Reformulation and

reconstruction: tasks that

promote ‘noticing’
Scott Thornbury

In various guises reformulation and reconstruction tasks have a long tra-

dition in ELT methodology. Since both task types foreground meaning,
they fit well into a task-based model of instruction, and because the
starting point in both cases is whole texts, their use is consistent with a
discourse-oriented view of language. However, their potential for focusing
learners’ attention on form (that is, noticing both what is present in input
and absent in output) has received little attention. This article rehabilitates
techniques that exploit both the meaning-driven and form-focused po-
tential of these two task types.

Noticing The role of noticing in second language acquisition has been the subject
of some attention recently (see for example Batstone 1996, Schmidt
1990). It has been suggested (Schmidt and Frota 1986) that two kinds of
noticing are necessary conditions for acquisition:
1 Learners must attend to linguistic features of the input that they are
exposed to, without which input cannot become ‘intake’.
2 Learners must ‘notice the gap’, i.e. make comparisons between the
current state of their developing linguistic system, as realized in their
output, and the target language system, available as input.
‘Matching’ is the term used by Klein (1986) for this second type of
noticing: ‘the learner must continuously compare his current language
variety with the target variety’ (1986: 62). Ellis (1995) prefers the term
‘cognitive comparison’, since this ‘better captures the fact that learners
need to notice when their own output is the same as the input as well as
when it is different’ (ibid.: 90). Noticing operations occupy a key role in
Ellis’s model of second language acquisition, facilitating the process
whereby explicit knowledge becomes implicit knowledge. In short: ‘No
noticing, no acquisition’ (ibid.: 89).
It follows that language teachers should try to promote noticing, by
focusing their learners’ attention on the targeted language in the input,
and on the distance to be covered between the present state of their
interlanguage, on the one hand, and the target language, on the other.
In the classroom, the first kind of noticing is customarily promoted
through activities and procedures involving input enhancement

326 ELT Journal Volume 51/4 October 1997 © Oxford University Press 1997

articles welcome
 (Sharwood Smith 1993), whereby targeted features of the input are
made salient in order to facilitate their becoming intake. The
presentation stage of the traditional Presentation-Practice-Production
(PPP) model of instruction is designed to do just that. However, the
effectiveness of this kind of approach has been called into question on a
number of grounds, not least because, given the current state of our
knowledge of acquisition order, this kind of pre-emptive strike on
targeted forms ‘could only be appropriate by chance’ (Allwright 1979:
170).
The second kind of noticing is traditionally mediated through corrective
feedback. Evidence suggests, however, that, when it comes to correction,
there is a considerable mismatch between teacher intentions and learner
outcomes, and that ‘the greatest error teachers make may be the
assumption that what occurs as ‘correction’ in classroom interaction
automatically leads to learning on the part of the student’ (Chaudron
1988: 152). Schmidt’s own experience of learning Portuguese suggested
that, in order to benefit from correction, he had to know he was being
corrected: implicit correction techniques such as clarification requests
made no impression, whereas hearing the correct version immediately
after making an error allowed him to match his present level with the
target (Schmidt and Frota 1986).
Given, then, the somewhat hit-and-miss nature of both presentation and
correction, what other kinds of activities and procedures might be
conducive to noticing and matching?
Tasks that provide opportunities for noticing are ones that, even if
essentially meaning-focused, allow the learner to devote some atten-
tional resources to form, and, moreover, provide both the data and the
incentive for the learner to make comparisons between interlanguage
output and target language models. Two generic classroom task types
that meet these criteria are, I will argue, reformulation tasks and
reconstruction tasks.
Opportunities for noticing alone are not enough, however, if the
learners lack the strategies to take advantage of them. Since noticing is a
conscious cognitive process, it is theoretically accessible to training and
development. This suggests that the teacher’s role is to develop noticing
strategies that the student can apply independently and autonomously.
This, too, is an issue I will address.

