k. philip choong and zhaohong han teachers college, columbia university [email protected]...
TRANSCRIPT
K. PHILIP CHOONG and ZHAOHONG HANTeachers College, Columbia University
[email protected]@tc.columbia.edu
Task Complexity and Output Complexity:
An Exploratory Study
Motivation
Cognition Hypothesis
Question: What is the relationship between task complexity and output complexity?
Is there a relationship between task complexity and output complexity?
If so, what is the nature of the relationship?
Operationalization of Task Complexity
Dimensions
Contextual support Reasoning demands Single/Dual Task
DimensionsDimensions
Contextual support
Dimensions
Contextual support
Dimensions
Reasoning demandsContextual support
Dimensions
Reasoning demandsContextual support
Dimensions
Single/Dual TaskReasoning demandsContextual support
Dimensions
Task: Story narration
Task conditions
+/-picture
+/-sequence
(C1)+ picture/+sequence
(C2)- picture/+sequence
(C3)+ picture/-sequence
(C4)-picture/-sequence
-Complex
+Complex
Participants
Experimental group: 10 native Japanese speakers from advanced ESL classes in New York City Average age – 38 Gender- 9 females and 1 male
Comparison Group 1: 5 native speakers of American English Average age: 33 All females
Comparison Group 2: 10 native speakers of American English
Procedure
Task: Story narration under 4 different conditions
4 stories vis-à-vis 4 conditions Instructions
I am going to show you a set of pictures that tell a story. Please take as long as you like to look over the pictures, then tell me the story as if I cannot see the pictures. We will do this twice. The first set will be practice, just to make sure you understand the instructions. There is only one “correct” story for these pictures.
Measures of Output Complexity
Syntactic complexity # of T-units per narration
Content complexity # of idea units per narration
Analysis and Results – Experimental group
Friedman Test and Kendall’s W Test of Mean rank
Both produced same results, significant at .05 level
Friedman Test of Mean Rank (Kendall’s test similar)
Condition Mean RankC1_T-unit 2.45C2_T-unit 2.00C3_T-unit 5.15C4_T-unit 3.90C1_I-unit 4.55C2_I-unit 4.25C3_I-unit 7.55C4_I-unit 6.15
Condition Mean RankC1_T-unit 2.45C2_T-unit 2.00C3_T-unit 5.15C4_T-unit 3.90C1_I-unit 4.55C2_I-unit 4.25C3_I-unit 7.55C4_I-unit 6.15
Results- Form ComplexityCondition 3 (+reasoning demands, +contextual support) most complex
Condition 2 (-Reasoning demands, -contextual support)
Least complex, least variation
Condition 4 (+Reasoning demands, -contextual support) shows most variation
C4_TC3_TC2_TC1_T
18
16
14
12
10
8
6
95
% C
I
Form Complexity and Variation by T- Unit
Results-Content complexity
Greater variation in content complexity than in form complexity
In line with Friedman’s and Kendall’s W tests of mean rank.
Also supported by paired sample t-test
C4_IC3_IC2_IC1_I
24
21
18
15
12
9
95
% C
I
Content Complexity and Variation by Idea Unit
Results – paired samples t-test
Significant differences between conditions 1 and 3, 2 and 3, and 2 and 4
Both for Form complexity (t-units) and for content complexity (idea units)
Contrastt-value df Sig. (2-tailed)
C1_T - C3_T-3.160 9 .012
C2_T - C3_T-4.371 9 .002
C2_T - C4_T-2.785 9 .021
C1_I - C3_I-2.918 9 .017
C2_I - C3_I-6.050 9 .000
C2_I - C4_I-2.293 9 .048
Results – individual
Greater number of idea units may suggest that participant was more focused on content than form.
4321
Condition
30.00
25.00
20.00
15.00
10.00
5.00
0.00
Me
an
IdeaUnit
T_Unit
Shihoko
Results-Comparison Group 1
Native speaker results are parallel to non-native speaker results in terms of:
Syntactic/Content Complexity:
C3>C4>C1>C2
Discussion
+contextual support+reasoning demands
-contextual support+reasoning demands
Finding 1: More complex conditions produce more complex output.
Reasoning demands
C3 C4
Discussion
Finding 2: Gap between the “contrived complexity order” and the “observed order of output complexity”
Contrived complexity: C4>C3>C2>C1Observed complexity of linguistic output:C3>C4>C1>C2
Task intrinsic complexity: C3>C4>C1>C2
TaskInternal
complexity
+/-picture
+/-sequence
Discussion
Finding 3: Patterns of variation differ for the experimental group vs. comparison group 1. Experimental group:
Syntactic complexity: C4 C2 Content complexity: C1 C2
Comparison group 1: Syntactic complexity: C1 C2 Content complexity: C4 C2
Content complexity and form complexity are unequal.
Limitations and Next Steps
Further data analysis Fluency as well as accuracy Lexical complexity Syntactic complexity in terms of S-nodes per T-unit
Conclusion
The jury is still out.Task complexity is a complex notion requiring
finer-grained analysis than has generally been given. More conceptual work is needed. In examining the relationship between task complexity
and output complexity, there is a need to differentiate between content complexity and form complexity, and more importantly, to investigate how attention is allocated to form and content during task performance.
There is a need to track down the differential impact of task-intrinsic complexity and contrived complexity
More attention should be given to task-intrinsic complexity