During silent reading, readers take longer to read sentences with repeated word-initial phonemes (Twenty toys were in the trunk) compared to matched sentences with unrepeated phonemes (Several games were in the chest) (McCutch & Perfetti, 1982). In a recent poster, Taylor, Eskenazi, and Folk (2015) found evidence that readers with high working memory do not show as much disruption as readers with lower working memory. This provided evidence that both sound codes and working memory is involved in word identification during silent reading. In the current follow-up experiment we used eye-tracking to determine whether the disruption occurs early or late in sentence reading and whether this slower reading is also related to comprehension difficulty. Participants read 21 sentences split into three conditions: six repeated phonemes, three repeated phonemes, and zero repeated phonemes. Participants’ eye movements were monitored using an EyeLink 1000 Plus eye-tracker. Participants answered comprehension questions after each sentence then completed a reading span task to measure working memory. We found that reading times slowed with each additional repeated phoneme, which indicates that sound codes are used in identifying words. We also found greater disruption in reading comprehension in sentences with more repeated phonemes. This experiment provides further evidence that sound codes are used to identify words during silent reading, and that sound codes are also used in comprehending the meaning of sentences.