They may be short but the 140-character messages posted on Twitter can tell researchers something about the state of mind of English-speakers around the world.
Using values assigned to words contained in 4.6 billion messages, also known as tweets, posted from Sept. 9, 2008, to Sept. 18, 2011, researchers were able to track how levels of happiness rose and subsided.
It turns out the last months of each year, Saturdays and the early mornings are happy times, while January, the first days of the week and late-nights are not. Holidays and individual events, like a royal marriage or the death of a celebrity, can also have significant effects on mood, the team of researchers from the University of Vermont found. Previous research examining global mood swings through the lens of Twitter has found similar patterns.
Collective mood swings
But recently, the overall trend is a gloomy one, their analysis indicated.
"After a gradual upward trend that ran from January to April, 2009, the overall time series has shown a gradual downward trend, accelerating somewhat over the first half of 2011," they wrote in a study published online Dec. 7 in the journal PLoS ONE.
Average happiness, they found, tends to increase during the last months of the years studied, dropping off in January. Happiness also rose over the weekend, peaking on Saturday, then reaching its nadir on Tuesday. [Happiest States Revealed]
Daily happiness generally peaked at 5 a.m. to 6 a.m. local time, then declined, at first rapidly, then more gradually, to the average low of 10 p.m. to 11 p.m., returning to its peak overnight, they found. These results contradict those made by other studies using information from blogs and the social networking site Facebook, the researchers note.
People tweeting after 6 a.m. are more likely to be rising for the day, rather than staying up late, leading to a change in the mental states of those tweeting, they write.
They also found that daily profanity use followed a pattern roughly opposite to that of the daily happiness cycle; for example, an analysis of five common expletives revealed that their use tended to peak during the late night, then plummet in the early hours of the morning, the hours when average happiness rose.
Good days and bad
There were also days when average happiness rose or sank out of pace with nearby dates. Among outlying dates, happiness was highest on Christmas Day, followed by Christmas Eve. In general, it appeared people became much happier on the holidays.
The only singular, non-annual event to stand out as a positive day was that of Prince William and Catherine Middleton's wedding on April 29, 2011, they wrote.
There were gloomy days, too. The bailout of the U.S. financial system, enacted in early October 2008, induced a multi-week depression among twitterers. Michael Jackson's death on June 25, 2009, produced the largest single-day drop.
Words as clues
The researchers turned to a crowd-sourcing service called Mechanical Turk — which pays people small amounts of money to complete certain tasks — and on a scale of 1 to 9 to rate the happiness conveyed by the 10,000 most common English words. When all of the scores were averaged together, the word "laughter" ranked 8.5, while "terrorist" got a 1.3, for example.
The team then applied these values to the words contained in the tweets, which also carried date and time, and sometimes other demographic information, to get a sense of the collective mood.
This approach allows researchers to look over the "collective shoulder of society," said Peter Dodds, an applied mathematician at the University of Vermont and the lead study researcher. [Twitter's Global Impact: Infographic]
"We get a sense of the aggregate expressions of millions of people," said Chris Danforth, a mathematician who worked on study, while noting twitterers are communicating in a "more natural way."
This technique allows them to look at a specific kind of happiness, they said in the study.
"There is an important psychological distinction between an individual's current, experiential happiness and their longer-term, reflective evaluation of their life," the researchers write, "and in using Twitter, our approach is tuned to the former kind."