Learning spoken word forms is a vital part of second language learning, and CALL lends itself well to this training. Not enough is known, however, about how auditory variation across speech tokens may affect receptive word learning. To find out, 144 Thai university students with no knowledge of the Patani Malay language learned 24 foreign words in a CALL program for receptive vocabulary learning. Word meanings were presented in line drawings, and their spoken forms presented by either one or three Patani Malay speakers. Several other variables were included for control. Logistic regressions with and without adjustment by other factors suggested learners recognized somewhat fewer word meanings when trained with three Patani Malay speakers, especially under varied image conditions. This finding implies fragile initial auditory representations, and limited processing resources. The application of current results and their interpretation to CALL may be to limit sources of variation that would otherwise direct attention away from word forms.