Swarachakra Malayalam is an open-source keyboard for Android. It generates a steadily growing database of Malayalam words. This word database could be a resource that can be used to fuel the development of new tools for the language. But the database (corpus) contains incorrect or unusable words (for certain contexts). Tagging these words becomes an important task to make this corpus usable by ‘cleaning’ the corpus.

Due to the complexity of the language's grammar paired with its agglutinative property, it is challenging to programmatically categorise the words. But while this may be challenging for a computer, it is easier and even enjoyable for a person who knows the language. But due to the large number of words in the corpus, it becomes a huge task. The aim of this project is to crowdsource, through gamification, the cleaning of the corpus.

During the course of the project, the corpus cleaning activity was broken down into multiple steps and turned into minitasks. Then multiple possible ways of gamifying these tasks were looked into. After weighing the pros and cons of each, one of the ideas was designed, detailed, and developed into functioning prototypes. The prototype version 1 had minimal gamification elements (only level scores and player levels). The prototype version 2 has more gamification elements like scores, player levels, achievements and badges, leaderboards, etc. The proto V1 acts as a benchmark against which player engagement levels of proto V2.

While the proto-V2 doesn’t have all the gamification elements that were explored, it lays a foundation upon which the others can also be added. The effectiveness of the game in cleaning the corpus and the effect of these gamification elements on player engagement were evaluated using a functioning prototype. Out of the gamification elements that were tried out in the prototype, the tutorial levels, game stats, achievements, and leaderboards seemed to have a direct positive impact on the players’ engagement levels. Identifying the impact of the other gamification elements will require a longer evaluation with a larger user base. In the bigger picture, the outcome of this project would be one of the several layers of filters that can be used to clean up the existing database and create a comprehensive database of words.