Datorbaserad bedömning i språk forskningsläge, erfarenheter och implikationer
To support uniform and fair assessment of Swedish students? English language competences, mandatory National tests are provided both for secondary and upper secondary education. The current thesis may be seen as a contribution to the further development and validation of these tests, with regard to the likely introduction of computerized testing methods. The aim is to gain further knowledge from current research in the field, to study and elucidate international experiences of implementation, and to gather and analyze feedback from students and teachers. Based on my findings, some tentative conclusions will be drawn, and recommendations made concerning future developments. During the last few decades technology has become an integrated part of language use in most domains. When reading, writing and communicating in other ways, computers are replacing traditional paper-and-pencil methods. Assessment has to follow and adjust to this (r)evolution. Computer based language materials for teaching and assessment are widely spread and many countries have introduced large-scale computer based language testing. In Sweden this journey has cautiously begun and it is probably more a question of how, than if the Swedish national tests of English will be computer based in the near future. The study shows that the most important issues in research about computer based assessment concern different aspects of validity. The impact of rubric on the assessed construct is essential, that is the impact of ?the how? on ?the what?. Depending on the extent to which software, innovative formats and the Internet are involved in the assessment process, the test construct must be revised and redefined accordingly. A possible new construct, as well as validity aspects of equity/fairness, practicality, accountability and good practice are emphasized in the literature. Washback-effects at different levels are discussed, and several researchers raise warnings against letting technology drive the development of good assessment towards ?efficiency? and overuse of simple formats. The empirical study undertaken was carried out at six Swedish schools, with more than 200 students and their 12 teachers. Contacts with the University in Bergen made it possible to use national computer based tests of English reading proficiency, already implemented in Norway. The study was performed using a case study approach. Each school was visited twice, for introductory meetings and for observation of the test taking process; field-notes were taken on these occasions. Questionnaires were answered, by students before and after taking the test and by teachers and head teachers in retrospect, with similar questions to allow comparisons. Data were analyzed inductively from both a quantitative and a qualitative perspective, and allow certain triangulation. The results indicate that the participating schools do not (yet) have the required resources to perform national computer based tests in a secure way, and the respondents agree that computers are seldom used in their English classrooms. All respondents are positive to the idea of computer based testing of English and to the actual tests as such, younger students, and boys, being the most motivated. The students also point out motivational aspects concerning, for instance, anxiety and show valuable insight in validity related aspects. The conclusion of the thesis is that the [likely] implementation of computerized testing methods in the Swedish school system must build on research and needs to take into account opinions and suggestions of future users, not least those expressed by students.