Towards a definition of intake in second language acquisition

Introduction
This  article  looks  at  input,  output,  and
interaction  as  sources  of  language  data  for
second  language  learners  to  draw  on.  The
second  part  of  the  article  looks  at  how
learners  extract  linguistic  information  from
this data and considers a crucial stage in the
input-to-acquisition process, namely intake.  
 
The study of input
The role of language input is acknowledged
by researchers with different perspectives on
second  language  acquisition.  All  agree  that
some  form  of  input  is  needed  for  language
learning  to  occur (cf.  Gass,  1997).  How
input  is  related  to  learning  is  an  area  of

contention:  ‘it  is  uncontroversial  that  a
learner  needs  input  in  order  to  acquire  a
language […]. Unfortunately the consensus
stops  about  there.  How  much  input  is
necessary? What kind of input? Under what
conditions  need  it  be  provided?’  (Gregg,
2001, p.167).  
 
What  is  input?  Sharwood  Smith  defines
input  as  ‘the  potentially  processable
language  data  which  are  made  available  by
chance or by design, to the language learner’
(1993, p. 167) and thus emphasises the point
that  input  need  not  be  used  (processed),  but
potentially  can  be.  Also,  the  intent  with
which  the  language  data  are  presented  or
sought  out,  does  not  determine  what  is  and
what  is  not  input.  Carroll  (2001)  makes  a
distinction  between  stimuli  as  ‘all  […]
observable  instantiations  of  the  second language’

  (p.  8)  and  input,  which  she
reserves  for  stimuli  that  have  entered  the
brain.  To  her,  input  is  not  physical
information  (sounds,  visual  data),  but  a
mental  representation,  available  for  internal
processors to use.  
 
Not  all  definitions  of  input  are  quite  as
neutral  as  the  above.  Gregg’s  definition
(2001)  of  input is ‘information that is fed
into  an  input-output  device;  the  output  is
grammar’, (p. 167). Gregg takes a mentalist
view of language learning that sees language
as  innate  and  its  development  as  set  or
‘triggered’ by the language data available to
the learner. Input here is seen as ‘evidence’.
Several  researchers  see  the  role  of  input  as
the  main  source  of  information  from  which
language  develops.  Krashen  (1982,  1985)
makes  a  distinction  between  input  in  a
general sense and comprehensible input, i.e.
language  data  that  are  understandable  to  a
learner. Comprehensible input in Krashen’s
view  is  sufficient  for  acquisition  to  take
place.  When  talking  about  input,  authors
such  as  Krashen  (but  see  also  Faerch,
Haastrup,  &  Phillipson,  1984;  VanPatten
1996)  thus  often  take  it  to  mean  a  specific
kind  of  (qualitatively  different)  language
data.  
 
Beebe  (1985)  argues  that  what  constitutes
input  is  determined  to  a  large  extent  by  the
learner:  
 
Studies  of  input  in  second  language
acquisition  must  view  non-native  speakers
not  simply  as  passive  recipients  of
comprehensible  or  incomprehensible  input
from  native  speakers,  but  as  active
participants  in  choosing  the  target  language
                        
  This  appears  to  differ  somewhat  from  the
commonly  held  view  in  cognitive  psychology  that
‘stimuli’  are  ‘anything  in  the  environment  we
respond to’.
models  they  prefer  and  thus  acquiring  'the
right stuff' according to their values. In other
words,  learners  have  'input  preferences'  (or
'model  preferences')  in  the  sense  that  they
consciously  or  unconsciously  choose  to
attend to some target language models rather
than others. (p. 404)  
 
Another  way  of  saying  this  is  that  input  is
what  learners  pay  attention  to  in  the
language.  This  appears  similar  to  what  Sato
&  Jacobs  (1992)  write:  ‘in  the  present
perspective, input is viewed as the object of
the  learner's  attention’  (p.  269).  This  is
clearly  different  from  Sharwood  Smith’s
‘language data that are potentially  available
for  processing’  in  that  it  combines
information with the process of selecting it.  
 
The  different  approaches  to  (research  into)
input  were  described  by  Schachter  (1986)
who  identified  1)  a  data-oriented  approach
which  observes,  records,  and  transcribes
actual learner - native speaker interaction, 2)
a  language-model  approach  which  attempts
to  describe  language  itself  (attempting  to
answer  questions  such  as  ‘what  does
knowledge of a language consist of?), and 3)
a  processing-model  approach  which  focuses
primarily  on  the  processing  learners  engage
in  when  interacting  with  the  second
language.  Although  these  three  approaches
are  all  valid  ways  of  investigating  second
language  acquisition,  most  researchers
would  probably  agree  with  the  basic
meaning  of  the  word  input  as  ‘linguistic
data’. This is also the working definition I
adopt here.  
 
Positive and negative evidence
Research  has  shown  that  the  input  learners
receive  does  not  provide  all  the  information
they need to learn a language. This has been
referred  to  as  the  ‘logical  problem’  of
language  acquisition  (Bley-Vroman  1989).

Chomsky  (1959,  1965),  most  prominently,
suggested  that  humans  make  use  of  internal
building blocks (‘universal grammar’), and
that the input they receive serves as evidence
of  what  is  and  what  is  not  possible  in  the
language that is learned. Language is thus to
a  large  extent  innate  and  the  input  triggers
its  development.  In  second  language
acquisition  research  the  most  vocal
subscriber  to  this  view  is  Krashen  who  has
argued for the importance of comprehensible
input as a necessary and sufficient condition
for  acquisition  to  take  place  (1982,  1985).
Language  develops  on  the  basis  of  positive
evidence,  i.e.  examples  of  actual  language
use.  
 
Others,  however,  have  argued  that  such
comprehensible  input  alone  is  not  sufficient
for learning to take place; it can be used for
comprehension  only  (e.g.  Faerch  &  Kasper,
1980)  without  affecting  change  in  the
learner’s  interlanguage.  ‘Paradoxically,
comprehensible  input  may  actually  inhibit
learning  on  occasion,  because  it  is  often
possible  to  understand  a  message  without
understanding  all  the  structures  and  lexical
items  in  the  language  encoding  it,  and
without  being  aware  of  not  understanding
them  at  all’  (Long,  1996,  p.  425).
Chomskyan  researchers  have  proposed  that
in  addition  to  universal  rules,  learners  are
also  endowed  with  a  set  of  language
constraints,  limiting  the  number  of
grammatical  possibilities.  However,  if
participants make use only of the input they
receive  and  the  positive  evidence  contained
in  it,  then  how  is  it  possible  that  they  make
mistakes  they  themselves  have  never
encountered before?  
 
