CALLIDUS - Computer-Aided Language Learning: Vocabulary Acquisition in Latin using Corpus-based Methods

What are the benefits of corpus-based methods for lexical acquisition like those that are used in the teaching of modern languages for the instruction of the Latin language? The Callidus project will address this question by means of a controlled intervention study. The project will re-use, extend, and combine existing Latin corpora, search tools and a teaching platform. Learning the vocabulary in any foreign language is a complex task because of a high degree of polysemy, the semantic interdependence of words and multi word expressions, and the linking of lexical material with grammatical constructions. The traditional (and widely and often exclusively used) method of memorization of word equations is therefore not very effective. A more promising method is to present lexical entities in many different contexts. Corpus-based methods allow the production of teaching materials (examples and exercises) complementing the textbook in an ad-hoc manner. In order to fulfill that goal, the corpora need to be annotated on several layers (lemma, part-of-speech, frequencies, dependencies) and must be searchable across these levels. Different types of exercises will be produced that can flexibly combined with various lexical and grammatical materials. Several cycles of a controlled, design-based intervention study will be conducted in 10th forms in high schools (Gymnasium) in Berlin and Munich. A control group taught by traditional means will be compared to a group which is instructed using corpus-based materials. Corpora of classical Latin compiled and annotated for lingustic research in the LangBank project (http://sfs.uni-tuebingen.de/langbank) will be supplemented with curricular information and semantic relations. Exercises produced on the basis of the corpora will be presented in the Moodle learning platform which will also contain the data from the intervention study (testing materials, scores) and serve as a communication platform for the user groups. The corpora will be hosted in the LAUDATIO repository (http://www.laudatio-repository.org/repository/) and are searchable by the powerful multi-layer search engine ANNIS (http://corpus-tools.org/annis/) by use of a domain specific corpus query language (ANNIS Query Language, AQL). As part of the project a specific component (Machina Callida) will be developed that will extend AQL and connect the corpus to the Moodle system.