We might be told that life expectancy is rising, poverty and disease are falling, and that even crime and violence are on the decline, but it's common to feel the world is getting worse, and not just this year. Why else would people vote to restore past greatness? Linguistic analysis suggests that these perceptions are creeping into Americans' language, and the trend has been going on for two centuries.
Linguists have noticed that, across many languages, people use more words with positive associations than negative ones. The phenomenon was initially dubbed the “Pollyanna hypothesis”, based on the theory that it indicated a tendency to look on the bright side of life, but it's now been renamed the “Linguistic Positivity Bias” or LPB.
In the Proceedings of the National Academy of Sciences, a team led by Dr Rumen Iliev of the University of Michigan decided to test if the strength of LPB is affected by external conditions. A selection of 907 words were classified as positive (examples given include “clever”, “confident”, and “grace”) or negative (“fear”, “hatred”, “grief”). The authors then examined Google Books' 1.3 million titles published by Americans between 1800 and 2000, seeking the number of times words from each category appeared. Similar analysis was applied to issues of the New York Times from 1851 to 2015.
Google Books' ratio fell from almost 2:1 to 1.6:1, and the trend in the Times was in the same direction, albeit much less steep.
If the trend had been consistent, this might simply indicate linguistic changes unrelated to the national mood, so Iliev sought out more short-term wiggles in the data, examining what happened in times of national crisis such as economic depression or war.
The Civil War and both World Wars were associated with sharp drops in the LPB in both books and newsprint, indicating that LPB really does measure general mood. A weaker association was seen with economic conditions.
If the LPB falls in hard times, what does it mean when the long-term trend is so clearly downward? Such a question is beyond the scope of Iliev's paper, but opinion polls asking people to assess their subjective happiness have been conducted since 1946, and these too show a correlation with the LPB.
Nevertheless, the last few years of the 20th century saw a modest recovery among the books Google has made available, and the New York Times database suggests that this may have continued in recent times (a drop possibly associated with 9/11 aside).
As the first published study to consider LPB changes with time, the paper opens up the opportunity for similar research on other countries to see if growing pessimism is a uniquely American feature.