Sequence Memory Constraints Give Rise to Language-Like Structure through Iterated Learning

Abstract

Human language is composed of sequences of reusable elements. The origins of the sequential structure of language is a hotly debated topic in evolutionary linguistics. In this paper, we show that sets of sequences with language-like statistical properties can emerge from a process of cultural evolution under pressure from chunk-based memory constraints. We employ a novel experimental task that is non-linguistic and non-communicative in nature, in which participants are trained on and later asked to recall a set of sequences one-by-one. Recalled sequences from one participant become training data for the next participant. In this way, we simulate cultural evolution in the laboratory. Our results show a cumulative increase in structure, and by comparing this structure to data from existing linguistic corpora, we demonstrate a close parallel between the sets of sequences that emerge in our experiment and those seen in natural language.

Chunk‐Based Memory Constraints on the Cultural Evolution of Language

“…Mirroring the reuse of chunks reported by Cornish et al. (), baboon performance on the memory task also significantly increased over generations (reaching approximately 72% by the final generation), which suggests that tetrominos may be easier for the baboons to spatially chunk and recall relative to other configurations. This idea parallels evidence from the human literature, which demonstrates that the presence of recurring visual statistical regularities can allow learners to compress information into chunks, which in turn strengthens visual working memory representations of that information (Brady, Konkle, & Alvarez, ).…”

“…But can repeated processes of chunking across generations of language learners actually shape aspects of linguistic structure? A recent study by Cornish, Dale, Kirby, and Christiansen () attempted to answer this question using an experimental version of cultural evolution known as “iterated learning” (see Scott‐Phillips & Kirby, , for a review). Iterated learning experiments can be thought of as a lab‐based version of the childhood game of telephone: Participants are organized in “generational” chains, in which what one person learns is passed on to the next one in a chain to simulate the cultural transmission of linguistic structure.…”

“…Cornish et al. () interpreted their analyses as suggesting that cultural evolution, both in the lab and in the real world, may favor the selection of structures that attenuate the challenges posed by the Now‐or‐Never bottleneck. To ensure that these patterns were in fact reflective of chunk‐based memory processes, and not simply a consequence of the strings being generated from a limited set of elements, network analyses were repeated on three additional types of human‐generated sequences: word frequencies, passwords, and random numbers.…”

Population turnover facilitates cultural selection for efficiency

“…Such memory constraints have been long recognized as necessary for both learning and cultural evolution [50,51,52]. Great tits appear to have exceptional memory for socially learned behaviors, remembering preferences years after learning a similar foraging puzzle [41].…”

Alien forms for alien language: investigating novel form spaces in cultural evolution

“…Together, these results suggest two major findings. First, even under conditions where forms are not mapped to any meanings, and copying accuracy is the only performance directive for participants, forms emerge which have internal structure (Cornish et al 2013); see also, Cornish et al (2017). This indicates that form structure facilitates learning, even without the anchor of a structured meaning space.…”