DILVI – A Platform to Build Language Training Simulation Environments

Abstract

An enabling environment always facilitates and motivates a person to strive for more practice. Learning a foreign language is a good example. Each of us gained proficiency in our mother tongue, spontaneously, since the environment enabled us the opportunities for practicing that language for many years. A similar reasoning can be arrived at in learning other languages such as English – an enabling environment, makes practicing and eventually learning a new language easier. Among the existing tactical learning enabled software, majority has been developed within the context of language training for military purposes. Hence, most of the codes and frameworks are protected and not made available for the general purpose users and researchers. This has hindered the growth in the domain of tactical language training. Thus, a burning requirement exits within the researchers and developers in the area of tactical language learning simulators, for an underlying framework which could be used in developing such tools. DILVI is an abstract framework which enables the developers and researchers to focus more on advanced modules rather than concentrating on the underlying framework. Therefore, allowing the developed modules to be plugged in to the abstract framework with less hassle. Built as an open source modular framework; DILVI follows the plug-in architecture developed on Java platform adhering to Open System Gateway Interaction (OSGi) specification. The framework consists of five modules, Role-Play Director, Simulation Engine, Dialog to Speech, Speech to Dialog and Pedagogical Agent. First two provides set of boilerplate implementation which is common in most of the times when developing tactical language learning simulators. The latter modules define a set of APIs enabling speech recognition, text-to-speech and pedagogical support. Once the framework is built, it was made available for a team of researchers and developers in the area of ESL tool development. Comprehensive effort estimation was obtained from the respective developers using DILVI framework and efforts required to build the same from the scratch. The results were then compared and aggregated in identifying the effectiveness of the framework. Furthermore, an application was developed using the abstract framework and a user survey has been carried out in indentifying the effectiveness of it. The evaluation of DILVI has proven to be a successful as it has shown 70% reduction of development efforts significantly.