91��Ƭ��

‘People have been interested in personality and language for a long time, but it’s really hard to get somebody to sit down and write 100,000 words. The nice thing about bloggers is they write a lot, often over very long periods of time’

Our words convey meaning, but our choice of words also conveys details about our personalities, new research confirms. Extraverts are likely to use the word “mouth” frequently, and “open” personalities are likely to use words like “folk,” “poetry” and “universe.”

In one of the largest studies on the matter to date, Tal Yarkoni, a psychology and neuroscience post-doctoral fellow at the University of Colorado at Boulder, explores what the words you write reveal about you.

His work also rebuts the widely held belief that people can maintain distinctly different offline and online personalities.

Yarkoni’s research was recently published in the Journal of Research in Personality and was funded by the National Institutes of Health.

Several previous studies have identified a connection between language usage and personality. While the approaches to these studies have varied, they share several limitations—for instance, using writing or speech samples that were limited in size, and focusing exclusively on relatively broad dimensions of personality.

By studying language use in a large sample of bloggers, Yarkoni surmounted these limitations.

“People have been interested in personality and language for a long time, but it’s really hard to get somebody to sit down and write 100,000 words. The nice thing about bloggers is they write a lot, often over very long periods of time,” Yarkoni said.

The data Yarkoni collected bear this out. On average, each of the nearly 700 bloggers in his sample provided more than 115,000 words of text, representing more than a novel’s worth of material, and many times what previous studies have used.

These larger samples “provided the ability to identify not only really large effects, which smaller studies can do, but also more subtle relationships,” Yarkoni said.

“And they let us look at things you can’t really look at with small writing samples. For instance, are there specific words that people with different personalities tend to use more? Previous studies have really only been able to look at broad categories made up of many words.”

For instance, Yarkoni was able to determine that highly neurotic people tend to use specific negative adjectives like “terrible” and “worse” more, rather than just observing that they use more negative words generally.

To obtain his uniquely large dataset, Yarkoni identified potential participants using Google’s popular Blog Search engine, and emailed bloggers who publicly listed an email address. The response was surprising—and gratifying—he notes.

“I was expecting to get a 1- to 2-percent response rate. It ended up closer to 10 percent. I sent out 5,000 emails, and about 500 participants came from those emails.”

After agreeing to participate, bloggers were directed to a web site where they submitted information about their personal background and filled out a personality questionnaire.

To ensure that participants didn’t get bored and drop out of the study, they could choose between filling out a shorter, 100-item personality questionnaire and a longer, 315-item questionnaire.

“I didn’t want people to say, ‘That’s going to take too much time,’” Yarkoni notes. “So, the shorter one is a standard ‘Big Five’ personality measure, while the longer one gives additional information about more specific ‘facets.’”

The “Big Five” personality traits are openness, conscientiousness, extraversion, agreeableness and neuroticism. Previous studies have shown that these five traits influence people’s behavior in many important ways. For example, highly conscientious people tend to be healthier, live longer and attain greater career success.

The top five words used by each of the Big Five traits varied as follows:

For those showing neuroticism, the top five were: “awful,” “though,” “lazy,” “worse” and “depressing.”

Among extroverts, the top hits were: “bar,” “other,” “drinks,” “restaurant” and “dancing.”

Among those showing openness, the top five were: “folk,” “humans,” “of,” “poet” and “art.”

Agreeable personalities most often used these words: “wonderful,” “together,” “visiting,” “morning” and “spring.”

And conscientious personalities used these most often: “completed,” “adventure,” “stupid,” “boring” and “adventures.”

Yarkoni’s analyses identified many correlations between the Big Five traits and the rate at which bloggers used different word categories, many of which substantiated the results of previous studies. However, the study also found several unexpected correlations that initially appeared to contradict previous research.

For instance, agreeable people often used sexual words. That finding was counterintuitive, because highly agreeable people tend to be more polite and socially deferential; they usually avoid doing things that might offend others, like using sexually charged language.

It turned out that many of these counterintuitive findings had a simple explanation: the word categories that people had used in previous studies were too broad, and masked differences at the level of individual words.

The “sexual words” category, as it turned out, contained distinct clusters of words that reflected either affection (“love,” “loving,” “hugs” etc.) or sex (e.g., “gay,” “porn” etc.). Agreeable people were more likely to use the former, but much lesslikely to use the latter.

When all the words were lumped together, however, it looked like agreeable people talked more about sex.

Similarly, Yarkoni found correlations between “excitement-seeking” bloggers and words from the “school” category. Upon closer review, he found that sports-related words from the school category (“football,” “basketball,” “team” etc.) correlated with excitement-seeking. But words relating to the studious aspects of school (“books,” “desk” etc.) were not often chosen by thrill-seekers.

In other analyses, Yarkoni found more evidence that many associations between personality and word use are more specific than previous studies have suggested. For example, while previous studies found that extraverts use more words conveying positive emotions, Yarkoni’s results suggest that this is true only for some facets of extraversion.

People who report high scores on facets like cheerfulness, gregariousness and friendliness were indeed more likely to use positive-emotion words; however, people with high scores on excitement-seeking were not. In fact, if anything, excitement-seekers tended to use more negative-emotion words.

“There are a lot of neat findings like that,” Yarkoni said. “Some of them make obvious sense, and some require some story-telling. For instance, extraverts use the word “mouth” more than others do.

“That seems strange at first, but these are people who like to talk, so maybe they think of the mouth as a particularly salient feature of the social world.”

But he cautions that without more data, such “story-telling” interpretations are pure speculation, and need to be replicated by future studies.

More generally, Yarkoni notes that while his study improves upon previous studies in some respects, it also shares important limitations. Most importantly, it focuses exclusively on how frequently different words are used, without considering how those words are put together or what their context is.

In other words, no consideration is given to the underlying meaning of the text.

“Understanding gist is a very difficult problem,” Yarkoni said. “It’s one of the biggest problems cognitive scientists face: how can you get a machine to understand what a sentence really means, as opposed to just recognizing its literal interpretation? If I tell you that I can’t come to your party because I have to stay home and watch paint dry, you would probably be offended and stop talking to me, whereas a computer might conclude that you had just painted your house.”

The meaning of a word often can’t be determined without knowing its context—yet Yarkoni’s study has no way of discerning context.

Despite its limitations, Yarkoni’s study successfully confirmed and expanded on numerous previous findings.

“The results converge with other recent findings suggesting that, contrary to popular wisdom, people do not present themselves in an idealized and overly positive way online, and maintain online identities that reflect the way they genuinely see themselves and are seen by others,” he explained in the paper.

Yarkoni also sees this new study as an affirmation that relationships between personality and language need to be explored on multiple levels. Previous studies that have focused exclusively on the Big Five personality traits, and on broad word categories, may be missing some critical information.

In future work, Yarkoni hopes to focus on the relationship of personality not only to word use, but also other types of online behaviors.

“There’s actually a lot of information you can get online,” he said. “You can look at what time of day people post to their blogs; you can look at how many online connections they have. It turns out that there’s a relationship between extraversion and the number of online friends you have—even in cases where they’re people you don’t already know off-line. You can reasonably predict how extraverted someone is based on the number of online connections they’ve formed.”

Yarkoni points out that this overlap between our offline and online selves shouldn’t be too surprising.

“If you’re sociable and like to seek out people offline, you’re probably going to do the same thing online. If you complain a lot when you’re around your offline friends, you may very well complain about similar things in your online blog. Our personalities don’t dramatically change just because we’ve turned on our computers.”