III. ISSUES SURROUNDING THE DEVELOPMENT OF ITSS
Designing each of the components of an ITS for language learning requires a careful consideration of the needs of both learners and teachers. In the light of that and based on available reports from literature on specific language-learning ITSs, we point out some of the issues that may arise during the development process, briefly explain them, and, where appropriate, offer some suggestions as to possible solutions.
One of the frequently noticed issues in language- learning ITSs is over-restricting the learning domain, both horizontally and vertically. Horizontal domain restriction refers to a system’s tendency to focus on a single linguistic skill (listening, reading, speaking and writing) or form (grammar, vocabulary) rather than a combination of them. Including a whole range of skills/forms or even all of them into the ITS would be more in line with the nature of language learning, which emphasizes the integration of both productive and receptive skills, visual and auditory input, written and spoken output, form and communicative competence, etc [27]. Conversely, vertical domain restriction refers to the inability of an ITS to cater for learners with different levels of language proficiency (e.g., CEFR proficiency levels A1 – C2), which decreases the number of potential system users and, consequently, system’s (potential) financial value. We maintain that a possible solution and a more sensible approach would be to offer learning content designed for a wide range of skills, forms and ability levels, and use the knowledge about the learner, kept by the ITS, to detect which skills/forms are needed by each learner. Moreover, learner data should be used to identify the learner’s current level of proficiency, offer learning materials that are appropriate for the identified level, and decide when the new level of proficiency has been reached.
The issue of user input evaluation is also a significant one. Whereas evaluation of certain linguistic skills and forms can be automated and independently done by the system (e.g., reading, listening and grammar), some skills are notoriously difficult for the computer to evaluate on its own. This particularly holds for productive skills, namely speaking and writing, which involve unrestricted user output. Even though natural language processing and speech recognition techniques have come a long way in terms of accurately evaluating user output, they maintain to exhibit certain flaws [13]. At present, a computer is still unable to fully replace a human being when it comes to evaluating unrestricted oral or written output. However, restricting user output to a particular domain with specific vocabulary (e.g., describe a room or a person, explain a shopping list) might make a significant difference in rising the accuracy of evaluating output.
Another issue, related to vertical domain restriction, is standardization and comparability between systems. Most of the ITSs reported in literature do not use some generalized framework of ability reference (such as CEFR or ILR scale) to define the level of learner proficiency for which the learning content of the ITS is intended, but specify such a level in terms of school grade or educational level only (e.g., college freshmen course). This significantly affects learner mobility between different ITSs, mostly because learners are unable to
objectively assess how appropriate the language learning system is for their proficiency level. Furthermore, the results of using an ITS for language learning should be observable as a shift in the proficiency level, quantified in familiar terms, and available for analysis for third parties. Aligning the content with any of the internationally set and verified standards for language learning proficiency levels can be a way to successfully address such issues.
Designing learning materials that will be presented to a learner during the educational process involve several issues that need to be acknowledged. The most notable among those is the issue of interactivity of learning materials. When using languages for communication, learners are engaged in an interactive process of sending and receiving messages through a number of communication channels. In certain cases, the meaning of messages cannot be fully grasped simply on the basis of the sum of words and their meanings, but other elements, such as tone of voice, gestures and body language, and contextual information, need to be taken into consideration. Hence, learning materials should reflect real world communicative situations, which is possible to achieve through the use of multimedia. Another issue related to learning materials is how to keep learners motivated during the learning process, and how to actively involve them to work with materials at hand. One possible way is to give small-scale task to the learner, such as “cli