How language can shape our attitudes and behaviours

Consider an example from the insurance world. Back in the 1930s, executives at the Hartford Fire Insurance Company in Connecticut realized that warehouses which contained oil drums kept blowing up. Nobody knew why. The company asked a fire-prevention engineer named Benjamin Whorf to investigate. Although Whorf was a trained chemical engineer, he had also done research in anthropology and linguistics at Yale, with a focus on the Hopi Native American communities. So, he approached the problem with an anthropologist’s mindset: he observed warehouse workers, noting what they did and said, trying to absorb everything without prior judgment. He was particularly interested in the cultural assumptions embedded in language, since he knew these could vary. Consider seasons. In English, “season” is a noun, defined by the astronomical calendar (“summer starts on June 20,” people say). In the Hopi language and worldview, “summer” is an adverb defined by heat, not the calendar (it feels “summer(y)”). Neither is better or worse; but they are different. People cannot appreciate this distinction unless they compare. Or as Whorf observed: “We always assume that the linguistic analysis made by our group reflects reality better than it does.”

This perspective solved the oil drum mystery. Whorf noticed that the workers were careful when handling oil drums marked as “full.” However, workers happily smoked in rooms that stored drums “empty.” The reason? The word “empty” in English is associated with “nothing”; it seems boring, dull, and easy to ignore. However, “empty” oil drums are actually full of flammable fumes. So, Whorf told the warehouse managers to explain the dangers of “empty” to workers and explosions stopped. Science alone could not solve the mystery. But cultural analysis-with science-could. The same principle (namely using antho-vision to see what we ignore) is equally valuable when mysterious problems erupt in modern bank trading floors, corporate mergers, or pandemics, say.

That is because, “the least questioned assumptions are often the most questionable,” as the nineteenth-century French physician and anthropologist Paul Broca reputedly said. It is a dangerous mistake to ignore the ideas we take for granted, be that about language, space, people…

“Empty” is printed on the drum. A worker reads “harmless”. The flammable vapour remains.

People respond to what a word normally implies during the task. Empty means harmless, available, finished. Premium means better. Basic means deficient. Cancel means losing something. Those associations can direct action before anyone begins a careful analysis. Product teams often defend a confusing label by proving it is technically correct. Lovely. The customer has still pressed the wrong control. Internal accuracy achieves little when the ordinary interpretation repeatedly produces danger or failure.

The useful research happens at the moment of use. Ask what people expect the label to do, then watch what they do next. If interpretation varies, add the concrete state or consequence. “Empty—flammable vapour” changes the relevant meaning. “Delete permanently” is clearer than a cheerful bin icon. “Pause payments” should say what happens to access and existing funds. Training, equipment and incentives also shape behaviour; in hazardous settings a better label works alongside proper controls. Small words are often left untested because insiders stopped noticing them years ago.

In the storage room, the revised label now states the risk directly before anybody acts on the ordinary meaning of “empty”. A worker approaching with a cigarette sees “flammable vapour” at the same moment they see the drum. The altered words have changed the immediate decision.