Language proficiency testing: a comparative analysis of IELTS and TOEFL

Abstract

There is a general belief that British and North American EFL proficiency tests represent radically different approaches to language test development. The North American tradition in language testing is heavily based on psychometric properties of tests such as reliability, and concurrent and predictive validity, whereas the British tradition is more focused on the specification of test content and expert judgement. Language proficiency tests, in either the .American or British tradition, are designed to serve different purposes, so they may not be comparable in terms of defined purposes. Nevertheless, the term 'language proficiency', no matter how it is defined, implies that we are referring to a monolithic concept. In the real world, test results are often used for screening purposes; the candidates' ability to cope with the future language medium is predicted by the proficiency criterion. If it is the case that language proficiency tests are used for similar purposes, i.e., measuring the general language ability of the candidates, comparability of such tests is a legitimate matter.This study compares two English language proficiency test batteries: TOEFL and IELTS. The main objectives of the research were to investigate the extent to which TOEFL and IELTS are comparable in terms of: a) the operational definitions of language proficiency on which the two tests are based, b) the degree to which the two tests provide similar information concerning the abilities of the testees. Analysis of test content suggests that both batteries are based on the notion that proficiency is divisible by skill (e.g. reading) and element of language (e.g. syntax), thus we have tests of reading, writing, listening, speaking, as well as tests of grammar and vocabulary. However, the tests differ in their representation of the scope of skills and elements of language proficiency. The analysis also shows that the TOEFL differs from the IELTS in its method of testing. Despite these differences in test methods and scope, to a great extent both tests measure a common aspect of the subjects' language ability, therefore their internal structures are unifactorial. A g-factor (general language proficiency) comprises much of the total variance in both tests. Additional information provided by each test regarding the competence in reading, listening, writing, and knowledge of vocabulary and grammar, does not seem to contribute much to the total variance. A content analysis of the two tests indicates that, in fact, there are more similarities between the reading and listening comprehension sections of the two batteries than differences. This is supported by the factor analysis of the test scores.