Thread by @emilymbender

I was feeling a little ranty this morning, but there's actually also some interesting points about context and pragmatics here, for when we write (or cause machines to write) text that will be interpreted in contexts we aren't directly participating in:

https://twitter.com/emilymbender/status/1322164196654551040

Surely, from the platform's point of view, WeCNLP is starting at 7am. For them, WeCNLP refers to an event with "start" and "end" times they have to program into their platform, so that people who have registered can access the platform during those times. >>

But for people *attending* WeCNLP (the addressees of that email), WeCNLP refers to an event with a specific internal schedule, of talks and informal meeting times. >>

The author of this kind of text (or programmer of the template, one step further removed even) can only influence the shape of the text at the point they write the text (or the template). >>

The reader has no access to the author (& might not even know who the author is, though I'm guessing the average #WeCNLP2020 attendee would suspect, like me, that this came from the platform and not the conference) and therefore limited info from which to guess @ author intent.>>

So for things to go smoothly authors have to put in extra effort to model the reader's context. In cases like creating signage for public spaces (think airports), I think it's generally clear that this is the case. >>

Probably less so when writing the templates behind auto-generated text.