<b>Acoustic Variability: Comparing L1 and L2 Computer- and Human-Directed Speech</b>

<p dir="ltr">This study investigated how L1 English speakers and L2 English speakers with Farsi as their first language modify their speech when interacting with a human teacher versus a virtual assistant. The analysis focused on segmental features associated with clear speech – vowel space, vowel tenseness, voice onset time (VOT) contrast, and vowel duration as a function of final voicing – to assess whether speakers adjust their speech differently based on the perceived interlocutor. English speakers showed a vowel space expansion in the computer condition, which aligns with our predictions. However, most acoustic measures did not show the expected effects of condition. There were no significant main condition effects for vowel duration or for absolute and relative VOT. Contrary to our predictions, voiceless stops had longer relative VOTs in the human-directed condition, and English speakers produced greater spectral contrast between tense and lax vowels when speaking to a human than a computer. Overall, the results suggest that the expected clear speech adaptations in the computer-directed condition were either absent or reversed. Several factors may have contributed to these outcomes, including the cognitive demands of the production task, vocal fatigue over the course of the recording session, participants’ attitudes toward the virtual assistant, or some other aspects of the experimental setup.</p>