Dr Watsons Problem

1

• Dr Watsons Problem

presented by M.A.Sharwood Smith Heriot-Watt
University April 2000 NWCL Workshop, Salford
University
2
Overview of Presentation

• Plato and Watson
• The Nature of L2 Knowledge
• Getting Using Knowledge the Dualist
  Interpretation
• The Metalinguistic Level
• Enhancing Input
• The Virtual Input Hypothesis
• Validating input

3
Plato and Watson
PW
4
With so little evidence, how come we know so
much?
Platos Problem
5
Within generative linguistics, this is a
well-known formulation of the problem faced by
the immature, ignorant first language acquirer
6
Within generative linguistics, this is a
well-known formulation of the problem faced by
the immature, ignorant first language acquirer
Limitations Attention span, Conscious, analytic
abilities
7
Within generative linguistics, this is a
well-known formulation of the problem faced by
the immature, ignorant first language acquirer
Limitations Knowledge of the World Little or no
corrective, grammatical feedback
8
With so much evidence, how come we know so
little?otherwise known, in Chomskys
political writings, at Orwells problem
Dr Watson Problem
9
This is my formulation of the problem faced by
more cognitively mature
10
This is my formulation of the problem faced by
more cognitively mature and knowledgeable
acquirers of a second language
11
We know that much more relevant information about
the target language is available and,
potentially, accessible to L2 learners
12
AVAILABLE L2 INFORMATION

• Positive evidence (as in L1A)
• PLUS EXTRA RESOURCES
• internal other tongue knowledge
• external supplementary evidence

13
Typically, the available other language
resources comprise
Other Tongue Knowledge

• L1 representations (knowledge)
• PLUS
• Computational (on-line) mechanisms
  subserving those L1 representations

14

So, with these other tongue resources, the L2
acquirer appears to have a headstart in
understanding and producing L2 utterances..
15

Lets assume for convenience that the other
tongue is the mother tongue (L1)...
16

L1 system
L2 item
L2 item
Newly developed L2 material may be meshed with
the L1 morphosyntactical and phonological system
almost from the word go...
Crosslinguistic influence
17

Observation and research do support this idea of
the early deployment of L1 resources in L1
performance ...
18

• suggesting the use of the L1 both as a
• L2 performance facilitator
• and also (more controversially) as a basis for a
  preliminary representation of the L2 system
  itself
• the L2 initial state

19
But even if L2 acquirers were to rely, for their
primary linguistic data, on the L2 input alone...
Supplementary evidence
20
..in combination with internal learning
mechanisms, of course...
21
..and never recruited the L1 in the course of
their L2 development...
No crosslinguistic influence in development of L2
knowledge
22
. they could in principle make use of a great
variety of additional or enhanced resources
compared to the child L1 acquirer.
Additional resources
23

• supplementary positive evidence
• learner-elicited or independent confirmations
  by authoritative sources
• various kinds of negative evidence
• learner-elicited or independent
  disconfirmations by authoritative sources

Confirming and disconfirming information
24
We know that, in principle, L2 learners can
become, or be made aware of all this extra
information but....
i
Evidence!
LOOK!
25
But we also know that most L2 learners are not
like Sherlock Holmes!
26
A Sherlock Holmes learner would not only have
swift access to this wealth of extra information
provided by the outside world.
27
.but also be able to USE the information to
build up a faithful interpretation of the crime
scene, that is, the complete native L2 system
including the grammar.
28
In fact, most of us behave much more like Dr
Watson!
29
We dont have swift access to this extra
information about the L2 grammar
30
And even when Holmes makes us aware of the
evidenceand explains his interpretation, all
we can do is marvel.
Elementary, dear Watson
31
I dont know how you do it, Holmes!
Watson never develops Holmes exceptional ability
no matter how much he is shown a) what and where
the evidence is and b) what use he should make
of it.
32
THE NATURE OF L2 KNOWLEDGE
K
33
Knowing is Possessing

• Knowledge as a possessed mental system
• to know Portuguese is to possess a Portuguese
  mental system

