Textbooks teach 'Prescriptive Grammar' (how language should be), but Treebanks are built on 'Descriptive Grammar' (how language is actually used, including errors and ...
Textbooks teach 'Prescriptive Grammar' (how language should be), but Treebanks are built on 'Descriptive Grammar' (how language is actually used, including errors and slang). How do modern parsing models handle the conflict between strict grammatical rules and the 'messy' reality of human speech?