Measuring the extent speech-to-text functionality aids speaking fluency (breakdown & speed-fluency) in EFL learners using a mobile assisted language learning app

测量语音转文本功能有助于口语流利程度的程度（细分

• 批准号: 19K00803
• 负责人: Robert Cvitkovic
• 金额: $ 2.91万
• 依托单位: Teikyo University (2021-2022)Tokai University (2019-2020)
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 2019
• 资助国家: 日本
• 起止时间: 2019-04-01 至 2024-03-31
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-19K00803/
• 关键词: speed fluency; breakdown fluency; speaking fluency; shadowing; STT; MALL; speech to text; TEL

A one-way independent ANOVA with 4 factors was performed for Breakdown fluency on a total of 182 participants. This gave an F statistic of F(3,178)=43.02, p<.000, with an overall effect size Cohen D = 0.87 which is considered large. Simple contrasts with a Bonferroni correction showed that the reduction in breakdown fluency for the 20 hour treatment group, T20 (M=-0.159) was significantly better than the control group of the same time, C20 (M=-0.108, t(94)=5.45, p=.001, r=0.480). Similarly, the contrast was calculated for the control group at 5 hours, C5 (M=-0.050) and the treatment group at the same time, T5 (M=-0.118, t(84)=1.661, p=.001, r=0.726).Low and intermediate level (CEFR A2-B1) Japanese EFL learners have high breakdown fluency and low to medium speed fluency. The longest treatment group of 20 hours showed the most reduction in breakdown fluency compared to the control group of the same time. All groups improved their performance over time, i.e. demonstrated a reduction in breakdown fluency and an improvement in speed fluency compared as the length of time increased. This was true with the control groups as well from C5 to C20. Most notable is that T5 group showed a statistically equivalent amount of improvement as the C20 group. This is the strongest indication that immediate feedback, a high degree of interactivity, progress tracking and overall gamification of the digital treatment can be a more effective way of improving speaking fluency in low to intermediate learners.

对182名参与者进行了4个因素的单因素独立方差分析。这给出了F（3，178）=43.02的F统计量，p<0.000，总体效应量Cohen D = 0.87，被认为是大的。与Bonferroni校正的简单对比显示，20小时治疗组T20（M=-0.159）的崩溃流畅性的降低显著优于相同时间的对照组C20（M=-0.108，t（94）=5.45，p=.001，r=0.480）。同样地，对照组在5小时时C5（M=-0.050）和治疗组在相同时间T5（M=-0.118，t（84）=1.661，p= 0.001，r=0.726）的对比计算。与相同时间的对照组相比，20小时的最长治疗组显示出最大的故障流畅性降低。随着时间的推移，所有小组的表现都有所改善，即随着时间的增加，分解流畅性有所降低，速度流畅性有所改善。从C5至C20，对照组也是如此。最值得注意的是，T5组显示出与C20组统计学上相当的改善量。这是最有力的迹象表明，即时反馈，高度的互动性，进度跟踪和数字处理的整体游戏化可以是提高低到中级学习者口语流利性的更有效的方法。

项目成果

Quantifying Speaking Characteristics in Intermediate Japanese Learners

量化中级日语学习者的口语特征

• 发表时间: 2022
• 作者: Cvitkovic;Robert; Massoud;O.
• 通讯作者: O.

Improving speaking fluency with speech-to-text shadowing tasks

通过语音到文本跟踪任务提高口语流利度

• 发表时间: 2022
• 作者: Cvitkovic;Robert
• 通讯作者: Robert

Quickly create effective auto-grading quizzes with Google Sheets+Forms

使用 Google 表格快速创建有效的自动评分测验

• 发表时间: 2022
• 作者: Cvitkovic;Robert
• 通讯作者: Robert

Measuring the extent speech-to-text shadowing aids speaking fluency in EFL learners using MALL

使用 MALL 测量语音到文本的影子对 EFL 学习者口语流利程度的帮助

• 发表时间: 2022
• 作者: Cvitkovic;Robert
• 通讯作者: Robert

Measuring the Extent Speech-To-Text Can Affect Speaking Performance

测量语音转文本对口语表现的影响程度

• 发表时间: 2022
• 作者: Cvitkovic;Robert
• 通讯作者: Robert