Josef Fruehwald

interesting developments in the American borrowing of “queue”

i understand how this happened, but it looks chaotic

ok they say “any difference of opinion”, but but i couldn’t find any shares in “the Phonological Component must be Serial and Rule‐Based”.

i thought this was cool when I first saw it, but for some reason i just realized *i* could be Skip Fruehwald

Rstudio wrapped

bsky.app/profile/sal....

i go where the data takes me

orrrr they’re underlyingly “bike” and “trike” with velar softening: bike-icle, trike-icle

one thing they both have in common is they seem like they're in permanently bad moods, maybe even angry?

Wait, do you mean you don’t raise in tire-noun, or tire-verb?

I hope you're medium comfortable with IPA, cause otherwise that's just a bunch of weird character's I've dumped in here.

I don't *think* you have this phonological process, but a lot of people in the north raise /ay/ before voiceless consonants to [ʌi], so "write" is [ɹʌit] and "ride" is [ɹɑɪd]. But it's more complex than just that, because "bicep" is [bɑɪsɛp] while "bicycle" is [bʌisɨkl̩]

I've been low-key obsessed with pre-voiceless /ay/ for nearly 20 years, and still somehow I have "oh yeah... wierd" moments with my intuitions.

Not raised:
- bicep
- bisexual

Raised:
- bicycle
- dichotomy

Unsure:
- diverticulitis

YouTube video by 6abc Philadelphia Action News Theme Song - Move Closer to Your World (with lyrics)

bum bum, bum bum, bum da dee dum, ba da da dum dum!

there was a “do you experience yourself moving through time, or time moving past you?” meme personality question a while ago, and undoubtedly time is violent screaming torrent constantly blasting directly into my face

a man with a beard is surrounded by mathematical equations on his face ALT: a man with a beard is surrounded by mathematical equations on his face

its hard when their examples and questions do this to me, but I also need to keep this train on the rails

me: What words rhyme with “cute”?
student: “astute”
me: Oh, that’s a good example, but l want to cover some other stuff and come back to it.
…
me: Actually, I just want to *think* about it and not talk about it

AH! It's because C is the descendent of the collider, so by conditioning on it, it opens the collider path

This was roughly my understanding as well, but still leaves me scratching my head why dagitty is identifying as part of the minimal adjustment set. If I remove the B -> D arrow, B is still a precision parameter, but isn't identified by dagitty for adjustment.

ok, causal inference people: Lets say I have the following DAG. I see the backdoor path through C, so I adjust for C. dagitty says I *also* need to adjust for B, but I'm not sure why since there's a collider along its path?

Most academics strongly opposed to using AI in REF 2029 While new technologies seen as ‘game changer’ for national-level research assessment, study finds vehement opposition, particularly among humanities scholars

Higher Education in the US is a dumpster fire, but I'm still grateful to be free of the REF wretchedness, and what I assume will be intense pressure from the higher-ups to run everything through BritGPT or some such

On the other hand, if you included a question about "How often do you use new pronouns", 3sg they acceptability and neo-pronoun acceptability both cause the usage rate, so you should definitely *not* include it in the model (collider bias)

without controlling for age, your estimate of the direct effect of they->neo-p could really be mostly reflecting the common cause of both (confounding bias)

(forgive me if trying to specialize it for your work is too pandering!) If you had a question like "Does the acceptability of 3sg they for a person have a causal effect on the acceptability of other neopronouns?" You'd need to include variables that could be a common cause of both, like age

not really. These DAGs are more for formalizing the presupposed causal relationships between variables to help you figure out what you should and shouldn't include in a model

cc @timoroettger.bsky.social or @bodowinter.bsky.social ?

has anyone seen Directed Acyclic Graphs utilized for causal inference in a quantitative linguistics paper?

(p.s. I know syntax trees are usually DAGS)

second person plural they

i also really admire {lubridate} but don’t work with complex datetime info very often

oh yeah. and {here}

I think {stringr} is my favorite basic utilities package.