Development of an Environmental ICAI System for English Conversation Learning using Speech Recognition and Synthesis Techniques

在这篇报告中，我们描述了一个环境ICAI系统的开发，“HyperEnglish”，用于以模拟呈现的情景为中心的会话学习。该系统有一个自由的学习环境，允许学生在学习中发挥主动性，并通过模拟过程的控制不引人注意地引导他们。近年来多媒体领域取得的显著进展使构建利用声音和图形来形成逼真的学习环境或准语言学习环境的模拟成为可能。因此，目前配备了多媒体学习环境的计算机辅助教学系统可以有效地替代真实的对话体验。然而，在模拟演示中使用多媒体技术并不能保证学习。在我们的研究中，我们考虑了一个用于英语会话学习的学习环境，并提出了一个配备角色扮演游戏风格的模拟环境的环境ICAI系统的体系结构。对于环境的实现，我们提出了一种知识表示方法--对话图，用于构建对话模拟。这种知识表示可以实现作为教育功能的模拟的“系统对话进度控制”。模拟控制是基于一种试图诱导学习者失败的归纳策略。通过参考学习者的操作历史和对他们在基础知识、情景识别和短语生成等方面的能力的诊断，系统利用该归纳策略选择合适的话语和回答。我们使用HyperCard和C语言在MacintoshTM&gt；计算机上实现了该系统。

In this report, we describe the development of an environmental ICAI system, "HyperEnglish", for conversation learning which is centered upon a situation presented through simulation. The system has a free learning environment which allows students to use initiative in learning and unobtrusively guides them through the control of the simulation progress.The remarkable progress which has been made in the field of multimedia recently has make it possible to construct simulations which utilize sounds and graphics to form a realistic learning environment, or quasi-environment for language learning. Therefore, current CAI systems equipped with a multimedia learning environment can be an efficient substitute for real conversation experiences. However, the use of multimedia techniques in the presentation of simulations will not guarantee learning. Without an appropriate methodology, we cannot expect the same learning effects as those which can occur with a human teacher.In our research, we have considered a learning environment which for English conversation learning and proposed an architecture of an Environmental ICAI system which equipped with a role-playing game style simulation environment. For a realization of the environment, we have proposed a knowledge representation, "Dialogue Graph", which for a construction of dialogue simulations. This knowledge representation can realize a "system dialogue progress controls" of the simulation as an educational function. The simulation control is based on an induction strategy which attempts to induce the learner's failure. By referencing the learners operation history and a diagnosis of the their abilities in fundamental knowledge, situation recognition and phrase generation etc., the system selects appropriate utterances and replies utilizing this induction strategy.We have implemented the system using HyperCard^<TM> and C language on a Macintosh^<TM> computer.