Is language a vehicle for happiness?

Sep 2, 2011 by

In the last few postings, we’ve discussed several examples of “amateurish linguistics”, chiefly involving the assumption that language is a cause rather than vehicle of conceptualizing the world in a certain way. Here is another example that comes not from laymen and not from professional linguists, but from scholars in another field trying to “do linguistics”.

A group of mathematicians from Cornell University and the University of Vermont conducted a large-scale study of texts in order to find out if English speakers use more positive or negative words. This is how it worked: the researchers examined huge amounts of English texts: 3.29 million books on Google Books, 20 years worth of New York Times, 2 years worth of tweets and the lyrics of 295,000 popular songs. For each of the sources they compiled a list of the 5,000 most-used words and produced a master list of 10,122 words. Each word has been evaluated by 50 people (using Amazon’s Mechanical Turk labor-outsourcing service) who scored each word from negative to positive on a scale of 1 to 9. The word “terrorist”, for example, received an average score of 1.30, while “laughter” merited an 8.50, the highest of any word. The results: altogether, positive words outnumbered the negative, and were used more frequently.

Great! So English speakers overall are an upbeat, optimistic, positive-thinking crowd. I could tell you that without spending a dime on a study. But this group of researchers jumped to far-reaching conclusions about human nature. One of the principal researchers, the University of Vermont’s Isabel Klouman wrote that the findings

“suggest that a positivity bias is universal… In our stories and writings we tend toward pro-social communication.”

(Pro-social behavior includes sharing, altruism, etc.)

Honestly, I don’t see how this conclusion is supported by the actual study. How is it that Dr. Klouman sees the positivity bias as universal if the only language that was studied is English (biased towards American English at that!). What about other languages? Would this positivity bias be found in other languages as well? Or are there “positive” and “negative” languages? That’s a big question.

But in addition to this grand posturing, there are significant problems with this type of what I call “accounting linguistics”, a branch of “amateurish linguistics”. Applying mathematical methods to language is all well and dandy. But math must be a tool to understanding linguistic issues, not a self-serving goal. And a solid foundation of basic linguistics is necessary before any mathematical methods can be applied. (By the way, I wonder if these researchers consulted with their colleagues in the linguistics department, such as the one at Cornell, one of the best linguistics departments in the country).

But from the point of view of basic linguistic knowledge, the study by Klouman and her colleagues is rather flawed. One thing to note is that even though connotation in general is an important part of a word’s meaning, most words have a neutral connotation. In fact, the most frequent words of any language — which includes such grammatical words as conjunctions (and), prepositions (in, on), pronouns (I, he), negation markers (not), as well as verbs (to be), to speak, to know) and nouns (man, day, house) — are neutral in connotation. Second, many words receive a connotation only in context. Take the word bear, for example. In the context of a zoo or a plush toy, it’s a positive word, but in the context of hikers in Alaska — not so much!

Even synonym pairs that seem to differ in connotation, such as hide and conceal, thrifty and stingy, overweight and fat, often have additional differences, whether in the denotation component of meaning (e.g., a fat person can be much heavier than someone who is just overweight) or in register/style. Take, for example, the following quote from Joseph Brodsky’s Watermark:

“A smell is, after all, a violation of oxygen balance, an invasion into it of other elements… Depending on that invasion’s intensity, you get a scent, a smell, a stench.”

Note that synonyms of smell differ in connotation: for instance, scent is positive and stench is negative. But they also differ in their denotational meaning: scent is more subtle than stench. Synonyms of smell also differ with respect to their register: compare, for example, effluvium (higher register) and stink (lower register). And the word smell itself is rather neutral in connotation, outside of context, that is.

Given the need to control for context, I attempted to replicated the study of Dr. Klouman and her colleagues with just a pair of words that differ (mostly) in their connotation. Also, to test their universal positivity bias conclusion, I’ve chosen a pair of words from Russian, a language (and culture) often blamed for a negative bias. The two words are: shpion and ravedchik, both meaning ‘spy’. The former word has a strong negative connotation and is most often applied to an enemy spy, whereas the latter word has a strong positive connotation and is typically applied to ‘our spy’. In short, shpion is a bad guy and ravedchik is a good guy. For example, Francis Gary Powers, an American spy shot down by the Soviets in 1960, is a shpion, and Willie Fisher, aka Rudolf Abel, who was exchanged for Powers in Berlin 1962, is a ravedchik.

If Klouman et al. are correct in saying that the positivity bias is universal, we expect to find that ravedchik is more frequent in Russian texts than shpion. But quite the opposite is the case. A Google search brought back 12,900,000 hits for shpion and only 7,220,000 hits for ravedchik. The frequency bias towards the negative word is even more pronounced in Yandex searchers (is it because Yandex is a Russian-made search engine, and Google an American-made one?): 17 million hits for shpion and only 7 million hits for ravedchik. Finally, a search in the National Corpus of Russian brought back 1,447 hits for shpion and only 838 hits for ravedchik. The clear frequency bias towards shpion is particularly interesting in light of the fact that only the word ravedchik can be applied to an inanimate ‘spy’: for example, the U-2 plane that Gary Powers flew of the USSR is samolet-razvedčik (literally ‘aircraft-spy’).

And speaking of connotation and spies, let me conclude with a quote from a spy thriller Die another day:

James Bond: I’m looking for a North Korean.

Raul: Tourist?

James Bond: Terrorist.

Raul: One man’s terrorist is another man’s freedom fighter.

P.S. A quick Google search revealed that the negative terrorist trumps the positive freedom fighter by nearly 13 to 1 (96,500,000 hits for terrorist and 7,550,000 hits for freedom fighter). That much for the positivity bias there!

Subscribe For Updates

We would love to have you back on Languages Of The World in the future. If you would like to receive updates of our newest posts, feel free to do so using any of your favorite methods below: