English for Communication Studies

Download Report

Transcript English for Communication Studies

The effects of increasing
cognitive complexity on
L2 narrative oral production
Roger Gilabert
[email protected]
Blanquerna Communication Studies Department
Universitat Ramon Llull
Barcelona, Spain
Leuven 2005
Context
2 main research agendas into task features:
-
Interactionist perspective: how certain conversational
episodes (e.g. CR or CC), caused by task design, may
lead to acquisition (Long, 1985, 1989, 2000).
-
Information-processing approach: how manipulating the
cognitive features of tasks can lead to differentials in the
fluency, complexity, and accuracy of learners’
performance .
Context: research into cognitive task features
degree of familiarity: (Bygate, 1999, 2001; Foster & Skehan,
1996; Plough & Gass, 1993; Robinson, 2001)
number of elements: (Kuiken & Vedder, 2004; Robinson, 2001)
single and dual task performance: (Niwa, 2000)
pre-task and on-line planning time: (Crookes, 1989; Ellis, 1987;
Foster & Skehan, 1996;
Mehnert, 1998; Ortega, 1999;
Skehan & Foster, 1997;
Wigglesworth, 1997; Yuan &
Ellis, 2003)
degree of complexity along displaced,
past time reference: (Iwashita et al. 2001; Robinson, 1995;
Rahimpour, 1997).
Concerned with: how Task Complexity affects performance; how balanced
performance may lead to better use and acquisition; and how tasks can be best
sequenced according to their cognitive complexity
Context: Skehan / Robinson
Most studies concerned with the issue of competition for attention
during task performance.
Predictions for performance:
competition exists (Skehan, 1998; Skehan & Foster, 2001 )
it depends: resource-directing vs. resource-dispersing
(Robinson, 2001; 2003; 2005)
+ cognitively complex = - fluency
+ complexity
+ accuracy
Claims based on studies which have manipulated task features in
isolation (e.g. planning time studies and +/- here-and-now studies).
GOAL: of this study is to provide further evidence regarding the two widely
researched variables of planning time and +/- Here-and-Now, and to explore
the synergistic effects of manipulating them simultaneously.
Previous findings
Planning time studies:
Fluency increases (Foster & Skehan, 1996; Skehan & Foster, 1997; Ortega,
1999; Yuan & Ellis, 2003)
Higher structural complexity (Foster & Skehan, 1996; Ortega, 1999; Yuan &
Ellis, 2003; only a trend in Skehan & Foster, 1997)
No significant effects on lexical complexity: (Ortega, 1999; Yuan & Ellis, 2003).
Mixed results for accuracy: higher accuracy (Foster & Skehan, 1997)
no differences (Foster & Skehan, 1996; Yan & Ellis,
2003)
mixed restuls (Ortega, 1999)
Previous findings
Here-and-Now/There-and-Then studies:
Fluency decreases: (Robinson, 1995; Rahimpour, 1997)
Increased lexical complexity: (Robinson, 1995; Rahimpour, 1997)
No differences in structural complexity: (Robinson, 1995; Rahimpour,
1997)
Higher accuracy: (Robinson, 1995; Rahimpour, 1997; Iwashita et al.)
Research question
How does manipulating Task
Complexity simultaneously along
planning time and the +/- hereand-now variables affect
production?
Hypotheses
I.
Pre-task planning time will positively affect the areas of fluency and
structural complexity, with no effects on lexical complexity or accuracy.
- cognitive complexity
+ cognitive complexity
+ fluency
+ structural complexity
= lexical complexity
= accuracy
II.
- fluency
- structural complexity
= lexical complexity
= accuracy
More complex tasks, in the there-and-then, will trigger more accurate and
complex speech at the expense of fluency.
- cognitive complexity
+ fluency
- structural complexity
- lexical complexity
- accuracy
+ cognitive complexity
- less fluency
+ structural complexity
+ lexical complexity
+ accuracy
Experimental design
Condition 1:
Condition 2:
Condition 3:
Condition 4:
Planned Here-and-Now
