Giving Help and Praise in a Reading Tutor with Imperfect Listening--Because Automated Speech Recognition Means Never Being Able to Say You're Certain

Mostow, Jack; Aist, Gregory

doi:10.1558/cj.v16i3.407-424

Cited by 33 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…LISTEN is one of these systems that provides various types of assistance when a child clicks for help or the system decides the child needs help. The feedback form provided on a word is randomly selected from the available options [10,4]. Research shows that the children's reading fluency and comprehension improved after practicing with the online tutor, compared to Sustainable Silent Reading (SSR) and regular classroom instruction [11,12].…”

Section: Research Backgroundmentioning

confidence: 99%

The Effects of Implicit and Explicit Feedback in an ASR-based Reading Tutor for Dutch First-graders

Hubers¹,

Cucchiarini²,

Hout³

et al. 2022

Interspeech 2022

View full text Add to dashboard Cite

Literacy skills are pivotal for children to become schooled and educated and to engage with written texts in everyday life. Learning to read is a primary skill that children acquire at school. Supervised, supportive guided reading aloud may help children improve their reading skills. However, such a supportive context is usually difficult to realize at school since teachers do not have enough time to give directed individual feedback. An ASR-based reading tutor could be a solution in such a pressing situation, as such a tutor can 'listen' to children reading, provide individual feedback on errors, and store information on reading practice into logfiles.In this study, we investigated the effectiveness of an ASRbased Reading Tutor for Dutch first graders, in which different forms of feedback were implemented. We collected data from 525 first-graders in 44 schools, with 263 pupils who received explicit feedback and 262 pupils who received implicit feedback. Analyses based on mixed linear regression models indicate positive effects of both feedback forms on reading accuracy and a trade-off relationship between accuracy and speed.

show abstract

Section: Research Backgroundmentioning

confidence: 99%

The Effects of Implicit and Explicit Feedback in an ASR-based Reading Tutor for Dutch First-graders

Hubers¹,

Cucchiarini²,

Hout³

et al. 2022

Interspeech 2022

View full text Add to dashboard Cite

show abstract

“…Additionally, ASR plays a vital role in continuous learning and learning assessment by allowing users to practice pronunciation and conversation, providing immediate feedback on accuracy, and creating student models to evaluate learning proficiency and performance. Mostow et al identified three areas of knowledge necessary to support the ASR engine: the acoustic model, the pronunciation lexicon, and the language model [24].…”

Section: Modelsmentioning

confidence: 99%

Procedural content generation for VR educational applications: The investigation of AI-based approaches for improving learning experience

Wang

2023

ACE

View full text Add to dashboard Cite

In recent years, artificial intelligence (AI) has witnessed significant advancements in the field of education, with its ability to personalize and adapt content to individual student needs. In parallel, virtual reality (VR) has emerged as a powerful tutorial tool, providing immersive and interactive experiential learning experiences which has the advantages of improving students' motivation and engagement. Previous researchers have demonstrated the potential of ML algorithms, particularly RL, for generating educational content and VR environments. To create high-quality content, researchers have started exploring the integration of Machine Learning (ML) and Reinforcement Learning (RL) algorithms into Procedural Content Generation (PCG) methods for automatically generating both textual and non-textual content such as practice questions, quizzes, VR learning environments, etc., which have the potential to increase the efficiency and effectiveness of educational interventions. Nonetheless, the development of these techniques requires addressing several challenges. Significant advancements are yet to be made in developing and refining these algorithms to produce high-quality and effective educational content for VR applications. This article provides a comprehensive overview of the current state of research in reinforcement AI learning content generation for VR educational applications. For each area, it discusses the state-of-the-art techniques, applications, limitations, and challenges faced in development, covering the use of natural language processing, reinforcement learning, and machine learning algorithms. The review concludes by highlighting some of the key opportunities for future research in this field, including the development of more sophisticated models and the exploration of new applications of machine learning in educational technology.

show abstract

“…• The feedback provided by the RT is subtle: It merely does not highlight incorrectly read text as the reader proceeds; further, it signals that the program is waiting for improved input only if there is a major delay or error; • Feedback provided by the RT is imperfect in the sense that its ASR technology does not catch every single reading error and stop to offer help (Mostow & Aist, 1999). Consequently, discourse flow is not frequently interrupted, thus helping to balance contending fluency and accuracy performance goals.…”

Section: Moderating Effect Of Good Quality Classroom Instruction With...mentioning

confidence: 99%

Speech Recognition Software Contributes to Reading Development for Young Learners of English

Reeder

Shapiro

Wakefield

et al. 2015

International Journal of Computer-Assisted Language Learning and Teaching

View full text Add to dashboard Cite

Thirty-six English language learners aged 6;8 to 12;6 years received practice with The Reading Tutor, which uses speech recognition to listen to oral reading and provides context-sensitive feedback. A crossover research design controlled effects of classroom instruction. The first subgroup worked with the software for 3.5 months, and following a week's crossover period, the second subgroup worked for a subsequent 3.5 months. Both groups were assessed to obtain comparable gains both in regular classroom with English as an Additional Language (EAL) support and in the classroom condition with EAL support plus the Reading Tutor. Oral reading fluency was assessed by the DIBELS measure. Fluency was also calculated by the program, and grade level of materials mastered was assessed by the software's logs. Both groups made significant gains in oral reading fluency and grade level of materials mastered, according to measures internal to the software. For one period, gains in fluency following experience with the program appeared to have been slightly larger than gains with regular classroom instruction and EAL support only.

show abstract

Giving Help and Praise in a Reading Tutor with Imperfect Listening--Because Automated Speech Recognition Means Never Being Able to Say You're Certain

Cited by 33 publications

References 7 publications

The Effects of Implicit and Explicit Feedback in an ASR-based Reading Tutor for Dutch First-graders

The Effects of Implicit and Explicit Feedback in an ASR-based Reading Tutor for Dutch First-graders

Procedural content generation for VR educational applications: The investigation of AI-based approaches for improving learning experience

Speech Recognition Software Contributes to Reading Development for Young Learners of English

Contact Info

Product

Resources

About