Hovy argues NLP 'ignored the social aspects of language'

Natural language processing has 'ignored the social aspects of language,' Dirk Hovy wrote in a 2018 position paper hosted at aclanthology.org.

The abstract puts the diagnosis plainly. Historical data and model limitations, Hovy writes, 'narrowed our focus and biased the tools demographically.' His prescription pairs two things that were newly abundant at the time: datasets carrying socio-demographic information, and 'more expressive (neural) models.'

'I argue that this combination can broaden the focus of NLP to solve a whole new range of tasks, enable us to generate novel linguistic insights, and provide fairer tools for everyone,' Hovy argues in the abstract.

The paper runs pages 42 to 49 of the Proceedings of the Second Workshop on Computational Modeling of People's Opinions, Personality, and Emotions in Social Media. It is a position essay rather than an empirical study. The abstract names no datasets, no benchmarks, no measured outcomes. Two researchers in our Who's Who directory shared the link.