Are humans inherently happy, sad, or somewhere in between? A new study suggests that, at least when it comes to our vocabulary, we tend to look on the bright side of life.
A team of mathematicians, computer scientists, and linguists at the University of Vermont and the MITRE Corporation combed through 10 languages' literature, movie subtitles, music lyrics, and, of course, Web pages and social media feeds, collecting an estimated 100 billion words from Twitter alone. The team used this data – from 24 types of sources in all – to draw up lists of the 10,000 most common words in each language.
Then, the researchers had native speakers of each language rate their emotional response to each word on a 9-point scale, from saddest to happiest. For each of the 10,000 words of each language, the scientists collected 50 ratings, for a total of about 5 million scores.
The result? Every source averaged above 5. Our words, which the study's authors describe as "the atoms of human language," reveal a "universal positivity bias."
"In every source we looked at," said UVM mathematician Peter Dodds in a press release, "people use more positive words than negative ones.”
All of the languages seemed to skew positive, but some did so more than others. In descending order of happiness, they are: Spanish, Portuguese, English, Indonesian, French, German, Arabic, Russian, Korean, and Chinese.
The team's result supports what linguists call the Pollyanna Hypothesis. First conceived in 1969 by University of Illinois psychologists Jerry Boucher and Charles Osgood, who examined data from 13 languages, the Pollyanna Hypothesis posits the existence of a universal human tendency to use positive words.
Of course, the word "microprocessor" didn't exist in 1969, and Google's founders had yet to be born. By necessity, Boucher and Osgood could collect only a tiny sample of each language's overall textual output. Their work, write the authors of this new study "could only be regarded as suggestive."
The authors' Big Data approach, by contrast, not only offers more robust qualitative support for the Pollyanna Hypothesis, but it also can be used to measure a population's shifting moods. In 2013, the researchers found that the positivity of geographically-tagged English-language Twitter feeds closely tracked Gallup's well-being surveys. The authors' Hedonometer collects random samples of 10 percent of Twitter's daily output, filters out non-English tweets, and uses them to create a graph showing the overall positivity of the Twitterverse.
According to the Hedometer the two saddest days in 2014 were January 23, which saw the arrest of Justin Beiber in Miami, and November 24, which saw major protests against police in Ferguson, Ohio. The happiest day was Christmas. Overall, however, every day of 2014 was more positive than not on Twitter. (The Hedometer can also be used to track the emotional valence of fictional stories, such as Moby Dick.)
Still, just because we tend to use words associated with happiness, does that mean that we actually are happy? What's the relationship between our words and our thoughts? Are the limits of our language, at Wittgenstein famously said, the limits our world? Or do we have, in addition to our public languages, a kind of private "mentalese," whose words, if they can be called that, carry the true content of our feelings, distinct from what we say or write out loud?
Linguistics has gone back and forth on this. One of the best-known proponents of linguistic relativity, that is, the belief that the structure of a language drives how native speakers of that language think, was the self-taught linguist Benjamin Lee Whorf, who in the 1930s argued that differences between the grammars and vocabularies of languages account for differences in the way speakers of those language conceptualize the world. Whorf's hypothesis was largely repudiated by Universalist linguists beginning in the 1960s, who, following MIT linguist Noam Chomsky, described evidence suggesting that all languages share the same underlying structure. Relativity, however, has made something of a resurgence in recent decades.
The authors of this most recent study acknowledge both camps, but avoid explicitly taking sides. They do write, however, that their findings suggest a "a deep imprint of human sociality in language."
"Human language reveals a universal positivity bias" appears in the current edition of the Proceedings of the National Academy of Sciences.