Why This Language? Historical Symmetry Breaking through Cultural Transmission

Version 2.0 (5 September 2026) corrects Version 1.0 after an independent data verification: a shared adjusted-Rand-index implementation replaces local copies that scored identical singleton partitions as 0 (this changes several second-architecture partition verdicts, in both directions); the capacity–coverage test (K12) is reported for its registered subgroup, which does not replicate equivalence; the reconstruction model's persistence condition (T4) was reimplemented as registered and does not reproduce the neural result; one supplementary value, one figure statistic and all figure error bars were corrected. The core findings (K16, K17c, K13a, K15, K18, C4, A3′, A4′-target) are unchanged. All changes are listed in the Version 2 Revision Notes included in this record. Version 1.0 remains available as the previous version of this record. In iterated learning of emergent languages, cultural transmission can align semantic category systems and steer learners toward structured, efficient regions of language space. We show that limited transmitted evidence can also break symmetry among multiple structured solutions, biasing independent learners toward a shared, historically contingent reconstruction despite substantial item-level turnover. The target and strength of coordination are empirically distinguishable: developmental provenance can shift what learners reconstruct without necessarily changing how strongly they coordinate. In a controlled referential-game model, anchors captured from an immature snapshot of a parent's language steer descendants toward the immature organization; in the main GRU learner, this occurs without a detectable weakening of coordination under a pre-specified criterion, but at a cost to systematicity. We call this the Snapshot Effect. A second architecture and a minimal reconstruction model recover the same symmetry-breaking and provenance-target pattern. The local propagation mechanism is regime-dependent: in the compressed languages of the GRU learner, class-matched anchors preferentially stabilize untaught neighbouring meanings that use the same convention, whereas this mechanism does not carry over to a near-injective learner. Here, symmetry breaking refers to historical evidence resolving underdetermination among multiple viable structured reconstructions, not to competition among labels. Culture need not create structure. It makes one structured possibility historical. Working paper, version 2.0. 14 pages plus a 19-page supplement (full methods, every pre-registered outcome including failures, the corrected reconstruction model, figure legends) and 4-page revision notes. Code, registrations, results and the corrected analyses: https://doi.org/10.5281/zenodo.22305563 (all versions). Generative AI tools assisted with code, execution, review and editing; the author takes full responsibility.