Professional Documents
Culture Documents
Feedback
JOHN BITCHENER
Framing the Issue
This entry discusses the potential of direct, indirect, and metalinguistic written
corrective feedback (CF) to facilitate second language (L2) development.
Written CF refers to the feedback that teachers (and sometimes other native
speakers of the target language) give second language learners on the accuracy
of the target language used in their written texts. In some of the early literature,
written CF has also been referred to as error/grammar correction. Over the past
40 years, our understanding of the purpose of providing written CF has
evolved: in the early literature, the primary purpose was to provide feedback
so that L2 learners could edit their own texts. This focus characterized many of
the early revision written CF studies. In more recent years, the interface
between L2 writing and L2 acquisition has generated a body of theoretical and
research literature about the potential of written CF to facilitate L2 develop
ment/learning/acquisition. In the literature, the terms L2 development, learning,
and acquisition have sometimes been used interchangeably even though each
has its own defining characteristics. In this entry, the term L2 development is
used because it refers to the developmental stages in the learning or acquisition
of an L2 from the point when written CF is provided to the point when a learner
demonstrates through a modified or accurate use of the L2 that the develop
ment process has begun.
Written CF can be provided to L2 learners in several ways. Direct feedback or
direct error correction refers to some form of explicit correction of linguistic form
or structure above or near the linguistic error. It may involve the crossing out of an
unnecessary word, phrase, or morpheme, the insertion of a missing word, phrase,
or morpheme, or the provision of the correct form or structure. Indirect feedback
or indirect written CF refers to that which indicates an error has been made but it
does not provide a correction. It may take the form of underlining or circling an
error (thereby indicating and locating the error) or recording in the margin the
number of errors in a given line (thereby indicating only that an error has been
made somewhere in the line of text). In recent years, metalinguistic explanation
has been discussed as a separate category of written CF. It may include grammatical
explanations/rules as well as examples of accurate L2 use and may be provided by
assigning each error a number and, at the bottom of the page of text or at the end
of the full text, presenting the metalinguistic explanation(s)/example(s) beside the
relevant number assigned to the error category in the learner’s text. Metalinguistic
explanation can also be provided in the form of an error code (e.g., PT for Past
Tense).
It can be seen, then, that each of these types of written CF occupies a specific
place on the explicit spectrum. At the least explicit end of the spectrum, we would
expect to find indirect written CF options. Arguably, at the most explicit end of the
spectrum we would expect to find either direct error correction and/or meta
linguistic explanation. If metalinguistic explanation is provided, it is most often
accompanied by direct error correction, so this combination would be considered
the most explicit form of written CF. Given these different approaches to provid
ing written CF, it is not difficult to understand why teachers, as well as researchers
and theorists, have been interested in finding out whether or not one type of
written CF is more effective than another for L2 development.
Making the Case
To understand the extent to which different types of written CF have the poten
tial to facilitate L2 development, it is important to (1) consider whether or not a
theoretical case can be made to support the assumption that written CF in gen
eral as well as specific types of written CF can play this role, and (2) examine
what the research evidence has revealed. Prior to Truscott’s (1996) call for the
abandonment of written CF, teachers had generally assumed to some extent at
least that it was a worthwhile practice and that L2 learners would develop their
L2 knowledge as a result. However, Truscott’s theoretical, empirical, and peda
gogical claims against the practice highlighted the absence of a sufficiently
developed theoretical case and of research evidence for or against the effective
ness of the practice.
Perhaps the most complete theoretical case in support of the practice can be
found in Bitchener and Storch (in press), a case which is situated within SLA
perspectives on the cognitive processing of explicit information (provided in the
form of written CF). The central goal of L2 learning is the acquisition of native or
near-native speaker competence, but as Krashen (1985) explains, acquired compe
tence is different from learned competence. It is generally agreed that acquired
competence draws upon implicit, unconscious L2 knowledge, but it is more con
troversial whether learned competence, resulting from a conscious development
of explicit L2 (afforded by explicit written CF), can be converted to implicit
knowledge. If it can be converted, as many “interface” theorists suggest, it can be
hypothesized to play a role. On the other hand, non-interface theorists maintain
that conversion is not possible. Skill acquisition theories are useful for explaining
how explicit L2 knowledge can be proceduralized through meaningful, contextu
alized practice over time (that is, through the process of consolidating new L2
knowledge gained from written CF) to a point where it is converted to implicit,
acquired knowledge.
