We wrote a position paper for ACL’s theme session on linguistic diversity: From text to talk: Harnessing conversational corpora for humane and diversity-aware language technology (Dingemanse & Liesenfeld 2022).
We were pleasantly surprised when Thamar Solorio, one of the speakers in the Next Big Ideas plenary session at ACL, highlighted a key line from the conclusion of our paper as her Takeaway Message: “We need language models that are representative of the actual ways in which people use language [and that] give people the feeling they do not have to leave their own linguistic identities at the door”.
![Screenshot of slide from Thamar Solorio, quoting the following line from our paper: "We need language models that are representative of the actual ways in which people use language [and that] give people the feeling they do not have to leave their own linguistic identities at the door".](https://markdingemanse.net/elpaco/wp-content/uploads/sites/3/2022/12/image-6-1024x576.png)
By the way, one of the more puzzling ACL reviewer comments we got was precisely about that line (among others), and featured a serious charge that @a_liesenfeld and I now often lob at each other: 🚨 “figurative language in evidence” 🚨!