Color Overmodification Emerges from Data-Driven Learning and Pragmatic Reasoning
This work addresses the challenge of studying language acquisition and pragmatic reasoning in humans by using computational models, though it is incremental as it builds on existing theories with new simulations.
The paper tackled the problem of understanding why speakers use redundant modifiers in language by using neural networks as learning agents to simulate varying environmental conditions, finding that overmodification is more likely with infrequent or salient features, and these results align with a probabilistic pragmatic model.
Speakers' referential expressions often depart from communicative ideals in ways that help illuminate the nature of pragmatic language use. Patterns of overmodification, in which a speaker uses a modifier that is redundant given their communicative goal, have proven especially informative in this regard. It seems likely that these patterns are shaped by the environment a speaker is exposed to in complex ways. Unfortunately, systematically manipulating these factors during human language acquisition is impossible. In this paper, we propose to address this limitation by adopting neural networks (NN) as learning agents. By systematically varying the environments in which these agents are trained, while keeping the NN architecture constant, we show that overmodification is more likely with environmental features that are infrequent or salient. We show that these findings emerge naturally in the context of a probabilistic model of pragmatic communication.