Alexander Kilpatrick

I do cool stuff with maths and language.

University of Aizu
Centre for Language Research
Director of the Language, Cognition, and Computation Lab

[Alexander Kilpatrick @alex98516712]

This is my new website: https://t.co/w5VXk7he3I

[Cognition @CognitionJourn]

Language isn’t optimized just for efficiency. This study shows that phonological surprisal systematically marks vivid meanings—slowing processing but strengthening memory. Surprisal isn’t noise; it’s how language directs attention.

https://t.co/Q7d0wD60cE

[Alexander Kilpatrick @alex98516712]

American English words that are specific, imaginable, or concrete tend to be made up of surprising phonological sequences. We propose the Attentional Optimization Hypothesis to explain this.

Say it like you mean it https://t.co/S9iwgeqvbq

[Alexander Kilpatrick @alex98516712]

Clarification: Summaries of my paper “What AI Might Teach Us About the Origin of Human Language” (arXiv:2301.06211) have appeared on ResearchBunny and in their promotion material. I was unaware of them until recently and I did not commission, fund, or collaborate on these posts.

[Alexander Kilpatrick @alex98516712]

Languages are efficient—but not always. My latest work shows how surprising sounds help highlight vivid meanings in English. We call it the Attentional Optimization Hypothesis. Video here: https://t.co/MGCGtWo66c

[A.Benítez-Burraco @abenitezburraco]

Words made up of improbable phonemes are processed with cognitive disadvantage, but that extra processing effort enhances their retention in long term memory
https://t.co/9CnWCAbrLT

[Alexander Kilpatrick @alex98516712]

🔍 What happens when the sounds in a word are unexpected?
Our new study explores how phonemic surprisal—the unpredictability of phonemes—shapes how we process, remember, and evolve words.
📄 Published in @PLOSOne
🔗 https://t.co/sDFx5LUy9a
#psycholinguistics #informationtheory

[Alexander Kilpatrick @alex98516712]

Obligatory publication post: American English words with negative valence carry more Shannon's information. Both increased information and negative valence are associated with higher accuracy in a memory recognition experiment. https://t.co/u1TrcED6CI

[A.Benítez-Burraco @abenitezburraco]

Negative words carry more Shannon's information and are better recalled https://t.co/vhy5i4fO3b

[Alexander Kilpatrick @alex98516712]

My research on "The Negativity Bias is Encoded in Language" is now in press! It shows that negative words carry more information & are more memorable due to higher surprisal & iconicity. Dive into the details here: https://t.co/qWZEbMfqXA #CognitiveScience #Language

[Alexander Kilpatrick @alex98516712]

My latest research, "The Negativity Bias is Entrenched in Language," shows that American English reinforces negative valence, making words with negative meanings more surprising and memorable. Read it here: https://t.co/5hUEiFhznA

[Thomas Van Hoey 司馬智 Semantics = lexical semantics @Simazhi]

Melissa Ebert with Kilpatrick and @cwiekaleks on size #soundsymbolism with #xgboost based on cross-linguistic data

#ICL2024

[Alexander Kilpatrick @alex98516712]

Four months ago I tweeted about GPT-generated text in two published articles. Coincidentally, both were later retracted (one of duplicate publication, the other for issues of consent). Perhaps there is a lesson to be learnt here.

https://t.co/2yYFTN6fyY
https://t.co/VqN57Zi2Q0

[Alexander Kilpatrick @alex98516712]

Chatgpt response in first line of the introduction of this published paper. The authors, the reviewers, the editors... Did anyone actually read?

https://t.co/bIgEDduFbN

[Alexander Kilpatrick @alex98516712]

My favorite iconicity seminar, ICOSEM (@ian_joo_korea), just released it's call for papers. Deadline is 30th September. Hope to see you there.

https://t.co/NZ97RO6Yj3

[Alexander Kilpatrick @alex98516712]

Another published study from science direct with GPT responses in text.

https://t.co/0X4NflcfPC

[Alexander Kilpatrick @alex98516712]

Chatgpt response in first line of the introduction of this published paper. The authors, the reviewers, the editors... Did anyone actually read?

https://t.co/bIgEDduFbN

[Alexander Kilpatrick @alex98516712]

Our new study examining the difference that extreme gradient boosting makes when classifying Japanese given names according to gender. https://t.co/4i044dJgHB

[Alexander Kilpatrick @alex98516712]

https://t.co/Hd90jrYiPN
@PeerJCompSci
Read this. Or don’t. I mean it’s up to you.

[Uju Anya @UjuAnya]

University of Pennsylvania kicked Dr. Katalin Karikó off their tenure track saying her work “wasn’t faculty quality.” She left Penn after they refused to reinstate her. Now Dr. Karikó won a Nobel Prize, Penn is putting out press releases sounding like an O’Jays song.