Grammar induction is a field that attempts to find syntactic generalizations in unannotated corpora. First, I will focus on work done during my Master's, in which I evaluate current shallow methods (ABL, Emile, Adios) independently. Consequently, I will argue that this line of work does not have a good outlooik, and that one should focus on deeper and more informed methods. A few directions of work are presented, following the 'corpus-oriented grammar development' approach exhibited by Miyao et al. (2004).