Forgetting of Foreign‐Language Skills: A Corpus‐Based Analysis of Online Tutoring Software

Abstract

We explore the nature of forgetting in a corpus of 125,000 students learning Spanish using the Rosetta Stone foreign-language instruction software across 48 lessons. Students are tested on a lesson after its initial study and are then retested after a variable time lag. We observe forgetting consistent with power function decay at a rate that varies across lessons but not across students. We find that lessons which are better learned initially are forgotten more slowly, a correlation which likely reflects a latent cause such as the quality or difficulty of the lesson. We obtain improved predictive accuracy of the forgetting model by augmenting it with features that encode characteristics of a student's initial study of the lesson and the activities the student engaged in between the initial and delayed tests. The augmented model can predict 23.9% of the variance in an individual's score on the delayed test. We analyze which features best explain individual performance.

Predicting University Students’ Exam Performance Using a Model-Based Adaptive Fact-Learning System

“…The adaptive fact-learning system may be extended with an analytics dashboard-similar to Duolingo and Rosetta Stone (Settles & Meeder, 2016;Ridgeway et al, 2017)-indicating a learner's progress on each lesson. Such a dashboard could also show students the items that are most difficult for them, along with their current estimated memory strength.…”

“…For example, Duolingo, a languagelearning tool, shows learners an overview of their mastery of each lesson in a dashboard (Figure 1(a) in Settles & Meeder, 2016). Rosetta Stone, another language-learning tool, has a similar dashboard and includes a suggested next study activity (Ridgeway, Mozer, & Bowles, 2017).…”

Second Language Acquisition Modeling

“…Settles and Meeder (2016) and Ridgeway et al (2017) recently proposed non-linear regressions that explicitly encode the rate of forgetting as part of a decision surface, however none of the current teams chose to do this. Instead, forgetting was either modeled through engineered features (e.g., user/token histories), or opaquely handled by sequential RNN architectures.…”

“…Relatedly, the spacing effect (Dempster, 1989) is the observation that people will not only learn but also forget over time, and they remember more effectively through scheduled practices that are spaced out. Settles and Meeder (2016) and Ridgeway et al (2017) recently proposed non-linear regressions that explicitly encode the rate of forgetting as part of a decision surface, however none of the current teams chose to do this. Instead, forgetting was either modeled through engineered features (e.g., user/token histories), or opaquely handled by sequential RNN architectures.…”

Commentary: Principles, Approaches and Challenges of Applying Big Data in Safety Psychology Research

“…Nevertheless, the paper provides input to wonder why we should limit our scope to the safety domain and not including other ones into the big data of psychology discourse. Examples of big data applications are identifiable in numerous areas of psychology, such as organizational (Guzzo et al, 2015;Tonidandel et al, 2016), educational (aka learning analytics; Watson and Christensen, 2017;Maldonado-Mahauad et al, 2018;Viberg et al, 2018;Elia et al, 2019;Shorfuzzaman et al, 2019), marketing (Hopp and Vargo, 2017;Matz and Netzer, 2017;Erceg et al, 2018;Ibrahim and Wang, 2019), personality (Bleidorn et al, 2017;Boyd and Pennebaker, 2017;Gerlach et al, 2018;Hinds and Joinson, 2019), emotion (aka affective computing; D'Mello et al, 2018; Chatterjee et al, 2019; Gruda and Hasan, 2019), psycholinguistics (Ridgeway et al, 2017;Johns, 2019;Luo et al, 2019), clinical (Anestis et al, 2016;Russ et al, 2018), cognitive (Medina and Fischer-Baum, 2017;Bhatia and Walasek, 2019), community (O'Brien, 2016), group (Guadagno et al, 2018), music (Greenberg and Rentfrow, 2017), political (Ma-Kellams et al, 2018), and positive psychology (Luhmann, 2017;Yaden et al, 2018). Additionally, the representation of BDSP as being the intersection of safety science, data science, and psychology, seems to equally fit other psychology branches.…”