pith. sign in

arxiv: 2606.12748 · v1 · pith:H3RFQ7P2new · submitted 2026-06-10 · 💻 cs.CL

Agent-based models for the evolution of morphological alternation patterns

Aravinth Kulanthaivelu , Richard Sproat This is my paper

Pith reviewed 2026-06-27 09:26 UTC · model grok-4.3

classification 💻 cs.CL
keywords agent-based simulationmorphological alternationlanguage evolutionsocial networkslarge language modelshistorical linguistics
0
0 comments X

The pith

Multi-agent simulations show scale-free networks and Bernoulli adoption produce more plausible morphologies

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops multi-agent simulations in which alternate word forms spread through populations via phonological changes or lexical alternatives, leading to entrenched stem and inflectional alternations. Agents adopt novel forms with varying probabilities and can extend them to related paradigm slots, using realistic lexicons and phonological rules on populations of tens to hundreds of agents. To assess realism, the authors introduce the AI Historical Linguist, an LLM system that simulates debates between two historical linguists to score real, disguised, and simulated morphologies. Results indicate that scale-free social networks combined with random Bernoulli adoption yield morphologies judged more plausible than those from other topologies or policies. The work also runs case studies on attested historical changes to explore alternative outcomes.

Core claim

In these simulations, scale-free social networks and random Bernoulli adoption of forms generate morphologies that the AI Historical Linguist rates as more plausible than those arising under other network structures or adoption rules, when benchmarked against real language morphologies and disguised controls.

What carries the argument

Multi-agent simulation supporting naturalistic lexical forms, phonological rules, multiple network topologies, diffusion patterns, and adoption policies, evaluated by the AI Historical Linguist LLM debate system.

If this is right

  • Scale-free networks favor more plausible morphologies than other topologies.
  • Random Bernoulli adoption favors more plausible morphologies than other policies.
  • The simulation framework can reproduce specific attested historical alternations.
  • Altering network or adoption parameters allows testing of counterfactual historical scenarios.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Social network structure may influence which irregular forms persist over time in actual languages.
  • LLM debate evaluation could be applied to test plausibility of other simulated language changes.
  • Random adoption policies might better capture diffusion processes observed in real populations.

Load-bearing premise

The AI Historical Linguist system gives a valid and unbiased measure of morphological plausibility through simulated linguist debates.

What would settle it

If the AI Historical Linguist assigns equivalent or lower plausibility scores to morphologies from scale-free networks with Bernoulli adoption than to those from other conditions when compared against real morphologies.

Figures

Figures reproduced from arXiv: 2606.12748 by Aravinth Kulanthaivelu, Richard Sproat.

