The corpus linguistics brief has already been raised in oral argument (by Justice Comey-Barrett). supremecourt.gov/oral_arguments…

Lots of folk linguistic claims being bandied about like 'your brain autocorrects 'and' to 'or'' in certain contexts.

🤖🧠NEW PAPER🧠🤖

Language models are so broadly useful that it's easy to forget what they are: next-word prediction systems

Remembering this fact reveals surprising behavioral patterns: 🔥Embers of Autoregression🔥 (counterpart to 'Sparks of AGI')

arxiv.org/abs/2309.13638
1/8

Zining Zhu Shunyu Yao Dan Friedman matt hardy Griffiths Computational Cognitive Science Lab We thought that too - but it turns out that GPT captures the spelling of its byte-pair tokens! (see image for how we tested this).

Here's a thread about a great paper that describes how GPT might learn this info: twitter.com/kmahowald/stat…

Our big language fMRI dataset is now officially published!

📰 The paper: nature.com/articles/s4159… (free pdf: nature.com/articles/s4159…)

🧰 Code to download data & build models: github.com/HuthLab/deep-f…

💾 The dataset: openneuro.org/datasets/ds003…

📣 Today we launched an overhauled NLP course to 600 students in the online MS programs at UT Austin.

98 YouTube videos 🎥 + readings 📖 open to all!
cs.utexas.edu/~gdurrett/cour…
w/5 hours of new 🎥 on LLMs, RLHF, chain-of-thought, etc!

Meme trailer 🎬
youtu.be/DcB6ZPReeuU

🧵

New paper alert 📣. In this paper (published in Cognition Cognition, link at tinyurl.com/deictics), we (@rljfutrell Kyle Mahowald and I) explain spatial demonstrative systems across languages using the Information Bottleneck, plus some new constraints. thread. 1/16

Yes vote for UT #SXSW2024 panels, including ours on The Art of Crosswords! panelpicker.sxsw.com/vote/134394

Send COLA to #SXSW2024 and #SXSWEdu by voting for our 12 proposed panels by Sun, Aug 20!

Find the list of COLA panels and how to vote here: liberalarts.utexas.edu/news/vote-for-…

From the always wonderful Open Source, a useful exchange (from 2017) for people thinking of using LLMs trained on 100B words to inform theories of language acquisition (radioopensource.org/noam-chomsky-a…):

This paper is very interesting and introduces to the literature (I assume) the ngram “ants formicating meaninglessly in the sand”.

The text below is from Christopher Potts wonderful paper on PiPPs. For me at least, it carves out a natural middle path between hardcore generative nativism and black box functionalism: generative grammars are clearly 'real' but learnable with very few innate learning mechanisms.

1/2

Surprising though I know it may be that a paper on the English Preposing in PPs construction could be not only linguistically interesting but also sharply insightful about LLMs and amusing and moving, Chris's paper lingbuzz.net/lingbuzz/007495 that goes with this model is all of that.

Now that you’ve no doubt solved your Sunday crossword puzzle, looking to read about crosswords and linguistics? In The Atlantic theatlantic.com/science/archiv…, Scott AnderBois, Nicholas Tomlin, and I talk about what linguistics can tell us about crosswords and vice versa. Thread.

If you solve a lot of crosswords, then you’re fluent in a grammar that you cannot fully describe. Scott AnderBois, Kyle Mahowald, and Nicholas Tomlin explain the hidden rules that govern clues: theatlantic.com/science/archiv…