This study investigates the role of hesitation pauses in spoken teacher discourse and their duration in the perception and short-term memorization of lexical items. While previous research has suggested that certain disfluencies, including hesi tation pauses, can facilitate comprehension and memorization, the specific conditions under which this occurs remain unclear. To address this gap, an experimental task based on auditory recognition was conducted among 72 schoolchildren aged 13–16. The stimuli consisted of authentic teacher utterances extracted from a corpus of Russian- speaking teachers’ speech and varied in terms of hesitation pause presence before target words. Participants were asked to identify whether the target word in an audio fragment matched the written version or had been replaced with a similar lexical item. A gen eralized linear mixed- effects model was used to analyze the accuracy of recognition, considering variables such as pause presence, pause duration, word frequency, word length, post-target pause, and the Levenstein distance between the original and the sub stituted word. Results showed that the presence of a hesitation pause before a word did not sig nificantly affect the recall accuracy. The duration of the pause, however, had a negative impact: longer pauses led to lower recognition rates. In contrast, shorter pauses, the pres ence of a pause after the target word and the stimuli in which target words were more frequent than the words for which they were substituted were positively associated with better recall. These findings suggest that hesitation pauses are not universally beneficial; their effectiveness depends on their temporal features and position. The study refines existing theories of speech disfluency processing by demonstrat ing that excessively long pauses may overload working memory or disrupt discourse coherence. These insights contribute to both the psycholinguistic theory and to optimi zation of spoken communication in pedagogical settings.