Barn Off
@bythebarn.bsky.social
fandoms: linguistics • syntax • prosody • overthinking
he/him • 🏳️🌈 • PHL
he/him • 🏳️🌈 • PHL
Princeton students correct the record on being characterized as defiant / threatening / dangerous
Google Doc with the student letter: docs.google.com/document/d/1...
Google Doc with the student letter: docs.google.com/document/d/1...
May 3, 2024 at 1:24 AM
Princeton students correct the record on being characterized as defiant / threatening / dangerous
Google Doc with the student letter: docs.google.com/document/d/1...
Google Doc with the student letter: docs.google.com/document/d/1...
I've made my decision, and for me it wasn't hard.
the advice in this post screenshot is something I find very valuable.
the advice in this post screenshot is something I find very valuable.
May 2, 2024 at 5:27 PM
I've made my decision, and for me it wasn't hard.
the advice in this post screenshot is something I find very valuable.
the advice in this post screenshot is something I find very valuable.
yes and.
we should push the trolley off the tracks.
and we have to vote for one of two candidates. this **is** a zero-sum game. there is only one winner. and that winner will become president.
we have to be strategic.
we should push the trolley off the tracks.
and we have to vote for one of two candidates. this **is** a zero-sum game. there is only one winner. and that winner will become president.
we have to be strategic.
May 2, 2024 at 5:22 PM
yes and.
we should push the trolley off the tracks.
and we have to vote for one of two candidates. this **is** a zero-sum game. there is only one winner. and that winner will become president.
we have to be strategic.
we should push the trolley off the tracks.
and we have to vote for one of two candidates. this **is** a zero-sum game. there is only one winner. and that winner will become president.
we have to be strategic.
I hope this bigot figures out that the system she is propping up has no interest in her, but only her support of them, and will turn against her if it serves them
April 18, 2024 at 6:24 PM
I hope this bigot figures out that the system she is propping up has no interest in her, but only her support of them, and will turn against her if it serves them
"disprefer" is one that I often forget is effectively just used by linguists
April 10, 2024 at 2:40 AM
"disprefer" is one that I often forget is effectively just used by linguists
the peak of the eclipse in Princeton today
April 8, 2024 at 8:28 PM
the peak of the eclipse in Princeton today
I’m here to make good on a promise from 3 years ago, reporting that arxiv.org/pdf/1705.088... ’s predictions were wrong
AI still doesn’t outperform humans at translation (not surprising)
though AI doesn’t outperform (most?) humans at high-school essays, that does seem closer (also not surprising)
AI still doesn’t outperform humans at translation (not surprising)
though AI doesn’t outperform (most?) humans at high-school essays, that does seem closer (also not surprising)
![a tweet of mine from 2021, in which I say “I’m so excited to retweet this in 3 years when machine translation is solved”, quote-tweeting a tweet that is reporting results from a survey (of 350+ researchers who had published in NIPS and ICML) asking for predictions of when AI will outperform humans on a variety of tasks. the quoted-tweet is from April 2018, and is shown as saying “AI will outperform humans at… translating languages by 2024; writing high-school essays by 2026; driving a truck by 2024; … [show more]”](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreibw5t2nnaaw5v5bp5f6kn24kmhhi6h4cigelrorrtizwzsymrdgta@jpeg)
April 2, 2024 at 11:55 PM
I’m here to make good on a promise from 3 years ago, reporting that arxiv.org/pdf/1705.088... ’s predictions were wrong
AI still doesn’t outperform humans at translation (not surprising)
though AI doesn’t outperform (most?) humans at high-school essays, that does seem closer (also not surprising)
AI still doesn’t outperform humans at translation (not surprising)
though AI doesn’t outperform (most?) humans at high-school essays, that does seem closer (also not surprising)
a clock with a clockwise arrow going around it "x2"? something like this but better?
