Category-specific and system-wide preferences in competition: Evidence from noun phrase harmony Author(s): Holtz, Annie; Culbertson, Jennifer; Kirby, Simon | Abstract: Typological data show a tendency for languages to exhibit harmonic (i.e., consistent) ordering between heads and dependents. Pre...

🚨🚨🚨 New paper out by @anniesotropic.bsky.social, Jennifer Culbertson and @simonkirby.bsky.social: Category-specific and system-wide preferences in competition: Evidence from noun phrase harmony doi.org/10.5070/G601...

@sabinebluetgen.bsky.social, @simonkirby.bsky.social, Jennifer Culbertson: Do humans have an internal drive to share what is on our minds? Testing Mitteilungsbedürfnis. See evolangconf.github.io/2026/proceed...

Shira Tal, @kennysmithed.bsky.social , @inbalarnon.bsky.social , Jennifer Culbertson: Communicative efficiency develops with age. See evolangconf.github.io/2026/proceed...

@luciewolters.bsky.social, @simonkirby.bsky.social, @inbalarnon.bsky.social: Children's narrow learning bottleneck accelerates the emergence of the statistical properties of language. See evolangconf.github.io/2026/proceed...

Ponrawee Prasertsom , Andrea Silvi, Jennifer Culbertson, Devdatt Dubhashi, Moa Johansson, @kennysmithed.bsky.social: Regularity and processing complexity in recursive numeral systems. See evolangconf.github.io/2026/proceed...

@mandolinguist.bsky.social, Jennifer Culbertson, @simonkirby.bsky.social: Participants in a communication game balance production effort and comprehension accuracy (but not always optimally). See evolangconf.github.io/2026/proceed...

@maisyhallam.bsky.social, @simonkirby.bsky.social, Fiona M. Jordan, @kennysmithed.bsky.social: Individual preferences during classification align with kinship semantics across languages. See evolangconf.github.io/2026/proceed...

One week until #Evolang2026 starts in Plovdiv! Here are some of our talks: 🧵

Silvi, Prasertsom, Culbertson, Dubhashi, Johansson, Smith: Evaluating the relationship between regularity and learnability in recursive numeral systems using Reinforcement Learning https://arxiv.org/abs/2602.21720 https://arxiv.org/pdf/2602.21720 https://arxiv.org/html/2602.21720

Ponrawee Prasertsom, Andrea Silvi, Jennifer Culbertson, Moa Johansson, Devdatt Dubhashi, Kenny Smith: Recursive numeral systems are highly regular and easy to process https://arxiv.org/abs/2510.27049 https://arxiv.org/pdf/2510.27049 https://arxiv.org/html/2510.27049

Cultural Transmission Promotes the Emergence of Statistical Properties That Support Language Learning Language is passed across generations through cultural transmission. Prior experimental work, where participants reproduced sets of non-linguistic sequences in transmission chains, shows that this pr...

New paper with @inbalarnon.bsky.social and @simonkirby.bsky.social! Learnability pressures drive the emergence of core statistical properties of language–e.g. Zipf's laws–in an iterated sequence learning experiment, with learners’ RTs indicating sensitivity to the emerging sequence information.

I'll be presenting this work at the @uoe-cle.bsky.social in Edinburgh on the 22nd January. I look forwards to robust discussion!

The New episode of PPLS Perspectives is available now!

Maisy Hallam chats with Prof. Jenny Culbertson about how #technology, #AI, and global change are reshaping the future of language — and why #linguistic diversity matters more than ever.

Listen now: edin.ac/4r7axdA

The lexicon adapts to competing communicative pressures: Explaining patterns of word similarity Cross-linguistically, lexicons tend to be more phonetically clustered than required by the phonotactics of the language; that is, words within a langu…

⚠️ New paper! Why do words sound so similar? In an agent-based model + communication game, we show that production/comprehension pressures trade off to shape lexicon structure.

In @cognitionjournal.bsky.social w/ @simonkirby.bsky.social & Jenny Culbertson.

www.sciencedirect.com/science/arti...

Exploiting individual differences to bootstrap communication Author summary A fundamental question in the field of language evolution is how a communication system can emerge in a population that is initially unable to communicate. One theory is that human comm...

Interacting agents can create shared communication systems without ever knowing if signals are successfully received. The conventionalisation process involves a curious interaction between joint attention and individual behavioural differences. doi.org/10.1371/jour.... Work with Casimir Fisch.

