no
difference was found at Stage 1, but differences existed among the other three stages. The
differences mostly originated between the feedback groups and the control group. The
feedback groups made more changes originating from Stage 2 and Stages 1+2 compared to
the control group. As regards the differences between the feedback groups, the post-hoc
analyses revealed that the EC group and the model group showed no difference in revisions
reported at Stage 2 and Stags 1+2. However, the model group (M = 1.66) incorporated
more lexical items which were previously unreported compared to the EC group (M =
0.55; z = -2.000, p = .045).
16 Ji Hyun Kim
TABLE 6
Frequency of Features Incorporated into the Revisions Across Stages and Feedback Types
Treatment Features Stage 1 Stage 2 Stage 1+2 Not rep. Total
All pairs Lexis 10(8%) 34 (27%) 61 (49%) 20 (16%) 125 (62%)
Grammar 4(8%) 25 (50%) 19 (38%) 2 (4%) 50 (25%)
Spelling 3(27%) 2(18%) 5 (45%) 1 (9%) 11(5%)
Ideas 3(19%) 0 4 (25%) 3 (19%) 16 (8%)
Total 20 61 89 26 202
Error correction
Lexis 4(7%) 17 (31%) 29 (53%) 5 (9%) 55 (58%)
Grammar 2(7%) 16 (53%) 10 (33%) 2 (7%) 30 (32%)
Spelling 0 2 (33%) 3 (50%) 1 (17%) 6 (6%)
Ideas 0 1 (25%) 3 (75%) 0 4 (4%)
Total 6 36 45 8 95
Models Lexis 4(7%) 12 (22%) 23 (43%) 15 (28%) 54 (71%)
Grammar 0 5 (50%) 5 (50%) 0 10 (13%)
Spelling 2(100%) 0 0 0 2 (3%)
Ideas 1(10%) 5 (50%) 1 (10%) 3 (30%) 10 (13%)
Total 7 22 29 18 76
Control group
Lexis 2(13%) 5 (31%) 9 (56%) 0 16 (52%)
Grammar 2(20%) 4 (40%) 4 (40%) 0 10 (32%)
Spelling 1(33%) 0 2 (67%) 0 3 (10%)
Ideas 2(100%) 0 0 0 2 (6%)
Total 7 9 15 0 31
For grammar revisions, overall 25% of changes were made. In the EC group, out of 30
changes, 53% were reported at Stage 2, 33% were at Stages 1+2, 7% were at Stage 1, and
another 7% were not reported. In the model group, among 10 changes, half originated from
Stages 1+2, and the other half were from Stage 2. In the control group, 40% of corrections
made to the revised texts were those reported at each Stage 2 and Stages 1+2. The
difference among the groups was found in the incorporation made from Stage 2 (χ2 =
11.792, p = .003). Post-hoc analyses revealed that the difference appeared between the EC
group and the model group (z = -2.903, p = .004) and the control group (z = -2.874, p
= .004). There was no difference between the model group and the control group (z = -.222,
p = .924)
For changes made to spelling and ideas, no group differences were found in spelling,
and as Table 7 shows the incidents of changes were small in all three groups. The number
of incorporated features related to ideas was also small, but the difference among the
groups was found in the changes originating from Stage 2 (χ2 = 8.148, p = .017) and in
those which were previously unreported (χ2 = 6.159, p = .046). Post-hoc analyses showed
that the model group more often incorporated the features reported at Stage 2 and those
previously unreported into the revision compared to the EC group and the control group.