Reformulation Reformulation has gained currency in recent years as a technique in the

activities development of students’ writing skills: rather than simply correcting a
student’s composition, which usually involves attention to surface
features of the text only, the teacher reformulates it, using the content
the student has provided, but recasting it so that the rewritten draft
approximates as closely as possible to a putative target language model.
It is then available for comparison with the student’s own draft. (See
Hedge 1988 for a more elaborated description of this procedure.) The
technique has also been promoted for the teaching of speaking skills in
Tasks that promote noticing 327

articles welcome
 one-to-one classes (see Wilberg 1987) since it is an effective way of
tailoring instruction to the individual student’s needs and interests: ‘The
content is dictated by the student, the form only by the teacher, in
contrast to most language learning where the student has little or no
control over the content’ (ibid.: 27).
Reformulation, then, reverses the order of traditional models of
instruction, which move from accuracy to fluency, as, for example,
when learners are required to imitate model texts (as in a product
approach to writing) or to drill pre-selected structures for subsequent
use in ‘freer practice’ activities (as in the PPP model). Reformulation is
consistent with a fluency-to-accuracy, or task-based, model of instruc-
tion, that is, one that ‘encourages learners to make the best use of
whatever language they have. It assumes that learners will find ways of
encoding the meanings they have in order to achieve the outcome’
(Willis 1990: 128). Once encoded by the learner, these meanings are
then Ye-encoded (or reformulated) by the teacher.

Reformulation and Why should reformulation be conducive to noticing? Johnson (1988)

noticing drawing on skill acquisition theory, argues that exposing learners to the
target behaviour after the event - rather than providing a model
beforehand - has greater psychological validity, in that the learners
are predisposed to look out for (and notice) those features of the
modelled behaviour that they themselves had found problematic in the
initial trial run (or first draft). Moreover, it allows for learners at
different stages and with different needs to notice different language
features, in contrast to the convergent and exclusive focus of the
presentation (accuracy-to-fluency) model of instruction: ‘When refor-
mulation takes place, it may be that the most useful feedback comes
from those areas of mismatch which students are themselves able to
identify, because those areas will accord with the stage of their skill (or
interlanguage) development.’ (ibid.: 93)
In Johnson’s instructional model, reformulation follows an initial trial,
and is in turn followed by a re-trial, into which noticed features of the
reformulated behaviour may be incorporated in full operating condi-
tions: ‘Reformulation provides a model of what the behaviour should
look like; and though its clearest use is for writing, there is no reason
why spoken language should not be reformulated.’ (ibid.: 92)

Community A methodology, one of whose core pedagogic principles is the

language reformulation of spoken language, is Community Language Learning
Learning (CLL), sometimes called Counselling-Learning (Curran 1976). In its
original conception, CLL was predicated heavily on principles of
humanistic counselling, and its borrowing of some of the procedures
and jargon of group therapy has given it an undeserved fringe status.
However, it is relatively easy to bypass the psychotherapeutic features of
CLL and to uncover a methodology that is not only reformulative, but
also incorporates explicit form-focused, noticing-type procedures.
328 Scott Thornbury

articles welcome
 In orthodox CLL the learners, seated in a closed circle around a tape
recorder, jointly construct and record a conversation, line by line, on a
topic or topics of their choosing. The teacher, outside the circle, is
available on request simply to model what the learners want to say,
either by translating from their Ll, or by reformulating, where
necessary, their flawed L2 trial runs. In this way, the content is provided
by the learners, the form by the teacher. Since learners are constrained
neither by a pre-selected topic nor a pre-selected linguistic agenda, the
activity is inherently heterogeneous, in Ur’s (1988) sense that all
students, irrespective of level, may participate.
Once a suitable body of text has been recorded, it is then replayed in its
entirety, thereby giving the learners a sense of its overall cohesion. It is
then transcribed line by line onto the board (or a transparency) by the
teacher, who subjects it to some form of linguistic scrutiny, e.g. by
pointing out features of the language that he or she feels the students are
ready for. (For a fuller treatment of CLL see Stevick 1980, Bolitho 1982,
and Larsen-Freeman 1986.)

