Understanding the inferences that speakers rely on to communicate is a core part of listening comprehension and is, more broadly, an important aspect of communicative ability. As a result, theories of communicative language ability account for it, and language testers who try to gauge the proficiency of learners of a second language include it in their assessments. Within the field of language testing, much research has been conducted to better understand how different aspects of listening may contribute to difficulty for second-language learners. One area of investigation has been the notion of listening being separable into different subskills, such as listening for inferences as opposed to listening for specific explicit details or listening for main idea. However, there have been mixed results when attempting to determine the psychological reality of these subskills.

This study attempts to clarify this question via a listening comprehension instrument that was designed specifically to assess the comprehension of conversational implicature, or pragmatic inferencing, in contrast to non-implicature, or general comprehension. This balanced instrument was administered to 255 language learners in two item formats, multiple choice and constructed response. In addition, participants were administered short-term memory and working memory measures. A variety of analyses, including item response theory (Rasch), logistic regression, and confirmatory factor analyses were used to try to attain evidence for 1) the existence of a separable listening for conversational implicature subskill and 2) the validity of assessing this subskill through a multiple-choice format.

The results from the analyses generally converged to indicate that while conversational implicature contributes to difficulty, it is not a separable subskill. However, the results did show that the multiple-choice item format is a defensible method for targeting this skill. This leads to the conclusion that expending effort on assessing comprehension of conversational implicature in general language proficiency tests may not be necessary unless the test-use context places particular emphasis on this ability. Although it is an integral aspect of listening, from an assessment standpoint, performance on general listening items will likely give test users the information they need to make predictions about comprehension of conversational implicature ability.