Show Less
Restricted access

Evaluating Tests of Second Language Development

A Framework and an Empirical Study


Khaled Barkaoui

The book introduces a framework for examining the validity of tests that aim to assess second-language (L2) proficiency development over time and/or in relation to L2 instruction. It also reports the findings of a longitudinal study that aimed to examine the sensitivity to change of a test of L2 proficiency development. Specifically, the study examined changes over time in Progress scores and the linguistic characteristics of essays written in response to Progress by learners who took the test before, during and after a period of L2 instruction in different countries. The book furthers our understanding of the nature of L2 proficiency as it develops over time and in relation to L2 instruction and provides a framework that can be used in future endeavours to design and validate tests of L2 proficiency development. The book is intended for graduate students, test developers, and researchers doing research in applied linguistics and L2 assessment.

Show Summary Details
Restricted access

Chapter 7: Changes in Discourse, Sociolinguistic and Strategic Aspects


←128 | 129→

Chapter 7Changes in Discourse, Sociolinguistic and Strategic Aspects

This chapter reports findings concerning changes in the discourse, sociolinguistic and strategic aspects of the Progress essays. It also reports findings concerning the relationships between essay linguistic characteristics and essay scores.

7.1.Changes in Discourse Aspects

Tab. 7.1 reports descriptive statistics for the three measures of cohesion (connectives, argument overlap, and LSA given/new) by test level and test occasion.

Connectives: Tab. 7.1 shows that connectives incidence did not vary much across test occasion; the average number of connectives was M = 109.77 per 1,000 words (SD = 37.42) at the beginning of the course; M = 108.82 (SD = 34.24) at the middle of the course; and M = 108.09 (SD = 39.15) at the end of the course. However, as test level increases, connectives incidence tends to increase as well from M = 103.48 (SD = 38.71) for Progress 15–30 to M = 110.83 (SD = 33.98) for Progress 55–70. The autocorrelations for connectives incidence (Tab. 7.2) were low, ranging between .07 (for times 2 and 3) and .16 (for times 1 and 3). MLM Model 1 for connectives (Tab. 7.3) indicated that (a) only 11 % of the variance was between learners; (b) the intercept variance was significant; and (c) only 25 % of the variation in the intercept is potentially explicable by learner-level predictors.

Subsequent MLM models indicated that (a) time did not have a significant association with connectives incidence; (b) the relationship...

You are not authenticated to view the full text of this chapter or article.

This site requires a subscription or purchase to access the full text of books or journals.

Do you have any questions? Contact us.

Or login to access all content.