Eurosla 11, Paderborn, 26-29 September

1
Eurosla 11, Paderborn, 26-29 September
Re-evaluating Theoretical and Methodological
Aspects of Focus on Form Research

• Danijela Trenkic Mike Sharwood Smith
• Heriot-Watt University, Edinburgh

2
Background Focus on Form research

• the principle aim
• to investigate whether focusing learners
  attention to formal aspects of language in
  communicative context can (in some cases) promote
  SLA.
• theoretical underpinnings
• the noticing hypothesis (Schmidt 1990) only what
  has been noticed will be learned
• goals of research
• pedagogical
• theoretical

3
Typical design of FonF studies

• Classroom studies
• typically, two or more classrooms of FL/SL
  learners at the same (or similar) proficiency
  level are pretested, and then exposed to a
  different instructional treatment over a period
  of time.

4
Their improvement on a post-test is then compared
Control Group no FonF treatments
Post-test
Pre-test
Experimental Group receives some sort of FonF
treatment
5
Typical findings and generalisations

• Learners who receive FonF treatment perform
  better on the post-test
• FonF - an effective means of promoting noticing
• noticing the key factor of SLA
• FonF should be included in language teaching

6
Problems with FonF research design, findings and
generalisations

• Theoretical problems
• how do you define noticing?
• Schmidt (1990) noticing/understanding
  distinction
• the distinction is scalar, rather than
  categorical
• what is it that a learner has to notice?
• how do you measure noticing?

7
Methodological problems

• A whole plethora of uncontrollable variables
  involved
• impossible to interpret the results confidently
  (you are not sure why something worked, and even
  more importantly, why something did not work)
• impossible to compare studies
• impossible to draw generalisations

8
Major variables in FonF studies

• The choice of forms
• learners proficiency levels
• the size and nature of groups
• the length of treatment
• the number of post-tests
• the choice of testing materials
• the level of explicitness

9
The choice of forms

• Not all linguistic features are equally amenable
  to focus on form (cf. Williams and Evans 1998)
• the results obtained on one form are no guarantee
  that the same treatments would work equally well
  for another form

10
Learners proficiency levels

• What work for learners at a certain level of
  proficiency may not work for learners at another
  level

11
The size and nature of groups

• Groups not usually big enough for reliable
  statistical analysis and conclusions ( the
  intergroup differences in proficiency,
  motivation, etc. can be considerable).
• since the sample sizes were fairly small and the
  data were not normally distributed (Doughty and
  Varela 1998129)
• BUT the effects of FonF treatment are clearly
  interpretable from our results (ibid.)
• This sample is too small to provide convincing
  quantitative evidence (Williams and Evans
  1998151)
• BUT the results point to the fact that focus
  on form is indeed useful and should be integrated
  into communicative curricula. (ibid.)

12
The length of treatment

• A few days/several weeks/whole semesters
• Effects for instruction of any kind may be, and
  probably almost are, gradual and cumulative
  rather than instantaneous and categorical (Long
  and Robinson 199840)
• results are affected by the length of treatments

13
The number of post-tests

• There is a reverse side to the gradual
  accumulation of the effect of instruction
• FonF groups often improve their performance on
  the immediate post-test, but on (sufficiently)
  delayed post-tests, this improvement decreases,
  or even disappears altogether.
• White (1991) - found positive effects for a FonF
  instruction on the 5 week post-test, but not on
  the post-test administered a year later.
• The last post-test rarely exceeds 5 weeks
• BUT instruction that appropriately incorporates
  form-focused treatments into communication-oriente
  d language teaching can have lasting positive
  effect on L2 acquisition (Muranoi 2000661)

14
The choice of testing materials

• Testing materials come in a variety of forms
• oral and written reports on science experiments
  (Doughty and Varela 1998), short constrained
  narratives based on several pictures (Williams
  and Evans 1998, Muranoi 2000), more or less
  constrained sentence-completion tasks (ibid.),
  different varieties of grammaticality judgement
  tasks (ibid.), description of a short film-scene
  (Muranoi 2000), etc.
• different testing materials yield somewhat
  different results ? when comparing the results,
  no guarantee that like is being compared with
  like
• the central problem what are the testing
  materials testing?

