The way listeners perceive speech sounds is largely determined by the language(s) they were exposed to as a child. For example, native speakers of Japanese have a hard time discriminating between American English /ɹ/ and /l/, a phonetic contrast that has no equivalent in Japanese. Such effects are typically attributed to knowledge of sounds in the native language, but quantitative models of how these effects arise from linguistic knowledge are lacking. One possible source for such models is Automatic Speech Recognition (ASR) technology. We implement models based on two types of systems from the ASR literature—hidden Markov models (HMMs) and the more recent, and more accurate, neural network systems—and ask whether, in addition to showing better performance, the neural network systems also provide better models of human perception. We find that while both types of systems can account for Japanese natives’ difficulty with American English /ɹ/ and /l/, only the neural network system successfully accounts for Japanese natives’ facility with Japanese vowel length contrasts. Our work provides a new example, in the domain of speech perception, of an often observed correlation between task performance and similarity to human behavior.
more »
« less
Early phonetic learning without phonetic categories: Insights from large-scale simulations on realistic input
Before they even speak, infants become attuned to the sounds of the language(s) they hear, processing native phonetic contrasts more easily than nonnative ones. For example, between 6 to 8 mo and 10 to 12 mo, infants learning American English get better at distinguishing English and [l], as in “rock” vs. “lock,” relative to infants learning Japanese. Influential accounts of this early phonetic learning phenomenon initially proposed that infants group sounds into native vowel- and consonant-like phonetic categories—like and [l] in English—through a statistical clustering mechanism dubbed “distributional learning.” The feasibility of this mechanism for learning phonetic categories has been challenged, however. Here, we demonstrate that a distributional learning algorithm operating on naturalistic speech can predict early phonetic learning, as observed in Japanese and American English infants, suggesting that infants might learn through distributional learning after all. We further show, however, that, contrary to the original distributional learning proposal, our model learns units too brief and too fine-grained acoustically to correspond to phonetic categories. This challenges the influential idea that what infants learn are phonetic categories. More broadly, our work introduces a mechanism-driven approach to the study of early phonetic learning, together with a quantitative modeling framework that can handle realistic input. This allows accounts of early phonetic learning to be linked to concrete, systematic predictions regarding infants’ attunement.
more »
« less
- Award ID(s):
- 1734245
- NSF-PAR ID:
- 10211671
- Date Published:
- Journal Name:
- Proceedings of the National Academy of Sciences
- Volume:
- 118
- Issue:
- 7
- ISSN:
- 0027-8424
- Page Range / eLocation ID:
- e2001844118
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
In the first year of life, infants' speech perception becomes attuned to the sounds of their native language. Many accounts of this early phonetic learning exist, but computational models predicting the attunement patterns observed in infants from the speech input they hear have been lacking. A recent study presented the first such model, drawing on algorithms proposed for unsupervised learning from naturalistic speech, and tested it on a single phone contrast. Here we study five such algorithms, selected for their potential cognitive relevance. We simulate phonetic learning with each algorithm and perform tests on three phone contrasts from different languages, comparing the results to infants' discrimination patterns. The five models display varying degrees of agreement with empirical observations, showing that our approach can help decide between candidate mechanisms for early phonetic learning, and providing insight into which aspects of the models are critical for capturing infants' perceptual development.more » « less
-
In acquiring language, differences in input can greatly affect learning outcomes, but which aspects of language learning are most sensitive to input variations, and which are robust, remains debated. A recent modeling study successfully reproduced a phenomenon empirically observed in early phonetic learning---learning about the sounds of the native language in the first year of life---despite using input that differed in quantity and speaker composition from what a typical infant would hear. In this paper, we carry out a direct test of that model's robustness to input variations. We find that, despite what the original result suggested, the learning outcomes are sensitive to properties of the input and that more plausible input leads to a better fit with empirical observations. This has implications for understanding early phonetic learning in infants and underscores the importance of using realistic input in models of language acquisition.more » « less
-
Infants learn the sound categories of their language and adults successfully process the sounds they hear, even though sound categories often overlap in their acoustics. Most researchers agree that listeners use context to disambiguate overlapping categories. However, they differ in their ideas about how context is used. One idea is that listeners normalize out the systematic effects of context from the acoustics of a sound. Another idea is that contextual information may itself be an informative cue to category membership, due to patterns in the types of contexts that particular sounds occur in. We directly contrast these two ways of using context by applying each one to the test case of Japanese vowel length. We find that normalizing out contextual variability from the acoustics does not improve categorization, but using context in a top-down fashion does so substantially. This reveals a limitation of normalization in phonetic acquisition and processing and suggests that approaches that make use of top-down contextual information are promising to pursue.more » « less
-
null (Ed.)Abstract Neurobiological models of speech perception posit that both left and right posterior temporal brain regions are involved in the early auditory analysis of speech sounds. However, frank deficits in speech perception are not readily observed in individuals with right hemisphere damage. Instead, damage to the right hemisphere is often associated with impairments in vocal identity processing. Herein lies an apparent paradox: The mapping between acoustics and speech sound categories can vary substantially across talkers, so why might right hemisphere damage selectively impair vocal identity processing without obvious effects on speech perception? In this review, I attempt to clarify the role of the right hemisphere in speech perception through a careful consideration of its role in processing vocal identity. I review evidence showing that right posterior superior temporal, right anterior superior temporal, and right inferior / middle frontal regions all play distinct roles in vocal identity processing. In considering the implications of these findings for neurobiological accounts of speech perception, I argue that the recruitment of right posterior superior temporal cortex during speech perception may specifically reflect the process of conditioning phonetic identity on talker information. I suggest that the relative lack of involvement of other right hemisphere regions in speech perception may be because speech perception does not necessarily place a high burden on talker processing systems, and I argue that the extant literature hints at potential subclinical impairments in the speech perception abilities of individuals with right hemisphere damage.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1073/pnas.2001844118
