Comparing two phonemic category learning mechanisms in infancy and the factors that modulate their efficacy

在生命的第一年，婴儿成为感知语音差异的专家，用来区分母语中的一个词和另一个词（音素）。这个过程被称为“调谐”，涉及到对母语语音（语音）区别的感知敏感性的增加，以及对非母语语音的敏感性的降低。在这项研究中，我们比较了两种语音类别学习机制，并询问它们是否同样能产生音素库来区分有意义的单词。我们研究了多感官信息的影响，以及每种机制由关键时期控制的程度。我们提出了六个实验集，使用EEG和ERP指标来测量学习、神经可塑性、识别和语义使用。第一组比较了两种学习机制：第一种是使用输入语音中相对频率分布的统计学习，第二种是基于单词形式到意义的映射。在这里，以及在第2-4集，我们测试了三组婴儿：开始调谐的婴儿（5个月），即将结束的婴儿（9个月）和之后的婴儿（12个月）。在学习阶段，我们测量了临界期可塑性的振荡生物标志物，以确定是否有一个（或两个）学习机制是由临界期控制的。在第二组中，我们增加了视觉语音线索（说话的面孔），在第三组中，我们增加了口头运动操作，以探索多感官信息如何与学习和/或可塑性的时间相互作用。第4组通过测试重新组织的语音类别是否产生可交换的音素，可以用来区分有意义的单词，来询问这两种学习机制在更高水平的语言习得中是否同样有效。为了探索可塑性是否已经完全关闭或可以重新打开，在第5组中，我们测试了年龄较大的儿童和成人在这两个学习过程中的表现，在第6组中，我们询问外部牵引伽马振荡是否可以重新打开成人的可塑性。总之，这些研究质疑了人类口语所需要的基础，即“模式的二元性”的建立——形式层面的模式（例如，语音）如何支持意义层面的模式。在某种程度上，这一基础成就受到生命早期关键时期的限制，并受到父母-婴儿面部互动中固有的多感官信息的影响，我们对语言进化的基本力量的汇合有了更深入的了解。这些结果也将具有重要的应用意义，无论是对于那些被剥夺与照顾者进行面对面互动的婴儿（例如，那些经历过极端早期逆境的婴儿），还是对于那些天生就有感觉或运动障碍的婴儿（例如，那些有视觉或口腔运动障碍的婴儿），这些婴儿限制了他们对多感觉信息的体验。

During the first year of life, infants become specialists at perceiving the speech sound differences used to distinguish one word from another (the phonemes) in their native language. This process, termed `attunement', involves an increase in perceptual sensitivity to native speech sound (phonetic) distinctions and a concomitant decrease in sensitivity to non-native ones. In this grant, we compare two phonetic category learning mechanisms and ask whether they equally well yield a phoneme repertoire to distinguish meaningful words. We examine the influence of multisensory information, and the extent to which each mechanism is gated by critical periods. We propose six experimental sets, using EEG and ERP indices to measure learning, neural plasticity, discrimination, and semantic use. Set I compares two learning mechanisms: the first using statistical learning of relative frequency distribution of sounds in input speech, and the second rooted in word form-to-meaning mappings. Here, as well as in Sets 2-4, we test three groups of infants: those beginning attunement (5 months), those toward the end (9 months), and those after (12 months). During the learning phase, we measure an oscillatory biomarker of critical period plasticity to determine whether either (or both) learning mechanisms are gated by critical periods. In Set 2, we add visual phonetic cues (talking faces) and in Set 3, oral-motor manipulations, to probe how multisensory information interacts with learning and/or the timing of plasticity. Set 4 asks whether the two learning mechanisms interface equally well with higher levels of language acquisition, by testing whether the reorganized phonetic categories yield commutable phonemes that can be used to distinguish meaningful words. To probe whether plasticity has closed off entirely or can be reopened, in Set 5, we test older children's and adults' performance in these two learning procedures, and in Set 6, we ask whether externally entraining gamma oscillations can reopen plasticity in adults. Together, these studies query the very foundation of what spoken human language entails, the establishment of `duality of patterning' - how patterning at the level of form (e.g., speech) supports patterning at the level of meaning. To the extent that this foundational achievement is constrained by critical periods in early life and is influenced by the multisensory information inherent in parent-infant enface interactions, we gain deeper insight into the confluence of forces fundamental to the evolution of language. These results will also have important applied implications, both for infants deprived of engaging enface interactions with caregivers (e.g., those who experience extreme early adversity) and for infants born with sensory or motor impairments that limit their experience of multisensory information (e.g., those with visual or oral-motor impairments).