ArticlePDF Available

Testing the output hypothesis: Effects of output on noticing and second language acquisition

SSLA, 24, 541–577. Printed in the United States of America.
DOI: 10.1017.S0272263102004023
An Experimental Study
on ESL Relativization
Shinichi Izumi
Sophia University
This study investigates the potentially facilitative effects of internal
and external attention-drawing devices—output and visual input en-
hancement—on the acquisition of English relativization by adult En-
glish as a second language (ESL) learners. Specifically, the study
addresses: (a) whether the act of producing output promotes notic-
ing of formal elements in the target language (TL) input and affects
subsequent learning of the form; and (b) whether such output-
induced noticing and learning, if any, would be the same as that
effected by visual input enhancement designed to draw learners’ at-
tention to problematic form features in the input. These questions
were examined in a controlled experimental study in which the re-
quirements of output and exposure to enhanced input were system-
atically varied. A computer-assisted reconstruction and reading task
was used as the vehicle of presentation of the target input materials.
The major findings are: (a) Those engaged in output-input activities
outperformed those exposed to the same input for the sole purpose
of comprehension in learning gains; (b) those who received visual
An earlier version of this paper was presented at the annual meeting of the American Association for
Applied Linguistics held in St. Louis, MO, in February of 2001. This paper is based on part of the
author’s doctoral dissertation, which was completed at Georgetown University. I am thankful to
Catherine Doughty for her guidance throughout the entire process of my research project. My thanks
also go to Jeff Connor-Linton and Cristina Sanz for their critical comments and encouragement.
Thanks are also due to a number of people who were directly or indirectly involved in this research
project, including the directors, teachers, and students of the participating ESL programs as well as
people who were kindly involved in the creation, execution, and assessment of the tests and tasks
used in the study. Finally, I wish to thank the anonymous SSLA reviewers for their helpful comments
and insights.
Address correspondence to: Shinichi Izumi, Department of English Language and Studies, Sophia
University, 7-1 Kioi-cho, Chiyoda-ku, Tokyo 102-8554, Japan; e-mail:
2002 Cambridge University Press 0272-2631/02 $9.50
542 Shinichi Izumi
input enhancement failed to show measurable gains in learning, de-
spite the documented positive impact of enhancement on the notic-
ing of the target form items in the input; and (c) in view of the above,
no support was found for the hypothesis that the effect of input en-
hancement was comparable to that of output. The subsequent dis-
cussion centers on reexamining the construct of noticing and argues
for the need to consider levels and types of processing in order to
account for how sensory detection can lead to learning.
For many second language (L2) learners and teachers, producing language
(i.e., output) is generally considered to constitute an important part of L2
learning. However, precisely how beneficial it is to produce language is often
not so clear. If the question is considered in terms of the contribution of out-
put to the development of L2 knowledge as opposed to its utility in increasing
fluency, the answer may become even more vague and speculative. Moreover,
if output has any positive effect on learning, we may wish to ask whether it is
unique to output or if essentially the same effect can be obtained by some
external manipulation of input. These are the issues that are addressed in the
study reported on in this article.
The global consensus that has emerged from decades of research in SLA is
that input plays a crucial role in driving learners’ acquisition of an L2. Current
SLA research, however, goes beyond general interest in the need for compre-
hensible input (Krashen, 1985), which is considered necessary but insufficient
(see Ellis, 1994; Larsen-Freeman & Long, 1991; and Long, 1996, for reviews).
Instead, it seeks to obtain a more precise understanding of how learners pro-
cess, or interact with, input to develop their interlanguage (IL) competence.
Given that not all of the input that learners are exposed to is utilized as intake
for learning, recent research in cognitive psychology and SLA has examined
the role of attention in mediating input and learning. A general finding of such
research indicates that attention is necessary for learning to take place (see
Robinson, 1995; Schmidt, 1990, 1995, 2001; and Tomlin & Villa, 1994, for re-
views; see also Schachter, 1998; and Truscott, 1998, for skeptical views). Sim-
ply stated, “people learn about the things that they attend to and do not learn
much about the things they do not attend to” (Schmidt, 2001, p. 30).
This general agreement on the importance of attention notwithstanding,
disagreement exists as to the amount and type of attention needed for learn-
ing. Three positions have been put forth in the SLA literature. First, the Notic-
ing Hypothesis (Schmidt, 1990, 1995, 2001; Schmidt & Frota, 1986) claims that
“intake is that part of the input that the learner notices” (1990, p. 139). Notic-
ing, Schmidt argued, crucially requires focal attention and awareness on the
Effects of Output and Input Enhancement 543
part of the learner. A second position, proposed by Tomlin and Villa (1994),
claims that, of three interrelated processes of attention—alertness, orienta-
tion, and detection—only detection, which does not require conscious aware-
ness, is crucial for learning; the other two processes may help to increase the
chance of detection and, thus, learning. Third, Robinson (1995) defined
Schmidt’s noticing as what is both detected and then further activated as a
result of the allocation of attentional resources from a central executive. He
argued that different task demands stimulate different types of further cogni-
tive processing.
Pedagogical Attempts to Promote Learners’ Noticing of Form
Taking the central role of attention in learning as a starting point of investiga-
tion, recent SLA research has begun to explore whether and how the learners’
attentional processes may be influenced for the sake of their greater IL devel-
opment. Such consideration is indeed at the core of influential pedagogic pro-
posals known as consciousness-raising or input enhancement (Rutherford &
Sharwood Smith, 1985; Sharwood Smith, 1993) and focus on form (Doughty,
2001; Doughty & Williams, 1998; Long, 1991; Long & Robinson, 1998).
Two specific pedagogical approaches to draw the learner’s attention to
form have received considerable attention in recent SLA research. One is
known as visual (textual or typographical) input enhancement, and the other
is learners’ output. These approaches share a basic characteristic—namely,
an attempt to direct the learner’s otherwise elusive attention to problematic
aspects in the input to promote their acquisition. They differ, however, in how
this is achieved. Whereas attention in the case of visual input enhancement is
induced by external means (i.e., by highlighting selected input forms), atten-
tion in output arises internally through production processes, in that learners
themselves decide what they find problematic in their production and what
they pay attention to in the input (although external manipulation of task vari-
ables may intervene in this process). In other words, it may be argued that
visual input enhancement is an external attention-drawing technique, whereas
output is an internal attention-drawing device. Previous studies on these two
approaches will be examined in turn.
Visual Input Enhancement: Previous Studies and Remaining Issues
Visual input enhancement is an implicit and unobtrusive means to draw the
learners’ attention to form contained in the written input (Doughty & Williams,
1998). The basic method of the enhancement is simply increasing the percep-
tual salience of the target form via combinations of various formatting tech-
niques (e.g., bolding, capitalizing, or underlining), which may sometimes be
accompanied by an explicit mention to the learners to attend to the high-
lighted form. With a particular form chosen for the target, the enhancement
544 Shinichi Izumi
embedded in the overall reading lesson aims to achieve the integration of at-
tention to form and attention to meaning.
Previous studies on the effects of visual input enhancement—both those
that used short-term treatment with rather limited exposure to the input (Ala-
nen, 1995; Jourdenais, Ota, Stauffer, Boyson, & Doughty, 1995; Leow, 1997;
Robinson, 1997; Shook, 1994; Williams, 1999) and those that used longer-term
treatment with a greater amount of input exposure (Doughty, 1988, 1991; Jour-
denais, 1998; White, 1998)—produced quite mixed results. Four of these nine
studies (Doughty; Jourdenais et al.; Shook; and Williams) yielded positive find-
ings for the facilitative effect of the enhancement, whereas three studies (Ala-
nen; Robinson; and White) showed only limited effects and two (Leow;
Jourdenais) found no significant effect at all. Although various differences in
these studies make direct comparison among them difficult, an examination of
several factors—of which I focus on three—is instrumental in identifying the
directions for future research on the issue (for a detailed review of these stud-
ies, see Izumi, 2000).
First, although clearly a desirable design feature, not all studies incorpo-
rated the noticing assessment component in their research designs. Studies
that did use some kind of noticing measure provided interesting evidence as
to the degree and nature of noticing. For example, Alanen’s (1995) study sug-
gested that noticing seems to be an important factor in accounting for subse-
quent learning, but the cross-comparison of the noticing results and learning
outcomes suggests that noticing seemed to be induced by a variety of factors,
not the least of which was input enhancement. White (1998), on the other
hand, reported that many learners noticed the forms but were not sure of
their relevance or importance, which arguably accounted for the limited im-
provement by the enhancement group in her study. Given the alleged impor-
tance of attention in learning that is presupposed by most studies, and given
the uncertainty regarding the link between input enhancement, noticing, and
learning, the employment of a noticing measure, or better yet, multiple mea-
sures of noticing, seems critical in future studies.
Second, previous research has raised questions about whether input en-
hancement alone can really induce desired learning effects as intended by the
researcher. In this vein, Williams’s (1999) study, which showed the benefits of
visual enhancement, used a form-focused verbatim recall task in conjunction
with visual input enhancement. Such tasks may have served as a focused out-
put task, inducing greater noticing of the form than may have been possible
by visual enhancement alone. Likewise, Doughty (1988, 1991) found positive
effects of the meaning-oriented treatment that involved not only visual en-
hancement but also various forms of comprehension assistance given for each
single sentence. As such, her meaning-oriented treatment may be construed
to be more explicit, elaborated, and focused than a simple enhancement con-
dition. Thus, it seems that further investigation into the effects of a combina-
tion of instructional techniques may be worthwhile.
Finally, for any research on the effects of intervention treatment on L2
Effects of Output and Input Enhancement 545
learning, it is useful to examine a priori the learner’s level of proficiency, or
more specifically the developmental level vis-a
`-vis the target form, because it
is quite possible that the treatment effect is constrained by the learner’s de-
velopmental readiness (Pienemann, 1984, 1998). Among the nine studies re-
viewed here, only Doughty (1988, 1991; and perhaps White, 1998) assessed the
learnability of the target form for the selected group of learners prior to the
start of the treatment. A general lack of learnability considerations in many of
the studies is partly due perhaps to the inherent difficulty involved in deter-
mining the learnability of forms (but see Pienemann’s, 1998, processability
theory) and partly due to the debate surrounding how rigidly the learnability
notion can and should be applied in SLA classroom studies and pedagogical
practices (see, e.g., Spada & Lightbown, 1999). Such debate notwithstanding,
more studies taking the learnability notion into account would be desirable if
we are to be sure that the learners are indeed psycholinguistically capable of
learning the targeted form.
The present study takes these considerations into account and investigates
the effects of visual input enhancement on the noticing and acquisition of a
grammatical form by adult L2 learners. Another independent variable to be
investigated is learner output. The inclusion of these two variables permits
the investigation of how they may interact or contrast in promoting SLA.
The Role of Output in SLA
The currently popular view of output posits that it constitutes not just the
product of acquisition or the means by which to practice one’s language for
greater fluency but also a potentially important causal factor in the acqui-
sition process. In general, the importance of output in learning may be
construed in terms of the learners’ active deployment of their cognitive re-
sources. In other words, it is posited that the output requirement presents
learners with unique opportunities to process language that may not be deci-
sively necessary for simple comprehension. In proposing the Output Hypothe-
sis, Swain (1985) argued that producing the target language (TL) 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). In psycholinguistic terms, it may be assumed that grammatical encoding
and monitoring mechanisms (see Levelt, 1989) play particularly important
roles for learning purposes by functioning as an internal priming device for
grammatical consciousness-raising for the language learner (see Izumi, 2000,
for a detailed discussion of the psycholinguistic rationale of the Output Hy-
Of the four functions of output specified in the current version of the Out-
put Hypothesis (Swain, 1993, 1995, 1998), the present study focuses on its no-
ticing function. The noticing function of output posits that learners may notice
the gap in their IL knowledge in an attempt to produce the TL, which then
prompts them to solve their linguistic deficiency in ways that are appropriate
546 Shinichi Izumi
in a given context. For example, if learners are left on their own to solve the
immediate production difficulties, they may engage in various thought pro-
cesses that can consolidate existing knowledge or possibly generate some
new knowledge on the basis of their current knowledge (Swain & Lapkin,
1995). On the other hand, if relevant input is immediately available, the height-
ened sense of problematicity during production may cause the learners to
process subsequent input with more focused attention; they may try to exam-
ine closely how the TL expresses the intention that they just had difficulty in
expressing on their own. In either case, learning is believed to be enhanced
through the act of producing language, which, by its mechanisms, increases
the likelihood that learners become sensitive to what they can and cannot say
in the TL, which leads to the reappraisal of their IL capabilities.
Previous empirical studies on the noticing function of output have pro-
duced mixed findings. In a series of studies (Izumi & Bigelow, 2000; Izumi, Bi-
gelow, Fujiwara, & Fearnow, 1999), my colleagues and I investigated whether
output would indeed alter the learners’ subsequent input processing and pro-
mote their IL development. Focusing on the English past hypothetical condi-
tional, these studies compared a group that was given output opportunities
and subsequent exposure to relevant input and a group that received the
same input for the sole purpose of comprehension. This basic format of the
treatment was instantiated in two types of tasks—a text-reconstruction task
and a guided essay-writing task—that were delivered in reverse orders in the
two studies. The results of both studies indicated a significant improvement
on the form only after the second phase of the treatment (i.e., both types of
tasks), which suggested the importance of extended opportunities to produce
output and receive input in effecting substantial learning.
In terms of task effects, both studies found that the essay-writing task was
more susceptible to individual variation than was the text-reconstruction task.
