Not making sense? Listeners prefer words with a single, unambiguous meaning.

Law of Conversations Explained

Why do you hear the word “lucubration” so rarely? Most words, in fact, aren't used much at all, while a few, like “an” or “ScienceNOW,” crop up all the time. Linguists have puzzled over this pattern--which is shared by all human languages--for more than half a century. Now a mathematical model suggests that the reason may be tension between the needs of speakers and listeners. The finding points to a fundamental difference between communication in humans and other animals and may shed light on the evolution of human language.

The steep drop-off in word frequency is called Zipf's law, after the early 20th century linguist George Zipf, who first observed it. Linguists debate what, if anything, it means. Zipf himself proposed a “principle of least effort” to explain the phenomenon, suggesting that people try to minimize their effort in using language. However, he provided no evidence for how the pattern he observed would achieve this aim.

Linguists agree that speakers and listeners have their own goals for simplifying communication: Speakers prefer a small pool of words that they can quickly retrieve from memory, whereas listeners prefer words with a single, unambiguous meaning--and that requires a large vocabulary. Computer scientist Ramon Ferrer i Cancho and physicist Ricard Solé, both of Pompeu Fabra University in Barcelona, Spain, thought this might explain Zipf's law.

To test their idea, the researchers mathematically defined effort, assuming, for example, that it's harder to recall obscure words. Then they created a model in which they varied the relative effort exerted by the listener and the speaker, and observed the word frequencies that resulted. Zipf's law emerged only when both the hypothetical speakers and listeners slacked off as much as possible, the team reports online this week in the Proceedings of the National Academy of Sciences.

The study demonstrates that Zipf's law is a meaningful feature of human language, the authors argue. Further, they say, the research underscores a fundamental difference between human language and communication in other animals, which have smaller repertoires of signals that are all used with similar frequency. Zipf's law emerges in humans, in contrast, because our ability to amass a large vocabulary is offset by an inability to quickly retrieve obscure words from the recesses of our memory.

“This is a very original contribution and a fascinating new perspective on language evolution," says Martin Nowak, a theoretical biologist at the Institute for Advanced Study in Princeton, New Jersey. He notes, however, that other explanations for Zipf's law exist and that more research is needed to determine which explanation is best.

Related sites
Ramon Ferrer i Cancho's Web page
Ricard Solé's Web page
The International Quantitative Linguistics Association