{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,18]],"date-time":"2026-07-18T16:06:04Z","timestamp":1784390764910,"version":"3.55.0"},"reference-count":57,"publisher":"MIT Press","license":[{"start":{"date-parts":[[2023,8,16]],"date-time":"2023-08-16T00:00:00Z","timestamp":1692144000000},"content-version":"vor","delay-in-days":227,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["direct.mit.edu"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2023,8,15]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Artificial learners often behave differently from human learners in the context of neural agent-based simulations of language emergence and change. A common explanation is the lack of appropriate cognitive biases in these learners. However, it has also been proposed that more naturalistic settings of language learning and use could lead to more human-like results. We investigate this latter account, focusing on the word-order\/case-marking trade-off, a widely attested language universal that has proven particularly hard to simulate. We propose a new Neural-agent Language Learning and Communication framework (NeLLCom) where pairs of speaking and listening agents first learn a miniature language via supervised learning, and then optimize it for communication via reinforcement learning. Following closely the setup of earlier human experiments, we succeed in replicating the trade-off with the new framework without hard-coding specific biases in the agents. We see this as an essential step towards the investigation of language universals with neural learners.<\/jats:p>","DOI":"10.1162\/tacl_a_00587","type":"journal-article","created":{"date-parts":[[2023,8,16]],"date-time":"2023-08-16T19:47:06Z","timestamp":1692215226000},"page":"1033-1047","update-policy":"https:\/\/doi.org\/10.1162\/mitpressjournals.corrections.policy","source":"Crossref","is-referenced-by-count":2,"title":["Communication Drives the Emergence of Language Universals in Neural Agents: Evidence from the Word-order\/Case-marking Trade-off"],"prefix":"10.1162","volume":"11","author":[{"given":"Yuchen","family":"Lian","sequence":"first","affiliation":[{"name":"Faculty of Electronic and Information Engineering, Xi\u2019an Jiaotong University, China"},{"name":"Leiden Institute of Advanced Computer Science, Leiden University, The Netherlands. y.lian@liacs.leidenuniv.nl"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Arianna","family":"Bisazza","sequence":"additional","affiliation":[{"name":"Center for Language and Cognition, University of Groningen, The Netherlands. a.bisazza@rug.nl"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tessa","family":"Verhoef","sequence":"additional","affiliation":[{"name":"Leiden Institute of Advanced Computer Science, Leiden University, The Netherlands. t.verhoef@liacs.leidenuniv.nl"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"281","published-online":{"date-parts":[[2023,8,15]]},"reference":[{"key":"2023081619464947400_bib1","doi-asserted-by":"publisher","first-page":"1233","DOI":"10.1162\/tacl_a_00424","article-title":"On the difficulty of translating free-order case-marking languages","volume":"9","author":"Bisazza","year":"2021","journal-title":"Transactions of the Association for Computational Linguistics"},{"key":"2023081619464947400_bib2","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781139164894","volume-title":"Case","author":"Blake","year":"2001"},{"key":"2023081619464947400_bib3","doi-asserted-by":"publisher","first-page":"981","DOI":"10.18653\/v1\/D18-1119","article-title":"How agents see things: On visual representations in an emergent language game","volume-title":"Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing","author":"Bouchacourt","year":"2018"},{"key":"2023081619464947400_bib4","doi-asserted-by":"publisher","first-page":"4427","DOI":"10.18653\/v1\/2020.acl-main.407","article-title":"Compositionality and generalization in emergent languages","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Chaabouni","year":"2020"},{"key":"2023081619464947400_bib5","article-title":"Anti-efficient encoding in emergent communication","volume-title":"Advances in Neural Information Processing Systems","author":"Chaabouni","year":"2019"},{"issue":"12","key":"2023081619464947400_bib6","doi-asserted-by":"publisher","first-page":"e2016569118","DOI":"10.1073\/pnas.2016569118","article-title":"Communicating artificial neural networks develop efficient color-naming systems","volume":"118","author":"Chaabouni","year":"2021","journal-title":"Proceedings of the National Academy of Sciences"},{"key":"2023081619464947400_bib7","doi-asserted-by":"publisher","first-page":"5166","DOI":"10.18653\/v1\/P19-1509","article-title":"Word-order biases in deep-agent emergent communication","volume-title":"Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics","author":"Chaabouni","year":"2019"},{"key":"2023081619464947400_bib8","article-title":"Emergent