The use of computers has increased in speaking assessments; however, there are concerns about how the absence of an interlocutor affects performance on speaking tests.
This is an interesting research article, looking at the results obtained when students use either face-to-face or online testing. The conclusions highlight a number of ways in which the research design was limited, but this also points to perhaps doing the same research again with different students, contexts and researchers. It assumes that being able to test by computer-delivered modes is something to aim for; this could also be deconstructed a bit more. However, it opens a number of possibilities for online learning.