L.  White  found  (1989)  that  learners’
language  behaviour  supports  the  transfer
hypothesis. This states that second language
learners  primarily  make  decisions  of
acceptability  on  the  basis  of  their  first
language.  When  the  first  language  (e.g.
English)  is  more  restrictive  (for  example  in
requiring adverb adjacency) than the second
(e.g.  French),  learners  will  limit  themselves
to the more restrictive use found in their first
language.  To  avoid  this  type  of  transfer
White  has  argued  that  negative  evidence  is
needed  (e.g.  L.  White,  1989,  1991;  see  also
Bley-Vroman,  1986).  Negative  evidence  is
defined as ‘the type of information that is
provided  to  learners  concerning  the
incorrectness of an utterance’ (Gass, 2003,
p. 225) and may help by drawing learners’
attention  to  language  form.  Trahey  writes:
‘exposure to positive evidence can lead to
changes  in  linguistic  competence  when  the
structure  to  be  acquired  is  readily  available
in  the  input.  However  when  the  problem  is
one  of  'unacquiring',  as  in  the  case  of
SVAO,  positive  evidence  appears  not  to  be
sufficient’ (1996, p.134).  
 
The  role  of  negative  evidence,  however,  is
disputed.  Some  have  claimed  that  it  is  not
clear  that  it  occurs  at  all  outside  instruction
(Pinker,  1989).  However,  several  studies
have  shown  it  to  exist  in  both  first  (Farrar,
1992),  and  second  language  acquisition
(Oliver,  1995).  Mackey,  Oliver  &  Leeman
(2003)  investigated  48  native  speaker/non-native  speaker  and  non-native  speaker/non-native  speaker  dyads  consisting  of  both
adults  and  children  while  completing  an
information  gap  task.  Although  they  found
differences  between  the  various  groups
(with  native  speakers  notably  providing
more feedback than non-native speakers), in
all  cases  at  least  30%  of  errors  resulted  in
feedback  and  between  25%  and  41%  of
these resulted in modified output.  
 
It  has,  however,  been  argued  that  even  if
negative  evidence  occurs,  it  is  not  relevant
to learning. Schwartz (1993) writes:  

If  there  were  a  translation  algorithm  that
could  take  the  knowledge  that  results  from
being  told  “This  is  not  a  grammatical
sentence in this language” and convert it
into  information  that  is  in  a  form  the
processes  in  the  language  module  could
make use of, then ND [negative data] could
be  usable.  It  appears,  however,  that  there
exists no such mechanism’ (p. 158).  
 
Schwartz  does  not  argue  that  negative
evidence  cannot  result  in  learning,  but  says
that  it  does  not  result  in  the  type  of
knowledge  underlying  second  language
proficiency.  However,  evidence  exists  that
learners  can  in  fact  make  use  of  negative
data.  An  often  cited  study  in  this  respect  is
one  by  Bohannon  &  Stanowicz  (1988)
which  showed  that  children  do  benefit  from
negative  evidence  in  learning  their  first
language
.  Also  for  second  language
learning  there  appears  to  be  evidence
showing a facilitative effect on learning (see
below),  but  it  has  been  pointed  out  that  in
interpreting  the  results  of  such  studies  one
has  to  be  careful  to  distinguish  between
immediate  effects  (uptake)  and  delayed
effects (learning) (cf. Birdsong, 1989).    
 
Negative  evidence  can  be  provided  in
different  ways.  Long  (1996)  distinguishes
between  explicit  negative  evidence  (such  as
error  correction),  and  implicit  negative
evidence  (such  as  when  one  interlocutor
shows  not  to  have  understood  the  other,  or
reformulates  an  utterance  without  it
interrupting the flow of the conversation (i.e.
recasts).  Negative  evidence  can  perform
different  functions.  Firstly,  it  can  help
learners ‘notice the gap’ between the input
and  their  own  output.  R.  Ellis  (1995)
  Although  their  study  has  been  criticised  for  only
investigating  one  particular  socio-economic  stratum
of  society,  thus  questioning  its  generalisability  to
other groups of learners.
describes  this  as  making  a  cognitive
comparison.  As  a  result  of  realising  this
gap,  participants  can  then  attempt  to
reformulate  their  utterance  or  store
information  about  that  aspect  of  the
language.  It  may  also  result  in  quite  sudden
shifts  in  the  learner’s  interlanguage,  for
example  when  it  leads  to  a  realisation  that
certain  forms  cannot  be  used  in  the  target
language at all. Secondly, negative evidence
can also increase learners’ awareness of the
target  language  in  a  broader  sense.  By
drawing  attention  to  what  is  not  possible  in
the  target  language,  negative  feedback
necessarily  contrasts  different  linguistic
forms and encourages learners to understand
the differences (Schmidt, 1990).  
 
Research  on  various  types  of  negative
evidence  has  shown  facilitative  effects  on
learning.  Carroll  &  Swain  (1993)  for
example,  investigated  instruction  of  English
dative alternation combined with one of four
types  of  feedback.  Learners  either  1)
received  metalinguistic  information  about
their  responses,  2)  were  told  when  their
answer was wrong, 3) received a recast with
a  model  answer,  or  4)  were  asked  if  they
were  sure  about  their  answer.  The  results
showed  that  all  four  groups  improved
performance,  but  that  the  metalinguistic
group did best.  
 
Han  (2002)  investigated  the  effects  of
recasts  for  tense  consistency  provided
during  eight  sessions  over  a  period  of  two
months  in  written  and  oral  narration  tasks.
Feedback  was  given  individually,  and  was
consistent;  all  errors  were  recasted.  Using  a
pretest-posttest-delayed  posttest  design  she
  Although  this  term  appears  broader  in  that  it  need
not involve a ‘gap’ in the sense of an inability to
express  oneself,  i.e.  the  comparison  could  be  made
between two sources of input. Swain (1998) refers to
this as noticing a ‘hole’.
 
found recasts to lead to increased awareness
(and much greater awareness than in the no-recasts  group),  and  improved  tense
consistency both on the production tasks and
on  an  immediate  and  delayed  posttest
Recasts  were  given  in  individual  settings,
and  Han  suggests  that  this  may  have
contributed to the results.
 
A  number  of  other  studies  have  also  found
benefits  for  negative  feedback  on  second
language  learning  (e.g.  Pica,  1994;  Mackey
& Philp, 1998), however a small number did
not  (e.g.  Sanz  &  Morgan-Short,  2004).
Mackey,  Oliver  &  Leeman (2003) point out
that  a  range  of  variables  (types  of  dyads,
type  of  interaction,  age,  etc)  affects  the
provision,  uptake  and  learning  effects  of
negative  evidence  and  that  more  research
needs  to  be  done  to  isolate  and  investigate
those variables.

