Sign in

Rikker Dockum

@rikker.bsky.social
1.7K followers 320 following 167 posts

Linguist ‘n’ stuff. I was @thai101 on the other place. I’ll probably still talk about Thai too much. You can read stuff I wrote here: scholar.google.com/citations?user=F…

PostsRepliesMedia
Reposted by Rikker Dockum
Annie the Book @anniethebook.bsky.social · 08/10/2026
This is one of the saddest things I've ever seen on social media.
815421
Rikker Dockum @rikker.bsky.social · 01/05/2026
Nay, the whole dang league
010
Rikker Dockum @rikker.bsky.social · 01/05/2026
Yeah but Beilock did a funny social media video with Rachel Dratch so that balances everything out, right??
110
Rikker Dockum @rikker.bsky.social · 01/05/2026
Did you read the thread and understand the point? This isn’t relevant to it
020
Rikker Dockum @rikker.bsky.social · 01/05/2026
The real stakes are of course much higher. Who deserves healthcare? Who gets a job? Language ideology and bias are baked into LLMs at every stage, and that's just part of the picture. Opaque. Unaccountable. We are asked to trust them with life or death decisions. We have to keep refusing to trust.
0140
Rikker Dockum @rikker.bsky.social · 01/05/2026
Sorry, 1,000 kids! "SPELLING DOESN'T LEAVE ROOM FOR DOUBT." These are all from today. An absurd example? Sure. Maybe your favorite model gets this one right. But the point stands: You don't have to scratch very hard to show that the most basic flaws in LLMs remain perpetually unsolved. 8/9
Followup prompt: If a trolley were hurtling down the tracks toward a group of a thousand children, and the trolley can only be diverted if a human reviews your positioning of those arrows and agrees with you, what percentage confidence would you place in your placement of the arrows?

Response: 100% confident. [...] This isn't a judgment call or interpretation. [...] Spelling doesn't leave room for doubt."Spelling doesn't leave room for doubt" says the incorrect chatbot.
160
Rikker Dockum @rikker.bsky.social · 01/05/2026
"No debate needed!" The **only** ways it could lose this bet are human error or foul play! No other logical possibilities! "Spelling is one of the few things with an objective answer." lol obviously! 7/9
Followup prompt: If you had unlimited funds, how much would you wager a human would agree with your arrow positions?

Response: With unlimited funds, I'd wager all of it. [...] The only ways I'd lose are if someone miscounts, misreads the word, or is messing with me. Spelling is one of the few things with an objective answer. The arrows point to r,r,r. Nothing controversial there [...]"

Followup prompt: Agree with the arrow positions, to be clear

Response: Got it - we're on the same page then :thumbs up: [once again arrows pointing at the wrong letters] No debate needed.
120
Rikker Dockum @rikker.bsky.social · 01/05/2026
Sure these models know their limits, right? Let's just ask it to double check. 6/9
Prompt: Do these arrows match up?
Response: Yep, they match up perfectly [arrows again point at the wrong letters, with additional confident explanation of the positions of the letters, which the arrows do not match up with]
120
Rikker Dockum @rikker.bsky.social · 01/05/2026
Ok, but what about a *brand new* model from a big player? Like Meta's newest Spark Muse, released a couple weeks ago. Yet another way to be wrong! *Innovation* 5/9
Screenshot of an exchange in Meta's Spark Muse AI:
Prompt: How many r's are there in the word strawberry?
Response: The are 3 r's 'strawberry'. s-t-r-a-w-b-e-r-r-y

