How textbooks (and learners) get it wrong: A corpus study of modal auxiliary verbs

Authors

1 Shiraz University of Paramedical Sciences, Iran

2 Chulalongkorn University, Bangkok, Thailand

Abstract

Many  elements  contribute  to  the  relative  difficulty  in  acquiring  specific  aspects  of  English  as  a
foreign  language  (Goldschneider  &  DeKeyser,  2001).  Modal  auxiliary  verbs  (e.g.  could,  might),
are  examples  of  a  structure  that  is  difficult  for  many  learners.  Not  only  are  they  particularly
complex  semantically,  but  especially  in  the  Malaysian  context  reported  on  in  this  paper,  there  is
no direct equivalent in the students’ L1. In other words, they are a good example of a structure for
which  successful  acquisition  depends  very  much  on  the  quality  of  the  input  and  instruction
students  receive.  This  paper  reports  on  analysis  of  a  230,000  word  corpus  of  Malaysian  English
textbooks,  in  which  it  was  found  that  the  relative  frequency  of  the  modals  did  not  match  that
found  in  native  speaker  corpora  such  as  the  BNC.  We  compared  the  textbook  corpus  with  a
learner  corpus  of  Malaysian  form  4  learners  and  found  no  direct  relationship  between  frequency
of presentation of target forms in the textbooks and their use by students in their writing. We also
found a very large percentage of errors in students’ writing. We suggest a number of possible
reasons for these findings and discuss the implications for materials developers and teachers. 

Keywords

Main Subjects


Introduction   
Materials  play  a  key  role  in  most  language
classrooms  around  the  world  and  their
evaluation is therefore of prime importance.
Language  learning  materials  can  be
evaluated  at  the  pre-use  stage,  where  they
are  seen  as  workplans  or  constructs,  during
use,  when  they  are  judged  as  materials  in
process,  and  retrospectively,  which
considers  outcomes  from  materials  use
(Breen,  1989).  Ellis  (1997)  suggests  that
predictive  evaluation,  which  aims  to
determine  appropriateness  for  a  specific
context, is carried out either by experts or by
teachers  using  checklists  and  guidelines.  At
the  in-use  stage  ‘long-term,  systematic
evaluations  of  materials      .  .  .  are  generally
considered  to  be  successful’  (Tomlinson,
1998,  p.5).  These  include  ‘formative
decisions  for  improvement  through
supplementation  or  adaptation  and
[sensitising]  teachers  to  their  own  teaching
and learning situation’ (Nedkova,  2000,  p.
210).  In  this  study,  we  concern  ourselves
with retrospective evaluation in that we look
at materials that were in use on a large scale,                         
by  many  thousands  of  language  learners,  at
one  given  time,  to  learn  about  the  type  and
quality  of  the  language  input  contained  in
them.  In  order  to  do  this  we  drew  on
corpora,  the  use  of  which  in  ELT  and
language  learning  research  we  will  now
discuss.  
The role of corpora in ELT
The  use  of  corpora  for  both  teaching  and
research has increased significantly in recent
years.  The  motivation  for  using  a  corpus
approach  in  language  learning  research  is
related in part to the attraction of being able
to offer a description of language in use and
also  to  the  fact  that  previous  research  on
authentic  texts  have  revealed  significant
inconsistencies  between  the  use  of  lexical
items and grammatical structures in corpora,
and  those  found  in  traditional  language
textbooks  that  are  based  purely  on
introspective  judgments  (Campoy,  Belles-Fortuno,  &  Gea-Valor,  2010).  At  the  same
time, corpus explorations can be carried out
by  learners  themselves  and  can  be  used  as
an integral part of the learning process either
directly or indirectly to both foster learners’
and teachers’ needs (Romer, 2010).
 
As  a  result  of  this  growing  interest,  the  use
of  corpora  has  resulted  in  the  development
of  more  effective  pedagogical  materials
(Gabrielatos,  2005).  Material  writers  can  be
informed  of  the  differences  between  the
language used in textbooks and of that used
in  the  real  world.  Information  about  the
frequency  of  occurrence  of  linguistic
features  in  a  reference  corpus  can  also  be
very  helpful  when  it  is  compared  with
prescribed  pedagogical  materials.  While
many linguists and researchers have focused
on  the  advantages  of  corpus-informed
materials, there are also limitations that need
to  be  taken  into  consideration  by  textbook
writers.  
 
For  instance,  Howarth  (1998)  and
Widdowson  (1990)  have  questioned  the
pedagogical  usefulness  of  frequency  lists
generated  by  corpora  because  in  their  view
frequency  does  not  equate  to  importance.
However,  this  argument  has  been  strongly
rejected  by  many  linguists  such  as  Mindt
(1995),  Kennedy  (2002)  and  Romer  (2004)
because,  as  they  argue,  frequency
information  leads  to  the  identification  of
words  or  structures  that  are  central  in  a
language and that without this information it
is  difficult  to  decide  what  should  be
included  in  teaching  materials.  Kennedy
(1998),  among  others,  points  to  the  need  to
concentrate  initial  teaching  on  high
frequency items and to grade vocabulary and
structures  accordingly  and  Conrad  (2000)
emphasizes  the  importance  of  frequency
information  for  teachers  because  it  helps
them  to  decide  which  items  to  emphasize,
for  example,  to  provide  low-level  students
practice  with  the  items  they  are  most  likely
to hear outside class.  
 