Input and learnability  
For  some  researchers  the  primary  role  of
input is to trigger the development of innate
knowledge  of  language  (Krashen,  1981;
Schwartz,  1993;  L.  White,  1989).
Pienemann (1984, 1985, 1988, 1989, 1998),
sharing this viewpoint, argues that language
(including  a  learner’s  second  language)
develops  in  a  predictable,  and  pre-determined  way.  His  language  processing
theory  posits  a  number  of  determinants  of
the  relative  difficulty  and  acquisition  order
of  various  linguistic  features  in  the  input.
The  first  of  these  is  psychological
complexity and refers to the extent to which
‘the language  learner  must  re-order  and  re-arrange  linguistic  material  in  the  process  of
mapping  underlying  semantics  onto  surface

  It  has  to  be  pointed  out  that  the  participants  in  the
study  were  at  an  upper-intermediate  level  and
reported high motivation for learning English.  
 
forms’ (1988, p. 89). The second is saliency;
items  that  are  more  salient  are  easier  and
will  be  acquired  earlier.  Items  are  more
salient  if  they  come  in  sentence  initial  or
sentence  final  position.  The  third
determinant  is  the  distance  between  an  item
that  triggers  a  transformation  and  the  place
in  the  sentence  where  the  transformation  is
effected. The greater the distance, the harder
the  item.  The  most  important  premise  of
processing  theory,  however,  is  that  learners
must go through various “levels” which they
cannot  skip.  At  the  first  level  learners  are
unable to organise lexical material into word
classes  or  categories  and  cannot  identify
where the information in the sentence is. At
the second level they are partially able to do
so  but  generally  only  on  the  basis  of  salient
sentence  items  (e.g.  sentence  first  or
sentence  final).  At  the  third  level  learners
are  able  to  organise  some  of  the  lexical
material into categories but transformation is
limited  to  either  word  or  constituent  initial
or final position. At the fourth level they are
able to recognise all elements in a sentence.
The  fifth  level,  S-procedures,  deals  with  an
exchange  of  information  between  internal
constituents. Pienemann  sees these levels  as
an  ‘implicational  hierarchy  of  processing
procedures’. Implicational here means that
each  procedure  necessarily  builds  on  the
other;  ‘the  learner  cannot  acquire  what
he/she  cannot  process’  (p.  87).  Again,
language  development  is  seen  as
predetermined.  
 
Others see a more active role for the learner
and  more  possible  variation  in  the
acquisition  of  various  features  in  the  input.
Goldschneider  &  DeKeyser  conducted  a
metastudy  (2001)  of  second  language
research  on  the  acquisition  of  morphemes.
Using  multiple  regression  analysis  they
identified  five  determinants  explaining  a
large portion of the variation in performance
 
found in the individual studies. These were:
1)  perceptual  saliency,  or  how  easy  it  is  to
hear  or  perceive  a  given  structure

Semantic  complexity  or  how  many
meanings are expressed by a certain form. 3)
Morphophonological  regularity  as  the
degree  to  which  a  grammatical  feature  is
affected by its phonological environment. 4)
syntactic category as the difference between
lexical and functional and free versus bound
items. 5) frequency. All of these factors will,
to some extent, impact on the order in which
they are acquired
 
N.  Ellis  (2002a,  2002b)  specifically
investigated  point  5)  above  and  argued  that
frequency  largely  determines  acquisition.
Second  language  acquisition,  in  his  view,  is
to  a  large  extent  exemplar  based  and  not
based on abstract rules or structures (2002a):
‘the acquisition of grammar is the piecemeal
learning of many thousands of constructions
and  the  frequency-biased  abstractions  of
regularities within them’ (p. 168). He sees
second  language  learning  as  a  form  of
implicit  learning,  dependent  on  input.
However,  he  points  out  that  this  may  not
apply  to  initial  registration  of  language
forms,  which  may  require  attention  and
conscious  noticing.  What  gets  registered
initially gets grouped later, by ‘unconscious
processes of association to form larger units
        
  Saliency  is  a  term  also  used  by  other  researchers
with sometimes different or additional meanings. For
example Bardovi-Harlig (1987) defines it ‘in terms of
availability  of  data’ (p. 401), (which appears to be
similar  to  frequency)  and  uses  it  to  explain  findings
of  acquisition  order  (preposition  stranding  before
preposition  pied  piping)  which  appear  to  go  against
the order predicted by  universal  grammar (unmarked
construction  such  as  preposition  stranding  are
acquired first).  

 Goldschneider & DeKeyser point out that ‘we have
argued  that  these  five  factors  are  not  a  completely
heterogeneous  set,  but  can  all  be  seen  as  aspects  of
salience in a broad sense of the word’ (p. 37).
that  are  henceforth  used  in  pattern
recognition’ (p. 174).  
 
Other  aspects  of  the  input  have  been
identified as affecting noticing and learning.
VanPatten  (1985,  1990,  1996;  VanPatten  &
Cadierno,  1993)  argues  that  learners  assign
varying degrees of “communicative value”
to  different  aspects  of  the  input,  defined  as
‘the  meaning  that  a  form  contributes  to
overall sentence meaning’ (2001, p. 759)
and ‘the relative contribution a form makes
to  the  referential  meaning  of  an  utterance
[…] based on the presence or absence of two
features:  inherent  semantic  value  and
redundancy  within  the  sentence-utterance’
(1996, p. 24). The communicative value of a
form  is  greater  if  its  semantic  value  is
greater and if it is not redundant. In his 1990
study,  VanPatten  asked  subjects  to  mark  all
the  occasions  they  heard  the  Spanish  article
la,  the  third-person  plural  verb  morpheme  -n, or the word inflación. His results showed
that  attending  to  the  verb  morpheme  or  the
definite  article  resulted  in  lower
comprehension  levels.  He  attributed  that  to
their  lower  communicative  value  and
suggested that learners first look for content
words  in  the  input.  If  sufficient  attentional
resources  are  available  grammatical  forms
may then be processed. VanPatten’s concept
of  communicative  value  has  come  under
heavy  criticism.  DeKeyser,  Salaberry,
Robinson  and  Harrington  (2002)  argue  that
both  form  and  meaning  can  be  processed
simultaneously  and  that  to  expect  the
internal  parser  to  scan  all  content  for
communicative  value  first  while  saving
certain  parts  of  the  input  to  be  processed
later  is  not  congruent  with  current  thinking
and  findings  from  research  into  sentence
processing.  

Output and interaction
The  roles  of  output  and  interaction  on
acquisition  have  received  considerable
attention since the publication of Hatch’s
early  papers  (1978a,  1978b)  in  which  she
drew  attention  to  their  potential  benefits  to
learning.  For  some  (e.g.  Krashen,  1981)  the
role  of  interaction  is  predominantly  that  it
provides learners with comprehensible input
(i.e.  input  that  is  attuned  to  their
developmental level). For others, interaction
and  the  output  it  generates  directly
contribute  to  learning.  Long  is  well-known
for  his  interaction  hypothesis  (1981,  1983,
1996)  which  emphasises  the  crucial  role  of
the  process  of  negotiation  on  learning.
Negotiation,  or  the  ‘modification  and
restructuring of interaction that occurs when
learners  and  their  interlocutors  anticipate,
perceive,  or  experience  difficulties  in
message comprehensibility’ (Pica, 1994, p.
493),  has  a  number  of  beneficial  effects.
Firstly,  it  aids  in  increasing  understanding,
and  thus  results  in  the  learner  receiving
more,  and  more  comprehensible  input,
necessary  for  learning  to  take  place.
Negotiation  exchanges  are  said  to  result  in
‘denser’  than  average  speech,  with  more
repetitions,  reformulations,  expansions,
extra stress, and a range of other features, all
of which increase frequency and saliency of
aspects  of  the  input.  Learners  are  also  more
likely to benefit from this enhanced input as
they  have  at  least  partial  control  over  the
semantic  content  of  the  interaction  and  can
thus  free  up  attentional  resources  to  pay
attention  to  form  in  the  input.  They  are
likely  to  be  alert  as  they  try  to  get  their
meaning  across  and  as  a  problem  in  the
communication occurs.  
 
