PhD Projects
During my PhD at the Max Planck Institute for Psycholinguistics in Nijmegen, I investigated a variety of research questions related to talker familiarity, speech perception, speech production, and the link between the latter two. Below is an overview of these projects and links to related publications. My PhD thesis can be found here. Expand the tabs below to find out more about each project!
Convergence (also known as alignment, accommodation, [speech] imitation, etc.) refers to the tendency of interlocutors gradually sound more like each other over time. While convergence has been observed in sentence structure, word choices, and even gestures, my research has focused on the sound level. That is, I was concerned with the mechanisms that drive how interlocutors "sound" more like each other at the acoustic-phonetic level. As a stepping stone for later projects, I ran an experiment that tested whether participants would converge to a subtle shift in one acoustic feature of one speech sound in Dutch. Namely, I manipulated the second formant (F2) of the Dutch vowel /e/ (e.g., in "deeg") to match that of the Dutch vowel /ø/ (e.g., in "deugd"). As all other acoustic features of /e/ remained identical, this change did not result in participants perceiving the vowel /e/ as /ø/. More importantly, though some participants did lower their F2 after exposure to manipulated speech, I found no robust convergence at the population level. These results challenged "resource-free" accounts of convergence at the phonetic level and suggested that the perceptual salience of changes in incoming speech may influence whether these changes are converged to at the phonetic level.
Publications
Uluşahin, O.*, Bosker, H.R., McQueen, J.M., Meyer, A.S. (2023). No evidence for convergence to sub-phonemic F2 shifts in shadowing. Proceedings of the 20th International Congress of the Phonetic Sciences (ICPhS 2023). pp. 96-100. Link
Some of the variability in incoming speech is addressed by mechanisms that are not talker-specific. In fact, they might not even be language-specific, as hearing non-linguistic stimuli can easily affect one's perception of speech as well. Such "normalizations" are often contrastive, meaning that they rely on the "popping out" of individual features of incoming speech. For instance, if someone has a high pitched voice, hearing their voice biases the perceptual system such that incoming speech sounds all appear to have lower pitch after exposure to this higher pitch. This adjustment is observable in lab experiments as effects persist over short periods of time. In a series of 6 experiments, I asked the following questions: Can similar adjustments be facilitated through pre-existing talker pitch information (as opposed to just the pitch information in the incoming signal)? Are these adjustments contrastive like in signal-driven normalization? And what happens if the pitch information in the signal doesn't match the pitch information one has about a particular talker? The data revealed that listener adjust their perception based on what they know about a talker's typical way of speaking, in addition to the acoustic features of incoming speech. Crucially, I only found this utilization of talker information when acoustic information incoming speech was ambiguous or otherwise insufficient for disambiguation.
Publications
Uluşahin, O.*, Bosker, H.R., Meyer, A.S., McQueen, J.M. (2026). Contextual F0 cues may outweigh talker F0 cues in fricative perception. (In Press) Attention, Perception, & Psychophysics.
Uluşahin, O.*, Bosker, H.R., McQueen, J.M., Meyer, A.S. (2024). Knowledge of a talker's F0 affects subsequent perception of voiceless fricatives. Proceedings of Speech Prosody 2024. pp. 432-436. Link
How does knowledge about someone's typical way of speaking affect whether (and to what extent) you converge to their speech? What does this say about the nature of perception-production interaction in the brain? To answer these questions, I ran a series of three experiments: In the first, participants first provided their (baseline) mean F0s by reading 40 sentences out loud. Later, they performed a synchronous speech task in which they were instructed to try to speak at the same time as a model talker's recordings. The model talker's speech had been manipulated to have high or low pitch, depending on the experimental condition a participant was assigned. I found that participants in the high F0 talker group increased in their pitch, and that participants in the low F0 talker group decreased it. Thus, both groups converged to pitch. In the second experiment, I added a familiarization task between the reading and synchronous speech tasks. In this familiarization task, participants heard the model talker at her "typical" F0 for their talker group. This meant that participants who heard the model talker at high F0 would later perform the synchronous speech task with high F0 as well. Compared to Experiment 1, Experiment 2 resulted in a lower magnitude of convergence, but did not significantly reduce the probability of convergence. Finally, in Experiment 3, I modified the design of Experiment 2 such that the familiarization F0 and the synchronous speech F0 of the model talker would be different. So, if a participant had been familiarized with the model talker at high F0, they would have to do the synchronous speech task with her low F0 version. So, in Experiment 2, participants had congruent talker F0 information, and in Experiment 3, they had conflicting talker F0 information. Once again, compared to Experiment 1, Experiment 3 resulted in both a smaller magnitude of convergence and a significantly smaller probability of convergence. These results indicated not only that information about a talker's speech characteristics acquired solely through listening are used in speaking, but also that the reliability of this talker information has the ability to modulate phonetic convergence.
Publications
Uluşahin, O.*, Hasan, E. P., Bosker, H.R., McQueen, J.M., Meyer, A.S. (2026). The role of talker information on phonetic convergence. (Under revision).
During my PhD, I was affiliated with the following research groups: