• The combination of complex and multidisciplinary requirements of the peda- gogical and computational concepts during the design phase of the ILTS. These requirements need to be revisited during different cycles of development and use (Nagata, 2010; Heift, 2010a; Schulze, 2010).
• The meaningful integration of ICALL into current foreign language teaching and learning practice (Schulze, 2010; Amaral and Meurers, 2011).
• The exploitation of the data produced by learners, actual learner production, but also learner activity tracking data, during the the use of ICALL systems. Use the collected data to deepen our knowledge on fundamental research and practice questions such as the characterisation of learner interlanguage, or the use of learner performance as a measure for task complexity and as a criterion to inform changes in activity design. And, finally, to assess the usefulness of learning assistance tools such as different help options and feedbacks of different nature (Heift, 2010a; Schulze, 2010; Amaral and Meurers, 2011).
This thesis’s goals, presented in Chapter 1, focus on the first two of the challenges: • Our first goal is to propose a methodology that helps both FLTL practitioners and NLP specialists find a common framework to describe FL learning activities, the responses that they are expected to elicit, and the assessment procedures. This will be a contribution to further characterise the viable processing ground including insights from FLTL, CALL and NLP.
• Our second goal is to design and evaluate an infrastructure for FLTL practition- ers to author and employ FL learning activities including NLP-based automatic feedback generation without the need of programming abilities. Crucially, the responses to the activities authored will be limited in length, but they will be more complex responses to those required in build-a-sentence or fill-the-gap ac- tivities. This will pursue the goal to foster the meaningful integration of ICALL in real-world instruction settings, as well as to facilitate learner individual work using computer-assisted instruction without reducing teacher control or auton- omy.
In the following two sections we develop further the goals of the thesis, which, as we will see, presuppose a tight relationship and information transfer between FLTL, CALL and NLP, at the research level and at the practical level.
2.4.1
The feasibility of ICALL
Our first goal can be worded as characterising the feasibility of ICALL. This goal is in line with the argument of Amaral and Meurers (2011: pp. 9–11) that in order to develop effective ICALL systems there has to be a clear identification of the re- lationship between activity design and restrictions needed to make natural language processing of learner responses tractable and reliable. As the authors propose, de- spite the most straightforward way to constrain learner production is by explicitly
requiring the learner to use certain linguistic constructions, it is more challenging to “investigate how the input can be constrained implicitly in order to provide more space for negotiation of meaning”.
In this context, the notion of viable processing ground introduced by Bailey and Meurers (2008: pp. 107–108) is crucial. The processing ground is that set of FL learning activities that (i) combine elements of comprehension and production, (ii) are meaningful and suitable for an ICALL setting, and (iii) require responses that exhibit “controllable” linguistic variation. However, for this to be practical, we need to know the kinds of activities that can be actually found in the viable processing ground. We propose a methodology to classify and analyse the kinds of activity designs that can be correlated with particular expected ranges of responses and particular assessment needs.
Our methodology is conceived as part of a framework to design ICALL materials that is informed by principles of the design of robust NLP resources for the processing of natural language, and particularly for the processing of learner language. Our approach is informed by relevant theoretical frameworks for activity characterisation in the fields of Task-Based Language Teaching and for the characterisation and design of test tasks under TBLT approaches. The relevant concepts in both areas will be introduced in Chapters 3 and 4. With such an interdisciplinary strategy we pursue to generalise the practice and the principles followed by NLP and FLTL researchers in those projects where there was a true cooperative process, where TBLT provided well-defined designs with clear sets of linguistic constructions that “facilitate[d] the restriction to a linguistic domain which is ‘manageable’ ” (Schulze, 2010: p. 79).
2.4.2
An autonomous use of ICALL in class
Our second goal investigates the autonomous use of ICALL in class, that is, in foreign language courses in real-world instruction contexts. Feedback provided with NLP technology significantly improves the level and fruitfulness of the student- textbook interactions (Heift and Schulze, 2007: p. 3 and pp. 25–29, Nagata, 2010: p. 461). However, ICALL materials, or ICALL systems, are still used in very few instruction settings. As we described, the implication of the actual teachers in the design of development of these systems is as significant as the collaborative and cross-disciplinary nature of the different areas of expertise used. The development of an authoring tool and the corresponding methodological aids thus is one of the necessary steps to facilitate the integration of ICALL in FL instruction contexts.
This goal is also linked to one important argument in the current research in CALL and ICALL, namely that CALL materials are more easily integrated in the course program when they tend to be designed “with the needs and resources of the individual learner in mind” (Levy and Stockwell, 2006: pp. 11-12). This argument is also claimed by researchers in ICALL, as in Amaral and Meurers (2011: p. 4), who defend that “pedagogical considerations and the influence of activity design choices” condition the successful integration of ICALL systems into FLTL practice.