Lawson  (2001)  argues  that  insights  from
corpus  linguistics  cannot  only  provide
information  about  the  frequency  of
occurrence of linguistic features in naturally
occurring  language,  but  also  about  register
variation,  that  is  about  how  the  use  of
particular  linguistic  features  varies  across
different contexts and situations of use. This
information,  according  to  Kennedy  (1998)
can  be  of  direct  application  to  textbook
writers.  Furthermore,  it  is  argued  that
corpus-based  analysis  can  provide
information  about  the  salience  or  scope  of
particular  features  which  otherwise  are
difficult  to  acquire  (Lawson,  2001).  Stubbs
(1996) summarises:  
 
There  may  be  the  illusion  that  they
[lists  of  collocations]  could  have
been provided, after a bit of thought,
by intuition alone. But this is indeed
an illusion. Intuition certainly cannot

provide  reliable  facts  about
frequency  and  typicality.  And  whilst
a  native  speaker  may  be  able  to
provide some examples of collocates
(which may or may not be accurate),
only  a  corpus  can  provide  thorough
documentation. (p.250)
 
In  our  study  we  use  corpus  linguistics  not
primarily  to  inform  materials  development,
but  to  learn  about  materials,  information
which,  subsequently,  may  be  useful  for
further development  
 
Methodology
The target structure  
We  chose  modals  for  this  study  for  several
reasons.  Firstly,  modal  auxiliary  verbs  are
particularly  challenging  for  language
learners  (Decapua,  2008)  and  also  for
Malaysian  English  learners  (e.g.  Manaf,
2007; Wong, 1983; De Silva, 1981). Perhaps
as  a  result  of  this,  they  do  not  receive  as
much  attention  as  part  of  the  school
curriculum  as  before.  As  De  Silva  (1981)
observes: ‘the modal auxiliary system used
in  the  Malaysian  schools  has  been  altered
and  functionally  reduced  through  the
continued  use  of  fewer  and  semantically
salient  modals  that  serve  multi  functionally
across  notions  (p.  12).  Wong  (1983)  argues
that  the  limited  exposure  of  Malaysian
learners  to  different  forms  of  modal  verbs
and  their  functions  has  resulted  in  an
overuse  of  one  form  or  function  over  the
others  by  teachers.  As  modal  auxiliaries  are
so difficult, they are likely to be particularly
influenced  by  the  quality  of  the  input  and
instructions learners receive on them and we
were  therefore  particularly  interested  to  see
how this feature is presented to learners.  
 
We  also  chose  modal  auxiliaries  because
they  play  an  important  role  in  learners’
language  use.  Many  Malaysian  learners
aspire  to  study  through  the  medium  of
English  and  good  use  of  modals  plays  an
important  role  in  successful  social
interaction  (Celce-Murcia  &  Larsen
Freeman,  1999).  In  other  words,  it  is  an
important  feature  of  the  language,  not  just
from  a  linguistic  point  of  view,  but  also  for
the  learners  themselves,  from  a  social-interactional  point  of  view.  Modal  auxiliary
verbs  are  also  common  and  we  therefore
thought  it  would  be  likely  that  we  would
find many exemplars to analyse.  
 
The  final  reason  for  the  selection  of  modal
auxiliaries is that previous studies conducted
in  other  countries  have  reported  that
textbooks  do  not  present  this  structure
accurately (Hyland, 1994; McEnery & Kifle,
2002).  In  summary,  modal  auxiliaries  are  a
difficult,  common  and  important  (to
learners)  structure  that  has  often  been
misrepresented  in  English  language
textbooks.  
 
Modal  auxiliary  verbs  and  Malaysian
learners
Malaysian  learners  have  been  observed  as
having  great  difficulty  with  the  modal
auxiliary  system.  Examples  (1)  to  (8)
provide  illustrative  evidence  for  existing
problems  concerning  the  appropriate  use  of
modal  can  with  its  various  functions  by
Malaysian students (Wong, 1983, p.137):  
 
1)  You  can  have  this  book  today.
(“permission”)
2)  You can drive? (“ability”)
3)  Can  lend  me  your  bike  or  not?
(“willingness”)
4)  Can also/ Sure can. (“agreement”)
5)  Can do. (“moderate approval”)
6)  You  come  with  me.  Can  or  not?
(“affirmation)
 
Hughes and Heah  (1993) made very  similar
observations  based  on  learner  data  and
report on problems Malaysian learners  have
36                               How textbooks (and learners) get it wrong                              
with  the  use  of  modals.  The  correct  use  of
modals,  according  to  them,  was  always
among one of the most problematic areas for
Malaysian learners (Hughes & Heah, 1993).
Furthermore,  in  their  study  of  students’
errors  in  Form  4  students’  composition,
Rosli  and  Edwin  (1989)  found  that  verb
forms and the verb aspects of modals are the
most  problematic  for  Malaysian  learners.
Twenty  years  since  Rosli  and  Edwin’s
(1989)  study,  the  same  observation  was
made  by  Manaf  (2009),  who  analyzed  the
modal  auxiliary  verbs  in  the  Malaysian
learner  corpus  (EMAS).  According  to  her,
students  were  not  only  uncertain  about
which  modals  to  use  to  express  modality
(inaccuracies  at  the  syntactic  and  semantic
levels), but also had difficulty to use modals
with  appropriate  verb  form  in  a  sentence
(Manaf,  2009).  Although  the  lack  of  equal
counterparts  between  the  English  modal
system  and  those  in  Bahasa  Melayu  might
be  the  reason  for  this  confusion  for  Malay
learners,  Romer  (2005)  believes  that  this
problem is due to the teaching materials.
 