Secondly,  interaction  takes  place  in  a
context  that  is  meaningful  to  the
interlocutors.  From  this  context  learners
derive a degree of support which helps them
in  their  understanding  as  well  as  in  getting
their  meaning  across.  They  also  derive
support  from  their  conversation  partners
who  may  supply  words,  or  restate
utterances,  and  in  so  doing  provide
scaffolding,  allowing  learners  to  express
meaning they would otherwise be unable to.  
 
Next,  interaction  can  also  lead  to  the
occurrence  of  negative  feedback  (one  form
of  negative  evidence,  see  above),  i.e.
information  about  what  is  and  is  not
understandable and/or correct in a speaker’s
output:  
 
Negative feedback is generally facilitative of
L2  acquisition,  and  necessary  for  the
acquisition of specifiable L2 structures (such
as  the  English  adverb-placement  example
for  French  speakers)  for  which  positive
evidence  will  be  insufficient.  A  mechanism
is  posited  whereby,  while  correct-form
meaning  associations  are  strengthened  both
by  positive  evidence  and  that  negative
feedback  that  contains  positive  evidence,
incorrect  associates  are  weakened  and  in
some  cases  ultimately  relinquished
altogether  as  a  result  of  both  negative
evidence  and  prolonged  absence  of  support
in the input. (Long, 1996, p. 430).  
 
Gass  (1997;  Gass  &  Varonis,  1994)  has
argued  that  since  such  negative  feedback  is
situated  in  a  communicative  context  and  is
thus linked to actual communicative goals, it
is more likely to be usable to the learner.  
 
However,  it  is  not  entirely  clear  how  much
negotiation  of  meaning  takes  place,  with
some  claiming  that  it  is  substantial  (Pica,
1994),  and  others  that  it  is  not  (Skehan,
1989).  It  has  also  been  pointed  out  that
although  interaction  can  have  beneficial
effects, conversational success in itself does
 Towards a definition of intake  
not  necessarily  result  in  learning  (Faerch  &
Kasper, 1980).  
 
Swain  has  argued  for  the  important  role  of
learner production in learning. Her “output
hypothesis”  (1985)  developed  from
observations  of  Canadian  immersion
students  who,  despite  years  of  receiving
exposure  to  the  second  language,  did  not
fully  develop  in  particular  certain
grammatical  aspects  of  the  target  language.
Swain  found  that  the  immersion  classes
were characterised by a lack of opportunities
for output and afforded few opportunities for
“pushed output”, i.e. output that required
them to ‘stretch their  interlanguage’. Many
students  were  able  to  get  by  using
communication  strategies  to  get  their
meaning  across  and  were  never  challenged
to  further  develop  their  language.  Swain
suggested  that  by  requiring  learners  to
produce  comprehensible  output,  they  would
be  pushed  to  be  more  accurate  and  to  pay
attention  to  both  form  and  meaning,  and  in
so  doing  move  from  semantic  to  syntactic
processing.  In  addition,  Swain  (1998)
suggested  that  output  would  1)  induce
noticing,  2)  allow  for  hypothesis  formation
and  testing  (see  also  R.  Ellis  &  He,  1999;
Pica  1988),  and  3)  give  opportunities  for
meta-talk.  The  effect  of  output  on  1)
noticing,  was  investigated  in  another  article
(Swain  &  Lapkin,  1995)  in  the  context  of  a
writing task with a think-aloud protocol. The
authors found that learners do become aware
of  problems  in  their  writing  and  engage  in
strategic  thought  processes  to  solve  those
problems.  Swain  (1985)  has  suggested  that
output  can  also  serve  to  help  with
developing  automaticity  (referred  to  as  the
fluency function of output) and this seems to
have been corroborated by research showing
that  when  producing  the  language,
connections  in  the  brain  are  strengthened,
aiding  the  process  of  automatisation  (cf.  de
Bot,  1992,  1996).  Izumi,  in  several  studies,
investigated  the  effect  of  output  on  noticing
(2002,  2003;  Izumi,  Bigelow,  Fujiwara,  &
Fearnow,  1999).  He  suggested  (2002)  that
Swain & Lapkin’s study may have found an
effect for the noticing induced by the output
because  their  measurement  was  immediate
task performance. To investigate if there was
also  a  delayed  effect  he  made  use  of  a
written  reconstruction  task  to  measure
noticing  of  English  relative  clauses,
followed  by  a  series  of  posttests.  He  found
that in comparison with a control group who
received  a  receptive  (meaning-focused)  task
only,  output  did  have  an  effect  on  both
noticing and learning.     
 
Benefits have been found for non-interactive
language production. In the general learning
domain,  Baddely  (1990)  writes  about  the
effects of producing an item: ‘the  act  of
successfully  recalling  an  item  increases  the
chance  that  that  item  will  be  remembered.
This is not simply because it acts as another
learning  trial,  since  recalling  the  item  leads
to better retention than presenting it again; it
appears that the retrieval route to that item is
in  some  way  strengthened  by  being
successfully used’ (p. 156).  
 
In  the  second  language  acquisition  domain
N.  Ellis  &  Sinclair  (1996)  found  that
subjects  encouraged  to  rehearse  foreign
language  utterances  were  better  than  both
silent  controls  and  subjects  prevented  from
rehearsal  by  articulatory  suppression  at  a)
comprehension  and  translation,  b)  explicit
metalinguistic  knowledge  of  the  detailed
content  of  grammatical  regularities,  c)
acquisition of the foreign language forms of
words  and  phrases,  d)  accuracy  in
pronunciation,  e)  some  aspects  of
productive,  but  not  receptive,  grammatical
fluency and accuracy. 
 
Although not rejecting  a role for output and
interaction,  VanPatten  (1996,  2002;
VanPatten  &  Cadierno,  1993)  has  argued
that  the  role  of  input  and  input  processing
are  crucial  for  language  development.
VanPatten  &  Cadierno  (1993)  compared
traditional  form-focused  instruction  (rule
presentation followed by output practice) on
direct  object  pronouns  with  ‘Processing
Instruction’ which ‘involves explanation and
practice/experience  processing  input  data,
taking  learner  strategies  in  input  processing
as  the  starting  point  for  determining  what
explicit  instruction  should  look  like’  (p.
225).  In  the  study,  participants  receiving
processing  instruction  were  given  an
explanation of the target structure as well as
‘explanations of important points to keep in
mind  about  the  position  of  object  pronouns
of Spanish’ (p. 231). This was followed by a
reading  or  listening  exercise  in  which
participants  had  to  demonstrate
understanding  of  the  structure,  and  an
activity  in  which  they  had  to  respond  to  the
content  of  the  input.  VanPatten  &  Cadierno
found that input processing led to significant
gains  for  both  comprehension  and  output
skills,  compared  with  a  significant
improvement  for  output  skills  only  for  the
traditional  group.  VanPatten  argued  that
second  language  instruction  should  include
an  increased  focus  on  improving  the  way
learners  process  the  input,  as  opposed  to
focusing  primarily  on  output  practice.
However,  in  a  replication  study,  DeKeyser
&  Sokalski  (1996;  see  also  DeKeyser  et  al.,
2002),  found  that  the  relative  effectiveness
of production versus comprehension practice
depends  on  the  complexity  of  the  target
structure  and  on  the  delay  between  practice
and testing; there may an immediate, but no
lasting  effect  for  comprehension  practice.
Allen  (2000)  investigated  acquisition  of
French  causatives  and  compared  the  effects
of  processing  instruction  and  production-based  instruction.  She  did  not  find  an
advantage  for  the  processing  instruction
group  compared  with  the  production  group,
but both groups did improve compared with
a control group. Other studies have reported
similar  findings  (Erlam,  2003;  Salaberry,
1997).  
 
