bayesian coef plots inside latex tables
i want both tables and the coefficient whisker plots, but i dont like how many things just cut through the text. i wrote an automated script that makes nice tables with whisker plots in tikz
I’m a PhD candidate in Linguistics at the University of Maryland, working with Ellen Lau and Colin Phillips. My research investigates how different kinds of linguistic information—morphosyntactic structure and morphophonological form—are planned in production and retrieved during sentence processing, and how mismatches between them drive errors and create illusions in production and comprehension. I also work in formal properties of case and adjectival morphophonology, as well as question semantics and embedding facts.
I recently visited UMass to work with Shota Momma and Faruk Akkus. Previously, I received my MA from Boğaziçi University (advisor: Pavel Logačev) and was a visiting researcher at Masaryk University (advisor: Pavel Caha).
My favorite food is gata with koritz and my favorite icecream flavor is saffron and rose (5pm snack of Anthony Bourdain). In my freetime, I like doing calligraphy.
essays
news
- SCIL 2026: a talk on quantifying cross-linguistic syncretism with Transformer Attention and a poster on effect of frequency on syntax in suspended-affixation
- ELM 4 (UPenn): a poster on the role of modal strength in English · abstract
- Mayfest 2026: our yearly workshop this year focuses on production processes. · website
- FASL 35 (UCSC): a talk on status on (B/C/M/S[erbian]) prefixes in mental grammar · abstract
- CLS 62: a talk on probabilistic cues and structural cues in agreement attraction · slides
- TU+11 (MIT): two talks on the role of structural cues in sentence processing and effects on frequency on syntactic choices. · abstract1 · abstract2
- HSP 2026 (MIT): talk on the role of structural cues in processing, along with 3 posters! · talk abstract · posters on production · surface-level cues · frequency on syntax
- New preprint on insensitivity to surface cues and role of controllerhood association · preprint · lingbuzz
just don't forget people pursued alchemy for 2500 years
experiments are *necessary* and quite often *not misleading*, our blind trust in them is misleading. and I do not think the answer lies in being more careful with your stats, though that is also a must.
i promise this is the last one for a while: audio generation with elevenlabs
you do not have to hear yourself in experiments, and you can generate experiment instructions with ai. here is how.
unaccusativity syntax or picture difficulty?
i am obssessed with early advance planning, but let me make sure about some picture saliency
generating consistent linguistic experiment materials with AI
next step is to have a norming data on text-to-image-to-image generated standardized stimuli
what might they really mean when they say 'but it's not grammatical to me'?
on why that phrase is usually unhelpful and attempt to be a charitable
saving pcibex recordings to s3 with a python lambda
modern replacement for the outdated JS guide in the original pcibex website.
misunderstanding Kate's talk on nonuniformity of phonology and phonetics
Kate’s asymmetry meets Turkish rounding: evidence that’s gradient but syntactically bounded (also a rant about dual-process framings)
montreal forced aligner workflow for pcibex production experiments
step-by-step instructions for filtering, renaming, and aligning pcibex recordings with mfa
automating experimental recording management with bash
it got really annoying after a while, so i wrote this code
agreement attraction in turkish
attraction in turkish seems to be modulated by bias, register, task details, but not morpology
is /j/ really a sonorant in turkish? i guess not.
findings point to a fricative status for /j/, despite its occasional sonorant disguise
can ‘as if’ clauses be degree comparisons?
turkish as if clauses specifically target gradable adverbials and adjectives, pointing to a degree-based analysis
how to include r expressions in (quarto) yaml
using yaml with embedded r expressions for cleaner, reusable quarto configs
No matching items