It seems that the greater freedom in production in the essay-writing task
makes the IL output–TL input comparison difficult vis-a
`-vis the target gram-
matical form, leading different learners to attend to vastly different aspects of
the input. On the other hand, the reconstruction task may have an advantage
in promoting noticing the gap when a specific form is targeted, as these tasks
maximize the similarities between the learner’s production and the TL model.
In general, however, output opportunities in these studies had variable effects
on noticing and learning of the form for different learners, which resulted in
blurring the overall between-group advantage of the output group. One pro-
posed explanation for this result is that presenting a text-length passage to
read all at once may have been exceedingly taxing on the attentional capaci-
ties of some subjects and made attending to form difficult. Another proposed
explanation, which finds some support in the IL analysis by Izumi and Bigelow
(2000), is that the target form in these studies may not be particularly notice-
able through the process of output followed by relevant input. It was sug-
gested that a combination of formal complexity and functional expendability
of the form might have contributed to this result.
Effects of Output and Input Enhancement 547
Four implications may be drawn from these studies for future research.
First, it would be important to provide learners with extended opportunities
to produce output and receive relevant input to ensure maximal benefit from
the output-input treatment. Second, for output to promote noticing and learn-
ing of a specific form, relevant task characteristics need to be considered
carefully. Also of interest in this regard is how output may be combined with
other focus-on-form techniques to promote greater learning. Third, care needs
to be taken to make sure that the learners’ processing capacity is not over-
loaded during output and input processing to allow for adequate allocation of
attentional resources to forms. Finally, the effects of output need to be investi-
gated with other linguistic forms. The present study takes these points into
account and aims to further our understanding of the input-output relation-
ships. Specifically, building on earlier studies, the present study examines
whether output and (visual) input enhancement, in isolation or in combina-
tion, promote noticing and learning of an L2 grammatical form.
Target Form
The target form in this study, English relative clauses (RCs), was chosen for
several reasons. Most generally, the acquisition of RCs in English has been
studied extensively by a number of researchers working on both first language
(L1) acquisition and SLA (see Doughty, 1988, 1991, and Izumi, 2000, for re-
views of relevant studies). These studies provide a rich source of information
as to the structures of English RCs, the natural developmental sequence of
acquisition of these forms, the potential facilitative role of instructional inter-
vention on them, and processing problems that both L1 and L2 learners face
in producing and comprehending them.
More specifically, the typological markedness relationship that holds
among different relativization types and the putative existence of natural de-
velopmental sequences of RC acquisition generate three implications that are
important for determining the sequencing, timing, and assessing of the in-
structional treatment. For sequencing, knowledge of the developmental se-
quences and the markedness relationship allows for an informed decision on
which structure, among many relativization types available in English, should
be targeted for instruction. For timing, knowledge of the developmental se-
quences enables us to give careful consideration to the learnability of the
form for particular groups of learners. For assessment, the markedness rela-
tionship informs us of how to assess the extent of learning caused by the in-
structional intervention (i.e., the degree of the generalizability of the
acquisition of the targeted form to other related structures in the implica-
tional relationship). As in Doughty’s (1988, 1991) studies, the instruction of
the present study focused on the object-of-preposition (OPREP) type of RCs.
548 Shinichi Izumi
Table 1. Four treatment groups included
in the study
Input enhancement Required Not required
Enhanced +O+IE –O+IE
Unenhanced +O–IE –O–IE
Research Design
The study was a controlled experimental study with a pretest–posttest design,
involving four treatment groups and one control group (see Table 1). The
treatment groups differed with respect to output requirements (notated as
±O) and exposure to enhanced input (notated as ±IE) such that: The +O–IE
group was required to produce output and was exposed to regular, unen-
hanced input; the +O+IE group was required to produce output and received
enhanced input; the –O+IE group received enhanced input without output;
and –O–IE group received unenhanced input without any output requirement.
The control group participated only in the pre- and posttests.
Dependent variables of the study included the noticing measures, which
were used to address the extent of noticing induced by output and input en-
hancement, and acquisition measures, which were used to address the extent
of learning brought about by the respective treatments. As will be subse-
quently described, two types of noticing measures were used: note scores de-
rived from notetaking done by the subjects during the input exposure phases
of the treatment, and immediate uptake of the form demonstrated in the sub-
jects’ production during the output phases of the treatment.
Thus, the factors
included in the study were: (a) two between-groups factors—the output con-
dition with two levels (+O and –O) and the input enhancement condition with
two levels (+IE and –IE); (b) time as a within-group factor with two levels—a
pretest and a posttest; and (c) two dependent measures—noticing and acqui-
Research Hypotheses
The research hypotheses were based on the consideration of the results of
the prior theoretical and empirical research previously discussed. Hypotheses
1 and 2 are concerned with the noticing issue in relation to output and input
enhancement, respectively. Hypothesis 3 was formulated on the basis of the
assumption that the learners’ noticing would be greater if the two attention-
drawing techniques are combined than when they are used separately.
Hypothesis 1: The noticing of the target form in the input would be greater for the
output subjects than for the nonoutput subjects.
Effects of Output and Input Enhancement 549
Hypothesis 2: The noticing of the target form in the input would be greater for sub-
jects receiving enhanced input than for those receiving unenhanced input.
Hypothesis 3: The noticing of the target form in the input would be greater for +O+IE
subjects than for +O–IE subjects and –O+IE subjects.
The fourth through sixth hypotheses are concerned with the acquisition
issue. If, as Schmidt (1990, 1995, 2001) claimed, noticing of the form is re-
quired for learning that form, it should follow that greater noticing of the tar-
get form induced via output, input enhancement, or a combination of both,
should lead to increased learning of the form.
Hypothesis 4: Output subjects would demonstrate greater learning of the target form
than would nonoutput subjects.
Hypothesis 5: Subjects receiving enhanced input would demonstrate greater learning
of the target form than would subjects receiving unenhanced input.
Hypothesis 6:+O+IE subjects would demonstrate greater learning of the target form
than +O–IE subjects and –O+IE subjects.
Hypothesis 7 is postulated based on the logical inference that, if the underly-
ing variable for the treatments is the focused attention given to the formal
features in the input (i.e., noticing), how it is induced—via output or input
enhancement—should make little difference in the learning outcomes as long
as noticing is somehow induced.
Hypothesis 7: By virtue of their attention-drawing effects, the two conditions of +O–IE
and –O+IE would produce a comparable amount of noticing and learning.
Subjects were recruited from among students enrolled in the ESL programs at
two major U.S. universities that were similar in both their programs and their
enrolled ESL student population. The selection of the subjects was determined
on the basis of a test of English relativization administered to intermediate-
level classes at the two institutions. This screening test also served as the pre-
test for the participating subjects. Following Doughty (1988, 1991), the present
study chose subjects who demonstrated emerging knowledge of relativization
but who had yet to acquire it beyond the initial subject relativization phase.
Specifically, students who demonstrated ample knowledge of the target struc-
ture or who failed to show any sign of knowledge of grammatical embedding
necessary for relativization were eliminated from the subject pool at this
point. An arbitrary cutoff score of 80% on the total pretest score was set to
exclude subjects from the former category. Students were determined to have
had at least rudimentary knowledge of relativization if they (a) produced tar-
getlike subject RCs at least once in the production tests, (b) produced other
types of RCs (i.e., direct object or OPREP types) in a targetlike way or in an IL
550 Shinichi Izumi
form (mostly with pronoun copies), or (c) responded to the items on the sub-
ject RC in the reception tests at better than chance scores.
Out of 110 students who took the pretest, 77 satisfied the learnability re-
quirements. Of these, 47 agreed to participate in the study and completed all
the treatments and the posttest. Fourteen randomly selected students who
satisfied the above requirements served as the control group and participated
only in the pre- and posttests. Thus, including the control group, the number
of the final pool of subjects was 61. To ensure pretreatment equivalence of the
subjects in their knowledge of English relativization, subjects were assigned to
different groups using a stratified random assignment procedure and the pre-
test results as the basis of stratification. This resulted in the following group-
ing: 11 for the +O+IE group; 12 each for the +O–IE, –O+IE, and –O–IE groups;
and 14 for the control group. To summarize the subject information for all
groups combined, the subjects’ L1s included Arabic (24), Chinese (6), French
(2), Japanese (3), Kazah (1), Korean (11), Persian (1), Polish (1), Portuguese
(1), Spanish (6), Thai (4), and Turkish (1). There were 34 males and 27 fe-
males. The subjects’ mean age was 26.2 years (SD =8.22). Their mean length
of residence in the United States was 6.3 months (SD =6.17). Almost all sub-
jects had completed their high school education prior to coming to the United
States. The mean length of English instruction was 6.6 years (SD =3.6). All of
these factors were divided roughly equally in each group.
Experimental Schedule
The experimental schedule of the study is as follows. First, the researcher vis-
ited each class and administered the pretest. The test was subsequently
scored for initial screening. The mean interval length between the pretest and
the commencement of the treatment for the subjects was 7.13 days. The treat-
ment, conducted in the written mode, consisted of six sessions and took place
over a 2-week period. During this period, subjects were requested to come to
the computer laboratory outside of class hours and complete the “lesson of
the day.” The researcher, present at the lab throughout the treatment, super-
vised and assisted subjects with procedural matters. After signing a consent
form and filling out a background questionnaire, subjects engaged in a prac-
tice session, the purpose of which was to familiarize them with the activities
that were done mainly on the computer and with production work done using
paper and pencil. Subjects were not told that this first session was a practice
session. The target sessions started with the second session. The topics of all
texts were related to relationships and marriage. In the few cases in which
subjects could not come to the session on six different days, they were al-
lowed to complete two sessions in one day. After the treatment period, the
researcher visited the students’ classes once again and administered the post-
test. The mean interval length between the last day of the treatment and the
posttest administration was 3.43 days. After the posttest was completed, all
the treatment subjects filled out a posttreatment questionnaire.
Effects of Output and Input Enhancement 551
In an attempt to control for the outside exposure to the target form, the
teachers of the participating classes were requested not to teach RCs and not
to answer any questions about RCs from their students during the experimen-
tal period; all teachers accepted and followed this request. Subjects were
asked not to discuss the treatment activities with other people. As a further
check, a posttreatment questionnaire probed whether any subjects consulted
with anyone or anything about the treatment activities and whom or what
they consulted if they did. The analysis of the questionnaire data and the cor-
responding analyses of the test scores did not reveal any obvious differences
in the subjects’ performance in this regard (see Izumi, 2000, for details).
Although it is assumed that L2 learners are equipped with psycholinguistic
mechanisms that underlie the Output Hypothesis (e.g., grammatical encoding,
monitoring, noticing mismatches; see Izumi, 2000), the earlier discussion sug-
gested that not all production circumstances provide ideal grounds on which
to promote sensitization to language forms. In particular, the capacity limita-
tions of human attentional resources are likely to affect the efficiency of the
encoding and monitoring processes in production.
Such effects, moreover,
are likely to be more pronounced for language learners than for mature L1
users, owing to the greater need of the former group to exercise controlled
processing that requires attentional control (Kormos, 1999). Precisely because
of the limited and selective nature of attention and the availability of an es-
cape hatch from focus on form in the production process, it is important to
ask not only whether output can enhance learning in general terms but also
under what conditions it contributes to language learning. The treatment
tasks described in the following subsections aimed to create an optimal condi-
tion for language learning.
Text Reconstruction Task: Rationale for Use and Major Design Features.
The task used in this study was a modified version of a text-reconstruction
task used in previous studies (Izumi & Bigelow, 2000; Izumi et al., 1999). Sev-
eral reasons underlie this decision. First, a reconstruction task by its very na-
ture has the potential for promoting comparisons between the IL output and
the TL input. It is essentially a meaning-based pedagogical activity that allows
learners to devote some attentional resources to form and that provides both
the data and the incentive for the learners to make IL–TL comparisons
(Thornbury, 1997). It is, in a sense, a unique “linguistic problem-solving task”
(Brett, 1994, p. 332). These are important features of the task in light of the
notions of focus on form, in which integration of form and meaning is empha-
sized. More specifically, one advantage of the reconstruction task lies in its
control over the content and form that learners produce. Maximizing the
equivalence between the learners’ output and the target input should promote
direct comparisons between their IL-output and the TL-input forms. Further-
552 Shinichi Izumi
more, a reconstruction task that provides a second opportunity for exposure
to the original input and for reconstruction permits the investigation of a shift
of learners’ noticing from the first to second input exposures and their uptake
of the form from the first to second outputs.
Overall, the following features characterized the design of the reconstruc-
tion task used in this study. First, the entire reading-based treatment was com-
puterized using the Libra program (Farris, 1993), which permitted both a
convenient data-collection schedule and a strict control over the time of input
exposure and sequence of the activities. Second, as a major modification on
the previously used reconstruction task, the input texts were divided into sev-
eral shorter, semantically coherent subsections to lighten the processing load
on the learners. Each subsection was made short enough for subjects to re-
member its content but long enough to make verbatim memorization difficult.
For the most part, each subsection consisted of three sentences. Third, the
input texts, which were specially created for this study, contained numerous
examples of the target form used in contexts, thereby creating a so-called in-
put flood condition for all groups (see Trahey & White, 1993).