Despite  VanPatten’s  suggestions,  there
appears to be evidence of a facilitative effect
for  output  and  interaction
.  In  sum,  output
and interaction can:
-  provide additional input
-  result in comprehensible input which
impacts on learning  
-  enhance  fluency  by  allowing
participants  to  produce  the  target
language
-  facilitate form-meaning connections  
-  result  in  negotiation  of  meaning
which in turn can raise  awareness of
the target language
-  provide  opportunities  for  negative
feedback
-  impact  on  learning  directly  as  a
result of verbalisation
 
Definitions of intake
The term intake ‘has taken on a number of
different meanings, and it is not always clear
what a particular investigator means in using
it’ (McLaughlin, 1987, p.13). That was true
nearly  two  decades  ago,  and  it  still  is  true
today. The purpose of this section, then, is to
review these different meanings and identify
the  commonalities  and  differences  between
them.  
                               
  VanPatten  (2002)  has  recently  clearly
acknowledged  a  number  of  important  roles  for
output, such as that it can function as a ‘focusing
device’, drawing learners to mismatches between the
input and their own output, and he has acknowledged  
its role in the development of fluency.
 
Definitions  of  intake  come  into  three  broad
categories: those that see intake as a product,
those that see it as  a process, and those that
see  it  as  a  combination  of  the  two.  Coming
into  the  first  of  these  categories  is  Corder
(1967)  who  provides  the  earliest  recorded
definition  of  the  term  as:  ‘a  mental
representation  of  a  physical  stimulus’  (p.
165).  For  Corder,  intake  is  thus  something
that  has  been  detected  but  has  not  yet  been
integrated  into  the  learner’s  developing
second language system as it is still linked to
the physical stimulus. Others also see intake
as  a  product,  but  give  it  quite  a  different
meaning. Krashen, for example, on the basis
of  an  investigation  of  caretaker  speech
(1978)  concludes:  ‘  “intake”  is,  simply,
where language acquisition comes from, that
subset  of  linguistic  input  that  helps  the
acquirer learn the language‘ (1981, p. 101)
and:  ‘intake  is  first  of  all  input  that  is
understood’ (p. 102; emphasis in original).
Interestingly,  Krashen  talks  of  input  in  first
language  acquisition  containing  ‘a  high
proportion  of  intake’  (p.  102,  1981;
emphasis  in  original)  by  which  he  means
language  input  of  which  a  great  deal  is
understood.  This  is  interesting  because  it
shows  that  for  Krashen  the  occurrence  of
intake  is  something  that  is  not  so  much
dependent on the learner as on the quality of
the input. It appears that in Krashen’s view
one  cannot  help  but  understand
appropriately  used  input,  and  thus  be
provided  with  intake.  It  must  be  noted  that,
perhaps  as  a  result  of  this  interpretation,
Krashen  appears  to  use  the  terms  input  and
intake  somewhat  arbitrarily.  For  example,
on  a  different  occasion,  he  talks  about
language acquisition developing better when
the intake is communicative and understood.
Finally, Krashen claims that intake ‘builds’
acquisition,  but  how  this  happens  is  not
elaborated on.  
 
Also Faerch & Kasper (1980) see intake as a
product  but  make  a  distinction  between
intake  for  communication  and  intake  for
learning.  Intake  for  communication  is
detected  input  that  the  learner  has
comprehended (maybe partially on the basis
of  non-linguistic  aspects  relating  to  the
communication that takes place), whereas on
the basis of intake for learning ‘the learner
forms her hypotheses about the L2 rules and
tests them out subsequently’ (p. 64). Intake
for learning is clearly processed more deeply
as  it  requires  the  learner  to  (consciously  or
not)  make  a  comparison  between  current
knowledge  and  new  information,  whereas
this  is  not  the  case  with  intake  for
communication.  Loschky  &  Bley-Vroman
(1993)  see  intake  for  communication  as
depriving  the  learner  of  the  potential  for
feedback  and  thus  a  chance  to  notice  a
difficulty with his/her performance:  
 
Thus,  it  may  be  possible  to  (1)  comprehend
native  speaker  input,  or  (2)  make  one's
interlanguage  output  comprehensible  to  a
native  speaker  without  (3)  focusing  on  or
using the target form of instruction. […] this
is certainly possible through use of strategic
competence.  Second,  as  a  consequence  of
this,  negative  feedback  which  could
potentially  destabilize  one's  target  language
hypotheses  may  be  either  absent  or  non-salient. The learner may never 'notice a gap'.
(p. 131).   
 
Sharwood  Smith  (1986)  does  not
specifically  discuss  intake  but  does  make  a
similar  distinction  as  do  Faerch  &  Kasper.
He  talks  about  input  having  dual  relevance,
for  immediate  communicative  purposes  but
the  input  may  also  ‘contribute  to  the
substantiation  or  reflection  of  some  current
hypothesis about the target language system’
(p.  243).  Input  may  not  be  relevant  for
acquisition  where  the  learner  is  not
 
developmentally  ready  or  where  the
demands of the communicative exchange are
heavy,  even  though  the  input  can  be
interpreted.  Sharwood  Smith  later  defined
intake  as  ‘that  part  of  input  which  has
actually  been  processed  by  the  learner  and
turned into knowledge of some kind’ (1993,
p. 8). He specifies this by saying ‘input is, as
it  were,  the  goods  that  are  presented  to  the
customer,  including  the  articles  that  the
customer picks up to look at.  Intake is what
is  actually  bought  and  taken  away  from  the
shop,  i.e.  what  passes  into  the  ownership  of
the customer’ (pp. 8-9).  This  is  ambiguous.
What  is  meant  by ‘ownership’ here? Is it a
hire-purchase which may  be returned  at any
time, are ‘the goods’ consumed before the
buyer  even  arrives  home,  are  they  shelved
for future use, or are they used every day?   
 
Carroll  (2001)  also  sees  intake  as  a  product
but  makes  a  very  clear  distinction  between
comprehension  and  intake.  She  describes
comprehended speech as a ‘speech signal
which  has  been  successfully  parsed  and

  re-encoded in semantic terms’ (p. 9). Carroll
rejects  the  view  that  intake  consists  of
comprehended speech as it would mean that
all learning would involve concept learning.
To her, comprehending speech is ‘something
that  happens  as  a  consequence  of  a
successful parse of the speech signal’ (p .9;
emphasis in the original). She sees intake as
a  subset  of  the  input;  stimuli  that  are
perceived  by  the  learner.  She  defines  it  as
‘that which is taken in by the hearer’ (p. 10).
In  addition,  perceived  stimuli  are
characterised  as  “transduced  stimuli”,  or
stimuli  neurally  available  for  processors  to
extract  further  information  from.  Carroll
emphasises  that  intake  is  not  input  to  the
learning  mechanisms,  but  input  to  speech
parsers.  Her  view  of  intake  thus  diverges
from that of the other authors cited here.
 