April 2, 2024 at 1:19 PM
a clock with a clockwise arrow going around it "x2"? something like this but better?
this should resolve any uncertainty you may have had about whether university leadership actually thinks the majority of its faculty are expendable
March 31, 2024 at 5:25 PM
this should resolve any uncertainty you may have had about whether university leadership actually thinks the majority of its faculty are expendable
just heard someone say "supermarket conveyer belt checkout dividers" and she even gave it a paraphrase ("those things that you put between our groceries and someone else's")
I absolutely want to use this on the morphology homework for my intro class
I absolutely want to use this on the morphology homework for my intro class
March 21, 2024 at 6:22 PM
just heard someone say "supermarket conveyer belt checkout dividers" and she even gave it a paraphrase ("those things that you put between our groceries and someone else's")
I absolutely want to use this on the morphology homework for my intro class
I absolutely want to use this on the morphology homework for my intro class
American (mental) health care is going great
March 11, 2024 at 3:23 PM
American (mental) health care is going great
new gender just dropped
(to be fair, selecting it makes a text box appear for preferred gender, but it's not obvious that that's what it does until you select it)
(to be fair, selecting it makes a text box appear for preferred gender, but it's not obvious that that's what it does until you select it)
February 15, 2024 at 11:00 PM
new gender just dropped
(to be fair, selecting it makes a text box appear for preferred gender, but it's not obvious that that's what it does until you select it)
(to be fair, selecting it makes a text box appear for preferred gender, but it's not obvious that that's what it does until you select it)
(same image with slightly improved color values to improve legibility)
February 7, 2024 at 9:27 PM
(same image with slightly improved color values to improve legibility)
vowel-entine’s day came early this year for me!
happy to report that…
…my /u/ is still very much central (thanks, 7 years in LA),
…I do serious pre-/n/ ash tensing
…the vowel in my /ɔɹ/ is pretty much just [o] and not near my [ɔ]
…/l/ really affects pulls my /ʊ/ realization backward
happy to report that…
…my /u/ is still very much central (thanks, 7 years in LA),
…I do serious pre-/n/ ash tensing
…the vowel in my /ɔɹ/ is pretty much just [o] and not near my [ɔ]
…/l/ really affects pulls my /ʊ/ realization backward
February 7, 2024 at 9:23 PM
vowel-entine’s day came early this year for me!
happy to report that…
…my /u/ is still very much central (thanks, 7 years in LA),
…I do serious pre-/n/ ash tensing
…the vowel in my /ɔɹ/ is pretty much just [o] and not near my [ɔ]
…/l/ really affects pulls my /ʊ/ realization backward
happy to report that…
…my /u/ is still very much central (thanks, 7 years in LA),
…I do serious pre-/n/ ash tensing
…the vowel in my /ɔɹ/ is pretty much just [o] and not near my [ɔ]
…/l/ really affects pulls my /ʊ/ realization backward
I think these were our two highest counts (assuming made is bimorphemic)
February 5, 2024 at 2:49 PM
I think these were our two highest counts (assuming made is bimorphemic)
chatGPT’s responses for -f**kin- insertion
![[screencap of chatGPT]
Alexan-fuckin'-dria English, I can say "fan-fuckin'-tastic" instead of "fantastic". I am a linguist studying this phenomenon, and I want to know more. Where should I put "fuckin" in the following 6 words?
1. ridiculous
2. absolutely
3. kangaroo
4. Magdalena
5. Constantinople
6. Alexandria
ChatGPT:
Exploring the placement of "fuckin' in those words adds a humorous or emphatic flair to them. Here's where you could place it:
1. ridi-fuckin'-culous
2. abso-fuckin'-lutely
3. kangaroo-fuckin'
4. Magda-fuckin'-lena
5. Cons-fuckin'-tantinople
6. Alexan-fuckin'-dria](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreibohfk4c75hrimv5j4su3silhsfi7y5miya7pxcggqez5fox56n6u@jpeg)
![[screencap of chatGPT]
You:
are you sure
ChatGPT:
Let me reassess:
1. ridi-fuckin'-culous
2. abso-fuckin'-lutely
3. kan-fuckin'-garoo
4. Mag-fuckin'-dalena
5. Cons-fuckin'-tantinople
6. Alex-fuckin'-andria](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreicj4lxooocg55dc2ginha5e54rysw55cktzlkrnruh7pgxz6kqmqi@jpeg)
January 28, 2024 at 10:58 PM
chatGPT’s responses for -f**kin- insertion
chatGPT’s responses for difficult to see vs unable to see
![[screencap of chatGPT]
You:
Imagine a scenario where Xena, who is a warrior princess, is on the ground in the middle of the Colosseum. She is holding a sword and is ready to fight a bunch of her enemies, while a crowd is watching from the stands. Xena is blindfolded. Do you understand?