15
The level of explicitness of FonF treatment

• it has not been clear exactly what it means to
  draw a learners attention to form or how this is
  to be accomplished. (Williams and Evans
  1998139)
• Accomplished in many ways from the most implicit
  ones (e.g. the flood of positive evidence), to
  the most explicit ones (e.g. stating a rule)

16
The level of explicitness

• Original proposal FonF should occur incidentally
  and be fairly implicit, so as not to distract
  learners from their communicative goal (cf. Long
  1991).
• a quintessential element of the theoretical
  construct of focus on form is its dual
  requirement that the focus must occur in
  conjunction with - but must not interrupt -
  communicative interaction. (Doughty and Varela
  1998114)
• more explicit procedures may cause stress and
  anxiety, and so preclude fluency. This is because
  they do not add attention to form to a primarily
  communicative task but rather depart from an
  already communicative goal in order to discuss a
  linguistic features (ibid.)

17
The levels of explicitness

• The general trend emerging from FonF studies
  employing a whole range of FonF techniques seems
  to be that the more explicit the treatment, the
  more marked the gain on the post-test.

18
Some examples

• Williams and Evans (1998) considered English
  participial adjectives
• found that ESL learners who received a flood of
  positive evidence, plus explicit instruction,
  plus feedback, significantly outperformed the
  group which only received a flood of positive
  evidence, which in turn, outperformed but not
  significantly, the control group which did not
  undergo any FonF treatment.
• Muranoi (2000) considered English articles
• found that Japanese EFL learners show much better
  results in using E articles after receiving an
  implicit interaction enhancement treatment, but
  even better when the treatment is supported by
  explicit formal instruction.

19
Result a split in the theory of FonF research

• Long and Doughty - still advocate exclusively
  implicit techniques
• Lightbown - argues for a role for grammatical
  instruction that is separate from communicative
  activities, and is yet integrated in the lesson
  as a whole. (1998194)
• DeKeyser - advocates explicit instruction at
  first, and believes that declarative knowledge
  acquired through explicit FonF instruction can
  eventually become fully automated (199847).
  Communicative interaction is vital for the
  process of proceduralising declarative knowledge.

20
Failing on the pedagogical aim

• Due to a great number of uncontrollable variables
  in research, all teachers can be told is yes,
  the results show that focusing your students
  attention to form may work, but you have to work
  out what will work for YOUR students.
• Not exactly helpful or revealing

21
Failing on the theoretical aim

• Not much has been revealed about the process of
  SLA.
• Since it is not properly defined what noticing is
  or how it is to be measured ? the hypothesis is
  not falsifiable

22
Failing on the theoretical aim

• Further, findings from FonF research show that
• the more explicit the instruction, the more
  marked the effect on a post-test
• effects are not preserved in spontaneous
  production (cf. the choice of testing materials
  above), or on sufficiently delayed post-tests
• overgeneralised uses of the treated form are a
  regular by-product of FonF research.
• These are characteristics of meta-linguistic
  learning/knowledge!

23
Implications of these findings

• FonF treatments, irrespective of their level of
  explicitness, actually manipulate meta-linguistic
  knowledge, that is knowledge about language,
  rather than knowledge of language (cf. Truscott
  1998).
• Similarly, meta-linguistic knowledge is what FonF
  research testing materials test.
• The only safe conclusion noticing does promote
  learning, but of meta-linguistic type. It does
  not seem to promote learning OF language (and we
  believe there are good theoretical reasons why it
  doesnt)

24
The future of FonF research?

• More research of the same type?
• OR defining a viable model of SLA with a clear
  set of testable predictions?
• We would advocate the second choice.
• The most theoretically grounded and
  methodologically worked-out line of research
  within the FonF framework so far, has been that
  of DeKeyser (1998), in Andersons ACT framework.
  The idea is to see whether declarative knowledge,
  acquired through FonF instruction, can be
  proceduralised by engaging in target behaviour
  while temporarily leaning on declarative
  crutches (199849)

25
The future of FonF classroom practice

• We believe that there is a place for FonF
  instruction and feedback in language classroom,
  despite the fact that it produces meta-linguistic
  knowledge.
• Learners may have very limited exposure to the TL
  (especially FL learners) they may learn in huge
  groups, a few hours a week, and be taught by a
  non-native speaker of that language ? there may
  not be enough input to develop knowledge of
  language.
• Learners have practical goals - e.g. to pass the
  exam, get a job, etc., and these goals can be
  achieved by developing meta-linguistic knowledge
  (it is actually more than likely that
  meta-linguistic knowledge is what is going to be
  tested by the testing materials).
• There is a possibility that it can ultimately
  influence knowledge of language - a question to
  be theoretically and empirically addressed by
  future FonF research.