Gass’s  (1997)  model  of  second  language
acquisition  consists  of  a  number  of  stages
starting  from  raw  input.  Several  factors
(including  time  pressure,  frequency,  affect,
salience,  associations  and  prior  knowledge)
influence  whether  input  gets  noticed,  or
apperceived. Apperception is conceptualised
as  a  priming  device.  It  prepares  the  learner
for  the  possibility  of  subsequent  analysis.
Some  or  all  of  the  noticed  input  may  be
comprehended, with comprehension relating
to  a  continuum  of  properties  of  the
apperceived  input,  from  meaning-related
properties  to  deeper,  grammatical  features.
What  gets  comprehended  may  (depending
on a range of factors) become intake, which
Gass defines as the ‘process of assimilating
linguistic material’ (p. 5). Intake can thus be
conceptualised as apperceived input that has
been  further  processed.  This  further
processing  can  take  the  form  of  hypothesis
testing,  rule  strengthening,  storage  for  later
use,  or  the  intake  may  remain  unused.  It  is
interesting  to  note  that  unlike  Faerch  &
Kasper  (1980),  Gass  sees  comprehension
(comprehended  input)  as  a  prerequisite  for
intake to take place.  
 
VanPatten’s (1996) definition of intake is
similar to Gass’s: ‘intake is the subset of
filtered  input  that  serves  as  the  data  for
accommodation by the developing system. It
is the input that has been processed in some
way  by  the  learner  during  the  act  of
comprehension.  Intake  […]  are  the  data
made  available  for  further  processing  (e.g.
internalization)  once  the  input  has  been
processed’ (p. 10). More recently VanPatten
(2002) defined intake as ‘the linguistic data
actually  processed  from  the  input  and  held
in working memory for further processing’
(p.  757).  For  VanPatten,  as  for  Gass,  intake
follows  or  occurs  simultaneously  with  the
process  of  comprehension.  Leow  holds  a
similar view: ‘intake, […] is that part of the
 
input  that  has  been  attended  to  by  second
language  learners  while  processing  the
input’ (1993, p. 334).
 
Others have approached intake as a process,
rather  than  as  a  product.  Chaudron’s
definition  of  intake  (1985),  for  example,
encompasses the processes Carroll, Gass and
others refer to: ‘intake is processing of target
language input’, or ‘the mediating process
between  the  target  language  available  to
learners  as  input  and  the  learners’
internalized set of L2 rules and strategies for
second  language  development’ (p. 1). And:
'in  speaking  of  intake  we  are,  in  effect,
referring not to a single event or product, but
to  a  complex  phenomenon  of  information
processing  that  involves  several  stages,
roughly characterized as (1) the initial stages
of  perception  of  input,  (2)  the  subsequent
stages  of  recoding  and  encoding  of  the
semantic  (communicated)  information  into
long  term  memory,  and  (3)  the  series  of
stages  by  which  learners  fully  integrate  and
incorporate  the  linguistic  information  in
input into their developing grammars’ (p. 2).
Chaudron  refers  to  this  process  as  a
continuum  from  preliminary  to  final  intake,
although  he  concedes  that  the  two  ends  of
the  continuum  constitute  very  different
categories  of  cognitive  activity  (1983,  pp.
438-439).
 
Also,  Boulouffe  (1987)  conceptualises
intake  as  a  process  and  calls  it  ‘the
notoriously  impenetrable  interval  between
input and output’ (p. 245). She describes
intake as ‘the locus of the learner’s active
search for inner consistency’ (p. 246) by
which  she  means  a  process  of  equilibration
through  accommodation  or  assimilation  of
new  knowledge.  She  gives  examples  of
students  receiving  feedback  on  an  incorrect
production of a target sentence. The number
of  attempts  represent  the  intake  process
whereby  the  new  knowledge  (the  correct
sentence  structure)  is  assimilated.  She
clearly sees intake as the process of learning,
hypothesis testing, transfer etcetera. It is also
a process that requires the learner’s active
participation  as  it  is  something  that  is
subject to control.  
 
Similarly,  Hatch  (1983)  writes:  ‘if  the
learner “casts a net” into the input, the result
is supposedly intake’ (p. 79) and ‘for me, all
input is intake if the learner does respond in
some way to it’ (p. 81).  
 
We  might  say  that  input  is  what  the  learner
hears and attempts to process. That part that
learners  process  only  partially  is  still  input,
though  traces  of  it  may  remain  and  help  in
building  the  internal  representation  of  the
language.  The  part  the  learner  actually
successfully  and  completely  processed  is  a
subset  called  intake.  That  part,  then,  is  the
language  that  is  already  part  of  the  internal
representation. (p. 81).
 
Yet  others  acknowledge  that  intake  can  be
seen  as  both  a  product  and  a  process.
Kumaravadivelu  (1994)  provides  an
overview  of  the  preceding  25  years  of
theorising  into  intake  and  suggests  that  a
focus  on  intake  as  either  a  product  or  a
process  is  flawed.  He  proposes  a  synthesis
of  a  range  of  intake  factors  (age,  affective
factors, negotiation etc) and intake processes
(structuring,  inferencing,  transfer  etc)  that
dynamically  interact  and  are  co-requisites
for  intake.  Intake  factors  determine  which
aspects  of  the  input  get  engaged.  These
receive  attention  and  as  a  result  of  this,
consciously or unconsciously, a mismatch is
detected between that aspect of the input and
existing  knowledge  the  learner  has.  As  a
result  of  this  mismatch  a  range  of  intake
processes  execute  that  form  the  process  of
learning.  Like  Chaudron  then,

Kumaravadivelu  includes  within  intake  the
whole  process  from  detection  to  final
acquisition.  It  is  difficult  to  see  how  this
proposal of intake can be distinguished from
one of learning in general.    
 
Finally,  although  not  concerning  itself  so
much  with  attempts  to  define  intake,  some
recent  neurobiological  research  has  made
efforts  to  come  to  a  more  precise
understanding of what constitutes intake and
where  in  the  brain  the  product  or  process  is
localised.  Sato  &  Jacobs  (1992)  identified
the nucleus reticularis thalami (NRT) as that
area  of  the  brain  that  seems  to  facilitate
processing in other areas of the brain known
to  be  involved  in  (language)  learning  and
production,  including  the  hippocampus,  the
cerebellum,  the  basal  ganglia,  and  the
cingulate  gyrus.  The  authors  propose  that
the NRT functions as a “gating mechanism”
that  allows  or  inhibits  information  flow  to
these areas. As such the NRT can be said to
facilitate intake from input in a literal sense.
Their  assertion,  however,  that  ‘the  key
assumption  here  is  that  information
ascending  through  the  NRT  to  the  cerebral
cortex constitutes intake (or at least potential
intake)’ (p. 287) casts some doubt on their
own  claims  as  it  shows  that  by  adding  the
afterthought  between  brackets,  the  authors
make  a  distinction,  perhaps  implicitly,
between  what  enters  the  system  and  aspects
of  that  information  that  may  be  used  for
language  learning.  Although  the  authors
admit  that  ‘the  NRT's  posited  role  in
language  acquisition  is  at  present  neither
directly observable nor testable’ (p. 287),
identifying  the  neurobiological  correlates  of
the  intake  process  is  a  promising  approach
to arriving at a more precise, and meaningful
interpretation  of  the  processes  that  the
concept of intake is said to embrace.  
 