Figure 1
Figure 1. Figure 1: Schematic of within-step adoption. During one communication round, an agent hears forms from neighboring agents and may replace its current entry for the relevant lexeme-slot cell. In (a) the main adoption mechanism is Bernoulli adoption, optionally augmented with mild biasing terms such as novelty weighting, while (b) showcases the stem-as-target setting, which replaces not only the specific slot but all … view at source ↗
Figure 2
Figure 2. Figure 2: One simulation time step. Optional exogenous events are applied first, after which agents communicate on the graph and adopt heard forms. Compression is then applied synchronously, and the resulting state is recorded for later analysis. Each time step proceeds as follows: 1. Optional exogenous events are applied. These may alter the graph or change the forms held by a subset of agents, allowing us to model… view at source ↗
Figure 3
Figure 3. Figure 3: Rate of suppletion, number of stem alternation signatures, and number of paradigms for the distinct experiment configurations at the terminal state. June 10, 2026 17/51 [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Signature-count summaries for the population-splitting intervention. The fixed population is divided into non-interacting components, changing the global interaction structure rather than only the local transmission parameters [PITH_FULL_IMAGE:figures/full_fig_p018_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Signature-count summaries for the cascade-targeting intervention. Cascades are applied to the whole population or to subsets selected at random or by connectivity degree. June 10, 2026 18/51 [PITH_FULL_IMAGE:figures/full_fig_p018_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Heatmap displaying the mean normalized impact relative to the baseline of each factor for various summary statistics. June 10, 2026 19/51 [PITH_FULL_IMAGE:figures/full_fig_p019_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Sample prompt for the AI Historical Linguist with an artificially evolved language. paradigms having only one example shown. • Patterns of stem alternation: The stem alternations present serious issues: – There’s a consistent pattern of accent shift (single to double accent marks: ’ to ”) in positions 4-5 across most paradigms, which is phonologically plausible – However, the extreme suppletion is highly p… view at source ↗
Figure 8
Figure 8. Figure 8: Top panel: Bar plot showing the mean scores assigned by the assessor and critic over ten runs of the AI Historical Linguist. Here the assessor is Gemini-2.5-Pro and the critic is Claude-Sonnet-4-20250514. The morphologies evaluated are for Estonian, Finnish, French, Hungarian, Italian and Latin in both real and disguised versions; and six sample artificial languages from the set evaluated more fully in the… view at source ↗
Figure 9
Figure 9. Figure 9: Relationship between fitted rank-frequency slope and AIHL plausibility for the main AIHL-scored experiment sweep. Each point represents an experimental condition, colored by the parameter family varied relative to the baseline. The horizontal axis gives the fitted power-law exponent for the terminal stem-alternation signature distribution, while the vertical axis gives the mean AIHL plausibility score. The… view at source ↗
Figure 10
Figure 10. Figure 10: AIHL plausibility scores by experimental condition. Bars show mean final score across AIHL evaluations, error bars show standard deviation, and points show individual runs. Higher scores indicate greater judged morphological plausibility. June 10, 2026 29/51 [PITH_FULL_IMAGE:figures/full_fig_p029_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: AIHL plausibility scores by experimental condition, ordered by decreasing AIHL score. Bars show mean final score across AIHL evaluations, error bars show standard deviation, and points show individual runs. Higher scores indicate greater judged morphological plausibility. June 10, 2026 30/51 [PITH_FULL_IMAGE:figures/full_fig_p030_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: AIHL plausibility scores for individual components in the population-split simulations. Bars show mean final score across AIHL evaluations, error bars show standard deviation, and points show individual runs. June 10, 2026 31/51 [PITH_FULL_IMAGE:figures/full_fig_p031_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: AIHL plausibility scores for cascade-targeting conditions. Bars show mean final score across AIHL evaluations, error bars show standard deviation, and points show individual runs. June 10, 2026 32/51 [PITH_FULL_IMAGE:figures/full_fig_p032_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Numbers of lexemes in a given category and language that show suppletive stem alternations from the Surrey Suppletion Database [225]. ‘receive’. So systems with ten or more lexemes in a given category showing suppletion are not unheard of, though relatively rare. Honorific systems are particularly ripe for lexical substitutions. Javanese, for example, is famous for its multiple speech levels. Alongside th… view at source ↗
Figure 15
Figure 15. Figure 15: The three adoption methods applied to a common inbox for a single focal cell. Bernoulli treats each received token as an independent coin flip and may leave the cell unchanged; degree-scaled biases adoption toward high-degree senders; cumulative aggregates per-form counts across the inbox and deterministically adopts the most heard form. June 10, 2026 2/37 [PITH_FULL_IMAGE:figures/full_fig_p053_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: The three adoption-target modes applied to the same heard token. The default mode updates only the heard lexeme-slot cell; stem-as-target additionally propagates the heard stem across every slot of the target lexeme’s paradigm; affix-as-target propagates the heard affix to every cell in the lexicon that previously carried the listener’s old affix. June 10, 2026 3/37 [PITH_FULL_IMAGE:figures/full_fig_p054… view at source ↗
Figure 17
Figure 17. Figure 17: The three compression policies applied to a shared population of seven stem-alternation signatures. No compression leaves the signatures untouched; class consensus snaps minority signatures within an alternant-set class to the modal pattern, preserving the alternant count; global drift snaps each signature to a dominant nearby one and can drop alternants altogether. June 10, 2026 5/37 [PITH_FULL_IMAGE:fi… view at source ↗
Figure 18
Figure 18. Figure 18: Decision-tree compression illustrated on a Korean nominative example. A classifier is fit to (environment, affix) pairs in the agent’s lexicon and applied back, learning that vowel-final stems take -ka and consonant-final stems take -i, and rewriting the outlier accordingly. June 10, 2026 6/37 [PITH_FULL_IMAGE:figures/full_fig_p057_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Representative social graph topologies used in the model: complete, Erdős–Rényi, scale-free, and log-normal. The complete graph maximizes connectivity; the Erdős–Rényi graph provides a homogeneous random baseline; the scale-free graph introduces hubs through preferential attachment; and the log-normal graph produces heterogeneous degree structure without the same hub dominance. Social graphs The social st… view at source ↗
Figure 20
Figure 20. Figure 20: Example of a split and merge operation applied successively to a single population. Under a split instance the population is split into a set of connected components, the number of which is specified by the user. Under a merge operation disjoint components are merged by adding edges at random between the components. June 10, 2026 10/37 [PITH_FULL_IMAGE:figures/full_fig_p061_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Birth/death turnover under the two persistence regimes. In the persistent regime each learner inherits its mentor’s social position and the graph topology is preserved across generations; in the non-persistent regime the edges within each connected component are re-sampled, shown here by new solid edges in the post-turnover graph and the previous edges drawn as faded dashed lines. June 10, 2026 11/37 [PI… view at source ↗
Figure 22
Figure 22. Figure 22: Signature histogram figures for the rule-based and LLM-based segmentation outputs. June 10, 2026 18/37 [PITH_FULL_IMAGE:figures/full_fig_p069_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: A run of the simulation of the change from the nominative marker -i to -ka after vowels in early Modern Korean. The population has 50 agents, exchanging 400 utterances at each tick. The probability of adoption is set to 0.1, and the probability of the initial change from -i to -ka after nouns ending in /i/ is also 0.1. The novelty parameter is 0.6. shows decision trees at two stages, the first from shortl… view at source ↗
Figure 24
Figure 24. Figure 24: Decision trees for an agent at step 420 (left) and 820 (right) for the simulation presented in [PITH_FULL_IMAGE:figures/full_fig_p072_24.png] view at source ↗
Figure 25
Figure 25. Figure 25: A run of the simulation of the change from the nominative marker -i to -ka after vowels in early Modern Korean. The population has 50 agents, exchanging 400 utterances at each tick. The probability of adoption is set to 0.1, and the probability of the initial change from -i to -ka after nouns ending in /i/ is also 0.1. Here the novelty parameter is 0.2, and rather than spreading the system collapses. June… view at source ↗
Figure 26
Figure 26. Figure 26: Evolution of forms of ambulāre ‘go’, barrıre ‘trumpet’, disquırıre ‘discourse’, farcıre ‘force’, flōrıre ‘flower’. Each row corresponds to the evolution of the prevalence of a particular pattern for the verb. The left panel represents the period prior to the dialect split, and the right two panels represent the ‘French’ and ‘Italian’ branches respectively. The curve corresponding to the historically corre… view at source ↗
Figure 27
Figure 27. Figure 27: Continuation of [PITH_FULL_IMAGE:figures/full_fig_p079_27.png] view at source ↗
Figure 28
Figure 28. Figure 28: Continuation of [PITH_FULL_IMAGE:figures/full_fig_p080_28.png] view at source ↗
Figure 29
Figure 29. Figure 29: Evolution of Lenition from Proto-Celtic to the two Insular Celtic branches for the case where the examples are balanced to roughly represent the proportion of vowel and consonant stems in Proto-Celtic. The number of agents is 50. Probability of adoption is 0.2, novelty is 0.6, and probability of mutation is 0.2. Mutation rules were introduced to the 20% of agents with the highest degrees of connectivity. … view at source ↗
Figure 30
Figure 30. Figure 30: Evolution of Lenition from Proto-Celtic to the two Insular Celtic branches for the case where the examples are biased towards having as many consonant stems as vowel stems. The same settings were used as in [PITH_FULL_IMAGE:figures/full_fig_p086_30.png] view at source ↗
Figure 31
Figure 31. Figure 31: Decision tree for one of the Goidelic-speaking agents at tick 1000 for the balanced high-degree case. For masculine-singular and feminine-plural nouns, the tree predicts no lenition, whereas the other two classes induce lenition. June 10, 2026 36/37 [PITH_FULL_IMAGE:figures/full_fig_p087_31.png] view at source ↗
read the original abstract

Why is the past of English "go" the apparently unrelated "went"? Such alternations are frequent in languages. They neither aid communication nor learnability, yet they can be persistent, surviving over centuries or millennia. We present a multi-agent simulation of the emergence of morphological stem and inflection alternations. Alternate forms arise by phonological changes or, as with "go/went", from lexical alternatives associated with a subset of the population. When an agent 'hears' another agent use a novel form for a slot in the paradigm of a word (say, the past tense of go), they will with some probability adopt that form, possibly spreading its use to other slots in the paradigm that shared the same original form. Thus alternative forms can spread through the population and become entrenched as stem or inflectional marker alternants. Unlike many previous computational studies, our system allows for naturalistic lexical forms, realistic phonological rules, lexicons with hundreds or thousands of entries, and agent populations in the tens or hundreds. It supports several network topologies, diffusion patterns and agent adoption policies. One issue with such simulations is evaluation: how realistic is the resulting morphology compared to those of real languages? We introduce the AI Historical Linguist, a novel Large Language Model-driven system that models a debate between two historical linguists. We use this to compare a set of real language morphologies, disguised morphologies, and experimentally evolved morphologies. The results suggest that among the factors that favor more plausible morphologies are scale-free social networks and random Bernoulli adoption of forms. We also present three case studies modeling attested historical changes, allowing us to test what might have happened if history had been different. All code and data are released.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper presents a multi-agent simulation of morphological alternation emergence, incorporating phonological changes and lexical alternatives that spread via adoption probabilities across agent populations and network topologies. It introduces the AI Historical Linguist—an LLM system simulating debates between historical linguists—to evaluate plausibility of simulated morphologies against real and disguised language data. Results indicate scale-free networks and Bernoulli adoption yield higher plausibility scores; three case studies of attested changes are included, with code and data released.

Significance. If the evaluation holds, the work provides a scalable framework for testing social and diffusion factors in morphological evolution, with strengths in handling realistic lexicons, phonology, and large populations. The open release of code and data supports reproducibility, and the counterfactual case studies offer a falsifiable modeling approach. The LLM-based judge is a novel but unvalidated component.

major comments (1)
  1. [Evaluation section / abstract] The central claim that scale-free networks and Bernoulli adoption produce more plausible morphologies (abstract) depends entirely on plausibility rankings from the AI Historical Linguist. No validation is reported—such as correlation with human expert ratings, inter-annotator agreement, prompt ablations, or checks for training-data leakage—which is load-bearing because systematic LLM biases (e.g., favoring attested alternation types) could invert the reported ranking of network topologies and adoption rules.
minor comments (1)
  1. [Abstract] The abstract refers to 'disguised morphologies' without defining the disguise procedure or how they control for surface similarity, which would clarify the evaluation baseline.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback, particularly on the need to validate the AI Historical Linguist. We agree this is essential for supporting the central claims and will revise accordingly.

read point-by-point responses

  1. Referee: [Evaluation section / abstract] The central claim that scale-free networks and Bernoulli adoption produce more plausible morphologies (abstract) depends entirely on plausibility rankings from the AI Historical Linguist. No validation is reported—such as correlation with human expert ratings, inter-annotator agreement, prompt ablations, or checks for training-data leakage—which is load-bearing because systematic LLM biases (e.g., favoring attested alternation types) could invert the reported ranking of network topologies and adoption rules.