The number of incorporated language features in each stage across proficiency levels
was examined in each group. In the EC group, the high-proficiency learners incorporated
noticed lexical features reported at Stages 1+2 into the revised texts more often than the
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 17
low-proficiency learners (z = -2.121, p = .034). No differences were found between the
high- and medium-levels or between the medium- and low-levels. For changes made to
grammatical features, a difference was found in Stages 1+2 (χ2 = 6.054, p = .048), and this
was from the difference between the high-level group and the other two groups (medium-
level: z = -2.023, p = .043/ low-level: z = -2.023, p = .043); there was no group difference
between medium-level and low-level learners (z = -1.650, p = .099). For ideas and spelling,
no group differences were found. In the model group, no difference across proficiency
levels was found in the areas of grammar, ideas, and spelling. But it was found that high-
proficiency learners made more lexical changes of the features they did not report
compared to medium-proficiency learners (z = -2.121, p = .034) and the low-proficiency
learners (z = -2.121, p = .034). In the control group, no differences were found across
proficiency levels in any language areas.
The major findings can be summarized as follows: (a) the learners’ attention was drawn
to language features while they produced output, and this self-initiated noticing was mainly
for lexical problems; (b) the high-proficiency learners paid attention to grammatical
features more frequently than the medium/low proficiency learners while they composed a
text; (c) the learners who received WCF, especially EC, paid more attention to language
features compared to those who did not receive CF; (d) the learners’ attention to language
was mediated differently by different types of WCF: while EC triggered noticing of
grammatical and spelling errors, the model text directed learner attention to ideas; (e) the
learners’ proficiency levels were related to their noticing of grammatical corrections but
were not related to lexis, spelling, and ideas; (f) the learners’ textual revisions were mostly
lexical, and EC triggered more grammatical revisions than the model text.
5. DISCUSSION
5.1. Learner-generated Noticing During the Writing Process
The study found that the child EFL learners did notice gaps in their IL knowledge while
they composed a text, and a majority of this self-generated noticing was for lexical
problems. The learners very often reported their lack of ability to find appropriate words to
express their ideas. This result is consistent with the findings some previous studies
reported (Hanaoka & Izumi, 2012; Kim, 2013; Qi & Lapkin, 2001). For instance, in Swain
and Lapkin (1995), the analyses of think-aloud data from 18 students revealed that half of
all language-related episodes involve lexical problems. In addition, Williams (2001) found
that learners were able to initiate attention to form, and an overwhelming 80% of all
language-related episodes were lexical. This is also in accordance with learners’ general
18 Ji Hyun Kim
tendency to focus on lexis (Williams, 2012).
As for learner proficiency levels, differences were not observed in the areas of lexis,
ideas, and spelling. However, the high-proficiency pairs more frequently reported problems
related to grammatical features compared to the medium- and low-proficiency pairs,
suggesting that learners’ proficiency may be related to their noticing of holes in grammar
while producing output. This outcome corroborates Qi and Lapkin’s (2001) claim that
high-proficiency learners may be able to generate more attention to language forms than
low-proficiency learners. However, Coyle and Roca de Larios (2014) and Hanaoka
(2007a) did not find any differences in learner-initiated noticing across proficiency levels.
In light of the lack of research into learner-generated noticing across proficiency levels and
language areas, it seems too early to reach any conclusion based on these inconsistent
results. Colye and Roca de Larios attributed the disparity to the different methodological
measures: Qi and Lapkin (2001) employed verbal reports while Coyle and Roca de Larios
(2014) and Hanaoka (2007a) relied on note-taking. Colye and Roca de Larios explained
that note-taking might be too demanding for the child learners. This may be valid, as it is
possible that note-taking may be an incomplete tool for measuring learner noticing data,
especially, for child learners. However, considering the fact that Colye and Roca de Larios
and the present study both employed note-taking as a means of collecting learner noticing
data, and verbal reports also present many limitations in examining learner noticing
(Mackey & Gass, 2005), the different findings may not be solely due to measurement.
There is no single explanation for the noticeable differences in the findings of previous
studies, and this one. One possible explanation, among many, may be learner orientation.
Since proficiency levels can be interpreted in many different ways, learners with a high-
proficiency level may be understood in a different way and may have different abilities
across different instructional settings. In the present study, although classes were mainly
meaning-based, the instruction might be more form-oriented compared to those in Colye
and Roca de Larios (2014). In such instructional settings, learners who are concerned about
language forms may be high-proficiency level learners and, indeed, this type of learners
may develop L2 more effectively than others who are less concerned with accuracy. Thus,
in the present study, the high-proficiency pairs might be more sensitive to grammatical
problems than the other groups. Needless to say, there may be other possible factors which
mediate learner-generated noticing when producing output, and exploring such factors will
certainly require more research.