Fourth, notetak-
ing was required of subjects as they read the input text. Notetaking was
chosen as an online measure of noticing in this study due to its relative advan-
tage in terms of task compatibility (i.e., the measure does not interfere with
the reading task) and the preciseness of the measure (i.e., the degree to which
the measure excludes items that are not attended to), despite its potential
weakness in completeness (i.e., the extent to which the measure includes all
items that are attended to). Given that no known measure of noticing (e.g.,
think-aloud, underlining, or retrospective interview) is perfect in all relevant
criteria, it was decided that the nonintrusive, task-compatible measure of
notetaking would be appropriate to tap the noticing phenomenon in this
study. Reconstruction was also used as another measure of noticing to par-
tially compensate for the weakness of the notetaking.
Treatment Procedures.
The goals of the task, as conveyed to the subjects
at the beginning of the treatment, were (a) to read and understand the text,
(b) to reconstruct the text as accurately as possible (for the output groups)
or to answer questions (for the nonoutput groups), and (c) to demonstrate
comprehension by writing a recall summary in their L1. Thus, comprehension
of the reading text was emphasized for all groups. Subjects were informed in
advance that they would have only a limited amount of time to read the given
section and could not go back to earlier sections; output-group subjects were
also told that they could not go back to their previous productions. Each of
the four treatment conditions followed the sequence depicted in Figure 1. Sub-
jects in all groups were first given the entire text and were instructed to read
it to get the gist of the text. They were given 1 minute for this skimming phase.
During input exposure 1, they were presented with the first subsection of the
text for closer reading. At this point, subjects in the two output groups were
instructed to “take notes of any and every word” that they thought was “par-
Effects of Output and Input Enhancement 553
Figure 1. Treatment sequence for each group.
ticularly important, necessary, or useful to reconstruct the text as accurately
as possible.” Subjects in the two nonoutput groups were instructed to “take
notes of any and every word” that they thought was “important, necessary, or
useful to understand the text.” All of the groups were given note sheets to use
for notetaking purposes. To discourage copying of all words in the text, sub-
jects were told not to write down entire sentences but rather to take notes
of only those words that they found important, necessary, or useful for their
respective purposes. The control for time of exposure was also set in such a
way that direct copying was not feasible.
During postexposure task 1, subjects in the output groups were instructed
to reconstruct the subsection as accurately as possible in both content and
grammar by writing on the reconstruction sheets. They were allowed to use
the notes they had taken to help them with their reconstruction. Subjects in
the nonoutput conditions, on the other hand, were presented with an exten-
sion question that asked for their opinion related to the topic discussed in
the text. Extension questions were used in lieu of comprehension questions to
prevent inadvertently drawing the learners’ attention to form, which was a de-
sign problem found in previous studies (Izumi et al., 1999; Izumi & Bigelow,
2000). Furthermore, the multiple-choice format of questions prevented inad-
vertent production of the target form by nonoutput subjects. An example of a
subsection of the reading text and ensuing extension questions is presented
554 Shinichi Izumi
in (1). Question 1 followed the first exposure to the text; question 2 followed
the second exposure.
(1) He arrived one hour late for the date last Friday. He said that the business meeting
which he participated in lasted a long time. Furthermore, the paper which he was
working on at the office took him a long time to finish.
Question 1: Do you show up for a date on time?
1. Always 2. Usually 3. Sometimes 4. Never
Question 2: Have you ever made up reasons for being late for a date?
1. Always 2. Often 3. Sometimes 4. Never
No time limit was set for either reconstructing for the output groups or an-
swering extension questions for the nonoutput groups.
After putting their note sheets (and reconstruction sheets for output sub-
jects) in the envelopes provided, subjects were shown the same subsection of
the text a second time (input exposure 2) and were directed to take notes on
another note sheet for their respective purposes. The enhanced groups were
presented with the enhanced text, whereas unenhanced groups were given
the same unenhanced text. Subjects in the enhanced condition were directed
to attend to the underlined (i.e., enhanced) portion of the text to examine how
the complex NPs were formed. To increase the salience of the target form in
the enhanced input, the head noun and the relative pronoun were highlighted
by a combination of bolding, shadowing, and different fonts and font sizes.
The preposition in the RC was also similarly highlighted to indicate the under-
lying site from which the relative pronoun was extracted. Additionally, the
whole NP containing the RC was underlined to make the identification of the
complex NP easier.
During postexposure task 2, the output subjects were asked to reconstruct
the given subsection on another reconstruction sheet, whereas the nonoutput
subjects answered another extension question. As soon as they completed
their respective tasks, they were told to put their note sheets (and reconstruc-
tion sheets for output subjects) in the given envelopes.
Each step was then repeated for each subsection until the entire text of the
day was covered. On completion of the entire text, subjects were directed to
write a recall summary (in their L1) on their recall-summary sheets for the
day’s material. The L1 was used here to further control for the L2 production
of the target form. This task was untimed.
In general, it took about 30–60 minutes for the output subjects to complete
each session, whereas it took about 30–45 minutes for the nonoutput subjects
to complete their lessons each day. The longer time spent by the output sub-
jects was due to the generally longer time required to reconstruct the text, as
compared to answering multiple-choice questions. It should be emphasized,
however, that the time of exposure to the input text was exactly the same for
all groups.
Effects of Output and Input Enhancement 555
Testing Instruments
To assess the subjects’ knowledge of English relativization before and after
the treatments, four different written testing measures were used. Two of
them—a sentence combination test and a picture-cued sentence-completion
test—examined the subjects’ productive knowledge; the other two tests—an
interpretation test and a grammaticality judgment test—aimed at testing their
receptive knowledge. These tests were given in the above order. In all tests,
three of the six relativization types represented in the Noun Phrase Accessibil-
ity Hierarchy (Keenan & Comrie, 1977) were included in equal number: subject
(SU), direct object (DO), and object-of-preposition relatives. Half of these
items had RCs embedded in the subject position and half in the object posi-
tion. To standardize the testing procedure, the entire test session was di-
rected by a recorded guide. The exact time allocated for the test, including
the time interval between test items, was determined on the basis of the re-
sults and feedback obtained from a series of pilot tests.
Sentence Combination Test.
The sentence combination test (SCT) was
adapted from Doughty (1988, 1991), who had adapted it from Gass (1982). In
this test, subjects were directed to combine two sentences in such a way that
the underlined words in the first sentence would be specified or identified by
using the information contained in the second sentence (e.g., The student went
downtown +I talked to you about the student The student who I talked to you
about went downtown). Subjects were told to begin their combined sentence
with the first sentence and not omit any information contained in the two sen-
tences. They were prohibited from using any coordinating conjoiners such as
and,because,so,orwhen. The test included 18 items and was completed in 15
Picture-Cued Sentence Completion Test.
The picture-cued sentence com-
pletion test (PCSCT) was also adapted from Doughty (1988, 1991), who had
adapted it from Hyltenstam (1984). Unlike these earlier studies, this test was
used in the present study as a written elicitation measure for logistical rea-
sons. For each item in this test, subjects saw a picture and heard a sentence
that described what was happening in each picture in a simple declarative
form (e.g., The man kisses the woman). They were then asked to complete a
sentence that identified the person in the picture, starting with a prompt (e.g.,
Number X is the woman ...).Tofurther encourage the use of particular relati-
vization types, subjects were told to follow the instruction about the focus for
each sentence (e.g., “focus on what the man does to the woman”) and not to
use the word and in their sentences. The test consisted of 12 items, with 15
seconds to complete each item.
Interpretation Test.
The interpretation test (IT) was specially created for
this study and was inspired by several sources: the interpretation-type tests
used in a series of studies by VanPatten and his colleagues (see VanPatten,
556 Shinichi Izumi
1996, for summaries of these studies); studies conducted within the frame-
work of the competition model, which underscores the importance of compe-
tition of different cues for sentence interpretation (e.g., Bates & MacWhinney,
1989; MacWhinney, 1987); and L1 research on children’s comprehension strat-
egies, which found children’s predominant use of the canonical order strategy
or the conjoined-clause analysis in interpreting RC sentences (e.g., Roth, 1984;
Tavakolian, 1981). In each item of this test, subjects heard a sentence (e.g.,
The cat bites the bird which chases the mouse) that was also written on the test
sheet. They were then given 15 seconds to decide which picture best de-
scribed the sentence among the three choices given. The incorrect alterna-
tives depicted some of the most common interpretations possible for the
sentence, which were carefully chosen based on the results of L1 acquisition
studies on English RCs previously mentioned (for closer analyses of the data
obtained in this test, see Izumi, 2000). The test consisted of 18 items.
Grammaticality Judgment Test.
The grammaticality judgment test (GJT)
was adapted from Doughty (1988, 1991) and consisted of 36 sentences. Each
sentence was presented via audiotape and was also written on the test sheet.
Subjects were given 15 seconds to decide whether a given sentence was cor-
rect or incorrect. If they decided that the sentence was incorrect, they were
asked to correct the error. Following Gass (1982), four error types were
included in the incorrect items: pronoun retention, nonadjacency, incorrect
relative-marker morphology, and inappropriate relative-marker omission. Sub-
jects were told that the who–whom distinction was not important for the pur-
pose of this test.
For the posttest, the same format was followed except that test items were
shuffled for all but the PCSCT, and words were substituted for other equally
simple words for the GJT and the SCT. Only these minor changes were made
because it was felt that more than 2 weeks of interval between the two testing
sessions were sufficient to counteract any possible practice effect, which was
checked and confirmed by examining the performance of the control group
(see the section entitled “Results”).
Scoring and Analyses
To measure noticing of the target form, scores were computed by tallying the
number of various categories of words written down on the subjects’ note
sheets and reconstruction sheets, deriving note scores and reconstruction
scores, respectively. Note scores were computed for the following categories
of words: words inside the relativized NP, words outside of it, content words,
function words, head nouns, relative pronouns, and prepositions that accom-
pany the predicate of the clause (e.g., look for). The analysis of the reconstruc-
tion was carried out by tallying the number of the following items found in
the reconstruction: any type of RC attempted, head nouns, relative pronouns,
prepositions associated with the predicate of the RC, and the correct use of
Effects of Output and Input Enhancement 557
the target RC. Because of space constraints, and in view of the overall similar-
ity of the results obtained in each of these analyses, this paper reports on
only the results pertaining to the three RC words (for notetaking) and to the
attempts and targetlike use of RCs (for the reconstruction).
All of the test data were scored by giving one point for a correct response
and zero for an incorrect response. The data from each testing measure were
scored separately and were then combined to obtain a composite test score.
Scoring was also done separately for three different RC types and for different
RC positions in the matrix clause, though these results are not reported here
(see Izumi, 2000). For the production tests, only the production of the RC that
was targeted in the given item was considered correct. Errors related to arti-
cles, tense, and spelling were ignored as long as they did not pertain to the
formation of the RC. Because omission of the relative pronoun is permissible
in the DO and OPREP types of RCs (but not for the SU type), it was considered
correct as long as everything else relevant to RC formation was correct. For
the IT, answers were scored as either a correct choice or an incorrect choice.
Given the problems a GJT can pose, such as the uncertainty of why a sub-
ject judged a sentence to be ungrammatical (see Sorace, 1996), the following
procedure was used. First, the subjects’ correction of errors was examined to
decide whether each item was judged on the basis of the RCs. If the correc-
tions indicated that subjects were judging the test sentences for reasons other
than RC-related problems (e.g., tense, articles, and prepositions), these items
were determined to be nonapplicable and were eliminated from further analy-
sis. RC-related corrections typically involved movement of the RC to a posi-
tion adjacent to the head noun if it was not already there, change of word
order inside the RC, change of relative pronouns, insertion of a relative pro-
noun when there was none (applicable only for the subject relativization), and
insertion or deletion of pronoun copies. Any items that were not judged or
that were judged incorrect but without corrections were excluded from the
analysis. The remaining items were then scored for the correctness of the
judgment. Owing to the elimination of some test items, which was caused by
subjects’ failure to complete the test in time or to follow the specified proce-
dure (e.g., not providing corrections in the GJT), test scores were calculated
for each subject by dividing the total correct scores by the total number of
applicable items. Each subject, therefore, received a percentage score for
each test.
Noticing Results
Results of the Note Scores.
Table 2 displays descriptive statistics on the
number of the three RC words noted by the four treatment groups. This table
displays the mean number of each category of words noted during the first
and second input exposures as well as the difference (i.e., increase) between
558 Shinichi Izumi
Table 2. Descriptive statistics for note scores: Head nouns (HN), relative
pronouns (RP), and prepositions (Prep) noted per text
Required Not required Total
1st 2nd Mean 1st 2nd Mean 1st 2nd Mean
Input enhancement input input diff. input input diff. input input diff.
HN 3.86 5.15 1.29 2.61 4.99 2.38 3.21 5.06 1.86M
SD 1.62 1.74 1.79 1.67 3.30 2.51 1.73 2.61 2.22
RP 2.66 4.29 1.64 0.40 2.83 2.43 1.48 3.53 2.05M
SD 2.27 2.27 1.95 0.40 3.40 3.31 1.94 2.94 2.73
Prep M2.78 4.27 1.49 1.74 3.54 1.80 2.24 3.89 1.65
SD 1.83 2.31 1.54 1.01 2.60 2.07 1.52 2.44 1.77
HN 2.74 3.02 0.28 3.11 3.23 0.12 2.93 3.12 0.20M
SD 1.11 1.40 1.27 1.63 2.11 0.99 1.38 1.75 1.12
RP 0.54 1.19 0.65 0.65 0.78 0.13 0.60 0.99 0.39M
SD 0.82 1.17 0.66 0.92 1.12 0.49 0.85 1.14 0.63
Prep M2.15 2.38 0.22 2.00 2.23 0.23 2.08 2.30 0.23
SD 1.40 1.63 0.77 1.48 1.85 0.98 1.41 1.71 0.86
HN 3.27 4.04 0.76 2.86 4.11 1.25M
SD 1.46 1.88 1.59 1.64 2.86 2.19
RP 1.55 2.67 1.12 0.53 1.81 1.28M
SD 1.96 2.35 1.48 0.70 2.69 2.61
Prep M2.45 3.28 0.83 1.87 2.89 1.02
SD 1.62 2.16 1.34 1.24 2.31 1.74
Note. Mean difference values for Mand SD were obtained by subtracting the value of the first input from that of the
second input for each individual subject and then dividing the total of all subjects by the number of subjects.
the two. Also displayed in this table are the results for the output versus non-
output groups (shown at the bottom) and for the input enhancement versus
nonenhancement groups (shown at the far right) to facilitate the comparison
along these two independent variables of the study.
