Learner mining of pre-task and

task input
Jeremy Scott Boston

The findings reported in this article suggest that learners inevitably ‘mine’
wordings contained in pre-task and task materials when performing tasks, even
when the teacher did not explicitly draw learner attention to these features.
However, this was found to be true only with written materials, and learners
did not appear to mine specific wordings from audio pre-task materials. Learner
mining of language input from written pre-task/task materials opens the
possibility of deliberately embedding specific language items into such materials.
While the article acknowledges that such an approach to task design is
controversial, with some arguing that it is inconsistent with principles behind
task-based learning, the article leaves it to the reader to decide if finding ways
of harnessing learner mining of wordings from pre-task input is in harmony
with the principles of a task-based approach.

A suggested task- A long-established methodology in ELT is to begin lessons with an

based lesson explicit focus on pre-selected wordings and language targets presented
sequence directly by the teacher or via student analysis of a written or audio text.
This focus on language items is eventually followed by performance of
spoken activities where learners are expected to use the language taught
freely and ‘communicatively’.
The more recent task-based learning (TBL) approach places the (initial)
focus of the lesson on the task rather than on the language. A general
definition of ‘task’ is:
A task is an activity which requires learners to use language, with
emphasis on meaning, to obtain an objective.
(Bygate, Skehan, and Swain 2001: 11)
Tasks can simply be used to stimulate the use of language in some general
performance area, such as fluency, or an aspect of communicative
competence. However, tasks can also prepare the ground for activities
that draw learner attention to the link between meanings and wordings
(Bygate, Skehan, and Swain: ibid.). This latter use of tasks rests on the
principle that:
In natural S LA circumstances, learners begin by wanting to mean . . . and
then go on to seek or notice wordings that express those meanings . . . It
follows that materials we offer learners should allow them to focus

66 E LT Journal Volume 62/1 January 2008; doi:10.1093/elt/ccm079

ª The Author 2008. Published by Oxford University Press; all rights reserved.
 first on meanings in context and then go on to look at the wordings that
realize the meanings.
(Willis 2000: 7)
In Willis’s (1996) task-based model, lessons would begin with a spoken task,
for example, ‘Find three things you did last week that your [task] partner(s)
did not’. Learners are successful if they can achieve the outcome (for
example, find three differences) regardless of the accuracy of the wordings
they chose to employ. A focus on wordings then follows the task where
learners analyse recordings and transcripts of fluent target-language
speakers performing similar tasks. Tasks intend to set up learners to
notice features from the lesson’s post-task language focus stage, because
they had a need for such features while attempting the task (Thornbury
1999: 134).

Language input in Tasks need setting up with a pre-task stage, at the very least to give task
pre-task and task instructions. One purpose of pre-task preparation is to establish for learners
materials what the task’s communicative goal and communicative context are.
Another is to give learners ideas on how to approach the task. Willis (op. cit.)
suggests possibly playing recordings of fluent TL speakers performing
tasks learners are about to undertake.
An unresolved issue is whether a task-based approach precludes the use of
linguistically enhanced pre-task input. Nunan (2001b) believes that tasks
can include ‘reproductive’ language use, where learners reproduce
language models provided by the teacher, textbook, or some other source:
. . . communicative tasks can also involve reproductive language use.
For example, a classroom survey in which a student has to identify
classmates’ food preferences is both reproductive (the speaker will be
using the predictable form ‘Do you like hamburgers?’) and also
communicative (he or she doesn’t know how the interlocutor will
respond).
(Nunan 2001b: vi)
Willis (1990) believes that providing language models prior to tasks is likely
to cause learners to concentrate more on remembering and replicating the
formulae as accurately as possible than on communicating meaning:
The danger with focusing mechanically on form too early [in a lesson]
is that students see what follows not as an opportunity to use language
for communication, but rather as an opportunity to produce the
prescribed form as often as possible.
(Willis 1990: 73)
Willis (ibid. 64) notes that while during the pre-task stage some language
input is inevitable, but believes that in TB L exposure to any particular
language forms is intended to be incidental.

