Issues in Promoting Multilingualism. Teaching – Learning – Assessment

Page 322

Computer Assisted Language Assessment

321

Some of the major concerns with adaptive tests are encapsulated by Chapelle and Douglas (2006), who point out that in order to design a reliable adaptive test a large item bank has to be compiled and pretested, and in order to ensure that the target construct will truly be measured a wide range of test tasks has to be designed. They further observe that the construction of an appropriate item selection algorithm is a challenge and requires a proper understanding of Item Response Theory (IRT), on which the mechanics of an adaptive test should be based. In addition, certain language abilities, such as communicative language ability, cannot as yet be tested through adaptive tests, and it may be decades before adequate technology appears. Finally, Chapelle and Douglas (2006) question the arguments of test security and reduced learner stress, put forward by the advocates of adaptive testing. They claim that test security may be compromised by testees memorising critical items and sharing them with others, and add that testees may feel depressed by the number of difficult items they are exposed to in the course of an adaptive test before their ability is diagnosed. Although web-based authoring / delivery systems dedicated to adaptive testing, e.g. SIETTE (GuzmĂĄn et al. 2005), have already been created, they do not eliminate the above-mentioned issues. An example of a web adaptive test is The Business Language Testing Service, which ‘allows on-demand assessment of all levels of language ability in English, French, Spanish and German’ (BULATS). It tests reading and listening skills, and its sample is available on the BULATS website at https:// bulatsdt.bulatsonline.org/appAdmin/dummyTest.html?method=language Selection.

5. CALT advantages and constraints CALT cannot be treated as the ultimate assessment mode as, despite the benefits it brings, it has numerous limitations. Lists of both were proposed by e.g. Silye and Wiwczaroski, who had reviewed the considerations of Brown, Bock and Mislevy, Stevenson and Gross, and Henning (Silye & Wiwczaroski 2002), as well as Chapelle and Douglas (2006). However, dividing the characteristics of CALT dichotomously into advantages and disadvantages is an oversimplification, because what seems advantageous in computerised assessment is potentially constrained either by the very


Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.