MAGRID - FROM DEVELOPING A LANGUAGE-NEUTRAL LEARNING APPLICATION TO PREDICTIVE LEARNING ANALYTICS

Mathematical proficiency serves as one of the foundations that must be solid if learners are to succeed in the classroom. However, the hierarchical nature of mathematical development means that basic math skills must form the groundwork for subsequent mathematical constructs. As a matter of fact, a fragile mathematical foundation leads to some unavoidable roadblocks further ahead. That being said, early childhood education and the preschool years are highly foundational, and thus, it is crucial to train early mathematical abilities.

If we bear in mind that teaching and testing in schools rely on communication between the teacher and learner, fluency in the language of instruction exerts a tangible influence on this process. Consequently, students who are not proficient in the language of instruction are set up to have a poor mathematical foundation that will likely hold them back in relation to their peers. This is a shortcoming that can drive a gap between learners within heterogeneous school settings, which becomes a hurdle for students and teachers alike.

Because digital devices are now more available and present in today's classrooms, digital interventions (e.g., computer software and tablet applications) are one channel that can be used to bridge such a gap through the introduction of language-neutral training and testing programs.

Students who are not pro ficient in the language of instruction would be enabled to acquire and maintain the mastery of early mathematical skills when the emphasis is placed on visual content rather than language skills in the teaching of mathematical concepts. Moreover, when digital devices are used as learning platforms, learners' interactions (behavior) with the training content can be recorded in log files. Analyzing the log data can provide teachers with detailed information on learners' progress, which may ultimately remove the necessity to administer tests and examinations.

The present dissertation explores the extent of educational technologies in the teaching and testing of early math abilities. In this regard, we address two main areas of focus in research: The first one is to examine the possibility of developing a language-neutral application for learning mathematical abilities for young learners so that they do not need to depend on their language proficiency to acquire mathematical skills.

Findings of empirical studies have demonstrated that students who participated in the early mathematics training program using the MaGrid application performed significantly better on various measures of early mathematical abilities. Therefore, the MaGrid application can provide an efficient way to reinforce the math knowledge of all preschoolers, including segments (second-language learners) that have traditionally been underserved.
The second focus is on studying the possibility of inferring young learners level of mathematical competencies on the basis of their interactions in a learning platform. The idea behind analyzing log data of learners behavior was to evaluate the alternative approach (behavioral-based assessment) to formative assessment for measuring students improvements and competency levels. Along these lines, we derived system-specific behavioral indicators using the presented systematic approach, and then we evaluated the predictive power behind them. Findings illustrate that assessing learners' level of knowledge could be carried out through the analysis of traces students leave in an e-learning environment and without interrupting their learning process. Results further unveil how digital devices can be used in schools to enhance learning outcomes.