5.2. Learner Noticing Caused by WCF
The comparison stage (Stage 2) increased the number of features noticed in both EC and
model groups. This confirms the noticing function of output (e.g., Swain, 1995). At Stage 2,
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 19
the learners noticed some new problems with their initial writing while studying the
teachers’ EC or model text. Thus, it may be claimed that output not only brings about the
immediate recognition of linguistic problems (noticing the holes), but also results in further
noticing of problems (noticing the gaps) in subsequent processing of target language input
(Izumi, 2002, 2003; Swain, 1995).
Similar to the self-initiated noticing, the majority of instances of noticing involved
lexical features, and no differences were found between the groups. This finding was
somewhat expected. At the comparison stage, the learners were asked to compare their
initial versions with the model text or the versions with EC. In this process, it was natural
that the learners’ attention went to the problems they had noticed in the writing process.
This outcome also corroborates the learners’ lexical-oriented tendency from Hanaoka
(2007b) and Coyle and Roca de Larios (2014), and suggests that both EC and model texts
are useful in promoting child learners’ noticing of lexical features.
For grammatical features, in the EC group, the learners attended to grammatical features
twice as often as the writing stage, while the learners in the model group and the control
group did not show much increase between stages. In other words, EC was effective in
promoting learner noticing of grammatical problems in their writing. This result can be
explained by the nature of EC. Since the teacher’s EC directly targets linguistic problems
committed by learners, they do not need to search for differences (Sachs & Polio, 2007).
Thus, this result came as no surprise. However, it was somewhat unexpected that no
difference was found between the model group and the control group, and no difference in
the learners’ noticing in the model group between the stages. Indeed, since most of the
previous research into learner noticing caused by WCF did not include control groups, it
was difficult to judge if the small amount of increased noticing was valuable and/or if it
was actually caused by the model texts (Hanaoka, 2007b). In this regard, the results of the
present study may suggest that models are not be capable of facilitating noticing gaps in
grammar, at least for child learners (see Kim, 2013 for adult learners). For them, it may be
too cognitively demanding to look for grammatical problems by comparing their original
writing with a model text.
While the model text was found ineffective in promoting noticing of gaps in grammar, it
led the learners to notice the difference in ideas. This result is similar to Martínez Esteban
and Roca de Larios (2010) which reported that learners attended more to the content of the
models than language forms. In addition, in Hanaoka’s (2007b) study, learners reported
their attention to ideas expressed in model texts more frequently than they did for grammar,
although the researcher did not discuss this finding.
When the learners’ proficiency was considered, the amount of learners’ noticing of lexis,
ideas, and spelling did not differ across proficiency levels in any groups. However, learner
noticing in grammar features resulted in complicated findings. The learners’ proficiency
20 Ji Hyun Kim
levels were not related to noticing in the model group. However, in the EC group, the high-
proficiency/medium-proficiency pairs and the low-proficiency showed differences between
the groups, but no difference was obtained between the high- and medium-proficiency
learners. This outcome needs to be interpreted with caution. That no difference was found
in the model group does not indicate that the model text was equally effective. As a matter
of fact, it suggests the opposite (i.e., equally not effective). If the learners reached
intermediate proficiency levels, they seemed to be able to notice gaps in grammar if error
corrections were provided. However, as shown in the result from the control group, only
advanced learners were capable of generating attention to form during self-reflection
process.
