CompCog: Infant Vocalization as Foraging for Caregiver Responses

在生命的第一年，婴儿开始发声，他们的咕咕声和咿呀学语变得越来越复杂和复杂，为说话铺平了道路。婴儿的发声通常是一串或一串相似的发声。甚至在婴儿会说话之前，照顾者就经常对婴儿发出的声音做出反应，而成年人的反应也会根据婴儿发出的声音类型而有所不同。这些成人的反应似乎会影响婴儿随后的发声。这项研究通过借鉴我们所知道的动物如何觅食的想法，系统地探索环境来寻找食物，为婴儿发声提供了一个新的解释视角。觅食时，动物们会在重新回到已经找到食物的熟悉地点和尝试可能导致未发现资源的新地点之间取得平衡。该项目将通过考虑语言前婴儿作为觅食者，探索各种可能的发声范围以寻找有趣或有用的反馈，包括来自成年人的积极反应，来促进对婴儿如何完善其发声能力的理解。全天的家庭录音将从3、6、9和18个月大的学习英语和西班牙语的婴儿那里收集。婴儿发声将被自动识别和声学分析，然后位于一个抽象的声学信息空间。对于声学空间的每个区域，将确定接收成人响应的可能性。以前开发的用于分析物理空间中资源觅食模式的方法将被应用。成人发声声学及其与婴儿发声觅食反应的关系也将进行类似的分析。婴儿和成人发声觅食的计算模型将被用来探索婴儿和成人发声在一天中观察到的时间动态的可能机制。将测试声音觅食措施在年龄和社会经济地位方面的变化。研究人员还将测试婴儿第一年的觅食模式能否预测其18个月大时的语言能力。这项工作有可能为自动获得的标记提供新的候选物，这些标记物可能有助于早期识别有语言迟缓或障碍风险的儿童。

During the first year of life, infants begin to vocalize, and their cooing and babbling become increasingly more sophisticated and complex, paving the way for speech. Infants' vocalizations often occur in strings or clusters of similar-sounding vocalizations. Caregivers frequently respond to babies' vocalizations even before they can talk, and adults vary their responses depending on the types of vocalization babies produce. These adult responses appear to influence babies' subsequent vocalizations. This research offers a new explanatory perspective on infant vocalization by borrowing ideas from what we know about how animals forage, systematically exploring an environment to find food. When foraging, animals strike a balance between revisiting familiar locations where food has been found and trying out new locations that may lead to undiscovered resources. This project will advance understanding of how infants refine their vocal abilities by considering prelinguistic infants as foragers who explore the range of possible vocalizations in search of interesting or useful feedback, including positive responses from adults.Daylong home audio recordings will be collected from English- and Spanish-learning infants at 3, 6, 9, and 18 months of age. Infant vocalizations will be automatically identified and analyzed acoustically, then situated in an abstract acoustic information space. For each region of acoustic space, the likelihood of receiving an adult response will be ascertained. Methods previously developed to analyze patterns of foraging for resources in physical space will be applied. Adult vocalization acoustics and their relationship to infant vocal foraging responses will be similarly analyzed. Computational models of infant and adult vocal foraging will be employed to explore possible mechanisms that underlie the observed temporal dynamics of infant and adult vocalizations over the course of the day. Vocal foraging measures will be tested for changes across age and with socioeconomic status. Whether foraging patterns during the first year predict speech-language abilities at 18 months will also be tested. The work has the potential to unveil new candidates for automatically obtainable markers that could be useful for early identification of children at risk for speech-language delays or disorders.