bergelsonlab
LightNews LightNews
banner
bergelsonlab.bsky.social
bergelsonlab
@bergelsonlab.bsky.social
7.3K followers 1.4K following 580 posts
Official account for the Bergelson Lab at Harvard, sporadically maintained by the PI:).
Just a lab, trying to figure out how babies learn language, somehow caught in the crosshairs of gov't admin battles.
Posts Replies Media Videos
bergelsonlab.bsky.social
ppl often ask if I rec Baby Sign to support early language. my 1-liner is "probably doesn't do much, but it doesn't hurt"
@sblldtrch.bsky.social & co new paper supports my 1-liner doi.org/10.1177/0142...
[n.b. NOT the same as sign lang access for DHH kids i.e. full lg access from birth, which is A+]
abstract screenshot with last sentence highlighted: Abstract Baby sign programs aim to support communication between parents and preverbal infants by pairing signs with spoken words. This study addresses a notable gap in the literature by evaluating the large-scale impact of baby sign on both lexical development and caregiver behavior. Using an online questionnaire in 1348 typically developing, French hearing children (50% female) aged 10 to 28 months, 723 children exposed to baby sign at home were compared with 625 non-exposed peers. After controlling for socioeconomic status and parent–child activities, results revealed weak to no effect of baby sign on vocabulary development or caregiver behavior. These findings suggest that while baby sign is not detrimental to vocabulary development, it does not enhance spoken vocabulary as often claimed.
January 23, 2026 at 7:24 PM
bergelsonlab.bsky.social
no they do not, thanks for coming to my tedtalk.
new york times headline that says "Dogs Build Their Vocabulaies Like Toddlers"
January 8, 2026 at 9:55 PM
bergelsonlab.bsky.social
a little too on the nose there, door.
Revolving door with text saying “to conserve energy please use revolving door” and then a taped up laminated sign saying revolving door is permanently closed
November 26, 2025 at 3:23 PM
bergelsonlab.bsky.social
Swirly eye face emoji
November 17, 2025 at 7:03 PM
bergelsonlab.bsky.social
@laurelmack.bsky.social 🐦🐦, glad they called in the experts !
screencap from nyt "Mamdani has an ‘M’ next to a ‘D’, and that’s hard for English speakers,” said Professor Laurel MacKenzie, a co-director of the NYU Sociolinguistics Lab. “Our tongues are just not used to making that specific sequence of sounds.”"
October 22, 2025 at 9:01 PM
bergelsonlab.bsky.social
we once made a Tart Tatin(ic) #MyHeartWillGoOn
yours still looks delicious!
a heap of cooked dough and peaches in nothing even close to resembling the intended tart.
October 9, 2025 at 1:08 PM
bergelsonlab.bsky.social
I hate the use of AI instead of actual thinking more than most, but the idea that NIH could reliably & definitively tell that a grant had substantial AI input is, imho, naive in all but the most BLATANT cases (hallucinated ref's, leaving prompt in text). ask all the students using it.
New NIH Policy Warns Against AI Use and Limits Applications A new National Institutes of Health (NIH) policy (NOT-OD-25-132), effective for applications submitted to the September 25, 2025, receipt date and beyond, opens the door to research misconduct findings if an application is found to have been “substantially developed by AI” or to “contain sections substantially developed by AI,” as these are not considered to be the original ideas of the applicant. If the detection of AI is identified post award, the NIH may refer the matter to the Office of Research Integrity to determine whether there is research misconduct as well as take enforcement actions including disallowing costs, withholding future awards, wholly or in part suspending the grant, and termination.
September 25, 2025 at 7:04 PM
bergelsonlab.bsky.social
man looking annoyed with text 'humans screaming into the internet void when llms can't count rs: I feel bad for you) man looking belittling with text 'LLMs & those who profit off of them: i don't think about you at all)
August 8, 2025 at 1:59 PM
bergelsonlab.bsky.social
July 29, 2025 at 1:04 PM
bergelsonlab.bsky.social
recently found myself again wondering just how predatory MDPI journals are/n't. Came across this amazing discussion/analysis from veterinary medicine. #AcademicSky
avmajournals.avma.org/configurable...
screencap of following text (cuts for space) “articles published in traditional journals undergo rigorous critical review by qualified peer reviewers... & that a pub'd article has been deemed “acceptable” by peers... A previous analysis8 has demonstrated that, at least for published scientific articles, the time from submission to acceptance remained constant, w/ median of ~100 days. However, we IDed a new category of ostensibly peer-reviewed journals that review & process articles in <60 days, something that has been previously noted by other investigators.7 Furthermore, the 2 journals in this category (Animals and Veterinary Sciences, both published by MDPI) published more research articles in 2024 than any other journals except 1 (Frontiers in Veterinary Science with 1,789 articles), which had a median total time to pub of ~3 months. Indeed, the sum total of articles published by Animals and Veterinary Sciences (4,319) in 2024 exceeded the total published by the other 22 journals (excepting Frontiers in Veterinary Science) by some 100 articles. Assuming that the journals rejected approximately 50% of submitted manuscripts,9 either prior to review or after an initial review, and assuming that each article required 2 independent qualified reviewers, this would mean these 2 journals recruited approximately 17,000 reviewers in 2024 alone! Even if they accepted close to 100% of the articles submitted, this would still have required 8,600 reviewers. Given the difficulties expressed by... editors in finding qualified ...reviewers ... coupled with ... review times for most clinical vet journals ...did not change, we have to wonder what incredible strategy these 3 journals have IDed that allows such rapid, critical review. It behooves these journals to share their methods of attracting qualified critical reviewers in the diverse fields of veterinary medicine who can provide reviews within a few days of the invitation (allowing time for revisions, etc),...
