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

  • 2021-09-10 09:30:35
  • Yuchen Lian, Arianna Bisazza, Tessa Verhoef
  • 0

Abstract

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

 

Quick Read (beta)

loading the full paper ...