CAA Validation in NIECELT and some Pedagogical Issues in Question Design and Content of an Innovative Cloze

The National Institute of Education (NIE) in Singapore has developed a comprehensive testing software known as “NIE Computerised English Language Test” or, simply, NIECELT, that is capable of administering a test or a number of tests to any specified number of examinees at the same time or at different times. Inherent in NIECELT are a number of interactive test questions which allow the examinees to craft some of their answers in response to a set of questions that assess the examinees’ proficiency in language structure, grammar, vocabulary, reading-comprehension and the use of language in context. An issue challenging NIECELT as a computer-assisted assessment tool pertains to the defensibility of inferences made from the obtained test scores. Related empirical questions include: 1) How well does performance on NIECELT reflect examinees’ language proficiency compared to that measured by the English Language GCE ‘O’ level examination? 2) What are the relationships among the different test questions in NIECELT presented to examinees? This paper attempts to provide evidences of both the internal and external structures of NIECELT. Implications of the findings are discussed in terms of the pedagogical issues in question design and content of one interactive question type – a modified cloze procedure that tests essentially language in context. The innovation in question type in the cloze will also be clarified.