The teaching buddy: Speech and language technologies for assisting and assessing instructional practice

Abstract

In this work we present a new tool, the Teaching Buddy, based on automatic speech and natural language processing technologies primarily intended to facilitate the observation and analysis of instruction in support of teachers' professional portfolio development and assessment of instruction. From the education point of view, our system is predicated on the theory that teachers develop best in a community of practice. The Teaching Buddy is a tool that enables this community of practice to reach deeper and more meaningful levels of analysis by allowing the professional development team to provide better, consistent, and evidence-based insights and feedback to the practitioner. From the technology point of view, the Teaching Buddy leverages existing state of the art automatic speech recognition (ASR) and natural language understanding (NLU) technologies together with instructional discourse analysis frameworks as well as mature and well established instructional assessment frameworks. The Teaching Buddy is structured in five layers: the data capture layer, the speech recognition layer, the natural language understanding layer, the evaluation and scoring layer and finally, the presentation layer. The results of the analysis can be used by an expert, mentor, or professional development team to provide the practitioner with constructive, evidence-based feedback. In this paper we first introduce the Teaching Buddy and describe in detail its five layers. We then illustrate how our system works by following the analysis process that the Teaching Buddy carries out using a brief lecture segment of a Discrete Mathematics course at the college level. We conclude our paper with a summary of what we believe are the most promising areas of future research including a brief survey of potential target delivery frameworks, including Eclipse and Sakai.