An examination of the noting patterns of the three RC words revealed that
the two enhancement groups had a greater increase from the first to second
exposures in the number of each of these words noted than did the nonen-
hancement groups. These data were submitted to a 2 ×2×2×3 ANOVA, with
output and input enhancement entered as the two between-subjects factors
and time and word type entered as the within-subjects factors. The ANOVA
revealed significant main effects for input enhancement, F(1, 43) =10.84, p<
.01, time, F(1, 43) =18.36, p<.001, and word type, F(2, 86) =124.81, p<.001, as
well as significant interaction effects between time and input enhancement,
F(1, 43) =9.64, p<.01, word type and input enhancement, F(2, 86) =8.40, p<
Effects of Output and Input Enhancement 559
Figure 2. Mean note scores for the highlighted RC words by
the ±input enhancement groups.
.001, time and word type, F(2, 86) =13.47, p<.001, and time, word type, and
input enhancement, F(2, 86) =8.45, p<.001.
To identify the sources of the three-way interaction involving time, word
type, and input enhancement, an interaction plot was examined for the num-
ber of three RC words noted by the enhancement and nonenhancement
groups at the two input exposures (see Figure 2). A visual inspection of the
figure revealed that the enhancement groups experienced an increase from
the first to second exposures in all three types of words, whereas the nonen-
hancement groups did not change much for any of these words. Additionally,
a post hoc analysis of the word-type effect using paired t-tests (with an ad-
justed alpha level of p<.17 for three-way comparisons) revealed that the
three RC words were noted in significantly different amounts. Head nouns
were noted significantly more than prepositions: 3.57 versus 2.62, t(46) =
13.84, p<.001. Prepositions were noted significantly more than relative pro-
nouns: 2.62 versus 1.63, t(46) =5.64, p<.001. Though not strikingly different, a
relatively greater increase in the number of relative pronouns noted than for
the other two words (see Table 2 also) may have contributed to the significant
three-way interaction.
Within-group comparisons using paired t-tests generally confirmed these
observations. A significant increase in the number of head nouns was not ob-
served for the nonenhancement groups although it was observed for the two
enhancement groups: for +O+IE, t(10) =2.39, p<.05; and for –O+IE, t(11) =
3.28, p<.01. For relative pronouns, a significant increase was found for +O+IE,
t(10) =2.79, p<.05, for +O–IE, t(11) =3.41, p<.01, and for –O+IE, t(11) =2.53,
p<.05. For prepositions, a significant increase was observed only for the two
560 Shinichi Izumi
Table 3. Mean reconstruction scores for the number
of attempts at RCs and targetlike use of OPREP relatives
by the two output groups
First output Second output Mean diff.
Group M SD M SD M SD
+O+IE group
Attempts 5.29 1.26 8.04 0.60 2.75 1.10
Targetlike use 2.07 1.46 4.38 1.74 2.31 0.97
+O–IE group
Attempts 4.97 1.92 6.73 2.08 1.77 1.13
Targetlike use 2.02 1.78 3.58 2.34 1.57 1.25
Note. Mean difference values for Mand SD were obtained by subtracting the value of
the first output from that of the second output for each individual subject and then
dividing the total of all subjects by the number of subjects.
enhancement groups: for +O+IE, t(10) =3.21, p<.01; for O+IE, t(11) =3.11, p<
Thus, it seems that the groups that received visually enhanced texts took
a greater number of notes of the RC words as they engaged in the reading or
reconstruction tasks. To the extent that notetaking served as an adequate
measure of noticing, these results indicate a positive role played by the visual
input enhancement. In comparison, the effect of output was not evident here.
The next section focuses on the reconstruction produced by the two output
groups during the treatment.
Results of the Reconstruction Scores.
Table 3 displays the mean recon-
struction scores for the number of attempts at the RC and the targetlike use
of the OPREP relatives. Figure 3 shows these scores graphically.
Both groups increased in the number of attempts at RCs and the targetlike
use of the OPREP relatives from the first to the second output. The trend
seems to be for the +O+IE group to improve more than for the +O–IE group in
both the attempt and the targetlike use. In terms of the percentage of the num-
ber of attempts out of the total number of RCs included in the input texts, 56%
and 80% of the RCs were found to have been attempted by subjects in both
groups in the first and second reconstructions, respectively. For targetlike
use, 22% and 43% of the RCs embedded in the texts were reconstructed in a
targetlike manner in the first and second reconstructions, respectively. If the
amount of targetlike use is considered out of the number of attempts at the RC,
there was an average of 40% successful use of the target relatives in the first
reconstruction and 54% in the second reconstruction.
The data for the number of attempts and targetlike use were submitted sepa-
rately to two-way, mixed design ANOVAs using group (+O+IE vs. +OIE) as a
between-subjects factor and time (first vs. second output) as a within-subjects
factor. For the number of attempts, the ANOVA revealed a significant main
effect for time, F(1, 21) =94.40, p<.001, and a significant interaction between
Effects of Output and Input Enhancement 561
Figure 3. Mean reconstruction scores for the number of at-
tempts at RCs and targetlike use of the OPREP relatives by the
two output groups.
time and group, F(1, 21) =4.44, p<.05. Figure 3 reveals that the interaction
effect was caused by the greater increase over the two reconstruction oppor-
tunities for the +O+IE group than for the +O–IE group. For the targetlike use,
the ANOVA revealed a significant effect only for time, F(1, 21) =67.92, p<.001,
which suggests that subjects improved in their accurate use of the target rela-
tives regardless of the enhancement factor.
To summarize, although the enhancement seemed to have some positive
impact on the number of attempts at the RCs, in general the two output
groups behaved similarly in their overall success at reconstruction, which
may be taken to suggest that the two output groups did attend to the target
form in the input and in their output, attempting and succeeding in producing
a fair number of RCs in their reconstructions.
Test Results
Turning now to the test results, Table 4 displays the descriptive statistics for
the total test scores on the pretest, posttest, and the gains made from the pre-
to posttests by each treatment group. Shown in the bottom rows are the re-
sults pertaining to output versus nonoutput groups. The columns on the far
right show the results for the enhancement versus nonenhancement groups.
Before examining the effects of the treatments, pretreatment equivalence
of groups in their knowledge of English relativization was checked by sub-
mitting the pretest scores to a two-way multivariate analysis of variance
(MANOVA) in which output and input enhancement were used as the two be-
tween-subjects independent variables. The data were also submitted to a one-
way MANOVA comparing the pretest scores across the five groups (including
562 Shinichi Izumi
Table 4. Descriptive statistics for total test scores
Required Not required Total
enhancement Pretest Posttest Gain Pretest Posttest Gain Pretest Posttest Gain
M54.42 68.17 13.75 50.84 56.24 5.40 52.55 61.94 9.39
SD 12.25 14.71 9.83 16.48 18.71 7.23 14.40 17.62 9.41
M50.99 64.91 13.93 54.06 59.44 5.38 52.53 62.18 9.65
SD 14.41 16.11 9.41 14.32 15.86 5.74 14.14 15.88 8.79
M52.63 66.47 13.84 52.45 57.84 5.39
SD 13.23 15.19 9.38 15.19 17.04 6.41
Note. Gain values for Mand SD were calculated as in Tables 2 and 3. The results for the control group are: pretest
(M=53.82; SD =15.17); posttest (M=56.99; SD =13.82); gain (M=3.17; SD =6.15).
the control group). The results of these MANOVAs indicated no significant dif-
ference among the groups. Therefore, any difference that may be observed on
the posttest or on the pre- to posttests gains is attributable to the respective
The overall test results displayed in Table 4 reveal that the two output
groups experienced greater improvement on the posttest than did the two
nonoutput groups. To test for the statistical significance of these differences,
the gain scores were submitted to a two-way MANOVA. The results revealed a
significant main effect for output: Wilks’s Lambda .696 and Pillai trace .304,
F(4, 40) =4.36, p<.01. However, there was no significant main effect for input
enhancement and no significant interaction effect. This indicates that subjects
engaged in output–input activities gained significantly more on the tests of
relativization ability when compared to subjects engaged in input comprehen-
sion activities. The results also indicate, though, that subjects exposed to the
enhanced texts did not demonstrate significantly better gains on the test as
compared to those exposed to the unenhanced texts. A one-way MANOVA
comparing the gains of the five groups (including the control group) also re-
vealed similar statistical results in that a significant main effect was found for
group: Wilks’s Lambda .569 and Pillai trace .482, F(16, 162) =2.06, p<.05. A
Fisher’s PLSD test indicated that the +O+IE and +O–IE groups outperformed
the –O+IE, –O–IE, and control groups, whereas there was no significant differ-
ence between the two output groups, between the two nonoutput groups, or
between the two nonoutput groups and the control group. Thus, it seems that
the output–input treatment had a significant impact on the learning of English
relativization, whereas input enhancement and a comprehension-based treat-
ment failed to induce an equally beneficial effect on learning.
As a further test for the changes made by different groups, paired t-tests
Effects of Output and Input Enhancement 563
Figure 4. Mean gain scores on the four testing measures by the five
were performed on the total pre- and posttest scores for each group. These
tests revealed significant improvement on the relativization ability over the
two testing sessions for all groups except for the control group: for +O+IE,
t(10) =4.64, p<.001; for +O–IE, t(11) =5.13, p<.001; for –O+IE, t(11) =2.57, p<
.05; for –O–IE, t(11) =3.24, p<.01; and for the control group, t(13) =1.93, p=
.076. The fact that not only the two output groups but also the two nonoutput
groups improved significantly seems to indicate some positive impact of the
exposure to the input flood during the treatment.
To examine the effectiveness of the treatment, effect sizes were computed.
The effect size estimates the magnitude of change attributable to instructional
treatments and gives us valuable information as to the practical importance of
the treatment effect without regard to the sample size of the study (Kramer &
Rosenthal, 1999). Following a recent study of meta-analysis of L2 instruction
studies by Norris and Ortega (2000), Cohen’s dwas used to calculate the ef-
fect sizes for the experimental groups as compared to the control group. Us-
ing the composite gain score means, the following effect sizes were obtained:
1.12 for +O+IE, 1.14 for +O–IE, 0.34 for O+IE, and 0.37 for –O–IE. When these
values are compared with the mean effect size observed in Norris and Orte-
ga’s study for treatments of the kind that the current study adopted (i.e., fo-
cus-on-form implicit treatments: d=0.69), we can see that the figures obtained
for the present output–input treatment are quite large, whereas input-only
treatments had a small magnitude of effect.
The MANOVA results previously reported indicated a significant main effect
for output. This prompted further statistical analyses on this variable. Conse-
quently, univariate analyses of variance (ANOVAs) were performed on the gain
scores obtained from each testing measure to determine in which test(s) the
significant effect for output was obtained (see Figure 4). The ANOVAs revealed
564 Shinichi Izumi
Table 5. Summary of the results
Hypothesis Prediction Results
H1 Noticing of the target form: Not supported
+output >–output
H2 Noticing of the target form: Supported
+enhancement >–enhancement
H3 Noticing of the target form: Mixed; on note score, +O+IE, –O+IE >
+O+IE >+O–IE, – O+IE +O–IE; on reconstruction score, +O+IE
>+O–IE in RC attempt
H4 Learning of the target form: Supported
+output >–output
H5 Learning of the target form: Not supported
+enhancement >–enhancement
H6 Learning of the target form: Mixed; +O+IE =+O–IE >–O+IE on test
+O+IE >+O–IE, – O+IE gains
H7 Noticing and learning: Not supported; on note score, +O–IE <
+O–IE =–O+IE –O+IE; on test gains, +O–IE >–O+IE
a significant effect for output on the SCT, F(1, 43) =7.32, p<.05, and on the IT,
F(1, 43) =4.99, p<.05. To include the control group, another set of ANOVAs
was also run on the same data set using Group as the independent variable. It
indicated a significant effect for group for the SCT, F(4, 56) =3.48, p<.05. A
Fisher’s PLSD test revealed that the +O+IE group outperformed the control
group, and the +O–IE group outperformed all other groups except for the
+O+IE group.
Thus, although not all testing measures showed equally strong discrimina-
tion among groups and over the two testing sessions, it appears that the bene-
fits of output were not limited to one mode of relativization ability but
manifested in both production and comprehension measures.
The summary
of the results in terms of the seven hypotheses is presented in Table 5.
Further Analysis: Comprehension of the Reading Texts
One of the most important features of focus on form as advocated by Long
(1991) and others (Doughty & Williams, 1998; Long & Robinson, 1998) is its
aim at achieving integration of attention to form and meaning as learners en-
gage in meaning-oriented activities. In view of the positive learning outcomes
for the output groups, the question may be raised as to whether any deterio-
ration of comprehension occurred for the output subjects as a result of their
greater attention to and processing of the form. To investigate this question,
recall summaries written by all treatment subjects were analyzed using the
procedure adopted by Doughty (1988, 1991). In this procedure, the researcher
first prepared a list of propositions representing the major ideas contained in
each text. Then, native or near-native speakers of the given language exam-
ined whether each proposition was understood fully, partially, or not at all.
