Planning and second language development in task-based synchronous computer-mediated communication
This dissertation explored the effect of two planning conditions (the multiple planning condition with rehearsal and online planning time, and the single planning condition with online planning time only) on L2 production complexity and accuracy and the subsequent development of these two linguistic areas in the context of written synchronous computer-mediated communication (SCMC).
Participants were 33 intermediate to advanced adult ESL learners recruited from an intensive English program at a U.S. university. They completed four picture-based narrative tasks that were equivalent in task complexity (Ellis, 2003; Robinson, 2001) under the two planning conditions over a two week period via text-chat. Two tasks were used as experimental tasks to gauge immediate planning effect on L2 production complexity and accuracy and the other two as immediate posttest tasks to gauge L2 development. Production complexity was analyzed using measures including number of clauses per AS-unit, mean length of AS-unit, mean length of clause, variety of grammatical verb forms, index of Guiraud, and frequency of difficult words used. Production accuracy was analyzed using measures including percentage of accurate use of grammatical verb forms, percentage of error-free AS-units and percentage of error-free clauses. Parametric paired-samples t-tests and non-parametric Wilcoxon Signed Ranks tests were performed.
Results showed that both types of planning condition supported immediate production complexity but the multiple planning condition was more effective in leading to more accurate use of grammatical verb forms in the immediate term. With regard to the subsequent development of production complexity and accuracy in new contexts, the multiple planning condition was more effective in leading to improvement on clausal complexity (as indexed by mean length of clause) as well as control over the use of grammatical verb forms and avoidance of general errors.