PPLS Linguistics Professor talks of the importance of understanding language structure and evolution in the age of AI | School of Philosophy, Psychology & Language Sciences | School of Philosophy, Psy... Last month Personal Chair of Experimental Linguistics Prof Jennifer Culbertson spoke at the Geneva Science and Diplomacy Anticipator (GESDA) Summit on Science of Language and Communication and how new...

How will language shape the future of AI, education & society? At the GESDA Global Summit last month, Prof Jennifer Culbertson explored why understanding language structure & evolution is vital in the age of AI. Read more and watch her talk: edin.ac/3X8WDJY
#Linguistics #AI

Predictive Structure Emerges During the Generalisation of Kin Terms to New Referents Abstract. Despite cross-linguistic diversity in how kin relations map to terminology, there are constraints on which kin may be categorised together. But what are the constraints on kin term variation...

New paper in @openmindjournal.bsky.social with @simonkirby.bsky.social, @kennysmithed.bsky.social and Fiona Jordan! Kinship terms overwhelmingly exhibit predictive structure - terms in one part of the system help us predict other terms - a pattern which emerges because it helps us generalise better.

Late bilinguals whose L1 allows subject omission tend to be more explicit in reference: they use more overt pronouns and full NPs than null pronouns? Why is that? We test two hypotheses (a desire to avoid ambiguity or ease processing) using online eye-tracking. (16/)

Yajun Liu, Antonella Sorace, @kennysmithed.bsky.social: Do bilinguals avoid ambiguity? An experimental study of lexical ambiguity in spoken Mandarin (Poster Session 2, P2-L-114, 1 August @ 10:30). Poster here: tinyurl.com/mr288rn8 blurb below 👇 (15/)

Bilingual experience may influence native speakers’ choice of referring expressions. We explore the impact of Crosslinguistic influence on reference production when both languages permit subject drop yet differ typologically. (14/)

Yajun Liu, Antonella Sorace, @kennysmithed.bsky.social: Beyond Crosslinguistic Influence: Mandarin Speakers with Exposure to Null-subject Languages Nonetheless Use Fewer Null Pronouns in Mandarin (Talk 12: Lang 2, 31 Jul @ 14:59). Blurb below 👇 (13/)

Find out more from @psydelle.bsky.social & @ivegner.bsky.social's poster: July 31 @conll-conf.bsky.social Paper: edin.ac/4kdIzYP (2/2)

Semi-compositional language like “kick the habit” takes longer for humans to process than both fully-compositional AND fully-idiomatic units! Also, a memory model with sBERT embeddings fails more on the same items. What's so special about these units? (1/2) 👇

Learn more at @marthaflinders.bsky.social @psydelle.bsky.social @ivegner.bsky.social's talk+panel @aclmeeting.bsky.social Jul 30, 9am Hall A! tinyurl.com/3bb3umhc (2/2)

Different ML datasets use wildly different definitions of “systematic generalization”. To REALLY tell how well a model generalizes, we must look for systematicity of the model's representations. (1/2) 👇

🧠 Check out our #CogSci2025 poster examining why languages avoid words like “peeb” or “gok”?
Our study found that people struggle to recognise words where consonants share the same place of articulation. This processing cost could contribute to why such words are rare across languages 🌏 (12/)

Annie Holtz @anniesotropic.bsky.social, @kennysmithed.bsky.social, Mitsuhiko Ota: The Impact of Similar Place Avoidance on Novel Word Learning in Adults (Virtual Poster Session 1, V1-H-87, 31 Jul @ 12:00). Poster here: tinyurl.com/bdf9zrsc blurb below 👇 (11/)

RSA succeeds in explaining phenomena in language use, but it oddly predicts a preference for ambiguous costly words in learning. We test this surprising prediction using artificial word learning and lexicon rating tasks. (10/)

Ponrawee Prasertsom, @kennysmithed.bsky.social, Jennifer Culbertson: Testing counterintuitive predictions about cost-based inferences in learning from the Rational Speech Act model (Poster Session 1; P1-P-159, 31 July @ 14:15). Poster here: tinyurl.com/3uxd7krp blurb below 👇 (9/)

Languages differ dramatically in how they categorise family members - what drives this semantic variation? For 1215 languages, our computational model reveals a hierarchy of semantic features for kinship, suggesting that kin term semantics are optimised for efficient communication. (8/)