Effects of Output and Input Enhancement 565
Subsequently, these evaluations were converted into comprehension scores
by giving one point if the proposition was evaluated to have been understood
fully, one half point when understood partially, and no point when not under-
stood at all. A mean score was then calculated over all the texts that were
read and summarized for each subject, yielding an overall comprehension
The mean comprehension score obtained for the two output groups was 65.0
(SD =12.2; for +O+IE, M=64.4, SD =10.6; for +O–IE, M=65.5, SD =13.9), and the
mean score for the two nonoutput groups was 67.5 (SD =17.0; for –O+IE, M=
60.7, SD =16.8; for –O–IE, M=75.0, SD =14.6). Although there are some between-
group differences in the descriptive statistics, a two-way ANOVA revealed no
significant main effect for either output or input enhancement and no signifi-
cant interaction effect; these results were also corroborated by a one-way
ANOVA using Group as the independent variable. This indicates that the output
subjects obtained their significant gains in learning while maintaining a com-
parable level of comprehension to that of the nonoutput subjects. The mean
comprehension score of 65% for the output groups also suggests that these
subjects, like subjects in the nonoutput groups, achieved a fairly high level of
comprehension of the reading passages (especially considering the demand-
ing nature of the recall summaries that test not only the learners’ comprehen-
sion but also their memory of the passage content). Thus, it may be concluded
that the output subjects successfully learned the target form without sacrific-
ing comprehension. In focus-on-form terms, it appears that the output–input
treatment in this study succeeded in achieving an adequate level of integra-
tion of attention to form and meaning. Nonoutput groups, on the other hand,
achieved a comparable level of comprehension to that of the output subjects
yet failed to demonstrate as much learning of the form. This seems to corrob-
orate findings of earlier studies that comprehensible input or even compre-
hended input is not sufficient to ensure automatic learning of the form
embedded in the input.
Three major findings are evident in this study. First, those engaged in the
output–input treatment outperformed those exposed to the same input for the
sole purpose of comprehension in learning English relativization. Although
note-score analyses did not reveal any significant impact of output, the imme-
diate uptake of the form evidenced in the reconstruction suggested that out-
put subjects attended to the target form in the input and in their own output.
It was also observed that a relatively high level of comprehension of the input
texts accompanied the superior learning by these subjects. Second, in con-
trast to the positive effect of output, visual input enhancement failed to show
any measurable effect on learning. This was so, despite a clear indication in
the note scores that the enhancement had a significant impact on the noticing
of the target-form items in the input. Third, in view of the above, it did not
566 Shinichi Izumi
appear that the effects of output on noticing and learning were comparable to
those of input enhancement. In what follows, an integrated account of these
findings will be sought in reference to the theoretical underpinnings of—and
previous research on—noticing, output, and input enhancement.
The Effect of Output on Learning English Relativization:
The Noticing Function of Output
The positive effect of output demonstrated in this study is consistent with the
hypothesized function of output in SLA. In particular, the noticing function un-
derscores the interconnectedness of input and output processes in SLA.
Based on previous research that indicates insufficiency of comprehensible in-
put in driving L2 development and on research that points to the importance
of attention in learning, it was hypothesized that learners’ attention somehow
needs to be drawn to the crucial form features to promote their learning. Out-
put was considered to be one means to achieve this aim by prompting the
learners to find problems in their IL through their production attempt. It was
further assumed that, on exposure to relevant input immediately after their
production experience, the heightened sense of problematicity would lead
them to pay closer attention to what was identified to be a problematic area
in their IL. In short, pushed output can induce the learners to process the
input effectively for their greater IL development.
The results of the learners’ reconstruction lend some support for the atten-
tion-inducing effect of output. The test results further indicated superior
learning of the form by the output subjects as compared to subjects who were
exposed to the same input for the sole purpose of comprehension. That such
superior performance of the output subjects was found not only in a produc-
tion test but also in a comprehension test reinforces the contention that the
output treatment affected the subjects’ IL knowledge, not just their use of
some production strategies. Both the process and product measures used in
this study, therefore, lend some empirical evidence for the noticing function
of output with respect to the acquisition of English relativization by adult L2
Reexamination of Noticing and Its Relationship to Input
Enhancement, Output, and SLA
In contrast to the positive results of output, why did visual input enhance-
ment fail to induce greater learning of the form? It was previously argued
that one of the differences between input enhancement and output is that,
whereas attention in the former is induced by external means, attention in
the latter arises internally through production processes. Interpreting the
present results in this light, it appears that these two approaches to focus
on form do not promote learning with an equal level of efficacy even though
they both supposedly involve learner attention as the crucial variable. Out-
Effects of Output and Input Enhancement 567
put, as an internal priming device, clearly worked better to promote learning
in this study.
Another difference between output and visual input enhancement may be
that pushed output, by virtue of producing utterances, can place the learner
in an ideal position to make a cognitive comparison between the IL and TL
forms (Doughty, 2001; Nelson, 1987; Saxton, 1997a, 1997b), whereas this may
not be the case with input enhancement. As Saxton’s Contrast Theory claims,
SLA, as with L1 acquisition, may be promoted when the learners’ error in their
output is followed immediately by a juxtaposed targetlike form that shows
contrast to the learners’ preceding utterance. This, Saxton claimed, may cre-
ate a favorable condition for the learners to make a cognitive comparison be-
tween their IL form and the TL form, leading them to expunge the
nontargetlike form from their developing IL in favor of the TL form. Recent
studies on the role of recasts in SLA show that L2 learners can indeed benefit
from being exposed to the targetlike use of form in discoursally contingent
positions (e.g., Doughty & Varela, 1998; Long, Inagaki, & Ortega, 1998;
Mackey & Philp, 1998).
It is possible that the cognitive comparison may be induced by visual input
enhancement as well. However, the enhancement, which is solely concerned
with drawing the learners’ attention to form, does not necessarily encourage
further cognitive processing that may be necessary for acquisition. This con-
dition of input enhancement might render form learning essentially a hit-
or-miss affair, with only some of the learners likely to benefit fully from it
(perhaps those with most form-conscious tendencies, those with metalinguis-
tically sophisticated prior knowledge, or both). Thus, the contrasting results
of the output and the input enhancement in this study may be explained by
positing that the input enhancement was not sufficient to induce noticing of
the IL–TL mismatches. Output, in contrast, promoted both the processes of
noticing the form and noticing the mismatches, which enabled the output
learners to attain successful and superior learning of the form. With these gen-
eral accounts as a backdrop, further illumination of the learning processes is
sought in the following section through explanations of a more psycholinguis-
tic nature.
Amount of Processing versus Levels, Depth,
and Type of Processing
The lack of a significant advantage of enhancement on learning was particu-
larly striking on consideration of the fact that there was strong evidence in
the notetaking measure that the enhancement subjects paid attention to vari-
ous form elements in the input. To find a psycholinguistically informed ac-
count for these results, the basic logic behind the hypothesized effect of input
enhancement needs to be reexamined. The logic is generally set up in two
steps. First, the perceptual salience created by highlighting the input will draw
the learner’s attention to the highlighted forms. Second, once the first step is
568 Shinichi Izumi
successful, learning of the attended form will occur based on the premise that
attention is what mediates input and intake.
Regarding whether the first step in this logic was met in this study, one
may argue that subjects in the enhanced groups did not in fact notice the tar-
get forms—that is, because of measurement problems associated with note-
taking, one cannot assume that taking notes of form-related words equals
noticing of these forms. However, the problem with notetaking in this study
seems to be with underestimation of actual noticing rather than with overesti-
mation. For example, one can argue that the absence of RC words written as
notes does not necessarily mean that no attention was paid to these words,
given the incompleteness of the measure. However, it is difficult to argue that
the presence of words written down as notes does not mean that at least
some attention was paid to these words; it is difficult to imagine a situation
where one bothers to write down words without paying any attention to them.
This is all the more true when notetaking of these words occurs rather consis-
tently and significantly, as was the case for the enhancement subjects in this
study. Given this understanding, then, the significantly greater amount of
form-related words noted by the enhancement subjects can be taken to indi-
cate that their attention was indeed drawn to the form. (See also Izumi, 2000,
on the responses to the posttreatment questionnaire, which lend further sup-
port to this conclusion.) If the minimum requirement of noticing, as defined
by Schmidt (1990, 1995, 2001), is to pay attention to key grammatical elements
in the input with greater than a threshold level of subjective awareness (i.e.,
reportable subsequent to the experience), then these data indicate that the
basic requirement of noticing has been met, in Schmidt’s sense of the term.
The second step in the logic states that success in drawing the learners’
attention would lead to learning the attended form. An implicit assumption
seems to be that what matters with learners’ attention is its focus and quan-
tity; the basic premise is that the more attention to the form, the better its
chance of acquisition. The method of notetaking that was used in this study
was based on this basic premise (i.e., the quantitative analyses of the notes
would reflect the degree of learners’ noticing and, thus, learning of the form).
As a result of this emphasis on quantitative aspects of attention in the com-
monly used attentional research paradigm in SLA, what has been left aside is
the consideration of the qualitative aspects of attention—namely, what levels
and type of attention and processing are engaged with the forms in focus. On
the basis of the results obtained in this study, however, it appears that these
are precisely the issues that need to be examined more critically.
Levels and Depth of Processing.
In cognitive-psychology literature, per-
ception is considered to involve the rapid analysis of stimuli at a number of
levels. Preliminary stages are concerned with superficial features, such as
physical aspects of the stimuli, whereas later stages are more concerned with
matching the input against the already stored knowledge base. In their semi-
nal paper on memory research, Craik and Lockhart (1972) referred to this no-
Effects of Output and Input Enhancement 569
tion of a series of processing stages as depth of processing, where depth is
defined in terms of the relative degree of semantic and cognitive analysis and
elaboration done on the input stimuli. Craik and Lockhart argued that “analy-
sis proceeds through a series of sensory stages to levels associated with
matching or pattern recognition and finally to semantic-associative stages of
stimulus enrichment” (p. 675). In this levels-of-processing framework of hu-
man memory, the persistence of memory traces is understood to be a function
of the depth of analysis, with deeper levels of analysis leading to more elabo-
rate, longer lasting, and stronger traces. Maintaining information at one level
of processing by rehearsing it repeatedly or by sustaining continued attention
to certain aspects of the stimulus will not, by itself, lead to improved retention
unless a shift to deeper levels of analysis occurs.
Considering the findings of the present study within this framework, it is
possible that attention evidenced by the enhancement took place at a rela-
tively shallow level without necessarily shifting to deeper and more elaborate
processing levels. If so, it should be no surprise, based on Craik and Lock-
hart’s (1972) argument, that the evidence of the large amount of attention to
certain form elements alone is not directly related to learning of the new asso-
ciative connections existing in the input stimuli. In other words, it does not
matter how much attention is devoted to the form unless the quality of atten-
tion somehow changes to involve deeper and more elaborate processing.
What causes deeper levels of processing? Craik and Lockhart argued in this
regard that
the depth at which primary memory operates will depend both upon the
usefulness to the subject of continuing to process at that level and also
upon the amenability of the material to deeper processing....There are at
least three sources of the failure of processing to reach this [deep] level:
the nature of the material, limited available processing capacity, and task de-
mands....Manipulations that influence processing at a structural level
should have transitory, but no long-term effects....Long-term recall
should be facilitated by manipulations which induce deeper or more elabo-
rate processing. (pp. 679–680; emphasis added)
In the current study, the “nature of the material” refers to the input materials
and the target form, both of which were equivalent for all groups of subjects.
The target form, moreover, was carefully chosen in light of the learnability
considerations so that it would be learnable by the selected subjects. The
“limited available processing capacity” was also carefully controlled by pre-
senting only a manageable batch of input sentences at a time, which was also
held equivalent for all groups. On the other hand, task demands were varied
in the present study along the two independent variables, output and input
enhancement. Thus, in this framework of analysis, it could be argued that the
input enhancement may have caused mere recirculation or rehearsal at the
same, relatively shallow processing level, which led the learners to experience
only a short-term retention of the attended form. On the other hand, the
570 Shinichi Izumi
greater learning evidenced by the output subjects suggests that output trig-
gered deeper and more elaborate processing of the form, which led them to
establish a more durable memory trace.
Type of Processing.
Another concept that is closely related to the levels
of processing is the notion of type of processing. In the previous SLA litera-
ture, type of processing (or of attention) has been discussed in terms of, for
example, focal versus peripheral attention, implicit versus explicit learning, in-
tentional versus incidental learning, and noticing versus understanding (e.g.,
Schachter, 1998; Schmidt, 1990, 1995, 2001; Tomlin & Villa, 1994). Other types
of processing important to SLA are discussed by Robinson (1995), who, in de-
fining noticing as detection plus rehearsal in short-term memory, argued that
learning outcomes are determined by the level of activation of information in
short-term memory, which is caused by rehearsal and elaboration.