CLL and noticing The procedure thus allows for noticing at least two stages:
Firstly, incidental noticing may occur during the initial ‘conversation’,
when the teacher reformulates individual learner utterances and the
learner commits these to the tape (as in writing activities, the formal
requirement of recording the utterance, plus the time spent rehearsing
the utterance prior to recording, will require the learner to allocate more
attention to form than if the conversation had been unrecorded and
unrehearsed (see Skehan 1996)). Secondly, focused noticing is the aim of
the final stage, when the teacher explicitly directs learners’ attention to
language items in the transcript. Of course, this stage need not be
teacher-led: the learners can be encouraged to ask questions themselves
about features of the transcript, consistent with the principle that what
the learners choose to notice is more likely to become intake than what
the teacher chooses to highlight (Slimani 1989). However, if the teaching
programme is grammar oriented, it requires only a little ingenuity to
structure the conversation in such a way that targeted grammatical
structures are likely to occur, and are therefore available for subsequent
analysis. For example, to elicit exponents of futurity, the teacher might
introduce the topic: ‘the next summer holidays’. Likewise, when
reformulating students’ utterances, the teacher can choose either to
introduce or to avoid certain language items, according to their
appropriacy in the course programme.
However, from the point of view of noticing, there is one possible
weakness in the CLL procedure. Since learners do not have on record
their original, unreformulated utterances, there is no opportunity to
compare ‘before’ and ‘after’ stages, to notice the mismatch, as there is in
the reformulation of students’ writing drafts, for example. However, in
Bolitho’s (1982) variation, learners can choose whether or not to have
their utterances reformulated during the recording stage. This means
Tasks that promote noticing 329

articles welcome
 that the recording includes more ‘first draft’ errors than if each utterance
had been painstakingly reformulated and rehearsed. The transcribed
conversation, errors included, is subject to reflection, analysis, and
reformulation, and the final ‘draft’ is then available for comparison with
the original. Nor need the analysis be explicit, as Johnson (1988: 93)
points out: ‘One of the benefits of reformulation is that if, without
comment, one merely presents students with a model performance to be
compared with their flawed performance, it is left up to them to note and
learn what they will from the comparison’.
And, finally, on the subject of reformulation, Stevick (1989), concluding
his case studies of seven successful language learners, describes one of
his own preferred strategies:
Another of my favourite techniques is to tell something to a speaker
of the language and have that person tell the same thing back to me in
correct, natural form. I then tell the same thing again, bearing in mind
the way in which I have just heard it [i.e. having noticed the gap]. This
cycle can repeat itself two or three times ... An essential feature of this
technique is that the text we are swapping back and forth originates
with me, so that I control the content and do not have to worry about
generating non-verbal images to match what is in someone else’s mind
(Stevick 1989: 148).

Reconstruction Unlike reformulation activities, in which the learner’s text is reformu-

activities lated by the teacher, the starting point for reconstruction activities is the
teacher’s text (or, at least, a text provided by the teacher) which the
learner first reads (or listens to) and then reconstructs. The recon-
structed version is then available for ‘matching’ with the original.
In reconstructing a text, learners will deploy their available linguistic
competence, which (depending, of course, on the choice of text) is likely
to fall short of the target model. This process alone, it is argued (e.g. by
Marton 1988), in forcing attention on form, activates bottom-up
processes that, in comprehension, and communicative activities, are
not necessarily engaged. Moreover, the extra effort involved may in
itself trigger noticing. ‘The activity of producing the target language may
prompt second language learners to consciously recognize some of their
linguistic problems; it may bring to their attention something they need
to discover about their L2’ (Swain and Lapkin 1995: 373).
But the real benefit may be in the matching: the comparison by learners
of their version with the model provides them with positive evidence of
yet-to-be-acquired language features, and this process of noticing,
theoretically at least, converts input to intake, and serves to restructure
the learner’s developing linguistic competence.
Many tried-and-true classroom activities fall within this generic type. For
example, copying (an innovative use for copying for cognitive
comparison purposes is described by Porte 1995); memorization and
recitation of texts; dictation; ‘rhetorical transformation’ (see Widdowson
330 Scott Thornbury