5.3. Learner Incorporation of Noticed Features into Their Revised Texts
The learners’ incorporation of noticed features into their revisions provided evidence
supporting the facilitating role of child learners’ processing and retention of information
provided in the form of feedback. More than half of the changes were lexical, and half of
them were the features noticed at both Stage 1 and Stage 2. The learners in both feedback
groups incorporated more noticed features than the control group. This is consistent with
the role of output as a trigger for learner noticing: producing the TL can encourage learners
to recognize what they can and cannot express (Stage 1), and this eventually leads learners
to actively look for relevant information in the input (feedback groups at Stage 2) and use
the information in their subsequent production. Such findings also corroborate the findings
from Qi and Lapkin (2001), Hanaoka (2007b), and Coyle and Roca de Larios (2014).
When the feedback groups were compared, no differences were found in the lexical
revisions originated from the writing stage and the comparison stage. However, it was
revealed that the learners in the model group, especially high-proficiency pairs, made
lexical changes which were not reported at the previous noticing stages more often than the
learners in the EC group. The unreported changes may be considered evidence for the
limitations of collecting learner noticing data (i.e., incomplete reporting of noticing) (Coyle
& Roca de Larios, 2014). However, this may also indicate that the model provided an
opportunity to be exposed to content that they could not draw on to express meanings on
their own.
For changes made in grammar, the EC group, especially the learners of high and
medium-level proficiency, incorporated the noticed features more often than the model and
control group. In addition, the fact that the number of grammatical revisions originating
from Stage 2 was significantly greater than for the other groups may imply that EC helped
the child learners to identify linguistic inadequacies in their writing texts, although they did
not recognize the problems during their composition and eventually led them to use the
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 21
information in the subsequent output.
Thus far, the findings from the research into WCF, including the present study, have
shown that EC is beneficial in triggering learner noticing of gaps and incorporating the
information delivered in the form of CF into the subsequent writing regardless of learner
ages. Yet studies have also shown that model texts are advantageous in that they help
learners to notice something beyond what they could produce on their own. This suggests
that EC and models have distinctive functions and they can play complementary roles in
L2 writing.
6. CONCLUSION
The present study provided an understanding of noticing in the process of written output
of child EFL learners who have generally been overlooked by researchers. The study found
that child EFL learners in instructional settings noticed holes during output production.
They also took advantage of feedback by comparing their original written texts with the
texts with EC and a model text. These findings confirm the facilitative role of output in
generating learner-initiated attention to language and the effects of WCF as a means of
learner noticing of gaps.
The role of noticing in child L2 development has rarely been dealt with under the
assumption that children may benefit more from implicit instruction rather than explicit
instruction. To some extent, this may be true. However, this does not mean that there is no
room for explicit instruction for child L2 learners. Self-stated errors reported by the control
group actually showed that child learners actually used explicit knowledge even in the
situations where no feedback was offered. Exploring the conditions where explicit
instruction works better for learners is one area that requires further investigation. In this
regard, the findings of the present study allow for the following pedagogical suggestions.
Both EC and model texts can facilitate child learners’ noticing of lexical features regardless
of proficiency levels. However, they also have distinctive roles. While error correction may
be more useful for directing learners’ attention to grammatical features if learners have
certain levels of proficiency, models do not seem effective even for learners of high-level
proficiency. In contrast, models may more effectively provide lexical and content
information regardless of learners’ current repertoires, which EC may not be able to offer.
Based on these findings, it can be suggested that both EC and model texts could be used
together with different purposes in L2 writing classes.
The present study has many limitations. Specifically, four major problems must be taken
into account for future research. First, a larger sample is needed to provide more
representative results. In the present study, only twenty-six pairs participated, and this
22 Ji Hyun Kim
number may be too small to bring about reliable findings. Secondly, the collaborative
writing was employed in the process of collecting data, and this might affect the outcomes
of the study. In addition, the present study only examined the learners’ immediate use of CF.
In order to claim the role of noticing and CF in L2 development, delayed incorporation of
feedback needs to be explored in future studies. Lastly, the study employed note-taking as
a means of measuring learner noticing. This certainly captures many of the problems
learners have experienced. Nevertheless, it is not possible to assert that it measures a
complete picture of learners’ noticing process. Since no single measure can completely
capture learners’ cognitive process, future studies may need to employ multiple tools. One
needs to keep the well-known research maxim that an absence of reports of noticing cannot
be equated with the evidence of an absence. Triangulating the findings of one measure with
other measures may provide a more reliable and valid outcome.