Mining language The purpose of this paper is not to take sides on the issue of whether it
input from pre-task is ‘permissible’ to linguistically enhance pre-task and task input. The
and task materials primary focus here is simply to find whether learners use language items

Learning mining of task input 67

 found in pre-task and task materials during task performance when
teachers do not explicitly draw learner attention to these language features.
‘Incidental exposure’ to language input during the pre-task stage may
underestimate how much language input learners take, or ‘mine,’ from
pre-task/ task materials. Samuda (2001) recorded a group of four high
beginner/low-intermediate learners performing a task that required them
to look at the alleged contents of a person’s pockets, and to hypothesize the
person’s name, sex, age, and marital status. The group was required to
register the degree of probability/possibility of their four hypotheses under
the following headings.

figure 1
Extract from task chart
(Samuda 2001: 127)

Samuda (ibid. 127–8) found that during the learner performance of the
task, out of 124 learner expressions of probability/possibility, 33 (26%) of
these expressions were conveyed using items ‘mined’ from the task
materials. These were ‘(it’s) possible’, ‘(it’s) probable’, ‘90%’, ‘certain’, and
‘50%’. The remaining 91 (74%) of learner expressions of possibility/
probability were items mobilized from their existing interlanguage.
These were ‘maybe’, ‘(I’m) sure’, ‘(I’m) not sure’.

The study Firstly, what input do learners mine on their own initiative? Do learners
Research questions mine input only from written task materials, or do they mine input from
audio pre-task recordings as well?
Audio recordings, whether scripted or authentic, are used in the
classroom often for the reason that they contain repeated examples of
certain language features. Do learners mine these features from recordings
and incorporate them into their task production?
Secondly, do learners mine specific language features deliberately
embedded by the teacher into pre-task/task materials? Do learners notice
and use these features even when the teacher does not explicitly draw
learner attention to them?

Subjects and During a semester of teaching a course comprised of task-based lessons,

procedure I recorded two classes of low-level (false beginner), Japanese, vocational
college students performing the same task. Over a three-week period,
each week a different task was performed. One class listened to two to
four audio recordings prior to tasks (labelled ‘R P T’), while the other class
listened to the recordings after tasks (labelled ‘RAT’). In the first week, I
recorded one pair of RPT and one pair of R AT learners, in the second week
a different pair of RPT and RAT learners, and in the third week another
pair of R P T learners and three pairs of R AT learners.

Minimizing learner Task-activities taken from the students’ coursebook were edited to ensure
focus on wordings that specific language features in, for example, the form of model
during pre-task dialogues or ‘speech bubble’ prompts, were not embedded in the
activities

68 Jeremy Scott Boston

 materials. The task instructions and materials given to the students are
shown below exactly as presented to the learners.
Pre-task listening activities did not overtly focus learner attention on specific
language structures contained in the audio recordings. Take, for
example, the coursebook listening activities below.

figure 2
Example listening task
from Expressions Book 1
(Nunan 2001a: 57)

Only listening activities similar to steps ‘A’ and ‘B’ above were performed
during the pre-task stage, as these focus only on displaying
comprehension (which can be displayed non-verbally), whereas, step ‘C’
and the ‘Try this’ step require students to recall and focus upon specific
wordings from the audio recordings.

Learning mining of task input 69

Tasks 1 and 2: mining Task 1:
written, not audio, 1 Task instructions and materials:
input

figure 3
Task activity from
Expressions Book 1
(Nunan 2001a: 108)

2 Example textbook recordings (2 of 4, ‘Let’s [verb]’ and ‘How about [verb] +

ing,’ recurring in alternate recordings).

figure 4
Recordings from
Expressions Book 1:
Teacher’s Annotated
Edition (Nunan 2001b:
T105)

Learner performance When performing Task 1, both the R P T and RAT pairs mined the words
of Task 1 ‘suggest’ and ‘suggestion’ from the written task instructions to make gift
suggestions. (See Figure 5 below.) While the recordings contain the
phrases ‘Let’s [verb]’ and ‘How about [verb] + ing’ to make suggestions, the
RPT pair seemed uninfluenced by this, instead using the mined words
‘suggest’ and ‘suggestion’. All learners initiated exchanges by stating
what a person likes (for example, ‘Connie likes going to gym’) or by asking
‘What does [name] like?’ whether they listened to pre-task recordings or
not, making it less likely that the R P T pair mined these wordings from
the audio input.

