Word Games

Ying-Cheng Lai’s research is not like playing a game of Scrabble or completing a crossword puzzle. However, the work does give him new justification for playing word games.

Lai is an ASU professor of mathematics and electrical engineering. He and his colleagues discovered that word association can link just about any two common words in the English language.

Lai worked with industrial engineer Nong Ye and Partha Dasgupta, an associate professor of computer science and engineering. The ASU researchers devised a method that uses an average of three or four steps and follows the same “six degrees of separation” principle known for connecting actor Kevin Bacon with other film stars.

The scientists presented results for the English language. However, Lai says their method will work for other languages as well, because the fundamental role of any language is the communication of ideas.

To build their conceptual network of language, the ASU scientists used an on-line English thesaurus. The database had 30,000 entries and listed an average of more than 100 words per entry.

Lai’s team created 900,000 different word pairs from the original 30,000 words. They then tried to link the words in each pair via a chain of related words. Lai defined two words as being related if they expressed a similar concept—that is, if they were listed together in any thesaurus entry.

“We constructed a conceptual network from the entries in the thesaurus. We considered two words to be connected if they express similar concepts,” Lai explains. “The network is clearly evolving and sparse. We believe and argue that these findings are important not only for linguistics, but also for cognitive science.”

The secret to the idea of a conceptual network lies in words with multiple meanings. Such words act as short cuts, connecting remote concepts together. This reduces the number of steps needed to get from one word to another.

A few word pairs, however, can’t be linked very closely. For example, Lai says that it took eight steps to get from “octagonal” to “appendectomy” using the link: octave, tone, purity, sterility, birth control, and vasectomy.

Even though the ASU study focused on the meanings of words, Lai says that building the conceptual network of language required an interdisciplinary approach.

“Our conceptual network involves physics, language, biology, and computer science,” he says. “It’s not just about studying the English language. Our work examines a person’s memory or ability to remember the links of words and their meanings.”

Human memory is associative. Lai says this means that information can be retrieved by connecting similar concepts.

“From the standpoint of the retrieval of information in an associative memory, the small-world property of the network represents a maximization of efficiency,” Lai says. “Similar pieces of information are stored together due to the high clustering, which makes searching by association possible.

“However, even very different pieces of information are never separated by more than a few links, which guarantees a fast search,” he adds. “We speculate that associative memory has arisen in humans partly because of a maximization of efficiency in the retrieval by natural selection.”—Manny Romero