Lexical sophistication measures and writing proficiency: The case of Indonesian learners of Japanese (Published)
The present study tests two measures of lexical sophistication in writing proficiency (moving-average morphological richness and moving-average mean size of paradigm for testing the lexical diversity) and mean word length for testing writing form (plain, humble, honorification). The findings suggest that the three metrics work reliably. Regarding lexical diversity, moving-average of morphological richness (MAMR) and moving-average mean size of paradigm (MAMSP) of Indonesian Japanese learners-written texts are close to native Japanese-written data. Lexical complexity measured by word length by Indonesian Japanese learners is characterised by slightly less richness than native Japanese data but remains very close. Word length-frequency relationship in the Indonesian-written data presents outstanding fitting results to nine models, including the Poisson Model families and Binomial Model families, with 0.9918 as the lowest and 0.9987 as the highest determination coefficient R2. It is hoped that this study’s outcome may help develop an automatic evaluation of the writing proficiency of agglutinative languages with diverse writing forms.
Keywords: L3 acquisition, dependency direction, mean dependency distance, word length