Adult speech perception is tuned to efficiently process native phonetic categories, causing difficulties with certainnon-native categories. For example, Japanese has no equivalent of the distinction between American English /r/ and /l/ and na-tive speakers of Japanese have a hard time discriminating between these two sounds. Here, we ask whether standard AutomaticSpeech Recognition (ASR) systems trained on large corpora of continuous speech can make correct quantitative predictionsregarding such non-native phonetic category perception effects. By training an ASR system on language L1 and evaluatingit on language L2, we obtain predictions for a native L1 speaker tested on L2 phonetic contrasts. Using a variety of L1 andL2, we show that ASR models correctly predict several well-documented effects. Beyond the immediate results, our evaluationmethodology, based on a machine version of ABX discrimination tasks, opens the possibility of a more systematic investigationof computational models of phonetic category perception.

It has long been assumed that infants' ability to discriminate between languages stems from their sensitivity to speech rhythm, i.e., organized temporal structure of vowels and consonants in a language. However, the relationship between speech rhythm and language discrimination has not been directly demonstrated. Here, we use computational modeling and train models of speech perception with and without access to information about rhythm. We test these models on language discrimination, and find that access to rhythm does not affect the success of the model in replicating infant language discrimination results. Our findings challenge the relationship between rhythm and language discrimination, and have implications for theories of language acquisition.

In acquiring language, differences in input can greatly affectlearning outcomes, but which aspects of language learning aremost sensitive to input variations, and which are robust, remainsdebated. A recent modeling study successfully reproduced aphenomenon empirically observed in early phonetic learning—learning about the sounds of the native language in the firstyear of life—despite using input that differed in quantity andspeaker composition from what a typical infant would hear. Inthis paper, we carry out a direct test of that model’s robustnessto input variations. We find that, despite what the original resultsuggested, the learning outcomes are sensitive to properties ofthe input and that more plausible input leads to a better fit withempirical observations. This has implications for understandingearly phonetic learning in infants and underscores the impor-tance of using realistic input in models of language acquisition.

In the first year of life, infants’ speech perception becomesattuned to the sounds of their native language. Many accountsof this early phonetic learning exist, but computational modelspredicting the attunement patterns observed in infants fromthe speech input they hear have been lacking. A recent studypresented the first such model, drawing on algorithms proposedfor unsupervised learning from naturalistic speech, and tested iton a single phone contrast. Here we study five such algorithms,selected for their potential cognitive relevance. We simulatephonetic learning with each algorithm and perform tests onthree phone contrasts from different languages, comparing theresults to infants’ discrimination patterns. The five models dis-play varying degrees of agreement with empirical observations,showing that our approach can help decide between candidatemechanisms for early phonetic learning, and providing insightinto which aspects of the models are critical for capturing in-fants’ perceptual development.