utku turk

PhD candidate @ University of Maryland
stats and psycholing janissary
fan of decomposing and phrasal spellout

[utku turk @UtkuTurkLing]

people think that I am only harsh on their topic and talk. here's my take on experimentation, and the people's misleading trust in them. I feel like alchemy is a good allegory and a reminder for all of us to think. https://t.co/NRHjUv0U6Q

[utku turk @UtkuTurkLing]

Will be in Boston until the Sunday for HSP, would be happy to meet with folks that is around!

[utku turk @UtkuTurkLing]

New and a lot more comprehensive version of this preprint is out, now includes bayes factor, prior sensitivity, and response bias analysis. As well as a lot more discussion on how memory might be tracking 'Association with Controllerhood': https://t.co/jPRjJl3pUM

[utku turk @UtkuTurkLing]

New preprint: https://t.co/UA8rH6daCQ! We know surface/phonological overlap can matter for reading & memory. But does it bleed into dependency resolution? In English, “pseudo-plurals” like /s/ in *cruise* don’t induce agreement errors—but Russian ones have been argued to. (1/3)

[utku turk @UtkuTurkLing]

Have you ever wondered when Lunar New Year - Ramadan - Ash Wednesday is going to be on the same day? It is 3571. And after next year, the next time they are going to be in the same week is 2127. Here's the code for it: https://t.co/G7hNCWLUDf

[utku turk @UtkuTurkLing]

@lakens I agree with it entirely, I am just afraid sometimes the tone might be disencouraging. Not in your case though, your materials/work helped me a lot for sure!

[utku turk @UtkuTurkLing]

(side note the bayesian analysis including an ordered model is ran with pyro)

[utku turk @UtkuTurkLing]

New blog post: https://t.co/VjMjBIZR1c Ran an analysis on whether some of the pictures used in previous advanced planning is sus. Used a CLIP similarity model to quantify picture<->target sentence similarity and how difficult to get the subject from the picture.

[utku turk @UtkuTurkLing]

I do not want to hype anyone about what LLMs are doing or what they can tell us. as I say in the blog post, what these models are doing and what humans are doing is of course different.

[utku turk @UtkuTurkLing]

They don’t. Across two experiments: nominal plural attractors reliably induced attraction, while plural verbal attractors don’t—despite identical surface -lAr. Takeaway: agreement retrieval is gated by controllerhood, and surface heuristics do not bleed into retrieval. (3/3)

[utku turk @UtkuTurkLing]

New preprint: https://t.co/UA8rH6daCQ! We know surface/phonological overlap can matter for reading & memory. But does it bleed into dependency resolution? In English, “pseudo-plurals” like /s/ in *cruise* don’t induce agreement errors—but Russian ones have been argued to. (1/3)

[utku turk @UtkuTurkLing]

An alternative: it’s not phonology, rather controller-eligibility. Turkish provides a test: the same suffix appears on nouns and verbs, both can be subjects, but only nouns can control agreement. If “looks plural” drives retrieval, plural-marked verbs should attract too. (2/3)

[Hiram Ring @hiramring]

Are most sentences unique? I investigated the Chomskyan claim (as summarized by @sapinker) using some corpora, inspired by a conversation with @langofmind: https://t.co/pYz0TL40ZS

[utku turk @UtkuTurkLing]

how to generate experimental pictures that has a consistent style using ai: https://t.co/3Dn7B4br9r

[utku turk @UtkuTurkLing]

I changed my website and have been writing blogposts about linguistics and tutorials on experiments: https://t.co/VR8dViqcnr. Here's some recent ones, feel free to comment and disagree or circulate to people who would disagree (which is one of the most things in linguistics)!

[utku turk @UtkuTurkLing]

a frustration post on unsolicited 'this is ungrammatical to me' comments when someone gives a linguistic talk, and an attempt to be charitable: https://t.co/3isMlKb6mx