Modal  auxiliaries  in  Malaysian  grammar
and textbooks  
There  are  six  modals  which  are  required  to
be  taught  in  Kurikulum  Bersepadu  Sekolah
Menengah  (KBSM)  syllabus  for  lower  and
upper secondary students namely, must, will,
should,  can,  may  and  might.  The  frequency
of  could,  would  and  shall,  however,  is  also
investigated in this study in order to see how
many  times  these  modals  are  presented  to
students  implicitly  throughout  the  texts
during  four  years  of  study.  According  to
KBSM, in the Form 1 textbook, students are
supposed  to  be  exposed  to  and  taught  the
three modals must, will and should. In Form
2  can,  will,  must,  may  and  might  are  added
and repeated in Form 3. In Form 4, should is
added.  The  prescribed  Malaysian  English
language textbooks used in schools are often
reported as being prepared through a process
of  material  development  involving  intuition
and  assumption  (Mukundan  &  Roslim,
2009;  Mukundan  &  Khojasteh,  2011).
Existing textbooks therefore appear to lack a
broad empirical basis.  
 
Corpus selection
In  order  to  answer  our  research  questions,
we  used  two  corpora;  a  pedagogic  corpus
and a learner corpus. A pedagogic corpus, as
coined  by  Willis  (1993)  and  defined  by
Hunston  (2002),  is  a  collection  of  data  that
‘can consist of all the course books, readers
etc.  a  learner  has  used’  in  an  ESL/EFL
language  learning  program  (p.16).  In  this
study  the  population  of  our  pedagogic
corpus  was  sourced  from  four  Malaysian
English  language  textbooks  currently  used
for secondary Malaysian students of Form 1
through  Form  4,  with  a  total  of  just  under
230,000  words  (Mukundan  &  Aneleka,
2007)
.  According  to  the  researchers  each
page  of  the  books  mentioned  above  was
photocopied and scanned and converted into
a Tagged Image File (TIF) format. This was
then  saved  and  processed  with  Optical
Character  Recognition  (OCR)  software,
which  converted  all  TIF  files  into  text  files
(.txt).  The  txt  files  were  then  checked  for
errors  before  saving  and  renaming  them
according  to  the  respective  units  of  the
textbook.
 
The  learner  corpus  we  used  was  sourced
from two written essays produced by Form 1
and  Form  4  Malaysian  students  as  part  of  a
previous  study  (Arshad,  Mukundan,
Kamarudin,  Rahman,  Rashid,  &  Edwin
2002).  In  the  study,  approximately  600
                                                          
   The  original  corpus  consisted  of  5
Malaysian  English  language  textbooks  used  in  the
secondary  cycle  (311,214  running  words).  However,
in order to suit the textbook data with our learner data
we  decided  to  only  include  Forms  1,  2,  3  and  4  and
eliminate  the  Form  5  data  from  this  pedagogic
corpus.  Hence,  the  remaining  running  words  in  this
corpus consist of 229,794 running words.  

Malaysian  learners  from  across  the  country
were required to write one essay on the topic
of  ‘The  happiest  day  of  my  Life’  and
another  based  on  a  given  picture.  Students
were given one hour to write the essays and
were  not  marked  or  given  credit  for  them.
Although  perhaps  not  ideally  representative
of Malaysian learners’ language proficiency,
it was decided to use this corpus because of
its  very  large  size  and  the  fact  that  it  does
give a broad indication of language learners’
writing  ability  across  the  whole  of  the
country.  
 
Analysis  
As our benchmark corpus we used the BNC,
the  British  National  Corpus.  This  corpus
consists  of  100  million  word  collection  of
samples  of  written  and  spoken  language.
Among  all  reference  corpora  available,  the
insights  on  modal  auxiliary  verbs  were
sought  from  BNC  because  the  samples  of
written  and  spoken  language  used  for  this
corpus  were  designed  to  represent  a  wide
cross-section of British English (BrE) which
is  the  closest  English  variety  used  in
Malaysia  (Mukundan  &  Roslim,  2009;
Mukundan  &  Khojasteh,  2011).  A  previous
study  by  Kennedy  (2002)  looked  at  the
occurrence of modal auxiliary verbs and we
draw  on  his  findings  here  for  our
comparisons  with  the  results  from  the
textbook  corpus  and  the  learner  corpus.  In
the  latter  two,  we  retrieved  modal  auxiliary
verbs  using  the  software  package
WordSmith  and  in  particular  its  Concord
tool  to  locate  all  references  to  modal  verbs
within both corpora. In order to examine the
first  research  question,  content  analysis  was
carried  to  retrieve  absolute  frequencies  of
occurrences  for  nine  core  modal  auxiliary
verb forms from all written and spoken texts
in  the  four  Malaysian  secondary  English
language  textbooks.  Then,  the  results  were
added  up  and  compared  with  the  frequency
and  rank  order  of  the  same  modals  in  the
BNC  in  order  to  see  if  there  were  any
discrepancies.  Next,  discourse  analysis  was
carried  out  at  the  sentence  level  in  order  to
examine  the  accuracy  of  the  way  in  which
the  modals  were  presented  at  both  syntactic
and semantic levels.  
 
