Although many people think of computer technology as a recent development, computer technology has been used in language assessment for a long time. Large mainframe computers have been used since the 1960s for the analysis of test data and for the storage of items in databases, or item banks, as well as for producing reports of test results for test users. More recently with the advent of the personal computer, it is now common to use word-processing software for the creation and modification of test tasks, as well as for all of the tasks previously done on a mainframe computer. Perhaps the most striking change that computer and information technology has brought to language assessment, however, is the potential for delivering a wide variety of test tasks online anywhere in the world, and providing immediate feedback, 24 hours a day, seven days a week. This potential for expanding the kinds of tasks that we can deliver to test takers has been accompanied by everincreasing capacity for scoring responses by computer. While selectedresponse items have been scored by optical scanners for half a century, recent advances in natural language processing and latent semantic analysis, along with improvements in scanning technology, have made it possible to score written responses to constructed response tasks – both open-ended and short-answer tasks and responses to composition prompts – by computer.

However, along with these potential advances come potential problems and concerns. To what extent might the use of multimedia in assessment tasks introduce features that actually diminish the validity of the interpretations and uses we make of assessment results?