Eyeblink activity was examined based on a discrete-trial paradigm in which 17 subjects engaged in a semantic priming task. They were presented a series of pairs of words (prime and target) associatively related or associatively unrelated and were required to name them as quickly as possible. Cognitive load or difficulty of naming of the target word was also manipulated. The latency data for naming of the target word showed a typical priming effect, namely, facilitation of naming speed by the associatively relate prime word. The eyeblink rate changed synchronically to onset of stimuli as a function of prime-target relationship and cognitive load during the task. Specifically, the eyeblink rate was suppressed during presentation of the target word and peaked just after that. This study showed that suppression of eyeblinks was longer on trials with high cognitive loads than on those with low cognitive loads and the peak of eyeblink rate was higher on trials for which the prime and target were unrelated than on those for which they were related. These results suggested that the eyeblink activity is influenced by inner cognitive processes of word recognition.