    Authors: We agree that the absence of validation for the AI Historical Linguist represents a genuine limitation, as unexamined biases could indeed affect the topology and adoption rule rankings. In revision we will add prompt ablations testing debate format variations, training-data leakage checks via held-out languages and rare alternation patterns, and inter-annotator agreement metrics across repeated LLM debates. We will also include a small-scale human expert rating study on a subset of outputs to provide initial correlation evidence. These steps directly address the load-bearing concern without overclaiming the current results. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained against external benchmarks

full rationale

The paper simulates morphological evolution via agent-based rules on networks and adoption policies, then evaluates outputs by comparing LLM-assigned plausibility scores against real-language morphologies and disguised controls. No quoted equations, parameters, or steps reduce the reported ranking of network topologies or adoption rules to a fit, self-definition, or self-citation chain. The AI Historical Linguist is introduced as an external judge rather than derived from the simulation results themselves. Evaluation therefore rests on independent real-language data rather than internal construction.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Ledger constructed from abstract only; full implementation details unavailable.

free parameters (1)
  • adoption probability
    Agents adopt novel forms 'with some probability'; this parameter controls diffusion but no specific value or fitting procedure is given in the abstract.
axioms (1)
  • domain assumption When an agent hears a novel form for a paradigm slot, it may adopt it and spread the form to other slots sharing the original stem.
    This mechanism is presented as the core process allowing alternations to entrench.

pith-pipeline@v0.9.1-grok · 5839 in / 1251 out tokens · 26929 ms · 2026-06-27T09:26:21.943421+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

251 extracted references · 62 canonical work pages

  1. [1]

    Computational Morphology

    Kirov C, Sproat R. Computational Morphology. In: Bernardy JP , Blanck R, Chatzikyriakidis S, Lappin S, Maskharashvili A, editors. Probabilistic Approaches to Linguistic Theory. Stanford, CA: CSLI Publications; 2023

    2023

  2. [2]

    Historical Syntax and Synchronic Morphology: an Archaeologist’s Field Trip

    Givón T . Historical Syntax and Synchronic Morphology: an Archaeologist’s Field Trip. In: Chicago Linguistic Society. vol. 7; 1971. p. 394-415

    1971

  3. [3]

    Human Languages Trade off Complexity against Efficiency

    Koplenig A, Wolfer S, Rüdiger JO, Meyer P . Human Languages Trade off Complexity against Efficiency. PLOS Complex Systems. 2025 02;2(2):1-42. Available from: https://doi.org/10.1371/journal.pcsy.0000032. doi:10.1371/journal.pcsy.0000032

    work page doi:10.1371/journal.pcsy.0000032 2025

  4. [4]

    A Mathematical Theory of Communication

    Shannon CE. A Mathematical Theory of Communication. The Bell System T echnical Journal. 1948;27(3):379-423

    1948

  5. [5]

    Prediction and Entropy of Printed English

    Shannon CE. Prediction and Entropy of Printed English. The Bell System T echnical Journal. 1951;30(1):50-64

    1951

  6. [6]

    Human Behavior and the Principle of Least Effort

    Zipf GK. Human Behavior and the Principle of Least Effort. Addison-Wesley; 1949

    1949

  7. [7]

    From Pure Phonology to Pure Morphology the Reshaping of the Romance Verb

    Maiden M. From Pure Phonology to Pure Morphology the Reshaping of the Romance Verb. Recherches linguistiques de Vincennes. 2009;38:45-82

    2009

  8. [8]

    The Intramorphological Meanings of Thematic Vowels in Italian Verbs

    Da T os M. The Intramorphological Meanings of Thematic Vowels in Italian Verbs. Università degli Studi di Padova; 2012

    2012

  9. [9]

    Morphome Death and Transfiguration in the History of French

    Esher L. Morphome Death and Transfiguration in the History of French. Journal of Linguistics. 2017;53(1):51-84

    2017

  10. [10]

    Romance Root Suppletion and Cumulative Exponence: Fusion, Pruning, Spanning

    Pomino N, Remberger EM. Romance Root Suppletion and Cumulative Exponence: Fusion, Pruning, Spanning. Languages. 2022;(7):161-85

    2022

  11. [11]

    A Morphological Typology of Non-Root Alternations: Invasion, Suppletion, and Allomorphy

    Kühlert NT . A Morphological Typology of Non-Root Alternations: Invasion, Suppletion, and Allomorphy. Harvard University; 2023

    2023

  12. [12]

    On Suppletion

    Mel´čuk I. On Suppletion. Linguistics. 1976;17:45-90. June 10, 2026 37/51 Preprint

    1976

  13. [13]

    Morphology by Itself: Stems and Inflectional Classes

    Aronoff M. Morphology by Itself: Stems and Inflectional Classes. No. 22 in Linguistic Inquiry Monographs. Cambridge, MA: MIT Press; 1994

