Jack Grieve

New paper alert! "The Sociolinguistic Foundations of Language Modelling" Co-authored by a big team of researchers in @Eng_Lang_UoB Has now been published in @FrontiersIn AI! frontiersin.org/journals/artif… doi.org/10.3389/frai.2…

@JWGrieve @danaroemling @BodoWinter @EDACS_UoB @artsatbham @Eng_Lang_UoB @UoBGradSchool @CCR_UoB @unibirmingham Check it out for free with a GiHub account labs.iximiuz.com/playgrounds/cu…

@rljfutrell Hey looks great! Looking forward to reading. Our recent paper — coming from a different perspective — may be of interest. doi.org/10.3389/frai.2…

We conclude: “We don’t see ourselves as calling for radical change, but rather as acknowledging the interdisciplinary work that is already thriving in the 21st century. “ (FYI: Check out the paper for much more on Norm and Claudette!) </thread>

We can mess with the architectures and objectives, in order to figure out what inductive biases are necessary for language learning. We can probe internal representations and change them to see how the output changes.

There are twin threats to psychology: the well-known HARKing (Hypothesising After the Results are Known) and the newly coined MASKing (Making Assumptions based on Skewed Knowledge). New paper led by @SakSGhai in @CommsPsychol nature.com/articles/s4427…

New tutorial paper published today with PhD student @danaroemling and @BodoWinter 'Visualizing map data for linguistics using ggplot2: A tutorial with examples from dialectology and typology' doi.org/10.1017/jlg.20… 🔓 Hope people find this useful!

In December 2023, I defended my dissertation, “A geographic analysis of lexical variation in North American English using Reddit corpora”. It’s available to read for free on ProQuest. In this thread I’ll go over what I did and some of the most interesting things I found! (1/9)

@gwalkden Halliday and functionalist in general think language is mainly for thought as opposed to communication?

"Kuhn describes an immature science, in what he sometimes calls its ‘pre-paradigm’ period, as lacking consensus. Competing schools of thought possess differing procedures, theories, even metaphysical presuppositions. ...

Excited to share my open-access paper published in Research Methods in Applied Linguistics (@RMALJournal)! It critically explores the limitations of traditional linear regression models (assuming normal error distribution) for analysing syntactic complexity measures. 1/10

@adawan919 @SaraBartl @MatteoFuoli @jegraf1 @Weihang_Huang_ @mrkm_a @BodoWinter @X lol

@JWGrieve @SaraBartl @MatteoFuoli @jegraf1 @Weihang_Huang_ @mrkm_a @BodoWinter @X factually and scientifically. 4/4

New paper alert! "The Sociolinguistic Foundations of Language Modelling" Co-authored by a big team of researchers in @Eng_Lang_UoB Has now been published in @FrontiersIn AI! frontiersin.org/journals/artif… doi.org/10.3389/frai.2…

How can AI be reconfigured to avoid spreading misinformation and harmful social biases? According to a new paper from Professor Jack Grieve and his team of linguistics researchers, the answer lies in the application of sociolinguistics. birmingham.ac.uk/news/2025/unde…

Our position paper on the sociolinguistic foundations of language modeling is out now in Frontiers in AI! frontiersin.org/journals/artif… We argue that sociolinguistics holds the key to understanding and improving large language models.

@adawan919 @SaraBartl @MatteoFuoli @jegraf1 @Weihang_Huang_ @mrkm_a @BodoWinter I think the claim, like you're maybe saying, is essentially tautological. The point of the paper is just to make it, as nobody else has, and to consider its implications. It's not an empirical study but it's just one paper but neither are your tweets.

@JWGrieve @SaraBartl @MatteoFuoli @jegraf1 @Weihang_Huang_ @mrkm_a @BodoWinter When I try to get one to "evacuate NLP/the 'language' space", I'm not trying to be mean. One cannot merely compute/evaluate with textual values/meanings alone. There's been some (cross-generational) mistake/oversight. (How to help me with this thankless task? Please just listen.)

@tobyordoxford frontiersin.org/journals/artif…

The Scaling Paradox: AI capabilities have improved remarkably quickly, fuelled by the explosive scale-up of resources to train the leading models. But the scaling laws that inspired this rush actually show very poor returns to scale. What’s going on? 1/ tobyord.com/writing/the-sc…

Some media coverage of our new paper msn.com/en-us/money/ca…

Artificial intelligence must be made less racist and sexist, scientists say 🖊️ @hannagsslr #DailyExpress express.co.uk/news/science/1…

Excited to share that our new paper has been published in Frontiers in AI!

Please let me or my co-authors (@SaraBartl, @MatteoFuoli, @jegraf1, @Weihang_Huang_, Alex Jawerbaum, @mrkm_a, Marcus Perlman, @BodoWinter) know if you have any questions!

In this position paper we argue that LLMs are inherently modelling what are known in sociolinguistics as "Varieties of Language" and that this insight has widespread value to the development and deployment of language models...