In  addition  to  looking  at  the  frequency  of
use  of  modal  auxiliary  forms,  we  were  also
interested  in  looking  at  the  grammatical
accuracy of learners’ use of this form. In
order  to  do  this,  all  sentences  in  the  learner
corpus that included modals were  examined
using  Mindt’s  (1995)  modal  verb  phrase
structure  framework.  According  to  Mindt
(1995),  word  categories  can  colligate  with
modals in five different structures:  
 
1)  modal  +  bare  infinitive  (e.g.  You
won't regret it!)
2)  modal  +  passive  infinitive  (e.g.
Something should be done)
3)  modal  +  progressive  infinitive  (e.g.
Define  what  you  will  be  talking
about)
4)  modal  +  perfective  infinitive  (e.g.
The number of the students will have
increased) 5) modal + perfect passive
infinitive  (e.g.  I  know  it  must  have
been hard for her).  
 
To this we added ‘modal alone’, a category
suggested by Kennedy (2002).  
 
Results
Here  we  present  the  results  of  our  study.
First  we  show  the  results  of  the  analysis  of
the  textbook  corpus,  followed  by  the
analysis  of  the  learner  corpus.  Finally,  we
present  our  analysis  of  the  errors  in  the
learner corpus.  
 
Modal auxiliary verbs in the textbook corpus
Figure  1  shows  the  frequency  of  the  modal
auxiliary  forms  (including  their  negative                           
forms)  in  the  four  English  textbooks  in
descending order.

There  were  altogether  2,807  instances  of
core  modals  in  the  textbook  corpus.  As  can
be seen above, there is a large frequency gap
between  can  and  will  on  the  one  hand  and
the  other  seven  modals  on  the  other.  There
are  1398  occurrences  of  can  and  will  and  a
total  of  1401  for  should,  may,  would,  must,
could,  might  and  shall.  The  most  frequent
modals  can  and  will,  therefore  account  for
almost  50  %  of  all  modal  tokens  in  the
corpus.  
 
Modal auxiliary verbs in the learner corpus
Figure  2  shows  the  order  of  frequency  in
which  students  used  modal  auxiliary  forms
on the writing tasks.

(13.59%)  and  175  (12.51%)  occurrences
respectively.  
 
Errors  in  modal  auxiliary  verbs  in  the
learner corpus
Next, we analyzed the accuracy of learners’
modal auxiliary use in their writing. Figure 3
shows  the  number  of  accurately  and
inaccurately produced modals.

In  descending  order,  the  lowest  percentage
of  syntactical  inaccuracy  was  for  shall
(80%)
,  can  (54%),  would  (46%),  could
(45%), might (41%), will (22%), may (11%)
and should (8%).  
 
Out  of  only  five  shall  modals  used  by  the
learners, four were used with progressives or
past  tense  forms  of  the  verb.  Examples  (1)
and  (2)  are  sample  sentences  of  inflected
modals:
 
(1) She  also  don't  know  how  what
she shall doing.
(2) "Shall  we  invited  John  join  with
us?" I asked Ahmad again.  
 
More  than  half  of  all  can  instances  used  by
Malaysian  learners  were  used  inaccurately.
149  occurrences  were  used  with  structure
one  (modal  +  bare  infinitive)  but  with  the

   But  note  the  small  number  of  total
occurrences

past tense of the verb. Examples (3), (4) and
(5) are sample sentences of such errors.  
 
(3) I can saw many kind of tress.  
(4) He  can  spoke  fluently  in  Malay
language.  
(5) She hope that Raj, Ah Seng, and
Ramlee can heard her.
 
There were also many incidences of the use
of  a  non-English  word  after  the  modal  and
combining  two  modals.  Furthermore,  many
of  the  negative  sentences  constructed  by
students using can were ungrammatical:
 
(6) I  hope  I  can  will  visit  this  place
again.
(7) She can’t swam.
 
Would  was  used  inaccurately  87  times  by
Malaysian  learners.  Although  most
sentences  were  still  comprehensible,  81  of
the  inaccurate  instances  had  the  modal
would followed by the past tense form. This
was  the  same  for  those  who  had  used  this
modal  in  structure  4.  In  only  six  cases  was
the verb after the modal would missing:
 
(8) I  felt  something  joyful  would
happened later.  
(9) If  they  call  me,  they  would  told
me that the enjoyable day of their
life was when they were in 3A1.   
(10)  Probably  they  would  have
broke some records if we were to
take the time.
 
The same tendency can be seen in the usage
of  could  where  in  all  cases  the  verb  that
follows the modal was in the past form:
(11)  and  we  could  entered  the  semi-final  because  our  compenen  had  a
stomachache during the competition.
(12) My heart beat was beating faster
and  faster  as  I  could  found  nobody
around.
Over-generalization  of  the  past  tense  was
also found in the use of might:
 
(13) I didn't tell my husband because
I  scared  that  I  might  lost  them
especially my children.
(14)  One  day,  when  I  came  back
from  school,  my  heart  felt  not  very
well  and  seemed  that  something
might happened.
 
