Thread 3: A corpus-assisted discourse analysis of linguistic transphobia on Mumsnet

[VitoCorleoneOfMNMafia]

In which we continue to discuss the Aston scrapists.

• The Site Stuff thread: https://www.mumsnet.com/talk/sitestuff/5057903-mumsnet-corpus
• FWR thread #1: https://www.mumsnet.com/talk/womensrights/5057460-a-corpus-assisted-discourse-analysis-of-linguistic-transphobia-on-mumsnet
• FWR thread #2: https://www.mumsnet.com/talk/womensrights/5060410-thread-2-a-corpus-assisted-discourse-analysis-of-linguistic-transphobia-on-mumsnet

Interesting that the money is given in dollars (US, presumably).

Remind me who in the women-vs-tra debate is allegedly funded by deep-pocketed americans?

Remind me who in the women-vs-tra debate is allegedly funded by deep-pocketed americans?

Hmm, that's a tricky one.

[scratches head]

Both people leading on that project for Aston have used the MN data (Kredens in the infamous ‘sandbox’ presentations, and Grant in at least one paper which used the adoption threads).
The project is about authorship attribution.

So a) they might not be trying to identify that Talulahalula is Mrs PissedOff from Large Town in the U.K., but they are trying to create a tool which identifies who wrote what and they used MN posts to train their LLM to do so. And b) Aston have gained to the tune of £11.3 m from work and reputations which that data scrape contributed to. And if that is not commercial gain, I am not sure what is. It was also contribute to the grant income section of the REF and again, reputation, which will contribute to PGT student fees. Etc.

Someone up thread mentioned an Aston University event next week.

I thought that had been cancelled.

If not is it anything to do with this department and is it online?

I did try and search but couldn't find anything.

Thanks

This is next week’s event:

https://www.eventbrite.co.uk/e/forensic-linguistics-and-artificial-intelligence-in-person-tickets-855866318197

Speakers include the original 2 Super Automatic Dox Machine Scrapers and the Head Honcho at Aston Forensic Linguistics Dep.

The PhD student is yet to rearrange the talk they/them cancelled due to ‘illness’.

https://www.eventbrite.co.uk/e/forensic-linguistics-and-artificial-intelligence-online-tickets-855861834787

Online tickets are supposedly sold out

Screenshots

Here’s the Eventbrite index for events by the FL department: https://www.eventbrite.com/o/34137004949

It actually says that you need to apply for an online ticket. Which suggests to me that - in an application of foresight and intelligence which has until now been lacking - they anticipate an influx of scraped and furious vipers.

happy to print Scraped and Furious Viper on a t-shirt if anyone fancies it

Yes, It’s possible that the Eventbrite for the online version was ‘sold out’ from the off!

Which suggests they are going to be discussing the corpus

Ooh my second mortgage is looking bigger by the day

Thanks for the links - interesting that the scrapping on mumsnet would have fitted in with this (or is this not the event our student withdrew from?). Quote:

" ... stylistic approaches to authorship analysis, which involve manual feature discovery at the token level, and stylometric approaches to authorship analysis, which involve the automated analysis of pre-defined feature sets at the type level. ... "

is that how they say they can identify users?

In the mean time I would have been interested in being an online scraped and furious vipers

Or a sc-raped and furious viper.

There is also the conference in June that the PhD student was due to present her research at. Her topic was removed from the programme but she's still down to present something. It's up to £450 to attend so I doubt any of us could afford to go. However, we could book a table or two at Asha's restaurant in Birmingham to join them at their conference dinner. 26th June if anyone is interested.
I've been to Asha's a couple of times. Foods expensive and nice but nothing special.

And if that is not commercial gain, I am not sure what is.
@Talulahalula IME academics are very good as not seeing any money that might flow in their general direction as 'commercial gain' because they're all about the furthering of knowledge and making any £££ out of it is the furthest thing from their innocent, non-capitalist brains.

I read papers on canine behaviour and training (because I am a nerd) and the number of people who sign off as having no conflicts of interest while producing research that supports the way they work in private practice outside academia is quite unnerving.

I wouldn’t be surprised if the university refuses to delete that recent scrape under ‘exemptions’ despite MN T&Cs, that they will defend the PhD student, and just ask EP to simply retitle their thesis. However, none of us ought to be surprised when the ‘results’ of the analysis remain the same ie MN/FWR is transphobic.

The supervisor, IMO, clearly reacted quickly and negatively to MN posters’ objection of misrepresentation, misuse of data, poor methodology, and confirmation bias, which is suggestive of being quite reflexively antagonist to criticism. Perhaps it wasn’t just Justine’s query that prompted a reaction and that even now software is used to scrape / continuously monitor MN? Whereas the Uni might assume that in 3 years time we’ll have forgotten about EP. Altogether I’m not holding out hope of any reflection, apologies or acceptance of poor research any time soon.

edited to add Justine

There’s obviously all sorts of back and forth going on, but I’m intrigued. If Aston has accepted whatever the MN/legal rationale was to delete the larger scrape, why didn’t that same rationale apply to the smaller one. It points to something being different about the more recent scrape. The difference could well be the recent scrape has a different “owner” of the data, or something else entirely. No doubt it will all be revealed eventually.

And just to say also that I could understand that if you were nearly at the end of your first year of your/your students PhD, and suddenly facing the prospect of it all going tits up and having to start again, you’d fight quite hard for that not to happen. However my sympathy is somewhat mitigated by the poor research hypothesis and the unethical approach to obtaining that data. I’d say to EP and Nicci that they should accept the sunk costs are sunk and the sooner they come up with something else, the better it will be for EP. EP should end up with a higher quality thesis with an unbiased hypothesis and an ethically obtained data set.

'EP should end up with a higher quality thesis with an unbiased hypothesis and an ethically obtained data set.'

Why should Edenself be constrained by these white, male, heteronormative ideas of what a thesis should be? Gotta queer those concepts! 🤐

I don’t want to exist in a world where ‘consent’ has been ‘queered’!

Wasn't that Foucault's main aim?

It's horrific.