Future Now
The IFTF Blog
Language Mining is the New Health (and Marketing) Tool
I've been really enjoying James Pennebaker's new book The Secret Life of Pronouns, which provides a great, readable overview of how subtle shifts in word choice--frequently, shifts in the use of pronouns from "we" to "I"--can reveal significant differences in emotional, and consequently, physical health. What's particularly intriguing about his book is that thanks to email and other text archives, businesses, doctors and researchers will have all sorts of capabilities to understand and reach people based on these once imperceptible language choices.
Speaking with Scientific American, Pennebaker describes some of the findings:
In the 1980s, my students and I discovered that if people were asked to write about emotional upheavals, their physical health improved. Apparently, putting emotional experiences into language changed the ways people thought about their upheavals. In an attempt to better understand the power of writing, we developed a computerized text analysis program to determine how language use might predict later health improvements. In other words, I wanted to find if there was a healthy way to write.
Much to my surprise, I soon discovered that the ways people used pronouns in their essays predicted whose health would improve the most. Specifically, those people who benefited the most from writing changed in their pronoun use from one essay to another. Pronouns were reflecting people’s abilities to change perspective.
Perhaps most strikingly, these sorts of language shifts are intriguing because we don't notice them in daily interaction, but as tools to mine everything from speech to email to social media streams continue to emerge, we're on the verge of gaining an unprecedented ability to understand people's mental states through their unconscious language choices.
Pennebaker isn't the only researcher to begin to find health applications from language mining. Last year, a group of psychologists announced that they had developed an algorithm that could identify depressed bloggers by scanning archives of blog posts. The algorithm was in agreement with a panel of psychologists nearly 80 percent of the time. Similarly, a different study found strong correlations between language use and psychopathy.
The power of these sorts of tools to pull meaningful information from seeming noise points toward a need to think far more broadly about what we mean by health data and puts us in a world where e-mail and Facebook pages offer legitimately valuable health information.
Language, in other words, can be a tool for identifying who needs help. Beyond that, though, I think it offers a potential tool for understanding how to help--based on personality type.
Pennebaker, for example, has found a variety of clues to personality that we can deduce from text:
One area this is useful is in personality research. As you might guess, different patterns of function words reveal important parts of people's personalities...
Formal writing often appears stiff, sometimes humourless, with a touch of arrogance... Formality is related to a number of important personality traits. Those who score highest in formal thinking tend to be more concerned with status and power and are less self-reflective. They drink and smoke less and are more mentally healthy, but also tend to be less honest. As people age, their writing styles tend to become more formal.
Analytical writing, meanwhile, is all about making distinctions. These people attain higher grades, tend to be more honest, and are more open to new experiences. They also read more and have more complex views of themselves.
Narrative writers are natural storytellers... People who score high for narrative writing tend to have better social skills, more friends and rate themselves as more outgoing.
By watching how people use function words, we gain insight into how they think, how they organise their worlds and how they relate to other people.
Of course, at some level, this is pretty theoretical. Most of us would probably be reluctant to let our doctors and health insurers search through our Facebook pages, even as that same source of information gets re-purposed for marketing and other efforts.
Health isn't the only application of this research, of course--Pennebaker has used language mining to find relationships between pronoun use and age, social class, and social status, some of which have been long standing marketing tools, while others, like social status are just now emerging as powerful tools for targeting sales pitches.
This will put us in a somewhat disappointing place in the next decade: One where valuable information to improve our well-being is trapped, even as it gets used to sell us things we may or may not want.