aiqbench analysis
ChronoLens introduces a novel framework using frozen multilingual models and sparse representations to quantify historical language evolution across morphology, syntax, semantics, and pragmatics over nearly two centuries of parliamentary records. By aligning feature spaces from incompatible linguistic levels, the system reveals that while languages change at comparable magnitudes within specific domains, their trajectories differ significantly in direction and timing. This approach resolves a key limitation in computational linguistics where dense embeddings failed to capture structured multidimensional shifts across diverse historical periods.
ChronoLens: Measuring Language Change Across Time, Languages, and Linguistic Levels
arXiv:2608.03507v1 Announce Type: new Abstract: Historical language change affects morphology, syntax, semantics, and pragmatics, yet computational studies typically examine these levels with incompatible representations and therefore cannot determine whether they evolve together across languages. We address this problem by asking how the magnitude and direction of change vary across linguistic levels, languages, and historical periods within a single analytical space. We introduce ChronoLens, a framework that combines frozen multilingual language models, feature-aligned crosscoders, and post-hoc linguistic interventions, and apply it to 44.98 million documents and approximately 17.2 billion tokens from five parliamentary traditions spanning 1803--2026. The resulting sparse representations agree substantially more strongly with linguistic statistics than dense embeddings or a pooled sparse autoencoder ($\rho=0.72$ versus $0.29$ and $0.28$), and reveal that morphology, syntax, semantics, and pragmatics generally change by comparable amounts within a language, while languages differ markedly in when, how far, and in which direction they change. These findings show that historical language change is a structured, multidimensional process: similar magnitudes can conceal different trajectories, and meaningful cross-linguistic comparison requires measuring both distance and direction.
Abstract:Historical language change affects morphology, syntax, semantics, and pragmatics, yet computational studies typically examine these levels with incompatible representations and therefore cannot determine whether they evolve together across languages. We address this problem by asking how the magnitude and direction of change vary across linguistic levels, languages, and historical periods within a single analytical space. We introduce ChronoLens, a framework that combines frozen multilingual language models, feature-aligned crosscoders, and post-hoc linguistic interventions, and apply it to 44.98 million documents and approximately 17.2 billion tokens from five parliamentary traditions spanning 1803--2026. The resulting sparse representations agree substantially more strongly with linguistic statistics than dense embeddings or a pooled sparse autoencoder ($\rho=0.72$ versus $0.29$ and $0.28$), and reveal that morphology, syntax, semantics, and pragmatics generally change by comparable amounts within a language, while languages differ markedly in when, how far, and in which direction they change. These findings show that historical language change is a structured, multidimensional process: similar magnitudes can conceal different trajectories, and meaningful cross-linguistic comparison requires measuring both distance and direction.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2608.03507 [cs.CL]
(or arXiv:2608.03507v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2608.03507 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 270 words
Related coverage
Not Deciphering The Voynich Manuscript
LessWrong · Oct 11, 2026
Models Can't Remember Their Training. Neither Can You.
LessWrong · Jul 19, 2026
Models don’t seem to be dishonest in the way humans are
LessWrong · Jul 22, 2026
Newer Models, Same Advantage
Hugging Face · Jul 16, 2026