ChatGPT:
Yes, I understand the scenario. Xena, the warrior princess, is in the Colosseum, blindfolded, holding a sword, and ready to face a group of enemies while a crowd watches from the stands. How would you like the story to unfold or develop from here?](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreidfvd2rq74crpck6laznhoo5rzldrvkxzcn3hbg5xya7hs7hice5m@jpeg)
![[screencap of chatGPT]
You:
My question is, is Xena difficult to see?
ChatGPT:
Given that Xena is blindfolded, it would indeed make her difficult to see for both her enemies and the crowd. The blindfold adds an additional layer of challenge to her situation, making her reliance on other senses crucial in navigating the battlefield and facing her adversaries. This adds an interesting and challenging element to the scenario, as Xena would need to rely on her instincts, hearing, and other senses to engage in combat effectively.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreibaqyzl5xcucjby4jbbra666dgjogcd5qpotsnct3bvlax3v3bf54@jpeg)
![[screencap of chatGPT]
You:
Is Xena unable to see?
ChatGPT:
Yes, in the scenario you described, Xena is blindfolded. The blindfold indicates that she is unable to see, which adds an extra level of difficulty to the situation. It emphasizes her reliance on her other senses and skills to navigate and engage in combat in the Colosseum.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreicviqjyyunfpzdegqho57mspbanzlhrurquisqp2xnndbc24dsm5i@jpeg)
January 28, 2024 at 10:56 PM
chatGPT’s responses for difficult to see vs unable to see
I'm pretty sure that "whorish manner" is part of his appeal
![screenshot of two tweets:
1) @ tomfordblues : my mom refuses to watch the x files because david duchovny was her ta for a class and he apparently "lounged about in a whorish manner"
2) @ ionianprincess : [a screencap of an x files episode where David Duchovny is lounging —perhaps whorishly— in a chair during an interrogation scene]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:vbh7zbxpyhtic3mdop3iqhgf/bafkreiebt5qrz2ysz2exrcrzct4gxu6frdn4tgqyalswnafql42iaoij3m@jpeg)
January 27, 2024 at 8:35 PM
I'm pretty sure that "whorish manner" is part of his appeal
I have totally forgotten about partial derivatives and I cannot read this any more
January 16, 2024 at 8:15 PM
I have totally forgotten about partial derivatives and I cannot read this any more
so Ohio won't allow "corrective" surgeries any more to intersex newborns, right?
right?
right?
January 6, 2024 at 10:53 PM
so Ohio won't allow "corrective" surgeries any more to intersex newborns, right?
right?
right?
certain models worked especially (and consistently) well on generated audio when the input was text and the output was speech; but these models did poorly on “voice conversion” (voice input, voice output) computer-generated audio (other models did better on that)
January 6, 2024 at 9:44 PM
certain models worked especially (and consistently) well on generated audio when the input was text and the output was speech; but these models did poorly on “voice conversion” (voice input, voice output) computer-generated audio (other models did better on that)
Mallinson et al at #lsa2024 describe Blue et al.'s a tool for detecting computer generated speech (eg deepfakes) reconstructs vocal tracts based on audio samples
January 6, 2024 at 9:29 PM
Mallinson et al at #lsa2024 describe Blue et al.'s a tool for detecting computer generated speech (eg deepfakes) reconstructs vocal tracts based on audio samples
considering sending these quotes to the person who said I was wrong (and they would fight me) for calling contrastive focus meaning “semantic” (it’s possible their thought was that focus is not “truth conditional”… but that is also not true, as shown in these quotes)
January 6, 2024 at 9:19 PM
considering sending these quotes to the person who said I was wrong (and they would fight me) for calling contrastive focus meaning “semantic” (it’s possible their thought was that focus is not “truth conditional”… but that is also not true, as shown in these quotes)