Testing of Inferencing Behavior in a Second Language

Abstract

The term ‘inferencing’ has been used in many texts and teaching books to mean a process or a discrete skill in reading and implies the process of gap-filling. Other texts call this ‘pragmatic inferencing’, meaning the incorporation of world knowledge into the meanings reconstructed during the processing of a text. This paper utilizes the term after Winne et al. (1993) to mean everything a reader does in the process of reconstruct-ing the meaning of a text. Our definition is synonymous with reading. Inferencing is a complex process and testing its products may never be accurate or even simple. The problems of testing SL inferencing may result from assumptions made by testers on the nature of reading, or test types to the presumptions and problems that readers bring into the testing situation. The study used 300 final year secondary school students who are SL speakers of English and administered two reading tests, one culturally familiar and the other culturally unfamiliar, based on three narrative texts per test. Four categories of inferences were tested in four different sections ‘A’ to ‘D’. The results showed that certain inference types were more difficult to make. Even Short-Answer Questions presented peculiar problems. Readers did significantly better on culturally familiar texts than culturally unfamiliar texts. The ability to identify the locus of an answer was not an adequate requisite for arriving at an acceptable answer.