In the paper, we claim that analyzing eye movement patterns while reading aloud can provide new evidence on multichannel processing. The current study focuses on the unfilled (silent) hesitation pauses that occur during unprepared reading in Russian as a first and Japanese as a second languages. We analyzed the pauses in the oral texts produced by 10 native Russian speakers while reading two text fragments in Japanese and their Russian translations. The eye movements were recorded while the participants were reading the texts. According to the results, the pause duration in L1 can be related to the time spent on the text processing both before and during the hesitation, whereas in L2 the hesitation duration is associated only with the processing time of the following segment. Moreover, when reading aloud in L2, where the number of pauses within the sentences exceeded the number of hesitations at the sentence boundary, the participants spent more time processing the text that followed the hesitation within a sentence, compared to sentence boundaries. In L1, however, we did not find any difference in text processing considering the location of the hesitation, although there were more pauses at sentence boundaries than within one sentence in Russian. The results suggest that in Russian as L1, the readers spend the same time after encountering a difficulty within a sentence and when preparing the next one, whereas in a second language the mid-sentence planning requires more time.