As compared to these various conceptualizations of the types of cognitive
processing, little attention has been paid in the SLA literature to the notion of
integrative processing. Graf (1994) explained it as follows:
Integration focuses on connections among the units that define an individ-
ual item, such as a word, an object, or a sentence; these kinds of connec-
tions are formed or strengthened when the subject either perceives
coherence among separate stimulus components (e.g., under the guidance
of preexisting representations or gestalt laws like proximity or common
fate) or conceives a structure for processing target features as a single en-
tity. (p. 685)
Graf and Schacter (1989) argued that so-called unitization, which is a prod-
uct of integrative processing, does not happen automatically when an item is
presented for study but instead depends on study activities that involve con-
ceiving the presented stimuli as coherent entities. The kind of activities that
promote integrative processing and unitization were discussed by Graf and
Schacter in the context of their investigation—that is, learning or remember-
ing paired words in one’s L1 by requiring subjects to generate a sentence that
connects the two words in a meaningful manner (e.g., from book–forest, one
produces The book included a picture of a forest). It is argued that semantic,
elaborative processing thus invoked “provides the ‘glue’ (the structure) that
unitizes sensory-perceptual and other features of paired words that are pro-
cessed concurrently during the study trial” (p. 939).
The notion of integrative processing is of great relevance to SLA because it
underscores the importance of not only attending to individual form elements
that make up a structure but also of perceiving or conceiving the relationship
among them to cause structure (grammar) learning. In a study on the roles of
memory and attention in SLA, Williams (1999) invoked the notion of integra-
tive processing to account for the results of his study. He argued that knowl-
edge of distributional rules, such as morphological agreement among the
determiner, noun, and adjectives in Italian, does not simply emerge out of
Effects of Output and Input Enhancement 571
memory encodings of the relevant forms but rather depends on how atten-
tional resources are appropriately allocated over the relationships among re-
lated form elements. In other words, unless one perceives the relationship
among related form elements, morphological concordances may never be ac-
quired. In a similar vein, applying the notion of integrative processing in the
present study, it may be argued that, to learn English RCs, one needs to not
only pay attention to key form elements in the input, such as a head noun, a
relative pronoun, and a preposition, but also to focus one’s attention on how
they are related to one another (e.g., the relative pronoun is associated with
the head noun as being coreferential with it; its extraction site is identified).
Attending to the individual items like which or who, no matter how intensely
one does so, will not by itself lead to the acquisition of the RC structure un-
less these items are grasped in relation to other related items in the same
clause or sentence.
Furthermore, if semantic or elaborative processing induced via a sentence-
generating task prompts unitization in remembering paired words by adult
native speakers, it may be argued in the context of the present study that at-
tention to meaning and the semantic processing that was induced in the com-
prehension task served only as a rough guide directing L2 learners to how the
individual form elements are related to one another. On the other hand, out-
put processing, required in the output treatment, pushed these learners fur-
ther in their cognitive processing and prompted them to perceive or conceive
the unitized structure. This occurs by virtue of the grammatical encoding op-
erations performed during production. As a consequence, the output task
served effectively both as the stimulator of integrative processing and as the
glue to connect individual form elements, which, one might say, were only
vaguely related to one another during the comprehension process.
Rethinking the Roles of Output and Input Enhancement in SLA
Based on the foregoing discussion, I posit that the cognitive processing neces-
sary to acquire a complex structure like the English RC involves (a) sensory
detection of key form elements such as the head noun and relative pronoun,
and (b) further cognitive processing—notably, integrative processing—that
connects and organizes related form elements and captures them as a coher-
ent structural set. Following Robinson (1995), noticing can be defined here as
sensory detection plus further cognitive processing on the detected data, and
these two processing operations together are hypothesized to lead to
Schmidt’s (1990, 1995, 2001) sense of noticing that is both necessary and suffi-
cient for learning. Cursory and superficial processing of input does not lead
to learning of the target structure, no matter how consciously or intensely one
attends to a particular form.
To summarize, then, output, especially pushed output, promotes not only
detection of forms but also integrative processing to conceive a coherent
structure among the detected elements. This, together with the juxtaposition
572 Shinichi Izumi
of the IL–TL forms, creates a favorable condition for the learner to notice the
mismatches between these two versions. The significant learning of the struc-
ture for the output subjects in this study can thus be accounted for. On the
other hand, visual input enhancement encourages sensory detection by
means of highlighting, but this does not necessarily lead to integrative pro-
cessing. The automatic link from sensory detection to further cognitive pro-
cessing cannot be assumed. Seen in this way, the discrepancy between the
notetaking results and learning outcomes found in this study may be ex-
plained by assuming that the notetaking reflected the learners’ detection of
individual form elements in the input but not whether or what kind of connec-
tions or integrations they have achieved among them. In other words, notetak-
ing of the kind used in this study may have been too gross a measure to
capture learners’ noticing, if noticing is to mean detection plus further cogni-
tive processing.
This line of argument also finds support in the findings of previous SLA
studies on visual input enhancement, which seems to work best in combina-
tion with other forms of assistance (e.g., semantic elaboration, as in Doughty,
1988, 1991; a focused production task, as in Williams, 1999; or activation of
prior knowledge, as in Shook, 1994) but not so well when used in isolation. It
appears that highlighting alone is often not enough to prompt the learners to
go beyond simple detection of forms, and additional instructional assistance
of some sort may be needed to trigger further cognitive processing, at least
for some complex grammatical forms.
To recapitulate, the advantage of pushed output may lie in promoting three
related processes that are important to learning complex grammatical forms
such as English RCs: (a) detection of formal elements in the input via priming
induced by internal feedback and monitoring mechanisms engaged during
production processing; (b) integrative processing of the target structure,
prompted by the grammatical encoding operations engaged during production
processing; and (c) noticing of the mismatches between one’s IL form and the
TL model, which is fostered by the engagement of these processes and by
the juxtaposition of the two versions that highlight any differences between
the two uses of the form.
Suggestions for Further Research
Some of the issues remaining to be studied include the following. First, long-
term effects of the output–input treatment need to be examined. Second, for
both output and input enhancement, possible interactions between the treat-
ment type and form type need further investigation. For this, in addition to
consideration of learnability and other factors, such as formal complexity and
functional importance attributed to a given form, it will be necessary to con-
sider the kind of cognitive processing that may be needed for the learner to
acquire the form. For example, does the form to be acquired need to be pro-
cessed at the word level (e.g., content words), phrasal level (e.g., adjective–
Effects of Output and Input Enhancement 573
noun agreement in Spanish), clausal or sentential level (e.g., RCs), or
suprasentential level (e.g., articles, choice of two verb tenses in a given situa-
tion)? The form type may also be considered in terms of whether the learners
already possess at least partial knowledge of the form or whether the form is
completely new to them. Further investigation into the effectiveness of differ-
ent combinations of instructional techniques will also be useful. In particular,
future studies need to consider carefully what type of enhancement proce-
dures stimulates the right kind of cognitive integration that is necessary for
the form to be acquired. Modality is yet another issue: Will the same results
be obtained if the modality is changed from the written to the oral or aural
modes? The prevalence of oral language use in naturalistic and classroom con-
texts suggests the importance of tackling this issue.
This study set out to investigate whether output and input enhancement, in
isolation or in combination, can promote noticing and learning of English re-
lativization by adult ESL learners. The results found in favor of output sug-
gested a three-fold facilitative impact of pushed output on L2 acquisition. That
is, under some circumstances, output promotes (a) detection of formal ele-
ments in the input, (b) integrative processing of the target structure, and (c)
noticing of the mismatches between one’s IL form and the TL input. The lack
of any significant impact of visual input enhancement also suggested that
these advantages of output may not be shared by the superficial external ma-
nipulation of the target form in the input, which, without any additional in-
structional assistance, may help only with detection of the highlighted form
items but does not necessarily engage the learner in further cognitive process-
ing. Without denying the essential role of input in SLA, the present study pro-
vided empirical evidence for a psycholinguistically motivated role of output in
L2 development. It is expected that further research on the roles of output
and various forms of input enhancement will be pursued with rigor and in-
creasingly refined methodology so that SLA research can be usefully and ef-
fectively applied to language teaching.
(Received 26 September 2001)
1. Thus, the reconstruction task was used both as a treatment task for the output groups and
as a noticing measure. Another measure—a posttreatment, retrospective questionnaire—was also
used to probe the attentional focus of the subjects during the treatment. Its results, however, are
not reported in this article (see Izumi, 2000, chap. 6).
2. It should be noted that recent theories of attention (e.g., Wickens, 1989) assume that there
are multiple pools of resources available in human attentional mechanisms. According to these theo-
ries, effective time-sharing or dual-task performance is considered possible if the task draws on dif-
ferent pools of resources (e.g., verbal vs. spatial, as in talking while driving). See Robinson (1995) for
a discussion of this issue as applied to SLA.
3. Given that an input flood is a type of input enhancement, one could say that all treatment
574 Shinichi Izumi
groups received some enhanced input. In the target texts, there were a total of 46 target sentences
out of 73 sentences in 890-word-long passages. Roughly half of the RCs contained the marker who
and half which. Also, about half were embedded in the subject position and half in the object position
in the matrix clause. As in most other studies on English relativization, this study used the preposi-
tion-stranding construction (e.g., the student who(m) the teacher is looking for) rather than the pied-
piping one (e.g., the student for whom the teacher is looking).
4. The reliability of all tests was checked using the Kuder-Richardson 20 (K-R20) and Kuder-
Richardson 21 (K-R21) reliability measures. In all cases, the values obtained were greater than .80
for K-R20 and greater than .78 for K-R21, which were taken to be reasonably high.
5. It might be useful to compare these effect sizes with those obtained in a similar study by
Doughty (1988, 1991), from which the present study obtained many insights. The effect sizes of the
meaning-oriented group and the rule-oriented group in her study were 1.34 and 1.16, respectively.
Although we need to take into account various differences in the treatments used in the two studies
(notably the amount of time for input exposure, which is substantially larger in Doughty’s study; see
Izumi, 2000, chap. 7 for a detailed comparison), their comparison underscores the effectiveness of
the present output–input treatment.
6. The different results obtained for the first and second sets of univariate analyses may be
explained by the fact that subdividing the output variable by the input enhancement variable re-
sulted in a smaller nfor each group, which led to reduction in the statistical power.
7. A possible explanation for the failure of the GJT to show any between-groups discrimination
may be due to a ceiling effect because all groups started out with already relatively high scores on
this test at the time of the pretest. The failure of the PCSCT to discriminate among groups and over
time seems related to the general failure of this test to elicit the target RCs, despite the efforts to
improve this test through piloting.
8. The data from four subjects were eliminated due to their failure to follow the given directions.
The resultant Nfor this analysis was 43. The reliability of the scoring was checked by having two
raters evaluate half of the recall summaries independently. An interrater correlation of r=.84 was
obtained, which was taken to be reasonably high for the measure.
9. It is relevant to note that later work in cognitive psychology (e.g., Graf & Mandler, 1984) found
that, although the level-of-processing effect is quite robust on explicit memory tests (e.g., cued recall
test), it was not evident on implicit memory tests that do not require conscious effort to recollect
specific prior episodes (e.g., stem completion test). In this regard, it is of interest to note that in-
structional effects in SLA studies are often measured using instruments that call on more explicit
memory-based performance (Norris & Ortega, 2000). This study is no exception, and it is possible
that different outcomes would have been obtained if more implicit types of assessment instruments
had been used. Proper investigation of this issue would require careful consideration of at least
three issues: (a) What type of tests constitutes implicit memory tests in SLA? (b) How applicable to
SLA are the findings of adult subjects’ showing automatic priming effects with familiar L1 words in
implicit memory tests? and (c) What is the practical value of such an automatic activation effect if
any such effect is obtained in SLA at all?
Alanen, R. (1995). Input enhancement and rule presentation in second language acquisition. In
R. Schmidt (Ed.), Attention and awareness in foreign language learning (Tech. Rep. No. 9, pp.
259–302). Honolulu: University of Hawai‘i Press.
Bates, E., & MacWhinney, B. (1989). Functionalism and the competition model. In B. MacWhinney &
E. Bates (Eds.), The cross-linguistic study of sentence processing (pp. 77–117). New York: Cam-
bridge University Press.
Brett, P. (1994). Using text reconstruction software. ELT Journal, 48, 329–336.
Craik, F., & Lockhart, R. (1972). Levels of processing: A framework for memory research. Journal of
Verbal Learning and Verbal Behavior, 11, 671–684.
Doughty, C. (1988). The effect of instruction on the acquisition of relativization in English as a second
language. Unpublished doctoral dissertation, University of Pennsylvania, Philadelphia.
Doughty, C. (1991). Second language instruction does make a difference. Studies in Second Language
Acquisition, 13, 431–469.
Doughty, C. (2001). Cognitive underpinnings of focus on form. In P. Robinson (Ed.), Cognition and
second language instruction (pp. 206–257). New York: Cambridge University Press.
Doughty, C., & Varela, E. (1998). Communicative focus on form. In C. Doughty & J. Williams (Eds.),
Effects of Output and Input Enhancement 575
Focus on form in classroom second language acquisition (pp. 114– 138). New York: Cambridge Uni-
versity Press.
Doughty, C., & Williams, J. (1998). Pedagogical choices in focus on form. In C. Doughty & J. Williams
(Eds.), Focus on form in classroom second language acquisition (pp. 197–261). New York: Cam-
bridge University Press.
Ellis, R. (1994). The study of second language acquisition. Oxford: Oxford University Press.
Farris, M. (1993). Libra: An authoring environment for multimedia lessons on the Macintosh (Version
1.1) [Computer software]. San Marcos, TX: Division of Media Services, Southwest Texas State
Gass, S. (1982). From theory to practice. In M. Hynes & W. Rutherford (Eds.), On TESOL ’81: Selected
papers from the 15th annual conference of Teachers of English to Speakers of Other Languages (pp.
