Complex working memory and written task performance

sub-ordinate clauses the most frequently are the ones with an average working memory span. Even if no significant differences could be detected in the case of the other linguistic variables, if we inspect the mean values for the ratio of error-free relative clauses, error-free past-tense verbs and the D-value in the cartoon description task, we can see a similar pattern: students with the lowest and highest backward digit span achieving the lowest mean values and learners in the middle range of backward digit span the highest ones. These results seem to indicate that there is no linear relationship between working memory capacity and narrative performance in this task, and students who can perform best in terms of task-specific aspects of accuracy, lexical variety and the use of relative clauses are the ones who have an average backward digit span. Although my sample is small, it might be speculated that too low backward digit span might not be advantageous in certain oral tasks because learners cannot allocate their attentional resources efficiently. Very high backward digit span does not seem to be beneficial either as these learners perhaps try to pay attention to too many aspects of performance at the same time, which results in lower lexical variety, accuracy and less frequent use of complex structures such as relative clauses in certain tasks.

These results show considerably smaller effect of working memory capacity in using L2 oral language than those reported by Kormos and Sáfár (2008), who investigated the link between working memory and end of year language test scores. It might be possible that working memory capacity has more influence on how L2 learners process input, i.e. how much attention they can devote to noticing various linguistic features presented to them in the language classroom than on how learners produce oral output in a particular task. We might assume that working memory capacity is therefore related to input processing mechanisms and consequently to general learning outcomes and might not play a noticeable role in regulating attention in oral task performance.

Complex working memory capacity appears to influence the structural complexity of the written narratives produced by the participants. Similarly to oral production, the cartoon description task imposed a heavier cognitive load at the formulation level than at the conceptualization level of message generation despite the fact that writing allowed participants to plan their message online. The results showed that learners with higher BW spans managed to retrieve more of the specific vocabulary required by the task than those with a lower BW score. Therefore, the findings suggest that greater working memory capacity supported the retrieval of the necessary vocabulary items. Similarly, in case of the subordination ratio, participants with a BW span of 7 used more subordinate clauses than those with a BW score of 4 and 5.

This result also indicates that complex working memory assists the use of complex structures when the task pushes learners to focus on the formulation of their message. In contrast, participants with an average BW span of 5 produced more complex language in terms of the relative clause ratio in the picture narration task which required them to devote a substantial amount of attention to conceptualizing the content of their message. Apparently learners with higher BW scores did not benefit from this advantage.

One way of accounting for this finding may be that learners with higher working memory capacity tried to pay more attention to the development of the story and various aspects of performance at the same time.

As a closer examination of the results reveals, they tended to slightly outperform the others in most other aspects of performance, although these relationships did not reach a level of significance. An alternative explanation may be that those participants whose working memory capacity was higher managed to coordinate their cognitive processes more efficiently and tailor their story to match their linguistic resources in such a way that they did not have to stretch their language abilities considerably to tell the story. The storyline of their message being more thought out, they might have been able to express themselves adequately by using a smaller number of relative clauses for the identification of characters and places in their story.

7.3.2 The relationship between foreign language aptitude and narrative task performance

7.3.2.1 The relationship between aptitude and oral task performance

The findings indicate that students with a given cognitive profile do not perform consistently in narrative tasks. In the oral cartoon description task, where students were not required to conceptualize and linguistically encode their message at the same time, high grammatical sensitivity and high level of language learning aptitude might have helped students to successfully encode complex and difficult syntactic structures such as relative clauses. In the oral picture narration task, however, where the HUNLAT correlations with lexical diversity and the accuracy of relative clauses were negative, the task demands in terms of conceptualizing the content of the story might have caused difficulties to students with high grammatical sensitivity in employing their linguistic knowledge efficiently. We might argue that grammatical sensitivity is a cognitive ability that aids the conscious and explicit acquisition of complex syntactic knowledge (Robinson, 2005a) and that this ability might also help learners with high aptitude to consciously employ their explicit knowledge when they have sufficient attentional capacities that they can devote to linguistic encoding. This might be the case in the oral cartoon description task, in which students did not have to conceptualize the content of the narrative, and hence learners with high grammatical sensitivity might have been able to apply their explicit knowledge of relative clauses efficiently.

From the perspective of students who scored low on the grammatical sensitivity component of HUNLAT, the findings indicate that they might be less able to encode complex syntactic structures accurately in oral tasks in which the given content requires the use of specific syntactic structures and are thus high in linguistic encoding demands. They might, however, be advantaged in the oral picture narration task, in which they have the opportunity to avoid complex structures and in which attentional demands on conceptualization are high. A possible explanation for this might be that learners with lower levels of grammatical sensitivity might rely more on implicit knowledge and memorized chunks and thus might be more efficient in employing implicit knowledge of complex syntactic structures than students with high grammatical sensitivity.

It is interesting to note the negative relationship in the oral picture narration task between deductive ability and lexical variety as measured by the D-value. We might tentatively argue that learners with high deductive ability might have devoted their attentional resources to some other aspects of performance in this task. Niwa’s research (2000) also found that in the cognitively more complex task students with high aptitude scores performed worse in fluency, which she explained with reference to the fact that they might have paid more attention to the linguistic and conceptual encoding required by the task. Although the correlation between deductive ability and ratio of error-free clauses (r = .27 p =.07) is not significant, its direction is positive. Despite being weak, this correlation might indicate that students with high level of deductive abilities prioritized accuracy over lexical variety.

A significant positive correlation emerged between the ratio of error-free past-tense verbs and rote learning ability in the oral cartoon description task (r =.38, p = .02). Rote learning ability helps in memorizing words and irregular morphological information. A great proportion of past-tense verb forms at the low-frequency level of vocabulary that the participants tended to employ are irregular, and consequently, rote-learning ability helps their accurate acquisition. The results indicate that students with

high score in rote learning ability tended to be able to accurately retrieve the past-tense verb forms in the oral cartoon description task, in which they had sufficient attention available for linguistic encoding.

Finally, I would like to point out tendency-level positive correlations between speech rate and grammatical sensitivity in both tasks (r = .31, p<.05) in the cartoon description task,

r = .27, p < .08 in the picture narration task). Based on these results, students with higher grammatical sensitivity scores tended to be more fluent especially in the cartoon description task, where they were more focused on linguistic formulation than on conceptualizing the content of their story the latter being aided by the clear structure of the task. Grammatical sensitivity assists the process of linguistic formulation which explains why this ability may affect speech rate.

7.3.2.2 The relationship between aptitude and written task performance

Similarly to the oral mode, grammatical sensitivity was found to be positively related to certain aspects of performance in the written cartoon description task, but in writing the negative relationship of aptitude components and accuracy and lexical diversity was not observed in the picture narration task.

The finding that learners with high grammatical sensitivity produced longer clauses in the written cartoon description task might be related to the complexity demands of this writing task. This task relieved students from the cognitive load of having to conceptualize the story but made high demands on the participants in the linguistic encoding phase because they had to express a given content with their existing resources. The comparison of the mean values for clause length in the two types of narratives also shows that students in general used shorter clauses in the cartoon description task than in the picture narration.

Considering the linguistic encoding demands of the task, it seems that students with higher grammatical sensitivity might have devoted more attention to clausal complexity than learners with lower levels of cognitive ability who might have prioritized other aspects of linguistic processing such as the encoding of