LEARNING TO TAKE THE NEXT STEP: KNOWLEDGE ACQUISITION OF ORTHOGRAPHIC PATTERNS IN SERIOUS GAMES WITH IMMEDIATE FEEDBACK

This paper describes the analysis of learning progression in a serious game that is part of the iRead EU Project. The goal is to teach students vowel duration in German orthography. Students are asked to identify words with long vowels and receive immediate feedback on their choices through game mechanisms. The game presents the player with a series of multiple choices to the same question (find the long vowel) with randomly generated answers and sets of distractors. 48 children played once a week for four weeks. The results contribute to deeper understanding of game analytics for larger data sets.

Serious games are considered a motivating way to teach and learn. Within the framework of the iRead project (www.iread-project.eu), we look at L1 (first language) learning of reading and writing skills by playing provided games.

In previous research, we have shown that skills gained in this study are transferable to writing and reading skills outside the game. The time a student takes to make the decision, decreases with experience. This plays an important role in the student's ability to transfer a skill learned in the game to related tasks like writing.

This publication forms part of a series of work on evaluation of games. In this paper we focus on the learning of vowel duration in German orthography through selected games, asking the following questions:
• Did learning take place in the game?
• How does learning progress?
• Are there differences between grades?

The students are divided into two groups:
(i) First grade students taking their first steps beyond reading single graphemes.
(ii) Second grade students who have already been exposed to unstructured input over a year of schooling, without any explicit teaching of the concept.

iRead offers a large number of game mechanics and presents these to the learner with adaptively generated content.
For the purpose of this experiment, we only chose one particular game mechanic for studying the impact of feedback on learning.
In around 5 steps the player is presented with a choice of 3-5 words.

Our results show that despite the small amount of total minutes spent playing, the amount of input in terms of words seen and multiple choice questions answered with immediate feedback was large enough to indicate some learning within the game when looking at average values across students grouped by grade.

We also observe differences between grade 1 and grade 2 students which indicate that prior knowledge interferes with learning in the beginning.
Grade 1 students show a steady improvement, while grade 2 students exhibit a u-shaped learning curve.
More detailed analysis with regard to different distractors and feedback indicate that frequency and the type of feedback also have an impact on learning.

The averaged learning curves compared between the two grades seem to indicate that some topics are more fruitful when taught earlier. Vowel length may b e one of those items.

The data that is used here was transcribed manually. In future, game analytics will yield better information automatically for larger data sets. The preliminary look at data presented in this paper is useful to gain insights into larger data collections.

Through automating that process of learner analytics, adaptive feedback and distractor selection can be used to optimize the learning efficiency. Longer sets of learning curves can then be analyzed both on average as well as individually.