Measuring accuracy and speed of L2 kanji word identification

Research output: Chapter in Book/Report/Conference proceedingConference contribution


This paper investigates the quality of interlanguage lexical representations in reading kanji (Chinese characters) compound words. Within second language (L2) Japanese pedagogy, it has been well established that learning to read and write Japanese orthography, especially kanji, is an outstandingly difficult task for learners from non-kanji background. This study includes the first demonstration of frequency effects on naming latencies of kanji compounds for English-speaking (EJ) learners of Japanese at different proficiency levels: literate novice learners (n=18), and skilled advanced learners (n=11).

In order to test accuracy and speed of word pronunciation, we need to consider the frequency of the words, as this is a fundamental factor in word recognition (e.g. high frequency words are pronounced faster and more accurately than low frequency words). This created a methodological problem for the present study. First of all, there is no previous literature in L2 acquisition of Japanese that deals with frequency issues of kanji compound words. Next, it is difficult to be convinced that L1 word frequency databases are relevant to our literate beginners, whose language exposure is limited to classroom learning (i.e. tutored acquisition). Accordingly, this study employs an original L2 frequency database, from which test words are drawn for two reading tasks (the pronunciation task and the naming task). I found several lines of evidence for the well-documented L1 frequency effect in the L2 naming task. Firstly, a significant negative correlation confirmed that EJ readers' naming latencies decreased as word frequency increased, and vice versa. Secondly, we found evidence that the rate of correct pronunciation was correlated positively with word frequency in the novice group. This line of evidence suggest that word frequency affects the quality of early interlanguage lexical representations. Thirdly, I found evidence that a low-frequency ambiguous word caused an L2 reading difficulty, which may not be associated with a particularly high latency (i.e. speed-accuracy tradeoff). Since the Second Language Acquisition (SLA) literature has just begun exploring the frequency effect on L2 reading, this finding suggest a potential interest in the use of Psycholinguistic-oriented methodologies for SLA research such as measuring Reaction times of L2 readers.
Original languageEnglish
Title of host publicationVocabulary knowledge and use: measurements and applications, Bristol
Publication statusUnpublished - Jan 2004


  • interlanguage
  • lexical representations
  • kanji


Dive into the research topics of 'Measuring accuracy and speed of L2 kanji word identification'. Together they form a unique fingerprint.

Cite this