articles welcome
 1978: 146); translation and re-translation (see, for example, Edge 1986);
and Storyboard-type computer games (see Brett 1994 for a discussion of
these games). Linguistic heterogeneity may be a priority in all of these
activities, and this can be achieved simply through the use of authentic
texts. On the other hand, if need be, the dice can be linguistically loaded
to meet the needs of a traditional, form-driven syllabus: a targeted item
can be deliberately embedded in the text, on the assumption that the
learners are ready for it.

Dictogloss A reconstruction activity that has been popularized recently is known

variously as dictogloss, ditto-camp (dictation/composition), or grammar
dictation (Wajnryb 1990). The basic procedure consists in the learners
simply listening to a short text once, maybe twice, in its entirety, and
reconstructing it from memory, either individually or in pairs or groups.
The reconstructed text is then compared with the original, a distinction
being made between differences that are acceptable - that is, where the
propositional content is the same, albeit differently realized - or
unacceptable: where either the propositional content is different, or,
as is more often the case, the realization is linguistically flawed.

Like dictation, dictogloss involves both top-down and bottom-up

processing: whereas the input might be decoded through attention to
content words and activated background knowledge, the encoding stage
will require lower level syntactic processing, where the remembered
lexis is ‘grammaticised’ (Rutherford 1987), as the following sequence
demonstrates.

Students were told they were going to hear a short text once, that they
were not to write while listening, and that, immediately afterwards, they
had to try and reconstruct the text from memory. This was the original
text, which was read aloud at natural speed:
There was a young woman of Riga
Who went for a ride on a tiger
They returned from the ride
With the woman inside
And a smile on the face of the tiger
This is how one (mid-intermediate) student reconstructed the text,
working alone:
There was a woman from Riga who were on a tiger to make a ride.
When they come back the woman was in tiger and the tiger were
smiling.
Working with two other students they then came up with this version:
There was a young woman from Riga who go for ride on a tiger. The
tiger come back with the woman inside and a smile on the tiger.
Finally, the class (of nine students) working together came up with this
version:

Tasks that promote noticing 331

articles welcome
 There was a young woman of Riga who went for a ride on a tiger. The
tiger returned with the woman inside and a smile on its face.
This steady improvement in accuracy during each successive draft
confirms Wajnryb’s (1990: 12) contention that ‘in the reconstruction
stage, specifically in the group effort to create a text, learners expand
their understanding of what options exist and are available to them in
the language’.
The noticing stage is equally important, either as a confirmation that
input and output are matched (as they almost are in the case of the
limerick example above) or in order to highlight mismatches, which, if
the theory is correct, may then trigger restructuring. There is an element
of ‘down the garden path’ learning in dictogloss (Tomasello and Herron
1988): texts can be chosen that contain examples of an item known to be
unfamiliar to learners so that errors of omission are virtually guaranteed.
The cognitive comparison stage forces attention on these errors: ‘In
learnability terms they [the learners] are made to generate their own
negative evidence with a little help from the teacher’ (Sharwood Smith
1993: 24). Another advantage of the dictogloss technique is its built-in
heterogeneity: different learners, depending on the state of development
of their interlanguage, as well as their interest and motivation, will notice
different things.
Adaptations of this technique include exposing students to the text in its
written, rather than spoken, form, e.g. by means of an overhead
projector, before they attempt to reconstruct it. Similarly, text
reconstruction software often provides the option of seeing the text
first, as well as allowing the user to call up the initial letters of words
(Brett 1994). The ‘partial dictation’ technique, of providing some of the
text (e.g. content words) eases the memory load, thereby allowing more
attention to be allocated to syntactic processing: either students read or
hear the complete text first or they attempt to reconstruct it unseen,
using only the existing prompts. As in all reconstruction tasks, they then
compare their version with the original.

