) showed that, when asked to choose a word to complete a sentence fragment that was either predictive of the follow-up word (e.g., For commuting to work, John got a 10-speed … was predictive of bike/bicycle ) or neutral (e.g., Last week John finally bought himself a new … ), participants more often chose a shorter word in the predictive context than in the neutral context. Across three experiments, we examined the broader role this principle has in language production and comprehension. In Experiments 1 and 2, we found that participants were more likely to supply a shorter word in the predictive than neutral context. Using self-paced reading, Experiment 3 found no facilitation in reading times when shorter words were used in a predictive context and longer words were used in a neutral context. In all, these findings suggest that information on short/long word patterning affects language production but not comprehension.