Interesting paper, though, once again, I'd maintain that a sentiment != an emotion. At best, it is a rough proxy; at worst, not even that. The same applies for

Interesting paper, though, once again, I'd maintain that a sentiment != an emotion. At best, it is a rough proxy; at worst, not even that. The same applies for instance to the following paper that uses lexemes: https://news.ycombinator.com/item?id=38504719 https://news.ycombinator.com/item?id=38504719 More generally, I'd argue that we'd need a careful longitudinal validation of the 'sentiment := emotion' assertion with human subjects (readers). Note also that a culture and a national language matter a lot here. It would be also interesting to do a similar validation with writers of the stories sampled (cf. what Rowling herself would think about Fig. 2). But note that this criticism is essentially only about terminology. And kudos to citing Vonnegut's classical presentation!