Learning languages from parallel corpora

This work describes a blueprint for an application that generates language learning exercises from parallel corpora. Word alignment and structures allow the automatic assessment of sentence pairs in source target languages, while users continuously improve quality data with their interactions, thus crowdsourcing material. Through triangulation, can be transferred to other than original ones if multiparallel corpora are used as source. Several challenges need addressed such work, we will discuss three them here. First, question how adequate material identified has received some attention last decade, detail what structure implies selection. Secondly, consider which type generated automatically they foster keep learners motivated. And thirdly, highlight potential employing users, is both teachers learners, crowdsourcers help