Feedback visualization in a grammar-based e-learning system for German: a preliminary user evaluation with the COMPASS system

Abstract

COMPASS 3 is an e-learning system that can visualize grammar errors during sentence production in German as a first or second language. Via drag-and-drop dialogues, it allows users to freely select word forms from a lexicon and to combine them into phrases and sentences. The system's core component is a natural-language generator that, for every new word the user wishes to attach to the current string (as an extension of this string or as a replacement of a substring), checks whether this tentative attachment is grammatically well-formed or not. On this basis, the system can compute and display online the grammatical structure of input strings in the form of syntactic trees, and identify and diagnose input errors. In the following, we focus on the crucial question of how to present the feedback to the learner. We propose tutored visualizations with animations of pedagogical agents. We briefly report the results of a preliminary user evaluation study in which the participants judged the well-formedness of prefabricated input sentences. The data, collected by means of eye-tracking and a questionnaire, show that L1 learners who are exercising an unfamiliar and error-prone grammatical structure, pay due attention to, and can profit from, this type of visualized error feedback.

Personalisation methods in e‐learning‐A literature review

“…Several typical examples of successful implementation of adaptive feedback are Live Interactive Feedback system [252], COMPASS3 [116], SQL Quizbot [275], Student Coursework Repository [47], Pgtracer [145], MCQ [13], ASSET [74] UCOM [258].…”

“…Multiple studies suggest that learners are engaged and motivated when provided with adaptive feedback [265] and that such feedback can promote learning attributes such as participation and engagement in voluntary learning activities [90]. However, most of the research about providing feedback suggests identical feedback to all students, only linked to their performance and did not focus on providing adaptive feedback [116]. The authors in Hafidi and Lamia [115] researched how learners' feedback and their multiple intelligence link to learning outcomes and examined the impact of the feedback principles on students' behaviour.…”

“…Different approaches exist and have been used in the literature for providing adaptive feedback. Most common is providing generic textual feedback based on the performance and cognitive level of the student [275], while the limited number of studies incorporate more specific approaches such are: video recordings with expert feedback [74,90], envisioning grammar mistakes during sentence construction [116] and incorporation of learning agents [288]. According to Bimba et al [48] adaptive feedback is generated based on the four adaptation characteristics: means (Which information about the learner is used for adaptation?…”

“…Early personalisation AHS MetaDoc [52], ISIS Tutor system [53], PUSH [131], AHA! [79], INSPIRE [213], ELM-ART [280], DCG [268], MetaLinks [201], E-school [260], CleverUniversity [51], EVODU [242], SRES [33] Content, Learner interface, Course structure, Learner model, Learning items 12 Adaptive feedback OFES [122], LIF [252], COMPASS3 [116], SQL Quizbot [275], SCORE [47], Pgtracer [18,145,288], MCQ [13], ASSET [74], UCOM [258] Learning…”