Research  into  brain  activity  may  help  make
such processes ‘observable’. One relevant
example is the work of Buckner (2000) who,
quoting himself, writes:  
 
Neuroimaging data suggest a pattern relating
localized  brain  activity  and  memory
encoding.  Several  neuroscientific
hypotheses  have  proposed  that  certain
regions  within  the  frontal  cortex  participate
in  the  short  term  maintenance  and
manipulation  of  information  over  brief
periods of time, as would be required during
many kinds of information processing tasks.
Deep  processing  tasks  and  intentional  tasks
make use of such processing, while shallow
processing tasks do not (Buckner & Tulving
1995).’ (p. 285)
 
Buckner  relates  this  to  the  formation  of
memory traces and writes: ‘one speculation
would be that the critical cascade that drives
episodic  human  memory  formation  occurs
when  frontal  activity  provides  a  source  of
information (input) to medial temporal lobes
and functions to bind together the outcomes
of  information  processing  from  frontal  and
other  cortical  regions  to  form  lasting,
recollectable memory traces’ (p. 285). The
same  could  apply  to  language  learning
where  frontal  processing  provides  input  to
the rest of the developing system. This is an
interesting,  but  as  of  yet  little  explored
avenue for research into the topic of intake.  
 
 
A working definition of intake
Above  a  wide  range  of  existing  definitions
have  been  discussed.  Next  I  turn  to  what  I
believe  are  the  key  elements  in  these
definitions  and  drawing  on  these  I  will
propose a working definition of intake.
 
First, I make a distinction between input and
stimuli,  where  stimuli  are  seen  as  the
 Towards a definition of intake   01   
 
language  potentially  available  to  a  learner,
and  input  as  those  stimuli  that  have  entered
the learner’s system. Secondly, and in line
with most of the authors mentioned above, I
see  intake  as  resulting  from  that  subset  of
the  input  that  is  detected  by  the  learner.
Following  Tomlin  &  Villa  (1994),  I  do  not
equate  detected  input  with  noticed  input.  A
definition  of  intake  or  an  operationalisation
of  it,  should  probably  leave  open  the
question  of  whether  or  not  intake  can  only
be  derived  from  noticed  input,  until  greater
evidence  for  either  position  has  been  found
(cf.  Schmidt,  2001).  For  some  (e.g.  Carroll,
2001;  Sato  &  Jacobs,  19929)  this  detected
input equals intake. However, detected input
can be used for comprehension only and this
need not involve any attention to the formal
aspects  of  the  input,  nor  does  there  have  to
be  any  link  with  long-term  memory.  To
resolve  this,  Faerch  &  Kasper  (1980)
proposed  a  distinction  between  intake  for
comprehension  and  intake  for  learning.
However,  this  seems  to  create  unnecessary
confusion,  as  these  two  terms  refer  to
different  representations  and  subsequent
uses  of  input.  It  may  be  more  accurate  to
distinguish  between  detected  input,  the
influence  of  which  does  not  reach  beyond
the  moment,  and  intake,  the  influence  of
which is potentially lasting.  
 
Tomlin  &  Villa  (1994)  define  detection  as
‘the process by which particular exemplars
are  registered  in  memory  and  therefore
could  be  made  accessible  to  whatever  the
key  processes  are  of  learning,  such  as
hypothesis formations and testing.’ (pp. 192-193). Assuming that ‘memory’ in the above
definition refers to ‘working memory’, I see
intake  as  a  subset  of  this  detected  input,
which is accessible (and not merely could be
                                 
 Sato & Jacobs specify this by suggesting that only
detected input that enters certain areas of the brain
affect language learning.  
made  accessible)  to  ‘whatever  the  key
processes are of learning, such as hypothesis
forming and testing’. Which exemplars or
what  subset  of  the  detected  input  becomes
intake  depends  on  a  large  range  of  factors,
including  (but  perhaps  not  necessarily)  the
amount  of  attention  given  to  that  subset  of
the  detected  input.  This  in  turn  depends  on
aspects of the input, such as its saliency, on
the  learner  and  the  state  of  her
interlanguage,  and  on  other,  non-cognitive
factors such (e.g. motivation).  
 
Intake  is  thus  detected  input  that  goes
beyond what is held in working memory for
immediate  recognition  and  comprehension.
It establishes a link with long-term memory.
In  case  of  a  subsequent  encounter  with  the
particular linguistic phenomenon these links
and/or  connections  are  strengthened.  This
would  explain  frequency  effects  (N.  Ellis,
2002a,  2002b).  The  degree  to  which  this
happens, again, depends  on a large range of
factors  including  the  amount  of  attention
paid  to  the  input,  the  strength  of  existing
connections, as well as learning aptitude etc.
As  a  working  definition  then,  I  propose  the
following:  
 
Intake  is  a  subset  of  the  detected  input
(comprehended  or  not),  held  in  short-term
memory, from which connections with long-term  memory  are  potentially  created  or
strengthened.
 
The above is quite similar to how Tomlin &
Villa  (1994,  p  .196)  describe  what  precedes
potential  acquisition  by  a  learner.  They
write:  
1.  The  learner  must  discern  the
presence  of  some  element  of
grammatical form.
2.  The learner must discern that there is
a  new  or  unusual  character  to  the
event representation witnessed.
Applied Research in English: 1(2)     29
 
3.  The learner must discern that there is
a  relationship  holding  between  these
two  levels  of  grammatical  form  and
mental representation.  
4.  The  learner  must  send  those
observations  off  for  further
processing  (hypothesis  formation
and testing).  
 
The  result  of  stages  1-3  is  intake.  However,
in an attempt to accommodate connectionist
views on acquisition, the working definition
includes  the  possibility  of  the  strengthening
of  existing  knowledge  as  opposed  to  the
learner discerning only a ‘new or  unusual
character to the event representation’.  
 
Operationalising intake
Intake  has  been  operationalised  and
measured  in  a  number  of  different  ways.
Several  authors  operationalise  intake  as  a
change  in  performance.  Zobl  (1985)  simply
sees  changes  in  a  learner's  rule  output  after
an  exposure  session  as  evidence  of  intake.
Similarly,  Rosa  &  O’Neill  (1999)
recommend  performance  measures  when
attempting  to  measure  intake  such  as  recall
protocols,  cloze  tests,  grammaticality
judgements,  and  rule  formation,  all  to  be
administered  soon  after  the  treatment  or
exposure  to  the  target  input.  They  accede
that intake tasks that introduce production as
a factor may be inadequate in that ‘there is
some  potential  for  interference  from
inappropriately  automatized  production
routines’ (p. 286). In their own study they
made  use  of  a  multiple-choice  recognition
task. Interestingly Rosa & O’Neill write in a
footnote to their 1999 article: ‘in order to
minimize  the  possibility  of  learners
performing  the  posttest  on  the  basis  of
memorized material, all of the test sentences
containing the target structure were different
than  the  sentences  included  in  the  treatment
task’  (p.  549).  This  appears  to  measure
learning, not intake.
 