129–139). Washington, DC: TESOL.
Graf, P. (1994). Explicit and implicit memory: A decade of research. Attention and Performance, 15,
Graf, P., & Mandler, G. (1984). Activation makes words more accessible but not necessarily more
retrievable. Journal of Verbal Learning and Verbal Behavior, 23, 553–568.
Graf, P., & Schacter, D. (1989). Unitization and grouping mediate dissociations in memory for new
associations. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 930–940.
Hyltenstam, K. (1984). The use of typological markedness conditions as predictors in second lan-
guage acquisition: The case of pronominal copies in relative clauses. In R. Andersen (Ed.), Sec-
ond languages: A cross-linguistic perspective (pp. 39–58). Rowley, MA: Newbury House.
Izumi, S. (2000). Promoting noticing and SLA: An empirical study of the effects of output and input en-
hancement on ESL relativization. Unpublished doctoral dissertation, Georgetown University,
Washington, DC.
Izumi, S., & Bigelow, M. (2000). Does output promote noticing and second language acquisition?
TESOL Quarterly, 34, 239–278.
Izumi, S., Bigelow, M., Fujiwara, M., & Fearnow, S. (1999). Testing the output hypothesis: Effects of output
on noticing and second language acquisition. Studies in Second Language Acquisition, 21, 421–452.
Jourdenais, R. (1998). The effects of textual enhancement on the acquisition of the Spanish preterit and
imperfect. Unpublished doctoral dissertation, Georgetown University, Washington, DC.
Jourdenais, R., Ota, M., Stauffer, S., Boyson, B., & Doughty, C. (1995). Does textual enhancement pro-
mote noticing? A think aloud protocol analysis. In R. Schmidt (Ed.), Attention and awareness in
foreign language learning (Tech. Rep. No. 9, pp. 183–216). Honolulu: University of Hawai‘i Press.
Keenan, E., & Comrie, B. (1977). Noun phrase accessibility and Universal Grammar. Linguistic Inquiry,
8, 63–99.
Kormos, J. (1999). Monitoring and self-repair in L2. Language Learning, 49, 303–342.
Kramer, S., & Rosenthal, R. (1999). Effect sizes and significance levels in small-sample research. In R.
Hoyle (Ed.), Statistical strategies for small sample research (pp. 59–79). Thousand Oaks, CA: Sage
Krashen, S. (1985). The input hypothesis: Issues and implications. London: Longman.
Larsen-Freeman, D., & Long, M. (1991). An introduction to second language acquisition research. Lon-
don: Longman.
Leow, R. (1997). The effects of input enhancement and text length on adult L2 readers’ comprehen-
sion and intake in second language acquisition. Applied Language Learning, 8, 151–182.
Levelt, W. (1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press.
Long, M. (1991). Focus on form: A design feature in language teaching methodology. In K. de Bot, R.
Ginsberg, & C. Kramsch (Eds.), Foreign language research in cross-cultural perspective (pp. 39
52). Amsterdam: Benjamins.
Long, M. (1996). The role of the linguistic environment in second language acquisition. In W. C. Rit-
chie & T. K. Bhatia (Eds.), Handbook of second language acquisition (pp. 413–468). San Diego,
CA: Academic Press.
Long, M., Inagaki, S., & Ortega, L. (1998). The role of implicit negative feedback in SLA: Models and
recasts in Japanese and Spanish. Modern Language Journal, 82, 357–371.
Long, M., & Robinson, P. (1998). Focus on form: Theory, research, and practice. In C. Doughty & J.
Williams (Eds.), Focus on form in classroom second language acquisition (pp. 15–41). New York:
Cambridge University Press.
MacWhinney, B. (1987). The competition model. In B. MacWhinney (Ed.), Mechanisms of language
acquisition (pp. 249–308). Mahwah, NJ: Erlbaum.
Mackey, A., & Philp, J. (1998). Conversational interaction and second language development: Recasts,
responses, and red herrings? Modern Language Journal, 82, 338–356.
576 Shinichi Izumi
Nelson, K. (1987). Some observations from the perspective of the rare event cognitive comparison
theory of language acquisition. In K. E. Nelson & A. van Kleek (Eds.), Children’s language (Vol. 6,
pp. 289–331). Mahwah, NJ: Erlbaum.
Norris, J., & Ortega, L. (2000). Effectiveness of L2 instruction: A research synthesis and quantitative
meta-analysis. Language Learning, 13, 417–528.
Pienemann, M. (1984). Psychological constraints on the teachability of languages. Studies in Second
Language Acquisition, 6, 186–214.
Pienemann, M. (1998). Language processing and second language development: Processability theory.
Amsterdam: Benjamins.
Robinson, P. (1995). Attention, memory, and the “noticing” hypothesis. Language Learning, 45, 283
Robinson, P. (1997). Generalizability and automaticity of second language learning under implicit,
incidental, enhanced, and instructed conditions. Studies in Second Language Acquisition, 19, 223–
Roth, F. (1984). Accelerating language learning in young children. Journal of Child Language, 11,89
Rutherford, W., & Sharwood Smith, M. (1985). Consciousness-raising and Universal Grammar. Applied
Linguistics, 6, 274–282.
Saxton, M. (1997a). The contrast theory of negative input. Journal of Child Language, 24, 139–161.
Saxton, M. (1997b). Negative evidence and negative feedback: Immediate effects on the grammaticality
of child speech. Unpublished manuscript, University of London.
Schachter, J. (1998). Recent research in language learning studies: Promises and problems. Language
Learning, 48, 557–583.
Schmidt, R. (1990). The role of consciousness in second language learning. Applied Linguistics, 11,
Schmidt, R. (1995). Consciousness and foreign language learning: A tutorial on the role of attention
and awareness in learning. In R. Schmidt (Ed.), Attention and awareness in foreign language learn-
ing (pp. 1–63). Honolulu: University of Hawai‘i Press.
Schmidt, R. (2001). Attention. In P. Robinson (Ed.), Cognition and second language instruction (pp.
3–32). New York: Cambridge University Press.
Schmidt, R., & Frota, S. (1986). Developing basic conversational ability in a second language: A case
study of an adult learner of Portuguese. In R. Day (Ed.), Talking to learn: Conversation in second
language acquisition (pp. 237–326). Rowley, MA: Newbury House.
Sharwood Smith, M. (1993). Input enhancement in instructed SLA: Theoretical bases. Studies in Sec-
ond Language Acquisition, 15, 165–179.
Shook, D. (1994). FL/L2 reading, grammatical information, and the input-to-intake phenomenon. Ap-
plied Language Learning, 5, 57–93.
Sorace, A. (1996). The use of acceptability judgments in second language acquisition research. In
W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of second language acquisition (pp. 375–405). San
Diego, CA: Academic Press.
Spada, N., & Lightbown, P. (1999). Instruction, first language influence, and developmental readiness
in second language acquisition. Modern Language Journal, 83, 1–22.
Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehen-
sible output in its development. In S. Gass & C. Madden (Eds.), Input in second language acquisi-
tion (pp. 235–253). Rowley, MA: Newbury House.
Swain, M. (1993). The output hypothesis: Just speaking and writing aren’t enough. The Canadian Mod-
ern Language Review, 50, 158–164.
Swain, M. (1995). Three functions of output in second language learning. In G. Cook & B. Seildlhofer
(Eds.), Principles and practice in applied linguistics: Studies in honour of H. G. Widdowson (pp.
125–144). Oxford: Oxford University Press.
Swain, M. (1998). Focus on form through conscious reflection. In C. Doughty & J. Williams (Eds.),
Focus on form in classroom second language acquisition (pp. 64–81). New York: Cambridge Uni-
versity Press.
Swain, M., & Lapkin, S. (1995). Problems in output and the cognitive processes they generate: A step
towards second language learning. Applied Linguistics, 16, 371–391.
Tavakolian, S. (1981). The conjoined-clause analysis of relative clauses. In S. Tavakolian (Ed.), Lan-
guage acquisition and linguistic theory (pp. 167–187). Cambridge, MA: MIT Press.
Thornbury, S. (1997). Reformulation and reconstruction: Tasks that promote “noticing.” ELT Journal,
5, 326–335.
Effects of Output and Input Enhancement 577
Tomlin, R., & Villa, V. (1994). Attention in cognitive science and second language acquisition. Studies
in Second Language Acquisition, 16, 183–203.
Trahey, M., & White, L. (1993). Positive evidence and preemption in the second language classroom.
Studies in Second Language Acquisition, 15, 181–203.
Truscott, J. (1998). Noticing in second language acquisition: A critical review. Second Language Re-
search, 14, 103–135.
VanPatten, B. (1996). Input processing and grammar instruction in second language acquisition. West-
port, CT: Ablex.
White, J. (1998). Getting the learners’ attention: A typographical input enhancement study. In C.
Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 85–
113). New York: Cambridge University Press.
Wickens, C. D. (1989). Attention and skilled performance. In D. H. Holding (Ed.), Human skills (pp.
71–105). New York: Wiley and Sons.
Williams, J. (1999). Memory, attention, and inductive learning. Studies in Second Language Acquisition,
21, 1–48.
... As seen, this type of noticing focuses much more on the relevance of output as L2 learners have to notice a hole to produce a form based upon this hole, thus triggering noticing in the input to look for this form. On this basis, also puts forward that a previous step to noticing a form may be noticing a hole, a claim which has been tested empirically by a range of studies (see Hanaoka, 2007Hanaoka, , 2012Izumi, 2002;Izumi, Bigelow, Fujiwara & Fearnow, 1999). Minimally noticing feedback is necessary before restructuring the interlanguage, and as reviewed by Leow (2020), if an L2 writer notices a mismatch between his output and a different language content in the WCF, stored in the STM, thus this restructuring is more likely to happen. ...
... Bearing this in mind, it is posited that comprehensible output "may be the trigger that forces the learner to pay attention to the means of expression needed" (Swain, 1985, p. 249) so that when difficulties arise, they are pushed into perfecting their output in such a way to make it more precise and coherent. Throughout time and subsequent revisions, Swain's hypothesis (1985Swain's hypothesis ( , 1993Swain's hypothesis ( , 1995 has experienced extensions, and its scope has been enlarged, leading to three specific functions of output positively benefitting SLA (Izumi et al., 1999), all of which will be next looked into greater detail. ...
... Through this, L2 learners can test how comprehensible and accurate their interlanguage is by comparing it with the feedback provided in the L2. As well, this way of regarding output is in close relation to comprehensible input since learners are somewhat forced to get involved in the process of negotiation of meaning, which gives rise to more accurate output (Izumi et al., 1999). ...
Full-text available
L2 writing has been recently one of the major areas of interest (Leki et al., 2008; Manchón & Matsuda, 2016), with much of the research focusing on the written product, the language learning affordances of written corrective feedback (WCF) (Bitchener & Storch, 2016), or L2 writing processes from an SLA perspective (Gánem-Gutiérrez & Gilmore, 2018). Research on WCF has been concerned with whether the provision or not of different types of feedback may affect the quality of the L2 text, as well as providing empirical evidence to the disciplinary debates on the effects of the provision and processing of WCF (Truscott, 1999; Ferris, 2004). Few studies have explored the role of models as WCF and how the resulting noticing of linguistic features may affect writing processes. Novel methodological procedures using keystroke-logging tools (Lindgren & Sullivan, 2019) have expanded ways to observe the writing process and pausological behavior unobtrusively. Our study intends to add new empirical evidence to: (1) the temporal distribution of writing processes in an underrepresented population (young learners), in both studies with writing processes, and in digital environments, and (2) the implementation of model texts as WCF in an attempt to observe how the management of writing processes vary in terms of lexical choices and the generation of ideas. We attempted to shed light on the extent to which the provision of model texts as WCF affect the management of young learners' writing processes and children's pausological behavior. We designed a study with 18 Primary school children (aged 10-11), through a three-stage feedback classroom-based experimental research with an experimental (n= 10) and a control (n= 8) group. Using Inputlog 8.0. (Leijten & Van Waes, 2013), children wrote their initial texts on the computer. The experimental group was subsequently provided with a model text, and the control group self-edited their texts without feedback. In the third stage, children were asked to rewrite their initial texts. In the analysis, we operationalized the writing process in terms of planning, formulation and revision. To analyze the pausological behavior, we used a wide range of measures as in previous research including time on task, pausing time, pause frequency, pause duration, pause distribution, and pause location. Our results revealed marked differences in the writing processes. In the case of planning, young learners receiving WCF made more frequent use of the planning process and increased the time spent on it in comparison with participants in the control group (who did not have access to feedback). This has indicated that children spend more time to planning than what resaerch with adult or high school EFL learners has revealed. As for the effects on formulation, the greatest impact was observed on the frequency of formulation episodes, which was larger in the group receiving WCF, and on the number of edits and words produced during formulation. The effect of the WCF was more clearly observed in the measure of frequency of revision episodes, and the experimental group increased the time spent on macroscopic revisions. All in all, our study has revealed that children strategically decided where and when to locate each of these writing processes. Regarding pausological behavior, our results revealed that the effect of WCF was more marked in pauses at word boundaries as well as sentence boundaries. WCF seems to mediate a great part of the aforementioned behavioural patterns. These findings were discussed from the perspective of i) enhancing our knowledge about young learners’ writing processes and pausological behaviour and thus to complement related previous theories, based on adult learners; ii) shedding light on the potential role of models as WCF to mediate such processes
... ). Effective as it has been shown, this kind of feedback does have several drawbacks. First, teachers' or peers' focus does not always match learners' actual focus, which might lead to unfulfilled expectations (e.g., Izumi et al., 1999;Long & Robinson, 1998). Second, teacher feedback is not always available as teachers have to undertake a heavy workload, especially in large classes (Lee, 2003). ...
