My name is Satoru Ozaki. I am a PhD student in the Department of Linguistics at UMass Amherst.
Before that, I graduated from the Master of Language Technologies (MLT) program at Language Technologies Institute at Carnegie Mellon University. My advisor was Prof. Eric Nyberg. I also worked extensively with Prof. Lori Levin.
My interests are:
- Theoretical linguistics
- Syntax/semantics (esp. binding)
- Constraint-based grammars, because I was converted at UMass
- Computational linguistics
- Parsing and grammar induction (phonology, syntax)
- Evaluating how well neural language models understand natural language syntax
I write about theoretical/computational linguistics on this website.
email: sozaki [snabel-a] umass [punkt] edu
LaTeX
To each their own, but here’s my favorite setup:
mlmodern Good font, but doesn’t work with XeLaTeX because it uses fontenc (see here for usage) langsci-gb4e An example numbering package. Though, the numbering seems to break after 4 levels of embedding. tabularray A nice modern package for tables. Tips Put formulas in numbered examples You can put inline math in numbered examples like this:
\ea \( x^2 = 2 \) \z What if you want aligned equations?…
Read more ⟶
KaTeX
Setting up this website This website uses Hugo with the theme Archie. I’m using the font Inter. There is also KaTeX support (see here here and here).…
Read more ⟶
Stanfun
Fun with Stan Some notes on trying to fit mixture of multivariate Gaussian models in Stan.
Setting things up Get CmdStan. Unzip, cd into the cmdstan-x.xx.x directory and run make. This doesn’t build anything, it just prints you some info on how to compile and use CmdStan.
CmdStan is a compiler that takes a statistical model description written in the Stan language (extension .stan) and builds a binary that you can use to fit the model on some data, diagnose the model, etc.…
Read more ⟶