investigation of the predictors of vocabulary knowledge across proficiency levels

in the past decades, we have witnessed a renascence of interest in vocabulary assessment (bachman, 2000). however, most researchers have focused on testing breadth rather than depth of vocabulary or receptive-productive dimensions of vocabulary knowledge (read, 2000; nation, 2001; wesche & paribahkt, 1996). the present research investigates the predicting variables of vocabulary knowledge across two (high and low) proficiency levels, using the three dimensional framework of henriksen (1999) which is composed of breadth, depth, and receptive-productive dimensions. 73 iranian efl students studying at universities or language institutes participated in this study. they were divided into a high and a low group based on their mean scores and standard deviation on toefl. breadth of vocabulary knowledge was investigated using predictors of vocabulary size (nation & beglar, 2007) and knowledge of words from different frequency bands (zareva, 2005). receptive-productive dimension was examined by a new measure developed for this study, and depth was studied using depth of vocabulary knowledge test developed by qian (2002). the results of stepwise regression indicate that the model which best predicts total vocabulary knowledge for the low group is a single predictor one consisting of vocabulary size, while the best model for the high group is a single predictor one consisting of knowledge of words from different frequency bands. the article concludes with a discussion of the findings and their possible implications for designing vocabulary tests.