Portuguese Knowledge
34
Knowing is Modular

• Knowledge as an assembly of separate systems
• not all of these systems behave like the grammar
• they may each implicate different learning
  mechanisms

lexicon
grammar
Etc.!
pragmatics
35
D
GETTING USING KNOWLEDGE THE DUALIST
INTERPRETATION
36
submeta and metacognition
H o l m e s d o m a i n

• Implicit, tacit
• subconscious
• not directly accessible to conscious manipulation
• no conscious involvement in development

• Explicit
• conscious
• accessible to conscious manipulation
• conscious involvement in development

No bidirectional interface?
37
M

• METACOGNITION

15
38

• THE METALINGUISTIC LEVEL
• language as an object of conscious inspection
• metacognition systematicity and coverage

Metacognition
39

Metalinguistic awareness
Metalinguistic knowledge

• Highly systematised
• detailed technical vocabulary
• (metalanguage)

A matter of degree (how systematic)
40

• Metaknowledge ABOUT language in general
• Metaknowledge ABOUT the L2 specifically
• Metaknowledge about relationships between L1 (or
  other tongues) and the L2 in question

A matter of degree (how extensive)
41
All metacognitive activity is typically viewed as
involving on-line computations that

• are relatively slow and costly
• yield results that are available for
  conscious introspection

42
it would take an exceptional ability to be so
metalinguistically skilful that no one would be
able to tell if metacognition was driving
someones fluent performance.
43
A Sherlock Holmes language learner would be
exceptional in that he would be able to conduct
metalinguistic operations with such apparent ease
as to fool the outside observer into thinking he
was using all aspects of the L2 straightforwardly
without conscious effort or involvement and had
learned them in the same way
44

• So, a Sherlock Holmes learner would be able to
  fake it.
• He would be a virtual native speaker.
• Under normal circumstances (outside the
  experimental lab) his performance would conceal
  any L2-specific grammatical disadvantages either
  in processing or knowledge representation

45
DISCLAIMERGiven the relative dearth of
rigorous experiments into the effect of
metalinguistic techniques ...the jury is still
out.
46
The significance of success Note that the
discovery of a successful metalinguistic
technique facilitating the development of L2
skills would still require us to establish
whether we had made our experimental subjects
into Sherlock Holmes learners (virtual native
speakers) or whether we had undermined the
(strong) interface hypotheisis
47
An empirical question
It still remains an empirical issue as to where
(in which modules of L2) and to what degree
appeals to the L2 learners metalinguistic skills
might facilitate L2 development and also what the
precise nature of that development might be.
48
Enhancing input
IE
49

• It is convenient to distinguish between
• a. FOCUS ON FORM
• b. INPUT ENHANCEMENT
• c. ATTENTION (NOTICING, AWARENESS, MEMORISING,
  ETC)

FonF, IE and Attention
50
FonF

• FOCUS ON FORM the hypothesis, rationale, or
  dogma of the outsider (researcher or teacher)
• Should we or should we not focus on form ..at
  all, or, if so, when, how, and how much?

51
IE

• INPUT ENHANCEMENT the various
  metalinguistic techniques used when attempting
  to make forms salient for the learner

52
IE
A general definition

• It is the principled manipulation of the
  learners environment with the aim of affecting
  development.
• It involves attempts to make given target
  features salient so that the learner may, by
  attending to their form, notice and (somehow!)
  incorporate them in their developing L2 systems.

Implying an empirical question
53

• NOTICING, ATTENTION, AWARENESS, MEMORISING,
• The various actual responses of the learner.
• Did they notice it?
• Did they attend to it?
• Were they aware of it?
• Did they commit it to memory?

