The advantage of ambiguity

Cognitive scientists develop a new take on an old problem: why human language has so many words with multiple meanings.


Press Contact

Caroline McCall
Email: newsoffice@mit.edu
Phone: 617-253-2700
MIT News Office

Why did language evolve? While the answer might seem obvious — as a way for individuals to exchange information — linguists and other students of communication have debated this question for years. Many prominent linguists, including MIT’s Noam Chomsky, have argued that language is, in fact, poorly designed for communication. Such a use, they say, is merely a byproduct of a system that probably evolved for other reasons — perhaps for structuring our own private thoughts.

As evidence, these linguists point to the existence of ambiguity: In a system optimized for conveying information between a speaker and a listener, they argue, each word would have just one meaning, eliminating any chance of confusion or misunderstanding. Now, a group of MIT cognitive scientists has turned this idea on its head. In a new theory, they claim that ambiguity actually makes language more efficient, by allowing for the reuse of short, efficient sounds that listeners can easily disambiguate with the help of context.

“Various people have said that ambiguity is a problem for communication,” says Ted Gibson, an MIT professor of cognitive science and senior author of a paper describing the research to appear in the journal Cognition. "But the fact that context disambiguates has important ramifications for the re-use of potentially ambiguous forms. Ambiguity is no longer a problem — it's something that you can take advantage of, because you can reuse easy [words] in different contexts over and over again."

Lead author of the paper is Steven Piantadosi PhD ’11; Harry Tily, a postdoc in the Department of Brain and Cognitive Sciences, is another co-author.

What do you ‘mean’?

For a somewhat ironic example of ambiguity, consider the word “mean.” It can mean, of course, to indicate or signify, but it can also refer to an intention or purpose (“I meant to go to the store”); something offensive or nasty; or the mathematical average of a set of numbers. Adding an ‘s’ introduces even more potential definitions: an instrument or method (“a means to an end”), or financial resources (“to live within one’s means”).

But virtually no speaker of English gets confused when he or she hears the word “mean.” That’s because the different senses of the word occur in such different contexts as to allow listeners to infer its meaning nearly automatically.

Given the disambiguating power of context, the researchers hypothesized that languages might harness ambiguity to reuse words — most likely, the easiest words for language processing systems. Building on observation and previous studies, they posited that words with fewer syllables, high frequency and the simplest pronunciations should have the most meanings.

To test this prediction, Piantadosi, Tily and Gibson carried out corpus studies of English, Dutch and German. (In linguistics, a corpus is a large body of samples of language as it is used naturally, which can be used to search for word frequencies or patterns.) By comparing certain properties of words to their numbers of meanings, the researchers confirmed their suspicion that shorter, more frequent words, as well as those that conform to the language’s typical sound patterns, are most likely to be ambiguous — trends that were statistically significant in all three languages.

To understand why ambiguity makes a language more efficient rather than less so, think about the competing desires of the speaker and the listener. The speaker is interested in conveying as much as possible with the fewest possible words, while the listener is aiming to get a complete and specific understanding of what the speaker is trying to say. But as the researchers write, it is “cognitively cheaper” to have the listener infer certain things from the context than to have the speaker spend time on longer and more complicated utterances. The result is a system that skews toward ambiguity, reusing the “easiest” words. Once context is considered, it’s clear that “ambiguity is actually something you would want in the communication system,” Piantadosi says.

Tom Wasow, a professor of linguistics and philosophy at Stanford University, calls the paper “important and insightful.”

“You would expect that since languages are constantly changing, they would evolve to get rid of ambiguity,” Wasow says. “But if you look at natural languages, they are massively ambiguous: Words have multiple meanings, there are multiple ways to parse strings of words. … This paper presents a really rigorous argument as to why that kind of ambiguity is actually functional for communicative purposes, rather than dysfunctional.”

Implications for computer science

The researchers say the statistical nature of their paper reflects a trend in the field of linguistics, which is coming to rely more heavily on information theory and quantitative methods.

“The influence of computer science in linguistics right now is very high,” Gibson says, adding that natural language processing (NLP) is a major goal of those operating at the intersection of the two fields.

Piantadosi points out that ambiguity in natural language poses immense challenges for NLP developers. “Ambiguity is only good for us [as humans] because we have these really sophisticated cognitive mechanisms for disambiguating,” he says. “It’s really difficult to work out the details of what those are, or even some sort of approximation that you could get a computer to use.”