70 Jeremy Scott Boston

figure 5
Excerpts of Task 1 pair
performance

Similar results were found from learner performance of the self-made task
below. Students mine input from written task materials in lieu of audio
pre-task input.

Task 2:
1 Task instructions/materials:

figure 6
Written instructions for
Task 1

2 Example textbook recording (1 of 4, ‘Do you know’ and ‘What does ____
look like’ in all recordings).

figure 7
Recordings from
Expressions Book 1: TAE
(Nunan 2001b: T25)

Learning mining of task input 71

Learner performance Neither R P T pair learners used the structure ‘Do you know [name]?’ or
of Task 2 ‘What does [name/he/she] look like’ while performing Task 2; despite
hearing it repeated in all four recordings. Rather, both RPT and RAT pairs
asked, ‘Who is [name]?’ or mined the categories under which they had to
place student names: ‘Names I forget’ and ‘Names I remember’. Often,
pairs took the words ‘forget’ and ‘remember’ and attempted to form indirect
questions (for example, ‘I forget who is Hitomi’, for, ‘I forget who Hitomi
is’). (See Figure 8.)

figure 8
Excerpts of task 2 pair
performance

Summary and With such a small sampling of student task production, any conclusions will
reflections on learner have to be tentative. Nevertheless, the data above appears to indicate that
performance of low-level learners do not mine whole structures or phrases from pre-task
Tasks 1 and 2 audio recordings. That the learners did not mine input from the audio
recordings was somewhat surprising, given that the RPT learners listened
to the audio recordings 3–4 times before being able to complete the listening
comprehension activities. Playing pre-task recordings, therefore, may not
result in low-level learners focusing on and employing specific language
features from the audio input, unless teachers explicitly draw learner
attention to these features.

Task 3: manipulating Learner performance of Tasks 1 and 2 above, show use of input mined
language input in from the written task materials. This opens up the possibility of covertly
written task ‘seeding’ written task materials with wordings for learners to mine and use
materials during task production. For example, in Task 1 above, the coursebook

72 Jeremy Scott Boston

 deliberately included the sentence, ‘He already has a lot of . . .’ which
learners did mine. (See also Figure 3 above.)

RAT Pair
S2 I suggest you to buy exercise clothes.
S1 She already has a lot of work out clothes, is same mean as exercise
clothes.

RPT Pair
S2 I suggest tennis balls
S1 No
S2 Why? Already has?
In the following task, I recorded one RP T pair, and three R AT pairs. Each
RAT pair were given differently worded task instructions, all other written
task materials being the same for all three R AT pairs.

Three differently Task 3:

worded task 1a Task instructions for the RPT Pair and RAT Pair 1.
instructions

figure 9
Task instructions for RPT
pair and R AT pair 1

1b Instructions for R AT Pair 2.

figure 10
RAT pair 2 task
instructions

1c Instructions for R AT Pair 3.

figure 11
RAT pair 3 task
instructions

Learning mining of task input 73

 Task materials

figure 12
Task activity from
Expressions Book 1
(Nunan: 2001a: 92)
Example textbook recordings (1 of 2, ‘going to . . .’ recurring in both
recordings).

figure 13
Recording from
Expressions Book 1 TAE
(Nunan 2001b: T89)

Learner performance of Task 3

Both R P T pair 1 and R AT pair 1 mined the word ‘schedule’ from the task
instructions and used it frequently during production, asking questions
like, ‘What is Joan’s schedule?’ However, neither pair mined ‘next week’
despite it being underlined. More interestingly, both pairs used ‘will’ rather
than the present progressive (‘I’m going . . .’). The RPT group did so despite
the present progressive being the ‘future tense’ heard on the recordings.

figure 14
Excerpts of R P T pair and
RAT pair 1, Task 3
performance

74 Jeremy Scott Boston

 RAT pair 2 and RAT pair 3 also mined input from their differently worded
task instructions. R AT pair 2 mined ‘be going to’ from their task
instructions; RAT pair 3 mined the word ‘plan(s)’ from theirs. Both pairs
employed ‘will’ from their existing interlanguage. (See Figure 15 below.)

figure 15
Excerpts of Rat pair 2 and
RAT pair 3, Task 3
performance

(Note: I am sure readers will notice the flaw in the design of Task 3, 1b.
The mismatch between the linguistic focus ‘be going to’ the writer uses
in the written instructions and the present progressive (albeit using the
lexical verb ‘go’) in the recording. Nevertheless, the fact R P T Pair 2 mined
‘be going to’ from the rewritten task instructions still stands.)