communication at scale","volume-title":"International Conference on Learning Representations","author":"Chaabouni","year":"2022"},{"key":"2023081619464947400_bib9","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.3555","article-title":"Empirical evaluation of gated recurrent neural networks on sequence modeling","author":"Chung","year":"2014","journal-title":"CoRR"},{"key":"2023081619464947400_bib10","volume-title":"Language Universals and Linguistic Typology: Syntax and Morphology","author":"Comrie","year":"1989"},{"issue":"3","key":"2023081619464947400_bib11","doi-asserted-by":"publisher","first-page":"306","DOI":"10.1016\/j.cognition.2011.10.017","article-title":"Learning biases predict a word order universal","volume":"122","author":"Culbertson","year":"2012","journal-title":"Cognition"},{"key":"2023081619464947400_bib12","doi-asserted-by":"publisher","first-page":"2970","DOI":"10.1109\/ICCV.2017.321","article-title":"Learning cooperative visual dialog agents with deep reinforcement learning","volume-title":"2017 IEEE International Conference on Computer Vision (ICCV)","author":"Das","year":"2017"},{"key":"2023081619464947400_bib13","doi-asserted-by":"publisher","first-page":"381","DOI":"10.1007\/1-4020-3395-8_17","article-title":"Computer modelling as a tool for understanding language evolution","author":"De Boer","year":"2006","journal-title":"Evolutionary Epistemology, Language and Culture: A Non-Adaptationist, Systems Theoretical Approach"},{"key":"2023081619464947400_bib14","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1911.01892","article-title":"Focus on what\u2019s informative and ignore what\u2019s not: Communication strategies in a referential game","volume-title":"EmeCom Workshop at NeurIPS 2019","author":"Dess\u00ec","year":"2019"},{"issue":"2","key":"2023081619464947400_bib15","doi-asserted-by":"publisher","first-page":"179","DOI":"10.1207\/s15516709cog1402_1","article-title":"Finding structure in time","volume":"14","author":"Elman","year":"1990","journal-title":"Cognitive Science"},{"key":"2023081619464947400_bib16","article-title":"Emergent communication in a multi-modal, multi-step referential game","volume-title":"International Conference on Learning Representations","author":"Evtimova","year":"2018"},{"issue":"44","key":"2023081619464947400_bib17","doi-asserted-by":"publisher","first-page":"17897","DOI":"10.1073\/pnas.1215776109","article-title":"Language learners restructure their input to facilitate efficient communication","volume":"109","author":"Fedzechkina","year":"2012","journal-title":"Proceedings of the National Academy of Sciences"},{"issue":"2","key":"2023081619464947400_bib18","doi-asserted-by":"publisher","first-page":"416","DOI":"10.1111\/cogs.12346","article-title":"Balancing effort and information transmission during language acquisition: Evidence from word order and case marking","volume":"41","author":"Fedzechkina","year":"2017","journal-title":"Cognitive Science"},{"issue":"1","key":"2023081619464947400_bib19","doi-asserted-by":"publisher","first-page":"29","DOI":"10.1146\/annurev-psych-122216-011653","article-title":"A mechanistic framework for explaining audience design in language production","volume":"70","author":"Ferreira","year":"2019","journal-title":"Annual Review of Psychology"},{"key":"2023081619464947400_bib20","first-page":"91","article-title":"Quantifying word order freedom in dependency corpora","volume-title":"Proceedings of the Third International Conference on Dependency Linguistics (Depling 2015)","author":"Futrell","year":"2015"},{"issue":"11","key":"2023081619464947400_bib21","doi-asserted-by":"publisher","DOI":"10.3389\/fnhum.2011.00011","article-title":"Experimental semiotics: A review","volume":"5","author":"Galantucci","year":"2011","journal-title":"Frontiers in Human Neuroscience"},{"key":"2023081619464947400_bib22","article-title":"Emergent communication for understanding human language evolution: What\u2019s missing?","volume-title":"Emergent Communication Workshop at ICLR 2022","author":"Galke","year":"2022"},{"issue":"5","key":"2023081619464947400_bib23","doi-asserted-by":"publisher","first-page":"389","DOI":"10.1016\/j.tics.2019.02.003","article-title":"How efficiency shapes human language","volume":"23","author":"Gibson","year":"2019","journal-title":"Trends in Cognitive Sciences"},{"key":"2023081619464947400_bib24","volume-title":"Deep Learning","author":"Goodfellow","year":"2016"},{"issue":"11","key":"2023081619464947400_bib25","doi-asserted-by":"publisher","first-page":"818","DOI":"10.1016\/j.tics.2016.08.005","article-title":"Pragmatic language interpretation as probabilistic inference","volume":"20","author":"Goodman","year":"2016","journal-title":"Trends in Cognitive Sciences"},{"key":"2023081619464947400_bib26","volume-title":"Universals of Language","author":"Greenberg","year":"1963"},{"key":"2023081619464947400_bib27","first-page":"2146","article-title":"Emergence