July 7, 2025 at 7:11 PM
bergelsonlab.bsky.social
How dare you.
“Dear colleagues, I affirm you are well! The department of…” screencap from email
June 8, 2025 at 4:26 PM
bergelsonlab.bsky.social
idk, but they're white & gold, and spell out YANNY.
screen cap of a NYT article titled "Do you see craters or bumps on the moon's surface" with a picture of the moon that can be differently interpreted as having little bumps or craters on it, depending on the inference people make about whether light is coming from above vs. below.
June 5, 2025 at 3:47 PM
bergelsonlab.bsky.social
ooh i've had the opposite, and the main tell was they *left the prompt in the email*. It took all my adulting not to mention how much i appreciated their positive and friendly tone in my reply.
screencap from an email that says: "Sure, here's a more positive and friendly version: Hi Elika, Thank you so much for following up on this"
May 20, 2025 at 1:46 PM
bergelsonlab.bsky.social
Today we celebrated our srs & our work on the basic science of how babies learn language. Then I went to a Future of Fed Funding mtg ab how aggressive gov't interference is decimating our ability to do this work. We'll keep swimming but we'll need help #SaveOurScience #AcademicSky #NIH #NSF #Harvard
group of 17 smiling scientists in the sun
April 25, 2025 at 1:32 PM
bergelsonlab.bsky.social
Big #BragelsonLab hooray to Nicole Calderon for successfully defending her sr. thesis! Nicole looked at word form recognition bilingual infants. Such a delight to watch her ownership & mastery over this work grow! The work will keep going beyond this milestone, thnx to nicole's push! #CogSciSky
four smiling women delighted by science
April 18, 2025 at 5:42 PM
bergelsonlab.bsky.social
Isn't that just like an iguana, always just steamrolling ahead with their plans, never just stopping after a mile or two. Do less iguanas, do less. 🐦🐦
screenshot from NYT of a picture of an iguana and a headline "These Iguanas Got Carried Away and Ended Up 5,000 Miles From Home"
March 18, 2025 at 6:09 PM
bergelsonlab.bsky.social
Going from #AcademicSky 6/6
photos of a group of smiling lab members sitting on stairs
February 19, 2025 at 2:44 PM
bergelsonlab.bsky.social
Ooh love this, brb ~running~ to add it to my cognition course this spring.
I do have a bit of a soft spot for this illusion !
Picture of the back of a head with muller-lyer illusion shaved into the undercut
January 26, 2025 at 1:45 AM
bergelsonlab.bsky.social
#SmolSnowcreatures strike again
Truly tiny snowman on a stoop
January 20, 2025 at 1:42 PM
bergelsonlab.bsky.social
#SmolSnowcreatures are the most charming
tiny snowperson no higher than 1.5 steps smiling coyly with a small sand-shovel that comes up to their 'arm'
January 12, 2025 at 3:23 AM
bergelsonlab.bsky.social
hey look, even newspaper headlines are made exponentially more clear with pictures:)
nyt headline: "a dutch baby is the best breakfast" with a photo of a cast iron pan and something looking like a doughy filled pastry situation inside)
December 19, 2024 at 2:57 PM
bergelsonlab.bsky.social
TFW you're reading a response to reviews & you can tell that the authors think that another reviewer is you, but it's not, but you can't say so because you purposely obscured whether you were you in your own review for anonymity's sake #AcademicSky #PsychSciSky
the spiderman meme where there are 3 people in spiderman suits and each is pointing at the other two
December 4, 2024 at 5:06 AM
bergelsonlab.bsky.social
hi hi! we looked@parent report vs. home rec's to estimate first words, points, & steps. for 1st words, parents & experimenters did not differ in their estimate of when the first word happened overall but they also did not correlate; gets better with more words.
pubmed.ncbi.nlm.nih.gov/31094558/
screen grab of fig. 2 & 3 from linked article (Moore, Dailey et al, 2019, DevPsych). fig 2 shows 3 correlations, observed 1stpoint/walk/word vs. reported point/walk/1st word, correlation for point is okay-ish, for walk is very strong, and for word is ~noisy~!. fig 3 shows an increasingly precise correlation for observed vs. reported vocabulary size each month, 12-17month (it's a 6 panel graph with one set of dots & correlation/month)
November 27, 2024 at 3:14 PM
bergelsonlab.bsky.social
Timeline cleanse, happy friday from Peri #BirdSky
peregrine falcon on a dirty window ledge giving a piercing look
November 22, 2024 at 9:25 PM
bergelsonlab.bsky.social
Wow Gmail is really misreading all my emails about ways we can expand recruitment and outreach for infant participants for our research studies #PsychSciSkt #DevPsychSky
Screen cap of spam inbox from “hard sex.43” about meeting “local babes”
November 15, 2024 at 4:41 PM