REFERENCES
Adams, R. (2003). L2 output, reformulation and noticing: Implications for IL development.
Language Teaching Research, 7, 347-376.
Bitchener, J. (2012). A reflection on “the language-learning-potential” of written CF.
Journal of Second Language Writing, 21, 348-363.
Bruton, A. (2009). Designing research into the effects of grammar correction in L2 writing:
Not so straightforward. Journal of Second Language Writing, 18, 136-140.
Cambridge ESOL. (2012). Cambridge young learners English tests 7: Movers. Cambridge:
Cambridge University Press.
Cohen, A. D. (1982). Writing like a native: The process of reformulation. Paper presented
at the annual convention of Teachers of English to Speakers of Other Language,
Honolulu, HI. (ERIC Document Reproduction Service No. ED 182 465)
Coyle, Y., & Roca de Larios, J. (2014). Exploring the role played by error correction and
models on children’s reported noticing and output production in a L2 writing task.
Studies in Second Language Acquisition, 36, 1-35.
Egi, T. (2007). Recasts, interpretations, and L2 development. In A. Mackey (Ed.),
Interaction in second language acquisition: A series of empirical studies (pp. 249-
267). Oxford: Oxford University Press.
Ferris, D. (2010). A framework for investigating oral and written corrective feedback.
Studies in Second Language Acquisition, 32, 335-349.
Han, Z-H. (2001). Fine-tuning corrective feedback. Foreign Language Annals, 34, 582-599.
Hanaoka, O. (2007a). Noticing from models and reformulations: A case study of two
Japanese EFL learners. Sophia Linguistica, 54, 167-192.
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 23
Hanaoka, O. (2007b). Output, noticing, and learning: An investigation into the role of
spontaneous attention to form in a four-stage writing task. Language Teaching
Research, 11, 459-479.
Hanaoka, O., & Izumi, S. (2012). Noticing and uptake: Addressing pre-articulated covert
problems in L2 writing. Journal of Second Language Writing, 21, 332-347.
Izumi, S. (2002). Output, input enhancement, and the noticing hypothesis: An experimental
study of ESL relativization. Studies in Second Language Acquisition, 24, 541-577.
Izumi, S. (2003). Comprehension and production processes in second language learning: In
search of the psycholinguistic rationale of the output hypothesis. Applied
Linguistics, 24, 168-196.
Jang, S-S. (2012). Types of information in written corrective feedback and its efficacy on
L2 acquisition. English Teaching, 67(3), 3-25.
Kim, J-H. (2011). Enhancing learner recognition of recasts in EFL classrooms. Foreign
Languages Education, 18(4), 1-21.
Kim, J-H. (2013). Learner understanding of written corrective feedback and its relationship
with immediate uptake and retention in EFL classrooms. English Teaching, 68(3),
109-130.
Kim, J-H., & Han, Z-H. (2007). Recasts in communicative EFL classes: Do teacher intent
and learner interpretation overlap? In A. Mackey (Ed.), Conversational interaction
in second language acquisition: A series of empirical studies (pp. 269-297).
Oxford: Oxford University Press.
Long, M. H. (1996). The role of linguistic environment in second language acquisition. In
W. Ritchie & T. Bhatia (Eds.), Handbook of second language acquisition (pp. 413-
468). San Diego, CA: Academic Press.
Mackey, A., & Gass, S. (2005). Second language research: Methodology and design.
Mahwah, NJ: Lawrence Erlbaum.
Mackey, A., Gass, S., & McDonough, K. (2000). How do learners perceive interactional
feedback? Studies in Second Language Acquisition, 22, 471-497.
