Word recognition of spontaneous speech is influenced by many cues in the signal. The present study investigates the effects of duration on human processing of spontaneous speech. Participants were presented with visual and audio representations of sentences that had one to four words taken out and replaced with a silent gap. Silent gaps were either kept the same as the original duration or manipulated to be 0.5 times shorter or 2 times longer than the original length of the words that had been replaced. Participants were asked to respond using a keyboard to type which word or words they believed fit in the gap. Subsequent responses were classified as either correct or incorrect. The results demonstrate that having access to any contextual auditory information increases correct response rates. This indicates that there are likely cues in the auditory context that allow the listener to resolve the target items that are not present when the trials are presented visually. We also find a trend in the data which seems to indicate that shortened gap targets are harder to predict than the other two conditions (original length and lengthened gap), and that the lengthened gap targets are consistently easier to predict. We interpret this result as indicating that when listeners have additional processing time it allows them to more accurately identify the missing words. Further, we believe that there are durational constraints in the mental lexicon during word selection processes which suggest a mechanism of activation and competition that has been used in various models of spoken word recognition.