iiinova — where your digital self keeps evolving

grushaprasad.bsky.social @grushaprasad.bsky.social · 06/10/2026

I have somehow taken on the role of making instagram posts for my department account (against better judgement). I made my first IG post ever today where the image ended up being just a blank white page 🤦‍♀️ #clanker

060

Reposted by grushaprasad.bsky.social

Shravan Vasishth @shravanvasishth.bsky.social · 01/10/2026

Applications are open for the 2027 summer school on statistical methods for linguistics and psychology: All details are here: smlp.science

smlp.science

The Eleventh Summer School on Statistical Methods for Linguistics and Psychology

01718

Reposted by grushaprasad.bsky.social

Jane Li 🦖 @janeli.bsky.social · 17/09/2026

🦀New preprint! (w/ @najoung.bsky.social)🦞 Is grammaticality a major organizing principle of NLM representations? We show that many NLMs exhibit abstract rep. separation for grammaticality. We believe this work addresses debates about confounds in measuring model gram. knowledge. [1/10]

12010

Reposted by grushaprasad.bsky.social

Elliot Murphy @elliot-murphy.bsky.social · 15/09/2026

New work out today! The result of a 15 year project to offer genuine linking hypotheses between linguistics and neuroscience: a revised research program, and its first result - a novel neural binding operation ('Meld') capturing core properties of natural language syntax 🧵 arxiv.org/abs/2609.14384

arxiv.org

Formal Properties of Language as Constraints on Neural Dynamics

What must a neural system be capable of to implement language? Current research annotates stimuli with linguistic variables and tests which electrodes, voxels, or language-model layers predict neural ...

26915

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Yes sorry, should have clarified! The study is entirely in English. So in addition to replicating this with other phenomena and paradigms, it is important to replicate with other languages! Would be very interesting if there were between language differences for same phenomenon!

020

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Does native speaker filtering impact effect sizes: Compared alternative definition of native speaker in terms of language usage and AOA to more commonly used self-report. No differences for SONA-lab, but there were differences in SONA-online. n/n

030

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Does it matter when in the quarter participants were recruited? No clear evidence for this. 9/n

130

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Prolific yields higher quality data than MTurk? No evidence for this when using masters. BUT recruiting these participants was PITA. So… can just keep using Prolific for convenience. 8/n

140

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Implication: qualitative patterns are likely to be replicated on web-based platforms with enough power. But when evaluating quantitative predictions, same experiment can result in different empirical magnitudes depending on the platform. Read on for practical implications. 7/n

140

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

These differences cannot just be from population differences because SONA-lab and SONA-online came from the same participant pool 6/n

130

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

In exploratory analyses with fillers, we also looked at other commonly used predictors and found differences in position, frequency, and spillover effects of surprisal. 5/n

130

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

We find strong evidence that garden path effects are smaller across all web-based platforms compared to lab. Adaptation effect is also numerically smaller, but there was insufficient evidence for this (likely cause this was a three-way interaction) 4/n

120

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

To evaluate this we replicated prior work on syntactic adaptation to garden path effects (Prasad & Linzen, 2021) on four platforms with relative large number of participants: SONA-lab (507), SONA-online (517), Prolific (517), MTurk (masters; 240). 3/n

120

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Most data used to evaluate quantitative predictions (e.g., from surprisal theory) comes from web-based platforms like Prolific or MTurk. This assumes that effects in these platforms are comparable to lab effects. But is this true? 2/n

120

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/08/2026

Excited to share a preprint of work done in collaboration with @emilynguist.bsky.social ! osf.io/preprints/ps... TLDR: some effect sizes are larger in lab than on web-based platforms. So when evaluating quantitative predictions, considering the testing platform can be important! 1/n

2388

grushaprasad.bsky.social @grushaprasad.bsky.social · 04/08/2026

Why is the submission process for some journals so painful? 😩 I’ve now entered the title and author and funding and credit contributions in 10 different ways.. and now been stuck in “building pdf” phase foreverrrr

061

grushaprasad.bsky.social @grushaprasad.bsky.social · 25/07/2026

Yea agreed! I think was getting tripped up cause i was imagining a person trapped inside a wall that needed cut free. But actually it’s maybe more nested? a person_i behind the walls inside person_i who needs cut free?

120

grushaprasad.bsky.social @grushaprasad.bsky.social · 25/07/2026

#linguistics hive mind: what is the structural analysis of this line from the song other side “I’ll cut you free out of the drudgery and walls you keep in” I’ve always thought the walls were something that kept the person in.. but that doesn’t work right!? So is the person keeping a wall in them?

