You’re not the POS of me: part-of-speech tagging as a markup problem
Bethan Siân Tovey · Balisage series on markup technologies · 2019
Part of speech tagging, labeling every token in a text with its grammatical category, is a complicated business. Natural language is messy, especially when that language consists of social-media conversations between bilinguals. The process can be done with or without human intervention, in a supervised or unsupervised manner, on a statistical basis or by the application of rules. Often, it involves a combination of these methods. It is, on the one hand, an obvious markup problem: mark up the tokens with appropriate grammatical categories. But it is also much richer than that. Theoretical problems that have been identified in the domain of markup can throw light on the problem of grammatical category disambiguation. Topics considered include subjectivity and objectivity, the semantics of tag sets, licensing of inference, proleptic and metaleptic markup, and the interesting characteristics of the Welsh “verbnoun”.