You sit down for a cuppa with your friend. Think about the tone of the conversation and the words you use. Do you talk about family and share your positive experiences, or do you talk about politics and sports? Is your conversation warm and friendly, or cold and objective?
In a new study published in PLOS ONE, we found that gender influences the words that people use on Facebook – in both intuitive and insightful ways.
The study arose from the World Well-being Project, an interdisciplinary collaboration between psychologists and computer scientists, based at the University of Pennsylvania. As one of the team’s primary social scientists, I’ve been involved in the project over the past five years, first at U Penn and then at the University of Melbourne for the last two years.
Our project examines the language that people use on social media – like Facebook and Twitter – to study characteristics of individuals and communities.
We see differences based on personality and age. For example, extroverted individuals are more likely to talk about partying and friends, and neurotic individuals note feeling depressed and lonely. At the community level, language can distinguish regions with higher versus lower risk for heart disease.
In this new study, we analysed the language of over 67,000 Facebook users. Across a two-year period (2009-2011), these users wrote about 15.4 million status updates. They were mostly American, with several thousand from Australia, the UK and other English-speaking countries.
the methodology
Using methods from computer science, we first analysed the language and found about 1300 topics, or groups of words. For example, one topic included the words cute, baby, adorable, puppy and aww, and another topic included the words government, freedom, rights, country, political, democracy and power. Then we looked at which topics were used more on average by men versus women. The top female categories included words such as excited, adorable, family, friends and love, while the top male categories included words such as government, politics, winning, battle and football.
To take things a step further, we aligned the topics with a psychological theory that is commonly used to characterise gender differences. The interpersonal circumplex model suggests that gender differences occur along two dimensions: 1) affiliation and warmth (versus interpersonal distance and coldness) and 2) assertiveness and dominance (versus submission and passivity).
Computer algorithms automatically classified the different topics along the two dimensions. For instance, an affiliative topic included the words family, friends, wonderful, blessed, amazing, thankful and loving, while an assertive topic included party, rockin, town, poppin, club and homies.
We then considered which topics were used the most by females and which were used most by men, and how they aligned along these two dimensions.
Reflecting other research as well as common stereotypes (at least in the US), women used topics that were warm, compassionate and personable in nature, whereas the men used more topics that were cold, distant and hostile.
Equally assertive
Unlike other studies, we found that men and women were equally assertive. A look at the topics suggests that for women, this was a positive assertiveness, expressing considerable positive emotion (for example love, amazing, wonderful). For males, the assertive topics were more critical in nature, and included many more swear words.
In many ways, the topics that were most used by women versus men are not surprising. We naturally classify people into different groups, as a mental shortcut to make sense of the massive amount of information all around us. But by looking at the words themselves, it hints at how our minds make these distinctions. The computational methods make visible what the human mind does automatically to categorise the people and things that we encounter in our everyday life.
Gender is a complex, multi-faceted and fluid concept, but as a whole, the study shows that self-reported gender does influence the way people express themselves on Facebook. By bringing together computer science with psychological theory we can test psychological theories at large scale. At the same time, looking at the patterns we see in the language can help us refine our theories.
The study highlights the value of language. We were able to use technology to identify words that are warmer and colder and more or less assertive. Think about how you talk with others, or perhaps your own posts on social media. Do your words offer a sense of warmth and connection, or are you a detached observer? What words do we teach and encourage our children to use?
The words we use say a lot about our attitudes and perspectives, and influence how others think about us. As we come to understand the language, we can be more deliberate in the words we use, and perhaps have a positive impact on both our own lives and those of the people around us.