Training ‘You can lead a horse to water but you can’t make it drink.’ What if
‘noticing’ learners do not, in fact, notice items in the input, or do not notice items
missing from their output? While O’Malley and Chamot, in their review
of learning strategies (1990), do not specifically identify as strategies
either noticing or noticing the gap, the metacognitive categories they
label as ‘selective attention’ and ‘self-evaluation’ seem, respectively, to
share some of the characteristics of the two kinds of noticing.
Selective attention involves ‘attending to specific aspects of language
input during task execution’ (ibid.: 137), while self-evaluation includes
‘checking the outcomes of one’s own language performance against an
internal measure of completeness and accuracy’ (ibid.: 137). Substitute
‘external’ for ‘internal’ in this definition, and it is consistent with
definitions of matching. (In fact, it is arguably easier, and more easily
verifiable, to check one’s output against an external model than against
332 Scott Thornbury

articles welcome
an internal one). If, as O’Malley and Chamot claim, learning strategies
are accessible for development and can be ‘used to assist learning
instead of being relegated to the uncertainty of unconscious mechan-
isms’ (ibid.: 91), then it follows that the two kinds of noticing are also
amenable to training.
Below is a short and by no means exhaustive list of possible awareness-
raising activities targeted at noticing strategies, with special reference to
reformulation and reconstruction tasks.
l Provide opportunities in class for silent study and reflection. Noticing
requires the marshalling of attentional resources, which, in turn,
requires time and the absence of distractions.
l Introduce the term ‘noticing’ into classroom metalanguage: for
example, ‘What differences do you notice?‘, ‘Did you notice...?‘, and
‘This is a noticing activity’.
l Exploit vocabulary acquisition experiences to introduce the value of
noticing: having once learnt a new word, many students may be
familiar with the experience of encountering it with surprising
frequency in different naturalistic contexts.
l Develop selective attention by, for example, playing tapes and asking
students to count the number of instances of a particular word or
structure.
l Develop text-scanning skills, e.g. spot the difference between two
similar texts, or between a recorded and a written text.
l Develop proof-reading skills, e.g. ask students to mark the
differences between first and reformulated drafts (by underlining
or circling),or between an original text and its reconstructed version;
they then exchange their work to double-check that no differences
were missed.
l Ask students to report on the differences between drafts, using fully
formulated sentences, e.g. ‘In the original text they used ‘would’ but I
used the past simple...‘.
l Repeat reconstruction activities (e.g. dictation) after an interval of
time, and then ask students to compare the number of differences
between first and second attempts.
l Supply students with dictionaries and student grammars so that they
can research the differences between their versions and the original
(in the case of reconstruction), or their version and its reformulation,
and explain these differences to the class. They could be asked to
classify the differences according to whether they are mistakes or
simply different ways of saying the same thing.
l Suggest to students that they keep a list of significant differences they
have noticed as a result, for example, of doing reconstruction activities
such as dictogloss.
Tasks that promote noticing 333

articles welcome
 Conclusion I am not necessarily advocating the exclusive use in the language
classroom of reformulation and/or reconstruction activities. My claim is
that these kinds of activities, because of their built-in noticing potential,
can be integrated into existing approaches and successfully deployed for
consciousness-raising purposes in conjunction with the development of
the appropriate cognitive strategies. A further attraction is that, since
the objective of tasks involving reformulation and reconstruction is the
production of whole texts or stretches of discourse, they allow for
consciousness-raising at a whole range of levels: discoursal, syntactic,
lexical, and phonological. Moreover, by the manipulation of task design
or the choice of text, they can be harnessed to the needs of an essentially
grammar-driven programme. Alternatively, they can form a sub-set of
task types within a meaning-driven, task-based syllabus.