    1994

  14. [14]

    Suppletion

    Carstairs-McCarthy A. Suppletion. In: Asher RE, editor. Encyclopedia of Language and Linguistics. vol. 8. Oxford: Pergamon; 1994. p. 4410-1

    1994

  15. [15]

    Regular morphology and the lexicon

    Bybee J. Regular morphology and the lexicon. Language and Cognitive Processes. 1995;(5):425-55

    1995

  16. [16]

    Suppletion: Frequency, Categories and Distribution of Stems

    Hippisley A, Chumakina M, Corbett GG, Brown D. Suppletion: Frequency, Categories and Distribution of Stems. Studies in Language. 2004;28(2):387-418

    2004

  17. [17]

    A Historical Consideration of the Irregular Verb Go

    Koike K. A Historical Consideration of the Irregular Verb Go. The Journal of J F Oberlin University Studies in Language and Culture. 2019;(11):1-14

    2019

  18. [18]

    On Rules of Referral

    Stump G. On Rules of Referral. Language. 1993;69(3):449-79

    1993

  19. [19]

    Pynini: a Python Library for Weighted Finite-state Grammar Compilation

    Gorman K. Pynini: a Python Library for Weighted Finite-state Grammar Compilation. In: Proceedings of the SIGFSM Workshop on Statistical NLP and Weighted Automata; 2016. p. 75-80

    2016

  20. [20]

    Finite-State T ext Processing

    Gorman K, Sproat R. Finite-State T ext Processing. Cham: Springer; 2021

    2021

  21. [21]

    Exploring Network Structure, Dynamics, and Function Using NetworkX

    Aric A Hagberg DAS, Swart PJ. Exploring Network Structure, Dynamics, and Function Using NetworkX. In: Gäel Varoquaux TV, Millman J, editors. Proceedings of the 7th Python in Science Conference (SciPy2008). Pasadena; 2008. p. 11-5

    2008

  22. [22]

    Linguistic Capacity was Present in the Homo sapiens Population 135 Thousand Years Ago

    Miyagawa S, DeSalle R, Nóbrega VA, Nitschke R, Okumura M, T attersall I. Linguistic Capacity was Present in the Homo sapiens Population 135 Thousand Years Ago. Frontiers in Psychology. 2025;Volume 16. Available from: https: //www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2025.1503900. doi:10.3389/fpsyg.2025.1503900

    work page doi:10.3389/fpsyg.2025.1503900 2025

  23. [23]

    The Evolution of Language: T owards Gestural Hypotheses

    Żywiczyński P , Wacewicz S. The Evolution of Language: T owards Gestural Hypotheses. Berlin: Peter Lang; 2019

    2019

  24. [24]

    Principles of Historical Linguistics

    Hock HH. Principles of Historical Linguistics. 3rd ed. De Gruyter Mouton; 2021

    2021

  25. [25]

    Quantifying the Evolutionary Dynamics of Language

    Aiden E, Michel JB, Jackson J, T ang T , Nowak M. Quantifying the Evolutionary Dynamics of Language. Nature. 2007 11;449:713-6

    2007

  26. [26]

    Brevia —Languages Evolve in Punctuational Bursts

    Atkinson Q, Meade A, Venditti C, Greenhill S, Pagel M. Brevia —Languages Evolve in Punctuational Bursts. Science (New York, NY). 2008 03;319:588

    2008

  27. [27]

    Languages with More Second Language Learners T end to Lose Nominal Case

    Bentz C, Winter B. Languages with More Second Language Learners T end to Lose Nominal Case. Language Dynamics and Change. 2013;3(1):1-27

    2013

  28. [28]

    Rate of Language Evolution is Affected by Population Size

    Bromham L, Hua X, Fitzpatrick TG, Greenhill SJ. Rate of Language Evolution is Affected by Population Size. Proceedings of the National Academy of Sciences. 2015;112(7):2097-102

    2015

  29. [29]

    Conjugation Class from Latin to Romance: Heteroclisis in Diachrony and Synchrony

    Kaye SJ. Conjugation Class from Latin to Romance: Heteroclisis in Diachrony and Synchrony. Oxford University; 2015

    2015

  30. [30]

    Analogy in Morphological Change

    Sims-Williams H. Analogy in Morphological Change. University of Oxford; 2016

    2016

  31. [31]

    Self-Organization in the Spelling of English Suffixes: The Emergence of Culture out of Anarchy

    Berg K, Aronoff M. Self-Organization in the Spelling of English Suffixes: The Emergence of Culture out of Anarchy. Language. 2017;93(1):37-64

    2017

  32. [32]

    Language Change and Morphological Processes

    Hamans C. Language Change and Morphological Processes. Yearbook of the Poznan Linguistic Meeting. 2017;3:1-23. Available from: https://doi.org/10.1515/yplm-2017-0001 . June 10, 2026 38/51 Preprint

    work page doi:10.1515/yplm-2017-0001 2017

  33. [33]

