Unveiling linguistic competence by facilitating performance

Abstract

This dissertation looks at the question of whether the source of language learners' poor performance is a deficit in knowledge or a performance problem. Two constructions are examined, both being cases in which learners have been reported to show deficits in linguistic competence: (1) Korean OSV (e.g., Elmo-lul Big Bird-ka anayo. 'Elmo, Big Bird is hugging.'), where knowledge of both scrambling and case markers is necessary for target performance; (2) the English tough construction (English TC; e.g., The dolli is easy to see ei), where knowledge of tough movement is necessary for target performance (i.e., the object interpretation only). If poor performance (operationalized as chance performance or below) is due to a failure to apply knowledge rather than the absence of knowledge, good performance (i.e., significantly-above-chance performance) should emerge with the effective manipulation of non-grammatical factors to facilitate learners' application of knowledge. I experimentally tested this hypothesis in two studies, employing both (a) sentence-internal manipulations and (b) sentence-external manipulations. The Korean OSV study collected data from Korean Heritage children (KHC) (n = 31) as well as two groups of native Korean-speaking children, a younger group (n = 21) and an older group (n = 23). The English TC study collected data from native Korean-speaking adult L2ers of English (n = 49) and native Korean-speaking child L2ers of English (n = 30), along with adult native speaker controls (n = 10). Study 1: Knowledge of scrambling and case in Korean was assessed, first, via picture-selection comprehension tasks that (a) manipulated the prosodic salience of case markers and (b) manipulated context to make the direct object a natural scrambled topic, and, second, via a production task eliciting case markers. The results suggested three possible sources of poor performance. Many participants showed good performance only in the manipulation conditions (vs. the baseline condition), suggesting that their poor performance in the baseline condition could be due to a performance problem (e.g., perception failure or a heavy processing load). Some participants showed absence or errors of case in the production task, which suggests that their poor performance could come from a deficit in knowledge of case or from a mapping problem between a Case feature and case morphology. Study 2: Knowledge of the English TC was assessed via Truth-Value Judgment Tasks (TVJTs) that (a) manipulated verb transitivity to make the infinitival object gap more salient or less salient and (b) manipulated context to avoid or strengthen bias toward the (erroneous) subject interpretation. A Korean TVJT was also conducted to find the interpretation tendency for the Korean translation equivalents of the English TC. Comparisons between the Korean and the English TVJT results revealed a strong subject-interpretation bias in both languages for lower proficiency adult and child L2ers but only in Korean for higher proficiency adult L2ers, which suggests that L2ers do transfer their L1 subject interpretation but can subsequently unlearn it----despite an (ostensible) absence of negative evidence. However, (most of) the high proficiency adult L2ers still showed (below) chance performance in the error-inducing conditions, which suggests that non-grammatical factors can conceal knowledge of the English TC that some of them actually have. The two studies show that poor performance on the part of language learners----here, Heritage learners and L2 learners----is not necessarily due to a deficit in knowledge, which in turn suggests that linguistic competence can be obscured due to performance reasons.