Unplanned Here-and-Now
Planned There-and-Then
Unplanned There-and-Then
Story 1
Story 2
Story 3
Story 4
Group A
+ planning time
There-and-Then
- planning time
There-and-Then
+ planning time
Here-and-Now
- planning time
Here-and-Now
Group B
- planning time
There-and-Then
+ planning time
Here-and-Now
- planning time
Here-and-Now
+ planning time
There-and-Then
Group C
+ planning time
Here-and-Now
- planning time
Here-and-Now
+ planning time
There-and-Then
- planning time
There-and-Then
- planning time
Here-and-Now
+ planning time
There-and-Then
- planning time
There-and-Then
Simple
Group D
Complex
+ planning time
Here-and-Now
Experimental design:
example of comic strip
Begin the story like this: YESTERDAY Mr. Brown was shopping at the supermarket. He
was checking his shopping list and looking at prices. An employee was putting price
tags on the products.
Participants
48 volunteers among lower-intermediate, first- and
second- year university students.
Ages 18-22.
Similar number of years of instruction.
No significant differences in proficiency (C-Test)
Measures
Fluency:
Structural
Complexity:
Rate A (syllables x minute in unpruned speech)
Rate B (syllables x minute in pruned speech)
S-Nodes per T-unit
Lexical Complexity:
Percentage of Lexical Words
Ratio of Lexical / Function Words
Guiraud’s Index of Lexical Richness
Accuracy:
Error-free T-units
TLU of Articles
Percentage of Self-repairs
Repaired to Unrepaired Errors
Why self-repairs?
Self-repairs are the result of wrong formulation (Levelt, 1989), and may be used
to correct an inappropriate syntactic structure, a lexical problem, faulty
morphology, or a phonetic error. Self-repairs, whether other-initiated or
self-initiated Schegloff’s (1977), denote students’ awareness of form and can
be interpreted as learners’ attempts at being accurate. Self-repairs require
conscious attention. Some of the functions of self-repairs are:
Learners automatize the retrieval of target language knowledge they already
have.
They revise their hypotheses about the target language (Lyster and Ranta,
1997:57).
They noticing a hole in their own interlanguage that may direct their attention to
relevant input (Swain, 1998:66; Dörnyei & Kormos, 1999).
They check their speech, both internal and overt, against their receptive knowledge
(De Bot, 1996:551).
Statistical analysis
Sphericity of data achieved by means of detecting (by means of box
plots) and eliminating outliers from the calculation.
Repeated-measures ANOVAS for stories and conditions.
Post hoc Scheffe’s comparisons to identify exact location of differences.
Transcription and coding
Intrarater and interrater measures used for transcription and coding
Intrarater agreement 97%
Interrater agreement on 10% of the data reached 93.7%
Fluency
Results
Hypothesis 1
+/- Planning time
Hypothesis 2
Here-and-Now/There-and-then
160
160
140
140
120
119.4
100
120
115.4
115.7
111.8
100
80
115.7
115.4
111.8
80
60
Speech Rate A
40
20
60
40
20
0
Condition 1
Planned
Here-andNow
Condition 2
Unplanned
Here-andNow
Condition 3
Planned
There-andThen
0
Condition 4
Unplanned
There-andThen
Condition1
Planned Hereand-Now
140
Condition 3
Condition 2
Planned ThereUnplanned
and-Then
Here-and-Now
Condition 4
Unplanned
There-and-Then
140
120
100
119.4
120
107.81
100
97.08
96.33
89.75
80
60
96.33
80
Speech Rate B
40
107.81
97.08
89.75
60
40
20
20
0
0
Condition 1
Condition 2
Condition 3
Planned
Unplanned Planned ThereHere-and-Now Here-and-Now
and-Then
Condition 4
Unplanned
There-andThen
Condition 1
Planned Hereand-Now
Condition 3
Planned Thereand-Then
Condition 2
Condition 4
Unplanned
Unplanned
Here-and-Now There-and-Then
As predicted, fluency is negatively affected by increases in cognitive complexity
along both variables
Lexical
Hypothesis 2
Complexity Here-and-Now/There-and-then
Results
Hypothesis 1
Planning time
