Multilingual Lexicalized Constituency Parsing with Word-Level Auxiliary Tasks

We introduce a constituency parser based on a bi-LSTM encoder adapted from re- cent work (Cross and Huang, 2016b; Kiperwasser and Goldberg, 2016), which can incorporate a lower level character bi- LSTM (Ballesteros et al., 2015; Plank et al., 2016). We model two important in- terfaces of constituency parsing with aux- iliary tasks supervised at the word level: (i) part-of-speech (POS) and morpholog- ical tagging, (ii) functional label predic- tion. On the SPMRL dataset, our parser obtains above state-of-the-art results on constituency parsing without requiring ei- ther predicted POS or morphological tags, and outputs labelled dependency trees.