Ů��С��Ƶ

The words used in millions of books show how US culture has generally become more permissive, leading to both innovation and risky behaviour.

That is according to an analysis of books published in the US between 1800 and 2000. However, the study overlooks the fact that books only encapsulate certain sections of society.

Who gets to write books is far from a level playing field. In 2016, 86 per cent of authors and writers in the US were white, according to the US Census Bureau. Many groups, including women, people who are poor and the LGBTQ community, are also under-represented. Publishing has mostly been the preserve of a privileged few, so using it to make generalisations about cultural norms doesn’t paint the full picture.

In the study, Joshua Jackson at the University of North Carolina at Chapel Hill and his colleagues searched through 155 billion words from books published in the US for clusters of words they associated with tighter cultural conventions, such as “prevent” and “uniformity”, and those associated with looser norms, such as “openness” and “encourage”.

They used an algorithm to identify passages where any of eight key words in both categories occurred close to one another, and then extracted the final list of “tight” and “loose” words. They then looked at how often these words were used over time, associating the trends with changes in societal norms.

But language isn’t so straightforward. The word “prevent” can indeed be used to call for banning some action – say, electing a woman as president – but it can also be used to ask why anyone would prevent such a thing. We also use language sarcastically or to joke. Algorithms struggle to make such distinctions.

Jackson says that looking at language over such a long period of time, and studying how these words are used together, can negate contextual idiosyncrasies.

Increasing creativity

He and his team found an association between loosening norms over time and societal changes. They found that creativity increased, as judged by people choosing less common names for their children, and more patent registrations, trademark applications and feature films being produced.

They also found decreases in conformity or societal order, as measured by higher rates of teen pregnancy and household debt, and fewer teenagers attending high school.

It is unclear whether these proxies for creativity and social order are accurate. The recent explosion of movies, for example, includes many remakes, so could instead be a sign of risk-averse capitalism not creativity.

The authors admit the potential flaws in their work. “It should be noted that linguistic trends may not be indicative of cultural trends more generally,” they write, though Jackson says this can be overcome by combining linguistics with data on other cultural trends.

The idea that book publishing can capture culture at large is unconvincing. Jackson says he and his team are planning to analyse local newspapers, as they have been published for a long time and may capture wider cultural norms. Although many of the same problems are likely to remain.