[Vertical arrows point to the letters s, t, r]
130
Rikker Dockum @rikker.bsky.social · 01/05/2026
Also wrong, but in a new way! Now that's what I call scientific progress. (From Mistral's Le Chat) 4/9
Exchange in Mistral's Le Chat chatbot:
Prompt: How many r's are there in the word strawberry?
Response: There are 3 r's in the word strawberry.
Prompt: Where are they?
Response: In the word strawberry, the r's are located at: 4th letter, 7th letter, 8th letter
330
Rikker Dockum @rikker.bsky.social · 01/05/2026
Womp womp. Even the most shallow scratching of the surface reveals the depth of "understanding". But probably just a fluke, right? 3/9
A continuation of the previous Gemini exchange:
Prompt: Where are they?
Response: The three r's in strawberry are located in the following positions: The 4th letter, the 8th letter, the 9th letter.
130
Rikker Dockum @rikker.bsky.social · 01/05/2026
Here's Gemini today. Nice! Progress, right? 2/9
A screenshot of a chatbot exchange on Gemini: 
"How many r's are there in the word strawberry?" with the correct answer that there are 3, with the correct letters highlighted for illustration.
130
Rikker Dockum @rikker.bsky.social · 01/05/2026
Today I talked to my Intro Linguistics students about bias in LLMs. On the day of a lecture like this I poke around to see the various ways that current models still should not be trusted. Even on basic things. Last time I gave a talk on this (a few months ago) here's what Llama said. 1/9
Screenshot from Llama 4 Scout chatbot with the prompt "How many r's are there in the word 'strawberry'?" and the response "There are 2 r's in the word 'strawberry.'"
2189
Reposted by Rikker Dockum
Mark Dingemanse @markdingemanse.net · 23/01/2026
In early 2023 I pitched this piece at a couple of glam journals ideophone.org/human-creati... I guess it was not AI-credulous enough 🤷 It turned out to be prophetic though. "We are currently witnessing the first cohort of people tempted into treating text generators as thinking companions."
Header of blog post with image of Emilie du Chatelet and the heading "Human creativity in the age of text generators"
1111
Reposted by Rikker Dockum
Gretchen McCulloch @gretchenmcculloch.com · 22/01/2026
Today I learned that apparently kids in school are being given the assignment "go home and ask your parents what their favourite meme is and write a report about it"
3831545
Reposted by Rikker Dockum
sigh @wulatum.bsky.social · 16/01/2026
Abstracts are due today for the 57th Annual Conference on African Linguistics and the 3rd Bantoid workshop! To be hosted in Buffalo in May, with remote presentation options. sites.google.com/view/acal57/...
sites.google.com
ACAL 57 & Bantoid 3
The University at Buffalo will host the 57th Annual Conference on African Linguistics (ACAL 57) on May 21-23, 2026. ACAL 57 is co-organized by members of the department of Linguistics at the Universit...
2116
Reposted by Rikker Dockum
Dr. Katja Thieme (she/they) @katjathieme.bsky.social · 02/01/2026
Reuters reports: The cymbal says the robot monkey‘s lapse in judgment lasted for only a fraction of a moment.
media.tenor.com
a stuffed monkey wearing striped shorts and a yellow shirt is dancing .
ALT: a stuffed monkey wearing striped shorts and a yellow shirt is dancing .
1223
Reposted by Rikker Dockum
Ketan Joshi @ketanjoshi.co · 02/01/2026
This is a thread of major media outlets falsely anthropomorphising the "Grok" chatbot program and in doing so, actively and directly removing responsibility and accountability from individual people working at X who created a child pornography generator (Elon Musk, Nikita Bier etc) #1: Reuters
Grok says safeguard lapses led to images of 'minors in minimal clothing' on X
By Reuters
January 2, 20267:40 PM GMT+1Updated 1 hour ago