Ninety-nine  out  of  the  syntactically
inaccurate  uses  of  will  were  either  followed
by  progressives  or  the  past  tense  of  the
verbs.  The  rest  were  either  preceded  by  the
verb  with  the  intervening  to  infinitive  or  a
combination of two modals:
 
(15) My parents will to stay with me
for a few days.
(16)  I  will  can  remember  this  party
forever in my life.
 
May  and  should  were  the  only  modals  in
which  students  did  not  produce  many
inaccurate sentences.  
 
Discussion  
In  the  preceding  section  we  presented  the
results of our analysis of the 1) frequency of
modal  auxiliaries  in  the  textbook  corpus,  2)
the  frequency  of  modal  auxiliaries  in  the
learner  corpus,  and  3)  the  errors  in  modal
auxiliary usage in the learner corpus. In this
section  we  will  discuss  and  attempt  to
explain these findings.  
 
The analysis of the textbook corpus showed
that there were altogether 2,807 instances of
core  modals  in  the  textbook  corpus.
Particularly  noticeable  were  the  large
frequency  gap  between  can  and  will,
accounting  for  nearly  50%  of  all  modals,
and  the  other  seven  modals.  We  were
interested  to  establish  to  what  extent  the
order  of  occurrence  of  the  modals  matches
                       
that found in native speaker corpora. To this
end,  we  compared  our  findings  with  data
from the British National Corpus (BNC), the
corpus  of  Survey  of  English  Usage  (SEU),
the  Lancaster-Oslo/Bergen  Corpus  (LOB),
and  the  Longman  Grammar  of  Spoken  and
Written  English  (LGSWE)  corpus.
According to Kennedy (2002), the four most
frequent  modal  auxiliaries  in  the  native
speaker  corpora  are  will,  would,  can  and
could,  accounting  for  72.7%  of  all  modal
tokens.  Similarly,  Coates  (1983)  reported
would,  will,  can  and  could  as  the  most
frequent  modals,  accounting  for  71.4  %  of
all  modal  tokens.  Will  is  therefore  only  the
second  most  common  form  (Biber,
Johansson,  Leech,  Conrad,  &  Finegan,
1999), while in the textbook corpus it is the
first.  Likewise,  can  is  only  the  third  most
common  modal  in  the  above  corpora,  but  it
the  most  common  in  the  textbook  corpus.
An  even  greater  discrepancy  is  found  with
the  modal  could,  which  is  the  4th most
common modal in the above corpora, but the 7th
  most  common  modal  in  the  textbooks.
Should  is  over-represented  as  the  3d  most
common  modal  in  the  textbook  corpus  but
(according  to  Kennedy  2002,  and  Quirk,
Greenbaum,  Leech,  &  Svartvik,  1985)  it  is
only sixth in the major corpora. May is more
frequent  in  the  textbook  corpus  than  could
and  would,  while  in  the  native  speaker
corpora this is not the case.  
 
In summary, the order of frequency of most
modals in the Malaysian textbooks does not
match  that  found  in  native  speaker  corpora.
In  some  cases  the  differences  are  in  fact
quite  significant.  This  points  to  the
likelihood  that  the  textbook  development
was  not  informed  by  corpus  data  but  was
based, at least in part, on the intuition of the
textbook writers.  
 
When looking at the frequency of modals in
the  learner  corpus,  we  found  that  it  did  not
match  that  of  the  modals  in  the  textbook
corpus.  A  significant  difference  was  found,
for  example,  for  the  modals  would  and
could,  which  were  among  the  four  most
frequent  modals  in  the  learner  corpus  but
which  were  not  very  common  at  all  in  the
textbook  corpus.  What  could  explain  these
differences?  One  possibility  is  that  the
frequency  of  occurrence  in  the  textbooks
does not match the extent to which they are
explicitly  dealt  with;  in  other  words,
although  a  modal  might  be  used  in  many
different  texts  throughout  the  book,  perhaps
there is no instruction in it, or vice versa. A
previous  study  by  Khojasteh  and  Kafipour
(2012)  looked  into  the  amount  and  type  of
instruction  on  all  nine  modals  in  the
textbooks  and  found  that  in  the  case  of
would  and  could  these  were  not  explicitly
dealt with at all in the textbooks. That leaves
two  possibilities;  teachers  instruct  learners
in  this  modal  in  class,  even  though  it  is  not
part  of  the  course  book  (which  seems
unlikely),  or  learners  are  exposed  to  this
form  elsewhere,  which  leads  them  to  use  it
more often.
 
On  the  other  hand,  should  did  not  appear
much  in  the  learner  corpus,  although  it  was
somewhat  common  in  the  textbook  corpus.
One  of  the  reasons  for  this  may  be  that  the
nature  of  the  writing  topics  that  the  learner
corpus  was  drawn  from  (see  above),  which
did  not  require  students  to  use  either  the
obligation  or  the  logical  necessity  meaning
of  the  modal  auxiliary  should.  However,
further  research  is  needed  to  establish  why
we found these discrepancies.  
 