Received February 1996

References Marton, W. 1988. Methods in English Language

Allwright, R. 1979. ‘Language learning through Teaching: Frameworks and Options. Hemel
communication practice’ in C. Brumfit and K. Hempstead: Prentice Hall.
Johnson (eds.). The Communicative Approach O’Malley, J. and A. Chamot. 1990. Learning
to Language Teaching. Oxford: Oxford Uni- Strategies in Second Language Acquisition.
versity Press. Cambridge: Cambridge University Press.
Batstone, R. 1996. ‘Noticing’. ELT Journal 50/3: Porte, G. 1995. ‘Writing wrongs: copying as a
273.
strategy for underachieving EFL writers’. ELT
Bolitho, R. 1982. ‘CLL: A way forward?’ in P.
Journal 49/2: 144-51.
Early (ed.). Humanistic approaches: An empiri-
Rutherford, W. E. 1987. Second Language Gram-
cal view. ELT Documents 113. London: The
British Council. mar: Learning and Teaching. Harlow: Longman.
Brett, P. 1994. ‘Using text reconstruction soft- Schmidt, R. 1990. ‘The role of consciousness in
ware’. ELT Journal 48/4: 329-36. second language learning’. Applied Linguistics
Chaudron, C. 1988. Second Language Classrooms: 11/2: 129-58.
Research on Teaching and Learning. Cam- Schmidt, R. and S. Frota. 1986. ‘Developing basic
bridge: Cambridge University Press. conversational ability in a foreign language: A
Curran, C. 1976. Counseling-Learning in Second case study of an adult learner of Portuguese’, in
Languages. Apple River, Ill.: Apple River R. Day (ed.). Talking to Learn. Rowley, MA:
Press. Newbury House.
Edge, J. 1986. ‘Acquisition disappears in adul- Sharwood Smith, M. 1993. ‘Input enhancement in
tery’: interaction in the translation class’. ELT instructed SLA’. Studies in Second Language
Journal 40/2: 121-4.
Acquisition 15:165-79.
Ellis, R. 1995. ‘Interpretation tasks for grammar Skehan, P. 1996. ‘A framework for the imple-
teaching’. TESOL Quarterly 29/1: 87-106.
mentation of task-based instruction’. Applied
Hedge, T. 1988. Writing. Oxford: Oxford Uni-
Linguistics 17/1: 38-62.
versity Press.
Slimani, A. 1989. ‘The role of topicalization in
Johnson, K. 1988. ‘Mistake correction’. ELT
Journal 42/2: 89-96. classroom language learning’. System 17:
Klein, W. 1986. Second Language Acquisition. 223-34.
Cambridge: Cambridge University Press. Stevick, E. 1980. Teaching Languages: A Way and
Larsen-Freeman, D. 1986. Techniques and Princi- Ways. Rowley, Mass: Newbury House.
ples in Language Teaching. Oxford: Oxford Stevick, E. 1989. Success with Foreign Languages.
University Press. Hemel Hempstead: Prentice Hall.

334 Scott Thornbury

articles welcome
Swain, M. and S. Lapkin. 1995. ‘Problems in Wilberg, T. 1987. One-to-One. Hove: Language
output and the cognitive processes they gen- Teaching Publications.
erate: a step towards second language learning’. Willis, D. 1990. The Lexical Syllabus. London:
Applied Linguistics 16/3: 371-91. Collins.
Tomasello, M. and C. Herron. 1988. ‘Down the
garden path: inducing and correcting overge-
The author
neralization errors in the foreign language
classroom’. Applied Psycholinguistics 9: 237-46. Scott Thornbury directs the teacher training
Ur, P. 1988. Grammar Practice Activities. Cam- department at International House, Barcelona.
bridge: Cambridge University Press. He has an MA TEFL from the University of
Wajnryb, R. 1990. Grammar Dictation. Oxford: Reading. He has written ELT materials for
Oxford University Press. Longman and Heinemann, and has recently
Widdowson, H. 1978. Teaching Language as published a book of language analysis tasks,
Communication. Oxford: Oxford University About Language (Cambridge University Press
Press. 1997).

Tasks that promote noticing 335

articles welcome