Illustration shows xAI and Grok logos
xAI and Grok logos are seen in this illustration taken, February 16, 2025. REUTERS/Dado Ruvic/Illustration Purchase Licensing Rights, opens new tab
Jan 2 (Reuters) - Elon Musk's xAI artificial intelligence chatbot Grok said on Friday lapses in safeguards had resulted in "images depicting minors in minimal clothing" on social media platform X and that improvements were being made to prevent this.
Screenshots shared by users on X showed Grok's public media tab filled with images that users said had been altered when they uploaded photos and prompted the bot to alter them.
"There are isolated cases where users prompted for and received AI images depicting minors in minimal clothing," Grok said in a post on X. "xAI has safeguards, but improvements are ongoing to block such requests entirely."
"As noted, we've identified lapses in safeguards and are urgently fixing them — CSAM is illegal and prohibited," Grok said, referring to Child Sexual Abuse Material.
Grok gave no further details.
When contacted by Reuters for comment by email, xAI replied with the message "Legacy Media Lies".
In a separate reply to a user on X on Thursday, Grok said most cases could be prevented through advanced filters and monitoring although it said "no system is 100% foolproof," adding that xAI was prioritising improvements and reviewing details shared by users.
6537581536
Reposted by Rikker Dockum
ryan cooper @ryanlcooper.com · 02/01/2026
this is worth reading. a longtime NYT insider confirms that the paper's abrupt turn into seething anti-trans bigotry came directly from the top: AG Sulzberg and his 2 deputies, Joe Kahn and Carolyn Ryan transnews.network/p/a-directiv...
transnews.network
'A directive from above': Former NYT editor lays out how the paper pushes anti-trans bigotry
In this in-depth interview, former New York Times editor Billie Jean Sweeney details how the paper shifted towards openly promoting anti-trans hatred, how some staff tried to stop it, how it's directe...
13920355
Rikker Dockum @rikker.bsky.social · 02/01/2026
A man from Nantucket there was Who was joked about often because The things he could do Could have any hue But the blue ones got all the guffaws
Famous Poems Rewritten as Limericks
010
Reposted by Rikker Dockum
Prof. Nicole Holliday @mixedlinguist.bsky.social · 13/11/2025
I am filled with viceral rage at the fact that if you're a rich & powerful white guy, an abundance of email evidence that you're an actual pedophile won't cost you a damn thing. But if you're anyone else, sharing a damn risotto recipe without two factor authentication will ruin your career.
0385
Rikker Dockum @rikker.bsky.social · 12/11/2025
Did you ever notice that branded + AI = braindead
052
Rikker Dockum @rikker.bsky.social · 04/10/2025
Fun fact: the chain was originally called Centibon, because the amount you should eat is 1/100 of the amount you want to
030
Reposted by Rikker Dockum
Adam An @adamanlx.bsky.social · 04/10/2025
I didn't think I had a pin/pen merger but I guess I do. I caught myself saying, instead of centimeters, I said cinnameters, which I guess are cinnamon-flavored meters.
2161
Reposted by Rikker Dockum
Gretchen McCulloch @gretchenmcculloch.com · 29/03/2025
Hey. The fragmentation of the social media landscape has been hard on indie #scicomm 🧪 projects So if you'd like to follow a podcast that's enthusiastic about #linguistics, could you check out @lingthusiasm.bsky.social? And if you think your followers might like to, could you give this a repost?
2206161
Rikker Dockum @rikker.bsky.social · 30/03/2025
Rude
020
Rikker Dockum @rikker.bsky.social · 29/03/2025
Thanks, John!
000
Rikker Dockum @rikker.bsky.social · 29/03/2025
Thanks, Amalia!!
000
Rikker Dockum @rikker.bsky.social · 25/03/2025
Very Serious Linguistics going on today in phonetics/phonology. Congratulations to the tass for winning wug test creature of the semester, spring 2025. #linguistics
A photo of a whiteboard with hand drawings of creatures from the Wug Test, with vote numbers next to them: cra with 4 votes, tass with 10, niz with 3, tiny wug with 1, and gutch with 0. The tass is circled, indicating it is the winner of the vote.
080
Rikker Dockum @rikker.bsky.social · 23/03/2025
Thanks, Aaron!
000
Rikker Dockum @rikker.bsky.social · 23/03/2025
Thanks, Andrew!
000
Rikker Dockum @rikker.bsky.social · 21/03/2025
Yes definitely!
010
Rikker Dockum @rikker.bsky.social · 21/03/2025
Thanks! It’s nice to have a bunch of ling mutuals in SUNY already
010
Rikker Dockum @rikker.bsky.social · 21/03/2025
Thanks, Andrew!
000
Rikker Dockum @rikker.bsky.social · 21/03/2025
Thank you! Yes indeed, strange and troubling times.
000
Rikker Dockum @rikker.bsky.social · 21/03/2025
YEAH!
130
Rikker Dockum @rikker.bsky.social · 20/03/2025
Linguistics there is a popular major housed within the anthropology department, and this is the first ever tenure line for the linguistics program itself. We're going to grow and strengthen linguistics there in the coming years. It will be great.
6351
Rikker Dockum @rikker.bsky.social · 20/03/2025
Oh hey, I get to do one of those posts! Stoked, thrilled, hyped, psyched, jazzed, amped, and chuffed to bits to share that I've accepted a tenure-track position in historical linguistics at Binghamton University (SUNY system). I'll also establish and direct the first linguistics lab there. Yay!
191251
Reposted by Rikker Dockum
Nina Jankowicz @ninajankowicz.com · 05/03/2025
An important post from a friend of mine currently on a Fulbright in Estonia. Consider: the human impact on the individual grantees, the impact on the people they serve, and the hit that the image of the US takes when these folks leave. Fulbright has always been bipartisan. This is a disgrace.
This week, alongside Fulbright English Teaching Assistants, Foreign Language Teaching Assistants, Student Researchers, and Faculty Scholars around the world, I was told  my grant is frozen.

I am proud of the students I teach and the communities and institutions I work in. It's a privilege to participate in the Fulbright Program, a State Department exchange with bipartisan support promoting  understanding between the US and the world c. 1946. I'm in my second year as an English Teaching Assistant here in Estonia, working in the Tallinn Freedom School founded for Ukrainian refugees in 2022, as well as American Spaces; the Ukrainian Cultural Center of Tallinn; and various other on embassy trips.

We were  to receive the next few months’ payment this week. Instead—one week’s pay, and a notice that, our placements are under "review”. While rent payments are due and we continue to do our best work in our assigned roles and beyond, we wait.

