Complexity, Accuracy, and Fluency as Properties of Language Performance: The Development of the Multiple Subsystems over Time and in Relation to Each Other

Applied linguists have identified three components of second language (L2) performance: complexity, accuracy, and fluency (CAF) to measure L2 development. Many studies researching CAF found trade-off effects (in which a higher performance in one component corresponds to lower performance in another) during tasks, often in online oral language performance. Trade-off effects are attributed to the inability of the learner to simuletaneously attend to all components at the highest level possible. Although cross-sectional research has suggested that students at different proficiency levels sacrifice performance in one CAF area while improving in another, there has been little longitudinal research about CAF (Ortega & Iberri-Shea, 2005). As such, previous research could not address if CAF grows linearly over time nor if the rate of CAF growth is the same for all learners. The current study explicitly addresses how language performance in CAF changes over L2 development in an instructed environment. This longitudinal study analyzed English L2 oral data from sixty-six students from Arabic, Chinese, and Korean language backgrounds over 3-9 months in the English Language Institute at the University of Pittsburgh. Elicited speeches were transcribed, coded, and assessed with three measures of structural complexity, a measure of lexical variety, two measures of accuracy, and three measures of fluency. The scores were then analyzed with hierarchical linear modeling (Singer & Willett, 2003) to investigate how each student’s performance changed over time for each measure and to determine predictive variables. Although individual differences were found in initial scores (often proficiency differences, but not for all measures), growth trajectories were the same for all measures, except one grammatical complexity measure (length of AS unit) where slopes differed by gender. All measures showed growth, and only two measures (lexical variety and a mean length of fluent run) showed non-linear growth. Trade-off effects found in cross-sectional studies were not found in these longitudinal data even though within-individual and between-individual correlations were also calculated. Additionally, the results may suggest that instructed language performance growth is uniform, rather than along individual paths. The research also serves to evaluate the measures, which has research and pedagogical implications.