Tuesday, June 4, 2013

94% of Communication is Non-verbal? Actually, no.

I was recently reading the book Present Shock: When Everything Happens Now, and in it the author mentions several times the statistic that 94% of communication is non-verbal. That is, the vast majority of information we transmit to others during face to face communication is non-verbal cues, such as body language, facial expressions, tone of voice, and so on, while only a very small fraction of the information we transmit is the actual words that we speak.

I've come across this statistic several times in the past in various places, as I'm sure have most other people (if not the specific 94% number, then probably something very similar). So I finally decided to do a little research into where it comes from, and if it's actually true.

94% is fiction

As you probably already guessed, the 94% figure is not true (if it was, I probably wouldn't be writing this post!). The source of this specific figure is... Kramer. Cosmo Kramer. The character in the TV show Seinfeld. That's right, if you ever see the specific value of 94% mentioned in this context, it has almost certainly come from an episode of Seinfeld.

It's quite amusing to think that various serious books, communications workshops, etc. are using a statistic pulled from a fictional character, but surely this value came from somewhere respectable, right? Well, actually yes, but with some very big caveats that change the value of the statistic significantly.

93% is not much better

The original source of high non-verbal communication numbers like 94% is a paper written by Albert Mehrabian in 1972. In the studies on which the paper is based, he found a value of 93%. That seems pretty damn close, right? What the hell am I being so picky for?

Well, as it turns out, this value of 93% only occurred under very specific circumstances: when subjects were reading out single words with positive, neutral (ambiguous) or negative connotations in either positive, neutral or negative tone of voice, or with positive, negative, or neutral facial expressions. So, in other words, when saying a single word with forced (rather than natural) tone of voice or facial expression, people used the tone or facial expression to judge the overall content of the message much more than the word itself.

Or, to put it simply, circumstances that have absolutely no relation to people conducting natural conversations in the real world.

To be fair, Mehrabian never claimed that his studies were providing broad, general answers to this question. That has been done by other people since, who, in a fashion we see so often with the application of scientific data, have taken a paper's results way out of context and subsequently misled other people as a result.

So what is the answer?

It shouldn't be a surprise that there is no general answer to this question. The amount of information transmitted verbally and non-verbally in conversation is going to be highly dependent on the topic being discussed, the conversational nature of the person talking, and also on the person being spoken to. Some people are much more emotive with their tone of voice and body gestures while talking. Some people are better at reading body language and verbal cues.

As we've all learned by writing emails and SMSes, it's often necessary to provide extra markup to plain text (such as smilies or sarcasm quotes) in order to get intention across, so it's not surprising that non-verbal components play some role in communication, and maybe even a big role. But anyone who throws a specific figure at you or suggests that there is a single value for how much of communication is non-verbal, well, that person <sarcasm> is really clever and insightful </sarcasm>.


  1. Of course if 94% was true, or even close, the telephone as an invention would never have taken off. Imagine it taking 20 times longer to tell someone something over the phone than in person!

    There clearly is extra information being carried non-verbally, and I for one feel uncomfortable when missing out that extra information. But as you suggest it's complex and likely not 94% on average.

  2. It's interesting. I'm not at all a fan of telephones, but I don't mind as much communicating via email/text, even though there is clearly less non-verbal information in the latter. I think the problem might be that people talk on the telephone much like they do face to face, so the reduced non-verbal cues make it clearly inferior, while with text, we tend to communicate in a different mode, aware of the need to explain more or provide explicit cues.