When we looked at learners’ errors in their
use  of  the  modal  auxiliaries,  we  found  that
shall,  can,  would  and  could  in  particular
proved  to  be  difficult.  Interestingly,  shall,
would and could were the only three modals
out  of  the  nine  that  were  not  dealt  with
explicitly  in  the  textbooks.  For  could  and
would  we  have  further  evidence  from
Khojasteh  and  Kafipour  (2012)  that  they

also  not  taught  explicitly  at  primary  and
secondary  levels  in  Malaysian  textbooks

All  this  may  help  to  explain  why  learners
struggle  with  these  forms.  In  the  case  of
would  and  could  we  speculate  that,  due  to
the  lack  of  explicit  instruction,  students  did
not  fully  learn  how  to  differentiate  between
the  present  and  the  past  forms  of  these
modals.  The  tasks  given  to  the  learners
(‘describe one of the best days of your life’,
and the picture story task) were more likely
to require learners to use the past tense form
of the modals, leading to a relatively higher
number  of  errors.  However,  this  does  not
help  to  explain  why  their  comparative
frequency  in  the  learner  corpus  is  so  much
higher than in the textbook corpus.  
 
Conclusion and limitations
From  this  study  we  can  draw  a  number  of
conclusions,  each  of  which  carries
implications  for  further  research  as  well  as
teaching practice. One of the most worrying
observations  is  that  the  textbooks  in  our
study  expose  learners  to  input  in  which  the
frequency  of  the  modal  auxiliaries  simply
does not match that found in native speaker
corpora.  Although  there  are  sometimes
sound  pedagogical  reasons  for  emphasising
or  reducing  the  focus  on  a  particular  form,
that  does  not  appear  to  be  an  adequate
explanation  here.  The  most  common  forms
in the native speaker corpora are will, would,
can  and  could  and  there  is  no  apparent
reason,  for  example,  why  should  is  a
reasonable  replacement  for  could.  We
believe instead that our findings point to the
likelihood  that  the  development  of  the  four
textbooks in this study was not informed by
corpus  data  but  was  based,  at  least  in  part,
                                                
   Although  Thornbury  (2004)  has  indicated
that  the  most  frequently  occurring  items  are  not
always the  most useful ones in terms of teachability,
and  that  they  may  be  better  delayed  until  relatively
advanced  levels,  in  the  case  of  this  textbook  corpus
the  modals  could  and  would  are  taught  neither  at
lower nor higher secondary levels.
on  the  intuition  of  the  textbook  writers.
Unfortunately,  this  is  (still)  not  uncommon.
Barbieri  and  Eckhardt  (2007)  indicate  that
despite  more  than  two  decades  of  language
teaching  aimed  at  fostering  natural  spoken
interaction  and  written  language,
instructional  textbooks  still  neglect
important  and  frequent  features  of  real
language  use  (see  also  Hyland  1994,
Harwood,  2005).  Of  course,  our  study  only
looked at one (albeit important) grammatical
feature,  and  we  need  be  careful  not  to
generalise  our  findings  to  the  rest  of  the
textbooks.  Nonetheless,  if  a  central
grammatical  feature  is  handled  in  this  way,
it  does  raise  concern  and  further  research
should  be  done  to  establish  whether  our
findings  apply  to  other  grammar  and
vocabulary too.  
 
For  teachers,  our  findings  point  to  the  need
to be vigilant and, where feasible, to extend
coursebooks  with  other  materials,  to  give
students  broad  exposure  to  target  language
input. Many corpus tools are now freely and
easily  accessible  (for  example  the  BNC;
http://www.natcorp.ox.ac.uk/), and these can
help teachers to ensure appropriate weight is
given  to  each  grammar  point.  Another
finding is that learners’ production of modal
auxiliaries does not match their presentation
in the textbooks in terms of frequency. Some
modals that are common in the textbooks are
not frequently used in the learners’ writing
and  vice  versa.  Why  would  this  be  so?  At
this  point  we  are  unclear  and  further
research  will  need  to  be  done,  for  example
to  establish  the  interaction  between
frequency,  instruction,  and  learners’
exposure  to  these  features  outside  of  class.  
Of  course,  as  we  have  pointed  out  above,
frequency  of  input  is  only  one  element
contributing  to  L2  knowledge.  The  amount
and  type  of  instruction  play  an  important
role  as  well.  Interestingly,  we  found  that
those  modals  that  learners  did  not  receive
               
explicit  instruction  in  were  the  same  ones
they  produced  more  errors  on  in  their
writing.  What  this  shows  is  the  relationship
between  instruction  and  accuracy  in
language  production  and  the  importance  for
teachers  to  be  very  much  aware  of  what  is
and  what  is  not  covered  in  the  textbooks
they  use,  and  to  adapt  or  supplement  this
where necessary.  
 
There are, however, a number of limitations
to  our  study,  which  we  would  like  to
acknowledge  here.  Firstly,  not  much
information  is  available  about  the  methods
for  obtaining  the  learner  corpus.  For
example, official publications do not specify
the  precise  instructions  that  learners  were
given as part of the writing tasks. Similarly,
little  information  is  known  about  the
students  themselves.  Nonetheless,  we  feel
that the sample is sufficiently large to allow
us  to  draw  conclusions  on  the  basis  of  the
learner corpus.  
 