Various SLA theorists explain the specific cognitive, information-processing
stages that need to occur if written CF is to help learners progress from the
point where they receive the feedback to the point where they are able to mod
ify accurately the grammatical error that it has targeted and produce new,
accurate output in new pieces of writing. They explain that the learner must (1)
attend to the written CF input, (2) notice with awareness and understanding the
difference between their inaccurate output and what the feedback is saying, (3)
internalize and (4) integrate the new knowledge in order to produce accurate L2
output. If new accurate output is produced, it is likely that the learning process
has been successfully initiated. Ongoing opportunities for using the same
grammatical form or structure are needed if the learner is to consolidate the
knowledge and work toward automatizing that use as native speakers do. It
has been hypothesized that this computerized model of the development pro
cess can be interrupted by the presence of individual and contextual factors
and that these are likely to also interact in different ways with the type of writ
ten CF a learner receives. Before we consider the potential effect of different
types of written CF on the development process, it is important to examine the
empirical research that has investigated whether or not these theoretical claims
are valid.
To determine empirically whether or not written CF can help learners pro
duce modified and new, accurate uses of the L2 on a range of occasions after
the first accurate use has been identified, a pre-test, treatment, post-test research
design has most often been employed. Once a pre-test piece of writing has
identified the grammatical error to be targeted, written CF (the treatment) is
provided on the error categories. An immediate post-test piece of writing is
then examined to see if the feedback has enabled the learners to produce accu
rate uses of the targeted grammatical forms or structures. Any number of post-
test pieces of writing can then be examined over time to see whether the initial
level of improved accuracy in the immediate post-test has been retained and
therefore whether or not new knowledge is in the process of being consoli
dated. To determine whether or not any improved accuracy is, in fact, the result
of written CF, the written texts of learners who do not receive written CF are
examined, and if there is a significant difference between those who received
the feedback and those who did not, written CF is considered to have been
effective.
The research (Sheen, 2007; Bitchener, 2008; Bitchener & Knoch, 2008; Ellis, Sheen,
Murakami, & Takashima, 2008; Sheen, Wright, & Moldawa, 2009; Bitchener &
Koch, 2010a, 2010b; Frear, 2012; Shintani & Ellis, 2013; Rummel, 2014; Shintani,
Ellis, & Suzuki, 2014; Stefanou, 2014; Qi, 2015) has found consistently that written
CF on certain targeted grammatical forms/structures at least is effective. However,
the researchers are not claiming that written CF will always be effective. Only a
few grammatical categories of error have been investigated: functional uses of the
English article system, simple past tense and present perfect, prepositions,
and hypothetical conditional. Whether or not more complex forms/structures and
more idiosyncratic, item-based forms/structures (i.e., those that are not governed
by rules) are equally “treatable” with written CF is not yet known. Nevertheless, it
can be claimed from this evidence that written CF has the potential to facilitate L2
development. Only one study (Bitchener & Knoch, 2010a) has examined its effec
tiveness over an extended period (10 months), so more longitudinal studies are
needed to confirm (1) its effect on consolidating what is being learned, and (2) the
extent to which learners are able to use the L2 knowledge automatically (without
conscious attention).
L2 teachers have always been interested in finding out whether or not certain
types of written CF may be more effective than others. As a result, this question
has dominated the written CF research agenda for more than 40 years. From a
theoretical point of view, it is an important question to ask because if noticing with
understanding is a condition of L2 development, it might be argued that certain
types of written CF (e.g., more explicit types) will provide learners with better
understanding of why their errors are not target-like and of how they can be cor
rected. Those who believe that the less explicit, indirect written CF types (e.g.,
underlining) are more effective suggest that these types invite the learner to
“engage in guided learning and problem solving” (Lalande, 1982), that is, they
invite the learner to do the work and determine what needs to be modified when
an error has been made. To benefit from this type of written CF, learners need to
draw on their existing knowledge of the linguistic form or structure in question
(stored in their long-term memory), consider it together with the new written CF
input in their working memory, and form a hypothesis about what constitutes
accurate L2 output. If this is the case, it would seem that indirect written CF may
be more beneficial for learners with partially acquired knowledge, whereas for
learners without that level of knowledge it would seem that more explicit written
CF in the form of either direct error correction or metalinguistic explanation, or
both, would be more instructive.
