Abstract We introduce a robust and shallow approach for grammatical role labeling (dependency labeling) where data-driven and theory-driven aspects are combined in a principled way. A classifier provides empirically justified weights, linguistic theory contributes well-motivated global restrictions, both are combined under the regiment of optimization. The empirical results of our approach are promising. However, we have made idealized assumptions (small inventory of dependency relations and treebank-derived chunks) that clearly must be replaced by a realistic setting.