    Morphological Change

    Trips C. Morphological Change. Oxford University Press; 2017. Available from: https://oxfordre.com/linguistics/view/10.1093/acrefore/9780199384655.001.0001/ acrefore-9780199384655-e-260

    work page doi:10.1093/acrefore/9780199384655.001.0001/ 2017

  34. [34]

    New Thoughts on an Old Puzzle: The Italian Alternation Type dissi, dicesti; feci, facesti

    Maiden M. New Thoughts on an Old Puzzle: The Italian Alternation Type dissi, dicesti; feci, facesti. Revue Romane. 2017;53(2):217-60

    2017

  35. [35]

    The Origins of Inflectional Classes

    Bach X. The Origins of Inflectional Classes. Oxford University; 2018

    2018

  36. [36]

    Memetic Epidemiology: A Mathematical Model for Tracking the Spread of Linguistic Features across Western Europe

    Shavitz B. Memetic Epidemiology: A Mathematical Model for Tracking the Spread of Linguistic Features across Western Europe. City University of New York; 2023

    2023

  37. [37]

    Semantic Change in Adults is not Primarily a Generational Phenomenon

    Kamath G, Yang M, Reddy S, Sonderegger M, Card D. Semantic Change in Adults is not Primarily a Generational Phenomenon. Proceedings of the National Academy of Sciences. 2025;122(31):e2426815122. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.2426815122

    work page doi:10.1073/pnas.2426815122 2025

  38. [38]

    Linguistic Change, Social Network and Speaker Innovation

    Milroy J, Milroy L. Linguistic Change, Social Network and Speaker Innovation. Journal of Linguistics. 1985 sep;21(2):339-84

    1985

  39. [39]

    Social Network and Social Class: T oward an Integrated Sociolinguistic Model

    Milroy L, Milroy J. Social Network and Social Class: T oward an Integrated Sociolinguistic Model. Language in Society. 1992;21(1):1-26

    1992

  40. [40]

    Centers and Peripheries: Network Roles in Language Change

    Fagyal Z, Swarup S, Escobar AM, Gasser L, Lakkaraju K. Centers and Peripheries: Network Roles in Language Change. Lingua. 2010 08;120:2061-79

    2010

  41. [41]

    Language Structure Is Partly Determined by Social Structure

    Lupyan G, Dale R. Language Structure Is Partly Determined by Social Structure. PloS one. 2010 01;5:e8559

    2010

  42. [42]

    Population Size and the Rate of Language Evolution: A T est across Indo-European, Austronesian, and Bantu Languages

    Greenhill SJ, Hua X, Welsh CF , Schneemann H, Bromham L. Population Size and the Rate of Language Evolution: A T est across Indo-European, Austronesian, and Bantu Languages. Frontiers in psychology. 2018;9:576

    2018

  43. [43]

    Simpler grammar, Larger vocabulary: How Population Size Affects Language

    Reali F , Chater N, Christiansen MH. Simpler grammar, Larger vocabulary: How Population Size Affects Language. Proceedings of the Royal Society B: Biological Sciences. 2018;285(1871):20172586. Available from: https://royalsocietypublishing.org/doi/abs/10.1098/rspb.2017.2586

    work page doi:10.1098/rspb.2017.2586 2018

  44. [44]

    2013.How to Build a Brain: A Neural Architecture for Biological Cognition

    Christiansen MH, Kirby S. Language Evolution: The Hardest Problem in Science? In: Language Evolution. Oxford University Press; 2003. Available from: https://doi.org/10.1093/acprof:oso/9780199244843.003.0001

    work page doi:10.1093/acprof:oso/9780199244843.003.0001 2003

  45. [45]

    Lecture Notes: Language and Evolution; 2007

    Stabler E. Lecture Notes: Language and Evolution; 2007. Available from: https://www.academia. edu/24306695/Lecture_Notes_Language_and_Evolution?email_work_card=thumbnail

    arXiv 2007

  46. [46]

    Evolution of Linguistic Diversity in a Simple Communication System

    Arita T , Koyama Y . Evolution of Linguistic Diversity in a Simple Communication System. Artificial Life. 1998 01;4(1):109-24. Available from: https://doi.org/10.1162/106454698568477. arXiv:https://direct.mit.edu/artl/article-pdf/4/1/109/1661594/106454698568477.pdf

    work page doi:10.1162/106454698568477 1998

  47. [47]

    The Evolution of Language

    Nowak MA, Krakauer DC. The Evolution of Language. Proceedings of the National Academy of Sciences. 1999;96(14):8028-33. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.96.14.8028. arXiv:https://www.pnas.org/doi/pdf/10.1073/pnas.96.14.8028

    work page doi:10.1073/pnas.96.14.8028 1999

  48. [48]

    The Evolutionary Language Game

    Nowak MA, Plotkin JB, Krakauer DC. The Evolutionary Language Game. Journal of Theoretical Biology. 1999;200(2):147-62. Available from: https://www.sciencedirect.com/science/article/pii/S0022519399909815. June 10, 2026 39/51 Preprint

    1999

  49. [49]

    Syntax Without Natural Selection: How Compositionality Emerges from Vocabulary in a Population of Learners

    Kirby S. Syntax Without Natural Selection: How Compositionality Emerges from Vocabulary in a Population of Learners. In: Knight C, Studdert-Kennedy M, Hurford J, editors. The Evolutionary Emergence of Language: Social Function and the Origins of Linguistic Form. Cambridge University Press; 2000. p. 303 –323