of language with multi-agent games: Learning to communicate with sequences of symbols","volume-title":"Proceedings of the 31st International Conference on Neural Information Processing Systems","author":"Havrylov","year":"2017"},{"issue":"2","key":"2023081619464947400_bib28","doi-asserted-by":"publisher","first-page":"151","DOI":"10.1080\/15475441.2005.9684215","article-title":"Regularizing unpredictable variation: The roles of adult and child learners in language formation and change","volume":"1","author":"Hudson Kam","year":"2005","journal-title":"Language Learning and Development"},{"key":"2023081619464947400_bib29","doi-asserted-by":"publisher","first-page":"55","DOI":"10.18653\/v1\/D19-3010","article-title":"EGG: A toolkit for research on emergence of lanGuage in games","volume-title":"Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP): System Demonstrations","author":"Kharitonov","year":"2019"},{"key":"2023081619464947400_bib30","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.6980","article-title":"Adam: A method for stochastic optimization","volume-title":"3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7\u20139, 2015, Conference Track Proceedings","author":"Kingma","year":"2015"},{"issue":"2","key":"2023081619464947400_bib31","doi-asserted-by":"publisher","first-page":"102","DOI":"10.1109\/4235.918430","article-title":"Spontaneous evolution of linguistic structure-an iterated learning model of the emergence of regularity and irregularity","volume":"5","author":"Kirby","year":"2001","journal-title":"IEEE Transactions on Evolutionary Computation"},{"key":"2023081619464947400_bib32","doi-asserted-by":"publisher","first-page":"108","DOI":"10.1016\/j.conb.2014.07.014","article-title":"Iterated learning and the evolution of language","volume":"28","author":"Kirby","year":"2014","journal-title":"Current Opinion in Neurobiology"},{"key":"2023081619464947400_bib33","doi-asserted-by":"publisher","first-page":"886349","DOI":"10.3389\/frai.2022.886349","article-title":"Emerging grounded shared vocabularies between human and machine, inspired by human language evolution","volume":"5","author":"Kouwenhoven","year":"2022","journal-title":"Frontiers in Artificial Intelligence"},{"key":"2023081619464947400_bib34","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2006.02419","article-title":"Emergent multi-agent communication in the deep learning era","author":"Lazaridou","year":"2020","journal-title":"arXiv preprint arXiv:2006 .02419v2"},{"key":"2023081619464947400_bib35","article-title":"Emergence of linguistic communication from referential games with symbolic and pixel input","volume-title":"International Conference on Learning Representations","author":"Lazaridou","year":"2018"},{"key":"2023081619464947400_bib36","doi-asserted-by":"publisher","first-page":"7663","DOI":"10.18653\/v1\/2020.acl-main.685","article-title":"Multi-agent communication meets natural language: Synergies between functional and structural language learning","volume-title":"Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics","author":"Lazaridou","year":"2020"},{"key":"2023081619464947400_bib37","doi-asserted-by":"publisher","first-page":"1192","DOI":"10.18653\/v1\/D16-1127","article-title":"Deep reinforcement learning for dialogue generation","volume-title":"Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing","author":"Li","year":"2016"},{"key":"2023081619464947400_bib38","doi-asserted-by":"publisher","first-page":"10121","DOI":"10.18653\/v1\/2021.emnlp-main.794","article-title":"The effect of efficient messaging and input variability on neural-agent iterated language learning","volume-title":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","author":"Lian","year":"2021"},{"key":"2023081619464947400_bib39","article-title":"On the interaction between supervision and self-play in emergent communication","volume-title":"International Conference on Learning Representations","author":"Lowe","year":"2020"},{"key":"2023081619464947400_bib40","first-page":"6437","article-title":"Countering language drift with seeded iterated learning","volume-title":"International Conference on Machine Learning","author":"Yuchen","year":"2020"},{"key":"2023081619464947400_bib41","doi-asserted-by":"publisher","first-page":"596","DOI":"10.4324\/9781315782379-138","article-title":"Case, word order, and language learnability: Insights from connectionist modeling","volume-title":"Proceedings of the Twenty-Fourth Annual Conference of the Cognitive Science Society","author":"Lupyan","year":"2002"},{"key":"2023081619464947400_bib42","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v32i1.11492","article-title":"Emergence of grounded compositional language in multi-agent populations","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence","author":"Mordatch","year":"2018"},{"key":"2023081619464947400_bib43","article-title":"Automatic differentiation in pytorch","volume-title":"NIPS 