54
If enhancement involves forms coming into focus
then, logically speaking, IE can be seen as
having two possible sources
Two Sources for
IE
55

• 1. Internally generated
• 2. Externally generated
• the usual meaning of input enhancement

56
Internally generated Enhancement

• SUBCONSCIOUS
• The learners acquisition mechanisms locate
  and fix on formal features following principles
  not consciously dictated by or accessible to the
  learner

57

• SUBCONSCIOUS
• For example, the learner (learning
  mechanisms) may be naturally and automatically
  attracted/sensitive to highly frequent words,
  useful words, short words, unusual sounding
  words, emotionally loaded words, or L1/L2
  cognates. etc.

lexical acquisition
58

• UNORDERED?
• With the lexicon, strictly predetermined
  schedules are hard to imagine. Some might suggest
  L1-L2 cognates come early, or high frequency
  items, or communicatively useful items, etc., etc.

Lion - lion Tigre - tiger Help Dont
I, you
59

• SUBCONSCIOUS
• The learner (learning mechanisms) may be
  naturally and automatically driven by mechanisms
  designed to look for specific formal features
  rather than others
• 1) always
• or
• 2) at a given moment in development

grammatical acquisition
60

• CONSCIOUS
• The learner consciously and deliberately
  locates and fixes on formal features

61

• For example, the learner may decide to
  memorise a list of vocabulary items devoted to
  cooking, computers, sport or pay special
  attention to words when watching a particular TV
  show, or memorise advertisments or learn a new
  word order by consciously thinking up and
  practising sentence after sentence manifesting
  that order..

62
Externally generated Enhancement

• The experimenter/teacher manipulates the
  learners environment in order to direct the
  learners attention to formal features.
• Means of doing this vary ALONG A CLINE from
• (relatively implicit techniques)
  avoiding/attempting to circumvent high levels of
  metalinguistic awareness and thereby influence
  subconscious developmental processes
• to
• (relatively explicit techniques) deliberately
  engaging high levels of metalinguistic awareness
  this includes the promotion of conscious learning
  strategies (internal conscious IE)

63

• Lexical examples
• (IMPLICIT)
• Le LION a mangé le petit chien. Le voilà
• Comme il est méchant, ce lion
• L OISEAU a bu leau. Le voilà
• Comme il est beau, cet oiseau!

64
There is a tradition of research into L2 grammar
that discounts metalinguistic operations (stored
representations and associated computation) as
IRRELEVANT to grammatical development (Krashen,
Schwartz, etc.)
65
?
?
meta
submeta
NO interface?
66
Despite popular conceptions of L2 learning, much
L2 grammatical development appears to be sealed
off or encapsulated as far as consciousness is
concerned.
67
Factors suggesting irrelevance

• Observed developmental stages in L2 cannot be
  manipulated (Krashen, Pienemann)
• the observed ineffectiveness of negative evidence
  in syntactic development (Trahey and White,
  Truscott)

68
Some recent opinions

• The conclusion.then, is that research on
  form-focussed instruction has produced
  essentially no evidence that it is helpful and
  has produced considerable evidence that it is
  ineffective (though the latter is not entirely
  conclusive) Truscott 1997 p.121
• .. intact classroom studies with immersion
  children appear to show that explicit
  presentation of structure with negative input do
  not offer much advantage over the long run. But
  the jury is still out on the effects on adults,
  partly because of the many conceptual and
  methodological problems that plague language
  studies Schachter 1999578-579

69
As mentioned earlier, there is still

• The lack of sophistication in metalinguistic
  techniques actually used in current
  experimentation
• some suggestive evidence that consciously
  apprehended constructions can affect subsequent
  development when noticed during spontaneous
  language use (Schmidt and Frota)

70
Experimental Approaches a Simple Example
Control Group No explanation No input
enhancement Naturalistic exposure to target
features in communicative contexts
Post-test 2
Post-test 1
Pre-test
Experimental Group focus on target features
(directly or indirectly) with or without
explanation or correction (negative input)
71
Enhancement techniques that must be investigates
are many and varied..
72
Typical kinds of external input enhancement
Control Group or Reported Research

• Experimental Group
• input flood
• typographical
• phonological
• Metalinguistic
• explanations

