The aim of the work described in this paper is to develop methods for automatically assessing the pronunciation quality of specific phone segments uttered by students learning a foreign language. From the phonetic time alignments generated by SRI’s Decipher HMM-based speech recognition system, we use various probabilistic models to produce pronunciation scores for the phone utterance. We evaluate the performance of the proposed algorithms by measuring how well the machine-produced scores correlate with human judgments on a large database. Of the various algorithms considered, the one based on phone log-posterior-probability produced the highest correlation with the human ratings, which was comparable with correlations between human raters.
Education & learning publications
Communication is a central aspect of human learning. Using the Probability Inquiry Environment (PIE) as an example, we examine how external representations (both textual and iconic) mediate face-to-face conversations among students, and support productive mathematical discourse. We provide quantitative data that suggests that seventh grade students who used PIE learned some of the basic principles of probability. Two cases studies are that illustrate how communication supported by computer-mediated representations
contributed to this success. The first case study demonstrates how the computer can actively prompt student conversations that lead to learning. The second case study examines how an animated graphical representation supported these productive conversations.
ECHOS is a voice interactive language training system being developed to foster improvement in French comprehension and speaking skills, incorporating speech recognition and pronunciation evaluation. Speech recognition allows students to navigate through units using oral communication with various types of system feedback. The pronunciation scoring being developed is validated by expert human raters. We will discuss the motivation for the program, the nature of the interdisci plinary effort, and the resulting system architecture. Challenges and trade-offs of designing activities using unscripted material and aspect of speech research as related to this application will be described. Finally, we discuss opportunities to use speech recognition on other platforms.
Scatter/Gather is a cluster-based browsing technique for large text collections. Users are presented with automatically computed summaries of the contents of clusters of similar documents and provided with a method for navigating through these summaries at different levels of granularity. The aim of the technique is to communicate information about the topic structure of very large collections. We tested the effectiveness of Scatter/Gather as a simple pure document retrieval tool, and studied its effects on the incidental learning of topic structure. When compared to interactions involving simple keyword-based search, the results suggest that Scatter/Gather induces a
more coherent conceptual image of a text collection, a richer vocabulary for constructing search queries, and communicates the distribution of relevant documents over clusters of documents in the collection.
This work addresses the problem of helping learners develop concepts and skills needed to be productive in the fast-changing technical workplace. It marries the kind of informal, on-demand learning preferred by workplace professionals and advocated by learning theorists with the support for learning at a distance afforded by networked computing environments (Schlager, Means, O’Day, & Poirier, 1994). Our cognitive mentoring model of collaborative learning (Schlager, Poirier, & Means, 1996) is based on the idea of individually tailored assistance
provided at the learner’s request when an impasse is encountered in problem solving. We have studied such interactions in the workplace (as well as classroom interactions) and developed a synchronous collaboration/simulation environment called Distant Mentor (DM) to facilitate mentoring from a distance (Schlager et al., 1994). This report describes a study that demonstrates how our prototype technology can support cognitive mentoring over a computer network.