Those more in favor of direct error correction suggest that it can (1) reduce the
type of confusion that learners may experience if they do not understand how to
respond to indirect types of written CF, (2) provide them with information to
help them resolve more complex errors, (3) offer more explicit feedback on
hypotheses that learners may have tested, and (4) be more immediate. While this
type of feedback may be more helpful for lower-proficiency learners with less
developed knowledge in their long-term memories, it may not always be the
most beneficial for more advanced learners who, to consolidate and automatize
their use of L2 forms/structures, need to be challenged to see if they can draw
upon their partially developed knowledge to formulate a hypothesis about what
may be a correct form or structure. Because the provision of direct error correc
tion alone does not inform a learner about why an error has occurred and what
is needed to correct the error, written CF in the form of metalinguistic
information may be more instructive for learners who have little or no partially
stored knowledge to draw upon. Considering these arguments, it would seem
that some learners might be more likely to develop their L2 knowledge better if
they are given more explicit types of written CF while others might benefit more
from less explicit types.
Studies that have investigated whether or not certain types of written CF are
more effective than others in contributing to the development of L2 knowledge
have tended to compare (1) direct error correction and less explicit types, (2)
direct error correction and direct error correction combined with more explicit
types, and (3) metalinguistic explanation and other types of written CF. The
findings of the first group of studies (Lalande, 1982; Semke, 1984; Chandler,
2003; Van Beuningen, De Jong, & Kuiken, 2008, 2012) are inconclusive. The first
three have been critiqued for design and/or execution flaws so need to be read
with caution. Although the fourth study (pilot and main study) reported an
advantage for direct error correction, the findings of one study are not sufficient
evidence from which to draw conclusions. In the second group of studies
(Bitchener, Young, & Cameron, 2005; Sheen, 2007; Bitchener, 2008; Bitchener &
Knoch, 2008, 2010a; Stefanou, 2014), the findings are again inconclusive. Some
of these studies report no difference between feedback types, while others
report an advantage for direct error correction combined with other more
explicit types of written CF. These differences are more the result of intervening
variables across the studies than design or execution flaws. Thus they under
score the need to hold variables constant if further research is to find out whether
a particular variable, including a particular type of written CF, is more effective
than another. In the third group of studies (Bitchener & Knoch, 2010b; Shintani
& Ellis, 2013; Rummel, 2014; Shintani et al., 2014; Qi, 2015), the findings are
again inconclusive, for the same reasons as those assigned to the second group
of studies. In the third group, it was particularly noticeable, when metalinguis
tic information was provided with one or more additional types of written CF,
that the combinations of types may have prevented the individual effect of one
variable within the combination being revealed. In considering whether or not
one type of written CF might be more effective than another or others, the jury
is out until the issues identified here are taken into consideration by future
researchers.
Additionally, individual differences (both cognitive and affective/motivational)
may also have a mediating effect on the type of written CF that is most helpful for
an individual’s L2 development. For example, learners who have a strong feeling
about which type of feedback is most helpful for them may respond negatively if
different types of written CF are given. For example, they may decide not to attend
to the feedback at all. Recent theoretical thinking and empirical evidence have also
shown that individual cognitive factors such as language analytic ability and
grammatical sensitivity, as well as individual contextual factors such as learners’
beliefs about written CF, may also interact with different types of written CF and
mediate its effectiveness. Given that the available research evidence on different
types of written CF has not revealed findings that can be generally applied, it is
not clear what advice can be given to teachers when making decisions about g
iving
written CF.
Pedagogical Implications
SEE ALSO: Correcting Errors; Error Analysis; Explicit Versus Implicit Grammar
Knowledge; Grammar and Second Language Writing; Written Versus Spoken
Grammar
References
Suggested Readings
Bitchener, J., & Ferris, D. (2011). Written corrective feedback in second language acquisition and
writing. London, England: Routledge.
Journal of Second Language Writing, (2012). Exploring L2 Writing–SLA Interfaces. Special
Issue Papers. Journal of Second Language Writing, 21(4), 321–436.