Fluency, Working Memory, and Second Language Proficiency in Multicompetent Writers

Abstract

The present study compared pausing behavior while writing in the L1 and the L2 of emergent bilinguals and investigated the role of WM and proficiency level, as well as observed the cognitive processes that occurred during these pauses using a mixed-methods approach. Learners from a 5th semester (N = 17), 7th semester (N = 16), and graduating seniors (N = 14) produced argumentative tasks in their L1 and L2 and all writing behavior was recorded using Inputlog (Leijten & Van Waes, 2013), a keystroke logging program designed specifically for writing studies. Pause frequency and length was compared within and across languages and regressions were run to test the effects of the different components of WM on pausing behavior in the L1 and L2. Proficiency level was also included for the L2. Stimulated recalls were collected for a subset of participants (N = 5) and these were coded according to Kellogg’s (1996) model of writing to compare cognitive behavior.

Results show that in both languages pause length increased by constituent level but that all pauses were longer in Spanish than English. The relationship of WM and pauses was different for each language, in English visual spatial WM was related to pause frequency between sentences while the updating function was related to pause length between words and pause length and frequency between sentences in Spanish. Proficiency level was found to have the most effects on pauses, with those at higher proficiency levels pausing less frequently between words and sentences and pausing for shorter periods of time. Stimulated recall comments showed that participants did engage in the cognitive processes described by Kellogg (1996) in both languages, however there were qualitative differences. Processes related to formulation were the most prevalent in both languages, but there were more behaviors related to planning in English and translation was the most frequent behavior in Spanish. Some differences were also seen by proficiency level, those with a lower proficiency level did not show evidence of monitoring in Spanish while those with a higher proficiency level did.