The Gender Artifact: Predicting Speaking Proficiency in Russia-bound College Students

Heather Smith, Princeton University

This paper examines gender as a predictor of speaking proficiency ratings in Russia-bound college men and women. Previous research on study-abroad data (e.g., Brecht et. al., 1993) has shown that gender is significantly associated with in-country gain in speaking proficiency. Overall, women achieve significantly smaller improvements in proficiency ratings than their male colleagues. However, previous studies found no gender-based difference in initial speaking proficiency and, when discussing discrepancies in in-country gain, even rely upon the apparent absence of gender bias in pre-program ratings. (ibid., Brecht and Robinson, 1993) A study of government subjects also showed no significant gender discrepancy in speaking proficiency (Ehrman and Oxford, 1995).

This study examines data from the ACTR Study Abroad Database on 625 undergraduate students who participated in ACTR semester and year-long programs from 1986-1996 (excluding Fall 1994 and 1994-1995). The subjects studied included a large population of both men (255 of 625, or 40.8%) and women (370 of 625, or 59.2%), with a mean age of 20.6. Speaking proficiency, measured with Oral Proficiency Interviews (OPIs), is captured by two variables. An ordinal variable reflects the broad range of OPI ratings. A second, binary variable indicates whether the student has crossed the so-called "critical threshold" between Intermediate and Advanced speaking proficiency.

The results indicate that, on average, women received a pre-program OPI rating 0.28 lower than men (t-test, p<.001). Multiple regression models were used to examine the effects of adjustment for age, years of college and high school Russian, previous study abroad in Russia or the former Soviet Union, year of participation, and previous study of Slavic and non-Slavic languages. The final model gave an adjusted mean difference of 0.30 (multiple linear regression, p<.001). In addition, only 4.8% of women, compared with 8.6% of men, crossed the critical threshold to achieve a speaking proficiency rating of Advanced or higher (chi-square, p=.056). This finding was strengthened by logistic regression models that adjusted for the above variables. The final, adjusted model estimates that males are 2.2 times as likely as females to achieve an Advanced rating (multiple logistic regression, p=.03).

The OPI provides a global rating: it indicates the highest level at which a subject consistently completes target-language tasks. The OPI is not designed to describe the subjects relative strengths and weaknesses; investigation of specific differences in men and women's OPI performances therefore lies outside the scope of this analysis. Moreover, it is impossible to determine from this data alone whether the gender discrepancy is an artifact of bias in the OPI testing procedure or whether the OPI ratings reflect a real imbalance in the speaking proficiency levels of the population studied (a discrepancy that would presumably be an artifact of some aspect of the language classroom). Rather, the primary emphasis is to establish the existence of a substantial, significant gender-based discrepancy in OPI ratings of Russia-bound college students.