    2000

  50. [50]

    Spontaneous Evolution of Linguistic Structure—an Iterated Learning Model of the Emergence of Regularity and Irregularity

    Kirby S. Spontaneous Evolution of Linguistic Structure—an Iterated Learning Model of the Emergence of Regularity and Irregularity. IEEE Transactions on Evolutionary Computation. 2001;5(2):102-10. doi:10.1109/4235.918430

    work page doi:10.1109/4235.918430 2001

  51. [51]

    In: Cangelosi A, Parisi D, editors

    Cangelosi A, Parisi D. In: Cangelosi A, Parisi D, editors. Computer Simulation: A New Scientific Approach to the Study of Language Evolution. London: Springer London; 2002. p. 3-28. Available from: https://doi.org/10.1007/978-1-4471-0663-0_1

    work page doi:10.1007/978-1-4471-0663-0_1 2002

  52. [52]

    Natural Language From Artificial Life

    Kirby S. Natural Language From Artificial Life. Artificial Life. 2002 04;8(2):185-215. Available from: https://doi.org/10.1162/106454602320184248. arXiv:https://direct.mit.edu/artl/article-pdf/8/2/185/1661892/106454602320184248.pdf

    work page doi:10.1162/106454602320184248 2002

  53. [53]

    Natural Selection and Cultural Selection in the Evolution of Communication

    Smith K. Natural Selection and Cultural Selection in the Evolution of Communication. Adaptive Behavior. 2002 01;10:25-45

    2002

  54. [54]

    Progress in the Simulation of Emergent Communication and Language

    Wagner K, Reggia JA, Uriagereka J, Wilkinson GS. Progress in the Simulation of Emergent Communication and Language. Adaptive Behavior. 2003;11(1):37-69. Available from: https://doi.org/10.1177/10597123030111003

    work page doi:10.1177/10597123030111003 2003

  55. [55]

    Least Effort and the Origins of Scaling in Human Language

    i Cancho RF , Solé RV. Least Effort and the Origins of Scaling in Human Language. Proceedings of the National Academy of Sciences. 2003;100(3):788-91. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.0335980100

    work page doi:10.1073/pnas.0335980100 2003

  56. [56]

    Why Synonymy Is Rare: Fitness Is in the Speaker

    Hurford JR. Why Synonymy Is Rare: Fitness Is in the Speaker. In: Banzhaf W, Ziegler J, Christaller T , Dittrich P , Kim JT , editors. Advances in Artificial Life. Berlin, Heidelberg: Springer Berlin Heidelberg; 2003. p. 442-51

    2003

  57. [57]

    Computer Models of The Evolution of Language and Languages

    Livingstone D. Computer Models of The Evolution of Language and Languages. University of Paisley; 2003

    2003

  58. [58]

    Modelling Language Evolution; 2003

    Cucker F , Smale S, Zhou DX. Modelling Language Evolution; 2003. Available from: https://home.ttic.edu/~smale/papers/language-pdf.pdf

    2003

  59. [59]

    Modelling Language Origins and Evolution; 2003

    Vogt P , de Boer B, Belpaeme T . Modelling Language Origins and Evolution; 2003. IJCAI Tutorial. Available from: https://www.academia.edu/2864168/Modelling_language_origins_and_ evolution?email_work_card=title

    arXiv 2003

  60. [60]

    The Role of Environment Structure in Multi-Agent Simulations of Language Evolution

    Bartlett M, Kazakov D. The Role of Environment Structure in Multi-Agent Simulations of Language Evolution. In: AISB 2004 Convention, Adaptive Agents and Multi-Agent Systems. York, UK; 2004. p. 1-8

    2004

  61. [61]

    A Computational Framework to Simulate the Coevolution of Language and Social Structure

    Gong T , Ke J, Minett JW, Wang WSY . A Computational Framework to Simulate the Coevolution of Language and Social Structure. In: Artificial Life IX: Proceedings of the Ninth International Conference on the Simulation and Synthesis of Living Systems. The MIT Press; 2004. Available from: https://doi.org/10.7551/mitpress/1429.003.0027

    work page doi:10.7551/mitpress/1429.003.0027 2004

  62. [62]

    Decoding Least Effort and Scaling in Signal Frequency Distributions

    Ferrer i Cancho R. Decoding Least Effort and Scaling in Signal Frequency Distributions. Physica A: Statistical Mechanics and its Applications. 2005;345(1):275-84. Available from: https://www.sciencedirect.com/science/article/pii/S0378437104008611

    2005

  63. [63]

    Language as an Evolutionary System

    Brighton H, Smith K, Kirby S. Language as an Evolutionary System. Physics of Life Reviews. 2005;2(3):177-226. Available from: https://www.sciencedirect.com/science/article/pii/S1571064505000229. June 10, 2026 40/51 Preprint

    2005

  64. [64]

    Computational Modeling on Language Emergence: A Coevolution Model of Lexicon, Syntax and Social Structure

