From SPMRL to NMRL: What Did We Learn (and Unlearn) in a Decade of Parsing Morphologically-Rich Languages (MRLs)?

It has been exactly a decade since the first establishment of SPMRL, a\nresearch initiative unifying multiple research efforts to address the peculiar\nchallenges of Statistical Parsing for Morphologically-Rich Languages\n(MRLs).Here we reflect on parsing MRLs in that decade, highlight the solutions\nand lessons learned for the architectural, modeling and lexical challenges in\nthe pre-neural era, and argue that similar challenges re-emerge in neural\narchitectures for MRLs. We then aim to offer a climax, suggesting that\nincorporating symbolic ideas proposed in SPMRL terms into nowadays neural\narchitectures has the potential to push NLP for MRLs to a new level. We sketch\nstrategies for designing Neural Models for MRLs (NMRL), and showcase\npreliminary support for these strategies via investigating the task of\nmulti-tagging in Hebrew, a morphologically-rich, high-fusion, language\n