Abstract：
Children\'s own language production has a role in structuring the language of their conversation partners and influences their own development. Children\'s active participation in their own language development is most apparent in the rich body of work investigating language in natural environments. The advent of automated measures of vocalizations and movement have made such in situ research increasingly feasible. In this chapter, we review recent research on children\'s language development in context with a particular focus on research employing automated methods in preschool classrooms for children between ages 2 and 5 years. These automated methods indicate that the speech directed to preschool children from specific peers predicts the child\'s speech to those peers on a subsequent observation occasion. Similar patterns are seen in the influence of peer and teacher phonemic diversity on the phonemic diversity of children\'s speech to those partners. In both cases, children\'s own speech to partners was the best predictor of their language abilities, suggesting their active role in their own development. Finally, new research suggests the potential of machine learning to predict children\'s speech in group contexts, and to transcribe classroom speech to better understand the content of children\'s conversations and how they change with development.
摘要：
儿童自己的语言生成在构建对话伙伴的语言方面具有作用，并影响他们自己的发展。儿童积极参与自己的语言发展在自然环境中研究语言的丰富工作中最为明显。发声和运动的自动测量的出现使这种原位研究变得越来越可行。在这一章中,我们回顾了最近关于儿童语言发展背景的研究，特别关注在2至5岁儿童的学前教室中采用自动化方法的研究。这些自动化方法表明,从特定同伴针对学龄前儿童的语音预测了在随后的观察场合儿童对这些同伴的语音。在同伴和教师音位多样性对儿童对这些伴侣的语音多样性的影响中可以看到类似的模式。在这两种情况下，孩子们自己对伴侣的言语是他们语言能力的最佳预测指标，表明他们在自身发展中的积极作用。最后,新的研究表明，机器学习在群体环境中预测儿童语音的潜力，并转录课堂演讲，以更好地理解儿童对话的内容以及它们如何随着发展而变化。
