Early phonetic learning without phonetic categories: Insights from large-scale simulations on realistic input
Early phonetic learning without phonetic categories: Insights from large-scale simulations on realistic input
复制标题
DOI：
10.1073/pnas.2001844118
复制
发表时间：
2021-02-16
影响因子：
11.1
通讯作者：
Dupoux，Emmanuel
中科院分区：
文献类型：
--
作者：
Schatz，Thomas;Feldman，Naomi H.;Dupoux，Emmanuel
在他们说话之前，婴儿就已经适应了他们所听到的语言的声音，比非母语更容易处理母语的语音对比。例如，在6 ~ 8个月和10 ~ 12个月之间，学习美式英语的婴儿比学习日语的婴儿更善于区分英语和[l]，比如“rock”和“lock”。对这一早期语音学习现象的有影响力的解释最初提出，婴儿通过一种被称为“分布学习”的统计聚类机制，将声音分组为母语元音和辅音类语音类别，如英语中的和[l]。然而，这种学习语音类别的机制的可行性受到了挑战。在这里，我们证明了一个分布式学习算法的自然语音操作可以预测早期语音学习，在日本和美国英语的婴儿，这表明婴儿可能会通过分布式学习毕竟。然而，我们进一步表明，与最初的分布式学习建议相反，我们的模型学习的单位过于简短，过于细粒度的声学对应的语音类别。这挑战了婴儿学习的是语音类别的有影响力的想法。更广泛地说，我们的工作介绍了一种机制驱动的方法来研究早期语音学习，以及一个定量建模框架，可以处理现实的输入。这使得对早期语音学习的描述与关于婴儿调音的具体、系统的预测联系起来。
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 thisearly phonetic learningphenomenon 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 amechanism-drivenapproach 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.