Newfangled words aren’t just for the young!
Read the full paper “Semantic Representations Are Updated Across the Lifespan Reflecting Diachronic Language Change” at: doi.org/10.1162/opmi...
The new norms we collected and the full analysis pipeline are available through OSF: osf.io/q7j9n/overvi...
This suggests that meaning representations are continuously updated, regardless of age.
Even though older adults have experienced previous decades with older meanings, that history doesn’t weigh them down. We all rapidly adapt to the modern usage of language to coordinate with our community.
Taking a closer look at the relatedness ratings:
When judging “changed” words, both younger and older adults rated the 1990s neighbors as more related than the 1950s neighbors. Crucially, ratings were similar across groups: older adults didn’t show a preference for meanings they learned early on.
Verdict? We found that older adults do not look like time capsules of past meanings in language.
In both studies, older adults’ semantic representations were best predicted by recent usage patterns (1990s), not the ones from their youth.
If the semantic structure of language is constantly shifting, what do older adults do?
If older adults retain earlier meanings, their semantic spaces should be more similar to when they were younger. If they update continuously, their spaces should look like the 1990s/2000s (like younger adults).
First, we verified that the "semantic structure" of language shifts enough within a lifespan to matter.
Using historical word embeddings, we found that English semantic structures measurably changed over the 20th century. Word relations in the 1990s are much more like the 1980s than the 1900s.
We explored this in two ways: 1) comparing meanings from historical word embeddings to word associations from people aged 20-90 y.o. 2) collecting relatedness ratings from 1,300+ adults (younger and older) for words that have and have not changed in meaning (e.g., “broadcast”).
Tl;dr: Even though word meanings in language change over time, individuals across different age groups keep their representations up-to-date!
Check it out here: doi.org/10.1162/opmi...
New paper w/ @ryskin.bsky.social in Open Mind!
Words change: “broadcast” once meant scattering seeds; “tweet” was just a bird sound. Do older adults keep earlier meanings, or update as language evolves?
Our new paper investigates how semantic representations differ across age groups. 🧵👇
We’ve recently updated our collaborative open-access book, “Neural Networks in Cognitive Science”, adding a few new authors, chapters, and lots of content.
downloads.jeffyoshimi.net/NeuralNetwor...
Looking forward to seeing everyone at #CogSci2025 this week! Come check out what we’ve been working on in the LInC Lab, along with our fantastic collaborators!
Paper 🔗 in 🧵👇
We found that there were common scene compositions that accompanied naming events, parent naming followed a Zipfian distribution, and the cross-situational statistics were comparable to lab experiments. See our paper for more details!
New paper w/ @ryskin.bsky.social and Chen Yu: We analyzed parent-child toy play and found that cross-situational learning statistics were present in naturalistic settings!
onlinelibrary.wiley.com/doi/epdf/10....
I'll be presenting some work with @tylermarghetis.bsky.social and Elizabeth Geballe at #CES2024 on translation as transmission. Stop by if you'd like to chat about digital humanities and cross-cultural transmission!
Link: ellisc.dev/pdf/ces_2024...
I’ll be presenting work I did with @ryskin.bsky.social in the CMCL workshop at #ACL2024 (8/15 at 14:20 in Lotus Suite 4). Come check it out, and please find me if you’d like to discuss!
aclanthology.org/2024.cmcl-1.19
The preprint includes an OSF link to behavioral norms from relatedness judgments for words that have and have not changed over time, collected from two age groups (18-33 y.o. and 63-92 y.o.)!
Excited to share a new preprint with @ryskin.bsky.social: Individuals continually update their meaning representations to match those of the current time period, regardless of their age!
Preprint can be found at: ssrn.com/abstract=488...