Full-text available
Achieving a sufficient IELTS band score for academic purposes has been a major goal of many L2 learners around the world, especially those in Asia. However, IELTS writing scores were consistently reported to be the lowest when compared to the scores in speaking, reading, and listening. Despite a growing body of research in IELTS writing, little focused on the role of model essays and noticing hypotheses. The present study aimed to fill in this gap by examining whether or not the implementation of both noticing hypothesis and model essays had a discernible influence on learners’ IELTS task 2 writing. To reach this goal, a quasi-experimental design including a pretest and a posttest was conducted with the voluntary participation of 52 undergraduates. These participants were divided into two groups: control group (CG, n = 25), learning in the conventional method (peer feedback and teacher feedback), and experimental group (EG, n = 27), using the noticing-model essays method. Following this, semi-structured interviews were performed to gain insights into the quantitative data. The results from this mixed-methods approach showed that there were significant gains in the overall performance and in the lexical resources subscale in the EG while no considerable changes were observed in the CG. Additionally, the other subscales (task response, grammatical range and accuracy, and cohesion-coherence) did not witness any significant differences between the two groups. Several pedagogical implications and recommendations for future research, especially in the Asian context, were also discussed.
... On account of the research on the immersion learners, Merrill Swain argued that learners in immersion classes actually engage in little language production and comprehensible input is not sufficient for L2 learners. Therefore, the comprehensible output is required for L2 acquirers who prefer to be fluent and proficient in the target language (Izumi et al., 1999). ...
... The results from the studies related to the hypothesis-testing function and the metalinguistic function of output (Nobuyoshi & Ellis, 1993;Pica, 1988Pica, , 1992Pica, HoUiday, Lewis & Morgenthaler, 1989) show that learners often modify their output in response to the linguistic demands of NS signals of comprehension difficulty and also suggest that pushing learners to produce more comprehensible output may have a long-term effect. Izumi et al. (1999) designed a study as follows: in treatment phase 1, participants reconstructed a short passage after being exposed to it, followed by a second exposure to the same input material and a second reconstruction opportunity. In phase 2, participants wrote on given topics, followed by the presentation of a model given by a NS. ...
p>This study analyses the Undergraduate conversation class from two different perspectives: firstly as a social setting with its own characteristics. Secondly, the interactions between the Native-Speaker and the Non-Native speakers, occurring within this particular setting have been investigated. Even though the conversation class is normally a formal part of the language teaching curriculum in Modem Languages Departments in England, very httie is known about it as a setting, and about its usefulness in terms of linguistic development. To this author's knowledge, no study has been carried out to investigate how the conversation class works, and what its benefits might be for advanced learners. This study explores one main hypothesis concerning the potential hybridism of this setting, with features belonging to the casual conversation, and to the traditional language class. This hypothesis is linked to the claim in the literature that in order to become a proficient language user, learners need to engage in complex normal speech encounters. The conversation class could thus be one of the very few settings in a university context, where opportunities for speech-encounters, aiming to reflect language authenticity, arise. Chapter One presents the two theoretical frameworks used in the study: the Ethnography of Communication, and the Input and interaction theory. The first one reviews the literature on features of casual talk and of formal lessons, while the second framework focuses on aspects and patterns of NS-NNS interaction. Chapter Two introduces the study carried out, the data collection, and the procedures for data analysis. The study focuses on three groups of undergraduates in their first or second year of studies; twenty-five lessons have been recorded; five variables, (turn-taking, NNS- NNS interaction, the use of learners' LI, negotiation, and negative feedback) have been analysed in order to investigate the hypothesis outlined above. Chapter Three focuses on the results, and the analysis of these variables, where some evidence shows hybridism in the oral class conversation. Issues of equality between participants, and symmetry in interaction are discussed, leading to a discussion of potential linguistic development for the learners. Finally, Chapter Four revisits the research areas, and provides ideas for pedagogical implications of this research, such as ways to increase equality among the participants (e.g. lowering the status of the NS), and symmetry in the interactions (e.g. genuine information gaps), thanks to good practice found in the data. Suggestions for further research in oral settings are also being offered.</p
... At the grammatical level, Nobuyoshi and Eills carried out experimental studies on the English past tense , and show that by making requests for clarification, the teacher can motivate students to produce more precisely, which contributes to acquisition [17]. Izumi and Bigelow did an experiment about the reconstruction of a short passage to test the Output Hypothesis, they investigates the effectiveness of output in promoting attention and learning knowledge, and the required conditions of it in prompting language acquisition [18]. To study the noticing function of output in a more detailed way, Izumi conducted an experiment on the essay-writing tasks, and results suggest that extended opportunities to produce output and receive relevant input are crucial in improving learners' use of the grammatical structure [19]. ...
... Por tanto, estos alumnos tuvieron la ventaja de 31. Izumi et al. (1999) e Izumi & Bigelow (2000) son dos ejemplos de estudios que ponen a prueba la hipótesis del Output y su papel activo en la ASL, los cuales apoyan la noción de que «el output puede tener efectos beneficiosos en el desarrollo lingüístico sumado -y no opuestamente-al papel crucial del input» (Morgan-Short & Wood Bowden, 2006: 38). 32. ...
Full-text available
Mood selection is the one of the most difficult aspects of learning Spanish as a FL/L2, and it is considered to be one of the last features ever acquired, if at all, because of its complexity. In addition, present-day instruction has changed little over the years and it approaches mood with lists of verbs that trigger indicative or subjunctive in an ever-growing number of situations. New proposals that bridge the gap between linguistic theory and teaching in an actual classroom are needed. In this study, a cognitive grammar approach to teaching the relationship between mood selection and its meanings was used in a language course. It was aimed at helping learners understand mood in order to produce an unlimited number of utterances without having to learn each usage individually. The target structures were relative, temporal, and concessive clauses. All students had an intermediate level at the beginning of the study. For the practice part, the so-called processing instruction was adjusted to have students both interpret and produce target structures. Research with this methodology has shown that learners who are able to make form-meaning connections during online input processing considerably improve their learning over other types of instruction. The results show that the combination of cognitive grammar and processing instruction has excellent effects on how students identify mood selection to construe meanings in both input and output learning situations.
... The findings contrast with the study conducted by Izumi, Bigelow, Fujiwara, and fearnow (1999) which indicated that output production could not provide any significant effects on noticing of the form. The findings of the study are also in contrast with what Leow (2001) and Leow et al. (2003) found. ...
Full-text available
The idea of encouraging awareness in classrooms is not new, but research into awareness is beginning to encourage those involved in language teaching to think more systematically about how language presentation facilitate language awareness. Awareness can be promoted through focus on form activities as it triggers important cognitive processes in L2 acquisition. The effectiveness of various input-and output-based focus on form instructions on the acquisition of different grammatical structures and the role of awareness in each type is a matter for debate. The present study qualitatively investigated the effects of Processing instruction, Textual enhancement, and Text editing on L2 learners" cognitive processes and the relationship between the learners" level of awareness and their abilities to interpret English inversion structures. To do this, learners" think-aloud verbalizations during instruction were recorded, transcribed, and coded. Criteria to decide which level of awareness they would fall into were slightly adapted from Leow, Hsieh, and Moreno (2008) to fit with the type of tasks employed in the study. Pretest-posttests design was also employed to measure learners" interpretive abilities. The findings indicated that each instructional technique promoted different levels of awareness and depth of processing. The findings also provided explanations for the non-significant differences in performances between the Processing instruction and Text editing groups on an immediate posttest and outperformance of the Processing instruction group on a delayed posttest. Given the benefits that Processing instruction and Text editing brought about in the present study, both might be incorporated into a curriculum and serve as complementary tools for language teachers.
... Some of them (Alanen, 1995;Lee & Benati, 2009;Leow, 2001;Leow, Nuevo, & Tsai, 2003), for example, have attempted to measure the effect of textual input enhancement on noticing in vocabulary learning. The noticing function of Output Hypothesis (Swain, 1993(Swain, ,1995(Swain, , 1998 has also attracted little attention in SLA research and as Russell (2014) claimed ''thus far, only three studies have been found ) Izumi, 2002;Izumi & Bigelow, 2000 ;Izumi, Bigelow, Fujiwara, & Fearnow, 1999) that have attempted to test the noticing function of output'' (p. 28). ...
Full-text available
The present study is an attempt to examine the psychological processes of noticing, retrieval, and generation and their possible contribution to the process of vocabulary learning and retention among intermediate students. The research method was experimental. One hundred and twenty intermediate students were randomly assigned into four groups, namely Noticing through Input enhancement (n=30), Input Enhancement plus Input-Based Reviewing (n=30), Input Enhancement plus Output-Based Reviewing (n=30) and Input Enhancement plus Input-Based and Output-Based Reviewing. The Academic Words contextualized in Focus on Vocabulary 2: Mastering the Academic Word List (Schmitt, Schmitt, & Mann, 2011) were the target words of the study. A pretest composed of VLT items was administered to the participants. The first group encountered the target words that have been already highlighted to absorb their attention. Encountering the already highlighted words, the second group reviewed the words through researcher-made word cards. The third group, besides encountering the already highlighted words, reviewed the words through rewriting the sentences including the target words. The fourth group experienced noticing through input enhancement; retrieval through using researcher-made word cards; and generation through rewriting the sentences containing the unknown words. One week after the last treatment session, an immediate posttest, and after two weeks, a delayed posttest were administered. Based on the results of four one-way repeated measures ANOVAs and three one-way ANOVAs, it was revealed that all types of input-based, output-based and input+output-based reviewing have positive effect on vocabulary learning. However, their positive effect on vocabulary retention was fairly vague. Moreover, the group treated through input enhancement+input- and output- based reviewing outperformed the other groups.
This book presents in‐depth explorations of uncontrolled interactions for speaking practice in three English as a foreign language (EFL) classrooms in a university in Mexico, where learners study language and teaching modules in order to become language teachers. The book crosses the traditional methodological boundary associated with interactionist research which focuses on interactional patterns and presents an alternative approach which involves both interactional and perceptual evidence to explore interactions in EFL classrooms. Specifically, the explorations in this book draw attention to the role of teaching and learning ideologies in language learning outcomes, showing how teachers’ and learners’ diverse and sometimes conflicting beliefs shape the structure and nature of classroom interactions. In particular, these explorations address how teachers’ and learners’ interactional- and teaching and learning-related choices and beliefs are influential on three aspects of learner talk: language performance indicated by levels of fluency, complexity and accuracy, use of discourse functions, and negotiations of meaning. The book concludes that there is an interrelated set of cognitive, practical and interactional factors which shape classroom interactions and in turn learners’ language achievement. The contribution of this book is then threefold. Examining uncontrolled interactions in EFL classrooms makes an original contribution to the field of foreign language teaching, providing a research- and methodological- based account of the interrelated (cognitive, practical and interactional) factors that have an impact on learners’ linguistic, discoursive and interactional skills during classroom interactions. Secondly, by formulating some pedagogical implications, it provides a great opportunity to advance our understanding of how interactions in EFL classrooms can be enhanced in order to promote learners’ speaking skills. Thirdly, it suggests a detailed analytical framework (i.e., the Framework of Interactional Strategies in Foreign Language Interaction) which EFL teachers may find useful for exploring the effectiveness of their classroom interactions. This book is a valuable resource for anyone involved in the process of EFL teaching and learning (i.e., pre- and in-service teachers, teacher educators, and education administrators). Specifically, it would be useful for those who are experiencing difficulties in promoting the effectiveness of EFL interactions and learner achievement during speaking practice.
Full-text available
One of the theoretical rationales for focus on form is the notion that second language (L2) learning requires noticing of what is to be learned. Various techniques of input enhancement have been developed with the hope that they promote noticing of target forms. While their effectiveness has been evaluated in terms of subsequent acquisition of the target forms, few attempts have been made to investigate whether enhanced input is processed differently by learners. The purpose of this study was to determine whether one input enhancement technique — textual modification — can make L2 forms more noticeable and affect learner on-line processing of forms. In this study, native speakers of English in a second semester Spanish class at Georgetown University were assigned to enhancement and comparison groups. Participants in the enhancement group received a sample text in Spanish with all preterit and imperfect verb forms highlighted; participants in the comparison group received the same text with no typographical modification. Think-aloud protocols were collected during a subsequent task in which participants wrote a picture-based narrative similar to that presented in the sample text. Analysis of the data revealed that enhancement participants’ protocols contained more episodes related to selection and conjugation of preterit and imperfect verbs than did those of the comparison participants. Enhancement participants also produced more target features in their written production. The results indicate that textual enhancement promotes noticing of target L2 form and has an effect on learners’ subsequent output.
The influence of cognitive processing on second language acquisition (SLA), and on the development of second language (SL) instruction, has always been a subject of major interest to both SLA researchers and those involved in SL pedagogy. Recent theoretical research into SLA and SL pedagogy has shown renewed interest in the role of cognitive variables such as attention, short, working, and long term memory, and automaticity of language processing. This volume first examines the theoretical foundations of research into the cognitive processes underlying SLA, and then describes various implications for pedagogically oriented research and for SL classroom practice. The blend of research from the cognitive sciences and applied linguistics make it an excellent introduction to applied linguists and language teachers interested in the psycholinguistic processes underlying SLA.