Phonological recoding in sentence-level Chinese character recognition by advanced adult L2 Chinese learners

To form a descriptive basis for establishing the learning potential in adult second language acquisition (SLA), many SLA researchers have urged that studies on second-language (L2) ultimate attainment should identify the domains in which adult L2 learners are or are not able to attain native-like proficiency levels. The empirical study reported in this dissertation aims to explore whether a native-like lexical processing system can indeed be attained by advanced adult L2 learners. To this end, the study adopted the advanced-learner approach, employing multiple screening criteria to recruit 23 adult L2 Chinese learners whose non-native features could not be easily perceived by native speakers and 23 native controls. To probe their underlying lexical processing procedure and representational knowledge involved in sentence-level Chinese character recognition, this study drew on the theoretical and methodological insights garnered from the phonological recoding research (e.g., Xu, Pollatsek, & Potter, 1999), in which four online reading tasks were designed for and administered to the participants.; Analyses of the participants' online performance data revealed that, while the L2 learners and native speakers were comparable in terms of their overall Chinese reading ability (as verified by a Chinese proficiency test), similarities and differences co-existed between them with regard to their underlying lexical processing procedure and the nature of the activated lexical information. With respect to the similarities, both the native speakers and the L2 learners appeared to employ the same processing procedure in initial semantic activation and in subsequent semantic integration: phonological recoding was consistently performed by both groups to assist semantic activation long before the semantic code of a Chinese character was fully activated and continued to mediate the subsequent semantic integration process. Nevertheless, while both tonal and segmental information were engaged in the native speakers' semantic activation and integration processes, tonal information was available late and only effectively involved in the L2 learners' semantic integration process. In other words, the L2 learners were still not quite native-like in terms of the efficiency with which tonal information was activated online in sentence-level Chinese character recognition. Furthermore, the native speakers outperformed the L2 learners across all tasks, both in terms of the processing time that was required for semantic access (mean= 0.4847 seconds vs. 0.8748 seconds) and the accuracy rate with which a Chinese character was identified (mean= 92.89% vs. 88.24%).; Note that the observed non-target features (e.g., delayed temporal availability of tonal information in semantic activation) of the L2 learners' lexical processing system were only perceived in the laboratory experimental settings. Outside of the laboratory settings, these L2 learners all appeared to be native-like in comprehending Chinese characters. Thus, following Hyltenstam and Abrahamsson's (2000) definition of near-nativeness---"second language proficiency levels that are not identical to native-like levels but that fall short above the limit of perceivable non-nativeness" (p. 163; see also Hyltenstam & Abrahamsson, 2003), the L2 learners may operate on a near-native lexical processing system. Based upon the findings, pedagogical implications for L2 (Chinese) reading instruction were provided.