Cognitive Daily

i-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifJim just started playing this year for his school’s junior varsity lacrosse team. As a beginner, he doesn’t see a lot of action, but it’s nonetheless exciting to watch the games — they are fast-paced, with plenty of scoring and a few hard hits. Most junior varsity teams don’t have the equipment budget of a varsity team, so they don’t have separate home and away uniforms like varsity teams do. This means game officials have a hard time remembering which team is which, so instead of referring to players as “home” or “away,” they use the color of the uniforms to distinguish between teams. Apparently “maroon” isn’t in a referee’s vocabulary, because they never use that word to refer to Jim’s school’s burgundy apparel. Instead, they just say “red.”

This got some of us parents to thinking. What if our school played a school with truly red uniforms? Since that team is clearly “red,” would someone have to explain to the officials the concept of “maroon”? A few days ago, we found out — during a game against a team with bright-red uniforms, our school was called “black” (the uniforms do actually have a black stripe on them — I’d never noticed before). But I still hesitated a bit whenever the officials called a penalty against “red” — I’d grown accustomed to thinking of “red” as our team.

When you think about language, very few items are actually referred to in the same way by everyone, all the time. Sometimes I call my car a “car,” sometimes it’s my “station wagon,” sometimes it’s a “VW,” and sometimes it’s a “lousy pile of $#@#!” But what if I’m sending Jim into the garage to get something from the car and I need to make sure he checks the right one (we have both a Volkswagen and a Plymouth)? How does he know which car I mean? He probably remembers that I usually call the VW the “car” and the Plymouth a “van.” But what mechanism is he using to remember that?

Over the past few decades, two competing explanations have emerged. One explanation, introduced in 1975 by P.H. Grice, suggests that Jim knows I’m cooperating with him: he’s aware that I want to communicate the specific car, and so relies on his memory of that fact. He might even be aware that I call the van the “old car” when talking to Greta, but knows that I don’t use that word with him (since he is too young to remember when we bought the van). A newer explanation, suggested by S.E. Brennan and H.H. Clark, argues that Jim isn’t assuming cooperation at all; he merely expects speakers to be consistent in the terms they use. Hadas Shintel and Boaz Keysar believe they have designed an experiment which distinguishes between the two explanations.

Shintel and Keysar bring two people into a room with a shelf divided into 16 cubbies, with 7 objects placed randomly in the cubbies. One person is the “director” and the other is the participant. The director tells the participant to move objects around. The key to the experiment is that the director is following a careful script, unknown to the participant, who believes the director to be another paid study participant. The participant is equipped with a head-mounted video camera and eye-tracking device, so the direction of his gaze can be precisely determined at each point during the study.

Before the study begins, the participant watches a “training” video showing the director giving instructions like “move the elephant rattle to the left” to another participant. Half the time, the participant is told not to let the director know he saw her in the video, and that the study is about how effectively the director can communicate. The other half the time, the director and the participant watch the training video together.

Then when the actual experiment begins, the director systematically describes some of the objects from the video using a different name. For example, the “elephant rattle” in the video is referred to as the “baby rattle.” The question is, when the participant thinks the director doesn’t know he watched the video, does his reaction time change? Here are the results:


The graph charts the time from when the director said the name of the object to the time the participant first looked at the object. If the cooperation hypothesis proposed by Grice is true, then when the participant thinks the director knows he saw the video, the participant should be slower to react when the director uses a new name for the object. But when he thinks she doesn’t know he saw the video, he shouldn’t take any longer to respond to a new name. In fact, participants were slower in both cases, and the biggest difference occurred in the “director does not know” condition — the opposite of what Grice’s explanation predicts.

Thus, Shintel and Keysar argue, the simpler explanation must be what’s occurring in this situation: we expect a speaker to be consistent in the terminology she uses to describe objects, and when she’s not consistent, it takes us longer to process her utterances.

But of course, language is very complex, and so are the social circumstances surrounding its use. We’re not surprised when a speaker refers to a “flat” when speaking to a Londoner but an “apartment” when speaking to a New Yorker. There are many situations when we expect speakers to use different vocabulary to refer to the same items. Yet this work shows that when everything else is controlled, consistency is expected more than cooperation is.

That said, would it kill our lacrosse refs to learn the color “maroon”?

Shintel, H., & Keysar, B. (2007). You said it before and you’ll say it again: Expectations of consistency in communication. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(2), 357-369.