Summary of learner The findings from Task 3 support previous observations that low-level
performance of learners do not appear to mine whole structures from pre-task audio
Task 3 input. Rather, it seems that low-level learners mine linguistic input
from written pre-task and task materials to incorporate into their spoken
task production. From the R P T pair 2 task performance, it appears
possible to plant specific input into written pre-task/task materials for
learners to mine.

Discussion and The debate in TB L over whether tasks can include linguistically
conclusion enhanced pretask and task input materials may never be resolved. As
Skehan (1998: 96) notes, ‘[the] two underlying characteristics of tasks,
avoidance of specific structures, and engagement of worthwhile meanings,
are matters of degree, rather than being categorical’.
I suggest that if the most natural task instructions happen to include specific
structures that also comprise the lesson’s language focus, so be it. For
example, in my experience, task instructions such as ‘Find out who has

Learning mining of task input 75

 been to the most foreign countries’ causes learners to attempt to use the
present perfect, both in their questions and in responses, yet there is really
no other way to word these instructions. The same was true in the task
materials for Task 1 above (Figure 3: the ‘Make-a-suggestion’ task). Due to
the somewhat artificial nature of this task, the materials had to provide
learners with a ready-made reason to reject suggestions. Learners were told
that, ‘He/She already has a lot of . . .’ and learners mined all, or part, of this
sentence during task production. I would also argue that deliberately
embedding pre-selected language formulae into natural sounding
instructions falls far short of ‘mechanically focusing’ on form.
Learner mining of pre-task/task input is an understudied factor to
consider when designing tasks for the classroom and this research needs
repeating with a larger sampling of students at a variety of learner levels.
It does appear however, from this study, that not only is some language
input inevitable during the pre-task and task stage, it appears that learner
mining of, at least, written language input is inevitable as well.
Therefore, teachers may also to be able to ‘seed’ written pre-task input with
pre-selected wordings in the deliberate hope that learners mine and employ
those wordings. While this will strike some as perhaps not being in the
‘spirit’ of a task-based approach, this paper leaves it to individual task
designers to decide if finding ways of harnessing learner mining of
wordings from pre-task input is in harmony with the principles of a task-
based approach.
Final revised version received September 2005

References Thornbury, S. 1999. How to Teach Grammar. Harlow:

Bygate, M., P. Skehan, and M. Swain. (eds.). 2001. Addison Wesley Longman.
Researching Pedagogic Tasks: Second Language Willis, D. 1990. The Lexical Syllabus. London: Collins
Learning, Teaching and Testing. Harlow: Addison CO BUILD.
Wesley Longman. Willis, J. 1996. A Framework for Task-Based Learning.
Nunan, D. 2001a. Expressions: Meaningful English Harlow: Addison Wesley Longman.
Communication, Book 1. Boston: Heinle and Heinle. Willis, J. 2000. ‘A holistic approach to task-based
Nunan, D. 2001b. Expressions: Meaningful English course design’. The Language Teacher 24/2: 7–11.
Communication, Book 1. Teacher’s Annotated Edition.
Boston: Heinle and Heinle. The author
Samuda, V. 2001. ‘Guiding relationships between Jeremy Scott Boston has been trying to make task-
form and meaning during task performance: the based learning palatable for Japanese EF L students
role of the teacher ’ in M. Bygate, P. Skehan, and for over 10 years. He currently teaches at Hiroshima
M. Swain (eds.). Shudo University, Japan.
Skehan, P. 1998. A Cognitive Approach to Language Email: [email protected]
Teaching. Oxford: Oxford University Press.

76 Jeremy Scott Boston