Differences between task types

The second research questions of this study asked how manipulating cognitive task complexity affected general and task-specific measures of fluency, grammatical complexity, accuracy, and lexical complexity in L2 narrative production. Motivated by Skehan’s Trade-Off Hypothesis (Skehan, 2005;

Skehan & Foster, 2001) and Robinson’s Cognition Hypothesis (Robinson, 2003a, 2005a), it was expected that the cognitive load imposed by the narrative task types used in this study would influence the various aspects of participants’ performance differentially. In order to examine participants’ linguistic output on the two narrative task types, paired samples t-tests were used to compare the means of different performance measures in terms of fluency, accuracy, complexity and lexical diversity. The results indicate that the two oral tasks (see Table 35) generated similar output in terms of global measures of speech rate, accuracy and syntactic complexity, which finding is similar to that of Robinson’s (1995) research, where he also used narrative tasks. For the participants of this study it seems that the increased cognitive load that was imposed on them by having to invent the storyline for Task 2 did not reduce the amount of attention available for the accuracy and global syntactic complexity of their output, nor did it decrease the fluency of participants’ performance. However, they used significantly more varied vocabulary in the

cartoon description task (Task 1) than in the picture narration task (Task 2). This confirms that the two tasks imposed an increased cognitive load at different points in message generation: the cartoon description task pushed participants to stretch their vocabulary knowledge and employ less frequently used words in their descriptions. They were able to do so because the more structured task did not require them to conceptualize the storyline itself, leaving much needed cognitive resources available for the formulation phase of message production. In line with this, Task 1 required the use of certain infrequent vocabulary items such as binoculars, uninhabited etc, which resulted in higher D-values in Task 1 than in Task 2. This finding reveals that certain tasks do not only promote syntactic development but also contribute to students’ vocabulary learning processes. It seems that task requirements can push students to use a wider variety of words, which might otherwise primarily constitute their passive vocabulary.

The participants were not more accurate in the task-specific linguistic aspects, namely the use of past tense and relative clauses on either of the task types. Albert (2008) used the same tasks in her research, but with more advanced students than our participants. In her study, students were significantly more accurate in Task 2, which imposed higher conceptualizing demands on the participants. However, her participants, just as ours, did not perform differently on the two tasks in terms of syntactic complexity and fluency. It seems that for English language majors, who constituted the examined population of Albert’s study, the task condition in which students had to invent a story directed their attention to accurate speech production. Albert explained this by arguing that in the cognitively less complex task in terms of conceptualization, the protagonists of the stories were depicted in the pictures together with the changes in location and the order of events and therefore, the participants’ attention might not have been directed to the accurate and precise use of personal pronouns and tenses. In Task 2, however, accuracy plays an important part in storytelling in the sense that protagonists, changes in location and the timeline must be kept clear for the listeners to be able to follow the story. Results indicating no significant

difference pertaining to the correct use of past-tense verbs and relative clauses also suggest that students who are at lower levels of proficiency are not able to benefit from this resource-directing aspect of the conceptually more demanding task. The significant correlations between accuracy and clause length in the two oral tasks as well as between speech rates in the two task types also seem to indicate that for these two narrative tasks, a certain proficiency threshold may be needed for a task effect to be detectable.

Another related explanation for the lack of task-effects in terms of the resource-directing dimensions might be that the two tasks made attentional demands on the participants in different ways. In the cartoon description task, even though the content was given, participants were forced by the task to narrate a given story with their available linguistic resources, whereas in the picture narration task, they could conceptualize their storyline taking their own linguistic resources into account and adjusting their story to their existing linguistic means. This implies that they might have used vocabulary which was easily accessible from their mental lexicon. As a consequence they might have had more attention available for the accurate linguistic encoding of their message. The lack of significant differences in fluency between the two tasks also seems to lend support to this trade-off effect and might suggest that the overall processing load at the linguistic encoding stage of the tasks might have been similar.

As regards the effect of task type in writing, the results indicate that the picture narration task in writing elicited syntactically more complex language, as assessed by clause length and the ratio of relative clauses, than the cartoon description task. The results also show that the participants’ performance on the two tasks was similar as regards lexical variety and accuracy as the differences observed did not reach the level of statistical significance. This finding is seemingly in contrast with that of Kuiken and Vedder (2008), who found no effect of task complexity in the written mode. Kuiken and Vedder (2008), however, used only a subordination ratio as the measure of syntactic complexity. This confirms - in line with recent calls for using task-specific performance measures (e.g., Norris & Ortega, 2008; Robinson, 2007b) – that

using linguistic measures relevant to tasks when examining the effects of task complexity on the quality of L2 output may yield more reliable results. As the findings of this dissertation also suggest, relying exclusively on general measures may not enable us to fully capture the effects of different task types.

Based on these results concerning syntactic complexity in different types of tasks in the two modes of performance, we can hypothesize that in writing, where the resource-dispersing dimension of task complexity might play a different role, the picture narration task, which requires students to conceptualize their own stories, has the potential to direct students’ attention to syntactic complexity. In the oral version of this task, however, the demand that students need to conceptualize and linguistically encode their narrative at the same time acts as a resource dispersion factor, and students do not seem to have sufficient attentional resources for producing syntactically complex language.