  1. #1 Kevin
    April 10, 2007

    That’s an interesting study, Dave, and thanks for highlighting it. As an aside, I can’t find a reference, but I remember asking about the color naming by refs when I was playing basketball in high school. Our school color was orange, and we were always called red. When I asked why, the ref told me that they are taught to always choose 1 syllable colors for the teams. And that was to decrease the chance of stumbling over words and to make it easier to communicate when yelling to be heard by score-keepers, side-lines, etc. I don’t think the refs are incapable, but as you said, people just prefer consistent.

  2. #2 ian
    April 10, 2007

    Which is why your school should have chosen puce as their colour. One syllable goodness 🙂

  3. #3 RCK
    April 10, 2007

    this seems a very limited interpretation of Grice and it is important to realize that the experiments, if anything at all, only show that this very limited interpretation of Grice may not be right, not that Grice was not right.

  4. #4 DavidD
    April 11, 2007

    Are all the differences in data significant? If so, then each difference is presumably the result of those subjects who viewed the tape with the “director” being put in a cooperative, flexible mindset, while those poor, deprived subjects who watched the tape alone had the mindset of being a cog in the machine. So flexible means being slower at the expected words than the cog in the machine, but a little more quickly adapting to unexpected words.

    Now I have experience at being a soccer ref, and I would say from this that being a good ref means being that lonely cog, being a mere instrument of the laws of the game, egoless but simple and fast. Egoless is good given that fans can be critical even over what color the ref calls a team. Fortunately the laws of the game don’t change at all in the middle of a game.

    I also remember reading a paper a few years ago about the names of colors in different cultures. The paper focused on the few cultures in the world that have little exposure to bright colors and therefore have few such words. The fewest number is two, one word for light things and another for dark things. The cultures that have three words all have a word for light things, a word for dark things, and red, red being the color of blood, as vital as that is.

    Of course our advanced cultures have many names for colors, but it’s amazing how many languages have exactly 11 common names for colors. In English there’s white, black, red, orange, yellow, green, blue, purple, gray, pink, and brown. Many languages have different words for exactly the same concepts. Russian apparently has a 12th word for light blue as pink is for red. If I spoke Russian, I might be able to name those colors you used for your data quickly instead of saying the column on the right is blue, and the one on the left is what? Light blue? Periwinkle? Where’s my Crayola box when I need it?

    Now a ref can’t afford such indecision. I’ve heard “silver” instead of “gray”, but I have my doubts about how good that ref was. “Maroon”? What, are you kidding? I am a mere blade of grass, an instrument of the laws of the game. I know 11 colors. Maybe I’d be a better ref if I simplified even more and called light orange “yellow” and dark orange “red”. It’s not the syllables. It’s the simplicity of mind that’s important, the simplicity that lets one be decisive, right or wrong. If all the data is significant here, that simplicity makes us even slower if you confuse us.

    Imagine all the biological and cultural evolution that went into that “red”. And you still want “maroon”?

  5. #5 Mirweis
    April 11, 2007

    Thank you for pointing me to that study. The interest for my own research work are twofold: the pragmatic issue highlighted here is referring to the “grounding” notion of Herbert Clark; the use of eye-tracking technologies in social flavoured cog-sci.

    Kevin’s comment reminds me the notion of “least collaborative effort” in Clark’s psycholinguistic model (In conversation, the participants try to minimize their collaborative effort. Namely the work that both do from the initiation of each contribution to its mutual acceptance). In this case, it’s not conversation in the proper definition of it. The addressee (in this case the parents) cannot really contribute by giving feedback about their misunderstanding to the speaker (referee). following Clark’s model, it can be assumed that the speaker count on the “visibility” and the “synchronicity” of the situation to assume that the addressee might be able to disambiguate the utterance. In other words, people tends to make utterances implying less effort when they know that the situation can be disambiguated and that the misunderstandings are broadly free of cost.

  6. #6 MT
    April 20, 2007

    It’s H. Paul Grice not Paul H. Grice…and that seems like a lame experiment to me. You can interpret words consciously or unconsciously. Most it’s the latter, but you’ll resort to the former when nonplussed–as by a speech act or any other event that doesn’t conform to whatever unconscious expectations you had. Who cares how people consciously interpret language? I doubt Grice had conscious interpretation of language in mind when he proposed his theory. It’s not a job for a philosopher.

New comments have been disabled.