Leow  (1993,  1995)  also  used  multiple-choice  recognition  tasks  and  gave
participants  very  limited  time  to  complete
their  tasks,  which  were  administered
immediately after exposure.
 
To  measure  learners’  intake  of  linguistic
items  in  the  input,  a  multiple-choice
recognition  assessment  task  was  carefully
designed  to  address  only  the  linguistic  item
that  had  been  attended  to  by  the  learners  in
the  input.  The  three  factors  crucial  to  this
assessment  task  were  a)  the  administration
of  the  assessment  task  immediately  after
exposure to the input, b) the limited amount
of time learners had to complete the task (cf.
Chaudron,  1985),  and  c)  a  single,  final
answer. (1993, p. 337).
     
The  fact  that  only  items  that  had  been
attended to by the learners in the input were
included  in  the  recognition  test  may  have
raised  participants’  awareness  of  those
items.  In  a  later  study  (2001a)  Leow  also
made  use  of  think-aloud  protocols.  In  that
study  he  aimed  to  investigate  the  effects  of
awareness  on  acquisition  and  recorded
correct verbal production of the target form.
Chaudron  (1985)  warns  that  production
measures  need  to  be  used  with  caution  as
they  could  cause  interference  from  previous
knowledge.
 
Shook  (1994)  made  use  of  both  production
tests (cloze test, sentence completion)  and a
recognition  test  (multiple-choice  sentence
completion)  all  of  which  were  administered
immediately  following  the  exposure.  Shook
claims that ‘it is most improbable that the
data collection procedures used could reflect
anything  except  the  immediacy  of  Process  I
[the  input-to-intake  stage],  and  thus  this
 Towards a definition of intake   02   
 
study does not reflect any  acquisition of the
grammatical input’ (p. 85).
 
The  above  measures  of  intake  have  in
common  that  they  attempt  to  probe  beyond
what  is  held  in  short-term  memory  (and  as
such  aim  to  measure  intake  as  opposed  to
detected  input).  Likewise,  they  attempt  to
avoid  measuring  (performance  based  on)
previous  knowledge.  A  measure  of  intake
should  also  avoid  measuring  acquisition;
any  measure  that  requires  retention  of
knowledge  for  extended  periods  of  time  is
an  indication  of  knowledge  in  long-term
memory,  not  intake.  Intake  tests,  then,  can
only  be  administered  after,  but  reasonably
soon  after  exposure  to  the  second  language.
This  does  not  preclude  task  performance  as
a  measure  of  intake,  however,  exposure  to
the  target  language  (i.e.  a  listening  or
reading  passage)  needs  to  be  separated  in
time  from  the  activities  participants  are
asked to perform on that input.
 
Forced  recognition  tests,  grammaticality
judgment  tests,  as  well  as  measures
containing a degree of production, including
fill-in-the-gap,  jumbled  sentences  tests
etcetera,  are  all  potentially  valid  measures,
despite their individual drawbacks (provided
they  are  administered  not  too  long  after
exposure  to  the  input).  However,  more
subtle  measures  such  as  the  forced
recognition  tests,  are  more  likely  to  be
sensitive  to  intake  in  the  early  stages  of  the
learning  process.  Measures  requiring
production  need  to  be  used  with  care.  Free
production, and measures such as fill-in-the-gap  without  multiple-choice  options  are
more  likely  to  measure  integrated
knowledge,  and  can  sometimes  be
awareness-raising. Verbal protocols can help
to  provide  an  additional  measure  of  intake
and  can  act  as  a  comparison  with
performance measures.  
Conclusion
Although  there  are  great  differences  in  the
various  positions  researchers  have  taken  in
relation  to  the  roles  of  input,  output,
interaction,  and  intake  in  second  language
acquisition,  there  is  certainly  also
considerable  agreement.  First,  there  appears
to  be  a  consensus  that  language  learning
cannot take place without input (although, as
mentioned  above,  input  can  affect  the
development  of  aspects  of  the  language  not
contained  in  the  input  itself).  There  appears
to  be  a  reasonable  degree  of  consensus  that
certain types of input are more favourable to
learning  than  others,  and  that  input  at  the
very least has to be comprehensible. A range
of  characteristics  of  the  input  determine
what is acquired (within the constraints of a
predetermined  developmental  order)  and  a
manipulation  of  those  can  affect  learning.
There  is  evidence,  at  least  for  second
language  learning,  that  both  output  and
interaction to some extent facilitate learning,
either  through  the  provision  of  more
comprehensible  input,  or  by  drawing
attention  to  certain  aspects  of  the  input,  or
the learner’s own output.  
 
Of  all  the  information  available  to  the
learner,  only  some  remains.  This  is  where
the  road  forks.  Connectionists  (cf.  Plunkett
&  Elman  1997).  argue  that  information  that
is  not  detected  is  discarded,  but  all
information  that  is  detected,  affects  the
learner’s developing system. It is unclear
what the role of intake in such a system is, if
indeed there is any; it is perhaps telling that
I  was  unable  to  find  a  clear  account  in  the
literature  of  how  such  a  process  would  fit
into  a  connectionist  approach  (but  see  N.
Ellis  2002a,  2002b,  for  a  discussion  of  the
role  of  noticing  in  essentially  frequency
drive  implicit  learning).  Many  second
language  researchers  working  from  an
information-processing  perspective  view

this differently: information may be detected
but that does not mean that it has any lasting
effect.  It  can  be  used  for  communication
purposes  or  be  lost  before  it  is  stored.
Although  this  group  of  researchers  holds
very  different  views  on  the  exact  process
affecting  what  information  is  incorporated
and what is not, they do appear to agree that
there is a level of processing that takes place
on  the  input,  that  determines  whether
aspects  of  it  may  potentially  be  learned  or
not.  Some  have  labelled  one  or  more  of  the
various  stages  in  this  processing as ‘intake’
(one  exception  appears  to  be  Carroll,  for
whom intake equals something more akin to
input,  as  defined  by  most  other  second
language researchers), others (e.g. Sharwood
Smith, 1993) appear to be talking about this
stage  without  using  the  same  term.  Leow
points  out that  the  concept  is  useful  to  have
and apply: ‘the distinction between input
and  intake  has  theoretical  value  because  it
proposes  that  there  is  at  least  one
intermediate  stage  of  input  processing
through  which  the  input  second  language
learners  receive  must  pass  before  any  or  all
of it can become part of learners’ developing
linguistic system’ (1993, p.334). Although
some refer to the actual processing as intake,
others to its product, and  yet others to both,
investigating this processing or its product is
crucial in understanding how input becomes
part of a learner’s system.