From Text to Lexicon : The Annotation of Pre-Target Structures in an Italian Learner Corpus