Factors affecting the incidental formation of novel suprasegmental categories

Loading...
Thumbnail Image

Date

2021-11-23

Authors

Wright, Jonathan

Journal Title

Journal ISSN

Volume Title

Publisher

University of Oregon

Abstract

Humans constantly use their senses to categorize stimuli in their environment. They develop categories for stimuli when they are young and constantly add to existing categories and learn novel categories throughout their life. A key factor when learning novel sound categories is the method a person uses to acquire the novel sound categories. Different learning methodologies interact with different neural processes and mechanisms, leading to diverse learning outcomes. However, auditory learning research has only recently begun to focus on the ways that various auditory processing structures interact with different learning methodologies. This dissertation investigates the acquisition of novel tone categories using natural tokens and an incidental learning paradigm. Throughout the experiments we demonstrated that native English participants with no prior experience with the target tone categories, from 18 to 66 years old, can use an incidental learning paradigm with natural tokens to form four novel tone categories after 30 minutes of training with very high, even perfect, accuracy. These findings confirm results from previous studies that suggest that participants can effectively learn novel sound categories through incidental learning paradigms, and we extend the investigation of factors impacting incidental learning into natural speech sound categories. Across the four experiments we examined factors known to impact novel sound category acquisition. We demonstrated that high variability of tokens within trials resulted in greater learning than when the variability was spread out across trials. We also demonstrated that training on a single talker results in robust learning to novel tokens but a sharp decline when generalizing to novel talkers. By contrast, if participants are trained on multiple talkers during training, there is less learning, but there is little or no difference when generalizing learning to novel talkers. We also demonstrated that the presence of an unfamiliar vowel in the auditory stimuli did not impact the incidental formation of novel tone categories during perception only training. Further, we demonstrated that producing the tokens on each trial destroyed perceptual learning, and we presented multiple hypotheses regarding the nature of the disruption for future investigation. We also demonstrated that the presence of an unfamiliar vowel did not further disrupt perceptual learning over training with familiar segments. Thus, as a whole, this dissertation illustrated that incidental learning paradigms are an effective and efficient means for learning novel tone categories and investigating factors known to impact novel sound category acquisition.

Description

Keywords

Incidental learning, Novel sound category learning, Reinforcement learning, Speech perception, Speech production, Tone category learning

Citation