Abstract
Mochizuki's (1998) Vocabulary Size Tests (VSTs) have three forms, which are made up of seven word-frequency levels (1,000, 2,000, 3,000, 4,000, 5,000, 6,000, and 7,000 word-frequency levels) to measure how many of the most frequently-utilized 7,000 words an examinee knows. In this study, the author will examine whether the rationale of estimating receptive vocabulary size of Japanese learners of English based on each word-frequency level, which adopts the stratified sampling method, is appropriate (in other words, it will partially examine the validity of Mochizuki's VSTs) from the psychometric viewpoint as well as reviewing the previous studies utilizing or analyzing Mochizuki's VSTs. University students majoring in English took six sections (2,000-7,000 word-frequency levels) in each of the three forms of Mochizuki's VST. Since six variables (six subtotals of correct answers in each section) were obtained in each form, Pearson's product-moment correlation coefficients among the six variables were calculated in each form, and exploratory factor analyses were conducted. The results showed that six sections (2,000-7,000 word-frequency levels) had high correlations with one another (especially with the closer word-frequency-level sections) and that only one factor was extracted in each of the three forms. This may indicate that no section (no word-frequency level) is heterogeneous and that each word-frequency-level section is consecutively connected. Theoretically no section (no word-frequency level) should be heterogeneous and each word-frequency level section should be connected successively. In this study, these theoretical corollaries may have been partially confirmed empirically from the psychometric results obtained through the interaction between the examinees and the VSTs.