But, as Gibson says, computer scientists have long been aware of this problem. The new study provides a better theoretical and evolutionary explanation of why ambiguity exists, but the same message holds: “Basically, if you have any human language in your input or output, you are stuck with needing context to disambiguate,” he says.


Topics: Brain and cognitive sciences, Computer science and technology, Human-computer interaction, Language, Linguistics, Natural language processing

Comments

I wish they had included French in the study. It has several instances of short words that sound almost exactly alike, mean different things, and cannot easily be disambiguated. For example, "below" and "above"; and "two" and "twelve" when followed by certain other words. This property of the language gave me a lot of trouble both when I was learning it, and later in normal conversation.
Instead of trying to imitate the way the human mind disambiguates, why not create an algorithm that would allow an intelligent computer to learn more and more context clues as it is exposed to new information? Disambiguation would happen naturally thereafter through association with previous experiences.
Euro-non-tonal language researchers ?? They are unaware of "Dragon Naturally Speaking" & tonal spoken languages (never intended to be in print copy; e.g. Cantonese = 9-tones). Wikipedia is trying to track evolving "simple" & "plain" languages. But they will not allow language developers to participate; only accredited journalists.
I recommend the critical discussion of this article on the linguistics subreddit: http://www.reddit.com/r/linguistics/comments/op8pm/mit_the_advantage_of_ambiguity_why_human_language/
Miguelb, the French words for 'above' and 'below' or 'two' and 'twelve' may sound 'almost exactly alike' to you, but they use quite different vowels. They are not ambiguous for native speakers or for anybody with a good understanding of French phonology.
Japanese has many fewer phonetic sounds and nearly all words are ambiguous. Use of automatic translation software often leads to ludicrous results since most programs ignore context. Among European languages, the collected ambiguities are often similar amongst one another, but these collections of ambiguities are different in Japanese (eg. spider & cloud are both KUMO; floor (level of a building) and time (counting occurences), shell, sea all KAI.)
Ambiguity exists because people want to deceive. Precision avoids deception. Think of the used car salesman. He says "previously owned." I say "used." Think of the precision in a legal contract. The contract seeks to avoid misunderstanding. At least we hope it does. How many contracts might have been written to ensure future legal actions, how many to confuse and hide the real agreements? Now let's think of the political world where EVERY phrase, clause and sentence is meant to appeal to voters yet be unenforceable in consequence. Ambiguity thrives in our political system where all truth is relative. Even the media presents news in terms of "how do we spin this?" I think English as a language was invented to guarantee ambiguity...it is good for my personal economy, the lawyer's value to society and the politician's reelection. What a unique and vibrant world we live in where even "truth" is ambiguous!
I wonder if the research might benefit from what is already known in compression coding. Audio compression -- both modern ADPCM (for your mobile) and MP3 -- uses knowledge of human perception to create efficiency: a lot of data is omitted because it has only minimal effect on your ability to understand the speaker or hear the music. The context - human aural processing - is known so the language saves bits WLOG. In high-end video compression, the algorithm can be tweaked to "spend more bits" on some things instead of others. Interestingly, this is culture-dependent, with some places preferring that bits be spent on the pretty girl's face and others preferring that those bits detail the fountain behind her. Context is culture dependent!
It can be hard learning another language, but seriously, Spanish doesn't differentiate 'his',' her', or 'its' (among other things), so in a sentence like "she used his car" you have to wait til "he got mad about that" to know whose car it was! I can read a short story and only when things stop making sense do I realize the characters I thought were talking aren't even in the picture. I know my understanding of the syntax isn't that great, and old Spanish is a bit different from what I've learned, but some things are ridiculous. Being brief only works if the context is clear and close to the ambiguous word. Otherwise, it's a fool's goose chase for new speakers/readers. I guess in a short story, the author crams in as much information in as few words as possible, and there isn't much time to clear up ambiguity a native speaker can understand.
Towards the end of the article its mentioned that," The speaker is interested in conveying as much as possible with the fewest possible words, while the listener is aiming to get a complete and specific understanding of what the speaker is trying to say". Being longwinded, perhaps inherently so, my sentences aren't always the concisest and its actually quite enjoyable when the person I am paying attention to, takes me on an oratory roller coaster, you know, as long as we still get there, to the point that is. I would even suggest that conversation, both monologue and dialogue, offer up a stroll through the meadows of others minds. Though, at times,"Honey, pass the sugar" ,is enough.
Back to the top