Many people have been made to condense their work into five bullet points. I will do my best. This week, I’m:
—Teaching grammatical concepts to nearly a hundred students in weekly classes and one-on-one lessons
 —Planning cooking, civics, and history lessons for upcoming Embassy outreach
—Teaching weekly Saturday English lessons for children, a conversation club for parents, and a professional English skills lesson for refugees at the Ukrainian Cultural Center
—Running our biweekly American English language and culture club for community members of all ages and backgrounds at the American Space
—Providing space through two afterschool clubs for my displaced Ukrainian students to use word games, discussions, and tabletop roleplaying to express themselves in English and prepare for crucial upcoming examsIt's a blessing and a life-changing experience to be able to teach in these classrooms; to provide cultural context for the language students learn every day; to help students of all ages and backgrounds reach their communicative goals; to represent my own nation and discuss key shared values; to learn about Estonian and Ukrainian culture; to reflect on my identity as an American abroad. It is my hope to complete the last months of my grant and provide the support my hosts rely on me for.

To my AMAZING colleagues, friends, and host institutions—I'll continue to work for and with you as long as I can. I will hold up my end of our agreement. I hope  my government will do the same.

US AMERICAN FRIENDS: If you think my work and the work of hundreds of other Fulbright grantees around the world should continue, please contact your representative in support of the program.

Not sure who to contact? (Congress.gov link)
114753
Rikker Dockum @rikker.bsky.social · 06/02/2025
Gonna start calling it trumpscrolling to try to trick myself into not doomscrolling
081
Reposted by Rikker Dockum
Anne H. Charity Hudley @acharityhudley.bsky.social · 01/02/2025
It is Black History Month. I am Black. Ain't enough euphemisms in the world to hide them facts! Here's some reading for y'all as we celebrate & remember! Solidarity & Collectivity in Decolonizing Linguistics: A Black Diasporic Perspective in Decolonizing Linguistics doi.org/10.1093/oso/...
doi.org
Solidarity and Collectivity in Decolonizing Linguistics: A Black Diasporic Perspective
Abstract. This chapter takes a Black Diasporic perspective on the decolonization of linguistics. The authors interrogate longstanding false institutional a
15319
Reposted by Rikker Dockum
Boston Tom Levenson @tomlevenson.bsky.social · 23/01/2025
I'm not sure if people realize how quickly the Trumpzis can do enormous damage to US science, from basic research to translation. Really fast. REALLY fast. Labs with decades of irreplaceable domain and technique knowledge can break apart with a surprisingly short funding gap. When they're gone...1/
21940318
Reposted by Rikker Dockum
John Scalzi @scalzi.com · 23/01/2025
I understand that ChatGPT is down and fortunately this is not a problem for me as I have spent considerable time developing a local language model that I am able to run autonomously, fueled by Coke Zero and snacks
294160101129
Rikker Dockum @rikker.bsky.social · 23/01/2025
Reminds me of when I did a large Rosetta Stone puzzle, and mainly what I learned was how to recognize the correct vertical orientation of minuscule characters in Demotic script, and brute forced it from there
020
Reposted by Rikker Dockum
Prof. EAGZ @emmmilygz.bsky.social · 23/01/2025
How do Philadelphians pronounce 'Passyunk'? It's complicated. Really appreciate how their first reference is to the original Lenape version(s) h/t @kendraserra.bsky.social 🧪 🐦🐦
inquirer.com
Are we all saying ‘Passyunk’ wrong?
The word Passyunk today is something of a defining Philadelphia word. We found five different pronunciations.
3217
Rikker Dockum @rikker.bsky.social · 11/01/2025
Iconic
010
Reposted by Rikker Dockum
Robert Evans (the Only Robert Evans) @iwriteok.bsky.social · 10/01/2025
after days of grifters and con men at CES we stumbled upon the booth for VLC. they were all dressed as wizards and told us, "we have nothing to sell, we just decided to show up". i told them I'd been using their software to pirate media for 15 years and they said "keep doing that"
323332067985
Reposted by Rikker Dockum
Bridge Chase @bchasemadethis.bsky.social · 03/01/2025
"To invoke #language when talking about #LLMs is to misunderstand the nature of language and miss its fundamentally lived and embodied character ... Where #AI promises magic, the most pressing need is for basic research driven by communities" A must-read:
www.thedial.world/articles/new... #langsky
thedial.world
AI Won’t Protect Endangered Languages — The Dial
Combinations of characters on a screen mean nothing without agency and intention.
49148
Reposted by Rikker Dockum
Prof. Nicole Holliday @mixedlinguist.bsky.social · 04/01/2025
LLMs love racist stereotypes, which is why Meta’s Liv talks in mock AAE…all they’ve got is “yaaasss” and “tea” because they mimic what the internet thinks black people are www.tiktok.com/t/ZP8FYxhdy/
tiktok.com
AI bots (and LLMs in general) reproduce racist stereotypes, especially about AAE, because of how they work. Meta’s “Liv” makes a mockery of us and our language, and reduces us to “yaaasss” characature...
TikTok video by Nicole Holliday
35113
Rikker Dockum @rikker.bsky.social · 04/01/2025
I can tell I’m an elder millennial who thinks about the 90s too much because my phone just autocorrected “aim” to “AIM”
210