An adaptive scheme for handling spelling errors by an e-learner while responding to the e-learning system through typed-in single-word responses is presented in this paper. To simulate the behaviour of a human instructor, the system preprocesses the input word with respect to spelling errors due to wrong letter or missing letter. The appropriately encoded input is then to fed into a neural net that intelligently recognizes the correct response, in spite of minor spelling mistakes committed by the learner. Results show that the scheme intelligent recognises the misspelled words as expected to be done by a human instructor. © 2010 IEEE.