A  methodological  challenge  is  the  fact  that
learners  of  course  only  used  one  of  the
textbooks  in  their  schools,  but  the  textbook
corpus  is  an  average  of  all  four  state-selected  books.  In  other  words,  we  are  not
comparing  individual  students’  writing
against  the  specific  textbook  they  learned
from.  Although  it  would  have  been
interesting  to  make  direct  comparisons,  our
data  did  not  allow  us  to  do  this  as  the
original  learner  corpus  did  not  include  this
information.  Nonetheless,  we  feel  that  this
issue  is  not  of  major  concern  given  the  fact
that  the  learner  corpus  includes  data  from
students  who  used  all  four  books;  in  other
words, the average of all students’ modal
usage is compared to the average occurrence
of the modals in all four books.  
 
Finally,  the  results  allow  us  to  draw  a
number  of  conclusions,  but  do  not  allow  us
to  definitely  explain  why  we  found  these
results  in  the  first  place.  For  example,  why
was students’ performance so poor on the
writing tasks? Although we have made some
comparisons with the results from a previous
study  (Khojasteh  &  Kafipour,  2012)  which
may  give  some  of  the  possible  reasons,  a
more in-depth analysis of learners’ exposure
to  the  modals,  not  just  from  the  textbooks,
but  also  in  class  and  beyond  their  schools,
would be beneficial. We hope our study will
be  a  starting  point  for  such  further  research
in  future.  Furthermore,  to  date,  the  focus  of
most  pedagogic  corpus-based  research  has
been  either  on  international  type  of
textbooks  (e.g.  Meunier  &  Gouverneur,
2009),  or  on  national  textbooks  mainly  in
EFL  contexts  such  as  Germany  (Romer,
2004), Hong Kong (Lam, 2010) and Taiwan
(Wang  &  Good,  2007),  to  name  a  few.
Surprisingly,  however,  English  for  General
Purposes  in  Iran  has  been  the  exception  to
this  rule.  Aimed  at  filling  the  existing  gap,
this  study  suggests  doing  corpus-based
studies on tertiary Iranian English textbooks
in order to provide better picture of the ways
in which not only modal auxiliaries but also
other  grammatical  structures  are  treated  in
each learning cycle in the Iranian context.

Arshad,  A. Hassan,  Mukundan,  F.
Kamarudin, G. Rahman, Sh., Rashid,
J.  &  Edwin,  M.  (2002).  The  English
of  Malaysian  school
students (EMAS)  corpus.  Serdang:
Universiti PutraMalaysia.
Barbieri,  F.,  &  Eckhardt,  S.  (2007).
Applying  corpus-based  findings  to
form-focused  instruction:  The  case
of  reported  speech.  Language
Teaching Research, 1(3), 319–346.
Biber, D., Johansson, S., Leech, G., Conrad,
S.  &  Finegan,  E.  (1999).  Longman
grammar  of  spoken  and  written
English. Harlow: Pearson Education.
Breen,  M.  (1989).  The  evaluation  cycle  for
language  learning  tasks.  In  P.  Rea-Dickins  &  K.  Germaine  (Eds.),

Evaluation.  Oxford:  Oxford
University Press.  
Campoy,  M.C.,  Belles-Fortuno,  B.  &  Gea-Valor,  M.  L.  (2010).  Corpus-based
approaches  to  English  language
teaching. London: Continuum.
Celce-Murcia,  M.  &  Larsen-Freeman,  D.
(1999).  The  Grammar  book:  an
ESL/EFL teacher’s course.  Boston:
Heinle & Heinle.  
Coates,  J.  (1983).  The  semantics  of  the
modal  auxiliaries.  London:  Croom
Helm.  
Conrad,  S.  M.  (2000).  Will  corpus
linguistics  revolutionize  grammar
teaching in the 21st century? TESOL
Quarterly, 34 (3), 548–60.
Decapua, A. (2008). Grammar for teachers:
A  guide  to  American  English  for
native  and  non-Native  speakers.
New Rochelle: Springer.
De Silva, E. (1981). Forms and functions in
Malaysian  English:  the  case  of
modals. SARE, 3, 11-23.
Ellis, R. (1997). The empirical evaluation of
language  teaching  materials.  ELT
Journal, 5 (1), 36-42.  
Gabrielatos,  C.  (2005).  Corpora  and
language  teaching:  Just  a  fling,  or
wedding bells? TESL-EJ, 8 (4), 1-37.
Goldschneider,  J.  M.  &  Dekeyser,  R.  M.  
(2001).  Explaining  the  natural  order
of  L2  morpheme  acquisition  in
English: A meta-analysis of multiple
determinants.  Language  Learning,
51 (1), 1-50.
Harwood, N. (2005). What do we want EAP
teaching  materials  for?  Journal  of
English  for  Academic  Purposes,  4
(2), 149–161.
Hughes,  R.  &  Heah,  C.  (1993).  Common
errors  in  English:  grammar
exercises  for  Malaysians.  Shah
Alam: FajarBakti.  
Hunston,  S.  (2002).  Corpora  in  applied
linguistics.  Cambridge,  England:
Cambridge University Press.
Howarth, P. (1998). Phraseology and second
language  proficiency.  Applied
linguistics, 19(1), 24-44.
Hyland,  K.  (1994).  Hedging  in  academic
writing  and  EAP  textbooks.  English
for  Specific  Purposes,  13  (3),  239–
56.
Kennedy,  G.  (1998).  An  introduction  to
corpus  linguistics.  London:
Longman Publishing.
Kennedy,  G.  (2002).  Variation  in  the
distribution  of  modal  verbs  in  the
British  National  Corpus.  In  R.
Reppen,  S.  Fitzmaurica  &  D.  Biber
(Eds.),  Using  corpora  to  explore
linguistic  variation  (pp.  73-90).
Amsterdam: John Benjamins.
Khojasteh,  L.  &  Kafipour,  R.  (2012).  Non-empirically  based  teaching  materials
can be positively misleading: A case
of  modal  auxiliary  verbs  in
Malaysian  English  language
textbooks.  Journal  of  English
Language Teaching, 5 (3), 62-72.
Lam,  P.  (2010).  Discourse  particles  in
corpus  data  and  textbooks:  The  case
of  well.  Applied  Linguistics,  31(2),
260-281.
Lawson,  A.  (2001).  Rethinking  French
grammar  for  pedagogy:  The
contribution  of  French  corpora.  In
R.C. Simpson, & J.M. Swales (Eds.),
Corpus linguistics in North America:
Selections  from  the  1999  symposium
(pp.  179–194).  Ann  Arbor,  MI:  The
University of Michigan Press.
Manaf,  U.  (2007).  The  use  of  modals  in
Malaysian  ESL  learners’  writing.
(Unpublished  Doctoral  Thesis).
Serdang: Universiti Putra Malaysia.  
McEnery,  T.,  &  Kifle,  N.  A.  (2002).
Epistemic modality in argumentative
essays of second-language writers. In
J.  Flowerdew  (Ed.),  Academic
discourse  (pp.  182-195).  Harlow:

Longman.
Meunier,  F.  &  Gouverneur,  C.  (2009).  New
types of corpora for new educational
challenges.  In  K.  Aijmer  (Ed.),
Corpora  and  language  teaching
(pp. 179-201).  Amsterdam: John
Benjamins.  
Mindt, D. (1995). An empirical  grammar of
the  English  verb:  Modal  verbs.
Berlin: Cornelsen.
Mukundan,  J.  &  Anealka,  A.  H.  (2007).  A
forensic  study  of  vocabulary  load
and  distribution  in  five  Malaysian
Secondary School Textbooks (Forms
1-5).  Pertanika  Journal  of  Social
Science  and  Humanities,  15  (2),  59-74.
Mukundan,  J.  &  Khojasteh,  L.  (2011).
Modal  auxiliary  verbs  in  prescribed
Malaysian  English  textbooks.
Journal  of  English  Language
Teaching, 4 (1), 79-89.
Mukundan,  J.  &  Roslim,  N.  (2009).
Textbook  representation  of
prepositions.  English  language
teaching, 2 (4), 123-130.
Nedkova,  M.  (2000).  Evaluation.  In  M.
Byram  (Ed.),  Routledge
encyclopedia  of  language  teaching.
London and New York: Routledge.
Nordberg,  T.  (2010).  Modality  as  portrayed
in  Finish  upper  secondary  school
EFL  textbooks.  (Unpublished
Master’s  Thesis).  University  of
Helsinki.  
Quirk,  R.  S.,  Greenbaum,  S.,  Leech,  G.,  &
Svartvik, J. (1985). A comprehensive
grammar  of  the  English  language.
Harlow: Longman.
Romer, U. (2004). A corpus-driven approach
to  modal  auxiliaries  and  their
didactics.  In J. Sinclair (Ed.), How to
use  corpora  in  language  teaching
(pp.  185-199).  Amsterdam:  John
Benjamins.
Romer,  U.  (2005).  Progressives,  patterns,
pedagogy.  A  corpus-driven  approach
to  English  progressive  forms,
functions,  contexts  and  didactics.
Amsterdam: John Benjamins.
Romer,  U.  (2010).  Using  general  and
specialized  corpora  in  English
language  teaching:  Past,  Present  and
Future.  In  M.C.  Campoy,  B.  Belles-Fortuno  &  M.  L.  Gea-Valor  (Eds.),
Corpus-based  approaches  to  English
language  teaching  (pp.  18-38).
London: Continuum.
Rosli, T. & Edwin, M. (1989). Error analysis
of  form  four  English  compositions.
The English Teacher, XVIII, 110-124.
Retrieved  from
http://www.melta.org.my/ET/1989/m
ain9.html
Stubbs, M. (1996). Text and corpus analysis:
Computer  assisted  studies  of
language  and  culture.  Blackwell:
Cambridge.
Thornbury,  S.  (2004).  How  to  teach
grammar.  Malaysia:  Pearson
Education Limited.
Tomlinson,  B.  (1998).  Materials
development  in  language  teaching.
Cambridge:  Cambridge  University
Press.
Wang,  J.  &  Good,  R.  (2007).  The  repetition
of  collocations  in  EFL  textbooks:  A
corpus  study.  Paper  presented  at  the
fourth  corpus  linguistics  conference
held at the University of Birmingham,
27-30.
Widdowson,  H.  G.  (1990).  Aspects  of
language teaching. Oxford: OUP.  
Willis,  J.  D.  (1993).  Grammar  and  lexis:
some  pedagogical  implications.  In  J.
M.  Sinclair,  G.  Fox  &  M.  Hoey
(Eds.), Techniques of description (pp.
83-93). New York, NY: Routledge.
Wong,  I.  (1983).  Simplification  features  in
the  structure  of  colloquial  Malaysian
English.  Singapore:  Singapore
University.