Ambiguous words communicate better, not worse
January 23, 2012
Noam Chomsky was wrong, say MIT cognitive scientists.
Words with multiple meanings (like “mean”) became popular because they were actually more efficient than using unambiguous new words, not as a side effect, they suggest.
Why? Because it’s more complex and time-consuming for a speaker/writer to invent new words (and explain them) than it is for the listener/reader to simply infer the meaning from the context — “a mean person” vs. “I mean” vs. “the mean age.”
So words with fewer syllables, higher frequency, and simpler pronunciation were preferred. Another reason why computational methods (used in this finding) are replacing conventional linguistics.
Ref.: Steven T. Piantadosia, Harry Tily, Edward Gibson, The communicative function of ambiguity in language, Cognition, 2012 [DOI: 10.1016/j.cognition.2011.10.004]
Comments (2)
by star0
Chomsky’s claim seems obvious to me; I had long though that myself.
On the other hand, Chomsky’s ideas about the “hard wired” aspects of language (a different matter altogether) seem less clear. If he is wrong, and humans learn language according to some general, simple statistical principles, then that is very good news for AI, because there is a chance that these principles can be discovered and used in applications. In support of this (and contra Chomsky), here is a short quote from the recent New Scientist article “Human Beings Are Learning Machines”:
[quote] Chomsky singularly deserves credit for giving rise to the new
cognitive sciences of the mind. He was instrumental in helping us
think about the mind as a kind of machine. He has made some very
compelling arguments to explain why everybody with an intact brain
speaks grammatically even though children are not explicitly taught
the rules of grammar.
But over the past 10 years we have started to see powerful evidence
that children might learn language statistically, by unconsciously
tabulating patterns in the sentences they hear and using these to
generalise to new cases. Children might learn language effortlessly
not because they possess innate grammatical rules, but because
statistical learning is something we all do incessantly and
automatically. The brain is designed to pick up on patterns of all
kinds.[/quote]
by Rob Fleming
Ambiguous words, like metaphors, engage more brain cells, making them more likely to be remembered and selected the next time a trigger event occurs.
Hmmmm, “like” is ambiguous. “as do metaphors” is what I would like to have said.