110

grushaprasad.bsky.social @grushaprasad.bsky.social · 24/07/2026

Lots of great points here! But this one really stood out 🔥

060

grushaprasad.bsky.social @grushaprasad.bsky.social · 16/07/2026

Since Hamilton NY is so tiny all AQI predictions are likely extrapolations.. It’s so interesting to see how sources that all broadly have the same predictions for nearby cities generalize wildly differently here. Accuweather thinks it’s 118 but apple weather thinks it’s 469 lol

040

grushaprasad.bsky.social @grushaprasad.bsky.social · 16/07/2026

The pervasiveness yellowness caused by the poor air quality makes me feel like I am in the Firefly show

010

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/07/2026

Also structuring the talk to provide a careful framing of the problems/challenges instead of presenting any "results" or "solutions" was a nice meta-level way of demonstrating one of the central points: the importance of focusing on the articulating questions we are trying to answer

060

grushaprasad.bsky.social @grushaprasad.bsky.social · 05/07/2026

Really loved Phil's keynote at #acl2026! The balancing act framing was very effective, and was itself a balancing act between articulating the ideological qualms about the state of the field and the material conditions of the world that contribute to sense of powerlessness wrt changing this state

191

grushaprasad.bsky.social @grushaprasad.bsky.social · 03/07/2026

Totally did not expect to get a beach towel and sunglasses in a conference welcome bag! High surprisal

2120

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/05/2026

I remember the same thing happening to me in undergrad/grad school, and I think more than once!

010

Reposted by grushaprasad.bsky.social

Anzi Wang @anziw.bsky.social · 25/05/2026

This paper was accepted as a talk at SCiL this year—please stop by if you might be interested! Meanwhile you can read our preprint here: osf.io/preprints/ps...

osf.io

OSF

181

grushaprasad.bsky.social @grushaprasad.bsky.social · 25/05/2026

As a more "syntax processing" person this was very different from anything I've worked on before, but also very fun! Thanks @anziw.bsky.social for exploring this space together, and @carolynanderson.bsky.social for sharing RSA wisdom!

2101

grushaprasad.bsky.social @grushaprasad.bsky.social · 24/05/2026

You just can’t get yourself to do something and keep putting it off? Will issue

050

grushaprasad.bsky.social @grushaprasad.bsky.social · 04/05/2026

In 2016, I started working on my undergrad thesis where I was looking at ERP signatures for singular 'they'. In 2026, the work is finally published! www.frontiersin.org/journals/psy... Kudos to Joanna Morris for working on this the past few years and pushing it past the finish line!

frontiersin.org

Frontiers | Distinct ERP signatures for singular “they” and gender violations

IntroductionThe English pronoun system is undergoing rapid change, with singular “they” increasingly used to refer to specific individuals. A key question is...

1156

grushaprasad.bsky.social @grushaprasad.bsky.social · 02/05/2026

One of the several reasons I dislike spring semester!

020

Reposted by grushaprasad.bsky.social

Kirby Conrod @kirbyconrod.bsky.social · 30/04/2026

new blog post about, like. why i teach syntax, i guess. may be a little bit inside baseball, curious about your thoughts kconrod.medium.com/build-me-a-b...

kconrod.medium.com

build me a bird

or, an apologia for (still) teaching syntax

6416

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/04/2026

Came here to complain at the exact same time I see

050

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/04/2026

"If you wish to register in this round we strongly encourage you to do so immediately on the 12th, because registration is capacity-limited and may sell out" Since when did trying to register for conferences become like trying to buy concert tickets??

190

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/03/2026

In #219 with @wtimkey.bsky.social, Kuan-Jung, @byungdoh.bsky.social, @sarehalli.bsky.social, @tallinzen.bsky.social, @linguistbrian.bsky.social we show that LM surprisal is sufficient to capture forward reading time but not regressive measures in eye-tracking!

040

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/03/2026

RSA modeling often assumes that meaning that speakers are certain about the meaning that they want to convey. Uncertainty comes from other sources. In #335 with @anziw.bsky.social we show modeling meaning uncertainty can account for differences between gumballs and elections that vanilla RSA can't!

150

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/03/2026

Graded structural priming is often used to argue for shared structure. In #365 @anids.bsky.social and @shotamomma.bsky.social use a simplified ACT-R based parser to show that this graded priming can emerge without shared structure because of *inhibition* of baselines that seem neutral, but are not!

180

grushaprasad.bsky.social @grushaprasad.bsky.social · 27/03/2026

Have three posters with my awesome collaborators today at #hsp2026! Read on more and come to the posters to talk about the importance of neutral baselines in priming, the need to model uncertainty in meaning in RSA, and the dissociation of prediction and structural processing in eye-tracking!

171

grushaprasad.bsky.social @grushaprasad.bsky.social · 24/03/2026

Extreeemely late to the party, but finally used energy gathered from sabbatical + HSP excitement to create a bluesky account!

0232