By about 1500 AD, the Middle English word deere had lost its original meaning. Once associated with animals or wild animals generally, deere has come to mean a very specific animal, relatively small, which we now refer to as a deer.
Linguistics offers thousands of documented examples of changes in the associations of words and their meanings. How word meanings evolve, and by which underlying social and cognitive processes, remains unknown, at least in a generalized sense.
Participants in a March SFI working group called for coordinated computational approaches to address the outstanding question of how word-meaning associations change through human history.
“We all agree that this is the time when we should sit down and really try to synthesize the evidence we have, and try to come up with laws of meaning change,” says SFI Professor Tanmoy Bhattacharya, who co- organized the Lexical Semantic Networks and Language Change working group with SFI External Professor Peter Stadler and SFI Professor David Wolpert. “Now that we have a frame of the problem, and participants know that other people are working on similar things, the ideas they share are likely to spark research collaborations.”
Given the distributed and quintessentially complex nature of language, the group has teed up research approaches from complex systems science to initiate their search for laws of meaning change. Similar approaches have already advanced researchers’ understandings of how word sounds change over time, and of the similarities in meaning structures across languages.
To approach the mechanisms for how word meanings evolve, the participants propose combining large linguistic datasets with agent-based models and perspectives from information theory, network science, machine learning, and distributed optimization.
Such research could advance the scientific understanding of the evolution of human language – and potentially find application in artificial intelligence – Bhattacharya says.