50
50
40
40
36.64
33.28
30
35.43
36.64
33.92
Percentage of
Lexical words
20
10
Condition 1
Planned Hereand-Now
Condition 2
Condition 3
Unplanned Here- Planned Thereand-Now
and-Then
33.28
33.92
20
10
Condition 4
Unplanned
There-and-Then
Condition 1
Planned Hereand-Now
80
Condition 3
Condition 2
Condition 4
Planned ThereUnplanned
Unplanned
and-Then
Here-and-Now There-and-Then
80
70
70
60
58.29
50.33
40
55.53
60
51.81
30
20
10
Ratio of Lexical to
Function words
0
50
58.29
55.53
50.33
51.81
Condition 2
Unplanned
Here-and-Now
Condition 4
Unplanned
There-and-Then
4.79
4.59
Condition 2
Unplanned
Here-and-Now
Condition 4
Unplanned
There-and-Then
40
30
20
10
Condition 1
Planned Hereand-Now
Condition 2
Unplanned
Here-and-Now
Condition 3
Condition 4
Planned ThereUnplanned
and-Then
There-and-Then
0
7
7
6
6
5
35.43
0
0
50
30
5.24
4
5
4.79
5.08
4.59
3
1
Condition 3
Planned Thereand-Then
5.24
5.08
4
Guiraud’s Index
2
Condition 1
Planned Hereand-Now
3
2
1
0
Condition 1
Planned Hereand-Now
Condition 2
Unplanned Hereand-Now
Condition 3
Planned Thereand-Then
Condition 4
Unplanned
There-and-Then
Against prediction, providing planning
time had a significant impact on lexical
complexity
0
Condition 1
Planned Hereand-Now
Condition 3
Planned Thereand-Then
Against prediction, increasing complexity along +/Here-and-Now had NO significant impact on lexical
complexity
Structural
Complexity
Results
Hypothesis 1
+/- Planning time
Hypothesis 2
Here-and-Now/There-and-then
2.5
2.5
2
2
1.5
1.55
1.5
1.46
1.45
S-Nodes
1.5
1
1
per
0.5
T-Units
0
Condition 1
Planned Hereand-Now
Condition 2
Condition 3
Condition 4
Unplanned Here- Planned ThereUnplanned
and-Now
and-Then
There-and-Then
1.55
1.41
1.5
1.45
0.5
0
Condition 1
Condition 3
Condition 2
Condition 4
Planned Here- Planned There- Unplanned
Unplanned
and-Now
and-Then Here-and-Now There-and-Then
Against prediction, structural complexity was not significantly affected by
planning time or by increasing along +/- Here-and-Now
Hypothesis 1: Planning time
35
30
25
20
23.65
23.36
Accuracy
Results
24.32
Hypothesis 2: Here-and-Now/There-and-then
35
30
25
21.53
20
15
Error-free
10
5
23.65
24.32
23.36
21.53
15
10
5
0
Condition 1
Planned Hereand-Now
Condition 2
Condition 3
Unplanned Here- Planned Thereand-Now
and-Then
Condition 4
Unplanned
There-and-Then
T-Units
0
Condition 1
Planned Hereand-Now
Condition 3
Condition 2
Condition 4
Planned ThereUnplanned
Unplanned
and-Then
Here-and-Now There-and-Then
120
120
100
80
86.09
84.41
83.43
100
82.27
60
80
TLU of articles
40
20
0
86.09
83.43
84.41
82.27
60
40
20
Condition 1
Planned Hereand-Now
Condition 2
Unplanned Hereand-Now
Condition 3
Planned Thereand-Then
Condition 4
Unplanned
There-and-Then
0
Condition 1
Planned Hereand-Now
Condition 3
Condition 2
Condition 4
Planned ThereUnplanned
Unplanned
and-Then
Here-and-Now There-and-Then
30
25
30
20
25
19.84
15
10
14.21
18.84
13.9
5
0
Condition 1
Condition 2
Planned Here-and- Unplanned HereNow
and-Now
Condition 3
Planned Thereand-Then
Condition 4
Unplanned Thereand-Then
Percentage of
self-repairs
19.84
18.84
15
10
14.21
13.9
5
0
Condition 1
Planned Hereand-Now
35
30
25
20
15
10
20
Condition 3
Planned Thereand-Then
Condition 2
Condition 4
Unplanned Here- Unplanned Thereand-Now
and-Then
35
30
25
25.46
22.23
17.19
17.63
5
0
Condition 1
Condition 2
Planned Here-and- Unplanned HereNow
and-Now
Condition 3
Planned Thereand-Then
Condition 4
Unplanned Thereand-Then
As predicted, planning time had no
impact on accuracy
Ratio of repaired
to unrepaired
errors
25.46
20
15
22.23
17.19
17.63
Condition 1
Planned Hereand-Now
Condition 3
Condition 2
Condition 4
Planned ThereUnplanned
Unplanned
and-Then
Here-and-Now There-and-Then
10
5
0
Partially in line with prediction, increasing
complexity along +/- Here-and-Now triggered more
attention to form
Discussion: Planning time
and fluency



