COLLABORATIVE RESEARCH: Social-Emotional Analysis of the Language Environment (SEAL): Key Word & Phrase Spotting in Early Childhood Care Settings

o本项目由“发展科学计划”及“促进竞争研究既定计划”联合资助。孩子们在学校和生活中的成功取决于他们早年是否学会了某些社交和情感技能。为了培养教师促进社会情感学习的能力，需要一种工具，可以为教师提供基于数据的关于社会情感学习的反馈。这个项目创造了这样一个工具，语言环境的社会情感分析（SEAL）。SEAL使用先进的语音处理算法，自动捕捉教师在幼儿教育课堂实践中支持社会情感学习的关键要素：他们在日常互动中使用的语言，特别是挖掘社会情感学习的单词和短语。这种从幼儿课堂录音中自动测量教师社会情感谈话的方法可以代表一种变革性的方法，帮助教师更好地支持儿童的社会情感发展（例如，理解、管理和表达情感）。反过来，这样的工具可以更好地确保幼儿获得技能，确保他们为上学、人际关系和生活做好准备。证明SEAL概念验证的项目活动包括：(1)开发一个社会和情感词汇和短语库，由全国调查和幼儿专业人员的焦点小组提供信息，并由专家咨询小组提供意见；(2)挖掘教师音频数据，完善现有语音处理算法，检测课堂成人社会情感关键词和短语；(3)通过将人工转录与自动检测语音数据的基线算法进行比较，检查单词和短语库的初始有效性。这项探索性研究提供了证据，证明未来完全制定的SEAL解决方案可以产生易于访问和解释的数据，并且可以被个别教师、教练和早期儿童项目使用，以改善旨在促进儿童社会情感结果的实践。最终，就像一个计算步数的自动监控器一样，我们设想SEAL可以被教师自己用来监控课堂上的社会情感谈话，并培养提高幼儿社会情感能力的能力。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

o This project is jointly funded by the Developmental Science Program and the Established Program to Stimulate Competitive Research (EPSCoR).Children’s success in school and in life depends on whether they have learned certain social and emotional skills in their early years. To build teachers’ capacity to promote social emotional learning, there is a need for tools that provide data-based feedback to teachers on talk that focuses on social-emotional learning. This project creates such a tool, the Social Emotional Analysis of the Language Environment (SEAL). SEAL uses advanced speech processing algorithms to automatically capture a key element of teacher practice in early childhood education classrooms known to support social emotional learning: their use of language, and specifically, words and phrases within everyday interactions that tap into social emotional learning. This automatic measurement of teachers’ social-emotional talk from audio recordings in toddler classrooms could represent a transformative approach to help teachers better support children’s social-emotional development (e.g., understanding, managing, and expressing emotions). In turn, such a tool could better ensure young children gain the skills that ensure they are ready for school, relationships, and life. Project activities to demonstrate proof-of-concept for SEAL include: (1) developing a social and emotional word and phrase bank, informed by a national survey and focus groups with early childhood professionals and input from an expert advisory panel; (2) exploring teacher audio data to refine existing speech processing algorithms to detect classroom adult social emotional key words and phrases; and (3) examining initial validity of the word and phrase bank by comparing human transcription to a baseline algorithm to automatically detected speech data. This probe study serves to create evidence that a future fully formulated SEAL solution could yield data that will be easy to access and interpret and can be used by individual teachers, coaches, and early childhood programs to improve practices aimed at promoting child social emotional outcomes. Eventually, like an automated monitor for counting steps, we envision SEAL could be used by teachers themselves to monitor classroom social emotional talk and build capacity to improve social emotional competence in young children.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.