Back in 2007, we were observing students attempting our short-answer free-text e-assessment questions in a Usability Laboratory. One student repeatedly typed ‘decease’ instead of ‘decrease’ and he didn’t realise he was doing it. At the time, the answer matching was linguistically based and so, on the basis of the fact that ‘decrease’ and ‘decease’ have different meanings, the student’s answers were all marked as incorrect, even when they weren’t. He was not amused. As a short-term fix we added ‘decease’ as a synonym for ‘decrease’, and paradoxically our current very much simpler answer-matching copes fine in its default setting (which allows one missing letter in the middle of a word – as well as one additional letter and/or one transposition), enabling appropriate marking and tailored feedback for responses that contain malapropisms of this sort:
There are some interesting lessons to be learnt. In particular, our pre-matching spell-checker also works by recognising words in its dictionary, so we can’t do away with the spell-checking within PMatch anytime soon.
I have a real bee in my bonnet about the importance of looking at student responses to our questions and spotting common spelling errors is another reason for doing so. Having said that, I’ve not found any other students who use ‘decease’ instead of ‘decrease’ : ‘to’ instead of ‘too’ is much more common but that is unlikely to interfere with answer matching. However we also use student responses in identifying words that are not in the dictionary used by the pre-matching spell-checker – not that there have been very many of these either; mostly geological terms, and they are easy to add.
So far I have been assuming that we want to give students credit if their answer is correct apart from a spelling mistake. Not everyone would agree with me. In particular, my biologist colleagues tend to want biological terms to be spelt correctly. One of the benefits for us of of PMatch over IAT is that we are in control of what we allow and what we don’t and it is trivially easy to alter this on a question by question basis; so decease can be allowed instead of decrease without allowing cystol instead of cytosol.