Conceptualization during pre-task
planning allows faster retrieval during
performance
Instantiation of words (i.e. lemmas and
forms) or chunks in WM
Problem-solving mechanisms, rehearsal,
and memorization (Ortega, 1999)
Discussion:
+/- Here-and-Now
and fluency


Lack of contextual support
Efficient scheduling and attentionallocation policies
Discussion:
planning time and lexical and
structural complexity



More elaborate conceptualization
Consideration of more alternatives
Access to less activated terms
Discussion:
+/- Here-and-Now and lexical and
structural complexity


Interpropositional coherence
Dependence on lexical meaning
Discussion:
Planning time and accuracy


It facilitates all dimensions of production
Does not necessarily draw attention to
form during performance
Discussion:
+/- Here-and-Now and accuracy




It draws attention to form
Changes in macro and micro planning
Stretch interlanguage (Klein & Perdue,
1992)
Comparison of how L1 and L2
grammatize notions (Talmy, 2000)
Discussion:
Simultaneous manipulation
of both variables
1)
2)
Attention may be allocated to complexity
and accuracy simultaneously
Keeping tasks simple along resourcedispersing dimensions and complex
along resource-directing one may be
beneficial for language development
Simultaneous
manipulation
Guiraud’s Index
Percentage of
self-reapairs
7
30
6
25
5
5.24
4
4.79
5.08
20
4.59
3
15
10
14.21
13.9
Condition 1
Planned Here-andNow
Condition 2
Unplanned Hereand-Now
19.84
18.84
Condition 3
Planned Thereand-Then
Condition 4
Unplanned Thereand-Then
2
5
1
0
0
Condition 1
Planned Hereand-Now
Condition 2
Unplanned Hereand-Now
Condition 3
Planned Thereand-Then
Condition 4
Unplanned
There-and-Then
Conclusions
Task Complexity is a robust and testable construct.
Competition for attention may only happen when Task Complexity is
increased along resource-dispersing variables.
Attention to both complexity and accuracy may be possible if tasks are kept
simple along resource-dispersing variable and complex along resourcedirecting variables.
More research is needed regarding the synergistic effects of combining
resource-directing and resource-dispersing variables.
Do the results in this study apply to other task types and other dimensions?
Thank you