    Gong T , Wang W. Computational Modeling on Language Emergence: A Coevolution Model of Lexicon, Syntax and Social Structure. LANGUAGE AND LINGUISTICS-TAIPEI-. 2005;6(1):1

    2005

  65. [65]

    The Role of Population Structure in Language Evolution

    Lee Y , Collier T , Stabler E, T aylor C. The Role of Population Structure in Language Evolution. 2005 01

    2005

  66. [66]

    Language Evolution and Population Dynamics in a System of Two Interacting Species

    Kosmidis K, Halley JM, Argyrakis P . Language Evolution and Population Dynamics in a System of Two Interacting Species. Physica A: Statistical Mechanics and its Applications. 2005;353:595-612. Available from: https://www.sciencedirect.com/science/article/pii/S0378437105001561

    2005

  67. [67]

    Language Evolution in Large Populations of Autonomous Agents: Issues in Scaling

    Vogt P , Divina F . Language Evolution in Large Populations of Autonomous Agents: Issues in Scaling. In: Proceedings of AISB 2005: Social Intelligence and Interaction in Animals, Robots and Agents. Hatfield, England: AISB; 2005. p. 80-7

    2005

  68. [68]

    Nonequilibrium Dynamics of Language Games on Complex Networks

    Dall’Asta L, Baronchelli A, Barrat A, Loreto V. Nonequilibrium Dynamics of Language Games on Complex Networks. Physical Review E. 2006 Sep;74(3). Available from: http://dx.doi.org/10.1103/PhysRevE.74.036105. doi:10.1103/physreve.74.036105

    work page doi:10.1103/physreve.74.036105 2006

  69. [69]

    Cultural Evolution of Language

    Smith K. Cultural Evolution of Language. Encyclopedia of Language & Linguistics. 2006 12

    2006

  70. [70]

    Sharp Transition T owards Shared Vocabularies in Multi-Agent Systems

    Baronchelli A, Felici M, Loreto V, Caglioti E, Steels L. Sharp Transition T owards Shared Vocabularies in Multi-Agent Systems. Journal of Statistical Mechanics: Theory and Experiment. 2006 jun;2006(06):P06014–P06014. Available from: http://dx.doi.org/10.1088/1742-5468/2006/06/P06014

    work page doi:10.1088/1742-5468/2006/06/p06014 2006

  71. [71]

    How to do Experiments in Artificial Language Evolution and Why

    Steels L. How to do Experiments in Artificial Language Evolution and Why. In: Cangelosi A, Smith ADM, Smith K, editors. The Evolution of Language (EVOLANG6). Rome; 2006. p. 323-32

    2006

  72. [72]

    Group Size Effects on the Emergence of Compositional Structures in Language

    Vogt P . Group Size Effects on the Emergence of Compositional Structures in Language. In: Almeida e Costa F , Rocha LM, Costa E, Harvey I, Coutinho A, editors. Advances in Artificial Life. Berlin, Heidelberg: Springer Berlin Heidelberg; 2007. p. 405-14

    2007

  73. [73]

    The Evolution of Conventions in Multi-Agent Systems

    De Vylder B. The Evolution of Conventions in Multi-Agent Systems. Unpublished doctoral dissertation, Vrije Universiteit Brussel, Brussels. 2007

    2007

  74. [74]

    Language Evolution as a Darwinian Process: Computational Studies

    Oudeyer PY , Kaplan F . Language Evolution as a Darwinian Process: Computational Studies. Cognitive processing. 2007 02;8:21-35

    2007

  75. [75]

    Stochastic Models of Evolution in Genetics, Ecology and Linguistics

    Blythe RA, McKane AJ. Stochastic Models of Evolution in Genetics, Ecology and Linguistics. Journal of Statistical Mechanics: Theory and Experiment. 2007 jul;2007(07):P07018 –P07018. Available from: http://dx.doi.org/10.1088/1742-5468/2007/07/P07018

    work page doi:10.1088/1742-5468/2007/07/p07018 2007

  76. [76]

    Computational Modelling of Evolution: Ecosystems and Language; 2008

    Lipowski A, Lipowska D. Computational Modelling of Evolution: Ecosystems and Language; 2008. Available from: https://arxiv.org/abs/0810.4952. arXiv:0810.4952

    Pith/arXiv arXiv 2008

  77. [77]

    Finite Populations Choose an Optimal Language

    Pawlowitsch C. Finite Populations Choose an Optimal Language. Journal of theoretical biology. 2008 01;249:606-16

    2008

  78. [78]

    Evolution of Communication and Language in Embodied Agents

    Nolfi S, Mirolli M. Evolution of Communication and Language in Embodied Agents. Springer Science & Business Media; 2009

    2009

  79. [79]

    Convergence Bounds for Language Evolution by Iterated Learning; 2009

    Griffiths TL, Klein D, Rafferty AN. Convergence Bounds for Language Evolution by Iterated Learning; 2009. Available from: https://api.semanticscholar.org/CorpusID:269866

    2009

  80. [80]

    Why are Some Word Orders more Common than Others? A Uniform Information Density Account

    Maurits L, Navarro D, Perfors A. Why are Some Word Orders more Common than Others? A Uniform Information Density Account. vol. 23; 2010. p. 1585-93

    2010

Showing first 80 references.