Manchón, R. M. (2009). Broadening the perspective of L2 writing scholarship: The
contribution of research on foreign language writing. In R. M. Manchón (Ed.),
Writing in foreign language contexts: Learning, teaching, and research (pp. 1-19).
Clevedon, UK: Multilingual Matters.
Martínez Esteban, F., & Roca de Larios, J. (2010). The use of models as a form of written
feedback to secondary school pupils of English. International Journal of English
Studies, 2, 143-170.
Qi, D. S., & Lapkin, S. (2001). Exploring the role of noticing in a three-stage second
language writing task. Journal of Second Language Writing, 10, 277-303.
Robinson, P. (1995). Attention, memory, and the “noticing” hypothesis. Language
24 Ji Hyun Kim
Learning, 45, 283-331.
Sachs, R., & Polio, C. (2007). Learners’ use of two types of written feedback on a L2
writing revision task. Studies in Second Language Acquisition, 29, 67-100.
Schmidt, R. (1990). The role of consciousness in second language learning. Applied
Linguistics, 11, 129-158.
Schmidt, R. (2001). Attention. In P. Robinson (Ed.), Cognition and second language
instruction (pp. 3-32). Cambridge: Cambridge University Press.
Storch, N. (2007). Investigating the merits of pair work on a text editing task in ESL
classes. Language Teaching Research, 11, 143-159.
Storch, N., & Wigglesworth, G. (2007). Writing tasks: Comparing individual and
collaborative writing. In M. P. Garcia Mayo (Ed.), Investigating tasks in formal
language learning (pp. 157-177). London: Multilingual Matters.
Storch, N., & Wigglesworth, G. (2010). Learners’ processing, uptake, and retention of
corrective feedback on writing: Case studies. Studies in Second Language
Acquisition, 32, 303-334.
Swain, M. (1985). Communicative competence: Some roles of comprehensible input and
comprehensible output in its development. In S. M. Gass & C. G. Madden (Eds.),
Input in second language acquisition (pp. 235-253). Rowley, MA: Newbury House.
Swain, M. (1995). Three functions of output in second language learning. In G. Cook & B.
Seidlhofer (Eds.), Principles and practice in applied linguistics: Studies in honor of
H. G. Widdowson (pp. 125-144). Oxford: Oxford University Press.
Swain, R., & Lapkin, S. (1995). Problems in output and the cognitive processes they
generate: A step towards second language learning. Applied Linguistics, 16, 371-
391.
Wigglesworth, G., & Storch, N. (2012). What role for collaboration in writing and writing
feedback. Journal of Second Language Writing, 21, 364-374.
Williams, J. (2001). Learner-generated attention to form. In R. Ellis (Ed.), Form-focused
instruction and second language learning (pp. 303-346). Malden, MA: Blackwell.
Williams, J. (2012). The potential role(s) of writing in second language development.
Journal of Second Language Writing, 21, 321-331.
Yarrow, F., & Topping, K. J. (2001). Collaborative writing: The effects of metacognitive
prompting and structured peer interaction. British Journal of Educational
Psychology, 71, 261-282.
The Role of Models and Error Correction in L2 Young Learners’ Noticing and Output 25
APPENDIX A
Model Text
One day, a family went to the beach. Sally ran straight for the water. Jim said, “Hold on,
Sally, let’s bring our new toy to the water.” They wave goodbye to their parents and run for
the ocean. Jim lets go of the toy dolphin and it starts swimming all around. Suddenly, a real
dolphin appears and starts playing with the toy dolphin! Sally says, “Look! That dolphin
has a new friend.” Jim laughs as the dolphin balances the toy on his noise.
APPENDIX B
Picture Story
Applicable Levels: Elementary
Ji Hyun Kim
Department of English Education, College of Education
Keimyung University
2800, Dalgubeoldaero, Dalseo-Gu
Daegu 704-701, Korea
Phone: 053-580-5136
26 Ji Hyun Kim
Fax: 053-580-5315
E-mail: [email protected]
Received in March 1, 2015
Reviewed in April 15, 2015
Revised version received in May 15, 2015
