The Effect of Efficient Messaging and Input Variability on Neural-Agent Iterated Language Learning

Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 10121–10129

Abstract

Natural languages display a trade-off among different strategies to convey syntactic structure, such as word order or inflection.This trade-off, however, has not appeared in recent simulations of iterated language learning with neural network agents (Chaabouni et al., 2019b).We re-evaluate this result in light of three factors that play an important role in comparable experiments from the Language Evolution field: (i) speaker bias towards efficient messaging, (ii) non systematic input languages, and (iii) learning bottleneck.Our simulations show that neural agents mainly strive to maintain the utterance type distribution observed during learning, instead of developing a more efficient or systematic language.

Authors 3

1. Yuchen Lian Aachen Leiden Institute of Advanced Computer Science

   Leiden University · Xi'an Jiaotong University

   Affiliation as printed

   Faculty of Electronic and Information Engineering , Xi'an Jiaotong University

   Leiden Institute of Advanced Computer Science , Leiden University

   Faculty of Electronic and Information Engineering, Xi'an Jiaotong University

   Leiden Institute of Advanced Computer Science, Leiden University

2. Arianna Bisazza

   University of Groningen

   Affiliation as printed

   Center for Language and Cognition , University of Groningen

   Center for Language and Cognition, University of Groningen

3. Tessa Verhoef corresponding Aachen Leiden Institute of Advanced Computer Science

   Leiden University

   Affiliation as printed

   Leiden Institute of Advanced Computer Science , Leiden University

   Leiden Institute of Advanced Computer Science, Leiden University