This study investigated the relative predictive power of vocabulary depth and reading fluency on the reading comprehension of advanced Korean EFL learners in college. By doing so, the scope of paradigmatic relations, as part of vocabulary depth, was extended to encompass associative vocabulary as well as synonyms and antonyms, and reading fluency at the discourse level was considered. For this study, 139 college students were tested on a range of vocabulary depth tests, as well as reading comprehension and text-level reading fluency. The findings revealed that although both vocabulary depth and reading fluency are significant contributors to reading comprehension abilities, the predictability of vocabulary depth was larger than that of reading fluency. In addition, associative vocabulary not only revealed additional predictive power for reading comprehension on top of reading fluency, synonyms and antonyms, but also showed stronger predictability compared to synonyms and antonyms. These results highlight that both vocabulary depth, especially the knowledge of how words are related together, and text-level reading fluency play a crucial role in boosting the reading comprehension abilities of even advanced L2 readers.

The purpose of the present study is to investigate what the memory representation of L2 text is like based on the Causal Network Model. In order to do that, 8 stories were read in English by Korean students and recalled in Korean. Their recall was analysed in terms of the number of causal connections each sentence has as specified in the model. And then it was compared with the results of Kim (2001) where Korean students read and recalled the same stories in Korean. The overall amount of recall was not different between L1 and L2 texts, but the pattern of recall showed differences in terms of the causal structure proposed by Causal Network Model. While the recall of L1 text was nicely accounted for by the number of causal connections specified in the model, the recall of individual goal statements in L2 text did not reflect the causal structure. Interesting was the finding that the more important goal among the two goal statements was recalled better for L2 than for L1 text.