2017 Autodiff Workshop","author":"Paszke","year":"2017"},{"key":"2023081619464947400_bib44","article-title":"Compositional languages emerge in a neural iterated learning model","volume-title":"International Conference on Learning Representations","author":"Yi","year":"2020"},{"key":"2023081619464947400_bib45","doi-asserted-by":"publisher","first-page":"335","DOI":"10.18653\/v1\/2020.conll-1.26","article-title":"\u201cLazImpa\u201d: Lazy and impatient neural agents learn to communicate efficiently","volume-title":"Proceedings of the 24th Conference on Computational Natural Language Learning","author":"Rita","year":"2020"},{"key":"2023081619464947400_bib46","article-title":"Emergent communication: Generalization and overfitting in lewis games","volume-title":"Advances in Neural Information Processing Systems","author":"Rita","year":"2022"},{"issue":"18","key":"2023081619464947400_bib47","doi-asserted-by":"publisher","first-page":"7361","DOI":"10.1073\/pnas.0702077104","article-title":"The emergence of simple languages in an experimental coordination game","volume":"104","author":"Selten","year":"2007","journal-title":"Proceedings of the National Academy of Sciences"},{"key":"2023081619464947400_bib48","doi-asserted-by":"publisher","first-page":"67","DOI":"10.1075\/slcs.94.06sin","article-title":"Complexity trade-offs in core argument marking","volume-title":"Language Complexity","author":"Sinnem\u00e4ki","year":"2008"},{"issue":"7","key":"2023081619464947400_bib49","doi-asserted-by":"publisher","first-page":"1317","DOI":"10.1111\/j.1551-6709.2009.01065.x","article-title":"Relationships between language structure and language learning: The suffixing preference and grammatical categorization","volume":"33","author":"St. Clair","year":"2009","journal-title":"Cognitive Science"},{"issue":"1","key":"2023081619464947400_bib50","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1075\/eoc.1.1.02ste","article-title":"The synthetic modeling of language origins","volume":"1","author":"Steels","year":"1997","journal-title":"Evolution of Communication"},{"issue":"1701","key":"2023081619464947400_bib51","doi-asserted-by":"publisher","first-page":"20150447","DOI":"10.1098\/rstb.2015.0447","article-title":"Agent-based models for the emergence and evolution of grammar","volume":"371","author":"Steels","year":"2016","journal-title":"Philosophical Transactions of the Royal Society B: Biological Sciences"},{"key":"2023081619464947400_bib52","doi-asserted-by":"publisher","DOI":"10.24963\/ijcai.2017\/385","article-title":"End-to-end optimization of goal-driven and visually grounded dialogue systems","volume-title":"IJCAI","author":"Strub","year":"2017"},{"key":"2023081619464947400_bib53","doi-asserted-by":"publisher","first-page":"105055","DOI":"10.1016\/j.cognition.2022.105055","article-title":"Redundancy can benefit learning: Evidence from word order and case marking","volume":"224","author":"Tal","year":"2022","journal-title":"Cognition"},{"key":"2023081619464947400_bib54","article-title":"The learnability of constructed languages reflects typological patterns","volume-title":"Proceedings of the Annual Meeting of the Cognitive Science Society","author":"Tily","year":"2011"},{"key":"2023081619464947400_bib55","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1515\/lity.2003.011","article-title":"Language type frequency and learnability from a connectionist perspective","volume":"7","author":"Van Everbroeck","year":"2003","journal-title":"Linguistic Typology"},{"issue":"3","key":"2023081619464947400_bib56","doi-asserted-by":"publisher","first-page":"229","DOI":"10.1007\/BF00992696","article-title":"Simple statistical gradient-following algorithms for connectionist reinforcement learning","volume":"8","author":"Williams","year":"1992","journal-title":"Machine Learning"},{"key":"2023081619464947400_bib57","article-title":"Bias and generalization in deep generative models: An empirical study","volume":"31","author":"Zhao","year":"2018","journal-title":"Advances in Neural Information Processing Systems"}],"container-title":["Transactions of the Association for Computational Linguistics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00587\/2154476\/tacl_a_00587.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/direct.mit.edu\/tacl\/article-pdf\/doi\/10.1162\/tacl_a_00587\/2154476\/tacl_a_00587.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,8,16]],"date-time":"2023-08-16T19:47:20Z","timestamp":1692215240000},"score":1,"resource":{"primary":{"URL":"https:\/\/direct.mit.edu\/tacl\/article\/doi\/10.1162\/tacl_a_00587\/117219\/Communication-Drives-the-Emergence-of-Language"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023]]},"references-count":57,"URL":"https:\/\/doi.org\/10.1162\/tacl_a_00587","relation":{},"ISSN":["2307-387X"],"issn-type":[{"value":"2307-387X","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2023]]},"published":{"date-parts":[[2023]]}}}