Post-test 2
Post-test 1
Pre-test
73
Enhancement techniques should be tested with
different modules..
74
Experimental Group 1 Pragmatic feature
enhancement
Experimental Group 2 Lexical feature enhancement
Post-test 2
Post-test 1
Pre-test
Experimental Group 3 Syntactic feature enhancement
75
For example
76
Experimental Group 1 Do sit down Sit down
pragmatic
Experimental Group 2 It sits on the floor It lies
on the floor
lexical
Post-test 2
Post-test 1
Pre-test
Experimental Group 3 There is sitting someone
inside There is someone sitting inside
syntactic
77
QUESTIONSIs development facilitated

• WHERE THE LEARNER IS DEVELOPMENTALLY READY FOR
  THE TARGET STRUCTURE?
• UNDER ANY CIRCUMSTANCES?
• ONLY WHERE STORAGE IS CONCERNED, E.G.
  IRREGULAR VERBS, IDIOMS, ETC, ETC?
• ALSO WHERE COMPUTATIONAL PHENOMENA ARE CONCERNED
  (rules and principles)?
  

78
Virtual Input Hypothesis
VIH
79
Schmidt and Frota

• In his diary, Schmidt attributes his
  accelerated acquisition of the Portuguese
  subjunctive to
• Prior discussion of the construction with
  language tutor
• Subsequent heightened noticing of the subjunctive
  in L2 input (PLD)
• Successful use normal communication with a native
  speaker

80
Schmidt and Frotas observations raise the
following possibility

• What is originally metalinguistic knowledge
  can, under certain circumstances, become
  validated as virtual input to the learning
  device.

81

• The immediate result is to enrich the learners
  metalinguistic representation of L2.

82

• The working hypothesis is that the learners
  metalinguistic representation of L2 may be
  exploited to create virtual input.

83

• Virtual input is input to the learning
  mechanisms that result in changes in the
  learners representations of L2.

84

• The input is virtual because it doesnt come
  from the standard source..outside the learner.
  The learning mechanisms are, as it were,
  fooled into thinking it does.

85

• Hence, the working hypothesis is that the
  learners metalinguistic representation of L2 may
  be exploited to create virtual input.

86
The Virtual Input Hypothesis
VIH

• Learner-produced utterances which involve
  internal
• metalinguistic operations on grammatical
  structure,
• by then being used primarily to express
  meaning,
• may come to function as input to the
  learners' own
• acquisitional mechanisms.
• In this way, they can play the same role in
  acquisition as
• native-speaker utterances from an external
  source
• 
  (Sharwood Smith 1980, 1996, 1997).

87
Validating Input
88
Validating Input

• "Noticing" some formal feature of L2 is different
  from the "validation" of that feature
• noticing does not automatically implicate
  development.

89
Pushed output

• The VIH is not exactly the same as the "pushed
  output" hypothesis (Swain 1985)
• although it might be seen as an extension of it
• If a teacher encourages production of a formal
  feature in order for the learner to notice and
  acquire it, this does not automatically implicate
  validation since something extra is required over
  and above production.

90
Validation Conditions

• The requisite circumstances implied by Schmidt
  and Frota are
• spontaneous language use by unaware
  interlocutor(s)
• (conscious) noticing by learner alone

6
91
A sample experimental design

• Recreate under controlled conditions situations
  where
• control group are instructed beforehand (cf White
  and Travis) highly explicit input enhancement
  techniques
• they are given a target structure to notice
  and/or use while communicating with native
  speakers
• they are required to perform tasks simulating
  spontaneous communication.
• The native interlocutors are unaware of the real
  goal of the experiment and act unreflectingly

92

• CONCLUSION
• A substantial body of serious research must
  (continue to) be carried out into all types of
  "input enhancement" , including the clandestine
  use of metalinguistic knowledge, within the
  framework of testable hypotheses.

93
Overview

• Plato and Watson L1 and L2.
• L2 Knowledge possession/modularised
• Getting Using Knowledge the Dualist
  Interpretation the interface issue
• The Metalinguistic Level specified
• Enhancing Input an empirical question
• Virtual Input Hypothesis
• Validating input

94

• THE END

PRESENTATION AVAILABLE SHORTLY
AT http//www.hw.ac.uk/langWWW/mss/MYPAGE.htm